This journal is © The Royal Society of Chemistry 2023 Chem. Commun. Cite this: DOI: 10.1039/d2cc07079k Continuum vs. atomistic approaches to computational spectroscopy of solvated systems Tommaso Giovannini and Chiara Cappelli * Molecular spectral signals can be significantly altered by solvent effects. Among the many theoretical approaches to this problem, continuum and atomistic solvation models have emerged as the most effective for properly describing solvent effects on the spectroscopic signal. In this feature article, we review the continuum and atomistic descriptions as applied to the calculation of molecular spectra, by detailing the similarities and differences between the two approaches from the formal point of view and by analyzing their advantages and disadvantages from the computational point of view. Various spectral signals, of increasing complexity, are considered and illustrative examples, selected to exacerbate the differences between the two approaches, are discussed. 1. Introduction The role of computational chemistry to assist experimental studies has grown substantially in the last decades. This is due to the increasing availability of user-friendly computational packages, which can be used also by non-specialists. Further- more, the success of computational approaches follows the development of methods that can treat with considerable accuracy large, complex, and realistic chemical systems. 1–5 In parallel, computational spectroscopy, i.e. the calculation of spectral properties by means of computational methods, has become an invaluable tool in most fields of chemical research. 6 In fact, thanks to the high level of accuracy that has been reached in the reproduction of experimental data in the condensed phase, it often provides a theoretical rationali- zation of many experimental findings, yielding truly synergistic investigations. 7–9 In fact, although many algorithms of increas- ing accuracy have been proposed and tested for systems in the gas phase, 10–12 the large majority of molecular spectra are routinely measured in the condensed phase, i.e. when chemical systems are dissolved in a solvent or other kinds of external environments (biological matrix, polymeric materials, crystal phase etc.). Indeed, molecular structure and response to exter- nal electromagnetic fields can be significantly altered by the environment, thus making the simulation for isolated systems mostly inappropriate for a reliable comparison with experi- mental findings. 13–16 Effective theoretical modeling of spectroscopy in the con- densed phase requires catching the physico-chemical features of the simultaneous interaction of a chemical system with the environment and the external radiation field. 1,17–20 Since spectroscopy arises from the interaction between the molecule and the radiation, which is in general driven by the electronic component, reliable modeling needs to resort to Quantum- Mechanical (QM) descriptions. However, while such appro- aches may be feasible for small-to-medium molecules in the gas phase, for systems in the condensed phase a ‘‘brute-force’’ description, i.e. which treats all atoms at the QM level, is unfeasible due to the enormous number of degrees of freedom that would need to be taken into account. However, such a description would be useless in most cases; in fact, the electro- nic structure of the environment would be described, however it commonly does not drive the spectral signal of chemical systems in the condensed phase. 21,22 In fact, it is universally recognized that molecular spectra are modified but not deter- mined by the environment, and this is indeed an important concept, which lays the foundations for the spectroscopic identification of molecular structures. Focused models 1,13,19,23–25 are the most successful class of approaches to computational spectroscopy in the condensed phase. There, the focus is the target molecule (solute in case of solutions) and the key is the accurate modeling of molecule/ environment interactions and their consequences on the mole- cular structure and properties. 17,18,20,26 The intrinsic spectral properties of the environment are not modelled, which corre- sponds to assuming molecular properties to be local properties of the molecule, which can be modified but not determined by the presence of the environment. The most renowned class of focused models belongs to the family of multiscale (multilayer) QM/classical approaches. 23 The latter have had great success in modern chemical research because they can be effectively coupled with most QM descrip- tions, ranging from semi-empirical methods 27,28 to Density Scuola Normale Superiore, Piazza dei Cavalieri 7, 56126 Pisa, Italy. E-mail: chiara.cappelli@sns.it Received 30th December 2022, Accepted 13th March 2023 DOI: 10.1039/d2cc07079k rsc.li/chemcomm ChemComm FEATURE ARTICLE Open Access Article. Published on 19 April 2023. Downloaded on 4/21/2023 9:32:47 AM. This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence. View Article Online View Journal