Identifying a pair of interacting chromophores by using SVD transformed CIS wave functions

https://doi.org/10.1016/j.cplett.2007.06.074Get rights and content

Abstract

It is discussed how the presence of a pair of weakly interacting similar chromophores in a molecule manifests in the framework of the recent scheme in which the description of individual CIS solutions has been simplified by using the singular value decomposition (SVD) method. The criteria are described which permit one to identify a pair of coupled excitations. It is also discussed that one can recover the individual occupied and virtual orbitals of two similar chromophores contributing to a complex overall excitation.

Graphical abstract

The manifestation of two similar chromphores is considered for the case of SVD transformed CIS wave functions.

  1. Download : Download full-size image

Introduction

In a recent Letter [1] it was described that any individual CIS solution can be dramatically simplified by using the singular value decomposition (SVD) method: after transformation it contains no more than n = min(nocc, nvirt) excitations. For that reason one should determine the appropriate nocc × nocc and nvirt × nvirt orthogonal (real unitary) matrices U and V, respectively, which transform the nocc × nvirt CIS coefficient matrix Cσ to a generalized diagonal form asΛσ=UTCσV=λ10000λ200000λn-10000λnHere we have assumed n = nocc < nvirt; if the opposite is true, then matrix Λ has a form transposed to that is shown above, and one has n = nvirt. Superscript σ (=α or β) indicates that we consider excitations of orbitals with spin σ. Note that the matrices U and V providing this generalized diagonalization can always be determined by standard linear algebraic techniques (e.g., [2]), as used in our program [3].

In terms of the original occupied and virtual canonic HF orbitals φio, φjv, the part of the CIS wave function which corresponds to excitations with spin σ can be written by using second quantization as|ΨCISσ=i=1noccσp=1nvirtσCipσφˆpv+φˆio-|HF(We indicate spin σ at the summations but not at the creation and annihilation operators φˆpv+ and φˆio-, respectively, in order to avoid very cumbersome notations.) By using the SVD method it is transformed to the form [1]|ΨCISσ=k=1nσλkψˆkv+ψˆko-|HFwhere the creation and annihilation operators correspond to some new orthonormalized orbitalsψrv=p=1nvirtVprφpvandψko=i=1noccUikφioobtained by performing separate unitary transformations within the occupied and virtual subspaces, respectively. Most recently Surján has showed [4] that these orbitals are nothing else than the natural orbitals of the multideterminant CIS wave function. This fact has a great importance for the considerations discussed below.

In this formalism the quantities λk play a central role: the relative weights of the individual determinants ψˆkv+ψˆko-|HF in the CIS wave function is given by 2λk2. (The factor 2 appears because the CIS wave function contains excited determinants for both spins, and their sum is usually assumed to be normalized to one, i.e., each |ΨCISσ is normalized to 12.) One may expect (or at least hope) that only one or a very few λk2 values differ significantly from zero, thus one or a few excitations dominate in every CIS wave function.

The first exploratory calculations indeed confirmed that expectation and indicated that the CIS solutions often are dominated by a single excitation exhibiting a relative weight somewhere between 90% and 99%. However, a first preliminary study [5] into a system which is expected to exhibit CD (circular dichroism) activity in its electronic spectrum, indicated that at least two excitations contribute significantly to some of the CIS solutions. This is due to the presence of at least two chromophores in the same molecule – CD spectra are often connected to a pair of interacting chromophores. Special effects may take place if the two chromophores are identical (very similar). Accordingly, the present letter is devoted to the problem of analyzing how the presence of a pair of weakly interacting similar chromophores in a molecule manifests in the framework of the present formalism, and how one can extract from the SVD transformed results some further information concerning the individual chromophores.

Although both approaches involve unitary transformations of the occupied and virtual orbitals, our scheme has little in common with the direct use of localized orbitals in discussing excitation processes [6], [7], [8]; however, at the end of our analysis we shall arrive to orbitals which can be expected not too far from those from which these authors are starting. It may also be mentioned that, by using a different formalism, the authors of Ref. [9] have also, in fact, derived the orbitals (4) but stopped just a step short from realizing that they correspond to transitions as described in (3).

Section snippets

Idealized case of non-interacting chromophores

Let us assume that the system contains two independent identical chromophores the interaction between which is negligible. Let each chromophore is characterized by its pair of occupied and virtual orbital χio and χiv, respectively. (Subscript i = 1 or 2 distinguishes the chromophores.) As noted above, the CIS wave function (3) obtained by using the SVD transformation is built up of the natural orbitals [4]. The natural orbitals of the CIS wave functions should correspond to the symmetry of the

Realistic case of weakly interacting chromophores

In the practice one does not have such idealized cases as discussed in the previous section, in which the CIS wave function consists of only two excitations of equal weights, as described by Eq. (6). One may, however, find cases in which the CIS wave function is dominated by a pair of related excitations:|ΨCISσλ1ψˆ1v+ψˆ1o-+λ2ψˆ2v+ψˆ2o-|HFThe larger the λi values are and the smaller is their difference, the idealized case is better approached. The deviation of the sum λ12+λ22 from the value 12

Identifying a pair of interacting chromophores: a sample calculation

According to the discussion in the previous section, one may identify a pair of interacting similar chromophores – in a chiral environment they may give rise to CD activity – on the basis of CIS calculations in the following manner.

One has to look for a pair of CIS solutions with a close energy, which satisfy the conditions:

  • (a)

    after the SVD transformation have (at least) two λ values significantly differing from zero. The larger are the λ values and the smaller is their difference, the idealized

Acknowledgements

The author is grateful to Prof. P.R. Surján for his useful comments on the manuscript and to Profs. J.G. Ángyan and G. Lendvay for calling his attention to Refs. [6], [7], [8], [9].

References (10)

  • I. Mayer

    Chem. Phys. Lett.

    (2007)
  • P.R. Surján

    Chem. Phys. Lett.

    (2007)
  • C.B. Moler

    Computer Methods for Mathematical Computations

    (1977)
  • I. Mayer, Program CIS-T, Budapest 2006. May be downloaded from the the web-site...
  • Zs. A. Mayer, private...
There are more references available in the full text version of this article.

Cited by (32)

  • Halochromic luminescent quinoxalinones as a basis for pH-sensing in organic and aqueous solutions

    2021, Dyes and Pigments
    Citation Excerpt :

    For CAM-B3LYP calculations all the excitations energies have been consistently red-shifted by 0.5 eV in order to better match the experimental spectra. In order to analyze the nature of the most important excited states Natural Transition Orbitals (NTOs) [57–59] have been calculated allowing for a better and more compact representation of the electronic density reorganization [36,60–65] upon excitation. The NTO representation, based on the singular value decomposition of the transition density matrix, allows to consider most TD-DFT electronic excited state as a transition from only one occupied NTO to a virtual one.

  • Stimuli-responsive emission of quinoxalinone-based compounds. From experimental findings to theoretical insight by means of multiscale computational spectroscopy approaches

    2021, Dyes and Pigments
    Citation Excerpt :

    Each simulated spectrum was obtained by broadening the vertical transitions with Gaussian function using full-width at half maximum of 0.3 eV. To analyze the nature of the involved excited states, Natural Transition Orbitals (NTOs) [71–73] have been calculated allowing for a better and more compact representation of the electronic density reorganization [37,74–79] upon excitation. Indeed, the NTO representation, based on the singular value decomposition of the transition density matrix, allows to represent any TD-DFT electronic excited state as a transition from only one occupied NTO to a virtual one.

  • Theoretical study of the excited state properties of luminescent phospholes

    2019, Dyes and Pigments
    Citation Excerpt :

    This can be clearly seen either via examination of the Khon-Sham frontier molecular orbitals, or even better through the NTO orbitals. Indeed the latter obtained as the singular value decomposition of the transition density matrix represent the most compact representation of the density rearrangement occurring in an electronic transition [47–50]. Remarkably, charge-transfer (CT) is never observed for any of the transition analyzed, as also confirmed by the use of topological indexes such as ϕS, that values close to 1.00 for all the excited states considered in this work.

View all citing articles on Scopus
View full text