Engineering Failure Analysis 7 (2000) 427±450 www.elsevier.com/locate/engfailanal Damage due to hydrogen embrittlement and stress corrosion cracking Jarmila Woodtli, Rolf Kieselbach* EMPA, Uberlandstrasse 129, CH-8600 Dubensorf, Switzerland Received 3 September 1999; accepted 20 September 1999 Abstract Damage of metals due to the in¯uence of hydrogen and to stress corrosion cracking is quite frequent and leads to dangerous failures as well as to loss of property and large compensational payments by insurance companies. One reason for this, is that some designers and engineers seem to lack sucient knowledge of the basic mechanisms of these phenomena and accordingly often have only vague ideas how to prevent such failure causes. Although the basic concepts can be found in a number of good text books it seems worthwile to recall them in a short comprehensive paper. 7 2000 Elsevier Science Ltd. All rights reserved. Keywords: Embrittlement; Environmental interaction; Hydrogen-assisted cracking; Stress corrosion cracking 1. Introduction A previous survey [1] has shown that approximately one third of all failure cases are caused by environmental eects like corrosion. This failure cause means more than simple rusting of iron and steel as is shown in this paper. Whereas this kind of corrosion is visible and thus can be easily detected, the damage caused by the phenomena discussed in this paper is normally invisible and unforeseeable and therefore much more dangerous. A ®rst example is the failure of a storage tank for compressed hydrogen. The consequences of this can be ascertained from Fig. 1. This failure was caused by the growth of large fatigue cracks which was induced by hydrogen in¯uence. The total damage paid by insurance in this case was approximately US$50 million. Hydrogen damage is more frequent than many people would suspect. Another important source of failure is stress corrosion. In this failure mechanism the damage is produced by simultaneous action of stress, corrosive substance and the properties of the material. An * Corresponding author. Tel.: +41-1-823-5511; fax: +41-1-823-4014. E-mail address: rolf.kieselbach@empa.ch (R. Kieselbach). 1350-6307/00/$ - see front matter 7 2000 Elsevier Science Ltd. All rights reserved. PII: S 1 3 5 0 - 6 3 0 7 ( 9 9 ) 0 0 0 3 3 - 3 428 J. Woodtli, R. Kieselbach / Engineering Failure Analysis 7 (2000) 427±450 example of a failure caused by stress corrosion cracking was the accident at the Uster indoor swimming pool in Switzerland. There, a dead concrete ceiling had been suspended over the pool by a great number of bars made from stainless steel (AISI 304, DIN 1.4301). After several years' service some of these bars broke, the roof collapsed and 15 swimmers were killed. The investigation showed that the stainless steel had been severely corroded (Fig. 2). By detailed metallurgical tests it was proven that the failure had been brought about by stress corrosion cracking. A typical indication for this kind of damage is the branched cracks of Fig. 3. Material damage due to hydrogen embrittlement and stress corrosion cracking can be classed as cases which cannot be clari®ed based on mechanical parameters alone. Both types of damage are caused by the ambient medium and are therefore controlled by physical, chemical or electromechanical processes. The appearance of the damage is also similar: in both cases delayed low ductility fractures or cracks appear and the material is embrittled only locally in locations where contact has occurred with the damaging medium. The role of hydrogen in the mechanism of stress corrosion cracking has been known since the 1970s so that preventative measures have since been introduced. Although hydrogen-induced cracking and stress corrosion cracking are similar phenomena, both types of damage are usually described and treated separately. The reason for this is the complexity of the damage mechanism which is dependent on the particular material and damaging medium. In some cases a combination of both mechanisms can be found which is then called hydrogen-induced stress corrosion cracking. The following paragraphs deal with this subject from a failure analysis point of view. 2. Hydrogen damage 2.1. General aspects Hydrogen embrittlement is usually understood as the unwanted delayed brittleness of a material Fig. 1. Factory after rupture and explosion of a storage tank for hydrogen. J. Woodtli, R. Kieselbach / Engineering Failure Analysis 7 (2000) 427±450 Fig. 2. Fracture surface of broken hanger of dead concrete roof, made from stainless steel. Fig. 3. Branched cracks of stainless steel bar of Fig. 2. 429 430 J. Woodtli, R. Kieselbach / Engineering Failure Analysis 7 (2000) 427±450 which is caused by the presence of hydrogen within the material. Practically all metal materials can be damaged by the absorption of hydrogen, if a sucient quantity can penetrate into the material. The sources of the hydrogen, the paths it takes to enter the material and the embrittlement mechanisms are extremely diverse. These factors must therefore be established in each particular case. Only then can preventative measures be taken at the correct location or during the correct stage of a process. The aim of this paper is to explain both the embrittlement mechanisms and the characteristic visible forms of the damage. These are illustrated by individual examples. For the failure analysis and the resulting preventative measures the following aspects of hydrogen embrittlement need to be understood: . . . . the source of the hydrogen; the absorption of the hydrogen into the metal; transport processes; the embrittlement mechanism and visual appearance. 2.2. Source of the hydrogen Hydrogen always enters metals as exogenous contamination. A component or system can be contaminated by hydrogen in various stages of its useful life. This begins in the metallurgical process [4], because the hydrogen solubility in the molten metal is much higher than when it is in the solid condition (see Fig. 4). This hydrogen loading is reversible to a great extent, because it is caused mainly by the storage of the hydrogen in the interstitial positions of the lattice. Irreversible damage only occurs if the hydrogen can euse or if it can accumulate as gas in the hollow spaces. This type of hydrogen damage is of importance in welding practice and is exhibited for instance in the development of cold cracks or ®sh eyes (or ¯akes). This type of damage is also produced during casting. Another source of hydrogen during the manufacturing process is the galvanising process or ¯ash pickling. The hydrogen is produced by the cathodic partial reaction of the electromechanical processes which is why the damage caused is designated as electromechanical induced hydrogen cracking. At ®rst the hydrogen diuses in the material (Fig. 5) and is then taken up into traps. In relatively soft steel or aluminium alloys, the hydrogen can recombine to form gas molecules on internal defects such as nonmetallic inclusions. Due to the high gas pressure [2] separations are produced parallel to the surface (bubble formation). In Fig. 6 the fracture surface of a tensile specimen is illustrated in which a large number of ®sh eyes can be detected. The fracture happened during a tensile test which had a reduced tensile strength from 487 to 529 N/mm2 (speci®ed value: 540 N/mm2) (Fig. 7). The hydrogen, which can be recognised fractographically from the cleavage aureole around the pores and the non-metallic inclusions, originated from humid feed material or badly dried oven lining. The greater the number of pores within the casting, the greater the amount of enclosed hydrogen (see Fig. 8). This type of hydrogen embrittlement is also called internal hydrogen embrittlement and is explained by the pressure theory based on thermodynamic principles. The equation for the balance between the atomic and gaseous hydrogen is: p 1 cH 135 pH2 ÿ 6500=RT, where cH is the hydrogen concentration in ppm and pH2 the hydrogen pressure in MPa in the solid phase of the adjacent gas phase. R is the gas constant and T is the absolute temperature. Accordingly, 1 ppm of dissolved hydrogen is balanced by gaseous hydrogen at room temperature and at a pressure of 2 105 MPa. The embrittlement is due to the fact that the excess dissolved hydrogen disperses into inclusions, pores or microcracks and, because of the high pressure (e.g. in the form of blistering ¯akes) J. Woodtli, R. Kieselbach / Engineering Failure Analysis 7 (2000) 427±450 431 or under external forces, results in cracks to the lattice. Even when it is considered that for kinetic reasons the actual pressures reached are one or two times lower, there would still be considerable pressure which could have an impact on the lattice. High tensile strength and cold drawn steels with apparent elasticity limits exceeding 650 N/mm2 are at particular risk. The embrittlement is approximately proportional to the logarithm of the hardness and in addition is related to the value of the diusion coecient for hydrogen (see Fig. 9) [3]. Martensitic structures are considerably more susceptible than bainitic structures, ferrite, austenite or precipitation-hardened steels. Moreover, coarse grained materials are more susceptible to brittleness than ®ne. In steel, the diusible hydrogen should be forced out during the surface treatment between 190 and 2208C [9] because the eusion rate is far too low at room temperature (see Fig. 10). Fig. 4. Solubility of hydrogen in iron as a function of temperature and of pressure. 432 J. Woodtli, R. Kieselbach / Engineering Failure Analysis 7 (2000) 427±450 Fig. 5. Coecient of diusion for hydrogen in ferrite (F) and austenite (A) respectively; Lattice diusion (G) [3]. J. Woodtli, R. Kieselbach / Engineering Failure Analysis 7 (2000) 427±450 433 Fig. 6. Fish eyes on the fracture surface. Fig. 11 shows blistering phenomena on a galvanised steel surface (unalloyed carbon steel tempered to 460 HV5). Apparently these washers were not fully degassed. They failed after a relatively short time in operation due to low ductility cracks and fractures even though the load at a torque of 70 Nm was very low. Fig. 7. Cleavage fracture of the aureole of a ®sh eye region. 434 J. Woodtli, R. Kieselbach / Engineering Failure Analysis 7 (2000) 427±450 Fig. 8. Solubility of hydrogen in iron having dierent porosities at dierent temperatures and pressures. The fracture surface (Fig. 12) indicates typical characteristics of hydrogen embrittlement in the form of a partially intercrystalline fracture with ductile markings on the grain boundaries (crow's feet). This fracture pattern may be a typical indication but it is nevertheless not a satisfactory proof of hydrogen embrittlement. In this particular case the residual hydrogen content of up to 1.49 ppm was determined by a gas chromatograph. This hydrogen concentration was above the lower brittleness threshold which lies between 0.3 and 3 ppm due to the particular material [5]. In general the contact between hydrogen gas and a steel surface at room temperature cannot be considered as a problem. On the one hand the dissociation constant of hydrogen at room temperature is very high (429.5 kJ/mol), and on the other hand the diusion coecient of 1.5 108 is very low (see Fig. 5 [3]). Therefore for almost 100 years containers, reactor tanks and gas bottles made of steel have been used successfully for the transport of hydrogen. Nevertheless the formation of cracks is possible, especially due to the combined impact of mechanical stress concentration and hydrogen dissociation. J. Woodtli, R. Kieselbach / Engineering Failure Analysis 7 (2000) 427±450 435 Fig. 9. Coecient of diusion for hydrogen in dierently structured psteels built up during hydrogen loading (solid line), eusion (dotted line). At right: time of diusion for a length of 1 cm l D t [3]. The high dissociation energy of the hydrogen is applied by a deforming metal surface. For this reason the hydrogen in a tension or ductile ®eld can penetrate into the metal lattice much more easily than into an intact tension-free surface, even if only microscopically. Therefore the atomic hydrogen diuses preferably on notched locations and crack openings with a three axis tension distribution and causes decohesion of the lattice. Especially in high-tensile steel, high crack propagation rates can occur (see Fig. 13). When the hydrogen is under pressure, the crack propagation rate is higher by a factor of 2 [6]. H2S can also cause an increase in the crack propagation rate [7]. At temperatures above 3008C the partial pressure is the decisive factor for the impact of the hydrogen (see Fig. 14). Above 3008C the hydrogen solubility in steel follows the Sievert law: p 2 cH kT PH2 The Arrhenius equation kT k0 exp ÿQ=RT can be used to determine kT, where Q is the heat of solution of 27.2 kJ/mol. In addition at this temperature there is no longer a problem of the high energy threshold for the 436 J. Woodtli, R. Kieselbach / Engineering Failure Analysis 7 (2000) 427±450 Fig. 10. Diusion of hydrogen after loading for 80 h in a saturated solution of H2S at room temperature. dissociation, adsorption and absorption of the hydrogen. After diusion of the hydrogen the damage at increased temperatures, however, is not caused by the material embrittlement but by the structure changes in the material. In carbon steels the combined carbon reacts as Fe3C with the hydrogen and forms methane. The steel loses its strength due to decarburization. The resistance to the formation of methane can be improved by suitable alloying additions or carbide components such as Cr, Mo, V, W or by using austenitic steels. 3. Stress corrosion cracking Stress corrosion cracking is understood as the formation of cracks in metals with simultaneous impact of certain corrosive media and tensile stress. Stress corrosion cracking does not only cause low ductility fractures in high tensile materials but also in ductile materials. In all cases of stress corrosion cracking three parameters with synergy impact are involved: . material . medium and . mechanical loading J. Woodtli, R. Kieselbach / Engineering Failure Analysis 7 (2000) 427±450 Fig. 11. Opened bubble on a galvanised steel surface. Fig. 12. Partially intercrystalline fracture with ductile marks on the grain-faces. 437 438 J. Woodtli, R. Kieselbach / Engineering Failure Analysis 7 (2000) 427±450 Fig. 13. In¯uence of hydrogen on the crack growth rate of a low alloy steel. J. Woodtli, R. Kieselbach / Engineering Failure Analysis 7 (2000) 427±450 439 Fig. 14. In¯uence of hydrogen pressure and temperature on the onset of damage by CIW [8]. Table 1 Some examples for stress corrosion cracking [21] Material Medium Remarks Low alloy ferritic steels Alkaline liquids (caustic solutions, carbonate and bicarbonate solutions) Among other things, so-called ``caustic embrittlement''. Precise composition and heat treatment of material as well as temperature and concentration of the solution are important As above. Test in a boiling standardised (DIN 50915) water and Ca (NO3)2ÿ solution as test for susceptibility to intercrystalline SSC Min. temperature of approximately 608C usually required, transcrystalline crack formation Nitrate solutions High alloy austentite steels Chloride solutions Caustic alkali solutions Nickel-based alloys Caustic alkali solutions Copper-based alloys Aluminiumbased alloys Titaniumbased alloys Ammonia solution, nitrate solutions Trans- or intercrystalline crack formation dependent on the electrochemical potential, concentration, heat treatment, etc. Intercrystaline crack formation on sensitised material (chrome depletion: intercryst. without tensile stress in Strauss test) Trans- or intercrystalline cracking depending on material (or condition) electrochemical potential, concentration, etc. ``Coriou'' eect; dependent on material, long incubation time, crack conditions have a negative eect Composition of the solutions is critical In particular solutions containing halogen Aqueous solutions containing chloride Intercrystalline crack formation, composition and heat treatment of the material is important Crack formation often only when sharp notches are present Oxidised high temperature water Pure high temperature water 440 J. Woodtli, R. Kieselbach / Engineering Failure Analysis 7 (2000) 427±450 Each of these parameters must be considered individually; therefore with respect to stress corrosion cracking, a material cannot be characterised solely on the basis of its chemical composition but also by its condition caused by heat treatment (see Table 1). In order to be able to classify the wide range of damage mechanisms, it is advisable to dierentiate between cathodic, anodic and expansion-induced crack formations. According to its particular mechanism, cathodic stress corrosion cracking can be classi®ed within the range of hydrogen embrittlement, as already discussed in Section 2.2. 3.1. Anodic stress corrosion cracking Anodic stress corrosion is selective. It occurs only in the passive condition of the material, i.e. when the metal surface is covered by a passive or protective layer. Most model concepts assume that an accelerated anodic disintegration of metal takes place on a crack tip kept free from protective layers by chemical and mechanical eects (see Fig. 15) [15]. According to this theory the freshly created crack edge must be immediately re-protected by the formation of passive or covering layers because otherwise crack fatigue could occur. In this way stress corrosion cracking spreads through the material like an ``electromechanical knife''. A general con®rmation of this concept is demonstrated by Parkin's theory (see Fig. 16) [16]. Fig. 15. Mechanism of common SCC by anodic solution. J. Woodtli, R. Kieselbach / Engineering Failure Analysis 7 (2000) 427±450 441 Fig. 16. Velocity of SCC and maximal current density [17]. According to this theory, the growth rates of stress corrosion cracks have a strong interrelation to the measured peak current densities with the anodic disintegration of various material/corrosion medium combinations. Intercrystalline stress corrosion cracking occurs in passivatable materials, for instance sensitised stainless steels, certain aluminium alloys and C steels in oxidised media (see Table 1) [21]. This corrosion occurs only in a limited potential range. Due to enriched impurities, the grain boundary needs a higher passivation current density than the internal grain. In addition the breakdown potential of the grain boundary is reached earlier than that of the internal grain. In this connection, mechanical tensile stresses play an important role. On the one hand they cause mechanical cracking of the grain boundary region and on the other hand they reduce the breakdown potential and thus make corrosion possible. The progress of intercrystalline stress cracking is therefore independent of the eects of mechanical stress. For this reason the residual tensile stresses located only on the outermost surface of a component are sucient to initiate stress corrosion cracking. 442 J. Woodtli, R. Kieselbach / Engineering Failure Analysis 7 (2000) 427±450 The large number of models written for transcrystalline stress corrosion cracks place emphasis either on the electromechanical or the physical metal mechanisms. The ®lm breakdown theory is based on the repeating sequence of the formation and destruction of the outer layers which is made possible by the gliding planes on the top surface. This process can be accelerated by the formation of elements between the passive walls of the crack and the active base of the crack (see Fig. 17) [18]. According to the tunnel theory, as a result of the interaction of a sliding band with passive layer, tunnel shaped pitting is produced which expands sideways under mechanical stress and later forms a feathery fracture structure (see Fig. 18). The appearance of stress corrosion cracking can be demonstrated by damage to components made of a+b brass. At the beginning of this century the phenomenon of stress corrosion cracking was observed on copper based alloys. The aggressive impact of ammonia, together with its compounds and derivatives, was very soon recognised [19]. In the recent past, cases of damage caused by the impact of industrial atmospheres have increased [20] which have also been shown by long-term outdoor investigations. For instance cold worked samples made of brass CuZn37 with various stress relieving anneal treatments indicate that SSC can still occur after a long incubation period dependent on ambient conditions and due to the very low internal tensile stress [22]. Fig. 17. Mechanism of the theory of ®lm breaking for SCC. J. Woodtli, R. Kieselbach / Engineering Failure Analysis 7 (2000) 427±450 Fig. 18. Mechanism of the tunnel theory for SCC. 443 444 J. Woodtli, R. Kieselbach / Engineering Failure Analysis 7 (2000) 427±450 Figs. 19 and 20 show plumbing parts damaged on the basis of crack formation in the lengthways and transverse directions. The fractographic investigations indicated an intercrystalline fracture progression (Figs. 21 and 22). A branched crack progression and a perfect microstructure consisting of a+b mixed crystals was discovered on the metallic microsections including the fracture area. Extreme cold formings on the thread ¯anks, which would have indicated an installation error, were not present. The blue colouring of the fracture surfaces as well as the proof of nitrogenated compounds on the fracture surfaces indicate that SCC had been triggered by ammonia or its compounds with the additional impact of moisture. In this particular case the mechanical stresses that triggered the fractures can be considered as low tensile stresses which are unavoidable during installation work. The tap extension with the transition of the cross-section also indicates a location with stress concentration. The modi®cation to the design (Fig. 19(D)) should help to reduce the stress peaks on the transition of the cross-section. 3.2. Strain-induced corrosion cracking The material and the corrosive medium are the decisive factors for conventional SCC. In contrast to this, for strain-controlled corrosion cracking, the mechanical load is a more central factor. The stress can be static but also cyclic. When the load frequency is high, the damage is caused by fatigue corrosion cracking. Strain-induced corrosion cracking occurs especially in alternating operating conditions. Bead chain type cracks are formed when repeated changes occur between corrosion processes caused during stationary conditions and mechanical loads in normal operation during which Fig. 19. Faucet extension made from brass showing fracture (A), (B) and cracking (C). J. Woodtli, R. Kieselbach / Engineering Failure Analysis 7 (2000) 427±450 445 Fig. 20. Longitudinal crack in brass ®tting. mechanical destruction of the top layer is produced. Magnetite layers on pipe bends and tube plates in hot water systems can be mentioned as examples. Thermal dilatation of the component is sucient as the mechanical load. Fig. 21. Microstructure of a+b-brass with branched cracks. 446 J. Woodtli, R. Kieselbach / Engineering Failure Analysis 7 (2000) 427±450 Fig. 22. Intercrystalline fracture surface. 3.3. Hydrogen-induced stress corrosion cracking (HISCC) This type of damage is a transition form between the classical hydrogen embrittlement and stress corrosion cracking and applies in particular to high-tensile steel in low aggressive water solutions, austentitic steel in diluted sulphuric acid, alloys in halogenide solutions and uranium alloys in water solutions [10]. When a building component or system is subjected to corrosion created by the development of hydrogen, the cathodically isolated hydrogen diuses into the material and causes local embrittlement. The hydrogen develops either in oxygen-free, neutral or alkaline mediums by the reduction of water molecules 2H2 O 2e ÿ 2OH ÿ 2H 3 or as a result of hydrolysis of corrosion products (in pitting locations, narrow cracks), for instance: 3Fe 2 4H2 O Fe3 O4 8H 2e ÿ 2H 2e ÿ 2Had 4 In acid solutions (pH < 4) the protons H+ are reduced to H2. In general, acid corrosion of steel in diluted halogenide solutions emanates from an acid corrosion mechanism in narrow gaps [11]. The diusion processes cause increased concentrations of halogenide and metal ions inside the cracks and, due to hydrolysis (Eq. (4)), the pH value in the crack sinks [10]. In addition, a decrease in the potential independent of the external polarisation can be detected in the narrow cracks and splits. This is caused J. Woodtli, R. Kieselbach / Engineering Failure Analysis 7 (2000) 427±450 447 by the development of H2 gas bubbles which leads to the production of a ``short circuit cell'' at the crack opening which to a great extent is decoupled from the surface. In English technical literature the expression ``occluded corrosion cell'' is used [12]. Molecular hydrogen is less damaging for steel than atomic hydrogen due to its high dissociation constant at room temperature. When kinetic inhibition of recombination exists, as is the case with the presence of, for instance, sulphides, arsenic, selenium and phosphorus compounds, considerable partial pressures can occur due to the accumulation of adsorbed hydrogen atoms on the surface. The interstitially dissolved hydrogen is particularly mobile in steel. The diusion coecient of steel is between 10ÿ4 and 10ÿ5 cm2/s [13] which is comparable with diusion coecients in ¯uids. Therefore as soon as the hydrogen has penetrated through the phase boundary into the metal, it is capable of penetrating to a depth of 100 mm into the material [14]. Thus a lattice deformation take place and also an accumulation of hydrogen on the boundary surfaces (e.g. at grain boundaries, pores and inclusions). It can be proved [15] that, with the increasing strength of the steel, a notched sample subjected to homogeneous hydrogen produces inhomogeneous distribution of the hydrogen. The hydrogen is concentrated in the range of the maximum hydrostatic stress in the ligament. If at ®rst the material is hydrogen-free, for the same reason, the hydrogen absorbed on the crack edges is distributed selectively into the crack ligament (see Fig. 23). In this case the corrosive hydrogen is concentrated where the most damage exists, which is in cracks, notches or other material defects that cause an increase in stress. Therefore this type of material damage should not be considered as global but as very local hydrogen embrittlement. A special type of this category of damage occurs with welded high tensile ®ne grained steel in fully demineralised or boiler feed water. At temperatures between 50 and 2508C a magnetite layer is created Fig. 23. Hydrogen diusion in a stressed specimen at externally (a) or internally (b) supplied hydrogen. The normal stress syy in the ligament and the crack opening displacement CTOD is shown on top. 448 J. Woodtli, R. Kieselbach / Engineering Failure Analysis 7 (2000) 427±450 (Schikorr reaction) caused by the development of hydrogen. After a long incubation time, cracks can occur in the heat aected hard zones of the weld joint, when they are not subjected to stress relief heat treatment. An example of hydrogen induced stress corrosion cracking is shown in Figs. 24±26. The shaft made of alloyed tempering steel ASM 6418 from a reducing gear system broke after reaching only one third of its expected useful life. It exhibited the typical features of an HCF torsion fatigue fracture. The central question was the early initiation of the cracks. As can be seen from the macrosection (Fig. 24) of the cross-section, the crack formation took place over the complete circumference. The cracks are branched and have a narrow opening. In the microscopic range a thin coating of corrosion products (Fig. 25) and the ®nest crack openings are recognisable in some locations. However the microcracks emanating from corrosion pits (Fig. 26) must be considered as the most important indication of hydrogen-induced stress corrosion cracking. The microstructure consists of ®ne martensite which has been tempered at very low temperatures. The hardness amounts to 490 HV 10, indicating a high tensile steel which is obviously not suitable for the application described. 4. Preventative measures 4.1. Hydrogen embrittlement In order to be able to produce eective measures, both the source of the hydrogen and the embrittlement mechanism must be known at least approximately. As important measures in the prevention of hydrogen damage, the following points can be mentioned: . material Ð selection of suitable material; . medium Ð not so easy to in¯uence, possibly by the addition of inhibitors; . design Ð avoid notches, slots and sharp transitions, avoid local plastic deformation Fig. 24. Cross-section of a shaft: multiple cracks, crack branching. J. Woodtli, R. Kieselbach / Engineering Failure Analysis 7 (2000) 427±450 449 Fig. 25. Branched cracks, partially ®lled with corrosion products. 4.2. Stress corrosion cracking Here the complete system must also be adequately assessed if damage caused by stress corrosion cracking is to be prevented. A pairing of the material medium can be tested in laboratory investigations Fig. 26. Cracks starting from corrosion pits. 450 J. Woodtli, R. Kieselbach / Engineering Failure Analysis 7 (2000) 427±450 only to a certain extent. If stress corrosion cracking is forced during laboratory tests, electrochemical and mechanic fracture measurements also have to be applied. In general, the following points need to be checked: . material Ð suitable choice of material for the medium in question, optimal heat treatment (e.g. with austentite steels, aluminium alloys), prevention of residual stress (tensile); . medium Ð determine the aggressive ions and if possible eliminate; screen using inhibitors, in some cases use cathodic or anodic protection or change the redox potential of the medium; . mechanical stresses Ð detect and optimise (assembling stresses, residual stresses). 5. Conclusions Hydrogen cracking and stress corrosion cracking produce delayed low grade deformation fractures even in high tensile materials although the adjacent material range may not exhibit any embrittlement characteristics. Both types of damage are dependent on many simultaneous parameters and are therefore accompanied by apparent unpredictability. A comprehensive analysis as a starting point for preventive measures is of the utmost importance both for safety and for economical reasons. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] Faria L. Cost of fracture. In: Proceedings of SPT5, Vienna, 1993. Gnirss G. Hydrogen and its eects in welding. TUÈ 1976;17(11):367±77 [in German]. Muster WJ. On hydrogen eects. EMPA Report No. 10848, 1982. p.2 [in German]. Adda I, Philibert J. La diusion dans les solides. Saclay, France: Institut National des Sciences et Techniques nucleÂaires, 1966. Simon H, Thoma M. Applied surface technology for metals. Munich/Vienna: C. Hanser, 1985 [in German]. Heinke G, Wagner GH. Material damage by hydrogen. Mat-Wiss und Werkstotechnik 1996;27:259±66 [in German]. Schmitt G, Savakis S. in: EUROCORR, 1987, p. 493. Nelson GA. Steels for hydrogen service at elevated temperatures and pressures in petroleum re®neries and petrochemical plants. Washington DC: API, 1990 [Publ. 941]. ISO 4042, Annex A: hydrogen embrittlement relief after electroplating. Stellwag B, Kaeschke H. Kinetics of H-induced stress corrosion cracking. Werkstoe und Korrosion 1982;33:274±80 [in German]. Ayeta GB, Pickering HW. J Electrochem Soc 1975;122:1018. Pickering HW, Frankenthal RP. J Electrochem Soc 1972;119:129. Johnson HH. Proceedings of the Conference on Eect of Hydrogen on the Materials Properties and Selection in Structural Design, Seven Springs, 1973. p. 35. Hirt JP, Johnson HH. in: NACE 32, 1, 1976. Liu HW. Trans ASME I, J Basic Eng 1970;92:6. Parkins RN. Environment sensitive fracture and its prevention. Brit Corr J 1979;14:5. Eertz PH, Forchhammer P, Hickling J. VGB Kraftwerkstechnik 1982;62(5):390 [in German]. Logan HJ. J Res Natl Bur Stand 1952;48:99. Speidel MO, Magdowski RM. Proceedings of the Second International Symposium on Environmental Degradation of Materials in Nuclear Power Systems Ð Water Reactors, American Nuclear Society, Monterey CA, USA, 1985. p. 267. Beavers JA. Stress corrosion cracking of copper alloys. In: Jones RH, editor. Stress corrosion cracking. Ohio: ASM Int, 1992. Hickling J. In: Lange G, editor. Systematic analysis of technical failure. DGM, 1983. p. 291. Landgren W, Mattson E. Brit Corros J 1976;11:80.