Origin isn't that good for spectral data analysis. It can do general deconvolution, but that quickly gets unwieldy after 4+ peaks. The regular peak selection, integration and other functions aren't there. I would also suggest software specialized to spectrum analysis.
There is indeed a point with Origin not being so user-friendly and reliable for XPS peak-fitting as compared to other (dedicated) software. However, if you are OK with writing scripts in C/C++/C# environment (not sure which one is used in Origin), you can construct your own fitting procedure for all the types of PE spectra you have, and it would work just fine, and is also convenient, since you will have all your data in one place - raw data, background/slope/energy corrected, and the fitted (in fact, I'd rather avoid using the term 'deconvolution' in respect to the processing of PE spectra). Another fantastic software for PE data processing (and spectra in general, and even much more suitable for processing images from 2D detectors) is Igor Pro (or some of the recent versions, not sure about the numeration). This is even more powerful than Origin, but requires, to my mind, more effort to understand the logic and start working with. Programmable as well. I saw some really AMAZING data treatment plugins written by non-professional programmers in Igor. However, if you aren't too much into programming, a dedicated peak-fitting software might work the best for you. Apart of those already suggested, I' d like to promote FitXPS from Aarhus University (distributed by Lunds University as well):
http://www.sljus.lu.se/download.html
This is very simple yet robust software. Free as well. At MAX-Lab synchrotron facility in Sweden we used to stick to this one.
Finally, you might want to buy/download CasaXPS - I haven't had an experience with this one, but I heard that it is a good one as well, much more powerful than FitXPS, for example, but is slightly more difficult to learn how to use, and not free.
One point to add. For whichever atom peak you wish to preform the deconvolution, you must have the peak determined in the same instrument for the clean pure material to use for the appropriate line shape in the deconvolution. For example: for an organic compound having bound carbon as compared to surface contamination the appropriate line shape will be for each type of carbon the same as for pure carbon and only the binding energy will be different. Thus it is very dangerous to use a generalized peak fitting program that may well fir the overall profile but most likely will use different line shape within the fit.
Thank you all for your kind suggestion. I have down loaded the XPSPEAK 4 and FitXPS software I will try to do it by software, instead Origin. If i need some clarification, let you know kindly give us your inputs.