Microwave Spectra and Molecular Geometries of Benzonitrile and Pentafluorobenzonitrile Mahdi Kamaee and Jennifer van Wijngaarden Department of Chemistry, University of Manitoba Work of my MSc student Mahdi Started as interest in using these as potential precursors for discharge expts

Fluorine substituted benzonitriles perfluoro monofluoro difluoro trifluoro We have previously reported some work on F substituted pyridines and the structural changes that occur upon fluorination at various sites. We decided to extend these studies to the benzonitrile rings and started by running ab initio calculations to identify possible trends that would guide our experiments. The talk today will focus on the parent benzonitrile (cyanobenzene) and the per-fluoro version although we’ve run calculations for all possible substitutions of one through four fluorine atoms. Some examples are shown here. 2

Geometry trends in the ring backbone MP2/6-311G++(2d,2p) At site of fluorination: C-C bonds shorten 0.005 – 0.008 Å Ring angle increases by >2o All C-C bonds shorten 0.003 – 0.009 Å Ring angle at C1 decreases by ~2o Based on our MP2 calculations, we’ve found a few potential trends to investigate. I’ll outline just a couple here. In the singly fluorinated species, we see a similar effect as in our earlier pyridine work. At the site of fluorination, the ring angle opens by 2-3 degrees, and the adjacent bonds shorten relative to the structure of benzonitrile. In the doubly fluorinated species, there are different effects depending on whether the substitution happens on neighbouring carbons or not. For substitution on neighbouring carbons, only the bond between them shortens but there are no noticeable angle changes at those sites… so the overall ring geometry is not largely distorted. For the non-neighbouring case, the geometry changes are analogous to what’s seen in the single substitution cases. In the perfluoro species, the entire ring backbone seems to contract relative to benzonitrile and the angle at the site of the cyano substituent decreases by about 2 degrees. Neighbour vs non-neighbour substituents neighbour: shared bond shortens ~0.010 Å No large angle changes non-neighbour: similar to single F case

Geometry trends near CN substituent MP2/6-311G++(2d,2p) There are potentially some interesting trends surrounding the cyano group as well. In terms of the C-C bond length from the ring to the cyano group, this bond contracts when the F is in an ortho position and this effect is doubled if you have 2 ortho substituents. We also expect a decreased ring angle in these cases. An ortho fluorine also distorts the angle at the sp hybridized C of the cyano group as shown here. Now, the question is whether microwave spectroscopy can provide us with accurate enough structures to see these trends as I’ve outlined.

Benzonitrile: Previous work Extensive MW studies (3-160 GHz) G. Erlandsson, J. Chem. Phys. 22, 1152 (1954). 14N hfs Dreizler and coworkers, Z. Naturforsch. 26a, 1124 (1981); 43a, 283 (1988). Dipole moment K. Wolfhart et al., J. Mol. Spectrosc. 247, 119 (2008). 13C isotopologues Bak et al., J. Chem. Phys., 37, 2027 (1962). Dreizler and coworkers, Ber. Bunsenges. Phys. Chem. 98, 970 (1994). The parent molecule, benzonitrile, has been extensively studied with about a dozen different microwave papers reported over the years which provide a wealth of information about the spectroscopic constants, dipole moment, 14N hyperfine structure etc. For the 13C isotopologues, which we need for our structural studies, there have been previous reports but we wanted a more complete data set including measurement of the hyperfine structure. The molecule has quite a large electric dipole moment along the a-axis and thus, only a-type transitions are observed. Jens paper: Large dipole makes this a good candidate for deceleration/trapping experiments µa=4.5152(68) D

Perfluorobenzonitrile: Previous work MW study (18-26 GHz) Sharma and Doraiswamy, Proc. Ind. Acad. Sci. 67a, 12 (1968). 14N hfs (7-13 GHz) Krüger and Dreizler, Z. Naturforsch. 47a, 865 (1992). No isotopic studies For the perfluoro analog, there are only a couple of previous studies and the 14N hyperfine structure was only reported for a few transitions. Furthermore, there are no reported 13C isotopologues. As in the parent species, PFBN has a considerable dipole component along the a-axis and because of the 5 fluorine atoms, the rotational constants are much smaller making it feasible to observe many transitions in the range of our instruments. µa=2.79 D MP2/6-311G++(2d,2p)

Balle-Flygare FTMW instrument 4-26 GHz FWHM ~7kHz Useful for: 14N hfs resolved 13C in natural abundance Sample prep: 1-2 atm of Ne/Ar bubbled through liquid sample Supersonic jet The BN experiments were done using BF-FTMW because the spectroscopic constants were previously well-determined and we were mostly focused on measuring hfs of the weak isotopologues of C. Fairly high boiling liquids so we used a bubbler to seed samples into the carrier gas. BP of BN 191oC, PFBN 161oC

Sample FTMW spectrum of PFBN 14N hyperfine structure

Chirped pulse FTMW instrument 8-18 GHz 6 GHz bandwidth FWHM ~100kHz Useful for: Identifying isotopologue transitions Sample prep: 1 atm of Ne/Ar bubbled through liquid sample Supersonic jet We used the cp-FTMW instrument to first identify the positions of the 13C lines of PFBN

Sample cp-FTMW spectrum of PFBN A small section of the spectrum showing these lines alongside the parent lines.

BN Spectroscopic constants Pickett’s SPFIT program, Watson A-reduction, Ir representation CD constants fixed to those of parent species   Parent BN C1 C2 C3 C4 C5 Rotational Constants /MHz A 5655.264(5) 5655.496(4) 5563.914(4) 5565.666(4) 5655.453(5) 5655.237(3) B 1546.8758(10) 1545.55191(6) 1546.80335(5) 1535.71300(5) 1523.65521(6) 1528.64085(6) C 1214.40407(9) 1213.60148(6) 1210.08975(4) 1203.37301(4) 1200.05798(6) 1203.13679(4) Centrifugal Distortion Constants /kHz ΔJ 0.0450(4) 0.0450 ΔJK 0.9373(19) 0.9373 ΔK -0.5(18) -0.5 δJ 0.0112(3) 0.01122 δK 0.59(4) 0.59 0.54 14N Nuclear Quadrupole Coupling Constants /MHz 1.5χaa -6.3560(8) -6.344(9) -6.367(8) -6.366(8) -6.348(15) -6.359(16) 0.25(χbb-χcc) 0.0848(3) 0.073(6) 0.070(5) 0.068(5) 0.074(5) 0.076(2) rms /kHz 0.95 1.2 1.1 1.0 no. lines 177 51 63 42 39 Our final data set includes 177 lines of the parent and 40-60 lines for each isotopologue. It was possible to measure more lines for the C2 and C3 isotopologues as there are two equivalent carbons in these positions. The overall fit looks quite good with the A rotational constant having the greatest uncertainty.

BN Geometry C1C2 1.381(6) 1.390(4) 1.392(3) 1.401 C2C3 1.415(12) rs, ro and reSE geometries calculated using KRA and STRFIT programs (http://info.ifpan.edu.pl/~kisiel/prospe.htm) ro geometry using B and C constants only reSE: rovibrational corrections were calculated at B3LYP/6-31G(d,p) re were calculated at MP2/6-311G++(2d,2p) rs ro reSE re C1C2 1.381(6) 1.390(4) 1.392(3) 1.401 C2C3 1.415(12) 1.381(5) 1.376(4) 1.393 C3C4 1.396(1) 1.404(2) 1.395(2) 1.396 C1C5 1.450(3) 1.455(6) 1.443(5) 1.435 (C1C2C3) 118.1(5) 119.4(4) 119.5(3) 119.5 (C2C3C4) 120.2(1) 119.8(6) 120.0(4) 120.1 (C3C4C3) 120.1(1) 120.2(6) 120.1(2) 120.2 (C2C1C2) 123.3(6) 121.5(6) 120.8(5) 120.5 We then used our rotational constants to estimate the geometry of ring. The results of the methods we’ve tried are summarized here along with our MP2 predictions of the geometry in the final column. There are fairly large uncertainties in the rs parameters which come from having some very small KRA coordinates. We seem to see the correct trends for the most part (in terms of which bonds are longer than others) but our uncertainties are really on the order of the geometry changes we’re expecting in most cases… so we have to work on this a bit more… Perhaps take out DEL K…. Ask Kisiel about weighting of reSE parameters… Ask Kisiel about methods for small coordinates…

PFBN Spectroscopic constants Pickett’s SPFIT program, Watson A-reduction, Ir representation CD constants fixed to those of parent species   Parent PFBN C1 C2 C3 C4 C5 Rotational Constants /MHz A 1029.36864(3) 1029.4002(3) 1026.38600(17) 1026.3606(2) 1029.4046(3) 1029.3699(4) B 764.595288(9) 762.9252(2) 764.32975(6) 763.68959(11) 761.7224(2) 756.6354(2) C 438.721848(6) 438.17764(10) 438.092216(9) 437.877115(9) 437.781210(10) 436.089680(11) Centrifugal Distortion Constants /kHz ΔJ 0.006169(18) 0.006169 ΔJK 0.04498(7) 0.04498 ΔK -0.0284(3) -0.0284 δJ 0.00237(10) 0.00237 δK 0.02955(7) 0.02955 14N Nuclear Quadrupole Coupling Constants /MHz 1.5χaa -6.5807(12) -6.58(4) -6.574(19) -6.58(2) -6.59(4) -6.63(4) 0.25(χbb-χcc) 0.1080(7) 0.105(3) 0.111(3) 0.108(3) rms /kHz 0.787 1.203 1.179 1.167 1.201 1.355 no. lines 753 81 99 96 78 For PFBN, we were able to measure many many lines…. 753 for the parent alone! And thus, our spectroscopic constants are better determined.

PFBN Geometry C1C2 1.400(2) 1.394(2) 1.389(2) 1.398 C2C3 1.360(4) ro geometry using A, B and C constants reSE: rovibrational corrections were calculated at B3LYP/6-31G(d,p) re were calculated at MP2/6-311G++(2d,2p) rs ro reSE re C1C2 1.400(2) 1.394(2) 1.389(2) 1.398 C2C3 1.360(4) 1.373(2) 1.373(3) 1.388 C3C4 1.391(1) 1.390(1) 1.383(2) 1.392 C1C5 1.436(1) 1.438(2) 1.430(3) 1.426 (C1C2C3) 121.6(2) 121.1(2) 120.9 (C2C3C4) 119.8(1) 119.7(2) 119.6(3) 119.6 (C3C4C3) 120.1(1) 120.3(3) 120.4 (C2C1C2) 117.5(2) 118.4(2) 118.3(2) 118.6 Again, we use our constants to derive geometrical parameters and come up with the following. In this case, the rs structure has slightly lower errors because there aren’t as many small KRA coordinates compared to that of BN… and again, we see the correct trends…

Geometry comparison of BN and PFBN Ab initio calculations predict: All C-C bonds shorten 0.003 – 0.009 Å Ring angle at C1 decreases by ~2o reSE  re (PFBN-BN) C1C2 1.392(3) 1.389(2) -0.003 C2C3 1.376(4) 1.373(3) -0.005 C3C4 1.395(2) 1.383(2) -0.004 C1C5 1.443(5) 1.430(3) -0.009 (C1C2C3) 119.5(3) 121.1(2) +1.4 (C2C3C4) 120.0(4) 119.6(3) -0.4 (C3C4C3) 120.1(2) 120.3(3) +0.2 (C2C1C2) 120.8(5) 118.3(2) -1.9 Now, to compare what we have so far on BN and PFBN using the semi-expt structures. The final column here shows the changes that are predicted by our ab initio (MP2) calculations. Again, we’re cautious about drawing conclusions because of the uncertainties in the parameters… but we seem to see the right tends. For the C1-C5 bond, we expect a significant change and we do see this in our data and it appears to be meaningful. This bond from the ring to the sp hybid C of the cyano group is shortened considerably with an ortho F substituent. We also see a contraction of the ring at C1 of ~2 degrees which also appears meaningful.

Geometry comparison with benzene Here, we show the ring backbone of each molecule superimposed on the ring structure of benzene (dotted lines) which shows the effect of the CN and F substitution on the ring. The coloured bars show the expected changes at the various angles relative to the CN group with alpha being the closest and delta the furthest. If we just focus on the semi-expt and ab intio structures (green, blue), we see that the larges changes are again around the CN group and the ortho position… and that interestingly, the entire ring in PFBN is contracted relative to that of benzene! This work is ongoing. We’d like to do a TD analysis on the 14N hfs…. But in the meantime

Monofluorinated BN μb μb μ μa μ μa µa=3.29 D µb=2.38 D µa=5.44 D (calc.) µa=5.44 D µb=0.64 D (calc.) We’ve completed measurements on the mono-fluoro species… and the spectra in these cases include both a- and b-type transitions… and the structural analysis is currently in progress while Mahdi starts expts on some di-fluoro species. Stay tuned…

Acknowledgements Dr. Ming Sun Mahdi Kamaee Questions??? vanwijng@cc.umanitoba.ca Ming is helping with TD analysis I’m looking for a PDF in MW or IR spectroscopy. Contact me for info.