Creator:Frank Dachille and Rustum Roy Date Created:June 6, 1959 Place Created:University Park, Pennsylvania Keywords:silica isotypes,phase changes Context:report about high-pressure regions of silica isotypes ************************************************** Sonderdruck aus: „Zeitschriffc fur Rristallographie", 111, 6, 1959 Herausgegeben von M. J. Buerger, M. von Laue, F. Laves, G. Menzer, I. N. Stranski Akademische Verlagsgesellsehaft m. b. H., Frankfurt am Main High-pressure region of the silica isotypes By Frank Dachille and Rustum Roy College of Mineral Industries, The Pennsylvania State University University Park, Pennsylvania* With 3 figures (Received June 6, 1959) Auszug Die Si02-Modifikationen, sowie BeF2, A1P04, A1As04, A1V04, MnAs04, GaP04, GaAs04, GaSb04, BPO„, BV04, BAs04, FeP04 und FeAs04 wurden im Bereich von Atmospharendruck bis etwa 60.000 bar und von Zimmertemperatur bis annahernd 600 °C untersucht. Die Hochdruck-Apparatur wurde an Hand der Quarz-Coesit-Umwandlungen auf Reproduzierbarkeit, Reversibilitat und Erreichbarkeit des Gleichgewichts sorgfaltig gepriift. Bei 500 °C wandeln sieh Si02 und BeF2 beiDrucken von 20.400 bzw. 21.600 bar aus der Quarzform in die Coesitform um und geht BP04 bei 46.000 bar aus der Cristobalit- in die Quarzform iiber. Die Umwandlung von MnP04, FeP04, GaP04 und A1P04 erfolgt um 450 °C und 55.000 bar. In diesem Bereich wurde keine Umwandlung von A1As04 und GaAs04 beobachtet. Tridymit und Cristobalit konnten nicht in neue meta-stabile Hochdruekformen ubergefuhrt werden. Die ytf-Kurven dieser Re-aktionen wurden fiir einen engen Zustandsbereich ermittelt. Abstract The region extending from ambient conditions up to approximately 600° C and 60,000 bars has been, examined for the various polymorphs of Si02, and for BeF2, A1P04, A1As04, A1V04, MnAs04, GaP04, GaAsO,, GaSb04, BP04, BV04, BAs04, FeP04 and FeAs04. Reproducibility, reversibility and attainability of equilibrium have been thoroughly explored for the high-pressure apparatus, using the quartz-coesite reaction. At 500° C, Si02 and BeF2 transform from the quartz form to the coesite form at 20,400 and 21,600 bars respectively, andBP04 transforms from acristo-balite to a quartz form at 46,000 bars. MnP04, FeP04, GaP04 and A1P04 all transform to new forms near 55,000 bars at 450'' C. There are no new forms of AlAs04or GaAs04 in this range. Tridymite and cristobalite cannot be transformed metastably into other new high-pressure forms. The p-t curves [for these reactions have been determined over a narrow range. * Contribution No. 58—26. 29* 452 Fank Dachille and Rstm Roy Introduction With the discovery of coesite or 'silica-C"1 an entirely new chapter of inorganic chemistry was opened, since it became clear that many new phases could be prepared under high pressure (> 10,000 bars). Since then some work has been done, especially by petrologists, on systems of petrologic interest at these pressures. This work has not, in general, led to a large number of new phases, since nature had already provided the high pressure to generate most of these phases. No other systematic work was known at these pressures and temperatures. Together with attempts to synthesize various phases, including diamonds, a program of systematic high-pressure studies was started in 1955. Some results of these studies have been reported by us at the annual meetings of the Geological Society of America in 1956, 1957 and 1958. An obvious choice for an area in which such systematic work would be meaningful was the silica structures. The quartz-coesite transition has been studied by numerous workers, the results of MacDonaud 2 being the most complete of those published. Nothing is known about the effect of pressure on the many analogous phases. Thus one might expect, on the basis of bond strengths alone, that BeF2 would transform, if it did, to a coesite structure at a much lower pressure than does SiOa; or likewise one would think that, since GaP04 has an average ^^ ratio much nearer the rutile ratio, that it would perhaps invert to rutile in this pressure range. Nothing quantitative is known about the effect of pressure on radius ratio, and hence one cannot predict the magnitude of pressures which might be required to go from one structure to another. An empirical approach is therefore not only necessary in this area of solid-state chemistry, but is virtually the only useful one. Apparatus The p-t range under consideration has been reached for sustained periods only in a very few laboratories. The industrial organizations (such as The General Electric Company and Swedish ASEA) which have prepared diamonds have not published any description of their apparatus. Coes 1, however, has done so, and a detailed drawing of his 1 Loreng Coes, A new dense crystalline silica. Science 118 (1953) 131 and personal communication (1954). 2 G. J. F. MacDonaijd, Quartz-coesite stability relations at high temperatures and pressures. Am. J. Sci. 254 (1956) 713—721. High-pressure region of the silica isotypes 453 apparatus is available3. It is the prototype for an entire family of apparatuses which have since been built for this range. Another family is based on the designs of Basset4 and of Bridgman5 as modified by Griggs and Kennedy6. The apparatus used in the present study belongs to the latter family which is herein called an (externally heated) uniaxial pressure device. The assembly consists of a pump delivering pressure (which is automatically controlled and recorded) to the low-pressure side of a hydraulic ram. The ram in a suitably strong frame brings the total Split furnace Control thermocouple Circular plate 'diameter 12" Sample i| thermocouple 20 ton hy-'draulic jack Pressure recorder controller "Nj (10000 PS.1.) [L Piston Piston Pt-lO%Rh ■ Fracture shields Fig. 1. Schematic layout of uniaxial high-pressure apparatus with accessories. Details of high-pressure pistons and sample assembly. force to bear on a small area piston and sample assembly. This is shown in detail in Fig. 1. The sample is pre-pelleted into a nickel ring 0.010 inches thick and surrounded by two platinum-rhodium sheets, thus giving an essentially noble-metal container. The temperature is measured by a thermocouple very close to the sample sandwich, and is easily fixed within ± 5 ° C. Several types of pistons are 3 Rustum Roy and O. F. Tuttle, Investigations under hydrothermal conditions. "Physics and Chemistry of the Earth", Vol. 1. Pergammon Press, London and New York (1956). 4 James Basset. Apparatus for carrying out physical or chemical experiments under pressure. Comptes rend. [Paris] 85 (1927) 343—345. 5 P. W. Bridgman, The physics of high pressure. G. Bell, London (1949). 6 D. T. Griggs and G. W. Kennedy, A simple apparatus for high pressures and temperatures. Am. J. Sci. 254 (1956) 722—735. 454 Frank Dachille and Rtjsttjm Roy in use, cemented carbides, hardenable steels, and sintered oxides (for higher temperatures at lower pressures). Water or a "mineralizer" solution is often added to the pelleted sample before enclosing in the platinum-rhodium. Either the pressure or temperature is raised first, depending upon the result sought, and the apparatus used is capable of maintaining these p-t conditions for periods of one week or more. Runs of 20—40 hours length were made standard to avoid problems of incomplete reaction or metastability, and this is an important innovation; many runs of week-long duration were made in this same range. For most rapid quenching a cold air blast was used to cool the samples rapidly before releasing the pressure. The samples were extracted and examined by x-ray diffraction and petrographic microscopy. Results We have discussed elsewhere7 our calibration of such devices. Briefly, the uniformity and reproducibility of pressure on the metal ring-sample surface were studied by determining the pressure of the Si02 quartz-coesite transition at 500 °C, using a series of sample assemblies which differed in their overall diameters and in their metal-to-sample area ratios. This pressure was found to be within ±0.4 kilobar of 20.4 kilobars even though the area ratios were changed from 3.5:1 to 0:1 and overall diameters were 0.25 and 0.50 inches. The uniformity of pressure was also evident in the consistent results obtained with different-diameter sample assemblies in high-pressure studies8 of the olivine-spinel transition in the system Mg2Ge04—Mg2Si04. The direct comparison of uniaxial pressure with hydrostatic pressure probably will never be realized over the full range of uniaxial pressures attainable. However, we have found that uniaxial and hydrothermal results at low pressures agree for the univariant p-t relation8 of the olivine-spinel transition of Mg2Ge04. Griggs and Kennedy6 also reported continuity of hydrothermal and uniaxial p-t relations for a number of systems. Suffice it to report that it is clear that in such apparatus the reproducibility and uniformity of the 7 Frank Dachille, E. C. Shafer and Rusttjm Roy, High pressure studies of the system GeOa— Si02. In press. 8 Frank Dachille and Rtjstum Roy, High pressure studies of the system Mg2Ge04—Mg2Si04 with special reference to the olivine-spinel transition. Am. J. Sci. (In press.) High-pressure region of the silica isotypes 45 pressure are within ± 5 per cent. Calibration of absolute pressure in the range of temperature and pressure has so far not been attempted in a serious way in any laboratory, and until this is done, absolute pressure values will remain somewhat questionable. The results obtained with each substance studied can be considered in turn, and a short list of critical data is given in Table 1. The results presented here are based on some 250 separate runs. 300 Fig. 2. Univariant pressure-temperature lines separating the fields of pairs of polymorphs for various compounds. (*Note—MnP04 quartz form reacts inconsistently at lower pressures than shown for the FeP04.) Si02. The p-t curve determined by a large number of runs in the calibration study is shown in Fig. 2. Coesite was prepared and then used as a starting material to demonstrate reversibility. The curve shown here agrees very well with that given by MacDonald1, but disagrees radically with scattered results reported by other workers9-10. In the attempts to determine the structure of coesite some difficulty was encountered by Ramsdell11 and by Zoltai and Buerger12 9 H. T. Hall, Proceedings of a symposium, High temperature, a tool for the future. Stanford Research Institute, Menlo Park, California (1956). 10 F. R. Boyd and J. England, Personal communication (1957), and Quartz-coesite transition. Carnegie Institution Yearbook 58 (1958/59) 87—88. 11 L. S. Ramsdell, The crystallography of coesite. Am. Min. 40 (1955) 975-982. 12 Tibor Zoltai and M. J. Buerger, The crystal structure of coesite, the dense, high pressure form of silica. Z. Kristallogr. Ill (1959) 129—141. 456 Frank Dachille and Rstm Roy Table 1. Short list of some experimental results Composition Starting material Pressure (kilobars) Temp. (°C) Time (hrs.) Results Si02 Silicic acid 20.5 499 48 Qtz. Silicic acid 20.7 500 20 Qtz. + tr. coes. Silicic Acid 21.7 494 36 Coes. + minor qtz. Silicic acid 24.0 586 48 Coes. + tr. qtz. Silicic acid 18.6 405 96 Qtz. + tr. coes. Silica glass 20 594 120 Qtz. Silica glass 22 502 48 Coes. + qtz. Tridymite 34 525 72 Qtz. + tr. coes. Tridymite 20.2 495 72 Qtz. + coes. Cristobalite 19.3 500 72 Qtz. Cristobalite 34.0 548 120 Coes. Coesite 19.3 590 60 95% qtz. Coesite 19.0 545 40 85% qtz. BeF2 Quartz form 21.7 575 72 Qtz. form Quartz form 22.7 570 72 Qtz. + coes. Coesite 17.3 450 80 Qtz. + minor coes. Glass 22.7 550 48 Coes. + minor^qtz. Glass 20.0 315 150 Coes. + minor qtz. Glass 19.3 348 80 Qtz. Glass 20.7 560 50 Qtz. bpo4 1:1 Acids 44.8 331 40 Crist. + tr. qtz. 1:1 Acids 46.2 475 55 Crist. + minor qtz. 1:1 Acids 42.8 330 40 Crist. 1:1 Acids 47.6 480 44 Qtz. + crist. 1:1 Acids 47.0 622 10 Crist. + qtz. Cristobalite 48.2 480 40 Qtz. + minor crist. Cristobalite 51.4 460 46 Qtz. BAs04 Cristobalite 34.5 453 20 Crist. Cristobalite 36.5 449 21 Qtz. + crist. Cristobalite 40.7 420 22 Qtz. Cristobalite 49.0 442 22 Qtz. aipo4 Gel 33 500 40 Qtz. Cristobalite 48 480 18 Qtz. Quartz 56 432 21 Qtz. Cristobalite 63 528 20 Qtz. + new phase Tridymite 55 469 24 Qtz. + new phase A1As04 Quartz 29 472 24 Qtz. Quartz 48 436 24 Qtz. Quartz 54 400 42 Qtz. High-pressure region of the silica isotypes 457 Table 1 (continued) Composition Starting material Pressure (kilobars) Temp. (°C) Time (hrs.) Results GaP04 Cristobalite 35 334 120 Qtz. Cristobalite 53 446 22 New phase + qtz. + tr. crist. Cristobalite 56 450 40 New phase + qtz. FeP04 ppt. Fe(N03)2 50 436 70 Qtz. 4" new phase and Na2HP04 53 446 20 New phase MnP04 Mn203 + H3P04, 27 388 40 Mainly crist. amorphous + reactants Mn203 + H3P04, 35 456 40 Qtz. + tr. new phase amorphous Mn203 + H3P04, 50 429 70 New phase + qtz. amorphous Cristobalite 53 413 24 New phase + qtz. -f crist. GaSb04 Rutile form 52 418 7 Rutile Rutile form 43 423 24 Rutile Abbreviations: Crist. = cristobalite, Qtz. = quartz, tr. = trace. The text summarizes pertinent results for other compositions. in fixing Z, the number of molecules/unit cell. We have redetermined the unit-cell parameters from an accurately measured and indexed powder pattern, and by a sink-float method also remeasured the density as 2.93 ± 0.02. When these corrections are made, as has been done in Table 2, Z is clearly established as 16. BeF2. This substance, which is nearly a perfect model13 for Si02, was found also to have a coesite polymorph. However, the pressure required for its formation* was considerably more than expected, and in fact is slightly above that required for Si02 (Fig. 2). Thus, although the melting behavior (BeF2 melts at 560°C) reflects rather directly the effect of the diminished interionic electrostatic forces, neither the atmospheric-pressure polymorphism nor the high-pressure polymorphism appears to fit this pattern. Only an average refractive index * Note added in proof, Recent work with a modified piston design shows that BeF2 coesite and the quartz forms of AlAs04 and GaAs04 persist at pressures up to 100,000 bars. 13 D. M. Roy, Rustum Roy and E. F. Osbobn, Fluoride model systems, III: The system NaF — BeF2 and the polymorphism of Na2BeF4 and BeF2. J. Am. Ceram. Soc. 36 (1953) 185-190. 458 Frank Dachille and Rstm Roy of the new form could be obtained: 1.345. Powder x-ray data are given in Table 2 for comparison with that of the Si02 coesite. BPOt. Another interesting case studied in some detail is BP04, which was known only in the cristobalite structure. In an earlier study14 it was noted that an anomalous possible quartz polymorph of BP04 Table 2 SiO, coesite BeF, coesite hkl Intensities Int.aaiti.. ob« eale Film Diffr. oba calc film Diffr. 020 6.217 6.198 ynrw 5 5.949 5.961 m 3 021 4.40* 4.383 TT* 5 4.19* 4.216 VTW 10 130, 111 3.432 3.439 ■■ 50 3.310 3.3«6 • 30 002, 040, 221 3.098 3.099 TTI 100 2.984 2.981 TTI 100 220,(041) 2.77* 2.772 ■ 15 2.668 2.6*6 «» 13 131 2.68» 2.705 ■ ■ 15 2.605 2.601 W 10 201, 241 2.350 2.343 W 5 2.256 2.233 n 5 112, 150 2.303 2.302 ■ 10 2.216 2.214 n 90 240, 223 2.195 2.191 ■ 10 2.108 2.107 . 60 151, 310, 132 2.034 2.038 ■ ■ 10 1.960 1.960 a* 40 330 1.846 1.848 10 1>76« 1.777 5 261 1.789 1.789 ■ 10 1.723 1.721 10 260, 222,(043) 1.716 1.717 Da 15 1.651 1.633 » 3 113, 352, 17T 1.711 1.703 m 10 1.640 1.638 V : a = c = 6 = Z = {? (meas.) 0 (x-ray)— 7.16 ± .01 A 12.39 A (axial ratio = 1.730) 16 2.93 ± 0.02 2.90 6.88 ± .01 A 11.92 A (axial ratio = 16 2.55 1.732) * Spacings marked by an asterisk were measured less accurately than the rest. Since the materials are pseudohexagonal, many reflections contribute to any one powder line. The reflections listed are those which appear to contribute most to the intensities, as judged by comparison with single crystal photographs, and are listed in order of decreasing contribution. Only the strongest reflection from each zone is listed; is should be remembered that the other reflections in that zone may also contribute. Indices in parentheses are of reflections which might theoretically contribute, but appear to be very weak or absent. appeared to be formed in some of the low-pressure hydro thermal runs. This phase had refractive indices and density incompatible with the expected increases from the cristobalite form. In the present study the authentic quartz form of BP04 was prepared at 50,000 bars and later a p-t curve was determined for its cristobalite-quartz transition which is shown in Fig. 2. Starting materials were 1:1 mixture of boric and phosphoric acids, and also the preformed cristobalite form. Even in the 5—6 mg samples used, very 11 E. C. Shafer, M. W. Shafer and Rustum Roy, Studies of silica structure phases, lis Data on FeP04, FeAs04, MnP04, BP04, A1V04 and others. Z. Kristallogr. 108 (1956) 263-275. High-pressure region of the silica isotypes 459 well-formed crystals with the characteristic habit of doubly terminated quartz were obtained, large enough for single-crystal study. The powder pattern is clearly analogous to that of Si02 quartz, and single-crystal rotation and Weissenberg photographs confirm this in detail. The unit cell with three BP04 molecules has a = 4.470 ± 0,005 A, c = 9.926 ± 0.01 A, with cja = 2.2, which is double the ratio for the quartz unit cell with 3 Si02. The indexed powder patterns and a more detailed discussion of the x-ray crystallography of the BP04 (and BAs04) quartz form are given by Dachille and Glasser15. Refractive indices obtained were Nm = 1.639 ± 0.002, NE = 1.647 ± 0.002. The density is 3.05 ± 0.02 by the sink-float method using a centrifuge. This compares with an x-ray density of 3.069. BAsO±. The quartz form was prepared from the mixed oxides at 40,000 bars and 420°C. A p-t curve was not worked out in detail, but it probably passes through a point as low as 36,000 bars at 450 °C. The samples are white and grind easily to a white powder. A microscopic examination shows that basal and prismatic cleavages are nearer perfect than in Si02 or BP04, and that the material has high birefringence for a quartz structure. Clear uniaxial-positive interference figures are obtained. Refractive indices are Nm — 1.730 ± .002 and Ne = 1.757 ± .002. X-ray powder and single-crystal data are given elsewhere15 so that it will be sufficient to report a = 4.562 ± 0.005 A, c = 10.33 ± 0.01 A. BVOi and AlVO4. The structures of these compounds are of interest because they could serve to set some limits to the fields assigned to quartz- and rutile-like phases in ionic radius-ratio plots of the type of Fig. 3. If radius ratios were the dominant factors, A1V04 should have the rutile structure, but Brandt16, Milligan17 and Shafer, Shafer and Roy14 have shown that this is not the case. The use of pressures up to 50,000 bars, which might be expected to shift the radius ratios further into the rutile field, has failed to produce such a phase. Pressures up to 33,000 bars also failed to produce quartz- or rutile-like phases of BV04. 15 Frank Dachille and L. S. Dent Glasser, High pressure forms of BP04 and BAs04; quartz analogues. Acta Crystallogr. 12 (1959) 820—821. 16 Karin Brandt, X-ray studies on AB04 compounds of rutile type and AB206 compounds of columbite type. Arkiv Kemi, Mineral., Geol., 17 A (1943) 1-8. 17 W. O. Milligan, L. M. Watt and H. H. Rachfobd Jr.,X-ray diffraction studies on heavy metal orthovanadates. J. Physic, and Colloid Chem. 53 (1949) 227-234. 460 Frank Dachille and Rstm Roy High-pressure forms of both these compounds—different from those obtained at atmospheric pressure—appear at pressures above 30,000 bars and 400 °C. The new forms have a metallic character which is more noticeable in the boron compounds. The small sample wafers are black, flexible but with a fibrous fracture, and present some difficulty in grinding because of the tendency of lath-like fragments or 0 Fig. 3. Ionic radius ratio plot of structure fields of ABX4 compounds closely related to some structures of AX2 compounds. Compounds of other structure types are not plotted, but their respective fields helped locate the three shown. Ternary compounds in the fluorite field may have more than one form, and some have "distorted" fluorite structures. flakes to adhere to the surfaces. There are indications that BV04 decomposes to the oxides at temperatures above 4000 C and pressures below 30,000 bars. Further work at higher pressures is in progress on these compounds. AlPOi; GaPOt, FePOi, MnPO4. These four compounds all invert to new forms near the limits of the apparatus, and could not be studied adequately. The p-t relations of these quartz-to-high-pressure forms are shown in Fig. 2. It will be noted that the pressure required for the Ionic radius ratio RA/RX High-pressure region of the silica isotypes 461 appearance of the high-pressure forms increase in the order Mn-Fe-Ga-Al phosphate. This may be a manifestation of the degree of "comfortable fit" in the quartz structure—that is, A1P04 (very definitely in the quartz field according to radius-ratio criteria) will maintain this structure to higher pressures than will MnP04, which is on the border of the quartz field (Fig. 3). X-ray powder patterns and single-crystal studies show that these new forms are not analogues of coesite, and further, that they do not appear to be related to each other. This is the first phase of A1P04 which does not match a phase of Si02. Details of the optical and x-ray crystallography of these phases and some others discussed herein will be reported later18. AlAsOi, GaAsOt, FeAsOv MnAsOt. The atmospheric-pressure quartz forms of A1As04 and GaAs04 remain stable to pressures of 50,000 bars at 400 °C. These conditions likewise failed to alter the non-quartz form of FeAs04 prepared by Shafer, Shafer and Roy 14. The MnAs04 composition as oxides yielded the pyroarsenate as the chief product at these conditions. Si02 (tridymite and cristobalite). Attempts were made to prepare new dense forms which may be related to tridymite and cristobalite, even if such forms would be metastable with respect to quartz or coesite. In our temperature range this proved impossible. In the shortest runs in which any transformation occurred, coesite or quartz were always formed from the tridymite or cristobalite. The results described herein reveal the very interesting possibility of preparing quenchable high-pressure forms of a high proportion of common inorganic compounds. Simple relationships, which might, for instance, be predicted between pressure of transition and radius ratio or ionic charge, do not appear to obtain. Acknowledgement We are indebted to Dr. L. S. Dent Glasser for checking and improving the coesite powder-pattern indexing against her own single-crystal data. This work forms a part of a study in crystal chemistry supported by the Chemical Physics Branch, US. Army Signal Corps, under contract No. SC 71,214 and SC-74,951. 18 L. S. Dent Glasser and Frank Dachille, High pressure forms of silica analogue compounds. In preparation.