I can see several options. The most obvious would be to perform a classical UV-Vis spectroscopy of polycrsystalline sample where several small crystals will be embedded in gel. If the gel can be solidified or polymerized, you'd get a transmission through a composite thin film. If it remains liquid, I'd try to fill 2-mm thick spectroscopic cuvette with a liquid gel and see (or hope) if snowflakes can be arranged flat along the illumination wall forming a quasi-monolayer. Then, you may get some order in the flat arrangement but still have to measure a response from several crystals.

You can also try to apply microscopy with spectroscopic detection. Boy, that can be real fun! You can select/illuminate an individual crystal, perhaps scan along it while sending the transmitted signal to a spectrometer. Then, you can also control the polarization of the exciting beam (now we talk about lasers) and check polarization properties vs wavelength. Ok, where do I stop?...

If you're up to even more fun and have resources for it, get near-field spectroscopic measurement with NSOM (or SNOM) or just do confocal microscopy.

To satisfy your non-linear appetite, add SHG, TPL to microscopy measurements. What have I missed?..Yes, just plane and simple luminescence (again, in confocal mode).

Hopefully, at least something from above can be useful:)

Good luck!

Serge       

I am very appreciated to your advice, Serge! All information was useful. We are intended to study optical properties of rare-earth cations as dopants in calcium oxalate crystalline matrice. The first stage of this work was to demonstrate that incorporation of dopants into grown in silica gel crystals is possible.  I think that microscopy is  prefferable but I afraid that such configuration of microscope is quite exotic. 

NSOM -yes, adding a fiber-optic collimator connected to a spectrometer - not really..It depends what microscope model do you have. For example, most Olympus models have a modular structure that allows takings some elements out and replacing them with custom-made components. For the simplest transmission mode with spectroscopic registration you would need to take the bottom (the 'inverted" ) part of the microscope out, check the optical path, remove unnecessary beamsplitters, filters etc. and think how would you mount a fiber-optic collimator (Thorlabs, Newport, other companies are making it too) that can bee hooked up to Ocean Optics USB 4000 or another model that you may already have.       

Once more opportunity is to compress the small crystals in a special press form. The compressed  plate should be transparent at least in the IR spectral range, where IR absorption could be observed. And who knows, may be it would be semi-transparent in the visible range. Is such press available near-by?

Dear collegues, thank you again. SNOM is an interesting method and i am searching possibilities. As to pellets, our amount of sample is too small to press them for ordinary spectrometer but I will analyse this possibility too.

Best wishes,

G.V.S.