So there is a known issue with vdW interactions:

van der Waals Parameter Scanning with Amber Nucleic Acid Force Fields: Revisiting Means to Better Capture the RNA/DNA Structure through MD

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10809421/

Also aggregation occuring in GLYCAM with glucose that can be fixed by adjusting LJ interactions:

Optimizing Solute−Solute Interactions in the GLYCAM06 and
CHARMM36 Carbohydrate Force Fields Using Osmotic Pressure
Measurements

https://outlookuga-my.sharepoint.com/:b:/g/personal/olgrant_uga_edu/ETa7PO39Sq9DnxC0jLw9zyIBmItFEEwsoPdV9sk_j8ciTQ?e=UOc34K

Maybe you can do this in Amber by editing the topology file LJ section. Rob doesn’t like how much they edit the values (see page 28 here):
https://pubs.acs.org/doi/suppl/10.1021/acs.jctc.5b01136/suppl_file/ct5b01136_si_001.pdf

As it’s unrealistic. I wonder if it’s because they are running in tip3p? Or that charges also need to be changed?