FTIR is any Fourier Transform Infrared spectroscopy method in any measuring geometry, may it be transmission, reflection or whatever.
ATR stands for attenuated total reflection and is developed in order to enhance the surface sensitivity since IR spectroscopy is a bulk method. So if you have an ATR unit available, use it.
If your substrate is transparent you can try a transmission setup, but I wouldn't hope for a lot of useful results from that.
ATR is one of many sampling methods for FTIR, which is the analytical technique itself. ATR allows you to acquire spectra from the very surface of the sample, by contact with a (more commonly) Germanium or Diamond window. It may be "single bounce" (one reflection) or "multi-bounce" (multiple reflections)
I concur with Herr Weipert : better try your luck using ATR sampling in your case.
Remember however that FTIR is not that sensitive compared to surface techniques such as XPS or SIMS.
Spectroscopically both gives almost same spectra only the principles are different. A very little variations you will have in intensity and peak positions.
Donya Moradi IR spectroscopy can be very surface sensitve - in many instances it is monolayer sensitive, in particular in reflectance, but also the transmittance mode can be monolayer sensitive. Even if not, there are methods to enhance the signal e.g. by interference or plasmonically. This certainly depends also on kinds of functional groups... can you provide us some more information what kind of changes you expect compared to the bulk?
During the polymer treatment process, new functional groups are created at the surface . I want to find out what new groups have been formed by comparing the treated and non-treated surfaces.
it all depends if the spectral contrast of these new functional groups is strong enough to distinguish it from the bulk. If the functional group is different enough, so that it does not spectrally overlap, your chances to detect it with ATR should be good. For that goal it might be of advantage to use Ge as ATR crystal and a higher angle of incidence, e.g. 60°, because this decreases the penetration depth, so that you get less signal from the bulk.
The choice of the ATR crystals like diamond or Ge or Si should also be due to optical parameters of your sample. In the case of organic compounds, it's usually not a problem most often they have a refraction index on the level of 1,5 so most crystals will do. But in case of high n values of the material (like for materials doped with metallic ions), Ge is better, and it has a very high refractive index itself (it should always be higher than this of your material to register the spectrum). Also, ATR comparing to transmission technique has an artefact connected to dispersion phenomena. In practice, the signal in low wavenumbers will be stronger than in high wavenumbers. So sometimes low intensity, high-frequency bands, may be hard to observe by ATR technique. But low frequency is often 'noisy' because of water vapour bands and low source energy in this region). Usually, in the transmission, you will have a lot more signal than in ATR in the whole range. Also, the analytical range is often wider for transmission technique. If I had such project, I would take the untreated polymer, measure it probably, on a diamond ATR, then using the same conditions I would measure the treated sample, and after I would try the spectral math and subtract the spectra. It's maybe the quickest way if your spectra would be of good quality you will see bands even of very low intensity. If it were not enough, I would try transmission. Best would be optically diluted sample using KBr or some other optical dilutant. Of course, if it's possible to prepare a powdered sample - would be hard in case of foil. But if your foil is of very low thickness, you may try transmission technique without dilution. Some polymers are almost transparent for IR frequencies (like PE). Wish You luck whit your experiment!
You can do transmission technique with FTIR where would not get multiple scattering caused by fine particles (in case if you have sub-micron particles), resulting in clearer diagnostic absorption features for the endmembers.