Well if you have measured the true transmission coefficient, you can use the Beer-Lambert law to obtain the copper thickness. For this, you must know the absorption coefficient of your glass substrate and that of copper at the measured wavelength. One difficulty, perhaps, is that part of the incident beam is reflected, which must be accounted when using the B-L formula, since the photon flux passing through the sample is only what has not been reflected.

One other thing that you can do is to protect a part of the copper layer with a scotch tape, and dissove the rest of the layer in a strong acid solution, such as HNO3 in water (the glass substrate will stay intact, but check before with a piece of glass), and, once done, use a "Talystep" (the ancestor of AFM) if you have one in your lab -  to measure the thickness at the edge of the copper film/glass substrate.  You will get at the same time its roughness. Good luck ! Ph.

As suggested by Dr. Parent, Beer- Lampert law can be used if you correct your transmission for reflection. Absorption coefficient of the film should also be known. Here will be a problem as at low thicknesses of the film, absorption coefficient and other optical parameters may depend on thickness. If you are in thickness range where optical constants are not thickness dependent then of course you can use this method. However, the accuracy of thickness calculation will depend on accuracy of many measured parameters for example transmittance, reflectance, absorption coefficient of glass and film. Hence, interference fringe method is generally preferred over Beer - Lambert law. Also this law is applicable only for homogeneous system.

As mentioned by Dr. Parent, one good solution is etching part of the film and having the other part intact to do AFM. Another thing you might be able to do and I have done in the past is to use an ellipsometer. With that you can measure film thickness on the glass very easily and quite precisely. Ellipsometer is in fact what is used for Display industries to measure the thickness polymer and/or Indium Tin Oxide on glass substrates.

What film thickness and Nanostructural composition?

XRR, X-ray Reflectivity, is a century aged, precise and unequivocal NDE technique to take advantage of :-)

https://www.researchgate.net/post/What_is_the_best_way_to_interpret_3D_reciprocal_space_data_from_XRD_rocking_curve_analysis2?_tpcectx=profile_questions

Though you can use different  technique like cross sectional SEM imaging or as per your interest you can use reflectivity technique like X ray (As said by Ravi Ananth) or Neutron reflectivity?

In addition to the excellent suggestions above it is also possible to model the thickness of your coating from T/R data, if you know the optical constants, using software such as SCOUT http://www.mtheiss.com/?c=1&content=products; much like you would have to using Ellipsometry. If you don't know the optical constants the database included with the software package contains optical data for many substrates and coatings which can be used as starting points. Personally I would prefer to measure the thickness in someway, e.g. profilometry or cross-sectional SEM or TEM, but if the apparatus is not available then modelling could be an option, once the model is set up however thicknes calculations will only take as long as measuring the T/R spectra.

Here is what the observed data and the modeled data look like in the case of XRR (X-ray Reflectivity). What would this comparison look like when you "model the thickness of your coating from T/R data"? Could anyone post some literature link or examples for illustration please? Thanks!

BTW what is the resolution & sensitivity of the "T/R data"? If any of you could discus this aspect for other thickness measuring methods as well, it would be highly instructional.

L. Marot! Does the thickness measurement need to be non destructive? If not TEM, SEM and AFM would work well depending on the magnitude of the thickness/thinness(?).

I'd like to understand (I hope others would too) the relative merits and practicality of each of the earlier mentioned techniques for film thickness measurements. Intuitively, I'd think that the interogating wavelength would be a limiting factor in both sensitivity and resolution of the thickness measurements for each method. What say you?

Do you have any estimation on thickness? It is very important. If you guess the thickness higher that 100 nm, you can not use XRR because you wont see any and only you can estimate density and roughness. 

With that data that you have, you can use Pointwise Unconstrained Minimization.