Photodissociation of carbon dioxide in singlet valence electronic states. I. Six multiply intersecting ab initio potential energy surfaces
Abstract
The global potential energy surfaces of the first six singlet electronic states of CO2, 1—3 and 1—3 are constructed using high level ab initio calculations. In linear molecule, they correspond to , , , and . The calculations accurately reproduce the known benchmarks for all states and establish missing benchmarks for future calculations. The calculated states strongly interact at avoided crossings and true intersections, both conical and glancing. Near degeneracies can be found for each pair of six states and many intersections involve more than two states. In particular, a fivefold intersection dominates the Franck-Condon zone for the ultraviolet excitation from the ground electronic state. The seam of this intersection traces out a closed loop. All states are diabatized, and a diabatic potential matrix is constructed, which can be used in quantum mechanical calculations of the absorption spectrum of the five excited singlet valence states.
I Introduction
This and the subsequentG13B paper (termed ‘paper II’) describe the results of an ab initio quantum dynamical study of the absorption spectrum and the non-adiabatic dissociation mechanisms of carbon dioxide photoexcited with the ultraviolet (UV) light between 120 nm and 160 nm. A brief account of this work has already been published.G12A Paper II gives the motivation behind the study and discusses its main result — the quantum mechanical absorption spectrum and its interpretation in terms of wave functions of metastable resonance states. The present paper sets the stage for paper II and describes the ab initio calculations and the topography of the potential energy surfaces (PESs) involved in photodissociation. This information, which is often stashed away in supplementary online sections,G12A ; GB12 is a central and, indeed, an indispensable ingredient of a reliable dynamics calculation. The construction of ab initio PESs and their diabatization — which, without much exaggeration, amounts to learning the topography of PESs and their intersections by heart — is often more challenging than the subsequent quantum dynamical calculation.
Photoabsorption from the ground electronic state of linear CO2 at wavelengths 120 nm — 160 nm is due to the first five excited singlet valence states , , and .HERZBERG67 ; RMSM71 ; OKABE78 In the group notation, appropriate for bent molecule, these states are and . UV light excites CO2 into the region of multiple electronic degeneracies, nuclear motion through which induces strong non-adiabatic couplings between electronic states. These couplings directly affect the observed absorption spectrum of the valence states and control distributions of photofragments over the final states. Their indirect influence apparently extends to shorter wavelengths where Rydberg transitions dominate: The combined experimental and theoretical analysis indicatesCJL87 that the manifold of coupled valence states acts as a ‘sink’ for the optically bright Rydberg states and affects their dissociation lifetimes.
Electronic degeneracies in the Franck-Condon (FC) region, which have been the focus of several studies in the past,EE79 ; KRW88 ; SFCCRWB92 ; G12A ; GB12 are of two types. Glancing intersections occur in the orbitally doubly degenerate and states which upon bending split into and components. Conical intersections (CIs) arise at the accidental crossing inside and symmetry blocks.KRW88 ; SFCCRWB92 ; G12A ; GB12 In fact, CIs between valence states are ubiquitous and found far outside the near-linear FC region, at strongly bent geometries.SFCCRWB92 ; G12A ; GB12 Together with local minima and saddles, these crossings are the principal features shaping the topography of the singlet valence states.
The outline of the paper is the following. Section II sketches the technical details of ab initio calculations. Next, the constructed adiabatic PESs are presented for the ground (Sect. III.1) and the excited (Sect. III.2) electronic states. Their intersections are discussed for near linear (Sect. III.3) and bent geometries (Sect. III.4), and put into perspective by a review of a network of closely spaced valence and Rydberg states (Sect. III.5).
If the solution of the Schrödinger equation for nuclei is out of reach, the description of adiabatic surfaces in Sect. III would be the last step in theoretical ab initio analysis. If, on the other hand, one intends to treat nuclear dynamics quantum mechanically, ab initio PESs featuring CIs have to be diabatized.C04 This is especially desirable if — as in paper II — a discrete grid is used to represent the nuclear Hamiltonian, because the diabatized potential matrix is free from either divergent off-diagonal couplings or non-differentiable potential cusps. While general schemes for constructing approximate adiabatic-to-diabatic transformations are established (see, for instance, Ref. K04, ), their application to the valence states of CO2 is complicated by the number of CIs to be simultaneously treated. Simplifications are called for, as described in Sect. IV, in which the diabatic representation is constructed separately for bent CIs (Sect. IV.1) and linear CIs (Sect. IV.2). Section V concludes.
II Electronic structure calculations
All ab initio calculations are carried out with the MOLPRO package.MOLPRO-FULL The Gaussian atomic basis sets used in this work are due to Dunning.D89 Previous studies indicate that diffuse functions should be added to the basis sets on oxygen and carbon atoms in order to account for the mixed valence-Rydberg character of the state.KRW88 ; SFCCRWB92 ; G12A A series of tests was conducted, in which , , etc. basis functions of triple and quadrupole zeta quality were selectively augmented by one or two diffuse functions. The pre-computed doubly augmented correlation consistent polarized valence quadrupole zeta (d-aug-cc-pVQZ) basis set,D89 as implemented in MOLPRO, was found computationally most stable and selected for calculations of global PESs.
Three-dimensional (3D) PESs of states and are calculated at the internally-contracted multireference configuration interaction singles and doubles (MRD-CI) level, based on state-averaged full-valence complete active space self-consistent field (CASSCF) calculations with 16 electrons in 12 active orbitals and 6 electrons in three fully optimized closed-shell inner orbitals. The electronic configuration of the ground state is . The dominant electronic excitations, leading to the lowest excited states, include (giving states and ) and (giving state ). Active orbitals in CASSCF comprised , ,, and . In symmetry, used in the calculations, these are and . In the MRD-CI step all 16 valence electrons were correlated. The maximum numbers of open shells allowed in the MRD-CI calculations were 8 in the reference space and 12 in the internal space. This lead to 38159928 contracted configurations. The Davidson correction was applied in order to account for higher-level excitations and size-extensivity.LD74
Adiabatic energies are calculated on a 3D grid of the two C–O bond lengths and the OCO bond angle : (step size equals 0.1 ), (step size varies between 0.1 and 0.4 ), (step size varies between 2∘ and 10∘). Additionally, many cuts in the plane are computed for angles between 60∘ and 0∘ within the continuing effort to construct a balanced description of both CO + O and C + O2 arrangement channels. At present, the grid comprises 4800 symmetry distinguishable points. The resulting energies were scanned in one and two dimensions for obvious errors. The list of corrected adiabatic energies was subsequently interpolated using 3D cubic splines and also used for constructing the quasi-diabatic representation. Missing energies for in the dissociation channels were obtained from those for using trigonometric extrapolation.
Absolute intensity calculations of paper II require transition dipole moments (TDMs) with the ground state . Components of the TDM vector are calculated, for each electronic state, on a 3D grid and covering the spot over which the vibrational ground state in is delocalized. The molecular axes in these calculations are chosen such that is orthogonal to the molecular plane, runs along one of the CO bonds and . For states, the in-plane components are generally non-zero, while for states, it is the component which carries the transition.
III Properties of the valence PES and their crossings
III.1 Ground electronic state
The eV deep adiabatic ground state PES supports three structural isomers: The familiar linear OCO molecule is the global equilibrium, while the carbene-like bent OCO and the linear COO are the two local ones. Table 1 summarizes the characteristic features of the state at the three equilibria and compares them with the previous ab initio studies and with the available experimental data.
The vicinity of the global minimum is of capital importance for the environmental chemistry.P11 The calculated equilibrium CO bond distance in linear OCO, , agrees well with the experimental value of 2.1960 . The accuracy of the vibrational zero-point energy (ZPE) and the vibrational transitions frequencies is assessed in Table 2 which compares energies of the low lying vibrationally excited states in rotating CO2 (the total angular momentum ) with experimentC79A and with recent electronic coupled cluster/vibrational configuration interaction calculations.RHYHTST07 ; YHH08 Each eigenstate is labeled using the quantum numbers of the symmetric stretch , the bend (with indicating the vibrational angular momentum, ), and the antisymmetric stretch . The calculated fundamental frequencies of the infrared active bend ( cm-1) and antisymmetric stretch ( cm-1) are accurate to within 1.5 cm-1. The zeroth order symmetric stretch frequencyNISTDATABASE1 cm-1 is about twice as large as the bending frequency , and the two modes are involved in the accidental Fermi resonance.F31 As a result, the vibrational spectrum is organized in polyads with the polyad quantum number ; states with , 3 and 4 are given in Table 2. In the original version of the PES, called ‘PES1’ in Table 2, the energies of states and , belonging to the lowest polyad , are underestimated by 20 cm-1 and the difference with the observed energies grows rapidly with . This systematic discrepancy is substantially diminished by slightly rescaling the symmetric stretch, , and the bend via
The vibrational energies in the scaled ‘PES2’ agree with their experimental counterparts to within 7 cm-1 and for the most states below 3000 cm-1 the accuracy is better than 3 cm-1. The results outperform even the highly accurate calculations of Refs. RHYHTST07, and YHH08, shown in the second column of Table 2, making ‘PES2’ one of the best available ab initio potentials of the state. Since the coordinate dependent dipole moment has also been calculated, the ab initio intensities of the infrared rovibrational transitions can be directly evaluated.G13C
The other two isomers in Table 1 have never been detected in the gas phase, and the only reference data stem from the previous ab initio studies. For the bent OCO, discovered by Xantheas and Ruedenberg,XR94 the present calculations confirm the symmetric equilibrium with the CO bond lengths of 2.51 and the OCO bond angle of 73.2∘. This minimum is located 6.03 eV above the global one, again in good agreement with the previous findings.XR94 ; HM00A The fundamental excitations in the OCO well, calculated for , are cm-1, cm-1, and cm-1. For the linear COO, the calculated CO and OO bond lengths are identical to the ones given in Ref. HM00A, ; both are elongated compared to free diatoms ( vs. 2.14 for CO and 2.45 vs 2.28 for OO). The calculations place the COO minimum at 7.35 eV, about 0.1 eV below the lowest dissociation threshold.
The ground electronic state correlates adiabatically with two dissociation channels,
(1) | |||||
(2) |
and the ZPE corrected dissociation energies are shown in Table 1. In channel (1), the calculated is 0.13 eV less than the experimental value. The deviation might reflect a large basis set superposition error introduced by the diffuse functions and as such is the downside of the highly accurate vibrational spectrum in Table 2. The error is independent of the arrangement channel, and in channel (2) [closed between 120 nm and 160 nm] is equally underestimated. The calculations of Hwang and MebelHM00A , using a noticeably smaller basis set, perfectly agree with the experimental dissociation energy for this channel.
One-dimensional (1D) cuts through the ground state PES are given for several angles in panels (a,c,e) of Figs. 1 and 2. Black solid circles are the raw adiabatic energies. The O + CO limit is reached smoothly and no barrier is detected towards the asymptote for any orientation of the CO diatom. The same is true for the C + O2 channel, as illustrated in Fig. 2(e); the potential well in Fig. 2(e) is the COO isomer. Angular dependence of the state is shown in panels (a,c,e) of Figs. 3 and 4 for two sets of fixed CO bonds. In Fig. 3, is fixed at the FC value; in Fig. 4, it is fixed close to the equilibrium of the bent OCO. Consequently, although the carbene-type minimum is perceptible in all panels, it is best seen in Fig. 4. As CO2 bends, the adiabatic state (black dots) forms a sharp narrowly avoided crossing with the state around (see, for example, Fig. 3). A dynamically meaningful representation for such nearly degenerate pairs is diabatic rather than adiabatic. In fact, the black line in Figs. 1 — 4 depicts the state locally diabatized at bent geometries as described in Sect. IV.1. This is the reason why the black dots sometimes switch away from the black line and why lines of different colors cross. In this locally diabatic picture, the bent OCO minimum correlates with the state [purple line]. The transition state separating the bent and the linear minima, analyzed by Xantheas and RuedenbergXR94 and Hwang and MebelHM00A is thus the signature of this two-state intersection. The two-dimensional (2D) contour plots of the PES of the locally diabatic state, used in the above calculations of vibrational states, are shown in Fig. 5.
III.2 Overview of the excited electronic states
Potentials of the excited electronic states are shown in Figs. 1 and 2 along one CO bond for and symmetries. As with the state, the solid circles indicate ab initio adiabatic energies. In the CO + O arrangement channel, which is the focus of the present investigation, all calculated states but one converge to the dissociation threshold (1). The state , correlating with at the FC point, reaches the higher lying threshold
(3) |
In the C + O2 arrangement channel [Fig. 2(e,f)], three electronic states, the state and the two components of the state, correlate with channel (2). Three other states ( and ) converge to the electronically excited fragments:
(4) |
Topographic hallmarks of the excited states can be exemplified using the states . In the FC region near linearity [Fig. 1(a)], the states and form two sharp avoided crossings near and . These crossings are in fact two CIs between states and .KRW88 ; SFCCRWB92 ; G12A ; GB12 The CIs are not independent: They are connected into a whole line, called a ‘CI seam’,Y04 unusual properties of whichGB12 are discussed in Sect. III.3. As the molecule bends, the gap between the adiabatic states grows. In the lower state , the intersection cone first turns into a broad barrier along the dissociation path [, Fig. 1(c)]. As decreases further, the local minimum near 2.3 deepens and the dissociation barrier disappears [Fig. 1(e) and Fig. 2(a)]. In linear COO [, Fig. 2(e)], the sharp avoided crossing between states reappears again, this time at . Similar to , the state supports a COO intermediate, although the local minimum lies 0.7 eV above the C + O2 threshold (2). The evolution of the uppermost state with decreasing angle is different, because its topography is very much influenced by a pronounced barrier located outside the FC zone near and separating the flat inner region from a steep decline towards the asymptotic limit. This barrier is distinct over a broad angular range and its sharpness suggests a CI with a higher lying state. The nature of this intersection becomes apparent in Sect. III.5 discussing valence/Rydberg crossings.
The states are similar in many respects. Near linearity, both symmetries mirror the topography of the orbitally degenerate states and , the states are involved in the same CIs in the FC region [Fig. 1(b)], and their behavior in the bent molecule up to closely follows that of . The state stabilizes with decreasing , while the state features a pronounced barrier around [Fig. 1(d,f) and Fig. 2(b)]. Similarities persist to [Fig. 2(f)]: The states form a CI at , and the state supports a local COO minimum. Differences with the symmetry are due to the state , which in the FC region correlates with and which has no counterpart among the calculated states. This state, which at linearity is accidentally degenerate with , has a single minimum near and approaches the dissociation limit (3) without a barrier. This simple shape is preserved through most cuts in Figs. 1 and 2. The state also supports a COO minimum [Fig. 2(f)].
Potential cuts along bending angle are shown in Figs. 3 and 4. The doubly degenerate and states split into and components for . Evolution of the electronic energies with decreasing was analyzed by Spielfiedel et al. using Walsh rules.SFCCRWB92 Based on this analysis, the adiabatic excited states are commonly classified as bent or linear. The energy of states and lowers as the molecule bends, while the energy of states and grows [see Fig. 3(a,b)]. The global minima of the ‘bent’ states lie near 120∘ () and 130∘ () and are located below the dissociation limit (1) [Fig. 3(a,b) and Fig. 4(a,b)]. Bent equilibrium of the state , which at geometries becomes , was predicted by Dixon in his analysis of the CO flame emission bands.D63 Finally, the state , which together with and is ‘linear’, is the least anisotropic of all states in a broad vicinity of .
Properties of various stationary points in the PESs of excited electronic states are given in Table 3. The point group notation is used to label adiabatic states; labels refer to the diabatic states at the FC point. Experimental reference data for the excited electronic states are scarce and fit into the footnote . For the bent states, the experimentalD63 ; CLP92 equilibrium CO distances and the bending angle are reproduced within and , respectively; the bending frequencies are accurate to within 30 cm-1 or better, and the calculated band origins lie within the experimental uncertainties. Results of the previous ab initio studiesKRW88 ; SFCCRWB92 are also shown in Table 3. Agreement in the equilibrium geometries is excellent, with the exception of the intersection-ridden uppermost states and , for which -restricted calculations of Refs. KRW88, and SFCCRWB92, miss the local minima in the upper CI cones. Vertical excitation energies are close to those of Ref. SFCCRWB92, . Slight differences are not surprising for near degenerate states whose ordering is sensitive to the details of the ab initio set up. Peculiarity of the spectral region 120 nm — 160 nm is that values are poor approximations to the positions of the absorption maxima because all transitions are electronically forbidden and the TDMs at the FC point are strictly zero. Finally, the ab initio dissociation energies in channels (1), (2), (3), and (4), also given in Table 3, agree with the experimental valuesHERZBERG67 ; AMS79 within eV irrespective of the particular arrangement or electronic channel.
III.3 Topography of state intersections in the FC zone
Intersections of electronic states in linear CO2 follow simple ‘symmetry rules’ KDC84 ; CMD84 which severely restrict
the intersection topography.
Two types of intersections with regard to their symmetry properties
are prominent in Figs. 1(a,b) and 3(a,b):
(1) Renner-Teller (RT) glancing intersectionsR34 ; HERZBERG67
involve the and components of the or
the state. In the rotating molecule, the
interaction
is proportional to the projections and of the electronic
() and the total angular momentum
()
on the molecular axis and diverges for
.R34 ; HERZBERG67 ; P88 ; GGH93
The transition
is best classified as linear-linear, and
the transition
is linear-bent. The ‘degeneracy manifold’ for the
intersecting states is
the whole plane defined by the condition
.
(2) Two nested CIs involve the and or states and stem from the accidental crossing. Both degeneracies are lifted linearly along the tuning and coupling modes spanning the common branching space.Y04 According to Fig. 1(a,b), the tuning mode is , i.e. a combination of the symmetric and antisymmetric stretch (their irreps are in the group); breaking the linear symmetry (irrep ) is the coupling mode. As a consequence of the ‘symmetry rules’, the CIs occur along a line in the plane at : The ‘degeneracy manifold’ is a 1D seam.HERZBERG67 ; T37 ; Y04
The CI seam, constructed separately for the and states on a fine grid,GB12 is depicted in Fig. 6. It has two remarkable properties. First, the intersection along the seam is fivefold. Two CIs and two RT intersections imply four degenerate states. The hitherto ignored state closely follows : The energy gap falls consistently below 300 cm-1 which is at the limit of the ab initio accuracy. Thus, the and the pairs cross at the same CO bond distances and the total degeneracy is five. Second, the calculated seam traces out a closed loop. Closed CI seams are rarely encountered,ARN97 ; GB12 although arguments have been devised to prove their ubiquity.LHVBZKB08 ; LHVBZKB09 A technical implication of closed or strongly curved seams is that local diabatization schemes failGB12 and global or semiglobal diabatizationKGM01 becomes necessary. The impact of closed seams on photodissociation dynamics has never been systematically investigated.GB12 In CO2, only a small portion of the loop is directly accessible to UV light. Nevertheless, paper II demonstrates that the seam topology deeply affects the observed absorption spectrum.
The curved intersection seam is responsible for a peculiar shape of the potentials in the FC region. The adiabatic PESs of the states in linear CO2 are shown in Fig. 6. The lower adiabatic state has a minimum at , and two symmetric saddles near or , separating the minimum from the dissociation asymptotics. Outside the area enclosed by the seam line, the state has the , inside the character. The saddles hide portions of cusp lines, along which the states intersect and which are washed out by the spline interpolation. The topography of the upper adiabatic state is a literal mirror image of the lower state; one finds a saddle at and two symmetric minima near and . The state character changes from outside the seam line to inside. The two minima are cusp-like and correspond to the upper cones of the CIs. The two other saddle points in the state, lying outside the FC zone near and , result from an avoided crossing with a higher lying state.
Changes of the electronic character of states across the intersection can be monitored using matrix elements of the electronic angular momentum .DK97 The matrix elements along the line passing twice through the closed seam are depicted in Fig. 7 for and . The states and act as ‘probes’ whose assignment in terms of or is known. Outside the seam area, most matrix elements are close to integer values of 0, 1, and 2. For for example, and so that is a state, while is a state. vanishes too, but the state is a state, as confirmed by the matrix element . As the CI seam is crossed at into the area interior to the seam loop, the three-state intersection induces violent changes in the electronic labels of the adiabatic states. The state becomes a state (); the state acquires character (), while the state acquires character (), and the non-integer values of the latter two reflect strong mixing of the near degenerate pair. The next reshuffling occurs as the seam loop is crossed outwards at . The state becomes again, while the matrix elements involving and vary until is reached, where finally vanishes indicating that emerges from the crossing region as a state converging towards the upper threshold (3), while becomes state correlating with threshold (1).
The positions of cusps, minima, and saddles in the PESs of the states at linearity are identical to those in states. Exception is the PES of the state , correlating in the FC region with . Its appearance, illustrated in Fig. 8, is very much simplified by it being tightly linked with the diabatic state. The PES has a single minimum at , is free from cusps seen in other adiabatic PESs, and provides a nature’s illustration of the shape of the diabatic PES.
CIs at linearity can also be recognized in the potentials plotted in the plane. An example including four electronic states at is shown in Fig. 9. Although the characteristic features are no longer symmetric, a maximum in the state at and and two minima in the state at and and are recognizable in panels (a) and (b). Close to linearity, the states [panels (c) and (d)] maintain the same contour maps as states.
III.4 Intersections and avoided crossings in bent CO2
As CO2 bends and all above degeneracies are removed, the primary topographic features (cusps, minima, and saddles) remain clearly visible up to . This is demonstrated in the plane in Fig. 9 and in the plane in Fig. 10. Further decrease of spawns new avoided crossings, a detailed map of which is given in the angular cuts in Figs. 3 and 4. All pairs of calculated states become near degenerate at various geometries. For example, the state forms avoided crossings successively with states at and at (Fig. 4). The state approaches closely at and at , and the states and cross between and . Solid lines in Figs. 3 and 4 cross because they refer to the states locally diabatized through the ‘bent’ avoided crossings (see Sect. IV.1). Potential curves of the states and bear evidence of strong interactions with the next higher states. Especially pronounced in Fig. 3(d,f) and Fig. 3(b,d) are the sharp near-intersections in around 155∘.
As a result of multiple avoided crossings, local bent equilibria are found in all calculated states. The diabatic origin of a particular local minimum is invariably different from the adiabatic one. Consider the carbene-like bent OCO with in Fig. 4(a). Purely adiabatically (solid dots), this minimum belongs to the ground electronic state. In the locally diabatic picture, bent OCO belongs the state (purple line). An avoided crossing between states and , recognizable in panel (a) around , implies that another diabatization might re-assign the bent OCO minimum to the state (brown line). Furthermore, the broad barrier in near indicates another avoided crossing with the next higher state which thus would be the true owner of the bent OCO minimum. Similar analysis applies to the bent OCO in the states in Fig. 4(b). Depending on the chosen representation, it can be ascribed to the fully adiabatic state (dots), to the locally diabatic state (brown line), or — via the sharp near-intersection around — to the next higher state. Other bent conformations in the excited states result from state interactions, too. One such isomer with the valence angle of 100∘ is created via an avoided crossing in [Fig. 3(c,e)]. Crossing of the same state with the next higher state between 60∘ and 70∘ leads to another high-energy bent minimum [Fig. 4(a,e)].
III.5 Beyond the first six states: Valence/Rydberg crossings
Many local barriers and minima in the states and are due to intersections with ‘invisible’ higher lying electronic states. Previous detailed ab initio studies exposed these ‘invisible’ intersection partners as mainly Rydberg states.EEW77 ; CJL87 ; SFCRW91 ; SFCRW93 The aim of this section is to illustrate how the Rydberg/valence interaction affects the potential profiles along the dissociation path and to outline the geometries at which valence states can effectively drain population from the Rydberg manifold. To this end, CASSCF calculations of the first 10 states, — and —, have been performed with the d-aug-cc-pVQZ basis set for the CO bond distances and the OCO bond angles . One CO bond is kept fixed at , so that only CO + O arrangement channel is covered.
The CASSCF potentials along are shown in Fig. 11. All five states and four states converge to thresholds (1) or (3). The states correlating with the lowest threshold (1) are of , , and symmetry. The states correlating with the highest threshold (3) are , , and . One state converges towards the asymptote
(5) |
At linearity, this state is the Rydberg state .NOTE-CO23-1 Compared to Fig. 1, many new avoided crossings are found in Fig. 11(a,b). The notation of a state in these panels is related to the expectation value which a diabatic state preserves across the intersections. Potential curves in Fig. 11(a,b) are color coded according to the values: states () are shown in green; states () are blue, and states () are red/orange. Clearly, most adiabatic curves change color more than once as the CO bond stretches: They are tailored out of several diabatic states.
Familiar from the discussion in Sect. III.3 are the fivefold crossings involving and the accidentally degenerate pair . These crossings are marked with arrows in Fig. 11(b). As in the MRD-CI calculations, the pair is easily recognizable using matrix elements of . At linearity, for these two states is either 0 or 4, but already a tiny deviation of 2∘ causes to collapse towards an average value of . This state mixing is stressed in Fig. 11(a,b) with the label and with the red/green color. The second edition of the mixed pair is found around 11.5 eV. The Rydberg state , shown with an orange line in Fig. 11(a,b), carries as a satellite the state [green line in Fig. 11(b)]. Again, for these states assumes a non-integer value between 1.5 and 2.5 in even slightly bent molecule.
The valence pair converges towards the uppermost threshold (3) and successively traverses the higher lying states. The prominent dissociation barrier in the and states near is a clear signature of the sixfold valence/Rydberg crossing involving , the state [green line in Fig. 11(a)] and a repulsive state [orange line in Fig. 11(a,b)]. is the optically bright Rydberg state responsible for the strong absorption band around 11.1 eV (111.7 nm).CJL87 The state corresponds to a pair of strongly repulsive states descending towards threshold (1) from very high energies. Two sections of this repulsive potential curve are seen in Fig. 11(a,b) in the intervals and . Thus, the asymptotic repulsive portions of the and states diabatically belong to . The sixfold crossing near is not the only one involving state. At higher energies and shorter CO bonds, intersects the Rydberg pair (cf. Fig. 11(a,b) near ). The diabatic state distinctly stands out because it strictly preserves the projection even for , while values for the other states become non-integer.
The above discussion is valid for both and states. The symmetry block in Fig. 11(b) contains one more state, namely the Rydberg state missing among the states where it would have been . This state, materializing out of nowhere at , is involved in the crossing — making it a sevenfold intersection.
While the adiabatic gaps in the valence/valence intersection grow as deviates from 180∘, all Rydberg/Rydberg and Rydberg/valence intersections not only remain recognizable in bent CO2, but clearly sharpen up. This is illustrated in panels (c) — (f) of Fig. 11 drawn for and 160∘. As a side result, the repulsive and states, deriving from , remain distinct at all angles, despite new ‘bent’ intersections in the and states [see, for example, Fig. 11(e,f)].NOTE-CO23-2 Via these intersections, carbon dioxide excited with vacuum UV light can — at bent geometries — reach any of the shown dissociation channels. Although a detailed analysis of these multiple pathways is beyond the scope of the present work, a brief description of intersections involving the optically bright state is added at the end of this section.
Potential curves along are shown in Fig. 12. The crossing patterns, expecially numerous among states, explain intricate topography of states and in Fig. 3. For example, the sharp barrier in the state at angles between 150∘ and 170∘ [Fig. 3(d,f) and Fig. 3(b,d)] originates from a complicated three state crossing around 165∘ involving states [Fig. 12(b)]. Diabatically, the decreasing branch of the state at belongs to — the state which at linearity merges into the Rydberg pair . This diabatic state can be followed to even smaller angles through another intersection, this time with near . Ultimately, as indicated with an arrow in Fig. 12(b), it is this diabatic state which the carbene-like OCO minimum belongs to. The avoided crossings in the symmetry block are similar and simpler. The three state crossing of is a mere ghost because the adiabatic gap is almost 2 eV wide and the barrier in the state is broad and low. The state , originating from , stabilizes upon bending and is easily traced to smaller angles through the intersection with near . Again, the carbene-like OCO correlates diabatically with the state originating from at linearity.
The analysis of broadening and splittings in the strong absorption band of the Rydberg state focused on the interactions of with valence states.CJL87 ; SFCRW91 ; SFCRW93 Present calculations reveal numerous avoided crossings with both valence and Rydberg states inside and outside the FC zone. At linearity, the state cuts twice through the Rydberg state [Fig. 11(a)], and both crossings persist to smaller angles [Fig. 11(c,e)]. Another crossing occurs near (for all calculated angles) and mixes with . This interaction strongly perturbs the ab initio values but vanishes beyond . The potential cut along in Fig. 12(a) exposes another avoided crossing at which involves the states and leads to a local minimum in (the state, correlating with at these RCO). Finally, the barrier in near implies interaction with the next higher state, which according to the analysis of Ref. SFCRW93, correlates with at linearity. While all these crossings can redirect population from the optically bright state along various linear and bent routes, Fig. 11 clearly demonstrates that the diabatic state, calculated at the CASSCF level, is repulsive at all geometries and thus can dissociate directly.
IV Quasi-diabatization of the valence states
As explained in the Introduction, the diabatic representation,NOTE-CO21-3 although not generally indispensable, is best suited for the particular implementation of nuclear quantum dynamics used in paper II. Rigorously speaking, diabatization has to be performed simultaneously on all six calculated valence states, because each pair of states intersects either in the FC zone or at bent geometries. Simplifications to this ‘Herculean task’QZGSCH05 stem from the expectation that the two groups of intersections influence photodissociation in different ways: While the ‘bent’ CIs affect later stages of the product formation, the electronic branching ratios, and/or the rovibrational photofragment distributions,CO2-5 the ‘FC’ CIs are directly responsible for the shape of the observed absorption spectrum. This distinction guides the practical construction of the diabatic states: The ‘bent’ CIs and the ‘FC’ CIs are analyzed at two different levels of detail, commensurate with their expected impact on the absorption spectrum. As a result, the complete multistate problem splits into several steps and the need for a global diabatization of six multiply intersecting states is obviated.
IV.1 Intersections at bent geometries: Local diabatization
The vicinity of ‘bent’ CIs is diabatized locally, using the energy-based scheme as described, for example, by Köppel.K04 All considered CIs are shown in Figs. 3 and 4 and include
-
(a)
the pair at ;
-
(b)
the pair at ;
-
(c)
the pair, with state diabatized in step (b), at .
The pair is considered as an example. At geometries, these states belong to and irreps, so that their CI is ‘symmetry allowed’.Y04 Its branching space includes as the tuning mode (irrep ) and the antisymmetric stretch as the symmetry breaking coupling mode (irrep ). The diabatic potential matrix is constructed on the ab initio grid of bond distances from the diagonal adiabatic potential matrix using the orthogonal adiabatic-to-diabatic transformation (ADT)K04
(6) |
with the transformation matrix parameterized by the mixing angle ,
(7) |
which varies between 0 and ; denotes the crossing point of the diabatic potentials (); the width of the function is evaluated from the adiabatic energy gaps and the average slope of the diabatic potentials ,
(8) |
at each grid point for . With the off-diagonal coupling in Eq. (6) neglected, the ground state becomes completely decoupled from other states. The remaining diagonal matrix element in is the potential waiting to be further diabatized at linearity. The diabatization is kept local by restricting the interval, over which varies from 0 to , to the vicinity of :
(9) |
The two switching functions ,HH00B
(10) |
with , , and , are adjusted to restrict the diabatization to angles 90∘—. If in , the so-called ‘diabatization by eye’ is recovered, which corresponds to a relabeling of ab initio energies at the crossing angle .LLYM94 ; GQZSCH07 ; G12A
The main advantage of local diabatization, the results of which are shown in 1D in Figs. 3 and 4, is that smooth potentials are created for strongly bent geometries at relatively low cost. The quality of the resulting PESs in 2D, either in the or in the plane is illustrated for all states in Figs. 5, 8, 9, 10, and 13. For the lowest excited states in each symmetry block, and , the global bent equilibrium at 120∘ or 130∘ [Fig. 9(a,c)] is not affected by the local diabatization. In contrast, the carbene-like bent OCO minima change the owner: They appear in the state [local minimum near 70∘ barely visible in Fig. 9(a) but pronounced in Fig. 13(a)] and in the state [Fig. 13(d)]. Barriers and local minima arising in the avoided crossings with Rydberg states are not diabatized within this scheme. Examples are the sharp barrier in near 150∘ [Fig. 9(d) and Fig. 13(d)] and the deep local minimum in near 100∘ [Fig. 9(b) and Fig. 13(b)].
IV.2 Intersections at linear geometries: Regularized diabatic states
In the second step, the CIs in the FC region are diabatized; kinematic singularities due to glancing intersections between and states are not considered. Accurate diabatization schemes rely on direct numerical differentiation of the electronic wave functions with respect to nuclear coordinates or on the orbital rotation methodDW93 ; SHW99 as implemented in MOLPRO.MOLPRO-FULL Both types of calculations become prohibitively expensive with the d-aug-cc-pVQZ basis set, and further approximations are needed in order to find the diabatic potential matrix, consisting of a block of and a block of states:
(11) |
The five diagonal matrix elements are the five diabatic states , , , , and , with the smooth intersecting potentials in the plane constructed to coincide at with the ab initio PESs for , , and , respectively. The semi-global diabatization scheme, akin to the vibronic coupling modelKDC84 ; K04 and adjusted to the topography of the closed fivefold CI seam, has already been introduced in Ref. GB12, . Diabatization proceeds in two steps. First, a model diabatic matrix of the form Eq. (11) with elements is constructed. Due to orbital degeneracy, and . Moreover, the accidental near-degeneracy of states and implies in a broad vicinity of the closed CI seam — the property which substantially simplifies diabatization of states. Deviations from linearity, measured by the coordinate , are included in the model via off-diagonal matrix elements represented as symmetry adapted expansions in :
(12) | |||||
(13) |
Couplings of the accidentally degenerate and states to are set equal, while the matrix element for the RT-like interaction is neglected. The model is complete after the expansion coefficients in are calculated from a non-linear least-squares fit to ab initio energies. In the second step, the regularized diabatic states approach is invoked,K04 and the matrix elements are used to define the orthogonal ADT matrix, which is applied to the adiabatic matrix via Eq. (6) giving the desired diabatic matrix elements of Eq. (11) on the full ab initio grid. Final interpolation is performed using 3D splines.
The diabatization is localized to the vicinity of the CI seam by modifying matrix elements () in Eq. (12),
(14) |
with being a radial distance from the center of a circle enclosing the seam; and . Adiabatic and diabatic states are forced to coincide if either bond becomes longer than , so that diabatization regions for linear and bent CIs are cleanly kept apart.
The constructed representation describes best the vicinity of linearity in which non-adiabatic transitions occur. The exact range of validity is determined by the length of the expansion in Eq. (12). The choice gives a model which fits ab initio data with a root mean square error of cm-1 for angles . The ADT, constructed using this model, is guaranteed to remove kinematic singularities at the CIs, but leaves the strength of residual non-adiabatic couplings unspecified.KGM01 The ultimate test of the scheme is the quantum mechanical absorption spectrum described in paper II. In order to assess the accuracy of the truncated expansion, three diabatic representations are constructed, based on the expansion coefficients obtained from fitting in three different angular ranges , , and . The corresponding absorption spectra are virtually identical. The spectra are also insensitive to small variations in and in Eq. (14) — the modifications take place too far away from the crossing seam to affect nuclear dynamics.
Another test of the constructed ADT is given in Fig. 14, in which the model mixing angle for states is compared with the ab initio one calculated using MOLPRO with a smaller cc-pVQZ atomic basis set. The dependence on the CO bond length has a characteristic bell shape: The closed CI seam is intersected twice giving rise to the ascending and the descending branch. The curve flattens out as decreases and CO2 leaves the degeneracy plane at 180∘. Agreement between the model and the ab initio results is satisfying for all angles. A constant shift of 25∘ applied to the ab initio mixing angle has no effect on the strength of non-adiabatic coupling proportional to .
One-dimensional cuts through the diabatic PESs (diagonal matrix elements corresponding to states and ) are shown in Fig. 15. They cross at all geometries and can be directly compared with adiabatic states in Fig. 1. The off-diagonal coupling matrix elements are large in the intersection region vanishing off towards the asymptotic channels. Diabatic matrix elements are further illustrated in Fig. 16 in the plane and in Fig. 17 in the plane. In all cuts, the diabatic PESs smoothly depend on internal coordinates. In the plane, the off-diagonal diabatic coupling stays localized in the inner region. In contrast, the coupling along bending coordinate is delocalized across a substantial range in the plane. As a result, the angular shape of the diabatic potentials is distorted compared to the adiabatic case: Diabatic potentials along the coupling mode are close to the average adiabatic potential .
V Conclusions
This paper describes properties of global PESs of the first six singlet electronic states of CO2 constructed from about 5000 symmetry unique ab initio points calculated with the d-aug-cc-pVQZ basis set using the MRD-CI method. The main results can be summarized as follows:
-
1.
Calculations accurately reproduce the known benchmarks for all states and establish missing benchmarks for the future calculations: Bond distances and bond angles are accurate to within 0.1%, known fundamental frequencies (mainly ground state ) are accurate to within 1.5 cm-1, the accuracy of the vertical excitation energies is expected to be better than 0.05 eV; dissociation energies agree with experimental thresholds within 0.15 eV for four covered arrangement channels.
-
2.
Local equilibria are abundant in the calculated states. Bent OCO isomer is found in the adiabatic states and . Linear COO is found in and . Their diabatic electronic origin is clarified, and the properties, including equilibrium geometries, excitation energies, and vibrational frequencies, are established.
-
3.
Near degeneracies can be found for each pair of six valence states, at linear or bent geometries, or at both. Avoided crossings and conical and glancing intersections literally shape the observed topography of the excited electronic states. Detected intersections are not limited to the valence manifold and the search for electronic origins of local minima and barriers involves valence/Rydberg and Rydberg/Rydberg intersections.
-
4.
Characteristic for state intersections in CO2, both conical and glancing, is that they include several states. In the FC region, a fivefold intersection between , , and states is found. The seam of this intersection forms a closed loop, spectroscopic manifestations of which are discussed in paper II. Outside the FC region at linearity, six- and sevenfold intersections are predicted, some of which persist over extended angular range in the bent molecule.
-
5.
Diabatic potential matrix, with all elements smoothly depending on internal coordinates, is constructed using two-step local diabatizations of linear and bent conical intersections.
It is tempting to try to infer the course of photodissociation and the principal features of the absorption spectrum — the outcome of a complicated quantum mechanical calculation — from the constructed PESs alone. Two issues have to be resolved if one deals with five interacting states. The first is the strength of diabatic (intra-symmetry) and RT (inter-symmetry) coupling. If the off-diagonal vibronic coupling is weak, the diabatic potentials should be chosen as ‘zeroth order guides’. If the vibronic coupling is strong, it is the adiabatic description which becomes relevant — and the difference between the two pictures is striking, especially along the bending angle as Figs. 9 and 17 demonstrate. As discussed in paper II, the vibronic coupling is strong, the RT interaction between and states is to a large extent quenched, and the adiabatic potentials can be used for qualitative analysis. The global minima of states or are bent and lie eV below the dissociation threshold (1) or eV below the FC point. The route connecting the FC region with these bent equilibria is barrierless. In contrast, there is a barrier to dissociation near linearity — the leftover of the lower cone of the CI. Thus, one expects the low energy bands in the absorption spectrum, associated with and states, to reflect highly excited bending motion. This interpretation is commonly given in the literature:RMSM71 Instead of dissociating directly, the molecule bends first. With growing photon energy, the contribution of direct dissociation through linear geometries will certainly grow, because the barrier is only about 0.2 eV high and is located eV below the FC point. Above eV, the valence states and will contribute to the observed spectrum. These ‘linear’ states are separated from the dissociation asymptote by a eV high and broad barrier, with the implication that CO2 in these states can decay only through non-adiabatic interactions with the dissociation continuum of the lower states. In other words, one expects to see a resonance-dominated absorption spectrum. The next qualitative change in the absorption spectrum within the valence manifold can be expected after the photon energy reaches the top of the dissociation barrier in the upper valence states and allows direct dissociation from linearity.
The above discussion is based on 1D and 2D potential cuts and the data in Table 3 — given the adiabatic representation is the adequate one. However, there is another important piece of information still missing, namely the TDMs with the ground state. As has already been mentioned in Sect. III.2, the electronic transitions in the wavelength range of 160 nm — 120 nm are forbidden. The bands are observed only because the TDMs are not constant but strongly change with molecular coordinates as one moves away from the high-symmetry FC point. This dependence is a manifestation of the Herzberg-Teller effectHERZBERG67 which plays the leading role in shaping the absorption bands, is at least as important as the potential profiles discussed above, and has to be considered on equal footing with the state intersections. The discussion of the coordinate dependence of the TDMs is deferred to paper II.
Acknowledgements.
Financial support by the Deutsche Forschungsgemeinschaft is gratefully acknowledgedReferences
- (1) S. Yu. Grebenshchikov, subsequent paper .
- (2) S. Yu. Grebenshchikov, J. Chem. Phys. 137, 021101 (2012).
- (3) S. Yu. Grebenshchikov and R. Borrelli, J. Phys. Chem. Lett. 3, 3223 (2012).
- (4) G. Herzberg. Molecular Spectra and Molecular Structure III. Electronic Spectra and Electronic Structure of Polyatomic Molecules. Van Nostrand, Princeton, (1967).
- (5) J. W. Rabalais, J. M. McDonald, V. Scherr and S. P. McGlynn, Chem. Rev. 71, 73 (1971).
- (6) H. Okabe. Photochemistry of Small Molecules. Wiley, New York, (1978).
- (7) C. Cossart-Magos, M. Jungen and F. Launay, Mol. Phys. 61, 1077 (1987).
- (8) W. B. England and W. C. Ermler, J. Chem. Phys. 70, 1711 (1979).
- (9) P. J. Knowles, P. Rosmus and H.-J. Werner, Chem. Phys. Lett. 146, 230 (1988).
- (10) A. Spielfiedel, N. Feautrier, C. Cossart-Magos, G. Chambaud, P. Rosmus and H.-J. Werner ans P. Botschwina, J. Chem. Phys. 97, 8382 (1992).
- (11) L. S. Cederbaum. In Conical Intersections, edited by W. Domcke, D. R. Yarkony and H. Köppel. (World Scientific, Singapore, 2004).
- (12) H. Köppel. In Conical Intersections, edited by W. Domcke, D. R. Yarkony and H. Köppel. (World Scientific, Singapore, 2004).
- (13) H.-J. Werner, P. J. Knowles, R. Lindh, M. Schütz, P. Celani, T. Korona, F. R. Manby, G. Rauhut, R. D. Amos, A. Bernhardsson, A. Berning, D. L. Cooper, M. J. O. Deegan, A. J. Dobbyn, F. Eckert, C. Hampel, G. Hetzer, A. W. Lloyd, S. J. McNicholas, W. Meyer, M. E. Mura, A. Nicklass, P. Palmieri, R. Pitzer, U. Schumann, H. Stoll, A. J. Stone, R. Tarroni and T. Thorsteinsson. Molpro, a package of ab initio programs, (2003). see http://www.molpro.net.
- (14) T. H. Dunning, J. Chem. Phys. 90, 1007 (1989).
- (15) S. R. Langhoff and D. E. Davidson, Int. J. Quant. Chem. 8, 61 (1974).
- (16) R. T. Pierrehumbert, Phys. Today 64, 33 (2011).
- (17) A. Chedin, J. Mol. Spec. 76, 430 (1979).
- (18) V. Rodriguez-Garcia, S. Hirata, K. Yagi, K. Hirao, T. Taketsugu, I. Schweigert and M. Tasumi, J. Chem. Phys. 126, 124303 (2007).
- (19) K. Yagi, S. Hirata and K. Hirao, Phys. Chem. Chem. Phys. 10, 1781 (2008).
- (20) S. E. Stein. ”Infrared Spectra” in NIST Chemistry WebBook, NIST Standard Reference Database Number 69, P.J. Linstrom and W.G. Mallard, Eds. National Institute of Standards and Technology, Gaithersburg MD, 20899, http://webbook.nist.gov.
- (21) E. Fermi, Z. Phys. 71, 250 (1931).
- (22) S. Yu. Grebenshchikov, unpublished (2013).
- (23) S. S. Xantheas and K. Ruedenberg, Int. J. Quant. Chem. 49, 409 (1994).
- (24) D.-Y. Hwang and A. Mebel, Chem. Phys. 256, 169 (2000).
- (25) D. R. Yarkony. In Conical Intersections, edited by W. Domcke, D. R. Yarkony and H. Köppel. (World Scientific, Singapore, 2004).
- (26) R. N. Dixon, Proc. Roy. Soc. A 275, 431 (1963).
- (27) C. Cossart-Magos, F. Launay and J. E. Parkin, Mol. Phys. 75, 835 (1992).
- (28) M. N. R. Ashfold, M. T. McPherson and J. P. Simons, Topics Current Chem. 86, 1 (1979).
- (29) H. Köppel, W. Domcke and L. S. Cederbaum, Adv. Chem. Phys. 57, 59 (1984).
- (30) S. Carter, I. M. Mills and R. Dixon, J. Mol. Spec. 106, 411 (1984).
- (31) R. Renner, Z. Phys. 92, 172 (1934).
- (32) C. Petrongolo, J. Chem. Phys. 89, 1297 (1988).
- (33) E. M. Goldfield, S. K. Gray and L. B. Harding, J. Chem. Phys. 99, 5812 (1993).
- (34) E. Teller, J. Phys. Chem. 41, 109 (1937).
- (35) G. J. Atchity, K. Ruedenberg and A. Nanayakkara, Theor. Chem. Acc. 96, 195 (1997).
- (36) C. Levi, G. J. Halasz, A. Vibok, I. Bar, Y. Zeiri, R. Kosloff and M. Baer, J. Chem. Phys. 128, 244302 (2008).
- (37) C. Levi, G. J. Halasz, A. Vibok, I. Bar, Y. Zeiri, R. Kosloff and M. Baer, Int. J. Quant. Chem. 109, 2482 (2009).
- (38) H. Köppel, J. Gronki and S. Mahapatra, J. Chem. Phys. 115, 2377 (2001).
- (39) A. J. Dobbyn and P. J. Knowles, Mol. Phys. 91, 1197 (1997).
- (40) W. B. England, W. C. Ermler and A. C. Wahl, J. Chem. Phys. 66, 2336 (1977).
- (41) A. Spielfiedel, N. Feautrier, G. Chambaud, P. Rosmus and H.-J. Werner, Chem. Phys. Lett. 183, 16 (1991).
- (42) A. Spielfiedel, N. Feautrier, G. Chambaud, P. Rosmus and H.-J. Werner, Chem. Phys. Lett. 216, 162 (1993).
- (43) CASSCF energies at , , and , relative to the minimum of the curve, are eV, 8.9 eV, and 11.2 eV for channels (3), (4), and (7), respectively.
- (44) Note that the avoided Rydberg/valence crossings between and states, investigated in Ref. SFCRW91, , lie outside the cuts shown in Fig. 11.
- (45) Whenever ‘diabatic representation’ is mentioned, a quasi-diabatic representation,K04 in which the kinematic singularities and the potential cusps at the intersections are removed, is meant.
- (46) Z.-W. Qu, H. Zhu, S. Yu. Grebenshchikov and R. Schinke, J. Chem. Phys. 122, 191102 (2005).
- (47) S.Yu. Grebenshchikov, in preparation (2013).
- (48) R. van Harrevelt and M. C. van Hermet, J. Chem. Phys. 112, 5787 (2000).
- (49) C. Leforestier, F. LeQuéré, K. Yamashita and K. Morokuma, J. Chem. Phys. 101, 3806 (1994).
- (50) S. Yu. Grebenshchikov, Z.-W. Qu, H. Zhu and R. Schinke, Phys. Chem. Chem. Phys. 9, 2044 (2007).
- (51) W. Domcke and C. Woywod, Chem. Phys. Lett. 216, 362 (1993).
- (52) D. Simah, B. Hartke and H.-J. Werner, J. Chem. Phys. 111, 4523 (1999).
Geometry | 111The ab initio ZPEs are: ZPE(CO2) = 0.314 eV, ZPE(CO) = 0.130 eV, and ZPE(O2) = 0.095 eV. | Reference | ||||
---|---|---|---|---|---|---|
OCO linear | 180.0 | 2.1991 | 2.1991 | 4.3982 | 0.0 | this work |
180.0 | 2.2050 | 2.2050 | 4.4100 | 0.0 | HM00A, | |
180.0 | 2.1924 | 2.1924 | 2.3848 | 0.0 | KRW88,; SFCCRWB92, | |
180.0 | 2.1960 | 2.1960 | 4.3920 | 0.0 | NISTDATABASE1, 222Experimental values. | |
OCO bent | 73.2 | 2.51 | 2.51 | 2.99 | 5.90 | this work |
72.9 | 2.52 | 2.52 | 2.99 | 6.04 | HM00A, | |
73.1 | 2.53 | 2.53 | 3.01 | 5.97333Energy without ZPE correction. | XR94, | |
COO linear | 0.0 | 2.20 | 4.65 | 2.45 | 7.35333Energy without ZPE correction. | this work |
0.0 | 2.20 | 4.65 | 2.45 | 7.14 | HM00A, | |
0.0 | 2.17 | 5.37 | 3.20 | 6.86 | XR94, | |
O+CO | — | 2.14 | 7.28 | this work | ||
2.175 | 7.64 | HM00A, | ||||
2.132 | 7.41 | HERZBERG67, ; AMS79, 222Experimental values. | ||||
C+O | — | 2.30 | 11.35 | this work | ||
2.356 | 11.49 | HM00A, | ||||
2.283 | 11.52 | HERZBERG67, 222Experimental values. |
State | Refs. RHYHTST07,; YHH08, | PES1 | PES2 | Exp |
---|---|---|---|---|
111The energy difference between the ground vibrational level and the potential minimum is 2516 cm-1 for PES1 and 2533 cm-1 for PES2. The ZPE = 2508.5 cm-1, given in the NIST database,NISTDATABASE1 is evaluated from the fundamental frequencies via . For PES1 and PES2, this value is 2503.6 cm-1 and 2512.5 cm-1, respectively. | 0.0 | 0.0 | 0.0 | 0.0 |
669.1 (-1.7) | 668.6 (-1.2) | 670.3 (-2.9) | 667.4 | |
1288.9 (-3.5) | 1265.3 (20.1) | 1284.1 (1.3) | 1285.4 | |
1339.6 (-4.5) | 1336.6 (-1.5) | 1336.8 (-1.7) | 1335.1 | |
1389.3 (-1.1) | 1373.5 (14.7) | 1389.2 (-1.0) | 1388.2 | |
1938.0 (-5.5) | 1913.2 (19.3) | 1933.4 (-0.9) | 1932.5 | |
2011.4 (-8.2) | 2005.9 (-2.7) | 2006.1 (-2.9) | 2003.2 | |
2080.0 (-3.1) | 2061.7 (15.2) | 2079.6 (-2.7) | 2076.9 | |
2349.2 (0.0) | 2350.6 (-1.4) | 2351.4 (-2.2) | 2349.2 | |
2552.0 (-8.6) | 2516.4 (27.0) | 2549.5 (-6.1) | 2543.4 | |
2676.3 (-5.2) | 2626.0 (45.1) | 2668.3 (2.8) | 2671.1 | |
2809.1 (-12.0) | 2761.8 (35.3) | 2790.2 (6.9) | 2797.1 | |
2589.8 (-4.7) | 2581.6 (3.5) | 2585.1 |
111Experimental data from Ref. SFCCRWB92, are: For , , eV, cm-1; for , , eV, cm-1. | 111Experimental data from Ref. SFCCRWB92, are: For , , eV, cm-1; for , , eV, cm-1. | 222For the and states, the geometries refer to the local minimum in the FC region. The vibrational frequencies are strongly perturbed by CI cusps, and the corresponding ZPEs are omitted in . | 222For the and states, the geometries refer to the local minimum in the FC region. The vibrational frequencies are strongly perturbed by CI cusps, and the corresponding ZPEs are omitted in . | ||
O+CO channel333Dissociation thresholds, labeled with electronic states of atomic/diatomic fragment, are correlated with the diabatic states of CO2. | |||||
C+O2 channel333Dissociation thresholds, labeled with electronic states of atomic/diatomic fragment, are correlated with the diabatic states of CO2. | |||||
5.36 | 5.39 | 7.95 | 8.67 | 8.67 | |
8.92 | 8.92 | 8.79 | 9.16 | 9.17 | |
, Refs. SFCCRWB92, | 9.00 | 9.00 | 9.19 | 9.28 | 9.28 |
2.37 | 2.36 | 2.41 | 2.25 | 2.25 | |
2.37 | 2.36 | 2.41 | 2.80 | 2.80 | |
, Ref. KRW88,; SFCCRWB92, | 2.38 | 2.38 | 2.40 | 2.29 | 2.29 |
127.3 | 117.9 | 176.0 | 180.0 | 180.0 | |
, Ref. SFCCRWB92, | 127.0 | 117.8 | 180.0 | 180.0 | 180.0 |
1283 | 1340 | 1015 | 520 | 550 | |
905 | 865 | 1118 | 1560 | 1550 | |
670 | 580 | 577 | 2290 | 3100 | |
444Experimental dissociation thresholds are given in Eqs. (1), (2), (3), and (4). Ab initio ZPEs are: ZPE[CO)] = 0.314 eV; ZPE[CO] = 0.130 eV; ZPE[CO] = 0.099 eV; ZPE[O] = 0.095 eV; ZPE[O] = 0.080 eV. | 7.27 | 7.27 | 11.31 | 7.27 | 7.27 |
444Experimental dissociation thresholds are given in Eqs. (1), (2), (3), and (4). Ab initio ZPEs are: ZPE[CO)] = 0.314 eV; ZPE[CO] = 0.130 eV; ZPE[CO] = 0.099 eV; ZPE[O] = 0.095 eV; ZPE[O] = 0.080 eV. | 11.34 | 11.34 | 13.59 | 13.59 | 13.59 |
![[Uncaptioned image]](https://cdn.awesomepapers.org/papers/3d831978-4eb9-42e1-abf4-d3f07fb55d46/x1.png)
Fig. 1
![[Uncaptioned image]](https://cdn.awesomepapers.org/papers/3d831978-4eb9-42e1-abf4-d3f07fb55d46/x2.png)
Fig. 2
![[Uncaptioned image]](https://cdn.awesomepapers.org/papers/3d831978-4eb9-42e1-abf4-d3f07fb55d46/x3.png)
Fig. 3
![[Uncaptioned image]](https://cdn.awesomepapers.org/papers/3d831978-4eb9-42e1-abf4-d3f07fb55d46/x4.png)
Fig. 4
![[Uncaptioned image]](https://cdn.awesomepapers.org/papers/3d831978-4eb9-42e1-abf4-d3f07fb55d46/x5.png)
Fig. 5
![[Uncaptioned image]](https://cdn.awesomepapers.org/papers/3d831978-4eb9-42e1-abf4-d3f07fb55d46/x6.png)
Fig. 6
![[Uncaptioned image]](https://cdn.awesomepapers.org/papers/3d831978-4eb9-42e1-abf4-d3f07fb55d46/x7.png)
Fig. 7
![[Uncaptioned image]](https://cdn.awesomepapers.org/papers/3d831978-4eb9-42e1-abf4-d3f07fb55d46/x8.png)
Fig. 8
![[Uncaptioned image]](https://cdn.awesomepapers.org/papers/3d831978-4eb9-42e1-abf4-d3f07fb55d46/x9.png)
Fig. 9
![[Uncaptioned image]](https://cdn.awesomepapers.org/papers/3d831978-4eb9-42e1-abf4-d3f07fb55d46/x10.png)
Fig. 10
![[Uncaptioned image]](https://cdn.awesomepapers.org/papers/3d831978-4eb9-42e1-abf4-d3f07fb55d46/x11.png)
Fig. 11
![[Uncaptioned image]](https://cdn.awesomepapers.org/papers/3d831978-4eb9-42e1-abf4-d3f07fb55d46/x12.png)
Fig. 12
![[Uncaptioned image]](https://cdn.awesomepapers.org/papers/3d831978-4eb9-42e1-abf4-d3f07fb55d46/x13.png)
Fig. 13
![[Uncaptioned image]](https://cdn.awesomepapers.org/papers/3d831978-4eb9-42e1-abf4-d3f07fb55d46/x14.png)
Fig. 14
![[Uncaptioned image]](https://cdn.awesomepapers.org/papers/3d831978-4eb9-42e1-abf4-d3f07fb55d46/x15.png)
Fig. 15
![[Uncaptioned image]](https://cdn.awesomepapers.org/papers/3d831978-4eb9-42e1-abf4-d3f07fb55d46/x16.png)
Fig. 16
![[Uncaptioned image]](https://cdn.awesomepapers.org/papers/3d831978-4eb9-42e1-abf4-d3f07fb55d46/x17.png)
Fig. 17