Pressure induced phase transition from wurtzite to rocksalt structure of Zn0 has been investigated up to 85 kbar and 1200℃. Phase identification at high pressure and temperature is based on the x-ray diffraction method using energy dispersive technique instead of the usual quenching method. The transition curve is approximately expressed by a linear relation, P(kbar)= -0.023 T(℃) + 74. The lattice parameters of both phases at various temperatures along the phase boundary were also measured. The volume change and thus calculated transition entropy were found to be almost temperature independent. The compressibility measurements at room temperature indicated the reduction of the axial ratio c/a with increasing pressure for the wurtzite type phase. These distortions of the hexagonal lattice, which are likely to be related to the phase transition in ZnO, are discussed with special reference to the theory of Keffer and Portis in term of the partial polar binding in wurtzite type compounds.

Reflections 200, 220, 311 and 222 of rocksalt structure are clearly resolved. Reflections of boron nitride sample holder, with which reflections from sample are often interfered in cases of low Z sample materials, are weak and negligible in this case because of the contrast of absorption in ZnO and boron nitride. Searching the wurtzite-rocksalt phase boundary, diffraction patterns were recorded in the interval of 100 deg. at various constant press loads. A series of patterns at 45 ton press load (corresponding to about 50 kbar) are shown in Fig. 2.

Fig. 2 Series of x-ray diffraction patterns to search the wurtzite-rocksalt phase boundary at 45 ton constant press load. patterns were recorded for 400 sec accumulating time sequentially from bottom to top. At 900℃ existence of both phases are clearly identified.

Exposure time for each pattern was only 400 sec, that seems enough to distinguish which phase is stable at the condition. Only Wurtzite phases are shown in the patterns from room temperature before heating up to 800℃, however, at 900℃ reflection 220 of rocksalt type ZnO appears suddenly between the reflections 110 and 103 of wurtzite type phase. Further heating to 1000℃ causes the increase of intensities of peaks from the new phase relative to those from low temperature phase. After subsequent heating up to 1200℃ there appears to be no trace of wurtzite phase at last. After rapidly quenched from 1200℃ to room temperature, only metastable rocksalt phase can be seen and there seems to be no reversion to wurtzite phase. Finally after releasing the pressure, rocksalt phase disappears and only wurtzite phase with broad reflections is observed. The results of phase identification at various P-T conditions are summarized in Fig. 3.

 

Fig. 3 P-T phase diagram for ZnO. The broken line shows the results of Bates et a1. by quenching method. At P-T conditions the arrows indicate, the lattice parameters of both phases, therefore volume changes were measured.

The results of Bates et al. (1962) are also plotted together in the figure. Since the old pressure scale was adopted in their original report, the correction was given based on the new fixed-point pressure scale calibrated by NaCl internal standard (Jeffery et al., 1966). A remarkable discrepancy between the Bates' results and the present ones is shown not only in the position in P-T space but in the sign of the slope of the phase boundary. In the quenching experiment the rocksalt phase could be retained only with ammonium chloride as a catalyst and furthermore, new phase has not been prepared free from contamination by Zincite even with long runs. The disagreement with our in situ phase identification is presumably coming from the failure of quenching in pure ZnO. The phase boundary determined in the present study can be expressed;
P(kbar) = -0.023 T(℃) + 74. (1)
The slope of the boundary is relatively small but definitely negative in comparison with the positive value of + 0.0425 kbar/deg in the quenching experiment. The lattice parameters of both phases along the equilibrium line are listed in Table 1.

The transition enthalpies and entropies are calculated by using Clausius-Clapeyron eq. to be almost temperature independent as shown in the table. Figure 4 shows a typical pattern used for the compression measurement of wurtzite type phase at room temperature.

Fig.4. Energy dispersive x-ray diffraction pattern of wurtzite type ZnO at 75 kbar and room temperature.

The effect of pressure on the volume, linear compressions and axial ratio, c/a for wurtzite phase of ZnO is shown in Fig.5 and 6,respectively.

Fig. 5 Pressure dependence of volume compression for wurzite type ZnO at room temperature. The solid line shows the ultrasonic equation of state.

Fig. 6 The effect of pressure on the lattice parameters and the axial ratio of wurtzite type ZnO at room temperature.

The solid line shows the ultrasonic eq. of state based on 2 parameter Birch eq. using the isothermal bulk modulus and its pressure derivative measured by Soga and Anderson (1967). The present results gives appreciably smaller bulk modulus than that determined in the ultrasonic experiment. Similar discrepancies are frequently encountered in many other high pressure x-ray diffraction experiments and have been partly accounted in terms of the stress inhomonogeneity in the aggregate of fine grains having different elastic properties (Sato, 1973). Accordingly it can be said that the present results on volume and linear compressions do not represent the intrinsic elastic properties of ZnO and probably there appears no correlation with the anomaly in shear property in polycrystalline ZnO. 0n the other hand, it should be noted that the pressure dependence of axial ratio, however, still remains valid in a qualitative sense.
Even at 1 atm, the axial ratio for ZnO is 1.602, far from the ideal value (1.633) for the hexagonal close packing. Such a small value in c/a in ZnO among the other wurtzite type compounds is considered to be coming from the large difference in ionic electronegativity with increasing pressure, c/a decreases monotonically to about 1.591 at 80 kbar in a further leaving sense from the ideal value. In the following discussion this anomalous variation in c/a will be related only qualitatively to the phase transition to rocksalt structure. Keffer and Portis(1957) have accounted the slight departure from ideal c/a ratio and from ideal sublattice displacement, u in wurtzite type compounds by the forces arising from partial polar binding and calculated these long range forces in terms of a postulated charge ±fe on ion and certain lattice sums in order to balance against the elastic, piezoelectric and dielectric contributions caused by the distortion.
For the departure of c/a from the ideal value, they derived the following equation;

where N is the number of ions of one sign per unit volume, fe, the assumed effective charge on each ion, s_ij and d_ij the elastic compliance and piezoelectric strain constants, respectively. In the case of ZnO, we evaluated f to be a reasonable value, about 0.78 at 1 atm by using the x-ray and single-crystal elastic data (Bateman, 1968). The accompanying piezoelectric effect have been assumed to play only a minor role to the lattice contraction and the effect of charge transfer have been also neglected in the calculation. It can be concluded, however, that ZnO is considerably ionic and the anisotropy in the Coulomb interaction is mainly responsible to the distortion from the ideal hcp lattice at 1 atm in wurtzite type ZnO. There is no direct measurement on the pressure derivatives of the second order compliance constants for ZnO, but it is probable to assume that the quantity in the square parenthesis in eq. (2) is roughly proportional to s₁₁ because s₁₂ and s₁₃ are small as compared to s₁₁ and s₃₃ and also the order of magnitude of ∂s₁₁ /∂p is about -1 x 10^-24 cm⁴/dyne².
Only less than 10% increase in the right side of eq. (2) was calculated at 50 kbar, for instance. The observed increase of about 25% at this pressure in the left side of eq. (2) i.e. the departure of c/a from the ideal value, cannot be explained without taking account of the change in ionic contribution with increasing pressure. Consequently the effective charge on each ion is expected to become larger at the higher pressures. In the ZnX (X=0, S, Se, Te) series, the large drops in the room temperature resistivity ZnS, ZnSe and ZnTe were found near 240-245, 165, 140-145 kbar, respectively (Samara et a1,1962) and confirmed to be the first order phase transitions from wurtzite structure to rocksalt one by means of high pressure x-ray diffraction technique (Smith and Martin, 1965). The sequence of the transition pressures in these three compounds is consistent with that of the transitions from diamond structure to βSn one in the group IV elements, where the element with lower atomic number undergoes phase transition at lower pressure. Relatively low transition pressure in ZnO observed in this study is violating the tendency mentioned above.
Jayaraman et al. (1963a,b) have pointed out in the generalization of the phase transitions in group IV elements, Ill-V and II-IV compounds that the wurtzite type compound with larger electronegativity difference tend to transform to rocksalt structure at lower pressure and to become metallic at higher pressure. It may be concluded that the relatively ionic nature of binding even at 1 atm among other covalent ZnX (X=S, Se, Te) and presumably the tendency of becoming more ionic at higher pressures are responsible to the transition to 6-coordinated rocksalt structure at such a low pressure.