Está en la página 1de 10

JOURNAL OF MOLECULAR SPECTROSCOPY

126, 183-192 (1987)

First Principles Calculation of the Molecular Constants of Hi, H2D+, D2H+, and Di
STEVEN MILLER AND JONATHAN Department TENNYSON

ofPhysicsand Astronomy,

University College London, Gower Street, London WCIE 6BT, United Kingdom

Ro-vibrational calculations on a recent ab initio potential energy surface for H: (W. Meyer et al., J. Chem. Phys. 84, 891 (1986)) are presented for the isotopomers H,_.D: (n = 0, 1, 2, 3). These calculations employ refinements recently developed for nuclear motion calculations and are thus of very high accuracy. The rotational levels with .I G 4 are fitted to standard vibrationrotation Hamiltonians due to Watson. The results are in excellent agreement with the experimental results where available. The complete set of molecular constants form probably the most complete and accurate determination of such parameters yet available. o 1987 Academic press. IX. I. INTRODUCTION

Although Hi has a comparatively long history (I), its spectrum was not observed until the work of Oka (2) and Shy et al. (3) in 1980. Since that date the intensity of spectroscopic activity has been great, with the last few years seeing the experimental characterization of all the infrared allowed fundamental bands of Hz and its deuterated isotopomers (4-9). The activity among theoreticians has been no less intense. Since the pioneering work of Camey and Porter (IO), many workers have been attracted by the challenge of a system with only two electrons, which thus lends itself to highly accurate electronic structure calculations (I I). The combination of theory and experiment has been particularly important in unraveling the complexities of the highly asymmetric mixed isotopomers HzDf (7) and D2H+ (8). This combination has led to the first tentative extraterrestrial sighting of H2D+ (12). The advantage that calculations, such as those presented here, have over observation is that it is in principle possible to obtain full information about all the levels in the system, uninhibited by selection rules and intensity considerations. This led Amano and Watson (4) to comment that the accuracy of the pure rotational frequencies they calculated from analysis of the uI band of H2D+ was “not greatly improved over the ab initio estimates.” In this paper we present new ab initio results for H: and its isotopomers which we expect to be of comparable accuracy to the experimental results for many situations and of greater accuracy in the cases where observation gives only partial information. We expect these results to be a significant improvement over previous ab initio estimates for two reasons. First, a new ab initio surface for H: has been calculated by Meyer, Botschwina, and Burton (MBB) (13). As shown by MBB, and confirmed by Table I, their best surface yields highly accurate theoretical estimates for band origins of H: and its isotopomers. As our analysis shows, this high accuracy is maintained for the
183 0022-2852187 $3.00 Copyljsht 1987 by Academic 0
All rights of reproduction

F’req Inc.

in any form resxved.

recent refinements have been proposed to the methods used to perform the nuclear motion calculations. k represents the projection of the total angular momentum. enabled us to converge all the levels calculated here to better than 0. 2. and the use of basis set selection according to the value of the diagonal matrix elements (14) in the first variational step. 0) In the above. J. The Hamiltonian for z embedded along rl is simply obtained by interchanging rl and r2 and pl and p2 in Eq. +L -r$-+-sm p2r-j( at-2 dr2 sin 6 a8 1 d 19% 11 (2) KvR = b&i 2 (J(J+ 1) . iC. The coefficients c:i.’ p:’ = m. are the usual step up and down coefficients c&= (J(J+ 1) -k(k+ 1))“2. In these coordinates the triatomic vibration-rotation Hamiltonian is represented as an atom interacting with a diatom: (1) with Kv=-- h2 z 1 a -rT 2 [ plrl ( drl a+ --sinBag dr. METHOD This work follows closely the approach used over a number of years by Tennyson and Sutcliffe (TS) (16-18) which is based upon the use of scattering coordinates. In this work the same technique is used to obtain highly converged results for levels with J < 4. This level of accuracy is sufficient to eliminate any significant splitting of degenerate levels caused by our neglect of the full symmetry of the symmetric . on the body-fixed z-axis.2k2) 2w? + ak’k+ I --+kcotB ) a ae +&‘k-. (3). la sin BdB a a d Id.’ + (Wzz m$‘. r2 the atom-diatom between rl and r2. This.’ + m. By this means they analyzed levels with J =S20. 2cLlr: (3) where rl is the diatom bondlength. + distance.184 MILLER AND TENNYSON rotational levels. The only term in H which couples states which differ in k is KVR.*($+kCOt6). Recently TS (25) solved the H2D+ ro-vibrational problem in two steps by (a) treating k as a good quantum number and (b) using the lowest solutions of step (a) to solve the full problem.’ = m. Second. The reduced masses are given by p.0 1 cm-‘. The above is for the z-axis embedded along rz. and 19the angle (4) where the atom has mass ml and the two diatom masses are mz and m3. In particular the use of more sophisticated methods of selecting basis functions (14) and of a two-step variational technique for rotationally excited states (15) has led to a significant enhancement in the accuracy and possible range of such calculations.

the improved representation of the ground state is also repeated for the vibrationally excited state parameters. all the results quoted here are for MBB’s corrected 87 CGTO potential which is their most accurate. Our calculations agree well with MBB’s and both give agreement with experiment to better than 0. but set to the value obtained from the J = 0 calculation as given in Table I. of our fits. In Tables II-V we present the results of our rotational analysis for each of the isotopomers. This failing was attributed to both inaccuracies in the potential and in the nuclear motion calculations (18). 17) for the ground state molecular parameters. the .CALCULATED CONSTANTS OF H: 185 isotopomers. For space reasons we have not given the energies of the individual levels (which can be obtained from the authors). The band origins were not included in our fitting. Only Coriolis interactions within the (Q. The empirical fits necessarily involved fitting levels simultaneously from both the ground and the excited states. In all tables. The most striking feature of the tables is the consistent good agreement between ab initio and experimental parameters for all the bands considered. The basis set for each calculation which treated k as a good quantum was chosen by selecting the lowest 600 functions for a J = 0 calculation. u. 5. In both steps we probably could have obtained good results with fewer functions. (7). This approach allows us to make a direct comparison between the molecular constants obtained ab initio and from experiment. vj) levels were conducted separately and each gave a separate standard deviation. The rotationally excited states were determined by using the lowest 300 vibrational levels for each k. It should also remove the irregularities in TS’s H2D+ results noted by Foster et al. 17. For all the ions considered our fits are better for the ground than the vibrationally excited states. CALCULATIONS For all calculations presented here we used Morse oscillator-like functions to carry the radial coordinates and associated Legendre functions for the bending coordinate. fairly accurate ab initio results had been obtained (4. 3. The radial basis set parameters were as optimized previously by TS for the relevant system (11. and (uZ. Although previously. but we were keen to ensure the accuracy of our calculations. In this work. 7-9). Q> manifold were explicitly accounted for in the fits.05%. All the calculations used program suite SELECT. It should be remembered that the standard deviation only measures the quality of the parameterization of the levels and gives no information about other errors in either the ab initio or the experimental results. vl. This is probably due to the increasing density of vibrational states and the resulting increase in Coriolis effects. Instead we present the results of fitting our calculated levels to the various model Hamiltonians developed by Watson and coworkers (4. 18). thus giving only a single standard deviation. In Tables II-V values are given for the standard deviations. The experimental results for the observed fundamentals are also given. Hence our fits for ground state. Table I compares our calculated zero-point energies and vibrational fundamentals with those predicted by MMB for the same potential. Although preliminary calculations were performed on several of the potentials due to MBB. TRIATOM. the agreement for the excited states was not so good. and ROTLEV (I 9).

The estimated experimental error is shown in brackets in units of the last figure for each band origin.) va(bend) v.096 2737.0 .3 2992. R3+ v.4 1967. (9) modified their previous parameterization procedure for the de- .669(l) 9 v. ) v.3 zPe 4363.302 1967.(es.212 3563.308(9) 5 ~*(R-tYPe) 2521. to determine Co by scaling TS’s ab initio value using the error in Bo.l)J (5).49(2) 4 2205. (5) in their analysis of the v2 band of Hz.250 1835.(bend) 2993. Because of the selection rules for infrared transitions in Hi.501 2521.OO 115 is very close to 1. it was necessary for Watson et al. the ratio of our Co to TS’s.060 3113.) zPe W+ v. Table II compares the ab initio and empirical molecular constants for Hi.825 2079.348 2521.449(9) 7 ".5 93+ u.733 1834.(A-me) 3178.OO 122.(s.4 3178.17(2) 2078.(as. Table III compares the ab initio and empirical molecular constants for D: .905 2992.2 2737.(s.str. Also given is the calculated zero-point energy (zpe) for each isotopomer. All the ab initio constants were derived using the parameterization utilized by Watson et al. Our calculations show that their scale factor of 1. For this ion Watson et al..869(8) 7 2335.str 2335.str.atr.282 4363. Table II also gives our values for the parameters qj and qk which give rise to the splitting of the G = 0 levels but which cannot be determined empirically as nuclear spin statistics forbid occupation of the levels with s = (.0 2206.) 2737. The small standard deviations for our fits show that this gave a good representation of our data.962 2206. which was based on Watson’s analysis of X3 systems (20).9 2079.186 MILLER AND TENNYSON TABLE I Comparison of ab Initio and Experimental Vibrational Frequencies in cm-’ IsotoDcmer Band Ab initio MBB(13) This work Experiment Band origin Ref.wtype) Is@+ v.41(2) 6 I3 8 zPe Note.244 2334. estimated error in each of the experimental constants is given in brackets in units of the last figure of the relevant constant. (A-type) 2301.00(3) 1968.986 3979.

88(9) 6.43 3.19 -7.7 44.399 -18.97 14.171 19.3 -5.95 5.74(4) 1.604 4.03(9) 16.i load" lO'h.1(4) -1.2 4.61(4) 3.98 0.9(4) 4.94 1.5 1.38 -0. 44.04(3) -7.CALCULATED CONSTANTS OF H.28(3) -10.605(3) 4.13 2.7 43.565(2) 20.6 -1. 1O'cr 0.651(4) -14.550 20.9 -5.009 4.1(l) 42.1(l) 10'0 1.340(6) -18.0(l) -1.1 -5.502 20.71 3.2 -0.2 Y1 1.226(3) 19.5 state ExPeriDFmt (5) Parameter Ab Initio B C C( lO%pJ lo%+ ~o~DJJ ~o~DJK lOaD* lo'@ q 1ozqj loaqk 10"dj lo'dk losj lo'& lO'h.77(19) 5.9 1.18(2) -7.4 .379 1.00 4.11 -8.2(l) -1.4 -0.84 -9.16(4) -5.602 -13.1(6) 3.34 4.7 1o’oJJ 10a~JK lOaD= 10Zd.0 -5.7(2) -4.4 0.372(3) 2. TABLE II 187 Vibration-Rotation Parameters for Hi in cm-’ Ground State "1 state Ab Parameter Ab Initio Experimemt Initio (5) B C 43.

97 ~o’DJJ IO'DJK lO=~~'f 10'dj lo'd" 10'h.0 -1. state Ab Initio Experiment Initio (9) B C 21.971 10.99 21.0 Y.3 -4.788 -1.3 -0.2 -0.0 state Experiment Parameter Ab Initio (9) B C CC 1O'd lo%+ 21.301 0.5104 0.995(2) 21.5 -1.425 10.125 -9.867(9) -0.9 1.7417(3) -2.75 0.35(E) ~o’DJJ 10'Dm IO'@ lo'/3 q loJqj lo'qk 1O'dj lo'dk io+J lo'& lO'h.241(9) 1.03 -0.2 0.1506(4) -9.04 1.8 -2.1 10'0 0.842 3.188 MILLER AND TENNYSON TABLE III Vibration-Rotation Parameters for Df in cm-’ Ground State Parameter Ab Y.08 -2.3 0.8240(3) 10.747 -3.1 -1. 10’0 0. 0.85 1.24 .74 0.01 1.515(2) -1. -1.22 3.8385(3) 3.39(l) -1.8 -2.511 0.7 1.818 10.92(l) 0.1 -0.082(6) 21.100(3) -1.7 0.734(5) 0.24 0.9748(3) 10.84 0.51(3) 1.

standard deviation 0. Amano and co-workers (4. The ab initio data of Table V are therefore fitted using the effective Hamiltonian the same as that of Table IV. In analyzing the (~2. Again experimental fits are available from analysis of both the v1 and (Q. while producing fits with similar or even slightly smaller standard deviations than the original parameterization. The discrepancy was largest for the set (C. may simply be due to differences in the fitting procedures adopted. but with the sextic terms represented by Pade-type expressions (7). We found that this modification. All the ab initio data on H2D+ were parameterized using the effective Hamiltonian (as opposed to supermatrix) of Foster et al. C3. Note also the large errors for the corresponding empirical constants. This Hamiltonian is based upon Watson’s I’ A-reduced asymmetric rotor Hamiltonian (22). without Pade approximants.005 cm-‘. vg) bands. as the fitting procedures for the v2 levels are no longer identical any close comparison between the ab initio and empirical parameters must be made with caution. However. had to scale ab initio values to obtain COand DtK. For H2D+ and DlH+. analysis of the u1band supplies an additional set of empirical constants for the ground state. of using three of the possible seven sextic constants (the a’s). Table V compares the ab initio and empirical constants for D2H+. The largest disagreement between previous ab initio and experimental determinations of the molecular constants of H2D+ was for terms related to the Coriolis coupling. .. For this ion spectra have been recorded and fitted for both the v1 and strongly coupled (v2.&) (7) apparently because TS’s calculations gave a relatively poor estimation of the gap between the v2. Pure rotational spectra have also been observed. but only for two key transitions (21). Q) bands.CALCULATED CONSTANTS OF H. 4. We followed the approach of Foster et al. However. Again for D:.04 cm-’ compared to 0. (8) slightly modified the effective Hamiltonian Coriolis constants used in their previous analysis of H2D+ (7). showed an extreme interdependence between the parameters which rapidly led to unphysical quantities. Of course. In fitting these transitions. due to correlation between these parameters. It is interesting to observe that many of our ground state parameters. 189 generate u2 band to make their Pade approximants “more physically meaningful. ~3)band.” However. the AG = 0 selection rule meant that Watson et al.and u3-band origins. we found that this parameterization gave a significantly worse. Table IV compares the ab initio and empirical molecular constants for H2D+. most of this difference remains in our more accurate redetermination and. (7). for example all three A’s for both ions. 6) used the I'A-reduced Hamiltonian (22) to sixth order. The good agree ment between their and our values confirm the validity of this procedure. lie in between the two sets of empirical values. Foster et al. CONCLUSIONS We have presented the results of a series of calculations on the low-lying rotational and vibrational states of Hz and its deuterated isotopomers using the accurate potential . We therefore used the same method as in Table II to represent our data. Using all seven constants was found to reproduce our data to better than the accuracy of the calculations. we would suggest that the values presented for the three sextic constants retained in our fits are of no great physical significance. fit to our data.

331 27.5 -1.440 29.02 3.14(l) 16.91(14) 1.8 Y.62 1.303 16.43(5) 2.30(3) 30.0(7) -30(5) 6(20) 42.20 -24.602(2) 16.27 -2.95 1.592 -5.63(14) 0.80 0.32 5.36(2) 16. state Ab Initio BxperiRnt (4) 41.00 0.6(7) 43.251 30.7(2) -13(6) -5(13) lo’+JJJ lo’+JJK lo’+Ju N(2) 120(70) -70(17) 10% 0.8 8.02 0.68 3.15 0.3 17.5(8) 4.2(8) 0.41(3) 14.4 -17.611(8) 1.31(l) 27.269 1.137(2) 29.4(7) 6.39 1.4 ExperiRnt (4) 41.603 1.42(6) 2.115 16.5(l) -1.0(4) 1.3(l) 3.8 Pammeter Ab Initio Experiment Znitio Experiment (7) A (7) 46.466(e) 43.24 VI state 1. state Ab 1.0 0.688 17.66 2.8(5) 0.TABLE IV Vibration-RotationParametersfor H2D+in cm-’ Or0und state Parameter Ab 1Ditio (7) A B C ~O'AJJ 1O'Am 1O'A~ i0v.i lo'& 43.0(7) -2.18 0.14 0.61 0.604 29.9(2) -2.7(2) 0.4(4) 1.38 2.39(l) 1.35(3) 3.6 46.3(8) 82(3) 1.08 0.1(3) 5(3) -4(24) 81(54) B C 10'6JJ lOsAm 1O'Am 10% lo'& 1o’tJJJ 1ovJJK ~o=tJm -22.32 0.59(7) -5.8(2) 0.24 .38(2) 2.0(6) -6(5) 10'0 1.52(7) 29.3(6) 0.382 14.18(g) -0.682(9) 16.256 0.2 35.8 13.6 42.71 0.25 2.5(l) 0.4 -2.402(3) 1.4 Y.36(7) 29.61 4.616 1.2 0.23 2.98(3) -1.4 0.389 1.8 -2.0(l) 0.8 0.6(7) 3.

103 20.19(2) 21.5 "1 state 3.067 0.813 23.5 34.3 36.066(5) 1.74 0.0 0.7(5) 0.3(l) 15.7 -43.06(4) 0.39(2) 0.62 -0.21 1.194 21.82(2) 10'0 0.0 .6(3) 2.erirnt (6) (8) A (6) 36.74 0.47 1.4bInitio Experiment Initio BxPcrimmt (53) (8) A 38.9(30) -16(18) 25(23) 34.8(2) 2.12 -0.3 50.25(28) 1.32(9) -3.47(2) 0.24(7) 0.78(3) -1.616 12.707 0.7(5) 7.39(9) -3.6(4) 0.877 13.32 2.03(2) 1.8(2) 1.11(2) 20.61 1.225(S) 1.3 -5.456 -0.219(8) 1.46 -0.935 0.73 1.58(l) 12.059 11.42(30) 3.4 "I state Ab 3.3 25.18(4) -0.0 -4.869(6) 13.4(3) 3.22 1.93(9) -0.17 1.29 1.6 2.4 2.74(8) 35.59 0.11 2.22 1.11 3.904(7) 13.96 0.98(l) 0.24(2) 21.83(l) 13.057(3) 0.5 .2(9) 105(l) B C 1o'bJJ ~O'AJS 10zAm load lo’& lo+JJJ lo’+JJ1( lO'+JJKK 10'0 Table V cont.08(l) 11.39 -3.3(E) 0.41(2) 21.9(7) -9.859 13.92(3) 1.3 38.9 8.419 21.TABLE V Vibration-Rotation Parameters for DzH+ in cm-’ "I state Ab Initio Ground State Pal-ameter Ab Initio BxDeriment bF.1(5) 15(11) 64(56) 35.12 0.17(39) 0.700 -4.68 3.47 -21.234 1.13(3) 2.0 -5.7 36.6(2) 6(3) -130(260) 300(600) B C 1o'dJJ 10'Am 10aAm 104 lo+ lo’+JJJ lo’+JJK 1ovJRK -22. Parameter 0.57 0.86(l) 23.7 7.

. G. Phys. WING. A. CROFTON. W.J. G. Mol. SUTCLIFFE. Phys. OKA.T. ALTMAN. H. Lett. l-36 (1986). T. The Netherlands. S. J. WATSON. 13.J. Sutcliffe. SUTCLIFFE. Structure and Chemistry” (T. Comparison with the available experimental results suggests that these calculations are highly reliable. 1. S. P. Chem.-T. S. H. J. 77. T. 82. AND T. K. TENNYSONAND B. DENIS. 20. Phys. J. BURTON. LAMB. G. 4. J. 2. K. R. 45. A. 5. AMANO. SUTCLIF’FE. D. J. 22.8746 A. W. WOODS.AND J. OKA. I. Chem. WATSON. LEMOINE. N.531-534 (1980). M. T. and Drs. A~~ANo. Eds. G. P. J. 1983. 9. T. PHILLIPS. TENNYSONANDB. 350-363 (1984). Mol. 8. WOODS. Mol.192 MILLER AND TENNYSON surface of Meyer et al. K. 81. AND B. Chem. Durig. W. FOSTER. 65. J. P/zJxRev. SUTCLIFTE. C. 16. C. HIROTA.C. SAITO. J. 4.891-900 (1986). Astrophys. K.J. F. CROFTON. TENNYSON. Chem. 14. S. TENNYX~N AND B. W. . E. T. G. MEYER. Commun. BLAKE.AND E.25 14 (1984). Phys. Phys. J. J. 62. KEENE. I12. FOSTER. Chem. 137. pp. J. .257-270 (1986). Spectrosc. 1886-1888 (1984). R. RECEIVED: January 15. Phys. FOSTER. MCKELLAR. North-Holland. 54.W. C. M.J. Phys. OKA in “Molecular Ions: Spectroscopy. L45-L48 (1985). W.38-46 (1987). Chem. L15L16 (1984). WATSON. Phys. Phys. 84.I. J. W. PAN. G. Sot.AND E. SO. T. F. S. R. DESTOMBES. 58. R.664-670 (1986). Chem. Phys. T. 51. C. 60. Undoubtedly they represent the most accurate first principles determination of the molecular constants of any polyatomic. 115 1162( 1986). Comput. AND R.-L. Lett. MCKELLAR. Canad. TENNYSONAND B.M. SHY. FARLEY. G. G. Watson et al. LLJBIC AND T. 21. J. It is our intention to perform further calculations on these ionic systems for highly rotationally excited states. Chem. J. W. Phys. IO. 103. J. OKA. PETRMICHL. R. M.425 1-4264 (1974). Phys. 1175-I 183 (1985). Rev. CONNER. A. Phys. CARNEYANDR. 6. PAN. AND T. KAWAGUCHI. The aim of this work will be to study spectral regions where the separation between rotational and vibrational levels becomes comparable and hence Coriolis interactions are of enhanced importance. T. 7. T. JR. Rep. WATSON. Astrophys.79.8743 A. 3.45-47 (1982). K. 56. Amsterdam.R. 81. Faraday Trans. TENNYSONAND B. “Vibrational Spectraand Structure”(J. Chap.J.A. Comput. S4. Miller and V. J.K. WATSON. 2 82. T. G. PORTER.). Watson and P. Lett. 45. 19. SUTCLIFFE.537-541 (1980). 6. Vol. K. T.AND W. 327 (1977). BOGEY. Phys. Phys. Finally we note that these calculations were performed on a potential energy surface whose minimum is at an Hz bondlength of 0. Phys. S. J. G.). Canad. 1987 REFERENCES 1. J. J. 1875-1885 (1984). Phys. C. (13).91-99 (1985). MCKELLAR. A. 18. 1I. J. Lett. ALTMAN. FOSTER. 141-144 (1985). WATSON. G.887-906 (1984). DEMYNCK. Botschwina for communicating results prior to publication. AND A. 15. yielding a complete set of parameters of accuracy similar to those derived empirically. WARNER. 85. Phys. 73-99. G.2869-2871 (1984). E. J. Bondybey. W. BERNATH. Chem. PETERKIN. S. 17. CHURCHWELL. Chem.43-5 1 (1983). Astron. H. Mol. 42. K. Canad. R. S. 294. BOTXHWINA AND P. WATSON. J. K. R. G. The good agreement between theory and experiment displayed by the results of this paper and the closeness of these values suggest that they must be near the true geometry parameter for this system. E.3547-3565 (1976). J. K. Elsevier. W. R. C. Ed. Phyx Rev.406 l-4072 (1982). MCKELLAR. S. 1977. 1067-1085 (1986). (9) extrapolated their results to a value of 0. Left 45. TENNYSON. B.R. J. 62. ACKNOWLEDGMENTS We gratefully acknowledge many helpful discussions with Dr. 65. J. AMANO AND J. 535-537 (1980).