HIGH-TEMPERATURE CRYSTAL CHEMISTRY
537
concentration, the effectiveness of the silica in discharging and recharging iron oxide surfaces increased with the silica ratio of the original samples. 6. The tendency of sodium silicates to form colloidal suspensions has been found to increase with increase in silica ratio. 7 . Evidence has been presented which supports the belief that silica sols are colloidal electrolytes. REFEREXCES (1) BAKERASD DEDRICK:U. S. patent 2,310,009 (February 2, 1943). (2) BAYLIS:Water Works & Sewerage 84, 221 (1937). (3) HARMAN: J. Phys. Chem. 32, 44 (1928). (4) HAY:J. Am. Water Works Assoc. 36, 626 (1944). ( 5 ) HAZEL: J. Phys. Chem. 46,516 (1942); (a) J. Phys. Chem. 42, 409 (1938); (b) J. Phys. Chem. 46, 731 (1941). (6) HAZEL A N D AYRES:J. Phys. Chem. 36, 2930 (1931). AND AYRES:J. Phys. Chem. 36, 3148 (1931). (7) HAZEL (8) HAZELAND MCQUEEN:J. Phys. Chem. 37, 553 (1933). (9) See McBain's discussion in Kraemer's Advances i n Colloid Science. Interscience Publishers, Inc., New York (1942). (10) STERICKER: Ind. Eng. Chem. 30, 348 (1938). (11) VAIL: Soluble Silicates i n Industry. The Chemical Catalog Company, Inc., New York (1928). (12) WILLEYASD HAZEL: J. Phys. Chem. 41, 699 (1937).
HIGH-TEMPERATURE CRYSTAL CHEIIISTRY OF A,BX, COMPOT:S DS WITH PARTICLLAR REFEREXCE TO CALCIUM ORTHOSILICATE A i . A. BREDIG Applied Chemicals Division, Vanadium Corporation of America, 420 Lexington Avenue, New York, S e w York
Received Julu 26, 1945 I. INTRODUCTION
Recently, an attempt has been made to apply the results of crystal-structure determinations of silicates and aluminates occurring in cement to an interpretation of their chemical behavior, such as the hardening by reaction with water (10). Brandenberger (2) has tried to explain in terms of crystal structure the greater reactivity of the high-temperature modifications of compounds such as calcium orthosilicate (cy and P ) as compared with the inertness of the low-temperature ( y ) form of this compound. A tendency of the calcium cation a t room temperature to surround itself with a larger number of negative ions, such as 02-,than at higher temperatures was assumed. Specifically, it was suggested that the coordination number of calcium for oxygen in the reactive /3-form
538
M. A . BREDIG
which is metastable below 675°C. might be 4, that is, less than the corresponding figure (6) for the inactive low-temperature (y) form, the crystal structure of which has lately been shown by O’Daniel and Tscheischwili (28) to be of the type of olivine. By all these authors a Ca04 group was assumed to have a greater tendency to add water molecules and thus to produce a normal, high oxygen coordination of 6 or 8 around calcium than a Ca06 group such as the one occurring in the olivine structure of the low-temperature form. S o actual experimental determination of the atomic arrangement in the reactive high-temperature forms of calcium orthosilicate Tvas available to support the assumption of C a 0 4 groups in stabilized p-calcium orthosilicate. On the contrary, the w i t e r (4, 7 ) has shown that the crystal structures of the two forms of calcium orthosilicate above 1420°C. most likely are the same as those of the two forms of potassium sulfate. In the low-temperature (p) form of potassium sulfate, potassiuni has nine to ten nearest neighbors, oxygen atoms belonging to tetrahedral SO4 anions (33). In high-temperature a-K2S04,shown (4) to have the structure of glaserite (18), the number of nearest neighbors of potassium is similar. Because of the isotypy of a- and a’-calcium orthosilicates, the forms stable above 142OoC., n-ith these two forms of potassium sulfate, calcium must necessarily haye the same number of nearest neighbors in these two crystal modifications of CaaSi04,that is, nine to ten. 173th the corresponding coordination number having been shown to be only G in the low-temperature (y) form of (‘a2SiOa, n e note that the tendency of calcium in regard to its coordination nith oxygen as a function of temperature appears to be just the reverse of that assumed in Brandenbcrger’s theory. The coordination around calcium actually is lowest at low temperatures and highest at high temperatures. Before a more detailed and a t the same time more general discussion of this problem \T ill be presented, some additional experimental evidence is examined in the following, Jvhich demonstrates the existence in calcium orthosilicate, at high temperatures, of an atomic arrangement in rvhich the coordination around calcium is 10 rather than 4. 11. THE STRUCTURE OF JIC~WISITE, A SOLID SOLCTIOX OF
lclg&Ol
ISa’-Ca28i04
The effect of additions, to calcium orthosilicate, of small amounts of phosphate or potassium silicate n-as considered to represent the stabilization of a heretofore unknown crystal phase of CanSiOr,designated the &-form by the writer, and isotypic with P-IZZSO, (4, 7 ) . It now appears that magnesium orthosilicate also belongs to the agents able to stabilize the orthorhombic, second-highesttemperature form (a’). Merninite, “3Ca0 .NgO 2Si02,” which occurs as a natural mineral (24) as well as an occasional constituent of slags (32), and which has also been recommended as a refractory material ( 2 5 ) , is believed by the writer to be nothing but the &-form of CaBi04, stabilized by and containing in solid solution a maximum of approximately one-third of one mole of MgzSi04per mole of CazSi04. In table 1, d values of an orthorhombic crystal lattice with the constants a0 = 5.20, bo = 9.20, and co = 6.78 b. ( a : b : c = 0.565:1:0.737), isotypic with 9
539
HIGH-TEMPERATURE CRYSTAL CHEMISTRY
P-Kk304, are compared with the experimental d values reported by Phemister (32). The agreement is considered rather satisfactory. The lattice constants, and also in general the intensities of the x-ray reflections, are quite similar to those of p-CaS‘aP04 with a0 = 5.215, bo = 9.32, and co = 6.83 A. TABLE 1 Interpretation of powder 2-ray pattern of merzcinite, a’-(Ca,SiO,~+Mg,SiO~) a0 = 5.20, bo = 9.20, co = 6.7Sw.; z = 4; specific gravity (x-ray) = 3.34* dexperimerital
dealdated
(Phemister (32))
(Bredin)
W.
8. 2.94 2.83
(2.94) 2.845
(130 KO) 102
m.
2.73
{;:;;
022
S.
.
2.65 2.41 2.30 2.20 2.16
w.
2.03
{;:E
m.
1.90
j 1.905
INTESSITY
w.
V.W.
w. w. TV
INDICES
hkl
8.
2.65 2.42 2.30 2.20 2.15
130 122 040 013 221 113) 023,
11.905
m.w. V.W.
1.87 1.75
v.w.
1.69
(1.89
123
1.88 1.74
222 150
i1.70
004 151
etc. Reflections h02, 1 = 2n
+ 1, and hkO, h + 6 = 2n + 1 are not observed; space group =
0;;.
* In the mineral merwinite, Larsen and Foshag (24) have found 3.15. Spurrite and gehlenite of considerably lower specific gravity were reported t o have been closely associated with merwinite in the sample. ( a : b : c = 0.560: 1:0.733), space group Di:,o which is also isotypic (4) with P-KzSO~,a0 = 5.76, bo = 10.05, co = 7.46 A. ( a : b : c = 0.573:1:0.742). The orthohexagonal lattice constants of solid solutions of the highest-temperature (CY) form of Ca2SiO4(4) are, for comparison, a0 = 5.40, bo = a 0 4 3 = 9.35, co = 6.996 (19) ( ~ : b : = c 0.577:1:0.749). In a brief discussion of the structure, H. J. Goldschmidt and Rait (16) re-
540
XI. A . BREDIG
cently suggested that merwinite was isotypic with perovskite, CaTiOa. Such an isotypy seems extremely improbable. It would require the same arrangement of the oxygen ions in relation to the metal ions. In perovskite titanium is surrounded by six oxygen atoms. It is not clear how the authors propose to reconcile with this fact the assumption of Si04tetrahedra. Just as in the cases of phosphate and potassium silicate additions to calcium orthosilicate, where formulas for ternary compounds such as ‘LCa,(P04)2(Si04)2” or L‘Ca23K2(Si04)12” could be discarded (4, 7 ) , present experimental facts do not warrant considering merwinite as a ternary compound, “3Ca0. MgO 2Si02.” This is true even though Osborn (30) and Parker and Nurse (31) have found merwinite to exist in equilibrium with the liquid, where it had not been observed originally in the first comprehensive study of the ternary equilibrium diagram by Ferguson and Merwin (14) and by Greig (20). In none of these studies mas any evidence presented which would exclude complete solubility in the solid state, in the temperature range between 1420°C. and, say, 1550”C., between CazSi04 (a’-form) and the composition of the alleged “Compound 3Ca0 .Mg0.2Si02.” According to the more recent investigations cited above (30,31), the phase called merwinite melts incongruently at 1580°C., decomposing into melt and CanSi04. At this temperature, pure Ca2Si04is most likely to be present in its hexagonal glaserite-type form ( a ) . On quenching, it will, however, be found in the &modification, owing to successive transformation into the a‘- and p-forms because of the absence of a sufficient quantity of stabilizers in solid solution. In a previously published representation of the equilibria of calcium orthosilicate with orthophosphate (7) the transition point a’ S a in pure Ca2Si04 had been placed tentatively near 1800”C., halfway between the long known transition point a t 1420°C. and the melting point. The latest data by Osborn and by Parker and Kurse, however, make it necessary (8) to place the transition a’ e a at a temperature between 1420” and 158OoC.,if the assumption of the identity of merminite as a solid solution of lIg2Si04 in a’-Ca2Si04,proposed by the writer, is to be maintained. These relations are indicated schematically in the diagram of figure 1, representing a pseudo-binary join from CazSiO4 towards CaMgSi04 (monticellite) and MgSSiO4(forsterite), in the ternary system Ca0-Mg0-Si02. They will haye to be tested experimentally in the region close to Ca2Si04. If a careful interpretation of x-ray intensities should show that magnesium is not randomly placed on the cation positions of the a’-Ca2Si04crystal lattice, but occupies only one type of these, it would be necessary to revert t o the concept of merwinite as a ternary compound in spite of complete isomorphism with a’-Ca2Si04. Such a condition has been assumed to exist in monticellite, CaMgSi04, of IIg2Si04 (olivine) structure (9), which forms an uninterrupted series of solid solutions n-ilh 1\IgnSiO4. Some observations on merwinite are in disagreement with the orthorhombic symmetry assumed here for a’-CazSiO,. If these observations are considered indicative of lower crystal lattice symmetry, it might be necessary to modify
HIGH-TEMPERATURE CRYSTAL CHEMISTRY
541
the assumption of the identity of the nierwinite specimens examined optically a t room temperature with the actual a’-CazSiOc solid-solution phase a t high temperatures: ,4transformation may have occurred on cooling which may have diminished the crystal symmetry from that of the orthorhombic 6-potassium sulfate structure of a’-CasSi04to that (monoclinic, triclinic) reported by Phemister and others for merwinite. In fact, the polysynthetic twinning reported by these authors seems to point t o such a possibility, just as did the twinning in
FIG.1. Probable phase relations of the system CaO-MgO-SiO1 in the vicinity of CazSiOc (schematic).
so-called a-Ca2SiO4,which was actually P-CazSi04 (Hansen (21) ; Insley (22)). The conclusions drawn both from the occurrence of the 6-potassium sulfate structure in Cassi04 containing phosphate or potassium silicate and from the striking similarity of Phemister’s x-ray pattern of merwinite with that of /%potassium sulfate, even though merwinite might actually be of lower symmetry, would remain valid regardless of the existence of such a transition. (The formation of a highly asymmetric monoclinic or triclinic structure from melts of simple orthosilicates a t high temperatures seems rather unlikely.)
542
M. -4. BREDIG 111. DISCCSSION
The oxygen coordination of 9 to 10 around calcium in the potassium sulfatetype, high-temperature, a’,crystal structure of CasSi04 represents a considerable increase from the Ca06 coordination of the olivine-type, lowest-temperature y-form. It is precisely this increase, rather than a decrease to Ca04 presumed by O’Daniel and Tscheischwili and others, that must actually be expected from theoretical considerations. The authors just mentioned pointed to the following rule pronounced by V. ill. GoldSchmidt (17) in 1926: ht high temperatures that crystal structure (of an X,BX, compound) is stable which could be also obtained, through a morphotropic change in structure, b j substituting a lower homologue for the cation (A) by which the BX, radical ion is LLcontrapolarized.” The term “contrapolarization” was introduced by Goldschmidt (17) in reference to the theory of the-mutual-dcformation (polarization) of the electronic shells of ions, advanced and extensively developed by Fajans (11-13). “Contrapolarization,” according to Goldschmidt, is the electrostatic influence (loosening) which A cations such as Ca2+,surrounding, in a crystal lattice, the BX, radical anions composed of strongly polarizing central B ions and strongly polarized (deformed) X ions (Si4+and 02-,respectively, in SiOt-) exert upon the state of polarization of the X ions, upon their distance from the central B ion, and, thus, upon the strength of the B-X bond. Applied to calcium orthosilicate, the above rule seemed to O’Daniel and Tscheischwili, Brandenberger, or Bussem, and others, to mean that thepresumedly-increased “contrapolarization” or expansion of the Si04 ion a t high temperatures would lead to a relatively more active r61e of the calcium ion in the crystal lattice, and t o a decrease from a coordination Ca06 to Ca04, somewhat similar to the change brought about by substituting for calcium the much smaller and stronger contrapolarizing beryllium (Be04 in phenacite, Be2Si04). V. M. Goldschmidt had been careful to limit the rule to those cases in which the transformation would actually be effected by no other of several possible causes but an increase in contrapolarization of the BX, radical ion by the cation A a t increasing temperature. It can be shown, however, that such cases must be quite rare, if not entirely lacking, and that in the examples given by Goldschmidt an increase in contrapolarization with increasing temperature does not occur, and therefore cannot be the cause of the polymorphous transition. On the contrary, it is both plausible and borne out by actual observation, that with increasing temperature the weakest binding (A-X, e.g., Ca-0) in L B X , compounds (e.g., CanSiOh)will be further weakened to a considerable degree long before the stronger bonds (B-X, e.g., Si-0) are materially affected by the thermal energy. This, hon-ever., means that the contrapolarization of the BX, radical ion (e.g., of SiOi-) by the cation d (e.g., Ca2+)must be expected to decrease with increasing temperature rather than to increase. The average distance between calcium and oxygen must be increased gradually and in steps (polymorphous transitions) until at the melting point “dissociation” into cal-
HIGH-TERIPERATGRE CRYSTAL CHEMISTRY
543
ciuni cations and Si04 anions takes place, mith the Si-0 bonds still largely intact. It is a fallacy to ascribe the increase in the amplitude of thernial oscillations almost entirely to the oxygen ions (O’Daniel and Tscheischnili) and to postulate a virtual decrease in the effective radius ratio Rca:Ro with increasing temperature. At a larger average distance iron1 the center of the calcium ion, a larger number of oxygen ions mag’ be thought to be located. Thus an increase in the coordination around A ions may be generally espected in compounds An,BX,,with increasing teniperature. Under these circumstances. let u s e\amine in greater detail the evidence upon n-hich the older rule n-as based. Of three examples of the effect, upon the crystal structure, oi a loosening of the B-X bond by increasing temperature, the group oi the alkali metal sulfates (KzS04,Rb2S04,and Cs2S04) n-as assumed t o take on phenacite structure n i t h AX4 beside BX4 coordination above their transition pointy a t 588”, 650”, and 660”C., respectively. However, the phenacite structure with an extreniely improbable KO4 coordination-Rb04 and CsO4 n-ould be even less plausibledoes not occur in potassium sulfate, but what is formed is the glaserite structure (4) with a rather high coordination around potassium, such as I