Hi,

Firstly, you have used Gaussian which is not free software. Secondly, I think quantum Espresso can do the task if you change some parameters and then compile it. You have also other options too. The best option is to use OpenMX. If it cannot (I hope it can) , you can check with Abinit program. My personal suggestion for you is to use OpenMX, which has been designed for large systems of nano-scale material.

Best Wishes

Amber (http://ambermd.org) is a good choice, it is almost free currently. It can carry out QM/MM calculation, Gaussian is supported as QM module, the Amber itself also has a built-in QM module supporting semi-empirical methods such as PM6 as well as DFTB.

GPU-accelerated QM/MM may be the most suitable choice.

Terachem, the latest version of Gaussian (Gaussian 16) support GPU computing.

Try them. They can deal with really large system.

Take a look at pDynamo. It's a library designed for QM/MM and it is free. It can use its own DFT or externally link to Orca, which is also free for academics.

https://sites.google.com/site/pdynamomodeling/

There is a wiki:

https://sites.google.com/site/pdynamowiki/

and a google group to answer your doubts:

https://groups.google.com/forum/#%21forum/pdynamo

For large systems I think Terachem might be a good choice, as a previous comment has already pointed out. Also for large systems, in particular when the QM region is large, I suggest you have a look at CP2K: https://www.cp2k.org/ , besides its efficiency as an ab initio MD engine, you can also run QMMM MD in cp2k using e.g. CHARMM FF, cp2k has an efficient implementation of the QM/MM electrostatic coupling that is suitable for large systems:

Article An Efficient Real Space Multigrid QM/MM Electrostatic Coupling

Article An Efficient Linear-Scaling Electrostatic Coupling for Treat...

you also have the option in cp2k to do a force mixing calculation where atoms are automatically included/excluded in the QM region when they diffuse into/out of it.

Article Hybrid atomistic simulation methods for materials systems

GPU-accelerated QM/MM is suitable choice for large systems.

While many people have thrown out software suggestions, I will throw out a strategy. I had a colleague who worked on polymers consisting of well over 400 atoms and she had to test many different chain configurations. Her solution was to use fast empirical potentials in LAMMPS to perform the MD runs and then perform expensive QM/DFT calculations on the structures.

Please let us know what software you chose and your experience with it because if you find something that it free, accurate, fast on a low number of cores, and user friendly we should probably all be using it =)

Gamess-US

Gaussian is not free, try pDynamo and Orca

GPU-accelerated QM/MM