Deviations of results: IMD5.02 vs IMD4.1.X

[Dmitry Kuznetsov]

Hello, David.

Trying to estimate the influence of carbon contamination of boron on 6.X reflectivity, we noticed that
the results by IMD5.02 are not physical: just 2% of B   (composition B98C02)  dropped reflectance by 6%.
Better example: for ideal La/B4C MLS of 220 periods one will end up with ~15% of R instead of ~75-78%.
Below is the file attached:

https://drive.google.com/file/d/0B6T92xyewYcpV1lRWXpKWVEzWk0/edit?usp=sharing

To be sure I re-done the file from scratch, and also Robbert tried to simulate that. Same results we have.

IMD 4.1.1 gives completely different (and closer to physically known) results for the very same structure
(period, gamma, all the layers set as density and at. scattering factors, AOI,...)

I checked scattering factors of different IMDs (by subtracting of columns in tables) and they completely the same.



P.S.
My colleague mentioned some problems for IMD 4.1 with calculating R for MLS if layers are mixed compositions,
set by f1f2 and density. I wll try ti figure the details - maybe, old bugs are "saved" in new IMD.


Good luck,
Dmitry.

[David Windt]

Hi Dmitry,

Please try manually reloading optical constants after you have changed the composition parameters, using the "Materials/Optical Constants -> Reload optical constants" menu item. I was able to replicate the problem you described and could resolve it this way. I haven't delved into the code just yet (I will soon), but I'm fairly confident that this problem is related to that one known bug I still have to correct, in which optical constants are not always recomputed when they should be. (This bug appeared as a side-effect after fixing a previous problem that Robbert had identified, wherein the optical constants were being re-computed too frequently during fitting. So I blame Robbert! ;-) Just kidding of course.) Anyway, please let me know if this workaround solves the problem or not.

David

[Dmitry Kuznetsov]

Yes, David, re-loading optical constants solves the problem.
I will not blame Robbert, but only because he has higher position in the lab ))

Good luck!
Dmitry