JOURNAL

OF MOLECULAR

SPECTROSCOPY

143,6 I-80 ( 1990)

Calculated Rotational and Rovibrational Spectra of D2S and HDS

STEVEN MILLER AND JONATHAN TENNYSON

Department of Physics and Astronomy, University College London, Gower St., London WCIE 6BT. United Kingdom PAVEL ROSMUS AND JORG SENEKOWITSCH Fachbereich Chemie. Universitat Frankfurt, D-6000 Frankfurt, Federal Republic of Germany AND IAN

M. MILLS

Department of Chemistry, University of Reading, Whiteknights, Reading RG6 ZAD, United Kingdom RovibraGonaI energy levels, transition frequencies, and Iinestrengths are computed variationally for the sulfur hydrides D2S and HDS, using ab initio potential energy and dipole surfaces. Wavenumbers for the pure rotational transitions agree to within 0.2 cm- of the experimental lines. For the fundamental vibrational transitions, the band origins for D2S are 860.4, 1900.6, and 19 12.0 cm _ for v2, Y, , and Q, respectively, compared with the corresponding experimental values of 855.4, 1896.4, and 1910.2 cm-. For HDS, we compute v2to be 1039.4 cm-, compared with the experimental value of 1032.7 cm-. The relative merits of local and normal mode descriptions for the overtone stretching band origins are discussed. Our results confirm the local modenature of the HIS. DzS, and HDS system. o 1990 Academic press, IX.

I. INTRODUCTION

In recent ab initio studies we have generated potential energy and dipole surfaces for Hz S (1) which yielded fundamental vibrational band origins within 8 cm- of those measured experimentally (2). The 20 overtone and combination bands measured up to 10 000 cm- all agreed with the calculated values to within 20 cm-. The calculated equilibrium bond length and angle differed from the experimental values (3) by less than 0.2% resulting in the leading rotational constants, A, B, and C, being reproduced to within 2%. The dipole surface, however, did not give quite such satisfactory results. A comparison of the experimental v2 bending fundamental spectrum measured by Mills (4) and that computed by Senekowitsch et al. (I) showed good agreement, after correction for the difference in band origins was made. But the dipole surface is extremely flat with respect to the stretching coordinates ( vl, symmetric stretch, and v3, asymmetric stretch) and the intensities of these modes could not be so well reproduced. In the last 4 years, there have been a number (5, 6) of experimental studies of the pure rotational and rovibrational spectra of the deuterated isotopomer D2 S. The pure
61 0022-2852190 $3.00
Copyright 0 1990 by Academic Press, Inc. All rights of reproduction in any form reserwd.

62

MILLER

ET AL.

rotational spectrum of the mixed species HDS has also been measured (5). Accurate intensity measurements accompanied the pure rotational spectra. Recently a limited study of the u2 band of HDS has been reported ( 7). There is, however, far less experimental data on the deuterated species than for H2S itself. A theoretical study of the deuterated isotopomers is useful, therefore, not only to see how well the theoretical surfaces reproduce the experimental data but also to make predictions for those bands not already investigated in the laboratory.
2. COMPUTATIONAL DETAILS

2.1. Rovibrational Energy Levels and Wavejiinctions The potential energy and dipole surfaces are fits to 50 values calculated for nearequilibrium geometries using 156 Gaussian-type orbitals using the coupled electron pair approximation (CEPA) . Further details of the surfaces may be found in Ref. ( 1) . Perturbation theory was used in the analysis of the potential energy surface ( 1) to derive spectroscopic constants. Fully coupled, variational rovibrational energy levels, transition wavenumbers, and linestrengths were computed using the TRIATOM program suite. These programs have recently been upgraded (8) for any general combination of two radial coordinates and the included angle, independent of the coordinate system in which the potential energy surface may be expressed. For these calculations, Radau coordinates (q, , q2, 6) ( 9) were used to represent the molecular structure even though the potential is written in bond-length, bond-angle coordinates. Radau coordinates are shown in Fig. 1. q, is the distance between atom A2 and the point R; q2 is the distance between atom A, and the point P. Then g, and g2 are the ratios of distances defined by P - A3 g = A2 _ A3 and R - A3 g2 = A, _ A3 . In the case that the molecule has C2, symmetry, g, = g2. For Radau coordinates,
g,=

ff++-CY,p

(2)

and
CY

g2=1 -p+(YpJ
where
a=

m3

ml+m2+m3

I/2

(3)

SPECTRA OF D2S AND HDS

63

A3

A2

FIG. 1. Coordinate system used in the TRIATOM program suite.

and
p=

m2
ml + m2

and ml, etc., are the masses of the respective atoms. For D2S and HDS, m3 is much larger than the other two atomic masses. For D2S, g, = g2 = 0.0297, and for HDS, with H at atom A,, g, = 0.0299 and g2 = 0.0152. Thus Radau coordinates differ very little from the bond-length, bond-angle coordinates in which the potential is written and the two sets of coordinates are related by a simple transformation. However, the Radau coordinates are orthogonal (IO), a fact which leads to considerable computational efficiency when computing rotationally excited states. Basis functions for the calculations consisted of products of Morse oscillator-type functions in q1 and q2 and Legendre polynomials in cos( 6). For the J = 0 band origins, a basis set composed of 1600 functions chosen on energy ordering (8) was used. For H2S and D2S, we also used the option to symmetrize the ql, ql basis functions. This splits the calculation into two sets of basis functions, one with a symmetric combination, the other with an antisymmetric combination: thus +,,,Jqi,
q2) = (1)2[4h(4iMn(92) + (-1P#bdchhMq2)1~ m a n + t

(4)

where m , n refer to the order of the Morse oscillator-like functions used in the radial basis set (II) and t = 0 for the symmetric combination and t = 1 for the asymmetric combination. Both m and n were allowed to take values up to 23. The angular coordinate 0 = (AIoA2) was represented by Legendre polynomial functions up to order 80. For all three molecules, all the band origins reported here were converged to 1 cm- or better. The fundamentals were converged to 0.01 cm-. Optimized parameters for the radial basis functions are given in Table I. The rotationally excited J > 0 calculations were performed by first computing the lowest 400 non-Coriolis coupled functions for each value of k from 0 to J as an

64

MILLER ET AL. TABLE I Optimized Parameters Used in Radial Basis Functions Molecule Co-ordinate
(IHS

&./au. 2.756 2.6tm 2.586 2.610

w,/a.u. 0.010 0.609 0.008 0.012

D.fa.u.
0.100 0.106 0.100 0.166

HzS DzS
HDS

qDS
rlars
qDS

intermediate basis set. The final basis was then composed of (J + 1) X 150 functions chosen from the intermediate basis on energy ordering (8). (N.B.: Eigenfunctions of odd Wang, time-reversal, parity require basis functions with k from - 1 to --.I which, in our basis, are identical to within a phase factor to the functions with k positive.) This gave rotationally excited levels for which the fully coupled rotation-vibration part of the calculation was converged to 0.001 cm- for the fundamental manifolds for J = l-l 5, for all values of K, and Kc. The energy levels reported here are an accurate representation of the Senekowitsch et al. surface. As well as studying the deuterated isotopomers of H2S, this work is also, therefore, a useful test of the efficiency of Radau coordinates for calculating fairly highly excited rotational levels. For future studies of dihydride systems and their deuterated isotopomers, in which even more highly excited rotational levels will be required (as in astrophysical applications), we consider that the use of Radau coordinates will prove extremely useful for the production of accurate, converged eigenenergies and wavefunctions. The Sutcliffe-Tennyson Hamiltonian (10) currently employed has to embed the molecular-fixed z axis along one or the other of the radial coordinates. When using either Radau or bond-length, bond-angle coordinates, therefore, it is not possible to keep the Cl symmetry of an A& molecule such as D2S for calculations with J > 0. Instead, wavefunctions corresponding to ortho and para spin states are computed together. 2.2. Transition Moments and Linestrengths The rovibrational wavefunctions generated in the above calculations were used to compute dipole transition moments and linestrengths (9, 12) for D2S and HDS. Care was taken to ensure that nuclear spin statistics were properly included in the calculations of intensities.
During the course of the calculations on HDS, an error in the TRIATOM modules TRIATOM and SELECT, as published in Ref. (9), was detected. p of Eq. (3) had been programmed as 1 - 0. This error has been corrected, and updated copies of these programs may be obtained from the authors on request. Please note, this error does not affect calculations on A& molecules.

SPECTRA OF D2S AND HDS 3. BAND ORIGINS

65

3. I. Spectroscopic Constants The spectroscopic constants calculated by perturbation theory are presented in Tables II and III for D2S and HDS. D2S is an oblate asymmetric rotor with a value of K = 0.350 compared with 0.504 for HIS. HDS is a prolate asymmetric rotor with K = -0.489. In Table IV, we compare the fundamental band origins and leading rotational constants in both species with those measured by Camy-Peyret et al. (5-7). Agreement with experiment is similar to that obtained for H2S. In Table IV we have arranged the band origins and integrated intensities for the

TABLE II Vibrational Constants of D2S and HDS (Perturbation Theory Calculations) DZS
w1 /

HDS

cm-

1956.9 885.3 1968.3

1962.7 1073.3 2734.0

w2 /

cm-

wj / cm-

zll

cm-

-14.5 - 3.2 -14.6 -10.4 -57.0 -12.0

-28.6 - 4.8 -55.5 -11.7 - 1.3 -21.8 2854.1 0.4 0.57763 -0.01286 0.81619

~22 / cm- 233 / cm- ~12 / 0-l ~13 / cm- ~2s / cm- G(OOJ) / cm-l

2337.3 -28.3 0.00000 0.03519 -0.99938

7 / cm-

&

st
Sk

cxf cm- 1
L-X$ cm- /
czf / cm- a; c$ af of at / cm-

0.058 -0.145 0.089 0.052 -0.055 0.030 0.030 0.024 0.018

0.011 -0.268 0.305 0.105 -0.081 0.001 0.042 0.035 0.031

CY~/ cm- / cm- / cm- / 0-l / cm-

Note that y = fk1,x3, where k, ,33 is the Darling-Denison Ref. (IS). bCoriolis constant.

constant defined by Mills and Robiette in

66

MILLER

ET AL.

TABLE III

Rotational Constants for DzS and HDS (Perturbation Theory Calculations)

DzS R,SD / A $SD;H RSA / degrees .i / A, / cm- B, / cm- C, / cm- n 10~ r-~ 10*x rcccc 10*X 7~88 10zX rg~cc / cm- / cm-l / cm- / cm- 5.4982 4.5178 2.4800 0.350 -0.20806 -0.11129 -0.00640 0.10044 -0.01310 -0.01207 0.26246 -0.43819 0.19174 -0.06836 -0.06049 0.39919 -1.25855 0.87537 1.41881 9.7700 4.9388 3.2805 -0.489 -0.23901 -0.05268 -0.01230 0.01002 -0.02211 -0.02252 0.04117 1.13122 -0.57488 -0.02002 0.02524 0.08122 0.89095 -0.37465 0.54663 1.3355 92.227 HDS 1.3355 92.227 1.3355

10*x ~BBBB / cm-l

10*x ~CCM / cm- 103x DJ / cm- 103x DJK / cm- 103x DK / cm- 103x R6 / cm- 103x
6J /cm-'

103x AJ / cm- 103x AJK / cm- 103x AK / cm- 103x 6K / cm-

lowest 60 vibrational modes for HzS, DzS, and HDS according to band type. Assignments were made on the basis of wavenumbers and transition moments between the bands in question. In this section we will comment only on the wavenumbers of the band origins. Expected mass effects reduce the wavenumbers of the bending only modes from H2S to HDS to D2S. We have grouped modes involving the stretches according to the total number of vI plus v3 quanta, n,. For n, = 1, we have bands with v2, the number of v2 bending quanta, ranging from 0 to 7. 3.2. Local versus Normal Mode Description Dihydride systems have been the focus of considerable attention (14, 15) with the respect to the appropriate description of the overtone stretching modes. Child and Halonen, in particular, have drawn attention to the fact that H2S and H2Se are very close to the local mode limit (14).

SPECTRA OF D2S AND HDS TABLE IV Comparison of Computed and Experimental Constants of D2S and HDS (Perturbation Theory Calculations) DzS
Computed Experimental Computed HDS Experimental

67

2377.3 5.4961 4.5210 2.4790 5.4911 4.5126 2.4418

2854.1 9.7699 4.9390 3.2803 9.7518 4.9321 3.2257

1894.1 5.4110 4.4391 2.4345

1896.4 5.4363 4.4647 2.4165

1904.6 9.7533 4.7804 3.2181

867.6 5.7154 4.5996 2.4445

855.4 5.6250 4.5866 2.4214

1038.8 10.1722 5.0602 3.2274

1032.7 10.0271 5.0246 3.1913

lw4.7 5.3645 4.4728 2.4527

1910.2 5.4055 4.4867 2.4266

2611.4 9.3119 4.9360 3.2342

DzS and HDS ground state constants from Ref. 5. DIS vl, 2 and 3 constants from Ref. 6. HDS y constants from Ref. 7.

In Table V we have given band assignments in both local and normal mode notation, and it is useful to analyze the computed eigenenergies and wavefunctions to ascertain which description is more appropriate for our calculations. For these A& systems we are suggesting that the labeling convention for local modes should place the local mode quantum numbers and symmetry ( + for Ai and - for BZ, in the (2~~ point group) in square brackets and the bend quantum number in parentheses, thus [ ni , n2, k ] ( v2). In addition, we have also changed the normal mode ordering of the levels of H2S from that given in Ref. (I), which was in error in several places, so that it now accords with the more usual ordering. The Morse oscillator-like basis functions used in the TRIATOM suite to represent the radial coordinate dynamics are essentially local mode functions. Normal modes should then be produced, in so far as they are relevant, as the appropriate combinations of these basis functions.

68

MILLER ET AL. TABLE V

Vibrational Term Values, G, and Vibrational Band Intensities from the Ground State for H2S and lsotopomers (The Intensities Were Calculated at 300 K, with Powers of 10 in Brackets; Variational Calculations)
LOG31
___ mode Normal mode ___ kl,%W) G (cm-l) (atm- Sb cm-z ) G (cm-) 860.4 1716.6 2568.0 3414.1 4254.7 5089.2 5917.4 6738.8 7553.2 (z&m- % cm-* ) E (cm-) 1039.2 2071.9 3097.3 4114.6 5123.0 6122.0 7110.7 8088.7 9055.9 (atm- Sb cm- ) w Da6 HDS

[~lr~Zl(W)

IWI (1) WI (2)

P?Ol 13) P#Ol (4) lOPI (5) IO,] (6) [OYOI(7) IO>01 (8) lOLI (9)

@L~1 Pm)
(0,3P) (%4P) P,5,0) (A6,O) (0,7,0) (OS>01 (0,QP)

1190.4 2372.0 3543.5 4703.7 5851.6 6986.3 8106.9 9212.5 10302.1

2.65(+0) 3.23(-l) 4.18(-2) 2.46(-3) 1.28(-4) 2.67(-6) 3.60(-E) 7.32(-10) 9.84(-12)

1.60(+0) 1.14(+1) l.oQ(-2) 3.55(-4) 1.85(-5) 3.61(-7) 8.84(-Q) 2.30(-10) 7.47(-12)

2.17(+0) 1.77(-l) 5.73(-Z) 9.716-4) 8.09(-6) 3.10(-7) 5.40(-Q) X71(-10) 3.81(-11)

11,0+1 (0) iw0-l (0) I2,0+1 (0) I2Pl PJI (0) (0)

(l>fAO) (OAl) Gvw (1>cv) KVA (3,OP) (2,W) (1,0,2) (OA3) (4,0,0) (3,W) (2,0,2) (1PP) (0,0>4)

2620.4 2631.0 5154.2 5155.5 5251.2 7589.3 7589.4 7768.4 7789.3 9929.0 9929.0 10208.3 10212.0 10308.8

1.40(+0) 1.54(+0) 3.60(-l) 6.12(-l) 3.65(-2) l.Ol(-2) 3.01(-2) 5.76(-3) 5.55(-3) 8.10(-4) 4.21(-4) 2.82(-4) S.oO(-5) 1.69(-5)

1900.6 1912.0 3760.4 3763.2 3814.4 5569.2 5569.6 5658.1 5679.7 7331.4 7331.40 7468.5 7476.1 7529.6

6.68(-l) 6.73(-l) 1.18(-l) 2.69(-l) 3.86(-3) 1.91(-3) 9.98(-3) l.Ol(-3) 1.41(-3) 1.32(-4) 1.38(-4) 4.17(-5) 8.12(-5) 1.02(-5)

1905.9 2625.6 3762.2 4530.2 5155.3 7058.8 6384.9 5569.2 7591.6 8912.5 8189.7 7326.7 9562.7 lco10.9

6.70(-l) 1.56(+0) 2.00(-l) 3.45(-2) 4.73(-l) 2.08(-3) 1.65(-3) 5.89(-3) 2.44(-2) 7.95(-7) 8.11(-5) 1.85(-4) 7.95(-S) 9.15(-7)

13>0+1(0) [3Pl (0)

12,1+1 (0) 12J-I (0) [4,0+1 (0) [4@1 (0) 13r1+1 (0) I3J-I (0) i2>21(0)

W+l WI WI ILO+ w-1

0) (1) (2) (2) (3)

0AO) (W,l) WJ) (1,2*0) (%3,1) (1,30) @,4,1) (1,4,0) (0,5J) (1,5,0) (OPJ) k6P) (0,7,1) U,7sJ)

3794.6 3799.8 4959.9 4960.1 6110.2 6115.6 7249.4 7259.6 8376.5 8391.6 9490.6 9510.2 10590.9 10614.6

1.77(+0) 4.04(+0) 5.48(-2) 6.39(-3) 2.30(-3) 2.07(-3) 3.92(-5) 5.26(-5) 2.69(-7) 1.13(-5) 4.65(-10) 2.64(-S) 3.65(-11) 1.00(-S)

2752.5 2761.3 3606.4 3600.2 4446.8 4443.3 5282.0 5281.2 6111.7 6113.4 6935.5 6939.6 7752.9 7759.4

6.88(-l) 1.55(+0) 1.56(-4) 7.67(-4) 4.19(-4) 3.50(-4) 3.58(-6) 4.88(-6) 9.79(-X) l.Ol(-6) 2.46(-10) 1.98(-8) 5.35(-U) S&0(-10)

2934.8 3645.6 4659.5 3957.2 5666.2 4972.4 6665.0 5979.5 7655.2 6977.7 8636.1 7966.4 9607.1 8944.9

1.33(+0) 2.44(+0) 2.12(-2) 1.61(-2) 1.41(-3) 1.13(-3) 3.76(-5) 2.66(-5) 4.27(-7) 2.64(-7) 7.14(-10) 8.23(-Q) 2.00(-S) 3.76(-10)

[1,0+1 (3) w1 W+l IW W+l (4) (4) (5) (5)

Penal (6) [LO+] WI (6) (7)

[Lo+1 (7)
L2,0+1 (1) [2c-I (1) Pll l2PI o (1) (2)

(2JP) (l,l,l) PA2) P,2>1)

6307.7 6307.7 6403.0 7451.5 are only accurate

4.65(-2) 3.09(-l) 3.06(-5) 4.72(-3) to km-.

4602.1 4603.9 4654.2 5440.5

1.38(-2) 8.32(-2) 1.95(-5) 8.29(-4)

4780.9 5539.6 6156.2 6542.8

5.97(-2) 1.14(-3) 1.33(-l) 5.22(-4)

These

band origins

SPECTRA

OF D2S AND

HDS

69

TABLE V-Continued
LOCal mode h,~11(9) NOrmal mode h,Y,YS) E (cm-l) 7452.1 7546.9 8585.3 8585.9 8681.5 9707.9 9708.0 9805.6 10817.4 10818.2 10917.9 8723.0 8723.1 8906.9 8917.4 9847.8 9848.0 10037.4 10037.4 Sb (atm- UT-* ) 1.26(-4) 4.71(-S) 2.71(-4) 6.86(-S) 1.20(-S) 1.03(-S) 1.05(-7) 1.09(-6) 2.80(-7) 5.28(-7) 3.39(-7) 1.34(-2) 1.85(-3) 1.44(-S) 1.47(-S) 3.77(-4) 1.53(-S) 6.46(-7) 9.76(-7) G (cm-)
5439.6

DzS

HDS

& (atm- cm- ) 5.06(-S) 4.26(-6) 1.9q-5) 5.77(-6) 1.25(-6) 5.21(-7) 6.64(-8) 5.48(-8) 5.83(-8) 1.21(-10) 1.30(-8) 1.49(-3) 6.65(-4) 6.29(-6) 1.09(-7) 2.50(-S) 5.85(-S) 4.66(-7) 1.12(-7)

E (cm-l)
5793.2

& (atm- cm- ) 3.08(-4) 1.25(-3) 2.18(-S) 4.24(-5) 4.36(-S) 5.35(-7) 1.15(-6) 6.65(-7) 5.75(-9) 6.14(-9) 6.28(-7) 3.36(-5) 2.47(-5) 1.45(-Z) 6.13(-3) 1.28(-6) 3.65(-6) 1.81(-5) 3.14(-7)

[2,0+1(2) IlJl (2)

WP) W,2)
(1,331) (2,3,0) (0,3,2) (1,4,1) (2,4,0) (0,4,2) (2,5,0) (1,5,1) (0,522) (2Al) (3AO) (lA2) (OA3) (2,2,1) (3,2,0) (1,2,2) (0,2,3)

5490.1 6272.4 6272.2 6321.7 7099.1 7099.3 7148.4 7920.6 7920.3 7969.9 6400.7 6400.6 6491.0 6507.8 7228.8 7227.8 7320.2 7331.4

7151.1 7539.0 6798.3 8139.2 8527.4 7795.4 9119.9 8783.7 9507.0 10093.2 7384.0 8049.0 6578.0 8578.1 8376.9 9033.2 7580.3 9562.1

w-1

(3)

[2,0+1(3) PA (3) 12,&l(4)

PA3+1 (4)
Ml (4)

PJJ+l (5) w-1 (5)

PJI (5) [3,0-l(1) [3,0+1(1) [2,1+1(1) [2,1-l(1) [3C-1(2) [3,0+1(2) [2,1+1(2) l2J-1(2)

In Fig. 2 we have plotted the wavefunctions for HDS for the n, = 4 vibrational levels as a function of the radial coordinates (where n, is the total number of stretchsymmetric plus asymmetric-quanta in the mode), with the angle 0 frozen at its ground state equilibrium value. A node in the wavefunction occurs where the amplitude changes sign and corresponds to one quantum of excitation. Nodes at right angles to one or the other of the radial coordinates represent excitations localized in that coordinate. These represent local modes. Nodes at right angles to the line bisecting the angle between the two axes (i.e., at 45 clockwise to q1) represent even linear combinations of the two local modes, and nodes parallel to this line, odd combinations. The plots shown in Fig. 2 all have nodes perpendicular to the radial axes, indicating that the mode involves only excitation of one or the other of the two radial coordinates, i.e., that it is local. This is as expected from the large frequency difference between v1 and u3 that results from the difference in mass between hydrogen and deuterium. Analysis of the vibrational wavefunctions for H2S and DzS are more difficult. In both these cases, it was appropriate to use the radial basis symmetrization described in Section 2. Wavefunctions for the n, = 4 manifold of D2S are shown in Fig. 3. The use of nodal structure to determine local mode versus normal mode character is not straightforward, since for both descriptions nodes may occur at right angles or parallel to the line bisecting the angle between the radial coordinates. In both descriptions, nodes at right angles to the 45 line represent symmetric combinations-in

4 a 3.5

b
,/ ....

1. 1

IS

115 q

3:5

5.5

1:s

3:5

H!31a 0
4 C d 3.5

3.5

3.

1.5

1.5 I.5

1.5

3.5

I. 5 .=I

215

3:5

et

.I+

1.5

3.5

FIG. 2. Plots of the wavefunctions for the n, = 4 manifold for HDS. Contours are given for 64, 32, 16, and 8% of the maximum amplitude of the wavefunction. The solid line countours enclose regions where the wavefunction has positive amplitude, and the broken lines, negative. The outer contour is the classical turning point for the relevant eigenenergy. Plots are in Radau coordinates with 0 frozen at its equilibrium value of 92. 70

SPECTRA OF D2S AND HDS

3.75 1

71
t

3.75
3.5 3.25 3. 2.75. 1.5. 2.25. 1.

I
1

2.25

2.5

2.75

3.25 3.5

3.75

1..

2.25 2.5

1.75

3.25 3.5

3.

qp*o

3.7
3.5 t

C
i

2.1,

1.5

2.15

3.25 3.5

3.

el

2.n

1.5

2.75

3.25 3.5

3.15

FIG. 3. Plots of the wavefunctions for the n, = 4 manifold for D2S. See the legend to Fig. 2 for details.

72

MILLER

ET AL.

local mode notation the [n, , n2, + ] combinations, in normal mode notation bands with v3 even. Nodes parallel to the 45 line are the corresponding asymmetric combinations, [n, , n2, -1 or bands with u3 odd. Thus the plots obtained do not distinguish between local and normal mode descriptions, but they do give information about the energy ordering of the final wavefunctions. To make sense of the resulting states, it is necessary to look at the energy structure of the various IZ,manifolds, and the coefficients of the radial basis functions comprising the individual eigenfunctions. Analysis of the energy levels alone indicates that both H2S and D2S exhibit almost ideal local mode behavior ( 1.5). For as low as n, = 2 the two lowest energy levels are almost degenerate and separated by a considerable energy difference from the highest level. In Table VI we report the result of decomposing the wavefunctions of the n, = 2 and 4 manifolds into their appropriate radial basis components, assuming that the Morse-like functions are identical to local modes. We also include the expected coefficients for the levels in the pure normal mode description. It is clear from the fact the expected radial basis functions do not account for the whole eigenfunction that the previous assumption is only partly valid. There are two reasons for this. First, the Radau coordinate does not lie along the S-H/D bond, and basis functions in this coordinate are clearly not identical with what is usually considered to be a local mode. The second reason is that states assigned as pure stretch modes do, in fact, contain a considerable amount of bend character. With these reservations, therefore, the eigenvector analysis broadly confirms the local mode character of the modes concerned. It certainly rules out a pure normal mode description of the states. The plots of the ~1,= 4 manifold wavefunctions are certainly consistent with this interpretation. In particular, the highest energy level in this manifold shows a nodal structure which is exactly what would be expected from the local [ 2, 21 designation. It is notable that the relation predicted by Mills and Robiette (15) for the local mode limit, namely xl1 = x33 = axl3 = &y (where y is the Darling-Dennison parameter, y = iki,33 in the notation of Mills and Robiette), is obeyed well by both H2S and D2S. The corresponding value of the Morse anharmonicity constant x for the SH diatomic moiety is 2x,, (or 2x33), i.e., about -29.0 cm- for D2S (see Table II). 3.3. Dependence of vJ/vI Splitting on VI Looking only at the homonuclear species, H2 S and D2 S, the levels with 1 quantum of vI , v2 quanta of v2, and 0 quanta of v3, denoted ( 1, v2, 0), are lower in frequency than the (0, v2, 1) levels only for v2 = 0, 1, and 2. For D2 S, the crossover point comes after v2 = 4. In fact, a plot of A~~;$ against v2, where
,%,,l
l,O~.O -

VO,tJ~,l -

vl,o*,o?
~2,

is, to a first approximation,

a linear function of

given by (6)

At:;;:, = A + B-v2.

SPECTRA OF D2S AND HDS TABLE VI Eigenvector Analysis for n, = 2 and 4 for H2S and D2S (Variational Calculations)
LOCd
mode basis W~V~~UIdX~ ( cd) Assignment Eigenvectors / Wavenumber ( cm-) Assignment Eigenvectors /

73

PUR normal mode

5154.2

P,0,+1
41 % 7%

3760.4

2,fJ+ Ll
5155.2

[2,0,+1 89 % 4%

(24
50 % 50 %

w,-1
41 %

3763.2

2,0,5251.2

[2,0,-l 94 % 11.11 5% 89 %

( y1+ 100%

us)

PA
16 % 30 %

3814.4

(2 YJ) 50 % 50 %

2,0,+ 171

9929.1 4,0,+ 3,1,+ 232

[4,0,+1 68 % 0% 0%

7331.4

[4,0,+1
41% 2% 0%

(4 h) 13 % 50 % 37 %

9929.1 4,0, 3,1,-

[4,0,-l 69 % 0%

7331.4

[4,0,-l 35 % 2%

(3 Yl + a) 50 % 50 %

10208.3 4,0,+ 3,1,+ 2,2 10212.0 4,0,3,1,-

[3,1,+1 2% 35 % 2% [3A-1 2% 40 %

7468.5

[3,1,+1 0% 67 % 0%

(2 w + 2w) 75 % 0% 25 % ( Qf 3w)

7476.1

[3A-I 0% 75 %

50 % 50 %

10308.3 4,0,+ 3,1,+ 222

I2A 0%
15 % 39 %

7529.6

[2,21 0% 8% 83 %

(4 US) 13 % 50 % 37 %

Percentages are the proportion of the Morse oscillator-like local mode basis functions occurring in the final eigenfunction. These do not add up to 100% for reasons discussed in text.

For H2S, A = 10.4 cm- and B = -5.2 cm-. For DzS, the plot is less well described by a straight line, but we get A = 11.8 cm- and B = -2.4 cm- (see Fig. 4). The slope of lines reported here may be explained using the approximate relationship

(v, +

JJ2v2)

(u2v2)

02

X2n,

(7)

74

MILLER

ET AL.

FIG. 4. Plot of the q/v,

splitting

against

2 for HzS and DIS. A?$;$ is in cm-.

where IZis 1 or 3, and the terms in parentheses represent the wavenumbers of the combination/overtone bands concerned. Then At,;;:, = v3 - vI + u2(xz3 - x2,). (8)

For HZS, ~23 = -21.8 and x21 = - 16.4, giving a slope of -5.4, in good agreement with our findings. But for D*S, ~23 = - 12.0 and x21 = - 10.4, giving a slope of - 1.6, rather shallower than the -2.6 our analysis suggests. It is clear, however, that for D2S, a relatively small increase in ~23 coupled with a small increase in x2, would improve the agreement considerably.
4. INTENSITIES AND SIMULATED SPECTRA

4.1. Band Intensities Vibrational band intensities (in units of atm- cmP2 at T = 298 K) are also given in Table V for the band origins presented, calculated from = 4.162 X lo- .(-$).(+)l,

( atm-cme2)

(9)

where wb is the band wavenumber and Rb is the vibrational transition moment. The D2S intensities are, in general, the lowest of the three isotopomers. In line with the findings of Senekowitsch et al., the bending mode is more intense than either of the two stretches for both D2 S and HDS, with vI more intense than v3. Coupling with bending vibration, however, causes the (0, 1, 1) bands to become more intense than the ( 1, 1,O) bands for both isotopomers. Where more than one stretching vibration is concerned the tendency is that the mixed vl / v3 modes are more intense for D2S and the pure z9 modes are stronger for HDS. In this way, the D2S behavior mirrors that of H2S.

SPECTRA OF D2S AND HDS 0.486

1 1

75
1111-10010 1011-12111 11 %-212 1 O12-922

b .3 i 2 0.000 f 54 cm -1

FIG. 5. Computed pure rotational spectrum for an equimolecular mixture of HDS and D2S at 300 K. Transitions of D2S are labelled V.

I
. 37-226 . i6

.. II

57

4.2. Rovibrational Transitions In Figs. 5 to 8 we present computed spectra for the regions for which Camy-Peyret
et al. have published detailed spectra (5- 7). For the pure rotational spectrum of their

mixture of HDS and D2S, our computed spectrum is a good representation in terms of both intensity and line position. In Table VII, we compare our intensities with those reported by Camy-Peyret et al. (5) for the pure rotational spectra for the most intense lines of Fig. 5. The wavenumbers are all reproduced to within 0.2 cm-. Agreement between the computed and experimental intensities is within 7%, probably within the experimental accuracy. At worst, our results indicate that the dipole surface is reproducing the permanent dipole to within 4%. Summing all energy levels up to J = 15, we compute the partition functions for DzS and HDS at 300K to be 1022 and 432, respectively, compared with the experimental values of 102 1 and 424 (5).
0.020

0%

92: cm -1 FIG. 6. Computed v2 spectrum of D2S at 300 K.

76
0.004 cu^ E! do b 0.002 5 h rz z J , 0.000
1950

MILLER ET AL.

1960 cm -1

1970

FIG. 7. Computed Y,/Q spectrum of D2S at 300 K.

In Figs. 6 to 8 we present sections of the simulated v2 and u1/v3 spectra of D2S (6), and the IQspectrum of HDS ( 7), for comparison with those presented in the experimental reports. The temperature was set to that of the experiment, 300 K. In all cases, the frequency distribution of lines (allowing for correction of the computed band origins) and the relative intensities are well reproduced. Experimental data on the absolute intensities of the rovibrational lines are not available. Since experimental data on the rovibrational spectra of HDS are still scarce, we present the full v2, vl , and v3 spectra in Figs. 9 to 11. All levels up to and including J = 15 have been included. In Table VIII, we have compared line strengths, S(f- i), for a number of transitions for D2 S and HDS. More detailed information is available from the authors on request.

0.042 cu^ B i !j 0.021 h .* : 3 0.000

900

1000

1100

lza0 cm -1

1300

FIG. 8. Detail of the computed 2 spectrum of HDS at 300 K.

SPECTRA OF DIS AND HDS TABLE VII Comparison of Computed and Experimental Rotational Transition Wavenumbers and Intensities for D2S and HDS (Variational Calculations)
vi, / cm- Intensitya / cm.mol-

77

Computed

Experimental

Computed

Experimental

54.083 54.653 54.689 54.746 54.782 55.923

54.140 54.834 54.868 54.922 54.957 56.094

1.05(-20) l.QQ(-20) 1.23(-20) X23(-20) S-00(-20) 1.16(-20)

9.87(-21) l.Ql(-20) 1.17(-20) 1.17(-20) 1.92(-20) l.OQ(-20)

DzS 856 - 744 11011 - 10110 11111 - 10010 1019 - 92s 10% - 91s 9s7 - 826 55.773 56.033 56.033 56.146 56.146 56.278 55968 56.179 56.179 56.293 56.293 56.423 1.23(-20) l-32(-20) 2.65(-20) 2.19(-20) lsxq-20) l-77(-20) 1.17(-20) 1.24(-20) 2.48(-20) 2.05(-20) 1.02(-20) 1.66(-20)

Pavers

of ten in brackets.

5.

CONCLUSIONS

The potential energy surface developed for HZS has proved itself to be equally good for the computation of rovibrational energy levels of D2 S and HDS. In the last 3 years Halonen and co-workers ( 16, 17) have developed a potential from spectroscopic data which, coupled with a perturbational Hamiltonian, gives even better agreement with

707-610

FIG. 9. Full Yespectrum of HDS at 300 K.

78
0.0050 c? E .-? k 0.0025 J h 5 i zi 0.0000 1600 1900

MILLER R

ET AL.

2000

2100 cm -1

2200

FIG. 10. Full v, (and 2~~) spectrum of HDS at 300 K.

the measured band origins for the Cl species. Our potential is purely ab initio, however, and has not been adjusted to fit the vibrational data, as was that of Halonen and coworkers. The drawback of their approach, however, is that the series representing the kinetic energy operator in the Hamiltonian they use is, of necessity, truncated. This, in turn, may have an influence on the terms reported in the potential energy, since these are fitted to the experimental data using the truncated Hamiltonian. One test of the extent to which this might be a problem would be to use the ab initio potential energy surface with the Halonen perturbational approach (or vice versa). From this it might be possible to improve the potential further. Senekowitsch et al. reported that while the dipole surface they calculated was a good representation of the bending coordinate mode, u2, it overestimated the strength

0.000 2500

2550

2600

2650

2700 cm -1

2750

2600

FIG.

I I. Full uj spectrum of HDS at 300 K.

SPECTRA OF D2S AND HDS TABLE VIII Sample Rovibrational Transitions for D2S and HDS (Variational Calculations)
Tkansition J,If;,x: J ) If., K, if / DzS HDS D* vi, / cm- S(f - i)b / D*

79

cm- S(f - i) /

(010 303 220 625 936 (020 303 220 625 936 (100 313 220 625 946 (001 313 220 625 946

WI
414 211 514 845 @JO) 414 211 514 845 000) 4 O/l 4 2 l/2 1 5 l/2 4 8 3145 000) 4 l/O 4 2 2/l 1 5 2/l 4 a 413 5 838.5 865.4 898.3 925.9 5.60(-4) 4.17(-4) 9.78(-4) 1.49(-3)
1008.9

1.47(-3) 7.28(-4) 1.79(-3) 6.58(-4)

1055.5 1094.7 1098.6

1694.7 1722.2 1755.6 1786.6

2.22(-5) 1.41(-5) 2.88(-5) 4.00(-5)

2042.0 2075.1 2129.8 2215.8

1.54(-5) 7.80(-6) 1.90(-S) 4.45(-5)

1878.2 1904.6 1935.8 1958.3

7.94(-5) 6.05(-5) 2.41(-4) 3.70(-4)

1877.3 1906.5 1951.0 1975.5

5.91(-5) 6.69(-S) 1.&q-4) 3.18(-4)

1889.7 1913.0 1947.5 1969.9

5.66(-5) 4.57(-5) 6.71(-5) 1.34(-4)

2597.6 2639.3 2677.3 2723.1

1.18(-4) 7.15(-5) 1.98(-4) 2.67(-4)

a Where two valuesof K. are given, the first refers to D2S.

b Powers of ten in

brackets.

of the stretch modes ( ul, v3). In the absence of experimental band intensities for either D2S or HDS, we are not able to test if this is still the case. It would, however, be surprising if it were not so. We conclude that while the intensities we have reported for the pure rotational and vz spectra are probably quite accurate, those involving the stretching modes are almost certainly an overestimation of what would be determined experimentally.
ACKNOWLEDGMENTS S.M. and J.T. acknowledge financial support from the Science and Engineering Research Council under Grant CR/F/ 14550. We thank James Henderson for plotting the wavefunctions. P.R. thanks Fonds der Chemischen Industrie and Deutsche Forschungsgemeinschaft for support during the course of this work. RECEIVED:

May 7, 1990
REFERENCES

I. J.

SENEKOWITSCH, S. CARTER, A. ZILCH, H-J. WERNER, N. C. HANDY, AND P. ROSMUS, J. Chem.

Phys. !20,783-794

(1989).

80
2. P.

MILLER ET AL.
BOTSCHWINA,A. ZILCH,H-J. WERNER, ROSMUS, P. AND E-A. REINSCH, Chem. Phys. 85, J.

5107

5118(1986). 3. R. L. COOKE,F. C. DE LUCIA,AND P. HELMINGER, Mol. Struct. 28,237-245 J. ( 1975). 4. 1. M. MILLS,private communication ( 1989). 5. C. CAMY-PEYRET, J-M. FLAUD, AND L. LECHUGA-FOSSAT,J. Mol. Spectrosc. 109, 300-333 ( 1985). 6. C. CAMY-PEYRET, J-M. FLAUD, A. NG~M, AND J. W. C. JOHNS, Mol. Phys. 65,649-657 ( 1988). 7. C. CAMY-PEYRET, J-M. FI_AUD, A. NGoM, AND J. W. C. JOHNS, Mol. Phys. 67,693-695 ( 1989). 8. J. TENNYSON AND S. MILLER, Comp. Phys. Commun. 55, 149-175 (1989). 9. F. T. SMITH, Phys. Rev. Lett. 45, 1157-l 161 (1980). 10. B. T. SUTCLIFI%AND J. TENNYSON, ht. J. Quantum. Chem. (in press). Il. J. TENNYSON AND B. T. SUTCLIFFE,J. Chem. Phys. 77,4061-4072 ( 1982). 12. S. MILLER, J. TENNYSON, AND B. T. SUTCLIFFE, Mol. Phys. 66,429-456 ( 1986). 13. S. CARTER AND N. C. HANDY, J. Chem. Phys. 87,4294-5007 ( 1987). 14. M. S. CHILD AND L. HALONEN, Adv. Chem. Phys. 57, l-58 (1984). 15. I. M. MILLS AND A. G. ROBIETTE,Mol. Phys. 56,743-765 ( 1965). 16. L. HALONEN AND T. CARRINGTON, JR., J. Chem. Phys. 88,4171-4185 (1988). 17. E. KAUPPI AND L. HALONEN, J. Phys. Chem. (in press).