Acta metall. Vol. 34, No. 8, pp. 1525-1531, Printed in Great Britain
$3.00 + 0.00 Journals Ltd
IDENTIFICATION OF FINE PRECIPITATE PARTICLES IN UNIDIRECTIONALLY SOLIDIFIED Ni-Al-MO EUTECTIC ALLOYS BY MEANS OF EDX AND SAD ANALYSES Z. HORITA, T. SAN0 and M. NEMOTO Department of Metallurgy, Faculty of Engineering, Kyushu University, Fukuoka 812, Japan (Received 10 September
1985; in revised form 12 December
Abstract-An attempt is made to identify the fine precipitate particles formed in the ~-MO fibers in the unidirectionally solidified Ni-Al-MO eutectic alloys. The energy dispersive X-ray (EDX) microanalysis and the selected area diffraction (SAD) analysis are used for this identification. Since all particles in the fibers are smaller than the electron probe size, the method proposed by Cliff et al. [Developments in Electron Microscopy and Analysis, p. 63. Inst. Phys., London (1983)] is adopted in the EDX analysis in order to exclude effects from the surrounding a-MO phase. In the SAD analysis, the diffraction patterns are taken at the particles in the transverse (OOl), and the longitudinal (1 lo), sections. These analyses show that the fine particles in the a-MO fibers are the y’ phase having the fixed composition identical to the y’ matrix around the cc-M0 fibers, the crystal structure of Ll, type and the N-W orientation relationship with the ~-MO phase. With reference to the recent phase diagram of Miracle et al. [MetaN. Trans. HA, 481 (1984)], this y’ formation in the a-MO fibers is understood as the stable phase formation and hence precludes any possibility of carbides, metastable phases or the detrimental b-NiMo phase. Discussion is extended for observing well-defined diffraction patterns in examining the orientation relationship between the f.c.c. y’ phase and the b.c.c. a phase. RCsunG-Nous nous sommes efforces d’identifier les fines particules de precipite qui se forment dans les fibres de MO-a des alliages eutectiques orient&s Ni-Al-MO. Pour cette identification, nous avons utilise la microanalyse de rayons X par dispersion d’energie (EDX) et la diffraction d’aire selectionnee (DAS). Toutes les particules dans les fibres &ant plus petites que la taille de la sonde Clectronique, nous avons adopt& la mtthode proposee en analyse X par Cliff et al. [Developments in Electron Microscopy and Analysis, p. 63, Inst. Phys., London (1983)] afin d’tliminer les effets provenant de la phase MO-~ environnante. Au tours de l’analyse par DA’S, les diagrammes de diffraction etaient enregistres pour des particules de section transverse (OOl), et longitudinale (1 lo),. Ces analyses ont montre que les fines particules dans les fibres de MO-a appartenaient a la phase y’, avaient la meme composition que la matrice y’ autour des fibres de Mo-a et presentaient la structure de type Ll, et la relation d’orientation de N-W avec la phase MO-a. Compte tenu du diagramme de phases recent de Miracle et al. [MetaN. Trans. 15A, 481 (1984)], nous pensons que la phase y’ qui se forme dans les fibres de Mo-a est la phase stable, ce qui permet d’ecarter les carbures, les phases metastables ou la phase nuisible NiMo-6. La discussion est elargie a l’observation de diagrammes de diffraction bien d&finis lorsqu’on etudie la relation d’orientation entre la phase c.f.c. y’ et la phase cc. a. Zusammenfaasung-Es wird versucht, die feinen Ausscheidungen in den a-Mo-F%den in der gerichtet erstarrten eutektischen Ni-Al-Mo-Legierung mittels energiedispersiver Rdntgenanalyse und Feinbereichsbeugung zu identifizieren. Da simtliche Ausscheidungen in den Faden kleiner als der Elektronenstrahl-durchmesser sind, wird die von Cliff et al. [Developments in Electron Microscopy and Analysis, p. 63. Inst. Phys., London (1983)] vorgeschlagene Methode der energiedispersiven Analyse angewendet; damit werden Einfliisse durch die benachbarte a-MO-Phase ausgeschaltet. Bei der Feinbereichsbeugung werden die Beugungsdiagramme der Ausscheidungen in transversalen (OOl),- und longitudinalen (IlO),-Schnitten aufgenommen. Aus der Analyse geht hervor, dat3 die feinen Ausscheidungen in den a-Mo-Flden die y’-Phase darstellen, die eine Zusammensetzung wie die der y’-Matrix urn die Flden herum aufweisen. Sie haben such die Kristallstruktur Ll ,-Typ und die N-W-Beziehung mit der a-MO-Phase. Nach dem neuen Phasendiagramm von Miracle et al. [Mefall. Trans. 15A, 481 (1984)] kann diese Bildung von y’ in den a-Mo-Flden als die Bildung einer stabilen Phase angesehen werden; daher ist jede Mijglichkeit eines Karbides, von metastabilen Phasen oder der schidlichen g-Phase ausgeschlossen. Die Diskussion wird auf die beobachteten, wohldefinierten Beugungsdiagramme im Zussammenhang mit der Orientierungsbeziehung zwischen der k.f.z. y’-Physe und der k.r.z. a-Phase ausgedehnt.
1. INTRODUCTION The in situ composites of Ni-Al-MO eutectic alloys consist of ~-MO fibers with rectangular cross section in a y’ or y + y’ matrix, where y’ is the Ll, type
ordered phase and y is the Ni rich solid solution phase. This type of composite has received considerable interest as materials for high temperature uses, because the composites are produced during 1525
HORITA et al.: PRECIPITATE PARTICLES IN Ni-Al-MO EUTECTIC ALLOYS
solidification and thus they have excellent morphological stabilities up to near their melting temperatures. It was reported that various types of metastable phases formed in the y matrix [l, 21 and that 6-NiMo phases at the cr-MO fibers [3-51 when the composites were subjected to certain quenching and aging treatments. Particularly, the 6 phase formation changes the morphology of the aligned U-MO fibers, causing a marked degradation of the creep properties . Therefore, it is important to perform detailed microstructural analyses for these composites. Close observations of the composite structure by transmission electron microscopy showed that fine particles were formed in the ~-MO fibers during the process of unidirectional solidification . Sriramamurty et al.  suggested that these fine particles might be composed of carbides. However, because the particles are so small, neither compositional analysis nor electron diffraction analysis has succeeded yet to give any detail about the particles. The present investigation is thus initiated in an effort to clarify these fine particles by conducting quantitative energy dispersive X-ray (EDX) microanalyses in an analytical electron microscope and selected area diffraction (SAD) analyses in a conventional transmission electron microscope. 2. EXPERIMENTAL
An alloy with a nominal composition of 27.0 wt% MO 8.0 wt% Al and 65.0 wt% Ni, (657at.%Ni, 17.6at.% Al and 16.7at.% MO) was prepared from hydrogen treated MO powder briquetts (99.95%), high purity Al ingots (99.99%) and high purity Ni ingots (>99.9%). This composition was selected such that the alloy exhibited two phases of CIand y’ . The alloy was first melted in a vacum induction furnace. Under exposure to an argon atmosphere, rods in - 200 mm length and - 5 mm diameter were obtained from the melt. The rods were then placed in a recrystallized alumina crucible and subjected to unidirectional solidification in a modified Bridgman furnace under a pressure of - lo-’ Pa at a lowering speed of -2.8 x 10-6ms-1. The melt temperature was controlled to -1773 K, having the thermal gradient of - lo4 Km-‘. A cylindrical piece in - 5 mm length and - 5 mm diameter was cut from the unidirectionally solidified rod and encapsulated in an evacuated quartz tube. This was homogenized at 1568 K for 3.6 ks (1 h) followed by quenching to iced brine. Then, for the purpose of particle size enhancement, an aging treatment was undertaken at 1173 K for 360 ks (100 h). This particular aging treatment was employed because a preliminary experiment showed that this temperature gave the largest particle size among the others in a range from 1173 to 1568 K where all specimens were kept for 1 h.
The sample, after the aging treatment, was cut both perpendicular and parallel to the fiber direction and subsequently made to -3 mm diameter discs with a thickness of -0.15 mm. These discs were then immersed in a solution of H,SO, (10%) plus C,H,OH (90%) and perforated using a modified twin-jet method . Microstructural observations and SAD analyses were conducted in a JEM-200B transmission electron microscope at an accelerating voltage of 200 kV. EDX microanalyses were performed using a JEM2000FX analytical electron microscope fitted with a Tracer Northern energy dispersive X-ray Si(Li) detector in HVEM Laboratory of Kyushu University. The JEM-2000FX was operated in the TEM mode at an accelerating voltage of 120 kV and emission currents of 1&20pA. The EDX microanalysis used a focused beam diameter of approx. 30 nm. The acquisition of emitted X-rays was continued for 200 s live time. Integration of the characteristic X-ray peaks, subtraction from background and calculation of the net intensities were achieved by a MTF program incorporated in a Tracer Northern 2000 multichannel analyzer. The characteristic X-ray Kcr intensities ZNi,ZAland Zr.,,,measured simultaneously for the elements of Ni, Al and MO, respectively, were converted to the corresponding weight fractions CNi, C,, and C,, through the following equations [lo, 111 CA,lC,i = k,,,i (ZAiIZNi)
CM,lC,i = k,,i-+ (ZMo/ZNi )
where the factors required for the conversion, kAINi and kMvloNi,were taken from the experimental values obtained by means of the extrapolation method shown by the present authors ; (k,,,i = 0.73 and kMoNi= 2.45). The absorption correction was made by the equation proposed by Goldstein et al. . Because the mass absorption coefficient of each characteristic X-ray in the sample and the sample density in the equation are functions of concentration, calculations were iterated until the concentrations converged to a certain value . The film thickness required for the absorption correction was measured by using the contamination spot separation method proposed by Lorimer et al. [ 151 and Knox [ 161. 3. RESULTS
Figure 1 shows an electron micrograph of the transverse section to the U-MO fiber direction. A magnification of an cr-MO fiber and its vicinity is shown in Fig. 2(a). Fine precipitate particles are clearly visible in the fibers of rectangular shape and they have a similar form of distribution for most of the fibers. A SAD pattern is shown in Fig. 2(b) taken
Fig. I. An electron micrograph of the transverse (001), section, showing fine precipitated particles formed in the rectangular shape ~-MO fibers and ~-MO precipitated particles formed in the y’ matrix. The former particles are distributed in a similar form for most of the fibers and the latter are mostly around the a-MO fibers.
from the area with -250 nm diameter which includes the needle-like particle marked A in Fig. 2(a). The pattern exhibits fairly intense streaks which seem to have intensity maxima or to consist of linear arrangements of fine spots. The details of the SAD analysis will be given later. Figures 1 and 2(a) show, furthermore, that there are many precipitate particles formed in the y’ matrix. They are likely to be found around the cr-MO fibers. The observation of these particles is consistent with earlier studies of the composites [2, 171and they are reported to be a MO-rich t[ phase [2, 171.
Fig. 2. (a) A magnification verse (OOl), section. (b)
of an ~-MO fiber in the transA SAD pattern taken at the
needle-like particle marked A in (a), consisting of diffraction spots from the surrounding a-MO phase and intense streaks having intensity maxima or linear arrangements of fine spots.
3.2. EDX microanalysis Since the diameters of spherical and needle-like particles in the U-MO fibers are less than the size of the electron probe, the compositions obtained through equations (l)-(3) are expected to have some contribution from the surrounding U-MO phase. In order to accomplish quantification of such small particles, a simple method proposed by Cliff et al. [ 181 is adopted in the present analysis. It was shown by Cliff et al. [ 18) and Lorimer et al. [ 191that, when both the particle and matrix are activated simultaneously by a primary electron beam, the measured fractions (e.g. G, C,,, and C,, in the present case) should lie on straight lines which extend from the matrix compositions toward those of the particles. Figure 3 shows CNi plotted against C,, and C,, in weight fraction. The results corrected for absorption are also included in Fig. 3. Furthermore, distinction has been made among the datum points obtained by positioning the focused beam on the particles, the E-MO phase without particles and the y’ phase encircling the E-MO phase. A typical X-ray spectrum
Al ond MO (wt %) Fig. 3. A relationship of the Ni fraction plotted against those of Al and MO in wt%. Distinction is made among the datum points obtained from the particle plus the surrounding ~-MO phase, the U-MO phase only and the y’ matrix encircling the M-MO fiber. Indicated also are those corrected and uncorrected for X-ray absorption.
HORITA et al.: PRECIPITATE
X - Ray
Fig. 4. A typical X-ray spectrum obtained from a particle in the U-MOfiber, indicating some effect from the surrounding ~-MO phase. The present analyses use the integrated intensity of K, line for each element. obtained from a particle is shown in Fig. 4, indicating that some contribution arises from the cr-MO phase surrounding the particle. It is seen that the datum points of the particles fall well on the straight lines, emanating from the composition of the cr-MO phase without particles. It is seen also that the extrapolation of the straight lines intersects well the datum points obtained from the y’ phase encircling the ~-MO fibers. From these results, it is suggested that the fine particles formed in the U-MO fibers may have the composition identical to the y’ phase. At least it is said that the fractional ratio of Ni to Al in the particles is the same as that in the y ’ phase. Cliff et al.  and Lorimer et al. [ 191 pointed out that, when small particles had variable compositions with respect to the size, the datum points did not lie on straight lines but they fell in a certain area determined by the degree of compositional variation. Despite the fact that the present analysis has covered a wide range of particle sizes, all datum points are well on the straight lines as shown in Fig. 3. This result is not altered even after correction for absorption. It is then concluded that the fine particles in the U-MO fibers have a fixed composition: C,i = 72.3 (69.4), CAl=9.6 (19.9) and C,,= 18.1 (10.7) in weight (atomic) percent. 3.3. SAD analysis From the SAD pattern shown in Fig. 2(b), it is difficult to obtain structural information of the particles because of the streak-like arrangement of fine spots. If the particles are the same as the y’ phase as suggested from the EDX microanalysis, it is necessary to consider the importance of the orientation relationship between f.c.c. y’ particle and b.c.c. cc-M0 phase for observing clearly defined diffraction patterns.
For the f.c.c./b.c.c. orientation relationships, commonly observed are Bain , Pitsch , Nishiyama-Wasserman (N-W) [22,23] and Kurdjumov-Sacks (K-S)  relationships. For any of these relationships, observation in the direction parallel to [ 110]b,cc, gives well-defined diffraction spots of both f.c.c. and b.c.c. crystals, but that in the direction parallel to [OOl],,,,,, is limited for the Bain and the N-W relationships. Since the transverse section shown in Fig. 2(a) is (OOl),, the longitudinal (1 lo), section cut in parallel to the c(-MO fiber turns out to be more appropriate for the SAD analysis of the particles. Figure 5(a) shows a micrograph of such longitudinal section and Fig. S(b) the SAD pattern from the area with -250 nm diameter which includes the needle-like particle marked B. The illustration of the pattern is given in Fig. 5(c) where the spots due to multiple diffractions are not depicted for simplicity. It is found clearly that the particle has the Ll, ordered structure and the N-W orientation relationship with the surrounding ~-MO phase given as (11 l),,ll(l lO), ]01~1,4w11, [2 1f],. II[1TO],]]long axis of needle. It is considered that there was no contribution from the y ’ matrix around the U-MO fiber to the SAD pattern shown in Fig. 5(b) and that in Fig. 2(b) as well. This is because the orientation relationship between the U-MO phase and the surrounding y’ matrix is known to be the Bain relationship which yields the diffraction patterns different from Figs 2(b) and S(b) . These results obtained from the SAD analysis are well consistent with those from the EDX analysis. It is thus concluded in a fairly confirmative form that
the fine particles phase having the the structure of relationship with
formed in the ~-MO fibers are the y’ composition same as the y’ matrix, Ll, type and the N-W orientation the @-MO phase. 4. DISCUSSION
Fig. 5. (a) An electron micrograph of the longitudinal (1 lo), section. A comparison with Fig. 2(a) outlines the particle morphology such that some are needle-like aligned in the [l IO], direction and others may be spherical or ellipsoidal. (b) A SAD pattern taken at thk needle-like particle marked B in Fig. 2(a), consisting of diffraction spots from the surrounding ~-MO phase and those from the needle-like particle. (c) Illustration of the pattern given in (b), showing that the y’ particle has the Ll, ordered structure and the N-W orientation relationship with the surrounding a-MO phase. This illustration does not include the spots due to the multiple
between the y’ particle phase.
and the CC-MO
The analytical method proposed by Cliff et al. [ 181 is not generally applicable for the compositional analysis of fine particles having the smaller sizes than the electron probe size: firstly the method does not specify the particle composition and secondly it is invalid for any binary alloy. Apart from the latter limitation, however, the former can be improved with the aid of electron diffraction analysis. A successful identification of fine particles can also be achieved with the presence of a known phase in the same alloy. The present study demonstrated that the straight lines intersected well the composition of the y’ phase encircling the ~-MO phase so that the fine particles were suggested to be the same as y’ phase. The possibility of 6-NiMo phase for the fine particles in the cc-M0 fiber can be excluded because there exists one datum point of the total Ni fraction greater than 38 wt% (50 at.%), the presence of which is not expected from the 6-NiMo phase. The SAD analysis supports this conclusion since the 6-NiMo phase is known to take a complex ordered orthorhombic pseudo-tetragonal structure  which produces the diffraction patterns quite different from those of y’ having the Ll, ordered structure. In fact, the diffraction studies by Nemoto et al.  and Yoshizawa et al.  showed this difference in diffraction pattern. An inspection of Figs 2(a) and 5(a) shows that not only needle-like particles but also spherical or ellipsoidal particles are present in the cc-M0 fiber. It is deduced that these needle-like particles lie in the direction parallel to the [ 1 lo], Although both needlelike and spherical particles have been examined in the EDX analyses, the SAD analyses have been conducted only for the former ones. The spherical or ellipsoidal particles are too small to give visible spots in the SAD patterns. Thus, the composition of these particles are said to be the same as that of the needle-like ones but uncertainty may remain for the orientation relationship between the spherical particles and the surrounding cc-M0 phase. It is necessary to discuss the reasons for the presence of the streak-like arrangements of fine spots in the SAD pattern taken from the transverse (OOl), section and the absence of them taken from the longitudinal (1 lo), section. Usually, the streaks can be accounted for by considering the morphology of particles. However, the particle morphology is less responsible in the present case because the particles examined for the SAD analysis are both needle-like. An alternative reason is suggested from the peculiarity of the orientation relationship between the y’ particle and the cc-M0 phase. Then, based upon the
HORITA ef al.:
’ tundomental spots superlattice *p0h . ~-MO 8~01s
Fig. 6. Illustration of the SAD pattern from [OOI],,constructed on the basis of the N-W orientation relationship between the y’ particle and the ~-MO phase. The streaklike arrangements of fine spots shown in Fig. 2(b) can be explained by the multiple diffractions between they’ particle and the ~-MO phase, though these spots are not depicted in the illustration. Note that the superlattice spots are so weak in the original negative that they cannot be seen in Fig. 2(b).
N-W relationship obtained, a possibility is examined for the presence of the streak-like arrangements of fine spots when the SAD pattern is taken from the transverse (OOl), section. The diagram of the pattern from [OOI],is illustrated in Fig. 6. Depicted there are only fundamental and superlattice spots but those due to multiple diffractions are not shown. It is noted that the superlattice spots from the 7’ particles are so weak in the original negative that they cannot be printed out visibly on the SAD pattern of Fig. 2(b). From Fig. 6 it is possible to explain the streak-like arrangements of fine spots as a consequence of the multiple diffraction between the y’ particle and the Z-MO phase. It turns out then that the SAD pattern taken from the [00 11,is in practice inappropriate for finding the N-W orientation relationship between the y’ particle and the a-MO phase although it should be possible in principle to analyze the relationship. It is interesting to note that the y’ matrix encircling the a-MO fiber has the Bain relationship with the Z-MO fiber  and the Pitsch and the N-W relationship with the Z-MO particles precipitated in the 7’ matrix [2,7]; whereas as observed in the present study the relationship between the 7’ particle and the surrounding a-MO phase is the N-W. Thus, despite the fact that two phases are made of the f.c.c. y ’ phase and the b.c.c. a-MO phase, the orientation relationship are found to take the different forms. Further studies are needed for clarifying this difference. Recently, Miracle er al. (271 constructed the phase diagram of Ni-AI-MO ternary system in a temperature range of 1200-I 533 K. Considering that the temperature 1200 K is comparable to 1173 K employed in this study, the present alloy falls well in the two phase region of y’-a in the diagram at 1200 K.
Thus, the y’ particles in the Z-MO fibers are reasonably understood as the formation of a stable phase without considering any possible formation of carbides, metastable phases or the detrimental 6-NiMo phase. In the phase diagrams of Miracle et al. , solubility of Ni in ~-MO phase increases as temperature decreases. However, if such y’ particles are formed in the ~-MO phase and, as the present preliminary experiment has shown, they grow with decreasing temperature, the solubility of Ni in the Z-MO phase should decrease as temperature decreases in contrast to the phase diagram of Miracle et al. . In the present study, identification of the particles in the a-MO fibers has been attempted exclusively in the two phase ~‘a composite. The particle formation in the a-MO fibers was also reported in the three phase y/y’-2 composite . Although the present study has not referred, it has been found that the similar relationship shown in Fig. 3 holds for the particles in the ~-MO fibers in the y/y’* three phase composite. This indicates that these particles are the same as those identified in the y’-u composite.
5. CONCLUSIONS 1. From the EDX and the SAD analyses, the fine precipitate particles formed in the ~-MO fibers arc identified as the ^J’ phase having the fixed composition same as the y’ matrix around the ~-MO fibers, the structure of Ll, type and the N-W otientation relationship with the ~-MO phase. 2. This y’ formation in the fibers is understood as the stable phase formation consistent with the phase diagram of Miracle et al. and hence precludes any possibility of carbides, metastable phases or the detrimental 6-NiMo phase. 3. When combined with the electron diffraction analysis, the method proposed by Cliff er al. is useful for identification of particles which are smaller than the electron probe diameters. 4. For observing the Nishiyama-Wasserman (N-W) orientation relationship, it is preferable to take diffraction patterns from the direction parallel to [ 110]t,c.,.rather than (00 l]bc.c.. Acknowledgements-The authors are grateful to Dr Y. Miura and Mr T. Hatano for valuable comments and Mr T. Manabe for generous cooperation. This research was supported partly by Grant in Aid for Scientific Research from the Ministry of Education, Science and Culture of Japan and partly by the Kurata Research Grant.
REFERENCES I. M. H. 2. M. H.
Nemoto, T. Suto, Trans. Nemoto, T. Suto, Trans.
Honda, Y. G. Nakagawa, Y. Saiga and Japan hr. Melah 21, 495 (1980). Honda, Y. G. Nakagawa, Y. Saiga and Japan Insf. Metals 21, 505 (1980).
3. H. Yoshizawa, K. Wakashima, S. Umekawa and T. Suzuki, Scripta metall. 15, 1091 (1981). 4. K. Wakashima, K. Higuchi, T. Suzuki and S. Umekawa, Acta metall. 31, 1937 (1983). 5. S. F. Dirnfeld and D. Schwam, Mat. Res. Sot. Symp. Proc., Vol. 21, Phase Transformation in Solids (edited by T. Tsakalakos), p. 775. Elsevier, Amsterdam (1984). 6. Y. G. Nakagawa and M. Nemoto, In Situ Composites IV (edited by Lemkey, H. E. Cline and M. McLean), p. 143. Elsevier, Amsterdam (1982). 7. A. M. Sriramamurty, D. Banerjee and S. N. Tewari, Acta mefall. 30, 1231 (1982). 8. F. D. Lemkey, NASC-76115-30, Final Report on NAVAIR Contract N62269-75-C-0129, United Technologies Research Center, East Hartford, Conn. (1975). 9. T. Sano, R. Nishihara, H. Yotsumoto, Y. Miura and M. Nemoto, Tech. Rep. Kyushu Univ. 56, 491 (1983). 10. G. Cliff and G. W. Lorimer, J. Microsc. 103, 203 (1975). 11. J. I. Goldstein, Introduction to Analytical Electron Microscopy (edited by J. J. Hren, J. I. Goldstein and D. C. Joy), p. 83. Plenum Press, New York (1979). 12. Z. Horita, T. Sano and M. Nemoto, J. Microsc. (1985). 13. J. I. Goldstein, J. L. Costly, G. W. Lorimer and S. J. B. Reed, SEM/1977(edited by 0. Johari), Vol. I, p. 3 15. IIT Res. Inst., Chicago, Ill. (1977). 14. N. J. Zaluzec, Introduction to Analytical Electron
16. 17. 18.
20. 21. 22. 23. 24. 25. 26. 27.
Microscopy (edited by J. J. Hren, J. I. Goldstein and D. C. Joy), p. 121. Plenum Press, New York (1979). G. W. Lorimer, G. Cliff and J. N. Clark, Deuelopments in Electron Microscopy and Analysis (edited by J. A. Venables), p. 153. Academic Press, New York (1976). W. A. Knox, Ultramicroscopy 1, 175 (1976). T. Ishii, D. J. Duquette and N. S. Stoloff. Acta metal/. 29, 1467 (1981). G. Cliff, D. J. Powell, R. Pilkington, P. E. Champness and G. W. Lorimer, Developments in Electron Microscopy and Analysis, p. 63. Inst. Phys., London (1983). G. W. Lorimer, G. Cliff, P. E. Champness, C. Dickinson, F. Hasan and P. Kenway, Analytical Electron Microscopy (edited by D. B. Williams and D. C. Joy), p. 153. San Francisco Press, San Francisco, Calif. (1984). E. C. Bain, Trans. metall. Sot. A.I.M.E. 70, 25 (1924). W. Pitsch, Phil. h4ag. 4, 577 (1959). Z. Nishiyama, Sci. Rep. Tohoku Univ. 23, 637 (1934). G. Wasserman, Archs Eisenhtitt. 16, 647 (1933). G. Kurdjumov and G. Sachs, Z. Phys. 64, 325 (1930). M. Nemoto, Bull. Japan Inst. Me/al.7 22, 132 (1983). C. B. Shoemaker and D. P. Shoemaker, Acfa crystallogr. 16, 997 (1963). D. B. Miracle, K. A. Lark, V. Srinivasan and H. A. Lipsitt, Mefall. Trans. 15A, 481 (1984).