I'm assuming you want to calculate the interaction energy between one molecule of solute and one explicit molecule of solvent. If the goal is to compute the solvation energy of a solute you might also consider using an implicit model, or a cluster model with several solvent molecules, however, I will focus on your specific question:

You can certainly build/draw the two molecules close to each other in GaussView and then perform a geometry optimization in Gaussian to generate a geometry for the solute/solvent complex. However, unless you are very fortunate, or your molecules are very simple indeed, this is likely to result in a local rather than the global energy minimum, and your computed interaction energy will be too small.

Therefore, you'll need to sample various configurations to find which is the most stable. In the literature you'll find various approaches, ranging from manual/chemical intuition (build a few different starting structures and optimize), sampling with Monte Carlo or molecular dynamics (probably with molecular mechanics), and some more recent protocols involving stochastic solvent placement (https://arxiv.org/abs/1909.06664).

I would recommend trying something like CREST with GFN2-xTB to sample lots of configurations and then using the most stable structures to start some Gaussian optimizations.

Robert Paton thank you for your answer. I agree with you. I'm also curious what do you think about the description of the calculation of binding energies in the literature supplied in attachment file. The reason can be attributed to their small molecules and fortune? In fact, most of the descriptions in literature are similar in my research field, only describing that the binding site was determined by the electrostatic potential surface, and then there is no mention of how the solute-solvent complex comes from

Dear Zhi Chen , there is a possible confusion. The text you attached are related to an interaction between two molecules, namely A and B and solvent is dealt with implicitly. On the contrary, you asked about a solute-solvent interaction, that is what Robert Paton replayed (correctly) to you about.

Dear Massimiliano Arca ，thank you for your reminder. I'm sorry that my description misled you. Exactly, I want to know how they deal with the solvent and build a solute-solvent complex for calculation.

Thanks for clarifying. The route is as follow:

1) draw you solute/solvent A in GaussView or build your z-matrix in any other way and optimize it at the level of theory of choice including implicit solvation. Add a frequency calculation to evaluate ZPE and thermochemical data.

2) do the same for the second molecule of solvent B (same level of theory, same basis sets, same implicit solvation model).

3) optimize the solvent/solvent or solute/solvent ensemble A···B (same level of theory, same basis sets, same implicit solvation model). For its starting geometry see what Robert Paton reported above. If possible, you can refer to existing XRD structures of solvates and statistically identify a likely interaction conformation. A search on the CSD helps for sure.

Now you can use thermochemical data for evaluate the interaction enthalpy and/or free energy. You may want to refer to this white paper: https://gaussian.com/thermo/

The calculation may be further corrected for BSSE. I have reported related interaction studies several times, especially for investigating the reactions of chalcogen donors with dihalogens.

Massimiliano Arca Robert Paton thank you