As already said by others, you can use ImageJ. But my experience with Image J and SEM analysis of particles shows that it works fine only if you have suitable images - with a good contrast between particles and background and without too much overlapping particles. If your images are not so good, then the only possibility I think is manual evaluation by measuring the size of at least few hundred individual particles and building the distribution for example in Origin. Of course, this manual method is much more time consuming.