Best Number of folds, on cross-validation, for a small dataset

Good Day all. I am fairly new to machine learning. I have a small dataset of 90 items and am running a simple binary classification on it. It is not possible to get that data set any larger as it is running on real world events that only happen a few times a year.

That being said, I can't find a solution on google to this. If I am running k-fold cross-validation on this data, what is the optimal number of folds to use?

Should I use 90-fold (aka leave one out)? 1x10 folds, 10x10 folds, 100x10 folds, 5x2folds, 2x5 folds?

Thanks