You want to make NP's aggregate? Increase the ionic strength of solution (concentration of salts) so that electrostatic screening depth and repulsion between NP's is decreased. Or change th pH of solution - depending on how NP's are functionalized in some pH range their zeta-potential will be close to zero and therefore they will start agglomerating.

Process of nanoparticles agglomeration is absolutely normal. The degree of agglomeration depended to concentration, size and kind of the particles. But for agglomerates destruction usually sufficient the action of ultrasound. What mechanism of nanoparticles deposition in your experiments? And what materials do you use?

Measure the zeta pontianzial of nanoparticle suspension. Above +30 or below -30 prevents the agglomeration

Agglomeration of nanoparticles can be prevented by either charge stabilization or steric stabilization. Usually , this takes place due to attractive forces like van der Waal's force. One has to counteract this with repulsive force. Charge stabilization is suitable if surrounding medium is polar like water. In colloid chemistry the process is known as peptization- where a drop of acid is mixed in the system. The steric stabilization is suitable for any type of medium-polar or nonpolar. Here , suitable surfactant layer is coated on each nano particle to counteract van der waal and other attractive forces. For example , in case of hydrocarbon based ferrofluids we use an appropriate amount of oleic acid to coat the nanomagnetic particles. For ionic ferrofluids we use Cgoulomb repulsion provided by charging each particle with an acid.

You do a size selective precipitation of the particles. This is the way by which we separate nanoparticles of different size and shape. In this process, due to separate weight of small particles (e.g. your monodispersed particles) and bigger particles (your aggregate) , when you do the centrifugation at low RPM, only the bigger clusters/particles will be precipitated at the bottom. This will leave the small monodisperse particles in the solution. Please note: you have add the precipitating solvent carefully dropwise. Donot add more. Just to fluctuate.

Another method is you can use size exclusion chromatography. Although this method is considered expensive.

Filtration might work depending on aggregate size (there are some very small pore-size filters available). Otherwise go for centrifugation as Sasanka Deka proposed.

By membrane filtration, we can separate the agglomerates from monodisperse nanoparticles. Selection of membranes are based on size of the nanoparticles

I agree with all the answers, but was not mentioned in the question what kind of raw materials resulted the nanoparticle.

In my case, I obtained nanoparticles lignocellulosic materials by controlled hydrolysis with sulfuric acid, because this type of acid, due to polarized groups which present after the hydrolysis is very helpful the separation of the nanoparticles.

i agree with mr. Ignacio Tudela. After hydrolysis and dialysis, sonication performed with various energies and different times.

All this helped me in the separation of the particles. The samples were analyzed by TEM.

Sonication can only provide so much shear force- in many cases the van der Waals forces in an aggregate of very small nanoparticles (tens of nm) will be much stronger than any microturbulence you generate using a sonicator. That being said, you can still try using a strong probe sonicator at high power to try to break up some of these agglomerates, followed by centrifugation to separate out the aggregates. As others have mentioned, perhaps capping agents could be added to increase the zeta potential or provide steric hindrance in order to prevent future agglomeration.

You can use size exclusion chromatography.

You have not provided sufficient information to answer this question. Centrifugation technniques will depend on the density and size of the particles and agglomerates. What is the material? What size range of primary and agglomerated particles are you trying to separate? How broad is the distribution? We have use field flow fractionation and ultracentrifugation to separate small discrete agglomerates (i.e., dimers, trimers, tetramers, etc...). But separation efficiency depends on the size, density and breadth of the distribution. You may also need to consider a serial process in which you remove the largest materials in a simple first step (filtraiton or centrifugation or even gravity settling for a period of time), followed by more sophisticated approaches to get down to the primary particle.

Sir, vincent i provide material now i.e. Chitosan nanoparticles of ciprofloxacin hcl and size range is 300nm agglomerates range is 600 -700nm . Thank you for answer

Simple way is to centrifuge at high ~12000rpm for few minutes.

Centrifugation is not a good idea to separate less mechanical strengthened materials. Because at higher rotation speed, the "g" force gets higher, the nano particle to particle interaction gets higher. thereby the Nanoparticles loses its strength. Depends on having the type of capping agent (if any) presented on the Nanoparticles which sustains at higher centrifugation force. If you are particularly wants to separate the Nanoparticles on it s size selective. I would suggest to make a phase separation mode i.e. by variation of liquid density. Do to so, please follow Mrs. Sasanka Deka proposed step up to the extract (of smaller size particles), then follow the phase separation mode. for further reference please see the attached article. Good luck

Note: you need to choose a appropriate liquid to avoid any structural damages to the Nanoparticles.

I match my opinion with dr. Sasanka Deka, i tried to measure the supernatant after centrifugation and the ppt after re dispersed in water.

the supernatant show particle size of 320 nm while ppt show 2000nm 

but how i can be sure about the % yield in supernatant (how i can calculate it) and how i can measure the drug content after that? 

i mean when i take the supernatant, how i can deal with it?

Is the Polydispersity Index value > 1?

The polydispersity index (PDI), together with the z-average diameter, is calculated

from the cumulants analysis as defined in ISO22412:2008 and is a dimensionless

estimate of the width of the distribution, scaled from 0 to 1.

A PDI value of 1 indicates that the sample has a very broad size distribution and

may contain large particles or aggregates that could be slowly sedimenting. If

this is the case, the sample may not be suitable for a DLS measurement.

Correlation function obtained from a sample consisting of a broad size

distribution containing large, sedimenting particles (note the noisy baseline).

This sample gave a PDI value of 1 and is a good example of a sample not suitable

for the DLS technique.

Often, when a sample gives a PDI value of 1.0, cumulants and distribution fit errors are also obtained.

ultracentrifuge at higher than 15000 rpm (4 C) for small NPs (below 100 nm) or dialysis for nanopolymer base structure.