ALANSNELSOK AKD KEXNETH S.PITZER
puted to be DO,,,(CaF) = 124.8 kcal./mole (5.41 e.v.), in excellent agreement with the value obtained from the homogeneous equilibrium. The value obtained from the heterogeneous equilibrium is considered less reliable than that from the homogeneous equilibrium because of the necessary use of the instrument pressure calibration and the uncertainty of the activity of CaF(s). Heats for various reactions in the Al-CaF, system are presented in Table V. TABLE
Reaction AlF(g) = Al(g) Ca(g) CaFW A 1 F ( d = Al(g) F(g) CaF(g) = C a ( d F(g)
+ l/zFz(g) = C a ( d +
F(d Ca(s) 4- '/zFz(g)
Ca(g) CaFz(s) = 2CaFfg) CaFzis) = Ca(g) 2F(g) CaF(g) = Ca(g) F(g)
THE ~Al-CaF2 ~
AHQzw (kcal./rnoIe) DOzu (e.v.) 3 4 . 6 rt 2 . 7 32.7 i 0 . 2 167.7 2 . 0 6.85 ==! 0 . 0 9 123.1 rt 4 . 7 5.35 f 0.20 125.0 i 2 . 2 5 . 4 4 i 0.10
61.1 -62.0 -63.9 120.7 370.3 124.8
Method Second law Third law Available data Secondlaw Thirdlaw
i 1.5 f 6.2 i3.7
Available data Second law Third law 0.8 Third law i2.0 Available data i 2 . 8 5 . 4 1 =I=0.12 T h i r d l a w
It is difficult to explain the difference in dissociation energies obtained fram spectroscopic and mass spectrometric studies if a true predissociation a t 3.15 e.v. was observed by Hellwege'O and Harvey.'l It may be, however, that what was observed was the forniation of headless bands. Such a head for a sequence does not mean a dissociation or predissociation limit, but does explain the intensity distribution observed. The possibility that a perturbation was responsible for the ob-
served intensity distribution cannot be excluded definitely either. Since the difference between the ionization potential of calcium and the electron affinity of fluorine is only 2.6 e.v., it is certainly possible that CaF may be an ionic (rather than atomic) molecule.9 The ions Ca+(%) F-(lS) may give the observed ground molecular state '9+. The non-crossing rule would then have this ionic state dissociate to the ground state atoms Ca(lS) F(2P)and the atomic Q+ would correlate with the atomic ions. Continuing work in this Laboratoryz9on the alkaline earth fluorides has yielded similar high values for the dissociation energies of SrF(g) and BaF(g), supporting 5.4 e.v. as the correct value for D(CaF). Acknowledgments.-The authors are pleased to acknowledge the financial support of this work by the National Science Foundation, the United States Atomic Energy Commission, and the Wisconsin Alumni Research Foundation. In addition, the helpful suggestions and advice of Drs. W. A. Chupka and J. Berkowitz of the Argonne National Laboratory are gratefully recognized. Mr. I. Reithmeyer's careful workinanship in machining the inner parts of the instrument deserves our highest praise. Above all, we wish to thank Professor M. G. Inghram of the Department of Physics, University of Chicago, without whose assistance the construction of the mass spectrometer would not have been possible. During part of the time these investigations took place, G. D. Blue mas a Shell Fellow.
REACTIONS ~ FOR ~
(29) G , D. Blue, J. W. Green. T. C. Ehlert, and J . L. Margrave. BUZZ. A m . Pfiys. f l o c . , 8, 112 (1063).
ISFRARED SPECTRA BY MATRIX ISOLATIOS OF LITHIUM FLUORIDE, LITHIUM. CHLORIDE, AND S0DIUR.I: FLUORIDE BY ALANSKELSON ASD KEXXETH S. PITZER~ Department of Chemistry and Lawrence Radiation Laboratory, University of California, Berkeley, California Received October 19, 1962 The matrix isolation method, which was adapted for infrared spectroscopy of molecular species existing at high temperatures by Linevsky, has been employed to examine the spectra of LiF, LiCl, and NaF. Rather complex spectra were found. Most of the features of these spectra must arise from polymeric alkali halide species. Spectra were taken for Ar, Kr, and Xe matrices and the effects of matrix material are considered. Isotope effects as well as spectral shifts after matrix diffusion assist in the interpretation of these data.
Introduction At the present time there is much interest in the vapor species in equilibrium with materials at high temperature. Mass-spectrometric techniques have been widely used to determine vapor pressures and equilibrium data for a large number of gaseous-solid systems. Unfortunately, this approach does not give any information concerning the structure or vibration frequencies of the gaseous species, which are of interest in their own right, and from which entropy data may be calculated. This latter quantity is often required to make full use of the mass-spectrometric approach, but usually has to be estimated through lack of experimental data. One of the oonventioiial methods for obtaining just (1) Rice University, Houston, Texas.
this information is optical spectroscopy. However, there are experimental difficulties in observing the spectra of high-temperature species in the gas phase. These are due mainly to the chemical reactivity of the vapor species with the containing vessel, the complexity of the spectra because of highly excited vibration and rotation interactions, and the difficulty of obtaining a sufficiently high concentration of the required species for the spectra t 3 be observed. These difficulties are largely over-come, if the matrix isolation technique, as developed by Becker and Pimentelz and adapted to high temperature systems by L i n e ~ s k yis, ~used to observe the spectra. A Knudsen cell is used to produce a molecular bean1 of the gaseous (2) E. D. Becker and G. C. Pimentel, J. Chem. Phus.. 25 224 (1956). (3) hI. J. Linevsky, abzd., 84, 587 (1961).
SPECTRA BY UATRIX ISOLATION OF LiF, LiCl,
species, which is simultaneously condensed with a stream of matrix gas on a window cooled to liquid hydrogen or helium temperature. The ratio of matrix to effusing species is kept sufficiently large to obtain effective isolation of the effusing species, and the spectrum is observed when a large enough amount ol the active species has been deposited. The matrix isolation technique is used in the present work to investigate the infrared spectra of lithium fluoride, lithium chloride, and sodium fluoride. Experimental Apparatus.-A molecular beam furnace and an optical system arc combined vith an all-metal cryostat of fairly conventional design. The cesium bromide window on which the matrix is formed, is housed in copper block, soldered into the bottom of a %in. diameter stainless steel tube, forming a hydrogen reservoir of about 1 1. capttc-ity. Good thermal contact between the cesium bromide window and its housing is obtained by compressing thin strips of indium foil between 1he window and thc copper block. The temperature of the window is measured by a copperconstantan thermocouple soldered to the houdng. The e.m.f. is recorded on a Leeds and Northrop d.c. Microvolt amplifier. The hydrogen reservoir is surrounded by a closed copper cylinder, cooled t o liquid nitrogen temperature t o minimize radiative heat leak to the hydrogen pot, in which suitable holes are cut to allow the matrix to be deposited and thespectraobserved. The entire low temperature assembly of the cryostat is supported on a, ballrace, and can be rohated with respect to the vacuum housing; the vacuum at the rotating joint being retained by two rubber O-rings compreseed radially. Two cesium bromide windows, sealed with O-rings to thev acuum jacket opposite the cooled window, permit spectral observations to be made on the matrix. The matrix gas enters the cryostat through a 2-mm. diameter jet and is directed a t an angle of 45' on to the cooled window. The matrix gas flow is metered by a "Nupro" vacuum valve. The molecular beam furnace is bolted t o a flange on the cryostat. The Knudsen cell made of 0.020 in. platinum sheet is cylindrical, 0.75 in. in diameter, and 0.75 in. long, with flat ends. The effusion hole in one end of the cell is 0.025 in. in diameter and 0.25 in. off center. The Knudsen cell is enclosed in a thick-walled tantalum receptor to minimize temperature gradients within the cell a t elevated temperatures. The receptor is mounted on the end of a thin-walled tantalum tube. A small hole drilled in the base of the receptor enables the temperature of the Knudsen cell to be measured by an cptical pyrometer, viewing the Knudsen cell through a window in the furnace housing, or by an iron-constantan thermocouple mounted in place of the window, the high temperature junction being trapped between the Knudsen cell and its housing. The Knudsen cell is heated electrically by two singleloop tungsten filaments. Resistance heating is used up to temperatures of goo", the power being supplied by a low-voltage (8 v.) 800 watt transformer. Above 900" electron bombardment heating from a three kilowatt (3,000 v. a t lamp.) d.c. poFer supply is used. A maximum temperature of about 2400" may be obtained. To minimize heat losses, the Knudsen cell is surrounded with tantalum radiation shields. The outer jacket of the furnace is water-cooled. The molecular beam is collimated by a single 5/16-in. diameter hole in a disk of sheet platinum located between the Knudsen cell and the cesium bromide window. The evacuation of the cryo8tat and furnace is provided by an oil-diffusion pump and a mechanical pump. The pressure in the system is measured by an ion gage mounted in the furnace housing. A Perkin-Elmer 421 double beam infrared spectrophotometer is used to record the spectra. The Dual Grating interchange is used for the region 4000-550 cni.-', and a cesium bromide interchange t o cover the region from 550 to 260 em.-'. The calibration of the spectrophotometer is checked during each experiment against atmospheric HzO and COZbands. General Procedure.-The experimental technique used for the matrix isolation of all the halides was essentially the same. The platinum KnudHen cell was filled with about 4 g. of sample, and the lid welded directly to the body of the cell. The Knudsen cell was placed in the furnace, evacuated and heated to a temperature such that about 0.03 g. of material effused from the cell per
hour. (This tvas found by trial and error). The cell was then heated from 50-100" above this temperature until about 1 g. of material had effused out. In this way it was hoped that any volatile impurities in the sample were evaporated off. To prevent the sample from being deposited on the cesium bromide window during the above process, the window was turned to a position in which the molecular beam was cut-off by the surrounding copper heat shield. Care was always taken a t this stage to keep atmospheric moisture out of the furnace and cryostat during any disassembly operation, and with this precaution a vacuum of about 1 X 10-5 mm. was obtained after three hours of pumping at room temperature. Prior to performing a matrix isolation experiment, the cryostat was aligned in the sample beam of the spectrophotometer, the matrix gas line connected, and the whole system left evacuating overnight. At the same time atmospheric moisture was removed from the spectrophotometer by continual flushing with dry nitrogen obtained from the evaporation of liquid nitrogen in a 25-1. dewar. The matrix gases, argon, krypton, and xenon, supplied by the Linde Company were all purified by passing over copper turnings at 600-700" to remove oxygen, Ascarite and magnesium perchlorate to remove COZ and HzO, respectively, and finally activated charcoal cooled to solid carbon dioxide temperature to remove any remaining adsorbable impurities. The gases were collected a t a rate of about 50 cc. per hour and stored in 10liter reservoirs. The purity of the gases was checked spectroscopically by condensing about 200 cc. on the cooled window, and observing the spectra of the entire region 4000-260 cm.-'. For a matrix isolation experiment, the cryostat was cooled to a liquid hydrogen temperature (20-21 "K.) and the spectra recorded the entire region from 4000 to 260 cm.-'. The Knudsen cell was heated a t the predetermined temperature until the pressure in the furnace was in the l o w 6mm. range (usually about 30 minutes). The matrix isolation was started by rotating the cooled window into a position perpendicular t o the molecular beam and starting the matrix gas flow. The pressure in the furnace during isolation was of the order 2-3 X 10-4 mm. The deposition of the matrix was interrupted periodically and the spectra of a particular absorption band recorded. When a suitable spectral absorption intensity has been obtained the deposition was terminated. After the spectrum had been observed over the entire region, the matrix was allowed to warm to a somewhat higher temperature to allow some diffusion of the trapped species within the matrix to occur, recooled to hydrogen temperature, and the entire spectrum again observed. Materials.-Lithium fluoride was obtained from spectroscopic grade windows supplied by the Harshaw Chemical Company. Lithium chloride and sodium fluoride were reagent grade supplied by Baker and Adamson, specified purity of 98 and 99%, respectively .
Results and Discussion I n all matrix isolation experiments the mole ratio of matrix gas ,If to halide H cmdensed on the window is calculated. A nominal value of 33 is computed on the assumption that all the matrix gas entering the cryostat is condensed on the cooled window. Although this assumption is not realized experimentally, as evidenced by a 3-4-fold increase in pressure in the cryostat during the isolation experiment, a more reliable estimate does not seem possible. Since the same method of computation is used in all experiments, internal consistency is maintained. The value of H was computed from the weight loss of the Knudsen cell during the experiment and the geometry of the system. As a check on this method of calculating H, an experiment was made in which lithium fluoride alone was condensed 011 a cool polished metal plate in place of the cesium bromide window. The mass of lithium fluoride deposited on the plate was computed from the resulting diffraction pattern and was in good agreement with the value calculated from the weight loss of the Knudsen cell. Preliminary experiments with lithium fluoride showed that isolation of the active species may be achieved with
X L - ~ SSELSON N AND KENNETH S. PITZER
This was shown by the absence of any absorption bands in the spectrum o n recooling to liquid hydrogen temperature. The spectrum of pure lithium fluoride condensate was observed with negative results. Typical matrix isolation spectra of lithium fluoride in argon, krypton, and xenon before and after warm-up are shown in Fig. 1 and 2 , respectively. The absorption frequencies are listed in Table I. TABLE I MATRIXISOLATION SPECTRA OF LITHICM FLUORIDE sea.
Fig. 1.-LiF 0.8 -
M/H: M / H : 550 06 06 0 Moles L i F :IxIO-'
c m-I. spectra of matrices as deposited.
LiF in Krypton M/Hx 3600 Moles L i F = l x 1 0 - 8
f 0.4 -
c m:'. epectra after alloring diffusion in the matrices.
M/H greater than 200. However, for a given value of H, values of M/H < 1000 resulted in a marked increase in the intensity of some absorption bands which previously at M / H > 2000 had been quite weak. To avoid complications due to poor isolation of the active species aiid to maintain consistency, 31," > 2000 has been used in most experiments. The effectiveness of the isolation achieved with the three matrix gases at J I / H less than 1000 was found to decrease in the order xenon, krypton, argon; and with xenon the spectrum recorded with M/H = 700 was not noticeably different from that obtained with M,'H = 2000. Presumably, xenon with the highest freezing point, forms a rigid matrix more rapidly a t liquid hydrogen temperature than krypton, and similarly krypton more rapidly than argon. Lithium Fluoride.-The lithium fluoride matrix isolation experiments were made with the Knudsen cell heated at 920 * 20". Isolation times of two or three hours were common. In warm-up experiments, the temperature of the window housing was raised 20, 30, and 45OK. above liquid hydrogen temperature for the argon, krypton, and xenon matrices, respectively. The higher the annealing temperature used, the greater the light scattering of the matrices became. Argon and krypton presented no problems in this respect, but with xenon it was difficult to raise the temperature sufficiently to obtain diffusion of the active species and yet still be able to record the spectrum. Effective isolation of the trapped species was destroyed at temperature approximately 15' higher than those stated above.
888 883 840 835 774 764 759 734 727 719 706 649 643 627 623 550 53 7 498
Frequencies, om. -I-----. Krypton
732 726 717
552 540 496
533 545 498
The fundamental frequency of the lithium fluoride in the gas phase has been recorded by T'idale4 a t about 900 c m . ~ ~Linevsky3 . has observed the spectra of lithium fluoride trapped in matrices of argon, krypton, and xenon, and reports values of 840 and 835, 830, and 823 cm.-l for the fundamental frequencies of the monomer in the respective matrices. The frequencies reported in Table I (sec. b), for the argon and krypton matrices are in agreement with the above values. However, in the present work, the monomer band in xenon is split, the major peak a t 820 aiid a minor one a t 805 cm.-l, both a t lower frequency than the single band observed by Linevsky. It is probable that even had the splitting been present in Linevsky's experiments, it would not have been observed since he worked a t low spectral absorption intensities. The frequency shift between the gas and matrix bands of lithium fluoride in xenon is about 1Oyoand it is conceivable that the small difference in frequency reported above, 3 em.-', may be due to an impurity in the xenon sample used by Linevsky. I n this work, the purity of xenon was >99.99yo according to a mass-spectrometric analysis. The absorption band occurring a t a in Fig. 1 is due to the monomer of lithium fluoride containing the Li6 isotope. The fundamental frequencies are listed in Table I (sec. a). The calculated value of wi]w where wi and w are the fundamental vibration frequencies of Li6F and L i T , respectively, is 1.059. This agrees well with the average value of 1.057 found experimentally for the three matrices. The remaining absorption bands shown in Fig. 1 (c, d, e, f , and g), are attributed to lithium fluoride polymers. It is w l l established6 that the vapor of (4) G. L. Vidale, J. Phgs. Chem., 64, 314 (1960). ( 5 ) R . F. Porter and R. C. Sohoonmaker, .I. C h e m . P h y ~ . ,29, 1070 (1958).
SPECTKA BY MATRIXISOLATIOS OF LiF, LiCl,
lithium fluoride contains a large proportion of dimer, Iii2F2, and also soime trimer Li,E',. Recent electron dif'fraction studies* have shown that lithium chloride dimer has a planar rhombic structure, Three infrared active frequencies are expected for such a lithium halide rlimcr of D21,symmetry. Two of the vilxations are inplane stretching motions (BZUand Baa), and the third (BIJ, an out-of-plane bending. The structure of the trimer is not kno~i~ii, but assuming a plane hexagon of D a h symmetry, there are seven infrared active frequencies (Az and 3E). I n Table I1 the abundances of the isotopic species of lithium fluoride dimer and trimer are listed.
Berkowitz7 predicts the bending frequency at 385 cm.-'. The gas phase spectrum of lithium fluoride has been reported by Klemperer and NorrisB; absorption bands a t 640 and 460 em.-l are attributed to the dimer stretching frequencies. This may be compared with the frequencies obtained in this work of approximately 620 TABLEI1 RELATIVEABUNDAKCIC O F DIMERTCB N D TRIMERIC LITFIIUM and 537 cm.-l. If the interpretations of the gas phase and matrix isolation spectra are correct, frequency shifts SPECIES of -20 and +80 cm.-' occur in the matrix. Kegative 857 0 Li72 Li$ 79 5% matrix frequency shifts are fairly common and may be in14 Li+JLiT2 Li6Li7 19 terpreted in terms of attractive interactions between the Li6,Li7 2 1 Lie2 Lie, 0 isolated species and the surrounding matrix. The net effect is to reduce the effective value of the force conThe frequency shift due to the mono-isotopic substistant between the vibrating atoms and thus lower the tution of one Li6 for Ii7 in the dimer cannot be calculated vibration frequency. Positive matrix shifts are less exactly, but a rough approximation may be made by common and in the present case may be due to the applying the Redlich-Teller product rule to calculate matrix cage restricting the amplitude of the vibrational the isotopic frequency shift for Li7,F2 and Li6,Fz motion and hence increasing the frequency. The form and assuming half this shift for the mono-substituted of B2, and Bsuvibrations is shown in Fig. 3 . dimer. This results in a value of wi/a = 1.03 for each of I n the lithium fluoride dimer, the amplitude of vibrathe infrared active frequencies, where wi is the frequency tion of the lithium ion is greater than that of the fluoride of the mono-isotopically substituted dimer. ion. It is the lithium ion in the B2, vibration moving Comparison of Fig. 1 and 2 shows that the absorption perpendicularly to the walls of the matrix cavity which bands e and f decrease noticeably in intensity after will suffer substantial restriction of movement, assuming warm-up, while those at c, d, and g remain about the the dimer is a "tight fit" in the cavity. The motion of same. Also the intensity of the absorption bands a t the lithium ion in the B Bvibration ~ is unlikely to be c, d, and g relative to those a t e and f could be init is directed tangentially to the walls of restricted as creased by working at low M / H values. Further examithe cavity. On this qualitative basis, the matrix nation of the absorption bands e and f in Fig. 1 shows absorption bands a t 620 and 537 cm.-l are tentatively that the stronger absorption w1 is always accompanied assigned to the Bau and B2, modes, respectively. a t higher frequencies by a weaker hand, w 2 . The ratios The assignment of the lower frequency to the Bzo mode w2/w1 for the absorption bands e and f are shown in is in agreement with the theoretical calculation of Table 111. These values may be compared with the Berkowit z .7 approximate isotopic frequency shift calculated for Previously, it was noted that after annealing the the dimers Li7,F2and Li7Li6Fzof 1.03. matrix the absorption bands a t e and f in Fig. 1 and 2 TABLE I11 decreased in intensity. However, in argon the shapes --8bsorption band-of bands e and f also change and the absorption e f Matrix maxima are shifted to higher frequencies by 7 and 1 032 1 024 Argon 42 cm.-l, respectively. A further anomaly present Krypton 1.032 1 023 in the argon matrix spectrum is the splitting of the 1 032 1 023 Xenon absorption band a t e into two peaks, a t 723 and 727 em. It is difficult to interpret this behavior, but, since On the basis of the above results, the absorption it only occurs in argon, it is perhaps connected with the bands e and f in Fig. 1 are assigned to the stretching trapped species occupying different types of sites in the frequencies of lithium fluoride dimer. No frequency argon matrix whereas in krypton and xenon matrices unequivocably attributable to the dimer bending mode only one type of site is involved. It is possible the was observed. An absorption band consisting of a absorption bond a t f is split in the same way as e major and minor peak, not shown in Fig. 1, occurred in noted above, but that the splitting is not observed due all three matrices a t approximately 274 and 283 cm. -l, to the low resolving power of the cesium bromide optics. respectively, in argoin and at slightly lower frequencies in Certainly the splitting of the absorption band e, which krypton and xenon, but always with a separation of was observed using the Dual Grating interchange, was 9 cm.-l between the major and minor peaks. about not resolved when using the cesium bromide interHowever it was not possible to show if this absorption change. band was due specifically to the dimer or higher polymeric species since the spectrometer was not very Of the remaining absorption bands in Fig. 1 (c, d, reproducible in t h k region. Theoretical calculation of and g) only g appears as a single peak in all three the vibration frequencies of lithium fluoride dimer by (7) J. Berkowite. zW., 29, 1380 (1958): 32, 1519 (1960). (6) 8, B, Bauer, T.Ino, and R, F, Porter, J , Chem. Phys,, 33, 685 (1960).
(8) W. Klernpeier and
IT.G, Norris, +bid,,34, 1071 (1961).
has SNELSON AND KESNETHS. PITZER
abundance of the halogen dimer isotopes is not given in Table V, since shifts in frequency due to the substitution of C135for C137in the dimer are too small to be observed.
LI C I in Krypton
TABLE IV MATRIXISOLATIOX SPECTRA OF LITHIUMCHLORIDE
0.81M / H ' 3 1 7 0
Moles LI CI:BxlO-'(
Frequencies, om. -IKrypton
300 Fig. 4.-LiCl
c in.-', spectra of matrices as deposited.
LLICIIn Argon '
362 352 298