Although the topic is not trivial, the procedure depends basically on the reasons behind the S0-T1 energy gap calculation.
1) The calculation of the gap may be required to ascertain what is the most stable ground-state multiplicity of a compound whose geometry is known, for example from XRD data. In this case the geometry of both the singlet and triplet configuration must be separately optimised (in the case of the triplet state, an unrestricted approach is required and the spin multiplicity 3 must be declared in the input file). Structural and optimised metric parameters can be subsequently compared.
2) The S0-->T1 ISC process is being investigated. In this case the triplet may (most likely) or may not have (less likely) time enough to relax, depending on the entity of spin-orbit coupling. In the (unlikely) former case, the geometry is the same as the S0 state. Here, the S0 configuration will be optimised and the geometry for the triplet state can be recovered from the chk file of the former calculation. Otherwise, see point 1).
In general, with the only partial exception described above for point 2), an S0-T1 gap calculated at the same energy is not meaningful.
A very particular case occurs when the S0 state derives from a diradical configuration (see below), but this is probably not the case.
If the above mentioned processes are compared to experimental results (for example deriving from luminescence measurements), solvation should be taken into account. In Gaussian this can be done, for instance, within the polarised continuum model (PCM).
****
About this question, I have been working often on such theme, and I would suggest a glance at the following two papers:
1) Article On the role of torsional dynamics in the solid-state fluores...
2) Article Diradical Character of Neutral Heteroleptic Bis(1,2-dithiole...
I am really very thankful for your important suggestion. I am trying to measure S0 to T1 and S1 to T1 gap for predicting fluorescence and phosphorescence. I will be really thankful again if you suggest me how to calculate S1 to T1 energy gap.
The S – T intersystem crossing (ISC) requires that the involved states are close in energy and are more likely if heavy atom are present, especially transition metal ions, to allow the otherwise forbidden state crossing by means of spin orbit (SO).
First, optimise S0 as usual: if your compound is neutral, the line above the z-matrix will be 0,1.
Then start from this optimized geometry and optimize the triplet state (the line above the z-matrix will be 0,3).