I agree with Glonaz, the smectite is an expandable clay and the XRD peaks will shift while the illite peaks will not.  In addition, illites usually incorporate 5-10% NH4 in the interlayer sites instead of K+.  So nitrogen content is a useful elemental parameter.

I think, Vincent asks compositional diagrams, rather than XRD methods. Smectite and illite can be evlauated a lot of compositional diagrams, but distinct differantiation can be possible on the basis of the decreasing Si(IV) or increasing tetrahedral Al(IV) (e.g., tetrahedral charge) versus increasing interlayer cations (K, Na) and octahedral charge  from smecttie to illite (e.g., Muscovite-Celadonit-Pyrohyllite ternary diagram (or tetrahedral charge vs. octahedral charge diagram), Si vs Na+K diagram, M+ - 4Si R2+ diagram). These differences should be affect their major and trace element composition. Chondrite or NASC normalized REE patterns also differentiate smectite to illite, by means of higher REE content of illites than smectites. Oxygen and hydrogen isotope compositions were also used for distinct differantiation, for example from smectite to illite, oxygen isotope values (deltha 18 O) decrease, but hydrogen (deltha D) values increase. Accordi,ng to these chemical data, you can also investigate relative changes with compaction as well as distinc differantiation.

Many thanks everyone for your answers. And thank you Omer, your answer is exactly what I was hoping for. I will try the compositional diagrams you suggest. Again, my problem is that I cannot use XRD because I am studying 50 - 100 µm thick deformation structures in clay sediments. They are way too small for sampling them and using XRD... Unless there is a way of making in situ XRD analyzes with a 10 µm spatial resolution ?

in agreement with Golnaz purpose, a simple method consisting in executing MEB analysis is recommended.