Your aim will be to reliably increase the distances between the particles. That calls for a reliable method to disperse them first and then e.g. dilute them in some polymer solution which could then be solidified.

In any case, you will need nome analytical tools (e.g. low energy SEM) to verify that the procedure is successful.

Since solubility is different depending on surface termination, I cannot think of a generic recipe. In case your particles are terminated by specific ligands, ligand exchange may be a strategy to achieve solubility in a desirent solvent or dispersion. Without comparison, it might then be hard to detect whether ligand exchange did something to the magnetic properties of the particles.

It can be a major effort to reliably generate samples with ensembles of quasi non-interacting particles.

There are magnetic test measurements, however, which can tentatively be used, in particular in the superparamagnetic regime (sufficiently high temperature). Even if the particles are non-identical, the magnetization must scale as M(H/T). Therefore, measuring isotherms M(H,T) and rescaling for the temperature as shown above must produce superimposing curves. If they don't, then either the particles are not superparamagnetic or there are considerable interactions.

If the Curie temperature is not very high, then you can choose proper polimer compound to be melted at T

Check this paper. There is a recipe for particle separation and testing the quality of the separation by the FMR. The same test can be performed using VSM. Lack of anisotropy of composite film - particles in a nonmagnetic matrix is an indication of the "off" dipole-dipole bonds between the particles.

http://scitation.aip.org/content/aip/journal/adva/1/4/10.1063/1.3657510

I have experience with this topic. Take Scotch tape and roll out a 20" strip. Deposit nanopowders on the sticky side.

Then cover with another 20" strip sticky side down. Rolll with pressure and then curl it up in a roll like toilet paper.

Use this as the core of a transformer and  create the hysterisis loop.

That should be a good start.

Scotch tape method reminds me early days of  graphene monolayers preparation)))

Grind it into a very very small particles, of which the thermal energy KT is comparable to the particle interaction. At this time, the magnetic properties of the particles may change from ferromagnetic to superparamagnetic at low temperature.

As you mentioned the interaction change the magnetic properties of samples. So if your samples are interacting then you should not change the powders. Otherwise if you want to have non-interacting samples, it is better to change the synthesis method (solvothermal using a polymer could be useful).

Now we assume that you prepared aggregated samples. And you want to minimize the interactions effect. You can use a gentile grinding by hand to separate the nanoparticles as possible. Also you can disperse the powders in a polymer (non-magnetic). Also using the method mentioned by Brian could be helpful.

I send you here some of my papers related to your question, and I hope they are useful for you.

http://www.sciencedirect.com/science/article/pii/S030488531100535X

http://www.sciencedirect.com/science/article/pii/S1567173911005852

http://www.sciencedirect.com/science/article/pii/S0304885315306296