Please change the mounted of the samples or deposit.

XRR requires very flat and smooth samples. Usually only float-glass and silicon wafers are flat and smooth enough for this type of analysis. You can try to use a very narrow parallel X-ray beam when your sample is not flat enough.

In principle a curved sample surface should behave like a divergent x-ray beam. If you take this into account in the modeling of youre data and the curvature is not to large you could still derive some information. around the critical angle, the steepness of the XRR signal drop will be lower due to the curvature and this might be used as a criterion for the modeling.

Good question! I need to know the answer as well. I'm beginning to learn a lot about XRR recently.  I hope to check the effect of curvature on the real time XRR signal as well sometime soon. I'll post details once I do so. Can't wait!

Meanwhile here are some examples of 2D XRR signatures from "flat" surfaces for two different "thin films" using a line source on both the Bruker and Panalytical diffractometers:

See the attached RG posts for more details and do share your opinions and experiences when convenient.

One of the challenges with the XRR method for this situation is the fact that the incident beam irradiates a humongous cross-section of the "flat sample" due to the shallow incident angle. In fact, this is the basis for being able to come up with a "flatness" rms value for the entire sample surface using XRR. As suggested by others it would make sense to include the curvature in the modeling parameters if possible. However, the curvature must be estimated then by another precise method to accomplish this. What method could that be?

If the curvature is "due to the process used to deposit the coatings", then it may be prudent to try the  coatings on a relatively thick & "flat" substrate and then try to compare XRR signals with the "curved". How "flat" or "smooth" do you estimate the plastic substrate to be prior to processing in your case?

What are the expected dimensions of film and substrate typically in your case? If the substrate & film are amenable to transmission XRD methods, you may be able to compute (estimate) diffracting volume of the film and hence its thickness by carefully monitoring the diffracted signal from the film in transmission and reflection modes. A "point source" may work better for this purpose. The additional complication introduced by any "preferred orientation" must also be included and considered for any such analyses, I'd think.

https://www.researchgate.net/post/Would_anyone_have_a_clue_as_to_the_origin_of_unexpected_spots_in_the_real_time_2D_XRR_signal_from_thin_film_samples_on_substrates

https://www.researchgate.net/post/What_is_the_best_way_to_interpret_REAL_TIME_3D_X_Y_o-2TH_o_ph_ch_reciprocal_space_relative_intensity_data_from_HRXRD_rocking_curve_analyses