Regarding freeware tools, one of my colleagues uses a program called Fytik, but I do not have any personal experience. Also some verison of Origin, which is a more general SW and might be available at your institution, enable peak deconvolution with various peak shapes, bandwidths, etc.
One can try Igor Pro software for deconvolution. It might be more flexible in terms of changing(otherwise locking) multiple parameters to get the best fit.
Gnuplot is one of the best and you can change all of the parameters. In this paper we used it and it works perfectly. In some part, Raman spectra had been fitted with 11 peaks.