STRESS CORROSION TESTING A symposium presented at the Sixty-ninth Annual Meeting AMERICAN SOCIETY FOR TESTING AND MATERIALS Atlantic City, N. 1, 26 June-1 July, 1966 ASTM SPECIAL TECHNICAL PUBLICATION NO. 425 List price $25.00; 30 per cent discount to members published by the Copyright by ASTM Int'l SOCIETY (all rights FOR reserved); Wed Dec 16 MATERIALS 15:53:43 EST 2015 AMERICAN TESTING AND Downloaded/printed by 1916 Race Street, Philadelphia, Pa. 19103 University of Washington (University of Washington) pursuant to License Agreement. No furth © BY AMERICAN SOCIETY FOR TESTING AND MATERIALS 1967 Library of Congress Catalog Card Number: 67-20038 NOTE The Society is not responsible, as a body, for the statements and opinions advanced in this publication. Printed in Baltimore, Md. Copyright by ASTM Int'l (all rightsDecember, reserved);1967 Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduct Foreword The Symposium on Stress Corrosion Testing was presented in four sessions during the 69th Annual Meeting of the Society, in Atlantic City, N. J., 26 June-1 July, 1966. The symposium was sponsored by Committee G-l on Corrosion of Metals. The symposium chairman was H. Lee Craig, Jr., Reynolds Metals Co. Presiding at the four sessions were M. A. Streicher, E. I. du Pont de Nemours & Co., Inc.; M. G. Fontana, The Ohio State University; G. J. Danek, Jr., U. S. Naval Marine Engineering Laboratory; and W. F. Gerhold, National Bureau of Standards. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductio Related ASTM Publications Stress-Strain-Time-Temperature Relationships in Materials, STP 325 (1962), $5.25 Stress-Corrosion Cracking of Titanium, STP 397 (1966), $14.00 Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further rep Contents Introduction Stress Corrosion Testing Methods Influence of Stress and Temperature on Short-Transverse Stress Corrosion Cracking of an Al-4.2Zn-2.5Mg Alloy—W. J. HELFRICH . . A Comparison of Three Precracked Specimens for Evaluating the Susceptibility of High-Strength Steel to Stress Corrosion Cracking— C. D. BEACHEM AND B. F. BROWN Application of an Accelerated Stress Corrosion Test to Alloy Development—E. E. DENHARD, JR., AND R. R. GAUGH Discussion A Rapid Stress Corrosion Test for Aluminum-Magnesium Alloys— H. L. CRAIG, JR., AND H. B. ROMANS Discussion Influence of Environment on Crack Propagation Characteristics of High-Strength Aluminum Alloys—J. H. MULHERIN Discussion Stress Corrosion Cracking of High-Strength Bolting—C. S. LIN, J. J. LAURILLIARD, AND A. C. HOOD Discussion Stress Corrosion of High-Strength Steel Alloys—Environmental Factors—A. GALLACCIO AND M. A. PELENSKY Stress Corrosion of Magnesium Alloys—Environmental Factors— 21 31 41 50 51 65 66 81 84 98 99 107 M. A. PELENSKY AND A. GALLACCIO Discussion Resistance of Ferritic Stainless Steels to Stress Corrosion Cracking— A. P. BOND, J. D. MARSHALL, AND H. J. DUNDAS Discussion Some Techniques Used in the Study of Stress Corrosion Cracking— H. L. LOGAN Discussion A Proposed Mechanism for the Stress Corrosion Fracture of a CopperBeryllium Alloy—W. D. SYLWESTROWICZ A Quantitative Stress Corrosion Test for Al-Zn-Mg Alloy Plate— Stress 1 3 115 116 125 127 142 145 H. ROSENTHAL AND H. R. PRITCHARD . 165 ROMANS 182 Corrosion Test Environments and Test Durations—H. B. Reactions Contributing to the Formation of Susceptible Paths for Stress Corrosion Cracking—D. A. VAUGHAN AND D. I. PHALEN . 209 Discussion 227 Critical Species in Stress Corrosion Phenomena—E. N. PUGH AND A. R. c. WESTWOOD 228 Circulating Autoclave System for Stress Corrosion Cracking Studies— R. W. STAEHLE Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 248 15:53:43 Stress Corrosion Cracking Rates of a Nickel-Brass Alloy Under Downloaded/printed by Applied Potential—NANCY MCKINNEY AND H. W. pursuant HERMANCE . .License . 274 University of Washington (University of Washington) to EST 2015 Agreement. N vi CONTENTS Discussion 291 Reporting and Evaluating Stress Corrosion Data—D. O. SPROWLS .... 292 Evaluation of Various Techniques for Stress Corrosion Testing Welded Aluminum Alloys—M. B. SHUMAKER, R. A. KELSEY, D. O. SPROWLS, AND J. G. WILLIAMSON Stress Corrosion Testing of 7079-T6 Aluminum Alloy in Various Environments—B. W. LIFKA AND D. O. SPROWLS Environmental Factors Affecting the Stress Corrosion Cracking Behavior of an Aluminum-Zinc-Magnesium Alloy—H. B. ROMANS AND H. L. CRAIG, JR 342 363 Discussion Copyright Downloaded/printed University 317 379 by by of STRESS CORROSION TESTING Introduction A recent survey by ASTM Committee G-l revealed that nearly one hundred laboratories are engaged in stress corrosion testing. This survey was part of a task group effort to determine the state-of-the-art concerning stress corrosion testing, organized into three areas: specimen design; test environments and duration; and data evaluation and reporting. The committee decided to provide a forum at which current practices would be presented and discussed by those active in the field. Papers were solicited from engineers and scientists who use these tests for specification purposes, for material selection guides, and for design and use criteria. Theoretical papers were included where the theory was shown to influence test methodology. The theme of the conference was the development of reproducible, standardized test methods. Nearly all structural metals were covered: high strength steels, stainless steels, and alloys of aluminum, copper, magnesium, and zirconium. Some new methods were fully discussed, including the precracked cantilever beam and tests on joints, such as bolts and weldments. The breadth of coverage shows the wide interest in this subject. Despite its longevity as a metallurgical phenomenon, stress corrosion cracking mechanisms are still being investigated. Three previous symposia have concentrated on this problem: in Philadelphia (1944), in Boston (1954), and in Pittsburgh (1959). Sufficient time has elapsed since the last one for those working in the field to realize that the complex nature of the problem has not yet yielded to the ever-growing amount of scientific and technical investigations into the causes. The designer, for. one, is faced with materials selection problems. Yet he must rely upon stress corrosion tests which are highly empirical and only poorly correlated with service performance. To select an example with which the writer is familiar, take highstrength aircraft alloys. Originally, these alloys were fabricated in sheet form. Stress corrosion tests, among others, were used to develop optimum alloys and tempers for the strength levels desired. Adequate service performance has been obtained from these materials. When designers and builders went to different forms of construction, using thick sections, Copyright by ASTM Int'l rights reserved); Wed Dec 16 15:53:43from EST 2015 they expected the(allsame adequate behavior these alloys. However, in Downloaded/printed by 1 to License Agreement. No further reproductions authorized. University of Washington (University of Washington) pursuant 2 STRESS CORROSION TESTING a few cases, it was not achieved, and problems with stress corrosion cracking grew in number with each new model aircraft. An unsuspected sensitivity to stress corrosion cracking in the short-transverse direction was discovered. Stress corrosion cracking was not well enough understood to foresee these problems. In addition, test methodology was such that development programs required months of testing. Designers are now able to deal with the shortcomings of these alloys hi a satisfactory fashion. This symposium reveals many such problems. The amount of work which is suggested on the basis of current investigations is staggering, if the desire to bring rigor into stress corrosion testing is to be fulfilled. The urgent and critical needs of industry, the defense effort, and the space program compel us to undertake this goal. From this point, or state-ofthe-art, the committee plans to launch its effort in fulfilling these needs. The task group for the symposium would like to acknowledge the efforts of the many persons who prepared papers, those who reviewed them, and the contribution of the ASTM staff for their help in making the symposium a significant step forward in fighting the problem of stress corrosion cracking. H. Lee Craig, Jr. Research supervisor, Department of Applied Chemistry and Mathematics, Reynolds Metals Co., Richmond, Va.; Chairman, Subcommittee VI on Stress Corrosion Testing and Corrosion Fatigue of Committee G-l on Corrosion of Metals Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprodu STRESS CORROSION TESTING METHODS REPORT OF TASK GROUP 1 ON STRESS CORROSION TESTING METHODS OF SUBCOMMITTEE X ON STRESS CORROSION OF COMMITTEE B-3 ON CORROSION OF NONFERROUS METALS AND ALLOYS* Stress corrosion tests are conducted for a variety of reasons, the reason frequently determining the type of test. Some of the reasons are: (1) Evaluate a metal or alloy, or various heat treatments of one alloy for susceptibility to stress corrosion in certain environments. (2) Compare stress corrosion susceptibilities of various alloys. (3) Evaluate environments which might accelerate stress corrosion cracking in various alloys. (4) Evaluate a specific service requirement with regard to the possibility of stress corrosion. (5) Evaluate the effectiveness of coatings or other protective measures for reducing stress corrosion of susceptible metals and alloys. cur, but complete fracture from stress corrosion also may not occur because of stress relaxation during the earlier stages of cracking. For (1) and (2), both types of tests should be conducted in order to obtain a complete picture of the susceptibility of an alloy to stress corrosion, preferably using specimens oriented in more than one direction with respect to the rolling direction. Behavior in the short-transverse direction is especially important for study. For (3) and (5), the most important considerations should be selection of a specimen type and a stressing method which provide reproducibility of stress pattern and stress level from one specimen to another, and selection of a material known to be susceptible to stress corrosion cracking. In materials that are notch sensitive, care should be Stress corrosion tests may be con- taken to eliminate notches, scratches, ducted under conditions of either con- etc., which might act as stress raisers. stant deflection or of constant load. An For (4), the part should be tested unimportant difference between them is der the type of stress it will encounter in that under constant load general corro- service. Table 1 lists some types of stress sion may cause a decrease in cross-sec- that prevail in service with some recomtional area with subsequent increase in mendations as to whether constant destress per unit area. This can lead to flection or constant load would more mechanical rather than stress corrosion nearly duplicate the stress condition. failure. Under constant deflection, the For (3), (4), and (5), specimens taken stress per unit area tends either to stay out in the short-transverse direction are constant or decrease. Therefore, failure preferable, since this is the direction by stress rupture is not as likely to oc- which shows the greatest susceptibility. The type of specimen to be used is * Reprinted from Proceedings, Am. Soc. Testing Mats., Vol. 65, 1965, pp. 182-197. frequently determined by the form in Committee B-3 has been replaced by Commit- which the metal is available (sheet, plate, tee G-l on Corrosion of Metals. Subcommittee by ASTM Int'l (allCommittee rights reserved); 16 15:53:43 EST 2015 bar, Dec tubing, wire, etc.). Table 2 relates XCopyright is now Subcommittee VI of G-l. Wed Downloaded/printed by 3 University of Washington (University of Washington) pursuant to License Agreement. No further reprod 4 STRESS CORROSION TESTING TABLE 1 Source of Sustained Tension in Service Residual stresses: Quenching! Forming / Misalignment Interference Bushings: Rigid Flexible Flareless fittings \ Clamps / Stressing Method Most Applicable constant deflection constant deflection constant deflection constant load either, but constant defection probably better Hydraulic pressure. . . constant load Dead weight constant load Faying surface corrosion constant load TABLE 2 Form in Which Material Is Available Possible Types of Specimens Sheet . bent beam, preform, tension specimen, Ubend Plate (less than 2 in. thick) C-ring, bent beam, tuning fork Plate (more than 2 in. thick) short-transverse tension specimen Bar (depending on thickness or diameter) C-ring, tuning fork transverse tension specimen Tubing C-ring, (Battelle system of internal pressure) Wire tension specimen, loop SPECIMENS FOR SIMULATING SPECIFIC CONDITIONS 1. Preformed specimens for simulating residual stresses 2. Welded assemblies 3. Interference rings for simulating pressed in bushings, fasteners, etc. specimen types to material form. Some of the most common types are described below. Bent Beam Specimen (Constant Deflection) : point loading; or by holding the ends in a rigid jig (snap-in bent beam). All three methods require jigs. The snap-in bent beam is the most commonly used because it requires the simplest fixture. The stress pattern and calculation are, however, the most complicated. If the broad assumption is made that the specimen makes a truly circular arc, then the theoretical relationship between stress on the outer fibers and specimen length can be calculated by Method 1-A described later. The calculated stress is always lower than the actual stress, and, if an accurate value is desired, strain gages should be used for stress measurement. A formula is also available which takes into account the fact that the arc inscribed by the specimen may not be truly circular. This is given in Method 1-B. A detailed stress analysis of the bent beam specimen has been published elsewhere.1 A photograph of the snap-in bent beam is shown in Fig. 1. Specimens that are stressed at low levels may be too short for easy insertion into the stressing jig, and a stressing device may be useful. Such a device is also shown in Fig. 1. A fixture employing four-point loading provides uniform distribution of stress between the two inner supports. The formula for calculating the deflection is given in Method 1-C. With three-point loading, a marked localization of the maximum stress exists at the central support. The formula for this is given in Method 1-D. This type of loading has been employed in an electrolytic test to determine rapidly the susceptibility to stress corrosion cracking of aluminum-magnesium This type of specimen may be stressed 1 G. Wed Haaijer and W. Longinow, Corrosion, Copyright by ASTM loading; Int'l (all rights Dec 16A. 15:53:43 EST 2015 in three ways: four-point three-reserved); Vol. 21, No. 4, April, 1965, pp. 105-112. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No furth REPORT OF COMMITTEE B-3 5 FIG. 1—Stressing device and jig for bent beam specimen. alloys. Details of this test are given in stant-deflection specimen useful for multiple testing has also been described.2 Method 1-E. Three- and four-point loading may also Tension Specimens (Constant-Deflection be of the constant-load type if the deor Constant-Load, Depending on the flection is accomplished by application Stressing Method}: of a dead load. Either flat or round tension specimens Precautions should be taken to insure that no galvanic corrosion can take place are suitable for stress corrosion testing. between the specimens and the supports. Direct loading may be accomplished by 2 G. J. Heimerl and D. N. Braski, "A Stress Also, with bent beams it is important Corrosion Test for Structural Sheet Materials," that the edges be broken. Materials Research & 16 Standards, Vol.EST 5, No. 1, Copyright by ASTM Int'l (all rights reserved); Wed Dec 15:53:43 2015 A self-loading, bent-beam type of con- January, 1965, pp.18-22. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. N 6 STRESS CORROSION TESTING use of compression springs, lever arms, or the tension rings shown in Figure 3. As pointed out by Godard and Harwood,3 however, dead loading by means of levers is difficult and requires proper placement and alignment of pivots to ensure that the desired load is being applied. With this type of test, uniaxial stress is increased by crack nucleation and propagation due to reduction in crosssectional area. Therefore, the rate of crack propagation should increase in a tension specimen. The increase will continue under constant load, while under constant strain the presence of small FIG. 2—Specimen holder for electrolytic test. The load to be applied is measured by calibrating the compression springs. The tension rings can either be calibrated for deflection under load, or a strain gage can be placed on the bolts and the bolt calibrated for strain. Constant deflection can be obtained by means of the fixture shown in Fig. 4. The | in. diameter tension specimen shown in Fig. 4 is frequently used for testing susceptibility of aluminum plate, forgings, and castings to stress corrosion in the short-transverse direction. The strain to be applied is best measured by a strain gage or an extensometer on the specimen. cracks may tend to cause relaxation in the specimen. C-Ring Specimen: The C-ring is useful for making transverse tests of a wide variety of products. If possible, the C-rings should be oriented in such a way that the direction of maxialso mum tensile stress will be parallel to the short-transverse direction of the piece. The recommended orientation and a drawing of this specimen are shown in Fig. 5. The formula to be used for stressing the C-rings is given in Method 2-A. A specimen similar to a C-ring but requiring less machining is shown in Fig. 3 6a. As the legs are drawn together by H. P. Godard and J. Harwood, Copyright by J. ASTM Int'l (allCorrosion, rights reserved); Wed Dec 16 15:53:43 EST 2015 February, 1955, p. 97t. tightening the bolt, the strain is measured Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No furth REPORT OF COMMITTEE B-3 with a strain gage mounted in the center of the surface being placed in tension. Two other types of "tuning fork" specimens are shown in Figure 6b. Formulas that can be used for stressing these specimens are given in Method 2-B. 7 posed, the face or root sides of both strips are stressed in tension, otherwise the face and root sides of the weld are stressed against each other (Fig. 8). The applied stress does not take into account Sheet-Type Preform Specimen: Sometimes methods of fabrication or assembly induce residual stresses in the metal. The effects of these stresses on corrosion can be determined by exposing the fabricated or assembled parts to a suitable corroding medium. Stresses induced by fabrication or assembly are likely to be of greater magnitude than the design stresses. One type of specimen that may be used to evaluate the effects of stresses induced by cold work is termed the "preformed" specimen (Fig. 7). A high elastic bending stress is superimposed upon residual stresses introduced by forming, giving a total sustained stress that is relatively high but not readily determined. In this specimen, a depression is formed in the center of a strip of metal 9 in. long machined to a width of 0.700 ± 0.005 in. The formed specimen can then be exposed without further treatment, or else the ends can be forced together and sprung into a jig or stressing plate with a 7f in. span. This latter condition is generally used and is a particularly severe test since it involves both residual and external stressing. Welded Beam Assemblies: FIG. 3—Tension rings for loading tension specimens. A bent-beam specimen can be used for testing butt welds. For such investigathe stress concentrations accompanying tions, strips full thickness by 1 in. wide by 8f ft long are sawed and machined the variations around the weld bead. The formula for calculating the defrom large welded panels. flection is given in Method 3. The specimens are stressed in pairs against each other using H-type saddles Interference Ring Specimen: and half-rounds and drawing up the The interference ring type of specimen bolts and nuts at both ends of the speciCopyright by ASTMspecimens Int'l (all rightsare reserved); 15:53:43 EST 2015 canDec be 16 used in special cases where it is men. When replicate ex- Wed Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions 8 STRESS CORROSION TESTING desired to develop a hoop stress in a particular part and where other more economical specimens are not suitable. This ring is particularly advantageous, for example, in circular die forged parts where the most critical grain structure tails of the method and the formula for stressing are given in Method 4. Tubing Specimen: A specimen design and jig for stress corrosion testing of tubing under con- FIG. 4—Tension specimen and stressing fixture. exists at the surface of the part. Other stant load is shown in Fig. 10. This advantages are that a relatively large method can be adapted to study a variety surface area of metal is placed under a of alloys under either aqueous or gaseous uniform tensile stress, and the assembled environments. Details of the method and unit simulates practical situations on the formula for calculating the hoop structures containing various types of stress are given in Method 5. interference-fit components. U-Bend Specimens: The ring is stressed by pushing it onto The Wed specimens have Copyrightplug. by ASTM Int'l (all Dec 16described 15:53:43 so ESTfar2015 an oversized A stressed ringrights and reserved); Downloaded/printed by in Fig. 9. De- involved stresses in the elastic region. stressing* tools are shown University of Washington (University of Washington) pursuant to License Agreement. No furth FIG. 5—C-ring specimen. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuantFIG. to License Agreement. No further reproductions authorized. 6—Tuning fork specimen. REPORT OF COMMITTEE B-3 U-bends are formed by plastic deformation from sheet stock. Tensile stresses are applied by closing the ends of the U with a nut and bolt. The U-bend specimen is highly sensitive to stress corrosion crack- 11 bend specimen. A detailed discussion of the U-bend has been published elsewhere.4 With very high-strength alloys it is sometimes impossible to form a U-bend FIG. 7—Preformed sheet specimen. FIG. 8—Welded beam stress corrosion assembly. ing, but the test is mainly considered to without fracturing the specimen. Such be qualitative since it combines a rela- materials can be bent to 150 deg in the tively unknown stress condition with a annealed condition and then heat treated relatively unknown condition of work before drawing the legs parallel. Care hardening. These specimens frequently 4 N. Wed Nathorst, Corrosion Stainless Copyright by ASTM Int'lthe (allpoint rightsof reserved); Dec "Stress 16 15:53:43 ESTof2015 crack at some distance from Steels," Welding Research Council Bulletin Downloaded/printed by highest load, that is, the top of the U- Series, No. 6, October, 1950. University of Washington (University of Washington) pursuant to License Agreement. No further 12 STRESS CORROSION TESTING PRESSURE AND CORRODENT INSIDE TUBE FIG. 9—Stressing tools for interference ringtype specimens. should be taken to avoid springback before closing the ends with a bolt. A stressing device to prevent springback has FIG. 10—Jig and specimen for stress corrosion testing of tubing under internal pressure. been described elsewhere.5 It is also important that edges be broken for this type of specimen. FORMULAS FOR STRESSING BENT BEAMS METHOD 1-A Stress-Strain Relationship: where: e = outer fiber strain, Copyright by ASTM Int'l (all rights reserved); Wed Dec 15:53:43 2015and W. L. Pearl, 6 D. C, 16 Vreeland, G.EST G. Gaul, Downloaded/printed by Corrosion, Vol. 18, No. 10, p. 368t. University of Washington (University of Washington) pursuant to License Agreement. No further reproduc REPORT OF COMMITTEE B-3 a = outer fiber stress, psi, and E = modulus of elasticity, psi. Strain on Outer Fibers When Flat Specimens Are Bent into Simple Arcs: where: t = material thickness, in., and r = radius of curvature, in. 13 h = distance between outer fiber and neutral axis, and rm = minimum radius of curvature of the bent beam. where: t = thickness of specimen and y = maximum deflection of the bent beam. FIG. 12—Four-point beam loading. FIG. 11—Diagram for calculation of specimen length. • Calculation of Specimen Length (see Fig. FIG. 13—Three-point beam loading. When y is much greater than t: 11): and where: S = specimen length, in., / = jig length, in., and 9 = central angle, radians. METHOD 1-B The basic equation for determining the stress, (r, in the outer fiber of a bent beam specimen is: The curve of a beam, bent by restricting the ends, closely assumes a parabolic shape. The semi-latus rectum of a parabola is the minimum radius of the curve. The length of the arc of a parabola which closely approximates the length, /, of the specimen may be calculated from the equation: where: Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 E = Downloaded/printed modulus of elasticity, by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions 14 STRESS CORROSION TESTING FIG. 14—Correction factor for stresses in C-ring specimen. where: d = jig span and rm = Et/2(T has been substituted into the equation. METHOD 1-D Three-Point Loading Stress Calculations for Bent Beams (Outer Fibers) (See Fig. 13): METHOD 1-C Four-Point Loading Stress Calculations for Bent Beams (Outer Fibers) (see Fig, 12): Symbols and units are the same as for Method 1-C. METHOD 1-E This test involves stressing a 3-in. where: strip cut from heavy sheet or plate as a y = deflection, in., simple beam in the holder shown in Fig. 2. A high strain value is used. The speciF = stress in outer fibers, psi, L = beam length, in., men is made the anode in an electrolytic A = distance between inner and outer cell. load points, in., The deflection of the middle of the E = modulus of elasticity, psi, and specimen is calculated Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 by ESTmeans 2015 of the D = specimen thickness, following equation: Downloaded/printed by in. University of Washington (University of Washington) pursuant to License Agreement. No further repr REPORT OF COMMITTEE B-3 15 FIG, IS—Dimensions of tuning fork specimen. where; y — deflection, in., « = strain in outer fibers, L — distance between end supports, in., and t = thickness of specimen, in. FORMULA FOR STRESSING C-RING SPECIMENS METHOD 2-A Procedure: 1. Measure with a micrometer to the nearest 0.001 in.: (a) The outside diameter parallel to the stressing screw (averaging the two ends of the ring). (b) The wall thickness. 2. Set up a table to calculate the final diameter (0Z?/) required to give the desired stress using the following equations: where: A = change of OD giving desired stress, in., / = desired stress, psi, OD = outside diameter, in., t D = wall thickness, in., = mean diameter (OD ~ t), in., = modulus of elasticity, = a constant (function of ring D/f), see Fig. 14, and ODf = final outside diameter of stressed C-ring. To simplify calculations, certain terms in the above equation may be combined into a constant that will be applicable for a group of rings of the same alloy and size. Let 4EZ/ir ~ K, a constant. Then A - fiy/Kt. If the alloy or the size of the ring is changed, a different K must be calculated. E Z NOTE—The main source of error 10 the stress determination lies in the measurements of the C-ring dimensions. If in a typical example of a 0.750 in. O.D. by 0.060 in. wall thickness C-ring the measurements are determined accurately to the nearest 0.001 in., the total error should not exceed about 6 per cent. FORMULA FOR STRESSING TUNING FORK SPECIMENS (JIG. 15) METHOD 2-B6 where: total amount of closure at tine ends, in., length of tines, in. (see Fig. 15), * Derived from deflection formula tor cantileverCopyright beams. by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions a 16 STRESS CORROSION TESTING / = thickness of each tine, in., E = modulus of elasticity, psi, and a = maximum stress in outer fiber of either tine, psi. STRESSING OF WELDED BEAM ASSEMBLY METHOD 3 The deflection used in stressing the welded beam assembly is obtained by the use of the following formula: where: da = change in the distance between the / E i L a plates along the longitudinal axis of the bolts, in., = the required stress, psi, = modulus of elasticity, psi, = thickness, in., = length (center to center of bolts), in., and = distance from center of bolts to nearest flange of "H" section, in. STRESSING OF INTERFERENCE RING SPECIMENS METHOD 4 Interference Ring Specimen: The nominal dimensions of this specimen can be varied to suit the part being tested. The following limitations should be kept in mind when specifying dimensions for the specimen: 1. The width of the ring should be at least four times the wall thickness to insure maximum uniformity of stress across the ring. (Note—There is no known maximum limitation to the width of the ring and probably this would be limited only by the ability to insert the plug into the ring.) 2. The stress varies through the thickness of the ring, being highest at the inside surface. stress on ID 3. The width of the plug should be two times the width of the ring + f in. 4. The tolerance in the plug diameter is 0.0005 in. This is adequate for interferences of 0.0080 in. or more, resulting in a maximum possible error of 6 per cent. For smaller interferences, it may be desirable to tighten this tolerance so as to maintain the maximum possible error at 6 per cent. Stressing: The following formula is used to calculate the interference required for stressing the ring. where: / = interference (on the diameter) between ring and plug, in., E = modulus of elasticity, psi, ID = inside diameter, in., OD = outside diameter, in., and F = circumferential stress desired at the OD, psi. For example, if the wall thickness of the ring is 0.042 times the OD, the maximum The ID of the ring must be measured and minimum stress in the ring will differ by about 10 per cent. For greatest as accurately as possible (usually to the Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 uniformity, the wall thickness should be nearest half mil) so that the diameter of Downloaded/printed by the plug pursuant can betodetermined in order to relatively thin. of Washington (University of Washington) University License Agreement. No further reprod REPORT OF COMMITTEE B-3 give the required interference as accurately as possible. A ssembly: The ring is stressed by pushing it onto the plug, using a lubricant to prevent galling. Liquid butyl stearate or petroleum jelly are good lubricants that are clean and easy to remove from the exposed surfaces after assembly of the 17 units. Figure 9 shows an assembled ring and plug and the stressing tools employed. In cases where it is not practicable to make the plug of the same material as the ring, the entire exposed surface of the plug and the fillet at the edges of the ring should be effectively coated to prevent galvanic action between the ring and plug. A suitable wax or paint may be used for this purpose. FIG. 16—Barlow correction factor. STRESS CORROSION TEST FOR TUBING (CONSTANT LOAD) METHOD 5 The hoop stress, which is uniform across the tubing thickness, can be calculated as follows : The test specimen is inserted into the jig and welded at each end; therefore, the material used for the jig should possess welding compatibility with the tubing material. The wall thickness of the tubing is reduced in the center as shown in where : Figure 10. The corrodent is introduced a = hoop stress, into the tubing there P Wed = internal pressure,EST 2015 Copyright by and ASTMmaintained Int'l (all rights reserved); Dec 16 15:53:43 under Downloaded/printed constant pressurebyfor the duration DI = inside diameter, and of the University test. of Washington (University of Washington) pursuant to License Agreement. No further repro / = wall thickness. 18 STRESS CORROSION TESTING Correct calculated stress by CF from the ratio of D0/D\ which can be obtained from curve in Fig. 16 (D0 = outside diameter). SUMMARY This report represents an attempt to provide the novice in stress corrosion testing with some idea of the principles and practical aspects that must be considered in planning a meaningful stress corrosion testing program. Some reasons for conducting stress corrosion tests are outlined. Differences between the two types of stressing, constant-deflection and constant-load, are pointed out, and some basis is provided for selecting one or the other. Some of the most common types of specimens are described in detail. Emphasis is placed on specimens that could be used for studying more than one type of problem, or that could be used for specific situations of more or less universal concern. The Subcommittee is indebted to the following task groups for developing and preparing this report: Sara J. Ketcham; Task Group Chairman. W. E. Berry W. W. Binger J. F. Eckel O. B. Ellis H. B. Romans R. S. Shane D. H. Thompson A. C. Willhelm Respectfully submitted on behalf of the Subcommittee, H. Lee Craig, Jr., Chairman ADDENDUM A more complete picture of the stress corrosion characteristics of an alloy is obtained if the effect of stress raisers is also taken into consideration. Stress raisers such as screw threads, sharp fillets, fatigue, and weld cracks may be built into a part, while others such as pits may naturally develop during exposure to a corrosive environment. Notched or precracked specimens are useful for this purpose. by changing the depth or the radius of the notch or both. Related values for these parameters are available from Ref 2. Precracked Specimens: The philosophy behind the use of precracked specimens, detailed test procedure, and treatment of the stress aspect by fracture mechanics have been set forth in several publications [J, 4\. Briefly, the purpose of this test procedure Notched Specimens: is to permit characterization of a metal Notched tensile bars or C-rings such in terms of the size of a preexisting flaw, as those shown in Fig. 17a and b can be which would be expected to initiate a used to investigate the effect of stress stress corrosion crack in that metal under raisers on susceptibility of an alloy to a specific stress in a particular environstress corrosion cracking. A detailed ment. description of the use of the notched CThe most common type of specimen is ring is available [7]7. Higher or lower stress shown in Fig. 17c. A notch is first maconcentration factors can be obtained chined into a rectangular bar, then the ' The italic numbers in brackets refer to the specimen is held against an eccentric bearing in 16 a machine lathe2015 and fatigued list of Copyright references by at the end of this ASTM Int'l (allpaper. rights reserved); Wed Dec 15:53:43 EST Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprod Copyright by ASTM Int'l rights reserved); Wed Dec 16 15:53:43 EST 2015 FIG.(all17—Notched and precracked specimens. Downloaded/printed by 19 pursuant to License Agreement. No further reproductions authorized. University of Washington (University of Washington) 20 STRESS CORROSION TESTING until the machined notch is extended. The specimen is then stressed as a cantilever beam. Methods for Wire, Rope, and Cable Products: The types of specimens most commonly used for these products are tension specimens, either with or without reduced sections. These can be dead loaded by means of calibrated springs or levers. The fixtures shown in Figs. 3 and 4 could be modified for this purpose. A free loop is used to stress wire into the plastic region. The test consists of looping wires around mandrels three times their diameters, permitting them to spring back. After exposure to a corrosive media, damage is detected by failure in unlooping. Cracking occurs on the inside of the loops, where the surface layers are in tension after springback. Use of such a specimen and the approximate residual stresses have been described [5, 6]. given in the report are accurate enough for most test work. However, for precise work, the experimenter can only be sure of the stress levels by determining the strain with strain gages. This particularly applies to specimens such as those shown in Figs. 6 and 8. Consideration should be given to the possibility of crevice corrosion developing at areas where specimens are in contact with a jig. In some cases, this can provide galvanic protection to the stressed area. In other cases, it may give rise to hydrogen embrittlement by the cathodic corrosion product, hydrogen. The areas where the jig and the specimen are in contact should then be either masked off or, where possible, the jigs suspended in such a way as to keep the contact area out of the corroding medium. Respectfully submitted on behalf of Task Group I, Sara J. Ketcham Naval Air Engineering Center, Aeronautical Materials Laboratory, Philadelphia, Pa.; Chairman Precautionary Notes: Stresses calculated from the formulas SUGGESTED REFERENCES F. A. Champion, Corrosion Testing Procedures, John Wiley, New York, 1952. TJ. R. Evans, The Corrosion and Oxidation of Metals: Scientific Principles and Practical Applications, chapter XVI, St. Martin's Press, New York, 1960. H. H. Uhlig, Corrosion and Corrosion Control, chapter 7, John Wiley, New York, 1963. Stress Corrosion Cracking and Embrittlement, W. D. Robertson, editor, John Wiley, New York, 1956. Symposium on Stress Corrosion Cracking in Metals, ASTM-AIME, 1944. Physical Metallurgy of Stress Corrosion Fracture, T. H. Rhodin, editor, Interscience, New York, 1959. CITED REFERENCES [1] ¥. S. Williams, W. Beck, and E. Jankowsky, "A Notched Ring Specimen for Hydrogen Embrittlement Studies," Proceedings, Am. Soc. Testing Mats., Vol. 60, 1960, p. 1192. [2] R. E. Peterson, Stress Concentration Design Factors, John Wiley, New York, 1953. [3] B. F. Brown and C. D. Beachem, Corrosion Science, Vol. 5, 1965, p. 745. [4] B. F. Brown, "A New Stress-Corrosion Cracking Test for High-Strength Alloys," Materials Research & Standards, Vol. 6, No. 3, March, 1966, p. 129. [5] G. T. Spare, Wire and Wire Products, Vol. 29, No. 12, 1954, p. 1421. [6] G. Brewer and H. C. Ihsen, Metal Progress, April, 1945, p. 707. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductio W. J. Helfrich1 Influence of Stress and Temperature on Short-Transverse Stress Corrosion Cracking of an AU.2Zn-2.5Mg Alloy REFERENCE: W. J. Helfrich, "Influence of Stress and Temperature on Short-Transverse Stress Corrosion Cracking of an Al-4.2Zn-2.5Mg Alloy," Stress Corrosion Testing, ASTM STP 425, Am. Soc. Testing Mats., 1967, p. 21. ABSTRACT: Stress corrosion cracking of an age-hardened Al-Zn-Mg alloy (7039) subject to continuous immersion in a 1 N NaCl solution was studied at stresses of 5 to 55 ksi and temperatures of 30 to 100 C and found to be an activated process. The rate, r, of stress corrosion (proportional to the inverse failure time) can be expressed as: r = /-o exp{ - [Q*(0) - sV*}/RT\ where Q* (O) is the activation energy in the absence of stress, s, and V* is the activation volume. Q* (O) has a value of 20 ± 0.8 kg-cal/mole, and V* is approximately 28 to 34 cmVmole for applied stresses of 25 ksi or higher. It is suggested that the rate-determining step in the stress corrosion of 7039 involves anodic dissolution of MgZn2 at grain boundaries. The mechanism of stress corrosion cracking is discussed with reference to previously derived models. KEY WORDS: aluminum alloy, stress corrosion, cracking, activation, electrochemical dissolution, pitting (corrosion), corrosion The effect of temperature on the kinetics of stress corrosion cracking of metals has not been studied in great detail. Yet, for a process in which electrochemical dissolution may determine the time dependence for failure, temperature should be a powerful rate-determining factor. Temperature effects undoubtedly account, in part, for the effectiveness of hot concentrated nitrates and alkalines in causing stress corrosion of mild steels, boiling 42 per cent magnesium chloride in cracking of austenitic stainless steels, and hot saline solutions on the reduced time-tofailure in certain aluminum alloys. In this latter connection, Gruhl [I]2 1 Research engineer, Department of Metallurgical Research, Kaiser Aluminum & Chemical Corp., Spokane, Wash. 2 The italic numbers in brackets refer to the list of references appended to this Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 paper. Downloaded/printed by 21 University of Washington (University of Washington) pursuant to License Agreement. No further reproductions aut 22 STRESS CORROSION TESTING has shown that stress corrosion cracking in an Al-Zn-Mg alloy was dependent upon temperature to the extent that the observed failure times were almost halved for each 10 C rise in testing temperature in the range of 25 to 70 C. However, not all aluminum alloys are similarly affected. The stress corrosion failure times of an age-hardened 2024 aluminum alloy are not measurably dependent upon the testing temperature in the range of 3 to 60 C [2]. In a like manner, the rate of intergranular corrosion in unstressed 2024 specimens is independent of testing temperature below 60 C but strongly temperature dependent above this temperature [3]. Although stress corrosion tests were not conducted at temperatures above 60 C, Nichols and Rostoker [3] questioned the compatibility between a process of intergranular penetration that is presumably independent of temperature and theories of stress corrosion cracking which are based on an intergranular corrosion mechanism. While it is well known that increasing stress shortens the observed time-to-failure in aluminum alloys, arguments against a mechanism for stress-enhanced anodic dissolution are based on the fact that elastic stress (strain) adds little to the thermodynamic driving force for dissolution [4]. That is, elastic strain energy causes a change in the anodic corrosion potential of aluminum of only 10~4 to 10~3 v per increment of 10,000 psi stress, which is negligible compared with normal reversible corrosion potentials of about 1 v. Thus, it appears that the problem is not one of whether stress accelerates stress corrosion cracking, but rather the manner in which the rate of cracking is dependent upon stress. Hillig and Charles [5] recognized that the rates of some chemical reactions involving solids are dependent upon the mechanical stress at or near the reaction boundary. Their theory for stress corrosion of an amorphous elastic solid fits the known experimental data of Mould and Southwick [6], Wiederhorn [7], and Charles [8] for static fatigue of soda-lime glass. The fracture process in glass has been shown to be stress dependent and thermally activated, consistent with a corrosion process involving transport of sodium ions to the reaction interface. Since it is the purpose of this paper to show that stress corrosion cracking in certain aluminum alloys is an activated process, new studies on the t6mperature and stress dependence for failure in an Al-4.2Zn2.5Mg alloy are presented. These results are discussed with reference to the model for stress corrosion proposed by Hillig and Charles and hi terms of the mechanism of stress corrosion cracking. Experimental A single lot of commercially available3 1-in. 7039-T64 alloy plate 3 Kaiser Aluminum and Chemical Corp., Oakland, Calif. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduction HELFRICH ON INFLUENCE OF STRESS AND TEMPERATURE 23 (66.1 ksi tensile strength, 57.5 ksi yield strength) was obtained for study. The chemical composition is given below. Weight per cent. Zn Mg Mn 4.2 2.5 0.24 Cr 0.16 Fe 0.30 Si Cu Ti V Al 0.14 0.03 0.01 0.01 balance Specimens for short-transverse stress corrosion testing were machined into C-rings measuring 0.948 in. in outside diameter, 0.750 in. in width, and 0.060 in. in thickness, then dressed with 180-grit emery, degreased in acetone, and stressed. Machining and the method of stressing have been described elsewhere.4 Prior to testing, all but the upper tension surface of each specimen was dip-coated in liquid neoprene to eliminate the possibility of dissimilar alloy corrosion between the C-ring and the aluminum fastener used in stressing the ring. Specimens in groups of five C-rings for each variable of stress (5, 10, 15, 25, 35, 45, and 55 ksi) and temperature (30, 40, 60, 80, and 100 C) were tested by continuous immersion in an aerated 1 N sodium chloride solution. Solutions were made up with reagent grade salt and distilled water with a pH of about 6.0 and resistance greater than 500,000 ohm • cm. All tests were conducted in 2000 ml of freshly prepared solution, in turn, contained in resin reaction kettles refluxed to prevent vapor loss. Solutions were changed every five to seven days. Temperature control to ±1.0 C was obtained with a mercury-in-glass thermoregulator, by agitating the solution with a magnetic stirrer and by regulating the power input to a 600-w Pyrex immersion heater. In general, all specimens were tested to failure, with the exception of those tested at or near the threshold or endurance stress. The criterion for failure was based on first visual evidence of cracking at five diameters magnification. Since it was found that the failure times were log-normally distributed, the geometric mean failure time was selected as representative of the stress corrosion performance for each group of test specimens [9]. Discussion General Approach Stress corrosion cracking is assumed to be a thermally activated process in which the rate of activation is dependent upon the local state of stress (driving force) at or near the reaction boundary. According to 4 See p. 3. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprodu 24 STRESS CORROSION TESTING Hillig and Charles [5] the activation free energy, Q*(s), for the corrosion reaction can be expressed by a Taylor series expansion at s = O where higher than first-order power terms of stress, s, are considered negligible. Q*(O) is the activation free energy for the corrosion reaction on a stress-free surface. The term dQ*/ds has the dimensions of volume and is termed the activation volume, V*, from analogy with terms such FIG. 1—Effect of applied stress at constant temperature on stress corrosion cracking of 7039-164. as (3G/dp)T r = V. In the case of tensional forces, it is expected that dQ*/ds will be negative [5]. The rate of activation, r, for the corrosion reaction in the direction of the applied force can be given by where R and T have their usual thermodynamic significance. The preexponential kinetic factor, r 0 , expresses the rate of corrosion of an unstressed planar surface and is exponentially dependent upon the zerostress molar free energy and the surface free energy at the reaction boundary. This latter term is associated with the surface energy at a concave solid-corrosion product interface and represents the surface Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 tension in retarding Downloaded/printed by crack motion. Since the curvature at the reaction University of Washington (University of Washington) pursuant to License Agreement. No further reprod HELFRICH ON INFLUENCE OF STRESS AND TEMPERATURE 25 interface may become very large, this term can become limiting when the applied force (stress) is small [5]. Elastic strain energy, which can also be associated with the driving force for the corrosion reaction, exerts a negligible thermodynamic effect and is neglected. Equation 1 describes how stress and temperature influence the rate of activation for corrosion of a homogeneous brittle solid such as sodalime glass [7]. Hillig and Charles [5] extended these arguments to include stress corrosion cracking in homogeneous plastic materials. It will be shown that this approach is applicable to stress corrosion cracking in a nonhomogeneous precipitation-hardened Al-Zn-Mg alloy as well. Stress Corrosion in Al-4.2Zn-2.5Mg The logarithms of the median failure times, as a function of the applied stress at constant testing temperature, are plotted in Fig. 1. It is apparent that the data obtained for stresses in excess of about 25,000 psi are adeTABLE 1—Activation volume V* for stress corrosion of Al-4.2Zn-2.5Mg at various temperatures and stresses between 25 and 55 ksi. Testing Temperature, deg C 100 80 60 40 30 V*, cm'/mole 27 ± 5 26 30 29 28 quately represented by straight lines and, assuming that the rate of reaction is proportional to the inverse of the failure time, satisfy Eq 1. That is, the log failure times at constant testing temperature are directly proportional to the applied stress. The slopes of the straight line portion of the curves in Fig. 1 are proportional to the activation volume V* which, in turn, should be independent of temperature. The data in Table 1 illustrate this point. The average value of the activation volume obtained from these data is approximately 28 ± 5 cm3/mole.5 (The error was computed for a two-sided 95 per cent confidence interval and log standard deviation of 0.12.) Real departures from linearity exist at applied stresses below about 25 ksi (Fig. 1). The time-to-failure increases with decreasing stress at a faster than exponential rate, and an approach to a threshold stress is 5 Presumably, during the early stages of stress corrosion, the stress at the reaction interface can be equated to the applied stress. However, stress concentration undoubtedly occurs with continued chemical attack. Since the acting interfacial stresses could approach values of several times the applied stress, it is not imposCopyright by ASTM Int'l volume (all rightsisreserved); Wed Dec 16 15:53:43 EST 2015 sible that the activation somewhat smaller than the value quoted here. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductio 26 STRESS CORROSION TESTING indicated. It is interesting, however, that for practical purposes, the threshold stress is sensitive to the testing temperature. Specimens stressed to 10 ksi and tested at temperatures above 40 C and specimens stressed at 5 ksi and tested at 100 C failed in a week or less, yet no failures were observed in specimens stressed at 10 ksi and tested more than 14 weeks at 30 C. While it was considered instructive to continue these latter tests, FIG. 2—Temperature dependence on stress corrosion cracking of 7039-T64 at applied stresses of 5 to 55 ksi. in that failures may have occurred, prolonging the test period had the undesirable side effect of causing general corrosion in the form of pitting attack. Reduced times-to-failure, with increasing temperature at constant stress, were noted because stress corrosion is a thermally activated process. The log of the median failure times for specimens subject to constant stress, as a function of the inverse of the absolute test temperatures, are graphically presented in Fig. 2. Excepting the results for specimens stressed at 5 and 10 ksi, the data are adequately represented by Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 straight lines, thebyslopes of which are a measure of the activation energy Downloaded/printed University of Washington (University of Washington) pursuant to License Agreement. No further reproduction HELFRICH ON INFLUENCE OF STRESS AND TEMPERATURE 27 for stress corrosion under constant loading. These results are given in Table 2. The approximately uniform decrease in the observed activation energy of 0.5 to 0.6 kg-cal/mole with increasing stress in increments of 10 ksi at or above 25 ksi again satisfies Eq 1 and is statistically significant within a 95 per cent confidence interval. The average change in the activation energy of 0.56 kg-cal/mole per 10 ksi gives a measure of the activation energy Q*(O) in the absence of stress, namely, 20 ± 0.8 kg-cal/mole. The data are not entirely self-consistent, however, in that a decrease of 0.56 kg-cal/mole per 10 ksi yields an activation volume V* of about 34 cmVmole, which is somewhat larger than the value quoted earlier (28 cmVmole). Mechanism of Stress Corrosion An activation energy Q*(O) of about 20 kg-cal/mole and activation volume V* of 28 to 34 cmVmole point to a rate determining step in the TABLE 2—Observed activation energy for stress corrosion of Al-4.2Zn-2.5Mg at temperatures of 30 to 100 C and stresses of 15 to 55 ksi. Q*(0) - sV*, kg-cal/mole Applied Stress, ksi 16.9 ± 0.8 17.5 18.1 18.6 20.4 45 35 25 15 55 process of stress corrosion of 7039 alloy which does not involve transport phenomenon, for example, diffusion of a chemically reactive species, such as magnesium or zinc, to the reaction boundary. In aluminum, the latter might generally be characterized by an activation volume in the neighborhood of 10 cmVmole [10] and an activation energy of 28 to 31 kg-cal/mole [11]. If, in fact, the rate of stress corrosion in Al-Zn-Mg alloys is controlled by electrochemical dissolution of an anodic phase, it may be possible to account for the values quoted above. The principal age-hardening phase in aluminum (3.5 to 6 per cent zinc, 2 to 4 per cent magnesium) alloys aged at temperatures of less than 200 C has been identified as MgZn2 . During aging, M-MgZn2 forms in the grain boundaries and, depending upon the degree of aging, GuinierPreston zones and platelets of M'- and M-MgZn2 form in the matrix [12-15]. Taking the measured lattice spacings [72] and the density [76] of the M phase, then by calculation the molar volume of MgZn2 is found to be approximately 30 cmVmole. Since the results of the present study indicate that V* equals 28 to 34 cmVmole, it is suggested that the rateCopyright byprocess ASTM Int'l rights reserved); Wed corrosion Dec 16 15:53:43 EST 2015 controlling for (all intergranular stress of 7039 alloy is Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further repro 28 STRESS CORROSION TESTING based on stress and thermally activated anodic dissolution of MgZn2 at grain boundaries. Unfortunately, comparative data relating to the zerostress activation energy for dissolution of Al+MgZn2 in aqueous salt solutions are not available. The suggested mechanism and the findings of other studies have yet to be related to the overall process of stress corrosion cracking in AlZn-Mg alloys. Hillig and Charles [5] suggest that corrosion in an ideally plastic solid increases the stress concentration near the crack tip until the flow stress, s f , a i s r e a c h e d . A s l o n g a s t h e a p p l i e d s t r e s s , s , i s l e s s t h a n st, the maximum stress in the vicinity of the crack tip will not exceed sf. To develop the stress concentration necessary for mechanical rupture, the value of smust locally attain larger values than the matrix yield stress. Probable factors favoring this condition in a nonhomogeneous (precipitation-hardened) solid are: 1. The presence of a high density of sessile dislocations in the vicinity of the corroding phase. Recently, Jacobs [13,14] proposed that dislocations associated with MgZn2 precipitates are in some manner responsible for the high susceptibility of 7075-T6. High dislocation densities are not observed in 7075-T73, and this product exhibits excellent resistance to ^stress corrosion cracking. As suggested by Hillig and Charles [5] in regions of linear dimensions less than the mean spacing between dislocations, the local stress could exceed sf. f. 2. The absorption of interstitial hydrogen. Haynie et al [17] have suggested that local cathodic charging of hydrogen at grain boundaries causes stress corrosion failures in 7079 alloy by an embrittling effect. Although aluminum typically exhibits a low solubility for hydrogen, stressed grain boundaries may absorb sufficient hydrogen for embrittlement. It is conceivable that either, or both, of these factors contribute to mechanical crack propagation during stress corrosion of aluminum alloys. However, as an alternative, we will consider the following model in which the specimen is being stressed in the elastic range and stressenhanced corrosion of an anodic phase (MgZn2 in 7039 alloy) has initiated a pit or notch at the grain boundaries. Again, according to Hillig and Charles [5], the stress ahead of the notch tip remains substantially constant up to a distance y from the tip equal to p, the radius of curvature at the notch tip. As tip sharpening by corrosion occurs, the level of stress at the notch rises dramatically until, at a distance y — y0 from the tip, the stress equals the matrix yield stress, s f . With further sharpening, local yielding would maintain the stress approximately constant at some distance from y0 to an elastic region where the stress would drop below the yield strength. Thus, the conditions for brittle rupture can exist, even when local yielding occurs, if p is less than y0 [5]. Copyright by ASTM that, Int'l (all 16 15:53:43 EST 2015 It is suggested inrights this reserved); instance,Wed y0 Dec would correspond to the mean Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions HELFRICH ON INFLUENCE OF STRESS AND TEMPERATURE 29 spacing between MgZn2 precipitates at the grain boundaries. Therefore, the significant parameters which would favor brittle rupture are (a) a small particle spacing at or near the grain boundaries and (6) a maximum value for the matrix flow stress. Clearly, the latter condition is obtained when the alloy has been aged or otherwise fabricated to produce maximum strength. Overaging or other practices which reduce the matrix flow stress or increase grain boundary ductility would favor increased resistance to stress corrosion. A minimum precipitate spacing at the grain boundaries would favor maximum susceptibility to stress corrosion cracking and, in all probability, would be a sensitive function of the quench rate from the homogenization temperature. Rapid cooling would favor the growth of large numbers of smaller precipitates, effectively reducing the particle spacing and the resistance to stress corrosion cracking. Of course, a continuous film of precipitate at the grain boundaries (nil-ductility) would presumably result in the maximum attainable susceptibility and may also account for the continuous rates of crack growth noted in 7079 alloy [IS]. The proposed model is entirely consistent with the statistical approach to stress corrosion cracking in aluminum alloys. As noted previously, the failure times observed in the stress corrosion of 7039 alloy are log-normally distributed. If, as suggested, the time dependence for failure is a function of the rate of corrosion, and if this rate is subject to the influence of a large number of small independent events which operate simultaneously, then a log-normal distribution of failure times would occur quite naturally [79]. Physically, the log-normal distribution of failure times describes a process involving the simultaneous development of a large number of small independent sites of corrosion (pits or notches) at grain boundary precipitates. These sites of corrosion then join, presumably by mechanical rupture, to form a macroscopic stress corrosion crack. Conclusions As a result of the present studies, the following observations are believed to be applicable to intergranular stress corrosion cracking of 7039 alloy: 1. Stress corrosion is a stress-activated and thermally activated process exhibiting a zero-stress activation energy of 20 ± 0.8 kg-cal/mole and an activation volume of 28 to 34 cm3/mole. 2. The rate-controlling, step in stress corrosion presumably involves anodic dissolution of MgZn2 at grain boundaries, in part, consistent with an electrochemical mechanism. Factors favoring rapid or easy mechanical crack propagation are believed to be associated with a minimum particle spacing and minimum ductility at the grain boundaries and a maximum Copyright by matrix ASTM Int'l (all stress. rights reserved); Wed Dec 16 15:53:43 EST 2015 value of the flow Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further repro 30 STRESS CORROSION TESTING References [1] Gruhl, W., 'The Temperature Dependence for Stress Corrosion Cracking in AlZnMgs," Zeitschrift fuer Metallkunde, Vol 53, No. 10, 1962, pp. 670675. [2] Nichols, H. and Rostoker, W., "Analogies Between Stress-Corrosion Cracking and Embrittlement by Liquid Metals," Transactions, American Society for Metals, Vol 56, No. 3, Sept. 1963, pp. 498-500. [3] Nichols, H. and Rostoker, W., "Intergranular Corrosion Penetration in an Age-Hardenable Aluminum Alloy," Journal of the Electrochemical Society, Vol 112, No. 1, Jan. 1965, pp. 108, 109. [4] Barnartt, S., "General Concepts of Stress-Corrosion Cracking," Corrosion, Vol 18, No. 9, Sept. 1962, p. 328t. [5] Hillig, W. B. and Charles, R. J., "Surfaces, Stress Dependent Reactions, and Strength," Report No. 64-RL-3756M, Oct. 1964, General Electric Research Laboratory, Schenectady, N.Y. [6] Mould, R. E. and Southwick, R. D., "Strength and Static Fatigue of Abraded Glass Under Controlled Ambient Conditions," Journal of the American Ceramic Society, Vol 42, 1959, pp. 542, 582. [7] Wiederhorn, S. M., "Effect of Environment on the Fracture of Glass," NBS Report 8901, June 1965, U. S. Department of Commerce, National Bureau of Standards, Washington, D.C. [8] Charles, R. J., "Static Fatigue of Glass: II," Journal of Applied Physics, Vol 29, No. 11, Nov. 1958, pp. 1554-1560. [9] Booth, F. F., Tucker, G. E. G., and Godard, H. P., "Statistical Distribution of Stress Corrosion Endurance," Corrosion, Vol 19, No. 11, Nov. 1963, p. 390t. [10] Butcher, B. M., Hutto, H., and Ruoff, A. L., "Activation Volume and Energy for Self-Diffusion in Aluminum," Applied Physics Letters, Vol 7, No. 2, July 1965, pp. 34-35. [11] Stoebe, T. J. et al, "Nuclear Magnetic Resonance Studies of Diffusion of Al(27) in Aluminum and Aluminum Alloys," Acta Metallurgica, Vol. 13, July 1965, pp. 701-708. [12] Embury, J. D. and Nicholson, R. B., "The Nucleation of Precipitates: The System Al-Zn-Mg," Acta Metallurgica, Vol 13, April 1965, pp. 403-417. [13] Jacobs, A. J., "Electron Microscopy and Stress Corrosion Cracking Studies on 7075 Aluminum Alloy," Research Report No: 65-10, March 1965, Rocketdyne Division of North American Aviation, Inc., Canoga Park, Calif. [14] Jacobs, A. J., 'The Role of Dislocations in Stress Corrosion Cracking of 7075 Aluminum Alloy," Transactions, American Society for Metals, Vol 58, No. 4, Dec. 1965, pp. 579-600. [15] Thomas, G. and Nutting, J., 'The Aging Characteristics of Aluminum Alloys: Electron Microscope Studies of Alloys Based on the Aluminum-Zinc-Magnesium System," Journal of the Institute of Metals, Vol 88, 1959-60, pp. 81-90. [16] Donnay, J. D. H., ed., Crystal Data: Determinative Tables, 2nd ed., American Crystallographic Assn., April 1963, p. 755. [17] Haynie, F. H. et al, "A Fundamental Investigation of the Nature of StressCorrosion Cracking in Aluminum Alloys," AFML-TR-65-258, July 1965 and Oct. 1965, Air Force Materials Laboratory, Research and Technical Div., AFSC, Wright-Patterson Air Force Base, Ohio. [18] Haynie, F. H. et al, Seventh Progress Report, AF 33(615)-1710, 15 December, 1965, Battelle Memorial Institute, Columbus, Ohio. [19] Booth, F. F. and Tucker, G. E. G., "Statistical Distribution of Endurance in Electrochemical Stress-Corrosion Tests," Corrosion, Vol 21, No. 5, May 1965, p. 173. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprodu C. D. Beachem1 and B. F. Brown1 A Comparison of Three Precracked Specimens for Evaluating the Susceptibility of High-Strength Steel to Stress Corrosion Cracking REFERENCE: C. D. Beachem and B. F. Brown, "A Comparison of Three Precracked Specimens for Evaluating the Susceptibility of HighStrength Steel to Stress Corrosion Cracking," Stress Corrosion Testing, ASTM STP 425, Am. Soc. Testing Mats., 1967, p. 31. ABSTRACT: The susceptibility of AISI 4340 steel to stress corrosion cracking in dilute NaCl solution is examined using the center-cracked and part-through-crack tension specimens and the edge-cracked cantileverbeam specimen. The data are treated in terms of linear elastic fracture mechanics. It is found that all three types of specimens can be used to determine the value of the plane-strain stress-intensity factor above which cracking will occur in the presence of this aggressive environment. KEY WORDS: stress corrosion cracking, high-strength steels, precracked specimens, cracking (fracturing), corrosion It has been conclusively shown [7]2 that significant susceptibility of a material to stress corrosion cracking (SCC) may escape detection unless precracked specimens are used. If cracks are not intentionally introduced, the test is usually dependent upon the formation of stress concentrators such as fatigue cracks or corrosion pits. Crack development at the bottom of a corrosion pit may not occur if the material does not undergo pitting attack. The use of precracked specimens, on the other hand, more closely approximates that critical part of a high-strength structure which contains a crack either built into it or developed during the testing or use of the structure. It is for this reason that the precracked SCC specimen was adopted at the Naval Research Laboratory. 1 Head, Micro-Mechanical Metallurgy Section, Physical Metallurgy Branch, and head, Physical Metallurgy Branch, respectively, Metallurgy Div., Naval Research Laboratory, Washington, D. C. Personal members ASTM. 2 The italic numbers in brackets refer to the list of references appended to this Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 paper. Downloaded/printed by 31 University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authoriz 32 STRESS CORROSION TESTING The precracked specimen must be properly designed, however, or it, too, may give misleading data [1]. Several steels and titanium alloys have been tested in different sheet and plate thicknesses, and the data show that whereas stress corrosion cracks may propagate readily under plane-strain conditions (in terms of linear elastic fracture mechanics), they often do not propagate in somewhat thinner sections. If the service application of the material will be in thick enough sections so that planestrain stress conditions will exist around the leading edge of a crack, the test specimen must also be designed to provide plane-strain conditions at the leading edge of its starting crack. The use of linear elastic fracture mechanics to analyze crack extension data permits appropriate characterization of relative propensity toward SCC. In addition, this analysis enables one to predict the cracking behavior of structures in terms of critical flaw sizes and stresses [1,2]. This paper gives the results of SCC experiments in which AISI 4340 steel tempered at 400 F was tested in plane strain, in the form of three types of crack propagation specimens in order to investigate further the usefulness of the cantilever-beam specimen. Material and Test Procedure AISI 4340 steel (air melted) was chosen for these specimen comparisons due to its wide use and high yield strength, with the latter making the data more amenable to linear analysis methods. Specimens 9 by 1 l/z in. were machined from a V^-m.-thick sheet with the long dimension oriented in the rolling direction. Holes l/z in. in diameter were drilled through the specimen center lines 1 l/a in. from the specimen ends. Some of these specimens were then centrally slotted at midlength and tensiontension fatigued to extend the slot to a total length of about 5/s in. in the RW direction. The general scheme of succinctly and unambiguously describing crack propagation directions in wrought plate and sheet may be found in Ref 3: RW indicates the crack propagated perpendicular to the rolling direction and parallel to the direction of the sheet width; RT indicates the crack grew perpendicular to the rolling direction and in the thickness direction. Some of these centrally cracked specimens were then slit lengthwise to make edge-cracked cantilever-beam specimens about 0.71 in. in depth. Other specimens were fatigued in bending against a sharp point located at midlength to form an RT surface crack at the root of a surface notch on the opposite surface. All the specimens were then austenitized together for l/z hr at 1550 F in argon, oil-quenched, and tempered at 400 F for 1 hr, giving a hardness of Rc 51. Containers were slipped over the central (cracked) portions of the specimens, and the 3.5 per cent sodium chloride (NaCl) solution was added after the specimens were loaded to the selected stresses. The salt Copyrightwere by ASTM Int'l (allbut rights reserved); Weddaily Dec 16 EST 2015 solutions not stirred were changed in 15:53:43 the long-term tests. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further repr FIG. 1—Specimen dimension identification and equations. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized. 34 STRESS CORROSION TESTING Specimens tested in tension were tested in either stress rupture racks or in a standard tensile machine, with the last being necessary for the higher stresses. The cantilever-beam specimens were tested in the racks described in Ref 1. Briefly, these racks individually consist of a sturdy upright support into which one end of a specimen is clamped (with the cracked edge up), a lever arm clamped to the other end of the specimen, and a system for adding weights to the end of the lever arm. Specimens of all three designs were tested at various selected stresses and the time-to-fracture recorded. If fracture did not occur after an extended period, the specimens were removed from the solution, dried, heated overnight in air at 400 F, and broken to make the extent of crack growth visible on the fracture surface. Specimen dimensions are given in Fig. 1, along with the equations. The stress-intensity factors from the cantilever-beam specimen tests were calculated from the Kies [4] equation where: KI = plane-strain stress-intensity factor, ksi m = bending moment, in-lb., B = specimen width (sheet thickness), in., D = specimen depth, in., and a = 1 — a/D, where a is the crack depth. The stress-intensity factors for the surface-cracked (or part-throughcrack) specimen results were calculated from the Irwin equation where: K! = plane-strain stress-intensity factor, ksi a = crack depth, in., <r = gross stress, ksi, (rys = the yield strength of the material, ksi, and $ = function of the crack shape. Values of <f> may be obtained from elliptical integral tables [5]. The stress-intensity factors from the center-cracked specimens were calculated using the equation [6] Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorize BEACHEM AND BROWN ON SUSCEPTIBILITY OF HIGH-STRENGTH STEEL 35 TABLE 1—Summary of data. CENTER-CRACKED SPECIMENS Specimen No. 57/^- Time, min ksi v in- 88.5 85.0 37.1 27.2 25.0 19.1 14.6 14.2 18 16 7 8 14 17 11 18 2 4 27 54 84 399 452 464 024 545 Comments WW, in. B, in. 2a, in. Load, Ib dry dry broke broke broke broke broke no break 1.473 1.499 1.503 1.493 1.499 1.500 1.508 1.473 0.123 0.125 0.126 0.127 0.126 0.125 0.127 0.123 0.750 0.598 0.587 0.617 0.600 0.617 0.640 0.740 12 150 15 000 6 500 4 630 4 330 3 150 2 460 1 970 SURFACE-CRACKED SPECIMENS SpecKT imen . . /'-.— No. ksiVm. 58.0 58.0 57.0 40.0 37.0 36.0 25.0 18.0 17.2 14.7 13.4 11.2 28 23 21 26 27 30 20 29 22 21 28 19 Time, min Comments WW, in. B, in. a, in. 1C, in. Load, Ib 39 15 86 130 451 630 3 042 1 452 26 706 dry dry dry broke broke broke broke broke broke no break no break no growth 1.465 1.466 1.467 1.465 1.466 1.465 1.465 1.466 1.465 1.465 1.465 1.465 0.125 0.126 0.125 0.124 0.126 0.126 0.125 0.124 0.125 0.125 0.125 0.129 0.031 0.048 0.034 0.049 0.031 0.045 0.040 0.033 0.034 0.031 0.029 0.020 0.120 0.195 0.125 0.191 0.134 0.196 0.163 0.131 0.154 0.125 0.120 0.115 34 350 27 600 33 550 20 700 24 150 17 250 13 800 12 000 9 960 9 357 8 822 8 490 CANTILEVER-BEAM SPECIMENS Specimen No. . . /-.— ksi V m. 1-B. 3-B. 2-B. 4-B. 5-B. 2-A. 3-A. 10-A 1-A. 68.4 50.6 34.3 22.0 17.8 14.6 12.8 12.7 11.6 Time, min Comments a, in. B, in. D, in. 5 20 36 144 330 1 144 21 600 1 200 dry broke broke broke broke broke no break no growth no growth 0.260 0.258 0.272 0.249 0.268 0.288 0.270 0.249 0.248 0.126 0.126 0.125 0.125 0.125 0.124 0.125 0.127 0.125 0.705 0.706 0.713 0.710 0.703 0.713 0.720 0.707 0.714 M, in. -lb • 640 486 320 215 159 124 121 126 100 where a is the gross stress, a is half of the crack length, and / (2a/W) is obtained from Paris [6]. Results The object of using the precracked specimens in these SCC tests was to Copyright determine KtInt'l value below which Wed SCCDecis 16 not15:53:43 observed, which is by the ASTM (all rights reserved); EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprod 36 STRESS CORROSION TESTING FIG. 2—Stress corrosion cracking tests of AISI 4340 steel, using cantileverbeam specimen in 3.5 per cent Nad solution. FIG. 3—Stress corrosion cracking tests of AISI 4340 steel, using center-cracked specimens tested in 3.5 per cent Nad solution. called KIscc. SCC • This critical value of Kt is found by loading a series of specimens at different stress intensities (KJ until the K^ level is found Copyright ASTM does Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 where the by specimen not break. Downloaded/printed by in Table 1 for all three specimen types. Data for The data are listed University of Washington (University of Washington) pursuant to License Agreement. No furt BEACHEM AND BROWN ON SUSCEPTIBILITY OF HIGH-STRENGTH STEEL 37 FIG. 4—Stress corrosion cracking tests of A1SI 4340 steel, using surfacecracked specimens in 3.5 per cent NaCl solution. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 FIG. 5—Surface-cracked specimen results plotted relative to a behavioral Downloaded/printed by curve predicted from the cantilever-beam test. University of Washington (University of Washington) pursuant to License Agreement. No further reprodu 38 STRESS CORROSION TESTING each specimen type are listed in decreasing order of plane-strain stress intensity, Kt. The Comments column indicates whether or not the crack grew. The data are plotted in Figs. 2 through 5. Figure 2 shows the data from the cantilever-beam specimen tests. The KTc in this test was found to be about 68 ksi \/in. Kle is the critical plane-strain stress intensity for fracture in the absence of the corrosive medium—a measure of the fracture toughness of the material. The curve in Fig. 2 is seen to become horizontal at a KIscc JSCC of 12.7 ksi \/ni! after about 1000 min, which means that (for this material and this specimen design) if a specimen does not break after one day it is likely not to break at all and may be taken out of the solution, broken, and examined for possible crack growth. Figure 3 gives similar data for the center-cracked specimens, only in this case the Klc is seen to be 88 ksi \/m. The Klscc is seen to be below 14.2 ksi -\/in. Specimen 18 was loaded to this last stress intensity for more than a month but did not break. However, a little crack growth was apparent on the fracture surface after the specimen was dried and broken. The curve in Fig. 3 is not horizontal after more than a month. Figure 4 shows the data from the surface-cracked specimens. The KIe is seen to be 58 ksi \/in., and the SCC KI is seen to be about 12 ksi \/in. with the curve becoming horizontal after several days or a week. Discussion of Results The cantilever-beam test appears to be a much more rapid test of a material's susceptibility to SCC than either the center-cracked or the surface-cracked specimens. Moreover, the cantilever-beam test results enabled accurate predictions of the test results from the other two specimen types. For example, the 12.7 ksi -\/in. value obtained from the cantileverbeam tests was used, along with an estimated a/2C value of 0.25 in the equation for the surface-cracked specimen, to draw a break-no break curve in Fig. 5. The cantilever-beam test results were thus used to predict the critical lengths (or depths since the ratio was rather constant) of cracks for specific gross stresses. No SCC was expected below the curve, but was predicted above the curve. The surface-cracked specimen data are plotted on the same graph, and one can see that the times-to-fracture (number in parentheses near data points) increase as one approaches the curve from above and that SCC did not occur below the curve. Since Specimen 19 had a relatively shallow crack, the break-no break curve was plotted using its a/2C ratio (0.174) and is seen as a dashed line to the right of the curve for a/2C values of 0.25. Of the three values of Klc found with these three specimen types, the value of 58 ksi \/in. is probably the most meaningful for service use. Copyright by ASTM Int'l rights reserved); Wed Dec 16 15:53:43 EST 2015 These differences in K(all lc results are most likely related to onset of fast Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduction BEACHEM AND BROWN ON SUSCEPTIBILITY OF HIGH-STRENGTH STEEL 39 fracture uncertainties, which would not influence the Klscc determinations. The surface-cracked specimen results are seen to be within the limitations [7] of (a) crack depth less than | the specimen thickness and (b) gross stresses less than the yield strength, as indicated on the ordinate and abscissa, respectively. It is interesting, however, to consider a limiting, but practical, approximation using the surface crack equation. For a long shallow crack, where it may be considered infinitely long, Eq 2 may be rewritten as At yield point stresses, this reduces to and the critical crack depth (of a long crack or sharp scratch) to cause SCC at yield point stresses in this material is seen to be 0.0007 in. The calculated plastic zone size at the tip of such a crack is approximately 0.001 in. Therefore, the value of 0.0007 in. for the critical flaw size for SCC does not have the indicated accuracy. The approximate value of 0.001 in. for a critical flaw depth for SCC- in this steel is real and significant, since this kind of flaw might easily be introduced in service and escape detection. The lesser fracture times in the cantilever-beam tests and the surfacecracked specimen tests are probably due to the more rapid increase of /sTj with crack size for these two specimens. Thus, once a crack starts to grow, the Kt increases rapidly, accelerating crack growth and lessening the time-to-fracture. Further consideration of some of the data in Table 1 provides a forceful argument for the use of stress intensities instead of net section stresses (uncracked area divided into the load) in the analysis of SCC data. The surface-cracked Specimen 19 and the center-cracked Specimen 18 both were stressed to give stress-intensity factors close to 12.7 ksi \/in. and reacted accordingly to their environments, with the surface-cracked specimen (slightly below the Klscc) not cracking and the center-cracked specimen (slightly above the ATISCC) very slowly cracking. The net section stresses for these two specimens, however, were quite different, with the center-cracked specimen being stressed at an = 21,800 psi, and the surface-cracked specimen stressed at <TN = 45,400 psi. Thus, while the stress intensities of the two specimens were nearly the same and close to Copyright ASTM Int'l (all rights reserved); Wed Decby 16 more 15:53:43 ESTa2015 the Klscc,bytheir net section stresses differed than factor of two. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduction 40 STRESS CORROSION TESTING Conclusions The cantilever-beam specimen test for the susceptibility of highstrength alloys to SCC yields the same Klscc as the surface-cracked specimen and probably the same as the center-cracked specimen. The cantilever-beam test not only is more efficient in permitting testing at high stress intensities but also allows a quicker determination of Klscc than the other two types of specimens. The critical crack size for SCC is seen to be about 0.005 in. for a structure containing a surface flaw which has an a/2C value of 0.25 and is stressed to its yield strength. Similarly, a critical flaw size of 0.017 in. is predicted for stresses of one half the yield strength. For a long shallow crack and yield point stress, the critical depth of a crack for SCC is seen to be about one mil. A cknowledgments The authors thank the Bureau of Naval Weapons for sponsoring this work and G. R. Irwin for his helpful and constructive criticism. References [7] Brown, B. F., "A New Stress-Corrosion Cracking Test for High-Strength Alloys," Materials Research & Standards, Vol. 6, No. 3, March 1966, pp. 129133. [2] Brown, B. F. and Beachem, C. D., "A Study of the Stress Factor in Corrosion Cracking by Use of the Pre-Cracked Cantilever Beam Specimen," Corrosion Science, Vol 5, 1965, pp. 745-750. [3] 'The Slow Growth and Rapid Propagation of Cracks," Second Report of a Special ASTM Committee, Materials Research & Standards, Vol 1, No. 5, May 1961, pp. 389-393. [4] Kies, J. A. et al, "Fracture Testing of Weldments," Fracture Toughness Testing and Its Applications, ASTM STP 381, American Society for Testing and Materials, Philadelphia, 1965, p. 328. [5] Irwin, G. R., "Crack-Extension Force for a Part-Through Crack in a Plate," Journal of Applied Mechanics; Transactions, American Society of Mechanical Engineers, Series E, Vol 29, No. 4, Dec. 1962, pp. 651-654. [6] Paris, P. C. and Sih, G. C., "Stress Analysis of Cracks," Fracture Toughness Testing and Its Applications, ASTM STP 381, American Society for Testing and Materials, Philadelphia, 1965, pp. 30-81. [7] Irwin, G. R., "Relation of Crack Toughness Measurements to Practical Applications," Welding Journal Supplement, Welding Research, Nov. 1962. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprod E. E. Denhard, Jr.,1 andR. R. Gaugh1 Application of an Accelerated Stress Corrosion Test to Alloy Development REFERENCE: E. E. Denhard, Jr. and R. R. Gaugh, Application of an Accelerated Stress Corrosion Test to Alloy Development," Stress Corrosion Testing, ASTM STP 425, Am. Soc. Testing Mats., 1967, p. 41. ABSTRACT: A direct-tensile type of stress corrosion test was used to develop a corrosion resistant alloy that is relatively lean in nickel content as compared to other alloys that are immune to cracking in boiling magnesium chloride. This new alloy also is readily hot worked and welded. The same type of test specimen was used to evaluate the susceptibility of Type 410 and 17-4 PH stainless steels in several heat-treated conditions to stress cracking in environments containing hydrogen sulfide. The resistance of these alloys to cracking was found to be improved by tempering or overaging to lower strength levels, confirming earlier findings obtained on other types of tests. When heat treated to equal levels of yield strength, 17-4 PH was somewhat more resistant to cracking than Type 410. KEY WORDS: corrosion, stress corrosion, corrosion resistant alloys, stainless steels, magnesium chloride, hydrogen sulfide, hot working, welding, heat treatment, yield strength It has been customary to specify the highly alloyed nickel-base materials for equipment that must be immune to chloride stress corrosion cracking. Typical is the extensive use of Inconel2 (15.8Cr-76Ni) piping, tubing, fittings, and hardware in heat exchange equipment in the nuclear industry. Although some of the less costly stainless steels display substantial resistance to stress corrosion cracking, these alloys have not met with general favor for critical applications. Thus, there appears to be a family of alloys with intermediate nickel content (35 to 60 per cent) which has not been given research attention. Copson [7]3 recognized that alloys containing 16 to 20 per cent chro1 Senior research engineer and research engineer, respectively, Research & Technology, Baltimore Laboratories, Armco Steel Corp., Baltimore, Md. Mr. Denhard is a personal member of ASTM. 2 Registered trademark of International Nickel Co. 3 The italic numbers in brackets refer to the list of references appended to this Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 paper. Downloaded/printed by 41 pursuant to License Agreement. No further reproductions authoriz University of Washington (University of Washington) 42 STRESS CORROSION TESTING mium and 45 to 50 per cent nickel seemed immune to cracking in boiling 42 per cent magnesium chloride (MgCl2). One of the serious shortcomings, however, for fully austenitic alloys of this type is the tendency for hot cracking or microfissuring to occur upon welding. The development of a relatively lean austenitic alloy, fully immune to chloride stress corrosion cracking, together with excellent weldability and hot workability, was the challenge that became the subject of the present paper. FIG. 1—Stress corrosion test specimen. FIG. 2—Stress corrosion test equipment. To meet the above objectives, a rapid, severe, and discriminating stress corrosion test method was needed to screen large numbers of experimental alloys. A notched tensile-type stress corrosion test with boiling 42 per cent magnesium chloride as a corrodent was used for this purpose. During the course of the investigation, it became evident that the test method would be useful for evaluating the effect of heat treatments and studying theASTM behavior of alloys in Wed corrosive mediaEST other Copyright by Int'l (all rights reserved); Dec 16 15:53:43 2015than boiling Downloaded/printed by These data also are included. magnesium chloride. University of Washington (University of Washington) pursuant to License Agreement. No further reproductions a DENHARD AND GAUGH ON ALLOY DEVELOPMENT 43 Experimental Procedure Lean Austenitic Alloy—Resistant to Chloride Cracking The experimental procedure consisted of making small induction melts, forging and rolling these to bar stock, and machining stress corrosion specimens. In the case of commercially available materials, stress corrosion specimens were prepared from small-diameter bar stock. All specimens were laboratory annealed at 2000 F, unless noted otherwise. The type of specimen and method of testing has been reported in an earlier publication [2]. Briefly, the specimens were 1A in. diameter by FIG. 3—Stress corrosion tests in boiling 42 per cent MgClz. 9l/2 in. long, containing four H-in.-radius circumferential grooves (Kt = 1.3) in the center portion, spaced at Vg-in. intervals (Fig. 1). The test cell (Fig. 2) consisted simply of a length of glass tubing fitted with top and bottom rubber stoppers, through which the specimen was inserted. Heat was supplied by a small coil of resistance wire which was wound around the bottom portion of the cell. During testing, two of the circumferential grooves were immersed in boiling 42 per cent MgCl2 solution; the other two grooves were exposed to the vapor. Temperature was monitored by means of a thermocouple well extending into the solution. The cell was equipped with a water-cooled condenser to maintain a constant concentration of the test solution. The test specimens were stressed by applying dead weights through a lever-arm system. Various levels of stress were applied, starting at just under the ultimate tensile strength of the alloy under study. If both by annealed annealed-plus-sensitized specimens lasted Copyright ASTM and Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 1000 2015 Downloaded/printed by hr, they were considered immune to chloride stress corrosion cracking. University of Washington (University of Washington) pursuant to License Agreement. No further 44 STRESS CORROSION TESTING Failure beyond this time invariably was caused by pitting or general attack due to the severe environment. Various existing commercial austenitic stainless steels and nickel-base alloys were tested first. These results are shown graphically in Fig. 3. The chemical analyses of the materials are shown in Table 1. Modifications of Type 330 containing substantial manganese and other small additions were studied next. Linnert's and Larrimore's earlier work [31 had shown the beneficial effects of manganese in preventing microTABLE 1—Chemical composition and annealing treatment of commercial alloysGrade Condition Type 304 Type 310 Type3306 Incoloy 800' . . Inconel 600C . . 2000 2000 2000 1950 1800 F, F, F, F, F, WQ° WQ WQ WQ WQ c Mn P s Si Cr Ni 0.048 0.102 0.076 0.046 0.044 1.18 1.78 0.47 0.92 0.26 0.022 0.011 0.022 0.013 0.003 0.011 0.006 0.009 0.008 0.003 0.60 0.40 0.52 0.35 0.30 19.27 27.31 15.11 19.67 15.06 9.26 24.47 35.39 31.51 75.5 0 WQ = water quench. Laboratory heat. c Registered trademark International Nickel Co. , b TABLE 2—Stress corrosion resistance of experimental alloys. Heat No. c Mn p s Time-to-Failure, hr» Si Ni Cr Cb Annealed R-2759 R-2760 R-4474 R-4475 033120 R-5481 035069 a 0.014 0.49 0.076 0.47 0.045 9.87 0.11 9.80 0.096 10.00 0.022 3.30 0.030 4.88 0.61 0.52 0.26 0.23 0.008 0.008 0.98 0.004 0.008 1.23 0.011 0.005 0.73 14.88 15.11 16.16 15.72 17.82 20.19 19.22 34.31 35.39 35.6 35.7 40.49 0.15 50.32 45.48 0.14 Sensitized 360 >1000 55 >1000 >1000 330 >1000 >1000 >1000 >1000 Tested at 75,000 psi in boiling 42% MgCl2 (154 C). fissuring in fully austenitic weldments, including alloys of Type 330 composition. This investigation was directed toward reassessing their findings in conjunction with the stress corrosion studies reported below. Table 2 contains the analyses and the results of stress corrosion tests of alloys with intermediate nickel content and manganese additions. Heats R-2759 and R-2760 represent Type 330 with low- and highcarbon content. The remarkable beneficial effect of higher carbon on stress corrosion is evident. This is again displayed in alloys R-4474 and R-4475, in which manganese was added to improve weldability. Up to this point, all alloys were tested in the annealed condition. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 After it was learned that Heat 033120 was subject to stress corrosion Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprod DENHARD AND GAUGH ON ALLOY DEVELOPMENT 45 in the sensitized condition (1250 F 1 hr air cool), all subsequent alloys were tested in both the annealed and annealed-plus-sensitized conditions. To meet this requirement, it was found necessary to lower the carbon and manganese and raise the nickel contents. Heats R-5481 and 035069 represent compositions with these changes. In the case of Heat 035069, a small columbium addition was made. The welding characteristics of Inconel, Type 330, and manganese modifications of intermediate nickel content alloys were studied using notched-type specimens of 0.050-in.-thick sheet. For this work, two coupons 2-in. square and one coupon 2 by 4 in. were arranged and fusion butt welded (Fig. 4). This procedure provided a fabricated notch to con- FIG. 4—Weld crack tests. centrate stresses at the weld bead. Inert-gas tungsten-arc welding was employed, with no filler metal added. Argon was used at the rate of 20 ft3/hr for shielding gas. The weld speed was 12 in./min using 75 amp at 9Vz v. Each alloy was rated on the basis of the resistance of the weld metal deposit to cracking. The notched area was judged by the severity of cracking at the root of the notch (heavy—3, medium—2, light—1, and no cracking—0). After these specimens were evaluated, longitudinal bend test coupons were cut as shown by the dotted lines of Fig. 4. These coupons were bent approximately 90 deg around a ?4 in. radius to emphasize any cracks present. Evaluation of these specimens was determined by counting the number of cracks observed. Both types of specimens were examined under 20-power magnification. The overall crack sensitivity was the Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 cumulative total for Downloaded/printed by the two types of tests. University of Washington (University of Washington) pursuant to License Agreement. No further repr 46 STRESS CORROSION TESTING Tables 3 and 4 show the analyses of the alloys tested and the results of the weld tests, respectively. All values are the average of two or more tests. The order of mechanical properties of the several alloys is shown hi Table 5. TABLE 3—Composition of weld test alloys. Grade Type 330 330 Cb Inconel 600 Experimental Experimental Experimental Heat No. . . R3524 . . R-3525 R-4475 R-4566 035069 C Mn Si Cr 0.062 0.060 0.032 0.11 0.058 0.030 0.67 0.60 0.22 9.80 9.62 4.88 0.50 0.72 0.25 0.23 0.56 0.73 18.15 18.02 15.77 15.72 16.26 19.22 Ni Cb 34.44 34.69 75.87 35.7 40.46 45.58 0.16 0 14 TABLE 4^-Weld crack sensitivity. Grade Type 330 330 Cb Inconel 600 Experimental Experimental Experimental Heat No. Crack Sensitivity at Notch Longitudinal Bend Overall Crack Sensitivity 3 3 2 1 1 1 0 4 1 0 3 7 3 0 1 R-3524 R-3525 R^475 R-4566 035069 1 TABLE 5—Mechanical properties. Grade Type 330 . Incoloy 800. . . Ultimate Tensile Strength, psi 80 000 min 75 000 to 100 000 Experimental. . . . 85 000 Inconel 600 . . . 80 000 to 100 000 0.2% Yield Strength, psi 40 30 50 32 30 50 000 min 000 to 000 500 000 to 000 Elongation in Reduc4 Diameters, tion of Area, % % Hardness, Rockwell B 30 30 to 50 30 90 49 30 to 50 71 65 78 65 to 85 max 88 max Other Applications of Accelerated Stress Corrosion Test Although the stress corrosion test referred to in this investigation was primarily used for austenitic alloys with boiling magnesium chloride as a corrodent, it has proven useful for other environments and purposes. For example, this accelerated test method has been used to evaluate the resistance of martensitic stainless steels hi various heat-treated conditions to stress cracking4 in simulated sour (that is, hydrogen sulfide (H2S) containing) oil and gas well environments. For this purpose, the Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 * The term "stressbycracking" is used advisedly, to embrace both stress corrosion Downloaded/printed cracking andof hydrogen possibilities.pursuant to License Agreement. No further reproductions University Washingtonembrittlement (University of Washington) DENHARD AND GAUGH ON ALLOY DEVELOPMENT 47 test cell was altered as shown in Fig. 5. This very simple arrangement uses the same type of specimen as for the MgCl2 test, and it allows testing to be performed at room temperature in any aqueous solution. Provision is made to allow charging the test solution with H2S (or any other gas desired). For this program, test specimens were machined from 1/4-in.-diameter bars taken from commercial heats of Type 410 and 17-4 PH.5 The chem- FIG. 5—Modified stress corrosion test equipment. TABLE 6—Composition of alloys used for simulated sour well environmental studies. Alloy Type 410 17^4 PH c Mn 0.50 0.12 0.038 0.28 P s Si Cr Ni Cu Cb+Ta 0.014 0.017 0.013 0.014 0.19 0.68 12.10 16.03 0.22 4.23 3.52 0.24 ical analyses of these materials are shown in Table 6. Larger sizes could have been used as well, but by starting with */4 -in.-diameter material, the necessary machining was greatly minimized. Since the resistance of martensitic stainless steels to stress cracking in hydrogen sulfide environments has been found to improve with decreasing hardness [4], the subject alloys were evaluated hi the tempered or overaged conditions. So heat treated, it was hoped that they would offer good resistance to cracking, while maintaining a moderate level of Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by 5 Registered trademark(University of ArmcoofSteel Corp. pursuant to License Agreement. No further reproduction University of Washington Washington) 48 STRESS CORROSION TESTING mechanical strength. Accordingly, the 17-4 PH was heat treated to Conditions H-1150 and H-1150 M, and the Type 410 was tempered at 1000 and 1100 F. These heat treatments were chosen to produce equal levels of yield strength (125,000 and 100,000 psi, respectively) for the two alloys. Details of the heat treatments are shown in Table 7. TABLE 7—Stress corrosion resistance of Type 410 and 17-4 PH steels. Alloy Heat Treatment 0.2% Yield Strength, psi Type 410. . . . 1800 F, % hr, oil quench + 1000 F, 4 hr, air cool 125 000 Type 410 . . . 1800 F, }4 hr, oil quench + 1100 F, 4 hr, air cool 100 000 17-4 PH 1925 F ]4 hr air cool + 1150 F, 4 hr, air cool 125 000 17-4 PH 1925 F % hr air cool + 1500 F, 2 hr, air cool + 1150F, 4 hr, air cool 100 000 Applied Stress, psi 75 50 25 75 50 25 75 50 25 75 50 25 000 000 000 000 000 000 000 000 000 000 000 000 Time-toFailure, hr" 1.3, 1.4 5.7, 6.5 11, 20 6.2, 18 27, 39 195, 845 6.1, 6.1 12, 18 83, 102 16, 22 50, 95 345, 1105 0 Tested at room temperature in 6% NaCl-0.5% acetic acid solution, saturated with H2S. FIG. 6—Stress corrosion tests in NaCl-acetic acid H^S solution. These tests were performed by submerging all four machined grooves of each specimen in a 6 per cent NaCl-0.5 per cent acetic acid aqueous solution saturated with H2S and loading in direct tension to several stress Copyright by ASTM Int'l is (allfrequently rights reserved); 16 15:53:43the ESTresistance 2015 levels. This solution usedWed forDec predicting of Downloaded/printed by stainless steels to cracking in sour well environments. The NaCl addition University of Washington (University of Washington) pursuant to License Agreement. No further reprodu DENHARD AND GAUGH ON ALLOY DEVELOPMENT 49 is necessary to increase the severity of the test and produce cracking in reasonably short lengths of time. The results of these tests, all performed at room temperature, are listed in Table 7 and illustrated graphically in Fig. 6. To resolve the question of the mode of failure in this test medium (whether stress corrosion cracking or hydrogen embrittlement), additional work is currently in progress following the method described by Brown [5], in which the stressed specimens are polarized either positively or negatively, with respect to their corrosion potentials, by means of a potentiostat. This area of endeavor also will provide information on the susceptibility of martensitic stainless steels to hydrogen embrittlement as a function of their heat-treated condition. Conclusions The accelerated stress corrosion test used in this investigation was an efficient means of evaluating the resistance of various alloys to cracking in hot chloride environments. The results obtained on commercial alloys (Fig. 2) closely parallel the findings of others, and of industrial experience. An experimental austenitic alloy, nominally containing 0.03C-20O45Ni-5Mn, was developed to meet the objectives of immunity to stress corrosion cracking and good welding characteristics. It was noted that Type 330 could be either susceptible or fully resistant to chloride cracking, depending on the carbon and nickel contents. For instance, at lower nickel levels, immunity to cracking was attained by increasing the carbon content. However, this approach had the concomitant disadvantage of decreasing the resistance to intergranular attack. These opposing effects were balanced by judicious adjustment of the carbon and nickel levels in a number of experimental alloys (not all reported herein), to arrive at the optimum composition. Manganese, when added in amounts of 3 to 10 per cent, greatly improved the welding characteristics of this alloy. Such additions did not materially affect the resistance to stress corrosion cracking. During this phase of the study, a noteworthy correlation between the hot working and welding characteristics was observed. These alloys exhibited not only improved weldability, but they also forged and hot rolled without the tearing or cracking that often occurred without such additions. Only minor modifications were necessary to adapt this stress corrosion test for the evaluation of Type 410 and 17-4 PH in aqueous NaCl-acetic acid-H2S solutions. The resistance to cracking of both of these alloys was improved by tempering or overaging to lower strength levels. This confirmed the earlier work of Bloom [4], who performed a similar study using other types of test specimens. Also, when heat treated to equal Copyright ASTM17-4 Int'l (all Wed superior Dec 16 15:53:43 EST 2015 strength by levels, PHrights wasreserved); somewhat to Type 410 in reDownloaded/printed by sistance to cracking in this environment. University of Washington (University of Washington) pursuant to License Agreement. No further reprod 50 STRESS CORROSION TESTING A cknowledgments The authors wish to thank J. F. Kreml, for developing much of the stress corrosion test data contained in this paper, and R. H. Espy and J. J. Junod, for carrying out and evaluating the welding tests. We also wish to thank G. N. Goller, manager, Baltimore Research Laboratories; M. E. Carruthers, director, Stainless Research; and T. F. Olt, vice president, Research and Technology, for permission to publish this paper. References [1] Copson, H. C. and LaQue, F. L., Corrosion Resistance of Metals and Alloys, 2nd ed., Reinhold, New York, 1963, pp. 518 and 519. [2] Denhard, E. E., Jr., "Effect of Composition and Heat Treatment on the Stress Corrosion Cracking of Austenitic Stainless Steel," Corrosion, Vol 16, July 1960, pp. 359t-369t. [3] U.S. Patent No. 2,894,833, 14 July, 1959. [4] Bloom, F. K., "Stress Corrosion Cracking of Hardenable Stainless Steels," Corrosion, Vol 11, Aug. 1955, pp. 351t-361t. [5] Brown, B. F., "Stress Corrosion Cracking and Corrosion Fatigue of High Strength Steels," DMIC Report 210, Problems in the Load Carrying Application of High-Strength Steels, 26-28 October 1964, pp. 91-102, Defense Metals Information Center, Battelle Memorial Institute, Columbus, Ohio. DISCUSSION R. W. Staehle1 (written discussion)—Regarding the many possible mechanisms for explaining stress corrosion cracking in a given alloy, I think that we should reconsider the detailed processes which are operating. Generally, the microprocesses of significance are: anodic reactions, cathodic reactions, adsorption and absorption processes (for example, of hydrogen), film formation, bulk transport of interstitial elements, dislocation-surface interactions, and internal dislocation interactions. I think that in most cases there is not sufficient definitive evidence to ascertain to what extent and which of these processes are operating. We should therefore be very careful about reaching conclusions concerning mechanistic interpretations. However, if we are ever to reach our ultimate objective of predicting cracking susceptibility as affected by imposed environmental conditions and by metal chemistry, we must proceed open-mindedly and in detail to determine quantitatively the nature of the microprocesses above and determine then* significance with respect to the specific stress corrosion process. 1 Assistant professor, Department of Metallurgical Engineering, Center, The Ohio State University, Columbus, Ohio. Corrosion Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further repr H. L. Craig, Jr.,1 and H. B. Romans1 A Rapid Stress Corrosion Test for Aluminum-Magnesium Alloys REFERENCE: H. L. Craig, Jr., and H. B. Romans, "A Rapid Stress Corrosion Test for Aluminum-Magnesium AUoys, Stress Corrosion Testing, ASTM STP 425, Am. Soc. Testing Mats., 1967, p. 51. ABSTRACT: The susceptibility of aluminum-magnesium alloys to stress corrosion cracking may be determined using a rapid, electrolytic test method. The stressed specimen, deflected as a beam, is made the anode in a 3.5 per cent salt solution by impressing a current of 40 ma/in.2 Prior to testing, a set of specimens may be heated, each one for an increased period of time over the previous one, to develop susceptibility in a part of the set of specimens. Test conditions have been chosen to maximize chances of failing materials with low degrees of susceptibility, while still not failing materials known to be nonsusceptible. Results with two commercially produced alloys are presented. Alloy 5083 shows less tendency to develop susceptibility than alloy 5456. The effect of cold work in accelerating the sensitivity to stress corrosion cracking is also shown. KEY WORDS: aluminum-magnesium alloys, stress corrosion, corrosion, aging, heat treatment, electrolysis, salt solutions During the past ten years, wrought aluminum alloys containing 4 to 5.5 per cent magnesium have become available on a commercial basis in the United States. Two such alloys are known as 5083 and 5456, the prime difference between the two being a slightly higher nominal magnesium content in 5456 alloy leading to slightly better mechanical properties in the annealed condition. They were foreshadowed in their appearance here by widespread usage in Great Britain and Europe [I].2 In addition to a high degree of corrosion resistance, these alloys are easily welded and possess good mechanical properties. This combination of factors places them in a position to compete with steel in such structural uses as the fabrication of railroad cars, barges, and ships. One specific 1 Research supervisor and scientist, respectively, Department of Applied Chemistry and Mathematics, Reynolds Metals Co., Richmond, Va. Personal members ASTM. 2 The italic numbers in brackets refer to the list of references appended to this Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 paper. Downloaded/printed by 51 University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized. 52 STRESS CORROSION TESTING example is a railroad gondola with a 271,000-lb gross weight capacity, made possible by using a 5083 alloy body for the car. In the literature, there are reports about the possibility of stress corrosion cracking occurring with these alloys when cold-worked, either as a result of exposure to temperatures in the range of 150 to 400 F, or aging at room temperature for many years [2]. Most published data have been developed using laboratory-produced, thin sheet specimens, while the commercial production mentioned above began with plate. There is a considerable difference in the fabrication and resultant metallurgical structure of sheet as compared with plate. The authors' laboratory has primarily been occupied with the testing of plant produced metal ranging from 0.125 to 1.25-in. gage heavy sheet and plate. This program was conducted to verify or disprove the above mentioned reports and interpret their significance in terms of commercial production. A conventional approach to this testing is to expose suitably prepared specimens to alternate immersion in 3.5 per cent sodium chloride solution, while they are stressed to 75 per cent of their yield strength. A device for accomplishing this objective is known as a "stress rack." The design for this rack was adapted from plans supplied by the National Bureau of Standards. The preparation of tension-test-type specimens for this method is comparatively expensive. The procedure also entails the determination of mechanical properties. When a large number of metallurgical variables are to be investigated, it is not feasible to test more than two or three specimens per variation. This is an acknowledged weakness of the test method, as there is a widespread scatter inherent in stress corrosion cracking data. If two specimens last a specified time in a given test, there is really no assurance that if a third one were tested, it would not fail. Furthermore, the corrosion resistance of the aluminum-magnesium alloys in salt solution is so high that it requires long-exposure times for a significant amount of attack to occur. Even then, it is not always clear whether a failure is due to pitting attack which reduced the cross section or to true stress corrosion cracking. To overcome these difficulties, the following criteria were used to develop a rapid laboratory stress corrosion test: 1. In testing closely related materials, it should be sufficient to stress them all to the same value, as inaccuracies in setting stress levels mask small differences in mechanical properties. 2. A large number of specimens of a given metal variation should be tested, including both specimens that fail and specimens that do not fail in the test. 3. Failure, when it occurs, should happen in a short time, so that pitting is not a significant factor in assessing the failure. 4. Thebytest should berights ablereserved); to pickWed outDec materials with low degree of Copyright ASTM Int'l (all 16 15:53:43 EST a2015 susceptibility. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduction CRAIG AND ROMANS ON ALUMINUM-MAGNESIUM ALLOYS 53 Degree of Susceptibility In alloy systems that are liable to stress corrosion cracking, various degrees of susceptibility are found. These degrees may be defined with respect to three terms of reference: (1) intensity of corrosion, (2) stress level, and (3) metallurgical factors. In the absence of corrosion attack, stress corrosion cracking cannot occur. On the other hand, cracking often occurs when very little general corrosion has taken place rather than when the metal is severely pitted or otherwise corroded. In long-term tests, failure of a stressed specimen may occur due to reduction in cross section by pitting attack. It is difficult then to decide if the specimen represents a low degree of susceptibility or not. In the case of early failures, a high degree is easily assessed. For purposes of this present test, the corrosion attack is made intense so that materials with "low degrees of susceptibility" as determined in less corrosive environments fail quickly. The same attitude is adopted toward the stress level: "no stress" should produce "no failures," and increasing stress levels should result in shorter and shorter lives. In testing similar materials with but slightly different mechanical properties, a high stress level above the yielding point tends to mask these differences, and it also eliminates the need for determining the yield strength for each specimen. It is possible to vary the metallurgical condition of some alloys to produce a range of susceptibilities to stress cracking. This is the procedure which was adopted for purposes of this test. Sensitization In the aluminum-magnesium alloy system, an important variable is the age of the metal. It is a well-documented fact that freshly prepared alloys with all the magnesium in supersaturated solution are not susceptible to cracking. However, prolonged aging at ambient temperatures or shorter times at slightly elevated temperatures (under the solid solubility curve in the phase diagram) result in precipitation of magnesium in the form of an intermetallic compound with aluminum. If this precipitate is distributed unfavorably, by being concentrated mainly in the grain boundaries, the structure is susceptible to stress corrosion cracking. This aging process is called sensitizing. This precipitation phenomenon is not completely understood, nor is the relationship between precipitation and stress corrosion cracking entirely explained on the basis of any simple mechanism. The role of dislocations has not been studied yet for this system. It is easy to demonstrate in a simple binary alloy a process that might be called the "classical example" of precipitation at grain boundaries. This condition results in stress corrosion cracking by intergranular attack Copyright by (all rights reserved); Wed or Deca 16 2015 localized by ASTM stress Int'l concentrations into one few15:53:43 paths EST of failure Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further rep 54 STRESS CORROSION TESTING across the specimen. However, it is more difficult, if not impossible, to correlate microstructure of complex alloys with their corrosion behavior. For this reason, it is necessary to test a large set of specimens that have been subjected to a variety of sensitizing treatments, or aging practices. A degree of susceptibility may then be defined as the time required to produce a susceptible structure from one that is nonsusceptible, by FIG. 1—Effect of time at temperature on susceptibility to stress corrosion cracking—NP6-Y4H. either aging at ambient or slightly elevated temperatures, for example, a material that fails in a stress corrosion test after being heated for one week at 212 F is more susceptible than one that does not fail until after it has been heated for one month at 212 F. Outline of Test Thus, a stress corrosion test for the aluminum-magnesium alloys consists of the following steps: 1. Heating a set of specimens at various times in a temperature region calculated to produce susceptibility. 2. Stressing the specimens to a high tensile stress value. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 3. Exposing the specimens to a corrosive environment. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions auth CRAIG AND ROMANS ON ALUMINUM-MAGNESIUM ALLOYS 55 Example Due to the favorable experience reported for British alloys, a lA-m. gage plate was obtained of a 5 per cent magnesium alloy under the British Standard Specification 1470 for Wrought Materials of Alloy NP6 - 1A H. The chemical analysis is: Si Fe Cu Mn Mg Cr 0.11 0.14 0.03 0.23 4.5 0.01 Ni Zn Ti <0.01 0.07 0.01 FIG. 2—Specimen position'in parent plate. The mechanical properties are: Tensile Strength, psi Yield Strength, psi Elongation, % 43 200 43 700 32 500 31 700 19.7 18.8 Longitudinal Transverse . Results of stress corrosion testing using direct loading tension test specimens, alternately immersed in a 3.5 per cent salt solution, gave the following results: Condition As received Specimens first heated one week at 212 F. . . . Specimens first heated one month at 212 F . . Days to Failure, Duplicate Specimens 942, one specimen not failed 157, 173 2,4 The alternate-immersion cycle was 10 min immersed in each hour. Specimens of the same plate were taken according to the procedure outlined in detail below. Figure 1 gives the results of 92 tests, each with Copyright by ASTM Int'lfirst (all rights reserved); Wedgiven Dec 16time 15:53:43 EST specified 2015 duplicate specimens heated for the at the tern- Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproducti 56 STRESS CORROSION TESTING perature. Note that susceptibility to cracking is rather easily induced in this alloy as revealed by this test through exposure to slightly elevated temperatures. The test itself lasts but one day, and the results may be compared with the data given previously, where it took an average of 165 days for the specimens sensitized for one week at 212 F to fail in the alternate-immersion test. Two other related points should be noted: whereas the specimen tested "as received" failed in the alternate-immersion test, another specimen taken from the same parent plate passed this test. Examination indicated that the as-received material failed due to a reduction of cross-sectional area by pitting. Then also, it should be noted that an extrapolation of an "imaginary" line between nonfailures and failures obtained by heating specimens does not follow a straight line, particularly at the ambient temperatures. FIG. 3—Specimen holder. The main conclusion to be drawn from this example is that the test used causes failures of specimens with very low degrees of susceptibility, as determined by other tests, but that it does not cause failures in nonsusceptible material. Experimental Procedure 1. Prepare specimens with machined (milled) transverse surfaces, 3.00 by 0.064 in. by the plate gage. See Fig. 2 for position of specimen with respect to rolling direction. If the gage exceeds 0.75 in., cut the specimens down to a width of 0.75 in. or less, retaining at least one rolled surface. 2. Smooth the 3.00-in.-long edges with a fine file or sandpaper. 3. Stencil or otherwise mark specimen on one end to identify it. 4. Sensitize according to the following schedule: (a) as received, (ft) one weekbyatASTM 212 F, at Wed 212 Dec F, and (d) longer periods and Copyright Int'l(c) (allone rightsmonth reserved); 16 15:53:43 EST 2015 other temperatures Downloaded/printed by as desired. University of Washington (University of Washington) pursuant to License Agreement. No further reproductio CRAIG AND ROMANS ON ALUMINUM-MAGNESIUM ALLOYS 57 5. Degrease sensitized specimens with acetone. Handle by ends from this point on. 6. Insert specimen in holder (Fig. 3) and tighten screw to hold in place. Do not tighten enough to deflect specimen. 7. Place holder in a clamp; bring dial gage up until the point touches specimen. 8. Adjust for zero displacement of specimen. Then turn down on screw until the necessary displacement is obtained. See the Discussion section below. A value of 8 X 10~3 in./in. for the nominal outer fiber strain has been used in this work. 9. Place holder in a 600-inl beaker containing 340-ml 3.5 per cent salt solution. Attach an alligator clip to the end of the specimen. (This amount of solution in a 600-ml beaker will immerse 2 in. of specimen in solution. This length is used hi the area calculation.) 10. Insert a piece of metal of the same alloy composition opposite the specimen to serve as a cathode. 11. Make electrical connection to a dry cell, storage battery, or rectifier through a variable potentiometer. 12. Adjust the current so that a value of 40 ma/in.2 of immersed specimen surface flows, with the specimen acting as the anode. (A drop of phenolphthaiein in solution will turn solution around cathode pink.) NOTE: For specimens wider than 0.250 in., it is necessary to reduce the current density to 25 ma/in.2, due to shielding of the reverse side of the specimen by the holder. The test is not very sensitive to changes in the current density in this range. 13. Leave on test for 22 to 24 hr, or until a crack appears in the specimen, whichever is sooner. 14. After the test period is over, remove the specimen holder and examine the specimen in place. A failure has occurred if a crack is visible. If no crack is visible, slowly deflect one end until it touches the opposite side of holder. In this manner, any incipient or microscopic cracks may become visible. 15. Report specimens: If the specimen fails as in Step 14, designate as F. If the specimen does not fail, designate as N. Discussion Specimen Design One of the most difficult problems in stress corrosion testing is sampling the metal. Especially in heavy plate, the structure may vary through the thickness. Distinct differences in susceptibility are found depending on the orientation of the grain structure with respect to the direction of tensile stress. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 A simple scheme Downloaded/printed by of three-point loading of a beam specimen with uni- University of Washington (University of Washington) pursuant to License Agreement. No further reproduct 58 STRESS CORROSION TESTING form cross-sectional area was selected [3]. It was then determined that the most sensitive direction with respect to stress corrosion was in the transverse plane, perpendicular to the direction of working the metal (Fig. 2). In addition to this feature of maximum sensitivity, machining specimens is a simple operation. This procedure also has the added benefit of testing the entire cross section, up to 0.75-in. gage plate. Heavier plate may be milled or sawed to reduce its thickness in half and then specimens taken. In the case of 3-in. or thicker plate, a short transverse specimen can be taken. At the other end of the gage scale, sheet as light as 0.125 in. has been tested successfully. Gages lighter than this are tested with the rolled surface being stressed. Allowances should be made for TABLE 1—Variation in response to stress corrosion test depending on plane of stress. Stressed Surface* Rolled Surface (transverse) 0 Condition As received Sensitized at 212 F, days 3.5... . 7 14 28 56 Perpendicular to Rolled Surface Specimen 1 Specimen 2 Specimen 1 Specimen 2 N N N N N F F F not tested N N N N not tested F F F F F N N F F F a Specimens from 5456 alloy sheet, 0.125-in. gage. Specimens 1 and 2 are not duplicates, but represent different fabricating practices, which were expected to respond with different degrees of susceptibility. 6 N — no failure; F = failure in test. reduced sensitivity when a comparison is made between these tests and those made of a transverse section. Table 1 gives two sets of data collected to show that the transverse direction is more sensitive than the rolled surface to stress corrosion cracking. Specimen Holder Design A 1 Vz-in.-diameter rigid poly (vinyl chloride) pipe is used for the holder body. For purposes of illustration, however, Fig. 3 shows a model of the holder with a body of clear plastic tubing. After some experimentation with four-point loading and several methods of three-point loading, the best deflector was found to be a screw machined from Vs-in.-diameter nylon rod. Studies of stress distribution using the Photostress technique showed an acceptable ratio of E\, the strain perpendicular to the end supports, to E2, the strain parallel to the supports. The end supports Copyright ASTM Int'lbolts (all rights Dec 16 15:53:43 2015 port consist of byaluminum withreserved); collars Wed of glass tubing. AnEST access Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprod CRAIG AND ROMANS ON ALUMINUM-MAGNESIUM ALLOYS 59 TABLE 2—Stress corrosion failure times in four test media, as a function of strain. Strain, in./in. 0.002 0.025 0.003 0.004 0.005 0.006 0.008 Alternate Immersion 3.5% NaCl F (26 days) F (13 days) F (7 days) Chromic Acid-6 Salt Solution F F F F (13 days) (25 days) (26 days) (11 days) F (1 day) Constant Immersion-ImpressedCunent0 3.5% NaCl Chromic AcidSalt Solution N (15 hr) F (280 min) F (90 min) N (20 hr) F (14 hr) F (13 hr) F (20 min) F (120 min) 0 A value of 40 ma/in.2 is used. 36 g/1 chromic acid (tech) plus 30 g/1 potassium dichromate (tech) plus 3 g/1 sodium chloride. 6 is drilled in the pipe wall opposite the point of loading for the dial gage point (not shown in Fig. 3). This holder is supported in a suitable clamp, the specimen inserted, and the deflector screw lowered until it just touches the specimen. The dial is raised until it touches the specimen opposite the deflector screw. The screw is then lowered until the desired amount of deflection is reached. This is calculated from the formula for the deflection of a simple beam, d = EL2/6t, where d is the deflection (in.); E, the nominal outer fiber strain (in./in.), L, the distance between end supports (in.); and t, the thickness of the specimen (in.). Two hundred holders were drilled from a jig, ten holders measured, and an average for L calculated to be 2.48 in. To simplify calculations, the constants in the above equation are collected together into a "deflection factor," which gives the deflection for 0.001 in./in. strain. Using the above figure for L, values of the deflection factor are: Thickness, in. Deflection Factor 0.050 0.055 0.060 0.065 0.070 0.0205 0.0187 0.0171 0.0158 0.0147 0.125 0.0082 Most specimens tested have been 0.064-in. thick, but specimens between 0.050 and 0.125 in. thick have been satisfactorily tested. To find the desired deflection, the thickness of the specimen is measured, the corresponding deflection factor is found from the table and multiplied by the desired strain value. A value of 8 X 10~3 in./in. has been used ASTM Int'lFor (all rights reserved); Wed Dec 0.065-in. 16 15:53:43thick EST 2015 inCopyright most ofbythis work. example, a specimen would be Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further repro 60 STRESS CORROSION TESTING deflected 0.1264 in. (8 X 0.0158). This is well above the yield point, and, consequently, the strain value quoted is not accurate as the simple beam formula does not hold; however, the value is reproducible and convenient for testing of alloys with yield strengths in the range of 30,000 to 40,000 psi. FIG. 4—Allay 5083 sensitized one week at 297 F, HNO, etch (XlOO). A comparison was made of constant deflection using this holder and constant load, using an adaptation of the holder, loading the specimen by means of lead weights placed on a pan riding on a rod hi place of the screw. The material tested in this case was an aluminum-7 per cent magnesium sheet, 0.050-in. gage, cold rolled 36 per cent, and aged for three years at ambient laboratory temperatures before this testing. In this Copyright ASTM Int'l was (all rights reserved); Wed Dec 16 15:53:43 EST 2015 instance,bythe strain varied: Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further repro CRAIG AND ROMANS ON ALUMINUM-MAGNESIUM ALLOYS Strain, in./in. 0.001. 0.002. 0.003. 0.005. Constant Load N F not tested not tested 61 Constant Deflection not tested N N F FIG. 5—Alloy 5083. As expected, the condition of constant load lowered the strain required for failure. Corrosion Medium Two variations in two solutions were tried before selecting the test: alternate immersion in a solution and constant immersion with an impressed current, using the specimen as the anode [4]. In this instance, a very susceptible material was used, whose susceptibility developed by natural aging at ambient temperature. The metal was a binary aluminum7 per cent magnesium alloy sheet, 0.050-in. gage, which had been cold rolled 22 per cent, and aged two years before these tests. The time-toCopyright by ASTM Int'l (all rights reserved); Dec 16 in 15:53:43 2015 failure as well as response to strain levelWed is quoted TableEST 2. A specimen Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduc 62 STRESS CORROSION TESTING of the material, solution heat treated 30 min at 930 F and quenched in cold water before the corrosion test, did not fail when strained at a value of 0.008 in./in. using impressed current in a 3.5 per cent salt solution. This behavior was due to a resolution of precipitate in addition to the effect of annealing and consequent loss of properties. This series of tests shows that an impressed current greatly accelerates the time-to-failure over an alternate-immersion test at equivalent strain values. The interesting thing is that the effect of a chromic acid-salt solution is reversed by the impressed current: in alternate immersion this latter solution is more aggressive than plain salt, whereas under the influence of an impressed current, the chromic acid inhibits cracking. These results indicate that testing with an impressed current is definitive of stress corrosion cracking, in spite of its apparent severity, since low strain values do not cause failures in susceptible material, nor does a high strain value cause failure in a nonsusceptible material. The function of the impressed current is to overcome the inherently good corrosion resistance of the aluminum-magnesium alloys. The dissolution appears to follow the grain boundaries, which are the site of the stress corrosion crack. In adopting this means of shortening the test time, the philosophy is analogous to that proposed by Brown [5] in his precracked cantilever-beam test. Whereas Brown's test is limited to materials with relatively high yield strengths, the present test only requires that a properly selective corrosive environment be found for the material to be tested. There is no limitation on the mechanical properties of the test alloys. Type of Failure Metallographic examination of a typical failure is shown in Fig. 4. The characteristic intergranular nature of the failure is evident. Only slight pitting has occurred during this exposure which was 30 min. A similar specimen left on test for the complete 22 hr was pitted as well as cracked. In susceptible metal, the cracks appear to form in preference to pits, while in nonsusceptible material uniform dissolution occurs by general, shallow pitting, or even fairly uniform etching of grain faces. This behavior of specimens in the test fulfills the third criterion, which is to allow susceptibility to stress cracking to be measured without the interference by reduction in cross-sectional area due to pitting attack. Results with Commercially Produced Plate The usefulness of this method can be demonstrated by the results obtained in comparing two alloys, in the form of 0.250-in. gage plate, mill fabricated. In addition to the alloy variable, the effect of cold work was studied: four levels—0, 5, 15, and 40 per cent cold reduction. Figure 5 shows alloy 5083, tested after sensitizing as described above Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 at four temperatures, from 1 to 120 days plus the as-received material Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further repr CRAIG AND ROMANS ON ALUMINUM-MAGNESIUM ALLOYS 63 after about 1100 days ambient temperature aging. The failed condition is represented by an "X," the unfailed result by a dot. A few anomalous results are evident, but the majority of the tests follow a logical pattern, showing the development of a susceptible condition after a period of time, depending on the specific temperature. Figure 6 shows the influence of increasing the magnesium content on this rate of precipitation. Alloy 5456 has, on the average, 0.75 per cent more magnesium (5.25 versus 4.5 per cent nominal). The effect of cold work was changed by the higher magnesium, as a comparison of the two FIG. 6—Alloy 5456. patterns of behavior at 15 per cent cold reduction readily shows. The same trend is evident at 5 and 0 per cent, also. Conclusions A rapid stress corrosion test for aluminum-magnesium alloys has been developed consisting of exposing a deflected-beam specimen, 3 by 0.064in. by the gage of the parent plate, to a 3.5 per cent salt solution under 2 anCopyright impressed current ma/in. while to a high by ASTM Int'l of (all40 rights reserved); Wed stressed Dec 16 15:53:43 EST strain 2015 value. To determine a degree of susceptibility for a given material, a series of Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduc 64 STRESS CORROSION TESTING specimens is heated at a temperature, 212 F, for periods of time calculated to produce some failures in the test. This test lasts but one day, in addition to the time required to sensitize the material to stress corrosion cracking. The conditions of the test have been chosen to maximize the chances of failure of materials with low degrees of susceptibility, while still precluding failure of nonsusceptible material. A cknowledgment The authors gratefully acknowledge the contributions of the staff of the Metallurgical Research Division in evolving this procedure. B. A. Niemeier aided in the development of the specimen holder, and J. S. Prestley contributed the metallographic studies. R. G. Connell, Jr. determined the effect on constant load. Numerous persons have kindly discussed these methods with the authors and have aided in their development. They are: S. J. Sansonetti, director of research technology and applied science, and L. E. Householder, chief metallurgist, both of Reynolds Metals Co.; F. A. Champion, British Aluminum Co.; Fred Reinhart, National Bureau of Standards, and especially H. P. Godard, Aluminum Laboratories, Ltd., who provided the initial suggestion of using an impressed current. References [1] Champion, F. A., 'The Interactions of Static Stress and Corrosion with Aluminium Alloys," Journal of the Institute of Metals, Vol. 83, 1954-1955, p. 390. [2] Dix, E. H., Jr., Anderson, W. A., and Shumaker, M. B., "Influence of Service Temperature on the Resistance of Wrought Aluminum-Magnesium Alloys to Corrosion," Corrosion, Vol. 15, 1959, pp. 55t-62t. [3] Fraser, J. P., Eldridge, G. G., and Treseder, R. S., "Laboratory and Field Methods for Quantitative Study of Sulfide Corrosion Cracking," Corrosion, Vol. 14, 1958, pp. 517t-523t. [4] Booth, F. F. and Godard, H. P., "An Anodic Stress-Corrosion Test for Aluminum-Magnesium Alloys," Proceedings, First International Congress on Metallic Corrosion, London, 1961. [5] Brown, B. F., "A New Stress-Corrosion Cracking Test Procedure for HighStrength Alloys," Materials Research & Standards, Vol. 6, No. 3, March 1966, pp. 129-133. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No fu DISCUSSION ON ALUMINUM-MAGNESIUM ALLOYS 65 DISCUSSION A. R. C. Westwood1 (written discussion)—What is the significance of the 22-hr test period adopted in this work? Could equivalent data be obtained in, say, a 4-hr test period? It is possible that the role of copper ions in the environment is to act as step-poisons, changing corrosion from general to specific and causing tunnelling or pitting. The copper may be present as simple ions or complexes. A discussion of such possibilities may be found in Corrosion Science,2 and on p. 228 of this symposium. H. L. Craig, Jr., and H. B. Romans (authors)—A 22-hr test period was selected on the basis of obtaining the maximum number of failures without causing failure of nonsusceptible material. Equivalent data are not obtained in 4 hr, as some failures occur after that time limit. We have not investigated the role of copper ions in this test, as the alloy contains less than 0.1 per cent, and either distilled or deionized water is used to make up the solutions. Later work with high-purity aluminum has shown the influence of trace amounts of copper during pitting corrosion, but its role in stress corrosion cracking has not yet been defined. See the paper by Romans and Craig on p. 363 of this symposium. M. B. Shumaker3 (written discussion)—This type of test has also been of considerable interest to us at Alcoa. Of particular concern has been the possibility of the test failing material having a microstructure only slightly outlined at the grain boundaries with aluminum-magnesium precipitate and having performed well in service. What evidence do the authors have as to whether this does or does not occur with their tests? H. L. Craig, Jr., and H. B. Romans (authors)—The correlation between this test and atmospheric stress corrosion testing is being presented in paper No. 27 at the Symposium on Atmospheric Testing, 25-30 June, 1967. Specimens which fail this test, in 22 hr, have failed after five years in the atmosphere. There have been no reports of service failure of 5083 alloy, except where inadvertent exposure to elevated temperatures has occurred. 1 Associate director, Research Institute for Advanced Studies, Martin Co., Baltimore, Md. 2 Westwood, A. R. C., Corrosion Science, 1966. 3 Research engineer, Chemical Metallurgy Div., Alcoa Research Laboratories, New Kensington, Pa. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further repro /. H. Mulherin1 Influence of Environment on Crack Propagation Characteristics of HighStrength Aluminum Alloys REFERENCE: J. H. Mulherin, "Influence of Environment on Crack Propagation Characteristics of High-Strength Aluminum Alloys," Stress Corrosion Testing, ASTM STP 425, Am. Soc. Testing Mats., 1967, p. 66. ABSTRACT: The stress corrosion behavior of four alloys of aluminum has been investigated. Using cantilever-loaded fatigue-cracked specimens, the crack propagation characteristics of 7178-T6, 2014-T6, 2024-T351, and 7075-T6 aluminum alloys were determined. The results are interpreted in terms of fracture mechanics parameters. Where susceptibility to stress corrosion exists, the rate of crack propagation is dependent upon stress intensity and environment. A method is also presented to obtain threshold values of stress intensity for stress corrosion attack with minimal experimental effort. The aqueous environments used were immersion in distilled water, immersion in 3Vz per cent sodium chloride, and alternate immersion in salt solution and air. The cantilever technique is evaluated and the results compared to other testing methods. KEY WORDS: corrosion, stress corrosion, cracking, crack propagation, fracture mechanics, aluminum alloys Nomenclature a ag o0 B BN d K Klc K0 L M m Crack depth, in. Crack extension in stress corrosion environment, in. Initial value of a, in. Gross specimen thickness, in. Net specimen thickness, in. Gross specimen depth, in. Stress-intensity factor, psi • in.1/2 Plane-strain fracture toughness, psi-in.1/2 Initial value of K, psi • in.1/2 Half-span length for three-point-loaded notch bend specimens, in. Applied bending moment, in • Ib Exponent in the correction factor for the side notches 1 Copyright by ASTM Int'l (all rights reserved); Dec 16 15:53:43 ESTPhiladelphia, 2015 Metallurgist, Department of Army,Wed Frankford Arsenal, Pa. Downloaded/printed by 66 University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authori MULHERIN ON HIGH-STRENGTH ALUMINUM ALLOYS 67 Force, Ib Failure time, min Poisson's ratio P // v In the usual testing method for the determination of stress corrosion susceptibility of aluminum alloys, a ring, bent beam, or U-bend specimen is used as discussed by Sprowls and Brown [I].2 Since these specimens are carefully fabricated to minimize stress raisers, they can be generally characterized as "nondefect" specimens. In the experimental procedure, TABLE 1—Nominal chemical compositions of the aluminum alloys investigated (weight per cent}. Alloying Elemc:nt Cu Mg Mn Si Zn Cr . . 2014 2024 7075 7178 44 4.5 1.5 2.5 2.0 5.5 0.3 6.8 0.3 0.4 1.5 0.6 0.8 0.8 2.7 TABLE 2—Mechanical and fracture properties. Alloy Designation Temper Plate thickness, in Direction0. ffu i t psi (fya , psi Elongation,6 % Average Kjc , psi- in.1'2 2014 T-6 y. LT 70 400 63 300 12 9 24 000 2024 T-351 1% ST 57 600 46 700 37 22 100C 7075 7075 7178 T-6 T-6 1% ST 79 700 69 100 4.0 23 200 T-6 y. LT 84 100 74 100 10 5 20 600 1/4 ^ LT 91 700 85 300 11.3 18 500 0 LT denotes the long-transverse direction and ST the short-transverse direction. 6 Two-inch gage length in the %-in. plate and ^-in. in the 1^-in. plate. c Lower bound value. the stressed specimens are exposed to either a natural atmospheric or laboratory environment. Failure is considered to occur when a crack apparent to the naked eye has developed. This test method concerns the period for crack initiation plus the time period for macroscopic crack development. In contrast with the conventional approach, this paper is concerned with the crack propagation characteristics of the material. There are advantages in using this approach. First, it is generally recognized that with the usual testing technique, crack initiation is sensitive to the surface condition of the test specimen. By starting with a precracked specimen, 2 The italic numbers in rights brackets refer Wed to the of references appended to this Copyright by ASTM Int'l (all reserved); Declist 16 15:53:43 EST 2015 paper. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions 68 STRESS CORROSION TESTING FIG. 1—Specimen configuration, nominal dimensions, and three-point loading locations. SHORT-TRANSVERSE SPECIMEN LONG-TRANSVERSE SPECIMEN FIG. 2—Schematic showing the location of the major specimen notch in relation to the rolling direction. this area of uncertainty is minimized. Second, it has been demonstrated that a titanium alloy was immune in a usual stress corrosion test using nondefect type specimens. However, in the presence of a crack, this material was very susceptible to stress corrosion cracking [2]. In a real structure, cracks can originate from a number of phenomena, for example, welding operations or fatigue. Therefore, the behavior hi the presence ofCopyright a crack, by regardless of origin, is pertinent to the evaluation and applicaASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 tion of the material. Third, by the use of specimens for which a stress Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No f MULHERIN ON HIGH-STRENGTH ALUMINUM ALLOYS 69 analysis is available, fracture mechanics provides a well-developed analytical approach by which the stress situation can be described [3, 4]. Finally, by eliminating the crack initiation phase in the study, possible additional insight into the mechanism of crack growth may be obtained. Experimentally, notched and fatigue-cracked bend-bar specimens were used. These specimens were cantilever loaded as in similar work reported by Brown [2]. A stress-intensity factor, K, from fracture mechanics is used to describe the macroscopic stress field at the crack front. The crack propagation characteristics of 2014, 2024, 7075, and 7178 aluminum alloys were studied in several laboratory aqueous environments. Experimental Procedure Materials The nominal chemical compositions of the four alloys used in this investigation, 2014, 2024, 1075, and 7178, are shown in Table 1. The temper designations, engineering mechanical properties, and fracture data in the direction investigated are summarized in Table 2. Test Method The specimens used were side-notched bend bars as shown in Fig. 1. Side notches were introduced in the plane of the anticipated crack propagation to approach plane-strain conditions for fracture toughness tests conducted under ambient laboratory conditions. For the sake of standardization, side notches were used through the entire experimental program. In the short-transverse direction, the L dimension was increased to 2 in. by the use of adapters for the fracture toughness tests. The orientation of the notches in relation to the plate is shown in Fig. 2. Fracture toughness data were obtained using three-point loading on a universal testing machine. Electrical resistance strain gages bonded to the crack tip area were used as the deformation sensing device to obtain load-deformation curves on an X-Y recorder. For the environmental tests, the method of loading was with a cantilever arrangement. These tests were run with a constant applied load. The environmental containers consisted of flexible polyethylene cups of approximately 150 cm3 volume. Although the liquid medium was not circulated, the entire notched area of the specimen was exposed to the environment. The environments used were constant immersion in distilled water, constant immersion in 3l/2 per cent aqueous solution of sodium chloride (NaCl), and an alternate-immersion cycle consisting of 10 min in 3J/2 per cent NaCl solution and 50 min in air. During the environmental tests, evaporation losses were restored with distilled water. Copyright by was ASTMmade Int'l (all rights reserved); Wed Dec 16 15:53:43humidity EST 2015for No attempt to control the ambient laboratory Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further r 70 STRESS CORROSION TESTING the alternate-immersion cycle. The environmental test facility has been previously described in Ref 5. Analytical Procedure In this paper, the stress is described by means of a stress-intensity parameter, K, from fracture mechanics. K is a parametric description of the magnitude of the stress field in the vicinity of the crack front. As a measure of the material response to this stress field, either a KIO or K0 designation is used. Klc represents the plane-strain fracture toughness obtained in a fast fracture test, and K0 represents the initial stress intensity applied to environmentally exposed specimens. The three-point loading equation of Srawley and Brown [6] was used as the analytical basis for the Klc computation. This equation, however, does not include the influence of side notches. Freed and Krafft [7] proposed a correction factor to account for the side notches through a ratio of the unnotched-to-notched specimen breadths, B and BN , respectively. The modified equation used as the basis for the calculation including the correction factor is Since the cantilever-loaded specimens have a shear stress, it appeared reasonable to use a three-point loading equation rather than a fourpoint loading equation for calculating the K0 values. However, the results are substantially the same. The equation can be modified to include a moment term, M, with consideration that M = % PL. Therefore, the equation used for calculating K0 including the side notch correction factor is Experimental Results and Discussion 7075-T6 Aluminum Alloy —Short-Transverse Direction For other materials such as steel and titanium, the relationship between the initial applied stress intensity, K0 , and the time-to-failure, // , has been presented in the literature [2,5]. Many of these data are characterized by extremely short failure times at high K0 levels. At lower K0 levels an "apparent" endurance limit is reached as indicated by an abrupt lengthening of the time-to-failure. This type of data was obtained for 7075 alloy in the solution treated and artificially aged conCopyright by the ASTM Int'l (alloriented rights reserved); Wed Dec 16 15:53:43 EST dition with specimens in the short-transverse direction. Downloaded/printed by University of Washington (University of Washington) pursuant to License 2015 Agreement. No MULHERIN ON HIGH-STRENGTH ALUMINUM ALLOYS 71 The plane of anticipated crack propagation in these specimens is parallel to the plate surface at midthickness (Fig. 2). The variation in the failure time as a function of the initial stress intensity for the three environments is shown in Fig. 3. For a constant immersion in 3^ per cent NaCl solution, the failure times gradually increased with decreasing initial stress intensity. No abrupt change was FIG. 3 —Delayed failure characteristics of 7075-T6 aluminum alloy in the shorttransverse direction. found in the relationship. In distilled water, the length of time necessary for specimen fracture was substantially longer than in the salt solution. However, over the limited stress-intensity range used in the distilled water experiments, the curves were of the same general shape. For the alternate-immersion cycle, in a range of K0 values between 11,500 and 15,700 psi-in.1/2, the failure times closely approximated the data obtained under constant immersion. Determinations were also made of the crack extensions which occurred Copyright by ASTM (all rightssalt reserved); Dec 16 15:53:43 EST 2015 during exposure to Int'l a constant water Wed immersion for selected time Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further 72 STRESS CORROSION TESTING FIG. 4—Crack propagation of 7075-T6 in the short-transverse direction for several exposure times. Also shown is the growth necessary to cause catastrophic fracture of the specimen. Arrows indicate the anticipated Ko at which failures of the specimens occur at 4000, 6000, and 8000 min (from Fig. 3). FIG. 5—Schematic showing the relationship between K and a for a constant Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 moment. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further r MULHERIN ON HIGH-STRENGTH ALUMINUM ALLOYS 73 periods. The resulting data (Fig. 4) show the relationship between the three variables, initial stress intensity, crack extensions, and exposure time. With an increment of crack extension, an incremental increase will occur in the stress intensity per Eq 2. The generalized relationship is shown in Fig. 5. With increases in a, related increases will occur in K until the catastrophic value of K is attained, Klc. From the schematic Fig. 5, it can be seen that a change in the initial crack length will cause a variation in the rate of change of K with a. This variation will also occur with a change in some other specimen dimensions such as the depth. Therefore, crack extension data as a function of exposure time and Ko shown in Fig. 4 are sensitive to the specimen configuration to a greater degree than the time-to-failure. A test to determine the time-to-failure FIG. 6—Photograph of fracture surfaces of the 7075-T6 aluminum specimens oriented in the short-transverse direction showing the crack propagation which occurred during 6000-min exposure in an environment of 3Y2 per cent NaCl solution. From left to right K = 3200; 3300; 6100; 9200; 9300; 12,000; and 12,000 psi-in.1/2 would cover a much greater range of both a and K and thus would be less sensitive to initial crack depth. Obviously, a variation in the rate of change of K as a function of the initial crack depth can also be substantial among the various types of specimens with different stress analyses. Beachem and Brown3 have reported the influence of specimen configuration on the tf parameter. Also shown on Fig. 4 is the extension of the crack necessary to cause catastrophic fracture as a function of the original stress intensity. These data were obtained from the specimens exposed to a constant immersion of 3^ per cent NaCl reported as tf data on Fig. 3. The arrows along the catastrophic fracture curve show the K0 level at which fracture of the specimen would be anticipated to occur after exposure for 4000, 6000, and 8000 min. These values were obtained from Fig. 3. Extrapolation of the exposure curves to these values shows substantial agreement. The successful use of this crack extension technique is contingent upon the delineation and measurement of the mechanically induced fatigue crack, the stress corrosion crack, and the catastrophic fracture 3 Copyright See p.by31.ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized. FIG. 7—Composite photomicrograph of a stress corrosion crack in a short-transverse specimen of 7Q75-T6 aluminum alloy showing intergmnular propagation. Reduced 31 per cent in reproduction. Arrow indicates approximate termination of the fatigue crack. Keller's etch (X/00). Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized. MULHERIN ON HIGH-STRENGTH ALUMINUM ALLOYS 75 crack. Immediately after exposure, these specimens were broken by an impact blow at right angles to the direction of anticipated crack propagation. The fracture surface can then be examined for the different crack propagation modes. For the 7075-T6 a definitive difference in macro appearance existed. Figure 6 is a photomacrograph showing the difference in crack extension as a function of stress intensity. A photomicrograph is shown in Fig. 7 of a cross section of an unbroken exposed FIG. 8—Comparison between the delayed failure characteristics of 2024-T351 and 7075-T6 aluminum alloys in the short-transverse direction in an environment of 3Vz per cent NaCl solution. specimen to show the transgranular nature of the fatigue crack and the intergranular nature of the stress corrosion crack. 2024-T351 Aluminum Alloy—Short-Transverse Direction In the T351 temper (solution treated and stress relieved by stretching \% to 3 per cent), the susceptibility of 2024 to stress corrosion attack in conventional testing has been reported to be approximately equivalent to the 7075-T6 material [7]. On the basis that the susceptibility was approximately equivalent ASTM Int'l (allwere rightsobtained reserved); to Wed Dec 16 a 15:53:43 EST 2015 toCopyright that of by 7075-T6, data develop K0 versus tf behavior Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduc 76 STRESS CORROSION TESTING pattern for comparison purposes. The values developed using a maximum exposure time between 25,000 and 30,000 min are shown in Fig. 8. These data show a substantially less stress corrosion cracking susceptibility when compared to the 7075-T6. The plane-strain fracture toughness measurements of this material using both the maximum load and departure from linearity (pop-in) from ^-in.-thick specimen were as follows: KIe maximum load, 24,300 lc pop-in, 20,300 and 23,900 psi-in.1/2 Based on the pop-in determinations, an average Klc value of 22,100 is reported in Table 2. These data indicate that plane-strain conditions were not achieved, FIG. 9—Crack length of 7075-T6 specimens oriented in the long-transverse direction after exposure to aqueous environments for 6000 min. No trend which demonstrates a dependency upon stress intensity or environment is evident in these data. due, most likely, to the higher Klc-to-ays ratio for this material. The Klc reported represents a lower bound value. Also, for the K0 data shown on Fig. 8, the data above approximately 14,000 psi-in.1'2 represent conservative values of initial stress intensity. These findings are consistent with the recent recommendation that the specimen thickness should be 2>£ times (K/ays)2 (see footnote 4). T With a higher level of fracture toughness, a lower level of susceptibility to crack extension might be anticipated, which has been found in the case of steel [5]. A comparison between the results of the 7075-T6 and 2024-T351 (Fig. 8) demonstrates this decrease in crack propagation susceptibility. * Brown, W. F. and Srawley, J. E., communication to Subcommittee I of ASTM Copyright by ASTM Int'l (all rightsofreserved); Wed1966. Dec 16 15:53:43 EST 2015 Committee'E-24 on Fracture Testing Metals, Feb. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No furthe MULHERIN ON HIGH-STRENGTH ALUMINUM ALLOYS 77 7075-T6 Aluminum Alloy—Long-Transverse Direction Since initial experimental probing indicated that failure of 7075-T6 specimens oriented in the long-transverse direction would not occur within 30,000 min testing times, an exposure time approach was used. The exposure time selected was 6000 min, and the three environments were used. After exposure the specimens were broken at 90 deg to the anticipated direction of crack propagation as previously reported. Subsequent examination of the fracture surfaces indicated that the most FIG. 10—Photomicrographs of the fatigue cracks in 7075-T6 aluminum alloy specimens oriented in the long-transverse direction after exposure to an alternate-immersion environment for 6000 min. (top) Ko = 17,400 psi-in.1'2 and (bottom) K0 = 15,200 psi-in v* Keller's etch (X700). reliable crack length determinations were from the specimen edge to the fast fracture portion. This includes the machined notch depth, the mechanically induced fatigue crack, and any potential stress corrosion crack propagation. Note that variation commonly exists in the size of the fatigue crack from specimen to specimen. Using this technique, the data obtained are shown in Fig. 9. It is apparent that no trend is present in these data which would indicate either a stress intensity or environmental dependency present. Metallographic examination did not show any clear evidence of intergranular crack extension. Photomicrographs of typical cracks are shown in Fig. 10 to demonstrate the crack characCopyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 teristics. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions auth 78 STRESS CORROSION TESTING FIG. 11—Crack length of 2014-T6 specimens oriented in the long-transverse direction after exposure to aqueous environments for 6000 min. FIG. 12—Photomicrographs of the cracks in 2014-T6 aluminum alloy specimens oriented in the long-transverse direction after exposure1/ato an alternate-immersion environment for 6000 min, (top) Ke = 19,300 psi-in. and (bottom) Ko = 14,100 psi-in.1'" Keller's etch (XlOO). Reduced one third for reproduction. 2014-T6 Aluminum Alloy—Long-Transverse Direction Using the same experimental procedure as was used for the 7075-T6 in the long-transverse direction, the stress corrosion susceptibility of 2014-T6 in the long-transverse direction was investigated. The values obtained for the crack length as a function of KO are presented in Fig. 11. Examination of these data also does not indicate any dependency upon stress intensity or environment. Copyright by ASTM Int'l (all rights reserved); Wed 16 15:53:43 2015 In some specimens, especially at high K0Dec levels, micro EST examination Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further rep MULHERIN ON HIGH-STRENGTH ALUMINUM ALLOYS 79 indicated a corrosive mechanism operative in the fatigue cracked area. This was indicated by a broadening effect in the main crack and possible crack extension. Since the orientation of the photomicrograph is in a short-transverse plane, this broadening would indicate propagation in the short-transverse direction. The photomicrographs showing this condition are presented in Fig. 12. 7178-T6 Aluminum Alloy—Long-Transverse Direction Similar behavior was observed on 7178-T6 aluminum alloy as was obtained on the 7075-T6. Therefore, a series was run with exposure time as variant instead of stress intensity and environment. A series of specimens was prepared using additional care to minimize any difference in fatigue crack length. Specimens were exposed to a distilled water enTABLE 3—Influence of time on the crack propagation of 7178-T6 aluminum alloy in the long-transverse direction. Exposure Time, min <1 <1 500 1 000 2 6 6 10 20 20 000. 000 000 000 000 000 a, iin. Unstressed 0.210 0.210 0.209 0.208 0.208 0.208 0.207 0.209 0.204 0.204 Stressed 0.206 0.209 0.209 0.210 0.207 0.206 Environment ambient ambient distilled water distilled water distilled water distilled water hydraulic oil distilled water distilled water hydraulic oil vironment in both the loaded and unloaded condition. For the loaded specimen, the KQ applied was 13,600 ± 200 psi-in.1/2 Exposure times used varied from 500 to 20,000 min. Two specimens were broken immediately after fatigue cracking to obtain a zero time crack length. Also, two specimens were exposed in the unloaded condition to an environment of light hydraulic oil for periods of 6000 and 20,000 min. The results obtained are compiled in Table 3. Examination of these data do not indicate any pertinent influence of exposure time up to 20,000 min. Metallographic evidence was similar to that obtained on the 7075-T6 in the long-transverse direction. Evaluation of Experimental Results The experimental results are evaluated by comparison to results obtained by a conventional testing method. As a basis of comparison, information is used from the reference by Sprowls and Brown [1]. From Copyright by ASTM (all rights reserved); Wed Dec EST 2015 this reference, both Int'l 7075-T6 and 2024-T351 in 16 the15:53:43 short-transverse Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No fur 80 STRESS CORROSION TESTING direction can be categorized as having low resistance to stress corrosion cracking. On the basis of sustained tensile load which did not cause failure in tension and ring type specimens, 7075-T6 has slightly better resistance than 2024-T351. In the data reported in this paper, however, a substantial difference in resistance to crack propagation is indicated between the two alloys. This observation, naturally, gives rise to a question on the relative importance of the crack initiation period as compared to the crack propagation phase. In the long-transverse direction, a perceptible difference exists on the resistance to stress corrosion cracking as shown in Ref 1. The lowest resistance is in the 2014-T6 alloy with increased resistance in the 7075-T6 and a further increase in resistance in the 7178-T6. In all cases, the resistance was substantially higher than in the short-transverse direction. No quantitative distinction was found in the present investigations between the three alloys within the experimental conditions investigated. Summary of Results 1. Substantial intergranular stress corrosion crack propagation occurred in 7075-T6 aluminum alloy specimens oriented in the shorttransverse direction. Susceptibility decreased with decreasing stress intensity. 2. Within the experimental range investigated, the material behavior patterns were similar in environments of distilled water, 3|/2 per cent NaCl solution, and an alternate-immersion cycle in the NaCl solution and air. Both the alternate and the constant immersion in salt solution were more aggressive environments than the distilled water. 3. A substantially lower level of susceptibility in the short-transverse direction was demonstrated in 2024-T351 aluminum alloy as compared to 7075-T6. A lower level would be anticipated in view of the higher fracture toughness of this material. 4. Using the crack length after exposure as the susceptibility criterion, stress corrosion crack propagation on long-transverse specimens from i4-in.-thick plate was not demonstrated on 7075-T6, 7178-T6, and 2014-T6 aluminum alloys. A cknowledgment The author wishes to express appreciation to his colleagues at Frankford Arsenal, especially H. Markus, H. Rosenthal, S. Lipson, and D. F. Armiento, for many hours of discussion, advice, and encouragement during the course of this investigation. References [1] Sprowls, D. O. and Brown, R. H., "What Every Engineer Should Know About Stress by Corrosion of Aluminum," Metal Wed Progress, Vol. 81, No. April and Copyright ASTM Int'l (all rights reserved); Dec 16 15:53:43 EST4,2015 May, 1962. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprod DISCUSSION ON HIGH-STRENGTH ALUMINUM ALLOYS 81 [2] Brown, B. F., "A New Stress-Corrosion Cracking Test Procedure for HighStrength Alloys," Materials Research & Standards, Vol. 6, No. 3, March 1966, pp. 129-133. [3] Irwin, G. R., "Fracture," Handbuch Der Physik, Vol. VI, Springer, Berlin, 1958. [4] Paris, P. C. and Sih, G. C. M., "Stress Analysis of Cracks," Fracture Toughness Testing and Its Applications, ASTM STP 381, American Society for Testing and Materials, Philadelphia, 1965, pp. 30-83. [5] Mulherin, J. H., "Stress Corrosion Susceptibility of High-Strength Steels in Relation to Fracture Toughness," Journal of Basic Engineering, Transactions, American Society of Mechanical Engineers, Vol. 88, Series D, No. 4, pp. 777-782. [6] Srawley, J. E. and Brown, W. F., Jr., "Fracture Toughness Testing Methods," Fracture Toughness Testing and Its Applications, ASTM STP 381, American Society for Testing and Materials, 1965, pp. 133-198. [7] Freed, C. N, and Krafft, J. M., "Effect of Side Grooving on Measurements of Plane-Strain Fracture Toughness," Journal of Materials, Vol. 1, No. 4, Dec. 1966, pp. 770-790. DISCUSSION D. O. Sprowls1 (written discussion)—The use of a stress-intensity factor, KI , to describe the stress that causes stress corrosion cracking of an alloy is very intriguing. This paper is significant, because it contains the results of some of the first testing done with precracked specimens of high-strength aluminum alloys. It is evident from the apparent lack of correlation of these limited test results on 7075-T6 and 2024T351 with the published results of conventional stress corrosion tests of these alloys, however, that additional tests are needed to evaluate this procedure for use with aluminum alloys. By way of caution it should be remembered that the susceptibility to stress corrosion of an alloy depends not upon the metallurgical structure and the attendant mechanical properties alone, nor upon the electrochemical characteristics of this structure alone, but upon these factors acting together in a specific environment. In the stress corrosion cracking process both the initiation and the propagation of cracking are important. An accelerated test that involves either aspect of the process to the exclusion of the other may not fully characterize the serviceability of an alloy from the stress corrosion viewpoint. E. H. Dix, Jr., in the introduction to the ASTM-AIME Symposium on Stress-Corrosion Cracking in 1944, said, truly, that "Accelerated corrosion tests, at best, 1 Assistant chief, Int'l Chemical Div.,DecAlcoa Research Laboratories, Copyright by ASTM (all rightsMetallurgy reserved); Wed 16 15:53:43 EST 2015 New Kensington, Pa.by Downloaded/printed University of Washington (University of Washington) pursuant to License Agreement. No further reprod 82 STRESS CORROSION TESTING are something of a hazard and accelerated stress-corrosion cracking tests are a hazard to the nth degree." It is well known that alloy 2024-T351 has greater notch toughness and fracture toughness than 7075-T6. Also it has been established that the fractures of notched and precracked specimens used for these tests are predominantly transgranular. On the other hand, stress corrosion cracks in these alloys follow a predominantly intergranular path. Thus, the susceptibility to propagation of intergranular stress corrosion cracks, in the opinion of this writer, must depend upon electrochemical reactions in the path of the crack as well as upon the fracture toughness of the metal. That such is the case is borne out, in part, by the fact that 2024T851 plate (artificially aged'2024-T351) is highly resistant to stress corrosion cracking with conventional stress corrosion tests of shorttransverse, as well as long-transverse, specimens, yet its fracture toughness is relatively low, even when compared to that of 7075-T6. B. W. Lifka2 (written discussion)—One point that is puzzling is the disparity between results obtained on the two alloys, 7075-T6 failing rapidly and 2024-T351 not failing. As mentioned by the author, stress corrosion tests using conventional specimens show these two alloys to be very similar; if anything, alloy 2024-T351 is the more susceptible of the two in the type of environment used. In 3.5 per cent sodium chloride by alternate immersion, both 2024T351 and 7075-T6 short-transverse conventional specimens (without any stress concentrator) will fail within one to two weeks even at very low stresses, such as 10 ksi (approximately 15 per cent yield strength). Since the principal purpose of a precracked specimen is to minimize the incubation period, it does not seem likely that its use should alter the relative performance of the two alloys so drastically. Another disparity with conventional stress corrosion tests of 7075-T6 alloy is that shorttransverse specimens fail by stress corrosion cracking considerably faster when intermittently immersed than when continuously immersed in 3.5 per cent NaCl solution. One possible explanation might be that the particular 2024 plate used by the author had higher than normal resistance for the -T351 temper. It is suggested, therefore, that a number of conventional stress corrosion tests also ought to be made to establish the quality of the material by this method. Any differences subsequently observed with use of precracked specimens could then more definitely be attributed to the change in testing procedure. It has been stated that use of precracked specimens "eliminates" the incubation period for stress corrosion cracking. This is undoubtedly an overstatement. Since the metal forming the walls of the fatigue crack 2 Research Div.,DecAlcoa Research Copyright by engineer, ASTM Int'lChemical (all rightsMetallurgy reserved); Wed 16 15:53:43 ESTLaboratories, 2015 New Kensington, Pa.by Downloaded/printed University of Washington (University of Washington) pursuant to License Agreement. No further reprodu DISCUSSION ON HIGH-STRENGTH ALUMINUM ALLOYS 83 has not been corroding, some time period, however slight, probably will be required before the corrosion process starts when the specimen is subsequently exposed. Furthermore, it is known from tests of conventional specimens and from service experience that stress corrosion cracks very frequently do not occur at the site of highest stress concentration. Rather they occur at the site where all factors—stress, environment, electrochemical, and metallurgical—are optimum in combination to generate a stress corrosion crack. No metal is ideally homogeneous, and this is true as regards the electrochemical activity of the metal surface. The likelihood of positioning the fatigue crack at the most active anodic sites is therefore a matter of chance. In one case, the root of the crack may be favorably located so that further propagation by stress corrosion cracking could occur almost immediately. In another case, the crack may be unfavorably located so that initially only general corrosion would occur, gradually widening and deepening the fissure. In tests of smooth specimens, the probability of developing a stress corrosion fissure will increase with increased area of exposure. Use of a precracked specimen tends to restrict this area of possible failure to the walls of the fatigue crack. Therefore, it must be established that the area and shape of the crack employed are such that the probability factor is comparable for all items tested. /. H. Mulherin (author)—The author wishes to express appreciation to Messrs. Sprowls and Lifka for their very interesting and informative comments and looks forward to the publication of the efforts of other investigators in this area. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further repro C. S. Lin,1 J. J. Laurilliard^ and A. C. Hood1 Stress Corrosion Cracking of High-Strength Bolting REFERENCE: C. S. Lin, J. J. Laurilliard, and A. C. Hood, "Stress Corrosion Cracking of High-Strength Bolting," Stress Corrosion Testing, ASTM STP 425, Am. Soc. Testing Mats., 1967, p. 84. ABSTRACT: Exposure of high-strength steel bolts (260,000 psi) to the natural environments of temperature changes, moisture, and salt air have resulted in a number of stress corrosion cracking failures. Cadmium plating of the bolts has not prevented these failures. This paper presents a study of bolts made from H-ll, 4340, and maraging steel heat treated to several strength levels. The authors have adopted the 3Va per cent NaCl solution, intermittent exposure test to high-strength bolting and have used it as a basis for comparison of bolt life. The influence of thread rolling sequence and applied and residual stresses, as well as electroplated and zinc primer coatings, were investigated. The results indicate that a H-ll high-strength steel bolt with threads rolled after heat treatment will fail in the shank. Cadmium plate with an undercoat of nickel increases the life by five to seven times. The comparison of steel types shows the maraging steel (300) to have the best life of the three alloys but not sufficient to overcome the environmental effects. KEY WORDS: stress corrosion, corrosion, bolts, steels, environmental testing, salt water, cadmium, nickel, gold, plating, rolling, shot peening Problem The needs of aerospace structural designers for higher strength-todensity ratios have resulted in demands placed on the fastener producers for bolting of higher and higher strengths. Today mechanical fasteners are being used with strengths in excess of 300,000 psi.2 Exposure of high-strength steels to the natural environment of temperature changes, moisture, and salt air have resulted in a number of service failures by stress corrosion cracking. The exposure of high-strength bolts to these same environments, even when the bolts were cadmium plated, 1 Research metallurgist, research engineer, and manager, respectively, SPS Laboratories, Standard Pressed Steel Co., Jenkintown, Pa. Messrs. Laurilliard and Hood are personal members ASTM. 2 Hood, A. C. and Sproat, R. L., "Ultrahigh-Strength Steel Fasteners," Structure and Properties of Ultrahigh-Strength Steels, ATSM STP 370, American Copyright ASTM and Int'l (all rights reserved); Wed DecOct. 16 15:53:43 Society forbyTesting Materials, Philadelphia, 1965, p.EST 208.2015 Downloaded/printed by 84 University of Washington (University of Washington) pursuant to License Agreement. No further reproductions aut LIN ET AL ON HIGH-STRENGTH BOLTING 85 TABLE 1—Chemical composition of bolt materials (weight per cent). Element Maraging (300) c 0.02 0.04 0.005 0.008 0.006 18.95 Mn P S Si Ni Cr Mo Fe Co Al Ti B Zr Ca 4.96 balance 9.09 0.15 0.66 0.0049 0.010 0.05 4340 H-ll 0.40 0.72 0.013 0.013 0.27 1.69 0.75 0.22 balance 0.40 0.29 0.006 0.012 1.00 5.11 1.33 balance TABLE 2—Heat treatment and tensile properties of high-strength bolts. Alloy Hardening Temperature and Time Maraging (300) . . .2050F, 5 min; 1050 F, salt quench H-ll 4340 1850 F, 20 min; atmosphere cool 1550 F, 20 min; oil quench Tempering Temperature and Time 975 F 2 hr + 1050 F 2 hr + 1020 F 2 hr + 1035 F 2 hr + 1080 F 2 hr + 1130 F 2 hr + 500 F 2 hr + 600F 2 hr + 800 F 2 hr + ProporUltimate tional Tensile Load, Limit, lb (avg) lb (avg) 2 hr + 2 hr 10 410 8000 2 hr + 2 hr 9 690 7200 2 hr + 2 hr 11 750 8350 2 hr + 2 hr 11 270 7500 2 hr + 2 hr 9 963 7350 2 hr + 2 hr 8 810 6400 2 hr 9 550 6550 2 hr 8 670 6200 2 hr 7 670 6100 resulted in similar failures. This prompted an investigation of the bolts in a laboratory test designed to simulate some of the worst conditions of environmental exposure. It was intended to learn what the influences of bolt material, bolt strength, surface condition, and applied stress would have on the resistance-to-failure by stress corrosion cracking. Manufacture of Bolts The standard 260,000 tensile boltWed made H-ll EST steel2015 in a size Copyright by ASTM Int'l (all psi rights reserved); Decfrom 16 15:53:43 Downloaded/printed V4-28 was used as by a control. This bolt is made as follows: University of Washington (University of Washington) pursuant to License Agreement. No further repro 86 STRESS CORROSION TESTING 1. 2. 3. 4. 5. 6. 7. 8. blank cut head hot forged headed blank annealed rough machined headed blank hardened and tempered body and thread roll diameters ground retempered after grinding for stress relief threads and head-to-shank fillet rolled FIG. 1—Load extension curve of lA-28 by 2 uncoated, 290,500 psi ultimate strength H-ll bolts. The normal practice after thread rolling is to dry-blast clean and vapor deposit cadmium on the bolt. Since several different coatings were to be applied, the uncoated bolt was used for a control. The other alloys, AISI 4340 and maraging steel (300), were also used in the investigation. They were made in the same fashion as the control bolt except that the heat treatment corresponding to the alloy was used. Chemical composition for three alloys is shown in Table 1, and heat treatment and strength are shown in Table 2. Proportional limits were established with an overall bolt extensometer and calculated as shown in Fig. 1. Copyright by confusion ASTM Int'l regarding (all rights reserved); Wed Dec 16 15:53:43 EST 2015for To avoid the cross-sectional area selection Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further re LIN ET AL ON HIGH-STRENGTH BOLTING 87 stress calculations, the applied loads are always given in pounds. Bolt strengths are indicated in pounds per square inch, and the stress is calculated using the area at 98 per cent of the basic pitch diameter. This is the conventional method used for this type fastener and is in accordance with Specification NAS1348. For the V4-28 thread used in these studies, an area of 0.0388 in.2 was used. Where unit stresses are shown for the shank of the bolts, an area of 0.0491 in.2 was used. Material and Bolt Strength Three steels, H-ll, maraging (300), and 4340, were manufactured as uncoated bolts in several strength levels to measure the influence of material and tensile strength on stress corrosion cracking. All threads and fillets were cold-worked after heat treatment. TABLE 3—Surface coatings. Type Average Coating Thickness, in. Remarks Vapor deposited cadmium (vacuum) 0.0005 Chromate conversion coatings on electroplated cadmium MIL-C-8837, no baking after deposition Cd 0.0005 Electroplated nickel 0.0005 Douglas proprietary cadmium and chromate coating AMS 2424, post plate baked for 3 hr at 630 F Electroplated nickel and electroplated cadmium (undiff used) . . . . Electroplated nickel and electroplated cadmium (diffused) Electroplated gold Zinc chromate Zinc chromate (encapsulated dry film) Ni 0.0003 Cd 0.0002 Baked after cadmium plating for 3 hr at 375 F Ni 0.0003 Cd 0.0002 0.00006 AMS 2416, diffused at 630 F for 3 hr MIL-P-8585 National Cash Register Co. proprietary process Thread Rolling Sequence Bolts of H-ll steel were manufactured as the control bolt except that threads were rolled before heat treatment. These bolts were compared with H-ll control bolts. Both lots were heat treated to several strength levels for a comparison of the influence of thread rolling sequence on resistance to stress corrosion cracking. Residual Stress Copyright by ASTM (all rights reserved);the Wed Dec of16residual 15:53:43compresEST 2015 Experiments were Int'l designed to measure effect Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No furthe 88 STRESS CORROSION TESTING sive stresses applied to the shank as well as to the threads of the control bolt. The control bolt is essentially free of residual stresses in the shank area since it is retempered after grinding. One group of bolts was shot peened over the entire bolt surface with O.Oll-in.-diameter cast steel shot to an Almen strip intensity value of FIG. 2—Loaded test cylinder used for stress corrosion. Note the slots which facilitate contact of the salt solution with the bolt shank. FIG. 3—Test equipment for stress corrosion studies. Note the fans used to accelerate drying during the 50-min portion of the stress corrosion cycle. 0.008 A. Another group was glass bead shot peened at 80-psi air pressure. The nozzle was held 6 in. from the bolt surface. Decarburization One of the groups of control bolts was made with the heat treatment atmosphere so adjusted that the finished bolt was decarburized. If surface hardness or carbon content influences the resistance of H-ll to stress corrosion cracking, then this experiment would measure it. The shank Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprod LIN ET AL ON HIGH-STRENGTH BOLTING 89 surface was decarburized to a hardness of VHN 506 (Rc 47.5) which increased to a core hardness of VHN 606 (Rc 53.5) at 0.020 in. below the surface. Surface Coatings Various coatings were applied to the H-ll control lot of bolts as shown in Table 3. Metallic coatings were applied by both electrodeposition and vapor deposition. Zinc chromate primer was also applied by two FIG. 4—Effect of applied load of the stress corrosion cracking susceptibility of /4-28 by 2 uncoated H-ll bolts. The bolts had an ultimate strength of 290,500 psi with threads rolled after heat treatment. All failures occurred in the bolt shank. l methods. Bolts were coated with a 1:1 solution of zinc chromate primer (Mil-P-8585) and allowed to dry while standing on the thread end. In the second method, bolts were coated with plastic encapsulated zinc chromate. The pressure of bolt tightening burst the spheres and applied the liquid in this fashion. The process of encapsulation is a development of the National Cash Register Co. and the U.S. Air Force.3 3 Hanny, J. F. and Price, J. E., "Dry Films of Encapsulated Zinc Chromate Primer," AFML TR-65-54, March 1965, Air Force Materials Laboratory, WrightPatterson Air Force Base, Ohio. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprodu 90 STRESS CORROSION TESTING Test Equipment and Methods The test equipment consisted of turned steel cylinders (Fig. 2) with slots milled in them to permit free flow of the corrosive media and air. Bolts were tightened on the cylinders to a predetermined percentage of the proportional limit. Bolts had previously been tested in a tensile machine to develop a load-elongation curve (Fig. 1). This was used to measure the proportional limit. A supermicrometer was used to measure the bolt elongation while tightening on cylinders was taking place. The bolted cylinders were hung from a test fixture (Fig. 3). The test fixture was immersed in a room-temperature solution of 3l/z per cent sodium chloride (NaCl) in distilled water. The fixture was immersed for 10 min of every hour and withdrawn for 50 min. While the fixture and cylinders were suspended over the bath, fans were turned on to facilitate drying with rapidly moving air. Previous investigation indicated that rapid drying would accelerate failure. Specimens were examined, and the test solution changed daily. Tests were continued until bolts failed or a minimum of 1000 hr. In comparison testing, bolts were loaded to either 90 or 100 per cent of the proportional limit. The test loads are indicated in the results. In one experiment, the H-ll control bolt was loaded to varying percentages of the proportional limit to see if the applied stress had an effect on the resistance to stress corrosion cracking. Test Results Effect of Applied Stress The H-ll steel control bolts, uncoated, but with threads and fillets rolled after heat treatment, were tightened to several applied loads in the cylinder, and the life was determined. Figure 4 shows the basic stressrupture type curve. The data appear to show that there is a lower level of applied stress, below which no failure will occur. All failures below the proportional limit occurred in the bolt shank, which would indicate that the problem is basically a material deficiency rather than something which is adversely affected by bolt manufacturing techniques. Effect of Material and Bolt Strength The control bolt of H-l 1 was compared with 4340 and maraging steel bolts manufactured in a similar fashion and loaded to the proportional limit for the respective bolts. The results hi Table 4 show that the maraging steel (300) is affected to a lesser degree than the other two alloys. The H-l 1 bolts are slightly better than 4340 at the same tensile Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 strength. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions LIN ET AL ON HIGH-STRENGTH BOLTING 91 TABLE 4—Effect of material and strength on susceptibility to stress corrosion cracking ofuncoated bolts." Material H-ll steel Ultimate Bolt Strength, psi Time-to-Failure,6 hr 63C 88' 302 800 290 500 256 800 227 100 Maraging (300) 268 300 249 700 4340 steel 246 100 222 200 197 700 74 106 106 136 1515* 227 1030* 87 196 1002d 136 185 252 103 117 124 140 184 164 1517d 504 1054d 90 220 1291d 167 1610d 603 1127* 353 1000d 1291d 0 Bolts are K-28 by 2 with rolled threads after heat treatment and loaded to 100% of their proportional limit. 6 Failures were shank failures unless noted. e Head failure. d No failure, test discontinued. TABLE 5—Effect of bolt strength and rolling threads before and after heat treatmen t on susceptibility to stress corrosion cracking ofuncoated H-ll steel bolts.° Thread Rolling Condition Rolled threads before heat treatment Rolled threads after heat treatment Ultimate Bolt Strength, psi 283 500 251 300 217 800 iPime-to-Failure!, hr I6 100* 140» 1516 185 600 75P 302 800 63d 88d 74e 290 500 256 800 227 100 106e 106« 136« 1515' 3.56 106* 1526 176* 8926 103e 117« 124e 140* 184' 164" 1517e 3.8* 177* 1343C 1343" 1344" 136e 185« 252* 167e 1610" 0 Bolts are ^-28 by 2 loaded to 100% of their proportional limit. Thread failure. c No failure, test discontinued. d Head failure. • Shank failure. 6 Effect of Thread Rolling Sequence The bolt life in hours as a function of bolt strength for both sequences of thread rolling is shown in Table 5. The influence of thread rolling after heat treatment, which shifts the failure to the bolt shank with a Copyright by ASTM Int'l (all rightsis reserved); Wed Dec 16 15:53:43 EST 2015 considerable increase in life, apparent. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions 92 STRESS CORROSION TESTING Effect of Residual Stress The results of shot peening and glass bead peening of H-ll control bolts are shown in Table 6. When the threads are rolled after heat treatment, further working of the bolt thread surfaces does not affect the results. Apparently the peening of the bolt shank provides insufficient compressive stresses to influence the environmental conditions. Effect of Coatings The results of the tests as shown in Table 7 indicate that nickel and nickel-cadmium coatings provided a considerable increase in the resistTABLE 6—Effect of residual stress on the susceptibility to stress corrosion cracking of H-28 by 2 uncoated H-ll bolts. Ultimate Strength, psi Applied Load, lba 284 800 6800 138 500 175 300 3 lc Standard bolt (rolled threads after heat treatment) 290 500 7500 152 700 192 300 7 Steel shot peened 290 500 7500 152 700 192 300 3 Glass shot peened 290 500 7500 152 700 192 300 5 74 106 106 124 120 123 128 102 114 123 Method of Stressing Standard bolt (rolled threads before heat treatment) Shank Stress, psi Thread Stress, psi Bolts Tested Time-toFailure,6 hr 3.5C 3.8" 140 184 252 133 271 a 100% of proportional limit. Failures were shank failures unless noted. c Thread failure. 5 ance of the standard bolting to stress corrosion failure. Vapor deposited cadmium offered some improvement over the bare bolt. Electroplated cadmium with a post plate dichromate more than doubled the resistance of vacuum cadmium but did not match the nickel-cadmium coatings. Gold plating, with a thickness of 0.00006 in., showed little improvement over uncoated bolts. Zinc chromate primer was approximately equivalent to vapor deposited cadmium whether dipped or applied by the encapsulated process. Effect of byDecarburization Copyright ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by The effect of decarburization of the H-ll control bolt was as follows: University of Washington (University of Washington) pursuant to License Agreement. No further reproductions a LIN ET AL ON HIGH-STRENGTH BOLTING Surface Treatment Neutral Decarburized Bolts Tested Time-to-Failure, hr 4 5 106, 106, 123, 124 74, 112, 114, 128, 138 93 All failures occurred in the bolt shank. Note that no significant difference in bolt life was observed. TABLE 7—Effect of coatings on the susceptibility to stress corrosion cracking of V±-28 by 2 H-ll bolts* Bolts Tested Bolt Coating Bare 8 Vacuum cadmium Time-to-Failure,6 hr 68 70 52 76 92 92 6 Douglas proprietary electroplated cadmium and chromate conversion coating. 186 242 256 291 329 329 6 Electroplated nickel 5 668 734 1010" 1010' 5093" 743 761 5093° 5093' 784 787 11 1587e 1587" 1587" 1587" 1587" 1587° 2581 Electroplated nickel and electroplated cadmium (undiffused) 995 1010^ 1010" 1252 4 Electroplated gold Zinc chromate primer (dipped) 3 4 Encapsulated zinc chromate primer 6 767 1512' 274 125d 21 ld 224 241 1512C 1512" 336 223d 268 246 379 Electroplated nickel and cadmium (diffused) electroplated 140 140 467 472 480 a Ultimate bolt strength = 290,500 psi; applied load = 6750 Ib (90% proportional limit); shank stress = 137,500 psi; and thread stress = 174,000 psi. 6 Failures were shank failures unless noted. c No failure, test discontinued. d Head failure. Mechanism of Failure From microscopic appearance, it was found that the location of stress corrosion cracking originates at the bolt surface (Fig. 5). This was related to the corrosion pits on the shank surface. Examination of the fracture surface on bare and plated bolts showed that considerable rust had Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 formed in the area of the corrosion pit prior to the initiation of the stress Downloaded/printed by corrosion crack (Figs.(University 6 and 7).of Washington) pursuant to License Agreement. No further reproductions University of Washington 94 STRESS CORROSION TESTING All primary stress corrosion cracking was found to propagate transgranularly and intergranularly at a direction approximately 90 deg to the direction of loading. In the majority of cases, stress corrosion cracking failures were located in the shank. Occasionally, failure occurred in the thread area, head area, and lightening-hole area. Head failures occurred only at a bolt strength FIG. 5—Corrosion pit and crack initiation in the shank of an H-ll steel bolt. White arrow indicates the direction of applied stress. of 302,800 psi on H-l 1 bolts with rolled threads after heat treatment Thread failures occurred only on bolts with threads rolled before heat treatment. The photomacrographs of thread, shank, and head failures are shown in Figs. 8 to 10. The lightening-hole failure (Fig. 11) was found only with plated bolts that were capable of resisting head, shank, and thread fracture for long exposures (> 600 hr) to the corrosive environment. All the lighteningCopyright by ASTM rights until reserved); Wedwas Dec applied 16 15:53:43 EST 2015the hole failures were Int'l not (all evident torque to remove Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further re LIN ET AL ON HIGH-STRENGTH BOLTING 95 FIG. 6—Stress corrosion fracture of H-ll bolt. Note the point source, corroded area with fracture radiating from this point. FIG. 7—Shank failure of H-ll steel bolt. Dark zones are patches of rust. FIG. 8—Thread failure of H-ll steelWed bolts threadsEST rolled Copyright by ASTM Int'l (all rights reserved); Decwith 16 15:53:43 2015 before heat treatment. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduction 96 STRESS CORROSION TESTING stress corroded bolts from the test fixture. Upon application of torque, the twelve-point serration sheared off. Figure 11 indicates that the crack path initiates near the base of the lightening hole and curves smoothly in a slight downward arc to the outside surface of the twelve-point area. The break is symmetrical around the bolt circumference. An experiment FIG. 9—Head failures of H-ll steel bolts with threads rolled after heat treatment. FIG. 10—Shank failures of H-ll steel bolts with threads rolled after heat treatment. was run to determine retention of bolt load after twelve-point serration failure. Bolts with strain gages were loaded in cylinders to the prescribed preload. The serrated section was cut off and the bolt load measured. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Measurements indicated that 97 to 99 per cent of the original applied Downloaded/printed by load was retained. University of Washington (University of Washington) pursuant to License Agreement. No further reproductions autho LIN ET AL ON HIGH-STRENGTH BOLTING 97 Discussions and Conclusions If one considers this investigation in the light of the use of high-strength steel bolting, then a number of interesting aspects are revealed. It is not unreasonable to assume that the 3l/2 per cent salt solution immersion is an accelerated version of what happens to structural bolts used in aircraft in many parts of the world. The loads imparted to the bolts in the cylinders are typical of what one would expect these bolts to carry in an aircraft structure. What appears to be the most significant discovery relates to the metallurgy of high-strength steels for structures. If the best possible metallurgical practices were used to produce a 260,000 psi threaded fastener and yet one lacked the technology of producing threads and fillets FIG. 11—Stress corrosion cracking of bolt head lightening hole. by rolling after heat treatment, then the life in our experimental test would average about 2 hr. Where threads and fillets are rolled after heat treatment, life is raised to at least 74 hr. The addition of vacuum cadmium plating more than doubles this Me. This is where standard high-strength steel bolting is today. It would appear that the most immediate improvement could be obtained by the addition of a nickel layer under the vacuum cadmium coating. It is possible that this may be sufficient to prevent stress corrosion cracking of high-strength bolting within the expected life span of today's aircraft, since it more than triples the predictable life. The other approach may be through the use of a bolting material which in itself has high resistance to stress corrosion cracking. The authors believe that a material with high resistance is the most foolproof soluCopyright ASTM Int'l (all rights reserved); Wed 15:53:43 EST 2015 tion to thebyproblem. However, other aspects of Dec bolts16must be preserved. Downloaded/printed by Such properties as notch strength, fatigue resistance, and resistance to University of Washington (University of Washington) pursuant to License Agreement. No further r 98 STRESS CORROSION TESTING hydrogen embrittlement must be considered, otherwise we solve the problem of stress corrosion cracking only to create others.2 It is believed that any future evaluation of high-strength bolt materials must include an assessment of their resistance to stress corrosion cracking. The present test appears to be adequate for bolts of this strength level. A cknowledgment The authors are grateful to R. L. Sproat, director of engineering, Standard Pressed Steel Co., for continued encouragement and guidance during the course of this investigation. We also wish to acknowledge the help and suggestions of W. Boyd and M. Epstein, Battelle Memorial Institute, provided during the course of a private research program for Standard Pressed Steel Co. DISCUSSION Anthony Gallaccio1 (written discussion)—You report that the nickel plate and the nickel-cadmium plate on H-ll steel bolts provided markedly increased resistance of the bolts to stress corrosion cracking. Can you explain why the nickel plate is so effective, and why the nickel plate is so much better than the cadmium plate? C. S. Lin, J. J. Laurilliard, and A. C. Hood (authors)—The authors appreciate the interest of Mr. Gallaccio in the stress corrosion cracking of high-strength bolts. The nickel is believed to offer greater resistance than cadmium to stress corrosion cracking primarily because of its higher resistance to pitting. We believe that when the pit progresses to the surface of the steel, then one of the several mechanisms advanced for stress corrosion cracking takes place. However, it takes longer for this to occur with nickel, and the advantage over cadmium is therefore one of time. 1 Chief, Protection and Preservation Branch, Pitman-Dunn Research Laboratories, Frankford Arsenal, Philadelphia, Pa. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduct A. GaUaccio1 and M. A. Pelensky1 Stress Corrosion of High-Strength Steel Alloys—Environmental Factors REFERENCE: A. GaUaccio and M. A. Pelensky, "Stress Corrosion of High-Strength Steel Alloys—Environmental Factors," Stress Corrosion Testing, ASTM STP 425, Am. Soc. Testing Mats., 1967, p. 99. ABSTRACT: Several years ago it became evident to members of ASTM Committee B-3, Subcommittee X (now Committee G-l, Subcommittee VI) on Stress Corrosion that a review of the reported information and, where available, work in progress on stress corrosion testing was needed. This was considered essential as a precursor to the proposed development of standards and criteria for determining the stress corrosion susceptibility and behavior of various alloys. The subcommittee established three task groups to accomplish a review of the subject. The task group assignments were as follows: TG-1, sample selection and method of stress; TG-2, environments under which stress corrosion occurs and duration of tests; and TG-3, reporting of stress corrosion results and interpretation of results. This report is part of the TG-2 assignment. It concerns a review of reported stress corrosion investigations relating to the effects of various environments on specific high-strength steels. KEY WORDS: corrosion, stress corrosion, steels, environmental testing A survey is presented of selected reported information from investigations performed to determine the effects of various environments or media on the stress corrosion of a variety of high-strength steels. The results of the stress corrosion tests were compiled with other pertinent related information. These included the specific environment, the composition and strength level of the alloy, the type of specimen used, the stress level employed, and the direction and method of stressing. Informational sources are cited. Based on the data reviewed and as deemed appropriate, comments and suggestions are offered. These concern the selection of environments, sampling, and the various integrated factors bearing on the methods of testing and the test results. 1 Chief and research chemist, respectively, Protection and Preservation Branch, Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Pitman-Dunn Research Laboratories, Frankford Arsenal, Philadelphia, Pa. Downloaded/printed by 99 University of Washington (University of Washington) pursuant to License Agreement. No further reproductions 100 STRESS CORROSION TESTING Environments and Materials The conditions of test and the materials covered in this survey are as follows. Environment (Unless otherwise indicated, ambient temperature and total immersion of the specimen apply.) Air: Indoor, 16 to 74% RH, 62 to 92 F Indoor, 100% RH, 140 F Indoor, 100% RH, 175 to 205 F Outdoor, industrial Outdoor, semi-industrial Outdoor, marine Waters: Distilled Distilled, 120 F Distilled, 160 F Tap Sea, natural Sea, synthetic Solutions: H2S, saturated MgCl 2 , 42%, boiling NaBr, 10.3% (1 Af) NaC2H4O2, 8.2% (1 Af) NaCN, 4.9% (1 Af) NaF, 4.2% (1 M} NaHCO 3 ,8.4% (1 Af) NaCl, 3%, sprayed, followed by exposure to air (dry) NaCl, 3%, sprayed, followed by exposure to air, ambient, indoors NaCl, 3%, sprayed, followed by exposure to air, 100% RH NaCl, 3% NaCl, 3%, boiling NaCl, 3^%, alternate immersion NaCl, 5%, alternate immersion NaCl, 5.7% NaClO3, 10.64% (1 Af) Nal, 15% (1 Af) NaN0 2 ,6.9% (1 Af) (NaNO2 + NaNO3), marquench, 1%, elevated temperature NaNO 3 ,8.5% (1 Af) NaOH, 4% (1 Af) NaPO 3 , 10.2% (1 Af) Na 2 HPO 4 , 14.2% (1 Af) Na2Cr2O7, 0.25 % Na2SO3, 10.6% (1 Af) Na2S04, 14.2% (1 Af) Na3P04, 16.4% (1 Af) Alloys: AFC 77 Cr (5 %); hot-worked die steel AM 350 D6AC (ladish); martinsite, low AM 355; cold-worked PH steel alloy Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions auth GALLACCIO AND PELENSKY ON HIGH-STRENGTH STEEL ALLOYS 101 17-4 PH 300M; silicon modified, 4300 series 4330M 4335M 4340 4340M Hll Maraging steel, 18Ni Maraging steel, 20Ni Maraging steel, 25Ni MBMC, No. 1 Ni (9%), Co (4%) Vascojet 1000 12MoV Data The tabulated data compiled in this survey are available through the authors. Table 1 is a sample of the data. TABLE 1—Sample of data. Alloy Yield Stress, ksi Test Specimen Stressed, % yield Direction of Stress Stress Method No. Specimens, time-to-failure Ref , 3%, AMBIENT TEMPERATURE D6AC. . . 18Ni maraging steel 0.23T1 0.52T1. 0.5Ti.. 0.5Ti.. 0.49T1. 0.4T1.. 0.55Ti. 0.62T1. 0.5Ti.. 0.62T1. l.OOTi. 0.5Ti.. l.OOTi. 197.5 197.5 222.5 222.5 235 f f f f f wr wr wr wr wr 75 dm 75 dm 75 dm dm dm dm dm bb ub bb ub bb 235 252 f wr f wr 75 dm dm dm ub bb 252 f wr dm dm ub 181.5248.2 249.9 255.4 269.7 278.0 279.1 283 302.5 323 323.3 331 354.4 f f f f f f f f f f f f f dm dm 75 dm dm 75 dm 75 75 75 75 75 75 dm dm dm dm dm dm dm dm dm dm dm dm dm ub ub bb bb ub bb ub bb bb bb bb bb bb wr wr wr wr wr wr wr wr wr wr wr wr wr Abbreviations: f wr = flat wrought ub = dm = data missing m = bb = bent beam avg = nf = nobyfailure Copyright ASTM Int'l (all rights 3 2 3 2 3 nf 21 days nf 198 days nf 21 days nf 198 days nf 21 days, 3 nf 104 days 1 62 days 3 nf 21 days, 3 nf 104 days 1 116 days, 1 19 days [7] 2 nf 4150 hr 3 1776 hr (m) 3 140 hr (avg) 2 nf 3600 hr 3 1174 hr (m) 2 nf 3600 hr 3 4704 hr (m) 3 36 hr (avg) 2 3200 hr (avg) 3 1580 hr (avg) 2 7 hr (avg) 2 nf 3600 hr 20 20 hr (avg) [4] U-bend mean average reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further repro 102 STRESS CORROSION TESTING Discussion Concerning the Data Atmospheres On the whole, indoor atmospheres are less severe than outdoor atmospheres for the stress corrosion cracking of the high-strength steels. The stress corrosion susceptibilities of D6AC and 300M alloys are shown to be greater at the higher temperature and humidity conditions in air, compared to a normal ambient indoor atmospheric condition [I].2 This observation would be expected to be evident for other alloys in similar investigations but is not borne out by the data. The independent effects of moisture or temperature conditions of the atmosphere are not clearly indicated for clean metals, but some results shown for specimens sprayed with salt solution and exposed indoors at different humidity levels indicate the effect of humidity [2]. The following alloys fail in a short time when exposed to marine environments (that is, Kure Beach 80-ft site): 4340, 218 to 254 ksi yield [2,3] MBMC No. 1, 218 to 254 ksi yield [2] Cr (5 per cent) steel alloy, 229 to 237 ksi yield [2] Kure Beach exposure results for 20Ni maraging and 4340 steels indicate these alloys are more readily stress corrosion cracked by marine exposure than by the industrial atmosphere at Bayonne [3]. Further, the Kure Beach atmosphere results for various alloys are observed to be comparable to those obtained with sea water immersion tests [3]. Waters Apparently distilled water more readily causes stress corrosion failures in certain high-strength steels than does NaCl, 3 per cent [4]. From immersion test data, this observation is applicable in the case of a few maraging, 18Ni steels, for example, 0.62 Ti (323 ksi yield); 0.5 Ti (249.9 ksi yield). Conversely the salt solution is more active in bringing on the stress corrosion cracking of other maraging 18Ni steel alloys, for example, the l.OTi (323.3 and 354.4 ksi yield) [4]. Alloy D6AC is shown to fail more readily in distilled water than in salt solution [1]. For maraging steels, 18 and 20Ni alloy steels, tested by the bent-beam method in distilled water, the influence of temperature at the high temperatures is evident. Failure times are shorter at the high temperature [4]. U-bend tests of the same material under the same environments were also found to fail more rapidly [4]. 2 The italic numbers in brackets refer to the list of references appended to this Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 paper. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprodu GALLACCIO AND PELENSKY ON HIGH-STRENGTH STEEL ALLOYS 103 Solutions Alternate immersion in NaCl, 3.5 per cent, using bent-beam specimens of alloys D6AC, 4340, 4335M, HI 1, AFC 77, and AM 355 caused more rapid failures than did outdoor exposures in a semi-industrial environment (Seattle) [5]. However, those materials first to fail during alternate immersion were not necessarily the first to fail in the atmosphere [5]. The frequency of rainfall or the persistence of condensate will materially influence the initiation or promotion of corrosion outdoors. Alloy 12MoV tested by immersion in NaCl, 3 per cent, at pH 1 through pH 11, with bent-beam stressing, reveals an inverse relationship of pH to stress corrosion cracking; with the increase of pH, stress corrosion failures decreased. No failures occurred at pH 11.5 or above [6]. In 1 M solutions of NaCl, NaNO3 , NaPO3 with bent-beam or U-bend stressing, the following order of increasing susceptibility to stress corrosion of the following high-strength alloys is revealed [7]: AM 355, D6AC, 300M, and Vascojet 1000. One molar solutions of NaCl, NaNO3 , and Na2SO4 to approximately the same degree attacked D6AC alloy of different yield strength levels. But NaPOs resulted in more aggressive attack of the alloy of 206 and 223 ksi strength levels [7]. The same solutions, on the whole, exhibited somewhat greater attack of the 300M and Vascojet 1000 alloys [7]. The aggressiveness of the Na2SO4 solution was second to the NaPO 3 , especially for 300M yield strengths 218 ksi and greater. The NaCl was most aggressive in the case of the Vascojet 1000, and the NaNO3 was least aggressive [7]. Alloys The susceptibility to stress corrosion failure of maraging steels increases with higher titanium content and with increasing yield strength [4]. In general, for a single alloy composition, the higher the stress level, the shorter the time to stress corrosion failure [7,8]. For a given yield strength level, the maraging steels, 18Ni, are less susceptible to stress corrosion failure than the conventional high-strength steels [4]. Some of the reported data permit comparisons of the stress corrosion behavior of maraging steel, 18Ni, to maraging steel, 20Ni, at the same U-bend stress levels and similar environments. The indication is that the 18Ni alloy is less prone to stress corrosion failure than the 20Ni alloy [3,4], but, for the same alloys and environments, the limited data for the bent-beam tests are less consistent in that respect [4]. Specimens of the 4340 steel drawn at 800 and 900 F have yielded results showing about 25 times greater resistance to stress corrosion crackCopyright by ASTM Int'l (all rightsand reserved); Dec 16 15:53:43 ESTare 2015 ing than those drawn at 475 700 F.Wed These comparisons based on Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprodu 104 STRESS CORROSION TESTING alternate-immersion tests in NaCl, 5 per cent, under sustained tensile loading. Resistance to stress corrosion cracking is poor when the ultimate tensile strength of the alloy exceeds 200 ksi [9]. Tempering temperatures above 1050 F used for triple drawing of Hll alloy (ultimate tensile strength near 260 ksi) results in improved resistance to stress corrosion cracking as determined by immersion in NaCl, 5 per cent, under sustained tensile loading [9]. Steel alloys of 9 and 18Ni are reported to be less susceptible to failure from stress corrosion than the 4340 alloy. The 9 per cent alloy is the better of the two nickel alloys as revealed by alternate immersion in NaCl, 5 per cent, with sustained tensile loading [9].3 Stress Methods Test results under all environmental exposures show that the U-bend stressing is more severe than bent-beam stressing [4,7]. Predominantly these were the test methods employed. Stressing by means of two- or three-point loading or by tensile loading was employed in some cases [3,9 JO]. Observations and Suggestions The data reviewed reveal that a wide variety of environments have been used for investigating stress corrosion susceptibility of high-strength steels. Frequently the choice of environment is based on an actual or an expected condition of service for the metal. Such choices are entirely proper. However, to attain a measure of universality or uniformity of data, at least at the earlier stages of stress corrosion testing of new or unusual alloys, the variety of environments—particularly the laboratory test media— might be selectively limited. Even on the basis of the incomplete and varying experimental data collected, a few of the environments show some consistency, specificity, and aggressiveness of stress corrosion attack of high-strength steels. Because of the variability of natural atmospheres from day-to-day, season-to-season, and year-to-year, tests of materials conducted in natural atmospheres should be repeated frequently enough and extended over sufficiently long periods of time to ensure acquiring adequate and dependable data. Frequently, test information obtained under laboratory controlled conditions are used to estimate or predict the service-life expectancy of a material. Sometimes the performance of materials under operating or natural exposure conditions fall severely short of such forecasts. 3 Data supporting the statements citing Ref 9 are not included in the tabular compilation. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions GALLACCIO AND PELENSKY ON HIGH-STRENGTH STEEL ALLOYS 105 In selecting a controlled environment for stress corrosion testing, as well as for general corrosion testing, the state or combinations of states of the environment, the concentrations of constituents in the medium, pH, density, surface tension factors, and temperature must be taken into consideration. It might be important to ascertain the role of a single factor, or of a combination of factors. Reasonable and functional criteria for corrosion testing methods, reproducibility of test results, and suitable confidence levels of reliability of the data should be established in order that some measure of standardization might be achieved. In addition to the factors mentioned above regarding environment, other factors must be acknowledged so that useful and reliable standardization might be realized. Essentially the factors to be considered regarding the specimen include sampling, the form or shape, grain orientation or direction of flow, the applied stress relative to the grain orientation, the level and mode of stressing, the rate of applied strain, and the number of replicates. It is interesting to note that specific information concerning these factors, in many cases, is not included in the published reports. In very many cases the data are highly scattered, and the replicates are few. The usual wide scatter of results associated with stress corrosion testing may be attributed in part to the inherent variability of the parent and specimen metal microstructure, composition, processing, and residual stresses. Sufficiently large numbers of replicates and statistical treatment of the test data should effectively help to establish an enhanced confidence in stress corrosion testing results. Further such treatment of the data, with controlled variations of the individual factors associated with the specimens, environments, and test methods, will expose those factors of predominant influence. References [1] "Stress Corrosion Cracking of High Strength Alloys," 0414-01-5 (Quarterly), Jan. 1962, Contract DA-04-495-ORD-3069, Aerojet, Asuza, Calif. [2] Phelps, E. A. and Loginow, A. W., "Stress Corrosion of Steels for Aircraft and Missiles," Corrosion, Vol. 16, July 1960. [3] Dean, S. W. and Copson, H. R., "Stress Corrosion Behavior of Maraging Nickel Steels in Natural Environments," Corrosion, Vol. 21, March 1965. [4] "Stress Corrosion Cracking of High Strength Alloys," 2914 (Final), Aug. 1964, Contract DA-04-495-ORD-3069, Aerojet, Asuza, Calif. [5] Dreyer, G. A. and Gallaugher, W. C., "Investigation of the Effects of Stress Corrosion on High Strength Steel Alloys," Technical Documentary Report ML-TDR-64-3, Feb. 1964, Air Force Materials Laboratory, Wright-Patterson Air Force Base, Ohio, AD 605 672. [6] Batt, H. J. and Phelps, E. H., "Effect of Solution pH on the Mechanism of Stress Corrosion Cracking of a Martensitic Stainless Steel," Corrosion, Vol. 19, Sept. 1961. [7] "Stress Corrosion of High Strength Steels and Alloys; Artificial Environment," Research Project 389-2 (Final), July 1960 to July 1962, Mellon Institute, Pittsburgh, Pa. Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Copyright by ASTM Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further 106 STRESS CORROSION TESTING [8] Davis, R. A., Dreyer, G. A., and Gallaugher, W. C., "Stress Corrosion Cracking Study of Several High Strength Steels," Corrosion, Vol. 20, March 1964. [9] Hildebrand, J. F., "Stress Corrosion Cracking of High Strength Nickel Alloys for Aircraft Applications," Materials Protection, Vol. 3, Sept. 1964. [10] Hildebrand, J. F., Turns, E. W., and Nordquist, F. C., "Stress Corrosion Cracking in High Strength Ferrous Alloys," Materials Protection, Vol. 2, Nov. 1963. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No furth M. A. Pelensky1 and A. Gallaccio1 Stress Corrosion of Magnesium Alloys—Environmental Factors REFERENCE: M. A. Pelensky and A. Gallaccio, "Stress Corrosion of Magnesium Alloys—Environmental Factors," Stress Corrosion Testing, ASTM STP 425, Am. Soc. Testing Mats., 1967, p. 107. ABSTRACT: Several years ago, members of ASTM Committee B-3, Subcommittee X (now Committee G-l, Subcommittee VI) on Stress Corrosion determined a need for review of work on stress corrosion testing prior to development of standards and criteria for determining stress corrosion susceptibility and behavior of various alloys. The subcommittee established three task groups to accomplish a review of the subject. Task group assignments were as follows: TG-1, sample selection and method of stress; TG-2, environments under which stress corrosion occurs and duration of tests; and TG-3, reporting of stress corrosion results and interpretation of results. This report is part of the TG-2 assignment. It concerns a review of reported stress corrosion investigations relating to the effects of various environments on specific magnesium alloys. KEY WORDS: corrosion, stress corrosion, magnesium alloys, environmental testing A survey is presented of reported investigations concerning environmental effects on stress corrosion failure of a variety of magnesium alloys. Reported test results were tabulated by environments (airs, waters, and solutions) with alloy compositions, strengths, stress levels, forms, direction, and methods of stressing where this information was reported. Informational sources are cited. Comments and observations concerning selection of environments, test methods, sampling, and results are presented. Some suggestions are offered. Environments and Materials The conditions of test and the materials covered in this survey are as follows: 1 Research chemist and chief, respectively, Protection and Preservation Branch, Copyright by ASTM Int'l (all rights reserved); Wed DecArsenal, 16 15:53:43 EST 2015 Pa. Pitman-Dunn Research Laboratories, Frankford Philadelphia, Downloaded/printed by 107 University of Washington (University of Washington) pursuant to License Agreement. No further reproductions aut 108 STRESS CORROSION TESTING Environment (Unless otherwise indicated, ambient temperature and continuous, total immersion of the specimen! apply.) Air: Indoor, ambient RH and temperature Indoor, 85 to 90% RH Indoor, 90% RH, ambient temperature; O2 added; SO2 added; CO2 added Indoor, 95% RH, ambient temperature; O2 added; SO2 added; CO2 added Indoor, 100% RH, ambient temperature; O2 added; SO2 added; CO2 added Outdoor, marine Outdoor, NBS (National Bureau of Standards) Outdoor, rural Waters: Distilled (partial and total immersions) Sea Solutions: NaCl (0.58 g/1) + K2Cr207 CsCl (101 g/1) (0.15 to 14.9 g/1) HF (0.20 g/1) HN03 (0.63 g/1) NaCl (29 g/1) + K2Cr207 (14.9 H2SO4 (0.010 to 12.25 g/1) g/1) KC1 (45 g/1) NaCl (35 g/1) + K2CrO4 (1 to KHF2 (5 g/1) (partial immer50 g/1) sion) NaCl (35 g/1) + K2Cr04 (20 K2CrO4 (20 g/1) + NaCH3COO g/1) (82 g/1) NaCl (58 g/1) + K2Cr04 (20 K2Cr04 (20 g/1) + Na2C03 g/1) (53 g/1) NaCl (40 to 220 g/1) + K2CrO4 K2CrO4 (20 g/1) + NaNO3 (5 g/1) (85 g/1) NaCl (200 g/1) + K2Cr04 (5 to K2Cr04 (20 g/1) + Na2SO. 200 g/1) (71 g/1) Na2CO3 (0.265 to 53 g/1) NaCH3COO (82 g/1) NaF (0.42 g/1) NaBr (62 g/1) Nal (90 g/1) NaCl (0.058 to 58 g/1) NaN03 (85 g/1) NaCl (0.1 g/1) (intermittent NaOH (0.40 g/1) immersion) (0.71 to 712015 g/1) NaCl(35g/l) H(all Copyright by ASTM+Int'l rights reserved); WedNa Dec 16 415:53:43 EST 2S0 2 S0 4 (4.9g/l) Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduction TABLE I—Sample of data. A»<* sJeSlsi Test Specimen Stressed, % yield ]Direction of Stress NACL (40 TO 220 AZ61 (AM-C57S-H) dm dm dm dm dm f f f f f wr wr wr wr wr >100 >100 >100 >100 >100 G/L) Stress Method dm dm dm dm dm f f f f f wr wr wr wr wr 100 100 100 100 100 dm dm dm dm dm dm dm dm dm dm Ref + K 2 CRO 4 (5 G/L) NA€L (200 G/L) --1- K2CRO 4 AZ61 (AM-C57S-H) No. Specimens, time-to-failure bb bb bb bb bb (5 TO bb bb bb bb bb dm 0 g/1 NaCl nf 980 hr dm 40 g/1 NaCl 17 hr (e) dm 80 g/1 NaCl 6 hr (e) dm 160 g/1 NaCl 3 hr (e) dm 220 g/1 NaCl 5 hr (e) [in 200 G/L) dm dm dm dm dm 5 g/1 K2CrO4 2 hr, 12 hr (e) 20 g/1 K 2 CrO 4 0.07 hr (e) 50 g/1 K 2 CrO 4 0.07 hr (e) 100 g/1 K 2 CrO 4 0.04 hr (e) 200 g/1 K 2 CrO 4 0.02 hr (e) dm dm dm dm dm dm 0.265 g/1 21 hr (e) 0.53 g/1 23 hr (e) 5.3 g/1 22 hr (e) 7.95 g/1 24 hr (e) 15.9 g/1 23 hr (e) 53 g/1 12 hr (e) (in NA2C03 (0.265 TO 53 G/L) AZ61 (MA3) Abbreviations: dm = data missing f wr = flat wrought bb = bent beam Copyright by ASTM Int'l (all dm dm dm dm dm dm dm dm dm dm dm dm dm dm dm dm dm dm nf = no failure hr = hour(s) e = estimated from graphs dm dm dm dm dm dm dm dm dm dm dm dm rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized. [9] 110 STRESS CORROSION TESTING Alloys The magnesium alloys reported in this review include the following: Ml AZ31 AZ31, Ml clad AZ51 AZ61 AZ80 ZK-60A-T5 Mg (high purity) Mg + 5A1 MWY1 (5A1, 0.3Mn, O.OBFe) MWY3 (5A1, 0.3Mn, 0.0019Fe) Mg + 8A1 Mg + l.SIMn, 0.15 to O.SCe Mg + 2Mn MA2 (3 to 4A1' °-2 0.15 to O.SMn) Mg + 14Li to °-8Zn> Data The tabulated data compiled in this survey are available through the authors. Table 1 is a sample of the data. Discussion Concerning the Data Air Environments No failures of ZK-60A-T5 alloy are reported for ambient indoor air exposures [I].2 In indoor air at ambient temperatures, the "critical" relative humidity at which AZ61 alloy undergoes cracking, is approximately 98 to 100 per cent [2]. Increasing concentrations of O2 or CO2 in the air decreases the "critical" relative humidity to 95 per cent, while increasing the concentration of SO2 has no apparent similar effect [2]. When the relative humidity is close to 100 per cent, the presence of significant amounts of SO2 and CO2 decrease the tune to cracking [2]. The incidence of cracking rises during periods of rain or of high humidity and temperature [3]. The seasonal time period of exposure is an important factor to be considered in attempting to compare results of specimens exposed outdoors [3]. AZ31, clad and unclad, and AZ61 alloys exposed in marine environments are shown to fail by stress corrosion [3,4,5], Results, however, do not indicate any failures of the Ml alloy in this environment [3]. The AZ31, AZ51, AZ61, and AZ80 alloys fail, after different periods of exposure outdoors at the NBS site, Washington, D. C. [4\. Comparison of AZ31 test results suggests that the outdoor exposure environment at the NBS site is more severe than the marine site at Hampton Roads, Va. M2 The italic numbers refer toWed the Dec list 16 of 15:53:43 references appended to this Copyright by ASTM Int'l in (allbrackets rights reserved); EST 2015 paper. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduc PELENSKY AND GALLACCIO ON MAGNESIUM ALLOYS 111 Results indicate stress corrosion failures of AZ31 and AZ61 specimens exposed to outdoor rural environment in the vicinity of Midland, Mich. [3]. Time-to-failure results of AZ61 specimens exposed at Kure Beach (800 ft from ocean) are shown comparable to those exposed to the rural-industrial Michigan environment [3]. Waters ZK 60A-T5 alloys fail in very short periods of time when exposed to sea water [7]. The sea water exposure is much more severe than exposure to distilled water [7]. The Mg + 5A1 alloys fail when partially immersed in distilled water, usually at or just below the water level [6]. Solutions ZK 60A-T5 alloy stressed at 90 per cent of the yield strength failed rapidly (in 7 min or less except for one specimen) in 0.6 N solutions of KC1 (45 g/1), CsCl (101 g/1), NaBr (62 g/1), NaCl (35 g/1), and Nal (90 g/1) [7]. As reported, the time preceding cracking of AZ61 (MA3) alloy in NaCl (35 g/1) + K2CrO4 (20 g/1) decreases as temperature is increased from 41 to approximately 104 F [7]. However, on further increase of temperature the time preceding cracking increases, and at temperatures above 104 to 122 F the alloy acquires a relatively high resistance to stress corrosion cracking [7]. The time-to-cracking of MA2 alloy in NaCl (35 g/1) + H2SO4 (4.9 g/1) is shown to decrease with increasing temperature from 50 to 158 F [7,8]. This relationship is also shown for AZ80 (MA5) [7]. In Na2CO3 , susceptibility of AZ61 (MA3) alloy to stress corrosion failure remains almost constant as the concentration of Na2CO3 increases from 0.265 to 15.9 g/1 but increases appreciably with concentration increases to 53 g/1 [9]. Results indicate that Mg (high purity) and Mg + 2Mn alloys fail by stress corrosion when partially immersed in KHF2 (5 g/1) [6]. No failures of AZ61 (MA3) alloy were reported to occur in 260 hr in the following solutions [70]: 2 KCrO4 (20 g/1) + NaCH3COO (82 g/1) 2Cr04 (20 g/1) + NaC03 (53 g/1) K 2Cr04 K (20 g/1) + NaC03 (53 g/1) A decreasing order of susceptibility of AZ61 (MA3) to stress corrosion failure in various solutions is indicated by the time-to-failure data as folCopyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 lows [10]: Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized. 112 STRESS CORROSION TESTING NaCl (58 g/1) + K2Cr04 (20g/l), 4 min K2CrO4 (20 g/1) + Na2SO4 (71 g/1), 9 min Na2SO4 (71 g/1), 40 hr NaN03 (85 g/1), 43 hr Na2CO3 (53 g/1), 107 hr NaCl (58 g/1), 265 hr NaCH3COO (82 g/1), 650 hr Time-to-stress corrosion failure of AZ61 (AM-C57S-H) is shown to decrease significantly with increasing concentrations of K2CrO4 (3 to 200 g/1) in NaCl (35 g/1) solution [11]. Similarly, increasing concentrations of NaCl (40 to 200 g/1) in K2CrO4 (5 g/1) solution result in increasing susceptibility of AZ61 to stress corrosion failure [11]. Increased susceptibility to stress corrosion failure of AZ61 (MA3) and Mg + 1.5Mn alloys with increasing concentration of NaCl solutions is also indicated [72,73]. Stress corrosion failures of various AZ31 and AZ61 alloys exposed to NaCl (35 g/1) + K2CrO4 (20 g/1) are shown in the tabulated data [3,4,77, 14]. Alloy AZ61 (AM-C57S-H) immersed in NaCl (35 g/1) + K2CrO4 (20 g/1) fails rapidly (8 min or less) by stress corrosion at pH 2 through pH 12; however, at pH 13 no failure is reported after 980 hr exposure [77]. Alloys The susceptibility to stress corrosion of various magnesium alloys increases with increasing stress [3,4,6,13,14]. Threshold stress levels are suggested for various alloys in certain exposures, for example, AZ61 in NaCl (35 g/1) + K2CrO4 (20 g/1) at approximately 60 per cent of the yield strength when tested by the bentbeam method [14]. AZ31 sheet when subjected to constant axial stress in a rural environment appears to have a stress corrosion threshold strength in the neighborhood of 20 ksi [3]. However, it is reported that under bending stress and exposed to marine environment the minimum stress to produce fracture is 25 ksi [3]. AZ61 is shown to be more sensitive to stress corrosion failure than AZ31. Threshold limits under axial loading appear to be of the order of 12 to 15 ksi [3]. Under constant bending, the threshold limit appears to be approximately 15 to 20 ksi [3]. At moderate stresses, resistance to cracking of wrought magnesium alloys (AZ31, AZ51, and AZ61) decreases with increasing aluminum content up to approximately 6.5A1 [4]. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized. PELENSKY AND GALLACCIO ON MAGNESIUM ALLOYS 113 Ml clad AZ31 alloy is shown to be more resistant to stress corrosion failure in marine environment at high stresses than unclad AZ31 [4]; no failure occurred after approximately 500 days for Ml clad AZ31 stressed to 30 ksi compared to failure at approximately 214 days for unclad AZ31 stressed to 20 ksi. Susceptibility to stress corrosion of Mg + 5A1 alloys in distilled water appears to increase with increase in iron content [6]. Results indicate that stress corrosion susceptibility of Mg-Li alloys in humid air is associated with aluminum content and with the condition resulting from rapid cooling of these alloys from high temperatures [75]. Heating and stabilizing at 300 F appears to result in stress corrosion resistance in these same alloys [75]. Mg-Li alloys with no significant aluminum content withstand this humid air exposure under the various conditions tested [75]. Hardened AZ31 alloys appear less susceptible to stress corrosion failure than do the annealed alloys exposed to the NBS site [4]. AZ61 (Jlh) alloy annealed at 350 F appears to have greater resistance to stress corrosion at a higher per cent yield strength in marine environment than do the cold-rolled (J1CR) and hot-rolled (Jlr) AZ61 specimens and also the AZ31 (FS1) alloys [3]. Stress Methods The stressing methods reported on include U-bend, bent beam, direct tensile, spring load, and eccentric loading. Where comparisons can be made, it appears that an axial tensile stress method may be more severe at lower stress levels than bent beam, for example, AZ31 (FSlh) and AZ61 (Jla) alloys exposed in a marine environment [3]. Observations A number of environments with their effects on a variety of magnesium alloys have been investigated. The alloys reported include those of different strengths and those which were subjected to different applied stresses (from approximately 11 to 43 ksi yield stress and stresses applied up to approximately 135 to 140 per cent of yield). It may be observed from the data that most of the environments (including airs, waters, and solutions) selected for testing result in stress corrosion failure of the magnesium alloys. The environments selected in some instances are intended to duplicate or at least closely represent service conditions. Other selections were made to accelerate testing but were intended also to aid in predicting actual service life. Results in some cases lead to specific conclusions regarding stress corrosion failure, other results This2015 latter Copyright by ASTMbut Int'lin(all rightscases reserved); Wedare Decinconclusive. 16 15:53:43 EST Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further rep 114 STRESS CORROSION TESTING situation may be due to the wide scatter of data in a number of instances, or the small number of specimens employed in testing, or to both. The large variations or scatter in some test results could be due to: differences in the alloy (variations in the as-received condition or resulting from preparation of specimens for testing), differences in environment (contamination, concentration, or temperature gradients, seasonal time of exposure, etc.), or the test method employed. It is difficult to attempt to make reliable predictions of stress corrosion service life based solely on accelerated testing results. Relative susceptibilities to stress corrosion failure are frequently estimated from laboratory tests. Caution should be used in attempting to relate accelerated test results with results of outdoor exposure and in predicting service life of materials based on results of exposure to outdoor environments. The latter reflect differences in test results attributable to daily and seasonal variations of temperature, humidity, and rain conditions. In setting up any stress corrosion test to evaluate environmental effects, many factors must be considered. The aims or purpose of the test should be clearly established prior to initiating the test. The alloy or alloys selected and their mechanical and metallurigical variables (composition, structure, form, treatment, strength, level, and types of stress) must be considered. The rate of strain also may be an important factor to consider. The specific environment or environments to which the material is to be subjected must be determined, including, as applicable, consideration of temperature, relative humidity, composition and concentration, and pH. Total or partial, continuous or intermittent immersion, and static or dynamic conditions may be involved. Finally, it is important that the experiments be designed to facilitate statistical analysis of the test results. References [1] Crossley, F. A., "Research on the Basic Nature of Stress Corrosion for Various Structural Alloys at Room and Elevated Temperature," ASD-TR-61-713, May 1962, Contract AF 33(616)-7612, Armour Research Foundation for Directorate of Materials and Processes, Aeronautical Systems Div., Air Force Systems Command, Wright-Patterson Air Force Base, Ohio. [2] Timonova, M. A., "Corrosion Cracking of Magnesium Alloys and Methods of Protection Against It," Intercrystalline Corrosion and Corrosion of Metals Under Stress, Levin, I. A., ed., Consultant Bureau Enterprises, Inc., 1962. [5] Loose, W. S. and Barbian, H. A., "Stress Corrosion Testing of Magnesium Alloys," ASTM-A1ME Symposium on Stress Corrosion Cracking of Metals, 1944, American Society for Testing and Materials, Philadelphia, 1945. [4] Logan, H. L. and Hessing, Harold, "Stress Corrosion of Wrought Magnesium Base Alloys," Journal of Research of the National Bureau of Standards, Vol. 44, March 1950. [5] Romanov, V. V., "Corrosion of Magnesium," Translation No. 1-8207, 11 April, 1963 (refers to Romanov, V. V., Corrosion of Magnesium, Academy of Sciences, USSR, 1961). [6] Ferryman, E. C. W., "Stress Corrosion of Magnesium Alloys," Journal of the Institute of Metals, Vol. 78, Sept. 1950 to Feb. 1951. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 [7] Romanov, V. V., "Corrosion of Magnesium," Translation No. 1-8207, 11 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No furthe DISCUSSION ON MAGNESIUM ALLOYS 115 April, 1963 (refers to collection of papers, Strength of Metals, IME T, Academy of Sciences, USSR, published by USSR Academy of Sciences, Moscow, 1956). [8] Romanov, V. V., "Effect of Some Factors on the Susceptibility of Magnesium Alloys to Corrosion Cracking," Inter crystalline Corrosion and Corrosion of [9] [10] [11] [12] [13] [14] [75] Metals Under Stress, Levin, I. A., ed., Consultant Bureau Enterprises, Inc., 1962. Romanov, V. V., "Corrosion of Magnesium," Translation No. 1-8207, 11 April, 1963 (refers to Zaretsky, E. M., Zhurnal Prikladnoy Khimii, 1951). Romanov, V. V., "Stress Corrosion Cracking of Metals," Israel Program for Scientific Translations, 1961, (refers to Romanov, V. V., Sbornik, IME T, AN SSR, 1958). Sager, G. F., Brown, R. H., and Mears, R. B., 'Tests for Determining Susceptibility to Stress-Corrosion Cracking," ASTM-AIME Symposium on Stress Corrosion Cracking of Metals, 1944, American Society for Testing and Materials, Philadelphia, 1945. Romanov, V. V., "Corrosion of Magnesium," Translation No. 1-8207, 11 April, 1963 (refers to Zaretsky, E. M., Doklady Akademii Nauk SSSR, 1947). Tomashov, N. D. and Modestova, V. N., "Effect of Stress on the Corrosion and Potentials of Alloys of the Magnesium-Manganese System," Intercrystalline Corrosion of Metals Under Stress, Levin, I. A., ed., Consultant Bureau Enterprises, Inc., 1962. Mears, R. B., Brown, R. H., and Dix, E. H., Jr., "A Generalized Theory of Stress Corrosion of Alloys," ASTM-AIME Symposium on Stress Corrosion Cracking of Metals, 1944, American Society for Testing and Materials, Philadelphia, 1945. Kiszka, J. C., "Stress Corrosion Tests of Some Wrought Magnesium-Lithium Base Alloys," M65-1-1, July 1964, Frankford Arsenal, Philadelphia, Pa.; Materials Protection, Feb. 1965. DISCUSSION F. A. Smith1 (written discussion)—I would appreciate your consideration of the possible heat-to-heat variation in chemical composition of magnesium alloys as a possible way to explain the large variation in timeto-failure for specimens discussed in your paper. M. A. Pelensky and A. Gallaccio (authors)—It is expected that heatto-heat differences in chemical composition could account for some of the variation in time-to-failure, since, as reported in this paper, variations in chemical composition of somewhat different alloys and even different heat treatments of the same alloy can result in different times-to-failure. However, the degree of influence of heat-to-heat variation in chemical composition is difficult to determine in view of other factors which can result in variations in time-to-failure. These other factors which must be considered include variations in physical microstructure, differences in concentration, contamination, temperature, and pH of the corrosive medium, and also nonuniformity of stresses, both residual and applied. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 1 Physicist, Argonne Downloaded/printed by National Laboratory, Argonne, 111. University of Washington (University of Washington) pursuant to License Agreement. No further reproductio A. P. Bond,1 J. D. Marshall? and H. J. Dundas* Resistance of Ferritic Stainless Steels to Stress Corrosion Cracking REFERENCE: A. P. Bond, J. D. Marshall, and H. J. Dundas, "Resistance of Ferritic Stainless Steels to Stress Corrosion Cracking," Stress Corrosion Testing, ASTM STP 425, Am. Soc. Testing Mats., 1967, p. 116. ABSTRACT: Stress corrosion cracking tests were carried out on 0.020in.-diameter wires under dead-weight loading in boiling solutions of MgCU , Ca(NO3)2 , and NaOH. The T430 and T434 ferritic stainless steels were immune to stress corrosion cracking in MgCU under conditions which caused various austenitic grades to fail in 6 to 46 min. Pitting corrosion in MgCl2 reduced the cross-sectional area of the ferritic steel specimens to the extent that they could not support the load, and they failed in a ductile manner. Stressed specimens of ferritic stainless steels did not fracture in Ca(NOs)2 or NaOH even after heat treatments which led to intergranular corrosion. Corrosion potential measurements were made on the stressed specimens and correlated with various stages in the stresscorrosion cracking process. KEY WORDS: corrosion, stress corrosion, steels, stainless steels, pitting, heat treatment The sensitivity of austenitic stainless steels to stress corrosion cracking in chloride-containing environments often limits the use of these steels. Since it has been reported that Type 430 ferritic stainless steel is highly resistant to cracking even in the presence of chlorides [./,2],4 the present work was undertaken to ascertain the combined effects of stress and corrosive environments on Types 430 and 4345 ferritic stainless steels. In addition to boiling magnesium chloride (MgCl2) solutions, boiling solutions of calcium nitrate (Ca(NO3)2) and of sodium hydroxide (NaOH) 1 Research group leader, Climax Molybdenum Company of Michigan, Ann Arbor, Mich. 2 Formerly with Climax Molybdenum Company of Michigan as research associate; now assistant metallurgist, Metallurgical Dept., W. S. Tyler Co., Cleveland, Ohio. 3 Laboratory associate, Climax Molybdenum Company of Michigan, Ann Arbor, Mich. 4 The italic numbers in brackets refer to the list of references appended to this paper. B Type 434 is the proposed designation for stainless steel containing 17 per cent Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 chromium + 1 per cent molybdenum. Downloaded/printed by 116 pursuant to License Agreement. No further reproductions authorized. University of Washington (University of Washington) BOND ET AL ON RESISTANCE OF FERRITIC STAINLESS STEELS 117 were used as test environments, since low-alloy steels are known to undergo stress corrosion cracking in hot nitrates or caustics [3]. Test Procedure Materials and Heat Treatment The materials evaluated during this investigation were commercial heats of stainless steels; their compositions are presented in Table 1. All of the steels were evaluated in the annealed condition. The annealing treatment for the austenitic steels consisted of heating the wires at 1950 F for 15 TABLE 1—Composition of alloys (weight per cent). Alloy Type c 201 + 1.7 Cu... 301 304 316 430, Heat A . . . . 430, Heat B 434, Heat A 434, Heat B 434, Heat C 0.095 0.045 0.058 0.074 0.043 0.065 0.066 0.059 0.077 0.069 201 Mn 6.32 6.50 0.60 0.75 1.57 0.49 0.40 0.49 0.51 0.42 Si 0.44 0.56 0.43 0.63 0.47 0.34 0.36 0.47 0.38 0.54 S p 0.007 0.009 0.002 0.005 0.015 0.007 0.011 0.010 0.006 0.007 0.029 0.030 0.017 0.024 0.029 0.015 0.017 0.018 0.022 0.016 Cr Ni Mo N2 Cu 16.53 4.83 0.098 0.039 1.73 16.50 5.40 17.56 7.19 0.23 0.037 0.189 18.38 9.40 0.27 17.52 13.05 ? 16 17.06 0.29 0.07 16.00 0.24 0.20 17.71 1.01 17.37 1.01 0.11 16.52 0.30 0.99 TABLE 2—Strength properties of the 0.020-in.-diameter ferritic stainless steel wires, psi. Type 430, Heat B Condition Annealed 1 hr, 900 F 15 min, 1800 F Type 434, Heat A Proportional Limit Tensile Strength Proportional Limit Tensile Strength 35 000 40 000 92 000 80 000 80 000 134 000 52 000 45 000 78 000 91 000 90 000 135 000 min; the ferritic steels were annealed by heating at 1500 F for 1 hr. The ferritic steels were also evaluated after sensitizing treatments of 1 hr at 900 F, or 15 min at 1800 F. The proportional limit and the tensile strength of the ferritic steels in the three different conditions are presented in Table 2. The majority of the tests were carried out on 0.020-in.-diameter wires of the steels. However, some tests on the Type 434 material were conducted on 0.175-in.-thick strip specimens. Stress Corrosion Tests Specimens, in the form of wires, were exposed to the corrosive environCopyright by ASTM Int'l a (allconstant rights reserved); Wed Dec 16 tensile 15:53:43 load; EST 2015 ments while under (dead-weight) a schematic Downloaded/printed by cell is shown in Fig. 1. A cell was constructed of glass drawing of the test University of Washington (University of Washington) pursuant to License Agreement. No further reproduction 118 STRESS CORROSION TESTING for use with Ca(NO3)2 and with MgCl2 environments. A similar cell, constructed of Teflon, was used for the tests with NaOH and for additional experiments with MgCl2 environments. The corrosion media were: MgCl2 solution, the concentration of which was adjusted to yield a boiling point of 284 F (140 C); 55 per cent FIG. 1—Stress corrosion test cell for wire specimens. TABLE 3—Failure times for 0.020-in.-diameter wire specimens of the various steels tested in MgClz solution boiling at 284 F (140 C). 4900 psi Stress Alloy Treatment 701 201 + 1.7 Cu.. 301 304 316 430, Heat A 430, Heat B 434, Heat A 434, Heat A 434, Heat A 434, Heat B annealed annealed annealed annealed annealed annealed annealed annealed 1 hr, 900 F 15 min, 1800 F 15 min, 1800 F 0 6 c StandAvg Time- No. ard of Devito-Failure. hr Tests ation, hr 0 .35 0 .37 1.53 1.68 2 .92 42 NF" 42 NF° 3 3 5 7 5 1 1 0.048 0.026 0.535 0.588 0.400 53,000 psi Stress Avg Time- No. of to-Failure, hr Tests 0. P 0. P 0. 475 0. 436 0. 776 1. 82^ 1.98' 45. 6' 60. 5C 327" 297' 3 3 2 63 6 3 1 1 1 1 1 Standard Deviation, hr 0.0 0.0 0.106 0.042 0.103 0.31 NF = no failure in indicated exposure time. Failure by stress corrosion cracking. Ductile failure by pitting corrosion. Ca(NO3)2 , boning at 242 F (117 C); and 25 per cent NaOH, boiling at 232 F (111 C). The MgCl2 solution was reused for up to ten tests with the austenitic stainless steels. The solution was checked from time-to-time by running a Type 304 specimen at 53,000 psi; a failure time near 26 min was taken as indicating that the solution was suitable for use. In all tests of the ferritic steels, a fresh solution was used for each experiment. During Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprod BOND ET AL ON RESISTANCE OF FERRITIC STAINLESS STEELS 119 some of the tests in MgCl2, potential measurements were made on the stressed specimens. For these potential measurements, a saturated calomel electrode was inserted in the side arm shown in Fig. 1. The potential as a function of time was autographically recorded. In addition to the runs on wire specimens, a few tests were performed on Type 434 steel using U-bend specimens made by bending l/z by 2J/2 by 0.175-in. strips over a 3/s-in. mandrel. The legs of the bend were secured by a steel bolt. The bolt and the adjacent areas of the specimen were covered with neoprene, and the U-bend was suspended in the solution in such a way that the coated bolt was not immersed in the liquid. FIG. 2—Longitudinal cross section of Type 434 Heat A specimen heat treated 15 min at 1800 F and exposed 327 hr in boiling MgClz at a stress of 53,000 psi. Unetched (X500). Results and Discussion The results of stress corrosion tests in boiling MgCl2 are shown in Table 3. Note that no stress corrosion cracking occurred in the ferritic steels, Types 430 and 434. The failures which did occur were of the ductile type, resulting from reduction of cross-sectional area due to pitting attack. Metallographic examination did not reveal any cracks in the failed specimens. The photomicrograph (Fig. 2) shows the appearance of a typical failure zone in ferritic specimens. Both ferritic steels were subject to pitting in this aggressive environment, but the molybdenum-containing Type 434 was much more resistant to pitting, as evidenced by its time-to-ductile failure being greater than that of the straight chromium steel. The Type 434 wires tested in the glass cells failed at the point where the wire entered the side arm. This suggests that a form of crevice attack was accelerating the pitting failure. The tests were repeated in Teflon cells hi which the wire entered a Teflon plug side arm Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduc 120 STRESS CORROSION TESTING which fitted the wire tightly, thus minimizing the crevice attack. The Teflon cell increased the time-to-failure due to pitting. All results shown in Table 3 for the ferritic steels were obtained in the Teflon cell. The heat treatments for the ferritic steels were selected to give several degrees of sensitivity to intergranular corrosion. The 1800 F treatment is known to sensitize ferritic steels to intergranular attack [4]. The 900 F treatment sensitizes Type 434 to a lesser extent but has little effect on Type 430, according to work in progress at this laboratory. Of course, the ferritic steels in the annealed condition exhibit the minimum sensitivity to intergranular corrosion. The depth of pitting of the ferritic steels was also affected by heat FIG. 3—Potential versus time for stainless steel specimens stressed at 53,000 psi while immersed in at 284 MgCl F (140 C). 2 solution boiling treatment. The 1800 F treatment caused a great increase in the time-tofailure (due to pitting) in MgCl2 . Apparently, this was mainly due to the initiation of a much larger number of pits on the sensitized Type 434, with the individual pits then growing more slowly than on the annealed steel as a result of the reduced cathode-to-anode area ratio. Thus, the large number of pitting sites introduced by heating at 1800 F had the effect of distributing the corrosion more uniformly relative to the annealed steel. Of course, the greater tensile strength of the specimens heat treated at 1800 F (Table 2) would also serve to increase the time-to-failure by pitting. However, this effect is probably too small to account for all of the observed increase in time-to-failure. Although the failures of the ferritic stainless steels under dead-weight Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 loading appeared to be completely ductile and gave no indication of stress Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduction TABLE 4—Results of exposing 0.020-in.-diameter specimens of ferritic steels to boiling 55 per cent Ca(NOz)z and to boiling 25 per cent NaOH at a stress of 53,000 psi. NaOH Ca(N08)2 Alloy 430, Heat B 430, Heat B 430, Heat B 430, Heat A 434, Heat A 434, Heat A 434, Heat A 434, Heat B Mild steel 0 Heat Treatment Exposure Time, hr Remarks Exposure Time, hr annealed 1 hr, 900 F 15 min, 1800 F 15 min, 1800 F annealed 1 hr, 900 F 15 min, 1800 F 15 min, 1800 F 1500 F 336 NF° very little corrosion very little corrosion intergranular attack intergranular attack very little corrosion very little corrosion intergranular attack intergranular attack failed intergranularly at 30,000 psi 336 NF" uniform corrosion 428 NF intergranular attack 336 NF 355 NF 336 NF 336 NF uniform corrosion uniform corrosion intergranular attack intergranular attack 336 NF 336 NF 362 NF 336 NF 336 NF 408 NF 500 NF 20.7 NF = no failure in indicated exposure time. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized. Remarks 122 STRESS CORROSION TESTING corrosion cracking, it was desirable to confirm this lack of stress corrosion cracking under conditions in which ductile failure would not occur. To do this, U-bend specimens of annealed Type 434, Heat C, were exposed to boiling MgCl2 solution. In this type of test, the maximum stress is somewhat greater than the yield strength but is not appreciably increased by FIG. 4—Longitudinal cross section of Type 434 Heat A specimen heat treated 15 min at 1800 F and exposed 336 hr in boiling 25 per cent NaOH at a stress of 53,000 psi. Unetched (X500). FIG. 5—Longitudinal cross section of Type 434 Heat B specimen heat treated 15 min at 1800 F and exposed 500 hr in boiling 55 per cent Ca(NO^ at a stress of 53,000 psi. Unetched (X500). pitting corrosion. Thus, ductile failure due to reduction of cross-sectional area cannot occur. Metallographic examination of two ferritic stainless steel specimens stressed as a U-bend and exposed for 75 days hi boiling MgCl2 revealed no evidence of stress corrosion cracks. All the austenitic stainless steels investigated were found to be very susceptible to stress corrosion cracking under the test conditions used. The Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 fact that the austenitic alloys cracked in very short times (see Table 3) Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprodu BOND ET AL ON RESISTANCE OF FERRITIC STAINLESS STEELS 123 emphasizes the importance of the absence of cracking in ferritic stainless steels evaluated under identical conditions. Potential-time relationships of the specimens stressed at 53,000 psi are shown in Fig. 3. The behavior of the austenitic steels is similar to that reported by other workers [5,6]. The initial rise in potential has been associated with partial film repair, while the region of potential drop corresponds to the crack propagation stage [5]. The initial rise in potential was not observed for the Type 201 steels tested at 53,000 psi, indicating that crack propagation began almost immediately. At a stress of 4900 psi, however, the initial potential rise was detected on all the austenitic steels. The variation of corrosion potential with exposure time was quite different for the ferritic steels. No peak in the corrosion potential was observed, although the Type 430 specimen did show a small potential drop as ductile failure occurred. The ferritic stainless steels were also tested in boiling 25 per cent NaOH and in boiling 55 per cent Ca(NO3)2 . Both of these media are known to cause intergranular stress corrosion cracking in plain carbon steels and in low-alloy steels which have undergone certain heat treatments [7,8], The results, summarized in Table 4, showed no failure of any of the test specimens. The specimens sensitized at 1800 F were subject to intergranular corrosion in both NaOH and Ca(NO3)2 , while the annealed specimens did not undergo intergranular corrosion. The extent of the intergranular corrosion is illustrated in Figs. 4 and 5. That intergranular corrosion and not stress corrosion cracking was the corrosion mechanism was confirmed by exposing unstressed specimens heat treated at 1800 F to the test environments. A very similar type of attack occurred on the unstressed specimens. One annealed Type 434 specimen (Heat A) was tested in boiling 50 per cent NaOH. Very rapid uniform corrosion caused ductile failure in 7 hr, so no further tests were made in this medium. It is not clear to what extent the resistance to stress corrosion cracking of the ferritic stainless steels, Types 430 and 434, is due to their crystal structure and to what degree to their chemical compositions. It has been suggested that the greater number of possible slip planes in the body-centered-cubic lattice as compared to the face-centered-cubic structure prevents stress concentration on any one glide plane reaching as high values in the body-centered-cubic as in the face-centered-cubic metals [9]. This cannot be the complete explanation of the resistance of the ferritic stainless steels to stress corrosion cracking, because some body-centered-cubic metals do undergo stress corrosion cracking under some conditions, for example, mild steel in nitrates. Another possibility is suggested by the observations of others [10,11] that some austenitic stainless steels containing less than 1 per cent nickel are immune to cracking in MgCl2 (as are Copyright by ASTM Int'l (all rights reserved); Wed amounts Dec 16 15:53:43 EST 2015 austenitic stainless steels containing large of nickel). Thus, it may Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductio 124 STRESS CORROSION TESTING be that the absence of nickel is responsible for the immunity of the ferritic steels to stress corrosion cracking. In fact, recent work with some experimental alloys has shown that fully ferritic 18 to 25 per cent chromium alloys containing as little as 1.1 per cent nickel do stress corrosion crack in boiling MgCl2 solution. Conclusions 1. Types 430 and 434 stainless steels do not undergo stress corrosion cracking even when stressed above their yield points in the following solutions: (a) MgCl2 boiling at 284 F (140 C), (b) Ca(NO3)2 boiling at 242 F (117 C), or (c) NaOH boiling at 232 F (111 C). 2. Type 430 stainless steel exhibits poor resistance to pitting corrosion in boiling MgCl2 solution. An addition of 1 per cent molybdenum to this steel (Type 434) increases the resistance to pitting corrosion. 3. Types 430 arid 434 steel heat treated at 1800 F are subject to intergranular corrosion in boiling Ca(NO3)2 and in boiling NaOH solutions. References [1] Scheil, M. A., "Some Observations of Stress-Corrosion Cracking in Austenitic Stainless Alloys," ASTM-AIME Symposium on Stress Corrosion Cracking of Metals, 1944, American Society for Testing and Materials, Philadelphia, 1945, pp. 395-410. [2] Scheil, M. A. et al, "First Report on Stress Corrosion, Corrosion Cracking of Stainless Steel in Chloride Solutions," Supplement, Journal of the American Welding Society, Oct. 1943, pp. 493-S-504-S. [3] Logan, H. L., "Stress Corrosion Cracking in Low Carbon Steel," Physical Metallurgy of Stress Corrosion Fracture, Rhodin, T. N. ed., Interscience, New York, 1959, pp. 295-309. [4] Baumel, A., "Relationship Between Heat Treatment and Corrosion Behavior of 17 Cr Stainless Steels in Boiling Concentrated Nitric Acid," Archiv fuer das Eisenhuttenwesen, Vol. 34, 1963, pp. 135-146. [5] Hoar, T. P. and Hines, J. G., 'The Electrochemistry of the Corrosion and the Stress-Corrosion Cracking of 18-8 Chromium-Nickel Steels in Hot Aqueous Magnesium Chloride Solution," Proceedings, 8th Meeting, International Committee of Electrochemical Thermodynamics and Kinetics, Madrid, 1956, Butterworths, London, 1958, pp. 273-291. [6] Van Rooyen, D., "Qualitative Mechanism of Stress Corrosion Cracking of Austenitic Stainless Steels," Corrosion, Vol. 16, 1960, pp. 421t-429t. [7] Long, L. M. and Uhlig, H. H., "Effect of Carbon and Oxygen in Iron on Stress Corrosion Cracking in Nitrate Solution," Journal of the Electrochemical Society, Vol. 112, 1965, pp. 964-967. [8] Uhlig, H. H. and Sava, J., "Effect of Heat Treatment on Stress-Corrosion Cracking of Iron and Mild Steel," Transactions, American Society for Metals, Vol. 56, 1963, pp. 361-376. [9] Denhard, E. E., discussion, Physical Metallurgy of Stress Corrosion Fracture, Rhodin, T. N., ed., Interscience, New York, 1959, p. 223. [10] Copson, H. R., "Effect of Composition on Stress Corrosion Cracking of Some Alloys Containing Nickel," Physical Metallurgy of Stress Corrosion Fracture, Rhodin, T. N., ed., Interscience, New York, 1959, pp. 247-267. [11] Riedrich, G. and Kohl, H., "Effect of Various Alloying Elements on the Susceptibility to Stress Corrosion of Austenitic Corrosion Resistant Steels," Bergund Hiittenmdnnische 108, 1963, Copyright by ASTM Int'lMonatshefte, (all rights Vol. reserved); Wed pp. Dec1-8.16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No fur DISCUSSION ON RESISTANCE OF FERRITIC STAINLESS STEELS 125 DISCUSSION /. F. Boyce1 (written discussion)—Exposure to boiling 42 per cent MgCl2 solution seems to have become a fad in stress corrosion testing. Is it not possible to be seriously misled by test results in this very severe environment in predicting behavior of the subject materials in sea water, tap water, or other mild environments, particularly those not containing appreciable concentrations of chloride ions? /. B. Marriott2 (written discussion)—The authors have shown the superiority of certain ferritic stainless steels over austenitic stainless steels in a number of test environments. In practice, however, the environments to which these steels are exposed may be much less corrosive, typical examples being chloride-contaminated high-pressure hot water and condensing steam. It would be desirable to know whether the results given here can be used when assessing these or other materials for mildly corrosion service. The authors also have referred to the effect of corrosion pit formation and growth on the life of their test specimens. Do they consider that this variable could be eliminated by the use of a precracked specimen, perhaps of the type described by Beachem and Brown? R. W. Staehte3 (written discussion)— Regarding possible objections to the use of boiling MgCl2 as a test environment, it is obvious that it is not a normal industrial fluid. However, the results obtained in this experiment with respect to alloy behavior are in good general agreement with work which we are conducting in very dilute chloride environments in the 200 to 300 C range. Furthermore, there is an enormous amount of information available from work in this environment, and I think that there is great value in having essentially an international standard for comparison among laboratories. A. P. Bond, J. D. Marshall, and H. J. Dundas (authors)—In reply to Mr. Boyce, a very large amount of data on the behavior of austenitic stainless steels in boiling MgCl2 has been obtained over many years. Thus, this solution seemed to be a logical one to use in comparing the stress corrosion cracking behavior of ferritic stainless steels with that of austenitic stainless steels. If chloride-induced stress corrosion cracking does not occur in this solution, it is usually considered that such cracking will not occur at lower temperatures or chloride-ion concentrations. Cer1 Metallurgist, Gillette Safety Razor Co., Gillette Park, Boston, Mass. The English Electric Co., Ltd., Central Metallurgical Laboratories, Cambridge Road, Whetstone, Nr. Leicester. 3 Assistant professor, Department of Metallurgical Engineering, Corrosion Center, The Ohio State University, Columbus, Copyright by ASTM Int'l (all rights reserved);Ohio. Wed Dec 16 15:53:43 EST 2015 2 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproducti 126 STRESS CORROSION TESTING tainly the occurrence of stress corrosion cracking in this severe test does not prove that it would occur under milder conditions, but it does indicate that it might. To obtain some information about the behavior of ferritic steels in environments free of chloride, testing was conducted in both boiling 55 per cent Ca(NO3)2 and in boiling 25 per cent NaOH. These environments were chosen because stress corrosion cracking of sortie mild steels had been observed in them. In reply to Mr. Marriott, the results in MgCl2 solution at 140 C do not guarantee that cracking might not occur at higher temperatures and lower chloride concentrations, especially under conditions in which the actual chloride concentration at the metal surface may be much greater than it is in the bulk environment. However, our results strongly suggest that the examined ferritic stainless steels would be at least highly resistant to stress corrosion cracking in the example cited. In this connection, Professor Staehle's comments are of great interest. Finally, it must be emphasized most strongly that pitting corrosion of the ferritic, Types 430 and 434, stainless steels in MgCl2 solution led to ductile failure when the reduction of the cross-sectional area caused the constant load to exceed the tensile strength of the materials. No evidence of stress corrosion cracking was found in this case. Thus the use of precracked specimens would serve no purpose. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductio H. L. Logan1 Some Techniques Used in the Study of Stress Corrosion Cracking REFERENCE: H. L. Logan, "Some Techniques Used in die Study of Stress Corrosion Cracking," Stress Corrosion Testing, ASTM STP 425, Am. Soc. Testing Mats., 1967, p. 127. ABSTRACT: Techniques used at the National Bureau of Standards in the study of stress corrosion cracking of metals are described together with precautions taken in these investigations. Especially designed specimens of low-carbon and stainless steels and a titanium alloy and supplementary techniques for obtaining data as to the mechanism of the stress corrosion process are also described. A specimen and technique recently used to determine whether hydrogen plays a part in the delayed failures of highstrength steels in chloride solutions is described. KEY WORDS: corrosion, stress corrosion, steels, titanium alloys, hydrogen embrittlement, chlorides, salt solutions, environmental testing, exposure testing The study of the stress corrosion cracking of metals has been a more or less continuous project at the National Bureau of Standards (NBS) for the past 25 years and a continuing project since 1950. This paper describes techniques that have been employed in the study of both the susceptibility of certain materials to stress corrosion cracking and the mechanism of cracking in certain alloy systems. Pitfalls that sometimes have not been avoided are described and certain interesting results, some of which have not been reported earlier, are mentioned. Early work during the war years (1941 to 1945) was devoted to problems such as: (a) determining the effect of elevated temperature aging on the resistance of the 2024 aluminum alloy to stress corrosion cracking [I],2 (b) determining the resistance of various high-strength aluminum alloys to stress corrosion cracking [2], and (c) determining the relative resistances of various magnesium alloys to stress corrosion cracking [3]. Subsequently, most of the investigations have had as their primary objective the 1 Physicist, National Bureau of Standards, Washington, D. C. Personal member ASTM. 2 The italic numbers brackets refer the16list of references Copyright by ASTM Int'l (allin rights reserved); WedtoDec 15:53:43 EST 2015 appended to this paper. Downloaded/printed by 127 pursuant to License Agreement. No further reproductions author University of Washington (University of Washington) 128 STRESS CORROSION TESTING study of the mechanism or mechanisms of stress corrosion cracking. These objectives have on occasion required special types of specimens. The specimens used for particular investigations have been described in papers giving the results of those investigations, but there has been no general discussion of specimen types, of the advantages or disadvantages of these specimens, or of precautions that may be either necessary or desirable in certain investigations. Similarly, specific methods of studying corrosion FIG. 1—Specimens in weather exposure racks at the National Bureau of Standards, Washington, D. C. Unstressed specimens were exposed between stressed specimens. Note grip supports for specimens and crossed flexure plates in pull rods. Common fulcrum for levers is seen on reverse side of rack. products have been discussed in individual papers but not in a review paper. Semiconventional Tension Specimens Because it was desirable to know the initially applied stress, much of the work has been done with specimens loaded in tension by means of lever systems and weights.3 Because stresses in bent beams can only be approx3 This stress is superimposed on any residual stresses from heat treatment or cold Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 work. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions a LOGAN ON TESTING TECHNIQUES 129 imated, bent-beam specimens have been used only where practical considerations rule out other types of specimens. (Stresses in U-bend specimens are unknown and for that reason they have not generally been used in our investigations.) Specimens Exposed in the Weather Ideally each specimen should have its own complete lever system. Such equipment is expensive, and some of the work at NBS has been done using racks having one continuous knife edge used by several levers for each of several specimens. A portion of a rack of this design used on the roof of the Northwest Building at NBS Washington site is shown in Fig. 1. Gripping devices for sheet specimens such as those used in tension testing are prohibitively expensive for multiple unit stress corrosion test racks. Pin devices, however, were found to be reasonably inexpensive and were satisfactory if properly designed and used (Fig. 1). The grip ends of Vs-hi. reduced section sheet tension specimens were either 1 in. or for some materials 11A in. wide. Because of the reduced stress concentration, a specimen that would fail in the pin area when supported by a V^-in.-diameter pin would not fail if a %6 or %-in. pin were used. The grip ends of pin supported thin sheet specimens (up to possibly l/s in. thickness with some alloys) must also be laterally supported to prevent failures at the pin holes. Axial alignment of specimens to prevent twisting or bending is very important. It was found that a slight twisting of magnesium alloy specimens in the corrodent produced failures either in the shoulders or the grip ends of the specimens, rather than in the reduced section, and at lower stresses than if specimens were properly aligned. The use of crossed chains or crossed flexure plates (Fig. 1) as a part of the pull rods can eliminate bending, but careful assembling of the specimen-support system, prior to loading, is required to prevent twisting. In the weather exposure tests (Fig. 1) unstressed specimens were placed adjacent to stressed specimens so that losses in tensile properties due to general corrosion and not to stress corrosion cracking could be evaluated. If resistances of different alloys are to be compared in the weather, either many specimens must be exposed or tests on all specimens should be initiated on the same day. Apparently weather conditions during the first 24 hr of exposure have a significant effect on the lives of the specimens. Loose and Barbian [4] reported that specimens placed in weather exposure racks in April had a much longer life than those initally exposed during the summer months. Further, it is our opinion that if the relative susceptibilities of different lots of material to stress corrosion cracking are to be compared, these materials should be statistically distributed in the weather exposure racks. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 It was noted in the NBS work with magnesium alloys that many speci- Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further repro 130 STRESS CORROSION TESTING FIG. 2—Sheet specimen exposed, dead loaded in tension, in a boiling corrodent. Air condenser and fan prevent evaporation of corrodent. Support system can be interchanged so as to take a cylindrical specimen with threaded ends. mens failed after a rain. Some specimens that were severely cracked after a rain, nevertheless, did not fail until the next rain. The postulated mechanism [5] was that rain water plus corrosion products temporarily destroyed the protective film at certain locations on the specimen surface. This permitted escape of dislocations with resulting slip and the exposure of narrow film-free surfaces to the moisture. These areas were anodic to the very much larger surface ofDec the 16 specimen. If the resulting Copyright by ASTM Int'lfilm (all covered rights reserved); Wed 15:53:43 EST 2015 Downloaded/printed by electrochemical-mechanical action was more rapid than the film repair, University of Washington (University of Washington) pursuant to License Agreement. No further reproduc LOGAN ON TESTING TECHNIQUES 131 the crack propagated to specimen failure. In some of the specimens, the film repair process eventually became sufficiently rapid to stop crack propagation. These specimens did not fail until the next rain. Specimens Exposed in Laboratory Corrodents If specimens are to be exposed to a specific environment rather than the weather, they must be enclosed in suitable cells. Figure 2 shows a typical cell used at the NBS. This was made from 40-mm Pyrex glass tubing with the ends drawn down to fit available stoppers. For elevated temperature work, the cell wall may be insulated with asbestos paper and a heating coil of resistance wire wound onto the wet paper. This is, in turn, covered with another thickness or two of asbestos paper. Heating tapes can of course be used in place of the coils described above. The cells are closed at the bottom and top with slotted rubber stoppers. Evaporation of the solution can FIG. 3—Four-point loaded specimens that may be exposed in a furnace. normally be prevented by the use of an air condenser and a small electric fan.4 If the specimens are to be subject to intermittent immersion, the solution may be raised into the cells from a central reservoir by means of either a mechanical pump or compressed air operated on a fixed cycle by means of a clock system. It is difficult to adjust the flow from a central reservoir to a number of cells so that there will be no overflows. For that reason an overflow tube, of slightly larger diameter than the input tube with its intake at the maximum desired solution level, should lead back through the lower stopper from each cell to the reservoir. Note again that if the corrodent produces any general corrosion at all in the material under test, unstressed specimens should also be exposed 4 The use of small 110-v a-c electric fans such as are available for cooling elecCopyright by ASTM (all rightsbyreserved); Wed Dec EST 2015 tronic equipment wasInt'l suggested E. H. Phelps and 16 A.15:53:43 W. Loginow, Applied Research Laboratory, U. Downloaded/printed by S. Steel Corp., and proved quite satisfactory. University of Washington (University of Washington) pursuant to License Agreement. No further reproduct 132 STRESS CORROSION TESTING under identical conditions, so that the damage resulting from general corrosion may be evaluated. Sheet specimens have also been tested as bent beams. Figure 3 shows a four-point loaded specimen of a salt-covered titanium alloy that had failed after exposure in air at 750 F for 64 days [6]. The stress in the outer fiber was computed from the deflection which was measured by the dial gage device shown in Fig. 4. The loading device also shown in Fig. 4 is a variation of the jig used by many investigators. Over a short length FIG. 4—Device for measuring deflection of bent-beam specimens in a 2-in. gage length and jigs for stressing bent-beam specimens. Knife edges rather than points are used in current models to avoid scratching of specimens. Upper jig designed so that specimens may be inserted into large mouth flasks. of arc, the specimen will approach circular curvature, and the strain can be computed by simple geometrical considerations. We have done little work, however, with bent-beam specimens except in a marine atmospheric exposure of stainless steels and the high-temperature exposure of saltcoated titanium alloy sheet material. Notched Specimens By changing the grip arrangements, the lever systems (Fig. 1) can, of course, be used for stressing specimens having circular cross sections. In the work on stress corrosion cracking of low-carbon steels in nitrate Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 solutions, conventional round tension specimens stressed to their yield Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further repr LOGAN ON TESTING TECHNIQUES 133 strengths failed in 300 to 1800 hr. Notched specimens, however, failed in 5 to 7 hr [7]. Sixty-degree notches were cut in Vs-m.-diameter rods, so that the specimen diameter at the root of the notch was 0.25 in. and the radius of the root of the notch was 0.005 in. This gave a notch concentration factor of approximately 4. To determine the loads to be used with these specimens, true stress-true strain curves5 were obtained on some of them, and the remainder were then loaded at a true stress just equivalent to that which produced a deviation from the linear portion of the true stress-true strain curve, that is, to a stress that would give a small amount FIG. 5—Upper chart shows sudden extension of low-carbon steel specimen indicating initiation of mechanical failure prevented from going to completion because of the work hardening of material at the crack tip. Change in potential results from exposure of bare material to the corrodent. Specimen was in corrodent at 100 C. of plastic deformation at the root of the notch. The extension of the specimen in the corrodent during an experiment could be plotted on a strip chart by using a differential transformer attached to the upper and lower shoulders of the specimen. The electrochemical solution potential of the specimens, in terms of a saturated calomel electrode, could be plotted on a.second channel of the recorder by connecting the electrode to the corrodent through an agar-agar bridge. Figure 5 shows a sudden extension of a low-carbon steel specimen that was accompanied by a change in potential in the active direction [7]. The extension of the specimen was believed to 6 Cross-sectional wererights calculated from diameter-gage readings at Copyright by ASTMareas Int'l (all reserved); Wed Dec 16 15:53:43 EST obtained 2015 specific loads. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further repro 134 STRESS CORROSION TESTING FIG. 6—Longitudinal section through root of low-carbon steel specimen. Note that a pair of cracks had developed (XlOO). FIG. 7—Longitudinal section through a notch in a stainless steel specimen that had been subjected to plastic deformation prior to exposure in the corrodent. Note that some of the cracks (out of the highly deformed region) were approximately parallel to the tensile axis (X50). be a mechanical fracture prevented from going to completion by work hardening and strain aging of the metal at the tip of the crack. This fracture exposed bare film-free metal to the corrodent, hence the change in potential. Cracks developed in pairs at the roots of notches in both low-carbon and stainless steelInt'l specimens. A longitudinal section through Copyright by ASTM (all rights reserved); Wed Dec 16 15:53:43 EST 2015the notch in Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions LOGAN ON TESTING TECHNIQUES 135 a low-carbon steel specimen shows that two cracks had developed, one on either side of the apex of the notch (Fig. 6). Figure 7 shows a longitudinal section through the root of a notch in a stainless steel specimen that had been pulled in tension to maximum load, unloaded, and subsequently exposed to a boiling magnesium chloride (MgCl2) solution while stressed at about one third the previously determined maximum load. Note that after cracks had penetrated through the severely cold-worked area they extended almost parallel to the nominal tensile axis [8]. Neither this behavior nor the fact that cracks develop in pairs at the roots of cylindrical FIG. 8—Stainless steel foil that had been used as an anode in the cathodic protection of stainless steel specimens in a boiling MgClz solution. Note the extensive stress corrosion cracking resulting from the applied anodic current and the residual stresses in the foil. notched specimens has been explained. A number of stress corrosion cracks penetrated to about the same depth in an annealed stainless steel specimen having a cylindrical reduced section and then split into pairs. One can postulate from this behavior that, when the stress concentrations resulting from the penetration of the individual cracks reached a certain value, further cracking was in pairs as was the case with the notched specimens. Plastic tape has been used effectively up to the boiling temperature of MgCl2 (154 C) to stop off parts of the specimen that might modify the electrochemical solution potential of that part of the specimen involved in the cracking reaction. Specifically, all of the area exposed to the solution exceptby the reduced of sheet round tension is Copyright ASTM Int'l (allsections rights reserved); WedorDec 16 15:53:43 ESTspecimens 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduc 136 STRESS CORROSION TESTING normally masked out. The connections to pull rods are also normally outside the solution. This avoids galvanic couples that might adversely affect the experiment. Hence, these do not have to be stopped off. The notched specimens are usually covered with tape up to the edges of the notch. In one instance, stop-off lacquer was used to cover a whole magnesium alloy specimen including the notched area. A circumferential cut was made through the lacquer at the root of the notch, and the specimen was exposed under stress in the corrodent. Its life was several times that of a specimen exposed subsequently with the whole notched area exposed. Exposing the whole notched area increased the cathodic area many times and hence the ratio of cathodic area to anodic area. A stainless steel foil is sometimes placed around the inside of the glass cell for use as an anode, if the effects of applied currents (anodic or cathodic) on a stainless steel specimen are to be evaluated. Because of the FIG. 9-—A hollow specimen used for investigating the resistance of materials to stress corrosion cracking at elevated temperatures and elevated or reduced pressures. Note that the specimen had failed completely, most probably after an earlier leak (marked by arrow) had developed. residual stresses in stainless steel foil, it would fail in a few hours by stress corrosion cracking when used to cathodically protect an austenitic stainless steel specimen in a boiling MgCl2 solution (Fig. 8). This points up the fact that anodic currents have also been used to accelerate stress corrosion cracking in materials such as low-carbon steel [7, 9], stainless steels [10], and aluminum alloys [77]. To avoid the difficulty of the foil anode failing in the magnesium chloride solution, a cell was constructed from a low-carbon ferritic steel tube. This cell served very well as an anode to stop stress corrosion cracking by cathodic protection in the austenitic stainless steel specimens. However, when the protection was removed, there was a marked difference using the different anodes. With the stainless steel foil anode, crack propagation would start again in a few minutes. With the ferritic steel anode, crack propagation did not start again for several hours. In the first case, a gelatinous magnesium hydroxide or hydrated oxide film formed on the specimen surface, but once protection was removed (current stopped) this was Copyright by ASTM rights MgCl reserved); Wed Dec 16 15:53:43 EST 2015 soon destroyed by Int'l the (all boiling 2. The coating formed on the speci- Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduct LOGAN ON TESTING TECHNIQUES 137 men protected by the ferritic anode could not be identified from its X-ray diffraction pattern. The Stainless Steel Specimen That Is an Autoclave A hollow specimen for elevated temperature work that could in itself be an autoclave was described at the 1962 annual meeting of ASTM [72]. The specimen is shown in Fig. 9, and a specimen, after it had been split so that the interior could be examined, in Fig. 10. Results of an investigation to determine the limits of chloride that could produce cracking at elevated temperatures in a Type 304 stainless steel were reported in Materials Research & Standards [13]. When an oxygen pressure was introduced above the solution, water containing 5-ppm chlo- FIG. 10—Hollow titanium alloy specimen sectioned for examination of the interior wall. Stress corrosion cracks are indicated by arrows. ride produced cracking at 575 F. If no oxygen was added, a water solution containing 20,000-ppm chloride at 575 F did not produce cracking. The Hollow Ti-8-1-1 Specimen This same type of specimen has been used in the study of the mechanism of salt stress corrosion cracking in titanium alloys. Some of the results obtained with the Ti-8Al-lMo-lV alloy were reported at the Seattle meeting of ASTM in November 1965 [6]. To coat the inner wall of the specimen with salt, a 10 per cent sodium chloride (NaCl) solution was poured into and then slowly pipetted out of the specimen maintained at a temperature >100 C. This left crystals of NaCl distributed, not necessarily uniformly, on the inner wall. The specimen was plugged, connected to a manifold through the small tube going through the plug, and placed inCopyright a creepbyfurnace. It was stressed in tension through pull rods connected ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 toDownloaded/printed the externally threaded ends. Prior to heating or loading the specimen by University of Washington (University of Washington) pursuant to License Agreement. No further reproductio 138 STRESS CORROSION TESTING cavity was evacuated to about 10~3 torr. Wet or dry oxygen, or wet or dry argon, was permitted to flow into the specimen cavity, and it was then heated to 750 F and stressed to 90 per cent of its yield strength at that temperature. If the specimen contained oxygen or water vapor, or if it had been heavily oxidized prior to the experiment, it would fail under these conditions of exposure. Figure 10 shows cracks extending into the interior wall of a specimen. FIG. 11—Fractured end of a specimen of high-strength 4340 steel showing area of brittle fracture resulting from hydrogen embrittlement that had initiated the failure. Evidence indicates that the brittle fracture had originated in the interior of the specimen wall (X12). Specimens for Investigation of Hydrogen Embrittlement Low-Carbon Steel Specimen Hollow specimens have also been machined from steel rod or steel tubing for the purpose of determining whether or not hydrogen was involved in the failure of the specimen exposed under tensile stress in a corroding medium. In one set of experiments, hollow-notched low-carbon steel specimens were simultaneously evacuated (to approximately 1 X 10~3 torr), exposed in a boiling aqueous 20 per cent NH4NO3 solution, and subjected to a tensile stress sufficient to produce a small amount of plastic deformation at the roots of the notches. When there was evidence that cracks were developing, the residual gases in the specimens were analyzed for hydrogen by mass spectrographic methods. There was no evidence that hydrogen had penetrated through the steel in these experiments [14]. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductio LOGAN ON TESTING TECHNIQUES 139 FIG. 12—Hydrogen bubbles emerging from a stress corrosion crack in a Type 304 stainless steel specimen exposed in a boiling 3Vi per cent NaCl-1 per cent NHtNOs aqueous solution. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductio 140 STRESS CORROSION TESTING High-Strength Steel Specimens In experiments having the same objective, specimens were machined from H-in.-outside diameter AISI 4130 and 4340 tubing. A reduced section about 1 l/z in. long was machined on each length of tubing leaving a wall thickness of approximately 0.035 in. The tubes were heat treated to give tensile strengths of about 180,000 psi for the 4130 and 255,000 psi or more for the 4340 steel. These specimens were simultaneously evacuated to about 10~3 torr, exposed to a boiling 3 l/z per cent aqueous NaCl solution, and stressed to approximately 75 per cent of their yield strengths. Once the specimen was stressed, the pumping system was cut off and the pressure buildup in the specimen noted. Periodically, the hydrogen concentration in the residual gas in the specimen was determined using standard hydrogen analysis techniques. The results have been reported in some detail elsewhere [75]. The fractured surface of a 4340 specimen is shown in Fig. 11. There is considerable evidence that the brittle fracture marked out by the arrows originated in the interior of the metal. Elongation measurements indicated that the failures occurred after an incubation period of several hours. Failure was usually complete within 30 sec after there were indications that it had been initiated. This behavior of the specimens and the fact that appreciable amounts of hydrogen were always measured in the residual gas prior to specimen failure indicate that failures in these steels under these conditions were the result of hydrogen embrittlement. Studies of Corrosion Products If one is to understand the mechanism of stress corrosion cracking, a knowledge of corrosion products is helpful. The fact that hydrogen diffused into the interior of the hollow high-strength steel specimens suggested that the failures resulted from hydrogen embrittlement. Figure 12 shows bubbles emerging from a developing stress corrosion crack in an austenitic stainless steel specimen exposed in NaCl, NH4NC>2 solution. Subsequent analysis of gases given off during the corrosion process indicated that these bubbles were hydrogen [16]. Chromium and nickel were found by chemical analyses in the liquid remaining in the interiors of the stainless steel autoclave specimens, which were removed before cracks penetrated entirely through the specimen walls, that is, before failure occurred [13]. Microspectroscopic analyses of the solid corrosion products associated with a crack in an austenitic stainless steel specimen indicated that iron, chromium, and nickel were present in approximately the same proportions as in the steel. This indicated that cracking was not due to the selective attack of the corrodent on a particular component of the steel. A gas sampling tube was included in the vacuum system connected to Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 the hollow titaniumbyalloy specimens exposed to dry NaCl. Mass spectroDownloaded/printed University of Washington (University of Washington) pursuant to License Agreement. No furthe LOGAN ON TESTING TECHNIQUES 141 graphic analyses of samples of the gas from the interior of the specimen did not reveal the presence of chlorine. It had been reported as a corrosion product by another investigator. At least one as yet unidentified solid corrosion product was removed from the interior walls of these specimens. Summary 1. Techniques developed during 25 years of investigating the susceptibility of various alloys to stress corrosion cracking are described. Certain interesting data, some of which have not been previously published, are also reported. 2. If results obtained on exposure of different materials to the weather are to be compared, specimens should be exposed at the same time. 3. Damage due to general corrosion should be evaluated in all stress corrosion testing, using unstressed specimens exposed under identical conditions, and taken into account in assessing damage due to stress corrosion cracking. 4. Notched specimens having cylindrical cross sections and stressed so as to produce a small amount of plastic deformation at the root of the notch may fail after exposure periods two orders of magnitude smaller than those for unnotched cylindrical specimens. Cracks at the root of the notch characteristically develop in pairs above and below the apex of the notch. 5. Hollow specimens have been designed that can contain the corrodent under either high or reduced pressures of oxygen, air, or an inert gas and that can be subjected to tensile stresses at elevated temperatures. This type of specimen has been used to investigate the resistance of stainless steels and a titanium alloy to special corrosive conditions. 6. Specimens machined from heavy-wall, high-strength alloy steel tubing were used to determine that hydrogen was a factor in the delayed failure of this steel in a hot chloride solution. 7. In efforts to understand the mechanism of stress corrosion cracking, gaseous and solid corrosion products have been identified by mass spectrographic, spectrographic, X-ray diffraction, and wet chemical analyses. References [1] Logan, H. L., Hessing, H., and Francis, H. E., Journal of Research, National Bureau of Standards, Vol. 38, 1947, p. 465, RP1788. [2] Logan, H. L., and Hessing, H., Journal of Research, National Bureau of Standards, Vol. 41, 1948, p. 69, RP1905. [3] Logan, H. L., and Hessing, H., Journal of Research, National Bureau of Standards, Vol. 44, 1950, p. 233, RP2074. [4] Loose, W. S., and Barbian, H. A., ASTM-AIME Symposium on Stress Corrosion Cracking of Metals, 1944, American Society for Testing and Materials, Philadelphia, 1945, p. 283. [5] Logan, H. L., Journal of Research, National Bureau of Standards, Vol. 61, 1958, p. 503,RP2919. Copyright (all rights reserved); Wed of DecTitanium, 16 15:53:43 EST STP 2015397, [6] Logan,byH.ASTM L. et Int'l al Stress Corrosion Cracking ASTM Downloaded/printed by for Testing and Materials, Philadelphia, 1966. American Society University of Washington (University of Washington) pursuant to License Agreement. No further rep 142 STRESS CORROSION TESTING [7] Logan, H. L., Journal of Research, National Bureau of Standards, Vol. 66C, 1962, p. 347. [8] Logan, H. L., The Stress Corrosion of Metals, Wiley, New York, 1966. [9] Pearson, E. E., and Parkins, R. N., Welding Research, Vol. 3, 1949, p. 95R. [10] Hoar, T. P. and Hines, J. G-, Stress Corrosion Cracking and Embrittlement, Wiley, New York, 1956, p. 122. [11] Booth, F. F. and Godard, H. P., First International Congress on Metallic Corrosion, Butterworths, London, 1962, p. 703. [12] Logan, H. L., "A Specimen for Use in Investigating the Stress-Corrosion Cracking of Metals at Elevated Temperatures," Materials Research & Standards, Vol. 2, No. 2, Feb. 1962, p. 98. [13] Logan, H. L., McBee, M. J., and Romanoff, M., "Stress-Corrosion Cracking of Type 304 Stainless Steel at 455 to 615 F," Materials Research & Standards, Vol. 3, No. 8, Aug., 1963, p. 635. [14] Logan, H. L. and Yolken, H. T., Second International Congress on Metallic Corrosion, NACE, Houston, 1966, p. 109. [15] Logan, H. L. and Wehrung, J. M., Corrosion, Vol. 22,1966, p. 265. [16] Logan, H. L. and Sherman, R. J., Jr., Welding Journal, Supplement, Welding Research, Vol. 35, No. 8,1956, p. 389-S. DISCUSSION F. A. Smith* (written discussion)—I would appreciate your comments on laboratory techniques for stress corrosion cracking of materials conducted at temperatures up to 1600 F in a sodium (liquid metal) atmosphere. The need for such testing is related to any liquid metal fast reactor program. One can project that liquid metal fast reactor technology cannot progress from laboratory devices to industrial power plants until the stress corrosion properties of a variety of alloys in sodium is well understood. Therefore, there exists a need for the development of "standard techniques" for stress corrosion testing in the new environment, sodium. E. N. Pugh2 (written discussion)—It seems to me that the term stress corrosion cracking is a generic term and that several different mechanisms of failure are operative, depending on the system and on the environmental conditions. I have the impression that you regard the electrochemicalmechanical mechanism, which you describe as a generalized model, applicable to all systems. Do you consider the model to be generalized? R. W. Staehle3 (written discussion)—The reason for the potential rise associated with deformation is undoubtedly rupture of the passive film 1 2 Md. 3 Physicist, Argonne National Laboratory, Argonne, 111. Staff scientist, Research Institute for Advanced Studies, Martin Co., Baltimore, Assistant professor, Department of Metallurgical Engineering, Corrosion CenCopyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 ter, The Ohio State by University, Columbus, Ohio. Downloaded/printed University of Washington (University of Washington) pursuant to License Agreement. No further reproduct DISCUSSION ON TESTING TECHNIQUES 143 by mobile dislocations and subsequent exposure of active metal. This nonprotected metal then reacts rapidly as shown in Fig. 13 (left). However, this process of film rupture and subsequent dissolution also occurs in alloys which do not crack (that is, pure nickel in chloride) as shown in Fig. 13 (right). Since this process of film rupture and dissolution occurs in alloys which crack, what do you consider to be the critical process in determining inherent susceptibility to cracking? H. L. Logan (author)—Dr. Smith's question is one that I am sure has received much thought by many people. One could use a hollow specimen such as I have described to contain the sodium at temperature. The problem is how to contain the hot sodium when the specimen fails. Two ideas FIG. 13—Thin foils are shown which have been exposed to boiling MgCh and subsequently examined in the electron microscope. Light areas are regions of dissolution: (left) Fe-15Ni-20Cr-1.5Si (15 min); (right) pure nickel. come to mind: (1) Could you use as a heating bath an alloy that will combine with liquid sodium in a manner that can be controlled? (2) Would it be possible to surround the hollow specimen with a metal bellows (by welding it to the shoulders), determine its effect on the stress applied to the specimen, and depend upon it to contain the sodium after specimen failure? Both Dr. Pugh's and Dr. Staehle's questions pertain to theory which I discussed in the presentation but is not included in the paper. My ideas are best presented elsewhere.4 As is well known, stress corrosion cracking can be prevented or, once * Logan, by H.ASTM L., The Stress Corrosion of Metals, Wiley, New York, 1966, pp. 33Copyright Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 36. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduc 144 STRESS CORROSION TESTING started, stopped in many systems at least by the application of cathodic protection. Conversely, stress corrosion cracking can be accelerated by the application of anodic currents. This would indicate that the process is in part, at least, electrochemical with small anodes at the crack sites, the remaining area forming large cathodes. It is a reasonable supposition that the function of the stresses is to rupture a protective film (visible or invisible) on the metal surface, exposing film-free areas which are more chemically active than the filmed surfaces, and become the anodes. Electrochemical attack of these anodes can trigger successive mechanical fractures until failure occurs. I believe that this is the most satisfactory explanation of the phenomenon of aqueous stress corrosion cracking as it occurs in many systems and, therefore, can be considered a generalized model. Yes, I think that we can present a valid generalized model for stress corrosion cracking. There are two possible reasons for the fact that the example posed by Dr. Staehle does not crack. There may not be a sufficient difference in the electrochemical solution potential of the filmed and film-free surfaces of nickel in chlorides to propagate a crack by an electrochemical process. We found that Type 321 stainless steel did not crack in a boiling lithium chloride (LiCl) solution unless tank oxygen was bubbled through it. The tank oxygen had the effect of making the electrochemical solution potential of the metal about 90 mv more noble and, hence, increased the potential difference between the film covered and film-free alloy by 90 mv. The second possibility is that plastic deformation of nickel in the chloride solution may result in much larger ratio of film-free-to-film covered areas than in materials susceptible to stress corrosion cracking. We postulate that if stress corrosion cracks are to propagate by an electrochemical mechanism, we must have small anodes associated with very large cathodes so that the current density at the anodes will be large. Specifically, then, if an alloy is to fail by stress corrosion cracking, there must be sufficient difference in the electrochemical solution potential between the film-free and film covered surfaces to promote rapid electrochemical attack. Second, plastic deformation must be sufficiently nonhomogeneous so that there will be relatively small regions of high stress concentration. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprodu W. D. Sylwestrowicz1 A Proposed Mechanism for the Stress Corrosion Fracture of a Copper-Beryllium Alloy REFERENCE: W. D. Sylwestrowicz, "A Proposed Mechanism for, the Stress Corrosion Fracture of a Copper-Beryllium Alloy," Stress Corrosion Testing, ASTM STP 425, Am. Soc. Testing Mats., 1967, p. 145. ABSTRACT: Polycrystalline specimens of a copper-beryllium alloy (1.8 per cent Be) were bent and exposed to an ammonia atmosphere. Two lots of polycrystalline material of similar chemical composition but of different grain size and distribution of segregates were investigated. Specimens were subjected to different values of plastic deformation and humidity conditions. Plastic deformation appears to be an essential factor in stress corrosion fracture. The importance of humidity conditions was emphasized by the results obtained. Both modes of fracture, transcrystalline and intercrystalline, were observed. It was found that the distribution of segregates determines which of these two modes of fracture will occur. A mechanism of stress corrosion fracture in a copper-beryllium alloy is proposed. KEY WORDS: stress corrosion, corrosion, copper-beryllium ammonia atmosphere, plastic deformation, fracture, humidity alloys, In essence the process of stress corrosion fracture is a propagation of a crack in a nonbrittle material with some features of brittle behavior. This process occurs when a stressed material is exposed to a specific corrosive environment. The value of stress under which the specimen fractures is usually much smaller than that for fracture in the usual static test. In the search for a model to describe stress corrosion fracture, a number of mechanisms have been suggested. They are based on a large volume of experimental data and can be divided into two general groups: electrochemical theories and mechanical theories. A comprehensive review of this field was given recently by Parkins [I].2 An attempt is made here to formulate a mechanism based on experiments with a copper-beryllium 1 Member of technical staff, Bell Telephone Laboratories, Inc., Murray Hill, N. J. The italic numbers in brackets refer to the list of references appended to this paper. by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Copyright 2 Downloaded/printed by 145 University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorize 146 STRESS CORROSION TESTING alloy, which takes into account both electrochemical and mechanical effects. The proposed model requires that the segregates form thin layers around the pileups of dislocations inside the grains or around the dislocations piled against the grain boundaries. The thickness of these layers is a critical factor. By an electrochemical process, the material is removed from these layers, forming narrow cracks with large stress concentrations at their tips. As a result of the diffusion of the solute and removal of the material, these layers become locally brittle, and with a large stress con- FIG. 1—Etched surface of a specimen from the first lot of material. Specimen was neither strained nor exposed to the corrosive atmosphere. Very small precipitates, probably of /3 phase, are visible (XlOOO). centration at the tip of the crack, cleavage occurs. But as the bulk material is not brittle, the propagation of the crack stops after proceeding a short distance with the formation of a number of fresh dislocations in front of the crack. Then the process repeats. The result is a discontinuous propagation of the crack. Material and Experiments The experiments described in this work were made on pressure cast copper-beryllium alloy of the following composition: 97.7 per cent copper, 1.8 per cent beryllium, and 0.3 per cent cobalt. Spectrochemical Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 analysis showed:bycobalt, magnesium, and iron from 0.01 to 0.3 per cent; Downloaded/printed University of Washington (University of Washington) pursuant to License Agreement. No further reproductions a FIG. 2—Etched surface of a specimen from the second lot of material. Specimen was neither strained nor exposed to the corrosive atmosphere. Large precipitates, probably of p phase, are present (XlOOO). FIG. 3—Etched surface of a specimen from the second lot of material. Specimen was neither strained nor exposed to the corrosive atmosphere. Large precipitates, probably of /3 phase, are present at the grain boundary, also broad segregates are shown along grainInt'l boundary (XlOOO). Wed Dec 16 15:53:43 EST 2015 Copyright by ASTM (all rights reserved); Downloaded/printed by 147 University of Washington (University of Washington) pursuant to License Agreement. No further reproductions author 148 STRESS CORROSION TESTING aluminum, magnesium, and silicon <0.03 per cent; silver < 0.005 per cent; copper, nickel, and tin < 0.001 per cent. The alloy was solution annealed at 800 C, quenched in water and aged at 355 C for 3 hr. From this material bars were machined 2 in. in length and % 6 by % 6 in. in cross section. The matrix of these polycrystalline bars consisted of the f ace-centered-cubic a phase in which was dispersed a small amount of the body-centered-cubic /? and y phases and cobalt beryllide particles. Two lots of material were used. Their chemical compositions were nearly identical, but the grain sizes and distributions of the precipitates and impurities differed. The grain diameter of the first lot was approximately 0.07 mm and of the second 0.3 mm. In the first lot, segregates were dispersed evenly through the grain with a comparatively small amount of segregation (probably y phase) at the grain boundaries (Fig. 1). In the FIG. 4—An oblique view of the bending jig. second lot, the /? phase formed large precipitates inside the grains (Fig. 2) and y-phase segregates at the grain boundaries (Fig. 3). Specimens from the first lot were used in experiments I and II; specimens from the second lot in experiments IV and V. The specimens after cutting from the bulk material were mechanically polished and then electropolished in 80 per cent phosphoric acid. Specimens were subjected to three-point bending in a specially designed jig (Fig. 4). By sliding the wedges one in relation to the other, the specimens were bent to the desired deflection. The deflection was measured with a microscope. With a known deflection, as long as a specimen is deformed elastically, maximum tensile stress can be easily computed from the measured deflection by the equation Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by where E is the elastic modulus, V is the deflection at the center of the University of Washington (University of Washington) pursuant to License Agreement. No further reproduction TABLE 1—Experimental data. Specimen No. 1-3 II-l II-2 H-3 H-4 IV4 IV-2 IV-3 V-l V-2 V-3 Lot 1 1 1 1 .1 2 2 2 2 2 2 ( Pf "max,6 kg/cm" 0.01 0.61 0.09 0.07 0.01 0.74 0.38 0.10 0.76 0.38 0.10 4.850 11.050 9.400 9.100 6.570 11.200 10.650 9.550 11.200 10.650 9.550 Water Pressure, mm Hg Ammonia Pressure, mm Hg 20 20 20 20 25 25 25 4.5 4.5 4.5 720 720 720 720 550 550 550 550 550 550 Crack 0.2 mm Long Crack 1.0 mm Long dl/dt,d Crack 3.0 mm Long dl/dt,d c Tex, hr dl/dt,d mm/hr 3 0.08 10 0.10 20 0.15 36 40 0.01 100 425 1.1 1.7 8.0 0.015 0.015 2.0 0.6 0.24 240 480 2.0 3.5 15.0 0.025 0.07 1« 1< 1 >60 >60 >60 c ,,_ T Tex nr ' a epi = plastic strain. omax = maximum tensile stress. Tex = time of exposure. d dl/dt = rate of crack propagation. 6 e Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized. mm/hr Tex ,c hr mm/hr 3.5 1.5 0.7 150 STRESS CORROSION TESTING specimen, c0 is the half-height, and / is the half-length of the specimen. For three-point bending, the value of the maximum tensile stress is in the outside layer in the middle of the specimen. The stress decreases toward and is zero at the neutral plane of the specimen. When the maximum stress is larger than the yield point, the calculation is more complex. For comparatively small values of the plastic strains, only a very thin outside layer of material near the central load point will be plastically deformed. The bulk of the specimen is deformed elastically. It is assumed, therefore, that for the computation of the total strain, the curvature of the deflected specimen is the same as for the specimen deformed elastically only. With FIG. 5—A relation between the length of the crack and the time exposure is shown for Specimen (1-3). this assumption and the measured deflections, total strain can be evaluated, and with the use of stress-strain curve for the material, the value of plastic strain in the outside layer can be determined. The values of strain for the specimens used in these experiments are given in Table 1. In the first series of tests, specimens were placed in a desiccator above a crystallizing dish containing ammonium hydroxide. After about 30 min of exposure, a black film developed on the surface of the specimen. To observe the propagation of cracks, specimens were taken periodically from the desiccator, cleaned in 10 per cent hydrochloric acid for a few seconds to remove corrosion film, and examined under a microscope. The structure of this black film was studied by an X-ray diffraction method. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 2015 hydroxide. Recorded reflections indicated a structure related toEST cupric Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduct SYLWESTROWICZ ON FRACTURE OF A COPPER-BERYLLIUM ALLOY 151 This was confirmed by examination with an electron-diffraction technique. Even after a very short exposure (20 min), observed lines corresponded to the cupric hydroxide structure with no indications of the presence of copper oxides. To check if the process of stress corrosion is sensitive to the humidity of the environment Specimen (I-3)3 was placed in a desiccator above a crystallizing dish containing ammonium hydroxide. After 60 hr of exposure, ammonium chloride was added to the dish to lower the vapor pressure of water in the desiccator with the result that a pronounced decrease in the rate of crack propagation occurred (Fig. 5). FIG. 6—The surface of Specimen (11-3) after 3 hr of exposure. Trans- and intercrystalline cracks are visible (XlOOO). In the second series of tests, Specimens (II-1, 2, 3, 4) were placed in an evacuated container (pressure below 0.1-mm mercury) at 25 C to which water vapor at 20-mm mercury pressure and ammonia at 720-mm mercury pressure were added. During the tests as some of the water reacted with the ammonia and copper, the water vapor pressure probably dropped below the initial value. The fourth series of tests (Specimens (IV-1, 2, 3)) was conducted at saturated vapor pressure of water (25-mm mercury) at 25 C with additional water supply in the system to keep this pressure 3 Specimen designation: first figure refers to the experiment and second to the specimen, for example, Specimen (1-3) refers to 15:53:43 the thirdEST specimen from number one Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 2015 experiment. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized. 152 STRESS CORROSION TESTING constant through the test. The ammonia pressure was 550-mm mercury. At the beginning of the tests in the fifth series (Specimens (V-l, 2, 3)), the water vapor pressure was 4.5-mm mercury at a test temperature 25 C and ammonia pressure of 550-mm mercury. After 60 hr of exposure even in the most prestrained Specimen (V-l), cracks were not observed, and the test was then discontinued. Observations and Discussion Initiation and Propagation of a Crack In these series of tests, a number of observations were made related to the initiation and propagation of the crack. As it was difficult at a FIG. 7—An electron microscopic replica taken from the surface of Specimen (1-3) after 3 hr of exposure (X20.000). very early stage to distinguish between an etched slip plane or an etched grain boundary and a genuine crack, the time when a crack obtained the length of 0.2 mm was considered as the time of initiation of the crack. At this size a crack was already well defined. The main observations resulting from these tests are summarized in Table 1. In the process of stress corrosion fracture, two stages will be distinguished: first the initiation of a crack, then its propagation. Prior to the appearance of initial cracks, an appreciable amount of etching takes place along the grain boundaries and along the intersection with the free surface of crystallographic planes inside of the grains. The initial cracks were either inter- or transcrystalline (Fig. 6). In the first lot of Copyright by ASTM (all rights reserved); Wed Dec 15:53:43 EST 2015 mode; material, cracksInt'l propagate predominantly in16the transcrystalline Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprod SYLWESTROWICZ ON FRACTURE OF A COPPER-BERYLLIUM ALLOY 153 FIG. 8—A crack in the Specimen (11-4) after 480 hr of exposure. FIG. 9—The surface of Specimen (II-3) after 3 hr of exposure (XlOOO). in the second lot, in an intercrystalline one. Transcrystalline cracks usually develop in the middle of the grain and then spread toward the Copyright by ASTM Int'l (all rights reserved); Wedcracks, Dec 16 15:53:43 EST form 2015 in crystalgrain boundary. These transcrystalline as a rule, Downloaded/printed by lographic planes nearly perpendicular to the axis of bending, in agree- University of Washington (University of Washington) pursuant to License Agreement. No further reproduction 154 STRESS CORROSION TESTING ment with the finding of Tromans and Nutting [2]. They found that in face-centered-cubic alloys transcrystalline cracks are in {111} planes. It is difficult to measure the radius of curvature of a crack at its tip, but a rough estimate can be made judging from the photograph of a surface replica in the electron microscope (Fig. 7). This radius seems to be smaller than 10~5 cm. A number of initial cracks appear along the central part of the stressed edge of the specimen. At this stage, the length of the crack does not depend on the position of the crack. This might be the result of local stress concentrations, crystallographic orientation of the grains, and grain boundary conditions. FIG. 10—The surface of Specimen (11-1) after 3 hr of exposure (XlOOO). The following observations were made on the propagation of transcrystalline cracks. Initially, cracks appear in the area stressed in tension. They are comparatively shallow (approximately 10 to 20/x), and the general direction of propagation is hi the plane perpendicular to the bending axis of the specimen (Fig. 8). A transcrystalline crack conveniently oriented to the bending axis of the specimen extends in its crystallographic plane. When the crack reaches a grain boundary, it passes into the adjacent grain again following crystallographic planes in that grain (Fig. 9). If the slip planes in the adjacent grain are unsuitably oriented to the propagation direction of the crack, while the grain boundary is Copyright ASTM Int'l of (allthe rightspropagating reserved); Wedcrack, Dec 16 the 15:53:43 EST 2015 close tobythe plane transcrystalline crack Downloaded/printed by changes into an mtercrystalline form until it encounters a grain with University of Washington (University of Washington) pursuant to License Agreement. No further reproduct SYLWESTROWICZ ON FRACTURE OF A COPPER-BERYLLIUM ALLOY 155 conveniently oriented slip planes. While the overall direction of spreading of a crack is well defined, locally the crack may deviate quite appreciably from this direction (Figs. 8 and 10). Once a crack spreads for some distance on one of the side faces of the specimen, its propagation slows down or stops until the crack spreads across the front face to the opposite side and penetrates through the whole depth of the specimen. Then it resumes its normal rate of propagation. This in itself results in changes of velocity of propagation of a crack measured at the surface of the specimen. If two cracks are formed close together, the stress field between cracks will be relieved and the crack in the less favorable condition will stop growing (Fig. 11). FIG. 11—The surface of Specimen (H-l) after 3 hr of exposure (left) and 7 hr of exposure (right) (X200). Reduced on the photograph by 2. Another factor contributing to changes in the rate of propagation of a crack is related to the proposed mechanism of spreading of a crack in a nonbrittle material. This mechanism will be discussed in what follows. It is not surprising, therefore, that a marked variation in the rate of propagation of the cracks is observed experimentally. Discontinuous motion of a crack was previously reported by Gilbert and Hadden [3] and Edeleanu and Forty [4]. Gilbert and Hadden attribute the changes in the velocity propagation of a crack to the sucessive fracturing and restoring of the surface film; Edeleanu and Forty to the presence of deformation bands in the specimens. Threshold Stress Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 It was confirmed in this study, as has been reported by many investiDownloaded/printed by gators, a threshold stressofisWashington) required pursuant to initiate a crack. In allNo tests, Universitythat of Washington (University to License Agreement. further reproductions 156 STRESS CORROSION TESTING cracks were restricted to the central part of the specimens in the regions of high stress. In the areas close to the cylindrical supports where the value of stress is small (zero at the supports), no cracks were observed. Also, the existence of a threshold stress for the propagation of a crack can be demonstrated in the following way. The value of the maximum tensile stress is given by Eq 1. If a stress concentration at the tip of a crack is taken into consideration the maximum stress is where L is the length of the crack; P is the radius at the tip of crack; 2c0 and 2c are initial and instantaneous heights of the specimen; and other FIG. 12—The time of exposure necessary to produce a crack 0.2 mm long as a function of plastic prestrain or stress in the second series of experiments. symbols have the same meanings as in Eq 1. With increasing length of the crack, c decreases; therefore, <rmax decreases also to the zero value when the crack approaches opposite face (c = 0). Assuming that a minimum value of stress is required to propagate a crack in the conditions described here, complete fracture should not occur. A crack should stop before reaching the opposite face unless it develops so large a velocity that it propagates by its momentum through the area of small stresses. In agreement with this prediction, complete fracture occurred in only one test, and this was in the specimen of the highest prestrain. In all other tests, cracks stopped a fraction of a millimeter before complete penetration of the specimen. Effect of Plastic Prestrain Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 The dependence of the time of initiation of a crack on the amount of Downloaded/printed by prior plastic deformation is shown hi Fig. pursuant 12. This relation indicates University of Washington (University of Washington) to License Agreement. No further reprodu SYLWESTROWICZ ON FRACTURE OF A COPPER-BERYLLIUM ALLOY 157 that the threshold stress may correspond to the value of stress necessary for the initiation of plastic deformation in the grains most favorably oriented to the applied stress field, as was suggested by Bakish and Robertson [5]. The rate of crack propagation is affected also by prestrain in the specimen. The values of time of exposure and of velocity of crack propagation are shown in Table 1. An exponential dependence was found for the velocity of crack propagation and plastic prestrain, as can be seen in Fig. 13 where values of velocity against plastic deformation plotted on semilogarithmic scale presents a straight line relation. No simple relation could be established between the time of exposure to fracture or velocity and value of prestress. This indicates that the main effect of prestraining a specimen is in the amount of plastic deformation produced FIG. 13—The velocity of crack propagation as a function of plastic prestrain for a crack 1 mm long (curve a) and 3 mm long (curve b) in the fourth series of experiments. and not in the value of stress. This does not imply that stress itself is not a factor in the process of stress corrosion. The existence of a threshold stress clearly indicates that imposed stress is a factor. The explanation of this seemingly inconsistent behavior may be in the fact that, after reaching the yield stress, increasing prestrain produces large amounts of plastic deformation with comparatively small increase of stress. In summary, there are three factors affecting the stress field around a crack: the shape of the crack, which is related to the amount of plastic deformation as will be discussed in what follows; the orientation of the crack in relation to the applied stress; and the value of applied stress. Among these, the shape of the crack is the predominant factor. Copyright ASTM Int'l (all and rights Segregates reserved); Wed Dec 16 15:53:43 EST 2015 Effect ofby Dislocations Downloaded/printed by Tromans and Nutting [6] associated transgranular cracking with deUniversity of Washington (University of Washington) pursuant to License Agreement. No further reprodu 158 STRESS CORROSION TESTING formations larger than 1.5 per cent (in 70:30 a-brass alloy) and intergranular cracking with smaller deformations. This dependence of the type of cracking on the amount of prestrain has not been confirmed for the copper-beryllium alloy investigated in this study. Specimens plastically prestrained by only 0.01 per cent developed transcrystalline cracks, and specimens of 0.7 per cent plastic prestrain showed some intercrystalline cracking. Therefore, it is concluded that the amount of prestrain is not related to the mode of formation and propagation of a crack. It is assumed here that the initiation of a crack is a result of removal FIG. 14—Etched surface of a specimen from the first lot of material. Specimen was neither strained nor exposed to the corrosive atmosphere. Very small precipitates with a tendence to align in the crystallographic directions are visible (XlOOO). of segregated material electrochemically and not by the mechanism proposed by Stroh [7], in which he postulates that dislocations in a pileup mechanically develop a crack. A dislocation acts normally as a sink for segregates. Segregates diffusing toward piled-up dislocations in their slip plane can form a thin plate around them. Through the removal of segregated material by an electrochemical process, a narrow crack is formed. The diffusion coefficients, temperature of test, amount of segregates, and disposition of dislocations will be decisive factors in determining the layer thickness of the segregates. With easy cross slip, tangles of dislocations will be formed. This results in broad cracks or rounded pits Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 and comparatively small stress concentrations around them. Likewise, a Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprod SYLWESTROWICZ ON FRACTURE OF A COPPER-BERYLLIUM ALLOY 159 prolonged heat treatment or large content of solute will result in broad segregates and produce cracks with small stress concentrations. This can explain why the susceptibility of some materials to stress corrosion fracture changes [8] when conditions of heat treatment are changed. With increasing prestrain, the number of dislocations increases, pileups become longer, and this leads to a larger number of long segregation plates. With a larger number of plates the chance of formation of a crack increases and the propagation of the crack is faster. In the copper-beryllium alloy studied, formation of transgranular cracks was restricted to the interior of the grain away from grain boundaries where large pileups of dislocation would be expected. This can be explained as follows: grain boundaries act as sinks for the solute, therefore, in their vicinity there might be insufficient solute to form a continuous layer in the slip planes and to form a crack. In some cases when a large amount of solute is present in the material, a partial migration into the grain boundaries might create conditions favorable for formation of thin cracks in the vicinity of boundaries. This interpretation of the effect of the distribution of segregates in the material on the form of fracture in the stress corrosion processes is supported by the results of the experiments described. Specimens from the first lot, with a very small amount of segregation at the grain boundaries, have segregations dispersed uniformly throughout the grains with a tendence to align in the crystallographic directions (Fig. 14). In these specimens, fracture is mostly transcrystalline. In the specimens from the second lot with an appreciable amount of segregates at the grain boundaries (Fig. 3), fracture is intercrystalline. Recently, Tromans and Nutting concluded [2] that change of mode of fracture is related to electrochemical potential, originating from a segregation of solute elements on dislocations and not to the value of stacking fault energy as they considered previously [6]. They believe that segregation is an essential condition for stress corrosion fracture. In the present proposal, it is also accepted that segregation of solute atoms is a fundamental condition for the process but with an important difference. Tromans and Nutting assumed that propagation of a crack occurs through "dissolution of dislocations," followed by the plastic shearing of material between dislocations. In the present work, it is postulated that material becomes locally embrittled as a result of diffusion of solute atoms toward pileups of dislocations and removal of material from this region. With applied stress and small enough radius of the crack, a large stress develops at the tip of the crack and cleavage occurs. It is implied in the presented model that at room temperature diffusion of solute atoms toward dislocations is possible at least through the distance of a few interatomic spacings. Recent investigations of the effect Copyright by ASTM Int'l rights reserved); fracture Wed Dec 16 EST 2015 show that of temperature on (all stress corrosion (to15:53:43 be published) Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduct 160 STRESS CORROSION TESTING by lowering temperature to — 20 C the process was stopped while general corrosion still occurred. Lowering temperature by 50 C does not embrittle tested material but strongly affects diffusion to which is attributed the slowing of the process of stress corrosion cracking. Effect of Humidity in a Corrosive Atmosphere The effectiveness of an electrochemical cell, among other factors, depends on the presence of a conducting liquid phase at the surface of the specimen [9]. This can be obtained from the water vapor in the corrosive atmosphere providing that a hygroscopic film is formed on the surface of the specimen. The condition of high humidity in the corrosive environment and formation of the hygroscopic film on the surface of the specimen are, therefore, essential for the initiation of a crack in the stress corrosion process. The effect of humidity is clearly shown in the experiment with a specimen exposed to the corrosive atmosphere in the test chamber with ammonium hydroxide. By addition of ammonium chloride, which reduces the vapor pressure of water in the test chamber, a very pronounced decrease of the rate of propagation occurs (Fig. 5). Even more striking are the results of the IVth and Vth series of experiments (Table 1) with vapor pressure of water 25 and 4.5-mm mercury, respectively. After 1 hr of exposure of Specimen (IV-1) a crack of 0.8 mm length was observed in the specimen. After 2 hr of exposure, the crack reached 3 mm length. In Specimen (V-l) of the same 0.7 per cent prestrain, after 60 hr of exposure, there was no evidence of formation of cracks. Effect of Corrosive Film The effect of humidity on the process of cracking is related to the presence of a hygroscopic film on the surface of the specimen. This is shown in the experiment with Specimens (II-2, 3). These two specimens had similar prestrain and were tested simultaneously. The surface film of Specimen (II-2) was left untouched while that of Specimen (II-3) was removed periodically (nine times). This periodic removal of the corrosion film resulted in the increase of exposure time required to obtain a crack of 4 mm length from 50 to 300 hr. In the mechanism of stress corrosion fracture which depends on film rupture [10-15], it is postulated that a surface film forms a protective layer over the specimen. The applied stress cracks this film exposing fresh material susceptible to the corrosive attack. This mechanism is not supported by the observations in the experiments described here. These show that removal of the surface film delays the process of stress corrosion (Specimens (II-2, 3)). In the studied copper-beryllium specimens, the surface film consists Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 of copper hydroxide of orthorhombic structure. It is unlikely therefore Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further repr SYLWESTROWICZ ON FRACTURE OF A COPPER-BERYLLIUM ALLOY 161 that cracks which are in crystallographic orientation in face-centeredcubic matrix coincide with cracks in the film which has an orthorhombic structure. To investigate the effect of the corrosion film, Tromans and Nutting [2] studied a carbon replica of the stress corroded surface of a specimen of Cu-30Zn alloy. They found a large number of cracks on the surface film of the grains. This suggests that transcrystalline cracking should be present in the matrix material if these cracks result from the cracks in the surface film. In fact, only intercrystalline cracks were observed. FIG. 15—Debris of the surface of a "dome" which collapsed under the heat of the illumination light in a microscope (X200). The cupric hydroxide, observed in this study, is not formed by direct reaction between copper and water. The reactions leading to formation of cupric hydroxide are not known fully, but the presence of ammonia is essential. A specimen exposed for three months to an atmosphere saturated with water vapor but without ammonia has not developed a cupric hydroxide film. A complex reaction probably occurs between copper, ammonia, and water. The product of this reaction may be a cupro-ammonium complex formed in the liquid phase from which cupric hydroxide precipitates. In some tests after longer exposures, a thin crystalline film formed over Copyright Int'lhydroxide. (all rights reserved); Wed Decof 16 spherical 15:53:43 EST"domes" 2015 the layer byofASTM cupric A number appeared Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions aut 162 STRESS CORROSION TESTING on the surface of the specimens. These domes appeared to be filled with a liquid. Due to the heat of the light during microscopic examination the thin surface of the dome collapsed, and liquid evaporated leaving debris of the surface film (Fig. 15) and exposing deep pits which had formed under the domes. The extrusions observed by some authors [2,16] on the surfaces of stress corroded specimens may in fact be this debris. It has also been suggested that wedging action of corrosion products is responsible for the propagation of a crack [76,77]. The fracture delay, caused by the removal of corrosion layers, could be considered as a confirmation of this mechanism. However, the observation that below a level of applied stress cracks do not propagate indicates that a wedging mechanism cannot be considered as a principal cause of crack spreading but only as a minor factor in this process. The results of Forty and Humble [72] showing that in a brass there is no increase of volume of the corroded material put in doubt the existence of a wedging action. The conclusion can be made that the importance of the surface film in the process of the stress corrosion fracture, at least in copper-beryllium alloy, lies in the hygroscopic properties of the film formed and not in the initiation of a crack in the matrix by a crack first developed in the surface film. Conclusions To summarize, the following model of a stress corrosion mechanism is proposed for copper-beryllium alloy. This model is probably applicable to a wider range of materials susceptible to stress corrosion fracture. A material under tensile stress is exposed to a corrosive atmosphere. As a result of the applied stress, a number of fresh dislocations are generated. These dislocations located in the slip planes act as a sink for solute atoms, which preferentially segregate around dislocations. These segregates will form thin layers in slip planes or at grain boundaries. At the same time as a result of corrosion, a film is formed on the surface of the specimen. This film is hygroscopic and absorbs water from the atmosphere to form a conducting layer. In the presence of the liquid, an electrochemical cell is formed between a layer of segregates and the matrix of the specimen. The effect of this electrochemical process will be the removal of the material from the zones around the slip planes or grain boundaries. With these two processes working simultaneously, material becomes brittle in these areas, and, in the most favorable location, a crack can be initiated on the surface of the specimen. Transcrystalline cracks lie in crystallographic planes (slip planes). At the tip of a crack which is roughly perpendicular to the direction of applied stress, a large stress concentration is present. In this embrittled area, conditions for cleavage propagation are produced and a crack starts to spread in the slip plane. But, as the bulk material is not brittle, the Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 velocity of crack will decrease as the crack proceeds into the nonbrittle Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprod SYLWESTROWICZ ON FRACTURE OF A COPPER-BERYLLIUM ALLOY 163 volume, new dislocations in front of the propagating crack will be formed, and the crack will stop. Then the process repeats. In this way a crack will spread through the specimen in a jerky motion. The essential parts of this model are: 1. Segregation of solute atoms to the dislocations in slip planes. The segregated material differs from the matrix material of the specimen in electrochemical properties. Also, as the result of diffusion of solute locally, the material becomes more brittle. 2. Formation of an electrochemical cell. This leads to removal of material from the areas already rich in solute atoms and to still more embrittlement of material. For the operation of the electrochemical process, the presence of a liquid phase is necessary. This liquid phase is produced as a result of the action of the corrosive atmosphere. 3. Formation of a crack with a small enough radius at its tip to produce stress concentration large enough to initiate propagation of a crack in locally embrittled material. A cknowledgments The author would like especially to thank H. C. Theuerer for the suggestions and helpful discussions on the chemical problems connected with this work; W. C. Ellis, T. D. Schlabach, and U. B. Thomas for critical reading of the manuscript; Mrs. A. M. Hunt for performing the electro-diffraction determination of the corrosion film structure; and Miss S. E. Koonce for taking electron microscopic photographs. References [/] Parkins, R. N., "Stress-Corrosion Cracking," Metallurigical Reviews, Vol. 9, 1964, p. 201. [2] Tromans, D. and Nutting, J., "Stress-Corrosion Cracking of Face-CenteredCubic Alloys," Corrosion, Vol. 21, 1965, p. 143. [3] Gilbert, P. T. and Hadden, S. E., "A Theory of the Mechanism of Stress-Corrosion in Aluminium-7% Magnesium Alloy," Journal of the Institute of Metals, Vol. 77, 1950, p. 237. [4] Edeleanu, C. and Forty, A. J., "Some Observations on the Stress-Corrosion Cracking of a-Brass and Similar Alloys," Philosophical Magazine, Vol. 5, 1960, p. 1029. [5] Bakish, R. and Robertson, W. D., "Structure-Dependent Chemical Reaction and Nucleation of Fracture in Cu3 Au Single Crystals," Acta Metallurgica, Vol. 4, 1956, p. 342. [6] Tromans, D. and Nutting, J., "Electron Microscope Studies of Stress Corrosion Cracking," Fracture of Solids, Drucker, D. C. and Oilman, J. J., eds., Interscience, New York, 1962, p. 637. [7] Stroh, A. N., "Crack Nucleation in Body-Centered-Cubic Metals," Fracture, Averbach, B. L. et al, eds., Wiley, New York, 1959, p. 117. [8] Uhlig, H. H. and Sava, J. P., "Origin of Delay Time in Stress Corrosion Cracking of Austenitic Stainless Steel," Proceedings, Second International Congress on Metallic Corrosion, 1963, National Association of Corrosion Engineers, 1966. [9] Uhlig, H., Corrosion and Corrosion Control, New York, 1963, p. 146. Copyright by H. ASTM Int'l (all rights reserved); Wed Dec Wiley, 16 15:53:43 EST 2015 [10] Mears, R. B., Brown, R. H., and Dix, E. H., Jr., "A Generalized Theory of Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprod 164 [11] [12] [13] [14] [15] [16] [77] STRESS CORROSION TESTING Stress Corrosion of Alloys," ASTM-AIME, Symposium on Stress-Corrosion Cracking of Metals, 1944, American Society for Testing and Materials, Philadelphia, 1945, p. 323. Logan, H. L., "Film-Rupture Mechanism of Stress Corrosion," Journal of Research of the National Bureau of Standards, Vol. 48, 1952, p. 99. Forty, A. J. and Humble, P., "The Influence of Surface Tarnish on the StressCorrosion of a-Brass," Philosophical Magazine, Vol. 8,1963, p. 247. McEvily, A. J., Jr., and Bond, A. P., "On the Initiation and Growth of Stress Corrosion Cracks in Tarnished Brass," Journal of Electrochemical Society, Vol. 112, 1965, p. 131. Forty, A. J., "Surface Films and Stress Corrosion Cracking," Proceedings, Conference on the Environment-Sensitive Mechanical Behavior of Materials, 1965, to be published. McEvily, A. J. Jr., and Bond, A. P., "On Film Rupture and Stress Corrosion Cracking," Proceedings, Conference on the Environment-Sensitive Mechanical Behavior of Materials, 1965, to be published. Nielsen, N. A., 'The Role of Corrosion Products in Crack Propagation in Austenitic Stainless Steel Electron Microscopic Studies," Physical Metallurgy of Stress-Corrosion Fracture, Rhodin, T. N., ed., Interscience, New York, 1959, p. 121. Pickering, H. W., Beck, F. H., and Fontana, M. G., "Wedging Action of Solid Corrosion Product During Stress Corrosion of Austenitic Stainless Steels," Corrosion, Vol. 18, 1962, p. 230. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductio H. Rosenthal1 and H. R. Pritchard1 A Quantitative Stress Corrosion Test for AI-Zn-Mg Alloy Plate REFERENCE: H. Rosenthal and H. R. Pritchard, "A Quantitative Stress Corrosion Test for Al-Zn-Mg Alloy Plate," Stress Corrosion Testing, ASTM STP 425, Am. Soc. Testing Mats., 1967, p. 165. ABSTRACT: The stress corrosion susceptibility of Al-4.25Zn-2.9Mg plate was determined in the short-transverse direction. C-ring type specimens were exposed to alternate immersion in a NaCl solution. It was shown that laboratory temperature and humidity must be controlled to obtain reproducible results. Time-to-failure results for a given lot of material have a normal distribution when failure times are converted to logarithms. Tests were also made in a marine environment and in an industrial environment. These outdoor results for various lots follow the same ranking for susceptibility as determined by the alternate-immersion tests. KEY WORDS: aluminum alloys, stress corrosion, corrosion, marine exposure, industrial exposure It is well known that the 7000 series alloys (Al-Zn-Mg-Cu) are susceptible to stress corrosion when heat treated to maximum hardness. The susceptibility is mainly confined to the short-transverse direction in rolled, forged, or extruded sections. A number of specifications require testing such material by subjecting specimens to intermittent immersion in a sodium chloride (NaCl) solution. Recently, there has been interest in using a copper-free weldable grade of an Al-Zn-Mg type of alloy plate for army vehicles. The data in this report include some of the information on stress corrosion characteristics of this alloy developed by Frankford Arsenal in cooperation with the aluminum industry. The objectives of the work were: (1) to define the laboratory test conditions such that the test results would be quantitatively reproducible and (2) to determine if there was a correlation between the laboratory tests and outdoor exposures. 1 Research advisor, and physical science technician, respectively, Frankford Arsenal, United States Army, Philadelphia, Pa. Mrs. Pritchard is a personal memCopyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 ber ASTM. Downloaded/printed by University of Washington (University of Washington)165 pursuant to License Agreement. No further reproductions authorized. 166 STRESS CORROSION TESTING Experimental Procedure Work described in this report covers alternate-immersion tests conducted under ambient conditions of laboratory temperature and relative humidity, as well as outdoor tests in marine and industrial environments. It also covers a later series of tests in which laboratory temperature and relative humidity were carefully controlled. Material Material consisted of aluminum plate conforming to Military Specification MIL-A-46063. Plate thickness ranged from SA to 3% in. Plates were in the solution-treated, quenched, and artificially aged temper. The chemical composition limits in weight per cent were as follows: zinc, 3.5 to 5.0; magnesium, 2.0 to 3.8; manganese, 0.10 to 0.70; chromium, 0.06 to 0.25; iron, 0.40 max; silicon, 0.30 max; zirconium, 0.20 max; copper, 0.10 max; titanium, 0.10 max; others, each 0.05 max, 0.15 total; aluminum, remainder. Minimum mechanical properties were as shown in Table 1. TABLE 1—Mechanical properties. Plate Thickness, in. Up to 1.500 inclusive Over 1.500 Tensile Strength, min, psi Yield Strength, 0.2% offset, min, psi Elongation, min, per cent 60 000 57 000 51 000 48 000 9 8 FIG. 1—Orientation of C-ring in relation to plate. Test Specimens C-ring type specimens were machined from the plate so that the centerline of the specimen, which is the area of maximum applied stress, represented the centerline of the plate. A sketch of the specimen orientation in relation to the rolling direction of the plate is shown in Fig. 1. Two sizes of C-rings were used, % and 11A in. diameter with wall thicknesses of 0.060 and 0.100 in., respectively. Three-quarter-inch rings were made from plate !J/4 in. and less in thickness, and lV4-in. rings from Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 plate over 1 x/4 in. Downloaded/printed bythick. All rings were % in. wide. University of Washington (University of Washington) pursuant to License Agreement. No further reprod ROSENTHAL AND PRITCHARD ON TEST FOR AL-ZN-MG ALLOY PLATE 167 Stress Corrosion C-Ring Specimen OD, in. t, in. Hole Dia (H) Hole Dia (H) for Bushing, w/o Bushing, Al Screw Size in. in. 0.750 0.060 0.257 0.203 10-32 1.250 0.100 0.257 0.203 10-32 Outside Finish - 63 FIG. 2—C-ring dimensions. FIG. 3—Two methods of preventing galvanic corrosion between screws and C-rings, (upper) Micarata bushing (ring partially sectioned) and flower) neoprene dip. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions 168 STRESS CORROSION TESTING Figure 2 shows the geometry of the specimens, with two alternate sizes of holes for the stressing screws. In early tests, a Micarta bushing was used to insulate the screw from the specimen to prevent galvanic corrosion. Later, no bushing was used and galvanic corrosion was prevented by dipping the bolt and nut area of the stressed specimen in liquid neoprene. Figure 3 shows two rings: one partly cut away to show the bushing and the other dipped in liquid neoprene. FIG. 4—Z correction factor used in formula by which C-ring stress is calculated. Stressing of Specimens Specimens were stressed by tightening a screw and nut to reduce the ring diameter. The following formula2 was used to determine the desired stress in terms of the properties and dimensions of the specimens. where: A = change of outside diameter required for desired stress, / = desired stress, psi, t = wall thickness, in., D = mean diameter (OD — i), in., 2 Sprowls, D. Int'l O.,(all inter-laboratory memorandum, Alcoa Research Laboratories, Copyright by ASTM rights reserved); Wed Dec 16 15:53:43 EST 2015 1954. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized ROSENTHAL AND PRITCHARD ON TEST FOR AL-ZN-MG ALLOY PLATE 169 E = modulus of elasticity, and Z = a constant (function of ring D/t, see Fig. 4). Specimen Preparation Specimens were degreased after machining. When no bushings were used, specimens were dipped in liquid neoprene to cover the bolt, nut, and surrounding specimen area and allowed to dry for 1 l/z hr. All specimens were wiped with acetone-soaked cotton after stressing and immediately put in test. FIG. 5—Schematic of alternate-immersion test equipment. Equipment A schematic drawing of the equipment is shown in Fig. 5 and a photograph in Fig. 6. Alternate immersion of the specimens is accomplished by movement of the salt solution rather than raising and lowering the specimens themselves. At the beginning of the immersion cycle, the solution is contained in the bottom tank. The cycle starts with the timer actuating the solenoid valve which causes air to enter the tank. The bleeder valve in the tank permits air to escape, but the air pressure rises, because the rate of leakage is much less than the rate at which the air is entering the tank. The rising pressure in the tank causes the liquid to flow into the tray above the tank. The second signal from the timer closes the solenoid valve and stops the flow of air into the tank. The pressure in the tank now reverts to atmospheric pressure because of the continued leakage from the bleeder valve. As the Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 pressure in the by tank becomes lower, the salt solution flows back into the Downloaded/printed University of Washington (University of Washington) pursuant to License Agreement. No further reproductions au FIG. Copyright by ASTM Int'l (all rights reserved); Wed Dec 6—Alternate-immersion 16 15:53:43 EST 2015 test equipment. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized. ROSENTHAL AND PRITCHARD ON TEST FOR AL-ZN-MG ALLOY PLATE 171 tank from the tray. The timer is so adjusted that the cycle for the specimens consists of 10 min immersion in the solution and 50 min drying time. The trays are tilted slightly to facilitate drainage and also to provide different depths of solution to cover test specimens of various thicknesses for the same period of time. Specimens in the trays were covered to a depth of approximately l/z in. Test Medium The test solution was 3.5 per cent (by weight) reagent grade NaCl and distilled water prepared weekly. The salinity of the solution was checked five days a week with a hydrometer, and distilled water was added to compensate for evaporation. No attempt was made to control pH, since it was 7.0 when the solution was made and 7.3 to 7.4 when the solution was discarded after a week of use. Outdoor Test Conditions Specimens prepared in the same manner as for alternate immersion were also exposed to marine and industrial outdoor environments. The marine site was located in southern New Jersey, approximately 500 ft from mean high tide. Specimens were supported on racks at an angle of 45 deg facing the ocean. The inland site was located on the roof of a Frankford Arsenal building (Philadelphia, Pa.). The atmosphere in the vicinity of Frankford Arsenal can be considered an industrial type atmosphere, since it is located in an area surrounded by heavy industry. Specimens here were also supported on racks at a 45 deg angle facing south. Examination for Cracks The presence of cracks was determined by visual inspection of the specimens. If doubt existed, visual examination was supplemented by examination with a binocular microscope at a magnification of X7. If doubt still remained, the area was noted, and subsequent examinations were directed to this area of the specimen to see if crack growth had occurred. The laboratory tests in alternate immersion were kept under observation except over the weekends. The observation period on weekdays was between the working hours of 7:30 a.m. and 4:30 p.m. The number of examinations carried out daily varied from two to four. Examination of specimens at the marine site was made weekly and at the industrial site every weekday. Tests Under Ambient Laboratory Conditions A series of alternate-immersion tests was made in which ambient lab- Copyright ASTM Int'l (alland rightsrelative reserved);humidity Wed Dec 16prevailed. 15:53:43 ESTDuring 2015 oratorybytemperature the winter Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions a 172 STRESS CORROSION TESTING TABLE 2—Results of tests in alternate immersion at different seasons of the year. Lot A B Stre ss . psi 21 21 39 39 000 000 000 000 Date Started 9 March 12 June 25 February 27 April Avg Temperature, degF Failure Time, days 82 36, 42, 56 4, 9, 10 6, 80, 80 7, 14, 14 87 83 85 FIG. 7—Time-to-failure at 35,000 psi of 14 lots of plate in three environments arranged in order of increasing resistance to alternate immersion. Tests started during summer months. heating months, the air temperature ranged from 77 to 84 F and the relative humidity from 30 to 50 per cent. In the summer, the air temperature varied from 77 to 101 F and the relative humidity from 30 to 90 per cent. The temperature of the test solution was from 4 to 7 deg less than the air temperature. Table 2 lists results of two lots of material tested in alternate immersion at different times of the year. These data show the effect on failure time of the temperatures prevailing during periods of winter heating and summer ambient conditions. Tests were run in triplicate, and individual failure times for specimens are shown in the table. It can be seen that failure time decreases with increasing temperature. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprod ROSENTHAL AND PRITCHARD ON TEST FOR AL-ZN-MG ALLOY PLATE 173 Comparative Summer Tests As was previously indicated, the variation of temperature and humidity was in a much narrower range in the winter than in the summer. It is therefore interesting to compare, in general, results of tests carried out in the summer months with those conducted during the winter. Figure 7 shows data obtained from tests started in the spring and summer, when laboratory ambient conditions showed their maximum variation. Data on stress versus time-to-failure were obtained on 14 lots of FIG. 8—Time-to-failure at 35,000 psi of 11 lots of plate in three environments arranged in order of increasing resistance to alternate immersion. Tests started during winter months. plate with at least 3 specimens per stress level. A curve of stress versus median time-to-failure was plotted for each lot. From these curves, a median time-to-failure at 35,000 psi was determined for each lot. These are arranged in Fig. 7 in order of increasing resistance to stress corrosion in the alternate-immersion test. Corresponding data developed simultaneously for the same lots in the marine and industrial environments are also shown. The ranking determined by the laboratory tests was considerably different from the ranking for the outdoor exposure tests (Fig. 7). Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduct 174 STRESS CORROSION TESTING Comparative Winter Tests Figure 8 shows data for eleven lots which were started during the winter heating months. These data are also arranged in order of increasing resistance to stress corrosion in the alternate-immersion test. The plots show good correlation between the alternate-immersion and the outdoor results except near the terminal part of the graph, where the median failure time in alternate immersion is in excess of twelve days. Data from Fig. 8 have been replotted in Fig. 9 in the form of per cent survival versus failure time. In this graph, the shape of the curve of the marine exposure is similar to that of the alternate immersion; in contrast, FIG. 9—Per cent survival versus time-to-failure at 35,000 psi in three environments. Specimens represent the 11 lots of plate on which tests were started during winter months. the industrial curve is nearly a straight line. Other investigators have noted that an industrial atmosphere is often extremely aggressive. Tests on Micarta Bushings In addition to the effects of temperature and relative humidity on reproducibility of results, it was found that the Micarta bushings used as insulators affected results in the alternate-immersion test, particularly at high stress levels. It was suspected that these bushings were deforming under stress and thus permitting a relaxation of stress in the specimens. An experiment was conducted to determine the extent of this effect on alternate-immersion test specimens and on two groups of control specimens. One control group was exposed to alternate immersion without bushings and the other group, with bushings, was exposed to air only. The air temperature was 80 ± 2 F and the solution temperature 76 ± 2 F. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Specimens were % in. diameter and were stressed at several levels. Diam- Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further rep ROSENTHAL AND PRITCHARD ON TEST FOR AL-ZN-MG ALLOY PLATE 175 eters were measured over a period of time to determine whether relaxation had occurred. It was found that the maximum change occurred after five days in the specimens with bushings exposed to alternate immersion and that the amount of relaxation increased with increasing applied stress. No change in diameter occurred in controls exposed to alternate immersion without bushings nor in controls exposed only to air at any stress level. Results for the alternate-immersion test are shown in Table 3. Apparently the Micarta absorbed moisture and became softened, permitting the specimens to expand and relax the applied stress. For this reason, the use of bushings was discontinued. Galvanic corrosion was thereafter avoided by dipping specimens in liquid neoprene to cover the screw, nut, and specimen contact areas. TABLE 3—Change in diameter of %-in. rings exposed to alternate immersion. Applied Stress, psi 20 000 30 000 40 000 No. of Specimens Increase in Diameter in 5 Days, in. Stress Relaxation, psi 2 2 4 0.001, 0.002 0.002, 0.003 0.002, 0.004, 0.004, 0.004 1600, 3200 3200, 4800 3200, 6400 Tests Under Controlled Conditions On the basis of the results cited, it was decided that constant temperature and relative humidity must be maintained in the alternate-immersion test to obtain significant and reproducible results. Therefore, the equipment previously described was installed in a room in which the air temperature was controlled to 80 ± 2 F and the relative humidity to 45 ± 6 per cent by the use of an air-conditioner, dehumidifier, and electric heater. With this combination of temperature and relative humidity, the solution had an average temperature of 76 ± 2 F. The temperature of the specimens during the 50-min drying cycle increased above the solution temperature by 1 to 2 F, as determined by thermocouples placed on specimens. At the relative humidity maintained, the specimens apparently dried completely between immersions. To determine whether quantitative and reproducible results could be obtained with the improved facilities, an experiment was designed in which two series of replicate specimens were tested in alternate immersion. All specimens were stressed to 35,000 psi. Details of these tests follow. Material Material was Al-Zn-Mg plate from two commercial lots in the solutiontreated, quenched, artificially aged One lotEST was a nominal Copyright by ASTM Int'l and (all rights reserved); Wedtemper. Dec 16 15:53:43 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprodu 176 STRESS CORROSION TESTING thickness of 1V4 in. and the other, 2 in. They were identified as Lots 1 and 2, respectively. These lots were selected because of their known low resistance to stress corrosion. Their composition is shown in Table 4. The mechanical properties in the long-transverse direction are shown in Table 5. Specimens Three-quarter-inch-diameter specimens were machined from Lot 1 and 1 V^-in.-diameter specimens from Lot 2. A total of 72 specimens was made TABLE 4—Chemical composition, weight per cent. Lot 1 Element Lot 2 0.1 to 0.2 0.1 to 0.3 <0.05 0.1 to 0.3 2.88 0.1 to 0.3 3.75 <0.05 none remainder Si Fe Cu Mn Mg Cr Zn Ti Zr Al 0.1 to 0.2 0.1 to 0.3 <0.05 0.2 to 0.5 2.97 0.1 to 0.3 3.57 <0.05 none remainder TABLE 5—Mechanical properties. T , Tensile Strength, psi 1 2 64 800 64200 Yield Strength, 0.2% offset, psi 57 000 55 900 Elongation, % 13.0 13.0 from each lot and tested in eight groups of nine specimens each. Specimens in each group were tested under identical conditions, with the start of the test for each group constituting a sequence with three- to seven-day intervals between starting times. Specimens were stressed, prepared, and tested according to methods previously described, except that no bushings were used as insulators. Instead, specimens were dipped in liquid neoprene as shown in the lower part of Fig. 3. Determination of Cracking Time Since examinations for cracks were not made during nights or weekends, there was necessarily a large uncertainty as to when cracks actually occurred in those instances when they were found after the weekend or night Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further repro TABLE 6—Failure time data on C-rings stressed to 35,000 psi and tested in alternate immersion under controlled conditions, days. Group No. Elapsed Time Averaged Time LOT No. 1—1J4-IN. PLATE, %-IN. C-RiNos 1 2.7, 3.7, 3.7, 3.7, 3.7, 3.7, 3.7, 3.7, 3.7, 4.7, 4.7, 5.0, 5.0, 5.7, 2.7, 3.0, 3.7, 4.0, 5.0, 5.0, 5.7, 4.0,4.7,4.7,4.7,4.7,4.7,4.7, 2.7, 3.0, 3.0, 3.0, 3.0, 3.0, 3.7, 2.7, 3.0, 3.0, 3.7, 3.7, 3.7, 4.0, 2.7, 2.7, 2.7, 2.7, 3.7, 3.7, 3.7, 2.7, 2.7, 2.7, 3.0, 4.0, 5.6, 5.6, 2 3 4 5 6 7 8 4.7, 5.7, 5.7, 7.0, 3.7, 6.7, 4.6, 6.7, 5.7 6.7 5.7 9.7 3.7 6.7 5.6 ...« 2.7, 3.4, 3.4, 3.4, 3.4, 3.4, 3.4, 4.4, 3.4, 3.4, 4.4, 4.4, 5.0, 5.0, 5.4, 5.4, 2.7, 3.0, 3.4, 4.0, 5.0, 5.0, 5.4, 5.4, 4.0,4.4,4.4,4.4,4.4,4.4,4.4, 7.0, 2.7, 3.0, 3.0, 3.0, 3.0, 3.0, 3.4, 3.4, 2.7, 3.0, 3.0, 3.4, 3.4, 3.4, 4.0, 6.4, 2.7, 2.7, 2.7, 2.7, 3.4, 3.4, 3.4, 4.3, 2.7,2.7,2.7,3.0,4.0,5.3,5.3, 5.4 6.4 5.4 8.4 3.4 6.4 5.3 6.4, LOT No. 2—2-iN. PLATE, I^-IN. C-RiNGS 2.7, 3.0, 3.7, 4.0, 4.7, 5.7, 6.7, 0.8, 2.8, 4.0, 4.0, 4.0, 6.7, 6.7, 3.7, 4.0, 4.0, 4.7, 4.7, 4.7, 4.7, 2.7, 3.0, 3.7, 4.7, 4.7, 5.7, 9.7, 4.0, 4.7, 4.7, 4.9, 4.9, 5.7, 9.7, 2.7, 3.7, 4.0, 4.7, 5.7, 6.7, 6.7, 2.7, 2.7, 3.7, 4.6, 4.6, 5.6, 7.0, 2.7, 3.0, 3.7, 5.0, 5.6, 6.7, 9.6, 1 2 3 4 5 6 7 8 0 6 12.0, 7.0, 11.0, 9.7, 9.7, 7.1, 9.6, 11.0, 24.0 14.0 14.0 11.0 9.7 11.0 9.6 13.0 2.7, 3.0, 3.4, 4.0, 4.4, 5.4, 6.4, . . . " 2.4, 4.0, 4.0, 4.0, 5.4, 5.4, 3.4, 4.0, 4.0, 4.4, 4.4, 4.4, 4.4, 2.7, 3.0, 3.4, 4.4, 4.4, 5.4, 8.4, 4.0, 4.4, 4.4, 4.9, 4.9, 5.4, 8.4, 2.7, 3.4, 4.0, 4.4, 5.4, 6.4, 6.4, 2.7, 2.7, 3.4, 4.3, 4.3, 5.3, 7.0, 2.7, 3.0, 3.4, 5.0, 5.3, 6.4, 8.3, Specimen lost. Omitted from analysis. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized. 12.0, 7.0, 11.0, 8.4, 8.4, 7.1, 8.3, 11.0, 24.0 14.0 14.0 11.0 8.4 11.0 8.3 13.0 178 STRESS CORROSION TESTING period. For the overnight cases, the true cracking time was assumed to be midway between the morning of discovery and the last previous examination of the preceding day. If the test was started on a Friday afternoon, as happened in many instances, the midpoint averaging method for cracks found the following Monday morning would be biased, because there is FIG. 10—Log™ of days-to-failure of 3A-in. C-rings, Lot 1, versus per cent failure plotted on normal probability paper. All specimens stressed at 35,000 psi. FIG. 11—Logio of days-to-failure of llA-in. C-rings, Lot 2, versus per cent failure plotted on normal probability paper. All specimens stressed at 35,000 psi. less likelihood of cracking during the early stage of the test. In these cases, the Monday morning cracking time was used. However, these values are identified hi graphs containing them. Table 6 shows the actual elapsed times when cracks were found and also the averaged times for the two lots of plate. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further repr ROSENTHAL AND PRITCHARD ON TEST FOR AL-ZN-MG ALLOY PLATE 179 Tests of Normality Figures 10 and 11 contain the data for Lots 1 and 2 on normal probability paper using the logarithms of failure times. Booth et al3 have shown that failure times are normally distributed when the times are converted to logarithms. The technique of plotting the points is that of Lewis.4 A straight line was fitted to the points visually, since the fit is quite simple. Because of this straight line relationship, the normality of the population is established. It is possible to determine from Figs. 10 and 11 various statistical quantities. Since the distribution is normal, the mean and the median can be considered to be identical. Other statistical data derived from these figures are discussed in the following section. TABLE 7— Statist ical data for Lots 1 and 2. Lot No. 1 2 a Median Failure Time Expected Median Min-to-Max Range0 Log 10 , days Days s Log 10 , days » = 7, days n = 9, days 0.590 0.720 3.9 5.2 0.130 0.220 3.0 to 5.1 3.3 to 8.2 3.1 to 4.9 3.6 to 7.8 » = is, days 3.3 to 4.6 4.0 to 6.9 95% confidence level. Determining Number of Specimens To Be Tested In developing a specification for stress corrosion susceptibility, the number of specimens to be tested must be kept as small as possible because of the costs of machining and handling. However, this desire must be balanced against the fact that the median failure time is more accurately determined with a larger group of specimens. The relevant relationship is: expected range of median failure times5 = median ± ts — where: s = estimated standard deviation, n = number of specimens, and t = a quantity dependent on n and the confidence level desired for the range of failure times. For 95 per cent confidence level, t = 2.36 (for n = 7), t = 2.30 (for n = 9), and t = 2.13 (for n = 15). 'Booth, F. F., Tucker, G. E., and Godard, H. P., "Statistical Distribution of Stress Corrosion Testing," Corrosion, Vol. 19, 1963, pp. 390T-395T. * Lewis, C. F., "Statistics—A Useful Tool for the Examination of Corrosion Data," Corrosion, Vol. 9, 1953, pp. 38-43. 5 Dieter, G. E., Jr., Mechanical Metallurgy, McGraw-Hill, New York, 1961. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further repro 180 STRESS CORROSION TESTING In Table 7, the estimated standard deviation and median failure times for Lots 1 and 2 are shown as determined graphically from Figs. 10 and 11. Also shown in the table are the calculated expected median failure times for 7, 9, and 15-specimen groups for a 95 per cent confidence level. The choice of a 9-specimen group appears reasonable since there is a worthwhile improvement in the calculated spread over a 7-specimen group. The added precision obtained with a 15-specimen group does not appear to FIG. 12—An analysis of data relating to Lots 1 and 2 and a number of other lots tested at 35,000 psi. The number of specimens tested for each lot is shown between the data points. Plotted on the ordinate scale is the range of median failure times calculated on the basis of nine specimens at 95 per cent confidence level; corresponding points on the abscissa are the median failure times. have good economic justification. With a group of nine specimens, the minimum median value for Lot 1 is approximately 20 per cent less than the experimental median; the corresponding value for Lot 2 is 30 per cent. The plots of experimental median versus 9-specimen minimum and maximum median are shown for Lots 1 and 2 in Fig. 12. Also plotted are a number of points corresponding to other lots tested under similar conditions. These lots represent a lesser number of specimens, but the calculations are based on a hypothetical 9-specimen group. The meaning of this graph can be further illustrated as follows: Assume an unknown lot of material is to be tested and that a group of nine specimens willbybeASTM used. Int'l It is(all desired a producer meet2015 a miniCopyright rightsby reserved); Wedthat Dec his 16 material 15:53:43 EST Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further re ROSENTHAL AND PRITCHARD ON TEST FOR AL-ZN-MG ALLOY PLATE 181 mum of 5-day median failure time with a confidence level of 95 per cent. The 5-day level is taken from the ordinate, and the corresponding minimum value on the abscissa must experimentally show a median of ll/z days to have less than 1 chance in 20 that the median of a group of 9 specimens falls below 5 days. When the consumer, on the other hand, accepts material having a median failure time of 5 days, there is some risk that he will accept material actually having a median time of 4 days. Conclusions When the laboratory temperature was controlled to 80 ± 2 F and the relative humidity to 45 ± 6 per cent (the conditions of these tests), results were reproducible. Using the controlled laboratory conditions for two lots of Al-Zn-Mg plate, the times-to-failure follow a normal distribution if failure times are converted to logarithms. Outdoor exposures in a marine atmosphere and in an industrial atmosphere gave similar results (on the basis of ranking) to the laboratory alternate-immersion tests carried out in the winter months. A cknowledgment The authors express appreciation to SP/7 B. Collins (U.S. Army) for preparation of the line drawings in this report; to the various staff members of the research laboratories of Alcoa, Kaiser, Reynolds, and Dow aluminum companies; and to General Motors personnel operating the U.S. Army Tank Automotive Plant (Cleveland). Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions author H. B. Romans1 Stress Corrosion Test Environments and Test Durations* REFERENCE: H. B. Romans, "Stress Corrosion Test Environments and Test Durations," Stress Corrosion Testing, ASTM STP 425, Am. Soc. Testing Mats., 1967, p. 182. ABSTRACT: The most frequent environments used for stress corrosion testing of aluminum and titanium are presented. The general effects of the various environmental factors which may affect the tests are described. These include temperature, humidity, pH, etc. The data on a definite test period are presented in a quite limited form due to the .scarcity of data in published papers on this subject. The results of a questionnaire which was submitted to all the members of Committee G-l concerning their testing methods for stress corrosion testing are presented. KEY WORDS: corrosion, stress corrosion, environmental testing, aluminum alloys, titanium alloys, temperature, humidity, pH, salt solutions, oxygen The types of environment in which stress corrosion cracking is best produced are those in which highly localized corrosion occurs with the absence of general surface corrosion. This makes the choice of the environment fairly specific for each of the metal families. Since no universal test environment can be adopted, it is necessary to discuss each of the metals separately. The test period beyond which a metal can be judged safe from stress corrosion cracking must depend on the end use being studied as well as the metal. At present, the test periods may range from one day to three years or until the test metal fails. In most cases, both of these considerations are left to the judgment of the persons conducting the test. Subcommittee VI of ASTM Committee G-l on Corrosion of Metals was formed for the purpose of setting up standard procedures for conducting stress corrosion tests. 1 Scientist, Department of Applied Chemistry and Mathematics, Reynolds Metals Co., Richmond, Va. Chariman, Task Group 2, Subcommittee VI, ASTM Committee G 1. Personal member ASTM. * Report of Task Group 2 of Subcommittee VI of ASTM Committe G-l on Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Corrosion of Metals. Downloaded/printed by 182 University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized. ROMANS ON TEST ENVIRONMENTS AND TEST DURATIONS 183 Task Group 2 of Subcommittee VI was formed for the purpose of compiling: 1. Data on test environments which are in general use for stress corrosion cracking tests. 2. The length of test required to give reasonable ensurance that a metal will perform satisfactorily in a given environment. The data presented in this report were obtained from three major sources. These were: (1) a literature survey by the chairman, (2) information supplied by the members of Task Group 2, and (3) a questionnaire submitted to all the members of Committee G-l. FIG. 1—Effect of stress on the time-to-stress corrosion failure. The report is divided into three general parts: I, General Philosophy of Testing; II, Stress Corrosion of Nonferrous Alloys; and III, Stress Corrosion Questionnaire on Ferrous Alloys. The results of the stress corrosion questionnaire are entered as part of the particular metal concerned. At the February 1966 Committee Week meeting in Washington, it was decided by the members of Subcommittee VI to present for publication only those metals which have been well covered. These include aluminum and titanium. The other metals will be added as supplements to the original report as they are completed. General Comments Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 It is not in thebyscope of Task Group 2 to cover in detail the methods of Downloaded/printed University of Washington (University of Washington) pursuant to License Agreement. No further reproductions 184 STRESS CORROSION TESTING applying the environment to produce stress corrosion. It is also considered that the publication of large amounts of data, test results, and test correlations would serve no useful purpose since it would be a repeat of published work. Therefore, the direction of this report has been aimed at discussing the environments used and the general methods of application. Titanium is presented in a more detailed form which does include test methods and periods. The work in this field is relatively new and the published work scarce; this makes a compilation useful. There has been no effort to describe specimen geometry or methods of stress as this was covered very well by Task Group 1. Interpretation of FIG. 2—Influence of the season of the year on rate of stress corrosion of brass. data and test results will be covered by Task Group 3 and is not necessary in this report. Comments on Test Periods Very little is mentioned in the literature as to a definite test duration. Most workers in the stress corrosion field use a test period which is based on the type test, test solution, material, convenience or space, and end use for the data. A number of practical considerations must enter into the selection of a test period other than those above. For instance, the stress level with respect to the yield strength of the material must be considered. As can be Copyright by ASTM (all rights reserved); Dec quite 16 15:53:43 EST seen in Fig. 1, the Int'l time-to-failure may be Wed affected radically by2015 small Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further ROMANS ON TEST ENVIRONMENTS AND TEST DURATIONS 185 differences in stress at low stress levels. At very high stress levels, the stress per unit area is not a very large factor [7].2 If outdoor exposures are made, the time of year the test is started must be considered in selecting the test period. Figure 2 shows the variation in time-to-failure which may be obtained by exposing at different months of the year. The corrosivity of the environment is another important factor in choosing a test period. Extremely corrosive environments may produce stress cracks at a slower rate than a mild environment. Careful consideration and judgment with regard to the general application of the metal may dictate the test period to be used. A consideration which may set the period of test is quality control. For instance, one quality control test requires that specimens pass a 5-day median test period. Another requires a 30-day test. PART I. GENERAL PHILOSOPHY OF TESTING Laboratory Testing for Stress Corrosion Susceptibility Stress corrosion tests conducted in the laboratory are generally under artificial environments conducted in such a way as to accelerate the corrosion and duplicate the type of failure experienced in service environments. Laboratory tests are advantageous since they can be performed under carefully controlled conditions. These conditions may include controlled atmospheres such as temperature, humidity, and air pollution. These are obtained by intermittent immersion, partial immersion, total immersion, wick and drip feed, and spraying. Only if the conditions are controlled to a high degree can accurate data comparisons be made. The test conditions must be specific to the alloy system under study since the stress corrosion of each metal is associated with a definite environment. Quite often an environment which develops stress corrosion hi one metal, inhibits the cracking of another due to severe general attack. Care should be taken to ensure that laboratory tests are not so severe that failures are erroneously assumed to be stress corrosion cracking when it is actually mechanical failure due to the reduction in cross-sectional area. This danger is minimized by careful selection of the corrodent, specimen size, and method of loading. Accelerated stress corrosion tests are most useful if the results correlate with service experience or natural environments. If the tests do not correlate with either of the above, they may still be useful for screening purposes. Variations in a given environment can cause a marked change in the 2 The italic numbers in brackets refer to the list of references appended to this Copyright paper. by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduct 186 STRESS CORROSION TESTING mode and type of cracking. For instance, in certain magnesium alloys, a change in pH of the corrodent (NaCl-K2CrO4) will cause the failure to change from transgranular, in high pH solutions, to intergranular, in low pH solutions [2], 'Various Environmental Factors Which Affect Stress Corrosion Test Results Physical Factors Temperature—The temperature of the test solution is probably the most important single factor in the role of stress corrosion cracking for some metals and alloys such as the aluminum-zinc-magnesium alloy. The rate of cracking increases with increase in temperature.3 A possible explanation is that the increase in temperature increases the conductivity of the solution. Also, diffusion at elevated temperatures is faster which allows increased activity in the corrosion cells. Humidity—The effect of humidity on the results of stress corrosion tests applies especially where the method of alternate immersion is used. There is general agreement that the specimens never become completely dry due to water absorption in the oxide layer. However, there is probably a critical amount of moisture which must be removed to get reproducible results. During periods of high humidity, the specimens can stay dripping wet for a long period and often do not dry between immersions. Some of this deficiency may be overcome if the air is circulated to dry the specimens. However, the test configuration often does not allow forced air drying. The literature indicates that the effect of humidity is maximum for aluminum at about 85 per cent and gives a marked decrease in the time-tofailure3 [3]. The stress corrosion cracking of all metals is affected to some extent by the relative humidity. The effect of this variable can and should be eliminated by controlling the condition of the test area. Atmospheric Environments For the most part, stress corrosion testing in the atmosphere is used to correlate laboratory tests and to determine how the metal will behave in a natural environment. Unfortunately, long periods of exposure are needed to produce failures in all but the most susceptible alloys, or when the superimposed load is very near the yield strength of the material. Moisture, temperature, and various atmospheric impurities, that is, seacoast, industrial, rural, etc., influence differently the failures of different metals. 8 Copyright See by p. ASTM 363. Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproducti ROMANS ON TEST ENVIRONMENTS AND TEST DURATIONS 187 Aluminum specimens may fail faster in an industrial atmosphere depending on the alloys3 [4]. Magnesium alloys subjected to a marine and a rural atmosphere failed in about the same time period. Stainless steels tested at Kure Beach, N.C., and Middletown, Ohio, showed failures occurring within a few weeks at Kure Beach, while the same alloys did not fail in two years at Middletown.4 Note that the rate of stress corrosion cracking under atmospheric conditions increases at high humidity and high temperatures. This means that exposures should be made at the same season of the year if accurate data comparisons are to be made. The time-to-failure can vary as much as a factor of eight by varying the month of exposure [5]. Chemical Factors Neutral Salt Solutions—The stress corrosion cracking of metals in neutral solutions of various salts, particularly sodium chloride, has been widely investigated. Many of the metal alloys are susceptible to stress corrosion cracking in distilled water at elevated and ambient temperatures. However, cracking occurs much more slowly in distilled water than in water containing a few parts per million of chlorides. An increase in sodium chloride (NaCl) concentration has been shown to cause a continual increase in the rate of stress corrosion cracking and in the rate of total corrosion of magnesium and some aluminum alloys. However, this does not hold true for all metal alloys.3 The addition of oxidizers (K2Cr2O7 or K2CrO4) to solutions of NaCl inhibits total corrosion and increases the rate of stress corrosion in some of the nonferrous alloys. This may be explained in part if we consider that the formation of a large number of pits on a metal surface hinders the development of isolated stress raisers into stress corrosion cracks. The Influence of Oxygen—Oxygen plays an essential role in the process of corrosion of unstressed metals. It probably has an even greater significance in the process of stress corrosion cracking. For example, an experiment was performed with Al-5Mg and Al-7Mg alloys in aerated and deaerated 3 per cent solutions of NaCl [6\. In aerated solutions, the specimens cracked after several hours and in deaerated solutions, no cracking was observed in 94 days. The influence of oxygen from the air was also confirmed under conditions of full immersion and spraying with a 3 per cent solution of NaCl. The immersion test produced failures in 365 days, while spraying produced failures in 55 days [7], It has been noted that in alternate immersion testing, specimens of Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 'Ellis, O. B., personal correspondence. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprodu 188 STRESS CORROSION TESTING aluminum-zinc-magnesium alloys at a shallow depth in the test tank failed faster than comparable specimens tested at a greater depth.5 Other work [7] states that for stainless steel, "salt spray stress corrosion testing produced failures while alternate immersion proved inadequate to cause stress corrosion failures in the materials tested." The Influence of pH—The rate of stress corrosion cracking of most metals generally decreases with increase in pH, especially in the alkaline range [8,9]. The pH of most tests is a function of the environment used unless an effort is made to control it. Most test environments will be neutral or very weakly acid. PART II. NONFERROUS ALLOYS Aluminum Laboratory Tests Laboratory tests are advantageous in that they can be performed in a convenient location under conditions that can be controlled. An ideal laboratory stress corrosion cracking test must be simple to control, rapid, and correlatable with service conditions. The Alternate-Immersion Test—This is a popular test for aluminum alloys, including aluminum-copper, aluminum-magnesium, and aluminum-zinc-magnesium, aluminum-zinc-magnesium-copper and aluminummagnesium-silicon types. Data obtained in this test may be correlated with seacoast atmosphere and seawater tests [4,10]. This test is more severe than inland industrial atmospheres for most alloys, the exception being those of the aluminum-zinc-magnesium family that contain little or no copper. In the case of the latter alloys, the 3.5 per cent NaCl alternate-immersion test is less effective in some cases than exposure to an industrial atmosphere. This test is fairly easy to control, but is not rapid enough for some purposes. The cycle generally used is 10 min in solution and 50 min out of solution. Continuous Immersion—Continuous-immersion tests are relatively simple to operate and control, but for some alloys they are not so effective as alternate-immersion types. More aggressive electrolytes have been used to make this procedure more effective. Immersion in boiling 1 N sodium chloride solution is an effective test for aluminum-zinc-magnesium alloys,3 but not so reliable for aluminum-zinc-magnesium-copper and other types of alloys as the 3.5 per cent NaCl alternate-immersion test. Acidified salt chromate solutions are effective in producing stress corrosion cracking, but they also produce severe localized intergranular or pitting attack in unstressed specimens. The corrosive attack is so drastic that "corrosion" failures will occur in relatively short exposure periods, Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 5 Task Group by 2 correspondence. Downloaded/printed University of Washington (University of Washington) pursuant to License Agreement. No further ROMANS ON TEST ENVIRONMENTS AND TEST DURATIONS 189 and the test data obtained will be very unrealistic unless pains are taken to separate the failures that are not the result of stress corrosion cracking. Salt Spray Test—This is reported to be an effective test environment and is reported to give results that are comparable to alternate-immersion tests on most alloys. In general, the spray type of test is not so. easy to control as an intermittent-immersion test. A salt spray test procedure is outlined by Chadwick and Grainger. It is also reported that an acidifie^ (pH 3) 5 per cent intermittent spray is effective with an exposure period of two weeks being sufficient for many aluminum alloys [11,12]. Electrolytic Tests—Electrolytic tests are highly accelerated stress corrosion cracking tests. They are conducted by impressing a current between the specimen as the anode and a suitable cathode in a sodium chloride solution. This type of test has been reported to be most successful for aluminum-magnesium type alloys. The method has been shown to place aluminum-magnesium alloys in the same order of stress corrosion susceptibility as the slower immersion test of stressed specimens in salt peroxide solution and in the same order as obtained by immersion in the sea and exposure in a seacoast atmosphere6 [13]. Highly accelerated tests such as this usually are most useful as screening tests for alloy development and quality control purposes. Environmental Factors Which May Affect the Stress Corrosion Test Results of Aluminum Humidity—An environment that will support the stress corrosion of aluminum must provide an electrolyte in which the electrode relationships of the readily corrodible paths, such as the grain boundaries, are anodic to the rest of the metal. It has been established that only very small amounts of moisture are required on the metal surface to provide the electrolyte3 [14]. Thus, stress corrosion cracking of susceptible aluminum alloys may occur in mildly corrosive environments such as the atmosphere and in distilled water. The moisture films which develop on a metal surface exposed to the atmosphere are acidic, especially in seacoast and industrial locations. The cracking susceptibility of aluminum alloys reaches a maximum at about 85 per cent relative humidity. Atmospheric Environments—Not all alloy types react the same to different environments. For example, aluminum-copper, aluminummagnesium, and aluminum-zinc-magnesium-copper alloys are more prone to stress corrosion cracking in seacoast atmospheres than in inland industrial or rural atmospheres. An aluminum 5 per cent magnesium alloy fabricated to be very susceptible to stress corrosion was exposed at a tropical marine site in Aruba, Netherlands Antilles. It failed in one day. At a marine site in Kure Beach, N.C., failure occurred in 32 days and at a rural atmosphere in Richmond, Va., in 258 days [75]. But Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by "Seep. 51. University of Washington (University of Washington) pursuant to License Agreement. No further reproduct 190 STRESS CORROSION TESTING aluminum-zinc-magnesium alloys with relatively low copper (7079) and no copper (7039) are just as susceptible in a less corrosive inland atmosphere as in the seacoast locations7 (also see Refs 4 and 16). Temperature—This is the most critical single factor affecting the rate of stress corrosion cracking for some aluminum alloys and should be carefully controlled. An increase in temperature decreases the time-tofailure. If the log of the life is plotted against the reciprocal of the absolute temperature, a straight line function is obtained. The straight line holds to about 10 C at which point a change in slope indicates that the time-tofailure at 0 C would be infinitely long.3 Work indicates that the effect of temperature on time-to-failure of aluminum alloys depends greatly on the alloy. An aluminum-zinc-magnesium alloy showed the time-to-failure was increased 12 min for each degree centigrade. In a less susceptible aluminum-zinc-magnesium alloy the time-to-failure was increased to 629 min per degree centigrade.3 Chlorides—The presence of certain anions, notably chlorides, tend to accelerate the stress corrosion of aluminum. For this reason, and because sodium chloride is so prevalent in natural environments, most stress corrosion tests are carried out in sodium chloride solutions. Constant deflection "loop tests" of sensitized Al-7Mg alloy sheet were conducted by immersing specimens hi sodium chloride solutions of different concentrations ranging from 1.5 to 25 per cent [6]. As the concentration was increased, the specimen life decreased from about 26 to 3 hr at 9 per cent NaCl after which the life remained constant. An experiment was performed by one research laboratory using a sensitized Al-5Mg alloy exposed by alternate immersion in sodium chloride solutions ranging in concentration from 3.5 down to 0.005 per cent (50 ppm). Tension specimens stressed to 75 per cent of the yield strength were subjected to bending strain and exposed. Stress corrosion cracking occurred in even the most dilute solution although the specimen life increased from 20 hr at 3.5 per cent NaCl to about 180 hr at 0.005 per cent [10]. In another laboratory an aluminum-zinc-magnesium alloy showed no difference in failure time when specimens were tested at 60 per cent of the yield strength in 0.01, 0.1, 1, and 2 N sodium chloride solution at 180 F. But the time-to-failure was increased five times when tested in distilled water.3 Hydrogen Ion Concentration (pH)—A change in the pH or other ions present in the corrosive medium may alter the electrochemical' relationship of the metal constitutents and thereby affect the type of corrosion, which in turn could influence the stress corrosion resistance. Work indicates that a change in mechanism occurs at about pH 3.6 (Fig. 1 from footnote 3). In a laboratory test of an Al-lOMg alloy stress corrosion cracking occurred in a neutral 53 g/liter NaCl + 3 g/liter H2O2 solution Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 7 Downloaded/printed by See p. 342. University of Washington (University of Washington) pursuant to License Agreement. No further reprod ROMANS ON TEST ENVIRONMENTS AND TEST DURATIONS 191 in 15 min. When the pH was raised by the addition of 10 g/liter NaOH, no cracking occurred because general corrosion predominated rather than intergranular attack [77]. In another experiment an Al-7Mg alloy failed in 5 min in 3 per cent NaCl solution at pH 1 and hi 60 days at pH 10.7; no failures occurred above this pH value. Oxygen—Oxygen plays an essential role in the stress corrosion cracking of aluminum alloys. Tests have been conducted hi aereated and deaereated 3 per cent solutions of NaCl, using Al-5Mg and Al-7Mg alloys. In aereated solutions, the specimens cracked after several hours, and in deaereated solutions, no cracking was observed in 94 days [6]. The influence of oxygen from the air was also shown by a comparison of full immersion versus spraying with 3 per cent NaCl solution. Spraying produced failures in 55 days hi contrast to 365 days for the immersion test. Results of Questionnaire on Stress Corrosion Testing of Aluminum The survey by Task Group 2 of the Committee G-l members turned up 37 investigators conducting stress corrosion tests on aluminum. The tests were conducted for one or more of the following purposes: 1. Quality control 18 2. Development research 31 3. Applied research 35 4. Pure research 20 All the investigators are involved in more than one of the above programs. There is no exclusive pure research being conducted on aluminum except through government-sponsored research. The groups doing this work were not contacted by this survey. The three general methods being used for testing are as follows: 1. Laboratory tests 32 2. Field tests 27 3. Service tests 16 It may be noted that most people conducting laboratory tests also conduct field tests. The method of exposing the specimens is by total or alternate immersion and high humidity. The number using each test is: 1. Total immersion 26 2. Alternate immersion 20 3. High humidity 14 The test environments are not so varied as in some of the other metals. They are divided up as follows: 1. Sodium chloride 24 2. Salt spray 5 3. Magnesium chloride 4 Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 4. Seawater by 4 Downloaded/printed University of Washington (University of Washington) pursuant to License Agreement. No further reproductions autho 192 STRESS CORROSION TESTING 5. Metal cleaning 2 Sodium chloride is certainly the most popular solution in use. The water used for the test solution shows that distilled water is preferred. The tabulated list reads as follows: 1. Distilled 24 2. Deionized 10 3. Tap 6 4. Seawater 2 5. Unknown 2 The survey showed that about half of the people use a written procedure. The exact figures are 18 using a written procedure, 17 not using a procedure, and 2 with "no comment." Most testing is conducted under controlled temperature conditions. The number reporting temperature control of the test environments are 27 and uncontrolled temperature of the test environments are 10. The grade chemicals used may be tabulated as follows: 1. Industrial 9 2. U.S. pure 12 3. Chemical pure 21 4. Reagent 1 5. Unknown 3 The test environments are changed as shown below: 1. Continuously 2 2. Daily 4 3. Weekly 8 4. Monthly 2 5. Each two days 1 6. Varies 8 7. After each test 1 8. Unknown 5 9. Other 5 There appears to be more disagreement about the period of solution change than any other factor. Most members were noncommittal on the questions of factors other than the above which affect stress corrosion. Some comments were, surface preparation, stress, specimen configuration, solution flow rate, and pH. Copper Alloys Questionnaire Results The survey results from 19 investigators on copper and copper alloy stress corrosion cracking are tabulated below. Tests are conducted for one or more of the following reasons: Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduction ROMANS ON TEST ENVIRONMENTS AND TEST DURATIONS 193 1. Quality control 11 2. Development research 14 3. Applied research 15 4. Pure research 8 The tests are conducted in: 1. Laboratory tests 16 2. Field tests 11 3. Service tests 6 The specimens are exposed to the environment by: 1. Total immersion 14 2. Alternate immersion 7 3. High humidity 5 The test environments are: 1. Sodium chloride 7 2. Magnesium chloride 7 3. Mercury compounds 5 4. Ammonia vapor 5 5. Molten bismuth 2 6. Mercury metal 1 7. Mattsson's solution 1 8. Seawater 2 9. Other 2 Eleven investigators use a written procedure while eight do not. Seven have a definite test period and twelve do not. The water used for solution make up is: 1. Deionized 3 2. Distilled 10 3. Tap 2 4. Seawater 2 The grade chemicals are: 1. Industrial 3 2. U.S. pure 5 3. Chemically pure 11 4. Unknown 2 Fourteen control the test temperature and five do not. The solution is changed: 1. Daily 1 2. Weekly 1 3. Constantly 4 4. Varies 4 5. Unknown 4 6. Other 4 7. As necessary 2 Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprodu 194 STRESS CORROSION TESTING Magnesium Results of the Questionnaire on Magnesium Alloys The survey results from three investigators on magnesium stress corrosion cracking are as follows: Tests are conducted for one or more of the following reasons: 1. Quality control 2 2. Development research 2 3. Applied research 3 4. Pure research 3 The tests are conducted in: 1. Laboratory tests 3 2. Field tests 3 3. Service tests 2 One investigator has a written procedure while two do not. None has a definite test period. The water used for solution makeup is: 3 1. Deionized 2. Distilled 3 The grade chemicals are: 1. U. S. pure 1 2. Chemically pure 2 Two control the test temperature and one does not. The solution is changed: 1. Daily 1 2. Weekly 1 3. Varies 1 4. Unknown 1 Nickel Results of the Questionnaire on Nickel Alloys The survey results from eleven investigators on nickel and nickel alloy stress corrosion cracking are tabulated below. Tests are conducted for one or more of the following reasons: 1. Quality control 7 2. Development research 10 3. Applied research 8 4. Pure research 5 The tests are conducted in: 1. Laboratory tests 9 2. Field tests 7 3. Service tests 4 The specimens are exposed to the environment by: Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 1. Total immersion 10 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduct ROMANS ON TEST ENVIRONMENTS AND TEST DURATIONS 195 2. Alternate immersion 5 3. High humidity 3 The test environments are: 1. Sodium chloride 5 2. Magnesium chloride 5 3. Seawater 2 4. Unknown 1 Five investigators use a written procedure while six do not. Three have a definite test period and eight do not. The water used for solution make up is: 1. Deionized 2 2. Distilled 5 3. Tap 1 4. Unknown 3 The grade chemicals are: 1. U.S. pure 2 2. Industrial 3 3. Chemically pure 6 4. Unknown 3 Eight control the test temperature and two do not. The solution is changed: 1. Constantly 1 2. Varies 7 3. Unknown 3 Titanium8 Service failures associated with stress corrosion cracking of titanium alloys have been attributed to hot salt, hot chlorinated hydrocarbons, hydrochloric acid, fuming red nitric acid, N2O4 , and methanol. Liquid metal embrittlement of titanium has been attributed to molten cadmium from overheated cadmium-plated fasteners. Fairly reproducible laboratory tests for detecting stress corrosion cracking of titanium alloys have been devised for all of the above environments except perhaps methanol. The cracking problem with methanol has arisen only recently during the pressure testing of a Ti-6Al-4V Apollo fuel tank. This failure is so recent that insufficient laboratory tests have been made to define the conditions. On the other hand, a laboratory test has been developed which causes stress corrosion cracking in salt (NaCl) solutions, although there have been no service failures of titanium in this environment. The test consists of propagating a crack which has been initiated in a bent-beam specimen. The details of the test are described in the section on 3.5 per cent NaCl (precracked specimens). Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 8 Downloaded/printed by Compiled by Warren Berry, Batelle Memorial Institute, Columbus, Ohio. University of Washington (University of Washington) pursuant to License Agreement. No further reproductions 196 STRESS CORROSION TESTING A summary of the conditions of each of the laboratory tests is presented in the following sections along with a listing of the sensitive alloys and a bibliography. The bibliography does not contain all the available references on the subject but includes selective references which are representative of those describing the particular test procedure. Hydrochloric Acid Tests The tests in hydrochloric acid have been conducted with bent beams and U-bends. Susceptible alloys include: Ti-6.8Al-2.2Sn, Ti-8Al-lMoIV, Ti-12Zr-7Al, and unalloyed titanium. The test is described in Table 1. TABLE 1 — Hydrochloric acid tests. Test solution Test temperature Specimen type Failure time Type of cracking Alloys which cracked Alloys which did not crack 5 to 10% HC1 in water 95 F 4-point loading, bent beam and U-bend instantaneous to >12 days intergranular (1) A-110AT (6.8A1, 2.2Sn) (2) A-55 (commercial purity) (3) A-70 (4) Ti-12Zr-7Al (5) Ti-8Al-lMo-lV (transgranular) C-110M (7.3 Mn) and Ti-3.36Mn. TABLE 2— Red fuming nitric acid. Test solution Test temperature Specimen type Failure time Type of cracking Alloys which cracked red fuming nitric acid (RFNA), 2.5 to 20% NO22 room temperature U-bends and Erichsen cups 3 hr to 1 week (168 hr) intergranular (1) 75A (commercial purity) in anhydrous RFNA containing 7% or more NO2 and in RFNA -20% NO2 containing <0.6% water. (2) Ti-8Mn in anhydrous RFNA containing 2.5% or more NO2 and in RFNA-20% NO2 containing <0.7% water. Alloys which did not crack : . . .. none listed Tests in Red Fuming Nitric Acid Stress corrosion cracking in red fuming nitric acid depends upon the NC>2 and moisture contents of the acid. Titanium 75A (commercial purity) cracks in anhydrous acid when the NO2 content is greater than 7 per cent, while Ti-8Mn cracks at NO2 greater than 2.5 per cent. Both resist cracking in acid containing 20 per cent NO2 if the moisture content exceeds 0.6 to 0.7 per cent. Additions of 1 per cent NaBr to red fuming nitric acid Copyright by ASTM Int'l corrosion (all rights reserved); Wedof DecTi 16 75A. 15:53:43 ESTtest 2015is described also inhibits stress cracking The Downloaded/printed by in Table 2. University of Washington (University of Washington) pursuant to License Agreement. No further reproduction ROMANS ON TEST ENVIRONMENTS AND TEST DURATIONS 197 Tests in Chlorinated Hydrocarbons Cracking of titanium in chlorinated hydrocarbons has been observed during stress relieving treatments at about 1150 F and hydrostatic testing at 600 to 700 F. The corrodent in the. stress relief failures was believed to be residual trichloroethylene from vapor degreasing, while that in the hydrostatic test was a proprietary high-temperature fluid. Laboratory tests with chlorinated hydrocarbons have been conducted in these two temperature ranges. Ti-5Al-2.5Sn was found to be susceptible to cracking at both temperatures. The tests are described in Table 3. TABLE 3 — Tests in chlorinated hydrocarbons. Trichloroethylene: Test solution Test temperature . . . . Specimen type Failure time Type of cracking Alloys which cracked Chlorinated Diphenyl: Test solution Test temperature Specimen type Failure time Type of cracking Alloys which cracked Alloys which did not crack as-received trichloroethylene 1150 to 1500 F U-bend and circle-cross patch welds 1 to 16 hr not given (probably intergranular) A-110AT (5Al-2.5Sn) as-received chlorinated diphenyl 600 to 700 F welds and circular patch welds with 0 to 70,000 psi applied stress V± to 3 hr not given (probably intergranular) RC A-110AT (5Al-2.5Sn) . none listed Hot Salt Cracking Tests The hot salt cracking problem has received considerable attention because the skin temperatures of the titanium wings in supersonic aircraft are expected to approach the 550 F temperature where cracking can occur and because of the possibility of salt contamination during heat treatment at higher temperatures (1100 to 1300 F). Laboratory tests have been devised using NaCl and sea salt (natural and synthetic) in the 550 to 950 F range and NaCl in the 1100 to 1300 F range. Recent tests indicate that NaCl produces cracking more readily than sea salt. The salt is applied by several cycles of dipping in a water solution or paste followed by drying prior to exposure at high temperatures. The temperature for drying the salt solution ranges from room temperature to 300 F. Some tension tests are run for long periods of time at applied loads of greater than 25 to 30 per cent of the 0.2 per cent offset yield strength. The specimens are then pulled in tension to determine whether incipient Copyright by ASTM Int'l (all reserved); Wed strength. Dec 16 15:53:43 EST 2015 stress corrosion cracking has rights reduced the tensile Downloaded/printed by The alpha-phase alloys such as Ti-5Al-2.5Sn, Ti-12Zr-7Al, Ti-8Al- University of Washington (University of Washington) pursuant to License Agreement. No f 198 STRESS CORROSION TESTING TABLE 4— Tests in hot NaCl (550 to 950 F). Starting test solution Test temperature Specimen type Test time Type of cracking Alloys which cracked Alloys which did not crack . . 3% NaCl to supersaturated boiling NaCl solution dried on specimen at room temperature or up to 300 F . . 550 to 950 F . . tension, bent beam, self-stressed spot welded beams, right angle bends . . 20 to 7000 hr . . intergranular . . Ti-8Al-lMo-lV, Ti-6AMV, Ti-12Zr-7Al, 2.5A1lMo-10Sn-5Zr, Ti-12V-llCr-3Al . . TMAl-3Mo-lV (stressed to 100,000 psi and exposed 7000 hr at 500 F) TABLE 5—Tests in sea salt (550 to 800 F). Starting test solutions Test temperature Specimen Type Test time Type of cracking Alloys which cracked Alloys which did not crack . . . . 6 parts NaCl-1 part MgCl 2 or 7 parts NaCl-1 part MgCl2 (synthetic sea salt solutions); natural sea salt slurry . . 550 to 800 F . . tension, precracked tension, self-stressed welded beams, tube-type tension, tension specimens with side hooks, notched and unnotched cantilever beams, bent beams . . 50 to 20,000 hr . . intergranular . . Ti-8Al-lMo-lV, Ti-6AMV, Ti-5Al-2.5Sn, Ti4A1-3MO-1V, Ti-5Al-2.75Cr-l.25Fe, Ti-13VHCr-3Al, Ti-6Al-6V-2Sn . . Ti-5Al-2.5Sn, TMAl-3Mo-lV, Ti-5Al-2.75Cr1.25Fe (precracked specimens exposed 50 hr at 600, 700, 800 F at 50 to 80 per cent of 0.2 per cent 100-hr creep strength), 6A1-4V (notched and unnotched cantilever beams exposed 20,000 hr at 650 F at 30 to 100 per cent of yield strength) TABLE 6— Tests in hot NaCl (1100 to 1300 F). Starting test solution Test temperature Specimen type Test time Type of cracking Alloys which cracked Alloys which did not crack . . . 100 ppm solution to salt slurry (NaCl) . 1100 to 1300 F . bent beam, U-bend, tension . . y% to 150 hr . . intergranular . . Ti-5Al-2.5Sn, Ti-5Al-5Sn-5Zr, Ti-12Zr-7Al, TiSAl-lMo-lV . . none listed IMo-l V, and Ti-5Al-5Sn-5Zr appear to be the most susceptible to hot salt cracking. The alpha-beta alloys are less susceptible, and the degree of susceptibility appears to increase with increasing aluminum content. However, the Ti-8Mn alloy, which contains no aluminum, is also susceptible ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 toCopyright hot saltbycracking. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduct ROMANS ON TEST ENVIRONMENTS AND TEST DURATIONS 199 Among the alloys with intermediate resistance to cracking are: Ti-5Al5Sn-5Zr-lMo-lV, Ti-6Al-4V, Ti-6Al-6V-2Sn, Ti-5Al-2.75Cr-l.25Fe, and Ti-13V-llCr-3Al. Alloys exhibiting the most resistance to hot salt cracking are: Ti-4Al-3Mo-lV, Ti-10Sn-5Zr-2.5Al-lMo, and Ti-4Mo4Zr-2Al. The laboratory tests are described in Tables 4, 5, and 6. TABLE 7— Tests in 3.5% Nad (precracked specimens). 3.5% NaCl solution pH 8 or synthetic sea salt0 ambient cantilever beam, notched, and with and without crack initiated at base of notch by fatigue less than 1 hr at critical load Failure time Type of cracking intergranular and transgranular unalloyed RS-70 (annealed) Alloys which are sensitive Ti-7Al-2Cb-lTa (annealed) Ti-7Al-3Cb (annealed) Ti-6Al-2.5Sn (annealed) Ti-5Al-2.5Sn (annealed) Ti-6Al-3Cb-2Sn (annealed) Ti-7Al-3Cb-2Sn (annealed) Ti-8Al-3Cb-3Sn (annealed) Ti-8Mn (annealed) Ti-8Al-lMo-lV (slightly) (annealed) Ti-6AMV (very slightly) (annealed) Ti-6.5Al-5Zr-lV (aged at 1100 F) Ti-6Al-4V-lSn (aged 1100 F) Ti-6Al-6V-2.5Sn (aged 900 F) Ti-6Al-2Mo (aged 1100 F) Ti-7Al-3Mo (annealed) Ti-13V-llCr-3Al (annealed) Alloys which are not sensitive.. Ti-65A (annealed) Ti-6Al-4V (annealed and annealed and aged) Ti-7Al-2.5Mo (annealed) Ti-6Al-2Mo (annealed) Ti-6Al-2Sn-lMo-lV (annealed) Ti-6.5Al-5Zr-lV (annealed) Ti-6Al-2Sn-lMo-3V (annealed) Ti-5Al-2Sn-2Mo-2V (annealed) Ti-6Al-2Cb-lTa-0.8Mo (annealed) Ti-4Al-3Mo-lV (age hardened) Ti-13V-llCr-3Al (age hardened) Test solution Test temperature Specimen type 0 ASTM Specifications for Substitute Ocean Water (D 1141 - 52). Tests in 3.5 Per Cent NaCl Solution or Seawater Titanium alloys have exhibited an exceptional resistance to stress corrosion cracking in salt solutions or seawater when exposed as bent beams, U-bends, or tension specimens. However, in recent months it has been shown that certain alloys show a marked tendency toward crack propagation in salt water if they are precracked prior to exposure. The test usually consists of machining a notch across the short edge of a rectangular cross-section beam. The beam is then fatigued until a small Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 crack is initiated. A Downloaded/printed by cup containing the salt solution is then sealed around University of Washington (University of Washington) pursuant to License Agreement. No further reprod 200 STRESS CORROSION TESTING the crack. The precracked specimen is then stressed as a cantilever beam. Sucessively larger weights are added until the crack propagates. A stress intensity factor is then calculated: where: m = moment at the notch (in-lb), B = horizontal thickness of the specimen (in.), D = vertical depth of specimen (in.), and a = 1 — (a/D) where a = total initial depth of notch plus fatigue crack (in.). TABLE 8— Testing in NZO± . Test solution Test temperature Specimen test Failure time Type of cracking Alloys which cracked Alloys which did not crack liquid N2O4 containing <0.18% H2O and <0.06% NO 85 to 165 F bent beams or pressurized tanks > 40 ,000 psi stress 1 to 20 days intergranular . Ti-6AMV; Ti-75A (commercial purity) . . none listed Similar tests are conducted in air (no salt solution) and the corresponding intensity factor is calculated. If the intensity factor for exposure to salt water is less than that for air, the alloy is judged to be sensitive to stress corrosion cracking. Evaluations of the implications of this test are now under way. Initial tests have indicated that a number of alloys are sensitive to stress corrosion cracking, but many are not (Table 7). In general, it appears that sensitivity to cracking in seawater is associated with aluminum content in the alloy, isomorphous beta stabilizers (columbium, molybdenum, and vanadium) and aging heat treatments (900 to 1300 F). Tests in N2O4 Service failures of Ti-6Al-4V tanks have occurred in cylindrical vessels (0.020-in. wall) which were pressurized to 250 psi at 105 F with propellant grade N2O4 (< 0,20 weight per cent moisture). Subsequent tests with pressurized tanks and bent-beam specimens have shown that cracking can occur at moderately high stresses in N2O4 containing less than 0.18 per cent H2O and < 0.06 per cent NO. Cracking is inhibited by additions Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further repro ROMANS ON TEST ENVIRONMENTS AND TEST DURATIONS 201 to the N2O4 of > 0.18 per cent H2O, > 0.06 per cent NO, or 7 per cent HNO3 . The laboratory tests are described in Table 8. Results of the Questionnaire on Titanium The survey results from eight investigators on titanium stress corrosion cracking are tabulated below: Tests are conducted for one or more of the following reasons: 1. Quality control 2 2. Development research 8 3. Applied research 7 4. Pure research 4 The tests are conducted in: 1. Laboratory tests 6 2. Field tests 7 3. Service tests 8 The specimens are exposed to the environment by: 1. Total immersion 5 2. Alternate immersion 4 3. High humidity 3 The test environments are: 1. Sodium chloride 4 2. Magnesium chloride 1 3. Solid salt 1 4. Seawater 2 Six investigators use a written procedure while two do not. Two have a definite test period and six do not. The water used for solution makeup is: 1. Deionized 1 2. Distilled 5 3. Seawater 2 The grade chemicals are: 1. U.S. pure 1 2. Industrial 1 3. Chemically pure 5 4. Unknown 1 5. ASTM sea salt 1 Four control the test temperature and six do not. The solution is changed: 1. Daily 2 2. Weekly 1 3. Constantly 1 4. Varies 3 Copyright Downloaded/printed University by by of 202 STRESS CORROSION TESTING PART III. FERROUS ALLOYS Mild Steel Results of the Questionnaire on Mild Steel The survey results from 14 investigators of mild steel stress corrosion are tabulated below. Tests are conducted for one or more of the following reasons: 1. Quality control 7 2. Development research 13 3. Applied research 12 4. Pure research 8 The tests are conducted in: 1. Laboratory tests 13 2. Field tests 10 3. Service tests 6 The specimens are exposed to the environment by: 1. Total immersion 11 2. Alternate immersion 7 3. High humidity 5 The test environments used are: 1. Sodium chloride 10 2. Magnesium chloride 5 3. Other 1 Four investigators have a written procedure and ten do not. Two have a definite test period and twelve do not. The solution makeup water is: 1. Deionized 5 2. Distilled 6 3. Tap 2 4. Seawater 2 5. Unknown 1 The grade chemicals used are: 1. Industrial 1 2. U.S. pure 2 3. Chemically pure 6 4. ASTM sea salt 1 5. Unknown 3 Ten control the temperature of the test environment and four do not. The solution is changed: 1. Weekly 2 2. Continuously 1 3. Varies 6 4. Unknown 5 Copyright Downloaded/printed University by by of ROMANS ON TEST ENVIRONMENTS AND TEST DURATIONS 203 High-Strength Steels Results o] the Questionnaire on High-Strength Steel The survey results from eight investigators on high-strength steel stress corrosion cracking are tabulated below. Tests are conducted for one or more of the following reasons: 1. Quality control 1 2. Development research 6 3. Applied research 7 4. Pure research 4 The tests are conducted in : 1. Laboratory tests 7 2. Field tests 6 3. Service tests 1 The specimens are exposed to the environment by: 1. Total immersion 6 2. Alternate immersion 4 3. High humidity 3 The test environments are: 1. Sodium chloride 5 2. Magnesium chloride 2 3. Salt spray 1 4. Other 4 Four investigators use a written procedure while four do not. Three have a definite test period and five do not. The water used for solution makeup is: 1. Deionized 2 2. Distilled 6 3. Tap 1 4. Seawater 2 The grade chemicals are: 1. U.S. pure 2 2. Chemically pure 5 3. Unknown 1 Three control the test temperature and five do not. The solution is changed: 1. Daily 3 2. Weekly 3 3. Constantly 3 4. Varies 1 Copyright Downloaded/printed University by by of 204 STRESS CORROSION TESTING Stainless Steels Results of the Questionnaire on Stainless Steels The survey by Task Group 2 of the members of Committee G-l turned up 51 investigators conducting stress corrosion tests on stainless steels. The tests are conducted for one or more of the following purposes: 1. Quality control 18 2. Development research 41 3. Applied research 42 4. Pure research 20 With five exceptions, each investigator includes more than one of the above reasons for conducting tests. Two conduct only development research, one conducts only applied research, and two pure research. The tests are conducted under three general headings: 1. Laboratory tests 47 2. Field tests 26 3. Service tests 23 Thirty-two people conduct two or three of the listed tests and seventeen use only one of the tests. The specimens are exposed to the test environment by either partial, total, alternate immersion, or high humidity. The number using each test are: 1. Partial immersion 3 2. Total immersion 39 3. Alternate immersion 20 4. High humidity 18 The greatest number of variables are the test environments. While sodium chloride and magnesium chloride appear to be the favorites, many others are used as can be noted below: 1. Sodium chloride 25 2. Magnesium chloride 30 3. Distilled water 1 4. Sodium sulfate and acetic acid 1 5. Selenium oxide and hydrochloric acid 1 6. Acetic acid and hydrogen sulfide and sodium chloride 3 7. Acetic acid and hydrogen sulfide 1 8. Hydrogen sulfide and sodium chloride.... , 1 9. Water and potassium hydroxide 650 F 1 10. Water and ferric chloride 650 F 1 11. Magnesium chloride and hydrogen peroxide 200 F 1 12. Salt spray 1 13. High purity water 1 Copyright by 14. Boiler water 1 Downloaded/printed University by of ROMANS ON TEST ENVIRONMENTS AND TEST DURATIONS 205 15. Process mixtures 2 16. Solid sodium chloride 1100-1400 F 1 17. Cathodic current in H2SO4 1 18. Seawater 5 19. Artificial seawater 1 20. Sodium hydroxide 1 With all the environments involved, it would seem that there would be wide ranging disagreements on the degree of susceptibility of various stainless steel alloys. The survey revealed that less than half the investigators use a written procedure. To be more specific, 20 use a written procedure, while 32 do not. There were also 38 of the polled who do not have a definite test period and 13 who do. The test solution makeup water shows that deionized and distilled water are the most popular. The tabulated list is given below: 1. Deionized 20 2. Distilled 24 3. Tap 6 4. Unknown 2 Most tests have temperature controls possibly due to the fact that a major portion are conducted at elevated temperatures. A total of 37 control the temperature while 16 do not. The grade chemicals used show that certified pure is most widely used. The tabulation is shown below: 1. Reagent 2 2. Certified pure 34 3. U.S. pure 13 4. Technical 1 5. Industrial 7 6. Unknown 4 The test environment was changed at intervals as shown below: 1. Constantly 5 2. Daily 8 3. Weekly 9 4. Monthly 2 5. Varies 18 6. Each specimen 2 7. Unknown 7 Most members were noncommittal on factors other than those above which affect stress corrosion tests. Some of the comments given were: sample configuration, stress application, surface preparation, specimen selection, solution flow rate, Copyright by ASTM Int'lpH. (all rights reserved); Wed Dec 16 15:53:43 EST 2015 oxygen content, and Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprod 206 STRESS CORROSION TESTING References [1] Harwood, J. J., "The Influence of Mechanical Factors on Stress Corrosion," University of Tennessee Corrosion Conference, 1-3 March 1955, University of Tennessee Press, Knoxville, Tenn. [2] ASTM-AIME Symposium on Stress Corrosion Cracking of Metals, 1944, American Society for Testing and Materials, Philadelphia, 1945. [3] Godard, H. P., "The Corrosion Behavior of Aluminum," Corrosion, Vol. 11, No. 12, Dec. 1955, pp. 542T-552T. [4] Sprowls, D. O. and Brown, R. H., "What Every Engineer Should Know About Stress Corrosion of Aluminum," Metal Progress, Vol. 81, No. 4, April and May 1962. [5] Bobylev, A. V., "Stress Corrosion Cracking of Brass," Metallurgizdat, 1955. [6]Ferryman, E. C. W. and Hadden, S. E, "Stress Corrosion of Al-7Mg Alloy," Journal of the Institute of Metals, Vol. 77, 1950, p. 207. [7] Croucher, T. R., "Northrop Norair ARTC Project 10-59." [5] Gilbert, P. T. and Hadden, S. E., "A Theory of the Mechanism of Stress Corrosion in Al-7Mg Alloy," Journal of the Institute of Metals, Vol. 77, 1950, p. 237. [9] Romanov, V. V., "Stress Corrosion Cracking of Metals, A Bibliography," The National Science Foundation, Washington, D.C. [10] Dix, E. H., Jr., Anderson, W. A., Jr., and Shumaker, M. B., "Development of Wrought Aluminum-Magnesium Alloys," Alcoa Technical Paper 14, Alcoa Research Laboratories, New Kensington, Pa., 1958. [11] Lifka, B. W. and Sprowls, D. O., "An Improved Exfoliation Test for Aluminum Alloys," Corrosion, Vol. 22, No. 1, Jan. 1966. [12] Chadwick, R. and Grainger, A. B., "Stress Corrosion of Wrought Ternary and Complex Alloys of the Al-Zn-Mg System," Journal of the Institue of Metals, Vol. 85, 1956-1957, p. 161. [13] Booth, F. F. and Godard, H. P., "An Anodic Stress Corrosion Test for Aluminum-Magnesium Alloys," International Congress on Metallic Corrosion, London, 1961, p. 8. [14] Dix, E. H., Jr., "Acceleration of the Rate of Corrosion by High Constant Stresses," Transactions, Institute of Metals Div., American Institute of Mining and Metallurgical Engineers, Vol. 137, 1940, p. 11. [15] Ailor, W. H., "World Wide Atmospheric Test Program," Corrosion Technology, Nov. 1965. [16] Aeronautical Systems Division Contract No. AF 33 (657)-8543, Douglas Aircraft Report 31421, 1 April, 1963. [17]Mears, R. B., Brown, R. H., and Dix, E. H., Jr., "A Generalized Theory of the Stress Corrosion of Alloys," ASTM-AIME Symposium on Stress Corrosion Cracking of Metals, 1944, American Society for Testing and Materials, Philadelphia, 1945, pp. 323-339. Bibliography on Titanium [1] Fontana, M. G., "Stress Corrosion in Titanium and Its Alloys," Industrial & Engineering Chemistry, Vol. 48, No. 9, 1956, pp. 59A and 60A. [2] Meredith, Russell and Arter, W. L., "Stress Corrosion of Titanium Weldments," Welding Journal Supplement, Welding Research, Vol. 36, 1957, pp. 415s—418s. [3] Kochka, E. L. and Peterson, V. C., "The Salt Corrosion of Titanium Alloys at Elevated Temperatures," Final Technical Report, 15 January, 1961, Crucible Steel Company of America. [4] Rittenhouse, J. B., "The Corrosion, Pyrophoricity, and Stress-Corrosion Cracking of Titanium Alloys in Fuming Nitric Acid," Transactions, American Society for Metals, Vol. 51, 1959, pp. 871-895. [5] Kiefer, G. C. and Harple, W. W., "Stress-Corrosion Cracking of Commercially Pure Titanium," Metal Progress, Vol. 63, No. 2, 1953, pp. 74-76. Copyright by ASTM (all rights reserved); Dec 16 2015 [6] Rittenhouse, J. Int'l B., "The Corrosion andWed Ignition of 15:53:43 TitaniumEST in Fuming Nitric Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduc ROMANS ON TEST ENVIRONMENTS AND TEST DURATIONS 207 Acid," paper presented at the Second Pacific Area National Meeting of ASTM in Los Angeles, Calif., 17-21 September, 1956. [7] O'Connor, Eugene, "Investigation to Determine the Reaction of A-110AT Titanium Alloy to Various Manufacturing Process Materials as Contaminants in an Air Atmosphere at Elevated Temperatures," Factory Laboratory Report F-4-3-59, 23 April, 1959, Solar Aircraft Co. [8]Brown, Hiram, "Stress-Corrosion of Ti-5Al-2.5Sn," Memorandum 60, 4 August, 1960, Defense Metals Information Center, Columbus, Ohio. [9] Braski, D. N. and Heimerl, G. J., 'The Relative Susceptibility of Four Commercial Titanium Alloys to Stress Corrosion at 550 F," NASA-Tn D-2011, National Aeronautics and Space Administration. [10]Covington, L. C. and Early, F. R., "Methods of Protecting Titanium Against Hot Salt Stress Corrosion," Progress Report 21, Aug. 1964, Titanium Metals Corporation of America, New York. [11] Pride, R. A. and Woodward, J. M., "Salt-Stress-Corrosion Cracking of Residually Stressed Ti-8Al-lMo-lV Brake-Formed Sheet at 550 F (561 K)," NASATM X-1082, National Aeronautics and Space Administration, April 1965. [12] Braski, D. N., "Preliminary Investigation of Effect of Environmental Factors on Salt-Stress-Corrosion Cracking of Ti-lAl-lMo-lV at Elevated' Temperatures," NASA-TM X-1048, National Aeronautics and Space Administration, Dec. 1964. [13] Heimerl, G. J. et al, "Salt Stress Corrosion of Ti-8Al-lMl-lV Alloy Sheet at Elevated Temperatures," paper presented at the Fifth Pacific Area National Meeting of the American Society for Testing and Materials, Seattle, Wash., 31 October-5 November, 1965. [14] Piper, D. E. and Fager, D. N., "The Relative Stress-Corrosion Susceptibility of Titanium Alloys in the Presence of Hot Salt," paper presented at the Fifth Pacific Area National Meeting of ASTM, Seattle, Wash., 31 October-5 November, 1965. [15] Donachie, M. J., Danesi, W. P., and Pinkowish, A. A., "Effects of Salt Atmosphere on Crack Sensitivity of Commercial Titanium Alloys at 600-900 F," Pratt and Whitney Aircraft, East Hartford, Conn. [16\ Rideout, S. P., Louthan, M. R., and Selby, C. L., "Basic Mechanisms of Stress Corrosion Cracking of Titanium," paper presented at the Fifth Pacific Area National Meeting of ASTM, Seattle, Wash., 31 October-5 November, 1965. [17] Boyd, W. K. and Fink, F. W., 'The Phenomenon of Hot Salt Stress-Corrosion Cracking of Titanium Alloys," NASA CR-117, National Aeronautics and Space Administration, Oct. 1964. [18] Avery, C. H. and Turley, R. V., "Chloride Stress Corrosion Susceptibility of High Strength Stainless Steel, Titanium Alloy, and Superalloy Sheet," MLTDR-64-44 Vol. II, May 1964. [79] Crossley, F. A., "Research on the Basic Nature of Stress Corrosion for Various Structural Alloys at Room and Elevated Temperature," ASD-TR-61-713, May 1962. [20] Logan, H. L. et al, "The Mechanism of Stress Corrosion of Titanium Alloys Exposed to Sodium Chloride at Elevated Temperatures," Report 8690, May 4, 1964, National Bureau of Standards, Washington, D. C. [21] Kirchner, R. L. and Ripling, E. J., "The Diffusion of Corrosion Products in Hot Salt Stress-Corrosion Cracking of Titanium," Materials Research Laboratory, Inc., Oct. 1965. [22] Martin, George, "Investigation of Long-Term Exposure Effects Under Stress on Supersonic Transport Structural Alloys," paper presented at the Fifth Pacific Area National Meeting of ASTM, Seattle, Wash., 31 October-5 November, 1965. [23] Newcomer, R., Tourkakis, H. C., and Turner, H. C., "Elevated Temperature Stress Corrosion Resistance of Titanium Alloys," Corrosion, Vol. 21, No. 10, Oct. 1965, pp. 307-315. [24] Braski,by D. ASTM N., "Preliminary Investigation Effect Environmental Copyright Int'l (all rights reserved); ofWed Decof16 15:53:43 ESTFactors 2015 on Salt Stress Corrosion Cracking of Ti-8Al-lMo-lV at Elevated TemperaDownloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further 208 STRESS CORROSION TESTING tures," NASA-TM X-1048, National Aeronautics and Space Administration, Dec. 1964. [25] Heimerl, G. J. et al, "Salt Stress Corrosion of Ti-8Al-lMo-lV Alloy Sheet at Elevated Temperatures," paper presented at the Fifth Pacific Area National Meeting of ASTM, Seattle, Wash. 31 October-5 November, 1965. [26]Faulkenberry, B. and lannucci, A., "Effects of Sodium Chloride on Stress Corrosion Cracking of Titanium Alloy During Stress Relieving," Convair Report MP 59-053, 19 May, 1959. [27] Myers, D. E., "DOD High Strength Titanium Alloy Sheet Research Corrosion Program," Report NA57H-527-18, 1 August, 1962, North American Aviation, Inc., Columbus, Ohio. [25] Murphy, J. F., and Carpenter, S. R., "Final Report On Alpha Titanium Manufacturing Development Sheet Program," RTD-TDR-63-4010, Oct. 1963. [29] Brown, B. F., "A New Stress-Corrosion Cracking Testing Procedure for HighStrength Alloys," paper presented at the 68th Annual Meeting of ASTM Lafayette, Ind. 13-18 June, 1965. [30] Lane, I. R., Cavallaro, J. L., and Morton, A. G. S., "Seawater Enbrittlement of Titanium," paper presented at the Fifth Pacific Area National Meeting of ASTM, Seattle, Wash., 31 October-5 November, 1965. [31] Brown, B. F. et al, "Marine Corrosion Studies, Third Interim Report of Progress," NRL Memorandum Report 1634, July 1965, Naval Research Laboratory. [32] Hatch, A. J., Rosenberg, H. W., and Erbin, E. F., "Effects of Environment on Cracking in Titanium Alloys," paper presented at the Fifth Pacific Area National Meeting of ASTM, Seattle, Wash., 31 October-5 November, 1965. [33]Dohogne, C. L. et al, "A Study of the Stress-Corrosion Cracking of Titanium Alloys in Seawater with Emphasis on the Ti-6Al-4V and Ti-8Al-lMo-lV Alloys," Research Report R471, Oct. 18, 1965, Reactive Metals, Inc., Niles, Ohio. [34] "Nitrogen Tetroxide/Titanium Alloy Stress Corrosion Investigation," Report 8271-928060, Vol. I, Bell Aerosystems Co., Buffalo, N. Y. Contract NAS 9-150, DMIC No. 65045. [35] "Nitrogen Tetroxide/Titanium Alloy Stress Corrosion Investigation," Report 8271-928060, Vol. II, Bell Aerosystems Co., Buffalo, N. Y., Contract NAS 9-150, DMIC No. 65046. [36] "Nitrogen Tetroxide/Titanium Alloy Stress Corrosion Investigation," Report 8271-928060, Vol. Ill, Bell Aerosystems Co., Buffalo, N. Y., Contract NAS 9-150, DMIC No. 65047. [37] "Nitrogen Tetroxide/Titanium Alloy Stress Corrosion Investigation," Report 8271-928060, Vol. IV, Bell Aerosystems Co., Buffalo, N. Y., Contract NAS 9-150, DMIC No. 65048. [38]Berry, W. E., White, E. L., and English, J. J., "Review of Recent Developments," Corrosion and Compatibility, Defense Metals Information Center, 30 November, 1966. [39] Mori, Kensi, Takamura, Akira, and Shimose, Takaaki, "Stress Corrosion Cracking of Ti and Zr in HCl-Methanol Solutions," Corrosion, Vol. 22, No. 2, Feb. 1966, pp. 29-31. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No furth D. A. Vaughan1 and D. I. Phalen1 Reactions Contributing to the Formation of Susceptible Paths for Stress Corrosion Cracking REFERENCE: D. A. Vaughan and D. I. Phalen, "Reactions Contributing to the Formation of Susceptible Paths for Stress Corrosion Cracking," Stress Corrosion Testing, ASTM STP 425, Am. Soc. Testing Mats., 1967, p. 209. ABSTRACT: Reactions between austenitic steels, martensitic steels, and ternary aluminum alloys have been investigated by optical and electron microscopy and by X-ray diffraction to evaluate metal-atomic hydrogen reactions as potential methods of generating susceptible paths for stress corrosion cracking of these materials. Significant structural changes are observed to result from cathodic charging treatments which have been compared with those that occur during stress corrosion. However, the reaction products are shown to be highly anodic to the uncharged metal and thus are very infrequently, if ever, observed in stress corrosion cracks. The mechanisms by which susceptible paths may be generated in stress corrosion cracking tests of these metals are discussed in terms of the observed metal-hydrogen reaction. Structural characteristics of these metals are shown to be contributing factors in the formation of these paths. KEY WORDS: corrosion, stress corrosion, cracking, steels, aluminum alloys, stainless steels The paths through which metals fail during stress corrosion have been difficult, if not impossible, to predict on the basis of preexisting metallurgical or structural characteristics. Therefore, the mechanism by which susceptible paths form prior to cracking in stress corrosion environments has been of considerable interest to the present authors for several years. Through an understanding of the phenomena for various alloys, it is anticipated that methods for preventing stress corrosion cracking will be derived and fuller utilization of the physical properties will result. Although alloy development or selection has been successful in minimizing overall external corrosion reactions, there are insufficient 1 Associate chief and metallurgist, respectively, Structural Physics Div., Bat- Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43Ohio. EST 2015 telle Memorial Institute, Columbus Laboratories, Columbus, Downloaded/printed by 209 University of Washington (University of Washington) pursuant to License Agreement. No further reproduction 210 STRESS CORROSION TESTING data on the nature of internal reactions to identify specific characteristics which lead to failure in stress corrosion environments. There are at least two structural characteristics which could provide susceptible paths: (1) lattice defects such as dislocation or stacking faults within the grains or at grain boundaries and (2) second-phase material. Either of these structural properties will undoubtedly form connecting paths through which failure may occur during stress corrosion. The density of defects is so very high in most commercial alloys that if they are the sources of susceptible paths, a corresponding high density of cracks might be expected to form during the initial exposure to stress corrosion environments. Even though this correlation does not appear to exist, the defect structure cannot be ignored, and some excellent work is being carried out by Pickering and Swann and others on the orientation of defects in material with different susceptibility to stress corrosion cracking. Alternatively, there are, in general, due to impurities or precipitation reactions, minor phases present in most commercial alloys, which provide sites for initiation of stress corrosion. A direct correlation between stress corrosion cracking and preexisting phases has not been attained. Thus, the studies to be reported here suggest possible mechanisms by which susceptible paths form during stress corrosion cracking tests for various alloys. The specific reaction investigated is that which occurs at cathodic sites, namely, the generation of atomic hydrogen on the metal surface. Experimental Approach Owing to the poor correlation between preexisting minor phases and the susceptible paths for stress corrosion cracking of most alloys, the approach taken by the authors has been directed toward understanding reactions which could result in the formation of susceptible paths during stress corrosion tests. The reactions of particular interest were those involving atomic hydrogen and the metal as would be expected to occur during aqueous corrosion. It has been established by Nielsen and others that hydrogen is produced during stress corrosion and that a rather large quantity of hydrogen is released from cracks. The hydrogen that escapes is molecular and would not be expected to react with the metal, but corrosion-produced hydrogen is initially in the atomic state, and reaction with the metal is quite probable. Although only a small amount of hydrogen pickup has been reported to result during stress corrosion cracking of metals, local concentration of hydrogen may be sufficient to produce a second phase or to alter the mechanical properties of the matrix metal. To increase the hydrogen concentration in bulk quantities of the metal, electrolysis of aqueous solutions has been employed to increase the atomic hydrogen concentration at the metal surface. This treatment increases the magnitude of the cathodic reaction that occurs at metal surCopyright by ASTM Int'l (all rights Wed Dec 16 15:53:43 2015 faces in corrosion processes. Since reserved); the anodic corrosion reaction EST is miniDownloaded/printed by of electrolytic cells, the products formed at the mized at the cathode University of Washington (University of Washington) pursuant to License Agreement. No furthe VAUGHAN AND PHALEN ON FORMATION OF SUSCEPTIBLE PATHS 211 cathode, that is the metal-hydrogen reaction products, can in some cases be retained on the metal for analysis by metallographic and X-ray diffraction methods. Through an analysis of these bulk reaction products, a better understanding of the potential cathodic reactions resulting from metal corrosion may be possible. Several types of austenitic stainless steel plus a high-strength (martensitic) steel and a high-strength aluminum alloy (Table 1) have been reacted with cathodically generated hydrogen, as described above, and the resulting products have been examined by optical and electron miTABLE 1 — Composition of alloys investigated (weight per cent). AUSTENITIC STAINLESS STEEL Alloy Commercial 304° Ab B° e'- er Ni Mn 18.8 17.7 17 6 17 3 9.4 8.0 17.8 17.9 1.6 1.5 15 2.5 N 0.023 0.006 0 036 0 010 C 0.06 0.075 0 017 0 055 Si P S Fe 0.58 0.50 0 49 0 24 0.04 0.028 0 023 0 007 0.016 0.012 0 021 0 Oil balance balance balance balance P S Fe HIGH-STRENGTH STEEL ALLOY Cr Commercial 4340° 0.80 Ni Mn Mo 1.82 0.67 0.21 C 0.41 Si 0.27 0.008 0.016 balance HIGH-STRENGTH ALUMINUM ALLOY Commercial 7079° D° 0 6 Zn Mg Cu Mn 4.36 4.05 3.14 0.65 0.23 3.01 ... Fe Cr Si 0.16 0.01 0.13 0.07 ... 0.01 Ti Al 0.02 balance balance Susceptible to cracking. Resistant to cracking. croscopy and by X-ray diffraction. Throughout these studies, a comparison was made of structural characteristics of the metals before charging with those obtained after charging. Furthermore, the comparison was extended to include structural characteristics of these materials after exposure to stress corrosion cracking environments. Results and Discussion Austenitic Stainless Steel The results on cathodic charging of austenitic stainless steels have been Copyright bypreviously ASTM Int'l (all Dec 16 15:53:43therefore, EST 2015 the present described byrights the reserved); authorsWed of this paper; Downloaded/printed discussion will bybe limited to the most pertinent results. Several alloys, University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authoriz 212 STRESS CORROSION TESTING FIG. 1—Photomicrograph of austenitic stainless steel specimens after cathodic charging for Vi hr; (a) susceptible and (b) resistant to stress corrosion cracking (X250). Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authoriz VAUGHAN AND PHALEN ON FORMATION OF SUSCEPTIBLE PATHS 213 both susceptible and resistant to stress corrosion cracking, were reacted with cathodically generated hydrogen and examined by the methods previously described. Distinct differences in the reaction products were observed between the two types of material, namely, the hydrogen formed a metastable solution in the austenite lattice for susceptible alloys but formed hydride phases immediately when reacted with the resistant alloys. This difference is revealed metallographically in Figs, la and b which show the surfaces of susceptible and resistant specimens after cathodically charging for l/2 hr. X-ray diffraction data are presented in Fig. 2 showing changes in lattice structure of the austenite due to cathodic FIG. 2—Graphical representation of X-ray diffraction charged stainless steel. results on hydrogen- charging of these materials. Plots A, D, F, and G of Fig. 2 represent single phases, which are present in varying amounts in plots B, C, and E. The hydrogen martensite indicated in plot B, so designated because of the broad diffraction lines and its formation on charging, is unstable and reverts back to austenite upon aging 8 to 16 hr at room temperature. This phase has a very small crystallite size as the breadth of the diffraction lines would suggest. Furthermore, there is no indication of its presence in the photomicrograph shown in Fig. la. The above reactions could be expected to occur to a limited extent in corrosion. In stress corrosion cracking, however, the applied stress is essential and is suspected as a contributing factor in the formation of the susceptible paths. Therefore, the cathodically charged specimens were deformed TheWed resulting Copyright by by bending ASTM Int'land (allreexamined. rights reserved); Dec 16photomicrographs 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further 214 STRESS CORROSION TESTING FIG. 3—Photomicrograph of cathodically charged stainless steel specimen after plastic deformation; (a) susceptible and (b) resistant to stress corrosion cracking (X250). Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduction VAUGHAN AND PHALEN ON FORMATION OF SUSCEPTIBLE PATHS 215 FIG. 4—Transmission electron micrographs of A1S1 steel quenched from 1600 F and tempered at temperatures indicated; (a) as quenched and (b) 400 F (X14875). Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions autho 216 STRESS CORROSION TESTING FIG. 4 (cent.)—Transmission electron micrographs of AISI steel quenched from 1600 F and tempered at temperatures indicated; (c) 800 F and (d) 1300 F (X14875). Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions autho VAUGHAN AND PHALEN ON FORMATION OF SUSCEPTIBLE PATHS 217 are incorporated in Fig. 3, which show that transgranular bands form perpendicular to the tensile direction in the susceptible alloy (Fig. 3d), but only grain boundary separation occurs in the resistant alloy (Fig. 3b). Thus, a potential mechanism for the formation of susceptible paths during stress corrosion cracking of stainless steel consists of corrosion produced hydrogen dissolving in the austenite lattice and stress-induced diffusion of this hydrogen to concentration sites, which are probably local regions of maximum tensile stress. Confirmation of the susceptibility of the transgranular bands, so generated, was established by exposing the charged and deformed steel to boiling 42 per cent magnesium chloride (MgCl2) FIG. 5—Change in microstress and crystalline domain size with tempering temperature. in which the bands were attacked very rapidly. Also, potential measurements revealed cathodically charged specimens to be 200-mv anodic to an uncharged specimen of the steel when the two are coupled in the MgCl2 solution. To evaluate the above proposed mechanism for stress corrosion cracking of the austenitic stainless steels and to design alloys resistant to failure by stress corrosion, it is recommended that factors controlling the type of reaction between atomic hydrogen and the metal need to be understood in more detail. For example, the effects of structural defects, of electron structure resulting from compositional changes, and of impurities in the metal on hydrogen solubility in the austenite lattice have not been investigated. Also, the hydrogen transport mechanism under applied stress has not been resolved. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprod 218 STRESS CORROSION TESTING FIG. 6—Transmission electron micrographs of AISI 4340 steel cathodically charged after tempering at indicated temperatures; (a) 400 F and (b) 1300 F (X14875). Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions autho VAUGHAN AND PHALEN ON FORMATION OF SUSCEPTIBLE PATHS 219 Martensitic High-Strength Steel In the case of high-strength steel (AISI 4340), the reaction with cathodically generated hydrogen is more subtle in that no second phases are produced or detected either by X-ray diffraction or by optical metallographic methods. The inability to detect a second phase by the latter method is not surprising in view of the fact that the martensitic grains are difficult to resolve by optical methods. X-ray diffraction analyses of the quenched and the 400 F tempered specimens show the domain size to be approximately 200 A units. This size was confirmed by transmission electron microscopic methods. The change in microstructure as a function of tempering temperature is shown in Fig. 4. Although the defect structure present in the as-quenched specimen (Fig. 40) appears to be partially removed by tempering at 400 F (Fig. 46), X-ray diffraction studies show little or no change in domain size. Structural changes become quite apparent by electron microscopic methods after tempering at higher temperature (Figs. 4c and d), which coincide with an increase in domain size on tempering at 600 and 800 F (Fig. 5). Included in Fig. 5 are the results of microstress analysis as a function of tempering temperature. The microstress decreases with tempering temperature, the largest change occurring in the temperature range 600 to 800 F as was observed for the domain size. The specimens described in Fig. 4 were cathodically charged and reexamined to determine structural changes resulting from the reaction between the steel and atomic hydrogen. Representative electron micrographs of the 400 and 1300 F tempered specimens after hydrogen charging are presented in Figs. 6a and b, respectively. The most interesting result in this study is the substructure produced in the 400 F tempered steel by cathodic charging. Little or no substructure is produced upon cathodic charging the 1300 F tempered steel, but the ferrite grain boundaries are widened to some extent. A comparison of Fig. 6a with Figs. 4a and b reveals that the structural defects of Fig. 4a are also present in Fig. 6a but not in Fig. 4b. This sequential investigation indicates that the defect structure, which appears to be partially removed upon tempering at 400 and 600 F, is regenerated by cathodic charging. Thus, hydrogen appears to enter defects in the martensite plates of this high-strength (284 ksi) steel, but it does not appear to enter the grains of the steel tempered at 1300 F, which is of considerably lower strength (120 ksi). As this change in defect structure is not observed upon charging the steel specimens tempered at 1300 F, compare Fig. 4d with Fig. 6b; the reaction with atomic hydrogen appears to differ from that of the higherstrength (low-temperature temper) material. In particular, the ferrite grain boundaries (Fig. 6b) are broadened by hydrogen charging. An examination of the metal surfaces," after charging with hydrogen, also reCopyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 vealed differences in reaction behavior for the steels tempered at low Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further repro 220 STRESS CORROSION TESTING FIG. 7—Electron micrographs of the surface of AISI 4340 steel cathodically charged after tempering at the temperatures indicated; (a) 400 F and (b) 1300 F (X8750). Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions author VAUGHAN AND PHALEN ON FORMATION OF SUSCEPTIBLE PATHS 221 and high temperature. The difference in surface reaction is illustrated in Fig. 7. For low-temperature tempered steel (Fig. Id), the surface structure remained unchanged, which is undoubtedly due to the easy entry of hydrogen through the defect structure in this specimen. In the case of higher-temperature tempered material, cathodic charging produced blisters on the metal surface (Fig. 76). This indicates that a lower void-space volume is present in the grain boundaries of the latter specimen than is available in the domain boundaries of the higher-strength steel. These structural changes resulting from metal-hydrogen reactions suggest that the cathodic portion of a stress corrosion reaction may be effective in altering the microstructure of this steel. These structural changes are probably restricted to the surface layers which would be in the tensile zone of stress corrosion test specimens, where the morphology of stress corrosion crack exhibits a transgranular path, namely, through the martensite plates of the high-strength steel. This path changes to an intergranular type fracture in the neutral and compressive zones of stress corrosion cracked specimens. In these latter zones, the fracture morphology indicates that the fracture path is the prior-austenite grain boundaries. In view of the above discussion on the probable effect of hydrogen in the tensile zone, it is likely that the intergranular failure in the other zones results from hydrogen diffusing to and concentrating in the prior-austenite boundaries. This is consistent with the mechanism proposed by Fetch and Stables on the lowering of grain-boundary energy by its adsorption of hydrogen. The structural modifications induced by hydrogen charging of the lower-strength (high-temperature temper) steel specimens are limited to ferrite grain boundaries, which in some respects are similar to those described above, even though failures do not occur as frequently in the lower-strength steel. This decrease in incidence of failure may be a result of partial elimination of the prior-austenite grain boundaries so that the neutral and compressive zones do not have rapid diffusion paths for hydrogen. The lack of a continuous grain boundary would not necessarily eliminate delayed failures by the metal-hydrogen reaction. However, at higher tempering temperatures, the structural defects within the martensite and ferrite grains are reduced or eliminated, so that hydrogen diffusion would be expected to be much slower. The studies of hydrogen reaction with 4340 steel specimens tempered at 800 and 1300 F indicate that hydrogen enters the ferrite boundaries formed by these heat treatments. As these grains are much smaller than the prior-austenite grains, the diffusion path would be appreciably longer. The hydrogen reaction with the surface of these specimens was found to differ from that observed for the low-temperature tempered specimens in that blisters formed on the former but not on the latter specimens. This difference was Copyright by ASTM Int'lthe (allfact rightsthat reserved); Wed Dec 16 15:53:43 ESTformer 2015 at a undoubtedly due to hydrogen diffuses into the Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further repro 222 STRESS CORROSION TESTING FIG. 8—Electron micrograph of (a) cathodically charged surface stress corrosion fracture of Al-4Zn-3Mg alloy (X8750). and (b) Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further rep VAUGHAN AND PHAUEN ON FORMATION OF SUSCEPTIBLE PATHS 223 slower rate, and the higher tempering temperature has reduced the defect concentration and thus the void space for hydrogen to enter the lattice. Improvement in the resistance of the high-strength 4340 steel to stress corrosion fracture may be attained (1) by preventing diffusion of hydrogen picked up during corrosion through eliminating the prior-austenite grain boundaries and (2) by reducing or changing the character of the defect structure within the martensite grains. FIG. 9—Electron micrograph of thin section of Al-4Zn-3Mg alloy; (a) as thinned (X12250). High-Strength Aluminum Alloy The fracture path for stress corrosion cracking of Type 7079 aluminum alloys has been established as the grain boundaries. Since the details of the failure mechanism are not completely understood, it was of interest to determine how the cathodic portion of the corrosion reaction might contribute toward understanding acceleration of the attack when stress is present. For this investigation, a special heat of Al-4Zn-3Mg ternary alloy was prepared which had failure times very nearly the same as those of commercial Type 7079 aluminum. As an initial analysis of the mechanism, the morphological characteristics of cathodically charged Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 specimens were compared with those in the fracture surface of stress Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further 224 STRESS CORROSION TESTING FIG. 9 (cent.)—Electron micrographs of thin section of Al-4Zn-3Mg alloy; (b) cathodically charged and (c) exposed to 3.5 per cent Nad solution without applied potential (X12250). Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions auth VAUGHAN AND PHAUEN ON FORMATION OF SUSCEPTIBLE PATHS 225 corrosion cracks. Figure 8 shows that the surface structures are very similar. While the structure on the surface of the charged specimen was essentially uniform from area to area, the corresponding pit structure (black dots) was observed only near the apex of the crack in the stress corrosion specimen. The remaining surfaces of the crack were relatively pit free and were similar to those observed when bulk specimens were made anodic in the test solution which contained 3.5 per cent sodium chloride (NaCl). The pits observed in the electron micrographs of Fig. 8 are believed to result from a two-step process: (1) reaction of the metal with atomic hydrogen and (2) preferential dissolution of the reaction product by the test solution. Thus, in aluminum alloys as in the case of austenitic stainless steel, the metal-hydrogen reaction products are attacked rapidly by the stress corrosion testing solutions. However, in view of the results on the steel-hydrogen reactions, it appeared that other structural changes could occur upon cathodically charging the aluminum in electrolytes other than the NaCl solution. Thin sections of the ternary alloy were prepared and 'examined (1) as thinned, (2) after cathodic charging in an arsenic saturated 5 per cent H2SO4 electrolyte, and (3) after exposure to 3.5 per cent NaCl solution in the absence of an applied voltage. Electron micrographs of thin sections of the ternary alloy in the above conditions are shown in Fig. 9. Although this alloy was given the T6 aging treatment, the grains exhibit little or no second phase, and only narrow strain lines are seen traversing the grains in the as-thinned condition. There are very small precipitates in the grain boundaries. The narrow width of the boundaries suggests that these adjacent grains have nearly the same crystallographic orientation. It would be of extreme interest to determine whether stress corrosion cracking paths propagate along high- or low-angle boundaries. After cathodic charging a thinned section of this alloy (Fig. 9Z>) was obtained. Here the pits, light spots on the print, are distributed rather uniformly over the grain except in a zone, approximately 2500 A wide, on either side of the grain boundary where the density of pits decreases. The structure within this zone differs from the matrix grain after charging in that the fine strain lines or subgrain structures that formed in this narrow zone appear to be oriented perpendicular to the grain boundary. As no stress was intentionally applied during charging, it is evident that these structural characteristics are inherent in the metal and when reacted with atomic hydrogen become detectable. This zone may result from depletion of the alloy during formation of the grain-boundary precipitate. It is quite likely that the physical and chemical properties are thus altered so that this narrow zone becomes the susceptible path for stress corrosion cracking. Figure 9c shows an electron micrograph of a thin section of this alloy Copyright by ASTM rights reserved); Wed One Dec 16 EST changes 2015 after exposure to 3.5Int'lper(allcent NaCl solution. of 15:53:43 the major Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further rep 226 STRESS CORROSION TESTING occurring as a result of this treatment is a broadening of the strain contour lines even though no stress was applied. This was also observed upon cathodic charging (Fig. 96). In addition, Fig. 9c shows that a portion of the grain boundary has been attacked. This attack appears to extend into the grain, particularly in the upper right corner of Fig. 9c, which would indicate an accelerated reaction in a narrow zone adjacent to the grain boundary. This attack switches from one side to the other along the boundary. Although the previously mentioned depleted zone is not as pronounced in Fig. 9c, there are some indications of its presence by the type of attack and by the shape of some strain contour lines as they approach the grain boundary. In this aluminum alloy, it would appear that the susceptible path may be present, but that the cathodic portion of the corrosion reaction may provide additional structural changes and thus increase the sensitivity of the grain-boundary zone to attack by the corrosion medium or to the propagation of cracks by an applied stress. Conclusions From the observations described in this research, it must be concluded that the type of metal-hydrogen reaction depends, to a large extent, upon the prior metal structure. The similarity between structural changes that occur by stress corrosion and by cathodic charging suggests that the cathodic portion of corrosion reaction may contribute significantly toward the generation of susceptible paths during stress corrosion cracking tests. In all cases, the cathodic reaction is limited to generation of the susceptible path, but an anodic reaction is necessary to produce a stress corrosion crack. It must be pointed out that the high-strength steel and, possibly, the aluminum alloy may fracture under cathodic charging alone; however, this type of failure is usually defined as hydrogen embrittlement. These alloys do not form hydride phases as does the austenitic stainless steel. The latter steel does not become embrittled, which is undoubtedly due to the short-range diffusion of hydrogen before the precipitate occurs. Acknowledgments The authors wish to acknowledge United States Steel Corp.; U.S. Navy, Bureau of Weapons; and U.S. Air Force, Materials Laboratory, Wright-Patterson Air Force Base; who contributed the support for this research. Bibliography [1] Swann, P. R. and Nutting, J., Journal of the Institute of Metals, Vol. 88, 1960, p. 478. [2] Swann, by P. R., Corrosion, Vol. 19, No. 3, 1963, p. 102t. Copyright ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further VAUGHAN AND PHALEN ON FORMATION OF SUSCEPTIBLE PATHS 227 [5] Pickering, H. W. and Swann, P. R., Corrosion, Vol. 19, No. 11, 1963, p. 373t. [4] Nielsen, N. A., Corrosion, Vol. 20, No. 3, 1964, p. 104t. [5] Vaughan, D. A. et al, Corrosion, Vol. 19, No. 9, 1963, p. 315t. [6]Vaughan, D. A. and Phalen, D. L, Metals Engineering Quarterly, Vol. 5, No. 3, 1965, p. 39. [7] Haynie, F. H. et al, 66-267, 1966, Air Force Materials Laboratory. [8] Fetch, N. J. and Stables, P., Nature, Vol. 169, 1952, p. 842. DISCUSSION Daniel van Rooyeri^ (written discussion)—It was interesting to hear of an 18Cr-8Ni stainless steel which is immune to stress corrosion cracking. Can the authors give some details of the tests in which the alloy was immune and also any metallurgical or other reason for this resistance to cracking in an austenitic 18Cr-8Ni alloy? Also, details of the 18Cr-18Ni alloy, which was mentioned, would be welcome. Was the electrochemical potential, used for charging specimens with hydrogen, in the range which may be expected at local cathodes during stress corrosion cracking of austenitic stainless steels? D. A. Vaughan and D. I. Phalen (authors)—The compositions of the stainless steel alloys used in this investigation, as given in Table 1, show the major difference to be in nitrogen content which is significantly higher for susceptible than for resistant material. The susceptibility to stress corrosion cracking was established during exposure to boiling 42 per cent MgCl2 solution for times up to 500 hr with some tests extending to 8000 hr. Specimens of these materials were used to determine the structural details of the metal-hydrogen reaction. The electrochemical potential used to generate hydrogen at the metal surface was of the order of 5 v, which may have exceeded that at local cathodes during stress corrosion as the absolute value is not well known. However, the present study was intended to produce sufficient reaction product to be detected by X-ray diffraction rather than just enough to initiate one or a few susceptible paths. Section supervisor, The International Nickel Co., Inc., Paul D. Merica Research Laboratory, Sterling Forest, Suffern, N. Y. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduct E. N. Pugh1 andA.R.C. Westwood1 Critical Species in Stress Corrosion Phenomena REFERENCE: E. N. Pugh, and A. R. C. Westwood, "Critical Species in Stress Corrosion Phenomena," Stress Corrosion Testing, ASTM STP 425, Am. Soc. Testing Mats., 1967, p. 228. ABSTRACT: Consideration has been given to the identification of the critical species in several stress corrosion systems. It is shown that in the a-brass/aqueous ammonia system, cupric complex ions of the type Cu(NH3)n2+ play a controlling role in the cracking process. Complex ions are also found to constitute the critical species in the embrittlement of silver chloride in certain aqueous environments. In the case of materials such as stainless steels and magnesium and aluminum alloys, which undergo stress corrosion cracking in chloride environments, the critical species may be the chloride ion itself or metal-chloride complexes. Attention is given to both the role of the critical species in the mechanisms of failure and the practical significance of these findings to stress corrosion testing. It is suggested that more attention to the chemistry of environments which cause stress corrosion cracking, with particular regard to the identification of the critical species, could be of significant practical value. KEY WORDS: corrosion, stress corrosion, cracking, brass, stainless steels, magnesium alloys, aluminum alloys It is generally accepted that, for any given material, stress corrosion phenomena occur only in certain specific environments. Thus, it is somewhat surprising that relatively little attention has been given to the chemistry of these environments, particularly to the identification of the particular species responsible for failure. Nevertheless, it is readily apparent that if the nature of the critical species and its role in the cracking process can be established, then the possibility of being able to suggest means of preventing or inhibiting stress corrosion cracking is much improved. This paper describes the results of recent studies which have led to the identification, and in one instance control, of the critical species in certain systems, and then discusses the implications of this work with regard to 1 Staff scientist and associate director, respectively, Research Institute for AdCopyright by ASTM Int'lMarietta (all rights Corp., reserved); Wed Dec 16 15:53:43 EST 2015 vanced Studies, Martin Baltimore, Md. Downloaded/printed by 228 University of Washington (University of Washington) pursuant to License Agreement. No further reproductions PUGH AND WESTWOOD ON CRITICAL SPECIES 229 the mechanisms of stress corrosion cracking and to the general problems of stress corrosion testing. For the purposes of this paper, the term stress corrosion cracking is considered, quite arbitrarily, to encompass the embrittlement of crystalline solids in aqueous environments. The term is also used in a generic sense, for it is becoming increasingly evident that no single unified mechanism of failure exists. Indeed, more than one mechanism may operate FIG. 1—Stress corrosion data for 70-30 brass specimens stressed in fresh or preconcentrated 15 N aqueous NHs. The copper exists in the preconcentrated solutions as C^NHs)^ ions [5]. in any given material/environment system. For example, it will be seen in the following section that in the classical a-brass/ammonia system at least two mechanisms are thought to operate, the determining factor being the chemical composition of the environment [I].2 The a-Brass/Aqueous Ammonia System An important contribution to our understanding of the role of the ammoniacal environment in the stress corrosion cracking of a-brass was 2 The italic numbers in brackets refer to the list of references appended to this paper. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduction 230 STRESS CORROSION TESTING made by Althof [2] in 1944, when he demonstrated that cracking did not occur until the testing solution turned blue and that stress corrosion life was markedly reduced when the test solutions were preconcentrated with copper before the test commenced. The significance of these observations has become apparent following their confirmation in more recent studies [1,3-6], for example, Fig. 1. It is known that in aqueous, oxygenated ammoniacal solutions, copper exists as cupric complex ions of the type Cu(NH3)n2+, the number of ammonia ligands, n, varying from 1 to 5 depending on the ammonia concentration of the solution FIG. 2—Effect of varying the volume of the testing solution on the time-tofailure of 70-30 brass specimens stressed in fresh or preconcentrated 15 N aqueous NH3 [5]. [7]. The presence of these complex ions gives rise to the blue color of the solution. If it is then assumed that such complex ions play an important role in the stress corrosion process, the differences in time-to-failure of specimens tested under a given stress in either "fresh" or preconcentrated solutions (Fig. 1) may be attributed to the fact that time is required in fresh solutions for the production, via dissolution of the specimen, of a sufficient concentration of these complex ions. In preconcentrated solutions, of course, this critical ion species is present at the beginning of the test. Data from experiments in which the volume of the testing solution was varied (Fig. 2) also may be explained on the basis of this hypothesis. In fresh solutions, the time required to build up a sufficient concentration ot complex ions to cause failure increases with increasing volume; in Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 preconcentrated solutions, time-to-failure is independent of volume [5]. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No furt PUGH AND WESTWOOD ON CRITICAL SPECIES 231 Some differences of opinion appear to exist concerning the exact complex ion which is responsible for failure. Complex ions containing 1.5 to 3.0 [8] and 4 [2-4] ammonia ligands each have been claimed to play major roles in the stress corrosion process, but these claims have not been substantiated by experimental evidence. Recently, Pugh et al [5] FIG 3 - —Effect of copper content of 15 N aqueous NH* on (a) time-to-failure and (b) rate of weight loss of 70-30 brass. In the stress corrosion data, the points represent the average of at least four tests and the bars indicate the highest and lowest values [1]. have combined studies of stress corrosion cracking of a 70-30 brass with spectrophotometric studies of the ammoniacal testing solutions. Experiments conducted in 15 N and 1 N solutions preconcentrated with copper established that rapid cracking occurred in each solution. Comparison of the absorption spectra of the test solutions with published spectroCopyright by data ASTM [7,9] Int'l (all rights reserved); Dec 16 15:53:43 EST 2015 photometric indicated that theWed predominant complex species Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further repro 232 STRESS CORROSION TESTING were Cu(NH3)52+ and Cu(NH3)42+, respectively. Work is in progress to determine whether cracking can occur when complexes containing fewer ligands predominate. To investigate the role of the complex ions in the cracking process, Pugh and Westwood have studied the effect of varying the complex-ion concentration on the behavior of 70-30 brass tested in oxygenated 15 N aqueous ammonia [1]. The time-to-failure, tF, rate of weight loss, and surface condition were each found to depend on the complex-ion concentration, in this case Cu(NH3)52+. Figure 3a illustrates the relationship between tF for specimens tested under a constant load and copper content of the solution which, in these experiments, was directly proportional to the concentration of Cu(NH3)52+ ions. It can be seen that tF decreased with increasing copper content, as expected from the earlier results, for example, Fig. 1, but that a well-defined inflection occurred at a critical concentration. This inflection coincided with a change in surface condition. Specimens tested in solutions of copper content exceeding the critical value were coated with the characteristic black oxide coating commonly termed the tarnish, while specimens tested in solutions of lower concentrations were apparently free from this coating. The relationship between rate of weight loss and copper content of the solution also exhibited a maximum at this critical concentration (Fig. 3b). These and other observations have led to the conclusion that two mechanisms of stress corrosion cracking are operative in this system, one in the presence of the tarnish and the other in the absence of this layer [1]. The inflection in the stress corrosion data (Fig. 3a) is considered to correspond to the transition between these mechanisms. The mechanism of stress corrosion cracking in solutions which do not cause tarnishing is not fully understood, but it is currently considered to occur by a dissolution-dependent mechanism involving the following autocatalytic reaction between the cupric complexes and copper atoms at the brass surface.3 Since this reaction does not involve zinc, then it might be expected that cracking also should occur in pure copper. Earlier work suggested that the pure metal is immune to this type of failure [10, 11], but recent studies have demonstrated that cracking does in fact occur provided that the ammoniacal solutions contain certain critical concentrations of Cu(NH3)52+ ions [12]. 3 This reaction accounts for the initial increase in the rate of weight loss with Copyrightcopper by ASTM Int'l (all reserved);(Fig. Wed 3b). Dec The 16 15:53:43 EST 2015 increasing content of rights the solution maximum in the curve ocby of the reaction does not occur at tarnished surfaces, dissolucursDownloaded/printed because stage (a) tionUniversity then proceeding mainly by reaction of ammonia withtothe oxideAgreement. [1], of Washington (University of Washington) pursuant License No further reproductions au PUGH AND WESTWOOD ON CRITICAL SPECIES 233 In tarnishing solutions, stress corrosion cracking is thought to occur by the tarnish-rupture mechanism first proposed by Forty and Humble [13]. According to this mechanism, cracking is confined to the tarnish and proceeds by the repeated formation and rupture of this brittle layer (Fig. 4). Evidence for this mechanism is convincing. For example, McEvily and Bond [14] have shown that the fracture surfaces of specimens FIG. 4—Schematic representation of the tarnish-rupture mechanism, (a)-(f), and resulting fracture surface (g). stress corroded in tarnishing solutions exhibit striations perpendicular to the direction of crack propagation. This establishes the discontinuous nature of the fracture process and is thus fully consistent with the tarnish-rupture model (Fig. 4g). Further, this mechanism predicts that cracking should also be produced by repeated cycles of immersion of an unstressed specimen in a tarnishing environment followed by stressing Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 inDownloaded/printed the absence of bythe environment. Pugh et al [15] have demonstrated that intercrystalline failure, identical in appearance occurring in No University of Washington (University of Washington) pursuanttoto that License Agreement. further repr 234 STRESS CORROSION TESTING standard stress corrosion tests, can in fact be produced by this procedure. This observation would appear to invalidate the claim that stress corrosion cracking requires the simultaneous action of stress and corrosive attack [16]. The cupric complex ions are considered to play a controlling role in the growth of the tarnish [7]. This layer, which consists largely of cuprous oxide [6,13], has been reported to exhibit a parabolic rate of growth [14]. Accepting the general view that the oxide is cation deficient [17], then it would appear that the rate of tarnish growth is controlled by the diffusion of cuprous ions across the oxide film, which is in turn dependent on the concentration of cation vacancies. The presence of zinc in the solid has been found to be a prerequisite for tarnishing [7]. It is probable that zinc exists in the oxide lattice as the divalent ion, occupying cation sites. To maintain electrical neutrality, each divalent zinc ion requires the presence in the lattice of a cation vacancy, so that it might be argued that the role of zinc in promoting tarnishing is simply to introduce these vacancies. However, if this were the case, then tarnishing in brass would be expected to occur in many oxygenated aqueous solutions, whereas, in fact, tarnishing has been reported only in ammoniacal solutions containing a sufficient concentration of cupric complex ions (Fig. 3). Accordingly, it has been proposed that the role of the complex ions is to react preferentially with zinc ions at the oxide-solution interface, and that the preferential removal of zinc in effect results in the injection of vacancies into the oxide [7]. The presence of these vacancies is considered to be primarily responsible for the high rate of cation diffusion necessary for rapid tarnish growth. Further work is in progress to determine the details of this model, but it is interesting to note that it is supported by recent studies using an electron microprobe analyzer, which established that the tarnish is severely depleted with respect to zinc [18]. The conclusion that two mechanisms of failure are operative in the a-brass/ammonia system, both dependent on the concentration of the cupric complex ions, raises the question of which is responsible for commonly observed service failures. While it is possible that both may occur under different conditions, it appears probable that failures in moist industrial atmospheres, that is, season cracking, proceed by the tarnishrupture mechanism.4 Under these conditions, shallow layers of adsorbed water formed on the surfaces readily pick up oxygen and ammonia from the atmosphere. The small volume of solution leads to the formation of large concentrations of cupric complex ions which, in the presence of a tensile stress, cause rapid cracking (Fig. 2). It is probable that the complex-ion concentrations produced in these circumstances would be suffi* Other copper-base alloys, not become tarnished, would EST be expected Copyright by ASTM Int'l (all which rights do reserved); Wed Dec 16 15:53:43 2015 to Downloaded/printed fail by the dissolution-dependent mechanism. by University of Washington (University of Washington) pursuant to License Agreement. No further re PUGH AND WESTWOOD ON CRITICAL SPECIES 235 cient to cause tarnishing. This view is supported by the observation that when specimens are immersed in fresh 15 N aqueous ammonia, removed, and then stressed in air while the surfaces are still wet (that is, corresponding to tests in small volume), they rapidly become tarnished, and cracking is observed [5]. If it is accepted that season cracking occurs by the tarnish-rupture mechanism, then it is apparent that tests for determining the susceptibility of brasses to this failure should employ conditions which lead to tarnishing. The long standing practice of simulating practical conditions FIG. 5—Effect of applied stress on time-to-failure of poly crystalline AgCl tested in aqueous NaCl solutions at room temperature [21]. by carrying out tests in air containing controlled partial pressures of water vapor and ammonia [11,19] probably satisfies these conditions provided that the adsorbed aqueous film does not become excessively thick. However, such tests can be carried out more simply and reproducibly by totally immersing the stressed specimens in tarnishing ammoniacal solutions. The studies by the authors suggest that oxygenated 15 N aqueous ammonia preconcentrated with >3 g/liter copper is suitable (note the reproducibility of data in the tarnishing range in Fig. 3d). Alternatively, the tarninshing solutions described by Mattsson [20], containing 0.05 g-atom/liter copper, 1 g-mole/liter ammonia, and of pH about 7, may Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 be employed. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprodu 236 STRESS CORROSION TESTING FIG. 6—Polycrystalline AgCl deformed in 6 N aqueous NaCl presaturated with AgCh3' ions, demonstrating initiation and subsequent growth of intercrystalline cracks. Transmitted light [23]. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further repr PUGH AND WESTWOOD ON CRITICAL SPECIES 237 Silver Chloride/Aqueous Chloride Systems The nonmetal silver chloride has been observed to undergo stress corrosion cracking in certain aqueous solutions [21-26]. For example, Fig. 5 illustrates the relationship between time-to-failure, tF , and engineering stress for polycrystalline specimens tested under constant tensile load either in air or in aqueous sodium chloride solutions of various concentrations [27]. Fracture in air and in solutions of concentration <3 N was ductile and transcrystalline, while brittle intercrystalline failure occurred in solutions of concentration >3 N. It can be seen from Fig. 5 that, as in the stress corrosion cracking of many metals [75] (compare Fig. 9), tF increased with decreasing stress and that a limiting stress existed below which cracking did not occur in finite times. The limiting stress was dependent on the chloride-ion concentration of the solution. The solubility of silver chloride in aqueous chloride environments is dependent on the concentration of chloride ions in solution. In water at room temperature, the solubility is about 10~5 N, whereas in 6 N sodium chloride solution it is about 10~2 N', the increase in solubility with increasing chloride-ion concentration is due to the formation of highly soluble complex ions, such as AgQ2~, AgCl32~, and AgQ43~ [27]. The importance of the most highly charged of these to the stress corrosion process can again be demonstrated by preconcentration experiments. For example, Fig. 5 illustrates the marked reduction in tF at a given stress caused by presaturating 6 N sodium chloride solutions with AgCl43~ ions. Conversely, increasing the volume of fresh 6 N sodium chloride solutions results in a significant increase in tv at a given stress (Fig. 5) [27]. Metallographic studies of specimens stressed in 6 TV sodium chloride solution presaturated with AgCl43~ ions have indicated that cracking is initiated where slip bands are arrested at grain boundaries of large misorientation [23]. For example, in Fig. 6a, intercrystalline cracks (arrowed), which appear dark in transmitted light, have been initiated at each of the blocked slip bands A, B, and C. Further cracks are formed at other blocked bands during subsequent stressing (Figs. 6b to e). Cracking is not observed when the stress field associated with the arrested slip bands is relieved by deformation in the neighboring grain, for example, Fig. 6a at S. Once initiated, an intercrystalline crack propagates in a relatively brittle fashion provided that the grain boundary containing it is approximately perpendicular to the tensile axis (note in Fig. 6 that crack propagation occurs more rapidly at L to K, where the grain boundary is approximately perpendicular to the tensile axis, T.A.) and that the embrittling solution is present at the crack tip. Tests on monocrystals have established that unnotched specimens are essentially immune to cracking in sodium chloride solutions [27,22], Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 but that deeply notched specimens undergo severe embrittlement [23,24]. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further repro 238 STRESS CORROSION TESTING Thus it may be concluded that embrittlement is not a consequence of some inherent property of the grain boundary, but that the boundaries play an important role by acting as barriers to glide dislocations, hence causing stress concentrations and facilitating crack initiation. Embrittlement of polycrystalline silver chloride has also been observed in other complex-forming solutions [21,25,26], The complex ions formed in these solutions can be positively or negatively charged, but the magnitude of the charge must be > 1 + or >2—; solutions in which complex ions of lower charge are produced do not cause embrittlement (Table 1). The degree of embrittlement has been found to increase with concentration of the critical species in the environment and with the charge on the complex ion (Table 1) and to be a function of the distribution of TABLE 1 —Relationship between charge on complex ions and their ability to embrittle polycrystalline silver chloride [21,25,26]. Solution 10 N NH4OH <0.1 N AgNO3 17 N AgNO3 0.1 AT NaCl 1 N NaCl 11 NCsCl 5.8 N KC1 6 N NaCl 11.8 NHCl 11.8 Af HC1, presaturated with CuCl2 18.8 ATLiCl 8 ATNaBr 20 AT LiBr 17 ATNaSCN 6 N NazSjOa Complex Ion AgCNIW AgsCl* Ag,Cls+ AgCl2AgCla2(AgCl43-)/nCs»+ (AgCl4*-)mK»+ AgCI4*AgCtfCuCls3AgCl4*AgBr4»AgBr4*~ Ag(SCN)4»AgC&O.,),*- Charge Embrittlement 1+ 1+ 2+ 12<3<333- no no yes no no no no yes yes 3 33335- yes yes yes yes yes yes the charge on the complex ion [25]. It is of interest to determine whether a "common" ion need be present in an adsorbed complex species to cause embrittlement of a given ionic-covalent solid. Work in progress suggests that this is not necessary, for adsorbed AgGU3" complexes have been shown to cause a marked reduction in the surface microhardness and, therefore, in the strength of magnesium oxide monocrystals [27]. The practical implication of this observation is that complex species produced by the dissolution of one structural component in a given environment might be responsible for the subsequent embrittlement of another material exposed to the same environment. It has been proposed that the embrittlement of silver chloride is associated with the adsorption of highly charged complex ions [21,25]. In earlier papers, the authors suggested that the presence of the adsorbed Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprodu PUGH AND WESTWOOD ON CRITICAL SPECIES 239 complex ions in the vicinity of strained surface bonds, for example, at sites where glide dislocations are arrested, or at crack tips, might perturb the distribution of bonding electrons, reducing bond strength and hence allowing rupture at reduced stress levels [21,25]. However, current work FIG. 7—Relationship between time-to-failure and temperature, T K, for polycrystalline AgCl tested at various stresses in aqueous solutions containing (a) positive and (b) negative complex ions [28]. favors an alternative mechanism involving the formation of adsorptioninduced charge double layers [25]. The predominant charge carriers hi silver chloride are Frenkel defects, that is, interstitial silver ions and silver-ion vacancies [29]. Thus the adsorption of negatively charged complex ions might be expected to induce a compensating positively charged surface layer in the solid, containing an excess concentration of niter- Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further re 240 STRESS CORROSION TESTING stitial silver ions, and a more diffuse, negatively charged layer, containing a greater than equilibrium concentration of silver-ion vacancies, extending into the bulk. Adsorption of positively charged complex ions would be expected to produce a double layer of opposite sense. FIG. 8—Inhibition of stress-corrosion cracking in poly crystalline AgCl in 6 N aqueous chloride environments by replacement of Na* ions by K+ or Cs+ ions. Applied stress 600 g/mm2 [25]. The presence of large concentrations of point defects in the surface layers might be expected to cause significant surface hardening, possibly leading to the formation of a brittle surface layer. Then, by analogy with the tarnish-rupture mechanism (Fig. 4), it might be imagined that cracking would proceed by the formation and rupture of these defect-hardened layers. Such a process might be expected to lead to striated fracture surfaces. Fractographic studies, using the optical microscope, have established that the fracture surfaces of embrittled monocrystals do in fact exhibit striations which are perpendicular to the direction of crack propagation [23]. Similar markings have been observed in intercrystalline fracture surfaces, but in this instance they were confined to certain regions [28]. Failure to detect striations in all regions of the fracture sur- Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further repr PUGH AND WESTWOOD ON CRITICAL SPECIES 241 faces may be due simply to the limited depth of focus and limited resolution of the optical microscope. Electron-microscope (replica) studies are being undertaken to clarify this point. Recent investigations of the temperature dependence of the fracture process provide strong support for this "film-rupture" model. According to this mechanism, the rate of cracking should be controlled by the rate of formation of the embrittled double layer ahead of the crack and hence by the rates of diffusion of the point defects. Studies of the effect of temperature on tF at a given stress have established that the relationship between log tF and l/T (note that tF is inversely proportional to the rate of cracking) is linear for specimens tested in solutions containing either negative or positive complex ions (Fig. 7). Calculations of activation energies from these data yield values of 0.08 and 0.39 ev, respectively, in reasonable agreement with reported values of the activation energies for the motion of interstitial silver ions, 0.05 to 0.15 ev, and silver-ion vacancies, 0.27 to 0.37 ev [29-37]. In addition, evidence for the postulated increase in surface hardness has been obtained from indentation studies. The identification of highly charged complex ions such as AgCl43~ as the critical species in the embrittlement of silver chloride allows for the possibility of inhibiting this phenomenon. Clearly, from Table 1, any addition to the environment which causes an effective reduction in the charge of embrittling complexes is likely to inhibit cracking. Thus the successive replacement of sodium ions in 6 N chloride solutions by either potassium or cesium ions has been found to be an effective means of preventing embrittlement [25] (Fig. 8). Such inhibition is believed to result from the formation of mixed complexes of lower charge (Table 1). Additions of Group III B cations Zn2+, Cd2+, or Hg2+ to 6 N sodium chloride solutions also inhibit embrittlement. In this case, inhibition was attributed to competition of these ions for chloride ions, again resulting in the formation of complexes of charge <3 — [25], for example. Such observations suggest the possibility of "built-in" inhibitors for systems such as these. Specific alloying elements could be incorporated in the solid which, following release to the environment during dissolution, would associate with or cause breakdown of the potentially embrittling species, thus preventing embrittlement. Other Systems In the systems discussed in the preceding sections, the critical species responsible for cracking are produced by dissolution of the material in the environment. Thus, in each case, preconcentration of the environment with critical species before the start of the test causes significant reductions Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprod 242 STRESS CORROSION TESTING in stress corrosion life. To investigate the generality of this effect, preconcentration experiments have been extended to several commercial alloys tested in aqueous chloride containing environments. The solutions were presaturated by stirring with filings taken from the alloy under test: AZ31 Magnesium (3Al-lZri)—Tension specimens were tested over a range of engineering stresses in solutions containing 35 g/liter sodium chloride and 30 g/liter sodium chromate. Times-to-failure for fresh and presaturated solutions were not significantly different (Fig. 9). 2024 Aluminum (4.5Cu-1.5Mg-0.6Mri)—Age-hardened tension spec- FIG. 9—Effect of applied stress on time-to-failure of AZ31 magnesium alloy stressed in fresh and presaturated NaCl-NazCrO*. imens were tested at an engineering stress of 17 kg/mm2 in solutions containing 53 g/liter sodium chloride, 50 g/liter sodium chromate, and hydrochloric acid (pH about 0.2). Specimens tested in fresh solutions failed after an average of about 160 sec compared to about 2100 sec for those tested in presaturated solutions. 304 Stainless Steel (18Cr-8Ni)5—Specimens were tested at a stress of 42 kg/mm2 in boiling 42 per cent magnesium chloride. Failure occurred after an average of 1200 sec in both fresh and presaturated solutions. The fact that presaturation of the test environments did not produce detectable reductions in tp for these systems indicates either that a sufficient concentration of the critical species is produced rapidly, or that 5 These by testsASTM were Int'l performed by reserved); E. E. Denhard, Jr.,16 Armco SteelEST Corp., Copyright (all rights Wed Dec 15:53:43 2015Baltimore, Md. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further re PUGH AND WESTWOOD ON CRITICAL SPECIES 243 the critical species is not generated by specimen dissolution but exists in fresh solutions. In the latter event, the chloride ion itself would appear to be the most probable species. While the mechanism or mechanisms of stress corrosion cracking are not understood in these systems, it is nevertheless interesting to speculate on the possible roles of this ion. Chloride environments are known to cause pitting corrosion in each of these alloy systems [52]. Moreover, recent studies suggest that pitting in aluminum results from the incorporation of the chloride ion in the protective oxide film, reducing the resistance of this film to the diffusion of ions [33]. By analogy with the case of the a-brass/ammonia system discussed above, it might be proposed that stress corrosion cracking in these systems proceeds by a similar oxide-rupture mechanism (Fig. 4) and that FIG. 10—Schematic representation of the mechanism of transcrystalline stress corrosion cracking proposed by Pickering and Swann [47] and Swann and Embury [48]. the role of the chloride ion is to modify the defect structure of the surface oxide, permitting its growth to brittle dimensions. Recent fractographic evidence might be considered to support this view, since stress corrosion fracture surfaces for each of these alloy systems exhibit striations similar to those observed in the case of a-brass and silver chloride [34-37]. It is evident, however, that further study is necessary to confirm that the oxide-rupture mechanism is operative in these systems. The observation that chloride ions adsorb strongly on stainless steel [35] suggests an alternative mechanism. It has been proposed that stress corrosion cracking in metals may occur by an adsorption-dependent mechanism [39-41], similar to that proposed for liquid-metal embrittlement of metals [42,43], but at the present time there appears to be no unambiguous evidence to support this view. On the other hand, the presence of strongly adsorbing species, acting as step poisons, may be releCopyright by ASTM Int'l (all rights reserved); 15:53:43 EST 2015 on vant to stress corrosion cracking by virtueWed of Dec their16significant effects Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further rep 244 STRESS CORROSION TESTING dissolution behavior [44-46]. To illustrate, in the mechanism for transcrystalline stress corrosion cracking proposed by Pickering and Swann [47] and Swann and Embury [48], it is postulated that cracking occurs by the ductile rupture of a slot weakened by many corrosion tunnels (Fig. 10). Such tunnels have been observed in Type 301 stainless steel exposed to 42 per cent magnesium chloride at 140 C [47]; magnesium-7 per cent aluminum exposed to sodium chloride-potassium chromate solutions [47], aluminum exposed to aqueous sodium chloride [49], and also in copper-25 per cent gold exposed to 10 per cent ferric chloride solution [47]. Little attention has been given to the role of the chemistry of the environment in the formation of the observed corrosion tunnels. However, similar tunnels have been observed in lithium fluoride following immersion in a slightly corrosive aqueous environment containing strongly adsorbing step poisons, for example, fatty acid molecules [46] or ferric fluoride complexes [45]. Tunnelling did not occur in the absence of such poisons. Westwood and Rubin [46] have suggested that tunnels form and grow because the dissolution process is less efficiently inhibited at the bottom of a pit or growing tunnel than at the external surface. This is a consequence of differences in the concentration of the step poison at the two sites. Westwood [50] has suggested that tunnel corrosion in metals may also be associated with the presence of strongly adsorbing step poisons. The adsorbing species may be simply the chloride ion itself, or, alternatively, a metal-chloride complex ion. The latter could be produced by dissolution of the material. Since concentrations of only 10~6 N of a sufficiently active step poison can markedly affect the dissolution behavior of a solid in a slightly corrosive environment, then the formation of the necessary concentration of metal-halide complexes would occur rapidly, thus rationalizing the absence of a detectable difference in tv in the tests carried out in fresh and presaturated solutions. For the same reason, the metal ion in the complex need not necessarily be one of the major alloying constituents of the alloy. It could be provided, for example, by dissolving inclusions. In this instance, tunnelling or pitting would be expected to be first observed in the vicinity of such impurity particles, as is often the case [48]. It would also follow that for a material cracking by a tunnelling mechanism, minor alloying additions could significantly influence stress corrosion life. Such an effect has been observed for a high-purity stainless steel containing 20 per cent nickel and 20 per cent chromium [51]. In this instance, the addition of 1 to 2 per cent molybdenum to the steel decreased the life in boiling 42 per cent magnesium chloride solution at a given stress from 56 to 9 hr. It should be possible, in terms of this approach, to reduce the degree of susceptibility to cracking by Copyright by ASTM Int'l (alltorights Wed Dec 16 15:53:43 2015or cause adding another element the reserved); solid, this being chosen toEST screen Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduct PUGH AND WESTWOOD ON CRITICAL SPECIES 245 breakdown of the "embrittling" complex species in solution (the "builtin" inhibitor approach). Thus it is interesting to note that additions of 2 per cent copper to the high-purity stainless steel increased the life to 139 hr [57].6 Conclusions It may be concluded from the above discussion that, in certain systems, the presence of critical ion species in the testing environment is primarily responsible for stress corrosion failure. In these systems, the critical species were produced by dissolution of the specimen during testing. From the standpoint of stress corrosion testing, realization of this point is most important, for in such systems variable test results will be obtained if the specimens are exposed to the environment for different times before stressing, or if testing solutions are reused. The volume of the testing solution used also will be a factor (Fig. 2). On the other hand, if the critical species can be identified, and the testing environments preconcentrated with this species, then extremely reproducible test data can be expected, as was illustrated in Fig. 3a. At the present time, however, knowledge of the chemical species existing in solution in most of the stress corrosion systems of technological interest is extremely limited. It is suggested that increased attention to this aspect of stress corrosion could be of immediate practical value to both stress corrosion testing and the more important problem of preventing stress corrosion cracking. A cknowledgment The authors are pleased to acknowledge financial support received from the U. S. Army Research Office (Durham) and the Office of Naval Research. References [1] Pugh, E. N. and Westwood, A. R., "Complex Ions and Stress-Corrosion Cracking in a-Brass," Philosophical Magazine, Vol. 13, 1966, pp. 167-183. [2] Althof, F. C., "Inter- and Intracrystalline Corrosion and Its Causes," Zeitschrift fur Metallkunde, Vol. 36, 1944, pp. 177-186. [3] Graf, L. and Richter, W., 'The Problem of Stress Corrosion of Homogeneous Solid Solutions," Zeitschrift fur Metallkunde, Vol. 52, 1961, p. 833. [4] Pugh, E. N. and Westwood, A. R., "Stress-Corrosion Cracking in Brass," High-Strength Materials, Wiley, New York, 1965, pp. 701-704. [5] Pugh, E. N., Montague, W. G., and Westwood, A. R., "On the Role of Complex Ions in the Season-Cracking of Alpha-Brass," Transactions, American Society for Metals, Vol. 58, 1965, pp. 665-671. [6] Hoar, T. P. and Booker, C. J., "The Electrochemistry of the Stress-Corrosion Cracking of Alpha-Brass," Corrosion Science, Vol. 5, 1965, pp. 821-840. 8 Note that in terms oxide-rupture mechanism, effects such as these may Copyright by ASTM Int'lof(allthe rights reserved); Wed Dec 16 15:53:43 EST 2015 be ascribed to changes in the defect structure of the oxide. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduc 246 STRESS CORROSION TESTING [7] Cotton, F. A. and Wilkinson, G., Advanced Inorganic Chemistry, Interscience, New York, 1962, p. 756. [8] Johnson, H. E. and Leja, J., "On the Potential/pH Diagrams of the Cu-NHrHsO and Zn-NHy-H2O Systems," Journal of the Electrochemical Society, Vol. 112, 1965, pp. 63 8-641. [9] Jorgensen, C. K., Absorption Spectra and Chemical Bonding in Complexes, Pergamon, London, 1962, p. 286. [10] Edmunds, G., "Season Cracking of Brass," ASTM-AIME Symposium on Stress-Corrosion Cracking of Metals, 1944, American Society for Testing and Materials, Philadelphia, 1945, pp. 67-89. [11] Thompson, D. H. and Tracy, A. W., "Influence of Composition on the StressCorrosion Cracking of Some Copper-Base Alloys," Transactions, American Institute of Mining, Metallurgical and Petroleum Engineers, Vol. 185, 1949, pp. 100-109. [12] Pugh, E. N., Montague, W. G., and Westwood, A. R., "Stress-Corrosion Cracking of Copper," Corrosion Science, Vol. 6, 1966, pp. 345 and 346. [13] Forty, A. J. and Humble, P., 'The Influence of Surface Tarnish on the StressCorrosion of a-Brass," Philosophical Magazine, Vol. 8, 1963, pp. 247-264. [14] McEvily, A. J., Jr., and Bond, A. P., "On the Initiation and Growth of StressCorrosion Cracks in Tarnished Brass," Journal of the Electrochemical Society, Vol. 112, 1965, pp. 131-138. [15] Pugh, E. N., Montague, W. G., and Westwood, A. R., "On the Mechanism(s) of Stress-Corrosion Cracking," Environment-Sensitive Mechanical Behavior, Gordon and Breach, New York, 1966, pp. 351-402. [16] Sutton, H. et al, Journal of the Institute of Metals, Vol. 71, 1945, p. xvii. [17] Kubaschewski, O. and Hopkins, B. E., Oxidation of Metals and Alloys, Butterworths, London, 1953, p. 30. [18] Forty, A. J. and Humble, P., "Surface Films and Stress-Corrosion Cracking," Environment-Sensitive Mechanical Behavior, Gordon and Breach, New York, 1966, pp. 403-420. [79] Jamieson, A. L. and Rosenthal, H., "Aqua Ammonia Test," ASTM-AIME Symposium on Stress-Corrosion Cracking of Metals, 1944, American Society for Testing and Materials, Philadelphia, 1945, pp. 36-46. [20] Mattsson, E., "Stress Corrosion in Brass Considered Against the Background of Potential/pH Diagrams," Electrochimica Acta, Vol. 3, 1961, pp. 279-291. [21] Westwood, A. R., Goldheim, D. L., and Pugh, E. N., "Embrittlement of Polycrystalline Silver Chloride," Discussions of the Faraday Society, No. 38, 1964, pp. 147-156. [22] Levine, E., Solomon, H., and Cadoff, I., "Fracture Characteristics of Polycrystalline AgCl Wet with Aqueous Solutions," Acta Metallurgica, Vol. 12, 1964, pp. 1119-1124. [25] Westwood, A. R., Goldheim, D. L., and Pugh, E. N., "Fracture Behavior of AgCl in Aqueous NaCl," Acta Metallurgica, Vol. 13, 1965, pp. 695-700. [24] Levine, E. and Cadoff, I., "The Embrittlement of Single Crystals and Bicrystals of AgCl in Aqueous Salt Solutions," Acta Metallurgica, Vol. 13, 1965, pp. 875-880. [25] Westwood, A. R., Goldheim, D. L., and Pugh, E. N., "Complex-Ion Embrittlement of Silver Chloride," Materials Science Research, Vol. 3, Plenum, New York, 1966, pp. 553-576. [26] Westwood, A. R. and Goldheim, D. L., "Embrittlement of Silver Chloride by Copper Chlorocomplexes," Proceedings, First International Conference on Fracture, Sendai, Japan, Vol. 2, 1965, pp. 1999-2014. [27] Westwood, A. R., Goldheim, D. L., and Lye, R. G., "Rebinder Effects in MgO," Philosophical Magazine, 1967, in press. [28] Westwood, A. R., Goldheim, D. L., and Pugh, E. N., "A Double-Layer Mechanism for the Complex-Ion Embrittlement of Silver Chloride," Philosophical Magazine,Vol. 15, 1967, pp. 105-120. [29] Abbink, C. and D. S.,reserved); "Ionic Conductivity Silver Chloride Copyright by H.ASTM Int'lMartin, (all rights Wed Dec 16of 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No fu PUGH AND WESTWOOD ON CRITICAL SPECIES [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45] [46] [47] [48] [49] [50] [5.7] 247 Containing Cadmium Chloride," Journal of Physics and Chemistry of Solids, Vol. 27, 1966, pp. 205-215. Ebert, I. and Teltow, J., "Zur lonenleitung und Fehlordnung von Silberchlorid mit Zusatzen," Annalen der Physik, Vol. 15, 1955, pp. 268-278. Abey, A. E. and Tomizuka, C. T., 'The Effect of Hydrostatic Pressure on the Ionic Conductivity of Silver Chloride," Journal of Physics and Chemistry of Solids, 1966, Vol. 27, pp. 1149-1159. Metals Handbook, 8th ed., American Society for Metals, Metals Park, Ohio, Vol. 1, 1961. Heine, M. A., Keir, D. S., and Pryor, M. J., 'The Specific Effects of Chloride and Sulphate Ions on Oxide Covered Aluminum," Journal of the Electrochemical Society, Vol. 112, 1965, pp. 24-32. Fairman, L. and West, J. M., "Stress-Corrosion Cracking of a Magnesium Aluminum Alloy," Corrosion Science, Vol. 5, 1965, pp. 711-716. McEvily, A. J., Jr., and Bond, A. P., "On Film Rupture and Stress-Corrosion Cracking," Environment-Sensitive Mechanical Behavior, Gordon and Breach, New York, 1966, pp. 421^53. Nielsen, N. A., 'The Role of Corrosion Products in Crack Propagation in Austenitic Stainless Steel. Electron Microscopic Studies," Physical Metallurgy of Stress-Corrosion Fracture, Interscience, New York, 1959, pp. 121-143. Logan, H. L., McBee, M. J., and Kahan, D. J., "Evidence for an ElectroChemical-Mechanical Stress Corrosion Fracture in a Stainless Steel," Corrosion Science, Vol. 5, 1965, pp. 729 and 730. Staicopolus, D. N., "Electrocapillary Studies on Solid Metals," Journal of the Electrochemical Society, Vol. 108, 1961, pp. 900-904. Uhlig, H. H., "New Perspectives in the Stress-Corrosion Problem," Physical Metallurgy of Stress-Corrosion Fracture, Interscience, New York, 1959, pp. 1-17. Coleman, E. G., Weinstein, D., and Rostoker, W., "On a Surface Energy Mechanism for Stress-Corrosion Cracking," Acta Metallurgica, Vol. 9, 1961, pp. 491-496. Nichols, H. and Rostoker, W., "Analogies Between Stress-Corrosion Cracking and Embrittlement by Liquid Metals," Transactions, American Society for Metals, Vol. 56, 1963, pp. 494-507. Westwood, A. R. and Kamdar, M. H., "Concerning Liquid Metal Embrittlement, Particularly of Zinc Monocrystals by Mercury," Philosophical Magazine, Vol. 8, 1963. pp. 787-804. Stoloff, N. S. and Johnston, T. L., "Crack Propagation in a Liquid Metal Enivronment," Acta Metallurgica, Vol. 11, 1963, pp. 251-256. Sears, G. W., Growth and Perfection of Crystals, Wiley, New York, 1958, p. 441. Sears, G. W., "Dislocation Etchings," Journal of Chemical Physics, Vol. 32, 1960, pp. 1317-1322. Westwood, A. R. and Rubin, H., "Etch-Tunnels in Lithium Fluoride," Journal of Applied Physics, Vol. 33, 1962, pp. 2001-2007. Pickering, H. W. and Swann, P. R., "Electron Metallography of Chemical Attack Upon Some Alloys Susceptible to Stress-Corrosion Cracking," Corrosion, Vol. 19, 1963, pp. 373-3891. Swann, P. R. and Embury, J. D., "Microstructural Aspects of Stress-Corrosion Failure," High-Strength Materials, Wiley, New York, 1965, pp. 327-359. Edeleanu, C., 'The Propagation of Corrosion Pits in Metals," Journal of the Institute of Metals, Vol. 89, 1960, pp. 90-94. Westwood, A. R., "Concerning the Possible Role of Adsorbed Step-Poisons in Stress-Corrosion Cracking," Corrosion Science, Vol. 6, 1966, pp. 381-384. Saxena, M. N. and Dodd, R. A., 'Transgranular Stress-Corrosion Cracking Mechanisms in High Purity Austenitic Stainless Steels," Environment-Sensitive Mechanical Behavior, Gordon and Breach, New York, 1966, pp. 455—479. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authori R. W. Staehle1 Circulating Autoclave System for Stress Corrosion Cracking Studies REFERENCE: R. W. Staehle, "Circulating Autoclave System for Stress Corrosion Cracking Studies," Stress Corrosion Testing, ASTM STP 425, Am. Soc. Testing Mats., 1967, p. 248. ABSTRACT: A circulating autoclave system has been constructed and operated in which time-to-breaking can be determined exactly at temperatures to 700 F, pressures to 5000 psi, a wide range of stresses, various concentrations of dissolved gases and ionic species, and for a large number of specimens exposed simultaneously. Preliminary experiments have studied commercial and specially melted iron-nickel-chromium alloys. Results have confirmed already established trends for the effect of environment and alloy composition at 500 F. KEY WORDS: corrosion, stress corrosion, cracking (fracturing), autoclaves, steels, stainless steels, Inconel, iron-chromium-nickel alloys There is a need to obtain reliable information on time-to-breaking due to stress corrosion cracking as a function of alloy composition, environmental composition, temperature, and stress. This information is necessary for both direct use in engineering applications and to furnish boundary conditions for the development of theories of stress corrosion cracking. In the past, it has been difficult to obtain quantitative information on time-to-breaking at temperatures above the solution boiling temperature or for any range of dissolved gas concentrations. Control of applied stress at the elevated temperatures and at pressures above atmospheric has also been difficult. Information available at the higher temperatures has generally been obtained by periodic examination of specimens at the end of specified test periods. Described herein is a circulating autoclave system capable of measuring time-to-breaking for up to 200 specimens at temperatures to 700 F, pressures to 5000 psi, dissolved oxygen concentrations to 1000 ppm, and stresses to any desired level. The system is capable of operating in the once-through or recirculating modes. At the outset it was expected that the oxygen and chloride concentra1 Assistant professor, Corrosion Center, Department of Metallurgical EngineerCopyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 ing, The Ohio State University, Columbus, Ohio. Personal member ASTM. Downloaded/printed by 248 pursuant to License Agreement. No further reproductions autho University of Washington (University of Washington) STAEHLE ON CIRCULATING AUTOCLAVE SYSTEM 249 tions required for cracking would be relatively high, that .is, several hundred parts per million. For these ranges of concentration, it was adequate to operate the system in the recirculating mode. Thus, a slight contamination due to recirculating the solution would exert little influence compared to the relatively high chloride concentration. However, it was found that the cracking was occurring in very short times at concentrations of chloride and oxygen less than 10 ppm. Therefore, it was decided that it would be necessary to change to a once-through mode of circulation in which the solution would be furnished from a distilling unit and dumped to drain after passage through the test region. The experimental results described herein are from preliminary work performed while the system was operating in the recirculating mode. As of the writing of this paper, the necessary equipment for the once-through system had been installed and preliminary shakedown runs begun. Significant components, procedures, and preliminary experimental results will be described. Emphasis is placed on features which are unique to measuring time-to-breaking and to controlling solution composition. Description of Circulating System and Components2 Design Objectives Circulation of the medium is required to prevent stagnation of the solution, replenish depleted reactants, remove tramp ions resulting from corrosion of the system, provide continuous monitoring of significant solution properties, and provide a means for rapidly heating the solution to test temperature. Since dissolved gases, especially oxygen, appear to exert a strong influence on the stress corrosion cracking phenomenon, it is necessary to introduce dissolved gases in quantities desired and according to whatever schedule is required to meet experimental objectives. The maximum quantity of gas introduced will depend on the solubility at temperature, requirements of the experiment, and maximum pressure capability of the system. The system was designed to operate up to 5000 psi with no component operating with a maximum internal applied stress of 10,000 psi. Materials exposed to the corrosive medium were selected to provide maximum resistance to corrosion and subsequent malfunction or contamination. Inconel 600 was selected as the main material of construction. A maximum temperature of 700 F was based on the temperature of liquid water being 705 F at the critical point. To obtain interpretable data, it was necessary to keep the corrosive medium single phase; twophase (boiling or gas bubble) environments were considered to produce Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 2 All high pressurebycomponents for this system were fabricated by Pressure ProdDownloaded/printed ucts Industries, Hatboro,(University Pa. University of Washington of Washington) pursuant to License Agreement. No further reproductions a Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized. FIG. 1—Schematic arrangement of circulating autoclave systems used for stress corrosion cracking studies. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized. FIG. 2—Schematic arrangement of electrical power for circulating autoclave system. 252 STRESS CORROSION TESTING more uncertainties than desirable. These requirements provided the incentive for the capability of operating at pressures above the equilibrium vapor pressure. (The vapor pressure of steam in equilibrium with water at 705 F is 3200 psi.) To obtain data of maximum usefulness, it was necessary to provide a procedure for indicating time-to-breaking of any specimen under FIG. 3—External view of test autoclave. Total height approximately 72 in. any condition of temperature and pressure without shutting down the circulating system. The components and system arrangements described in subsequent sections were designed specifically to meet the above objectives. Circuating System A schematic arrangement for the circulating system is shown in Fig. 1, and a schematic of the electrical power arrangement is shown in Fig. 2. Copyright by ASTM Int'lrecirculated (all rights reserved); Dec 16 15:53:43 EST 2015 The solution may be or mayWed be circulated on a once-through Downloaded/printed by basis. Assuming that the solution is circulated on a once-through basis, University of Washington (University of Washington) pursuant to License Agreement. No further repro STAEHLE ON CIRCULATING AUTOCLAVE SYSTEM 253 it is introduced at valve 2. The water coming to valve 2 is produced continuously in a 100 gal/hr still and prior to entering at valve 2 is mixed with the required chloride or other specie additions using positive displacement pumps. Dissolved gases, if required, are introduced at the main pump. The pump raises the liquid pressure from ambient to the desired operating pressure, usually about 4000 psi. The solution moves to the preheater in which it is heated to test temperature. Passing from the pre- FIG. A—Cross section of test vessel used for stress corrosion cracking tests. heater, the solution enters the test vessel from which it passes through a cooler and back pressure regulator. The cooler is necessary to prevent damage to the back pressure regulator valve and to prevent a large volume of hot steam from entering the drain system. The back pressure regulator together with the positive displacement pump acts to maintain solution pressure. Main Te,sf Fe-s.se/ Copyright by ASTM Int'l (allisrights reserved); Wed3 Dec EST 2015 The main test vessel shown in Figs. and 164. 15:53:43 The primary purpose Downloaded/printed by of the large size is to permit exposing a relatively large number of speciUniversity of Washington (University of Washington) pursuant to License Agreement. No further reprodu 254 STRESS CORROSION TESTING mens at the same time to identical environmental conditions, thereby enhancing the self-consistency of results. A shell-liner construction was used primarily to keep the stresses low on the inside surfaces of the liner. By shrink fitting the liner into the shell, the calculated tangentical stress on the inside surface is less than 1000 (negative) psi when the autoclave operates at 700 F and 5000 psi internal pressure. The liner was centrifugally cast Inconel 600, because this method of fabrication permitted some reduction in cost, and the shell was 4340 steel. A double- FIG. 5—Top of autoclave head showing electrical lead fittings in place. Plugs are placed in openings not used for electrical leads. end construction of the vessel was used to permit cleaning and for possible future use of the bottom head for bringing in electrical leads. The top head contains 15 openings through which electrical or other leads can be brought into the autoclave. Figure 5 shows the top head with both seals and electrical lead fittings. Figure 6 shows a schematic cross section of an electrical lead-through fitting. Each of these fittings permits up to 16 electrical leads to be brought through the pressure boundary to be used for determining time-to-breaking of wire specimens. These fittings are capable of withstanding up to 5000 psi without losing their pressure seal. The Teflon gasket is a standard Conax design, and the use of cooling fins maintains the temperature sufficiently low so that the Teflon does not flow. Housings were of Inconel 600, the insulators of Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 alumina, and the electrical leads of Teflon-coated Inconel 600. These Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further rep STAEHLE ON CIRCULATING AUTOCLAVE SYSTEM 255 FIG. 6—Schematic cross section of fitting for bringing leads through autoclave head. insulated electrical lead wires have been satisfactory to 600 F and have shown no evidence of significant deterioration below this temperature. Preheater The preheater consists of an Inconel 600 tube (%6 by % ap6 in.) Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 proximately 130 ft by long and is heated by using the tube itself as a reDownloaded/printed University of Washington (University of Washington) pursuant to License Agreement. No further r 256 STRESS CORROSION TESTING FlCr. 7—-Relationship among power, tube length, and voltage for preheater. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 FIG. 8—Assembled preheater showing insulation in place. One side removed. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprod STAEHLE ON CIRCULATING AUTOCLAVE SYSTEM 257 sistor. The relationship between power, tube length, and voltage is shown in Fig. 7. Self-heating was chosen because it was the most economical and involved only the cost of the transformer. The assembled and insulated preheater is shown in Fig. 8. Figure 2 shows the electrical arrangement for providing power to the preheater. The length of the preheater tubing was chosen to keep the thermal stresses to a minimum, with the maximum thermal stress at full power being less than 3000 psi in tension at the inside surface. Pump and Gas Injection A duplex positive displacement pump is used to circulate the solution and ha^a maximum flow capacity of 40 gal/hr against a pressure of 5000/psi. One end of the pump is shown in Fig. 9. FIG. 9—One end of duplex hydraulic pump used to circulate solution. Pump fluid system driven by 5-hp motor. Any dissolved gases desired are introduced at the pump. This procedure eliminates uncertainties which would result from direct injection of gas under pressure. Figure 10 shows the schematic arrangement for introducing the dissolved gas. The gas is trapped between the solenoid valves (8 and 9 on Fig. 1 and shown on Fig. 10) at a pressure higher than system pressure and in a known volume. The entrapped gas is released on each stroke of the positive displacement pump. Thus, on each half cycle of the pump, solution containing dissolved gas of a measured and required concentration is injected into the system. The pump cam and switches for activating the solenoid valves are shown in Fig. 11. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Cooler Downloaded/printed by The cooler is designed(University to cool solution from pursuant 700 F to room temperaUniversity of Washington of Washington) to License Agreement. No further re Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized. FIG. 10—Schematic arrangement for injecting gas and for the hydraulic system. STAEHLE ON CIRCULATING AUTOCLAVE SYSTEM 259 ture at a rate of 40 gal/hr. Tap water is used as a coolant. The tubing in the cooler, Inconel 600, is about 120 ft long and is spirally wound. An exterior view of the cooler is shown in Fig. 12. Specimen Stressing and Break Indication Important objectives in devising techniques for stressing the specimens are: There should be a positive indication when the specimen breaks; the applied load should be adjustable and controllable; the specimen should not be subject to selective chemical effects due to edges, corners, or crevices; a maximum number of specimens should be exposed in a single experiment; and the final result of time-to-breaking should be indicative of the early stages of the cracking process. FIG. 11—Cam on pump used for activating microswitches which control oxygen additives. The optimum result from considering the above objectives was to utilize a compression spring and 0.015-in.-diameter wire specimen. Essential features of the design are shown in Fig. 13. The wire specimens are first attached to the upper clamp, and a weight is attached to the other end. The magnitude of the weight is determined from the load required on the wire and the temperature affected change in the shear modulus. Since the spring material is Inconel X heat treated for 10 hr at 1350 F, there is essentially no relaxation of the spring material up to the 700 F maximum operating temperature.3 The adjustment for effect of temperature on shear modulus is included in the initial load.4 When the prescribed load is applied, the lower clamp is tightened. The deflection of the spring due to the weight is maintained when the 3 "Superalloy Wire," Sept.Wed 1963, Standard Copyright by ASTM Int'l Bulletin (all rights 1201, reserved); Dec National 16 15:53:43 EST 2015Co., Niles, Mich. Downloaded/printed by 4 Private ofcommunication from International Nickel Co. to License Agreement. No further reprodu University Washington (University of Washington) pursuant 260 STRESS CORROSION TESTING clamp is tightened; thus, the spring exerts the same tensile force as the weight. Important consideration in designing the springs are: minimizing spring volume to maximize the number of specimens; minimizing stress in the springs to reduce relaxation and chemical attack; maximizing FIG. 12—Cooler used to cool solution from test temperature to ambient. spring deflection rate to reduce errors in stress level; restrict length of springs to keep below the critical buckling length; and minimize solid height of the spring to improve circulation around the specimen. The foregoing requirements conflict and compromises are required. Since Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 there is no obvious way to derive an analytical expression for optimizing Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further rep STAEHLE ON CIRCULATING AUTOCLAVE SYSTEM 261 FIG. 13—Design of jig used for stressing specimens. spring size, it was arbitrarily decided that the autoclave should contain approximately 100 specimens. Using the standard equations for spring deflection and spring surface stresses as plotted in Fig. 14, together with Wahl's expression5 for the buckling criterion, the spring parameters of wire diameter (d), number of coils («), length (L), and average diameter (D) were selected. Figure 15 shows a close-up of the actual springs and electrical lead B Wahl, A. Mechanical Springs, McGraw-Hill, York, 1963; Cheronis, Copyright by M., ASTM Int'l (all rights reserved); Wed DecNew 16 15:53:43 EST 2015 N.Downloaded/printed P., ed., Spring Design by and Application, McGraw-Hill, New York, 1961. University of Washington (University of Washington) pursuant to License Agreement. No further re 262 STRESS CORROSION TESTING wires, and Fig. 16 shows an experiment ready to be placed in the autoclave. The middle zone contains no specimens in Fig. 16. All materials in the specimen support and testing jigs are Inconel 600. In Fig. 16 it should be noted that the autoclave head rotates, but the specimen rack does not. Recording Time-to-Cracking Times-to-cracking are recorded in an Esterline Angus event recorder as shown in Fig. 17. When a specimen breaks (Fig. 18), a circuit is closed and a pen in the event recorder moves to indicate that the circuit FIG. 14—Spring design curves relating deflection, load, stress in spring, spring wire diameter, number of coils, and stress in test wire for springs of 1.25 in. mean diameter. is closed. The event recorder has modules of 20 pens each, but the unit used contains two modules per cabinet so that 40 pens are available. In the present system, each pen can monitor several specimens simultaneously if a cyclic timing system is utilized. Thus, for one second out of a minute, circuit a would check all specimens on the a circuit, and during a subsequent interval the b circuit is checked. The individual pens may be used for double or triple duty. The circuit for multiple pen use is shown in Fig. 19. One of the difficulties associated with the use of the "no-contact and subsequent contact" system is the conductivity of the environment. If theInt'l environment is too conducting, there isEST little discrimCopyright by ASTM (all rights reserved); Wed Dec 16 15:53:43 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprodu STAEHLE ON CIRCULATING AUTOCLAVE SYSTEM 263 ination between the nonbroken and broken situations. This difficulty is circumvented by reducing the available area of the contacting electrodes or by reducing the conductivity of the solution. The latter approach is generally the most applicable, since critical studies of cracking are generally conducted at low ionic concentrations. FIG. 15—Close-up of assembled load of stressing jigs ready to be inserted into autoclave. Protective Circuits The nature of the circulating system is such that certain accidents are possible such as loss of pressure or overheating. Since it is obviously desirable to protect both the experiment and system, a safety circuit system (Fig. 20) is utilized. This system will shut off any of the major components on signals from either the pressure monitor or the multipoint temperature recorder. The multipoint temperature recorder contains six alarms which can be connected to any of the 24 available temperature positions. Copyright by are ASTM Int'l (all rights Dec 16system 15:53:43 2015 Personnel protected from reserved); the highWed pressure byEST a barricade Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further repr 264 STRESS CORROSION TESTING FIG. 16—Specimen stressing jigs attached to autoclave head and ready to be inserted into autoclave. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduction STAEHLE ON CIRCULATING AUTOCLAVE SYSTEM 265 FIG. 17—Dual event recorder for recording time-to-cracking. FIG. 18—Schematic showing configuration of stressing jigs with broken and unbroken wire specimens. (Fig. 21). This barricade has two thicknesses of %-in. plywood and one thickness of ^-in. boiler plate. All valves are mounted on the inside with their valve stems extending through the wall. A monorail host is used to transport the experiments over the safety barricade into the Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 autoclave. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions 266 STRESS CORROSION TESTING Miscellaneous Components All valves have Inconel 600 bodies. Valves for higher temperature portions of the system have their packings extended about 5 in. from the solution. The temperature drop to the packing is about 300 F at an operating temperature of 500 F. The system is protected by gold-covered stainless steel ruptive disks. Gold covering of the ruptive disks is essential to prevent their premature failure. FIG. 19—Wiring arrangement for multiple use of event pens. Joints are of the mechanical cone type using a collar and gland nut. These joints frequently lose their seal when there is appreciable temperature cycling. This requires the use of welded joints in critical applications such as the preheater. Source of Distilled Water The new distilled water source is a Barnstead Model SS-100 still having a capacity of 100 gal/hr. Two positive displacement pumps are used to mix the distilled water and the chloride. The water is pumped with a Clark Cooper c-p 2 positive displacement pump, and the chloride with a Beckman solution metering pump No. 746. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further rep STAEHLE ON CIRCULATING AUTOCLAVE SYSTEM 267 FIG. 20—Schematic wiring diagram for protective circuits. Experimental Results and Discussion Experiments performed to date have been conducted using the system in a recirculating mode. This initial work has been directed primarily toward determining the general behavior of the circulating system and the time-to-breaking circuitry. Such factors as oxygen concentration, Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No furth 268 STRESS CORROSION TESTING alloy composition, initial stress, metallography, and specimen mounting procedures have been investigated. For the recirculating experiments double distilled water was used and sodium chloride (NaCl) additions were made as required. Specimens were stressed as described previously. All specimens in these experiments were wires of 0.015 in. diameter. Results from the first series of experiments are tabulated in Table 1. Important features of the data in Table 1 are as follows: 1. With increasing nickel content, the time-to-breaking increases. FIG. 21—Safety barricade for circulating autoclave system showing instrumentation. 2. As the concentration of reactants in the environment decreases, there is a significant increase in time-to-breaking. 3. At high concentration of environmental reactants, the effect of stress is relatively small, but as the concentration decreases, there is an increasing difference between the high- and low-stress levels. Table 2 summarizes results from a preliminary experiment which was conducted to determine the general significance of alloy composition. The experiment involved 93 hr and 30 min of operation at 50 ppm Nad and 8 ppm oxygen, after which the oxygen was changed to 158 ppm and the experiment continued. The data in Table 2 have been arranged that portions of 16 the15:53:43 experiment Copyright assuming by ASTM Int'l (all the rightstwo reserved); Wed Dec EST 2015can be Downloaded/printed by All specimens were stressed by the same load. The treated individually. University of Washington (University of Washington) pursuant to License Agreement. No further reprod Type 304 Type 309 Type 310 Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized. Alloy 6 TABLE I—Time-to-breaking for three stainless steels at 500 F, min." 1000-158d 1000-158 21 000 36 800 21 000 23, 27, 28 22,27 26, 25, 27 24,25 25,26 25, 25, 25 36 800 21,25 24,24 91,20, 35 17, 17 35 NB (365) NB (365) 27,23 21 000 42,43 36 800 41,30 57, 35 NB (235) 33,30 NB (365) 3 specimens 26, 147 c Stress, psi 1000-8 10-8 50-8 50-158" 3581, 376, 634 186, 20 NB (4350) 1870 903 -6 4501 NB (4350) 3 specimens NB (4350) 2 specimens 30, 90, 685 43, 42, 58 NB (5610) 2 specimens 58, 53 NB (5610) 2 specimens 74, 59 " Time-to-cracking is given to nearest minute. The designation NB indicates that the specimen did not break for the duration of the experiment with the number in parentheses giving the length of the experiment. The flow for all experiments was 33.5 gal/hr, and the operating pressure was 3250 psi. Time-to-cracking is measured from a time 15 min after starting oxygen additions. The oxygen additions were started when the temperature of the preheater outlet was 330 F. Specimens broke both in the grips and between grips. Times-tobreaking were found to be generally the same for breaking in either location. Metallographic examination showed no differences in the morphology of the cracks. 6 Specimens in the as-received condition. c The stresses are corrected for change in shear modulus. No check was made later to determine the relaxation of the individual specimens which in many cases were broken and thus did not permit such a measurement. d The first number indicates amount of chloride in parts per million; the second the amount of oxygen. In cases where oxygen was listed as 8 ppm, it was assumed that the water was saturated with air prior to the experiment. The 8 ppm value is the approximate amount present in water in contact with air at 25 C. In experiments showing 158 ppm oxygen, the oxygen was injected as a measured amount on each stroke of the positive displacement pump. » Specimens previously exposed to 50 ppm NaCl and 8 ppm oxygen before oxygen was raised to 158 ppm. Because of the nature of the results, it was considered legitimate to divide the experiment into two parts. 270 STRESS CORROSION TESTING TABLE 2—Results from stress corrosion studies of iron-chromium-nickel base alloys at 500 F.a Breaking Times with 50 ppm NaCl and 8 ppm Oi Alloy* 304 30 310 54 54C 54Mo 54N 54P 54Pd 54P 54Re 54Si 54W 58 Total Individual Time, hrrmin 0:30, 1:30, 11:25 Breaking Times with 50 ppm NaCI and 158 ppm O: Average Total Individual Time, hr:min Total Average Average after Os Increase 4:30 94:11, 94:12, 94:28 94:28, 94:23 94:44, 94:29 94:36, 94:12, 94:44 94:25,94:21,94:18 94:18 94:03 93:57 93:53, 93:50 94:17 94:26 94:37 94:31 94:21 94:18 94:33 93:57 93:52 0:47 0:56 1:07 1:01 0:51 0:48 0:33 0:27 0:22 94:33 94:31 94:34 97:46 1:03 1:01 1:04 4:16 0:58, 0:58 0:43 81:40,76:20 0:58 0:43 79:00 2:50, 57:49 2:23 30:05 2:23 94:18,94:47 94:31 94:34 96:39, 96:21 101:58,96:06 0 All specimens loaded to 21,000 psi stress. Notes from Table 1 apply. Alloy 54 = Fe-15Ni-20Cr. Fourth component additions are made at 1.5 atomic per cent for metallic elements and 0.1 atomic per cent for nonmetallic elements. Alloy 58 = Fe-45Ni-20Cr. 6 TABLE 3—Time-to-breaking (tf)° as affected by surface preparation for Type 310 stainless steel at 500 F and 90 per cent of yield stress (0.2 per cent offset}. No. Specimens Tested No. Failed As-received6 Vacuum annealede Electropolished** 11 11 0 12 2 1 Mechanically polished* 12 6 Condition Times- to-B reaking, hr:min 0:29, 0:47 0:28 0:40,0:44 0:48, 0:49 18:00, 0:42 Avg, hr:min 0:38 0:28 0:44' 0 NaCl content: 0.0050 weight per cent, O2 content: 0.0010 weight per cent (approximately). All specimens broke at ends and inside Teflon sleeves. Cracking at this location has been determined previously to occur at about same time as outside sleeve. Also, failure occurs by cracking and not general oxidation. 6 Specimen held approximately 10 sec at 1100 C in strand anneal. « Vacuum annealed approximately 2 hr at 1090 C. d Vacuum annealed as in (c) and electropolished. • Vacuum annealed as in (c) and mechanically polished with 600-grit paper. (Note that questions of importance of local mechanical work, roughness; and bulk deformation are being checked in subsequent experiments.) f Did not include 18:00 hr breaking time in average. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprod Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized. FIG. 22—Effect of surface preparations on time-to-breaking of Type 310 stainless steel. 272 STRESS CORROSION TESTING most pronounced effect of composition is the effect of nickel. The best comparison is between alloys 54 and 58 (15 versus 45 per cent nickel, both with 20 per cent chromium). The breaking time for the alloys is 61 versus 256 min after the increase in oxygen concentration. There are other obvious trends with various fourth component alloy additions, but the significance of trends based on these observations is not presently apparent. Table 3 summarizes a preliminary experiment to compare the effect of surface preparation on time-to-breaking. Clearly, there is a significant FIG. 23—Effect of chloride and oxygen concentration on time-to-breaking of Type 304 stainless steel. difference among the preparations from the standpoint of total number of failures. Again the significance of this experiment is not presently obvious except to note that surface preparation is indeed important. These data should be compared with the results from similar experiments6 in boiling magnesium chloride (MgCl2) shown in Fig. 22. These results show the same general trends, that is, that the mechanically abraded surfaces tend to break earlier and tend to exhibit a tighter spread of cracking time. Results from experiments described above and summarized in Tables 1 to 3 are very similar to results from many previous experiments by a Cochran,byR., unpublished results, Corrosion Department of MetallurCopyright ASTM Int'l (all rights reserved); WedCenter, Dec 16 15:53:43 EST 2015 gical Engineering, Theby Ohio State University, Columbus, Ohio. Downloaded/printed University of Washington (University of Washington) pursuant to License Agreement. No further reprod STAEHLE ON CIRCULATING AUTOCLAVE SYSTEM 273 other investigators. For example, the beneficial effect of nickel, the relatively rapid cracking of Type 304 stainless steel, and the effects of chloride and oxygen concentration have been described many times. However, the unique aspect of the work described herein is the capability for measuring times-to-breaking exactly for any combination of significant environmental parameters. Subsequent experiments will study in detail the effect of alloy, temperature, stress, and environmental chemistry on time-to-breaking. Figure 23 summarizes time-to-breaking information for Type 304 stainless steel. Points plotted are averages of data from Table 1. The strong effects of chloride and oxygen concentrations are evident. Conclusions 1. A circulating autoclave system has been developed in which exact time-to-breaking can be determined for specimens stressed at designated stresses, various dissolved gas concentrations, various ionic species, temperatures to 700 F, and pressures to 5000 psi. 2. Preliminary stress corrosion cracking experiments show that the system performs as intended. These experiments show for the ironchromium-nickel alloys tested that time-to-breaking is increased by reducing the stress, reducing chloride and oxygen, and increasing nickel concentration. 3. From an engineering point of view, note that cracking of Type 304 stainless steel will occur in about 6 hr at the annealed yield stress for chloride contents of 10 ppm and oxygen concentrations of 8 ppm. For this same chemistry, cracking can occur in 20 min at 36,800 psi. A cknowledgments This work was sponsored by the U.S. Atomic Energy Commission, Nuclear Technology Branch, under the supervision of J. M. Simmons and A. E. Van Echo. Experimental work described herein was performed by P. J. Simmons and J. Frey. It is a pleasure to acknowledge the very considerable contributions to the design and construction of the high pressure equipment by Pressure Products Industries, Hatboro, Pa. Robert Wolf and Anthony Ostrowski of that organization were especially helpful. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further re Nancy McKinney1 and H. W. Hermance1 Stress Corrosion Cracking Rates of a Nickel-Brass Alloy Under Applied Potential REFERENCE: Nancy McKinney and H. W. Hermance, "Stress Corrosion Cracking Rates of a Nickel-Brass Alloy Under Applied Potential," Stress Corrosion Testing, ASTM STP 425, Am. Soc. Testing Mats., 1967, p. 274. ABSTRACT: A method was developed to test the susceptibility of nickelbrass alloys to stress corrosion cracking in the presence of nitrates or other salts. A wire of the alloy under test was stressed, and an electrical leakage path was provided to a cathode by a filter paper impregnated with the electrolyte being studied. Stresses, potentials, salt concentrations, relative humidities, and time were varied to evaluate these factors. In the case of ASTM Grade D nickel-brass alloy, using ammonium nitrate as the electrolyte, the rate of crack penetration increased with increases in the salt concentration in the leakage path, the applied stress, relative humidity, and temperature. There was no direct correlation with the applied potential. Other hygroscopic nitrates such as zinc, and to a lesser degree calcium and copper, also caused stress corrosion cracking. Chlorides and sulfates were ineffective under conditions in which nitrates produced cracking. From the data obtained, the expected failure times of nickel-brass parts at average ambient temperature and humidity could be estimated. KEY WORDS: copper alloys, nickel alloys, zinc alloys, corrosion, stress corrosion, electric potential, nitrates, humidity, sulfates, chlorides, salt water Failures of stressed nickel-brass parts of telephone equipment have been observed in central offices in the Los Angeles area. The parts that failed, wires of 0.023 in. diameter, were normally under about 6-g load and had a positive potential. An inspection showed that failure was due to a form of stress corrosion cracking. These failures often occurred within as short a time as two years. Large amounts of particulate matter had accumulated on these metal surfaces near areas that had cracked. This particulate material was an accumulation of air-borne dusts. It seemed possible that corrosive materials in this dust during periods of high humidity could have caused the cracking. Chemist and chemist (retired), respectively, Chemical Dept, Bell Telephone Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Laboratories, Inc., Holmdel, N. J. Downloaded/printed by 274 University of Washington (University of Washington) pursuant to License Agreement. No further reproductio McKINNEY AND HERMANCE ON NICKEL-BRASS ALLOY 275 The Los Angeles dusts are hydrophilic, containing large amounts of adobe dusts, which are fine claylike materials, as well as containing oxidized organic compounds and inorganic ions. About half of the solids obtained from an evaporation of a water extract of Los Angeles dust deliquesced at 55 per cent relative humidity. Table 1 shows that these air-borne particulates are composed of sulfates and nitrates; about 8.2 per cent by weight is sulfate; 8 per cent by weight is nitrate. Generally TABLE 1—Analysis of Los Angeles air-borne dust specimen (weight per cent).0 Material 4-g Specimen Benzene soluble Water soluble Sulfate Nitrate Sodium Ammonium Chloride Calcium Potassium Magnesium Insoluble-combustible Insoluble-noncombustible a 18 28 8.2 8.0 3.0 2.6 1.9 1.5 1.0 0.7 46 8 Specimen collected by high volume sampler from 18 February to 19 July, 1960. TABLE 2—Relative humidity over saturated solutions. Salt Calcium chloride Calcium nitrate Ammonium nitrate Sodium nitrate Sodium chloride Ammonium chloride Ammonium sulfate Sodium sulfate Calcium sulfate RH, % Temperature Reference, pp. 28.99 50.7 61.8 73.93 75.93 78.4 80.0 81 98 25 C 77.7 F 25 C 25 C 25 C 71 F 25 C 25 C 20 C 521-528° 519-520 507-517 521-528 521-528 519-520 507-517 507-517 675 a Wexler, A., ed., "Humidity and Moisture Measurement and Control in Science and Industry,". Vol. 3, Reinhold, New York, 1965. 6 International Critical Tables, 1st ed., Vol. 1, McGraw-Hill, New York, 1926. the nitrate salts are more hygroscopic than the sulfate salts as is shown in Table 2. Preliminary qualitative experiments were made to determine if either nitrate or sulfate salts could cause cracking of nickel brass. It was found that the nitrate salts such as ammonium and calcium caused rapid cracking at 75 per cent relative humidity but that ammonium sulfate did not. A laboratory was Wed thenDec begun to study Copyright by ASTM testing Int'l (all program rights reserved); 16 15:53:43 ESTquantitatively 2015 the effect of nitrate and other anions as well as to study the effect of apDownloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduc 276 STRESS CORROSION TESTING plied potential, relative humidity, temperature, and stress on the rate of stress corrosion cracking of nickel-brass wire. In addition, it was hoped that a method for testing the susceptibility of various alloys to this type of stress corrosion cracking could be developed. Experimental Details The unit shown in Fig. 1 was used. It permitted simultaneous testing of six wires so that average effects could be obtained. The frame, 2, was FIG. 1—Test unit. made of Plexiglas and contained holes through which the test wires, 1, were inserted. Clamping screws, 3, were used to hold the individual wires in place. Two metal plates of gold-plated brass that were fastened to the frame were used as electrodes. The plate, 4, served as the cathode and the plate, 6, as the anode. The openings in the cathode plate were symmetrical about the test wire. The anode contained small holes through which the wires were inserted, so that the wires became anodic during the testing. The vertical position of the anode could be adjusted to give the desired constant deformation on the test wires. Filter paper disks (0.005 in. thick, 0.06 in.2 area), 5, containing a desired amount ofCopyright electrolyte, wereInt'l centered about each Wed of theDecwires and were physical by ASTM (all rights reserved); 16 15:53:43 ESTin2015 Downloaded/printed contact with both bythe wire and cathode rim. This was accomplished by University of Washington (University of Washington) pursuant to License Agreement. No further reprod McKINNEY AND HERMANCE ON NICKEL-BRASS ALLOY 277 punching out the papers to a larger diameter than that of the cathode rim and using a central hole in the paper disk of a smaller diameter than the wire. The alloy studied was ASTM Grade D 65-12-23 copper-nickel-zinc alloy, spring tempered, 0.025 in. in diameter (Specification for CopperNickel-Zinc Alloy (Nickel Silver) Wire, B 206 - 56). The wire was straightened, cut to 4 to 5-in. specimens, and degreased before testing. The initial loads on the wire were set with the anode plate at either 0, 5, 10, or 15 g corresponding to cantilever stresses of 0, 19,000, 39,000, or 56,000 psi.2 All these test loads are under the proportional limit for this wire which is about 91,200 psi. Unless otherwise stated, the testing was done at 15-g load. Ammonium nitrate was generally used as the electrolyte, since it was the most active of the electrolytes we had tested in producing stress corrosion cracking. Moreover, these ions are present in large amounts in the Los Angeles dusts. The amounts of nitrate used per disk ranged from 360 to 0.4 p,g as nitrate ion. When no potential was applied, at least 60/*g nitrate per disk were used. To prepare these disks, a calculated amount of nitrate was incorporated into Whatman No. 1 filter paper by dipping papers into a nitrate solution and air drying them. The exact amount of nitrate per square inch of paper was then determined. Disks were punched out of these sheets and stored at low humidity until used. The assembled units were tested either at 43 or 75 per cent relative humidity. These humidities were chosen to be above and below 62 per cent relative humidity, which is the equilibrium humidity for a saturated solution of ammonium nitrate (Table 2). Two saturated salts were used to obtain these test humidities. A saturated solution of sodium chloride gives 75 per cent relative humidity at room temperature;3 a saturated solution of potassium carbonate gives 43 per cent relative humidity at 24.5 C.4 As many as eight units at one time could be placed in a relative humidity chamber containing the appropriate saturated solution for simultaneous testing. Before starting the tests, the air in the chamber was circulated by a blower to ensure a uniform relative humidity in the chamber. Except when otherwise indicated, a potential of 45 v connected through a protective 1-megohm resistor to the unit's terminals was used. Since the measured current was always much less than 45 ju,a, it was obviously limited by the resistance of the filter paper disks, which was 2 Stresses were calculated using the following equation: S = 3EdD/2l2, where S = stress, psi; E = Young's modulus, psi; d — diameter of wire, in.; D — deflection due to load, in.; and / = length of wire, in. 3 Wexler, A. and Hasegawa, S., "Relative Humidity-Temperature Relationships of Some Saturated Salt Solutions in the Temperature Range 0 C to 50 C," Journal of Research of the National Bureau of Standards, Vol. 53, No. 1, July 1954, pp. 19-26. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 * Handbook of Chemistry and Physics, The Chemical Rubber Publishing Co., Downloaded/printed by p. 2499. Cleveland, Ohio, 1960, University of Washington (University of Washington) pursuant to License Agreement. No further reproductio 278 STRESS CORROSION TESTING FIG. 2—Photomicrograph of a diametrical section of a test wire. This wire was exposed to 29-ng nitrate, 75 per cent relative humidity, 15-g load, for 5 days. The photo shows the transcrystalline nature of the crack. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproducti McKINNEY AND HERMANCE ON NICKEL-BRASS ALLOY 279 governed by the relative humidity of the test chamber and the nitrate concentration. Discussion Crack Development Within a few hours after starting a test run, a small band of dark corrosion products appeared on the surfaces of the wires directly under FIG. 3—Initial corrosion (points 1 and 2). the filter paper about the wire's maximum stressed area. As the test continued, the corrosion began to penetrate into the wire only within the area of the wire which was under tensile stress and formed small transcrystalline cracks, which may have followed lattice imperfections in the metal. A photomicrograph of a section of a test wire is shown in Fig. 2. These cracks are not as yet deep enough to cause wire yawning. With the passage of time, the corrosion continued to deepen and widen. When it reached about three fourths of the way through the wire, visible surface cracks appeared. To determine just how far corrosion did penetrate into each of the test wires, they had to be broken open through their corroded area by bendCopyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproducti 280 STRESS CORROSION TESTING ing them in the same direction as the applied stress. Sometimes the corroded areas in the wire appeared only slightly discolored but usually were dark brown-black. Figure 3 shows the beginning of visible wire penetration. This wire, after bending, contained two large pits on its surface with discolored edges. It also had other very small pits, above point 1 and a little to the left, which cannot be seen in this figure. Figure FIG. 4—Cross sections of test wires showing progressive stages of penetration. 4 shows increasingly deep penetrations. Estimates of damage to the interior of these wires are listed on each illustration. These estimates were based on the per cent of total cross-sectional area that appeared to be covered by the corrosion product. The accuracy of this estimate was about ±5 per cent. Assuming that a typical corrosion penetration had progressed into the test wire from three main corrosion points or deep pits, Fig. 5 illustrates probable successive stages of corrosion through its cross section Copyright by ASTM Int'lof (allwires rightsat reserved); Dec of 16 penetration. 15:53:43 EST 2015 based on observations variousWed stages The three Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprod McKINNEY AND HERMANCE ON NICKEL-BRASS ALLOY 281 hypothetical corroded areas eventually join into a larger more continuous area of corrosion which then continues to deepen until wire yawning occurs. This type figure has been used by Hines5 to illustrate the depths of corrosion reached in austenitic chromium-nickel steel wires exposed to boiling magnesium chloride under varying stresses. Penetration Rates at Fixed NO3-, RelativeHumidity Humidity and and Load Load CondiCon 3~, Relative tions The rate of corrosion penetration into wires under various conditions, such as nitrate concentration, relative humidity, and load, was studied. This was done by placing up to eight test units under equal loads and nitrate concentrations in the same relative humidity chamber. Each unit FIG. 5—Progressive stages of penetration starting from three areas. was removed after a different time interval, and the depth of corrosion penetration into the test wires was determined. The average depth as well as the maximum and minimum depths of penetration reached for each of these time intervals was recorded. From these data, plots of average penetration versus time and maximum penetration versus time were made. Average and Maximum Rates—Figure 6 shows the rate curves based on average values obtained at high humidity. The range from the highest to the lowest penetration obtained with the six test wires is indicated by the straight lines through the average points. The maximum penetration rates are also shown here by the dotted lines. The rate of penetration, as is clearly shown in this figure, is directly dependent upon the amount of nitrate that was in the paper disks. 6 Hines, J.by G.ASTM and Hugill, R. W., Physical Metallurgy of Stress-Corrosion FracCopyright Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 ture, Rhodin, T. N., ed., Interscience, New York, 1959, pp. 193-223. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No f 282 STRESS CORROSION TESTING FIG. 6—Effect of nitrate concentration on penetration rates at 75 per cent relative humidity. FIG. 7—Effect of nitrate concentration on penetration rates at 43 per cent relative humidity. These rate curves are parabolic. Since the 5, 11, and 29-^.g rate curves do not appreciably slow down, it could be assumed that these concentrations could probably cause penetrations into greater depths of metal than is afforded by the size that we tested. The fastest penetration rate in terms of area obtained in this work was about 0.4 /*2/sec calculated from the maximum rate curves shown for 29-ju.g nitrate. No penetration occurred after a month in wires exposed to paper disks without nitrate. Figure 7 shows the average and maximum rate curves obtained at 43 per cent relative humidity. The rates are much slower at this humidity Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 but are still dependent upon nitrate concentration. It was surprising that Downloaded/printed by penetration occur (University at the lower humidity,pursuant since ammonium nitrate No is University of did Washington of Washington) to License Agreement. further reprodu McKINNEY AND HERMANCE ON NICKEL-BRASS ALLOY 283 not appreciably hygroscopic until the relative humidity is 62 per cent (Table 2). A possible explanation might be that the small amount of moisture adsorbed by this salt plus the amount adsorbed by the filter paper was sufficient to permit a small amount of electrolysis and migration of ions to occur. Some of the products formed might have been an extremely deliquescent salt such as zinc nitrate. (A saturated solution of this salt at room temperature gives an equilibrium humidity of 42 per cent relative humidity.5) These could then provide the necessary moisture to continue the corrosion penetration reactions. Figure 8 shows the average rate curves for wires tested at different loads. All the wires were exposed to the same nitrate concentration at either the high or low humidity. No visible penetration occurred in wires FIG. 8 —Effect of load on penetration rate at high and low humidity, with 29-m nitrate. tested under no load after 20 days. The rates, as the figure illustrates, are dependent upon the load on the wire. The wires with 10- and 15-g loads were penetrated to about the same depths; the wires under 5-g loads were penetrated at a much slower rate. Other nitrate concentrations were also tested at 5- and 10-g loads at both humidities. A similar effect of load on the penetration rate was noticed. An empirical rate equation was derived based on the parabolic curves of Figs. 6 to 8. It is expressed as dx/dt = ci/(2jc + c2) where t = time in days, and x — per cent cross-sectional penetration. The constant c^ appears to depend upon the concentration of nitrate in the paper. The constant c2 is apparently dependent upon relative humidity and load. The rate equation therefore indicates that the rate of penetration is directly dependent upon the concentration of nitrate in the paper disk and Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 inversely dependent upon the area already covered by corrosion. Downloaded/printed by Iso-Penetration Levels— To anticipate the fastestpursuant rate at which penetraUniversity of Washington (University of Washington) to License Agreement. No fur 284 STRESS CORROSION TESTING tion moves through the wires under the fixed test conditions, various points were selected from the maximum rate curves (Figs. 6 and 7) and used to plot iso-penetration curves. Figure 9 shows such a plot for penetration levels of 5, 25, 60, and 80 per cent at high humidity and indicates the minimum time that the wires FIG. 9—Minimum time to reach given depths of penetration at 75 per cent relative humidity. FIG. 10—Minimum time to reach given depths of penetration at 43 per cent relative humidity. took to reach a given depth of penetration. The lines are almost parallel and apparently level off. The plots indicate that for various penetration depths to be reached within 100 days of laboratory testing certain definite amounts of nitrate would be required. For example, it took at least 2-ju.g nitrate to cause a penetration of 80 per cent and 0.4-/xg nitrate to cause a penetration of 25 per cent in this time. Since 80 per cent penetration is Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 deep enough for surface cracking to occur, amounts of nitrate in the paper Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduction 285 McKINNEY AND HERMANCE ON NICKEL-BRASS ALLOY disks as small as 2 /xg could cause permanent wire damage at high humidity. By extrapolation of points on the figure, the minimum amounts of nitrate that would be needed to cause penetrations such as 5 and 60 per cent can be estimated. Figure 10 shows a plot for penetration levels of 5, 25, and 60 per cent at the low humidity. The 60 per cent line was obtained by drawing a line through the available point (29 /xg, 101 days) parallel to the others. A rough estimate of the shortest time that a wire exposed to 5-/ig nitrate would take at low humidity to reach 60 per cent penetration using this line would be about 5l/z years. Contrasting 5M> years with the time required by 5-^g nitrate to reach 60 per cent penetration at high humidity which is about 16 days, the importance of reducing humidity as a means of control is readily apparent. Penetration Rates at Other Fixed Test Conditions Temperature—The rate of penetration is dependent upon temperature. Wires that were tested at temperatures near 0 C showed less penetration than wires tested under the same conditions of stress and humidity at room temperature. Wires tested at higher temperatures showed greater penetration than wires tested under the same conditions at room temperature. The results are listed in the following table: Concentration NOs~ , Mg 29. 29 Test Period, days 3 5 Average Penetration at 75% RH, % 3C Room Temperature 50 C 24 37 98 6 Applied Potential (relationship to leakage current, nitrate consumption) —Wires were tested in the units at high humidity without applying potential. Penetration did occur but only when disks contained very large concentrations of ammonium nitrate. This is shown in the following table: Concentration NOa~ , jig 360. 150. 60. Test Period, days Average Penetration (two or more wires), % 4 14 40 48 30 12 No potential apparently is needed if the concentrations of nitrate in the disks are very large. But to obtain penetrations with the smaller nitrate concentrations within a rights reasonable anEST anodic Copyright by ASTM Int'l (all reserved);length Wed Decof 16 time, 15:53:43 2015 current is needed. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions 286 STRESS CORROSION TESTING Tests were also made to determine if varying the applied voltage would vary the rate of penetration. The wires were exposed to the same nitrate concentration, 29 /ng, at high humidity. Table 3 shows that, after runs of 9 days, about the same average depth of penetration was obtained with voltages between 6 to 135 v. TABLE 3—Effect of applied potential on depth of corrosion penetration (after nine days at 75% RH, 29-v-g NOr). Potential, v 0 6 45 90 135 Avg Penetration, % none 57 60 57 65 FIG. "11—Electrolysis current through assembly containing 6 nickel-brass wires at various voltages at 75 per cent relative humidity, 11-fi.g nitrate. Figure 11 shows the amount of leakage current obtained at various applied voltages, when wires were tested with 1 l-/xg nitrate at high humidity. The initial current is highest with the highest voltage, but within 2 to 3 hr the current for all the voltages tested dropped to about the same steady value of 0.1 /u.a. Approximately the same number of coulombs passed in each case. Figure 12 shows the currents obtained with various nitrate concentrations at a fixed voltage of 45 v. The current, after an initial transient peak is apCopyright dependent by ASTM Int'l reserved); Wed Dec 16 15:53:43 EST 2015 parently on (all the rights amount of nitrate present in the paper but not Downloaded/printed by 11) on the applied potential. (as was seen in Fig. University of Washington (University of Washington) pursuant to License Agreement. No further reprod McKINNEY AND HERMANCE ON NICKEL-BRASS AUOY 287 FIG. 12—Electrolysis current through assembly containing six nickel-brass wires at various nitrate concentrations at 75 per cent relative humidity, 45v. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 FIG. 13—Rate of nitrate removal from filter paper originally containing either Downloaded/printed by 11, 5, or 2-fig nitrate. University of Washington (University of Washington) pursuant to License Agreement. No further reproduc 288 STRESS CORROSION TESTING Figure 13 shows that the nitrate in the filter paper disks is only slowly removed with a 45-v potential at high humidity. This indicates that the ion migration under these conditions is slow. At low humidity, this rate is still slower. For example, after 19 days of test, no measurable amount of nitrate was removed from the disks containing 29-ju,g nitrate, even though the test wires showed an average cross-sectional penetration of 11 per cent. TABLE 4—Effect of electrolytes on rate of corrosion penetration (75%RH,29ngN03~). 3~). Salt Zinc nitrate Ammonium nitrate Calcium nitrate Copper nitrate Ammonium sulfate0 Ammonium chloride 0 Test Period, days Avg Penetration, % 5 5 5 5 30 30 39 38 18 5 0 0 This salt, after 50 days at 96% RH, causes 15% average penetration. FIG. 14—Effect of ammonium nitrate and ammonium sulfate on penetration of wires at 75 per cent relative humidity. After 40 days, 1.2-ju.g nitrate were removed from these disks, and the wires showed a 33 per cent penetration. Electrolytes (individual, mixed)—Nitrate salts other than ammonium nitrate, such as zinc, calcium, and copper, also cause a similar type of stress corrosion cracking. These data are shown in Table 4. Ammonium sulfate and ammonium chloride did not cause any penetrations after one month's testing. The test wires were heavily corroded however. At over 96 per cent relative humidity, sulfate salts can cause a similar type corrosion cracking. This would be similar to nickel-brass cracking in sulfate soluCopyright by ASTM Int'lMendizza (all rights Wed of Dec 15:53:43 tions as was studied by A. and reserved); C. A. Holden Bell16Telephone Downloaded/printed by Laboratories, Inc. University of Washington (University of Washington) pursuant to License EST 2015 Agreement. McKINNEY AND HERMANCE ON NICKEL-BRASS ALLOY 289 Wires were tested using a mixed electrolyte of nitrate and sulfate in the paper disks. The wires were examined after 18 and 42 days. Figure 14 shows the results obtained from this study. The top line shows the rate of penetration with 5-//,g nitrate alone. As the amount of sulfate is increased, the rate of penetration is decreased. However, when this test was repeated with a nitrate concentration of 11 ^.g, the retarding effect of the sulfate ions was not as great. Wires were tested with paper disks containing a known amount of an aqueous extract of the Los Angeles dust specimen (Table 1). The concentration of nitrate in the disks was 10.4 j«,g, the ratio of nitrate to sulfate in this dust extract is about 1:1. It was found that the Los Angeles dust did cause a slower penetration rate than 11-ju.g nitrate as ammonium nitrate. Test Period, da 14 41 £j!S3? NH4NOs , 11 ng NOs- 66 100 32 70 Here it would appear that the sulfate and possibly other salts in the dust extract reduced the rate. Relationship of Laboratory Rates to Field Breakage Rates It was found that the rates generally correlated with field failures, if the differences in stress were taken into account. Normally, where the nitrate accumulations on equipment were the greatest, the failure rates were greatest under the same relative humidities. In some instances, where the relative humidity was lower, breakage was noticeably less, as would be expected. Evaluation of Other Alloys Other alloys were tested with 29-/*g nitrate at high humidity for 5 days. The results are shown below. The 80-20 alloy did not show corrosion penetration under the conditions of test, while the same type of alloy with active metal additives did exhibit evidence of corrosion. Alloy Composition 80Cu-20Ni 80Cu-20Ni with Al, Mg additives. No. Wires Tested 5 23 No. Wires with Penetration 0 1 Conclusions 1. A test apparatus was developed which was useful hi obtaining the Copyright byfactors ASTM Int'l (all reserved); Wed Dec effects of various on the rates ofrights stress corrosion penetration into 16 15:53: Downloaded/printed by wires of nickel-brass or other alloys. University of Washington (University of Washington) pursuant to 290 STRESS CORROSION TESTING 2. Of all the salts tested, only those containing nitrates were capable of causing cracking or serious corrosion penetrations into the test wires at 75 per cent or lower relative humidity. 3. Large concentrations of nitrate can cause cracking of the test wire even without an applied potential. With small amounts of nitrate, noticeable corrosion penetrations occurred only when the wire was held at a positive potential. 4. The rate of corrosion penetration was found to be directly dependent upon nitrate concentration, stress, relative humidity, and temperature. Increases in any of these variables increased the penetration rate. The rate of corrosion penetration appears to depend on the sign but not the magnitude of the applied potential. 5. Large amounts of Los Angeles air-borne dusts under periods of high humidity are capable of stress corrosion cracking nickel-brass parts. At low humidity, serious cracking would not be expected to occur since the penetration rates are extremely slow. A cknowledgment The authors would like to thank D. W. Long, for his help in constructing the equipment and his aid in obtaining experimental data; Miss B. Russiello, for the quantitative determination of the various anions; and W. A. Lawrence, Jr., for his assistance in the mathematical interpretation of the data. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduc DISCUSSION ON NICKEL-BRASS ALLOY 291 DISCUSSION W. C. Harding1 (written discussion)—The authors are to be congratulated on an extremely interesting and thorough analysis of a phenomenon that is puzzling to us also. We have experienced this same stress corrosion failure of spring temper 12 per cent nickel silver relay springs. However, our failures occurred in Texas and New Jersey. The authors mention Los Angeles only. Have any other failures been noted in other locations in the United States? Have any tests been conducted on plated or coated nickel silver? Have the authors conducted corrosion tests on 18 per cent nickel silver to determine whether this composition would be superior in resistance to stress corrosion to the 12 per cent grade? Nancy McKinney and H. W. Hermance (authors)—Other areas of the United States have not reported failures of this type. A recent study was made of telephone central office dust deposits in industrial areas to determine if conditions for causing nitrate cracking existed outside Los Angeles. No cities in Texas were included in this survey. Dusts containing the largest amounts of nitrates were found in cities along the eastern seaboard with New York City, Bayonne, N. J., Baltimore, Md., and Washington D. C. leading. But these eastern seaboard dusts and dust found in other coal and oil burning industrial cities contain large amounts of sulfate and are heavy in tarry, oily, carbonaceous material which have a lesser tendency to deposit as fine wettable material. Unless the dust contamination on nickel silver contains a very large amount of nitrate, enough to overcome the inhibiting effects of its sulfate content and nonwettable nature, and unless the humidity conditions have been met, cracking would not be expected to occur. For example, in New York City after about 15 years, some stress cracking has occurred in nickel silver component parts which were within 4 ft of the windows and were not under an applied potential. The dust and corrosion on these surfaces contained 360-yu.g nitrate/in2, 2280-/j,g sulfate/in2. Nickel silver components away from these same windows did not become stress cracked. Their dust deposit contained only 9-jug nitrate/in2. We have not tested plated or coated nickel silver. An 18 per cent nickel silver wire alloy (55Cu-18Ni-27Zn) was tested by us under the conditions described in this paper. It appears to be even more susceptible to cracking than the 12 per cent nickel silver possibly due to its greater zinc content. 1 Materials and process engineer, Westinghouse Electric Corp., Jersey City, N. J. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further r D. O. Sprowls1 Reporting and Evaluating Stress Corrosion Data* REFERENCE: D. O. Sprowls, "Reporting and Evaluating Stress Corrosion Data," Stress Corrosion Testing, ASTM STP 425, Am. Soc. Testing Mats, 1967, p. 292. ABSTRACT: Susceptibility of a metal to stress corrosion implies a greater deterioration of its mechanical properties through the simultaneous action of a static stress and exposure to a corrosive environment than would be produced by the combined effects of each factor operating separately. In recent years this point of view on stress corrosion and stress corrosion cracking (the ultimate in stress corrosion) has become widely recognized. However, confusion still exists, with the result that some work is published that has little value because of the poor design of testing procedure. It is still sometimes erroneously assumed that failure of a specimen under stress corrosion test conditions suffices to demonstrate susceptibility to stress corrosion cracking. This conclusion is unjustified, for the specimen may have become so weakened by corrosion, even without any acceleration by stress, as to fail under the applied load. Thus, it is most essential to select techniques (type of specimen and method of loading, test medium and period of exposure) that produce failure that is purely the result of stress corrosion cracking. Because of the marked effect that test procedure can have upon ordinary criteria, such as specimen life, percentage of specimen survival, threshold stress, etc., investigators should report details of procedure and technique as well as detailed data along with their analysis of the results. This is necessary to enable the reviewer to adapt the information to his need and to make possible a reanalysis of the data by some method that may be developed in the future. KEY WORDS: stress corrosion, cracking (fracturing), corrosion, data, evaluation, analysis Just to define "resistance to stress corrosion cracking" is a difficult and controversial task. Although it is relatively easy to stress corrosion crack a susceptible alloy in a suitable environment, a considerable * Report of Task Group 3 of Subcommittee VI of ASTM Committee G-l on Corrosion of Metals. Task Group members are D. O. Sprowls, chairman; E. G. Haney; J. F. Hildebrand; D. S. Neill; H. R. Pritchard; W. G. Renshaw; H. B. Romans; T. J. Summerson; and D. H. Thompson. 1 Assistant chief, Chemical Metallurgy Div., Alcoa Research Laboratories, New Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Kensington, Pa. Downloaded/printed by 292 University of Washington (University of Washington) pursuant to License Agreement. No further reproductio SPROWLS ON REPORTING AND EVALUATING STRESS CORROSION DATA 293 amount of testing is necessary to determine whether or not the degree of susceptibility constitutes a practical limitation upon the use of the alloy. The latter task generally involves testing at several stress levels under conditions of loading and corrosive environment that have an established relationship to the intended service of the alloy. As increasing numbers of alloys and tempers are developed, and as applications become more sophisticated, complex, and demanding, the need for information on stress corrosion behavior of metals has become increasingly greater. Comparisons of stress corrosion test data obtained by different investigators, however, are impossible unless the test procedures and the methods of reporting the results are similar. This situation has led to the need, not only for standardized test procedures, but also for accepted methods of analyzing stress corrosion test data. Scope The considerations of Task Group 3 have been directed along the following lines: 1. Critical review of the physical and mathematical significance of various methods of reporting and analyzing data of the types required of the various test purposes. 2. Sample evaluations of stress corrosion performance using selected criteria or "representative values." 3. Compilation of a pertinent bibliography. Definitions Dix [7]2 referred to stress corrosion cracking as: "Spontaneous failure by cracking of a metal under the combined action of high stress and corrosion." Most workers in the field of stress corrosion have agreed that stress corrosion cracking results from tensile stresses at the metal surfaces acting for prolonged periods of time. Champion [2] has expanded this definition and commented as follows: "Susceptibility of a metal to stress corrosion implies a greater deterioration in the mechanical properties of the material through the simultaneous action of a static stress and exposure to a corrosive environment than would occur by the separate but additive action of those agencies." In recent years these views on stress corrosion and stress corrosion cracking (the ultimate in stress corrosion) have become widely recognized. However, confusion still exists, with the result that some work is published which has little value because of the poor design of testing procedure. It is still sometimes erroneously assumed that failure of a specimen under stress corrosion test conditions suffices to demonstrate 2 The italic numbers in brackets refer to the list of references appended to this Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 paper. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further rep 294 STRESS CORROSION TESTING susceptibility to stress corrosion cracking. This conclusion is unjustified, for the specimen may have become so weakened by corrosion, even without any acceleration by stress, as to fail under the applied load. Thus, it is most essential to select techniques (type of specimen and method of loading, test medium and period of exposure) that produce failure that is purely the result of stress corrosion cracking. Verification of Stress Corrosion Stressed Versus Unstressed Specimens It follows from the definitions presented above that resistance to stress corrosion of an alloy with unknown susceptibility can only be established FIG. 1—Distinguishing stress corrosion failure from mechanical failure [3]. by exposing both unstressed and stressed specimens to corrosive conditions. Various investigators have emphasized the importance, especially with constant load tests, of distinguishing between failures resulting from stress corrosion cracking and failures resulting from a reduction in load supporting area by corrosion [3]. The difference between these two situations is illustrated by the diagram in Fig. 1. Three identical tension specimens may be assumed as examples. Specimen A is the unexposed, unstressed blank; its breaking load is represented by the height of Block A. Specimen B is exposed to a corrosive environment in the unstressed condition. Specimen C, exposed to the same corrosive environment, is stressed by a constant load equal, for example, to three quarters of the breaking load of specimen A. DurCopyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 ing the exposure, specimen C breaks under the combined influence of Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductio SPROWLS ON REPORTING AND EVALUATING STRESS CORROSION DATA 295 stress and corrosion, and at that time specimen B is tension tested. The closer the breaking load for specimen B is to that of specimen A, the greater is the evidence of stress corrosion of specimen C and the certainty that it failed as a result of stress corrosion cracking. On the other hand, if the breaking load for B is sufficiently close to the load on specimen C to be within experimental error, then the failure should not be attributed to stress corrosion cracking, but rather to the probability that the specimen merely had corroded to the point where it could no longer support the load. Under test conditions involving prolonged exposures at ambient temperatures or elevated temperatures that could result in weakening of the alloy, failure resulting from stress rupture could be encountered; in this situation, additional unstressed and stressed specimens should be subjected to the same conditions of temperature and time but without exposure to the corrosive medium, so that the loss in strength due to corrosion can be separated from the loss due to the simultaneous metallurgical change in the alloy. As a general rule, test procedures are designed to serve best the purpose of the investigation. For some purposes, of course, tests are performed on materials of known susceptibility to stress corrosion cracking, and there is no need to expose accompanying unstressed specimens. Examples are tests to evaluate protective coatings or specific structural assemblies and quality control tests. For the purpose of quality control, a rapid test is preferred, and the standards adopted frequently have no relationship to the serviceability of the alloy. Stress Corrosion Index Jones [4] has developed a "percentage loss of strength at failure due to stress corrosion" by relating the breaking loads in Fig. 1 as follows, (B — C)/(A — C). For practical purposes this relationship may be expressed in terms of stress, and Booth et al have termed this the "stress corrosion index" [5]. where: TSU = tensile strength (based on original cross-sectional area) of unstressed specimen exposed for the time-to-failure of the stressed specimen, TS = tensile strength of unexposed specimen, and <r = nominal stress applied to stressed specimen resulting in its failure. Higher index values indicate a greater tendency toward stress corrosion Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 cracking. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions 296 STRESS CORROSION TESTING It is evident from an examination of the formula that the corrosion loss of the unstressed specimen has a great deal of influence upon the stress corrosion index and, therefore, should be determined carefully. It is advantageous to test unstressed specimens after several periods of exposure and plot a TSu versus time curve, from which TSu values can be taken for individual stressed specimen lives (or suitable representative specimen lives). It may be readily appreciated also that the shape and distribution of sites of localized corrosion (stress concentrators) can have an appreciable effect on the load required to break a stress corroding or a corroded unstressed specimen. Thus, the stress corrosion index for an alloy can be expected to vary with testing conditions and is not very useful for ranking alloys except under specific test conditions. An exTABLE 1—Comparison of stress corrosion indexes for several aluminum alloys exposed at different stress levels and in different environments.11 3.5% NaCl Alternate Immersion Industrial Atmosphere Tensile Strength Median Specimen Life, days Tensile Strength, Unstressed, psi SCI Median Tensile Speci- Strength, men UnSCI Life, stressed, days psi 45 64 5 66 0.84 30 25 20 15 10 10 48 40 32 24 16 16 2 2.5 4.5 12 73 1 59 48 44 40 36 27 0.88 0.61 0.56 0.52 0.49 0.32 Applied Stress Tensile Strength, ksi ksi 2014-T65 70 2024-T35 63 Alloy % 623 62 0.60 0 Source: Unpublished data, Alcoa Research Laboratories. Test specimens were 0.125-in.-diameter by 2-in.-long tensile bars machined transversely from rolled rod. Stressed specimens were exposed in triplicate. ample of such variation is shown in Table 1, where the SCI varies markedly for a given alloy depending upon the test environment and upon the magnitude of the applied stress. Metallographic Examination One of the most useful methods of verification of stress-corrosion cracking in a specimen corroded under stress is the microscopic examination of sections taken parallel to the direction of the applied stress and including the fracture or major crack and an appreciable length of the adjacent tensile surface. Several important factors can be established: (1) the path of the crack, that is, intergranular or transgranular, which for many alloys indicates whether or not stress corrosion cracking has occurred; (2) identification of questionable surface indications of cracks, directional corrosion, etc.;Int'land thereserved); prevalent corrosionESTand Copyright folds, by ASTM (all (3) rights Wedtype Dec of 16 15:53:43 2015the presence of stress corrosion stringers (directional corrosion perpendicular Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further r SPROWLS ON REPORTING AND EVALUATING STRESS CORROSION DATA 297 to direction of tensile stress, not present in unstressed specimen). It frequently is necessary to examine also a similarly corroded unstressed specimen to establish the evidence of stress corrosion when cracks are not detected. In addition, electron microscopic examination of replicas of fracture surface can be helpful in establishing evidence of stress corrosion cracking. The consideration of the specific characteristics of stress corrosion cracks in various alloys and environments is considered beyond the scope of this report. FIG. 2—Effect of applied current on cracking time, USS 12MoV stainless steel in aerated 3 per cent NaCl solution [6]. Anodic and Cathodic Polarization For alloys that are susceptible both to stress corrosion cracking and to hydrogen embrittlement, polarization experiments can be used to distinguish between the two mechanisms of cracking. Phelps and Loginow [6] have described the procedure and discussed the interpretation of the results in the case of high yield strength steels. In an anodic-cathodic polarization experiment with USS 12MoV steel, anodic polarization accelerated failure (Fig. 2). Cathodic polarization at low current density greatly extended the time-to-failure. On the other hand, cathodic polarization at relatively high current densities again accelerated failure. These trends indicate that (1) with no current, (2) with anodic polarization, Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 and (3) with cathodic polarization at low current densities, the observed Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions 298 STRESS CORROSION TESTING cracking was caused by corrosion along an active path through the steel, that is, stress corrosion cracking; with cathodic polarization at relatively high current densities, hydrogen was evolved and hydrogen embrittlement was the probable cause. Reporting Stress Corrosion Data Stress corrosion testing technique can have a marked effect upon ordinary criteria of stress corrosion resistance; hence, it is desirable for investigators to describe the material evaluated and the test procedure in detail. Information should be given on: (1) test material, including composition, manufactured form and dimensions, particulars of heat treatment or other tempering practice, and method of sampling; (2) type of test specimen and its orientation relative to grain structure; (3) method of applying stress (or strain); (4) possible residual stress in the test specimen; and (5) test medium, giving such details as concentration, temperature, period of exposure, etc. It is desirable also that the raw test data be reported along with the interpretation of the investigator. The basic data are valuable to readers wanting to make other comparisons. Analysis of Stress Corrosion Test Data An adequate analysis of stress corrosion test data is of very great importance. The first problem is to determine the most useful measure of stress corrosion performance. Then, in order to compare one material with another, representative values must be selected. The most promising approaches to this problem found in the literature or gained from the experience of members of this task group are discussed in the sections that follow. Specimen Life Frequently, stress corrosion testing involves determining the lives of the specimens under specific test conditions. It is well known that considerable scatter in results occurs, and it is often found that the majority of specimens in a test fail rapidly, leaving a few which fail at much longer times, or even do not fail at all before the test is discontinued. This behavior raises considerable difficulties, both theoretical and practical, in deciding when to terminate a test, choosing a satisfactory "representative value," and in comparing such values. Criteria—Various criteria of the time-to-failure, or specimen life, have been used, depending upon the test procedure. Under direct-tensile loading with weights or springs, the initiation of a stress corrosion crack generally results in rapid fracture of the specimen, and the specimen life is easily determined. There is the necessity, mentioned above, of verifyCopyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 ingDownloaded/printed that the failure isbytruly a result of stress corrosion cracking. University of Washington (University of Washington) pursuant to License Agreement. No further repro SPROWLS ON REPORTING AND EVALUATING STRESS CORROSION DATA 299 Under direct-tensile constant-deformation type loading to a relatively high stress, the initiation of a stress corrosion crack also generally causes rapid fracture of the specimen. Because of the stiffness of such stressing frames,3 however, the problem of the tensile failure of a corrosion-weakened specimen is greatly reduced and occurs only under conditions when severe localized corrosion has developed. With relatively low applied stress the time required to initiate a stress corrosion crack will be longer, and under constant-deformation load, a specimen of an alloy that is susceptible to acceleration of general corrosion by stress may develop many stress corrosion stringers without resulting in actual cracking of the specimen. In this case, comparative tension tests with corroded unstressed specimens and microscopic examination must be relied upon to reveal the tendency for stress corrosion. Constant-deformation loaded bent specimens of various types, including beams, loops, U-bends, C-rings, etc. of high-strength alloys that are appreciably susceptible to stress corrosion cracking provide fractures that are sharply defined and easy to detect. Frequently, however, with lower-strength alloys, or more resistant alloys, or with relatively low applied stress, cracking will be initiated slowly and is difficult to detect. Cracks may initiate at multiple sites, and a real problem arises in deciding when to consider a specimen "failed" and when to terminate the test. Inasmuch as these specimens do not always fracture, it is preferable to report the first crack as the criterion of failure. It is common practice to make this inspection with the naked eye or a low-power (X10 to X15) magnifying glass. If there are indications noted that cannot be established definitely as a crack by this type of examination, the investigator should either (1) note the date of this first suspicion of cracking and continue the exposure of the specimen and watch for further growth that will confirm the first indication as the failure date, or (2) discontinue exposure of the specimen and perform a microscopic examination of a cross section taken through the suspected crack area to verify the crack. Because of this difficulty, some investigators have resorted to other criteria, such as a fixed amount of "sag," or permanent set, to denote failure. Representative Values—The arithmetic mean specimen life is widely used, because it can be manipulated algebraically and can be used in most standard statistical tests of significance. It should be remembered, however, that extremely large or extremely small values may seriously affect the mean and render it atypical of the average distribution. Moreover, in using the mean it is assumed that the population is normally or very nearly normally distributed. The median, on the other hand, has the advantage that it is influenced less by extreme values and has the important property of requiring no assumption about the population dis- Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 3 See p. 342. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further repr 300 STRESS CORROSION TESTING tribution. When practicable, the population distribution should be determined. Schikorr and Wasserman [7] have suggested that specimen lives do not follow a normal distribution and that the median is a more reproducible figure. These authors also point out that median values may be obtained much faster than arithmetic mean values because only about half the number of specimens exposed need to be tested to failure. The TABLE 2—Comparison of several representative values for ranking protective coatings for preventing stress-corrosion cracking.0' Coating 7, ZnCrO4 primer Specimen Life, L Specimen Life, L Days Log L Days 43 43 43 91 641 1.63347 1.63347 1.63347 1.95904 2.80686 Coating 5, Alumilite 205 14 43 56 56 63 Log L 1.14613 1.63347 1.74819 1.74819 1.79934 Mean 172 Range 43 to 641 86 Geometric mean . . . Median 43 Coating 11, polyurethane 43 63 63 91 117 Mean 46 1.61506 Range 14 to 63 41 Geometric mean Median 56 Treatment 2, shot peened 63 1.79934 1.63347 2.04922 112 1.79934 2.04922 112 1.79934 2.38021 240 1 .95904 2.38021 240 2.06819 Mean Range Geometric mean Median 1.85188 75 43 to 117 71 63 1.93326 Mean Range Geometric mean Median 2.13164 153 63 to 240 135 112 0 Source: Ref 11. Interference ring specimens of 2014-T6 alloy stressed at 75% of the yield strength exposed in quintuplicate to the seacoast atmosphere at Point Judith, R. I. median is used in a German specification [8], which provides additionally for the use of the geometric mean if the number of specimens is small. Booth et al [5] and Booth and Tucker [9] have investigated the statistical distribution of specimen lives (endurances) in accelerated laboratory tests of aluminum-magnesium alloys in which all of the highly stressed specimens failed. These authors observed that raw specimen lives were not satisfactory for use in statistical analysis of experiments using most standard tests of significance for two reasons: (1) the variance of the specimen lives was not independent of their arithmetic mean, and by ASTM Int'lwas (all markedly rights reserved); Wed Dec 16 15:53:43 EST 2015 (2)Copyright their distribution skew. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprod SPROWLS ON REPORTING AND EVALUATING STRESS CORROSION DATA 301 Both of these objectives are overcome if the logarithms of the endurances are taken as the experimental data. Besides producing a symmetrical distribution, taking logarithms produced a distribution that appeared to be normal. These considerations apply equally well even if no statistical analysis is contemplated, and they demonstrate that the arithmetic mean of raw endurance is not a very meaningful parameter. On the other hand, the geometric mean (which is the antilog of the arithmetic mean of the logarithm) is meaningful. FIG. 3—Comparison of various rankings of protective coatings for prevention of stress corrosion cracking (Table 2). Further investigation of the frequency distribution of specimen lives for other alloys and other test conditions is needed. It should be established, for example, whether or not the population distribution of specimen life will be the same for specimens that are highly resistant as for those that are less resistant under a given set of test conditions. Further examination of stress corrosion data may reveal an analogy to fatigue testing experience which has shown that, in the finite-life range of the S-N curve, normal distributions generally result from transformation to log cycle-life; however, at stresses near the fatigue limit, where runouts are observed, normal distributions are not obtained even after Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 various transformations are performed [10]. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproducti 302 STRESS CORROSION TESTING The problem of choosing a "representative value" for comparing one item with another is illustrated by the data tabulated in Table 2 and summarized graphically in Fig. 3. The one extremely long life of a specimen with Coating 7 badly distorts the ranking using the arithmetic mean. Including either the range or the time for the first failure for each coating helps to show a more complete picture, and it appears in this FIG. 4a—Stress/cracking-time curves for a mild steel immersed in boiling LiNOa solutions [12]. FIG. 4b—Stress-cracking-time curves of steels of different compositions [13]. example that the combination of geometric mean and first failure time gives the most representative summary. Another way of presenting these types of data is discussed under section on per cent survival curve. Stress-Life Curve More information about the resistance to stress corrosion cracking of a material can be obtained by testing specimens over a range of applied Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 stresses. Such data usually are presented graphically with stress plotted Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No furthe SPROWLS ON REPORTING AND EVALUATING STRESS CORROSION DATA 303 against specimen life. The primary interest is generally in the long-life portion of the curve (stress corrosion "threshold"); therefore, more tests should be made in this vicinity. If only a relatively small number of specimens or stressing frames are available, single specimens may be tested at each stress level. When more tests can be performed simultaneously, information can be gained sooner by exposing groups of specimens at several different stress levels. In the latter procedure, each group should consist of at least four specimens if it is desired to estimate the variability of the data. (Ten or more specimens are preferable to obtain some indication as to the shape of the distribution of specimen-life values.) Furthermore, to obtain approximately an equal degree of precision throughout the range of FIG. 5—Comparison of stress corrosion tests of Al-Zn-Mg 3 under constant tensile and bending load [14]. stress levels, more specimens should be tested in the long-life than in the short-life range. Smaller groups, of course, can be tested if the specimenlife distribution shape is known or if statistical comparisons are not desired. Additional guidance may be found in ASTM Recommended Practice for Choice of Sample Size to Estimate the Average Quality of a Lot or Process (E 122 — 58). Stress-specimen-life curves generally are of a hyperbolic type as shown for mild steel [12,13] in Fig. 4a. Similar curves have been observed for other metal alloy systems, including those of aluminum [14,15], magnesium [16], austenitic stainless steel [77], and copper [18]. When the specimen lives are plotted on a log scale (Figs. 4b and 5), the curves consist of two straight line branches with one being parallel to the time axis and designating a "threshold stress" for theDec particular test EST conditions. Copyright by ASTM Int'l (all rights reserved); Wed 16 15:53:43 2015 Downloaded/printed On the basis ofbytests made on dead-weight loaded specimens of an University of Washington (University of Washington) pursuant to License Agreement. No further 304 STRESS CORROSION TESTING aluminum-zinc-magnesium alloy, Brenner and Gruhl [14], and Gruhl [79] obtained curves such as those shown in Fig. 5 and derived the following equation for the higher stress level branch where: L = a = c = K = specimen life, applied stress, constant relating to the degree of stress dependency, and constant which depends upon temperature in accordance with the Arrhenius equation. Azhogin [20] has proposed an equation for the entire curve. where: <r = applied elastic stress <Tcr = critical stress (threshold) t = specimen life K = constant The curve is a hyperbola displaced along with the o-axis by the value of <r cr . When the applied stress is equal to or less than <r cr , the alloy is not susceptible to stress corrosion cracking. When (a — o-C7.) was plotted against / on log-log paper for experimental results on tests of highstrength steels, brass, and magnesium alloy, straight lines were obtained, thus satisfying the equation for the curve. To evaluate a commercial product it can be useful to plot the stresslife points for a large amount of data for which upper and lower limit curves can be drawn (Fig. 6). The lower limit represents the threshold stress. Another alloy can then be compared with the commercial alloy by superimposing data points on the performance band for the commercial alloy (Fig. 7). Per Cent Survival Curve A curve can be constructed using the values of per cent survival of a group of specimens after a given period of exposure that are observed for several (at least three) values of applied stress. This method of analyzing stress corrosion data is especially useful when some of the specimens "run out," that is, survive the duration of the test. An example is given in Fig. 8. Similarly, per cent survival can be plotted against time of exposure (Fig. 9). These curves are applicable particularly when relatively large size groups of specimens have been tested. Curves of type, which can bereserved); plotted Wed eitherDec as per cent survival or Copyright by this ASTM Int'l (all rights 16 15:53:43 EST 2015 per cent failure, provide one of the most significant comparisons of Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No furthe FIG. by 6—Specimen and section size can16affect resistance to stress corrosion cracking oj 7075-T6 extruded sections. Typical microCopyright ASTM Int'l orientation (all rights reserved); Wed Dec 15:53:43 EST 2015 structures are shown Downloaded/printed by with arrows indicating the direction of sustained stress [15]. University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized. Copyright byFIG. ASTM7—Tests Int'l (all rights reserved); Dec 16 15:53:43direction EST 2015 compare the resistance to stress corrosion cracking of several high-strength alu taken in theWed short-transverse Downloaded/printed by forgings. The -T73 Temper of 7075 alloy was developed specifically to provide high resistance to stress corrosion cracking minum alloy University[15]. of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized. SPROWLS ON REPORTING AND EVALUATING STRESS CORROSION DATA 307 FIG. 8—Effect of environment on the resistance to stress corrosion cracking of short-transverse specimens from aluminum alloy 7079-T6 forgings (see footnote 3). Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 FIG. 9—Effectiveness of various types of protective coatings in delaying stress Downloaded/printed by corrosion cracking of 2014-16 alloy [11]. University of Washington (University of Washington) pursuant to License Agreement. No fu 308 STRESS CORROSION TESTING materials that have different degrees of susceptibility to stress corrosion cracking. While the curve can be drawn on any type of graph paper, the per cent survival values often will lie along a straight line when the data are plotted on normal probability paper (Fig. 10) [21}. Confidence limits can be computed and also plotted on the graph. FIG. 10—Distribution of stress corrosion results for USS 12MoV stainless steel [221. Stress Corrosion Thresholds Suss [22] has pointed out the many pertinent factors that can influence the corrosion process and the actual tensile stress within a member. Some of these are: (1) variations in test specimens (size, shape, stress concentrators, internal stress, and method of load application), (2) variations in environments (chemistry, temperature, galvanic cells, etc.), and (3) variations in materials nominally within the same specification. In consideration of the many variables and of possible interactions Copyrightthem, by ASTM Int'l (all be rights reserved); Wed Dec 16 15:53:43 EST 2015specific among it would hazardous to attempt to establish Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduc FIG. 11—Relative resistance of extrusions in several high-strength aluminum alloys to stress corrosion cracking. The "highest sustained tension stress that did not cause Jailure" was obtained from the bottom limit of a band drawn in the same manner as those in Fig. 6. Arrows indicate no stress corrosion failures at the highest stress employed [15]. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized. 310 STRESS CORROSION TESTING parameters below or above which stress corrosion will not occur in service. Nevertheless, for the purpose of comparing alloys, various thresholds can be used advantageously. Critical Stress—Stress corrosion thresholds in terms of the sustained tensile stress (more accurately defined as the initially applied strain or load, with allowance for residual or internal stress in the metal) determined under specific test conditions can be useful as guide lines for comparing alloys and manufactured products [75]. A graph such as that shown in Fig. 6, which represents a summary of a great many tests, is of special significance. The threshold stresses so obtained for various materials can be compared as shown in Fig. 11. It should be emphasized, however, that so-called threshold stresses determined in laboratory or in field tests should not be regarded by designers as limiting stresses, TABLE 3—Effect of test solution composition on resistance to cracking of steel.0 Solution Atmosphere pH Specimens Critical train, Sc 9 N distilled water 0.01% acetic acid 0.01% acetic acid 0.01% acetic acid H2S H2S H2S H2S 4.02 3.64 3.16 2.93 12 8 12 31 2.9 1.6 1.2 0.7 N-80 0.1% acetic acid 0.5% acetic acid H2S H2S 3.16 2.93 12 42 3.8 2.8 Alloy 0 b Data excerpted from Table 2 of Ref 24. Standard error is about 0.1 X 10~3 in./in. in the same way that fatigue or endurance limits are used for dynamically loaded members. Azhogin and Pavlov [23] have used the "critical stresses," <r cr , to compare various high-strength steel alloys and to predict their stress corrosion performance in outdoor atmospheres based on laboratory stress corrosion tests. Critical Strain—Fraser et al [24] have developed a quantitative procedure using the probit method for studying sulfide corrosion cracking of steel, by determining the degree of loading for a specific alloy at which the probability of cracking failure with the test period is one-half. The number obtained is called the "critical strain," and it is a function not only of the alloy tested but also of the test environment and procedure. Very susceptible steels have low critical strain (Se) values, whereas nonsusceptible steels have high Sc values. Using alloys of varying Sc values, all tested at the same high strain, the relative severity of any given test environment can be measured. Analysis of the data from field tests yields a severity rating, Rs, which is the critical strain of an alloy which would be expected to give 50 per cent in test. High Rs Copyright by ASTM Int'l (all rights reserved); Wed Dec failures 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further repr SPROWLS ON REPORTING AND EVALUATING STRESS CORROSION DATA 311 values are associated with severe test environments and low Rs values with mild environments. Determining the stress (or strain) at which the probability of failure is one half is a more simple approach statistically and is just as useful for comparing the relative susceptibility of two or more alloys as determining the threshold stress. The authors preferred to use strain rather than stress because the loading necessary to produce failure exceeded the yield strength in many instances. The test procedure consists of exposing a number of specimens, usually about twelve, at a variety of different strain levels for a definite time period. The test strains are chosen as close as possible to the estimated critical strain so as to obtain maximum FIG. 12—Effect of initial stress intensity on time-to-fracture, AIS1 4340 steel heat treated to about 150 kg/mm3 yield strength [27]. information from each of the limited number of test specimens. The critical strain is then calculated by the statistical technique of the probit analysis (described in detail by the authors) [24]. The probit method also is described in detail as used for making fatigue tests in Ref 10. A sample of the authors' data illustrating the effect of test solution composition on the resistance to cracking is given in Table 3. Critical Stress Intensity—Tiffany [25] and Johnson and Willner [26] have concluded the apparent existence of a stress-intensity threshold value required to initiate stress corrosion cracking in a high-strength steel. Studies by Brown and Beachem [27] at the U.S. Naval Research Laboratory on a large variety of high-strength steels and titanium alloys have reinforced this conclusion. By inserting a fatigue crack in the specimen commencing the reserved); test, stress crackingEST can2015 be Copyright before by ASTM Int'l (all rights Wedcorrosion Dec 16 15:53:43 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further 312 STRESS CORROSION TESTING caused to initiate immediately upon application of sufficient stress. The stress at the root of the precrack can be described quantitatively by the stress intensity parameter K ± , and a threshold value of this required to initiate stress corrosion cracking is defined as ^Tlscc. An example is shown in Fig. 12. Percentage Stress Corrosion Susceptibility By combining the stress corrosion index and a stress specimen life curve, Jones [4] has arrived at another parameter, percentage stress corrosion susceptibility. This is obtained by plotting the index against the applied stress expressed as per cent of the tensile strength. The crosshatched areas in the graphs in Fig. 13 were obtained by extrapolation of FIG. 13—Effect alloys [4]. of stress on corrosion of strain-aged aluminum magnesium the plotted curves EF to E', the tensile strength, and to G, the threshold stress, for these test conditions. The percentage stress corrosion susceptibility, then, is an expression of the cross-hatched area as a percentage of the total area bounded by the axes. This parameter, while it has the advantage of combining a considerable amount of test data into a single number, has some of the same limitations discussed above for the stress corrosion index. Change in Mechanical Properties Another way to evaluate quantitatively the degree of susceptibility or resistance to stress corrosion cracking is to determine the effect of cracking upon some mechanical property or physical characteristic of the test specimen. Such tests are meaningful, however, only if it is established first that the in mechanical property the16result of stress corroCopyright by change ASTM Int'l (all rights reserved); WedisDec 15:53:43 EST 2015 Downloaded/printed by merely an effect of pitting. sion cracking and not University of Washington (University of Washington) pursuant to License Agreement. No further rep SPROWLS ON REPORTING AND EVALUATING STRESS CORROSION DATA 313 Relaxation—An example of this technique is a simple test described by Thompson [25] for determining the resistance of copper alloys to stress corrosion cracking in moist ammoniacal atmospheres and in natural environments. In this procedure the resistance to stress corrosion cracking is represented by the time to produce 50 per cent relaxation (permanent set) of loop specimens. This may be found from relaxation versus time data by interpolation. The data given in Fig. 14 show how a series of copper alloys was rated by this method. FIG. 14—Stress-corrosion cracking of brasses, and brasses containing a third element in moist ammoniacal atmosphere [28]. Ductility—The relative susceptibility of titanium alloy "self-stressed" beam specimens was assessed by Braski and Heimerl [29] using the loss of bend ductility shown by a compression test. An example of the author's data is given in Fig. 15, in which the presence of and the relative degree of stress corrosion cracking is represented by the amount of shortening of the specimen at fracture. Summary The selection a criterion of the degree of 16 susceptibility Copyright by ASTMofInt'l (all rights reserved); Wed Dec 15:53:43 EST or 2015resistance toDownloaded/printed stress corrosion cracking generally is determined by the purpose of by University of Washington (University of Washington) pursuant to License Agreement. No further reproductio 314 STRESS CORROSION TESTING the test. It should be recognized that there is a great deal of inherent variability in the results of stress corrosion tests, and it is hazardous to base comparisons upon small differences in the criteria. Statistical analyses have not been used to any appreciable extent because usually too few specimens are tested and also because of the difficult problem of choosing suitable representative values. Because of the marked effect that test procedures can have upon ordinary criteria, such as specimen life, percentage of specimen survival, threshold stress, etc., it is desirable for investigators to report details of FIG. 15—Results of compression tests for Ti-8Al-lMo-lV specimens stressed at 50 ksi [29]. (Shortening at fracture as a function of exposure time at 550 F.) procedure and technique as well as detailed data along with their analyses of the results. This is necessary to enable a reviewer to adapt the information to his need and to make possible a reanalysis of the data by some method that may be developed in the future. Except in simple tests for which a crack-no-crack type of answer is adequate, it is advantageous to test a material at several different levels of applied stress, including unstressed specimens. This procedure enables the proper identification of stress corrosion failures, and permits the use of the most significant comparison methods, such as stress-per cent survival curves and stress-specimen-life curves including estimates of the critical stress or strain (thresholds). Comparisons of specimen lives are Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 ofDownloaded/printed less significance,byexcept to suggest possible trends. University of Washington (University of Washington) pursuant to License Agreement. No further repro SPROWLS ON REPORTING AND EVALUATING STRESS CORROSION DATA 315 Selecting representative values for specimen lives should be done with caution: It appears that the geometric mean or the median may be more meaningful than the arithmetic mean, but in any case, it is advisable to cite also the range of lives or time until the first failure. Additional research is needed to establish the population distribution of specimen lives under various testing conditions. The percentage stress corrosion susceptibility and changes in mechanical properties are valuable under special conditions. The burden of reporting and interpreting comparative stress corrosion test results can be lightened only as additional research leads the way to fuller understanding of the general acceptance of standardized tests or recommended practices. References [1] Dix, E. H., Jr., "Acceleration of the Rate of Corrosion by High Constant Stresses," Transactions, American Institute of Mining, Metallurgical, and Petroleum Engineers, Vol. 137, 1940, p. 11. [2] Champion, F. A., Corrosion Testing Procedures, 2nd ed., Wiley, 1965, p. 133. [3] 'The Stress Corrosion of Metals," Metals Handbook, American Society for Metals, 1948, pp. 227, 228. [4] Jones, E. L., "Stress Corrosion of Aluminum Magnesium Alloys II—Methods for Expressing Stress Corrosion Susceptibility on a Comparative Basis," Journal of Applied Chemistry, Jan. 1954, pp. 7-10. [5] Booth, F. F., Tucker, G. E., and Godard, H. P., "Statistical Distribution of Stress Corrosion Endurance," Corrosion, Vol. 10, No. 19, Nov. 1963, pp. 390t-395t. [6] Phelps, E. H. and Loginow, A. W., "Stress Corrosion of Steels for Aircraft and Missiles," Corrosion, Vol. 16, 1960, pp. 325t-335t. [7] Schikorr, G. and Wasserman, G., Zeitschrift fuer Metallkunde, Vol. 40, 1949, p. 201. [8] Deutsche Normen, DIN 50908, "Prufung von Leichtmetallen Spannungskorrosion Versuche." [9] Booth, F. F. and Tucker, G. E., "Statistical Distribution of Endurance in Electrochemical Stress Corrosion Tests," Corrosion, Vol. 21, No. 5, May 1965, pp. 173-177. [10] A Guide for Fatigue Testing and Statistical Analysis of Fatigue Data, ASTM STP 91-A, 2nd ed., American Society for Testing and Materials, Philadelphia, 1964. [H] Sprowls, D. O. et al, "Investigation of the Stress-Corrosion Cracking of High-Strength Aluminum Alloys," Eleventh Quarterly Report, 20 Jan., 1966, Contract NAS 8-5340, George C. Marshall Space Flight Center, Huntsville, Ala. [12] Parkins, R. N. and Usher, R., "The Effect of Nitrate Solutions in Producing Stress-Corrosion Cracking in Mild Steel," First International Congress on Metallic Corrosion, Butterworth's, London, April 1961, p. 289. [13] Parkins, R. N., 'The Stress-Corrosion Cracking of Mild Steels in Nitrate Solution," Journal of the Iron and Steel Institute, Vol. 172, Oct. 1952, pp. 149-161. [14] Brenner, Paul and Gruhl, Wolfgang, "Stress-Corrosion Cracking Tests of Al-Zn-Mg 3 Under Constant Tensile and Bending Strain," Zeitschrift fuer Metallkunde, Vol. 52, No. 10, 1961, pp. 599-607. [15] Sprowls, D. O. and Brown, R. H., "Resistance of Wrought High Strength Aluminum Alloys toreserved); Stress Wed Corrosion," Technical Copyright by ASTM Int'l (all rights Dec 16 15:53:43 EST 2015 Paper 17, Alcoa Research Laboratories, New Kensington, Pa. 1962. Also, published under title, "What Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized. 316 STRESS CORROSION TESTING Every Engineer Should Know About Stress Corrosion of Aluminum," Metal Progress, Vol. 81, No. 4, April and May 1962. [16] Loose, W. S. and Barbian, H. A., "Stress-Corrosion Testing of Magnesium Alloys," ASTM-A1ME Symposium on Stress-Corrosion Cracking of Metals, 1944, American Society for Testing and Materials, Philadelphia, 1945, pp. 273-292. [17] Denhard, E. E., Jr., "Effect of Composition and Heat Treatment on the Stress-Corrosion Cracking of Austenitic Stainless Steels," Corrosion, Vol. 16, 1960, pp. 359t-369t. [18] Bulow, C. L., "Stress-Corrosion Testing of Copper-Base Alloys," ASTMAIME Symposium on Stress-Corrosion Cracking of Metals, 1944, American Society for Testing and Materials, Philadelphia, 1945, pp. 19-35. [19] Gruhl, Wolfgang, 'The Temperature Dependence of the Stress-Corrosion Cracking of Al-Zn-Mg 3," Zeitschrift fuer Metallkunde, Vol. 53, 1962, pp. 670-675. [20] Azhogin, F. F. /'Corrosion Cracking of High Strength Steels," Inter crystalline Corrosion and Corrosion of Metals Under Stress, Consultants Bureau, New York, 1962. [21] Loginow, A. W., "Specimens Used in Stress-Corrosion Testing of Alloys," Twenty-first Annual Conference of the National Association of Corrosion Engineers, St. Louis, Mo., 15-19 March, 1965. To be published in Corrosion. [22] Suss, Henry, "Practicality of Establishing Threshold Values to Eliminate Stress Corrosion Failures in Metals and Alloys," Corrosion, Vol. 17, No. 2, Feb. 1964, pp. 83-88. [23] Azhogin, F. F. and Pavlov, Yu. L., 'Tendency of Steel Toward Corrosion Cracking in Various Media," Korroziya i Zashchita Metallov, Sbornik Statey, Moscow, Oborongiz, 1962, pp. 112-117. (Translation FTD-TT 64-643, Foreign Technology Div., Air Force Systems Command, Wright-Patterson Air Force Base, Dayton, Ohio.) [24] Fraser, J. P., Eldredge, G. G., and Treseder, R. S., "Laboratory and Field Methods for Quantitative Study of Sulfide Corrosion Cracking," Corrosion, Vol. 14, No. 11, 1958, pp. 517t-523t. [25] Tiffany, C. F., "Progress in Measuring Fracture Toughness and Using Fracture Mechanics," Fifth Report of a Special ASTM Committee, Materials Research & Standards, Vol. 4, No. 3, March, 1964, p. 107. [26] Johnson, H. H. and Willner, A. M., "Moisture and Stable Crack Growth in a High Strength Steel," Applied Materials Research, Vol. 4, No. 1, Jan. 1965, p. 34. [27] Brown, B. F. and Beachem, C. D., "A Study of the Stress Factor in Corrosion Cracking by Use of the Precracked Cantilever Beam Specimen," Corrosion Science, Vol. 5, 1965, pp. 745-750. [28] Thompson, D. H., "A Simple Stress-Corrosion-Cracking Test for Copper Alloys," Materials Research & Standards, Vol. 1, No. 2, Feb. 1961, p. 108. [29] Braski, D. N. and Heimerl, G. J., 'The Relative Susceptibility of Four Commercial Titanium Alloys to Salt Stress Corrosion at 550°F," NASA Technical Note D-2011, National Aeronautics and Space Administration, Dec. 1963. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized. Af. B. Shumaker,1 R. A. Kelsey,1 D. O. Sprowls,1 and J. G. Williamson'i Evaluation of Various Techniques for Stress Corrosion Testing Welded Aluminum Alloys REFERENCE: M. B. Shumaker, R. A. Kelsey, D. O. Sprowls, and J. G. Williamson, "Evaluation of Various Techniques for Stress Corrosion Testing Welded Aluminum Alloys," Stress Corrosion Testing, ASTM STP 425, Am. Soc. Testing Mats., 1967, p. 317. ABSTRACT: In the development of high-strength, weldable aluminum alloys it is necessary to determine the resistance to stress corrosion cracking of experimental combinations of parent plate and filler alloys. Several types of specimen and methods of loading have been studied to evaluate techniques suitable for rapid screening of alloys and for demonstrating the serviceability of alloys. Stress corrosion tests have been conducted in 3.5 per cent sodium chloride by alternate immersion and in seacoast and industrial atmospheres, comparing beam and tension specimens. The beam specimens were loaded by constant deformation, and tension specimens were loaded both by constant deformation and by constant load. Welded specimens designed to investigate the effect of residual welding stresses upon both butt welds and fillet welds also are included. KEY WORDS: corrosion, stress corrosion, weldments, aluminum alloys, salt solutions, chlorides, exposure testing, residual stresses Stress corrosion has not been a problem with weldments of commercial aluminum alloys. Many years of satisfactory service can be cited for welded structures of various alloys of the aluminum-manganese, aluminum-magnesium, aluminum-magnesium-silicon, and aluminum-copper types. The search for ever stronger aluminum alloys as materials of construction, however, leads investigators into complex alloy systems for which the resistance to stress corrosion is a necessary consideration. Also, the trend toward larger and more sophisticated structures in the transpor1 Research engineer, Chemical Metallurgy Div.; research engineer, Engineering Design Div.; and assistant chief, Chemical Metallurgy Div., respectively, Alcoa Research Laboratories, New Kensington, Pa. 2 Head, Corrosion and Finishes Sections, Materials Div., Propulsion and VeCopyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Ala. hicle Engineering Laboratory, George C. Marshall Flight Center, Huntsville, Downloaded/printed by 317 University of Washington (University of Washington) pursuant to License Agreement. No further reprod 318 STRESS CORROSION TESTING FIG. 1—The extent of metallurgical changes in various aluminum alloys caused by ASTM Int'l (all rights reserved); Dec 16measurements. 15:53:43 EST 2015 byCopyright heat of welding as indicated by potential andWed hardness Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further re SHUMAKER ET AL ON WELDED ALUMINUM ALLOYS 319 tation and aerospace industries has resulted in some problems with residual and locked-in assembly stresses. Hence, there is a need for reliable methods for testing the resistance to stress corrosion of weldments of new alloys and tempers. Stress corrosion testing a weldment presents a number of complications that are not encountered in the testing of the parent metal. Fusion welding not only introduces a region of metal with a cast grain structure, but also creates a number of metallurgical and mechanical property Specimen dimensions for vdriOuS plate thicknesses , inches Formula for stressing Where Ad = deflection, inch f = nominal stress , psi £ = Young's modulus, psi J_ _o_ b_ 1/8 1/4 4 4 2 2 3/8 1/2 3/4 1 43/4 43/4 31/2 31/2 5'/2 H/2 6'/2 6 8 12 6 s. .k 10 10 13 13 17 12 12 15 15 19 20 25 22 27 FIG. 2—Beam stress corrosion assembly. changes of the parent metal in the heat-affected zone (HAZ). These changes have variable effects on the hardness and upon the electrochemical properties of different aluminum alloy systems (Fig. 1). Such changes are especially important in the case of heat treated alloys and can have a bearing on the choice of the most suitable test procedure. It is the purpose of this paper to compare the merits of several test procedures that have been used for evaluating aluminum alloys at the Alcoa Research Laboratories. This program was designed to provide information useful not only to the producers and the users of aluminum alloys but it is anticipated that this comparison of test methods will be of Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 value also to the producers and the fabricators of other metals. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduc 320 STRESS CORROSION TESTING Test Specimens and Methods of Loading Four techniques have been evaluated at the Alcoa Research Laboratories for stress corrosion testing welded aluminum alloys. A description of the types of specimens, methods of stressing, and a discussion of the advantages and disadvantages of each follows. Simple Beam Specimen The beam specimen (Fig. 2) has been in use for over 20 years and has been found satisfactory for all types of aluminum alloys. This specimen is simple to make from any thickness of plate and is easy to stress in a quantitative manner. Beams may be conveniently stressed in pairs (Fig. FIG. 3—Calibration tests of welded beam-type stress corrosion specimen. 2), or they may be stressed individually and grouped in racks as has been done by some other investigators [I].3 Beam deflections required to develop the intended tensile stress are calculated with the formula given in Fig. 2 and are then applied by bolting together the ends of the beams. The deflections are measured with a dial gage to within ±0.0005 in. Thus the error in stress application, if the beams were of homogeneous material arid the cross section were uniform, is within 2 per cent; the precision of the deflection measurement is within 0.5 per cent; arid the error in determining Youftg's modulus of elasticity (ASTM Method for Determination of Young's Modulus at Room Temperature, E 111-61) is within 1 per cent. The extent to which measured stresses in welded beam specimens agree with the nominal stress Was determined from tests on welded l-in.-thick plate. Strains were measured immediately adjacent to the weld bead arid 3 The italic numbers in brackets refer to the list of references appended to this Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 paper. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions au SHUMAKER ET AL ON WELDED ALUMINUM ALLOYS 321 in the parent material outside the HAZ by means of foil type electrical resistance strain gages, located as shown in Fig. 3. Calibration tests under tensile loading showed the average measured strains in the parent material to be in close agreement with the strain calculated by dividing the applied load by the cross-sectional area and Young's modulus. The results obtained in the bend tests on X7106-T6351 are shown in Fig. 3. In these tests the beam specimens were deflected by an amount calculated to stress the parent metal to 28.7 ksi (75 per cent of the 10- FIG. 4—Welded U-bend stress corrosion specimen. in. gage length yield strength of the weldment). For this case, involving essentially elastic strains, the stress in the parent material outside the HAZ was within 5 per cent of the nominal stress; while in the HAZ adjacent to the weld, the stress, concentrated by the close proximity of the weld bead fillet, was about 20 per cent greater than the nominal stress. The results of the bend tests of 2219-T8 type alloy beam specimens are shown in Fig. 3. In these tests it was desired to stress the parent metal to 30 ksi (75 per cent of the 10-in. gage length yield strength) for which the calculated beam deflection was 0.46 in. The calibration showed that a deflection of 0.46 in. resulted in an elastic strain in the center span, outside HAZ,Int'l of(all 2800 to a stress of 29 ksi (also Copyrightthe by ASTM rights|uin./in., reserved); equivalent Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions 322 STRESS CORROSION TESTING within 5 per cent of the nominal stress). In this instance, the material immediately adjacent to the weld bead underwent plastic deformation, the measured strain being 4500 /iin./in., equivalent to a stress of only 22 ksi. The latter stress was determined from a stress-strain curve obtained by a tension test of a full section welded specimen, on which strains were measured in the HAZ by electrical strain gages. These data show that the stresses outside the HAZ are in close agreement (within 5 per cent) with the nominal values, regardless of the strains in the HAZ. Although in tensile loading the stress in the HAZ will be at least equal (or higher, due to stress concentration) to that in the parent material, regardless of the amount of local yielding in the HAZ; it must be realized that in bending, local yielding may result in a lower stress in FIG. 5—Welded sheet tension specimen. the HAZ than in the parent metal. Thus, to make direct comparisons between bending and tension tests, where local yielding occurs, it is desirable to develop the same strain in the HAZ, accepting the fact that the stress in the parent material will be greater in the bent specimen. V-Bend Specimen The U-bend specimen (Fig. 4) is a qualitative, highly stressed specimen that has been widely used for various metals with and without welds. It is of limited value and is realistic only for certain applications that involve severe deformation of the welds. At the Alcoa Research Laboratories it has been used chiefly as a rapid screening of those welds that are sensitized by plastic deformation [2]. The U-bend consists of a rectangular strip, its length depending upon the desired bend radius, bent 180 deg around a mandrel. The bent specimen is allowed to spring back elastically to a stable position and then stressed by bringing it back to the 180 deg position and holding it in Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduction SHUMAKER ET AL ON WELDED ALUMINUM ALLOYS 323 FIG. 6—Constant-deformation type of stressing frame used for 0.125-in.-thick sheet tension specimens in direct tension. FIG. 7—Constant-load type of stressing frame for 0.125-in.-thick sheet tension specimens in direct tension. place by means of a bolt. This produces the same stress state that was in the specimen hi the original bent condition. Tension Specimen Tension specimens of the general type shown in Fig. 5, like the Ubend, have been used extensively for testing various materials. A quantitative stress can be applied by a variety of methods ranging from constant load to constant deformation. The stressing frames (Figs. 6 and 7) were designed and built at Alcoa Research Laboratories to use in a study of Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further re 324 STRESS CORROSION TESTING weld testing methods [3], Practical considerations limit these methods of stressing to relatively thin specimens. The constant-deformation type stressing frame (Fig. 6) is a modification of the one used by the Rocketdyne Division of North American Aviation, Inc.4 that, in turn, uses the principle used much earlier for round tension specimens at the Alcoa Research Laboratories [4]. To load the specimen in this stressing frame, the side struts are simultaneously forced inward toward the specimen until the desired strain in the specimen is achieved. A special loading device is used to obtain steady and equal movement of the side struts to assure uniaxial stressing of the specimen. FIG. 8—Application of load to tension specimen in the constant-load stressing frame. In both stressing frames (constant-deformation and constant-load), the strain equivalent to the desired stress was measured over a Vi-in. gage length to ±1 pin. by means of an electrical strain gage placed adjacent to the weld heat-affected zone. With both types of stressing frames, the strain, equivalent to the intended stress (according to Hooke's law), is measured by means of an electrical extensometer attached to the specimen just outside of the heataffected zone on one side of the weld. Because of warpage during welding the specimens may not always be perfectly flat and some bending stress may be developed during loading. This effect, usually very small, is averaged out by taking strain readings on both surfaces of the specimen (Fig. 8). Because of the high precision of the strain measuring technique, the error of stress application is governed by the error inherent in Young's modulus (less than 1 per cent). 4 Private communication, J. W., Rocketdyne North American Copyright by ASTM Int'l Greenwood, (all rights reserved); Wed Dec Div., 16 15:53:43 EST 2015 Aviation, Inc. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No furth SHUMAKER ET AL ON WELDED ALUMINUM ALLOYS 325 An investigation was performed to determine the extent to which the constant-deformation and constant-load fixtures (Figs. 6 and 7) approach the intended conditions of loading. First the relationship between load and fixture stiffness was experimentally determined as indicated in Figs. 9a and b. Load-deflection data were obtained for specimens slotted at their midlength to give reduction in area of 0, 20, 40, 60, and 80 per cent (Figs. 9a and c). The stress corresponding to a given slot (crack) depth was found by trial and error. Let P° be the force required to stress a FIG. 9—Procedure used for determination of fixture stiffness cracking on specimen load-deflection relationship. and effect of smooth specimen to the initial stress <T° and P' the force accompanying a given crack depth. Then to satisfy equilbrium requirements and where: s = specimen, and / = fixture. Let AL°g be the initial change in specimen length produced by P°s and Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 AL'S the change in length corresponding to P's. The change in specimen Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further rep 326 STRESS CORROSION TESTING FIG. 10—Effect of stressing frame stiffness (K) and increasing depth of cracking on average net section stress in sheet-type tension specimen. length, J(AL) S , may be found as shown in Fig. 9c. To maintain equilibrium, the incremental change in specimen length must equal the incremental deformation of the fixture or where the values of K° and K' being obtained from Fig. 9b. From the value of P' which satisfies the above requirements, the average net area stress is where A' is the net area for the given slot (crack) depth. The variation of net section tensile stress with the per cent reduction of net section through a propagating stress corrosion crack under various Copyright of by ASTM Int'lis(allplotted rights reserved); 16 15:53:43 2015 were inmethods loading in Fig.Wed 10.Dec Two limitingESTcurves Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduction SHUMAKER ET AL ON WELDED ALUMINUM ALLOYS 327 eluded, one for dead-weight loading when the stiffness, K = 0, and the other for an infinitely stiff stressing frame for which K = oo. It can be seen from the curves that the average tensile stress on the net section increases most rapidly with stressing frames having the lowest stiffness. The curve for the constant-load stressing frame, as expected, almost FIG. 11—Typical H-plate specimen for stress corrosion tests of weldments. Berry mechanical strain gage is in position for measurement just outside the heataffected zone of center section. duplicates the dead-load curve. The curve for the constant-deformation stressing frame, however, instead of duplicating the curve for an infinitely stiff stressing frame, tends to approach the curve for the constant-load stressing frame. This is because there is a certain amount of elasticity in the frame components. The similarity of the curves for the two stressing frames also is associated with the assumption of localized cracking, which is the usual case for susceptible weldments. In testing materials other Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 than weldments, however, generalized cracking or intergranular attack Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further rep 328 STRESS CORROSION TESTING may occur, and the similarity of the curves would be decreased. Calculations have shown5 that the curve for a similar constant-deformation stressing frame in the case of generalized cracking approaches closely the limiting curve for an infinitely stiff fixture. Residual Tensile Stress Butt Welds—The H-plate specimen (Fig. 11) was designed as a selfstressed specimen for testing thick plate in direct tension by utilizing the residual stresses developed in plates welded under constraint. Tensile stresses are developed across the weld as a result of the thermal mismatch and plastic deformation accompanying welding of the center section. The H-plate is a modification of a specimen developed at the Alcoa Research FIG. 12—Sandwich specimen simulating rigid structure. Note stress corrosion cracking in edge of center plate of this specimen immersed for 2 hr in 6 per cent boiling sodium chloride solution. Laboratories for the measurement of residual stresses developed in weldments. The advantages of this type of specimen are that special stressing frames are not required, any thickness of plate can be used, and the stresses are representative of those produced by constraint welding and can be measured. Moreover, the stresses can be adjusted by reducing the width of the center or side struts after welding. A disadvantage is that residual stresses are inherently variable and the reproducibility of the specimen is not as high as that of the previously described specimens. The magnitude of the tensile stress developed in the test section was determined from strain measurements made before welding and after final sizing. Figure 11 shows the Berry mechanical strain gage in position for measurement, just outside the heat-affected zone of the center section. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 5 See p. 342. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further SHUMAKER ET AL ON WELDED ALUMINUM ALLOYS 329 Fillet Welds—Most of the weldable aluminum alloys have a good resistance to stress corrosion cracking in the short-transverse direction; therefore, residual stress developed across the edges of the plate during fillet welding do not introduce a stress corrosion hazard. Care must be taken, however, to avoid exposure of the short-transverse grain structure when stressed in tension for two notable exceptions; namely, certain heat treated aluminum-zinc-magnesium alloys and improperly processed strain hardened aluminum-magnesium alloys containing more than 4 per cent magnesium. FIG. 13—Sections from sandwich specimens of 1.5-in.-thick 7039-T6 plate after exposure to boiling 6 per cent sodium chloride solution for 96 hr. Failure was observed after only 2 hr in test for specimen at left with no protection. The specimen at right which had the edges of the plate overlayed with 5356 filler alloy was free from cracking. To simulate rigid assemblies involving fillet or butt welds situated close to the edge of a plate, a sandwich type of specimen (Fig. 12) has been devised. This type of specimen consists of a center panel with machined edges, with slightly smaller outer panels welded on each side. Another type of specimen used for this purpose has been obtained by removing slices from a cruciform weld-cracking specimen [5]. Critical stub lengths (Fig. 13) for both this and the sandwich specimen range up to about 1.5 times the plate thickness. As in the case of the H-plate specimen, the fillet weld specimens are realistic in that they contain actual welding stresses, and elaborate testing fixtures are not required. A disadvantage, however, is that the residual stresses cannot be controlled readily, and the specimens do not provide a Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 high degree of reproducibility. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further 330 STRESS CORROSION TESTING Test Environments Test environments used by the Alcoa Research Laboratories include the 3.5 per cent sodium chloride (NaCl) alternate immersion (1-hr cycle with 10-min immersion followed by 50-min drying in air at ambient room temperature), the seacoast atmosphere at Point Judith, R.I., and the industrial atmosphere at New Kensington, Pa. In the atmospheric exposure, specimens were exposed at 45 deg angle on racks facing south. Test Materials A description of the composition and properties of alloys for which stress corrosion data are cited is given in Table 1. TABLE 1—Nominal chemical composition of aluminum alloys discussed in this paper (weight per cent). Alloy 5456 5556° 2014 40432219 2319° 7039 X7139 X7106 X5180-. 51830 Cu 4.5 6.3 Si 0.8 5.25 Mn Mg 0.8 0.8 5.25 5.25 05 0.8 Zn Cr 0.10 0.10 0.3 0.25 0.25 0.25 0.45 0.75 2.8 2.8 2.2 4.0 4.8 Zr 4.0 4.0 4.2 2.0 0.20 0.10 0.10 0.15 Ti V 0.10 0.15 0.15 0.15 0.10 0.10 0.15 0.15 0.10 0.10 Filler alloy only. Experimental Results Data from the Alcoa Research Laboratories' files were chosen to illustrate the utility and the engineering significance of the various techniques described above for stress corrosion testing welded aluminum alloys. Specimens Stressed in Bending Simple Beam Versus U-Bend—Simple beam and U-bend specimens have been used extensively in the development of aluminum-magnesium alloys. The simple beam specimen is used as a practical specimen to demonstrate the serviceability of plate filler combinations. The U-bend specimen is used as a screening test to investigate such effects as plastic deformation (strain), magnesium content, and long-time natural aging upon the aluminum-magnesium alloys. Plastic deformation (strain) will accelerate the precipitation of aluminum-magnesium constituent during natural aging of alloys of relatively Copyright by ASTM Int'l (allcausing rights reserved); Wed Dec 16 susceptible 15:53:43 EST to 2015 high magnesium content them to become stress Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further repro 331 SHUMAKER ET AL ON WELDED ALUMINUM ALLOYS corrosion cracking [2]. An illustration of the effect of plastic deformation of welds is given in Table 2 for 5456-H321 alloy welds. Fortunately, the specific sequence of forming, heating, and stressing that developed susceptibility to stress corrosion cracking in this test are not likely to occur in industrial uses of welded aluminum alloys. Under the more realistic sequence of heating, forming, and stressing, using both U-bend and simple beam specimens, excellent resistance to stress corrosion TABLE 2—Effect of plastic deformation on the resistance to stress corrosion cracking of Y^-in.-thick 5456-H321 alloy welded plate.° Beam Specimens Stressed 75% of Weldment Yield Strength (4-in. gage length) U-Bend Specimens (4< radius) Formed, Stressed, Formed, Heated 1 Week Formed and Stressed and Heated 61 Week at 212 Ffr, and Stressed 212 F F/N C Time in Test F/NC Time in Test F/NC Time in Test As Welded 1 Week 212 F6 Time in Test F/N C Time in Test 4 years 0/8 4 years F/NC 3.5% NaCl ALTERNATE IMMERSION 0/4 4 years 0/4 4 years 4/4 5, 5, 33, 63 0/4 days SEACOAST ATMOSPHERE—POINT JUDITH, R.I. 0/2 9 years 0/2 9 years 2/2 212, 436 days 0/4 0/8 11A years 1Y2 years INDUSTRIAL ATMOSPHERE— NEW KENSINGTON, PA. 0/2 9 years 0/2 9 years 2/2 859, 859 days 0 Butt-welded in three passes with parent filler by the consumable electrode process, direct current straight polarity-metal inert gas (DCSP-MIG). Tensile properties of weldment—tensile strength, 50.8 ksi; yield strength (0.2% offset in 4-in. gage length), 27.9 ksi; % elongation in 4-in., 15.7. 6 Heating 1 week at 212 F used to simulate long-time natural aging at room temperature. c F/N = number of stress corrosion failures over number of specimens exposed. cracking has been demonstrated. It is significant that aluminum-magnesium alloys containing 4 to 5.5 per cent magnesium have been used extensively for welded constructions of various types with eminently satisfactory results. Specimens Stressed in Tension Constant Deformation Versus Constant Load—The data in Table 3 obtained on aluminum-zinc-magnesium and aluminum-copper alloys Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 show that the constant-deformation and the constant-load stressing Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprodu TABLE 3—Resistance to stress corrosion cracking of DCSP-TIG welded %-in.-thick sheet of several heat treated aluminum alloys loaded by different methods. Original Properties °hcct \lloy 2014-T6 (as welded) 2219-T87 (welded as 2219-T37; aged 24 hr at 325 F) Mloy 4043 2319 Condition bead on Tensile Strength, ksi Yield Strength, ksi Elongation in 2 in., % 52.0 36.4 3 Loading Method0 Industrial Atmosphere 3.5% NaCl Alternate Immersion F/N5 Days 0 F/N & BCD TCD TCL 0/4 1/2 0/2 84 (1) 84 0/4 0/2 84 bead off 49.6 35.1 4 BCD TCD TCL 0/2 0/2 0/2 84 84 84 bead on 45.4 40.7 1 BCD TCD TCL 0/4 0/2 1/2 84 84 (43) 84 bead off 49.1 41.2 2 BCD TCD TCL 0/4 0/2 0/2 84 84 84 0/2 0/4 0/2 0/3 484 498 498 835 815 X5183 bead on 55.1 38.1 5 BCD TCD 0/4 0/3 180 180 0/4 0/3 7039-T6 (welded as 7039-T6; aged 8 hr at 225 F + 16 hr at 300 F) X5183 bead on 61 .4 50.1 6 BCD TCD 0/4 2/3 (137), (151) 180 0/4 3/3 BCD TCD 0/4 0/3 (143)" 180 180 0/4 2/3 TCL 0/2 (123)* 180 3/3 X7139-T6 (as welded) X5180 bead on 55.1 38.1 5 Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized. 484 498 498 7039-T6 (as welded) 180 Days" 812 (153), (421), (545) 835 (526), (625), 815 (481), (483), (530) 61.4 50.1 X7139-T6 (welded as X5 180 X7139-T6; aged 8 hr at 225 F + 16 hr at 300 F) bead on X7106-T6 (as welded) bead on 53.0 37.7 bead on 59.8 50.0 X7106-T6 (welded as X7106-T6; aged 8 hr at 225 F + 16 hr at 300 F X5180 bead off 51.0 45.4 6 BCD TCD 1/3 3/3 (6) 180 (2), (2), (2) 2/2 3/3 TCL 3/3 (D, (2), (2) 3/3 5 BCD TCD TCL 0/4 0/3 0/3 180 180 (129)* 180 0/4 0/3 1/3 835 815 (584) 815 6 BCD TCD 1/3 3/3 (13) 180 (2), (3), (5) 2/2 3/3 TCL 3/3 (4), (4), (5) 3/3 (83), (150) (45), (58), (58) (69), (73), (73) BCD TCD 0/4 0/3 180 180 0/4 2/3 TCL 0/3 180 3/3 3 (87), (164) (21), (34), (35) (47), (47), (54) 469 (113), (113), 478 (116), (165), (197) * Loading Methods: BCD = bending, constant deflection (Fig. 2); TCD = tension, constant deflection (Fig. 6); and TCL = tension, constant load (Fig. 7). In all cases parent metal outside of HAZ stressed to 75 per cent yield strength in 2-in. gage length. 6 F/N = number of stress corrosion failures over number of specimens exposed. c Number in parentheses is the time required for failure of specimen(s) that failed. d These failures, unlike those in the post weld aged aluminum-zinc-magnesium alloys which occurred along the fusion line, occurred in the heat-affected zone of the parent metal. At this location, severe undermining corrosion occurred causing concentration of the stress and initiation of small intergranular stress corrosion cracks extending parallel to the surface of the sheet. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized. 334 STRESS CORROSION TESTING frames were equally effective, and there was no tendency for specimen lives to be shorter under constant load, as might be inferred from a comparison of the curves in Fig. 10. Thus, the simpler and less expensive constant-deformation stressing frame is preferred for testing weldments when the application of direct tensile stress is desired. Residual Tension (butt welds)—The procedure used in making the Hplate specimens represented an extreme effort to develop maximum residual tensile stress across the 2-in. weld test section. The actual tensile stresses measured in as-welded H-plates of X7106-T6 alloy averaged about 15 ksi for tungsten inert gas (TIG) welded Vs-in. sheet, 20 ksi for TIG or metal inert gas (MIG) welded %-in. plate and 26 ksi for MIG welded 1-in.-thick plate. Post weld aging the specimens (8 hr at 225 F + 16 hr at 300 F) tended to reduce the above values only by about 2 to 3 ksi. Similar stresses were also obtained in Vs-in. and %-in.-thick H-plates of other aluminum-zinc-magnesium alloys and of aluminum-copper alloys 2014-T6 and 2219-T87. Stress corrosion failures have occurred, which indicate that a surface residual stress transverse to the weld of sufficient magnitude to cause stress corrosion cracking of post weld aged XV106 and X7139 alloys can be developed in H-plate specimens. Only a few H-plates have failed to date, suggesting that the stresses are close to stress corrosion thresholds for these alloys. This is the first investigation in which the H-plate has been used, and the information available to date indicates that this type of specimen is worthy of further trial, especially for relatively thick sections, about % in. and above in thickness. Residual Tension (fillet welds)—In sandwich specimens (Fig. 12) with test plate of 1.5-in.-thick 7039-T6, extensive stress corrosion cracking occurred in the plate edges after only 2-hr immersion in boiling 6 per cent NaCl solution. Using this specimen in an investigation of various protective measures to prevent this cracking, it was found that one of the most effective procedures was to "butter" the edges, that is, to lay down on the edges a protective coating of filler alloy (Fig. 13). Bending Versus Tension The first stress corrosion tests of welded joints providing a direct comparison of tension versus bending were performed on welded 2014-T6 sheet. Two tension specimens (Fig. 5) were machined from a butt welded panel 0.090 in. thick. One specimen was stressed as a beam, and the other was stressed in tension in a constant-deformation fixture similar to that shown in Fig. 6. The tension specimen was loaded to 25 ksi as measured in the parent metal outside the HAZ by means of a mechanical strain gage. The strain developed in the HAZ adjacent to the weld bead was Copyright by ASTM Int'la(all rightselectrical reserved); Wed Decgage. 16 15:53:43 EST 2015 also measured using VHn. strain The beam specimen Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprod SHUMAKER ET AL ON WELDED ALUMINUM ALLOYS 335 FIG. 14—Representative of failures of TIG welded X7106-T6 and X7139-T6 sheet (VI in. thick) employing X5180 filler wire (post -weld aged). Etch-.Keller's. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduct 336 STRESS CORROSION TESTING having a similarly located electrical strain gage was then loaded to the same local strain in the HAZ as determined for the tension specimen. When placed in the 3.5 per cent NaCl alternate-immersion test, the tension specimen failed in the HAZ adjacent to the fusion line after 2 days, but the beam specimen was still intact after exposure for 144 days. Metallographic examination of both specimens established the evidence of stress corrosion cracking in the tension specimen and the absence of cracking in the beam specimen. Welded Vs-in.-thick sheet of several aluminum-copper and aluminumzinc-magnesium alloys was evaluated using beam specimens and directtension specimens. In all cases the specimens were loaded by constant deformation to develop a tensile stress in the parent metal outside the HAZ equal to 75 per cent of the 2-in. gage length yield strength of the weldment. The test data are summarized in Table 3. Of the items that showed some susceptibility to stress corrosion cracking, the percentage of failures was higher and the specimen lives were shorter for the tension test (TCD) than for the beam test (BCD). A recap of the failures is as follows: No. Failures/No. Specimens Exposed Alternate Immersion 2Q14-T6 7039 -T6 PWA° X7139-T6, PWA X7106-T6, PWA Total 0 Tension Bending Alloy 0/4 0/4 1/3 1/3 Atmosphere 0/4 2/2 2/2 2/14 4/8 6/22 ( 27%) Alternate Immersion 1/2 2/3 3/3 3/3 Atmosphere 2/3 3/3 3/3 8/9 9/11 17/20 (85%) PWA = post weld aged. Thus, the tension test was shown to be more discriminating, particularly when testing an alloy in the vicinity of its stress corrosion threshold. It would be expected, therefore, that stress corrosion cracking threshold stresses indicated by tensile loading would tend to be slightly lower than those indicated by beam specimens. A typical stress corrosion failure of the post weld aged aluminum-zinc-magnesium alloys when loaded either by bending or tension is shown in Fig. 14. These results are in general agreement with the findings of Brenner and Gruhl [6] in tests on aluminum-zinc-magnesium alloy sheet (not welded). It is of interest to consider possible reasons for the difference in results of the bending and the tension tests. Although it is difficult to compare Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 the state of stress at the crack tip in bending and tension specimens, some Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further r SHUMAKER ET AL ON WELDED ALUMINUM ALLOYS 337 FIG. 15a—Distribution of yield strength and residual stresses in a longitudinally welded 5456-H321 plate 36 in. wide and Vi in. thick. FIG. 156—Residual stresses produced transverse to weld by multipass welding. (ll/4-in. 7039 aluminum alloy plate MIG welded with ^LQ-m.-diameter X5180 electrode). information may be gained by comparing the energy stored in the two types of specimens. Since no external energy is applied to the system during stress corrosion cracking, the energy required to propagate the crack must be supplied by the release of stored elastic-strain energy in the specimen and the stressing frame. There are at least two factors which Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 contribute to greater stored energy for specimens loaded in tension as Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprodu 338 STRESS CORROSION TESTING compared to those loaded in bending. First, with increasing crack depth, the potential energy corresponding to the tensile stress field remains greater in a tension specimen than in a beam specimen. Second, for tensile stressing frames of the types shown in Figs. 6 and 7, considerable energy is stored in the frames, whereas the only stored energy in the bending fixture shown in Fig. 2, is that stored in the supports at midspan and in the bolts. This is relatively small compared to that in the tensile fixtures. There are two other factors which could contribute to greater crack sensitivity in tension tests than in bending tests. First, both surfaces (and the edges) of the tension specimen are stressed in tension and hence are vulnerable to the initiation of stress corrosion cracking, whereas only the convex side of the beam specimen is stressed in tension and likely to initiate cracks. Second, in a direct-tension test, the specimen will fracture when the crack depth is sufficient to produce an average stress on the net section equal to the fracture stress of the material (under the existing stress state); however, in bending, only the extreme fibers in the vicinity of the crack tip are stressed to the breaking strength, so that failure will tend to occur progressively rather than suddenly. (This appears to be borne out in tension tests of notched bars; sharply notched tension specimens generally fail suddenly, whereas sharply notched beam specimens fail gradually by pulling the sections apart.) General Discussion The stress corrosion performance of a welded structure is influenced by (1) the compositions of the parent plate and the filler alloy; (2) the metallurgical conditions of the parent plate, the heat-affected zone, and the filler alloy; (3) the magnitude of the sustained tensile stress, which is determined by the algebraic sum of the stresses introduced on the exposed surface; and (4) the physical and chemical nature of the corrosive environment. In the evaluation of the stress corrosion performance of weldments, factors (1) and (2) can be markedly influenced by variations in welding procedure. The heat from welding produces metallurgical changes in the parent metal and reduces the strength in part of this heat-affected zone (HAZ). Examples of such effects are shown for both strain-hardened and heat treated alloys in Figs. 1 and 15. The yield strength across butt welds made in aluminum alloys, as determined at 0.2 per cent permanent set, depends upon the gage length used, the yield strength increasing with gage length. Structural designers recognize from past experience with riveted and bolted joints that a small amount of local yielding is characteristic and will not significantly affect the performance of a structure [7]. Thus, the yield strength value corresponding to 0.2 per cent offset on a 10-in. gage length is considered to by ASTM (all rights Wed Dec 16 structures 15:53:43 EST[8], 2015 beCopyright applicable to theInt'ldesign of reserved); welded aluminum because Downloaded/printed by a permanent deformation across the joint of the this offset represents University of Washington (University of Washington) pursuant to License Agreement. No further reprod SHUMAKER ET AL ON WELDED ALUMINUM ALLOYS 339 same magnitude commonly encountered in riveted or bolted joints at loads that cause an average net section stress equal to the yield strength of the plate. For weldments of most alloys, the proportional limit based on 10-in. gage length tests will be 50 to 70 per cent of the 10-in. gage length yield stress. At the Alcoa Research Laboratories, the practice of loading stress corrosion specimens to 75 per cent of the 10-in. gage length tensile yield stress has been adopted. This stress level will result in some inelastic deformation of the metal in the reduced strength zone. The justification for this is that locked-in assembly stresses developed in structural members welded under constraint may be as high as the yield strength of the weakest metal in the heat-affected zone. When the structure is subjected to external loads, the load stresses add to the locked-in stresses and generally result in some inelastic deformation of the metal in the reducedstrength zone. In addition to the assembly stresses developed in weldments made under constraint, it is possible for local residual stresses resulting from thermal mismatch and plastic deformation during welding to be developed in the immediate vicinity of the weld. These stresses may be developed parallel to the weld [9] (Fig. I5d) or transverse to the weld (Fig. 15b). The former stresses will be relieved when cross weld specimens are cut from the welded panel. However, the latter type stresses, which generally occur only in thick plate joined by multipass welds, will not be relieved when full-thickness cross weld specimens are cut from the panel. It is evident from the data shown in Fig. I5b that such stresses can be appreciable and variable and should be recognized when it is attempted to develop a specific stress in the test specimen. It is obvious that in attempting to make a quantitative stress corrosion test of a weldment, the mechanics of developing a known stress in the test specimen is complicated by the presence of the weld bead (or beads). Application of a load could be simplified in the case of light gage material by machining the weld beads flush with the plate or in the case of multipass welded plate by machining uniform thin (0.125-in.-thick) specimens from the weldment. Data in Table 3 for 0.125-in.-thick sheet tested "bead-off" versus "bead-on" both as beams and as tension specimens show that tests of bead-off specimens may fail to reveal an appreciable susceptibility to stress corrosion cracking. Although testing bead-off specimens may be advantageous for some purposes, these data raise a doubt as to whether or not such results would be representative of the majority of industrial weldments. Summary Several types of test specimens and methods of loading have been described, theInt'l investigator may choose a 16 test15:53:43 procedure that best Copyright byand ASTM (all rights reserved); Wed Dec EST 2015 Downloaded/printed suits his need. It by is important, however, that the complicating factors University of Washington (University of Washington) pursuant to License Agreement. No further repr 340 STRESS CORROSION TESTING associated with the testing of weldments be understood so that a procedure can be selected that will not give misleading results. Although more information is needed on this subject, the following observations are significant. Beam specimens with the weld bead(s) intact provide a versatile, reliable test for determining the serviceability of a weldment in varied environments. The test is reasonably quantitative provided that the specimens are not overstrained during loading. In testing beams of relatively thick plate containing multipass welds, consideration must be given to the presence of residual welding stress transverse to the weld. For laboratory studies that in some cases may require the most rapid and discriminating test, the direct-tension specimen is advantageous. Stressing frames of the constant-deformation and spring-loaded types are equally effective, and more practicable, especially the constant-deformation frame, than dead-weight loading equipment. Often there is a tendency for selective attack in the HAZ causing a reduction of the cross-sectional area, which in the case of the direct-tension test may result in failure because of tensile overload that could be confused with a stress corrosion cracking failure. To develop the maximum tensile stress that is tolerable in a welded structure, the specimen should be loaded to the point of developing a small amount of inelastic deformation in the HAZ. As a general rule, this can be accomplished by stressing the test specimen to 75 per cent of the 10-in. gage length yield strength of the weldment. The U-bend specimen is a qualitative, highly stressed specimen of limited value, because it is realistic only for certain applications that involve severe deformation of the welds. It can be useful for rapid screening of weldments that are sensitized by plastic deformation. Testing specimens with the weld beads machined off will permit a closer control of applied stress but may not provide as severe or as realistic a test as testing bead-on specimens. For testing butt welds in thick plate, the self-stressed H-plate specimen permits the combination of extreme assembly stress due to welding under constraint and local residual welding stress and is worthy of further evaluation. A cknowledgment A portion of the work was supported by the George C. Marshall Space Flight Center of the National Aeronautics and Space Administration (Contract NAS 8-5340) to whom the authors express their appreciation. The contract work was administered under the technical direction of the Propulsion and Vehicle Engineering Laboratory, Materials Div., George C. Marshall Space Flight Center, with J. G. Williamson acting Copyright by ASTMofficer's Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 as the contracting technical representative. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further r SHUMAKER ET AL ON WELDED ALUMINUM ALLOYS 341 References [7] Baysinger, F. R., "Investigation of Welding and Fabricability of Kaiser Experimental Alloy MR39A," Contract NAS 8-5065, Request PT2-82464, 21 May, 1964. [2] Dix, E. H., Jr., Anderson, W. A., and Shumaker, M. B., "Development of Wrought Aluminum-Magnesium Alloys," Technical Paper 14, Alcoa Research Laboratories, New Kensington, Pa., 1958. [3] Sprowls, D. O. et al, "Investigation of the Stress-Corrosion Cracking of High Strength Aluminum Alloys," Contract NAS 8-5340, summary report, 1 August, 1965, sponsored by the George C. Marshall Space Flight Center, Huntsville, Ala. [4] Sager, G. F., Brown, R. H., and Mears, R. B., "Tests for Determining Susceptibility to Stress-Corrosion Cracking," ASTM-AIME Symposium on StressCorrosion Cracking of Metals, 1944, American Society for Testing Materials, Philadelphia, 1945, p. 255. [5] Weiss, S., Ramsey, J. N., and Udin, H., "Evaluation of Weld-Cracking Tests on Armor Steel," Welding Journal, Vol. 35, No. 7, 1956, p. 348S. [6] Brenner, Paul and Gruhl, Wolfgang, "Stress-Corrosion Cracking Tests of AlZn-Mg 3 Under Constant Tensile and Bending Strain," Zeitschrift fuer Metallkunde, Vol. 52, No. 10, 1961, pp. 599-607. [7] Hill, H. N., Clark, J. W., and Brungraber, R. J., "Design of Welded Aluminum Structures," Transactions, American Society of Civil Engineers, Vol. 127, Part II, 1962, pp. 102-126. [5] Task Committee on Lightweight Alloys, "Suggested Specifications for Structures of Aluminum Alloys 6061-T6 and 6062-T6," Paper 3341, Proceedings, American Society of Civil Engineers, Journal of the Structural Division, Dec. 1962, p. 1; Alcoa Handbook of Design Stresses for Aluminum. [9] Hill, H. N., "Residual Welding Stress in Aluminum Alloys," Metal Progress, Aug. 1961. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduc B. W. Lifka1 andD. O. Sprowls1 Stress Corrosion Testing of 7079-T6 Aluminum Alloy in Various Environments REFERENCE: B. W. Lifka and D. O. Sprowls, "Stress Corrosion Testing of 7079-T6 Aluminum Alloy in Various Environments," Stress Corrosion Testing, ASTM STP 425, Am. Soc. Testing Mats., 1967, p. 342. ABSTRACT: Stress corrosion cracking of certain 7079-T6 aluminum alloy aircraft structures involving sustained tensile stress acting in an unfavorable grain direction has occurred sometimes in surprisingly mild environments. The resistance of short-transverse specimens of 7079-T6 products has been evaluated at different levels of applied stress in several different seacoast and inland industrial atmospheres. It was found that the atmosphere was more critical than the generally accepted alternate-immersion test using 3.5 per cent sodium chloride. The results of the outdoor tests are correlated with eight accelerated exposures; and consideration is given to the most suitable laboratory stress corrosion test environment for this alloy. A detailed analysis of the constant deformation methods of stressing is included. KEY WORDS: corrosion, stress corrosion, aluminum alloys, environmental testing, exposure testing, chlorides, salt solution, salt spray test, anisotropy Aluminum alloy 7079-T6 is attractive for aerospace and hydrospace applications because of its desirable mechanical properties in thick sections. In fact, 7079-T6 has become the alloy most widely used in the United States for large high-strength aluminum forgings. Excellent service over a number of years has been demonstrated in a wide variety of parts involving large tonnages of metal. This generally satisfactory service record is the result of the cooperation of materials engineers and designers. There is a strong incentive for such cooperation when it is realized that an understanding of the capabilities and the limitations of materials combined with necessary precautions in design and assembly [I]2 will permit the use of stronger alloys and tempers. 1 Engineer and assistant chief, respectively, Chemical Metallurgy Div., Alcoa Research Laboratories, New Kensington, Pa. 2 The italic numbers in brackets refer to the list of references appended to this Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 paper. Downloaded/printed by 342 University of Washington (University of Washington) pursuant to License Agreement. No further reproductio LIFKA AND SPROWLS ON ALUMINUM ALLOY 343 Service failures resulting from stress corrosion cracking occasionally have occurred in parts of 7079-T6 and other high-strength aluminum alloys. Invariably these failures have resulted from residual or assembly surface tensile stress acting continuously in the short-transverse direction (transverse in the case of round and square sections) relative to the grain structure. Longitudinal tensile stresses, on the other hand, have rarely caused stress corrosion problems. This record is consistent with the results of laboratory tests [2], which have shown that all of the commercial high-strength aluminum alloys are highly resistant to stress TABLE 1—Resistance to stress-corrosion cracking of 2-in.-thick 7079-T651 plate. Number of triplicate specimens failing and days-to-failure.a Test Environment Test Period, days Industrial atmosphere (New Kensington, Pa.). 1460 Seacoast atmosphere (Point Judith, 1460 R-i.) 3.5% NaCl alternate immersion 84 Acidified 5% NaCl intermittent 14 spray Acidified 5% NaCl intermittent 30 spray 4.55% NaCl + 1.16% CrO3 alternate immersion First run 30 Second run 30 0 Short-Transverse Stress, 10 ksi No. Days Long-Transverse Stress, 53 ksi No. Longitudinal Stress, 56 ksi No. Days 3 448, 644, 1202 0 0 3 126, 126, 716 3 716, 716, 826 0 0 0 0 3 6, 7, 10 0 0 ? 10, 17 3 10, 15, 17 2 19, 24 0 3 3 7, q, Q 7, 7, 8 2 2 Days 23, 28 17, 24 Type of specimen: % -in. -diameter tensile bar. corrosion cracking when stressed in the longitudinal or the long-transverse directions; but that many alloys, including 7079-T6, have a relatively low resistance when stressed in the short-transverse or the transverse (round or square sections) directions relative to the grain structure. A sample of data illustrating this directional behavior for 7079-T6 plate is given in Table 1. Purpose of Stress Corrosion Testing One of the limitations associated with all high-strength metals is stress corrosion cracking. "While it is relatively easy to determine if a product is 'susceptible to stress-corrosion cracking,' it is far more difficult to determine possesses a 'degree hamper Wed its Copyright if it by ASTM Int'lof susceptibility' (all rights that will reserved); Downloaded/printed University of by Washington (University of Washington) pursuant Dec to 344 STRESS CORROSION TESTING {**; (a) As heat treated 7079-T6 landing gear. Bored hole extends to point C, section AB shown below. (b) Section AB showing cracks on ID. (c) Shows depth of crack in parting plane, Flick's etch. (d) Tip of stress corrosion crack, Keller's etch (X100). (e) Shows interfragmentary nature of crack, Keller's etch (X500). FIG. 1—Example of a service failure. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further re LIFKA AND SPROWLS ON ALUMINUM ALLOY 345 general usefulness" [3]. The mechanism of stress corrosion of aluminum alloys and their resistance to stress corrosion cracking have been under continuous investigation at the Alcoa Research Laboratories for over 35 years. The objectives of this research have been to: (1) determine the factors which must be controlled in order that aluminum alloy products of superior strength may be successfully employed and (2) develop stress corrosion resistant compositions and tempers. A number of publications have been issued, some of which are identified in Refs 2 through 8. Use of Accelerated Tests It is readily recognizable that the most reliable data are those obtained in environments representative of the conditions the final assembly will encounter in use. For aerospace applications, typical environments are industrial or marine atmospheres; while total, partial, and intermittent immersion in fresh and sea waters are representative of hydrospace applications. Unfortunately, the prolonged exposure periods required for tests in these environments to be conclusive are often prohibitive and accelerated laboratory tests must be used. It is essential, however, that an accelerated test reproduce the same mode of failure as occurs in service environments. It should rate the percentage failures and the times-tofailure of various thermal treatments and anisotropic behaviors in the same order of merit as do natural environments. The accelerated stress corrosion test most widely accepted by the aluminum industry is the 3.5 per cent sodium chloride (NaCl) alternateimmersion test (1-hr cycle: 10 min immersion plus 50 min drying in air). Extensive evaluation over many years at Alcoa Research Laboratories [2,7] has established a good correlation between this procedure and extended seacoast and industrial atmospheric exposures for alloys and tempers of the aluminum-copper and aluminum-magnesium systems and for most aluminum-zinc-magnesium-copper alloys such as 7075 and 7178. Susceptible specimens of these alloys are just as likely, and frequently more prone, to stress corrosion crack in 3.5 per cent NaCl alternate immersion than in seacoast or inland industrial atmospheres. This information has enabled the industry to reliably test these alloy types in the alternate-immersion test without the continual necessity of back-up tests in natural environments. Deviation of 7079-T6 Alloy These same extensive tests have shown that alloy 7079-T6 deviates from the general pattern in that at low stress levels transverse and shorttransverse specimens are less likely to stress corrosion crack in the alternate-immersion test than in atmospheric tests. An example of this is illustrated by the case history of the 7079-T6 ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 dieCopyright forging by shown in Fig. 1. Solution heat treatment and quenching of Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No furthe 346 STRESS CORROSION TESTING FIG. 2a—Specimen used to simulate various levels of residual quench tensile stress at the parting plane of the forging shown in Fig. 1. FIG. 2b—Resultant data for specimen shown in Fig. 2a. the rough machined forging unavoidably resulted in circumferential tensile stress on the finish-machined inside wall of the cylinder. Cracking was revealed in the inside wall of a number of the parts by a penetrant inspection. Microscopic examination of the failed parts (Fig. 1) established that the cracks occurred in the parting plane of the forging where the residual stress a short-transverse grain structure. Copyright by tensile ASTM Int'l (allacted rights across reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further r LIFKA AND SPROWLS ON ALUMINUM ALLOY 347 The cracks were intergranular (and interfragmentary) in nature, and appeared as typical stress corrosion cracks. Measurements by layerremoval techniques on four parts that did not fail showed residual tensile stresses in the circumferential direction of about 10 ksi at most locations, with a maximum of 18 ksi. However, it is possible that higher stresses than these might have existed in»the parts that actually failed. A laboratory test was undertaken to duplicate the service failures using the C-ring specimen shown in Fig. 2a. Rings were machined from the forging, so that the parting plane was at the middle of the C. The wall FIG. 3—Comparison of the probability of survival of short-transverse specimens of 7079-T6 alloy with alloys 7075-T6 and 7039-T63 in two environments (21 to 30 specimens representing 7 to 10 lots of rolled plate as a basis for each point). thickness of the C-rings was machined to l/s in. by removing only the outer surface so that the test surface was the same as in the original part. The C was then spread to impose tensile stress on the inner surface, the level of stress being determined from electrical strain gage measurements. Exposures were made in 3.5 per cent NaCl alternate immersion and hi seacoast and industrial atmospheres. The resultant data are presented graphically in Fig. 2b. In the alternate-immersion test, failures occurred at 40 and 30 ksi but not at the 20 and 10 ksi stress levels. On the other hand, failures did occur at the 20 and 10 ksi levels with prolonged exposures to the atmosphere, indicating a need for an accelerated test correlating better Copyright by ASTMenvironments. Int'l (all rights reserved); WedofDec 15:53:43 ESTtests 2015conwith atmospheric The results the16atmospheric Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further r 348 STRESS CORROSION TESTING firm the residual stress determinations of 10 to 20 ksi, although the timesto-failure were relatively long (11A to 3 years). Investigation of the stress corrosion characteristics of 7079-T6 alloy and of various experimental aluminum-zinc-magnesium alloys has shown that this deviation of 7079-T6 alloy is related to its difference in composition from aluminum-zinc-magnesium-copper alloys such as 7075-T6, the most significant difference being the decrease in copper content. Data plotted in Fig. 3 for the 3.5 per qent NaCl alternate-immersion test show FIG. 4a—Type of die forging used for comparison of the effect of environment. FIG. 4b—Macroetched cross section shows parting plane structure and specimen location. that at relatively low stresses, such as 15 or 25 per cent yield strength, the percentage survival of 7079-T6 and 7039-T63 was far higher than that of 7075-T6. However, in the case of the atmospheric exposure data shown in Fig. 3, the percentage survival of 7079-T6 and 7039-T63 at stresses of 15, 25, or 50 per cent yield strength has reversed and is now far lower than that of 7075-T6. Investgation of Stress Corrosion Cracking of 7079-T6 in Various Environments Investigation was made to study the susceptibility of 7079-T6 to stress corrosion cracking in various types of atmospheric environments and to Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further LIFKA AND SPROWLS ON ALUMINUM ALLOY 349 FIG. 5—Shows the Vs-in.-diameter tension specimen, the various parts of the stressing frame, and the final stressed assembly. search for a more definitive laboratory accelerated test for 7079-T6 than the 3.5 per cent NaCl alternate-immersion test. Procedure Test Materials • The major portion of the investigation was conducted on three 7079T6 die forgings of the type shown in Fig. 4a. The principal reason for using this forging was that its uniform cross section permitted a large number of replicate specimens to be taken across the short-transverse parting plane grain structure as shown in the macroetched section in Fig. 4b. Tests were also made on transverse specimens from IVa-in.-diameter 7079-T651 rolled rod and in all three directions of 2-in.-thick 7079-T651 Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No furthe 350 STRESS CORROSION TESTING FIG. 6—ARL's synchronous loading device used to stress specimens. A stressed assembly and one assembled fingertight for stressing are shown to the left. plate. All items had composition within the specified limits and had been produced using commercial fabricating and heat-treating facilities. Aluminum Association composition limits for alloy 7079 are: Alloy Si Fe Cu Mn Mg Zn Cr Ti 7079 0.3 max 0.4 max 0.4 to 0.8 0.1 to 0.3 2.9 to 3.7 3.8 to 4.8 0.1 to 0.25 0.1 max Test Specimen and Method of Loading The data to be discussed hi this paper were obtained with Vs-in.diameter by 2-in.-long tension specimens, stressed hi the wedge-type frame shown in Fig. 5. Components of the frame are assembled fingertight, and the specimen is then stressed with ARL's synchronous loading Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 device shown in Fig. 6. The inward movement of the wedge-shaped side Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprod LIFKA AND SPROWLS ON ALUMINUM ALLOY 351 pieces induces uniaxial tensile stress in the specimen. Applied stresses usually are kept below the proportional limit, or about 75 per cent of the 0.2 per cent yield strength. Thus, only the straight line portion of the stress-strain curve for the test specimen is needed, since the applied stress is considered to be related directly to the measured strain by Hooke's law and Young's modulus. During the stressing operation, the strain is FIG. 7—Effect of loading method and extent of cracking on average net section stress, local cracking. measured with a Type F Huggenberger tensometer with a 0.5-in. gage length and a magnification of about 2000. The error of stress application at the point of measurement is probably less than about 2 per cent, since the precision of the strain measurement is within 0.5 per cent, and the error in determining Young's modulus is within 1 per cent (Method for Determination of Young's Modulus at Room Temperature, E 111-61). Because there is a certain amount of inherent eccentricity in mating threads, it is impossible to completely avoid some slight bending of the specimen. The likelihood of developing unequal stress around the circumference of the specimen is reduced howCopyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 ever by the use of spherically seated nuts threaded onto the grip ends Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduction 352 STRESS CORROSION TESTING of the specimens and the synchronous loading device that ensures equal and simultaneous movements of the side pieces of the stressing frame. Tests have shown that the uniformity in stress around the circumference is of the same order as is achieved in tension tests employing this same type of specimen. FIG. 8—Effect of loading method and extent of cracking on average net section stress, general cracking, or corrosion. A comparison of this method of loading with other techniques, such as loading with dead weight (constant load) or with an infinitely stiff fixture (constant deformation), is given in Figs. 7 and 8. The change in the average tensile stress on the net section caused by the initiation and propagation of a stress corrosion crack or by general cracking has been calculated and checked by a compliance experiment (Appendix I). With the initiation of localized cracking, the average tensile stress on the net section increases rapidly, as in the case of dead-weight loading, until the Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprod LIFKA AND SPROWLS ON ALUMINUM ALLOY 353 fracture strength is reached and the specimen breaks. The average stress does not increase quite as rapidly as with dead-weight loading, so the specimen life can be expected to be longer for specimens loaded in this frame than for dead-weight loaded specimens. In the case of general cracking or severe general corrosion, however, there is a considerable FIG. 9—Effect of environment on the resistance to stress corrosion cracking of 7079-T6 die forgings stressed in the short-transverse direction. difference in the effects of the loading method on the change in stress: whereas with dead-weight loading, a reduction in area of about 40 per cent will result in tensile failure, a specimen loaded in the stressing frame can be corroded completely away without actually breaking. Thus, this method of loading is similar in effect to dead-weight loading when localized cracks are initiated, and it has the advantage over dead-weight loading when generalized corrosion occurs in that fractures from tensile overload are not so likely to occur. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further rep TABLE 2—Individual specimen lives, in days, of the data shown graphically in Fig. 9. Triplicate, short-transverse, %-in. -diameter'by2'-in. -long tensile bars from 7079-T6 alloy die forging. Forging No. 1 No. 2 No. 3 atSOF New Kensington, Pa. at 120 F No. 3 Point Judith, R.I. lOksi 20 ksi 30 ksi 10 ksi 20 ksi 30 ksi 10 ksi 20 ksi 30 ksi 10 ksi 20 ksi 30 ksi 10 ksi 20 ksi 30 ksi 10 ksi 20 ksi 30 ksi 730° 730" 730° 187 200 730° 180 200 302 730° 730° 730" 59 391 535 81 300 339 173 200 730° 730° 730° 730° 303 303 730° 3 39 66 20 44 50 44 135 140 94 94 99 559 633 643 1460 1460 1460° 123 123 123 198 201 205 262 293 303 23 24 25 20 77 80 38 150 155 20 23 38 598 598 836 1460 1460° 1460° 221 221 305 221 221 221 221 221 221 221 221 221 221 221 221 221 221 221 221 221 589 920 1322 232 1460° 1460° 232 232 232 232 180 730° 730° 140 140 151 249 296 421 94 94 94 77 730° 730° 730° 296 339 339 1141 1460° 1460° 140 151 339 232 232 145 232 232 145 145 232 67 145 145 145 145 232 145 145 145 Forging No. 2 Point Comfort, Tex. Cleveland, Ohio Alternate Immersion — 12 Weeks No. 1 Seacoast Atmosphere—4 Years Industrial Atmosphere—4 Years 100% Relative Humidity 3.5% NaCl + 0.1% NHiHCCh 3.5% NaCl 10 ksi 20 ksi 30 ksi 10 ksi 20 ksi 84° 84' 29 29 42 35 48 64 4 4 6 11 11 84° 84° 15 21 84° 84° 20 28 80 82 28 32 84° 84" 84= 84" 11 84° 84" 13 6 8 10 4 4 11 30 ksi Total Immersion Synthetic Sea Water*" 10 ksi 20 ksi 84° 84° 60 84° 84° 35 42 84° 10 18 35 84° 84° 84° 84° 84° 84° 84° 3.5% CrOs + 0.3% NaCl + 2.9% KzCnOr— 12 Weeks 30 ksi 10 ksi 20 ksi 13 15 18 21 61 84° 61 84° 84° 84° 84° 84° 22 26 26 84° 84° 7 84° 84° 11 Boiling 6% NaCl 4 Days 30 ksi 10 ksi 20 ksi 30 ksi 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 3 7 2 2 2 4 21 28 4° 4° 4° 4° 4° 4° 4c Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 a Specimens still in test scheduled to complete 4 years. Downloaded/printed by 6 Composition of solution as per ASTM Specifications D 1141 without heavy metals. University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized. c Specimen did not fail. Acidified 5% NaCl Intermittent Spray at 120 F— 2 Weeks 10 ksi 7 11 11 7 14° 14° 14° 14° 14° 20 ksi 1 1 3 4 5 8 7 7 14° 30 ksi 1 1 3 2 3 3 3 3 4 LIFKA AND SPROWLS ON ALUMINUM ALLOY 355 Results and Discussion Survey of Various Environments The three die forgings were first tested in four different outdoor atmospheric environments: two inland industrial sites, New Kensington, Pa., and Cleveland, Ohio; a seacoast location in the northeastern United States, Point Judith, R.I.; and a seacoast location in a warm climate, Point Comfort, Tex. A summary of the test results (Fig. 9)3 shows similar performance in all four locations, with a markedly higher percentage of failures at the lowest stress level (10 ksi) than in the 3.5 per cent sodium chloride alternate-immersion test. It is noteworthy that failure times (Table 2) of specimens exposed to the seacoast atmosphere at Point Comfort, Tex., were not appreciably different from the lives of specimens exposed at Point Judith, R.I., where the average temperature is about 20 F lower. Exposures to atmosphere saturated with water vapor (100 per cent relative humidity) at 80 or 125 F caused failures of all specimens stressed to 30 ksi as did the exposures to the natural atmospheres, but in water vapor there were fewer failures at 20 ksi and no failures in over two years of testing at 10 ksi.4 In these tests in saturated atmospheres, a slightly greater number of failures was encountered at the higher temperature. Five laboratory test media were investigated in a search for an improvement over the 3.5 per cent sodium chloride alternate-immersion test. Synthetic sea water (Specifications for Substitute Ocean Water, D 114152) was even less discriminating than 3.5 per cent sodium chloride solution. The addition of 0.1 per cent ammonium bicarbonate (based on work of Farmery and Evans [9]) to the 3.5 per cent sodium chloride (Fig. 9) gave only a slight improvement. Continuous immersion in an acidified-salt-dichromate solution suggested by Sager et al [5] was not so effective as the 3.5 per cent sodium chloride alternate immersion. Continuous immersion in boiling 6 per cent sodium chloride solution [5] for four days was no more effective in producing failures at the low stress levels than the alternate-immersion test but had the advantage of being of shorter duration. A recently developed acidified 5 per cent sodium chloride intermittent spray at 120 F [10] (Fig. 9) appears to be the most promising, although this test still did not produce quite so high a percentage of failures at the 10 ksi stress levels as did the atmospheric exposures. Modifications o] the 3.5 per cent Nad Alternate Immersion Tests utilizing the 2-in. plate were made to study the effect of purity 3 Tabular data on which Fig. 9 is based are given in Table 2. * Tests in 100 per cent relative humidity are still in progress, scheduled to comCopyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 plete four years of exposure. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further repr 356 STRESS CORROSION TESTING of the salt and water and to evaluate alternate-immersion tests using various salt plus chromate solutions. These showed that the tendency to cause cracking increased slightly with increasing purity of both the salt and water. They also showed that a combination of sodium chloride and chromic acid would cause more and quicker failures than would salt by itself. However, tests of specimens taken in all three grain directions revealed a characteristic of the chromate-containing solutions that seriously detracts from their usefulness. It is not enough for an accelerated test to cause rapid failure of susceptible specimens. To be most useful it should correlate well with natural environments and service experience in all respects and should not be so severe as to cause failures that would not occur in normal use. FIG. 10—Relative resistance to stress corrosion cracking of three items of 2Vzin.-diameter aluminum alloy rolled rod. The stress levels employed corresponded to 75, 50, and 25 per cent of the respective transverse yield strengths. Alloy 7079-T6 shows appreciable susceptibility to stress corrosion cracking only to stresses acting in the short-transverse direction. Like most high-strength alloys, it is relatively resistant to stresses acting in the long-transverse and longitudinal directions. In fact, only a few isolated instances have ever been recorded of service failures due to stress in the latter directions. Tests on the 2-in. plate showed that when a chromate-bearing solution was aggressive enough to cause 7079-T6 short-transverse specimens to fail at low levels of stress, it also causes fractures in specimens from the long-transverse and longitudinal directions. An example of this is given in Table 1 for a 4.5 per cent NaCl +1.2 per cent CrO3 solution, which was one of the solutions most able to cause fractures in shorttransverse specimens at 10 ksi. Data obtained in the acidified 5 per cent sodium chloride intermittent spray test are also included to show that ASTM Int'l to (all be rights 16 specimens 15:53:43 ESTfrom 2015 the it Copyright was notbyconsidered tooreserved); severe. Wed All Dec three Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further rep LIFKA AND SPROWLS ON ALUMINUM ALLOY 357 short-transverse direction failed at a low stress (10 ksi—15 per cent yield strength) in all environments, excepting the standard alternate-immersion test. For highly stressed (53 ksi—75 per cent yield strength) long-transverse specimens, the industrial atmosphere (four years), the 3.5 per cent NaCl alternate immersion (twelve weeks), and the acidified intermittent spray (extending over two weeks) did not cause failure. Failures did occur, however, in the northern seacoast atmosphere and in the intermittent spray but only after relatively long exposures. On the other hand, the salt-chromic-acid solution fractured the long-transverse specimens in shorter time than the short-transverse specimens. Finally, none of the environments failed the longitudinal specimens except the salt-chromicacid solution. Consequently, use of acidic chromate solutions is not recommended because of their unrealistic aggressiveness. Comparison of Alloys The data obtained on the rolled rod [11] are shown in Fig. 10 and provide another comparison of alloy 7079 with aluminum-zinc-magnesium alloys of higher and lower copper content. Again the alloy with the higher copper content, 7178 (1.6 to 2.4 per cent copper), shows good agreement between results obtained in the alternate-immersion test and in the atmospheres; while the newer, copper-free weldable alloy, X70065 (0.1 per cent copper max), shows a disparity between results obtained in alternate immersion and the atmospheres, following a pattern similar to that of 7079. As was the case with the tests of forged material, the acidified spray test shows promise of being able to rapidly produce results similar to those obtained in atmospheres. General Considerations for a Laboratory Test Medium The search for a more satisfactory accelerated test has not as yet produced a method that is completely satisfactory. Thus far, the accelerated test showing the best correlation for 7079-T6 with atmospheric exposure, without being too severe, has been the acidified 5 per cent sodium chloride intermittent spray test. However, there are other factors that should be considered in the selection of a new test method. Applicability to Other Alloys It obviously would be advantageous to be able to test several alloys and alloy systems in the same test medium. Not only would this reduce the need for a multiplicity of test methods, but, more significantly, it permits data obtained on one alloy to be placed in their proper perspective to other alloys. Tests on nine alloys of aluminum-zinc-magnesium 5 X7006 has been replaced by a slight modification X7106, but the stress corroCopyright by ASTM Int'lalloys (all rights reserved); Wed Dec 16 15:53:43 EST 2015 sion resistance of the two is similar. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No furth 358 STRESS CORROSION TESTING (7000 series), aluminum-magnesium (5000 series), and aluminum-copper (2000 series) systems showed the acidified intermittent spray gave good correlation with atmospheric exposure for all alloys with the exception of copper-free aluminum-zinc-magnesium alloys, such as 7039-T63, at stresses below 25 per cent yield strength. This is unfortunate, because this is the only other major alloy group for which the standard alternate-immersion test is not effective. Note that the acidified intermittent salt spray test has a disadvantage in that it is relatively corrosive to most aluminum alloys. The general surface pitting may impair visual detection of fine cracks in specimens. Also, it can result in failure in prolonged tests by mechanical rupture due to reduction of the cross section. Because of this, a two-week test period is recommended as the maximum period suitable for all alloys and sizes of specimens. However, longer periods of three or four weeks may be employed for alloys with high resistance to general corrosion or for specimens for which the load-carrying area is large. The boiling 6 per cent NaCl solution test is the most effective accelerated test thus far developed for the copper-free aluminum-zinc-magnesium alloys. While it is not so effective on 7079-T6 at very low stresses as is the intermittent spray test, it is somewhat better than the standard alternate-immersion test. Unfortunately, the major drawback to this method is that it is not effective on aluminum-zinc-magnesium alloys with a copper content of 1 per cent or more, nor is it effective on the aluminum-copper and aluminum-magnesium alloy systems. The amount of general corrosion that occurs in the boiling 6 per cent NaCl solution test is very slight. This simplifies detection of cracks and makes the possibility of mechanical rupture extremely remote. Some relaxation of applied stress due to creep has been noted, but, in general, the amount has not been so large as to invalidate the test method. Because of this relaxation, some investigators favor use of total immersion in 6 per cent NaCl at 70 F rather than at 212 F. Obviously, additional effort is needed to develop an accelerated test that is completely satisfactory for copper-free aluminum-zinc-magnesium alloys and also suitable for other alloy systems. Method of Exposure The procedures for both the acidified intermittent spray and the boiling 6 per cent NaCl solution tests are outlined in Appendix II. The procedures for continuous spray tests are definitely established and well known. Use of an acidified solution and intermittent operation do not complicate these in any way. In all such spray tests, however, corrosion is not uniform on all surfaces of a specimen; corrosion is maximum on surfaces facing upward on which the dispersed mist can settle out. This Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further LIFKA AND SPROWLS ON ALUMINUM ALLOY 359 does not present any real problem for specimens where all surfaces are equally susceptible. It could be of major significance on complex shapes or assemblies if a particularly susceptible area is shielded from the mist. The simplicity of the boiling 6 per cent NaCl solution test is one of its principal advantages and results in a high degree of reproducibility. All that needs to be done is to suspend the specimens in such a manner that they do not touch the vessel or one another. Availability and Cost of Equipment The intermittent spray test is operated in equipment designed to comply with ASTM specification for continuous spray tests. Such equipment is commercially available in a wide range of sizes and can be readily and inexpensively modified for automatic cyclic operation. The boiling 6 per cent NaCl solution test requires nothing special and can be run with ordinary laboratory equipment. Other than a heat supply and any inert vessel, all that is required is a sealed lid equipped with a condenser to maintain the solution at a constant concentration. This is fortunate because it permits both tests to be run for maximum correlation with very little additional expenditure. Summary It has been established that alloy 7079-T6 rarely fails by stress corrosion cracking in 3.5 per cent NaCl alternate immersion at stresses below 20 ksi, whereas the alloy has failed by this mechanism in natural atmospheres at stresses as low as 7 ksi [2]. Various accelerated test media have been investigated in search of a laboratory test more closely paralleling results obtained in service-type environments represented by seacoast and industrial atmospheres. The best correlation for 7079-T6 was achieved with an acidified 5 per cent NaCl intermittent spray test. This test has the advantage of being short (two weeks) and applicable to most other aluminum alloys. The exception is that this method represents only marginal improvement over 3.5 per cent NaCl alternate immersion for the copper-free aluminum-zinc-magnesium alloys, such as 7039-T63. It has the disadvantage of causing severe generalized corrosion that could result in "corrosion failures" due to excessive attack if the period of exposure is too long. Because of its simplicity and low cost, the boiling 6 per cent NaCl solution test is recommended as a back-up test on 7079-T6 and to provide better correlation with copper-free aluminum-zinc-magnesium alloys. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further rep 360 STRESS CORROSION TESTING APPENDIX I Procedure Used for Determination of Fixture Stiffness and Effect of Cracking on Specimen Deformation The change in average net area tensile stress due to cracking (Figs. 7 and 8) was determined as follows: The relationship between load and fixture stiffness was experimentally determined as indicated in Figs. \\a and b. Load deflection data were then obtained for specimens slotted at their midlength to give reductions in area of 0, 20, 40, 60, and 80 per cent (Figs, lie and d). The stress corresponding to a given crack depth was found by trial and error. Let P° be the force required to stress a plain specimen to the initial stress <r° and P' the force on the specimen for a given crack depth. Then to satisfy equilibrium requirements Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 FIG. 11—Procedure Downloaded/printed by used for determination of fixture stiffness and effect of cracking on of specimen deformation. University Washington (University of Washington) pursuant to License Agreement. No further repro LIFKA AND SPROWLS ON ALUMINUM ALLOY 361 Let AL° be the initial change in specimen length produced by P° and AZ/ the change in length corresponding to P'. The change in specimen length, d(AL)s, can be found from the specimen load-deflection data (Fig. lid). To satisfy compatibility, the incremental change in specimen length must equal the incremental deformation of the fixture or where From the value of P' which satisfies the above requirements, the average net area stress is where A' is the net area stress for the given crack depth. APPENDIX II Test Procedure for Acidified Intermittent Spray and Boiling 6 per cent NaCl Solution Tests Acidified Intermittent Spray Test [10] A 5 per cent NaCl solution is made with salt of 99.7 per cent purity and distilled water and acidified with acetic acid to pH 3. Specimens are exposed in cabinets designed to meet the requirements of ASTM Method of Acetic AcidSalt Spray (Fog) Testing (B 287 - 62). The recommended length of exposure is two weeks, and specimens are inspected daily for failure. Test conditions are the same as those required by ASTM except for the following variations: 1. Operating temperature may be increased from 95 to 120 F. 2. Specimens are intermittently sprayed in 6-hr repetitive cycles, consisting of M-hr spray (ASTM Method B 287), 2 hr of dry-air purge, and S^-hr soak at high relative humidity. 3. Due to the small percentage of spray time, periodic measurements of pH and specific gravity are made only on the reservoir solution and not on condensate specimens. Boiling 6 per cent NaCl Solution Test This test employs a 6 per cent NaCl solution made up from salt of 99.7 per cent purity and distilled water. The solution is brought to a rapid boil, and the specimens are then freely suspended in the boiling solution. The usual exposure period is 96 hr (4 days) which is occasionally extended to 7 days. The majority of failures occur during the first few hours of test, consequently, specimens are inspected for failure quite frequently during the initial portion of the test. A typical inspection schedule is: Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 (a) Every 15 min for the first 2 hr. Downloaded/printed by (b) Every 30 min for the next 2 hr. University of Washington (University of Washington) pursuant to License Agreement. No further re 362 STRESS CORROSION TESTING (c) Every hour for the next 4 hr. (d) Three times a day thereafter References [1] "Metallic Materials and Elements For Flight Vehicle Structures," MIL-HDBK5, Section 3.1.1.3, Precautionary Notes on Aluminum (first appended in Aug. 1962). [2] Sprowls, D. O. and Brown, R. H., "What Every Engineer Should Know About Stress Corrosion of Aluminum," Metal Progress, Vol. 81, No. 4, 1962, pp. 79-85; Vol. 81, No. 5, 1962, pp. 77-83. [3] Dix, E. H., Jr., "Prevention of Stress-Corrosion Cracking in Service," Metal Progress, Vol. 56, Dec. 1949, pp. 803-806. [4] Mears, R. B., Brown, R. H., and Dix, E. H., Jr., "A Generalized Theory of the Stress-Corrosion of Alloys," ASTM-AIME Symposium on Stress-Corrosion Cracking of Metals, 1944, American Society for Testing and Materials, Philadelphia, 1945, pp. 323-339. [5] Sager, G. F., Brown, R. H., and Mears, R. B., 'Tests for Determining Susceptibility to Stress-Corrosion Cracking," ASTM-AIME Symposium on Stress-Corrosion Cracking of Metals, 1944, American Society for Testing and Materials, Philadelphia, 1945, pp. 255-272. [6] Nock, J. A., Jr., 'Today's Aluminum Aircraft Alloys," Transactions, Society of Automotive Engineers, Vol. 61, 1953, pp. 209-220. [7] Dix, E. H., Jr., Anderson, W. A., and Shumaker, M. B., "Influence of Service Temperature on the Resistance of Wrought Al-Mg Alloys to Corrosion," Corrosion, Vol. 15, No. 2, 1959, pp. 55-62. [8] Sprowls, D. O. and Rutemiller, H. C., "Susceptibility of Aluminum Alloys to Stress Corrosion," Materials Protection, Vol. 2, No. 6, June 1963, pp. 62-65. [9] Farmery, H. K. and Evans, U. R., 'The Stress Corrosion of Certain Aluminum Alloys," Journal of the Institute of Metals, Vol. 84, 1955-56, pp. 413422. [10] Lifka, B. W. and Sprowls, D. O., "An Improved Exfoliation Test for Aluminum Alloys," Corrosion, Vol. 22, No. 2, 1966, pp. 7-15. [11] Lifka, B. W. et al, "Investigation of the Stress-Corrosion Cracking of High Strength Aluminum Alloys," Contract NAS 8-5340, two-year summary report, 1965. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further H. B. Romans1 and H. L. Craig, Jr.1 Environmental Factors Affecting the Stress Corrosion Cracking Behavior of an Aluminum-Zinc-Magnesium Alloy REFERENCE: H. B. Romans and H. L. Craig, Jr., "Environmental Factors Affecting the Stress Corrosion Cracking Behavior of an Aluminum-Zinc-Magnesium Alloy," Stress Corrosion Testing, ASTM STP 425, Am. Soc. Testing Mats., 1967, p. 363. ABSTRACT: The stress corrosion cracking phenomenon in an aluminumzinc-magnesium alloy is affected by the environmental factors in a different fashion than the way these factors affect other corrosion processes. The presence of traces of moisture is sufficient for stress corrosion cracking to proceed; the absence of moisture prevents cracking. The time required to initiate stress corrosion cracking is affected mostly by the temperature of the environment. Wide variations in sodium chloride concentration have no effect on the time-to-failure. Changes in solution pH and contamination with dissolved copper alter the time required to produce failure, so that stress corrosion cracking is not a continuous function of either variable. Evidence is presented which shows that the manner of stressing a specimen and specimen configuration both affect failure time. These anomalies may be used to understand the lack of reproducibility of stress corrosion testing and the difficulties encountered in correlation studies, either among different tests or the same test carried out by different laboratories. KEY WORDS: corrosion, stress corrosion, environmental testing, cracking (fracturing), aluminum alloys, zinc alloys, magnesium alloys, sodium chloride, pH, copper ions, temperature, humidity, immersion tests (corrosion) Aluminum alloys subjected to stress corrosion tests are usually judged on the degree of stress corrosion susceptibility by the time required for a specimen to fail. In the past, tests have not been very reproducible and small differences in alloy behavior were usually unnoticed due to the larger scatter in data. There has been very little concern for this behavior since most people were interested with gross differences in alloy systems. However, isolated cases of service failures, combined with a large 1 Scientist and research supervisor, respectively, Department of Applied Chemistry and Mathematics, Reynolds Metals Co., Richmond, Va. Personal members Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 ASTM. Downloaded/printed by 363 University of Washington (University of Washington) pursuant to License Agreement. No further reproduction 364 STRESS CORROSION TESTING amount of publicity, have brought about an ever-increasing trend toward the writing of stress corrosion test specifications by some of the aircraft and military agencies of the government. One such specification designated as MIL-H-6088D for alloy 7075-T73 states: The 7075-T73 alloy shall be capable of passing the following tests without any stress corrosion cracking. Thirty-days' exposure by alternate immersion in 31A% NaCl solution (meeting the purity and pH requirements of method 811 of Federal Test Method Standard No. 151) at room temperature while stressed in any direction to 75% of the yield strength. The exposure cycle shall consist of 10 minute immersion in the solution and 50 minutes out of solution. A definite period of time is set as the criteria for an acceptable product. This means that reproducible tests are now a must in order for the specifications to be meaningful. Note that there is no specified test temperature nor is there any mention of humidity or purity of water used to make up the solution. In fact, most of the tests conducted today ignore these factors.2 To determine if these and other, environmental factors affected the time-to-failure, a series of tests was conducted on the effect of humidity, temperature, solution contaminants, variations in stress, and specimen configuration. Tests were also conducted between two different laboratories at controlled and uncontrolled temperature and humidity to determine the correlation of test results. Temperature has probably the greatest single effect on the time-to-failure, but it is seldom controlled. Contaminants from the air or heavy metal ions in solution probably have a large effect, although no numerical value has yet been placed on them. The effect of humidity is usually ignored, but actually this is the only factor needed to produce stress cracking in certain alloys. The purpose of this paper is to discuss the test environment and to show some of the effects of testing conditions on the stress corrosion cracking of a single alloy in the aluminum-zinc-magnesium system. Materials An aluminum-zinc-magnesium alloy with the following nominal composition (in per cent) was used for the experiments: Zn 4 Mg Fe Si Mn Cr Ti 2.8 0.4 0.3 0.3 0.2 0.1 Cu Others Each 0.1 max 0.05 max Total 0.15 max All specimens were from plate 1 to 3 in. thick. 2 Copyright by ASTM Int'l (all rights Dec 16 15:53:43 Stress Corrosion Questionnaire ASTMreserved); CommitteeWed G-l/VI Task Group 2. EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. ROMANS AND CRAIG ON ALUMINUM-ZINC-MAGNESIUM ALLOY 365 Two tempers for experimental purposes were prepared with several lots representing each temper. One temper, designated as Temper R, was prepared to provide good resistance to stress corrosion cracking. The other, designated as Temper S, was prepared to provide low resistance to cracking. The typical longitudinal mechanical properties were: tensile strength, 70 ksi; yield strength, 60 ksi; and per cent elongation in 2 in., 10 per cent. Procedure and Results The method of specimen preparation prior to testing, the test solutions, test equipment, and other details which apply to each test are given in the procedural details in the Appendix. Distilled Water Tests C-ring specimens of a sample in Temper S were stressed to 35 ksi for testing in total immersion in distilled water at 100 C. The following steps were taken to ensure that no contamination would be present in the test chamber: 1. The bolts used for stressing the specimens were fabricated from the parent sample to ensure the absence of dissimilar metal contacts and to eliminate the need to use foreign material, such as the wax usually used for insulation. 2. The specimens and bolts were cleaned as specified in the procedural details section. 3. An all-glass reaction flask was cleaned in sulfuric acid cleaning solution, rinsed in distilled water, dipped in nitric acid, and again rinsed. 4. After cleaning, everything in the test was handled with rubber gloves to prevent contamination from the hands. The specimens all failed from stress corrosion cracking within 15 hr (overnight). These failures occurred in the absence of any contamination other than that from the glassware and dissolved metal ions from the specimens. The distilled water was 8 X 105 ohm-cm prior to the test, and at the end of the test it was 2 X 105 ohm-cm. This shows a relatively high pickup of ions of which the major portion is assumed to be aluminum. Humidity To determine the effect of humidity, tests were conducted on C-ring type specimens from a sample in the S temper stressed at 35 ksi. Three to five specimens were exposed to each of the following environments: 1. A dynamic vacuum at an average of 30 C and 0 per cent relative humidity. 2. A desiccated atmosphere at an average of 30 C and 0 per cent relaCopyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 tive humidity. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further re 366 STRESS CORROSION TESTING 3. An air-conditioned laboratory atmosphere at an average of 23 C and 50 per cent relative humidity. 4. A normal laboratory atmosphere at an average of 30 C and 85 per cent relative humidity. 5. A controlled humidity test chamber at 38 C and 95 per cent relative humidity. Humidity alone may play a major role in the time required to produce stress corrosion cracking as proved by the following results. Note that in Table 1 failures occurred in all environments except the dynamic vacuum. The fact that failures occurred in the desiccated atmosphere may be explained by the ability of the oxide film to adsorb and hold moisture TABLE 1 —Effect of humidity on time-to-failure. Approximate Relative Humidity, % Average Temperature, degC 0 30 dynamic vacuum 0 30 desiccated atmosphere 50 23 air-conditioned atmosphere 85 30 laboratory atmosphere 95 38 controlled humidity test chamber Environment Days-to-Failure 5 specimens did not fail in 50 days 3 specimens failed within 12 days 3 specimens failed within 4 days 3 specimens failed within 3 days 4 specimens failed within 1 day in preference to the desiccant. This moisture can be removed under vacuum. Much more rapid failures occur in environments where the moisture is readily available. There appears to be a close relationship between the per cent humidity and the time-to-failure. There is a strong possibility that other contaminants in the atmosphere also affect time-tofailure. Effect of Solution Concentration Forty C-ring specimens from Temper S (Lot 1) were stressed at 35 ksi for testing in total immersion at four concentrations of sodium chloride at 100 and 80 C. Five specimens were tested to failure at each concentration of 0.01, 0.1, 1.0, and 2.0 M at each temperature. The time-tofailure at stress of 35 ksi was not affected by the sodium chloride content in the concentrations tested. All specimens failed within 20 min at 100 C and within 40 min at 80 C. This is the normal failure time for this temper for a 3.5 per cent sodium chloride solution. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further repro ROMANS AND CRAIG ON ALUMINUM-2NC-MAGNESIUM ALLOY 367 Temperature Tests to determine the effect of temperature on stress corrosion cracking were made on four different types of specimens fabricated from a sample of Temper S (Lot 1). The specimen description is listed below in the procedure. Procedure for C-Ring Jesto—Thirty C-ring specimens, similar to the one shown in Fig. 1, from Temper S (Lot 1), as well as 20 C-ring speci- , FIG. 1—C-ring specimen. FIG. 2—Subsize constant strain specimen and stressing frame. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No furth 368 STRESS CORROSION TESTING mens from Temper S (Lot 2) were stressed to 35 ksi. Five C-rings at each temperature of 100, 77, 50, 36, 8, and 3 C were tested, except Lot 2, which was not tested at 8 and 3 C. The specimens were totally immersed for the entire period of testing. The bolt and nut used for stressing was insulated from the specimen with a plastic dip coating to prevent galvanic effects. FIG. 3—Deflected-beam specimen and stressing fixture. FIG. 4—Constant-load specimen. Procedure for Subsize Constant Strain Tests—Thirty specimens similar to the one shown in Fig. 2 were stressed to 35 ksi. Five specimens were tested in total immersion at each of 100, 84, 80, 40, 20, and 3 C. The stressing frame was covered with a plastic dip coat to prevent contact with the test solution and prevent galvanic contact. Tests with other coatings demonstrated that this dip coat did not affect the results. Procedure for Deflected-Beam Tests—Twelve deflected-beam speciCopyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 mens, 0.060 by 0.25 Downloaded/printed by by 2.75 in. were stressed slightly above the yield University of Washington (University of Washington) pursuant to License Agreement. No further rep ROMANS AND CRAIG ON ALUMINUM-ZINC-MAGNESIUM ALLOY 369 strength. (This stress is used because previous experience has shown that this type of specimen gives erratic results if stressed below the yield strength.) Three specimens were tested at each temperature of 100, 70, 47, and 22 C. The specimens were tested using a standard deflected-beam holder3 as shown in Fig. 3. However, the holder for the elevated temperatures was fabricated from aluminum pipe so it would not be distorted. The specimens were electrically insulated from this holder with glass sleeves over the aluminum bolts. Procedure for Constant-Load Tension Specimens—Twelve specimens similar to the one shown in Fig. 4 were stressed to 35 ksi. They were FIG. 5—The effect of temper and temperature on the time-to-failure. C-rings stressed at 35 ksi. threaded into flexible grips and loaded by means of a lever system. Calibrated lead weights were added to obtain the desired stress. Three each were tested at 88, 70, 50, and 35 C; all specimens were tested to failure except those at 35 C. These were removed after 9 hr, since previous work has shown that in this test the life is considerably extended at temperatures below 50 C. The results of the above tests show that the temperature is the greatest single environmental factor governing the time-to-failure by stress corrosion. A linear function is normally obtained if the log of the time-tofailure is plotted against the reciprocal of the temperature in degrees Kelvin. This linear function holds true down to temperatures of around Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 3 See p. 51. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further re 370 STRESS CORROSION TESTING 8 C. An exception to this is discussed later. There are indications, as shown in Fig. 5, that, in the vicinity of 0 C, the time-to-failure would be infinitely long. Variations in heat treatment of the alloy change the slope of the curve. The data from the C-ring tests are plotted in Fig. 5. An analysis of variance was made on the data using the procedure set forth by Ref 1. The analysis shows that the variance at each temperature is linear when plotted on a log scale. This is represented by the broken lines in Fig. 5. The slope of the more resistant lot of Temper R is presented to show the differences obtained. The results from these tests also manifest some FIG. 6—Effect of specimen configuration and method of applying stress on the time-to-failure. of the effects of the method of stress application and specimen configuration. The data from the subsize constant-strain tests, the deflected-beam test, and the constant-load tension test are plotted in Fig. 6. The C-ring test data in Fig. 5 are repeated for comparison. All specimens except the constant-load tension type produce straight line functions when the log of the time-to-failure is plotted against the reciprocal of the temperature in degrees Kelvin. Note from the plot of the data, a different slope is obtained depending on the shape of the specimens and manner of applying stress. This prevents direct comparison of failure times among the different specimens. Effect of Stress Level on the Time-to-Failure Fifteen subsize constant-strain tension specimens (Fig. 2) were Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 prepared for testing Downloaded/printed byfrom each of Temper R and Temper S. Three speci- University of Washington (University of Washington) pursuant to License Agreement. No further repr ROMANS AND CRAIG ON ALUMINUM-ZINC-MAGNESIUM ALLOY 371 mens from each temper were stressed at each stress level of 10, 15, 25, 35, and 45 ksi and tested by total immersion at 100 C. A Huggenberger Model 892 extensometer was used to set the stress. A major factor affecting the time-to-failure by stress corrosion is the stress level at which the specimen is stressed. As would be expected, the higher the stress, the more rapid the failure. If the log of time-to-failure is plotted against the stress, the resultant data can be represented by two straight lines with different slopes (Fig. 7). At high stresses, there is little difference in time-to-failure due to differences in stress. However, at some point which may be called the critical FIG. 7—Effect of stress on the time-to-failure at 100 C constant-strain tension specimens. stress, a sharp break is noted and small differences in stress cause large differences hi time-to-failure. Effect of pH Five C-ring specimens from Temper S, Lot 1, were stressed at 35 ksi and tested at each pH level of 1.0, 2.0, 3.0, 3.6 (ten specimens were tested at this pH), 4.0, 4.7, 5.0, 6.3, 8.2, and 10. The tests were conducted in total immersion at 55 C. The pH was adjusted with either hydrochloric acid or sodium hydroxide and measured with a Beckman zeromatic pH meter. Solution potential measurements were made at each pH level except pH 10, using a recording potentiometer. A 0.1 N calomel cell with eight Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 junctions was used Downloaded/printed by so that the measurements could all be monitored at University of Washington (University of Washington) pursuant to License Agreement. No further reproductions a 372 STRESS CORROSION TESTING one time. The reference electrode was at room temperature; the specimens were at 55 C. The effect of pH on the time-to-failure varies with the pH value. A definite area of prolonged life occurs at a pH of 3.6 (Fig. 8). On either side of this value, the time-to-failure is much shorter. At a very alkaline pH of 10, the life is considerably extended. The solution potentials do not explain the stress corrosion behavior as a function of pH. The potential at pH 1 was —0.99 v and the remainder up to pH 8 were —1.03 to-1.04v. FIG. 8—Time-to-failure as a function of pH at 80 C. C-ring specimens stressed at 35 ksi. Solution Contaminants Copper Ion Additions—Eighteen C-ring specimens of Temper S (Lot 1) were stressed at 35 ksi for testing in total immersion at 55 C at five levels of copper. Three specimens were tested at each of 0.1, 0.5, 1.0, 3.0, 10, and 100-ppm copper, based on dilution from a stock solution prepared by a weighed addition of copper chloride. Three specimens were tested with no copper added to the solution. The solution was agitated during the test to maintain a flow over the specimens and thereby reduce the effect by of depletion therights diffusion layer. Copyright ASTM Int'lof(all reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further re ROMANS AND CRAIG ON ALUMINUM-ZINC-MAGNESIUM ALLOY 373 In short-term tests, the effect of copper ion additions is to accelerate the rate of cracking. For instance, with no added copper in solution, the specimens fail in an average time of 33 min, while with 0.1 to 10 ppm, the failures occur in an average of 20 min. If amounts are present in the range of 100 ppm, then cracking is retarded with failures occurring in about 40 min. However, in this range, there is noticeable plating out of copper on the surface; this apparently delays the reaction. The alternate-immersion test solution used for routine testing was monitored for copper ion contamination to determine the exact amount of the contamination and its variation as a function of time. The solutions were monitored in two tanks. One tank contained only copper-free alloys (Tank 1); the other tank contained only copper-containing alloys (Tank 2) (Table 2). Aluminum Ion Additions Twelve C-ring specimens of Temper S (Lot 1) were stressed at 35 ksi for testing in total immersion at 55 C at four levels of aluminum ion TABLE 2— Typical copper contamination in alternate-immersion testing (precision of measurement : ±0.02 ppm). Solution Age New 4 days old Tank 1, Copper-Free Alloys 0.07 0.04 Tank 2, Copper-Bearing Alloys 0.05 0.18 concentration. Three specimens were tested at each of 1, 10, 50, and 100-ppm A1+ + + ion based on dilution from stock solution. The solution was agitated during the test to maintain a flow over the specimens. The results of the tests were inconclusive. However, it was established that if the aluminum ion does affect the time-to-failure, it was not measurable in the test. Environmental Comparisons Data Correlation Between Two Laboratories Uncontrolled temperature and humidity alternate-immersion tests were conducted in two different laboratories on four lots of Temper R and two lots of Temper S stressed at 35 ksi. Nine specimens of each lot were tested in 3.5 per cent sodium chloride alternate-immersion tests at each laboratory. The average temperature in these tests was 32 C with an average humidity of 80 per cent. At the time that these tests were conducted, both laboratories conducted tests in an open laboratory area where all types of contaminants were in the air. There was a large spread inCopyright the seasonal temperature and reserved); humidity Wed sinceDec neither laboratory was by ASTM Int'l (all rights 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No furt 374 STRESS CORROSION TESTING air-conditioned. However, the tests were conducted to conform closely with Federal Test Method Standard No. 151. Controlled temperature (38 C) and humidity (45 per cent) tests were conducted between the same two laboratories on three lots of Temper R and three lots of Temper S with nine specimens of each lot tested in 3.5 per cent sodium chloride alternate immersion at each laboratory. These tests were conducted in a small insulated room of very similar construction at both laboratories. The rooms were air-conditioned and the humidity controlled with a dehumidifier. The main difference between the two tests was the water for solution makeup and surface preparation of the specimens. Laboratory A used deionized water and etched the specimens prior to testing as outlined in TABLE 3—Data comparisons between laboratories median failure time of nine specimens of each lot tested in 3.5% NaCl alternate immersion, days. Temper and Lot No. Temper R: 1 2 3 4 Temper S: 1 2 3 Laboratory B c Laboratory A 51° 43 27 29 II 6 12 40 10" 15 <6 7 86 6 13 19 10 6 7 7 <7 <6 4 4 4 0 Uncontrolled ambient temperature and humidity, approximately 90 F and 80% RH. 6 Temperature controlled at 38 C and humidity at 45%. c Surface preparation of specimens in Laboratory B consisted of degreasing in acetone only. They were not cleaned as outlined in the Appendix. the Appendix. Laboratory B used distilled water and degreased the specimens in acetone prior to testing. There is some correlation in the order of the ranking of lot numbers based on time-to-failure between alternate-immersion tests conducted on the same material in separate laboratories at uncontrolled temperature and humidity. However, the failures in one laboratory occur three to five times as fast as in the other laboratory. The correlation in the order of ranking as well as the time-to-failure is greatly improved when temperature and humidity are controlled. However, somewhat faster failures are still in Laboratory B (Table 3). Discussion of Results The number of factors which are shown to affect the time required to produce stress corrosion cracking should dictate more careful controls on future testing. The fact that moisture is the only requirement to initiate Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 cracking shows the sensitivity of this corrosion process to its environment. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further repr ROMANS AND CRAIG ON ALUMINUM-ZINC-MAGNESIUM ALLOY 375 Ansbacher [2]4 observed that the oxide film is capable of adsorbing water vapor as well as atmospheric contaminants. It is known from the experimental data and other sources [3] that the per cent moisture affects, directly, the time-to-failure which decreases with increasing humidity. It is reasonable to assume that air contaminants would also accelerate the cracking process. It has not been possible to place an exact numerical value on the effect of humidity due mainly to the interaction between it and temperature. It is shown that temperature is one of the biggest factors and should be the one most closely controlled. A linear function is obtained when the log of the time-to-failure is plotted against the reciprocal of the temperature in degrees Kelvin. This holds true for all the specimens used except those stressed by loading with a dead weight. The stress level at which a specimen is tested determines to a large degree the expected life. High stresses produce rapid failures with little differences in the time required. However, once a critical level is reached, the life is greatly extended by small decreases in stress. A plot of the stress versus life can be represented by two straight lines with different slopes. The critical stress is referred to as being at the intersection of the two lines. Brenner and Gruhl [4] state that, "Below the kink in the stress/life curve the life is practically unlimited." However, our data shows that, for the alloy tested, failures do occur below the break in the curve. There is a possibility of a second break below which failures would not occur even with this alloy. The data are sometimes represented by a curve in which the sharp break is eliminated [5]. It has been reported in the literature [6] that the failure time of aluminum-magnesium alloys is shortened by increasing the sodium chloride concentration. This is not the case with the aluminum-zincmagnesium alloy system which is insensitive to the effect of concentration. The pH affects the time-to-failure in a somewhat radical manner. A maximum life occurs at a pH of 3.6 in the range from pH 1 to pH 8. Investigations are underway to determine the reason behind this phenomenon. Mattsson [7] shows a similar relationship with brass but as a mirror image of the plot in Fig. 1. He shows a maximum life at a pH of 7.8 in the range 11.2 to 4.7. Copper ions in elevated temperature tests reduce the time required to obtain cracking. It is postulated by the authors that their role is to increase the potential difference between the cathodic and anodic sites which at elevated temperatures promotes cracking instead of pitting. Bell and Campbell [8] and Godard [9] found that copper in amounts as low as 0.02 ppm have a large effect in stimulating pitting corrosion. 4 The italic in (all brackets to theWed list Dec of references appended to this Copyright by numbers ASTM Int'l rightsrefer reserved); 16 15:53:43 EST 2015 paper. Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further re 376 STRESS CORROSION TESTING The test results show that correlation of tests between laboratories is greatly improved by controlling temperature and humidity. It is predicted that good correlation can be obtained if the tests are controlled to a high degree of accuracy. Conclusions Carefully prepared tests conducted in distilled water and in a desiccated atmosphere show that the only requirement for the initiation and propagation of stress cracks is the presence of moisture. The moisture does not have to be added intentionally. The oxide film is capable of entrapping or adsorbing sufficient moisture to allow stress corrosion cracking to occur. The surface moisture can be removed under a dynamic vacuum. This is evidenced by the fact that stress corrosion failures do not occur under this condition. The experimental data show that the per cent moisture affects the time-to-failure which decreases with increasing humidity. It was not possible to place an exact numerical value on the effect of humidity due to its interaction with temperature. Temperature is the biggest factor controlling the time required for stress corrosion cracking to occur. Extremely rapid stress corrosion failures occurring in minutes can be obtained at 100 C, while at 20 C the time required may be several days. There are indications that at 0 C, the time-to-failure would be infinitely long. Between 100 and 30 C, the log of the time-to-failure is linear if plotted against the reciprocal of the absolute temperature. This holds true for all specimens and configurations tested except the constant load which has a linear function only between 100 and 50 C. The fact that this test method does not produce linear data makes it undesirable from the standpoint of extrapolating to lower temperatures. The other three specimen configurations produce linear data but with different slopes. This will prevent direct comparisons between the different types of specimens. If the metallurgical condition of the alloy is changed, it will also change the slope of the curve. It is postulated that the reason for this behavior is that the fabrication practice can form a grain boundary precipitate which is much more active at elevated temperature than ambient temperature tests would indicate. The stress level at which a specimen is tested determines to a large degree the expected life. A stress-life curve can be represented by two intersecting straight lines. At high stresses, rapid failures are obtained and there is very little difference in the time required to produce stress corrosion failure as the unit stress is lowered. However, once a critical stress level is reached, there is a sharp break in the curve giving a large increase in life with small decreases in stress. Copyright by ASTM Int'l (all rights reserved); Wed in Decthis 16 15:53:43 EST 2015 The aluminum-zinc-magnesium alloy used investigation is inDownloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further ROMANS AND CRAIG ON ALUMINUM-ZINC-MAGNESIUM ALLOY 377 sensitive to changes in the sodium chloride concentration in amounts of 0.01 to 2 M. The pH of the sodium chloride test solution affects the timeto-failure in a somewhat radical manner. There is an increase in life at a pH of 3.6 in the range from pH 1 to pH 8. At the present time, there is no plausible explanation for this behavior. Copper ions in elevated temperature tests accelerate failures. It is postulated that their role is to increase the potential difference between the cathodic and anodic sites, which at elevated temperatures promotes cracking instead of pitting. The reproducibility of tests on the same materials between laboratories is poor. This indicates the need for better controls on environmental factors than is presently being used. The fact that the correlation of data between two laboratories was greatly improved by controlling only the temperature and humidity indicates that additional controls on water purity, grade salt, pH, and other factors should be beneficial. A cknowledgment The assistance of F. E. Loftin and Z. L. Vance in obtaining the data for this presentation is gratefully acknowledged. The authors gratefully acknowledge the permission of Reynolds Metals Co. to publish this paper. APPENDIX Procedural Details All of the specimens were prepared for testing according to the following procedure: 1. The specimens were machined from the sample in such a way as to orient the short-transverse direction of the plate parallel to the applied stress.5 The four types may be listed as the C-ring 0.750 in. outside diameter by 0.060-in. wall (Fig. 1); the constant-strain subsize tension specimen 1.875 in. long by 0.125-in. reduced section (Fig. 2); the deflected-beam specimen 3 in. long by 0.064 in. thick by plate gage (Fig. 3); and the constant-load tension specimen 3 in. long by 0.125-in. reduced section (Fig. 4). The largest portion of this work used a C-ring specimen. Exceptional care was taken to ensure the smoothest possible machined surface. No abrasive was ever used due to the danger of smearing the surface. 2. The specimens were degreased in acetone to remove residual oil and then etched for 30 to 40 sec in 5 per cent sodium hydroxide solution at 77 C. This was followed by a dip in concentrated nitric acid at 25 C to remove the smut. 5 Prestley, J. S., "The Effects of Specimen Orientation on Resistance to Stress Corrosion of Aluminum Alloys—A Mathematical Model," presented at the sixtyninth ASTM Annual Meeting, Atlantic City, N. J., 26 June-1 July, 1966, unCopyrightwork. by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 published Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No furt 378 STRESS CORROSION TESTING After the specimens were stressed to the desired level, all portions of the fixture were waxed to prevent any galvanic effect or solution contamination from the fixture. The specimens were always cleaned and stressed directly before testing to minimize surface contamination by laboratory atmospheres. Test Solutions All the test solutions used for the experiments reported in this paper were prepared from deionized water with an average resistance of 200,000 ohm-cm, unless otherwise noted. The sodium chloride was USP grade added to obtain a concentration of 3.5 per cent. The solution was aged for seven days prior to use to allow stabilization of pH. The solution is about pH 4 when first prepared and requires about five days for the CO2 content to reach equilibrium with atmospheric conditions. After five days, the pH will adjust to 6 to 6.5. The solution then meets the requirements of Federal Test Method Standard No. 151. Test Equipment Unless otherwise stated, the alternate-immersion tests reported have the temperature controlled at 28 ± 1 C. The humidity was controlled at 45 ± 2 per cent. The solution containers were 50-gal glass aquaria with Plexiglas immersion racks. The specimens were exposed in a single layer to prevent differences in drying rates or drainage onto a lower layer. All-plastic immersion pumps were used to circulate the solution to insure a homogeneous condition. When other alloys were tested, the ones containing deliberate additions of copper were placed in a separate container from the copper-free alloys. The total immersion tests were conducted in glass containers immersed in a water bath in which the temperature was controlled to ±1 C. A hot plate was used for the total immersion specimens tested at boiling temperature. References [1] Acton, F. S., Analysis of Straight Line Data, Wiley, New York. [2] Ansbacher, F., "The Effects of Water Vapor on the Electrical Properties of Aluminum," Nature, Vol. 171, 24 January, 1953. [3] Champion, F. A., 'The Interactions of Static Stress and Corrosion with Aluminum Alloys," Journal of the Institute of Metals, Vol. 83, 1954-55. [4] Brenner, P. and Gruhl, W., "Stress Corrosion Testing of Al-Zn-Mg Under Constant Tension and Bending," Zeitschrift fuer Metallkunde, Vol. 52, No. 10, 1961, pp. 599-607. [5] Symposium on Corrosion Fundamentals, University of Tennessee Corrosion Conference, University of Tennessee Press, March 1965. [6] Ferryman, E. C. and Hadden, S. E., Journal of the Institute of Metals, Vol. 77, 1950, p. 207. [7] Mattsson, E., "Stress Corrosion in Brass Considered Against the Background of Potential/pH Diagrams," Electrochemical Acta, Vol. 3, 1961. [8] Bell, W. A. and Campbell, H. S., "Aluminum in Fresh Waters," British NonFerrous Metals Research Association Laboratories, London N.W. 1, 8 April, 1961. [9] Godard, H. P., "The Corrosion Behavior of Aluminum," Corrosion, Vol. 11, No. 12, Dec. 1955, pp. 542T-552T. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further re DISCUSSION ON ALUMINUM-ZINC-MAGNESIUM ALLOY 379 DISCUSSION D. O. Sprowls1 (written discussion)—It is generally recognized that the specimen configuration and the manner of stressing affect both the probability of developing a stress corrosion crack and the rate of crack propagation in susceptible alloys. The data given in Fig. 6, however, provide an unusual opportunity of making a direct comparison of three different procedures, namely: (a) direct tension, constant load; (b) direct tension, "constant strain;" and (c) bending, constant strain (C-ring). (The deflected-beam specimen is excluded from this comparison because the beams were stressed at a much higher level.) One would expect from the distribution of stress in the specimens that the tension specimens, especially when placed under constant load, would fail more rapidly than the C-ring specimen. Brenner and Gruhl2 have, in fact, shown that aluminum-zinc-magnesium alloy sheet tension specimens stressed in tension failed more rapidly and at a lower stress level than similar specimens stressed in bending. This same comparison was shown for weldments by Shumaker et al.3 At the Alcoa Research Laboratories we regularly use large numbers of C-rings and 0.125-in.-diameter by 2-in.-long tensile bars stressed by "constant strain".4 Usually one specimen or the other is used, the choice depending upon the dimensions of the product to be tested. However, in several tests of experimental heat treatments for -T6 type 7039 core, rolled plates in the 2 to 4-in. range of thickness were stress corrosion tested in the short-transverse direction using both types of specimen. As a general rule, the tensile bars cracked sooner than the C-rings, and in cases of borderline susceptibility, the percentage of failures was higher for the tensile bars. A sample comparison is shown in Fig. 9. The comparisons shown by the authors in Fig. 6 are surprising in that: (1) the C-rings failed more rapidly than the tension specimens stressed in tension by "constant strain," and (2) at the lower temperatures the tension specimens stressed by constant load presumably would have failed after longer exposures than the C-rings. The expression "constant strain" is used with quotation marks in the case of the tensile bar, because with this particular tensile bar and stressing frame the specimen is 1 Assistant chief, Chemical Metallurgy Div., Alcoa Research Laboratories, New Kensington, Pa. 2 Brenner, Paul and Gruhl, Wolfgang, "Stress-Corrosion Cracking Tests of AlZn-Mg 3 Under Constant Tensile and Bending Strain," Zeitschrift fuer Metallkunde, Vol. 52, No. 10, 1961, pp. 599-607. 3 See p. 317. 4 Sprowls, D. O. and Brown, R. H., "What Every Engineer Should Know About Stress Corrosion of Aluminum," Metal Progress, Vol. 81, April and May 1962. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No furth 380 STRESS CORROSION TESTING not loaded purely in constant strain. An analysis of the loading characteristics of this system5 has shown that the strain energy developed in the specimen and the stressing frame during loading is sufficient that when localized cracking of the specimen occurs, the average tensile stress on the net section through the crack increases in almost the same manner as in the case of dead-weight loading. Therefore, our experience at Alcoa Research Laboratories indicates that the behavior of this "constant strain" tensile bar and the constant load tension specimen generally are FIG. 9—Comparison of stress corrosion test specimens, short-transverse tests of 7039 alloy plate. about the same and both constitute a slightly more severe test than the C-ring specimen. It should be pointed out that the "critical stress" identified by the authors in Fig. 7 should not be confused with the stress corrosion "thresholds" referred to by many investigators. The term "threshold stress" is generally used to denote the highest sustained tensile stress that did not cause failure by stress corrosion cracking under the conditions of the test, and this stress is not necessarily related to the shape of the stress specimen life curve. The threshold stress so determined will, of course, be related to the specific test conditions. Threshold stress and critical stresses as used by various investigators are discussed in more detail in 6 See p. by 342.ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Copyright Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reprodu DISCUSSION ON ALUMINUM-ZINC-MAGNESIUM ALLOY 381 the report of Task Group 3 of ASTM Subcommittee VI of Committee G-l on p. 292 of this volume. H. B. Romans and H. L. Craig, Jr. (authors)—In general, the authors can agree with the contention that the subsize constant-strain specimen provides faster stress corrosion failures than the C-ring. However, we have shown that this does not hold true for all materials under all test conditions. We believe this behavior should show that making a flat statement about anything in stress corrosion testing is really hard to justify. A. R. C. Westwood6 (written discussion)—Would you comment on the variation in slope of the log tp versus (1/T) curve with temper for a given alloy composition? If an activated process is controlling, this suggests that some change in mechanism is occurring. Messrs. Romans and Craig—The explanation for the slope variations may be explained by considering the potential differences which can exist between the grain boundary and the grain. If the volume of the precipitate particles in the matrix were equal to the volume in the grain boundary, then the potential difference would be small. However, if no matrix precipitate were present and there were a large volume of grain boundary precipitate, then there would be a very large potential difference. This would result in a highly anodic grain boundary. The specimens we were working with were between these two extreme cases. At the high temperatures the grain boundary of all of the specimens was attacked very rapidly. The temperature factor was great enough to overcome some of the potential difference due to precipitate distribution. This resulted in a narrow spread among specimens in time-to-failure. At low temperatures the potential difference between matrix and boundary was the controlling factor. This gave a wider spread in stress corrosion cracking results due to small differences in precipitate distribution resulting in different slopes for each specimen. 9 Associate director, Research Institute for Advanced Studies, Martin Co., Baltimore, Md. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further reproduct THIS PUBLICATION is one of many issued by the American Society for Testing and Materials in connection with its work of promoting knowledge of the properties of materials and developing standard specifications and tests for materials. Much of the data result from the voluntary contributions of many of the country's leading technical authorities from industry, scientific agencies, and government. Over the years the Society has published many technical symposiums, reports, and special books. These may consist of a series of technical papers, reports by the ASTM technical committees, or compilations of data developed in special Society groups with many organizations cooperating. A list of ASTM publications and information on the work of the Society will be furnished on request. Copyright by ASTM Int'l (all rights reserved); Wed Dec 16 15:53:43 EST 2015 Downloaded/printed by University of Washington (University of Washington) pursuant to License Agreement. No further repro