SPECTROSCOPY OF Ti0 AND Ti02 MOLECULES
3243
Spectroscopy of Titanium Oxide and Titanium Dioxide Molecules in Inert Matrices at P°K by N. S. McIntyre, K. R. Thompson, and W. Weltner, Jr.* Department of Chemistry, University of Florida, Gainesville, Florida 32601
(Received April 16, 19’71)
Publication costs assisted by the National Science Foundation
The lowest-lying TI state of Ti0 has been observed as a weak system at 8406 8 in absorption j n a neon matrix at 4’K. The ultraviolet bands recentoly observed by Pathak and Palmer i n the gas phase and attributed to Ti0 have also been observed at 3100 A in a neon matrix. As in the gas phase, the vibrational progressions seem to be due to two overlapping systems which are strongly perturbing each other so that a definite analysis for either Ti160 or Ti180 was not possible. The electronic transitions of Ti0 in the visible region have also been observed in krypton and xenon matrices. The infrared spectrum of trapped Ti02 yields the stretching frequencies, and a bond angle of -110’ is obtained from Ti isotope splittings. An emission system of TiO, was observed at -5300 A, terminating in the ground state.
Introduction
vious work.16 Also, the spectra have been observed in krypton and xenon matrices, where presumably the selection rules would be more easily broken down.I0 Although there are no definite indications of spinforbidden transitions, two new allowed systems have been observed, one of which is the uv system recently detected by Pathak and Palmer.20 The molecule TiOz has also been studied more thor-
The molecule T i 0 has been studied more thoroughly by optical spectroscopy than any other diatomic transition metal molecule. While the astrophysical importance of the molecule prompted earlier investigations, more recently theoretical c a l c ~ l a t i o n s ~have - ~ broadened the basis for a thorough understanding of the electronic structure of TiO. Five band systems, two triplet ( a , 7 ) and three singlet (0, 6 , p), have been rather thoroughly analyzed (1) A. Christy, P h y s . Rev., 33, 701 (1929). by gas phase emission s p e c t r o s ~ o p y . ~ ~ These - ~ ~ are (2) F. Lowater, Proc. Phys. SOC.(London), 41, 557 (1927). indicated in Figure 1. Emission bands observed in the (3) K. Wurm and H. J. Meister, Z. Astrophys., 13, 199 (1936). red-orange region4 have besn tentatively identified as (4) F. P.Coheur, Bull. SOC.Roy. Sei. Liege, 12, 98 (1943). arising from as many as three new system^.'^ One of (5) (a) J. G. Phillips, Astrophys. J., 111, 314 (1950); 114, 152 these systems, y’, has been confirmed to be a triplet(1951); (b) see ref 17 for the spectroscopic symbolism used here. (6) R. A. Berg and 0. Sinauoglu, J . Chem. Phys., 32, 1082 (1960). triplet transition by its appearance, along with the a (7) K. D. Carlson and C. Moser, J . Phys. Chem., 67, 2644 (1963). and y bands, in the absorption spectrum of T i 0 trapped (8) K. D. Carlson and R. K. Nesbet, J . Chem. Phys., 41, 1051 in solid neon and argon matrices.15 Two singlet states, (1964). lA and l8, observed as lower levels in emission (9) K. D. Carlson and C. R. Clayton, “Advances in High Temperastudies11,12are expected to lie close in energy to the 3Ar ture Chemistry,” Vol. 1, Academic Press, New York, N. Y., 1967, p 43. ground state,’ss and Phillips16 has placed the lowest IA (10) R. W. B. Pearse and A. G. Gaydon, “The Identification of level at about 581 cm-l above the X3A. However, the Molecular Spectra,” 3rd ed, Wiley, New York, N. Y., 1963. accuracy of that number is probably low, and it must (11) A. V. Petterson, A r k . Fys., 16, 185 (1959). be considered as uncertain by at least several hundred (12) A. V. Petterson and B. Lindgren, ibid., 22, 491 (1962). wave numbers. (13) C. Linton and R. W. Nicholls, J. P h v s . B ( A t o m . Molec. Phys.), 2, 490 (1969). Spin-forbidden transitions, which would definitely (14) B. Rosen, ”4th International Meeting on Molecular Spectroslocate the singlet manifold relative to the triplet manicopy, 1959,” Vol. 2, Pergamon Press, Elmsford, N. Y., 1962, p 533. fold, have not been identified in the gas phase spectra, (15) W. Weltner, Jr., and D. McLeod, Jr., J. Phys. Chern., 69, 3488 (1965). One might expect more ready relaxation of the spin (16) J. G. Phillips, Astrophys. J., 115, 567 (1952). selection rule in this heavier molecule than in many cases where such transitions have been o b s e r ~ e d . ~ ~ f(17) l ~ G. Heraberg, “Spectra of Diatomic Molecules,” Van Nostrand, Princeton, N. J., 1950. It was hoped that these weak transitions could be ob(18) G. Heraberg, M e m . SOC.Roy. Sei. Liege, Collect. 8, 17, 121 (1968) served in the less-crowded matrix spectra or that they (published 1969). (19) G. W. Robinson, J. Mol. Spectrosc., 6, 58 (1961); J. Chem. could be induced to appear in the solid state.lg For Phys., 46, 572 (1967). that reason the spectra of trapped T i 0 have here been (20) C. M. Pathak and H. B. Palmer, J . Mol. Spectrosc., 33, 137 measured over a greater energy range than in the pre(1970). T h e J O U T n d of Physical Chemistry, Vol. 76,N o . 81, 1971
X.
3244 SINGLET STATES
TRIPLET STATES D (I&(Str)-
- - - - -r -
s. MCINTYRE, E(. R. THOMPSON, AND W. WELTNER,JR. WAVELENGTHS
IN
A
“104.01
6697.07
-30 -25
Figure 2 . Absor.pt,ion bands neon matrix a t 4°K. Figure 1. Known energy levels and observed transitions of TiO. Arrow heads indicate whether transitions were observed in emission or absorption or both. For discussion of dashed levels, see text. For symbolism used to indicate electronic configurations, see ref 28.
oughly in the infrared, and an emission spectrum observed in the visible has been attributed to it.
Experimental Section The design of the furnace and cryostat assemblies used in this work has been described previously.15r21A molecular beam of Ti160was produced by the inductive or resistive heating of solid TiOz (Fisher reagent) in degassed tungsten or tantalum Ihudsen cells a t temperatures between 2200 and 2400°K. The beam was trapped on a CsI or CaF2 window at 4°K along with neon or argon gas in a concentration of -1:lOOO. Ti180 was prepared by passing 1802 (Miles-Yeda Ltd.) over an equimolar amount of powdered titanium (Spex Co.) at 1600°K in an inductively heated tungsten cell. High-purity neon, argon, and krypton gases (Linde) were precooled before deposition. The condensed TiOz system was observed to react with tungsten above 1900°K to produce various complex tungsten oxides with vibrational spectra in the region 1020-650 cm-l of the infrared spectrum.22 The intensity of these absorptions could be reduced, but not eliminated, by heating the loaded cell at operating temperature for about 1 hr. The conditioned cell was then reloaded with fresh TiOz prior to the actual deposition. To identify unambiguously the new electronic transitions in T i 0 and TiOz, the molecule was also evaporated from tantalum cells and the resultant tantalum oxide species were identified from an earlier matrix study of that system.23 Electronic transitions to states lying below 9000 8 in energy could be investigated using a CaF2 prism interT h e Journal of Physical Chemistry, Vol. 7 6 , N o . $1, 1971
(e
system) of Ti160 in a
change installed in a Perkin-Elmer 621 infrared spectrophotomoeter. The near-infrared region from 7000 to 10,000 A could be scanned at higher resolution using a Jarrell-Ash Ebert spectrometer equipped with an RCA 7104 photomultiplier tube cooled to 77°K. The uv region (2000-3500 A) was investigated using a deuterium lamp source and an RCA 7200 photomultiplier detector. Interchangeable gratings with blaze appropriate for each spectral region were calibrated with mercury emission lines. Emission of TiOz molecules in neon matrices was excited with an AH-6 high-pressure mercury lamp.
Results I . Near-InfTared and Visible Absorption Spectra of TZO. The visible absorption spectrum of a transparent neon matrix containing T i 0 molecules evaporated at 2300°K included the three strong band systems described previously.’j A strong infrared absorption at 997.2 cm-I could be identified with the T i 0 3A groundstate vibration frequency in the gas phase, AGII,“ = 1000.0 A new system of sharp absorption bands, labeled E , was observed with (0,O) band at 8405.5 in neon (see Figure 2). The E system, of lower overall intensity than the previously observed a, y, and Y‘ bands, also had a weaker (1,O) band than observed in previous systems. Table I lists band positions and vibrational analysis for both Til60 and Til*O. The ratio of AGl/,’(Ti180)/AGl,,’(Ti160) = 0.9539 is in good agreement with the calculated v’(Ti180)/v’(Ti160) = 0.9674. (21) W ,Weltner, Jr., “Advances in High Temperature Chemistry,” Val. 2, Academic Press, N e w York, N. Y., 1970, p 85; Science, 155, 155 (1967). (22) W. Weltner, Jr., and D. McLeod, Jr., J . Mol. Spectrosc., 17, 276 (1965). (23) 11‘. Weltner, Jr., and D. McLeod, Jr., J . Chem. Phys., 42, 882 (1965).
SPECTROSCOPY OF Ti0
AND
TiOa n/IOLECULES
3245
Table I: Infrared Absorption Bands of T i 0 in Neon and Argon Matrices a t 4°K (E System)
-Neon
7 -
A,
V'
0
-
Ti160
7
A
8405.5
AGv+1/z', V,
om-1
cm-1
,----
, -0 8 1 j T -,
r
x, A
8403.0
11,894
7807.1
12,805
2
7295.0
13,704
Argon--Til60 -
-
om-1
cm-1
AGv+'/z't A,
11,897
V,
om-'
8394
11,910
7798
12,820
870 7830.7
--.
_-_-
AGv+'/z', V,
911 1
~
om -1
910
12,767
899
TiI60in argon gives a relatively strong band a t 8394 weaker ones at 8297, 8166, and 7798 8. The 8394-A band does appear to be the counterpart of the 8406-8 band in neon so that the 7798-A band could be assigned as indicated in Table I. However, without the neon results, one would find such an interpretation of the argon spectrum rather dubious. There are also some bands in that region of the krypton spectrum, but an assignment is even more doubtful there. Although the y bands in neon and argon matrices are quite distinct and easily assigned in krypton and xenon, there is anomalous absorption in that region. The two spectra are rather similar, each showing only two definite bands, one relatively weak and the otherostrong and rather broad. The weaker band is at 6975 A in krypton and at 7097 8 in xenon, and it seems likely that it is the (1,O) bands of the y system. The (0,O) bands would then lie a t -7360 and -7580 8 in krypton and xenon, respectively, about where one would expect them to be shifted in these heavier matrices. There appears to be strong very broad background absorption in this region which does not occur in the lighter matrices. The two stronger bands occur at 6660 in krypton and 6797 8 in xenon and are not assignable to any of the three allowed systems, y , yl, or a. This could be a desired triplet absorption, but there is no confirming singlet evidence other than the possible intensity enhancement in the heavier matrices. The (1,O) bands for these transitions would lie at about the (0,O) positions of the y 1 system. The progressions in the y' system and especially the stronger a system are easily observed in the heavier matrices. All of the observed bands of T i 0 in krypton and xenon matrices are listed in Table 11. I I . Ultraviolet Absorption Spectrum of TiO. The ultraviolet absorption spectrum of a neon matrix containing T i 0 had a series of strong absorption bands in the spectral region from 3200 t o 2600 A. The major portion of the spectrum is shown in Figure 3. The spectrum has the shape of a broad Franck-Condon envelope with individual bands spaced about 500 cm-l apart and with increased complexity at the blue end of the spectrum. As Figure 3 indicates subtitution of l*O resulted in considerable alteration of the general appearance of the system, as well as shifts in individual band positions.
8 and
+
Table 11: Absorption Bands of T i 0 in Krypton and Xenon Matrices a t 4°K ---Kryp System
___-
t onaom-1
A,
u,
6926 6697 6644 6328 6013 5716 5484 5257 5043 4846 4658
14,434 14,928 15,047 15 798 16,626 17 490 18,230 19 017 19 ,824 20,630 21 462
a Bands were gen:rally determined to & I 5 A.
A,
Xenon5--v , om-'
7118 6829
14,045 14 639
6445 6118 5791 5590 5379 5162 4963 4772 4572
15 512 16,341 17 263 17 884 18 586 19 , 367 20,143 20,950 21,852
broad so that peak positions were
The ultraviolet spectrum of tungsten oxide species has very few bands in this regionlZ2and they are not part of the observed system. lroreover, these same spectra were observed when several evaporations of T i 0 were made in the presence of excess titanium, thus precluding the presence of TiOavapor on the basis of earlier mass spectrometric studies a t high t e m p e r a t ~ r e . ~ ~ The flame spectrum of T i 0 in this spectroscopic region has recently been reported by Pathak and Palmer.20 A triplet-triplet emission system, D-X3A, was proposed, having AG~,,' = 1035 em-l and AGl/," = 1004 cm-I. The latter is in approximate agreement with the ground-state value of 1000 cm-'. The mean wavelength for the four bands assigned to the (0,O) head is 3130 8 (31,940 cm-l). Other than the (O,l), (O,O), and (1,O) bands those authors were not able t o assign the vibrational progressions because the spacing was irregular. Unfortunately, the 4°K spectrum also shows the same kinds of features and seems to indicate that these bands arise from at least two perturbing electronic states. In analyzing the neon matrix spectrum the strongest bands have been treated as two overlapping systems belonging to TiO. The similar appearance and in(24) J. Berkowitz, W.A. Chupka, and M. G . Inghram, J. Phys. Chem., 61, 1569 (1957).
The Journal of Phvsical Chemistry, Vol. 75, A'O. 21, 1971
E.S. MCINTYRE, K. R. THOMPSON, AND W. WELTNER, JR.
3246
Tit60
~2879.0 r2969.6
I
I
r3117.5 r
/
3069.27
I
2900
3000
+A, Figure 3. Uhaviolet absoiption bands of TiI6O and
Ti180
2700
A
in a neon matrix at 4°K (wavelengths in hgstroms).
tensity pattern of alternate bands is taken as evidence of this. Table I11 lists band positions and differences between bands on the basis of such an assignment. Assuming the same magnitude of gas-neon matrix blue shift observed for other T i 0 triplet band systems,15 the first band in system I a t 3117.5 8 corresponds satisfactorily with the assigned (0,O) band in the gas phase study. The matrix value for AG,,,', however, does not agree with that for the gas phase, nor do the other differelices in either system I or I1 exhibit any regular anharrnonic behavior. Both systems thus appear to be perturbed a t several vibrational levels, as is perhaps also the case in the gas phase. The substituuseful in an earlier study for the retion of l80for l60, moval of perturbations, 2 2 apparently decreases the perturbation in one area, but intensifies it in another. Thc Journal of Physical Chemistry, Vol. 76, A.0. 21, 1971
2800
It does not appear to be possible to satisfactorily explain the appearance of the spectrum in Figure 3 by a simple energy level scheme based on mutual repulsion of vibrational levels in just systems I and 11. Also the (0,O) bands in the matrix are not clearly determined so that the numbering of vibrational levels in Table I11 is indefinite. There also appeared to be some weaker bands in the spectra which were not part of the system I and I1 bands. These have been listed a t the bottom of Table 111. These less intense bands were for the most part visible only in the regions where systems I and I1 appeared to be most perturbed. III. Spectroscopic Studies of Ti& The mass spectrometric study of the evaporation of T ~ O Z ( Sre)~~ ported the gas phase molecule TiOz as having an
3247
SPECTROSCOPY OF T i 0 AND Ti02 niIOLECULES
Table 111: Ultraviolet Absorption Bands of T i 0 in a Neon Matrix at 4°K
I
n' n' n' n' n' n' n' n" n'! n'' n" n'! n'
I1
A,
V'
+0 +1 +2 +3 +4 +5 +6 +0 +1 +2 +3 +4 +5
.i
3117.5 3013,l 2923.5 2842.6 2779.1 2695.7 2623.1 3061.0 2969.6 2879.0 2805.0 2712.1 2681.9 3178.5 3084.7
Other T i 0 bands
----
Til80----
7
Band system
2839.8 2764.5
2821.5 2737.6 2668.0
ACv+l/r'$ Y.
cm-1
32,068 33,179 34 , 196 35,169 35 ,972 37,084 38,111 32 660 33,665 34,724 35,640 36,861 37 698 31,452 32 ,409
,
35,212 36 , 162
35 ,432 36,518 37,470
9620 CM-'
9972 CM-'
9348 CM-'
Figure 4. Stretching frequencies in the infrared spectra of Ti0 and TiOz molecules trapped in a neon matrix a t 4°K.
enthalpy of sublimation slightly greater than that of TiO. In agreement with the previous matrix isolation study,14 evaporation of a fresh sample of Ti02 into a neon matrix resulted in the appearance of a strong infrared absorption band a t 934.8 cm-l. Closer examination (see Figure 4) revealed four weaker bands symmetrically bracketing the main peak that are
om-'
x, .i
Til80------Y,
om-'
7
Af&+'/Z', cm-1
1111 1017 973 803 1112 1027
3131.7 3024.3 2935.8 2850.0 2776.1 2698.3
31 ,922 33 ,056 34 ,052 35 077 36,011 37,049
1134 996 1025 934 1038
1005 1062 916 1221 828
3069.2 2979.5 2891.8 2813.9 2737.4
32 , 512 33 ,553 34 ,570 35 ,52 1 36,520
1041 1017 951 999
3008.3 2916.8
33 232 34,274
1042
3050.4 2964.8 2881 .O
32 ,773 33,719 34 ,700
946 981
,
979
947
1086 962
attributable to the less-abundant isotopes of titanium. The overall intensity distribution of the five peaks closely corresponded with the natural abundances of the titanium isotopes. Annealing of the matrix resulted in no isolated changes in this group of bands. Individual positions were 4eTi02,940.7 cm-l; 47TiOz, 937.6 cm-l; 48TiOe,934.8 cm-'; 49Ti02,931.8 cm-l, and 50TiOz, 928.7 cm-l. All measurements have a relative precision of A0.2 cm-l. I n addition to weak tungsten oxide bands, observed in earlier work, and the T i 0 band a t 997.2 cm-', a new infrared absorption of moderately weak intensity was found at 962.0 cm-l which only appeared when the 934.8-cm-1 band was present. For a bent triatomic oxide (C2usymmetry) two stretching frequencies are allowed. The strong band at 934.8 cm-l is assigned to vat the asymmetric stretch and the weaker band at 962.0 cm-l to vl, the symmetric stretch. The bending frequency, v 2 , was not found. Weak emission bands in the green region were observed when a neon matrix containing Ti02 was irradiated a t -3800 using an AH-6 mercury lamp (see Figure 5 ) . Analysis of the bands in Table IV indicates only one progression with AGII2'' = 959 cm-l. A bent-bent or linear-bent electronic transition between excited and ground states will lead to a progression in v1 and/or VZ. Thus the close agreement betwcen AGl,2't and the infrared band at 962.0 cm-' is considered as support for assignment of tho latter as VI, if, The Journal of Physical Chemistry, Vol. 76,IVO.21, 1971
S. MCINTYRE, K. R. THOMPSON, AND W. WELTNER,JR.
3248
%I.
-5578 5 I
Figure 5 . Emission spectrum of Ti02 in a neon matrix a t 4°K (wavelengths in Bngstroms).
Table IV : Emission System of TiOz in a Neon Matrix a t 4°K
as seems likely, the emission terminates in the ground state. The isotopic shifts observed for v3, the asymmetric stretching frequency, can be used t o calculate an approximate 0-Ti-0 bond angle. For a Czvmolecule, the angle a is obtained from the relationz5
41)
[p(l60)
+ pL(Ti(l))(l
- cos
dl
[La] = Il(“o)+p(Ti(2))(1 - cos a)] where p is the reciprocal mass of the isotope in question and w 3 is the harmonic asymmetric stretching frequency. Taking all possible combinations of the five isotopes, the calculated bond angle was 110 f 15”.
Discussion
several points t o be noted about the observations relative to this assignment. The upper state of the E-X transition in neon falls a t Too = 11,894 cm-l. Assuming that Phillips’ value of 581 em-l is correct for the alA-XaA spacing, the b111(9u)(47r) state (in the gas) is found to lie a t Too= 11,854 cm-I, which is then actually below the assigned E311state. The reverse is expected to be true, but this discrepancy can be removed if the a-X spacing is really larger than 581 cm-I, which could easily be the case (particularly in the matrix). However, it seems likely that the b1rI-E3rI spacing would probably be less than about 1000 cm-’ even then. Such a singlet-triplet spacing would then be of the same order of magnitude as the a1A-X3A energy of the (16)(9u) configuration, but it is considerably smaller than the C1@-A3@spacing of the (16)(4a) configuration (4200 cm-l). The relatively low-lying electronic states of T i 0 are associated with orbitals which are largely atomic 3d and 4s centered on the metal and these configurations can perhaps be more informatively designated as U4sT3d*, Uls63d, and 63d‘rr3d* [corresponding to (9u) (h), (16)(9u), and (16)(4a), respectively]. On this basis, similar correlation energies in the u4s~3d*and U4883d configurations as compared t o the 63d73d* seems reasonableS29 A second point to examine in the assignment of the E bands is the distribution of intensity in the observed vibrational progression in the neon spectrum. The (0,O) band is considerably stronger (see Figure 2) than the (1,O) whereas longer progressions have been observed in the other matrix systems.l5 However, upper state vibrational frequencies in these a, 7 ,and 7’ systems are all considerably lower, ranging from 866 to 838 cm-l, than the AG,,,’ = 911 cm-l (see Table I) derived from the E bands for the E311level. If the vibrational frequency can be used, as is often approximately true, as an indication of the rotational constant in that state, then one expects a B value in the E311state closer to that in the ground state. Then the intensity in the vibrational progression in the E-X transition should be shifted toward the (0,O) band relative to the other matrix systems, as observed. The remaining question is why the overall intensity of this system is so small. It is significant here that the E311+- X3A transition is the only one in the triplet manifold which, if classified jn terms of metal atom orbitals, involves a d-d transition; all others observed
TiO. The excited electron configurations and electronic states of the Ti0 molecule have been discussed p r e v i o u ~ l y , 1and ~ ~the ~ ~present ~ ~ ~ status is indicated in Figure 1. The Ban state has not been definitely (25) G. Heraberg, “Infrared and Raman Spectra of Polyatomic identified in the gas statej4rI4but a series of strong abMolecules,” Van Nostrand, Princeton, N. J., 1945. sorption bands do appear in neon matrix spectral6 and (26) C. J. Cheetham and R. F. Barrow, “Advances in High Temhave been designated as the y‘ bands, after C ~ h e u r . ~ perature Chemistry,” Vol. 1, Academic Press, New York, N. Y., 1967, p 7 . The only as pet unidentified triplet state expected t o be (27) L. Brewer and D. W. Green, High T e m p . Sci., 1, 26 (1969). located in the infrared region is the T I ( 9 ~ ) ( 4 a )and , ~ ~it (28) The symbolism for the electronic configurations of T i 0 used seems reasonable to assign the weak matrix E bands to here is that of ref 7, 8, and 9. this E311 + X3A transition. There are, however, (29) K. D. Carlson and C. Moser, J . Chem. Phys., 46, 35 (1967). The Journal of Physical Chemistry, 7’01.
76,NO.21, 1071
INFRARED
3249
STUDY OF BC13 CHEMISORBED ON SILICA GEL
involve excitation of the 4s electron into a ad or r d orbital. The ultraviolet spectra of T i 0 in matrices supports the findings of Pathak and Palmer,20indicating that there is a t least one triplet D level a t about 32,000 cm-l. The matrix bands indicate the probable presence of a t least two states lying close to one another. These two triplet states are indicated in Figure 1 by a dashed line and derived from a configuration (16)(5r) where the 5 r orbital is assumed to be largely titanium 4p. I n the isoelectronic ScF molecule two states, and 3@, have been observed to occur in this energy region and probably also arise from that configuraLarge Franck-Condon envelopes for the T i 0 bands indicate a large change in bond distance in the upper states, supporting Cheetham and Barrow's suggestionz6 that p electrons contribute little to the bonding in these molecules. TiOz. Electric deflection investigationao of molecular TiOz indicates that it has a permanent dipole moment and therefore a bent structure. This is corroborated by our infrared work. However, a comparison of the bond angle derived here (110 i 15') with
that of the isoelectronic molecule CaFz (140 i 5')3l indicates that there is little geometric similarity. CeOz may also be considered as isoelectronic with TiOz, and it is interesting that the emission systems of these two molecules observed in neon matrices are so much alike. The (0,O) band of CeOz 3 z occurs a t 5100 b as compared to 5300 b for TiOz, and the FranckCondon envelopes are quite similar. The emission lifetime, which in the case of CeOz is very long (-0.2 sec), was not measured for TiOz. It can also be noted that TiOz is another molecule, like CeOz, TaOz, Thoz, and ZrOz, which contradicts the general rule that v3 > v1 for the ground-state vibrational frequencies.32
Acknowledgment. This work was supported by the National Science Foundation under Grant NSF GP25411. (30) M. Kaufman, J. Muenter, and W. Klemperer, J . Chem. Phys., 47, 3365 (1967). (31) G. V. Calder, D. E. Mann, K. 5.Seshadri, M. Allavena, and D. White, (bid., 51, 2093 (1969). (32) R. L. DeKock and W. Weltner, Jr., J . Phys. Chem., 75, 514 (1971).
Infrared Study of Boron Trichloride Chemisorbed on Silica Gel' by Victor M. Bermudez U.S. Naval Research Laboratory, Washington, D . C. t203.90 (Received M a y 3, 2071) Publication costs assisted by the U.S. Naval Research Laboratory
The mid-infrared spectrum of BC13 chemisorbed on aerosil silica gel, in the form of a loose powder supported on a transparent plate, has been measured, together with the spectrum of the hydrolysis product. The spectra are quite rich in the 1500-500-~m-~region, and the effect of high-temperature evacuation of the substrate prior to chemisorption suggests that a distinction must be made between reaction with isolated and with geminal free hydroxyl groups. A proposed model for the chemisorption, whereby isolated hydroxyls form a nonbridging species =Si-O-BC12, geminal groups a bridging species =Si(-0-)2BCl, and hydrogen-bonded hydroxyls another bridging species (=Si-O-)2BCl, is shown to be consistent with the trends observed in the spectra and with earlier studies of BC13-Si02 systems and capable of resolving the discrepancies in the value of the free hydroxyl concentration obtained by different methods. An attempt is made to assign the observed absorptions to specific normal modes of each species, although the mid-infrared spectra alone provide insufficient data for a complete assignment. The experimental problems involved in obtaining spectra of small concentrations of highly reactive surface species are discussed in detail.
Introduction The reactions of the surface hydroxyls of silica gel and porous glass \vith inorganic and organic halides (particularly chlorides), both in gas phase and in solution, have been the object of numerous studies during the past decade.2 Much of the effort has been directed toward achieving an understanding of the surface
structure of silica-based adsorbents by utilizing these quantitative reactions to measure the concentration of surface hydroxyls and to distinguish among the possible configurations ( i - e . , vicinal, geminal, or isolated). (1) Supported in part by the Advanced Research Projects Agency, OrderNo.418. (2) See ref 3-7, 9, and io and references quoted therein.
T h e Journal of Physical Chemistrv, Val. 76, N o . $1, 1971
