J. CHEM. SOC. FARADAY TRANS., 1994, 90(5), 683-688 683 Configuration Interaction Studies on the S, Surface of H,CO: 2 'A'@, n*/z,n*)as Perturber of 1 lB2(n, 3s) Michel Hachey, Pablo J. Bruna and Friedrich Grein* Department of Chemistry, University of New Bruns wick, Fredericton, New Bruns wick, Canada €38 6E2 MRD-CI calculations reveal that the third singlet surface S, of H,CO exhibits two energetically close-lying minima. The more stable state at 7.09 eV is the experimentally known 1 'B2(n, 3s).The second minimum at 7.53 eV corresponds to the pyramidal state 2'A', of mixed CT, n* and IT, n* character near equilibrium. This S, valence minimum, which apparently has not been experimentally observed, is suggested to cause the blue shift of the S,+-So system of H,CO in condensed media (as well as in related aldehydes and ketones), ruling out n +CT& ('9,) as responsible for this blue shift.The 2'A' state is non-adiabatically coupled with the ground state, explaining the observed predissociation of 1 '9, into the CH, + 0 continuum. The planar (C2J and pyrami- dal (C,) minima can be connected only through a C,intermediary, requiring participation of the antisymmetric CH-stretch and CH, bend vibrations, besides the CO-stretch and out-of-plane modes. A dipole moment p = +3.45 D (1 D x 3.33564x C m) with polarity (H2C)-0' is predicted for 1 'B,(n, 3s)at equilibrium, at variance with an experimental ,u of +0.33 D. An experimental route for studying the valence region of S, is proposed. The photochemical decomposition of H,CO into H, + CO and H + HCO via the 1'A, +-% 'A, (S, tSo) transition at energies below 4.0 eV has been studied extensively by experimental' and theoretical group^.^ By contrast, almost nothing is known about the photochemical fragmentation into CH,(8 3B,) + O(3PJ.Since these products lie at 7.64 eV (experimental),,? rupture of the CO bond can only be affected via primary absorption into electronic states lying above 7 eV. In CZVsymmetry, the dissociation channel CH2(8 3B,) + O(3P,) correlates with 'v3,'[A,, A,, B,] states of H,CO. Focusing on the singlet manifold, the surfaces So, S, and S, (the lowest singlet surfaces irrespective of symmetry) correlate with the ground-state products CH, + 0.The lower states So (8'A,) and S, (1 'A,) can be ruled out as precursors of the CH, + 0 products since rupture of the CO bond would require a large amount of vibrational energy.Instead, the vibrationally highly excited states of So and S, decompose preferentially into H, + CO or H + HCO, as these are more stable than CH, + 0 by ca. 7.6 and 4 eV, respectively.'q2 In the vertical region, the S, surface corresponds to 1 'B,(n, 3s), a Rydberg state placed experimentally at 7.09 eV,4-9 only 0.55 eV below CH, + 0. The process 1 'B, +-2'A, (S, cSo) thus constitutes a good candidate for investi- gating the dissociation H,CO CH, + 0. Interestingly, in 1935 Price" assumed that the diffuse character of the S, tSo system was caused by predissociation into CH, + 0 products.A careful analysis of the experimental literature leads to the conclusion that the S, tSo transition is more complex than expected for an n+ 3s excitation of the type non-bonding -+ non-bonding. Besides the predissociation effects mentioned above, three additional experimental peculiarities are of interest, namely (i) the blue shift in rare-gas fluids, (ii) the anomalous electric dichroism and (iii) the irregular vibra- tional structure. On the first point, the S, tSo spectrum has been recorded several times by gradually increasing the Ar density." In these experiments, the maximum at 7.09 eV (relatively narrow t Af H"data (in eV) used: -1.085 (H,CO), 4.001 (CH, ,8'B,) and 2.558 (0,3P,). band) in the Ar-free spectrum is shifted to ca.7.40 eV (rather broad band) in the Ar-dense spectrum. Since Rydberg bands should not be observable in condensed media,', their appar- ent persistence in the spectrum is an indication of the pres- ence of a valence state lying close to the s, Rydberg minimum. Note that a similar coalescence into a 'valence' band in the condensed media has been reported for the corre- sponding 'B,(n, 3s) Rydberg state of CH,HCO and (CH,),C0.12 Although there is no direct proof to substan- tiate such an assignment, the experimental literature favours n +gzo (lB2) as the underlying valence ~tate.**'~.~~ On the second point, the relative change in dipole moment for the S, + So excitation was investigated by recording the modulated electric field spectrum (electrochromism).l4 These measurements show that the transition moment is strongly affected by the applied field, a feature also pointing out that 1'B,(n, 3s) is perturbed by a nearby valence state. Finally, the vibrational structure of the vertical transition S, tSo in both the optica16*8 and electron-impact7*'' spectra looks more complex than that of the photoionization spec- trum H2CO+(% ,B,) tH,CO (% Mental1 et aL6 con-cluded that the vibrational structure extending from 7.09 to 7.37 eV involves all three a, modes v,, v2, v3, (CH- and CO- stretch, and CH,-bending, respectively), as well as the b, mode v4 (out-of-plane). Later, the high-resolution spectrum from 6.89 to 7.62 eV was interpreted by Lessard and Moule' as arising essentially from excitations of all totally symmetric modes, though the participation of the vibrations ~4(b1) and Vs(b2, CH,-rocking) was not conclusively ruled out.At this point we can assume that the anomalies reported for the vertical transition S, tSo (1 'B, 8 'A,) are caused t by a common, up-to-date unidentified state. In general terms, the perturbing state should be of valence character, lie close (geometrically and energetically) to the planar 1 'B, Rydberg minimum, and be somehow connected with CH, + 0 pro-ducts. The search for a relatively low-lying singlet excited state of valence character [other than 1 'A,(n, IT*)] can also be interpreted as asking for the location and relative stabil- ities of three valence excitations: n -+ czo ('B,), CT +n* ('B,) and n + IT*('A').None of these excitations has as yet been characterized e~perimentally.'~~-'~~''~~ In this work, an ab initio investigation of the lowest Rydberg state 1 'B,(n, 3s) and close-lying singlet valence states is carried out. The CO-stretch potential curves for both planar (C,,) and pyramidal (C,)conformations are presented for states relevant to this work. Higher-lying singlet, triplet and quintet states are discussed elsewhere.18 As shown in the next sections, the valence state perturbing (in fact, mixing with) the Rydberg portion of the S, surface of H,CO can be identified as a bent 'A' species of mixed (r, n* and n, n* com-position. Contrary to expectations, the excitation n + azo does not interact with n + 3s near equilibrium.Technical Details For carbon and oxygen the lOs6p/5s4p basis sets from Huzinaga-Dunning were used. Additional d-polarization functions were added, with exponents of 0.75 for C2' and of 0.74 and 1.52 for O.,, A CO bond function of s-type was also included with a, = 1.15.,, Rydberg AOs of s, p and d type were added to each heavy atom. In that order, the exponents are 0.023, 0.021 and 0.015 for C, and 0.032, 0.028 and 0.015 for O.,' This basis set, containing 85 contracted AOs, will be called basis I. It has been used for all MRD-CI calculations on planar H,CO. A second A0 basis set (called basis 11) was derived from basis I by replacing the two d-polarization functions on the 0 atom by a single function with an exponent of 0.85; the d-Rydberg function on 0 was deleted.This basis set, contain- ing 73 contracted AOs, was employed to study the non- planar conformations. The majority of the CI calculations reported here were carried out with 1 'A,(n, n*)SCF MOs. The standard frozen- core approximation has been used throughout. The CI results were obtained with the multireference MRD-CI method developed by Buenker, Peyerimhoff and co-worker~.~~ The total and relative energies correspond to the so-called esti- mated full CI values, which are obtained by a generalized Langhoff-Davidson c~rrection.,~ Depending on the particular symmetry representation under study, the number of reference configurations varied from 40 to 60. In general, the selected CI spaces have dimen- sions from 20000 to 32000, out of a total generated space of the order of 3 x lo6.The bulk of the calculations was carried out by selecting simultaneously the six lowest roots (states) of each symmetry; the selection threshold used was 10 pE,. The geometry optimizations of the 1 'B, (planar) and 2'A' (pyramidal) states were carried out by using a smaller selec- tion threshold T = 5 pE, and by selecting only one ('B,) or two ('A)roots. The standard C,, orientation is used throughout. The z axis (CO bond) transforms like A,; the y component lies in the molecular plane and behaves like the B, representation; the x axis transforms like B,. Starting from planar H,CO, two different C,geometries can be generated where either the molecular or perpendicular plane is maintained. These planar Table 2 Adiabatic excitation energies (in ev) with respect to C-0 methods" 1 'B,(n, 3~)~ 2 1A(o,n*/n, n*) 2 'Alp; 3PJ (left minimum) 7.08 7.53 7.98 (ca.7.06) 8.40 -7.99 9.197.93 (7.97) 8.51; 7.70 8.14; 8.49 -; (8.47) J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 and pyramidal conformations are labelled as C,(yz) and C,(xz), respectively. Results A summary of the equilibrium geometries predicted for the 1 'B, and 2 'A' states of H,CO is given in Table 1. Literature data on the ground states of H,CO and H,CO+ as well as the 1 'A" state of H,CO are also given in Table 1 for com- parison. Adiabatic excitation energies (with respect to C-0) for the H,CO states 1 'B,, 2 'A1, 1 'B, (C2J and 2 'A' (C,, xz) are collected in Table 2.The corresponding CO-stretch potential curves are shown in Fig. 1 ;the energetic location of the products CH, + 0 is also indicated. CO-stretch Potential Curves for Planar H2C0 W2" Symmetry) 1 'B, State This state has n -+ 3s Rydberg character at equilibrium. As shown in Table 1, the equilibrium geometry of 1 'B, as pre- dicted in this work differs by <0.02 a, for the CO and CH bonds and 3" for <HCH from the experimental ground-state values.25 Our results for 1 'B, agree closely with those pre- dicted by Feller and Davidson26 for the jf2B, state of HzCO+, in particular R(C0) and <HCH. A prior GVB in~estigation,~on the 1 'B, state of H,CO obtained slightly larger values of R(C0) and <HCH.By contrast, two recent CIS studie~~~~~~ have reported opposite changes: shorter R(CO), larger R(CH) and smaller < HCH. Table 1 Ab initio equilibrium geometries of the 1 'B, and 2 'A states of H,CO and literature data as the 8 'A, and 1 'A" states of H,CO (experimental) and on the 8,B, state of H,CO+ (theoretical) ~ R(CO)/a, R(CH)/a, <HCHIdegrees eldegrees ref. H2C0, C2,, 1 'B,(n, 3s)2.288 2.090 119.0 - this work 2.324 2.060 124.0 - 27 2.123 2.138 2.218 2.220 100.2 100.5 -- 29 28 2.274 2.080 H2C0, CZv,fl: 'A,116.2 - 25 2.287 2.079 H2CO+,C,, , x ,B,118.0 - 26 2.888 H,CO, C,, 2 'A(on*, nn*) 2.040 111.3 ca. 44 this work 2.813 2.034 122.5 49.9 28 2.8 13 2.038 121.9 46.5 29 2.500 2.075 H,CO, C,, 1 'A(n, n*) 118.4 34.0 4 of selected low-lying singlet states of H,CO as predicted by ab initio ~ 2 'A,(n, IC*) (right minimum) 1 'B,(o, n*)c ref.7.95 8.27 this workd 8.50 (7.99) (< 8.32) 30' 8.61 -28f 8.65; 8.60 8.35; 8.79 29g a Planar C,, geometries except for 2 'A' in C,(xz) symmetry. * To, = 7.09 eV, experimental (ref. 1, 4-9, 11, 12, 15, 17). 'Saddle point in C,,. Estimated full-C1 results from MRD-CI treatment. See also ref. 18. 'Langhoff-Davidson correction in brackets. CIS results including zero-point corrections. CIS (first entry) and CIS-MP2 results (second entry). Value in brackets corresponds to the vertical transition energy. J. CHEM. SOC. FARADAY TRANS., 1994, VOL.90 9.5 8.5 > 7.5 G \Wd 6.5 5.5 4.5 1.8 2.3 2.8 3.3 3.8 R(CO1/a0 Fig. 1 CO-stretching potential curves for selected states of H,CO (full line, C,, symmetry; dotted lines, C, symmetry). Composition of the 2lA, state: n, 3p, (left minimum) and n, n* (right minimum). Composition of 2 'A' near equilibrium: 0, n* (45%) and n, n* (30%). The dissociation products CH,(8 3B,) + O(3P,)are also shown. The 1 'B,(n, 3s) minimum lies at 7.08 eV, compared with an experimental To, value of 7.09 eV.1*4-971 5,1 By keeping all other geometrical parameters fixed at the ground-state values, the corresponding R(C0) potential curve shows a maximum at approximately 3.05 a, and 8.9 eV (Fig. 1). Close to this maximum, the 1 'B, state undergoes a change in struc- ture, from n, 3s (bound, left side) to n, CT& (repulsive, right). Energetically, this C,, dissociation barrier lies ca.1.8 eV above the left minimum, or 1.25 eV above CH, + 0. Since this change in structure takes place geometrically and energetically far away from the n, 3s minimum, the exci- tation n -,CT:~can be ruled out as causing the S, (Rydberg) band anomalies. Moreover, it can be inferred that disso- ciation of H,CO into CH, + 0 via the vertical absorption S, tSo should involve intermediate valence configurations on the S, surface of lower than C,, symmetry (i.e. more accessible energetically). One is thus compelled to look for other low-lying valence states which may interact with the Rydberg portion of the S, surface. A closer look at the C,, potential curves in Fig.1 points out that 1 'B,(n, 3s) is crossed by 2 'A, and 1 'B, at R(C0) shorter (2.6-2.7 a,) and energies lower (8.1-8.3 eV) than cor- responding values of the 1 'B, maximum. It is suggested that either 2 'A, or 1 'B, (or both) are involved in the process So S, -+ CH, + 0 rather than the higher-lying excita- tion n -+O~~('B,). 2 'A, State As shown in Fig. 1, our study corroborates earlier predictions by Schaefer and co-~orkers~~ about the double-minimum character of the 2 'A, potential. This feature results from an avoided crossing near R(C0) = 2.55 a, between the non-adiabatic states n, 3py (Rydberg, left side) and n, n*/n2 (valence, right). The present MRD-CI data place the inner and outer minima at essentially the same energy, i.e.7.98 and 7.95 eV, respectively. In the valence region of 2 'A1, with increasing R(C0) the contribution of n2 to n, n* increases. For example, at R = 2.65 a,, one finds 50% n, n* and 25% n2; at R = 3.0 a, it is 45% n, n* and 40% n2; and at R = 4.0 a, 10% n, n* and 685 50% n2. The valence configuration CT,gz0 starts to contribute for R > 3.5 a,; at R = 4.0 a,, for instance, the 2 'A, state has ~~30% u, CT character. In other words, slightly above R = 2.5 a, the 2'A1 state is mainly of R, n* type, but at larger R it acquires a predominantly 'closed-shell' . . n2 character. The existence of such a heavy mixing between both configurations was first noticed by Allen and S~haefer.~" Related to this is the change in electronic structure under- gone by the ground state.Near equilibrium, xlAl has ca. 70% 'closed-sheT1' * * . n2 character, but the relative weight of this configuration decreases almost linearly with R(CO), to ca. 45% at R = 3.0 a,, and to ca. 10% at R = 4.0 a,. Com-plementarily, the 'open-shell' n, n* configuration increases its contribution to RIAl from 15 to 45 to 65% at the same R(C0)distances. In conclusion, for R(C0) >2.5 a, the R'A, and 2'A1 states are strongly coupled (non-adiabatically). The highest degree of mixing between -. . n2 and nn* occurs near 3.0 a,, close to the 2 'A, (pseudo) minimum. At that geometry both states are energetically separated by ca. 5.4 eV, corresponding to a region in which the ground state is vibrationally excited by ca.2.6 eV in the v,(CO) mode (roughly nv2 z 15). Since both minima of 2 'A, lie energetically slightly above the CH,+ 0 channel, a non-radiative decay of this excited state into the ground state may consequently cause the rupture of the CO bond. Other electronic states interacting with the n, n* configuration are also indirectly coupled with the ground state, so that the phenomenon of predissociation into CH, + 0" might affect several states of H,CO lying above 7.5 eV*l5,17 1'B, State The 1 'B,(o, n*) minimum lies at 8.27 eV [only 0.32 eV above that of 2 'A,(n, n*)] with an equilibrium R(C0) of 2.68 a,, ca. 0.4 a, longer than for the ground state but ca.0.24 a, shorter than for 2 'A1(n, R*). Other theoretical studies have estimated a C,, transition energy T,(1 'B,) also around 8.30 eV (Table 2). Earlier calculations by Buenker and Peyerimhoff 31 demon-strated that the planar minimum of 1 'B, actually corre-sponds to a saddle point as the configuration u,n* stabilizes into a pyramidal equilibrium structure through the out-of- plane bending mode v4. Moreover, a concomitant stabilizing effect with important photochemical consequences takes place by such a symmetry lowering. The C,, states 2 'A, and 1 'B, can mix since both species correlate with 'A' of C,(xz). A great deal of stabilization is expected to occur relative to the C,, potentials since both configurations o,n* (B,) and n, R* (A') favour non-planar conformations, and in addition their energetic separation is relatively small.CO-stretchPotential Curves for Pyramidal H,CO <c,Symmetry) 2 'A' State The optimized geometry at the MRD-CI level (Table 1) com-pares fairly well with that obtained by Foresman et aL2' and .~~Hadad et ~1 The only significant discrepancy between our results and the other studies lies in the HCH angle (ca. 10" deviation). When compared with the ground state, the major changes in geometry correspond to R(C0) and out-of-plane angle 0, with AR(C0) = 0.6 a, and A0 = 44".Since 2 'A' cor-relates with the 1 'B,(o, n*) and 2 'A1(n, n*) states [with R(C0) minima at ca. 2.7 and 3.0 a,, respectively], a relatively large R(C0) of ca. 2.9 a, predicted for 2 'A' is understand- able.Interestingly, the 2 'A'(S,) equilibrium geometry is not too different from that of the S, minimum, 1 'A"(n, n*).The data in Table 1 indicate that the major structural difference between them lies in R(C0) with AR(C0) x 0.39 a,. As shown in Fig. 1 (dotted potential curve), the 2'A' state lies well below 1 'B, and 2 'A, [for R(C0) > 2.4 a,]. Our data place the 2IA' minimum at 7.53 eV relative to the ground state, i.e. ca. 0.45 eV above 1 'B,(n, 3s) or ca. 0.10 eV below CH, + 0. Recent CIS (single-excitation) and MP2 studies by Fore- .~~ .~~sman et ~1 and Hadad et ~1 assigned to 2 'A' pure a, n* character and higher excitation energies (7.99 and 8.49 eV, Table 2). Both studies place the Rydberg states up to 1.4 eV too high, rendering the energetics useless for quantitative comparisons.The CO-stretching potential of 2'A' exhibits a rather broad minimum, a feature reflecting the mixed valence char- acter of this state. For instance, 2 'A' at equilibrium consists of 45% a, n*, 30% n, n* and 10% * -n2. At R(C0) = 3.6 a,, this state is mainly described by 8,n*, with some n'. At R(C0) < 2.4 a,, the 2 'A' state assumes n, 3py character; since this Rydberg configuration favours a planar conforma- tion, it is understandable why the 2 'A' potential lies above 2 'A, for R < 2.4 a, (Fig. 1). The relatively high contribution of n, n* to the 2 'A' state is of spectroscopic importance. Although the 2 'A' minimum lies practically at the same energy as the CH, + 0 products, both regions are separated by a huge energy barrier (Fig.1). In other words, despite the fact that the 2 'A' S, minimum is connected adiabatically with ground-state CH, + 0, a large amount of vibrational energy is needed to reach dissociation. Nevertheless, rupture of the CO bond can proceed without any additional vibrational excitation of 2'A' by the non- radiative decay S, -+ So, a process in which the energy-rich state S, (2 'A') releases its electronic energy into nuclear motion (CO-stretching in particular) of the ground state. Dipole Moments The ground state has an experimental dipole moment p = -2.33 D,32 with a polarity (H,C)+O-. The present MRD-CI study slightly overestimates this quantity, giving p = -2.56 D with 8 'A, SCF MOs and p = -2.42 D with natural orbitals.Other dipole moments studied here should similarly be accurate to ca. f10%. For 1 'B,(n, 3s) at equilibrium, we obtain p = +3.45 D, corresponding to a reversed polarity (H,C)-O+. This rela- tively high p value confirms an earlier GVB-CI result of +3.068 D.33Both theoretical estimates are at variance with an experimental result of +0.33 D obtained by electro- chromism determinations. l4 However, all studies agree in that the n -+ 3s excitation is accompanied by a shift of elec- tron density from oxygen to the CH, group. Quite interestingly, measurements of the Stark effect in H,CS led to dipole moments of -1.649 D for X'A, and +2.2 D for 1 'B,(n, 4s).4b These experiments indicate that a ground-state polarity (H,C)+S- is reversed to (H2C)- S+ in the 1'B, Rydberg state, a similar behaviour as in H,CO.Moreover, the relative change Ap/p(R 'A,) is of comparable magnitude in both molecules. For H,CS it is 2.33 (e~perimental),~'and for H,CO it is 2.43 (this work), 2.30 (ref. 33) and 1.14 (e~perimental'~). This discrepancy between theory and experiment in H,CO could be settled by carrying out on H,CO the same type of Stark-broadening measurements as have been done on H,CS.46,34 According to Li~tay,~' the major difference between dipole moments derived via Stark effect or electro- chromism resides in the procedure used to evaluate the experimental data (in both techniques the absorption spec- trum is analysed under the influence of an external electric J.CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 field). While the Stark effect focuses on the broadening and splitting of certain lines, electrochromism measures the field dependence of the molar absorption coefficient (i.e. band intensity). In other words, the variation of the transition moment with the applied field constitutes the key quantity for electrochromism, but is irrelevant for the Stark effect. The electrochromism experiment^'^ detected a pertur-bation on 1'B,(n, 3s) caused by a nearby state (assumed to be of 'B, symmetry, too) as the field strength was varied. In fact, 2 'B,(n, 3pz) lies only 0.9 eV above 1 'B,, so that 3s-3pZ mixing can occur. The dipole moment of 2 'B, has been pre- dicted to be ca. -2.2 D1*and -1.83 D,33 i.e.slightly smaller than for the ground state but having the same polarity (H,C)+O-. Assuming that the applied field induces signifi- cant 3s-3pz mixing, the dipole moment derived for 1'B, via electrochromism should not be as strongly (H,C)-O+ pol- arized as theory predicts because of the large contribution of the (H2C)+O- structure from 2 'B,. At its pyramidal minimum, the predicted dipole moment of 2 'A' is -3.47 D, with a polarity (H,C)+O- like the ground state or the planar 2 'A, state.30 The vector p has an out-of- plane angle 8 = 34", almost parallel to the CO bond (e = 440). This result points out the extent of the electronic charge reorganization taking place between the 1'B, (Rydberg) and 2 'A' (valence) states. Since we obtained p(1 'B,) = +3.45 D, it follows that in going from 1 'B, to 2 'A' the polarity is reversed, corresponding to a total change of ca. 6.9 D.This is understandable since for 1 'B,(n, 3s) one has a H,CO+ core plus a spatially extended 3s electron, whereas for 2 'A' (a, n*; n, n*) the electron cloud is obviously more compact. It is worth emphasizing that this drastic polarity reversal is caused primarily by a shift of the CO electron density from C-0' to C'O- for an increment AR(C0) of only 0.6 a,. Other electric properties of S, (quadrupole moment, pol- arizability etc.) are also expected to change significantly along the reaction path connecting the minima of 1'B, and 2 'A'.The same behaviour is expected for the transition moment of S, + So,so that the intensity distribution of this system is no longer governed by the Franck-Condon approximation. Discussion Earlier speculations about the existence of a nearby state per- turbing 1'B,(n, 3s) at 7.09 eV are confirmed by the present MRD-CI study. At equilibrium, the valence state 2 'A' (mixed a, n* and n, K*) lies only 0.45 eV higher than 1'B,. The n, a&( 'B,) state, thought by experimentalists as low-lying and perhaps responsible for some of the S, t So irregularities is seen to cross 1'B,(n, 3s) at energies and geometries far removed from the 1'B, minimum, and does therefore not qualify as perturber. Since the excitation spectrum of H,CO has been recorded several time~,'*~-~*l it appears somewhat astonishing that 'a1 none of these experiments claims the existence of a valence state (ie.2 'A') lying adiabatically near 7.53 eV (60 740 cm-'). However, this apparent failure is in part understandable since direct absorption 2 'A t2 'A, should consist of a long pro- gression of low-intensity peaks. Regardless of the magnitude of the transition moment, our statement is justified by the rather small Franck-Condon factors (FCFs) expected for a transition having such large changes in geometry. Note that the valence portion of the S, surface (i.e. 2'A') might have been detected indirectly during the Ar high- pressure experiments carried out by Messing et al." via verti-cal absorption into the Rydberg portion of S, .According to these authors, the n -+ 3s transition at 57 192 cm- ' reveals a vibrational structure which persists up to the liquid density; J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 moreover, the band becomes rather broad (half width ~3000 cm-') and the absorption maximum is blue shifted (ca. 2500 cm-') as the Ar pressure is increased. A broad band extend- ing from roughly 58000 to 61 250 cm-' (7.19 to 7.59 eV), with a maximum at 59345 cm-' (7.36 eV) compares quite well with the present theoretical estimate of T, = 60 740 cm-' (7.53 eV) for the 2 'A' tj?: 'A, excitation. The broadness of the band can also be correlated with the different equilibrium conformations of the upper and lower states. The discussion above sheds light on what is actually taking place in these high-pressure experiments : the non-verticality of the 2 'A + 'A, transition (poor FCFs) is partly compen- sated for by the vertical Rydberg excitation 1 'B, tj?: 'A, (favourable FCFs).This initial Rydberg population might evolve into valence population through internal conversion. The population flow from 1 'B, to 2 'A' is favoured with increasing Ar pressure by at least two effects: First, the S, surface gradually loses its Rydberg character (which ener- getically implies higher absorption energies) and secondly, the collision frequency increases, thereby facilitating intermode vibrational couplings. At this point, an alternative experimental route for directly studying the valence region of S, can be envisaged, namely by optical absorption from the S, (1'A", nn*) s~rface.~ As shown in Table 1, the states 2 'A and 1 'A have comparable equilibrium geometries and consequently their Franck-Condon factors are more favourable than those between 2 'A' and 8 'A,.The feasibility of experimentally observing the excitation 2 'A t 1'A (S, tS,) is additionally supported by a relatively long radiative lifetime of ca. 10 ps assigned to 1 'A" (S,).' However, a dificulty may arise from the concur- rent decomposition process s, + H, + CO,'., which could significantly reduce the 1'A population available for absorption. Fortunately, formaldehyde isolated in solid rare- gas matrices does not decompose into H, + CO after S, +-So absorptionL6 (as the coupling between S, and the disso- ciation continuum is essentially eliminated).Since the matrix cage at the same time suppresses the Rydberg states, the most practical experimental route to detect 2 'A' appears to be by 2 'A't 1 'A" absorption in a matrix. On the basis of an experimental T,(1 'A) = 3.50 eV,' the valence-valence tran- sition s, ts, is expected near 4.0 eV (ca.32 OOO cm-'). As the stationary minima of 1'B, and 2 'A' both lie on S, , the question arises about the vibrational mechanism connect- ing these two regions. The main dilemma to be solved is the following. By lowering the symmetry from C,, to C,(xz) via the v4 mode (out-of-plane), the 'B, planar state correlates with 'A" but not with 'A.The only possibility to obtain an intermediate conformation having a spatial symmetry which correlates with both 1'B, (C2J and 2 'A' (C,, xz) is to work within the C, point group. Such an intermediate C, structure can be visualized, for instance, by distorting pyramidal H,CO (C,, xz) through the two antisymmetric a" modes, i.e. CH-stretch and CH,-rocking. Similarly, starting from planar H2C0 (C,,), the C, complex can be obtained by the simultaneous activation of the antisymmetric vibrations b, (v4,) and b, (v5 or v6, or both). Moreover, since the CO equilibrium distances are quite different for 1 'B, and 2'A', the v2(C0) stretching motion should also play an important role along the path c,,-b c, --* C,(xz). In short, internal conversion between both S, minima through a C, intermediary requires the simultaneous activa- tion of several vibrational modes : CO-stretch, out-of-plane bend and at least one of the CH- and CH,-antisymmetric distortions.The existence of a strong coupling between various modes justifies the complex structure of the Rydberg portion of the S, tSo spectrum and the difficulties found by experimentalists in obtaining a satisfactory vibrational a~signment.~,~,~Obviously, the double-minimum topology of the S, surface of H,CO does not exactly match with the single minimum of the ground-state surface of H2CO+,18 a feature also explaining why the experimental band profile of the S, tSo transition is much broader and more complex than that of the photoionization spectrum H,CO+ tH,CO. Assuming that the 2 'A' basin of S, is populated indirectly from 1'B, via the vertical absorption 1'B, +-ft 'A1, several decay processes may act upon 2 'A'.For instance, radiative emissions such as 2'A'+ 1 'A" (AEoo~4.0eV) and 2 'A' -+ 'A, (AllOOx 7.40 eV) are possible. The former process is expected to generate vibrationally excited 1 'A" molecules, which in turn may decompose into H + HCO or H, + CO. On the other hand, the non-verticality of the emis- sion 2 'A' -+ j?: 'A, would generate So molecules vibrationally excited up to roughly 3 eV. Since such internal energy is smaller than the barrier for the thermal decomposition So +H, + CO,' this emission process would allow the study of vibrational relaxation within the So surface.Parallel to these radiative emissions, the 2'A' state may decay non-radiatively into 8 'A,. This process will lead directly to CH, + 0 fragments. Summary and Conclusion The observed spectrum for the transition 1 'B,(n, 3s) t% of H,CO shows diffuseness and a complex vibrational structure. Previously, the perturbing state was assumed to be n, a&. However, the CI calculations presented here find n, a& to be very high in energy. Instead, we propose the perturbing state to be the non-planar 2'A', consisting of a, n* mixed with n, n*.Its minimum is calculated to lie only 0.45 eV above the n, 3s minimum, shifted towards a larger CO distance by 0.6 a,. Since in pyramidal C, symmetry 'B, transforms into 'A", the internal conversion 1'B,(n, 3s) -+ 2 'A'(a, n*/n, n*) can only take place by further lowering the symmetry to C,.It is shown that in C, symmetry, 1 'B,(n, 3s) and 2'A'(a, 7c*/7c, n*) lie on the same surface, the third singlet surface commonly designated by S,. On that surface, a low barrier, estimated to be <0.7 eV above the 1 'B, minimum, exists between the two minima. This has to be overcome by vibra- tional excitation. Since the internal conversion 1 'B, + 2 'A' can only take place in C, symmetry, participation of anti-symmetric vibrational modes v5 and v6 (b,)and the out-of- plane motion v4 (b,) are expected, as well as the C-0 stretch motion v2 due to the required change in C-0 distance, and most likely also v1 and v3 (all a').Here the C,, notations are used. This explains the complex vibrational structure of the transition in question, and the observation of many, if not all, vibrational modes of H,CO, After internal conversion from 1 'B,(n, 3s) to 2 'A' has taken place, coupling with high vibra- tional levels of the ground-state %'A, may lead to disso- ciation into the products CH, + 0, which lie energetically slightly above the minimum of 2 'A'. Since both the ground state and the 2'A' state contain part of the n, n* configu-ration, the two states can couple non-adiabatically. The blue shift observed for the 1'B, tj?: 'A, system under higher pressure [also observed for CH3HC0 and (CH,),CO] can similarly be explained by internal conversion from the Rydberg state 1 'B, to the valence state 2 'A'.An experimental route to studying the valence region of S, ,by optical absorption from 1'A"(n, n*),is proposed. The authors thank NSERC (Canada) for financial support and Computing Services of the University of New Brunswick for the allocation of computer time. 688 J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 References 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 H. Okabe, Photochemistry of Small Molecules, Wiley, New York, 1978; C. B. Moore and J. C. Weisshaar, Annu. Rev. Phys. Chem., 1983,34,525. J. D. Goddard and H. F. Schaefer, J. Chem. Phys., 1979, 70, 5117; M. Dupuis, W. A. Lester, B. H. Lengsfield and B. Liu, J. Chem. Phys., 1983, 79, 6167; G. E. Scuseria and H. F. Schaefer, J. Chem. Phys., 1989,90, 3629, and references therein.D. L. Baulch, R. A. Cox, R. F. Hampson, J. A. Kerr, J. Troe and R. T. Watson, J. Phys. Chem. Ref. Data, 1984,13,1259. (a) D. C. Moule and A. D. Walsh, Chem. Reo., 1975, 75, 67; (b) D. J. Clouthier and D. A. Ramsay, Annu. Rev. Phys. Chem., 1983, 34, 31; (c)D. J. Clouthier and D. C. Moule, in Top. Curr. Chem., Springer Verlag, Berlin, 1989, vol. 150, pp. 167. G. Fleming, M. M. Anderson, A. J. Harrison and L.W. Pickett, J. Chem. Phys., 1959,30,351. J. E. Mentall, E. P. Gentieu, M. Krauss and D. Neumann, J. Chem. Phys., 1971,555471. M. T. Weiss, C. E. Kuyatt and S. Mielczarek, J. Chem. Phys., 1971,54,4147. C. R. Lessard and D. C. Moule, J. Chem. Phys., 1977,66,3908. P. Brint, J. P. Connerade, Ch. Mayhew and K. Sommer, J. Chem. SOC.,Faraday Trans.2,1985,81,1643. W. C. Price, J. Chem. Phys., 1935,3,256. I. Messing, B. Raz and J. Jortner, Chem. Phys., 1977, 23, 351; 1977,25, 55. M. B. Robin, Higher Excited States of Polyatomic Molecules, Academic Press, New York, 1974 and 1985, vol. 1-111. A. D. Walsh, Proc. R. SOC. London, Ser A, 1946, 185, 176; G. Lucazeau and C. Sandorfy, J. Mol Spectrosc., 1970,33,274. G. C. Causley and B. R. Russell, J. Chem. Phys., 1978, 68, 3797; J. Am. Chem. SOC., 1979,101,5573. S. Taylor, D. G. Wilden and J. Comer, Chem. Phys., 1982, 70, 291. B. Niu, D. A. Shirley and Y. Bai, J. Chem. Phys., 1993,98,4377, and references therein. 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 M. Hachey, P. J. Bruna and F. Grein, to be published. T. H. Dunning, J. Chem. Phys., 1971,55,716. T. H, Dunning and P. J. Hay, in Modern Theoretical Chemistry, ed. H. F. Schaefer, Plenum Press, New York, 1976, vol. 3. B. Roos and P. Siegbahn, Theor. Chim. Acta, 1970,17,199. K. H. Thunemann, S. D. Peyerimhoff and R. J. Buenker, J. Mol. Spectrosc., 1978,70,432. R. J. Buenker, S. D. Peyerimhoff and W. Butscher, Mol. Phys., 1978, 35, 771, and references therein; D. B. Knowles, J. R. Alvarez-Collado, G. Hirsch and R. J. Buenker, J. Chem. Phys., 1990,92,585. R. J. Buenker, S. D. Peyerimhoff and P. J. Bruna, in Computa-tional Theoretical Organic Chemistry, ed. 1. G. Csizmadia and R. Daudel, NATO AS1 C67, Reidel, Dordrecht, 1981. J. L.Duncan, Mol. Phys., 1974,28, 1177. D. Feller and E. R.Davidson, J. Chem. Phys., 1984,80,1006. S. Bell and J. S. Crighton, J. Chem. SOC., Faraday Trans. 2,1985, 81, 1813. J. B. Foresman, M. Head-Gordon, J. A. Pople and M. J. Frisch, J. Phys. Chem., 1992,%, 35. Ch. M. Hadad, J. B. Foresman and K. B. Wiberg, J. Phys. Chem., 1993,97,4293. (a) G. Fitzgerald and H. F. Schaefer, J. Chem. Phys., 1985, 83, 1162; (b)W. D. Allen and H. F. Schaefer, J. Chem. Phys., 1987, 87, 7076. R. J. Buenker and S. D. Peyerimhoff, J. Chem.Phys., 1970, 53, 1368. B. Fabricant, D. Krieger and J. S. Muenter, J. Chem. Phys., 1977,67,1576. L.B. Harding and W. A. Goddard, J. Am. Chem. SOC., 1975,97, 6293. W. Goetz, D. C. Moule and D. A. Ramsay, Can. J. Chem., 1981, 59, 1635. W. Liptay, in Excited States, ed. E. C. Lim, Academic Press, New York, 1974, vol. I. 17 M. Suto, X.Wang and L. C. Lee, J. Chem. Phys., 1986,85,4228. Paper 3/06639H; Received 5th November, 1993