J. Chem. SOC., Faraday Trans. I, 1982, 78, 1131-1 140 Change in the Phase Stability of Zinc Blende and Wurtzite on Grinding BY KENICHIMAMURA AND MAMORU SENNA* Department of Applied Chemistry, Faculty of Engineering, Keio University, 3-14-1 Hiyoshi, Yokohama 223, Japan Received 1 lth May, 1981 Two different zinc blende (ZB) powders were vibro-milled and the transformation process into the wurtzite (WZ) phase on subsequent heating in N, was observed. Vibro-milled ZB transformed rapidly into WZ at the beginning of the heating. Part of the WZ was then retransformed into ZB leaving a large amount of WZ, the reaction attaining an apparent equilibrium state. The transition took place even at 1023 K, which is more than 200 K lower than the lowest Z B e WZ transition temperature ever reported.The results were interpreted on the assumption of an increase in the chemical potential and a broadening of the chemical potential distribution within the powder mass caused by the mechanical treatment. Unlike liquids and gaseous substances, the chemical potential of a crystalline solid is generally structure sensitive. This often leads to uncertainty about the conditions necessary for phase stability. Reported values of the transition temperature between zinc blende and wurtzite, for example, are widely scattered, i.e. 1230,' 1243,2 1248,3 1293* or even 1423 K5 The variation in stability region caused by different degrees of activity will become more marked when a solid is mechanically treated. This is one of the fundamental principles of mechanochemical activation.6 There are very few systematic studies, however, on the mechanochemical effect on the polymorphic equilibrium.Zinc sulphide is a suitable material for a systematic study because of its high structural sensitivity. Gmk7 studied the mechanochemical activation of ZnS with regard to ore dressing. In his work, however, phase transformation played a subsidiary role. The main purpose of the present study is to elucidate the change in phase relation between zinc blende and wurtzite caused by preliminary grinding. EXPERIMENTAL MATERIALS Two different sorts of zinc blende were used as starting materials. (a) Synthetic ZnS8 (ZP): gaseous H,S was blown slowly into 0.1 mol dm-3 ZnSO, aqueous solution for 10 h. Precipitated ZnS was washed with water and acetone and was dried at room temperature under reduced pressure for 2 days.After deagglomeration in a mortar, the sample was tempered in N, at 673 K for 2 h and then at 753 K for 3 h. The temper conditions were chosen so as to obtain as good crystallinity as possible with no transformation into wurtzite. (b) Natural ZnS (ZN): A natural zinc blende ore from Aniya, Japan, was crushed under 100 mesh and tempered in N, at 473 K for 5 h. 11311132 PHASE STABILITY OF ZnS GRINDING The starting material was vibro-milled in an inert liquid, cyclohexane, in order to reduce the possible complicating effects of the moisture, oxygen and significant local heating. A 1 g sample, 100 steel balls of 6 mm diameter and 30 cm3 cyclohexane were put into a 50 cm3 cylindrical steel vessel of 35 mm diameter. Vibro-milling was carried out isothermally at 303 K using a small laboratory mill (Glen Creston) operated at 12 Hz.After grinding, the supernatant was decanted and the rest was dried for 15 h at room temperature under reduced pressure. The specimens are labelled ZP-n or ZN-n according to the starting material, with n being the grinding time in hours. HEATING OF THE SPECIMEN The samples were heated at predetermined temperatures iegulated within 1 K in a half-opened tube either of glass (< 873 K) or of quartz (2 873K) under a N, flow of 5 dm3 min-l. X-RAY DIFFRACTOMETRY Because of the superimposition of many diffraction peaks, it is not easy to determine the amount of each phase precisely for the 11-VI compounds which have zinc blende and wurtzite isomorphic ~tructures.~ The difference in the interplannar distance between zinc blende (1 1 1) and wurtzite (002) is only 5 x nm.Hence the diffraction peaks from these lattice planes could not be resolved. Instead, the total intensity of these two peaks, I,, was determined. On the other hand, the ratio of the intensity of wurtzite (loo), I , (loo), to I, was constant at 1.3 for all the samples obtained by heating at > 1273 K for a sufficiently long time. By assuming that the ratio I,(lOO)/~, = 1.3 corresponds to the pure wurtzite phase, where 1,(111) = 0 [cJ the ratio I,( 100)/1,,,(002) = 1.16, according to the ASTM card], the fraction of wurtzite phase, (1) x,, was determined by X, = 1,(100)/1.31I, where I, = Z,(002) + I,( 1 1 1). The lattice distortion and the crystalline size were determined by conventional methods according to Hall.lo RESULTS PHASE TRANSFORMATION OF GROUND ZnS A remarkable phase change process was observed on heating the ground ZnS isothermally in N, flow. As shown in fig. 1, there was a rapid phase transformation into wurtzite at an early stage of heating, showing a maximum wurtzite fraction, xwm. A sluggish retransformation then occurred, resulting finally in an apparent equilibrium state at the fraction of wurtzite, x,,. Note that the transformation was scarcely observed on very well crystallized zinc blende, ZN-0, even after heating at 1273 K for 3 h. This indicates that the initial rapid transformation into wurtzite is characteristic of ground zinc blende. With increasing heating temperature, both x, and x,, increased as shown in fig.2 and 3. However, the rate of increase with temperature was much larger for the ZP series than for the ZN series. The effect of grinding, particularly on x, was also much more pronounced for ZP than ZN. RELATION BETWEEN X, X, A N D THE LATTICE DEFORMATION The distortion of the zinc blende lattice, q-, increased, whereas the crystallite size, D, decreased with grinding time in the manner shown in fig. 4. Before heating, no mechanochemical phase transformation into wurtzite was observed. Within each series (ZP or ZN), x, increased monotonically with q, as shown in fig. 5 . The value of x, at the same q, however, was substantially higher for ZP than ZN. The relationK. IMAMURA AND M. SENNA 1133 0.71 0 10 20 30 heating time/h FIG.1.-Change in the fraction of wurtzite, x,, with heating time for ZN-8 (squares) and ZP-8 (circles). Curves 1 and 4, heated at 1023 K; curves 2 and 5, heated at 1123 K; curves 3 and 6, heated at 1173 K. 0.7 0.6 0.5 0 . 4 E Y 0.3 0.2 0 . 1 f' 1 0 1 " ' I " " I " " 1 ~ ' 1050 1100 11 50 temperature/K FIG. 2.-Relation between x, and temperature. Curve 1, ZN-4; curve 2, ZN-8; curve 3, ZP-0; curve 4, ZP-4; curve 5, ZP-8.1134 PHASE STABILITY OF ZnS 0.4 - 083 - -.---------:: 1 0.1 - -0 I , , I , , , I , , 0 1050 1100 1150 temperature/K FIG. 3.-Relation between x, and temperature. Curve 1, ZN-4; curve 2, ZN-8; curve 3, Zp-0; curve 4, ZP-4; curve 5, ZP-8. 3 0 1 2 3 4 5 6 7 8 melting time/h FIG. 4.-Variation of lattice distortion (filled symbols) and crystallite size (open symbols) with milling time, for ZN (0, m) and ZP (0, a).K.IMAMURA A N D M. SENNA 1135 0.7 0 . 6 0.5 0 . L E 2 0.3 0.2 0 . 1 lattice distortion.,'q (arb. units) FIG. 5.-Relation between x, and lattice distortion for ZN (squares) and ZP (circles). Curves 1 and 4, heated at 1023 K; curves 2 and 5, heated at 1123 K; curves 3 and 6, heated at 1173 K. 1 .o 0.5 5 0.2 0 . 1 8 10 20 50 80 100 crystallite size, D/nm FIG. 6.-Relation between x, and crystallite size. Curve 1, heated at 1023 K; curve 2, heated at 1123 K; curve 3, heated at 1173 K.1136 PHASE STABILITY OF ZnS FIG. 7.-Relation between x,, and lattice distortion for ZN (squares) and ZP (circles). Curves 1 and 4, heated at 1023 K; curves 2 and 5 , heated at 1123 K; curves 3 and 6, heated at 1173 K.8 10 20 50 80 100 crystallite size, D/nm FIG. 8.-Relation between x,, and crystallite size. Curve 1, heated at 1023 K ; curve 2, heated at 1123 K; curve 3, heated at 1173 K. between xwm and D, on the other hand, was fairly simple, irrespective of the starting material, as shown in fig. 6. The relations between x,, and v (fig. 7) and x,, and D (fig. 8) were similar to those for x, with the exception that x,, was practically constant when q was large. DISCUSSION EFFECTS OF A N D D O N THE RATE OF TRANSFORMATION As seen from fig. 5-8, the effects of and D are more pronounced on x, than on x,,, because the imperfection of the matrix phase could serve as a driving forceK. IMAMURA A N D M. SENNA 1137 for the zinc blende + wurtzite transformation, by enhancing the nucleation of the new phase, but not for the reverse process.The distorted zinc blende will recover to a less distorted state, which is recognized from the sharpening of the X-ray diffraction peak of the starting phase, as shown in fig. 9.* The major portion of the decrease in the line breadth occurs at the initial stage of the heating. The behaviour is compatible - 0 10 20 30 40 heating time/h FIG. 9.-Variation of X-ray diffraction peak breadth of zinc blende ( 1 1 1 ) with heating time, for ZP-8. Curve 1, heated at 1023 K ; curve 2, heated at 1123 K; curve 3, heated at 1173 K. with the observation represented in fig 1, in the sense that the transformation from zinc blende to wurtzite occurs mainly at the early stage of the heating.It indicates that the transformation from zinc blende to wurtzite characteristic of the ground zinc blende is achieved at the cost of the lattice imperfection of the matrix zinc blende phase. On the other hand, x, and x,, were larger for smaller values of D at the same values of q. This would mainly be owing to the larger sites for the formation of the new stacking, i.e. the nucleation of the wurtzite phase. This was similar for the aragonite -+ calcite transformation,ll in spite of the rather different transformation mechanism. CHANGE I N THE TRANSFORMATION TEMPERATURE OWING TO G R I N D I N G The extent of the maximum transformation, x, is determined not only by the balance of the rates of the forward and reverse transformations, but also by the stability of the phases.As already mentioned, the reported temperatures of zinc blende G= wurtzite equilibrium range from 1230 to 1243 K. In the present experiment, the transformation of zinc blende into wurtzite occurred ,at 1023 K, which is more than 200 K lower than the lowest equilibrium temperature reported previously. The formation of intermediate phases possibly assigned as y-ZnS l2 or polytypes such as 6H, 18H or 24H l3 was not observed in the present experiment. Even if the transformation takes place through such intermediate structures, it still seems necessary to postulate that the chemical potential of the zinc blende phase has been changed significantly because of the mechanical treatment, as will be discussed later. * Because of the peak superposition, a precise analysis to obtain q and D separately for the partially transformed samples was not possible.1138 PHASE STABILITY OF ZnS POSSIBILITY OF THE GRADUAL TRANSITION MECHANISM Besides the maximum degree of transformation, one of the most characteristic observations was the existence of the apparent equilibrium state of the mixed phases, even after prolonged heating.According to classical thermodynamics, coexistence of two immiscible equicomponent phases at a constant pressure and an arbitrary temperature is not allowed. If two phases were miscible, forming a mixed stacking due to the layer structure of ZnS, then a temperature region for the stable coexistence of both phases could still be expected, by assuming the gradual transition mechanism proposed by Allen and Eagles.14 Even if the gradual transition mechanism were working and the different value of x,, at different temperature were explained, it still leaves open the question of why the retransformation from wurtzite to zinc blende takes place.NECESSARY ASSUMPTIONS In order to explain the above behaviour with regard to the phase transformation, the following two assumptions are necessary: (i) the chemical potential of the zinc blende phase, pZ, is increased on grinding and (ii) the value of pz is not identical throughout the sample but has a certain distribution within a powder mass, the distribution being broadened when the sample is ground. The major part of the excess free energy in active solids is considered to be the enthalpy contribution.15 Since a number of experimental r e s u l t ~ l ~ - ~ ~ verify the increase in enthalpy through mechanical stressing, assumption (i) is generally acceptable, A direct measurement of the enthalpy increase of the present specimen using a differential scanning calorimeter was not successful, presumably because of the insufficient sensitivity of the instrument and the sluggish energy release.Assumption (ii) can also be accepted, since each particle has a different chance of being hit by the milling balls. Even if the chance could be averaged after prolonged grinding, local differences in the microdeformation within particles cannot be avoided. Moreover, even the particle size distribution alone could result in the free energy distribution since, according to Allen and Eagles,l* the chemical potential of the solid material is a function not only of the temperature, pressure and composition, but also of the lattice distortion and the particle size.If the liberation of excess energy were easily observable, e.g. by means of d.t.a. or d.s.c. thermograms, as with A1,20 CaFZ2l or y-Fe20,,22 it would be possible to estimate the broadness of the distribution. With these assumptions, the behaviour of zinc blende during vibro-milling and subsequent heating can be explained. VARIATION OF THE CHEMICAL POTENTIAL DISTRIBUTION The chemical potential distribution, m), of well crystallized zinc blende, mZ)O, is very sharp, as shown schematically in fig. lO(a). No phase transformation into wurtzite can take place at a temperature, qx, below the ‘true’ transformation point, cr, where the two chemical potential curves, & and ,u:, of zinc blende and wurtzite in well crystallized states, respectively, cross each other.After grinding, the chemical potential distribution is broadened between pz and pz,max, as shown in fig. lO(b). Since a part of the ground sample has a chemical potential higher than &, it is possible for the transformation to take place at temperatures between ItJr,rnin and cr, as shown in fig. lO(c). The resulting wurtzite phase could also possess a certain distribution of chemical potential owing to the insufficient growth of the new nuclei. After prolonged heating, the rest of the zinc blende, which has not transformed into wurtzite, recovers to make the distribution narrower again, so that no further transformation into wurtzite is possible, as shown in fig.lO(d). On the other hand,T P ( d ) T T FIG. 10.-Schematic representation of changes in chemical potentials and their distribution caused by grinding and subsequent heating. (a) Starting material (well crystallized zinc blende); (b) ground zinc blende; (c) after heating for a short time; ( d ) after prolonged heating.1140 PHASE STABILITY OF ZnS some insufficiently grown wurtzite has a chance to retransform into zinc blende. The rest of the wurtzite grows further resulting in well crystallized wurtzite, as is also seen from the sharpening of the X-ray diffraction peaks (not shown). Thermodynamically, it is still possible for wurtzite to transform into zinc blende, even after the phase has grown, when heated at temperatures lower than TFr.In the actual experiments this did not occur. Instead, an apparent equilibrium state was observed. This could be due to kinetic factors, and in particular the difficulty of nucleation, as in the case of the massicot to litharge transformation without mechanical aids.23 COMPETITIVE PROCESSES The excess free energy in the activated zinc blende could be used up as a driving force either in the transformation to wurtzite, or in the return to the more stable zinc blende. through recovery and recrystallization. A similar kind of competition should also take place in the latter part of the transformation process, where the insufficiently crystallized wurtzite phase could either retransform to zinc blende or remain as wurtzite and continue to grow.Which of the two processes is dominant is an important question when mechanochemical activation is applied to practical solid-state reactions. It depends on many complicated factors, one of the most important being the heating temperature, as partly elucidated in the case of mechanically treated y-Fe,0,.22 We thank Prof. H. Kuno for valuable discussions. A. Kremheller and A. Lavine, Sylvania Technol., 1957, 10, 67. H. Samelson, J. Appl. Phys., 1962, 33, 1779. N. N. Sirota and V. P. Sapelkina, Krist. Tech., 1971, 6, 381. E. T. Allen and J. L. Grenshaw, 2. Anorg. A&. Chem., 1913, 79, 2. A. Addamino and M. Aven, J. Appl. Phys., 1960, 31, 1. * R. Schrader and B. Hoffman, Festkorperchemie (VEB Deutscher Verlag, Leipzig, 1973), p. 522. ' E. Gock, Habilitationsschrijit (T. U. Berlin, 1977). B. Basak, D. R. Glasson and S. A. A. Jayaweera, Particle Growth in Suspensions, SOC. Chem. Ind. Mongr., 1974, 38, 143. !a G. Ohtani and M. Senna, Phys. Status Solidi A, 1980, 60, K35. lo W. H. Hall, Proc. Phys. Soc. London, Sect. A , 1949, 62, 741. l1 H. Momota, M. Senna and M. Takagi, J . Chem. Soc., Faraday Trans. I , 1980, 76, 790. l2 D. C. Buck and L. W. Strock, Am. Mineral., 1955, 40, 192. l3 C. J. Schneer, Bull. Geol. SOC. Am., 1958, 69, 1640. l4 J. W. Allen and D. M. Eagles, Physica, 1960, 26, 492. l5 K. Torkar, Proc. 4th Int. Symp. Reactivity of Solids, Amsterdam, 1960, ed. J. H. deBoer (Elsevier, lo G. S. Chodakov and L. I. Edelmann, Kolloidn. Zh., 1967, 29, 728. Amsterdam, 1961), p. 400. R. Schrader and B. Hoffmann, 2. Anorg. Allg. Chem., 1969,369,41. K. TkaEova, Silikaty, 1976, 4, 321. M. Senna and K. Schonert, Powder Technol., to be published. 2o L. M. Clarebrough, M. E. Hargreaves, M. H. Loretto and G. W. West, Acta Metall., 1960, 8. 797. 21 R. Schrader and W. Oese, J. Therm. Anal., 1970, 2, 349. 22 H. Imai and M. Senna, J. Appl. Phys., 1978, 49, 4433. 23 M. Senna and H. Kuno, J. Am. Ceram. Soc., 1971,54, 259. (PAPER 1/752)