I would recommend looking at the hand specimen of your rock, then perhaps making a thin section and only then start doing chemical analysis. The best free software to use is your brain ;-)

On a more helpful note: are you dealing with igneous, metamorphic or sedimentary rocks, or is that something that you do not know either? A weathered granite and an feldspathic arkose could give rather similar XRF results, but would deserve different names...

If you want examples of the chemical composition of igneous rocks, you can try http://georoc.mpch-mainz.gwdg.de/georoc/

If you need to predict the type of rock from its characteristics and obtain a confidence interval of your prediction, you can use a free software for data analysis or data mining "R" with good documentation available

Dear Pavlos,

There are some softwares regarding your needs. Examples include GCDkit, Minpet and Igpet.

But first of all you should use your geological knowledge as Marlina said.

Good Luck,

Masoud

It really depends on the type or rock you are wishing to identify; igneous, metamorphic or sedimentary. Sedimentary rocks tend to be classified on component composition rather than chemistry. There are separate classification systems for siliciclastic versus carbonate rock types. Both would require thin section (t/s) analysis in the first instance.

With metamorphic rocks it is typically the mineral assemblage and texture that is key to nomenclature - so again t/s analysis is the best starting approach.

With igneous rock one can either use mineralogy (t/s again), or use XRF (or similar analyses) using a Total alkali-silica (TAS) diagram - which is a purely chemical classification system. There is a useful introduction to this simple analytical approach on Wikipedia.

https://en.wikipedia.org/wiki/TAS_classification

However, whilst this works well for unaltered rock, as Marlina has pointed out, alteration effects can lead to misclassification because the alkali, and alkali-earth elements are easily leached during weathering of igneous rocks, thus could lead to misclassification. If alteration is suspected there are two further routes to try. First, assess the degree of alteration using a standard index like chemical alteration index (CIA); Second, if the rocks are revealed as altered by the CIA index, use an alternative classification method which employs immobile elements. A series of useful plots were devised by Winchester and Floyd (1977) Chemical geology 20, 325-343.

There are several free and proprietary plotting packages, such as IgPet by Rockware, that will take your XRF data and input it into these classification diagrams.

Good background in petrographic and thin sections solve many problems and help in understanding the best way to get many details about the chemical composition. 

Not aware of any "free" software that does this well - get what you pay for. Complex problem and solution, but certainly solvable. I have written code to do this commercially, but not for free.

I think nowadays the SEM solve many problems that can solve the mineralogy and petroglyphy composition of the hard rocks in general and also in detialed and the researchers can get a better understanding of the whole rocks.