ONIOM Investigation of the Second-Order Nonlinear Optical Responses of Fluorescent Proteins

The eXact integral simplified time-dependent density functional theory (XsTD-DFT)

In the framework of simplified quantum chemistry methods, we introduce the eXact integral simplified time-dependent density functional theory (XsTD-DFT). This method is based on the simplified time-dependent density functional theory (sTD-DFT), where all semi-empirical two-electron integrals are replaced by exact one- and two-center two-electron integrals, while other approximations from sTD-DFT are kept. The performance of this new parameter-free XsTD-DFT method was benchmarked on excited state and (non)linear response properties, including ultra-violet/visible absorption, first hyperpolarizability, and two-photon absorption (2PA). For a set of 77 molecules, the results from the XsTDA approach were compared to the TDA data. XsTDA/B3LYP excitation energies only deviate on average by 0.14 eV from TDA while drastically cutting computational costs by a factor of 20 or more depending on the energy threshold chosen. The absolute deviations of excitation energies with respect to the full scheme are decreasing with increasing system size, showing the suitability of XsTDA/XsTD-DFT to treat large systems. Comparing XsTDA and its predecessor sTDA, the new scheme generally improves excitation energies and oscillator strengths, in particular, for charge transfer states. TD-DFT first hyperpolarizability frequency dispersions for a set of push-pull π-conjugated molecules are faithfully reproduced by XsTD-DFT, while the previous sTD-DFT method provides redshifted resonance energy positions. Excellent performance with respect to the experiment is observed for the 2PA spectrum of the enhanced green fluorescent protein. The obtained robust accuracy similar to TD-DFT at a fraction of the computational cost opens the way for a plethora of applications for large systems and in high throughput screening studies.

Ultrafast Evaluation of Two-Photon Absorption with Simplified Time-Dependent Density Functional Theory

This work presents the theoretical background to evaluate two-photon absorption (2PA) cross-sections in the framework of simplified time-dependent density functional theory (sTD-DFT). Our new implementation allows the ultrafast evaluation of 2PA cross-sections for large molecules based on a regular DFT ground-state determinant as well as a variant employing our tight-binding sTD-DFT-xTX flavor for very large systems. The method is benchmarked against higher-level calculations for trans-stilbene and typical fluorescent protein chromophores. For eGFP, a quadrupolar chromophore and its branched version, the flavine mono-nucleotide, and the iLOV protein, we compare sTD-DFT 2PA spectra to experimental ones. This includes extension and testing of our all-atom quantum chemistry methodology for the evaluation of 2PA for a system of ∼2000 atoms, providing striking agreement with the experimental spectrum.

TDDFT Investigation of the Raman and Resonant Raman Spectra of Fluorescent Protein Chromophore Models

The off-resonance and resonant Raman spectra have been simulated for models of fluorescent protein chromophores, those of the green fluorescent protein (GFP, called FP1) and of DsRed (called FP2), which presents a longer π-conjugated path, with the aim of providing a systematic investigation of structural but also computational aspects. These were performed at the (time-dependent) density functional theory [(TD)DFT] level. The off-resonance intensities have been calculated from the derivatives of the frequency-dependent polarizability with respect to the normal coordinates while the resonant ones have been evaluated using Huang-Rhys factors determined from the gradients of the excitation energies with respect to the normal coordinates. When applied with the M05 meta-GGA exchange-correlation functional, this simple computational scheme can reproduce most of the experimental Raman signatures of FP1 in its protonated and deprotonated forms, the differences of vibrational signatures of the cis (Z) and trans (E) isomers, as well as their changes as a function of the excitation wavelength. On the other hand, testing the predictions made for FP2 would require new experimental work. It was also observed that simulations with methods that inadequately predict the resonant Raman spectra could nevertheless produce a UV-vis absorption spectrum that is quite similar to the one obtained with better methods, once realistic peak broadening has been applied.

Modeling the First-Order Molecular Hyperpolarizability Dispersion from Experimentally Obtained One- and Two-Photon Absorption

The search for optical materials, particularly organic compounds, is still an attractive and essential field for developing several photonic devices and applications. For example, some applications are based on light scattering with twice the energy of the incoming photon for selected compounds, that is, the nonlinear optical effect related to the second-order susceptibility term from the electronic polarization expression. The microscopic interpretation of this phenomenon is called the first-order molecular hyperpolarizability or incoherent second harmonic generation of light. Understanding such phenomena as a function of the incoming wavelength is crucial to improving the optical response of future materials. Still, the experimental apparatus, hyper-Rayleigh scattering, apparently simple, is indeed a challenging task. Therefore, we proposed a proper alternative to obtain the dispersion of the first-order hyperpolarizability using the well-known one- and two-photon absorption techniques. By the spectral analysis of both the spectra, we gathered spectroscopic parameters and applied them for predicting the first-order hyperpolarizability dispersion. This prediction is based on an n-level energy system, taking into account the position and magnitude of transition dipole moments and the difference between the permanent dipole moment of the n-excited states. Moreover, using the presented method, we can avoid underestimating the first-order hyperpolarizability by not suppressing higher-energy transitions. Quantum chemical calculations and the hyper-Rayleigh scattering technique were used to validate the proposed method.

First hyperpolarizability of water at the air-vapor interface: a QM/MM study questions standard experimental approximations

Surface Second-Harmonic Generation (S-SHG) experiments provide a unique approach to probe interfaces. One important issue for S-SHG is how to interpret the S-SHG intensities at the molecular level. Established frameworks commonly assume that each molecule emits light according to an average molecular hyperpolarizability tensor β(-2ω,ω,ω). However, for water molecules, this first hyperpolarizability is known to be extremely sensitive to their environment. We have investigated the molecular first hyperpolarizability of water molecules within the liquid-vapor interface, using a quantum description with explicit, inhomogeneous electrostatic embedding. The resulting average molecular first hyperpolarizability tensor depends on the distance relative to the interface, and it practically respects the Kleinman symmetry everywhere in the liquid. Within this numerical approach, based on the dipolar approximation, the water layer contributing to the Surface Second Harmonic Generation (S-SHG) intensity is less than a nanometer. The results reported here question standard interpretations based on a single, averaged hyperpolarizability for all molecules at the interface. Not only the molecular first hyperpolarizability tensor significantly depends on the distance relative to the interface, but it is also correlated to the molecular orientation. Such hyperpolarizability fluctuations may impact the S-SHG intensity emitted by an aqueous interface.

All-Atom Quantum Mechanical Calculation of the Second-Harmonic Generation of Fluorescent Proteins

Fluorescent proteins (FPs) are biotags of choice for second-harmonic imaging microscopy (SHIM). Because of their large size, computing their second-harmonic generation (SHG) response represents a great challenge for quantum chemistry. In this contribution, we propose a new all-atom quantum mechanics methodology to compute SHG of large systems. This is now possible because of two recent implementations: the tight-binding GFN2-xTB method to optimize geometries and a related version of the simplified time-dependent density functional theory (sTD-DFT-xTB) to evaluate quadratic response functions. In addition, a new dual-threshold configuration selection scheme is introduced to reduce the computational costs while retaining overall similar accuracy. This methodology was tested to evaluate the SHG of the proteins iLOV and bacteriorhodopsin (bR). In the case of bR, quantitative agreement with respect to experiment was reached for the out-of-resonance low-energy part of the β_HRS frequency dispersion. This work paves the way toward an accurate prediction of the SHG of large structures-a requirement for the design of new and improved SHIM biotags.

Unravelling the Effects of Cholesterol on the Second-Order Nonlinear Optical Responses of Di-8-ANEPPS Dye Embedded in Phosphatidylcholine Lipid Bilayers

Cholesterol is known for its role in maintaining the correct fluidity and rigidity of the animals cell membranes and thus their functions. Assessing the content and the role of cholesterol in lipid bilayers is therefore of crucial importance for a deeper understanding and control of membrane functioning. In this computational work, we investigate bilayers built from three types of glycerophospholipid phosphatidylcholine (PC) lipids, namely dipalmitoylphosphatidylcholine (DPPC), 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC), and dioleoylphosphatidylcholine (DOPC), and containing different amounts of cholesterol by analyzing the second-harmonic generation (SHG) nonlinear optical (NLO) response of a probe molecule, di-8-ANEPPS, inserted into the membranes. This molecular property presents the advantage to be specific to interfacial regions such as lipid bilayers. To unravel these effects, Molecular Dynamics (MD) simulations have been performed on both DPPC and DOPC lipids by varying the cholesterol mole fraction (from 0 to 0.66), while POPC was only considered as a pure bilayer. In the case of the structural properties of the bilayers, all the analyses converge toward the same conclusion: as the mole fraction of cholesterol increases, the systems become more rigid, confirming the condensing effect of cholesterol. In addition, the chromophore is progressively more aligned with respect to the normal to the bilayer. On the contrary, addition of unsaturation disorders the lipid bilayers, with barely no impact on the alignment of the chromophore. Then, using the frames obtained from the MD simulations, the first hyperpolarizability β of the dye in its environment has been computed at the TDDFT level. On the one hand, the addition of cholesterol induces a progressive increase of the diagonal component the β tensor parallel to the bilayer normal. On the other hand, larger β values have been calculated for the unsaturated than for the saturated lipid systems. In summary, this study illustrates the relationship between the composition and structure of the bilayers and the NLO responses of the embedded dye.

A systematic study of the effects of thionation in naphthalene dimide derivatives to tune their nonlinear optical properties

In the present study, the number and position of sulfur atoms on naphthalene diimide (NDI) is systematically investigated to tune its nonlinear optical (NLO) response properties. Our DFT calculations for third-order polarizability (γ) show that the thionation significantly influences the nonlinear optical property of NDI as it is seen among its several designed derivatives (NDI-1 to NDI-10). The smallest and the largest γ_zzzz amplitudes are 503.49 × 10^-36 and 1299.5 × 10^-36 esu for NDI-1 (having tetraone group) and NDI-10 (having tetrathione group), respectively. The increase in γ_zzzz amplitude for NDI-10 is 796 × 10^-36 esu, which is ∼150% from the γ_zzzz amplitude of NDI-1. A comparison of the γ_zzzz amplitudes of our designed derivatives are made with para-nitroaniline i.e. a prototype NLO molecule. The γ_zzzz amplitude of pNA is found to be 42.64 × 10^-36 esu at the same B3LYP/DZVP2 level of theory. Using two-level model, the origin of larger γ_zzzz amplitudes is traced in lower transition energy of NDI-10. Furthermore, the calculation of vertical ionization potentials (VIPs) shows that the thionation does not affect the stability of designed derivatives, where a slight difference of 0.06 eV is seen between the VIPs of NDI-1 (6.63 eV) and NDI-10 (6.57 eV). Thus, a systematic comparison of the third-order polarizability and other electro-optical properties of our designed derivatives shows that our derivative systems possess good potential for their practical realization in the field of optical and NLO materials.

Nonlinear-response properties in a simplified time-dependent density functional theory (sTD-DFT) framework: Evaluation of the first hyperpolarizability

Recent developments in nonlinear imaging microscopy show the need to implement new theoretical tools, which are able to characterize nonlinear optical properties in an efficient way. For second-harmonic imaging microscopy (SHIM), quantum chemistry could play an important role to design new exogenous dyes with enhanced first hyperpolarizabilities or to characterize the response origin in large endogenous biological systems. Such methods should be able to screen a large number of compounds while reproducing their trends and to treat large systems in reasonable computation times. To fulfill these requirements, we present a new simplified time-dependent density functional theory (sTD-DFT) implementation to evaluate the first hyperpolarizability where the Coulomb and exchange integrals are approximated by short-range damped Coulomb interactions of transition density monopoles. For an ultra-fast computation of the first hyperpolarizability, a tight-binding version (sTD-DFT-xTB) is also proposed. In our implementation, a sTD-DFT calculation is more than 600 time faster with respect to a regular TD-DFT treatment, while the xTB version speeds up the entire calculation further by at least two orders of magnitude. We challenge our implementation on three test cases: typical push-pull π-conjugated compounds, fluorescent proteins, and a collagen model, which were selected to model requirements for SHIM applications.