Infrared spectroscopy and molecular properties of chlorosyl fluoride, FClO

10 December 1999

Chemical Physics Letters 314 Ž1999. 396–402 www.elsevier.nlrlocatercplett

Infrared spectroscopy and molecular properties of chlorosyl fluoride, FClO Holger S.P. Muller ¨

)

I. Physikalisches Institut, UniÕersitat Strasse 77, 50937 Koln, ¨ zu Koln, ¨ Zulpicher ¨ ¨ Germany Received 30 September 1999; in final form 11 October 1999

Abstract The structure of the very reactive, yet fairly long-lived, FClO molecule has been determined experimentally for the first time. Analysis of the high-resolution infrared spectrum of its n 1 band provided the spectroscopic constants needed. FClO was obtained as the main product of the reaction between ClF3 and H 2 O under slow-flow conditions. A preliminary analysis of the n 2 band and a comparison of the FClO structural properties with those of related molecules are also presented. q 1999 Elsevier Science B.V. All rights reserved.

1. Introduction Chlorosyl fluoride, FClO, is a molecule of fundamental interest for basic and theoretical chemistry. It consists of only three atoms all of which are highly electronegative. Furthermore, the molecule contains hypervalent Cl. These aspects make the precise calculation of physical and chemical properties by quantum-mechanical means a challenging task. Consequently, the experimental characterization of these types of compounds will provide important reference points to test quantum-chemical models. In addition, FClO has been suggested to be of atmospheric interest. Most members of the series FClO N with 0 ( N ( 3 have been characterized fairly well, as can be seen,

) Tel.: q49-221-470-3554; fax: q49-221-470-5162; e-mail: [email protected]

for example, in the review article by Christe and Schack w1x. However, FClO had remained elusive for a long time. It was proposed as an unstable intermediate in the hydrolysis of ClF3 by Ruff and Krug in 1930 w2x, but the identity of the product obtained remained uncertain w1x. Studies of reactions involving ClF, ClF3 , or F3 ClO provide indirect evidence of the formation of FClO as a very reactive intermediate that is unstable with respect to disproportionation into FClO 2 and FCl w1,3x. Attempts to find direct evidence for the formation of FClO by complexing it with a strong Lewis acid, such as AsF5 to yield ClOAsF6 , were unsucessful w3x; the controlled hydrolysis of ClF2 AsF6 in HF solution resulted in ClO 2 AsF6 as the only non-volatile chlorine-containing product. Only two experimental studies have been reported on FClO previously. Cooper et al. investigated the hydrolysis of ClF3 in a flow reactor w4x. Using an excess of ClF3 , they observed a band centered at

0009-2614r99r$ - see front matter q 1999 Elsevier Science B.V. All rights reserved. PII: S 0 0 0 9 - 2 6 1 4 Ž 9 9 . 0 1 1 9 7 - 5

H.S.P. Mullerr Chemical Physics Letters 314 (1999) 396–402 ¨

about 1032 cmy1 with PQR band contour in addition to bands of FClO 2 and OClO. This band vanished with a half-life of about 25 s to yield FClO 2 and FCl, and it was provisionally assigned to FClO. When they used an excess of H 2 O, OClO was the main product, and no evidence for the formation of FClO was found. Andrews et al. co-condensed FCl and O 3 in an Ar matrix and photolyzed the mixture w5x. They observed all three expected fundamentals of FClO including some 16 Or18 O and 35 Clr37Cl isotopic shifts. From a force-field calculation, they concluded that the FCl bond in FClO is weaker than in FCl, and the ClO bond is stronger than in ClClO and even more so than in ClO. Ab initio calculations have been performed on FClO and related species. High-level coupled-cluster calculations with moderately large basis sets were reported by Lee w6x and Francisco w7x. In the present study, the hydrolysis of ClF3 was investigated by infrared ŽIR. spectroscopy under slow-flow conditions. The very reactive FClO molecule has been identified as the most prominent, fairly long-lived, product under favorable conditions. A brief analysis of the high-resolution spectrum of FClO in the region of n 1 , the ClO stretching band, is presented. It provides the spectroscopic constants needed to determine the structural parameters of FClO experimentally for the first time. A preliminary analysis of n 2 , the FCl stretching band, will also be given.

2. Experimental details The hydrolysis of ClF3 was initially studied at the Universitat ¨ Hannover using a ; 20 cm long stainless steel absorption cell equipped with AgCl windows. ClF3 and H 2 O were mixed at the entrance to the cell, and the mixture was flowed slowly through the cell; estimated flow rates are of the order of 25–50 mm sy1 . Low-resolution IR spectra Ž1 cmy1 . were taken on a Bruker IFS 66 v. Later, high-resolution spectra were recorded on a Bruker IFS 120 HR spectrometer at the Bergische Universitat ¨ Wuppertal using a 1.5 m long stainless steel absorption cell equipped with KCl windows for n 1 and n 2 and with polyethylene windows for the bending mode, n 3 . For the n 1 band,

397

two spectra were used for the analysis: one with 0.003 cmy1 resolution Ž120 scans coadded. for most of the lines, and one with 0.005 cmy1 resolution Ž64 scans coadded. for weaker lines.

3. Results and discussion Both in Hannover and in Wuppertal, highest FClO yields were obtained when about equimolar amounts of ClF3 and H 2 O Žup to 100 Pa each. were mixed at the entrance of the absorption cell and when the products were pumped through the cell slowly. These conditions are somewhat different to those of the previous study w4x in which an excess of ClF3 was found to give best results. Fig. 1 shows in a low-resolution spectrum that ClF3 bands were still quite prominent in the present study, whereas FClO 2 amounts were low and OClO was not observed under these circumstances. Experiments to estimate the lifetime of FClO have been undertaken in Wuppertal. When the flow of ClF3 and H 2 O was stopped and the total pressure inside the cell was about 100 Pa, 50% of FClO decomposed within approximately 3 min. This indicates a substantially longer half-life under suitable conditions than the 25 s reported previously w4x. FClO 2 was found as a decomposition product. FCl, another product, was not observed, partly because of the weakness of its lines and partly because it was hydrolyzed to HOCl; strong lines of the pure rotational spectrum of HOCl were seen around 300 cmy1 , the region of the weak bending mode, even under slow flow conditions. Absorptions that could be assigned unambiguously to HOClO or HOClO 2 , possible hydrolysis products of FClO and FClO 2 , respectively, were not observed, suggesting that the band intensities of these molecules are small or that the molecules have very short lifetimes under the conditions of the experiments. An excess of H 2 O caused the FClO bands to decrease in intensity, most likely due to hydrolysis. A slight excess of ClF3 lowered the FClO yields and increased the amount of FClO 2 , suggesting that ClF3 enhances the disproportionation of FClO. On the other hand, when ClF3 was flowed slowly through a glass cell at pressures of ; 1 Pa, the water adsorbed on the cell walls was sufficient to hydrolyze a significant part of the ClF3

398

H.S.P. Mullerr Chemical Physics Letters 314 (1999) 396–402 ¨

Fig. 1. Low-resolution IR spectrum of the hydrolysis products of ClF3 . P-, Q-, and R-branches of n 1 and n 2 of FClO are indicated. Also given are the assignments of bands of unreacted ClF3 and of FClO 2 , a decomposition product of FClO.

so that the pure rotational spectrum of FClO could be recorded in the microwave to sub-millimeter regions; these results and their interpretation will be presented elsewhere. The low-resolution IR spectrum of the hydrolysis products of ClF3 in Fig. 1 shows that n 2 of FClO is about twice as strong as n 1. Since FClO is a planar molecule, all three fundamentals are ab-type hybrid bands. The Cartesian displacements of the normal coordinates are shown in Fig. 2; they indicate that both n 1 and n 2 are more a-type in character, in agreement with the observed PQR band contour in Fig. 1. The n 3 band is expected to be predominantly a b-type band with a Q-branch not or barely visible in a low-resolution spectrum. The maxima of the n 1 and n 2 bands of F 35 ClO and F 37ClO Žnatural isotopic ratio, ; 3:1. are at about 1036.6, 1027.4, 595.9, and 589.9 cmy1 , respectively. Weaker features ; 3.2 and ; 3.9 cmy1 lower than the n 1 and n 2 Q-branch maxima, respectively, can be assigned tentatively to the Q-branches of the n 1 q n 3 y n 3 and n 2 q n 3 y n 3 hot bands. The n 1 band was recorded first, and its analysis is complete. The assignments proceeded in the following way: approximate band origins for the 35 Cl and

37

Cl isotopomers were obvious from both the lowand high-resolution spectra. Simulations of the a-type

Fig. 2. Cartesian displacements of the normal coordinates of FClO. The a- and b-inertial axes are also shown.

H.S.P. Mullerr Chemical Physics Letters 314 (1999) 396–402 ¨

399

Fig. 3. Detail of the n 1 Q-branch region of F 35 ClO, selected assignments of J quantum numbers for each K a sub-branch are given in the lower part. The short solid lines represent transitions used in the final fit; dashed and dotted lines have not been used due to unfavorable overlap of several lines; the latter ones indicate asymmetry splitting. Some assignments for a-type R-branch transitions of F 37ClO used in the final fit are given in the upper part.

spectrum using approximate rotational and quartic centrifugal distortion constants from an estimated structure and the harmonic force field, respectively, revealed the oblate paired R- and P-type transitions Ž D J s "1. with J s K c and J f 25 to be the strongest in the spectrum. These lines were quite readily identified. The change in spacing between successive transitions yielded an estimate of DC, the difference between the upper and lower state C rotational constants. With new simulations, an unambiguous J assignment was possible for these transitions. Subsequently, the J range of these transitions has been extended to higher and lower values, and some R- and P-type transitions with K a s 1 and K c s J q 1 were found. After assigning some Qbranch transitions Ž D J s 0. with low J and K a , the simulated spectrum was quite close to the observed one so that the assignment of further a-type transitions was straightforward. Part of the F35 ClO Qbranch region is shown in Fig. 3 with some assignments. Eventually, several b-type transitions were found; they had K a 0 9 and were at the low- and high-frequency sides of the spectrum almost exclusively. Simulations of the spectrum with different

relative magnitudes of the a- and b-transition dipole components suggested the latter to be about three times smaller than the former; therefore, the b-type transitions are weaker than the a-type transitions by a factor of ; 10. This finding is in accord with the Cartesean displacements of n 1 shown in Fig. 2. Even though the n 1 spectrum of FClO is not particularly dense, accidental overlap of lines is not uncommon. The transitions overlapping in one line were used in the fit as an intensity weighted average as long as the most intense lines were due to only one isotopomer and of the same symmetry Ž a or b .. Watson’s A reduced Hamiltonian was used for fit-

Table 1 Summary of data for fitting the n 1 band of FClO No. of lines No. of transitions Max. J Max. K a rmsP10 3 rcmy1 Weighted rms

F 35 ClO

F 37ClO

1134 2033 74 24 0.238 0.381

462 956 71 20 0.269 0.431

H.S.P. Mullerr Chemical Physics Letters 314 (1999) 396–402 ¨

400

Table 2 Ground-state spectroscopic constants C0

a

of FClO and changes DC1 for the Õ1 s 1 state Žcmy1 , MHz.

F35 ClO

F37ClO

Õs0

b c

Õs0

Õ1 s 1

b

n0 A B C D J P 10 3 D J K P 10 3 DK P 10 3 d J P 10 3 d K P 10 3 F J P 10 6 F J K P 10 6 F K J P 10 6 F K P 10 6 f J P 10 6 a

Õ1 s 1 1037.68938 y225.19 Ž3. y29.399 Ž4. y25.585 Ž2. 0.0937 Ž11. y2.10 Ž3. 7.12 Ž16. 0.0377 Ž7. y0.04 Ž4. c

36651.0 Ž7. 8345.29 Ž5. 6777.50 Ž4. 12.038 Ž17. y111.7 Ž4. 1197.8 Ž24. 3.394 Ž12. 34.7 Ž5. 0.028 Ž4. c y0.20 Ž13. c y15.6 Ž10. c 121.8 Ž26. c 0.013 Ž3. c

1028.55201b y217.14 Ž3. y28.962 Ž7. y25.213 Ž4. 0.0922 Ž18. y2.03 Ž4. 7.01 Ž15. 0.0374 Ž12. y0.04 Ž4. c

35689.6 Ž13. 8339.46 Ž8. 6739.74 Ž5. 11.944 Ž24. y106.8 Ž4. 1136.6 Ž29. 3.427 Ž15. 33.5 Ž10. 0.028 Ž4. c y0.20 Ž13. c y15.6 Ž10. c 112.9 Ž24. c 0.013 Ž3. c

y0.46 Ž12. c 1.66 Ž28. c

y0.46 Ž12. c 1.55 Ž26. c

Band origin in cmy1 ; rotational constants, quartic and some sextic centrifugal distortion constants. Uncertainty in the fit below estimated accuracy of ; 10y3 cmy1 . Common constants for F35 ClO and F37ClO, or fixed-ratio for F K .

ting of the spectra in the representation I r. Both isotopomers were fit simultaneously. The spectroscopic constants C determined are those of the ground vibrational state Ž C0 . and the differences between the ground and vibrationally excited states Ž DC1 .. High correlations between corresponding parameters are reduced by this fitting procedure compared to the more common determination of ground Ž C0 . and vibrationally excited Ž C1 . states parameters. A summary of the data is presented in Table 1; the spectroscopic constants derived from the analysis are given in Table 2.

The ground-state effective structure was derived from the rotational constants. A force-field calculation was performed using the vibrational wavenumbers and the quartic centrifugal distortion constants as input. Experimental and ab initio structural parameters and vibrational wavenumbers are presented in Table 3. A comparison of structural parameters and stretching force constants of FClO with those of related molecules is given in Table 4. The analysis of n 2 is currently under way. It is in weak Coriolis and Fermi resonance with 2 n 3 approximately 20 cmy1 higher. Preliminary, selected

Table 3 Experimental and ab initio structural parameters Žpm, deg. and vibrational frequencies Žcmy1 . of FClO Exptl. Gas phase r ŽClO. r ŽFCl. / v1 v2 v3 a b c d e

c

149.41 Ž35. 168.67 Ž32. 110.67 Ž6. 1037.7 596.9 ; 310 e

Matrix

1038 593.5 315.2

d

CCSDŽT. a

CCSDŽT.r6-311G . . .

TZ2P

Žd.

Ž2d.

Ž2df.

cc-pV5Zq d

151.8 174.6 111.3 994 540 281

153.4 180.6 113.5

152.1 175.7 112.9

150.1 170.8 111.5 1038 593 304

148.55 169.63 110.68 1061 618 318

Ref. w6x. Ref. w7x; number and type of polarization functions given in parentheses. This work; anharmonic values. Ref. w5x; anharmonic values. Derived from 2 n 3 .

b

B3LYP c

H.S.P. Mullerr Chemical Physics Letters 314 (1999) 396–402 ¨ Table 4 Bond lengths a Žpm. and force constants ŽN my1 . of FClO and related molecules FCl b r ŽFCl.

FClO c ClF3

d

168.7 170.1 160.1 r ŽClO. 149.4 /ŽFClO. 110.7 f ŽFCl. 250.8 295.0 445.2 408.0 f ŽClO. 701.0

FClO 2

e

FClO 3 f ClO g ClO 2

h

169.7

162.83

142.2 101.7 237.8 937.4

162.0 140.1 156.96 146.98 102.0 363.3 1013.7 469.4 703.5

a

r 0 parameters for FClO, r z parameters for ClF3 and FClO 2 , equilibrium parameters elsewhere. b Ref. w8x. c This work. d Ref. w9,10x. e Ref. w11,12x. f Ref. w13x. g Ref. w14,15x. h Ref. w16x.

data of this analysis are given in Table 5 to allow for a more detailed comparison with the Ar matrix and ab initio data. A more detailed analysis of n 2 and the analysis of n 3 will permit a better description of the n 2r2 n 3 resonance system and the determination of the FClO equilibrium structure–the molecular geometry in the potential minimum. The vibrational wavenumber of n 1 for the F35 ClO isotopomer from the previous gas phase study Ž1032 cmy1 . is slightly lower than the present value. Ar matrix values of stretching vibrations are usually lower than gas-phase values as has been found for n 2 of FClO; the values for n 1 are remarkably similar. The value of 2 n 3 suggests n 3 to be around 310 cmy1 , slightly lower than in Ar matrix, which is not uncommon for bending modes. The coupled-cluster ab initio structural parameters and vibrational wavenumbers obtained with smaller basis sets w6,7x presented in Table 3 show rather large deviations from experiment; the 6-311GŽ2d. and TZ2P basis sets are quite similar. Better agreement is obtained with the largest basis set employed w7x. B3LYP is a method from density functional theory ŽDFT. whose results quite often approach the quality of coupled-cluster calculations. The efficient use of computer time and memory permits the use of large basis sets. The results obtained in the present study using the program package GAUSSIAN98 w17x

401

are in good agreement with experiment showing that very large basis sets are needed for describing the FClO molecule well by quantum-chemical models and that B3LYP can yield good results for molecules containing hypervalent Cl. In general, structural parameters can be estimated from matrix isolation spectroscopy only by relating them to force constants or vibrational wavenumbers. Andrews et al. assumed 157 and 180 pm for the ClO and FCl bond, respectively, and 1208 for the bond angle, much larger than the present values w5x. Christe and Schack suggested the bond angle to be smaller than the tetrahedral angle of 109.5 w1x, probably based on simple VSEPR considerations. The present value is about 18 larger. However, it is not unusual for an O atom with double-bond character to be more repulsive than a free electron pair. For example, the OClO angles in FClO 2 and FClO 3 are 115.08 and 115.88, respectively w12,14x, while the FClO angles are 101.78 and 102.08, respectively, much smaller than the tetrahedral angle ŽTable 4.. Christe and Schack have argued that there are two limiting types of FCl bonds: one type is covalent, short and strong, and the second type of FCl bonds is much longer and weaker and belongs to the group of weak Žp–p ) . s bonds proposed by Spratley and Pimentel w19x. FCl, FClO 3 , and the equatorial bond in ClF3 are examples of more covalent bonds. As is shown in Table 4, they are about 160 to 163 pm long, and the force constants are about 400 N my1 . The bonds in FClO w1,5x, FClO 2 w1,18x, and the axial bonds in ClF3 w1x are, with ; 170 pm, much longer and, with force constants below 300 N my1 , much weaker. They are examples closer to the second limiting type of FCl bonds. It is interesting that the Table 5 Band origins Ž n 0 ., Fermi Ž F0 . and Coriolis Ž Gc . interaction constants Žcmy1 ., and changes in the rotational constants a ŽMHz. for the n 2 r2 n 3 band system of F35 ClO Õ 2 s1

n0 F0 Gc DA DB DC a

Õ3 s 2

596.898 Ž2.

617.404 Ž9. 0.849 Ž22. 0.01379 Ž6. y133.26 Ž13. 1368.6 Ž58. y57.815 Ž3. y38.98 Ž24. y47.113 Ž3. y54.78 Ž28.

With respect to the ground vibrational state; preliminary values.

402

H.S.P. Mullerr Chemical Physics Letters 314 (1999) 396–402 ¨

latter group of bonds shows very little variation in their length. However, for all of the FCl bonds there is little correlation between the force constant and the bond length, as shown in Table 4. In contrast to the initial Spratley–Pimentel bonding picture in which there is little change of the ClO bond between ClO and FClO, the F atom withdraws electron density from antibonding orbitals at the Cl, as does the O atoms bonded to Cl, thus, the ClO bond gets shorter from ClO to FClO, as it does from OClO to FClO 2 , as well as from FClO to FClO 2 to FClO 3 , and from ClO to OClO. For these molecules, there is a strong correlation between shorter ClO bonds and larger force constants.

Acknowledgements I would like to thank Prof. Helge Willner and Prof. Hans Burger for the opportunity to record ¨ spectra at the universities of Hannover and Wuppertal, respectively. I am also grateful to Friedbert Lucker for the help in recording the high-resolution ¨ spectra in Wuppertal and to Dr. Edward A. Cohen for some helpful comments. The work in Koln ¨ was supported in part by the Deutsche Forschungsgemeinschaft ŽDFG. via SFB 301 and by special funding from the Science Ministry of the Land Nordrhein–Westfalen.

References w1x K.O. Christe, C.J. Schack, Adv. Inorg. Chem. Radiochem. 18 Ž1976. 319, and references therein. w2x O. Ruff, H. King, Z. Anorg. Allg. Chem. 190 Ž1930. 270. w3x K.O. Christe, Inorg. Chem. 11 Ž1972. 1220. w4x T.D. Cooper, F.N. Dost, C.H. Wang, J. Inorg. Nucl. Chem. 34 Ž1972. 3564. w5x L. Andrews, F.K. Chi, A. Arkell, J. Am. Chem. Soc. 96 Ž1974. 1997. w6x T.J. Lee, J. Phys. Chem. 98 Ž1994. 3697. w7x J.S. Francisco, J. Chem. Phys. 108 Ž1998. 659. w8x R.E. Willis Jr., W.W. Clark III, J. Chem. Phys. 72 Ž1980. 4946. w9x S.T. Haubrich, M.A. Roehrig, S.G. Kukolich, J. Chem. Phys. 93 Ž1990. 121. w10x H.S.P. Muller, unpublished results. ¨ w11x A.G. Robiette, C.R. Parent, M.C.L. Gerry, J. Mol. Spectrosc. 86 Ž1981. 455. w12x H.S.P. Muller, J. Mol. Struct., 1999. ¨ w13x H.S.P. Muller, M.C.L. Gerry, J. Mol. Spectrosc. 175 Ž1996. ¨ 120. w14x R.K. Kakar, E.A. Cohen, M. Geller, J. Mol. Spectrosc. 70 Ž1978. 243. w15x J.B. Burkholder, P.D. Hammer, C.J. Howard, A.G. Maki, G. Thompson, C. Chackerian Jr., J. Mol. Spectrosc. 124 Ž1987. 139. w16x H.S.P. Muller, G.O. Sørensen, M. Birk, R.R. Friedl, J. Mol. ¨ Spectrosc. 186 Ž1997. 177. w17x GAUSSIAN 98, Revision A.5, M.J. Frisch, et al., Gaussian, Pittsburgh, PA, 1998. w18x C.R. Parent, M.C.L. Gerry, J. Mol. Spectrosc. 49 Ž1974. 343. w19x R.D. Spratley, G.C. Pimentel, J. Am. Chem. Soc. 88 Ž1966. 2394.