A particle can be made up of a number of fundamental crystallites that cannot be separated by 'normal' means such as sonication. The grain boundary is simply the boundary between grains - it's an area of weakness through which atoms potentially can diffuse.  You do not expect correspondence between XRD crystallite size and TEM particle size.  The techniques are not measuring the same thing in the same why that apples and pears are fruit but are not identical.  Also the XRD result is a single number whereas the TEM result could provide a distribution.  Note that with sectioning in TEM that every image is an artifact - consider a bunch of monodisperse spheres - the simple act of microtoming will produce a distribution of apparent discs where only the largest observed dimension reflects the true spherical diameter.   Clusters of particles can form aggregates (tightly bound) or agglomerates (loosely bound).  Another technique such as DLS will show the bulk size of the material in the form of its hydrodynamic diameter.  Is there no one in your school to answer this and you come onto ResearchGate?

Dear  Sir thanks for the explanation. Still you have not clarified about the grain size. And can you please give some references or can you suggest some papers, because so many people have answered this question in different ways in  ResearchGate(if you search you will get to know) and I'm a bit confused because of that.

Thanks