To get rid grain during deposition you must work in isolated reactor, and for your problem you must dissolve this grains by using proper solvents depends on grains nature. 

there is a misunderstanding, the loosely grains are from the same film materials. it is not contamination to dissolve these grains the film it self will be vanished. 

Dear Ashraf Ali,

The question is not very clear. It is not very clear whether there are loosely bound grains of Ag on the film or there are some grains (such as dust particles or some other particles attached to the cantilever, i.e., some contamination). 

But as clarified later that they are loosely bound grains from the same material, the question arises how do we know that they are loosely bound because if the cantilever is in contact mode then it is possible that some of the grains, i.e., tall islands of Ag may break and stick to the cantilever. Therefore, the AFM scanning may be done in non-contact mode. Another way to confirm this is to take SEM images of the the film to confirm whether there are loosely bound grains or some islands grown at certain places. If so then there is no need to remove them but do something to grow the film soother. This study would confirm the roughness of the film also. 

All the best!

--- P D Sahare  

Dear Prof Sahare,

the AFM images captured in the tapping mood, no contact. I changed the cantilever many times and get the same result in the slow scanning mood the images have very large amount of lines. but in the fast scanning mood, these line density is less.

SEM images indicates that the surface is very rough, this is expected for silver. also some kind of blocks or flakes appear beside the regular grains with nearly the same sizes.

regards 

Dear Dr. Ashraf

The loosely grains mean that; your thin film is not homogeneous, and this depends on your method to prepare it, so that, you can change or improve this method to avoid this grains (time of treatment, moving substrate or homogeneous target.... etc).

Did you find this grains after a certain time of treatment?

But if this grains make your thin film more sharply roughness it will be useful in some applications like micro discharge.

Attached video shows methods of measuring sample surface, I hope this useful for you.

Good luck.

Barakat

https://www.youtube.com/watch?v=fivhcWYEtkQ#t=818.852787

Thanks Dr Barakat for nice suggestion and informative video. 

I think the problem lies in the adhesion between Ag and Si substrate which permit silver to agglomerate in a vertical way on Si this arrangement make hills and trough resulting in a rough surface. the problem now is there is some loosely small grains on the surface that stuck to the cantilever and and changes the measuring conditions. I made many samples and get the same problem. some colleagues advice to rinse the samples with dry nitrogen gas or clean the samples with water in ultrasound cleaner, I didn't try this yet, but I am not sure about this.

Dear Ashraf Ali El-Saftawy,

a (sputter-) deposition of a (10nm) Ti buffer layer on the Si substrate, if possible, might help you.

Surface-properties relationship in sputtered Ag thin films: Influence of the thickness and the annealing temperature in nitrogen; http://www.sciencedirect.com/science/article/pii/S0169433214023150

Thickness dependence on thermal stability of sputtered Ag nanolayer on Ti/Si(1 0 0); http://www.sciencedirect.com/science/article/pii/S0169433207008136#articles

Thanks prof Samaras,

Your suggestion is so appreciated and I think the annealing step suggested in the attached article will work good. I will try to employ it. regards

The simplest solution is to blow on the surface with a jet of clean, dry nitrogen or another inert gas. Sonicating the sample may work as well, but if the film is poorly adhered to the substrate then it may lift off.

If that doesn't work, there are some products for surface cleaning that you might try (linked below). I've had great AFM results after using First Contact from Photonic Cleaning Technologies to clean single crystal surfaces. It brushes onto the surface like nail polish, dries in a few minutes, and is then peeled off with tape. You can also try replicating tape (cellulose acetate), but it seems to leave an organic residue that can cause problems for tapping mode AFM.

http://www.photoniccleaning.com/

https://www.tedpella.com/replicat_html/44840.htm