<<

KAMB-B 65-0860

Structure of VI Barclay Kamb

Repriated frent Science. Oct.her 8, IH.'). Vel. 150, Xe. 3693, ...." .i-", powder data, to search the capillaries for suitable for single- study, and to orient and collect com­ plete diffraction data from such crys­ tals. Ice VI, prepared and examined in this way, gives a powder pattern (Ta­ ble 2) in general agreement with data obtained at high pressure and also with data from other experiments on quenched samples (6). Single crystals are easily obtained and are found to be tetragonal with unit cell dimensions a = 6.27 ± 0.01 A, c = 5.79 ± 0.01 Structure of Ice VI A. The diffraction patterns show extinc­ tions for reflections of type hkO with

Abstract. Ice VI, a high-pressure form of density 1.31 g cm-J , has a tetrag­ h + k odd and for hhl with I odd. The onal cell of dimensions a = 6.27 A, c = 5.79 A, space group P4jnmc, each space group is thus uniquely determined cell colltaining ten lIlolecules. The structure' is built up of hydrogen-bonded as P4~ / nmc (7) . The foregoing cell c//{lins of water molecules that are analogs of the tectosilicate chains out of which accounts well for the observed powder the fibrous zeolites are constructed. The chains in ice VI are linked laterally' to spacings (Table 2). one another to form an 0 en; zeolite-like framework. The cavities in this r ne­ Recently single crystal data for ice war ' are filled with a second fralllewor I enflca Wit 1 t 7e rst. The two frame­ VI, obtained at high pressure by means l . pene ra e £It 0 not interconnect and the complete structure call thus of a pressure cell mounted on a pre­ be con.fldered a "sell-clathrate." 1 h~ctural feature is a natural way to achieve cession camera, have been published high density in tetrahedrally linked framework structures. by Block, Weir, and Piermarini (8). These authors report an orthorhombic The stability conditions of the quenched by immersing the cylinder in cell with dimensions a' = 8.38 A, b' known forms of ice (Fig. 1) are a re­ liquid nitrogen, the pressure is released, = 6.17 A, and c' = 8.90 A, and for flection of the thermodynamic relations and the sample is forced out of the this cell they report diffraction extinc­ among the various phases, and these cylinder and into a bath . of liquid tions for reflections of type h' k'O with relations are in turn a macroscopic re­ nitrogen. At the temperature of liquid h' odd (diffraction aspect p •• a). This flection of the molecular architecture nitrogen (-196°C), high-pressure ice identification of the unit .cell is errone­ of the individual forms. Conversely, samples are metastable and can be OLIS. The cell chosen by Block el al. the structural features of the indi­ kept indefinitely at atmospheric pres­ (8) is related to the true tetragonal vidual forms in relation to their stabil­ sure. When the capillaries are removed cell as follows. The b' axis is one of ity fields and to the thermodynamic into the open aii', they explode, because the tetragonal a-axes, and the a' axis variahles provide some insight into the of the volume increase upon inversion is the tetragonal cell diagonal [10 I], forces of interaction between water of the high-pressure ice to ordinary that is, a' = a + c. The c' axis chosen molecules, and in particular into the ice J. by Block et al. (8) should evidently disruption of these forces that is caused Individual capillaries were mounted correspond to the other cell diagonal by packing the molecules more closely for x-ray study in a precession camera, [lOT]. The angle between the a' and c' together, as is achieved under high where they were maintained at a tem­ axes is 94.5°, according to the dimen­ pressure. Some structural information perature of -175°C by a stream of sions of the tetragonal cell, rather than for most of the forms of ice is now cooled nitrogen gas. The precession 90° as apparently assumed by Block availahle (Table 1), and the detailed camera can he used to obtain x-ray et al. (8). structures of ice T. Ie, II, I If, and VTJ are known (1-5). I here describe the structure . of ice VI, determined by x­ Table I. X-ray structural data for ice polymorphs. ray diffraction, and give preliminary x-ray .structural information for Density" Crystal Space Cell dimen- Ice Zt Reference (Table 1). (g cm"" ) syste{J1 group sions (A)" The samples were prepared as fol­ lows. Short glass capillaries, filled with I 0.92 Hexagonal P6,, / nZlllc a 4.48, c 7.31 4 (I) l. Ie 0.92 Cubic Fd3m a 6.35 8 (2) water, were placed in a hollow steel II 1.17 Rhombo- R3 a 7.78, Ot Il3.to 12 (3) cylinder fitted with pistons at both hedral

ends. The chamber thus formed was III 1.14 Tetragonal P4,212 a 6.79, C 6.79 12 (4) filled with gasoline, to transmit pres­ IV 1.28 (14) sure from the pistons to the water sam­ V 1.23 Monoclinic A2/ a a 9.22, b 7.54 28 ~ C 1035, {J 109.2· ples. By loadirig the pistons and control­ VI 1.31 Tetragonal P4"/ llmc a 6.27. e 5.79 10 :j: ling the cylinder temperature appropri­ VII 1.50 Cubic . PI/3m a 3.41 2 (5) ately (Fig. 1), the desired i~e phase • At atmospheric pressure and -175°C. i Number of molecules in the unit cell. :j: Data given can be formed. The sample is then in accompany ing text. Fig. I. Phase diagram for water. The phases (ice polymorphs) are identi­ fied by the roman numeral designations assigned by Bridgman (14, 9, 15), upon 20 whose data the diagram is based. The field of ice IV in relation to ice VI and liquid is plotted by dotted lines within the ice V field, and is plotted by analogy to 0,0 ice, for which the field was ac­ tually measured (14); ice IV is everywhere unstable relative to ice V. Ice Ic ("cubic ; ice") is shown sc hematically below the tem­ If) 15f- 0 L perature of about _105 C, below which 0 -<> it forms by vapor condensat.ion (16) and 0 above which it inverts to ice I (17) : it is Y: not known to have afield of actual sta­

Fig. 2. Precession photograph about the c-axis of ice VI, made with unfiltered molybuen'um radiation and without a 0 layer-line screen, /L = 15 • The II kO plane shows as single spots in the central por­ tion of the photograph. ilk I and higher layers as double spots and hyperbolic streaks in the oute r portion. Note 4-folu symmetry about the c-axis.

2 temperature (quenched) form would tail. The parameter values x and z Table 2. X-ray powder data for ice VI. be more symmetrical (tetragonal) than given must be accurate (probably to hkl the high-temperature form (supposedly ± 0.002), since the "fractional dis­ orthorhombic) from which it inverted crepancy" ~ I ~F ! /~IF()I is only 0.07, 4.44 4.47 110 by a displacive transfor~ation upon which is about as low as is commonly 81z 4.3 4.27 4.26 101 81z 3.6 cooling. achieved for refined structures based 51t 3.4 The density of ice YI at 6175 at­ on photographically measured x-ray 13 3.12 3.14 3.14 200 m~ospheres and 0 2°e (the triple point intensities. 8 2.91 2.92 2.90 002 3 90 2.75 2.76 2.76 201 ice V-ice VI-Ii ui cm- Positions of the hydrogen atoms 50 2.63 2.64 2.63 102 . If the compressibility measured cannot be closely defined from the 100 2.51 2.53 2.52 211 over the range 6175 to 8710 atm (9) x-ray diffraction data because hydrogen 8 2.43 2.43 2.42 112 25 2.21 2.22 2.21 220 holds over the range from 6 L75 atm atoms scatter x-rays only weakly. The 13b 2.10 2.13 202 down to 0, the density at zero pressure hydrogen atoms doubtless lie some­ 20 2.01 2.02 2.01 212 would be 1.3 L g cm-a. This density where near the line of centers between 20 1.97 1.98 1.98 310 17b 1.85 1.85 1.85 103 and the cell dimensions imply that atoms hydrogen-bonded to one Ib 1.75 1.75 222 the unit cel\ of ice VI contains ten another. 1.66 321 25b 1.64 1.64 203 water molecules (calculated density The structure of ice VI is plotted { 1.64 312 1.314 g cm-~). These ten molecules in c-axis projection in Fig. 3 and is 20 1.56 1.57 400 1.49 322 must be located at special positions in drawn in Fig. 4 in such a way as to 25 1.48 J 1 1.48 330 space group P4j nme, since the gen­ bring out its component parts. Each 3 1.44 1.45 004 eral position in this space group has water molecule (the oxygen atom of 12 1.41 1.42 303 16-fold multiplicity (7). The particu­ which is represented by a ball in Figs. 402 U~ 313 lar special positions occupied can be 3 and 4) has four nearest neighbors 7b 1.37 1.38 114 deduced by a process of elimination at a distance of 2.81 A, and is evi­ { 1.36 421 1.35 412 based on the chemical expectation that dently hydrogen-bonded to these. The no two water molecules will be closer four neighbors form a rather highly • Intensity values (I) and observed spacings (d, than about 2.7 A. This condition leads distorted tetrahedron around the cen- in A.) from quenched samples reported by Bertie, Calvert, and Whalley (6); II indicates to a unique solution for the structure tral water molecule. The water mole­ lines in ,"egion masked by diffraction halo from cules are linked together by the hydro­ silica glass tube; b designates broad lines. of ice VI. Eight of the oxygen atoms t Spacings (A.) measured from powder patterns of the water molecules occupy position gen bonds (shown as sticks in Figs. in present work. Listing is incomplete because no attempt was made to measure the complete 8 g, having coordinates of the fOrIn 0, 3 and 4) to form chains parallel to pattern. Relative intensities agree with those re­ x, z with x = 0.276, z = 0.382, and the e-axis of the crystal (column A, ported in column I. with the exception of lines Fig. 4). The chains link sideways to reported at 3.6 and 3.4 A., which are missing. the remaining two oxygen atoms oc­ ~ Calculated spacing (deale) from unit cell given cupy position 2 a, at 0, 0, 0 and 1/2, one another. parallel to the crystal in text. 1/2, 112. The values of x and z given are chosen so as to make the distances from each oxygen atom to its four nearest neighbors equal. The structure so derived is tested by comparing structure factors calcu­ lated from it with structure factors determined from the intensities of x-ray reflections from ice VI. For this pur­ pose reflections of type hOl, obtained from precession photographs of {his plane, are sufficient since a tetragonal structure is completely defined in a-axis projection. The comparison (Table 3) shows agreement so close as to verify that the structure is correct in de- o

by a 90° rotation (42 axis) , can be hydrates the cavities commonly ac­ inter-framework contacts made by combined together to make the com­ commodate organic or noble gas mole-. each water molecule are just as short plete structure (column C, Fig. 4). cules. In ice VI, however, the cavities as the four hydrogen bonds (intra­ The cavities of one framework ac­ of one framework accommodate a framework linkages) made. by it. Ice commodate the chains of the second second framework identical with the VI contrasts in that the inter-frame­ framework, and the cavities of the first; hence the crystal is in a sense a work contacts are all considerahly

A c

Fig. 4. Structure of ice VI, shown by assembly of its component parts. In column A are shown hydrogen-bonded chains of water molecules running parallel to the c-axis of the crystal. The chain below is identical with the one above but is rotated through 90°. Oxygen atoms of water molecules are represented by balls, and hydrogen bonds by sticks; the hydrogen atoms are not shown. In column B such chains are linked sideways to form two framework structures and are placed in their proper location relative to the unit cell (outlined). In column C the two framework structures are combined within a. single cell to make the complete structure of ice IV. 4

------

Table 3. Structure factors for ice VI: com­ longer than the hydrogen bonds. Each the positions of the hydrogen atoms. parison of calculated (F. ) and observed (F.) values. Calculated values contain contributions molecule (both types 2 a and 8 g) has This is of interest in relation to the from oxygen atoms only, positions as given in eight next-nearest neighbors, all at a large distortion of bond angles in the text, temperature parameter B = 1.5 A2~ Ob­ distance of 3.51 A. These neighbors ice VI structure. the angle between served values from intensities VIsually mea­ sured on precession photogra~s. Units of represent the closest approach between some of the hydrogen bonds at a given electrons per unit cell. oxygen atoms of the two separate water molecule being as low as 76° hkl interpenetrating frameworks. and as large as 128°. The detailed The special structural feature found study will also, it is hoped, make pos­ 200 29.4 - 26.S 2.6 here-interpenetratj'~n of separately sible an interpretation of the fact that 400 59.0 1.7 60.7 linked frameworks-is a natural way a very weak but definitely nonzero 600 14.2 -13.2 1.0 to achieve high density in framework intensity is observed for the 001 re­ SOO 14.7 14.1 0.6 structures based on tetrahedral link­ flection, in violation of the extinction 101 7.3 6.4 0.9 age, in view of the rather open struc­ rules for space group P4'!,/ nmc. This 201 61.3 59.4 1.9 ture that tends to result from such reflection suggests the possibility that 301 19.2 -20.3 1.1 linkage. It is noteworthy that this fea­ the orientations of the water molecule 401 4.5 -3.5 1.0 ture is shown by the highest-pressure become ordered upon cooling to 501 5.3 -5.5 0.2 forms of ice (VI and VII), but has not -196°C; it ·may be possible to de­ 601 14.6 14.2 0.4 yet been reported for high-pressure termine the of this ordering 701 10.3 -9.3 1.0 forms of SiO:! or for tectosilicate from the x-ray data. SOL 4.2 -3.9 0.3 structures. BARCLAY KAMB 002 34.9 -31.4 3.5 There is, however, a significant re­ Division of Geological Sciences, 102 54.S -50.S 4.0 lation between the ice VI structure and California Institute of Technology, 202 19.4 IS.9 0.5 certain tectosilicate minerals, the fi­ Pasadena 91109 302 13.4 15.4 2.0 brous zeolites. The structures of these 402 12 .0 -14.5 2.5 minerals (J J) are based on a tecto­ Referen~es and Notes 502 26.7. -2S.5 I.S silicate chain that is topologically 602 1. S. W. Peterson and H . Levy, Acta Cryst. 5.7 6.7 1.0 identical with the chain, described 10,70 (1957). 702 <4.4 0.9 above, that occurs in ice VI. The tecto­ 2. G. Honjo and K . Shimaoka, ibid., p. 710; S(}2 M. Blackman and N. D. Lisgarten, Admn. 4.0 -4.5 0.5 silicate chains are linked together in Phys. 7, 189 (1958). 103 32.6 32.7 0.1 3. B. Kamb, Acta Cryst. 17, 1437 (1964). different ways to form the various dif­ 4. -- and S. K . Datta, Nature 187, 140 203 42.3 -43.5 1.2 ferent fibrous zeolite structures (J 2). (1960) . 303 34.5 -32.0 2.5 5. B. Kamb and B. L. Davis, Proc. Noll. A cad. The particular linkage in edingtonite Sci. U.S. 52, 1433 (1964). Structural data 403 4.3 3.0 1.3 for ice VIr were obtained at high pressure is analogous to what occurs in ice VI but in Table I are extrapolated to at­ 503 S.1 9.4 1.3 (Fig. 4, column B). In edingtonite, mospheric pressure by means of data on the 603 12.5 -13.5 1.0 quenched form published by Bertie et al. however, the chains twist relative to (6) . 703 15.1 -14.3 O.S one another in such a way as to col­ 6. J. E. Bertie, L. D. Calvert, E. Whalley, Can. S03 4.6 4.0 0.6 J. Chell!. 42, 1373 (1964) . lapse the cavities between them and 7. K . Lonsdale, Ed., Tnternational Tables for 004 35.1 34.S 0.3 X-ray Crystallogral,hy (Kynoch, London, to reduce the equivalent a-axial length 1959). vol. I, p. 237. 104 4.1 4.6 0.5 to about 5.93 A (compared with 6.27 A 8. S. Block, C. W. Weir, and G. 1. Pier­ 204 S.5 9.2 0.7 marini, Science 148, 947 (1965) . . in ice VI). In the zeolite, the collapse 9. P. W . Bridgman, Proc. Am. A cad. Arts 304 <4.1 -1.6 of the cavities is governed mainly by Sci. 47, 441 (1912) . 404 17.3 IS.2 0.9 10. L. Pauling, The Nature of tl,e Chemical Bond interaction between the negatively (Cornell Univ. Press. Ithaca. N.Y .• 1960), 504 4.4 3.5 0.9 charged framework and cations (Na+, p. 469; J. H. van der Waals and J . C. 604 <4.4 - 0.6 Pattceuw. Ad,·all. Chem. Phys. 2, 1 (1959). Ca + + , Ba + + ) contained in the cavi­ It. L. Pauling. Proc. Natl. Acad. Sci. U.S. 16, 704 <3.9 - 0.1 453 (1930). ties (13); there is also a slight de­ 12. W. H. Taylor, Proc. Roy. Soc. Londoll Ser. 105 17.7 - 19.6 1.9 pendence on the water content of the A 145, 80 (1934); W. L. Bragg, Atomic 205 14.6 - 15.6 1.0 Strllctllre of Millerals (Cornell Univ. Press, zeolite. In ice vr, the cavities have Ithaca, N.Y., 1937), p. 258 . 305 21.1 22.9 I.S to be of equal size to the chains, in 13. W. M. Meier, Z. Krist. 118, 430 (1960). 405 < 4.5 1.2 14. P. W. Bridgman, J . Chem. Ph),s. 3. 597 order to accommodate the second (1935) . Density of ice IV is estimated from 505 6.9 - 6.9 0.0 measured volume difference between 0,0 ice framework; hence there is no collapse. IV and ice VI by assuming that the same 605 6.7 - 6.5 0.2 The large openings between pairs of molar volume difference applies to the H,O 006 <4.4 0.3 forms, and assuming equal compressibilities linked chains, which in the zeolite for ice IV and VI. The density of ice IV 106 13.7 13 .2 0.5 structures allow ion-exchange and de­ can then be calculated from the known den­ 206 5.0 5.3 0.3 sity of ice VI as established crystallollraphi­ hydration (/3), are utilized in the ice cally. 306 4 .S - 5. 1 0.3 15. P. W. Bridgman, J. Chelll. Ph),s. 5, 964 VI structure for the cross-links be­ (1937). 406 <4.4 - 0.3 tween chains of the second framework. 16. M. Blackman and N. D. Lisllarten, Ad,·all. 506 13.6 12.7 0.9 Ph.v.\'. 7. 197 (1958) . Refinement of the ice VI structure, 17. J. E. Bertie. L. D. Calvert, and' E . Whalley, 107 <4.4 1.0 J. Chell!. PI,ys. 38, 840 (1963). which IS being undertaken with more 207 12.3 14.4 2.1 18. Division of the Geological Sciences, con­ complete x-ray data now at hand, will tribution No. 1373. The work was supported 307 < 4.1 1.4 permit any slight differences in length by the NSF and the Sloan Foundation. J 407 < 3.4 -1.2 thank Steven Card and Jon Evans for as­ among the three types of hydrogen sistance with the experimental work and OOS 13 .6 15.9 2.3 calculations, and Linus Pauling for helpful bonds present to be determined, and it discussion. lOS 2.8 2.6 0.2 may lead to some conclusions about 20 August 1965 5