I think there are some misconceptions in the optimization you are trying to run. First, Gaussian is a softwares based on localized orbitals, which are not the case for crystalline materials, leading to serious errors in the calculation very often, although it can be used sometimes for some exceptions.

DFT calculations on crystalline materials is better performed using the plane-waves approach, which is implemented in some open-source softwares like Quantum Espresso, SIESTA, CP2K, CASTEP, among ohters. Therefore, I would recommend you to perform a VC-Relax calculation in Quantum Espresso, which will alter both atomic coordinates and cell parameter. The calculation is quite simple and it converges in 10 - 20 cycles, so you can find lots of tutorials in YouTube and in the Quantum Espresso website.

Best wishes.

Convergence error occurs often with perovskite in Gaussian. You need to try different basis sets and functions. It may work.

Here is the link to errors in Gaussian and methods to correct them.

https://docs.alliancecan.ca/wiki/Gaussian_error_messages

Use iop(5/13=1).

You can refer this article which shows DFT optimization for almost same structure.

Article A Novel Organic-Inorganic Ionic Cocrystal - Piperazine-1,4-d...