Crystallographic Polarity of ZnO Crystals A. N. Mariano and R. E. Hanneman Citation: J. Appl. Phys. 34, 384 (1963); doi: 10.1063/1.1702617 View online: http://dx.doi.org/10.1063/1.1702617 View Table of Contents: http://jap.aip.org/resource/1/JAPIAU/v34/i2 Published by the American Institute of Physics. Additional information on J. Appl. Phys. Journal Homepage: http://jap.aip.org/ Journal Information: http://jap.aip.org/about/about_the_journal Top downloads: http://jap.aip.org/features/most_downloaded Information for Authors: http://jap.aip.org/authors Downloaded 05 Oct 2012 to 152.14.136.96. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions JOURNAL OF APPLIED PHYSICS VOLtiME 34. NUMBER i FEBR1JARY 1963 Crystallographic Polarity of ZnO Crystals A. N. MARIANO AND R. E. HANNEMAN Lincoln Laboratory,* Massachusetts Institute of Technology, Lexington 73, Massachusetts (Received 30 July 1962) The crystallographic polarity of the noncentrosymmetric material zincite (ZnO) has been determined by a rapid x-ray absorption edge method. The polarity results have been correlated to marked differences in etching behavior and crystal morphology in opposite polar directions of ZnO crystals. These differences are shown to be consistent with the proposed surface bonding model for AII-Bv! compounds. Crystal morphology is shown to be a useful criterion to indicate crystallographic polarity in well-formed wurtzite-type crystals. I. INTRODUCTION II. X-RAY THEORY AND TECHNIQUE ONSIDERABLE interest in recent years has focused on crystallographic polarity and consequent property differences on opposite polar surfaces of various materials having noncentrosymmetric structures. Particular emphasis has been placed on materials of significance in solid-state electronics of the ArnBv type and, to a more limited degree, the ArrBvI type. In the course of some of this recent work a model was conceived/,2 based on the surface bonding characteristics of the A and B surface atoms, to account for the marked differences in some physical and chemical properties of the opposite polar surfaces of these crystals. It is of considerable interest to investigate the applicability of the proposed surface bonding model to noncentrosymmetric materials which are oxides and those with larger electronegative differences (and greater ionicity) than heretofore examined. The present paper treats this region of interest by studying the crystallographic polarity of ZnO and its relationship to certain property differences on A and B polar surfaces, including etching behavior and crystal morphology. Recently, Cole3 and Stemple3 presented a lucid description of the effect of crystal perfection and polarity on x-ray absorption edges seen in Bragg diffraction by noncentrosymmetric crystals. The present work describes in some detail a modification of that x-ray method for rapid absolute polarity determination, with specific application to wurtzite-type materials. The crystal habit of many well-formed noncentrosymmetric crystals is distinctly different in the opposite polar directions of such materials. This study attempts to show that crystal morphology of euhedral ZnO crystals is consistent with the surface bonding model and that morphology can be used to determine crystallographic polarity in many cases. This criterion should be particularly useful for polarity identification in compounds not amenable to x-ray or piezoelectric studies. The spatial arrangement of atoms in the unit cell of a crystal and the differential scattering power of the individual types of atoms determine the modulus of the geometric scattering factor of such a crystal. James 4 and Cole3 have presented theoretical treatments relating the geometric scattering factor F to the intensity of x rays diffracted from perfect and mosaic crystals. If a crystal scatters as an ideally mosaic crystal the intensity is proportional to 1 F 1 2, whereas, in the case of strong reflections from a perfect crystal the diffracted intensity is proportional to 1F I. The geometric scattering factor of a general (hkl) reflection is given by C F(hkl) = Lj(fo+~1'+i~f")j X exp[27ri(hxj+kYj+lzj)], (1) where 10 is the atomic scattering factor, ~1' and ~1" are anomalous dispersion corrections,4.5 and (XjYjZj) are the spatial coordinates of each of the j atoms in the unit cell. The wurtzite structure contains 4 atoms per unit cell, two of the A species (i.e., Zn) at the positions (000) and (Ht) and the B species (i.e., 0) at (OOl) and (Hi). Substituting these values into Eq. (1) and regrouping: F(hkl) = [(fA'+i~fA")+ (f B'+it.fB")e!!ril] X {1 exp[27ri (ih+~k+tl)], + (2) where 1'= (fo+~f'). Multiplying by the complex conjugate of F(hkl) and simplifying, the following expression is obtained for the modulus of the geometric scattering factor: 1 * Operated with support from the U. S. Army, Navy, and Air Force. ., 1 H. C. Gatos and M. C. Lavine, J. Electrochem. Soc. 107, 427 (1960). 2 E. P. Warekois, M. C. Lavine, A. N. Mariano, and H. C. Gatos, J. App!. Phys. 33, 690 (1962). 8 H. Cole and N. R. Stemple, J. App!. Phys. 33, 2227 (1962). FF*(hkl) 1 = {2+2 cos27r(ih+~k+tl)} X {[jA' + fB' cOS!7rl- t.fB" sih!7rl]2 +[t.fA"+~fB" cos!7rl+ fB' sinhl]2}. (3) This equation is a general expression for any (hkl) reflection in any material with a wurtzite structure. For the determination of absolute configuration, or polarity, in noncentrosymmetric crystals in the wurtzite class, such as ZnO, the {OO ·l} family of reflections must be considered, as they are normal to the polar axis. Referring to Eq. (3), if lis odd, FF*{OO·I} is equal 4 R. W. James, Optical Principles of the Diffraction of X-rays (G. Bell and Sons, Ltd., London, 1958). 6 C. H. Dauben and D. H. Templeton, Acta Cryst. 8, 841 (1955). 384 Downloaded 05 Oct 2012 to 152.14.136.96. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions CRYSTALLOGRAPHIC to zero; if 1= 4n, where n is zero or an integer, F F* {OO ,1) is equal to FF*(oo·l). However, for ±1=4n-2(i.e., 002, or 002), FF*(OOl);;Z!FF*(OOl), and therefore the diffracted intensity of these reflections should be different on the A and B sides of the wurtzite crystal. (The A and B sides are defined as those opposite surfaces, which are parallel to {OO ·l} planes and they terminate with only A and B atoms, respectively.2) Using Eq. (3) this intensity ratio, which is used to identify ZnO crystal polarity in this paper, is given by: 1(00·2) FF*(OO·2) 1(00·2) FF*(00·2) 385 POLARITY OF ZnO CRYSTALS A- 80 o Zn SURFACE (00'2) SURFACE (00·2 ) >- !:: 60 II) z UJ >.., • 1.263A >"K" >"2" 1.303 A 1.283A ~ ~ 40 20 IBACKGROUND\ {[jA'+LlfB"]2+[LlfA"- fs']2} O~__~~~~~~__~~~~~~~b~l~'_~~__~ (4) 28 {[fA' -Llfs"]2+[LlfA"+ fB']2} ' where I (00·2)/1 (00,2) is the calculated intensity ratio for ideally mosaic crystals. (The ZnO crystals used were found to behave as almost ideally mosaic crystals from intensity measurements of several {hkl} planes.) Since the dispersion corrections, in effect, represent a differential absorption of incident radiation by the scattering atoms, the maximum effect will be noted at the short wavelength side of the absorption edge of one of the elements. The anomalous dispersion effect becomes quite small immediately adjacent to the long wavelength side of the edge. All other variables cancel in the calculated intensity ratios except FF* (00 ·1)/ FF*(OO·l). Thus a comparison of experimental and calculated l(OO·I)/l(Oo·i) near both sides of the absorption edge provides a highly sensitive and convenient method for unambiguous determination of crystallographic polarity. There are no higher-order {OO ·/} reflections superimposed on the {OO· 2} reflections near the corresponding K edge of interest because the x-ray tube is operated at sufficiently low voltage that: I\(Kedge»I\SWL>tl\(Kedge). I\SWL is the short wavelength limit of the continuous spectrum source corresponding to the operating voltage. The intensity of (00·2) and (00·2) reflections from ZnO are obtained as a function of A from the continuous spectrum of a copper tube in a Norelco x-ray unit. The sample is placed on the diffractometer specimen holder with {00·2} planes satisfying the Bragg relation; a Geiger counter is used to record intensity as a function of 28 (and hence A). Measurements are arbitrarily made at a convenient position very close to either side of the absorption edge of interest. In this case measurements and calculation of intensity were made at ±0.02 A from the Zn Kedge (i.e., approximately ±O.So 28 on either side of Zn Kedge). Background intensity is subtracted before calculating the experimental 1(00·2)/1(00·2) ratio. 29 28 29 28·FIG. 1. Continuous spectrum diffraction contours across the zinc K-absorptiop edge wavelength, from ZnO single crystal with (00·2) and (00·2) planes satisfying the Bragg condition. III. RESULTS AND DISCUSSION A. X-Ray Diffraction The diffraction contours from the continuous spectrum produced by the (00· 2) and (00· 2) planes of ZnO are shown in Fig. 1 with their differential absorption edge effect due to anomalous dispersion. The background x-ray level is indicated below the scans of both sides in Fig. 1. Referring to Table I, the calculated intensity ratio [FF*(00·2)/FF*(00·2)]c~lc at A=1.303 A is very close to unity (i.e., 1.01 to 1). Experimentally this condition is effected by slight adjustment to the tube current before the x-ray scan of either side of the crystal such that the [I(00·2)/1(00·2)]exptl at A=1.303 A is equal to unity as seen in Fig. 1. The 1(00·2)/1(00·2) at A= 1.263 A, however, is then observed to be appreciably less than unity as predicted by theory and consistent with [FF*(00·2)/FF*(00·2nalc in Table I. The deviations of calculated and observed differences of 1(00·2)/1(00·2) at A= 1.263 A are due to: (1) the fact that the crystals are really not ideally mosaic, and thus the intensity does not exactly have /F /2 dependence for strong reflections, (2) approximately a one percent error inherent in the calculated ratio by not including the Debye-Waller temperature factor in the atomic scattering factors, and (3) the degree of crystal perfection TABLE r. Calculated and observed values of 1(00·2)/1(00·2) for ZnO on either side of Zn Kedge. ExperiCalculated mental FF*(00'2) 1(00·2) X(A) 1.263 1.303 10(Zn) 10(0) 22.8 22.8 5.75 5.75 !llzn' !llo' -4.06 -4.37 +0.04 +0.04 ---- --- !llzn" tl/o" FF*(00·2) 1(00·2) 3.41 0 0.03 0.03 0.82 1.01 Downloaded 05 Oct 2012 to 152.14.136.96. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions 0.7 1.0 386 A. N. MARIANO AND which can vary from Zn to 0 surfaces. Nevertheless the agreement between calculated and experimental values is good and the polarity determination is unambiguous with Zn occupying the (00·1) surface, which had the lower intensity in Fig. 1 for t.. < 1.283 A, and the R. E. HANNEMAN t Zo<OO'l> FIG. 3. Photomicrograph (16X) of euhedral ZnO crystal etched with a 6:6: 1 solution (by volume) of fuming HNO., glacial acetic add, and H 20. Distinctly polar etch figures are evident on the (lOiO) prism faces. o atoms occupying the (00· i) surface. The relationship Ca) Zn(OOol) of these x-ray results is correlated to crystal morphology and etching behavior in the following section. (The experimental values shown in Fig. 1 and listed in Table I are for the natural ZnO used, but the other grades of zincite crystals used in the experiment gave almost identical results.) In the course of this work new accurate lattice constants were determined by powder methods for pure ZnO. These are: a= 3. 2499± 0.0001 A and c= S.206±O.001 A. B. Etching Behavior and Crystal Morphology The A and B surfaces of ZnO crystals show marked differences in their response to chemical etching. This is in agreement with earlier studies of other materials with the wurtzite structure.2 ,6,7 Specific etchants used for revealing differences in chemical reactivity on polar surfaces are listed in Figs. 2 and 3, and their effects are illustrated therein. From the etching experiments it was found that the (0001) surface (i.e., oxygen surface as identified by x rays) etched more rapidly in the oxidizing etchants used than did the Zn surface. In impure crystals etching action on the A (or Zn) face is primarily located where dislocations intersect that surface [see Fig. 2(a)]. This preferential etching effect is particularly noticeable in such ZnO crystals.8 The rapidly attacked B surface (oxygen surface) in these crystals, (b) O(oo.i) FIG. 2. Photomicrographs of ZnO 100· t I showing Zn and a surfaces after etching 90 sec in 20% HNO a solution (180X). (\ D. C. Reynolds and S. J. Czyzak, J. Appl. Phys. 31, 94 (1960). 7 R. Zare, W. R. Cook, Jr., and L. R. Shiozawa, Nature 189, 217 (1961). S Crystals used in these studies included: (1) laboratory pure vapor grown crystals and (2) cinderbank crystals from furnace residues; both from the New Jersey Zinc Company. These crystals had sufficiently well-developed external forms to correlate crystal morphology with x-ray polarity results. Other crystals used include ZnO grown in a PbF2 flux and the naturally occurring mineral zincite from Sterling Hill, Franklin, New Jersey. Downloaded 05 Oct 2012 to 152.14.136.96. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions CRYSTALLOGRAPHIC POLARITY OF ZnO CRYSTALS however, is etched with more uniformly distributed etch figures (hillocks). Since the formation of intermediate surface products does not appear to interfere with etching these factors are consistent with a surface bonding model proposed by Gatos and Lavine for III-V compoundsl and modified for II-VI compounds by Gatos et al. 2 As shown in Fig. 4 which illustrates this model, the surface atoms are triply bonded, and the bulk atoms are tetrahedrally bonded. The A atoms have no electron dangling bonds at the surface whereas the B atoms have a dangling bond with two electrons at the surface. As a result the A atom bonds are more severely distorted from tetrahedral Sp3 symmetry than the B atom bonds. Of the three bonds joining each surface atom to the crystal, two are considered largely covalent, and the third bond is considered coordinate, with resonance between each bond. Since the Zn and 0 atoms show a significant difference in electronegativity, an appreciable part of the so-called covalent bonds possess a partial ionic character which indicates that the zinc atoms tend to transfer some of their electrons to the oxygen atoms. In this case the layer of zinc surface atoms tends to have a positive charge and the layer of oxygen surface atoms a negative charge, in addition to the two dangling electrons per oxygen atom. A net surface dipole moment prevails which has recently been discussed by Gatos.9 The dangling electrons on the B surface account for the high etching rate on the B surface because of their susceptibility to reaction with electron-seeking agents in the etchant. The relatively uniform attack observed on the oxygen surface in Fig. 2 (b) is also consistent with the idea that dangling electrons largely control the surface etching action in certain kinds of etchants where the formation of intermediate products is not controlling. The slower over-all etching rate of the A surface is in accord with the absence of any dangling electrons on the A surface. Dislocation etch pits probably occur at t<OOO/) BULK "A" SURFACE o A A f \2100 B ,,'0..--' ,, ,, ,, : 0: '(.\ /\::.;\ ',' ~ 0i A A l 1 BULK BULK o "B" SURFACE 1<tIoo i> FIG. 4. Scehmatic of surface bonding mode of Au - BVI wurtzite compounds. 9 H. C. Gatos, J. App!. Phys. 32, 1232 (1961). 387 both surfaces due to etchant attack along the dislocations, but they are readily apparent only on the A (Zn) surface where their effect is not masked by the rapid etching action which prevails everywhere on the B surface. Etching behavior on the (lOiO) natural prism planes, shown in Fig. 3, also provides a means of unambiguous identification of polarity in ZnO. The apex of the large etch pits (which are depressed from the surface level) point toward the Zn surface. Conversely the small etch hillocks (which are elevated above the prism plane level) point toward the 0 surface. These effects are observed in both pure and impure ZnO. Another physical property, abrasion resistance, was found to be different on the Zn and 0 surfaces. While polishing with 600 grit garnet abrasive it was observed that one of the basal surfaces liberated a yellow ZnO residue much more rapidly than the other side. The side which abraded most rapidly was identified by x-ray results to be the (B) oxygen surface. This is in agreement with earlier findings on III-V compounds. lO Noncentrosymmetry in the polar direction is immediately revealed by the crystal morphology of ZnO, as is illustrated in Fig. 3. The crystals are characterized by a flat base at one end and by steep pyramidal forms at the opposite end. X-ray absorption edge data established that the large flat base is the B surface, or oxygen layer, and that the steeply terminated end is the A surface, or zinc layer. This crystal morphology can be explained with the bonding model described above. According to that model the A surface atom bonds are more severely strained than the B surface atom bondsl l and thus the effective surface energy should be higher on the Zn surface. Since a fast growing plane generally tends to disappear (leaving behind the slower growing forms with lower surface energy),12 the steeply inclined pyramidal faces should form on the fast growing, or Zn, end of euhedral (well-formed) ZnO crystals. There are a number of wurtzite-type Au-Bv! compounds which occur as natural or synthetic euhedral crystals with analogous crystal morphology to ZnO.13 In some of these materials it would be very difficult to determine crystallographic polarity by any x-ray method. In such an instance, Newkirkl4 and coworkers have identified the polarity of BeO by piezoelectric methods and found that the differences in crystal morphology and etching behavior are in agreement 10 E. P. Warekois, M. C. Lavine, and H. C. Gatos, J. App!. Phys. 31, 1302 (1960). 11 Recent experimental evidence on spontaneous bending of thin wafers of the Am-Bv material, InSb, supports this theory. R. E. Hanneman, M. C. Finn, and H. C. Gatos, J. Phys. Chern. Solids 23, 1553 (1926.) 12 F. C. Frank in Growth and Perfection of Crystals, edited by R. H. Doremus, B. W. Roberts, and D. Turnbull (John Wiley & Sons, Inc., New York, 1958), p. 4. 13 C. Palache, H. Berman, and C. Fronde1, The System of Mineralogy (Dana) (John Wiley & Sons, Inc., New York, 1955), 7th ed., Vo!' I. 14 H. Newkirk, D. Smith, and J. Kahn (private communication). Downloaded 05 Oct 2012 to 152.14.136.96. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions 388 A. N. MARIANO AND with the ZnO results reported above. In similar cases, especially if unambiguous piezoelectric polarity data are not attainable, crystal morphology may be especially useful as a rapid indication of the crystal polarity. R. E. HANNEMAN posite polar directions are entirely consistent with this model. (3) Crystal morphology can serve as a means of identification of polarity in well-formed crystals of any wurtzite-type materials. IV. CONCLUSIONS (1) The absolute crystallographic polarity of ZnO has been determined by a useful, rapid x-ray method and the results have been correlated to etching figures and crystal morphology. (2) The surface bonding model for An-Bv! compounds has been extended to a noncentrosymmetric material of the oxide type and with a larger electronegativity than previously investigated. Etching behavior and crystal morphology of ZnO crystals in op- JOURNAL OF APPLIED PHYSICS ACKNOWLEDGMENTS Appreciation is extended to T. B. Reed of Lincoln Laboratory and C. R. Bieling of New Jersey Zinc Company for supplying the ZnO crystals used in this work. The authors are grateful to E. P. Warekois and H. C. Gatos for valuable discussions. They wish particularly to thank H. Cole and H. Newkirk for furnishing results regarding the x-ray method and the BeO polarity, respectively, prior to publication. VOLUME 34. NUMBER 2 FEBRUARY 1963 Growth of Ferrous-Free Cobalt Ferrite Single Crystals A. FERRETTI, W. KUNNMANN, AND A. WOLD Lincoln Laboratory,* Massachusetts Institute of Technology, Lexington 73, Massachusetts (Received 21 May 1962) Single crystals of cobalt ferrite were grown from a melt having the composition 95% CoFe 20. - 5% NaFe02 by weight, under an oxygen pressure of 1650 psig and a temperature of 1590°C. Chemical analysis indicated that the crystals contained no Fe2+ and 3 parts per 10 000 sodium. INTRODUCTION been difficult to grow pure ferrite crystals from I Tthehasmelt because they dissociate at high temperatures with the loss of oxygen. Pure magnetite single crystals have been grown by a Bridgman method! and the ferrite region of the phase diagram Fe-Co-O has also been investigated. 2 Cobalt ferrite single crystals, which contained 1.3% Fe2+, were grown from melts by Ferretti, Arnott, Delaney, and Wold. a Ferrous-free cobalt ferrite single crystals have been grown from sodium ferrite flux by Kunnmann, Banks, and Wold. 4 It was believed that pure cobalt ferrite crystals could , also be grown from melts if sufficient oxygen pressure (ca. 1600 psig) was maintained to prevent decomposition of the ferrite. However, under these conditions stoichiometric cobalt ferrite melted at a much higher temperature than reported previously and platinum crucibles could no longer be used. Both iridium and platinum-rhodium crucibles were used and found to be unsatisfactory. The lack of suitable crucible materials * Operated with support from the U. S. Army, Navy, and Air Force. 1 J. Smiltens, J. Chern. Phys. 20, 990 (1952). 2 J. Smiitens, J. Am. Chern. Soc. 79, 4881 (1957). 3 A. Ferretti, R. J. Arnott, E. Delaney, and A. Wold, J. App!. Phys. 32, 905 (1961). • W. Kunnmann, E. Banks, and A. Wold, J. App!. Phys. 33, 1364 (1962). necessitated the development of a modified Bridgman technique in which crystals were grown from a melt, having the composition 95% CoFe 204-5% NaFe02 by weight, under an oxygen pressure of 1650 psig and a temperature of 1590°C. EXPERIMENTAL Preparation of Starting Materials The cobalt ferrite used in these investigations was prepared by the thermal decomposition of a cobalt pyridinate "precursor" COaFe6(CHaC02)170aOH·12C5H5N.5 The ferrite obtained was free from any impurities which would be introduced by ball-milling or other ceramic technique. A ratio of Fe/Co of 1.99±0.01 was computed from values obtained for total iron and cobalt. Cobalt ferrite prepared by this procedure did not melt at 1600°C and 1650 psig oxygen pressure, whereas the ferrite obtained by ball-milling and firing cobalt carbonate and iron oxide mixtures did melt under these conditions. The sodium ferrite was formed by the fusion of sodium carbonate with an equimolar mixture of ferric oxide. X-ray examination of the product showed the presence of several weak lines which indexed as sodium ferrate, 6 D. G. Wickham, E. R. Whipple, and E. G. Larson, Nuel. Chern. 14, 217 (1960). Downloaded 05 Oct 2012 to 152.14.136.96. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions J. Inorg.