Multiconfigurational Electron Propagator Research Articles

Investigation of negative-ion resonances using a subspace-projected multiconfigurational electron propagator perturbed with a complex absorbing potential.

The transient negative-ion resonances found in scattering experiments are important intermediates in many chemical processes. These metastable states correspond to the continuum part of the Hamiltonian of the projectile-target composite system. Usual bound-state electronic structure methods are not applicable for these. In this work, we develop a subspace-projection method in connection with an electron propagator (EP) defined in terms of a complete-active-space self-consistent-field initial state. The target Hamiltonian (Ĥ) is perturbed by a complex absorbing potential (CAP) for the analytical continuation of the spectrum of Ĥ to complex eigenvalues associated with the continuum states. The resonance is identified as a pole of the EP, which is stable with respect to variations in the strength of the CAP. The projection into a small subspace reduces the size of the complex matrices to be diagonalized, minimizes the computational cost, and affords some insight into the orbitals that are likely to play some role in the capture of the projectile. Two molecular (Πg2N2 - and 2Π CO-) and an atomic shaperesonance (2P Be-) are investigated using this method. The position and width of the resonances are in good agreement with the previously reported values.

Comparison of perturbative and multiconfigurational electron propagator methods

Ionization energies below 20 eV of 10 molecules calculated with electron propagator techniques employing Hartree-Fock orbitals and multiconfigurational self-consistent field orbitals are compared. Diagonal and nondiagonal self-energy approximations are used in the perturbative formalism. Three diagonal methods based on second- and third-order self-energy terms, all known as the outer valence Green's function, are discussed. A procedure for selecting the most reliable of these three versions for a given calculation is tested. Results with a polarized, triple ζ basis produce root mean square errors with respect to experiment of approximately 0.3 eV. Use of the selection procedure has a slight influence on the quality of the results. A related, nondiagonal method, known as ADC(3), performs infinite-order summations on several types of self-energy contributions, is complete through third-order, and produces similar accuracy. These results are compared to ionization energies calculated with the multiconfigurational spin-tensor electron propagator method. Complete active space wave functions or close approximations constitute the reference states. Simple field operators and transfer operators pertaining to the active space define the operator manifold. With the same basis sets, these methods produce ionization energies with accuracy that is comparable to that of the perturbative techniques. © 1996 John Wiley & Sons, Inc.

Theoretical study of C2 and C−2 : X 1Σ+g , a 3Πu , X 2Σ+g , and B 2Σ+u potentials

We employ multiconfigurational self-consistent field and multiconfigurational electron propagator methods to characterize the X 2Σ+g and B 2Σ+u states of C−2 and the X 1Σ+g and a3Πu states of C2 over a wide range of bond lengths (1.0–1.9 Å). We find a systematic difference of approximately 0.3 eV in the relative positioning of our anion- and neutral-state potentials compared to the anion–neutral spacing in the best curves constructed from experimental data. Once this energy shift is taken into consideration, all four of our computed potential energy curves are in reasonably good agreement with experimental information. However, there remains a substantial difference in the relative positioning of our B 2Σ+u and a 3Πu curves, compared to the best available experimental data, at larger bond lengths. The relevance of this discrepancy and of our other data to the present state of experimental knowledge on C−2 /C2 is discussed.

The ionization potentials of F2: A comparison of multiconfigurational electron propagator (MCEP) with other large scale methods using the same basis set

The multiconfigurational electron propagator technique (MCEP) gives reliable vertical ionization potentials (I.P.s) and electron affinities (E.A.s) for atoms and molecules, including open-shell and highly correlated systems. Shake-up and inner-valence I.P.s can be accurately obtained and characterized. In contrast, perturbative-type Green’s function (PTGF) approaches are useful for closed-shell systems with relatively little correlation. Perturbative-type Green’s functions cannot consistently reliably predict shake-up and inner-valence I.P.s. We have applied the MCEP method to F2 at 2.68 a.u. using 〈4s3p1d〉 and 〈5s4p2d〉 basis sets. In F2, the complete active space of all valence orbitals is small. Hence, reliable MCEP results should be obtained for all valence ionization processes. In addition, comparison calculations are given using other large scale techniques, i.e., ΔSCF, ΔMCSCF, PTGF, and Δ multireference CI using the same basis sets. The photoelectron spectrum of F2 below 30 eV is not well characterized experimentally except for the three lowest principal peaks, and apparently no reliable spectra have been reported for higher energies. We report and characterize several I.P.s in F2 0–60 eV, including several I.P.s that have not as yet been observed experimentally.

Multiconfigurational electron propagator (MCEP) ionization potentials for general open shell systems

We have developed a multiconfigurational electron propagator (MCEP) technique for the theoretical determination of ionization potentials for general open shell and highly correlated atomic and molecular systems. In order to do this, we have used and extended the generalized spin-symmetry adapted operators of Pickup and Mukhopadhyay. To properly account for correlation effects we have additionally included ionization and electron affinity operators analogous to the ‖Γ〉〈0‖ state transfer operators necessary in multiconfigurational linear response. MCEP ionization potentials and ionization process probabilities have been evaluated for both O2 and N2 and used to carry out detailed examination and interpretation of the respective PES and ESCA spectra. The MCEP results are extremely encouraging for both principal and shake-up I.P.′s. For example, using 〈5s4p1d〉 contracted Gaussian basis sets the principal valence ionization potentials to bound ionic states are calculated within ±0.3 eV of experiment for both N2 and O2. For several I.P.′s the breakdown of the single particle picture is calculationally demonstrated, e.g., the intensity of the 2σu electron removal process to give 2Σ−u states in O2 is almost equally divided between two ESCA peaks. One of these peaks is the previously unassigned A peak in the ESCA spectrum.