I think this  review will help you in specimen preparation for TEM 

see attached file 

Dear Uday, One has to use thinned  specimen as a test modulus  for TEM, which is the case for any set-up relying on the transmitted beam. As far as I remember the specimen preparation for TEM was an art and science to get really good results in early fifties and sixties. TEM I don't believe is a good choice for study grain microstructure or grain size as long as you don't deal with the thin films. SEM or AFM would be the best  techniques for that purpose.

I  have preferred to use AFM in combination with computer simulations to study of surface topographies,  especially the grain boundary triple junctions morphology to get some quantitative information related to the surface diffusion and surface and GB specific free energies.

TEM is very powerful tool to study dislocation domain structure, stacking fault width, small clusters, and the density and structure of  dislocations etc. 

If the thickness of your sample is larger than 100 nm you absolutely need to thin it for TEM. If it is 40 micron or less you may do it by ion-milling; with this technique you create holes at whose borders you can look at. Other possibiliities are electro- or chemical-polishing.

I think this  review will help you in specimen preparation for TEM 

see attached file 

Dear Uday,

usually for TEM you need thin foils, that also can be prepared from bulk alloys. These foils need FESEM surface quality at both sides!!! Thus, sample TEM sample prepartion is challenging. Our group focusses on metal alloys and the most characterization is done by SEM or FESEM - like determination of composition, microstructure or grain size.

However, if you have a very fine scaled microstructure (nanocrystalline or nano precipitates), you will need TEM to visualize and to characterize in detail. Also for determination of the dislocation density TEM is recommended. This has been discussed recently, see the link below:

https://www.researchgate.net/post/How_to_use_the_transmission_electron_microscopy_to_measure_the_density_of_dislocation_in_metal_material

In our quantitative studies on  the  dislocations and  stacking faults densities, where  XRD and NMR  techniques were employed, the   metallic ( Cu and Alpha Brass) bulk samples were cold-worked by filings and  deformed by  rolling, respectively. TEM also employed in connection with the determination of the stacking faults width, which was later used to get the dislocation densities from the measured stacking faults probabilities directly from XRD. measurements.  The broadening of the NMR line and its  satellites are very sensitive for the nuclear quadruple interactions associated with the electric field gradient generated by dislocations even the point defects. In heavily deformed metals dislocations are no longer distributed uniformly in the matrix but rather they form cell or domain structures. The dislocations are mostly populated at and in the vicinity of the cell walls. High stacking fault energy metal such as Aluminum  the cell walls are very thin and highly dense compared to low stacking faults energy metal such as copper, where dislocations are split into partials forming intrinsic stacking faults, the cell walls are not defined sharply but rather they look like fiber bundles  as can be easily observed in TEM. In the domain interior one observes very few  dislocations. Therefore in the statistical analysis of the heavily deformed specimens  the domain size and cell wall thickness  enter into the final calculations. 

The classical way to prepare TEM specimens if you do not have FIB consists of the following steps:

1 ) you start from the bulk samples (plates, for instance), which have to be mechanically thinned down to 0.1 mm

2) then, from these thin plates, you punch out disks having 3 mm in diameter

3) the disks have to be electro-polished in a one-step procedure, using for example a twinjet Tenupol under a set of conditions that depende on the material of the samples (for example, for some aluminum alloys it works an electrolyte of 25 vol.% nitric acid in methanol, at around 12 V and -22ºC)

Then you analyze the disks with a TEM with the required performance to achieve the magnification you need to make your observations.

Dear Uday,

if you want to determine dislocation density I recommend you to use XRD technique. TEM microscope is totaly improper scientific tool. The visibility of dislocation depends on diffraction settings. You can make dislocation invisible under well-known condition g . b = 0 and g . b x u = 0     for screw and edge dislocation respectively,

where g - reciprocal lattice vector, b - burgers vector, u - unit vector along dislocation line.

This means you can set up such diffraction condition that you will be able to see none dislocations or as such dislocation density as you wish. Beside this, the dislocations may be introducted to structure during sample preparation processes and manipulation.

If you need information about the grain size and type of microstructure, probably you don´t even need TEM. If the grain size is of hundrets of nanometers and structure features are not under (let´s say) 50 nm, then SEM or Light Microscopy is better choice considering the complicated sample preparation for TEM microscopy.