oppler, infrared laser spectroscopy of the propyne 2~~ band: Evidence of zmaxis Coriolis dominated intramolecular state mixing in the acetylenic WI stretch overtone Andrew M&of) and David J. Nesbitt Joint Institute for Laboratory Astrophysics, University of Colorado and National Institute of Standards and Technology and Department of Chemistry and Biochemistry University of Colorado, Boulder, Colorado 80309-0440 Erik R. Th. Kerstel,b) Brooks H. Pate,@ Kevin K. Lehmann, and Giacinto Stoles Department of Chemistry, Princeton University, Princeton, New Jersey 08544 (Received 3 September1993; accepted 5 November 1993) The eigenstate-resolved2q (acetylenic CH stretch) absorption spectrum of propyne has been observedfor J’ =0-l 1 and K==O-3 in a skimmed supersonicmolecular beam using optothermal detection. Radiation near 1.5 ,um was generatedby a color center laser allowing spectra to be obtained with a full-width at half-maximum resolution of 6 x low4 cm-’ ( 18 MHz). Three distinct characteristics are observed for the perturbations suffered by the optically active (bright) acetylenic CH stretch vibrational state due to vibrational coupling to the nonoptically active (dark) vibrational bath states. ( 1) The K=O states are observedto be unperturbed. (2) Approximately f of the observedK= l-3 transitions’are split into 0.02425 cm-’ wide multiplets of two to five lines. These splittings are due to intramolecular coupling of 21?,to the near resonant bath states with an averagematrix element of .( V2>“2=0.002 cm-’ that appears to grow approximately linearly with K. (3) The K subband origins are observedto be displaced from the positions predicted for a parallel band, symmetric top spectrum. The first two features suggestthat the coupling of the bright state to the bath states is dominated by parallel (z-axis) Coriolis coupling. The third suggestsa nonresonant coupling (Coriolis or anharmonic) to a perturber, not directly observedin the spectrum, that itself tunes rapidly with E, the latter being the signature of diagonal z-axis Coriolis interactions affecting the perturber. A natural interpretation of thesefacts is that the coupling betweenthe bright state and the dark statesis mediated by a doorway state that is anharmonically coupled to the bright state and z-axis Coriolis coupled to the dark states. Z-axis Coriolis coupling of the doorway state to the bright state can be ruled out since the y1 normal mode cannot couple to any of the other normal modes by a parallel Coriolis interaction. Basedon the range of measuredmatrix elementsand the distribution of the number of perturbations observedwe find that the bath levels that couple to 2vt do not exhibit Gaussian orthogonal ensembletype statistics but instead show statistics consistent with a Poisson spectrum, suggestingregular, not chaotic, classical dynamics. I. INTRODUCTION Vibrational energy flow in isolated polyatomic molecules has been a topic of ongoing interest for over two decadesdue to its importance in many fundamental chemical and physical processes.’From the first observationsof the benzene overtone spectrum by Reddy, Heller, and Berry, and their interpretation of the spectral widths as reflecting the time scale for intramolecular vibrational relaxation (IVR),2 there has been intense interest in elucidating the relaxation time scales and the mechanismsby which IVR occurs. With moderateexperimental resolution (0.1 cm-’ or worse), many investigations of IVR have concentrated on large systems at relatively high levels of excitation. For these systems it may occur that the IVR ‘)Present address: The Aerospace Corporation, Mail Stop MY754 P.O. Box 92957, Los Angeles, California 9ooO9-2957. “Present address: Laboratorie Europe0 di Spettroscopie Non-lineari (LENS), Largo E. Fermi no. 2 (ARCETRI), 50125 Fiienze, Italy. ‘)Present address: Department of Chemistry, University of Virginia, Charlottesville, Virginia 22901. induced homogeneousbroadening exceedsthe thermally induced, inhomogeneousbroadening. However, in these lower resolution studies it is often difficult to differentiate between a single, very broad vibrational band and a series of closely spacedbands in mutual resonance.3-7Although both results would indicate that vibrational energy redistribution is occurring, in the first casethe decay is irreversible while in the second multiple recurrences are present and the excitation remains localized in a small fraction of the total phasespace.For example,Quack and co-workers’ have found that in many CX,H molecules vibrational energy is exchangedbetween the CH stretch and bend on a time scale of 100 fs or less, while decay into the other modes of the molecules takes much longer. Despite the ambiguities, much has been learned about subpicosecond IVR from such modest resolution experiments. But in order to investigateIVR on time scalesof 1 ps or longer, time scalesmore relevant to most unimolecular reactions, methods of higher resolution are needed,possibly supplemented 2596 J. Chem. Phys. 100 (4), 15 February 1994 0021-9606/94/100(4)/2596/16/86.00 @ 1994 American Institute of Physics Downloaded 18 Mar 2002 to 128.112.83.42. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp Mcllroy et a/.: The propyne 2v, band by double-resonancetechniques, to reduce the inhomogeneous effects that complicate and often prevent a dynamical analysis of the spectra. As one considers longer time scales, new physical effects can becomeimportant. On ultrafast time scales(a few fs), only low order anharmonic interactions can be strong enough to cause such rapid population transfer. When relaxation is much slower, much more numerous, but weaker high-order anharmonic resonancesmay come into play. Further, for relaxation on the time scale of the rotational motion or longer, the interaction of vibration and rotation can modify the dynamics and also introduce new coupling mechanisms. In the regime of slower IVR ( 1 ps or longer), many central issueshave yet to be clarified by either experiment or theory. It is common to label interactions as “anharmanic” if their strength is independent of J, “Coriolis” if they scale linearly with one component (projection) of the angular momentum, and “centrifugal” if they scale quadratically with angular momentum components.gDespite the fact that we tend to think of these interactions as physically distinct, such a separation is not basis set independent and thus is ambiguous. In a harmonic oscillator basis, first-order Coriolis interactions are restricted to states differing by only two quantum numbers. But in anharmonic molecules, Coriolis interactions may lead to a small mixing of an off-resonant harmonic oscillator state, which may itself be coupled by anharmonic interactions with resonant states. In this case,one will have an effective direct “Coriolis” interaction between the initial state and the nearby levels which reflect both Coriolis and anharmonic interactions. In such cases we refer to the unseen intermediate state as a “doorway” state. Another example is that one can always choose as a basis set the eigenfunctions of the J=O effective vibrational Hamiltonian. In this case, one will still tind strong J&-dependent interactions between the states. While the anharmonic interactions have been diagonalized. Differences in vibration-rotation interactions, which are nearly diagonal in a normal mode basis set, now appear as “centrifugal” interactions between the vibrational eigenstates. In order to sort out these and other interesting IVR effects, one needs a fully eigenstateresolved and rotationally assigned spectrum. Experimentally, it is most convenient to perform fully resolved experiments at low energies where the transitions are stronger, specifically in the region of the vibrational fundamentals. Although it has recently been shown that, even in the region of the fundamentals, vibrational excitation can result in bond specific chemical reactions,1o7”there is much more practical interest in the IVR behavior of more highly vibrationally excited states, in the region from 1 to 5 eV above the ground state, which matches the activation energiesof most chemical reactions. It is therefore important to understand how the detailed information which can be obtained at low levels of excitation extrapolates to these higher levels. This information can be acquired most directly by studying a sequenceof vibrational states in as much detail as possible. This has been done in a number of three to five atom molecules,*2-18 2597 but few, if any, eigenstate-resolvedstudies have been possible at higher energiesfor larger systems.This is precisely what we have done in the present work where we have extended the previous eigenstate resolved u= 1 acetylenic CH stretch studies to v=2. It is worth noting that most spectroscopic investigations of IVR observe these “relaxation” effects not by the observation of the decay of an initially prepared vibrational state in the time domain, but by the detection of multiple line splittings in the frequency domain. The “fractionation” of a single, expected transition occurs due to the mixing of the zeroth-order state, the so-called bright state, to which oi>tical transitions are allowed, with the bath of vibrational states that are nearly resonant to the bright state. These bath, or dark, states are typically made up of combinations and overtones of the low frequency vibrations of the molecule and therefore are only very weakly optically connected to the ground vibrational state. The oscillator strength of the bright state, which can be considered to be the only state carrying a transition dipole moment, is then spread out over the eigenstatesformed by mixing the bright with the dark bath states. These eigenstates are what frequency domain experiments directly probe. The ideal frequency domain, eigenstate resolved spectrum contains all the information of the perhaps more familiar one color, pump-probe experiment performed in the time domain. The latter can be directly calculated from an autocorrelation of the spectrum combined with appropriate spectral and ensembleaveraging. But while an unresolved time domain experiment gives an ensembleaverage relaxation rate, an unresolved or unassignedfrequency domain spectrum gives a convolution of homogeneous structure (which reflects dynamics) and inhomogeneous structure which can only be predicted if extraneous assumptions are made. On the other hand, from the eigenstate resolved spectrum we can determine the relaxation for each rotational level of the excited state. From such information, lg we can determine the relative contributions of anharmonic and Coriolis effects to the root-meansquared (rms) coupling matrix element ( ( V’) ) “‘. In addition, the volume of phase space sampled during the relaxation process can be estimated and compared with the volume of available phasespacewhich can be derived from an estimate of the density of states.” However since the onset of IVR has been shown to occur typically at bath it has often been state densities of 10-100 states/cm-‘,21 difficult to experimentally resolve theseclosely spacedrovibrational eigenstates, especially in the presence of rotational congestion. Several groups have developed methods that have the resolution necessary to observe individual rovibrational eigenstatesof molecules either ( 1) at high energy in small molecules or (2) for the fundamental vibrations of larger molecules. For smaller polyatomics, double resonance methods such as stimulated emission pumping,22-2’ microwave-optical double resonance,26m or infrared (IR)visible double resonance27have allowed the rotational congestion problem to be overcome, thus permitting wide regions of the spectrum to be studied. For larger molecules, Downloaded 18 Mar 2002 to 128.112.83.42. Redistribution copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp J. Chem. Phys., Vol. subject 100, No.to4,AIP 15 license February or1994 2598 Mcllroy ef a/.: The propyne 2v, band the large room temperature partition functions make room-temperature, gas-phase double-resonance experiments less feasible.As a result, supersonicexpansioncooling (which alone reducesthe rotational partition function by a factor of - 103) have proved essential.Direct, highresolution IR absorption methods have beenapplied to the study of vibrational state mixing in larger polyatomics using a pinhole expansion by Perry and co-workers,28and using a slit expansion by McIlroy and Nesbitt.2g These direct absorption IR methods provide resolution down to 0.001 cm-’ at 3000 cm-‘, allowing state resolved IVR effects to be observed in CH stretch v= 1 excited molecules. Higher resolution (0.0003 cm-‘) IR studies of IVR have beenperformed using color center laser excitation in a skimmed molecular beam followed by bolometer detection-19*30 Furthermore, this method allows for an easier extensionto overtone excitation provided laser sources of suitable power levels are available. Using this technique,a number of fundamental and first overtone bands of terminal acetylenecompoundshave beenrecently measuredand analyzed.31 In this paper we report on the f&St overtone band of the acetylenic CH stretch (2~~) of propyne. In the companion paper, reinvestigation of the vi fundamental band of propyne on the same apparatus is reported. Following this work, Go and Perry3’have observedthe sameband of propyne using an IR-IR double resonancemethod that uses sequential vl==O+ 1-t 2 excitation. Using the same double resonancetechnique,Go and Perry3’-have also observedthe y1+ v6 state and have plans to study the vi + ‘v2 +2vg as well. At Princeton, we have also measuredthe fundamental and first overtone bands of CD3CCH,34and have recently observed the 3~~ and 2v1+lrg band of CH,CCH using double resonance vl =O+ 1-t 3 excitation.35Lower resolution studies of the full range CH overtone bands of propyne have been reported in the thesis of Hall,36and in the paper by Baylor, Weitz, and Hofmann.37 Hall was able to observeand fit the J structure in severalof the higher propyne bands,but eventhough limited only by room temperature Doppler broadening, he could not resolve the K structure. From the above it is clear that propyne is rapidly becoming an interesting model system for the study of IVR in intermediate size molecules. II. EXPERIMENT The 29 band of.propyne was measuredin a collimated molecular beam, using optothermal detection.38The apparatus has been described previously.3o The beam was formed by expanding a 10% mixture of propyne in He from a nozzle of 30 pm diam at the pressureof 4.5 x lo5 Pa (4.5 atm). For overtone excitation, the spectrometer employs a powerful ( - 100mW), continuous wave, high resolution (0.0001 cm - ’) F$ color center laser tunable from 1.4-1.6 ,um. The laser radiation is multipassed across the molecular beam using a pair of parallel mirrors. This design requires that the laser cross the molecular beam at an - lo” angle away from the normal, resulting in an instrumental resolution of 0.0006 cm-’ (18 MHz) at 6500 cm-‘, primarily due to residual Doppler broadening.The 64 65 66 67 68 Frequency 69 70 (cm-’ 71 72 ) FIG. 1. The propyne Zv, spectrum near 6570. cm-l showing the rotational progression between P( 7) and R ( 10). The spectrum was observed in a skimmed supersonic beam with color center laser excitation and bolometric detection. For clarity in labeling the frequency axis has 6500 cm-’ subtracted from the total frequency. need for such high resolution is demonstratedby the fact that at twice the resolution, the K= 1 and 2 subbranchesin the yl fundamental of propyne are only partially resolved.39Absolute frequencycalibration of the spectrum is basedupon the R ( 1) through R (6) lines of C,H, observed simultaneously in a static absorption cell. The acetylene transition frequencieswere taken from Baldacci, Ghersetti, and Rao,40and have a reported accuracy of f 0.005 cm- ‘.m Relative frequencycalibration is provided by the the transmission peaks of a - 150 MHz, confocal etalon, resulting in a precision of 2x 10e4cm-‘. The free spectral range of the Ctalon was constrained so that the K=O ground state combination differencesgive the values predicted from the known ground state rotational constants.41 III. RESULTS The first overtone absorption spectrum of the acetylenic CH stretch is a parallel transition with the usual AJ =0, f 1 and AK=0 selection rules.42This band was observedfrom P( 12) to R( 10) between6562 and 6574 cm-’ Fig. 1 shows P(7) through R( lo)]. At the molecular beam temperature of -20 K (determined from a Boltzmann analysis of the K=O transitions), the bulk of the intensity is found in transitions with K between0 and 3 (A=5.3 cm-‘), with K=O and 1, the lowest K values of the A and E nuclear spin modifications, being the most intense. Figure 2 shows an expandedscale of R(4) and P(6) lines of the spectrum, which are representativeof the fine structure found for other rotational transitions as well. The appearanceof many near-resonantperturbations is the significant qualitative changein the propyne spectrum upon going from y1 fundamental to overtone excitation. These additional lines in the spectrum result from weak couplings to the near-resonantbath states.The calculated density of states of propyne at 2v1 (p=66 states/cm-‘) is almost identical to that of trifluoropropyne at y1 (p=57 states/ J. Chem. Phys., Vol. 100, No. 4, 15 February 1994 Downloaded 18 Mar 2002 to 128.112.83.42. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp Mcllroy et al.: The propyne 2v, band ropyne a) TABLE 2~~ K=O T R(4) I. Observed propyne 2v, transitions. Assignment %,C10) QR,(lO) QRz(10) QR,(10) QR,W QR,(9) QR#4 QR,(9) QR,(8) QR~@) QR,(8) 6571.0000 6570.9500 6570.9001 a- Y (cm-,‘) b) QR,(S) QRr,(7) QR, (7) Q&(7) K=O P(6) K=l T Q&(7) Q&3(6) Q&(6) Q&(6) 6564:7500 6569:7000 +-u (cm-‘) 6564.6500 FIG. 2. Detail of the propyne 2vt spectrum P(6) and R(4) transitions, which share a common upper state, S = 5. The K assignments of the lines is by ground state combination differences determined from previous microwave studies. Note that K=O is a single transitions, but the K= l-3 transitions are each split by mixing with background states. This is typical of the J,K multiplets observed here. cm-‘), which was also found to contain extra lines due to near-resonantcouplings.l9 The quality of the data can be determinedby the standard deviation of the fit to the precisely known ground state combination differences.41Fitting 71 pairs of transitions sharing a common upper state we find that the data for 2~~ propyne yield a residual standard deviation of 7.3 MHz. Table I contains the complete list of assignedlines, including relative intensities. Basedon previous experience with the spectrometerthe relative intensity precision over a region of a single R(J) or P(J) transition group is estimated to be - 10%. Based upon the combination.differences, the precision of the frequencies listed should be -0.000 24 cm-’ ( -7 MHz), though relative spacingsin each clump are likely slightly better. The absolute frequency accuracy of the lines is limited by the acetylene calibration standard to iO.005 cm-‘. 2599 %(4) QR, (4) QRo(2) Y (cm -I)* 6574.2417 6574.2558 6574.2572 6574.2586 6574.3080 6574.2640 6573.7050 6573.7232 6573.7208 6573.5352 6573.7623 6573.7314 6573.1658 6573.1809 6573.2234 6572.9636 6573.2163 6573.1947 6572.6232 2572.6332 6572.6362 6572.3976 6572.6390 6572.6577 6572.6682 6572.6707 6572.6577 6572.0775 6572.0815 6572.0853 6572.0875 6572.0894 6572.1009 6571.9858 6572.1172 6572.1201 6572.1464 6572.1201 6572.1227 6572.0949 6571.5288 6571.5188 6571.5486 6571.5673 6571.5709 6511.2600 6571.4652 6571.5826 6570.9764 6570.9498 6570.99 5 1 6570.999 1 6571.0122 6570.9972 6571.0145 6571.0166 6571.0184 6571.0322 6570.9303 6571.0364 6570.4216 6570.4300 6570.4413 6570.4596 6570.3868 6570.5010 6569.8637 Relative intensityb 0.206 0.015 0.101 0.095 0.070 0.074 0.225 0.021 0.193 0.011 0.058 0.048 0.317 0.232 0.014 0.014 0.085 0.072 0.348 0.009 0.244 0.012 0.020 0.057 0.056 0.014 0.057 0.385 0.070 0.036 0.029 0.010 0.258 0.012 0.077 0.042 0.007 0.042 0.007 0.017 0.458 0.029 0.385 0.050 0.050 0.009 0.011 0.053 0.436 0.006 0.375 0.008 0.015 0.010 0.013 0.071 0.009 0.008 0.025 0.022 0.498 0.017 0.467 0.112 0.039 0.005 0.488 Downloaded 18 Mar 2002 to 128.112.83.42. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp J. Chem. Phys., Vol. 100, No. 4, 15 February 1994 Mcllroy et al.: The propyne 2v, band 2600 TABLE TABLE I. (Continued.) Assignment QR,W Q&(2) QMl) QR,(1) QRoVN QPdl) +x2) QP,(2) QPd3) QP,(3) Q&(3) Q&(4) QP,(4) Q&(4) QPd4) . QPo(5) QP,(5) QPd5) Q&(5) QPo(61 QP,(6) Qp2(6) Q4(6) Q&(7) Q&(7) Q&(7) Q(7) Q&(8) Q4@) QP2(8) QP3W QP4F3) Q&l(9) QP*(9) QPd9) QPd9) QP,ClO) Y (cm-‘)’ G 6569.8826 6569.9038 656918945 6569.3025 6569.3214 6568.7384 6567.6007 6567.0281 6567.0472 6566.4521. 6566.4711 6566.4871 6566.4895 6565.8731 6565.8921 6565.9043 6565.9137 6565.8436 6565.2905 6565.2990 6565.3 100 6565.3288 6565.2566 6565.3706 6564.7053 6564.6788 6564.7242 6564.7280 6564.7413 6564.73 10 6564.7439 6564.7461 6564.7478 6564.7615 6564.6602 6564.7665 .6564.1171. 6564.1073 6564.1370 6564.1562 6564.17206563.8495 6564.1720 6564.0549 6563.5256 6563.5297 6563.5336 6563.5359 6563.5378 6563.5492 6563.4355 6563.5661 6563.5694 6563.5950 6563.5694 6563.5721 6563.5455 6562.93 10 6562.9413 6562.9444 6562.7063 6562.9473 6562.9658 6562.9768 6562.9796 6562.9670 6562.3333 Relative intensityb 0.464 0.048 0.027 0.381 0.284 0.213 0.161 0.288 0.204 0.306 0.307 0.020 0.030 0.329 0.308 0.025 0.038 0.014 0.287 0.009 0.250 0.082 0.024 0.0090.269 0.005 0.232 0.012 0.009 0.006 0.008 0.051 0.006 0.007 0.017 0.019 0.209 0.016 0.181 0,029 0.024 0.005 0.025 0.006 0.320 0.040 0.03 1 0.024 0.006 0.251 0.012 0.050 0.055 0.006 0.050 0.008 0.012 0.269 0.007 0.152 0.009 0.009 0.043 0.046 0.011 0.013 0.192 I. (Continued.) Assignment QP,(W QP,UO) QP,(lO) QPlJ(ll) QPL(ll) QPdll) QPdll) Q&C12) QPl(l2) +2(12) Q&C12) Y (cm-‘)” 6562.3487 6562.3913 6562.1320 6562.3848 6562.3641 6561.7327 6561.7486 6561.7511 6561.5638 6561.7910 6561.7610 6561.1292 6561.1436 6561.1450 6561.1463 6561.1964 6561.1538 Relative intensityb 0.152 0.007 0.010 0.045 0.041 0.173 0.117 0.013 0.007 0.036 0.032 0.111 0.007 0.040 0.054 0.030 0.027 “The frequency uncertainty is 0.0002 cm-’ (-7 MHz). The absolute frequency accuracy is 0.005 cm-’ and is limited by the calibration gas frequencies (see text). bRelative intensities within a P(j) or R(J) transition set are accurate to - 10% (see text). The K assignmentof thesetransitions is obtainedusing precisely known ground state combination differences.4’ For a symmetric top, the energy difference between the eR(J-l,K) and QP(J+l,K) lines in the spectrum is [B - DJKK2](4Ji- 2). Thus, making K assignmentsof transitions is in generaldifficult due to the smallnessof DJR for most molecules.However, given the relatively large ground state DJK (5.451 17x 10V6 cm-‘) of propyne41and the low residual standard deviation (=+=2.4X10m4cm-‘) of the fit to ground state combination differences,it is possible to distinguish the K dependenceof the shifts, and thus obtain unique assignments for most of the transitions. K=O and 1 at low J have the smallest relative shifts and thus are most difficult to assign.The strongest K== 1 lines could be found in the Q branch ruling out a K=O assignment. Some ambiguity exists for the assignmentsof several weaker lines which we assignto K= 1. These assignments are made to K=l largely due to the fact that the K=O subbranch fits a rigid rotor term value formula within experimental error, arguing that it is at most only very slightly perturbed. In the analysis of symmetric top spectra that suffer perturbations it is usually advantageousto first analyze each individual K subband.If these results are sufficiently regular, the final step in the full analysis is simply to account for the subbandorderings. The subband orderings will often be anomaloussince first-order Coriolis splittings, in those bath states which have vibrational angular momentum, lead to large shifts with Kin the relative spacings of different vibrational states. As a result, near resonant interactions with different states will typically occur for each value of K. In some cases,like in the fundamental of CFsCCH which has a relatively small d rotational constant, the K subbandordering can be accounted for by a simple perturbation scheme.” J. Chem. Phys., Vol. 100, No. 4, 15 February 1994 Downloaded 18 Mar 2002 to 128.112.83.42. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp Mcllroy et al.: The propyne 2~~ band TABLE II. Rotational constants of propyne iv, .’ Constant Fitted valueb (a) Fit including only K=O data 6568.17155(16) i&lx 10s - 1.520 6(26) (b) Fit including 6 &4x103 ABxlO! PredictedC TABLE III. Rigid-rotor linear least squares fit to the K= l-3 subbands.a Previous workd (a) Fits to the K= 1 centers of gravity data 6568.19077(98) 0.283 507( 14) BAT “sub all K=O data and K= 1 data with J’= 1 and 2 6 568.19 568.171 6(2) 17.46(30) 17.6x lo-’ - 1.349 -1.54 - 1.520 6(24) vsub *All value5 in cm-‘. bValues from a least squares fit of all K=O transitions and K= 1 transitions terminating on P= 1,2. In the fit, lower state constant B value was fixed at the microwave value (Ref. 41). Values in parentheses are 20 error estimates for the parameters, assuming that the errors are statistical. ‘Values predicted based upon the change in rotational constants determined by McIlroy and Nesbitt (Ref. 29) analysis of the n, fundamental, K-O and 1 only. dThese values come from the Doppler-limited spectrum of Antilla et al. (Ref. 43). Since the K=O transitions are unsplit and show no obvious signs of perturbation, they may be usedto estimate B’ in a linear least-squaresfit using a rigid rotor expansion. In this fit, the ground-state rotational constants (B”,D,N) were fixed at the values determined via microwave measurements,41and the upper state distortion constant, D;, is held fixed to the ground-state value. The results of this fit are given in Table II. The rigid rotor expression fits the K=O data yield a residual standard deviation of the fit of 8.1 MHz, only slightly worse than the precision of the data. The value for a’=$( B”- B’) deducedfrom this fit is only 14% larger than that determined from the pi fundamental. Such a fit of a single K subband cannot determine an upper state A rotational constant. Since the lowest observedK= 1 upper states (J’ = 1 and 2) appear unsplit in the spectrum, we also fit.to a symmetric top equation including these upper states as well. The results of this fit are also shown in Table II. The fitted value for AA is almost exactly twice the fundamental AA value deducedby McIlroy and Nesbitt basedupon the K=O and 1 splitting observedin the fundamental.29However, as shown in the companion paper on the reinvestigation of the pi fundamental, this splitting in the fundamental is strongly perturbed.39 Fits to the higher K values in the y1 spectrum predicts a AA of -4X 10m4cm-’ as opposedto the +8.8 X 10B3cm-’ deducedfrom the K=O-1 splitting. Also compared in Table II are-the results of Antilla et a1.43 who studied the Doppler-limited spectrum of 2~~of propyne, which could not resolve the K structure. From the position of the y1 and 2~~bands, we can calculate that Xi1=$[E(2~1)-E(~I)]=-50.97 cm-‘, which is in good agreementwith the value of -5O* 1 cm-’ determined by Baylor et a13’ in a fit of the y1 through 6vi bands at low resolution. It is also in good agreementwith the value for Xii determined for a number of terminal acetylene compounds.36 For the K> 0 subbranches,fits are made to the centerof-gravityof all linesassigned to the sametransition.Table Predicted’ Experimentalb Constant 6568.19 - 1.54 ... -1.349 2601 (b) Fits to the K=2 %f (c) Fits to the K=3 vsub B& ... 0.283 719 data centers of gravity data 6568.205( 14) 0.283 55(19) 0.283 687 centers of gravity data 6 568.161(18) 0.283 92(24) 0.283 633 ... ... ‘All values are in cm-‘. The uncertainties are 20 of the fit. The term vsub is the subband origin. The term B&given by B& = B’ - D;KK2. %e ground state constants are held fixed to their literature values in the tits. 0; is held at the ground state value (Ref. 41). The predicted values are calculated using the aB constant reported in Ref. 39 and fixing D;K to the ground state value. III contains the results of these fits to the individual subbands. The D of the fits are 39, 510, and 570 MHz for K= 1,2,3, respectively, all much greater than the experimental precision. Despite mixing of the 2vi level with background states (which causesthe transition intensity to be distributed over more than one eigenstate), the centerof-gravity of all transitions of a given type [i.e., all RK(J) lines for a fixed J,K] should be at the unperturbedenergyof the 2v1 state, as long as the bath states have no absorption intensity themselves(single bright state assumption). Figure 3 shows the calculated minus observed frequencies, when the transitions are predicted using the constants listed in Table II, It is seenthat the K= 1 levels fit well, up 0.20 0 0.05 0 - 0 0 0 0 1 2 . * ~ e 3- 4 5 6 7 6 9 10 11 12 J’. FIG. 3. Propyne 2~~ perturbations. A plot of the deviations of the J,K multiplet spectral centers of gravity from the predictions based upon a tit to the K=O and 1 subbranches. The overall shifts of the K= 1 and 3 lines reflect perturbations of the subbranch band origins, while the curvature of the lines reflect long range perturbations which which perturb the fits to each individual subband. Notice that the perturbations grow in strength as a function of K. 1994 Downloaded 18 Mar 2002 to 128.112.83.42. subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp 100, No. 4, 15 February J. Redistribution Chem. Phys., Vol. Mcllroy et al.: The propyne 2v, band 2602 1.2 , 1.0 - cl o^ 6 0.8 - cl .tJ sd. 0.6 . a & 3 o.4 0 l- 2 3 4. 5 0” n y : ;..I 6 7. 8 9 10 11 J’ FIG. 4. Propyne 2v, intensity ratios. Predicted and observed R-branch total multiplet intensities rationed to K=O for a given J to minimize the effects of a non-Boltzmann intensity distribution. Note that no systematic loss of intensity is observed in contrast to the systematic frequency shifts of the spectral centers of gravity observed as a function of K. to through .T=7, and have~modesterrors above that. The K=2 and 3 subbranches,however, have large systematic shifts ( -0.04 and 0.15 cm-‘, respectively), and a large curvature as a function of J, The systematic shifts reflect the fact that, as in the fundamental, the AA value deduced from the K=O-1 splitting is much too large. In fact, the K=3 subband is shifted -0.05 cm-’ to the red of the K=O subbranch,not 0.17 cm-’ to the blue as predicted by the constantsin Table II. The curvature in the residualsin Fig. 3 is most likely due to perturbations by other, unobservedstates in the spectrum. Figure 4 shows a cotiparison of the experimental K= l-3 transition intensities normalized to the unperturbed K=O transitions, and the calculated ratios, for a 24 K Boltzmann distribution, determined from the K=O, 3”=0-8 data, with symmetric top HSnl-London factors and with spin degeneraciestaken into account. This analysis permits us to look for large ( > 10%) intensity losses due to missed transitions. Clearly, no systematic intensity loss is evident, suggestingthat the missed transitions are indeed fairly weak. Since the shifts induced by these perturbers (-0.05 cm-‘) are fairly large, the small intensity loss to these perturbers imply that they must be separated from the bright state by an energy gap significantly larger than the shift or the coupling matrix element. IV. ANALYSIS OF THE NEAR-RESONANT PERTURBATIONS The multiple transitions observed for individual J,K in pr6pyne in the first acetylenicCH stretch overtone indicate that the vibrational mixing leading to IVR has already becomesignificant, even in this simple molecule at this relatively low level of excitation. In this section, we consider several aspects of the state coupling that create this vibrational state mixing. We start with the assumption that at this time, the complex nature of this multiple state mixing is extremely difficult, if not impossible, to understand within the spectroscopicprecisionof the data; that is, states we are not able to develop a physically meaningful model Hamiltonian which can reproduce the data within 0.0002 cm-’ . We will insteadseekquantitative information which can help us understandthe broader, unansweredquestions concerning vibrational mixing. Such questions include the relative importance of Coriolis and anharmonic coupling mechanismsand the fraction of background states participating in the vibrational mixing. In the analysis that follows we are concernedwith incorporating the fact that the limited signal to noise of any measurement directly affects the observed spectrum. A good test of the completenessof a high resolution data set is to fit the spe@ralcenter-of-gravity positions of eachmultiplet to rigid ?otor formulas, with distortion fixed to the ground state values, for each subband.This approachhas been previously employed in the study of the acetylenic C-H stretch fundamental of CF3CCH.19 There it was found that the centers-of-gravity of the multiplets were well described by a rigid rotor expression, and that the rotational constantsreturned from thesefits were in agreement with that obtained from the fit to the unperturbed subbands. As discussedabove and shown in Fig. 3, the fits to a rigid-rotor expressionis excellent for K=O transitions, but for K> 0 the fits to the centers-of-gravity are quite poor and get progressively worse for the higher K subbands. Errors in the intensity measurementswill make the calculated center-of-gravity less precisethan the individual line frequencies,but such errors should be largely uncorrelated for the P and R branch. Thus the high correlation of the fit residuals found for the two branchesarguesthat the dominant effect is that we have failed to observe all the perturbers in the spectrum. The most likely perturbers to be missed are those with both large interaction matrix elements and detunings since these states “steal” relatively modest intensity but have a disproportionate effect on the center-of-gravity. Still thesestatesmust not be too far away ( d 1 cm-‘) or else their effect would vary smoothly with J and thus could be incorporated into an effective rotational constant and the fit would not be compromised. In fact, the effective rotational constants predicted by the fits increasewith K, opposite to those of the ground state and the y1 fundamental, indicative of just such nonresonant interactions. A. Density of coupled states A quantity of furidamental importance in understanding the IVR process is the fraction of the bath states that mix with the bright state. From this we may elucidate whether selectionrules createsome preferential coupling to a subset of states. For example, if anharmonic coupling aloneis responsiblefor the coupling, only vibrational states of A 1 and A, symmetry for J> 0 would be coupledto 2~~of propyne. (The il, states can couple since, as discussedin the precedingpaper,thesestatesonly occur in A, +A, pairs that are strongly mixed by a first-order parallel Coriolis interaction for K > 0.) Clearly two pieces of information are required to make such a comparison, the total density of statesand the density of statesobservedin the spectrum. J. Chem. Phys., Vol. 100, No. 4, 15 February 1994 Downloaded 18 Mar 2002 to 128.112.83.42. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp Mcllroy et al.: The propyne 2v, band For a small molecule such as propyne at the relatively modest energy of 6500 cm-‘, it is possible to obtain an excellent estimate of the total number of bath states by direct counting algorithms, such as that of Kemper et al., which we employ here.44Using the vibrational frequencies and diagonal anharmonicities compiled by McIlroy and Nesbitt,29 we find a total vibrational state density of ptotal=66 states/cm- * (including all vibrational angular momentum sublevels). In the C,, point group, this total state density breaks down to 11, 11, and 22 states/cm-’ of A *,A,, and E symmetry, respectively. Calculating the experimentaldensity of states,pcoupled, is more difficult. This quantity is most simply estimated from Pcouplcd = htes AE’ (1) where two quantities must be determined; nstates,the number states observedfor each bright state, and AE, the energy window in which these states appear. A simple method that has been applied in the past is to count the number of transitions in each multiplet and divide by the width of the multiplet. This simple algorithm truncates the frequency window on the outermost peaksof the multiplet, artificially enhancing the state density. This effect can be eliminated by using ( nstates-1) in place of nstatesin the above formula. This can be understood when one realizes that AE is an (nstit,- 1) nearest neighbor spacing. The above assumes,however, that we have detected all of the states in the energy interval AE. To the extent we miss levels, we will systematically underestimatethe true density of states. When many transitions are observed,so that an intensity distribution function can be described,then it is possible to try to estimate the fraction of observedlines by extrapolation down to zero intensity. This has been done for the NO2 spectrum for example.45For a weakly fractionated spectrum, such as the present propyne spectrum, such a method is, unfortunately, not applicable. We propose here an algorithm, applicable when the zero-order bright state character stays largely localized in a single eigenstate(sparsecaseIVR), which accountsfor the tlnite experimental signal to noise ratio and provides a properly weighted average of the observables, such as pcouPled as considered now. The basic ideas is as follows: Since the fractionation due to IVR is not sufficiently great to determine reliable averageproperties of the spectrum, we need another method for calculating useful quantities. Many quantities can be determinedif it is possibleto derive the distribution function of the observables,such as the coupling matrix elements.To obtain the distribution function we pool the information obtained over the entire spectrum, not just over a single IVR multiplet. We will assume that all the observedperturbations are to independentbath states and thus are statistically independent.This will allow us to perform averagesover all the observedperturbations without having to decide which are infact unique. Below, we will show that the expected relative tuning of the states with J supports this assumptionfor at least most of the observedtransitions.Sincethe deriveddistribution 2603 function will involve all of the data, which is of varied quality, it is necessaryto include the biasing effectsof limited signal to noise in the spectrum. This idea of using the information present in all measuredtransitions is now illustrated in the calculation of pcouPled. A perturbation will be observedin the spectrum only if the mixing betweenthe bright and bath states is sufficient to produce a transition whose intensity exceedsthe experimental signal to noise ratio (S/N). Given the relatively isolated nature of the perturbations observedin the propyne 2~~ spectrum (and more generally, for sparse case IVR), it is not unreasonableto consider the perturbations one at a time, i.e., as casesof isolated two state mixing. In the perturbative limit, the intensity of a dark state in the spectrum will be proportional to ( V/AE)2, where V is the coupling matrix element and AE is the energy difference of the zero-order energies.Thus a perturbation of strength V will only be observedif 1AE 1 < (S/N) lnV. We then define the maximum energy window for the observation of a perturbation as AE lnax(V) =2(S/N)“2V. (2) The probability of observing a perturbation, PobS, of a given size range [Vi, Vi+ A V], in a single transition is then given by the product of the most likely number of statesin the “visibility window” and the probability of the propersized coupling strength appearingin the probability distribution of coupling matrix elements [P( V)], Pot&V= [Wn,( F7)X~c,x,~ledlX [p( vi)AU. (3) Under the assumptionsdiscussedabove, each individually observedVi occurs for a bright-state/bath-state interaction at one and only one upper state. Therefore, when examining one of the NB observedupper states,the probability of finding a given matrix element, say V,, can be written as pobs( v,> = Mn,( vk> &mpled& v,) =& . (4) All that remains is to estimate P( V) and then the experimental value of the coupled density-of-states can be determined from Eq. (4). The value of this distribution function will be determined by summing over all measurements in the spectrum. Already the analysis above has included the effect of finite signal to noise through Eq. (2). However, we also note that the experimentally measured matrix elementswill be biased since stronger coupling matrix elements are more likely to produce observablesplittings. Thus the experimentally observeddistribution is actually proportional to V +P( V), where P( V) is the “true” matrix element distribution function required in Eq. (4). Therefore, the experimental estimate for P( V) is given by the normalized distribution ZS(v- Vj) p(~=y.~(yi)-l* (5) The sum extends over all dark states observedin the spectrum of NB independentbright states. The distribution of Eq. (5) is a primary result for much of the analysis in this Downloaded 18 Mar 2002 to 128.112.83.42. Redistribution subject to AIP or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp J. Chem. Phys., Vol. 100, No. 4, 15license February 1994 Mcllroy et al.: The propyne 2v, band 2604 paper and shows how the experimental bias for observing larger matrix elements is accountedfor in calculating the true distribution P( V) . Substituting for P( Vk) and AE,,, and solving for pcoupkd in Eq. (4) gives Z( Vi)-* PwuPld=&( S/N) l/f * (6) For the observed J,K multiplets of 2~~ propyne, we have N,=46 and the Vi are calculated using the deconvolution method of Lawrance and Knight46 (seebelow). The results of the Lawrance-Knight deconvolution are presented in Table IV. With an averageS/N=20 over the entire spectrum, Eq. (6) gives pcoUpled=42 states/cm-*. The A, symmetry 2~~ state has allowed anharmonic coupling to only other states of Ai+& symmetry (33% of ptotal, 22 states/cm-‘). The calculated value of pcoUpled suggeststhat at least most of the isoenergeticstates with correct vibrational symmetry for anharmonic interactions are, in fact, coupled. However, basedsolely on this calculation, we cannot say unambiguously whether the additional E symmetry vibrational states are coupled (which would have to be through x,y-Coriolis interactions). This question is better addressed by considering the J-dependenceof the coupling strengths. B. Coupling strengths The vibrational state mixing observedin this spectrum comes from two types of perturbations. First, nearresonantmixing splits many of the transitions into multiplets of two or more lines. Second,the spectral centers-ofgravity of these clumps are shifted due to the type of nonresonantcouplings discussedin the accompanyingpaper.39The latter type of coupling is most easily observedby the anomalous K subband ordering. In this section the near-resonantinteractions are analyzed. Nonresonant interactions are discussedin the next section. Near-resonantcouplings in propyne are characterized by small matrix elements ( <0.05 cm-‘) which only significantly mix statesthat are very local to 2~~.This type of coupling is observed in about 2/3 of the K=l-3 states. When the spectrum arises from a single bright state coupled to the dark vibrational states of the bath, the mixing can be deconvolutedusing the Green function deperturbation schemeof Lawrance and Knight.46In caseswhere this method can be used it is often possibleto obtain information about the rotational dependenceof the couplings if the deconvolution is performed at a number of different J values.19This is the approach we follow here for propyne. In contrast, molecules which possessinternal rotors are usually not amenableto this procedure since the spectrum is often a superpositionof two or more spectra from different torsional state symmetries.47 Given N positions and intensities of the eigenstates resulting from mixing of a single bright state, the deperturbation returns the (N- 1) matrix elements,Vi, between the bright state and the (N- 1) dark statesand the locally unperturbed bright and dark state energies. More pre- TABLE IV. Matrix elements determined by LawranceKnight deconvolution. Individually determined values of the matrix elements and centers of gravity for the perturbed transitionsa (i) R= Y=4 Center Matrix S=5 Center Matrix J’=6 Center Matrix J’=7 Center Matrix J=8 Center Matrix f=9 Center Matrix S=lO Center Matrix J’=ll Center Matrix 1 transitions of gravity= elements: 6570.4409 21 6565.3096 20 of gravity = elements: 6570.9951 6 55 32 6564.7241 8.5 63 31 of gravity= elements: 6511.5465 76 6564.1346 81 of gravity= elements: 6572.0949 26 26 34 64 6563.5445 '27 30 38 54 of gravity= elements: 6572.6361 5.5 6562.9443 6.3 a of gravity= elements: 6573.1833 10 6562.3505 9 of gravity =~ elements: 6573.2577 7 6561.7488 I of gravity= elements: 6574.2577 5 I 6561.1456 5 7 6569.9005 44 6565.9100 45 6571.0159 6 8 42 57 6564.7462 5.8 8.3 41 58 of gravity= elements: 6571.5691 18 6564.1633 18 of gravity= elements: 6572.1081 14 66 370 6563.5562 18 63 400 6572.6406 13 60 83 690 6562.9502 14 66 67 700 of gravity= elements: 6573.1805 88 6562.3388 97 of gravity= elements: 6573.7261 83 6561.1540 a4 6570.3869 362 6565.2877 501 (ii) K=2 transitions J’=3 Center of gravity= Matrix elements: J=5 Center of gravity= Matrix elements: J’=6 Center Matrix J”=7 Center Matrix ;7’=8 Center of gravity = Matrix elements: S=9 Center Matrix J’=lO Center Matrix (iii) K= 3 transitions J’24 Center of gravity = Matrix elements: J. Chem. Phys., Vol. 100, No. 4, 15 February 1994 Downloaded 18 Mar 2002 to 128.112.83.42. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp Mcllroy et al: The propyne 2v, band TABLE 2605 IV. (Contintied.) K=l Jf=5 Center Matrix .P=6 Center Matrix of gravity= elements: 6570.9800 53 6564.7163 53 of gravity= elements: 6571.5251 960 490 6564.1077 990 520 6572.1205 9 6563.5698 9 P=7 Center of gravity= Matrix elements: * 6 K=Z “The two columns are for the R (first column) and P branch (second column) data. The comparison of the matrix elements determined for the two branches gives a feeling for the precision of the relative intensities which enter into the deconvolution procedure. The matrix elements are 104x the determined value, i.e., a matrix element in the table reported as 25 is 0.0025 cm-’ ‘&mpling matrix element in the spectrum. 0.08 K=3 0.06 0.04 cisely, the deconvolution determines Vf, and thus the relative signs of the Vi are unknown. In the following, these signs are all taken to be positive, i.e., Vi in the formula is written for 1rfl. The unperturbed bright state p~ositionis just the spectral center-of-gravity. The results of this deconvolution for the first overtone of propyne are given in Table IV. Using the matrix element distribution function P( V) introduced in Eq. (5) of Sec. IV A, we can calculate (v> and ( Y2), corrected for fmite S/N, from the formulas (8) In this formula, ND is the total number of dark states extracted from- the deconvolution of all the bright state transitions in the spectrum, and the sum extends over all ND values of Vi determined by the deconvolution. The mean coupling matrix element calculated in this way ((v> =O.OOl 95 cm-‘) is an order of magnitude smaller than the result of a straight average.This implies that there are many more weak perturbations which are unobserved. Theoretically, the root-mean-squareof the matrix elements (V,,) is more interesting since-its value is invariant to the mixing of the bath states. However, the V,, statistic is strongly influenced by a-few large matrix elements and is poorly determined in the present small data set. Now that the matrix element distribution function, P( V), has been used to calculate the required averagevalues for the spectrum we briefly discuss the validity of this form of the distribution function. The above formulas for P~,,~I~, ( I VI > and ( V2> are derived under assumptions that are ‘only strictly valid in the sparse limit where Vmp,,pld(l. In order to check the argument, and accuracy of the formula as one approachesthe intermediate lid ( Vms~coupkd - 1), a numerical simulation was performed of the interaction.of a single bright state with a bath of 100Poissondistributed levels, coupled to the bright state by matrix elements obeying Gaussian statistics. The effectsof finite S/N wasincludedby only retainingin the ;:Le 0 12 3 0 0 4 5 . 6 J 7.8 9 10 11 Ii FIG. 5. Average propyne 29 matrix elements for each J,K multiplet determined from a deperturbation assuming a single bright state and a bath of prediagonaliied dark states. Note the apparent growth of the matrix elements with K, but the lack of a systematic growth with J’. analysiseigenstateswhosebright state population is greater than (S/N) -I. This choice definesthe “signal” as the intensity.expectedfor the bright state if unfractioned, i.e., the sum of intensity over all eigenstatesbelonging to a single bright state ,transition. All eigenstatesabove the “noise” were retained and subjectedto a Lawrance-Knight deconvolution to determine the “experimental” coupling matrix elements.These coupling matrix elements,along with the S/N, were then used with the above formula to calculate pc,,,,pled,( 1VI ), and ( V2) which could be compared with the correct values for the ensembleused in the simulation. It was found that for a S/N of between 100-1000,the above formulas are accurate to better than - 1% for Vmspcoupled < 0.3. For ;C~~spc0upled=0.57, the value of pcoupled was underestimatedby only l%, while the values of (IV/) and (V2) are correct within the -5% sampling errors of the finite simulation (100 spectra calculated). This is about the deducedstrength of the coupling for the K=3 levels, which are the most perturbed of the observed transitions. For Vmspcoupled= 1, the formula starts to break underestimatedby 29 ( 13) % with the down, with &oupled S/N= lOO(lOOO), while ( I VI ) and ( V2) are overestimated by 10% and 25%, respectively for S/N=lOO, but are within statistical error ( -2%) when the S/N is increasedto 1000. For this large a value of V~@c,-@edthe averagenumber of “observed” dark states was 12.5 or 45 with the S/N= 100or 1000,respectively, and thus existing methods for dealing with “intermediate” case molecules will be applicablefor this or more strongly coupled spectra. By looking for trends in the dependenceof the matrix elementson rotational state, the presenceof Coriolis coupling may be detected. However, no obvious trends are observedas a function of J (see Fig. 5). If the E symmetry J. Chem. Phys., Vol.subject 100, No. 15license Februaryor 1994 Downloaded 18 Mar 2002 to 128.112.83.42. Redistribution to 4, AIP copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp Mcllroy et al.: The propyne 2q band 2606 TABLE V. Average coupling matrix elements determined by LawrenceKnight deconvolution. Subband K=O K=l K=2 K=3 w (I VI)X103 cm-’ 0 14 16 5 ... ... 1.4 2.6 3.8 1.8 7.3 12.7 mX V. NONRESONANT INTERACTlONS AND THE POSSIBILITY OF A DOORWAY STATE MEDIATED VIBRATIONAL ENERGY REDlSTRlBUTlON 103cm-’ ‘Total number of matrix elements from the deperturbation of the spectrum for each K. vibrational states were coupled through perpendicular (ny) Coriolis interactions the matrix elements would be observedto increase as #(J+ 1) --K(K+ 1).48 No such increaseis apparent in our data so we conclude that the E symmetry states are not strongly coupled to the acetylenic C-H stretch overtone in propyne. Some weak x,y coupling could be responsiblefor the fact (see above) that our estimate for ~~~~~~is higher than the calculated density of statesof Al and A, symmetries. However, the (v> calculatedby averagingover eachK subbandseparatelydoes show a systematic and essentially linear growth with K (see Table V). The V,, increases evenfaster than linear in K, but more slowly than K’. The I’,, values are also listed in Table V for comparison. The increaseas a function of K is consistent with the increasingly poorer fit of the higher K subbandsto rigid rotor formulas. As mentioned previously, these poor fits probably result from rather distant states which are coupled through relatively large matrix elements. The observed K rotational dependencemay indicate the presenceof Coriolis mixing. The most obvious type of coupling that vanishesat K=O and grows linearly with K is a tist order parallel (z-axis) Coriolis coupling.48The matrix element for parallel Coriolis coupling of two rovibrational states 1u,r) and 1u’,r’) is given by (w-1 --P$J~~( u’,r’) = -24&K, (9) where c,,, is the Coriolis coupling constant betweenthe Al and A, vibrational states. This matrix element increases linearly in K, as does our averagematrix element. Most importantly, it also vanishes at K=O which agreeswell with our observationthat the K=O statesare unperturbed. However, for propyne which has no A2 symmetry normal modes, this interaction cannot directly couple the Al acetylenic stretch fundamental to the bath states.39This is because this interaction only occurs between states both of which are excited in one or more E symmetry modes. In order for parallel Coriolis matrix elementsto be active, the acetylenic C-H stretch must be coupled via anharmonic terms to another, likely distant, A, state which should have at least two-quanta in the E symmetry modes.Coupling of bath states near 2~~ to this distant Al state could be induced by a z-axis Coriolis interaction. If secondorder coupling of 2~~ with the bath states through this distant state dominates over direct coupling, then this state is acting as a doorway state for IVR. This result is discussedin the next section. In this section we discussthe possibility that the vibrational energy redistribution may occur as a sequentialprocess involving coupling to a distant, nonresonant state which is then very effectively coupled to the rest of the bath states.This model is often called the doorway model. It is a limitation of the high resolution technique that it is does not allow for a direct determination of the state-to-state pathway for the energy relaxation from the spectrum. All models, whether they are doorway models, tier models, or simply direct interactions to bath statesmodels, in so much as they produce upon diagonalization the same eigenstates and distribution of bright state character, cannot be distinguished on the basis of a single spectrum. These models, however, may differ in their predictions for the distribution of specific dark state character among the eigenstatesnear a bright state resonance.As such, distinguishing between such models requires two, color experimentsthat probe for dark state character in the secondstep. Lacking such measurementsat present,a discussionof the mechanismof the vibrational energy flow must be inferred from indirect effects and from the consistencyof a model with all that is known about a molecule. The single most distinguishing aspect of the 2~~ propyne spectrum is that, to the level of precision of the data ( - 10 MHz), the K=O statesare locally unperturbed.This is determined by the facts that there is only one K=O transition assignedfor each P(J) and R(J) and that these transition frequenciesfollow a rigid rotor energy expression to the full precision of the data. Although the K=O states are free of near-resonantperturbations, they may still be affected by nonresonantperturbations. For example, the rigid-rotor fit of the K=O states in 2~~ yields a value for aB of- 7.609(26) X 10v4 cm-‘, which is larger than the value, 6.65(4) x lo-” cm-‘, obtained from the vi fundamental spectrum. This difference is likely due to the perturbation of a nonresonantstate. Furthermore, it has been pointed out that the K subbandorigins do not follow the predicted ordering for an unperturbed symmetric top spectrum. As discussedin the previous paper, this behavior is indicative of nonresonantinteractions.39 Here we will first show that the lack of perturbation of all the K=O states is a significant result. Using the results presentedabovefor the statistical analysis of an eigenstate resolvedspectrum, including the effectsof limited signal to noise, we can define the energy window over which a perturbation would be detected.This expressionis AE=(S/‘N)1’2((V)~. .( 10) In the previous section the averagematrix elements were obtained. For example, over the whole spectrum the average matrix element was found to be 0.001 95 cm-‘. The average signal to noise for the whole spectrum is about 20:1. Using thesevaluesthe energy window for observinga perturbation is 0.0087cm-‘. Earlier it was also shown that J. Chem. Phys., Vol. 100, No. 4, 15 February 1994 Downloaded 18 Mar 2002 to 128.112.83.42. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp Mcllroy et a/.: The propyne 2v, band 2607 the measured coupled density of states is 42 states/cm-‘. If these states are randomly distributed, the probability of finding an unperturbed transition is p,&g-~Pcoupled. 1”“” (11) b M ‘ii We then find that the probability of observing an unperturbed state is 69%. We observe 39 upper states in the spectrum and 19 ( -50%) of them show perturbations, which is in good agreement with the above estimate. Using the same analysis we can then estimate the probability of finding all K=O states unperturbed. For the K=O states, which have the largest thermal population in the molecular beam, we have an experimental signal to noise ratio of -2OO:l. Using the same average matrix element given above, the probability of finding a given K=O unperturbed for one observation is 3 1%, with an observation window of 0.027 cm-‘. We measure 12 different K=O states (S=O-11) in the spectrum. Therefore, assuming uncorrelated events, the probability of observing no K=O perturbations in the entire spectrum is less than one in a million, clearly demonstrating the anomaly of the result and thus the need for an explanation of this fact. Before attempting to explain this result, let us examine the assumption made above of the independence for the measurementsof each K=O transition. Bath states tune in and out of resonance with the bright state, due to differences in rotational constants. The amount of the energy shift on going from J to Jf 1 is 2SB( Jf 1) , where SB gives the difference in rotational constant between the bright state and the bath state. The assumption of independence for the K=O measurementsis valid if the size of the typical energy shift is larger than AE, the energy width over which a perturbation can be observed. This assumption can be checked by considering the rotational constants of the bath states, which can be estimated by using the values of aB calculated in the previous paper. From a list of nearly resonating bath states, calculated by the direct count algorithm, a histogram of rotational constants was calculated and is shown in Fig. 6. It is seen that the rotational constant for 2~~ is on the low end of the distribution, as is generally true for acetylenic stretches. Therefore, a majority of the states have SB of -4X low3 cm-‘. The energy window AE for observing a K=O perturbation is -0.03 cm-‘. Thus, for J’ > 4 the measurementsare largely independent. If we lower the number of independent measurements to 5, conservatively allowing independencefor transitions with J’ > 6, we find that the probability of findtig K=O completely unperturbed is still < 0.3%. The conclusion basedon the observedspectrum is that the resonant coupling of the K=O states in 2~~ propyne is completely shut off. As mentioned above, paTalle1Coriolis interactions have matrix elements that depend linearly on K and thus these interactions are turned off for K=O. However, as discussedin the accompanying paper, the A1 acetylenic C-H stretch of the C,, cannot directly couple to AZ bath states through this mechanism. The first-order coupling occurs between AI+A2 pairs arising from the combinations of E states. This Spectroscopicpeculiarity of the vibrational energy relaxation as occurring directly to the near-resonant bath states through z-axis Coriolis coupling processes. Another possibility is that the acetylenic C-H stretch is coupled via anharmonic interactions to the Al component of an A, +A, pair of states that contain excitation in the E symmetry modes (the doorway state). This pair of levels (which will be strongly mixed for K> 0 by a “diagonal” Coriolis interaction) may have z axis Coriolis interactions with other E modes, making it possible for parallel Coriolis interactions between the doorway state and the bath states near 2~~ to dominate over direct anharmonic interactions. This would lead to an efictive parallel Coriolis interaction between 2~~ and its background of nearresonant states.The postulated doorway state is believed to be nonresonant for two reasons. ( 1) It is not observed in the measured spectrum. (2) The J structure of the subbands is fairly regular, even though the fits to the subbands for K#O are poor. This coupling model is shown schematically in Fig. 7. Still there are some problems with this interpretation that should be mentioned. First of all, this model alone does not’preclude the perturbation of K=O states. Even without z-axis Coriolis interactions, 2~~ is coupled to the bath states by both direct and indirect (though the doorway state) anharmonic interactions. Second, the doorway State, which must be an Al +A2 pair of states, will tune as a function of K by *AC&K. Since we expect I&- 1, the doorway state must be many cm -’ away not to completely tune in or out of resonancefor a change in K by one unit. The larger the detuning of the doorway state, the larger we require the product of the bright state-doorway state anharmonic coupling times the doorway state-bath state Coriolis coupling to be to produce the observed strength of effective bright-bath state coupling. It is particularly difficult to understand why these interactions should be so the acetylenicC-H stretchpreventsthe iriterpretationof large in the presentcase,whenparallelCoriolisinterao /ij 500 0 !~~$y~~g~~~~~~“~~~ B’ (wavenumbers) FIG. 6. A histogram displaying rotational constants of states expected between 6500 and 6600 cm-‘. States where located by a direct count, and rotational constants estimated from the $‘s compiled by McIkoy and Nesbitt (Ref. 29). The position of the rotational constant of the 2~) state is indicated by the arrow in the figure. Downloaded 18 Mar 2002 to 128.112.83.42. Redistribution to AIP or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp J. Chem. Phys., Vol.subject 100, No. 4, 15license February 1994 2608 Mcllroy ef a/.: The propyne 2-q band Doorway State Anharmonic \FIG. 7. A possible doorway state coupling scheme for propyne ZY, which the apparent z-axis Coriolis interactions observed in the spectrum between bright and dark states. The bright state (2~~) is coupled to an (A,& pair doorway state via an anharmonic interaction. The doorway state is in turn coupled to the bath states by a &axis Goriolis interaction. This requires that the doorway state, and the coupled bath states, must have excitation in E symmetry normal modes. tions are of apparently minor importance at most in the IVR of other terminal acetylenecompoundsthat have been While the above picture studied at both y1 and 2~~.2g931 appears to be the simplest model which can explain the spectroscopicobservations,it is particularly hard to understand physically how this parallel Coriolis interaction can be of dominant importance in~the present spectrum. The zero-order assignment of the bright state as 2vr implies that the momentum of the atoms is parallel to the symmetry axis, which implies that there cannot be any such parallel Coriolis interaction1 In conclusion, the 2~~ spectrum of propyne has a strongly rotationally dependent matrix element for its near-resonant couplings. The average coupling increases approximately linearly with K and completely disappears for K=O. This behavior strongly suggeststhat a parallel (z-axis) Coriolis interaction plays a prominent role in the vibrational energy redistribution. There doesnot appearto be any systematic variation of the matrix element as a function of J’, suggestingthat perpendicular (x,y-axis) Coriolis interactions are not operating or, are at most are of minor importance. Since the ‘4, symmetry acetylenic C-H stretch cannot directly participate in first order parallel Coriolis interactions with bath states, we conclude that there is an intermediate state that is playing the dominant role in mediating the vibrational energy relaxation, although the exact nature of the mediation is open to discussion. Since this state is (or possibly, few states are) not directly observedin the spectrum, and since the rotational progressionsfor each subbandare fairly well behaved,.the doorway state must couple to the bright state in a nonresonant manner. There is definite evidence of nonresonant perturbations in the spectrum revealedby the highly perturbed subband ordering. We believe this to be strong evidence of nonresonantvibrational states controlling the vibrational energy redistribution in isolated molecules. Since completing this work, Go and Perry33have reexamined the 2~~ state by IR-IR double resonanceusing sequential excitation through the pi fundamental. They found (for J=5, K=&2) a state -0.03 cm-’ higher in energy that has many of the characteristics expectedfor our proposed doorway state. Namely, this state is anharmonically coupled to 2v1, and also is coupled to its nearly resonantbath statesby what appearsto be parallel Coriolis interactions. The size of the observed2v,-“doorway state” mixing and the average“doorway state”-bath coupling are of the right size to explain the average strength of the 2v,-bath coupling. Given the uncertainties in thesequantities due to the small number of observedperturbations,it is not possibleto be precise,but this new state detectedby Go and Perry appearsto provide a significant fraction of the 2v,-bath coupling, and thus largely confirms the model presentedin the presentwork. However, this state detunes in energy very slowly with K, and this strongly suggests that this state is not part of a nearly degenerateA r , AZ pair. This implies that this new state also has no vibrational angular momentum in any of the E symmetry normal modes, a requirement for the “doorway state” which we have proposed.If this is the case,then this nearby doorway state must itself be coupled to yet another, likely further detuned, “doorway state” that provides the neededvibrational character to allow parallel Coriolis interactions. Detailed calculations, of the type recently presented by Stuchebrukov and Marcus,56are clearly neededto try to further elucidate the nature of the chain of couplings responsiblefor the observedperturbations in the propyne2v1 spectrum. Further, the complex nature of the interactions illustrates the pressing need to develop methods to probe the vibrational character of the bath states observed through mixing with a well defined bright state, such as in these measurements. VI. NATURE OF THE UNDERLYING DYNAMICS: CHAOTICVERSUSREGULAR The observation of coupling to all symmetry allowed levels suggeststhat the spectrum may be strongly mixed or “chaotic.” Further, the rapid fluctuations in the number of perturbations observedat each different J’ level might also be interpreted as indicating chaotic behavior. However, as we shall demonstrate in this section, this is not the case. We focus on two different statistical properties of the speo trum, the distribution of matrix elements and the second nearest-neighborlevel spacing of the bath states. In both caseswe find that our observationsare consistent with the properties expectedif the classical dynamics of this system are regular and not chaotic. The random matrix model for a spectrum (the Gaussian orthogonal ensembleor GOE model) assumesthat the distribution of coupling matrix elements is Gaussian.4g With such a distribution, we should essentially never observe matrix elements more than a few times the mean value. Even with a weighting factor proportional to the matrix element,to account for finite signal to noise effects, 99% of the observed matrix elements should be <3-4 times the mean. In contrast, we observematrix elementsup to 50 times the mean value. This suggeststhat approximate selection rules determine the magnitudes of the observed J. Chem. Phys., Vol. 100, No. 4, 15 February 1994 Downloaded 18 Mar 2002 to 128.112.83.42. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp Mcllroy et ab: The propyne 2q band matrix elementsand that the eigenstatesare not statistically distributed in phasespace.Such propensity rules for the coupling naturally explain the importance of coupling through a set of special, i.e., doorway, states observedin the spectrum. A frequently used test for a chaotic spectrum is the behavior of the level spacingsof eigenstatesforming a pure sequence. 5oThis statistical measureof a spectrum was used previously to show that the 2~~ spectrum of CF,CCH is chaotic.19Obviously this type of analysis cannot be performed on the propyne spectrum sincethere is never a pure sequenceof more than a few levels in the spectrum. Further, given the wide range of matrix elements,most eigenstatesare m issing from the spectrum. Instead, we consider a related statistic which can be used in spectra where only a few perturbations are observed.P”-‘(s) is the nth spacing distribution function and is defined as the probability density that given the energy of the ith eigenstateis E, the energy of the i+nth eigenstateoccurs at E+sX d, where d= l/p is the mean spacingbetweenlevels. By this deflnition, p(s) is the nearest-neighborspacing distribution, while P’(s) is termed the next-nearest-neighborspacing distribution. A related quantity is the probability that the spacing between levels Ei+nt- 1 and Ei is between 0 and s and is given by I”(s) = i-s J o Pn(s’)ds’. (12) In particular, we will consider I’ (s), which is the probability that three sequentialeigenstatesare found in the interval s. Sincewe would like to comparethis quantity for a regular spectrum vs that expectedfor a chaotic spectrum, the formulas for P’(s) are needed in both cases. For a Poissonspectrum which has uncorrelatedlevels,characteristic of a spectrum displaying regular dynamics, P’(s) =se-‘, as calculated from the convolution of p(s) =e-’ with itself. Thus, I’(s)=l-(l+s)e-‘-y s2 for ~41. (13) For the GOE spectrum, characteristicof chaotic dynamics, there are strong correlations in the spectrum,.such that whenever there is a small spacing, the next spacing is biasedto be larger than averageto maintain the “rigidity” of the spectrum. P’(s) wascalculated by Kahn,” who found that for small spacingsP’(s) = (?r4/270)s4(for s---0.5, this formula is -20% larger than the numerically calculated exact result). We note that if the nearest-neighborspacings are (incorrectly) assumedto be Wigner but uncorrelated, P’(s) - (d/24>? for small s which is much larger than the correct result. From P’(s), we get for the GOE spectrum, I’(s) = --s5 ,;;, for small s. (14) We thus seethat the existenceof essentiallyany closely spacedtriplets in the spectrum is a sign that the spectrum does not have GOE statistics. For example, for s=1/3, I’(s) =0.055 for a Poissonspectrum, but only 3 X 10e4 for 2609 a GOE spectrum.Thus evenif we observeevery state in an interval, for a GOE spectrum we have only a 0.03% chanceof three levels being in an energy interval of 0.3/p, while for a Poisson spectrum, there is a 5.5% chance of ‘such an occurrence.In the propyne spectrum, we observe clumps of three or more lines separatedby 0.3/p three times out of 39 upper statesobserved(7.7%). This number is clearly an underestimateof the true number of close lying triplets, sincesomesometriplets are lost due to m issing lines in the spectrum. If anything, our observations show even more level clustering than predicted by a Poisson distribution, and are clearly inconsistentwith a GOEtype spectrum. This extra clumping in the spectrum may reflect the high level of degeneracyexpectedin a symmetric top molecule at the harmonic lim it, which may survive even at the eigenstatelevel as enhancedfluctuations in the eigenstatedensity. VII. CONCLiJSlONS We have presenteda detailed analysisof the high resolution spectrum of the first overtone of the acetylenic C-H stretch of propyne. This spectrum exhibits fairly extensive near-resonantperturbations that are fully resolved. When such a spectrum is assignedit provides the opportunity to quantitatively study the vibrational energy redistribution process.Here we have presentedan analysis of the near-resonantperturbations which includes the instrumental effects of lim ited signal to noise. The remarkable feature of the spectrum is the very unlikely observation of the lack of perturbations for all K=O states.Furthermore, there is a strong K dependence of the averagematrix elementsfor the near-resonantcouplings. These observationsstrongly suggeststhat there is an additional, nonresonantstate that plays a fundamental role in the vibrational coupling, acting as a doorway for vibrational energy flow from the acetylenic C-H stretch to the near-resonantbath states.The observedK dependence of the coupling indicates that z-axis Coriolis interactions dominate the doorway state-bath state coupling. This behavior has not beenobservedin the spectrum of any of the other acetylenic compoundsstudied to date. The spectrum provides an interesting link betweenthe nonresonantperturbations and the local, resonant perturbations. It appearsto be generally the casefor symmetric top moleculesthat the K subbandordering is very sensitive to the long-range perturbations in the spectrum. The resonant interactions can be analyzed by studying each individual subband. In the case of propyne 2~~ we find that these two types of interactions are intimately connected. The coupling strength to the near-degeneratebath statesis dependenton the coupling strength of the acetylenic C-H stretch to a nonresonantstate. Presumablythis interaction is of relatively low order and thus stronger. This suggests that the matrix elementsto the near-resonantbath states could be calculated perturbatively through the lowest order interaction pathway. Such an approach was recently shown to be extremely successfulin the calculation of high order, local interactions in HCCF.52 Downloaded 18 Mar 2002 to 128.112.83.42. subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp 1994 J. Redistribution Chem. Phys., Vol. 100, No. 4, 15 February 2610 Mcllroy et ak The propyne 2v, band The above scenario carries the implication that for larger molecules, a tier model may be the appropriate physical picture of vibrational energy redistribution.53,54 The propyne overtone spectrum presentedhere would be a casewhere the first, most strongly coupled tier is so sparse that a single state (or very small number of states) controls the relaxation. Unfortunately, due to the limited information availabIe on the nonresonantinteraction (limited to the subbandorigins of the K=O-3 subbandsand possibly the slightly large value of czBfor the K=O subband) we cannot learn much about the identity of the hypothesized doorway state directly from the spectrum. If a tier model is appropriate there is still the the task of identifying the important states and their coupling strengths. It iS possible that the roughly similar behavior of all hydrocarbon acetyleneshas its origin in a common bath structure resuIting from the small differencesin the normal mode frequencies of these molecules.29~55 High resolution studies of other molecules undergoing IVR will provide valuable data for testing a tier model hypothesis. Recent work by Stuchebrnkhov and Marcus56 on the series of molecules (CX3),YCCH (X=H,D; Y-C,%) (Ref. 30) has shown that a tier model, including only cubic and quartic anharmanic coupling constants,is capableof a quantitative prediction of the IVR relaxation rate of both the fundamental and first overtone acetylenic stretching bands. This is exciting and suggestthat for propyne, for which the potential energysurface is much better known,39the exact nature of the important doorway statesmay be predictable from current theory. Lastly we have presentedevidenceof Poissonstatistics of the energy levels, which is taken to imply that the underlying classical dynamics of this system are regular at this energylevel. This conclusion is reachedon the basis of two features of the spectrum. First of all, a wide dynamic range of matrix elementsis found. Under the assumption of Gaussian distributed matrix elements, as might be expectedfor a chaotic system, it would be extremely unlikely for matrix elementsmany times the mean to be observed. Secondly, there are several observations of “clumps” of bath states. This clustering of bath state levels is found ‘to be more consistent with a Poisson distribution of bath states than with a GOE distribution of bath states. The observation of regular dynamics is in contrast to the observation of chaotic behavior found previously for the 2~~ spectrum CF3CCH.19 ACKNOWLEDGMENTS The authors would like to thank David Perry and Alexi Stuchebrukhovfor making their results availablebefore publication and for many useful discussions. This work was supported by the National ScienceFoundation. ‘(a) J. Jortner and R. D. Levine, Adv. Chem. Phys. 47, Part 1, 1 (1981); (b) J. D. McDonald, Annu. Rev. Phys. Chem. 30,29, (1979); (c) V. E. Bondybey, ibid. 35, 591 (1984); (d) E. Weitz and G. Flynn, Adv. Chem. Phys. 47, Part 2, 185 (1981); (e) C. S. Parmenter, J. Phys. Chem. 86, 1735 (1982); (f) C S. Parmenter, Faraday Discuss. Chem. Sot. 75,7 (1983); (g) R. E. Smalley, Annu. Rev. Phys. Chem. 34, 129 (1983); (h) S. A. Rice, Adv. Chem. Phys. 47, Part 1, 117 (1981); (i) R. A. Marcus, Faraday Discuss. Chem. Sot. 75, 103 (1983); (j) T. Uzer and W. H. Miller, Phys. Rep. 199, 73 (199 1). ‘K. V. Reddy, D. F. Heller, and M. J. Berry, J. Chem. Phys. 76, 2814 (1982). 3J. S. Wong and C. B. Moore, J. Chem. Phys. 77, 603 (1982). ‘H. L. Fang and R. L. Swofford, Chem. Phys. Lett. 105, 5 (1984). ‘5. M. Jasinski, Chem. Phys. Lett. 109, 462 (1984). 6J. E. Baggott, M.-C. Chuang, R. N. Zare, H. R. Dubal, and M. Quack, J. Chem. Phys. 82, 1186 (1985). ‘C. Manzanares, 1. N. L. S. Yamasaki, and E. Weitz, J. Chem. Phys. 90, 3953 (1986). *H. R. Dubal and M. Quack, Chem. Phys. Lett. 72,342 (1980); 80,439 (1981); J. Chem. Phys. 81, 3779 (1984); H. R. Dubal, T. K. Ha, M. Lewerenz, and M. Quack, ibid. 91, 6698 (1989). 9D. Papousek and M. R. Aliev, Molecular Vibrational-Rotational Spectroscopy (Elsevier, New York, 1982), Chap. III. 16. lo A. Sinha, M. C. Hsiao, and F. F. Crim, J. Chem. Phys. 94‘4928 (1991). “M. J. Bronikowski, W. R. Simpson, B. Girard, and R. N. Zare, J. Chem. Phys. 95, 8647 (1991). “M. Quack, Annu. Rev. Phys. Chem. 41, 839 (1990). “K. K. Lehmann, G. J. Scherer, and W. Klemperer, J. Chem. Phys. 77, 2853 (1982). 14G. J. Scherer, K. K. Lehmann, and W. Klemperer, J. Chem. Phys. 78, 2817 (1983). “K. K. Lehmann and S. L. Coy, J. Chem. Sot. Faraday Trans. 2 84,1389 (1988). 16J K. Holland, D. A. Newham, and I. M. Mills, Mol. Phys. 70, 319 &990). “D. M. Jonas, X. Yang, and A. M. Wodtke, J. Chem. Phys. 97, 2284 (1992). I’D. M. Jonas, S. A. B. Solina, B. Rajaram, R. J. Silbey, R. W. Field, K. Yamanouchi, and S. Tsuchiya, J. Chea Phys. 97, 2813 (1992). 19B. H. Pate, K. K. Lehmann, and G. Stoles, J. Chem. Phys. 95, 3891 (1991). “E. 3. Heller, Faraday Discuss. Chem. Sot. 75, 141 (1983). *‘H. L. Kim, T. J. Kulp, and J. D. McDonald, J. Chem. Phys. 87,4376 (1987). uC Kittrell, E. Abramson, J. L. Kinsey, S. A. McDonald, D. E. Reisner, R. W. Field, and D. H. Katayama, J. Chem. Phys. 75,2056 (1981); D. E. Reisner, P. H. Vaccaro, C. Kittrell, R W. Field, J. L. Kinsey, and H.-L. Dai, ibid. 77,573 (1982); E. Abramson, R. W. Field, D. Imre, K. K. Innes, and J. L. Kinsey, ibid. 80, 22 948 (19.84). uX. Yang, C. A. Rogaski, and A. M. Wodtke, J. Opt. Sot. Am. B 7, 1835 (1990). “Q. Zhang, S. A. Handel, T. A. W. Wasserman, and P. H. Vaccaro, J. Chem. Phys. 96, 1640 (1992). =Y. Chen, L. Hunziker, P. Ludowise, and M. Morgen, J. Chem. Phys. 97, 2149 (1992). 26K. K. Lehmann and S. L. Coy, J. Chem. Phys. 83, 3290 (1985); S. L. Coy and K. K. Lehmann, ibid. 84, 5239 (1986). *‘P. R. Fleming and T. R. Rizzo, J. Chem Phys. 95,865 (1991); 95, 1461 (1991). 28A. M. de Souza, D. Kaur, and D. S. Perry, J. Chem. Phys. 88, 4569 (1988); J. Go, G. A. Bethardy, and D. S. Perry, J. Phys. Chem. 94, 6153 (1990). =A. McIlroy and D. J. Nesbitt, J. Chem. Phys. 91, 104 (1989); 92, 2229 (1990). 30E R Th Kerstel, K. K. Lehmann, T. F. Mentel, B. H. Pate, and G. Scolds, J.‘Phys. Chem. 95, 8282 (1991). “K K Lehmann, 9. H. Pate, and G. Stoles, Laser Chem. 11, 237 (1991.). 32J. Go and D. S. Perry, J. Chem. Phys. 97, 6694 (1992). 33J. Go, T. J. Croain, and D. S. Perry, Chem. Phys. 175, 127 (1993). 34J. Gambogi, E. R. Th. Kerstel. K. K. Lehmann, B. H. Pate, and G. Stoles (in progress). ‘5J . E . Gambogi, J. H. Timmermans, K. K. Lehmann, and G. Stoles, J. Chem. Phys. 99,9314 (1993); J. E. Gambogi, E. R. Th. Kerstel, K. K. Lehmann, and G. Stoles, ibid. 100, 2612 (1994). “R . R . Hall, Ph.D. thesis, Rice University, 1984, University Microfilms International, 8416524. 37L. C. Baylor, E. Weitz, and P. Hofmann, J. Chem. Phys. 90, 615 (1989). 38T E. Gough, R. E. Miller, and G. Stoles, Appl. Phys. Lett. 30, 338 (i977). J. Chem. Phys., Vol. 100, No. 4, 15 February 1994 Downloaded 18 Mar 2002 to 128.112.83.42. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp Mcllroy et al.: The propyne 2v, band “E R. Th. Kerstel, K. K. Lehmann, B. H. Pate, and G. Stoles, J. Chem. Phys. 100, 2588 (1994). 40A. Baldacci, S. Ghersetti, and K. N. Rao, J. Mol. Spectrosc. 68, 183 (1977). “V. M. Horneman, G. Graner, H. Fakour, and G. Tarrago, J. Mol. Spectrosc. 137, 1 (1989). “G. Herzberg, Molecular Spectra and Structure II. Infrared and Raman Spectra ofPobatomic Molecules (Krieger, Malabar, 1945), Chap. IV. 43R. Antilla, S. Jaakkonen, and T. Sahlstrom, Spectrochim. Acta 28A, 1615 (1972). &M. J. H. Kemper, J. M. F. van Dijk, and H. M. Buck, Chem. Phys. L&t. 53, 121 (1978). 45K. K. Lehmann and S. L. Coy, Ber. Bunsenges. Phys. Chem. 92, 306 (1988). 46W. D. Lawrance and A. E. W. Knight, J. Phys. Chem. 89,917 (1985). “The Lawranc+Knight deconvolution method assumes a single bright state. However, many polyatomics, specifically molecules with -CH, and -NH, internal rotors, exist in multiple nuclear spin modifications. Frequently the barrier to internal rotation is sufficiently high that the nuclear spin splitting is undetectable at the 1 MHz level of resolution. 2611 In these cases, the spectrum consists of two (or more) overlapped spectra. Deconvolution cannot be used until the nuclear spin labels for each eigenstate is made. 48D. Papousek and M. R. Aliev, Molecular Vibrational-Rotational Spectroscopy (Elsevier, New York, 1982), Chap. 111.17. 4qT. A. Brody, J. Flares, J. B. French, P. A. Mello, A. Pandey, and S. S. M. Wong, Rev. Mod. Phys. 53, 385 (1981). “C. E. Porter, Statistical Theories of Spectra: Fluctuations (Academic, New York, 1965). “P. B. Kahn, Nucl. Phys. 41, 159 (1963). 52K. K. Lehmann, J. Chem. Phys. 96, 1636 (1992). 53E. L. Sibert III, W. P. Reinhardt, and J. T. Hynes, J. Chem. Phys. 81, 1115 (1984). 54A. A. Stuchebrukov, M. 2. Kuzmin, V. N. Bagratashvili, and V. S. Letokhov, Chem. Phys. 107, 429 ( 1986). 55K K. Lehmann, B. H. Pate, and G. Stoles, J. Chem. Phys. 93, 2152 (1990); B. H. Pate, Ph.D. thesis, Princeton University, 1992. 56A. A. Stuchebrukov and R. M. Marcus, J. Chem. Phys. 98, 6044 ( 1993). Downloaded 18 Mar 2002 to 128.112.83.42. subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp J. Redistribution Chem. Phys., Vol. 100, No. 4, 15 February 1994