Creator:Frank Dachille and Rustum Roy Date Created:April 1960 Place Created:University Park, Pennsylvania Keywords:Olivine-Spinel transition,magnesium system Context:article reprinted from the American Journal of Science ************************************************** [American Journal of Science, Vol. 258, April 1960, P. 225-246] American Journal of Science APRIL 1960 HIGH PRESSURE STUDIES OF THE SYSTEM Mg2Ge04-Mg2Si04 WITH SPECIAL REFERENCE TO THE OLIVINE-SPINEL TRANSITION* FRANK DACHILLE and RUSTUM ROY College of Mineral Industries, Pennsylvania State University, University Park ABSTRACT. The system MgzGeOi-MgsSiOi has been studied to the experimental limits of available hydrothermal and uniaxial high pressure apparatuses. The inversion temperature for the MgsGeO^tpinen-MgaGeOuoiivine) equilibrium is 810°C at atmospheric pressure. The AV of the inversion is 3.5 cc/mole; AH is 3690 ± 180 cal/mole. The inversion temperature is raised by 0.025°C/bar for the first 5500 bars. Infra-red absorption spectra, x-ray intensities and molar refractivities clearly show that MgaGeOj is an inverse spinel. Solid solution between MgaGeOj and Mg«SiO, is complete in the olivine phase at temperatures above that of the inversion in MgiGeO(. The maximum silicate content of the spinel solid solutions at lower temperatures increases steadily with pressure, from 10 mole percent at 700 bars to 50 mole percent at 60,000 bars at 542°C. Extrapolation places the spinel-olivine transition for MgaSiOj at 100,000 ± 15,000 bars. The change in the lattice spacings of the spinel solid solutions of Mga(Ge,Si)04 shows that MgsSiOtopinei) has a cell edge of 8.22A. Therefore the AV for this transition is 2.0 cc/mole. The pressure dependence of the transition in Mg2SiO( is estimated by extrapolation at 0.013°C/bar. Experiments show that substitution of Fe2+ for Mg+2 markedly increases the maximum silicate content of the spinel solid solution under corresponding p-t conditions. The geophysical implication of these results is that an olivine-spinel transition in the mantle of the earth does appear to be a reasonable explanation of the seismic and density discontinuities starting at 400 km. In a paper presented to the Royal Astronomical Society Dr. H. Jeffreys (1936) gave evidence of a discontinuity in the properties of the earth some hundreds of kilometers beneath the surface. He reported that at a depth of 480 ± 20 kilometers a change could be measured in the velocities of seismic waves. Specifically P-waves above this depth had a velocity of 9.08 km/sec, but below this depth the velocity was 9.8 km/sec. Calculations by K. E. Bullen (1936) for the density distribution 'in the earth also suggested a nearly discontinuous change in density at a depth of 300-400 km. Since no known materials could fit this data Dr. J. D. Bernal, in attendance at the meeting, was asked if it was likely that an olivine-like material could be converted to a new state at great pressures. A direct quotation from Observatory (1936) will serve to present important crystal chemical concepts in the paraphrase of Dr. Bernal's discussion: Dr. J. D. Bernal said at ordinary pressures olivine, which is magnesium ortho-silicate with part of the magnesium replaced by ferrous iron, is a hexagonal lattice of oxygen atoms in which the silicon and metals occupy the cavities, somewhat unsymmetrically. It appeared possible that when the lattice was much compressed the cavities might become too small to hold the silicon atoms, and a different * Contribution No. 58-117, The Department of Geophysics and Geochemistry, The Pennsylvania State University, University Park, Pennsylvania. INTRODUCTION 225 226 Frank Dachille and Rusturn Roy structure would have to be found. The structure of a crystal was largely a matter of the sizes of the atoms, the cavities having to be neither too large nor too small to hold them. Thus the effect of compressing the lattice as a whole would be similar to that of inserting a larger atom, such as a germanium in place of silicon. Magnesium germanate had been studied by Goldschmidt and found to exist at ordinary pressures in two forms, one isomorphous with olivine and the other cubic. The latter is about 9 percent denser, and therefore is the high pressure form. By-analogy it therefore seemed probable that at high pressure olivine would adopt a cubic form. The change of density suggested by the germanate would be of the right order of magnitude. The reference to V. M. Goldschmidt introduces a series of developments having to do with the reality of the transition noted by Goldschmidt (1931). In a footnote in this paper it was stated that magnesium orthogermanate, Mg2Ge04, was prepared in a spinel-like form. A year later Jander and Stamm (1932) placed the inversion temperature for the spinel-olivine polymorphs at 1063°C. Incidental to work on synthetic clay phases, Roy and Roy (1954) prepared reproducibly under high water pressures the Mg2Ge04 as euhedral spinel-structure crystals. Aware of the change in density of the polymorphs as reported by Bernal and restated by B. Mason (1952), Roy and Roy determined fairly accurately the volume change as 8.3 percent of the spinel volume. They gave the inversion temperature as 1005°C using dry techniques. This work has in general been overlooked since it appeared with much other material on a different subject. Subsequent to this time four reports appeared with the claim that the spinel polymorph of Mg2Ge04 could not be made. Thus Urey (1952) says . . . the observations of Goldschmidt were very approximate and it has proven impossible to prepare the cubic modification in this laboratory. Therefore, Bernal's suggestion rests only on pure hypothesis and it seems desirable to explore the possibility that iron phase is present in the mantle. Similarly Ringwood (1956) says The evidence . . . indicates that the olivine structure is probably thermodynamical-ly stable above 600° C. Accordingly the author would agree with Romeijn that Jander and Stamm were probably wrong about the transition of MgzGeO, at 1065 °C. It is of interest to note that although Romeijn (1953) had to conclude that the dimorphism did not exist, he did acknowledge that the absence of dimorphism in Mg2Ge04 was a serious flaw in his analysis of spinel structures. E. F. Bertaut in his early work with germanate spinels (1954 and personal communications) was likewise unable to make the spinel form. The location with respect to depth of the seismic discontinuity reported by Jeffreys has also been changed with subsequent investigations. The current belief is that two discontinuities exist, one at 413 km and one at 984 km. Mason (1952) summarizes the general implications of the hypothesis of solid phase changes in this manner: . . . olivine would begin to change to a higher-density polymorph at a depth of 413 km, and the change would become complete at a depth of 964 km. The change would be gradual through the intervening region, the transition depending on temperature, pressure and the magnesium-iron ratio in the olivine, all of which probably vary with depth. A polymorph change of this sort within the mantle also provides a mechanism for periodic orogeny; an increase in the amount of the denser polymorph during geological time would result in contraction of the earth as a whole, thus giving a plausible explanation for crustal shortening and orogeny. All this is highly speculative, however. As Adams (1947) remarked, speculation of this kind point up the desirability of devising means for the investigation of High Pressure Studies of the System MgGeMgSiO,227 phase changes in silicates at pressures of the order of 10E atm and temperatures above 1,000°. Obviously a direct experimental attack on the problem of a transition in natural olivine was not possible with known and available techniques and apparatus. However, it was decided that a fruitful and feasible approach to the problem would be a study of the system Mg2Ge04-Mg2Si04, with emphasis on solid solubility and on the polymorphic transitions of the olivine to the spinel type with a view to obtaining reliable data on the Mg2Si04 by extrapolation. The essence of the experimental approach and some results, including the effect of the substitution of some Fe2+ for Mg2+ were reported to the Geological Society of America, Minneapolis meeting (1956). Additional work concerned with some modification and calibration of high pressure apparatus was reported to the same group at the Atlantic City meeting (1957), and at the St. Louis meeting (1958) the results of work on the system Mg2Ge04-Mg2Si04 at pressures up to 60 kilobars were presented. EXPERIMENTAL PROCEDURES Work was begun on the system Mg2Ge04-Mg2Si04 using the hydrothermal techniques necessary to catalyze reaction in the solid state. Details of the basic apparatus for these techniques are given by Roy and Tuttle (1956) and Roy and Osborn (1952). The hydrothermal bombs may be used to provide desired and controlled pressure and temperature conditions on prepared samples. The practical upper limits of pressure are 60.000-70,000 psi and of temperature 900-1,000°C, but these limits are not coincident. The higher pressures lower considerably the usable temperature limits. A second major piece of apparatus necessary in the investigations is the uniaxial pressure device in which pressures can be attained an order of magnitude greater than those feasible in the test tube bomb apparatus. Generally the unit is a modification by Tuttle and Harker in our laboratory on designs of Bridgman along the lines of Griggs and Kennedy (1956). A description of equipment and procedures is given elsewhere (Dachille and Roy, in press a). In this type of device a large thrust is developed by a hydraulic ram which is intensified and transmitted into a small sample by the use of very hard small piston faces. These pistons have been made of cemented carbides, polycrystalline alumina or mullite, or of special hardenable steels. Directed pressure, that is, total thrust divided by reduced area, may approach 900,000 to 1,000,000 psi, but the effective coincident temperature limits fall very rapidly to about 600°C at these pressures. Pressure to the "jack" is supplied and automatically controlled by means of a Foxboro controller and Aminco oil pump. Temperatures are controlled by Honeywell millivoltmeter instruments. Many techniques for experiments at these pressures were worked out during this and allied studies with a view to studying reversible equilibria rather than only preparing phases. Thus in general we have made efforts to make runs from 1 day to 2 weeks long rather than several minutes to a few hours. It is perhaps accurate to say that no similar work has been described either with respect to range or detail other than that of MacDonald (1956). The petrographic microscope and the x-ray diffractometer were the primary analytical tools, the latter being much more useful to give not only the 228 Frank Dachille and Rustu Roy phases present but their composition from the spacings. A Perkin-Elmer Double Beam Model 21 infra red spectrometer was used to obtain absorption spectra in series of compounds with systematic substitution of one cation for another to indicate the coordination of the ion in the spinel phases. PREPARATION OF MIXTURES The method of preparation of most of the mixtures in this study was the direct mulling of the finely powdered oxides under absolute alcohol, then heating for 2-3 hours at 400°C. The magnesia was C. P. Reagent grade of Baker and the germania was the soluble form of Ge02 prepared by the Eagle-Picher Company. Silicon dioxide was used in the form of silicic acid, the water content of which was determined by dehydration at 1350-1400°C. A few mixtures used were prepared by the nitrate method and by the method of coprecipitation of germanium and silicon hydrous gels from solutions of their tetrachlorides (Roy, 1956). No significant advantages were found with these latter methods so that systematic work was restricted to mixtures prepared from the oxides. Table 1 X-ray data on Mg2Ge04 and Mg2Si04 spinel and olivine polymorphs SPINEL OLIVINE MgsGeOi MgaSiOi MgnGeOi Mg,SiO( ** hkl d(A) I (est.) d(A) (extrap.) hkl d(A) I(est.) * 111 4.77 40 4.74 020 5.15 20 5.11 26 220 2.92 30 2.907 110 4.44 40 311 2.492 100 2.480 021 3.91 80 3.88 69 222 2.384 2 2.374 101 3.81 10 3.73 25 320 2.292 2 2.280 111,120 3.57 10 3.487 21 400 2.066 10 2.056 121 3.06 10 3.000 17 330 1.947 5 1.938 130 2.619 20 2.768 53 331 1.896 10 1.886 131 2.546 100 2.513 73 422 1.687 50 1.678 112 2.488 90 2.458 100 333 1.590 50 1.583 041 2.364 10 2.348 9 * *** 210 2.390 2.316 9 a = 8.255A 8.22A 122 2.295 10 2.268 59 b = " 140 2.279 2 2.250 33 c = »* 211 2.220 20 2.161 15 v = 562.5A3 555.4 132 2.056 2 2.034 5 z = 8 8 042 1.956 2 1.945 4 150 1.898 2 1.878 5 113 1.825 1.811 2 151 1.810 1.792 3 222 1.783 20 1.748 60 241 1.730 2 1.670 13 061 1.652 5 1.636 12 133 1.632 5 1.618 15 152 1.603 5 1.589 2 043 1.581 4.915A 10.295 6.020 10 1.572 4.76A 10.20 5.99 10 304.6(609.2) 290.8(581.6) 4(8) 4(8) * Roy and Roy (1954) * * Swanson and Tatge NBS Circular 539, v. 1 *** Extrapolated from figure 2. High Pressure Studies of the System MgGe-MgSiOi 229 In the course of the investigation starting compounds were prepared by the hydrothermal treatment or even dry fusion of the oxide mixtures. Compounds so prepared were then used to test a criterion of reversibility in transitions or to seek clear evidence of exsolution and change in solid solubility under various conditions of pressure and temperatures. RESULTS Data on the phases encountered.—In table 1 are recorded the characteristic x-ray diffraction data for the olivine and spinel polymorphs of Mg2Ge04. Data for the olivine phase of Mg2Si04, forsterite. are included to permit a direct comparison with the germanate isomorph. Further x-ray data are given in figures 1 and 2 to demonstrate the effect of replacing some of the ger- Fig. 1. System MgaGeO^MgjSiOi. Change of 20 values with composition of the olivine polymorph. 230 Frank Dachille and Rustu Roy Fig. 2. System Mg2Ge01JMg2Si0<. Change of 20 values with composition of the spinel polymorph. manium with silicon in the olivine and spinel phases respectively. The shifts in these "d-spacings" approach the ideal Vegard relation for substitution in solid solution. They will permit a comparison of volume change between the olivine and spinel polymorphs up to almost 50 percent silicate content, the upper limit of the spinel solid solution prepared. Extrapolation of the d-spacings, figure 2, show that the spinel phase of Mg2Si04 will have a cell edge of 8.22 A and a unit cell volume of 555.4 A3. This volume is only 4.7 percent smaller than that of the olivine. Hence the volume change in the silicate end number is only 56 percent of that found in the germanate. Although such an extrapolation is considerably more reliable than various guesses in the past, the possibility of strong deviations from ideality in spinel solid solutions cannot be completely ignored. The transition in Mg^GeOt.-—The end member Mg2Ge04 provided an example for a direct study of the pressure-temperature dependency of the olivine-spinel inversion within limits set by the hydrothermal apparatus. Clean cut and easy reversibility of the transition was achieved under hydrothermal conditions (see table 2). At 10,000 psi the transition occurs at 823°C and the use of higher pressures raises the temperature rapidly to the point at which bomb damage or failure occur frequently. Nevertheless a number of runs were made at higher temperatures and pressures in order to follow this dependence. Similarly equipment problems limited the exploration by the uniaxial devices. It was possible however to make a number of runs at pressures and temperatures approaching 1,000°C and 100,000 psi in this equipment by using pistons made of polycrystalline A1203 or of mullite. Within these limits the pistons held up remarkably well with no distortion of the circular faces bearing the maximum stress. Unfortunately these surfaces were subject to pitting or flaking when they were separated to recover the sample wafer so that regrinding of the pistons was necessary after each run. Further, the pistons, which were made of rods 4 inches by 1 inch in diameter, on continued use did fail by shearing and High Pressure Studies of the System MgGe,-MgiO 231 splitting at pressures considerably below 100,000 psi. The information from these experiments is collected in table 2 and the pressure-temperature dependency deduced therefrom is indicated in figure 3. From the slope of this line and the measured difference in volume of the two polymorphs at some selected temperature, it is possible to calculate from the Clapeyron equation a change in enthalpy for the transition. The AH of transition obtained (olivine-spinel) is 19.9 cal/gm or 3,690 cal/mole with an error of ± 5 percent. These values will be reviewed later in the discussion. Fig. 3. Univariant p-t curve for the spinel-olivine transition in MgzGeOj. (Partially filled circles indicate persistence or formation of olivine. Curly division shows runs in which pistons fractured after start of spinel phase.) Mg2Ge04 MOL'/. MjjSiO, Fig. 4. System MgaGe0i-Mg2Si01—Isobaric section at 10,000 psi water pressure. 232 Frank Dachille and Rustum Roy Table 2 Data on the system Mg2Ge04-Mg2Si04 Pressure Dura-Run (°C) 1000 tion Composition (No.) Temp. psi Hrs. Phase MgaGeO* (Hydro- thermal) Oxides 7043 925 jj 7153 833 »» 7076 830 7081 828 11 7088 822 11 7089 801 »t 7094 808 7096 813 »» 7107 830 >» 7146 873 n 7148 864 Spinel Form 7135 865 Olivine Form 7136 865 Spinel Form 7028 820 7029 825 »» »» 7030 845 >» »» 7031 875 »» »» 7133 845 Olivine Form 7134 845 (Uniaxial) Oxides 7437 925 »» 7441 1088 »» 7444 906 » 7445 903 »» 7451 913 7454 938 Spinel Form 7457 958 » »» 7458 950 » »» 7460 918 s> 99 7485 925 n v» 7479 935 Olivine Form 7489 930 >» »» 7500 868 » 7501 863 Oxides + olivine seeds 7524 927 »» »» »» n 7529 934 » »» n 99 7532 910 (Hydrothermal) 5mol% MgaSiO* 7016 548 >» 91 7121 623 J» »> 7108 626 11 >» 7095 640 19 99 7021 670 19 99 7056 745 5J 99 7078 812 5 15 Olivine 10 70 " 10 70 Spinel & ol. 10 70 Sp. + trace ol. 10 120 Sp. 1 200 Sp. 1 44 Sp. 1 90 Sp. + trace ol. 26 70 Sp. 30 40 Ol. 39 40 Sp. + trace ol. 39 40 Sp. 39 40 Sp. + minor ol. 10 20 Sp. 10 40 Sp. + trace ol. 10 44 Sp. + minor ol. 10 40 Ol. 37 40 Sp. 37 40 Sp. + trace ol. 150 1 Split. Spinel, MgO, Ge02 ru. 90 % Split, trace ol. MgO, GeOa 69 6 Dry. MgO, Ge02 ru. 73 24 Sp. + trace Ge02, MgO 57 24 Sp. + ol. 65 40 Trace sp. + ol. + oxides 78 40 Sp. + ol. trace 36 2 Split. + ol. tr. 72 60 Sp. 62 40 Sp. + tr. ol. 62 40 11 _j_ 11 11 62 40 Spinel 32 40 Olivine 32 40 n 61 40 Sp. + tr. ol. 52 60 " -fol. 50 40 " + tr. ol. 10 47 Sp. + tr. ol. 10 80 10 40 10 96 »» 10 48 « ii ii 10 200 " + minor ol. 10 96 01. + tr. sp. High Pressure Studies oj the System Mg2GeOt~Mg2SiOj, 233 Table 2 (Continued) Pressure Dura-Run (°C) 1000 tion Composition (No.) Temp. psi Hrs. Phase 10% 7091 521 10 80 Sp. + serp. 11 »> 7583 529 10 72 " + tr. serp. 11 J» 7017 548 10 47 " + " ol. 7087 595 10 70 " + minor ol. 11 11 7079 814 10 96 Ol. 7063 821 10 72 " 24,8% 7018 548 10 47 Sp. + ol. 11 11 7057 750 10 240 ii 11 11 7064 772 10 70 Ol. 30% ^ 7164 665 10 90 H 7073 712 10 48 " + minor sp. 47% 7053 650 10 30 " + tr. sp. 11 11 7003 651 10 48 " + " " Ii 91 7023 670 10 48 11 11 11 53% 7554 613 10 70 11 »> 11 70% ;; 7161 503 10 120 " + " talc. 7583 529 10 72 " + " sp. & talc. 5 mol% MgsSiO, 7109 626 55 48 Sp. 10 7105 565 55 40 " 10 7104 584 54 72 " + tr. ol. 24.7 7155 745 67 10 " minor ol. 24.7 7157 796 52 29 01. + minor sp. 30 7156 745 67 10 ii ii ii 30 7158 796 52 29 ii 47 7162 737 54 48 47 7160 770 54 9 " 70 7183 652 45 24 ii 70 7184 595 53 72 " (Uniaxial) Oxides 10 mol% MgaSiOj 7548 543 122 70 ii 18 7573 540 282 21 19 18 7570 547 196 14 " + tr. ol. 18 7568 605 200 24 »» »> »> 24.7 7546 542 260 44 11 11 11 24.7 7530 542 450 48 " + " serp. 24.7 7520 570 570 46 24.7 7541 546 203 25 " + minor ol. 24.7 7476 560 350 24 (3/8" sample) 30 7563 537 515 23 " + tr. ol. 30 7569 545 553 16 30 7537 547 405 48 »» » »» 38 7571 537 740 12 38 7559 538 570 22 " + minor(Ge02) (rut.) 47 7562 530 750 11 " + tr. (GeO.,rut) 47 7567 530 849 6 " +tr. " tr. talc. 47 7527 541 585 48 " -(- minor ol. 47 7468 545 670 22 » " »» 47 7561 622 190 22 »» _; _ »» »» 53 5797 542 780 5 " + " 53 75,34 562 408 32 " +20% " 70 7596 536 950 3y2 " + ol. (3/8" s.) 80 7579 530 200 24 No H20. No reaction 80 7555 538 100 23 01. 80 7550 540 311 20 " + minor serp. 234 Frank Dachille and Rustu Roy The system Mg^GeO^-MgsSiOi.—Data from critical runs in the hydro-thermal and uniaxial pressure units are collected in table 2. From these data the phase diagrams shown in figures 4, 5. 6, and 7 have been constructed. The dashed portions in these figures indicate regions not accessible to the known and available techniques. However, the two rather complete isobars at 10,000 and 55,000 psi and the isotherm at 542° (fig. 8) indicate a self consistency in the pressure-temperature-composition relations of the system. Figure 9 is a three dimensional p-t-x diagram of the system Mg2Ge04—Mg2Si04 constructed from the data. IOOO 500 400 0 10 20 30 40 50 6 0 70 8 0 90 100 M0L% Mg2Si04 Fig. 5. System Mg3Ge0i-Mg2Si04—Isobaric section at 55,000 psi water pressure. 1000 900 400 0 10 20 30 40 50 60 70 80 90 100 MOL % M92Si04 Fig. 6. System Mg«Ge04-Mg2Si04—Isobaric section at 200,000psi (uniaxial pressure). High Pressure Studies of the System MgGeg2SiO 235 0 10 Z0 30 40 50 60 70 80 90 100 MOL V. M9,SiO, Fig. 7. System MgaGeOi-MgaSiOi—Isobaric section at 580,000psi (uniaxialpressure). Fig. 8. System MgaGe04-Mg2Si0i—Isothermal section at 542°C (uniaxial pressure). 236 Frank Dachille and Rustu Roy Substitution of Fe2+ for Afg*+.—The term olivine has the very definite connotation in mineralogy of an iron bearing magnesium orthosilicate. In fact, the word is applied as a group name for the solid solution series between forsterite, Mg2Si04 and fayalite, Fe2Si04. The importance of Fe2+ as an additional component in the system under study was recognized early in the experimental work. A number of runs were made with starting mixtures in which a portion of the Mg2 + was replaced by Fe2+. Fayalite compositions also were used in seeking a direct inversion of fayalite to a spinel-like form. The problem of maintaining the oxidation state of Fe2+ was tackled by the use of ferrous oxalate in the starting mixture, and the addition of powdered metallic iron. In the uniaxial pressure device the samples in some runs were encased in pure iron rings and disks. Data for this phase of the investigation are in table 3. Reaction products were often rather difficult to identify unambiguously so that working out phase boundaries in detail by phase identification would be a tedious job even in a study directed specifically at this system. However, the data do show plainly that in any isobaric section the spinel solid solution area is appreciably larger than that in the completely magnesian system. Increasing the Fe2+ content broadens further the spinel solid solution field so that one might expect the fayalite inversion pressure to be relatively low. The poor spinel-like x-ray pattern obtained on a fayalite composition at about 500,000 psi may very well be a confirmation of this observation.1 However, this could well be a solid solution between Fe2Si04 and Fe304. Quantitative work in this field was postponed in view of the elaborate arrangements which would be necessary to control the valence of the iron. Indeed it is not even sure whether Fe2 + and Ge4+ could exist in equilibrium (Muan, personal communication) although Bertaut claims to have made Fe2Ge04. Routine chemical analysis of 1 A. E. Ringwood subsequently appears to have confirmed this finding in his brief note (1958a) on the formation of a spinel form of Fe3Si0i at 400° C and 45 kilobars. A more complete description of his results in this area appeared recently (1958b). High Pressure Studies of the System M gsGeO ,,-M ggSiOi 237 Table 3 Data for the system Mg»Ge04-Mg2S i 0 4 with Fe substitution for Mg <°C) Composition Run Tem- 1000 psi Duration Mg/Fe Ge/Si No. perature Pressure (hrs.) Phases 9:1 7:3 7169 715 10 60 01. + Sp. + ? 9:1 7:3 7170 737 10 60 Added Fe. Less sp. »» 10:0 7171 699 10 80 Sp. complete SP. »» 7:3 7172 813 10 40 Added Fe. 01., sp. " 7173 915 10 20 Olivine »J 7174 544 10 60 Olivine, Serp. J» 7175 880 10 40 Trace sp. »» 7176 880 10 40 Added Fe. 01. 9:1 1:1 7177 704 10 40 " sp. 9:1 1:1 7179 745 10 40 Excess Fe. Sp. 9:1 7:3 7181 650 180 190 Oxides + trace sp. 9:1 85:15 7182 720 10 20 Sp. + ol. 9:1 85:15 7185 633 10 90 All spinel 9:1 7:3 7186 628 170 160 Sp. + tr. ol. Hematite, serp. 8:2 7:3 7187 655 11 70 Sp. + minor ol. 8:2 7:3 7188 592 11 70 Sp. + trace ol. 8:2 7:3 7189 744) 11 70 Sp. + ol. 8:2 7:3 7190 600 180 190 " hematite 8:2 7:3 7225 530 250 70 II 91 8:2 7:3 7229 520 350 90 " carbonates 9:1 1:1 7230 540 350 90 No reaction Natural Olivine 7236 515 250 250 Ol. + serp. + en- statite 2FeO. Si02 7249 542 180 70 Oxides »» 7391 550 720 40 Cubic phase, green these very small samples was not attempted in view of the problem of the added iron or contamination from the iron washers. DISCUSSION Effect of water.—All through the investigation water was used, and considered, strictly as a catalytic agent. In a few runs, however, it did enter into the composition of the products with the formation of small amounts of germanium-bearing serpentines and talcs together with either the olivine or the spinel phases. Since the main interest was in the phase relations of the olivine-spinel polymorphs conditions leading to the production of hydrous phases were avoided whenever possible. Hydrous phases appeared at temperatures below 520-523° in the lower pressure studies, but also appeared at temperatures close to 542° at the higher pressures. The latter appearances are somewhat suspect because of the possibility of the hydrous phases forming while quenching the runs, but they cannot be dismissed without further systematic study. It is interesting to note that the stability of the serpentine appears to be affected so little by pressure changes from Roy and Roy's value of 520°C at 1,000 atmospheres to nearly 60,000 atmospheres in this work. Evaluation of the catalytic effect of water in this system can be only qualitative because experiments were not set up to study rate phenomena. It has been pointed out that the Mg2Ge04 spinel-olivine reaction was found to take place only above 1,005°C in a dry 238 Frank Dachille and Rustu Roy system by Roy and Roy (1954). Extrapolating the hydrothermally determined pressure-temperature dependency of the transition back to atmospheric pressure the present data would place the transition near 810°C. An empirical observation may be made concerning a measure of the "catalytic effect" of water in this reaction. The presence of water enables it to proceed at a rate so that essentially complete conversion may be achieved at a temperature at least 200°C below the equivalent temperature in the absence of water. In this reaction the lower limit is set by the stable equilibrium temperature. In other cases such as the exsolution of A120:! from spinel solid solutions this catalytic effect in terms of temperature would be 400-500°C. An evaluation of the importance of water-catalysis in terms of time is gained from the following: A reaction of the dry oxides plus water will yield the spinel and forsterite phases in a matter of a few hours in the uniaxial devices. In the total absence of water, two weeks reaction time at 500,000 psi and 450°C failed to react the oxides, although the Ge02 did go to the rutile form. Numerous observations of this type were made on many reactions in the uniaxial devices which did not go to completion because of loss of added water due to splitting of the reaction capsule or to cracking of the piston surfaces. Thus it was observed that even at 880°C and 45,000 psi reaction which apparently had started the spinel phase growth of Mg2Ge04 was interrupted after a few minutes by the cracking of a piston face with the presumable loss of water. The pressure and the high temperature were maintained on the sample at least 10 hours without obtaining complete reaction. THE NATURE OF PRESSURE IN UNIAXIAL DEVICES Many observations in this and related studies give support to the interpretation that the pressures calculated, or at least their effects, are very close to those of hydrostatic systems. Considerable work was done expressly to evaluate the nature of the pressure (Dachille, Shafer, Roy, in preparation) but only an outline is given here: 1. Field boundaries in the system Mg2Ge04-Mg2Si04 as determined in a purely hydrostatic run, kindly made for us by Dr. H. S. Yoder, at 550° ± 5°C and 10,000 ± 50 bars water pressure agree with those determined in the uniaxial devices. 2. The formation of the CS2 "polymer" within 10 percent of the 42,000 bars reported by Bridgman (1941). 3. Uniformity of pressure and reproducibility were studied in the Si02 quartz-coesite reaction by varying the sample area-metal area ratios and by determining the p-t dependence of this reaction in the 350-650° interval. This dependence was found to be essentially the same as that reported by MacDonald who worked with a sample system entirely omitting a containing ring. The transitions in MgnGeOj, and Mg2Si04.—The reality and reversibility of the olivine-spinel transition (contrary to the assertion of Bertaut, 1954, 1956) has been amply demonstrated and the equilibrium temperature at atmospheric pressure set at 810°C. In figure 3 the pressure-temperature dependence High Pressure Studies of the System MgGe-Mg2Si 239 of the transition in Mg2GeO, is shown, obtained as a composite of runs from both the hydrothermal and uniaxial pressure systems. The slope of this line is equivalent to the reciprocal of 24.6°C per kilobar with an uncertainty of 1.2°C per kilobar. A precise calculation of the AH of transition by use of the Clapey-ron equation requires thermal expansion (and compressibility) data of both polymorphs at the inversion. The data are not available and therefore the AH has been calculated with the assumption that the volume difference at room temperature applies for the transition at one atmosphere and 810°C. The error in the value caused by neglecting the Aa term is unlikely to be more than 1 percent of the total AV term since thermal expansion coefficients of similar materials are about 100 X 10~7/°C. By far the largest error is involved in the graphical determination of the slope. The pressure-temperature dependence in the transition of the silicate end member is of geological interest but remains beyond direct experimental determination at present. An estimate may be made considering some aspects of the system Mg2Ge04-Mg2Si04. In various oversimplified calculations the slope may be assumed to be the same as that of the germanate end member. Therefore in the Clapeyron equation dt/dp = TAV/AH an identity of dt/dp for the germanate and silicate end members requires that the AV/AH ratios are the same for the same absolute temperature. However, the isotherm of figure 8 and the p-t-x of figure 9 suggest that at the pressures at which the transition in the silicate first appears possible the temperature is more than 2,000°C below that of the germanate member. If this difference were only 1,000°C the absolute temperatures of the transitions at 100 kilobars will be nearly 800°K and 1800°K for the silicate and germanate members respectively. The value of dt/dp for the silicate will be correspondingly higher unless the AV/AH value increases in the proportion of 1800/800. The AV of the silicate transition is only 56 percent of that found in the germanate. Therefore its AH value would have to be only about 25 percent that of the former, illustrating the danger of making simple assumptions. This is in agreement with the work of Majumdar and Roy (1958) which shows there are no data to justify assuming equality or any simple relation between AH, AV, or dp/dt even for analogous transitions between dimorphic substances, removing any justification that values from the Clapeyron equation can be extrapolated from one isotype to another. The observation by Mason (1952) that the unit cell of spinel Al2Mg04 was about 9 percent smaller than the corresponding Mg2Si04 olivine unit encouraged the above assumptions, but it appears that in addition to the above objection an interesting detail of crystal structure intervenes. Later in the discussion it will be shown that the germanate spinel is of the inverse type with Ge4+ in octahedral sites whereas in the silicate spinel 5i4+ is probably in tetrahedral sites. However, in both olivine structures the 4+ cations are in tetrahedral sites. Therefore, it is reasonable to expect that Ge1+ changing from 4 to 6 coordination will produce a greater volume decrease than found in the silicate member where the coordination is unchanged.2 3 The smaller unit cell of the AlJVfgOt may mean that in this spinel the Al3t is in fact in octahedral sites in agreement with neutron diffraction confirmation that AMVIgOi is a normal spinel. 240 Frank Dachille and Rustu Roy In view of the serious limitations of extrapolation of the Clapeyron relations an evaluation of dt/dp of the silicate transition directly by graphical means may be more accurate. Isotherms at 600, 660 and 730°C for the spinel-spinel + olivine boundary from the data and construction of the four isobaric sections are plotted on figure 10 on a logarithmic scale. The isotherm at 542° ipoopoo lopoo Mg2Ge04 10 MOL % 100 MgjSiO, Fig. 10. System MgaGeOvMgsSiOj. Logarithmic plot of isotherms constructed from data of the t-x and p-x sections. LOGARITHMIC PLOT OF ISOTHERMS 100,000 for the same boundary also is plotted to include the results at higher experimental pressures. From the results of graphical extrapolations to 100 percent the values of dt/dp obtained are seen to decrease from 24.6°C per kilobar for the germanage member to about 13°C per kilobar for the silicate member. High Pressure Studies of the System MgGe-Mg2SiOi 241 The experimental work on the substitution of some Fe2 + in the system does no more than indicate that the spinel phase is stable to higher temperatures (or lower pressures) than in the iron free systems. The effect of fayalite solid solution on dt/dp cannot be determined from our data, but for substitutions up to 20 mol percent an assumption is made that it is the same as that of the magnesian member. From these data and the few runs on the Fe2Si04 composition one can only confirm that which is easily guessed that the presence of Fe2+ in the olivine will lower the pressure necessary for its transition to a spinel. Certain geophysical implications of the data.—A consideration of the geophysical aspects of the results is summarized in figure 11. On this figure are Fig. 11. Pressure, temperature and p-wave velocity relations for a portion of the mantle on which are superimposed possible p-t relations of Mg2Si0j. shown for a portion of the earth 1) the variation of the P-wave velocity 2) the pressure-temperature relation of Turner and Verhoogen (1951) 3) the pressure-temperature relation of Daly (1943). The two p-t lines representing the limits of the uncertainty of the p-t dependence of the Mg2Si04 transition, have slopes of nearly 25 and 13°C per kilobar and are plotted passing through 542° and 100 kilobars (300 kilometers depth). An inspection of figure 11 shows that a transition of olivine to spinel could be the determining factor in the second order discontinuity just below 400 km depth. If the seismic discontinuity is accurately placed at 413 km depth and if the composition of the mantle is essentially forsterite with up to 10-20 mol percent fayalite, the main considerations then are the dt/dp of the transition and the p-t curve of the earth. It is clear that using either of the widely divergent earth p-t curves, one can reconcile the seismic discontinuity "reasonably" with the transition. The 242 Frank Dachille and Rustu Roy general agreement is certainly in the right order of magnitude since the experiments conclusively rule out 10 or 1.000 kilobars as the transition pressure at say 1,000°C. The problem to determine directly the p-t relation of the transition for forsterite in order to narrow the uncertainty of the extrapolation made here remains urgent. These refinements will then permit a closer examination of the temperature variation in the earth. CRYSTAL CHEMICAL OBSERVATIONS In work from this laboratory (Dachille and Roy, 1959) it has been shown that in the absence of the complete assignment of absorption frequencies in Cm-1 3000 2000 1500 1200 1000_800 700 WAVELENGTH- MICRONS Fig. 12. Infrared absorption spectra. a. MgsGeOi spinel form b. Mg2(Ge,Si)Oi spinel form (25 mol% silicate) c. Mg2(Ge,Si)04 spinel form (38 mol% silicate) d. Mg2(Ge,Si)04 spinel form (47 mol% silicate) High Pressure Studies of the System MgGe-MgSiO, 243 Cm-1 WAVELENGTH —MICRONS Fig. 13. Infrared absorption spectra e. Mg«GeO< olivine form f. Mg>(Ge,Sd) 0« olivine form (10 mol% silicate) g. Mga(Ge,Si)04 olivine form (25 mol% silicate) h. Mg»(Ge,Si)01 olivine form (47 mol% silicate) complex solids one can use the main absorption frequencies in the infra-red region to indicate primary coordination changes of the cations. Naturally the present substances presented excellent examples for checking such ideas. Figures 12, 13, and 14 present the infra-red absorption patterns for some of the spinel and olivine structure solid solutions. The following deductions are pertinent: 1. In MgoGeO.i spinel all the Ge4+ is in 6 c.n. (cf. Ge-0 absorption in Ge02 qtz at 11.5 and in Ge02 rutiie at 14.0). 2. In both the olivine forms the Si and Ge ions appear to be of two types with 244 Frank Dachille and Rustu Roy Cm-1 3000 2000 1500 120 0 1000_800 700 WAVELENGTH- MICRONS Fig. 14. Infrared absorption spectra i. Mgs(Ge,Si)0i olivine form (90 mol% silicate) j. MgaSiOi olivine form appreciably different Si-0 (Ge-O) distances. Indeed the patterns themselves suggest the highly unlikely possibility of an inverse olivine structure. 3. Although this is not clearcut, the spinel solution towards the silicate member appears to be changing from an inverse to a normal spinel. Unequivocal confirmation of the first deduction that Mg2Ge04 is an inverse spinel has been obtained from the x-ray intensities. In 1956 Durif-Varambon, Bertaut and Pauthenet came to precisely the opposite conclusion on the basis of intensities measured from a very poorly crystallized sample. They calculated that a distinctive criterion to judge between the various spinel arrangements is the ratios of the intensities of the 422 and 400 reflections. The following values were calculated: j422 • = For Normal, 1.67; For Inverse, 6.6; For Random, 3.3 ■Uoo Their conclusion was based on their measurement of the intensity ratio as 1.7. On examination of the ratio in random x-ray diffractometer mounts it was immediately noticed that the ratio was consistently very much higher than 1.7; indeed it never fell below 3.8 for the germanate end member, and values as high as 6.5 were found. Obviously it had not been possible to avoid all orientation in the slide preparations although the crystal size was extremely small. Since the more likely cubic habit or cleavage would tend to increase the 400 intensity in relation to 422 the true ratio -1- is seen to approach if not equal 1400 the 6.6 figure for the inverse spinel. High Pressure Studies of the System Mg;Ge(),,-Mg2SiOi, 245 With increasing Si4+ content there is a shift to lower values of the intensity ratios which suggests that as the Si4+ takes more tetrahedral positions the structure approaches the normal spinel. Molar refractivity (Rm in cc/mole), table 4, affords another line of evidence on cation coordination. (SafTord and Silverman, 1947; Roy, 1950; Dachille and Roy, in press b). The "measured" Rm of the olivine forms of Mg2Si04 and Mg2Ge04 are 15.87 and 18.33 respectively, and each value is 0.4 less than the value calculated from the "ideal" values for component oxides. Similar small differences have been noted for binary and ternary compounds. In the olivines the Mg ions are in VI, and the Si.Ge ions are in IV coordination. The measured Rm for Mg2Ge04 spinel (n = 1.768) is 17.62, a value which is 1.09 less than that calculated for the normal spinel and 0.59 less than for the inverse. This latter value is comparable with the usual differences between observed and calculated values and suggests an inverse spinel Mg2Ge04. It can then be shown that the relative refractivity contribution of Ge02 in the inverse spinel is about 14 percent less than that in a normal spinel, which difference is close to that found between the rutile and quartz forms of Ge02. Table 4 Molar Refractivities, Rm Substance Measured Calculated Difference MgaSiO, olivine 15.87 16.27 .40 MgaGeO* olivine 18.33 18.71 .38 MgsGeOi inverse spinel 17.62 18.21 .59 MgaGeO. normal spinel 17.62 18.71 1.09 Values used for oxides (see Refs. cited for cation coordination) Coordination VI IV MgO 4.54 5.18 SiOa — 7.19 GeOa 8.40 9.63 ACKNOWLEDGMENTS Financial assistance for the work was derived in part from the National Science Foundation Grant 4648 and the Corning Glass Works Foundation Fellowship. References Bertaut, E. F., Durif-Varambon, A. and Pauthenet, R., 1954, Proprietes cristallograpliiques et magnetique de quelques nouvelles series de spinelles mixtes: 3rd Internat. Congr. of Crystallography, Paris. --—, 1956, Etude des germanates spinelles: Ann. de Chimie, v. 13, tome 1, p. 525- 553. Bullen, K. E., 1936, The variation of density and the ellipticities of the strata of equal density within the earth: Royal Astron. Soc., London, Monthly Notices, Geophys. Supp., v. 3, p. 395, 1932-1936. Dachille, F. and Roy, R., 1956, System MgsSiO,—MgaGeO, at 10,000, 60,000 and about 300,000 psi.: Geol. Soc. America Bull., v. 67, p. 1682-1683. --, 1957, Silica structure studies. The high pressure region of model and half-breed derivatives: Geol. Soc. America Bull., v. 68, p. 11713. -, 1958, Experimental study of the olivine-spinel inversion in MgaSiOi: Geol. Soc. America Bull., v. 69, p. 1550. -, 1959, A new high pressure form of B-.03 and inferences on cation coordination from infra red spectroscopy: Am. Ceram. Soc. Jour., v. 41, p. 78-80. 246 Frank Dachille and Rustu Roy -, in press a, The high pressure region of the silica isotypes: Zeitschr. Kristal- lographie. -, in press b, Use of infra-red absorption and water refractories to check coordination: Zeitschr. Krystallographie. Dachille, F., Shafer, E. C., and Rov, R.. in preparation, High pressure studies in the system GeOs-SiOs. Daly, R. A., 1943, Meteorites and an earth model: Geol. Soc. of America Bull., v. 54, p. 401-455. Goldschmidt, V. M., 1931, Zur kristallchemie des germaniums: Nachr. gesell. wiss. Got-tingen, math. phys. Kl. fachgr. IV, v. 1, no. 184, p. 190. Griggs, D. T. and Kennedy, G. C., 1956, A simple apparatus for high pressures and temperatures: Am. Jour. Sci., v. 254, p. 722-735. Jander, W. and Stamm, W., 1932, The internal structure of solid inorganic compounds at higher temperatures. V. electrical conductivity, diffusion, and reactivity of magnesium orthosilioate and magnesium on the germanate in the solid state: Zeitschr. anorg. Allg. Chem., v. 207, p. 289-307. Jeffreys, H., 1936, The structure of the earth down to the 20° discontinuity: Royal Astron. Soc., London, Monthly Notices, Geophys. Supp., v. 3, p. 401, 1932-1936. MacDonald, G. J. F., 1956, Quartz-coesite stability relations at high temperatures and pressures: Am. Jour. Sci., v. 254, p. 713-721. Majumdar, A. J. and Roy, R., 1957, Experimental phase rule studies involving transitions of various kinds: Geol. Soc. America Bull., v. 68, p. 1762-1763. Mason, B., 1952, Principles of geochemistry: New York, John Wiley and Sons, Inc., p. 33. (In personal communication with Professor Roy, Dr. Mason explained that his estimate of volume change of about 9% was based on an assumption that the MgsSiO« spinel would have the cell dimensions of AlaMgOj). Observatory, 1936, On J. D. Bernal: London, no. 748, p. 267-268. Ringwood, A. E., 1956, The system MgaSi04-MgiGe04: Am. Jour. Sci., v. 254, no. 11, p. 707-711. -, 1958a, The system MgjSiO.-MgsGeO.: Geol. Soc. America Bull., v. 69, p. 129. -, 1958b, The system MgaSiOt-MgaGeOi: Geochim. et Cosmochim. Acta, v. 15, p. 18-29. Romeijn, F. C., 1953, Physical and crystallographic properties of some spinels, pt. 3: Phillips Research Reports, v. 8, p. S21-341. Roy, Rustum, 1950, Magnesium in fourfold coordination in glass: Am. Chem. Soc. Jour., v. 72, p. 3307. -, 1956, Aids in Hydrothermal experimentation. II. Methods of making mixtures for both "dry" and "wet" phase equilibrium studies: Am. Ceram. Soc. Jour., v. 39, no. 4, p. 145-146. Roy, D. M. and Roy, R., 1954, An experimental study of the formation and properties of synthetic serpentines and related layer silicates minerals: Am. Mineralogist, v. 39, p. 957-975. Roy, R. and Osbom, E. F., 1952, Some simple aids in the hydrothermal investigation of mineral systems: Econ. Geol., v. 47, p. 717. Roy, R. and Tuttle, O. F., 1956, Investigations under hydrothermal conditions in Physics and chemistry of the earth: London, Pergammon Press, v. 1, p. 138-180. Safford, H. W., and Silverman, A., 1947, Alumina-silica relationship in glass: Am. Ceram. Soc. Jour., v. 30, p. 203-211. Turner, F. J. and Verhoogen, J., 1951, Igneous and metamorphic petrology: New York, McGraw-Hill Book Co., Inc., p. 357-358. Urey, H. C., 1952, The planets: New Haven, Yale University Press, p. 69.