Francesco, try TMA (Thermal Mechanical Analyzer) furnished with a flat tip probe. Apply minimal reasonable load (say 5-10 mN) to the probe at room temp. Good TMA models (e.g. Perkin Elmer) easily provide 10 nm resolution in sample thickness measurements. Else, by running a simple temperature program you'll be able to measure the thermal expansion coefficients for your crystals (or phase transitions if any). Good luck!
With which precision do you need the thickness? Do you need in one point ore in many points (two surfaces are never parallel and the must not be flat at all). I work in Crystal optics for X-rays, so I know a bit about that.
I am a physicist and do not know K alum. does this material has a normal long name or a chemical formula?
We use an instrument to measure the thickness of silicon plates (up to 20 mm thickness) with a precision below 1 micrometer. Measuring time few seconds. It commercially available . It is laser based and senses the interfaces with a kind of michelson interferometer (Low Coherence Interferometric Sensor (LISE) from FOGALE nanotech (France) ).
But we also use an optical microscop with a good software and look at a side surface.
The crystals you describe seem large enough to be mounted on a spindle axis device on a conventional optical microscope and measured by using a graded eye-piece (which you have checked by a certified standard) to read off the thickness from an edge view directly.
Provided that you have a first measurement with a precision of the order of 1 µm and that the sample is more or less transparent to a given lambda ray, you could use interferometry to improve the accuracy.
If you want a high accuracy measurement , you can try to study the coefficient of transmission of some LASER light through your sample while tilting it by successive angle steps. By considering the interferences between the light reflected by both faces you should, at least in theory, come back to the refraction index and the depth of the sample.
An alternative way, is to mount the sample in a transmission spectrophotometer and to study the variation of the transmission whith the wavelength in a region of the spectrum where there is no strong absorption.
In my answer above I omitted to mention the obvious point, well addressed by Vincent Bourdin, regarding the precision (and accuracy, may I add) required. Given the dimensions of the crystals I assumed the measurements obtained by direct observation in an optical microscope (the most simple and readilyy available method), would be sufficient. I hope Francesco can comment on this and check the feasibility of my suggestion.
Jürgen: I would need a precision (not so important the accuracy) in the micrometer range. I am not very interested in the accuracy (provided that a hypothetical systematic error remains constant) of the measurement because I would need to record growth or no growth of the (111) face of these crystals in an experiment. The formula
Sometimes you can try microscopy- putting a crystal on the glass slide with the thin line (with a marker) at the top of a crystal and make a thin line on the glass substrate. Then with the use of micrometer screw of the microscope you can measure the thickness of the object. Contact for an advice at your place with the microscopists. If anybody has SEM around you -it can help as well.