Final Report Summary - DEDOM (Development of Density Functional Theory methods for Organic Metal Interaction)
The DEDOM project aims at developing new theoretical/computational methods to describe the electronic/optical properties of organic molecules interacting with noble metal surfaces and metal nanoparticles (MNPs). This research has a direct impact on the development of more efficient devices for organic optoelectronics, molecular electronics and molecular plasmonics. The main task of the DEDOM project is the development of exchange-correlation (XC) functionals of the density-functional theory (DFT). The exact XC functional is unknown and different successful approximations are available in literature, e.g. the famous Perdew-Burke-Ernzerhof (PBE) functional. However, many limitations are still present, especially for the description of hybrid organic-metal interfaces. A second main task of the DEDOM project is the development of multiscale computational approaches to describe the optical properties of hybrid interfaces and of MNPs interacting with neighbouring organic molecules, including lifetime modifications due to the coupling of molecular excitations to localized surface plasmon resonances (LSPRs). The main results of the DEDOM project can be subdivided in five groups:
1) Semilocal XC functionals. We developed several new non-empirical semilocal XC functionals: PBEint, a Generalized Gradient Approximation (GGA) which interpolates between different density regimes and yields very good accuracy for the description of hybrid interfaces and metallic clusters; APBE, a GGA derived from the semiclassical neutral atom (SCA) theory, which yields excellent performances for molecular systems; Q2D-GGA, an accurate GGA for quasi-two dimensional systems, including surfaces of transition metals; BLOC, a meta-GGA with localized correlation energy-density (to increase the compatibility with the exact-exchange), with excellent performances for different systems (molecules, metal clusters, surfaces, and bulk solids).
2) Orbital-dependent XC functionals. We extended and optimized an effective exact-exchange method, namely the P.I.’s workhorse Localized Hartree Fock (LHF) approach, and we applied it to study the electronic properties of metal clusters and hybrid interfaces. We found that the LHF method can correctly describe the density of states of gold clusters and the energy-level alignment at the metal-molecule interface. We developed a new correlated optimized effective potential (OEP) method based on the Scaled-Opposite-Spin (SOS) second-order correlation. This method has an accuracy/cost ratio much larger than other OEP methods based on the second-order correlation.
3) Frozen density embedding (FDE) theory methods. To describe large and complex systems such as molecule-MNPs, a subsystem formulation of DFT, such as the FDE theory, offers high accuracy and reduced computational costs. The FDE theory is an exact theory, except for the approximations needed for the non-additive kinetic-energy (KE) and for the XC functional. Concerning the former, we developed a new non-empirical semilocal KE functional (revAPBEK), which is based on the SCA theory and outperforms the current state of the art, and KE functionals which use the Laplacian of the density as an ingredient. Concerning the XC functional, we extended the FDE theory, previously limited to the use of GGA functionals, to hybrid and orbital-dependent functionals. We found that the use of exact-exchange improves the description of the ground-state electronic properties of different systems and leads to embedding errors smaller than with GGA functionals.
4) Multiscale methods for surfaces and interfaces. To describe the electronic properties of self-assembled organic monolayers (SAM) on metal substrates we developed an efficient ab initio embedding scheme, the Self-consistent Periodic Image Charge Embedding (SPICE) which allows the description of depolarization effects using accurate orbital-dependent methods. We found that GGA XC functionals rely on error cancellation between the concurrent overestimation of dipole and polarizability, while a correct description can be only obtained with the LHF approach or with hybrid DFT methods. We developed the Periodic Charge-Dipole Electrostatic Model (PCDEM) which allows to describe the linear response of metallic slabs in terms of charge- and dipole-type Gaussian basis functions: for higher accuracy a kinetic-exchange-correlation correction was also considered.
5) Computational plasmonics. To model the electromagnetic interaction between organic molecules and LSPRs, we describe the former as point-dipoles and the latter using the Discrete Dipole Approximation (DDA). We implemented an efficient and parallel tool to compute the near-field around a MNP of arbitrary shape and dimension. We investigated the metal-enhanced fluorescence of emitters near silver plasmonic substrates and we found that covering oxide layers lead to an unexpected increase of the near-field enhancement. We extended the DDA method to compute the radiative and non-radiative lifetimes of molecules near MNPs.
1) Semilocal XC functionals. We developed several new non-empirical semilocal XC functionals: PBEint, a Generalized Gradient Approximation (GGA) which interpolates between different density regimes and yields very good accuracy for the description of hybrid interfaces and metallic clusters; APBE, a GGA derived from the semiclassical neutral atom (SCA) theory, which yields excellent performances for molecular systems; Q2D-GGA, an accurate GGA for quasi-two dimensional systems, including surfaces of transition metals; BLOC, a meta-GGA with localized correlation energy-density (to increase the compatibility with the exact-exchange), with excellent performances for different systems (molecules, metal clusters, surfaces, and bulk solids).
2) Orbital-dependent XC functionals. We extended and optimized an effective exact-exchange method, namely the P.I.’s workhorse Localized Hartree Fock (LHF) approach, and we applied it to study the electronic properties of metal clusters and hybrid interfaces. We found that the LHF method can correctly describe the density of states of gold clusters and the energy-level alignment at the metal-molecule interface. We developed a new correlated optimized effective potential (OEP) method based on the Scaled-Opposite-Spin (SOS) second-order correlation. This method has an accuracy/cost ratio much larger than other OEP methods based on the second-order correlation.
3) Frozen density embedding (FDE) theory methods. To describe large and complex systems such as molecule-MNPs, a subsystem formulation of DFT, such as the FDE theory, offers high accuracy and reduced computational costs. The FDE theory is an exact theory, except for the approximations needed for the non-additive kinetic-energy (KE) and for the XC functional. Concerning the former, we developed a new non-empirical semilocal KE functional (revAPBEK), which is based on the SCA theory and outperforms the current state of the art, and KE functionals which use the Laplacian of the density as an ingredient. Concerning the XC functional, we extended the FDE theory, previously limited to the use of GGA functionals, to hybrid and orbital-dependent functionals. We found that the use of exact-exchange improves the description of the ground-state electronic properties of different systems and leads to embedding errors smaller than with GGA functionals.
4) Multiscale methods for surfaces and interfaces. To describe the electronic properties of self-assembled organic monolayers (SAM) on metal substrates we developed an efficient ab initio embedding scheme, the Self-consistent Periodic Image Charge Embedding (SPICE) which allows the description of depolarization effects using accurate orbital-dependent methods. We found that GGA XC functionals rely on error cancellation between the concurrent overestimation of dipole and polarizability, while a correct description can be only obtained with the LHF approach or with hybrid DFT methods. We developed the Periodic Charge-Dipole Electrostatic Model (PCDEM) which allows to describe the linear response of metallic slabs in terms of charge- and dipole-type Gaussian basis functions: for higher accuracy a kinetic-exchange-correlation correction was also considered.
5) Computational plasmonics. To model the electromagnetic interaction between organic molecules and LSPRs, we describe the former as point-dipoles and the latter using the Discrete Dipole Approximation (DDA). We implemented an efficient and parallel tool to compute the near-field around a MNP of arbitrary shape and dimension. We investigated the metal-enhanced fluorescence of emitters near silver plasmonic substrates and we found that covering oxide layers lead to an unexpected increase of the near-field enhancement. We extended the DDA method to compute the radiative and non-radiative lifetimes of molecules near MNPs.