i am getting crystallite size in the range 4 nm-20 nm corresponding to different XRD peaks of the sample using Scherrer's formula. What to say about average crystallite size?
As already mentioned above by Gerhard Martens, microstrain might contribute to the peak broadening that you observe. Williamson-Hall method is the simplest technique to separate grain size and microstrain effects. If this doesn't really work, you might need even more sophisticated methods (modified Williamson-Hall; Warren-Averbach; WPPM; etc.).
Besides, I hope that you corrected for the instrumental broadening before calculating the grain size (or better domain size). Any other contributions than grain size to peak broadening (microstrain, instrumental effects, etc.) will result in a too small calculated grain size if you apply the Scherrer equation without separating these other effects first.
As already mentioned above by Gerhard Martens, microstrain might contribute to the peak broadening that you observe. Williamson-Hall method is the simplest technique to separate grain size and microstrain effects. If this doesn't really work, you might need even more sophisticated methods (modified Williamson-Hall; Warren-Averbach; WPPM; etc.).
Besides, I hope that you corrected for the instrumental broadening before calculating the grain size (or better domain size). Any other contributions than grain size to peak broadening (microstrain, instrumental effects, etc.) will result in a too small calculated grain size if you apply the Scherrer equation without separating these other effects first.
Why u still using SF. It has no direct physical interpretation unless u introduce K that depends on the how the line profile width is determined, the shape of the crystallites, and the size distribution - which is usually unknown. All the FWHM base calculation dependent on K. An independent approximation intruduced by Laue where K could be assumed as 1 when use integral breadth. But as the profile itself usually contains Gaussian and Lorentzian function than a double voight approach has to be use.
The calculated value itself cud be area weighted or volume weighted, depend on methods of analysis. And things get complicated when u compare with TEM analysis, where the number weighted is the result.
I have seen calculations here on RG which use degree values for the 2theta FWHM value when calculating the crystallite size via Scherrer or Williamsen-Hall. One has to use FWHM in radians ( please see the link). Degree value as input in these equations will give a factor of about 60 (=180/pi) too less crystallite size !!!!. That's why you might have calculated too little crystallite sizes ( see the comments above of Alan and Gert) . However do not forget to subtract instrumental broadening.