2324
J . Phys. Chem. 1985,89, 2324-2330
TABLE IV: Main Results Obtained from the XAS Studies on Calcined Cu-Zn and Cu-Zn-AI Catalysts environment of catalyst cu Zn binary valence Cu2+;structure valence Zn2+; (Cu-Zn) has similarities with ZnO structure CuO; no significant amounts of Cu dissolved in ZnO Zn2+;only first shell ternary Cu2+;only first shell (Cu-Zn-AI) (4 oxygens) visible; Cu2' visible; structure different from ZnO presumably forms a mixed oxide phase with Zn2' and ZnA120, (and AI) In this connection, it is interesting that, from XPS9g'o and XRD4v5 studies, it was concluded that several percent of Cu2+ions may dissolve substitutionally in the ZnO lattice not only in the surface but also in the bulk of ZnO. Addition of A1 to the Cu-Zn catalysts is observed to lead to drastic changes in the local surroundings of not only the zinc atoms but also the copper atoms. Although the Cu absorption K-edge of the ternary catalysts shows essentially the same feature as the binary catalysts (Figure 7A), the Fourier transformed Cu EXAFS of the aluminum-containing catalysts shows fewer and much less intense peaks outside the strong nearest-neighbor oxygen shell (Figure 7B). This result indicates that copper in the ternary catalysts is present in smaller crystallites or in less well-ordered surroundings compared to the binary catalysts. A computer fit of the EXAFS contributing to the first-shell backscatter peak gives essentially the same parameters for this shell in the two catalyst systems (See Table 111). The influence of A1 on the local surroundings of the zinc atoms is evident even from a simple analysis of the Zn absorption Kedges. A comparison of the edges for the model compounds, ZnO and ZnA1204,and the T6 catalyst clearly shows that zinc is present neither as ZnA1204nor in a structure resembling that of wellcrystallized ZnO (Figure 5B). Further evidence for this result is found in the Fourier transformed Zn EXAFS of these compounds (Figure 6B). Only one strong peak corresponding to backscattering from the nearest-neighbor oxygen atoms is observed for the catalyst. Although this peak is very similar to the first shell peak in ZnO and ZnAls04, the complete absence of back-
scatter peaks from more distant coordination shells indicates that neither ZnO nor ZnA1204is present in any significant amounts in the ternary catalysts. Thus, the addition of A1 to a Cu-Zn catalyst seems to prevent the formation of any ordered zinc phase. This result is in qualitative agreement with the work of Gherardi et a1.* who reported an X-ray pattern quite similar to those shown in Figure 1. However, these authors also reported that by changing preparation parameters, sharp X-ray patterns could be obtained and separate phases of CuO, ZnO, and ZnA1204could be identified. This dependence of preparation parameters on the structures formed after calcination is probably the reason why quite different results on apparent similar systems have been reported in the literature (see e.g., ref 2,8-10). The absence of any distinct diffraction lines in the XRD pattern of the ternary catalysts is in agreement with the EXAFS results, showing that both the Cu and Zn atoms are present in microcrystalline or amorphous-like structures. The zinc-containing phase in the ternary catalysts exhibits such a poor crystallinity that the backscattering from metal atoms in the second shell is absent. In essence, the same holds for the copper-containing phase and it is not unlikely that Cu2+and Zn2+form a mixed oxide phase which is basically amorphous. This phase may also contain aluminum. Conclusion The present study has shown that X-ray absorption spectroscopy may be a valuable tool in characterizing both binary Cu-Zn and ternary Cu-Zn-A1 methanol catalysts. The unique advantage of the technique is due to the fact that the local environment of both the Cu and Zn atoms can be probed, and the presence of microcrystalline or amorphous-like structures poses no difficulties in these studies. In Table IV we have summarized the main conclusions of the XAS studies. Acknowledgment. We are grateful to HASYLAB for offering beam time on the synchrotron radiation facility and for access to the ROMO spectrometer. We are also grateful to J. Villadsen for assistance with the X-ray diffraction measurements, 0. Srarensen for the electron microscopy measurements, H. Topme, P. Stoltze, and P.E. Hajlund Nielsen for valuable discussions, and J. W. Ornbo for technical assistance.
Registry No. CuO,1317-38-0;ZnO, 1314-13-2; methanol, 67-56-1.
Phase Transitions in Nitrogen Observed by Raman Spectroscopy from 0.4 to 27.4 GPa at 15 K David Schiferl,* Steven Buchsbaurn,+and Robert L. Mills University of California, Los Alamos National Laboratory, Los Alamos. New Mexico 87545 (Received: December 4, 1984)
Raman scattering from solid nitrogen was measured in a diamond-anvil cell from 0.4 to 27.4 GPa at a temperature of 15 K. Two new low-temperature structures, 6(LTl)-N2 and 6(LT2)-N2, were discovered through changes in the Raman spectra. Both structures bear a resemblance to the known high-temperature 6-Nz phase, which has space group Pm3n. From a comprehensive study of correlationdiagrams for related structures, likely space groups for the new phases were found. 6(LTl)-N2, which forms from y-N, at 1.9 GPa and is stable up to about 21.0-24.5 GPa, probably has structure Rgc, in agreement with recent theoretical calculations. At higher pressures, 6(LT2)-N2exhibits a structure that is quite possibly R3c.
Introduction With the advent of high-pressure diamond-anvil cells and more powerful computer modeling techniques, there has been renewed +Present address, Department of Physics, University of California, San Diego, CA.
0022-3654/85/2089-2324$01.50/0
experimental and theoretical interest in the interactions between mOleCUleS as a function Of pressure and temperature. An important goal is the derivation of detailed intermolecular potentials from which reliable predictions can be made of the structures and space group of mole4Xh' crystals. Nitrogen Serves as an excellent subject for study over a wide range of P and T because the N 2 0 1985 American Chemical Society
Phase Transitions in Nitrogen molecule is quite stable, as a consequence of its strong triple bond, and is relatively simple to treat, by virtue of its position as a first-row element. Experimentally it is known that liquid nitrogen freezes under its own vapor pressure at 63.1 K into 0-N, (space group P63/ mmc), which has a hexagonal close-packed structure with a high degree of molecular disorder.' On cooling to 35.6 K a transition takes place to cubic a - N 2 (Pa3) in which the molecules are completely ordered.2 Pressurization of a-N, to about 0.4 GPa causes a transformation to y-N2(P4,/mnm), a tetragonal form which is also molecularly ~ r d e r e d . ~At . ~ room temperature nitrogen can be frozen into P-N2directly from the gas by applying a pressure5 of about 2.5 GPa. Further compression to 4.5 GPa gives 6-N2 (Pm3n), a cubic solid which, curiously, has molecules exhibiting two types of di~order.~,' In a previous paper8 we presented a phase diagram of N 2 that was determined from characteristic features in the Raman spectra y-, and 6-N2 measured in a diamond cell from 15 to 300 of 0-, K at pressures up to 52 GPa. We reported, however, that over wide regions of the 6-N2 phase, extra Raman-active lines were observed in both the vibrational and lattice modes, which were inconsistent with Pm3n point symmetry. It was suggested that these lines may have arisen from a distortion of the cubic Pm3n structure, although it was not clear whether the distortion was continuous with P and T or whether various discrete low-symmetry structures were formed. On the theoretical side, Kobashi and co-workers9 used a fitted exp-6-12 potential to predict that 7-N2would transform to the R3m structure at about 4.0 GPa. LeSar and Gordon,Io using an electron-gas model with short-range energy and an anisotropic c 6 term, also calculated that an R3m structure would form, but at the lower pressure of 2.5 GPa. These theoretical models were later refined to incorporate the experimental discovery8 that the low-temperature structure of N2,which forms at about 2 GPa, is not R j m , but more likely a slight distortion of the known high-temperature 6-N2 (Pm3n) structure. With an improved method for minimizing the lattice energy, Chandrasekharan et al." calculated that a transformation from y N 2 to rhombohedral R3c (subsequently changed', to RJ) would take place at 1.92 GPa, followed by a second transition to R3m a t 6.75 GPa. Recently, LeSar,I3 employing a model with damped-dispersion energy and terms up to Clo,found a transition from -y-N, to R3c at 2 GPa. The R3c structure was stable with respect to R3c at pressures up to at least 75 GPa. By generalizing a constant-pressure molecular-dynamics technique to study molecular crystals, Nos6 and KleinI4 calculated that at 7.0 GPa cubic Pm3n N2with eight molecules per unit cell would transform when cooled to 230 K into cubic 1213 with 64 molecules. In this new structure the disk-shaped disordered molecules are aligned parallel to the cube axes. At 140 K a second transition to R3c takes place with 64 molecules per cell and the spherically disordered molecules aligned along trigonal directions.
The Journal of Physical Chemistry, Vol. 89, No. 1 1 , 1985 2325 The present experimental study was undertaken to confirm the structural distortions that were observed8in 6-N2 and to determine their nature.
Experimental Section Several Merrill-Ba~sett'~cells were used to generate pressures up to 27.4 GPa at low temperature. The highest pressures were reached with a cell in which the triangular frames and the backing plates for the diamonds were constructed of hardened berylliumcopper. All cells incorporated low-fluorescence diamonds with 16 facets on both table and culet, and with culet tips 600 pm in diameter. The liquid-immersion techniqueI6 was used to load each cell, after which it was mounted on the cold-finger of a Janus Model S/T helium-flow cryostat. The lowest sample temperature that could be achieved was 15 f 5 K. It was necessary to remove the cell from the cryostat to make pressure changes. Cooling the cell from room temperature to 15 K caused the pressure to increase or decrease by as much as 2 GPa, depending on whether the pressure had been, respectively, raised or lowered just before cooling. Pressures were measured with the ruby-fluorescence method,17J8 assuming the relationI9 P = 3 8 0 . 8 ( [ ~ o ( T ) / ~ ~ (T )1) ]~
where P is the pressure in GPa, up( T ) is the wavenumber of the ruby R, line at pressure P and temperature T, and vo(T) is the corresponding wavenumber at the same T and zero pressure. For our sample of ruby, which contained 0.16 wt% Cr203,the shift in u o ( T ) could be fitted within experimental error ( f 0 . 4 cm-I) by the expression8
+
v O ( T ) = 14422.0 - 36.612(T/300)3/2 169.77(T/300)4/2 264.54(T/300)5/2 112.14(T/300)6/2cm-I (2)
+
for temperatures in the range 0 < T I300 K. Both the ruby-fluorescence pressure measurements and observations of the Raman spectra were made with a Spex 1403 spectrometer equipped with a periscope for viewing the sample image in the entrance slits. We moved the laser spot, which had a diameter of about 30 pm, around in the sample area to focus it either on a ruby chip for pressure measurements or on a ruby-free area for Raman measurements. To avoid heating the sample, the power incident on the cell was kept below 30 mW for pressure measurements and below 200 mW for Raman studies. To find the laser excitation that caused the least interfering fluorescence from the diamonds, we tested the 21 000-cm-I line from a krypton laser and the 20492- and 21 839-cm-' lines from an argon laser. The 21 839-cm-l line gave the highest ratio of sample signal to fluorescence background. In no case, however, were we able to observe lattice-mode spectra at sample temperatures above about 20 K.
Results (1) Streib, W. E.; Jordan, T. H.; Lipscomb, W. N. J . Chem. Phys. 1962, 37, 2962.
(2) Venables, J. A.; English, C. A. Acta Crystallogr.,Sect. B 1974,30,929. (3) Mills, R. L.; Schuch, A. F. Phys. Reu. Lerr. 1969, 23, 1154. (4) Schuch, A. F.; Mills, R. L. J. Chem. Phys. 1970, 52, 6000. (5) Mills, R. L.; Liebenberg, D. H.; Bronson, J. C. J . Chem. Phys. 1975, 63, 4026. (6) LeSar, R.; Ekberg, S . A.; Jones, L. H.; Mills, R. L.; Schwalbe, L. A,; Schiferl, D. Solid Stale Commun. 1979, 32, 131. (7) Cromer, D. T.; Mills, R. L.; Schiferl, D.; Schwalbe, L. A. Acta Crystallogr. Sect. B 1981, 37, 8. ( 8 ) Buchsbaum, S.; Mills, R. L.; Schiferl, D. J. Phys. Chem. 1984, 88, 2522. (9) Kobashi, K.; Helmy, A. A.; Etters, R. D.; Spain, I. L. Phys. Reu. B 1982, 26, 5996. (10) LeSar, R.; Gordon, R. G. J . Chem. Phys. 1983, 78, 4991. ( 1 1) Chandrasekharan, V.; Etters, R. D.; Kobashi, K. Phys. Reu. B 1983, 28, 1095. (12) Etters, R. D., private communication. (13) LeSar, R., private communication. (14) Nost, S.; Klein, M. L. Phys. Rev. Lett. 1983, 50, 1207.
The pressure dependence at 15 f 5 K of the stretching-mode Raman peaks is shown in Figure 1 and that of the lattice modes is shown in Figure 2. Qualitative changes in the two spectra were observed both at 1.9 GPa and over the pressure range 21.0-24.5 GPa, where phase transitions are assumed to occur. An experimental point at 436 cm-' and 27.4 GPa falls on an extension of the uL1 curve but was inconvenient to plot in Figure 2. Below 1.9 GPa we find the same lattice modes that Medina and Daniels2OS2land Thiiry et al.,, reported for y N 2 (P4Jmnm), (15) Merrill, L.; Bassett, W. A. Reu. Sci. Instrum. 1974, 45, 290. (16) Schiferl, D.; Crorner, D. T.; Mills, R. L. High Temp. High Pressures 1978. 10. 493. (17) Forman, R. A,; Piermarini, G. J.; Barnett, J. D.; Block, S. Science 1972. 176. 284. (18) Barnett, J. D.; Block, S.; Piermarini, G. J. Reu. Sci. Instrum. 1973, ~
44.> -1 .
(19) Mao, H. K.; Bell, P. M.; Shaner, J. W.; Steinberg, D. J. J . Appl. Phys. 1978, 49, 3216.
Schiferl et al.
2326 The Journal of Physical Chemistry, Vol. 89, No. 1I, I985
2410L
2400 -
i
250-
//'
2390-
2380-
--
-
-E
E,
Y
h
200-
2370 -
/ /
2360-
/
t..dl-
l / d
/
z
/
/ - /
2350-
B ,/ B
,
I
1
1
I
0
5
IO
15
20
25
2330
I
5
and have used their notation to describe the excitations. Above 1.9 GPa we have no symmetry information and have simply labeled the modes sequentially in Figure 1. In the lattice-mode Raman spectrum shown in previous work8 we found that the peaks at uL3, uLs, and uL6 became very weak above 6.5 GPa, and could not be observed at all in some runs. Since lattice modes in the phase stable above 24.5 GPa were not clearly observed, we have not labeled them specifically. There are seven lattice modes between 1.9 and 21.0 GPa, and as many as nine above 21.0 GPa, although it is not certain which phase to assign them to. In y N 2 (P42/mnm), both Medina and Daniels20 and ThiBry et a1.22point out that there should be two Raman-active N2 stretching modes with symmetries A,, and B2,, although they observed only one asymmetrical line. As reported earlier,s we observe a strong peak in this region with a well-defined shoulder on the higher-energy side. The shoulder becomes weaker with increasing temperature and cannot be distinguished above 50 K. These features are consistent with the type of asymmetry reported by Medina and Daniels2' Between 1.9 and 21.0-24.5 GPa, the stretching-mode spectra show two well-separated peaks, just as has been reported6S8above 4.9 GPa at 300 K. Accompanying the transition at 21.0-24.5 GPa, the u2 peak splits into two peaks u~~ and V2b. The sequence in Figure 3 shows the evolution of the splitting: in Figure 3a at 16.0 GPa the u2 peak is shown well below the onset of the transition; in Figure 3b a t 21 .O GPa the u2a peak appears in addition to u2; in Figure 3c at 22.8 GPa the V2b peak also appears, while the (20) Medina, F. D.; Daniels, W. B. J . Chem. Phys. 1976, 64, 150. (21) Daniels, W. B.; Mediha, F. D. "Physics of Solid Under High
Pressure"; Schilling, J. S., Shelton, R. N., Ed.; North-Holland: New York, 1981; p 23. (22) ThiE.ry, M. M.; Fabre, D.; Jean-Louis, M.; Vu, H. J . Chem. Phys. 1973,59. 4559.
I
1
I
15
20
25
,", ",
30
D ICDnl I
P (GPO)
Figure 1. Pressure dependence at 15 K of the Raman stretching-mode peaks in N,. Open symbols are for decreasing pressure and closed symbols are for increasing pressure: circle, 21 839-cm-' Ar+ laser excitation; square, 21 000-cm-' Kr+ laser excitation;triangle, 20 492-cm-' Ar' laser excitation. Lines are guides for the eye.
I
10
Figure 2. Pressure dependence at 15 K of the Raman lattice modes in N,. Symbols as in Figure 1.
intensity of u2 drops relative to that of uZa; in Figure 3d at 23.8 GPa the uz peak is much reduced relative to the uZa and V2b peaks; in Figure 3e at 25.4 GPa, uh and Y2b are strong, while u2 has almost completely disappeared; and, finally, in Figure 3f at 27.4 GPa the completely separated uk and Y2b peaks are shown well above the transition. It is not clear from Figure 3 whether Y2b grows relative to uk or whether the two peak heights remain in the ratio of approximately 2:3 over the transition region. Based on arguments given in the following discussion, we present in Figure 4 a provisional phase diagram, showing the new lowtemperature phase transitions in solid nitrogen.
Discussion The most striking result of this work is the discovery of a phase transition in the pressure range 21.0-24.5 GPa. As discussed below, the new phase appears to be yet another solid form of nitrogen that is closely related to the cubic Pm3n structure of 6-N2 found at 4.9 GPa and room temperature. It is difficult to devise a reasonable notation for all of the phases that seem to arise from distortions of the Pm3n cubic structure in 6-N2. From Raman spectroscopy we cannot determine unambiguously the number of phases or their crystallographic symmetries. The problem is further complicated by the fact that the molecules in 6-N2 (Pm3n) show orientational disorder, making group-theuretical analysis difficult. For convenience we refer here to the first low-temperature high-pressure phase derived from 6-N2 as 6(LT1)-N2. This is the phase stable at 15 K between the transitions at 1.9 and 21.0-24.5 GPa. The phase that appears above 21.0-24.5 GPa at 15 K we call b(LT2)-N2. It is not obvious how these phases relate to the room-temperature phases. A relationship between the cubic 6-N, (Pm3n) and the 6(LT1)and 6(LT2)-N2structures is apparent from similarities in their distinctive Raman stretching-mode spectra. For each of these structures the stretching mode is divided into two main branches, U, and u2, with the lower-frequency u2 considerably more intense. It is, therefore, reasonable to regard the 6(LT1)- and 6(LT2)-N2
The Journal of Physical Chemistry, Vol. 89, No. 11, 1985 2327
Phase Transitions in Nitrogen
1 (a)
v , p l ,I, , FLU ID
O.l0
50
100 150 200 TEMPERATURE (K)
250
300
Figure 4. Provisional phase diagram of solid nitrogen: continuous solid lines and dashed line, ref 8; solid-line segment and cross-hatched segment, present work.
TABLE I: Space Groups That May Be Distortions of the Pm3n Structure of Cubic 6-N2 Hermann-Maugin full symbol
P 42fm 5 2/n
P2fm3 R 3 2/c R3c R5 R3 P 42fm 2 f m 2/c p 42f m P42mc p42 Pam2 P42c
P4 P 2fm 2fm 2 f m Pmm2 P222
I
2380
:
:
:
:
23m
:
: R)*wI
‘
:
2380
’
:
:
’ I :: 2390
: J
SHIFT bl‘)
Figure 3. Splitting of the lower-frequency v2 stretching-mode peak in N2 as a function of pressure at 15 K. (Higher-frequency v I peak not shown.) (a) 16.0 GPa, well below transition; (b) 21.0 GPa, start of transition; (c) 22.8 GPa, transition proceeds; (d) 23.8 GPa, transition proceeds; (e) 25.4 GPa, end of transition; (f) 27.4 GPa, well above transition.
structures as distortions of the cubic 6-Nz (Pm3n) structure. We have considered correlation diagrams for the Pm3n cubic structure of 6-N,, as well as distortions from this structure having cubic, rhombohedral, tetragonal, and orthorhombic space groups. The notation for the correlations follows Fateley et al.,25 with correlations being drawn from their Appendix 111. The notation (23) Kobashi, K.,private communication. (24) Agnew, S. F., private communication. (25) Fateley, W. G.; Dollish, F. R.;McDevitt, N. T.; Bentley, F. F.
‘Infrared and Raman Selection Rules for Molecular and Lattice Vibrations: The Correlation Method“; Wiley: New York, 1972.
short symbol Pm3n Pm3 R3c
Schoenflies symbol
G TI D!d
c !”
Si P4,fmmc
Ddh q h
c cf
:s:;: Pmmm
Dih
D2
C:,
no. 223 200 167 161 148 146 131 84 105 17 115 112 81 47 25 16
centrosymmetric? Yes Yes Yes no Yes no Yes Yes
no no no no
no Yes no
no
for the site symmetries and space groups is from “The International Tables For X-Ray Crystallography”, Vol. I.26 For each distortion the centers of the molecules are located at sites with symmetries that are subgroups of the 6c(D2) and 2a( Th)site symmetries in 6-N2 (Pm3n). Furthermore, we consider only structures with the same number of molecules per primitive unit cell as are in cubic 6-Nz (Pm3n); superlattices are excluded. The space groups meeting these criteria are listed in Table I. It is not necessary to invoke structures of lower symmetry or greater complexity to account for the number of Raman lines observed in 6(LT1)- and 6(LT2)-Nz. Unfortunately, the two space groups, 12,3 and R3c, each with 64 molecules per unit cell, which were recently suggested by Nos6 and Klein,14 are too complex to be confirmed or rejected on the basis of our data. It is not possible to specify uniquely the structures of 6(LT1)and 6(LT2)-Nz from our Raman data alone. We can, however, rule out structures for which the correlation diagrams yield fewer Raman-active modes than are actually observed. We are able also to make reasonable speculations as to which are the most likely structures, with the aim of guiding further theoretical and experimental efforts. (26) Hamilton, W. C. “International Tables for X-ray Crystallography”; Kynoch: Birmingham, 1974; Vol. I.
2328 The Journal of Physical Chemistry, Vol. 89, No. 11, 1985 TABLE 11: Cubic, Space Group Pm3n ( 0 i), Site Za(Th)O Molecular
Site
Sy m e t r y
Symmetry
Symmetry
Th
oh
D%
ll(N-N stretch)?
Ag
TABLE I V Rhombohedral, Space Group R h ( D ",), Site 2b(S6) Mo 1ecu 1ar
Factor Group
-q::1:
Schiferl et al.
Site
Symmetry Activity
D%
Symmetry ?-
S6GCCji
Factor Group Symmetry
.
D3d
Activity
Raman
--
TABLE V Rhombohedral, Space Group R k ( D k),Site 6e(C,)
'Note that the C,+(R,)mode does not exist for a diatomic molecule; it is included for formal completeness.
Site
Symetry
Symmetry
D%
Dzd
-
Site
Symmetry
Symmetry
D=h
TABLE 111: Cubic Space Group Pm3n (0i), Site 6c(DU) Molecular
Molecular
+
c2
Factor Group Symmetry
-
D3d
Activity
Factor Group
Sy m e t r y
'5 '
h'
+Oh
Activity
Before we can discuss the structural implications of the Raman data, we must understand the origins of the stretching-mode branches, as well as the numbers of Raman lines predicted and observed. At room temperature, the higher-frequency vl peak arises from the molecules centered on the 2a(Th) sites of the cubic Pm3n structure. From Table 11, a single peak may be expected. This conclusion is borne out by calculations by KobashLZ3 It is also consistent with the fact that, in the observed distortions of cubic 6-N2, the u1 peak never splits or broadens asymmetrically. In 6-N2 (Pm3n) the v2 branch is observed to have a single, symmetrical peak. However, Table I11 shows that two Ramanactive peaks, the A,, and Eg,are to be expected. This discrepancy is not difficult to reconcile, because either of the predicted peaks could be extremely weak, or the two predicted peaks could be unresolvable. What is important to note is that these two modes have Raman-active derivatives in all of the lower-symmetry structures discussed below. Thus, it seems unlikely that derivatives of the A,, and E, peaks are the source of the abrupt splitting of v2 at onset of the 6(LT2) phase. We now discuss the highest-symmetry structures consistent with the Raman data for each phase. Of course, depending on the magnitude of the distortions, any lower symmetry structure may also be a possibility. Structure of 6-N2. In 6-N2, for which no lattice modes are observable at room temperature, the Pm3n structure can account for the Raman,6*8infrared?4 and X-ray' data within experimental limits of resolution. Structure ofs(LT1). For 6(LT1), the cubic Pm3n space group is incorrect, because it can account for only four Raman-active lattice modes, as can be seen from Tables I1 and 111. Figure 2
shows that there are seven such modes. The remaining highsymmetry structures to consider are Pm3 (cubic), R3c (rhombohedral, Tables IV and V), P4,lmmc (tetragonal), and Pmmm (orthorhombic). The Pm3 cubic structures may have up to nine Raman-active lattice modes. However, the v1 stretching-mode peak should be split, because it arises from molecules a t two independent sites, la(Th) and lb(Th). On the other hand, calculations by KobashiZ3 indicate that the splitting may be only 0.1 cm-' and, therefore, unobservable. The R3c structure may have eight Raman-active lattice modes according to Tables IV and V. The stretching modes should look like those observed for Pm3n. In 6(LT1)-N2, the similarity of v , and v2 to their counterparts in 6-N2 increases the probability that 6(LT1)-N2 has the R3c structure. We believe this to be the case, but it cannot be proven with only the data at hand. Calculations by LeSar" indicate that R5c is the most stable structure for low-temperature N 2 at 2 GPa. The P42/mmc tetragonal structure may have up to eleven Raman-active lattice modes. In this structure a and c are both parallel to fourfold axes in the Pm3m space group. Because only seven lattice modes are observed, this structure would seem less likely than either the Pm3 or R3c structures. Moreover, the stretching modes should be divided into three branches with up to five Raman peaks. In b(LT1)-N2 the similarity of v , and u2 to their counterparts in cubic 6-N2 strongly suggests that any 1. However, tetragonal distortion should be small with c l a there is no way to pack the N z molecules so the P42/mmc symmetry is maintained with this c / a ratio. The problem is that the molecules centered on site 2f(D2J must be arranged end-to-end in strings, giving a very inefficient packing. Not surprisingly, lattice-stability calculations by LeSarI3 show that this structure
The Journal of Physical Chemistry, Vol. 89, No. 11, 1985 2329
Phase Transitions in Nitrogen TABLE VI: Rhombohedral, Space Group R3c(C :,), Site 2a(C3) Factor Group
Molecular
Site
Symmetry
Symmetry
S y m e t ry
c3
c3v
-
%l
-
Activity
TABLE VII: Rhombohedral, Space Group R 3 c ( C t.), Site 6b(CI) Molecular
Site
Symmetry
Symmetry
D%
C1
Factor Group Symmetry c3v
Activity
Raman, ir
-_ Raman, ir
-_ I”%,
‘E(2)
Raman, ir
Ty )
-E>
E(&)
Raman, ir
ng(R,, Ry)
has a much higher energy than either the R3c or R3c structure at all pressures considered. For the above reasons, P4,/mmc is not a very plausible structure for 6(LT1)-N2. The P4,/m tetragonal form has molecules with the same packing problems. Thus, both of the centrosymmetric tetragonal structures can be ruled out. The Pmmm orthorhombic form may have 19 Raman-active lattice modes, which is considerably more than the observed seven, and thus is also an implausible structure for 6(LTl)-Nz. Worse yet, we do not see a splitting of the uI mode into two peaks as required by the two different la(&) and 1h(DZh)sites. In summary, we believe the RJc structure to be the most likely one for 6(LTl)-N2. The eight predicted Raman-active modes are in close agreement with the observed seven. The u1 and u2 stretching modes can be similar to their counterparts in 6-Nz (Pm3n). Moreover, cohesive energy calculations by LeSar” show the RJc to be the most stable structure at 2 GPa. Structure of 6(LT2)-N2. The key, new feature that must be accounted for in the observed Raman spectrum of this phase is the splitting of the uZ stretching mode peak into two peaks, uza and uZb The number of lattice modes is uncertain, but is at least seven and may be as high as nine. Many of them appear to extend across the 6(LT1)-6(LT2) transition boundary, which is an indication that the two structures are very similar. If we grant that the Raman-active Al, and Egmodes contribute only to the single observed uZ peak in 6-Nz (Pm3n) and 6(LT1)-Nz ( R ~ cthen ) , we cannot invoke the argument that these two modes or their derivatives give rise to the uza and uZb peaks. A new Raman-active mode is needed. This automatically eliminates the cubic structures, Pm3n and Pm3. (Of course Pm3n is also eliminated because it permits only four Raman-active lattice modes). The centrosymmetric, rhombohedral structures R h and R3 do not provide any new Raman-active u2 modes. The noncentrosymmetric structure R3c, however, provides an extra E mode, which is derived from the Fzuof the Pm3n structure when the symmetry is reduced. The correlation diagrams for the R3c structures are given in Tables VI and VII. We believe that the R3c structure is the most probable one of those considered for 6(LT2)-Nz, although this cannot be estabilished unambiguously from the present data. The R3c requires a single u , stretchingmode peak, as observed. It also allows 20 lattice modes which are both Raman- and infrared-active. At first sight this would appear to be far too many when no more than nine are observed. Nevertheless, it may be that the changes in the crystal field with the loss of centrosymmetry are not large enough to create many new modes with strong Raman activity. Changes across the 6(LT1)-6(LT2) phase boundary must certainly be small. This can be seen from Figure 1 where there is not even a discontinuity in dul/dP at the transition. Furthermore, the lattice modes, which are normally very sensitive to crystal-structure changes, seem generally to continue a c r w the phase boundary with only a change in slope, as can be seen in Figure 2. Calculations by LeSarI3 up to 75 GPa, however, never show the R3c as more stable than R3c. The other noncentrosymmetric rhombohedral structure is the R3. It allows the u2 peak to split, but also requires the u1 peak
to split. Both splittings arise because the molecules in each case are at two independent sites, making a total of four independent sites in the crystal. This structure is, therefore, not very likely. The tetragonal and orthorhombic structures may be dismissed for the same reasons given previously, that they are unsuitable candidates for the related 6(LT1)-NZ structure.
Conclusions The structures of 6-Nz, 6(LT1)-Nz, and 6(LT2)-N2 are very closely related. We have considered all possible distortions of the Pm3n cubic to other cubic, rhombohedral, tetragonal, and orthorhombic structures in which (1) the number of molecules per unit cell does not change and (2) the resulting space group is a subgroup of Pm3n. Within the resolution of Ramaq6S8infrared,23and single-crystal X-ray diffraction’ experiments, 6-Nz at room temperature has space group Pm3n. The structure of G(LTl)-N, is probably a distortion of Pm3n with space group Rgc. This is the highestsymmetry space group fully consistent with the Raman data and is supported by cohesive-energy calculations of LeSar.13 The structure of 6(LT2)-Nz probably has space group R3c and may be regarded as a distortion of R3c as well as Pm3n. Of course, other structures based on 6-N2 (Pm3n),but with various crystallographic axes doubled or tripled, cannot be conclusively ruled out. The structures favored here are those which are the simplest, have the highest symmetry, and are consistent with the data at hand. The proposed sequence of transitions 6-N, (Pm3n) 6(LT1)-Nz ( R k ) 6(LT2)-Nz (R3c) is also supported by the following arguments. The minimum distortion required to obtain the R3c structure is simply a small shift of molecular centers from the high-symmetry DZdsite of Pm3n to the Cz site of Rjc. A further distortion to R3c requires only an additional shift of the molecular centers from the S6 and Czsites of R3c to the C3 and C1sites of R3c. In principle, the lattice constants need not change, although the angle a would probably deviate slightly from 90’. The picture that emerges tentatively is that, at temperatures near 0 K and pressures between 1.9 and 21.0-24.5 GPa, the R3c structure is the stable one. With increasing temperature, the higher symmetry Pm3n becomes stabilized by the increasing entropy. With higher pressure near 0 K, the molecular centers shift to sites of lower symmetry, perhaps to accommodate a closer packing of distorted N z molecules, although the cause of this shift is not clear. Low-temperature infrared and X-ray diffraction experiments are required to establish these structures with certainty.
-
-
Acknowledgment. We acknowledge stimulating and helpful discussions with S. K. Satija, S. F. Agnew, and R. LeSar. R. LeSar and K. Kobashi kindly provided us with their unpublished
J . Phys. Chem. 1985, 89, 2330-2335
2330
calculations on 6-N2. We are also indebted to J. M. Neff for patiently typing the manuscript, especially the correlation diagrams. S.B. expresses appreciation for a Graduate Research Assistantship sponsored by the Los Alamos National Laboratory's Center for Materials Science. This work was performed under the auspices of the U.S. Department of Energy, Office of Basic Energy Sciences, Division of Materials Sciences.
Registry No. N 1 , 7727-37-9.
Supplementary Material Available: An Appendix to this paper, giving correlation tables for all of the cubic and rhombohedral space groups listed in Table I, as well as the centrosymmetric tetragonal and orthorhombic structures, is available (33 pages). Ordering information is available on any current masthead page.
Flash Photolysls of Transient Radicals. 1. X;
with X = CI, Br, I, and SCNt
V. Nagarajan and Richard W. Fessenden* Radiation Laboratory and Department of Chemistry, University of Notre Dame, Notre Dame, Indiana 46556 (Received: May 22, 1984; In Final Form: February 12, 1985)
The radicals CIT, Bry, 12-, and (SCN),- were prepared by photolysis of appropriate chemical systems with one laser and were subsequently photolyzed with a second laser. The first three species were photolyzed at 355 nm, and (SCN)2- was photolyzed at 532 nm; 12- was also photolyzed at 700 nm. In each case, dissociation into the fragments X and X- was detected by a bleach in the absorption of Xy. The quantum yield for this process is about 0.1 except for C12- where the value is 0.2. In no case was electron photodetachment observed. Observation of the recovery of the original absorption allowed the corresponding rate constant to be measured. The values for CI + CI- (8 X lo9 M-' s-l ) a nd SCN + SCN- (9 X lo9 M-' s-') have not previously been measured directly. In the case of C1Y the bleach in absorption does not completely recover and the loss of absorption is dose dependent. Because the presence of acid allows a more complete recovery, it can be concluded that the product C1 atom is photolyzed by a second photon to produce OH and C1-. The quantum yield was determined to be about 0.5. This photoreaction is direct experimental evidence that the absorption band of C1 involves charge transfer from solvent. Detailed analysis of the bleaching and recovery behavior at high Cl- concentrations showed no time lag which could be attributed to the 2P,i2 C1 atom, implying a short lifetime for this species.
Introduction There exist a number of techniques for investigating the properties and reactivity of stable molecules in the liquid phase which have not been applied widely to transient species such as radicals. One such technique is laser flash photolysis. The experimental setup needed is very similar to existing laser photolysis systems with the important addition of a second laser to photolyze the transient species. We have undertaken a program to investigate the photophysics and photochemistry of transient radicals in solution by time-resolved optical absorption spectroscopy. For a start, systems of the type X y in aqueous solution (where X = I, Br, C1, and SCN) have been studied. These systems have strong absorptions in the near-UV or visible regions, are structurally simple, and should have easily interpretable photophysics or photochemistry. One would like to know if resonant excitation of these species in solution would lead to either dissociation
-
x2- x i- xor photodetachment
X,-
-
X, i- e-
The absorption maxima in water for the first three species] are similar to but slightly shifted from those in the gas phase2 and halide crystal^,^ and the bands are also slightly broadened. The absorption spectrum of (SCN)z-4 in other than the solution phase has not been reported. In fact, its existence in the gas phase is not known. From experiments2a*csd and theoretical calculation^,^ all the excited states of X2- are known to be dissociative ((SCN), excluded). For C12- in the gas phase, the efficiency of path 2 was found to be nonzero but less than 5%. Lee et a1.2dconcluded that negligible photodetachment would occur since the vertical ionization potential from their potential curves is 4.0 eV and greater than the photon energy at the peak of the absorption (3.65 eV). The electron affinity values are in the order C12 > Br, > I, so +This is Document No. NDRL-2589 from the Notre Dame Radiation Laboratory.
0022-365418512089-2330$01.50/0
that the photodetachment efficiency could be larger for Br2- and 1,- than for C12-. In any case, the gas-phase behavior cannot be expected to hold exactly for the aqueous phase where there may be shifts and/or distortions in the potential energy curves for the ionic and neutral species in addition to extensive solvation of the ions. Resonanceenhanced Raman spectroscopy of Cly, Bry, and 1,- in solution6 finds vibrational frequencies and anharmonicity values in reasonable accord with values based on theoretical potential curves for the gas phase. However, the dissociation of XT species under equilibrium conditions in water is known to occur' with an equilibrium constant of about M. Thus, it is likely that the adiabatic difference in energy of X2- relative to X and X- in solution is less than implied by the gas-phase potential curves. The Raman experiment appears to sample the potential curve without solvent rearrangement. The gas-phase potential energy curves for C12- are given in Figure 1. The curves for Br2- and I, are similar but with larger separations between the 2Pl/z and 2P3i2 states of the atoms. Nothing is known about (SCN),-, but because of the pseudohalogen nature of SCN, one might expect a general behavior that is very similar to that of the other halogen species. In this case, excitation to a bound state and consequent relaxation are also possible. In addition to these points, photolysis of X
