Normally the flotation process is what is used to separate the graphite from the clay, if this is not happening you may need to adjust your reagents. Alternatively, if the clay is finely disseminated, classification will work after flotation.

Do you have the quality specifications - physical (e.g., sizing) and chemical (impurities), etc.?

You can try one step of de-slime in scrubber drum before the flotation. Or a tank to mix the reagents before dosages in cells. Try contacts with Clarient or Cytec reagents suppliers.

What do you mean by coated graphite? Do you mean clay coated graphite or graphite coatings on other minerals? If it is the former, as Wesley has suggested, very gentle attrition in the presence of a dispersant (e.g. sodium silicate, etc.). But then, how did it float - unless it was only a partial coating?

Basically my graphite is coated on clay. My question is how I remove my graphite so that ash% become low. Feed ash 76%. Flotation product ash % is 33 %. Now I want to reduce up to 10 %

I would think that you are not too worried about maximising the size of the graphite particles so gentle attrition is the way forward. You need to have a way to physically separate the clay from the graphite before attempting the separation, which would probably consist of classification (depending how fine the graphite has become) then flotation in the presence of dispersant at low percent solids.

It's a matter of how you design grinding and classification process before the flotation. If there is overgrinding problem in the process you will have graphite slime coating on clay particles and this makes those clay particles behave like graphite particles and will report to the floated products.

I suggest:

1. Go back to correct your grinding and classification step.(better be closed circuit).

2. Apply a proper dispersant and do the flotation at low % solids as Andrew suggested.

One recommended dispersant for not-wet particles is aerosol OT. You may ask Cytec how to get this kind of dispersant.