Computational Methods in Organometallic Catalysis. Yu Lan

      which is used to construct free‐energy profiles. The calculated free‐energy profiles for all of the possible reaction pathways are considered by energy criterion. The most energetically feasible pathway usually is considered to be the best reaction mechanism for a specific reaction. Herein, initial geometries are acquired by conformational searches with force fields. Input geometries of the transition structures are identified either from experience or through scanning along the reaction coordinate.

      If results are in agreement with experiment, we analyze the transition states and the structural factors responsible for asymmetric induction are proposed. Qualitative models are developed to understand and to make new predictions for catalyst and reaction design.

2.7 Overview of Popular Computational Programs

Relevant calculation programs are often needed to implement various calculation methods. In computation organic and organometallic chemistry, various programs are provided for theoretical calculations involving Gaussian, ADF, ORCA, GAMESS, Molpro, MOLCAS, NWchem, Q‐Chem, etc. [96–103].

      Undoubtedly, Gaussian series is the most frequently used program for computational chemistry, which is also the most popular one in computational organic and organometallic chemistry [96]. It is capable of predicting many properties of molecules and reactions, including molecular energies and structures, energies and structures of transition states, bond and reaction energies, molecular orbitals, multipole moments, atomic charges and electrostatic potential, vibrational frequencies, NMR properties, and reaction pathways. Computation can be carried out on systems in the gas phase or in solutions, and in their ground state or in an excited state. It is very flexible in the choice of basis sets, ECPs, and density functionals, and it has excellent geometry optimizers and initial guesses for the SCF iterations. It supports analytic Hessians for all functionals, and has classical dynamics capabilities and a very useful interface to external programs. Gaussian supports CPU parallelization with TCP Linda in the distributed memory multiprocessor environments, and GPU acceleration in the latest version Gaussian 16. Gaussian is commercially available with single, site, and Lina licenses. Website: https://gaussian.com/.

      Q‐Chem is a comprehensive ab initio quantum chemistry software for accurate predictions of molecular structures, reactivities; and vibrational, electronic, and NMR spectra [102]. The package is available either as a standalone code or integrated with the Spartan package. Fully integrated graphic interface IQmol includes molecular builder, input generator, contextual help, and visualization toolkit. The new release of Q‐Chem 5 represents the state of the art of methodology from the highest‐performance DFT/HF calculations to high‐level post‐HF correlation methods. Q‐Chem supports ONIOM model and an interface of CHARMM for QM/MM calculation. Q‐Chem is commercially available with academic, government, and industry licenses. Website: http://www.q-chem.com/.

      ORCA is an ab initio and semiempirical quantum chemistry program and has become one of the most widely used programs developed in the research group of Professor Frank Neese [98]. Various modern electronic structure methods have been implemented, including DFT, single‐reference correlation methods, multireference correlation methods, local correlation methods, excited state methods, solvation, solids and surfaces, and semiempirical quantum chemistry methods. ORCA has special features of robustness, efficiency, and focus on transition metals, and spectroscopic properties. QM/MM calculations could be performed using the interfaces of external programs, such as ASE, Gromas, NAMD, and Chemshell. Academic use version at academic institutions is freely available, and the commercial version is also available. Website: https://orcaforum.kofo.mpg.de/app.php/portal.

NWChem is a high‐performance computational chemistry code with ab initio, band‐structure, molecular mechanics, and molecular dynamics methods [103, 104]. This program achieves an efficient computational chemistry code with high parallel scalability of running on tens of thousands of processors. The ab initio methods available include HF, DFT, MPn, multiconfiguration self‐consistent field (MCSCF), CI, and CC. There are a large number of basis sets available, including ECP sets. Ab initio dynamic calculations can be performed at respective levels of theory. QM/MM minimization and dynamics calculations are also possible. This computational chemistry code is fully open source from Pacific Northwest National Laboratory. Website: http://www.nwchem-sw.org/.

      Turbomole is a software package for large‐scale quantum chemical simulations of molecules, clusters, and periodic solids [105]. Turbomole has electronic structure methods with excellent performance, such as time‐dependent DFT, second‐order Møller–Plesset theory, and explicitly correlated CC methods. These methods are implemented by efficient algorithms such as integral‐direct and Laplace transform methods, resolution‐of‐the‐identity, pair natural orbitals, fast multipole, and low‐order scaling techniques. The properties of optics, electrical density, and magnetics are accessible for ground and excited states. Recently, new methods were released such as post‐Kohn–Sham calculations within the random‐phase approximation, periodic calculations, spin–orbit couplings, explicitly correlated CC singles doubles and perturbative triples methods, CC singles doubles excitation energies, and nonadiabatic molecular dynamics simulations using time‐dependent density functional theory (TDDFT). This code is commercially available. Website: https://www.turbomole.org/.

      The general atomic and molecular electronic structure system (GAMESS (United States)) package has many electronic structure capabilities [99]. The electronic structure methods have been implemented, including Hartree–Fock method, DFT, generalized valence bond, and multi‐configurational SCF. Post‐SCF correlation corrections can be performed by configuration interaction, second‐order Møller–Plesset perturbation theory, and CC theory. Solvent effect can be considered explicitly by QM/MM calculations or inexplicitly by PCM (polarized continuum models). The third‐order Douglas–Kroll scalar terms were used to calculated relativistic corrections. For large systems, using fragment molecular orbital method is highly recommended. Particularly, it has the interfaces to the valence bond programs, like VB2000 and Xiamen valence bond (XMVB). This computational chemistry code is fully open sourced available. Website: https://www.msg.chem.iastate.edu/gamess.

      The Amsterdam Density Functional (ADF) is a program for first‐principles electronic structure calculations in understanding and predicting structure, reactivity, and spectra of molecules [97]. The program can run various types of calculation, including optimization of geometry, transition structure optimization, frequency analysis, and intrinsic reaction coordinate (IRC) calculation and electronic excited states. Solvation effects can be considered using the conductor‐like screening models (COSMO) method. External electric fields, point charges, and relativistic correction calculations can be also performed. Molecular properties such as ESR and NMR simulations, Raman intensities, and hyperpolarizabilities can be computed. The “BAND” module is used for periodic materials to perform a single‐point calculation. This code is commercially available. Website: https://www.scm.com/product/adf/.

Molpro is an ab initio program package for advanced molecular electronic structure calculations [100, 106]. The package has state‐of‐the art methods with efficient parallelized algorithms, such as DFT, high‐level CC and multireference wavefunction methods, such as MCSCF/CASSCF (complete active space self‐consistent field), CASPT2, multireference configuration interaction (MRCI), or full configuration interaction (FCI) methods, or response methods such as TDDFT, CC2, and EOM‐CCSD. Computing modules include molecular properties, geometry optimization, vibrational frequencies, and further wavefunction analysis. Density‐fitting approximations can speed up DFT and MP2 calculations for large systems. F12 explicitly correlated methods with triple‐zeta