Monte carlo simulation of BS coefficient

[Ben Buse]

Hi,

I've just run a quick comparison of the BS coefficient calculated using Win X-ray, casino and penepma for Ag and Fe metal (initially to check upbound fraction in penepma-res.dat was the bs coefficient-but then the results where interesting.)

Results as follows - best agreement Win X-ray/Penepma whereas casino considerable differ. My guess Win X-ray Penepma the best. (Although casino was much faster even than win x-ray at least through wine on linux.)



Ben

[Probeman]

I've just run a quick comparison of the BS coefficient calculated using Win X-ray, casino and penepma for Ag and Fe metal (initially to check upbound fraction in penepma-res.dat was the bs coefficient-but then the results where interesting.)

Results as follows - best agreement Win X-ray/Penepma whereas casino considerable differ. My guess Win X-ray Penepma the best. (Although casino was much faster even than win x-ray at least through wine on linux.)

Hi Ben,
Excellent!

It would be awesome to compare these calculations to actual measurements, but this is not easy as one has to apply a small bias voltage to an isolated sample to prevent additional signal from secondary electrons.

Attached below is a paper (you must be logged in to see attachments!) where I replicated Heinrich's BSE measurements in my SX51 at UC Berkeley.
john

[Probeman]

Here is the post with the Au BSE coefficients calculated for different densities attached:

http://probesoftware.com/smf/index.php?topic=126.msg508#msg508

The reason I mention the Au BSE calculations on different densities is because some time ago I found that putting in different densities in Casino caused it to produce significantly different BSE values. I pointed this out to Raynald Gauvin and he went "sacre bleu"!     

I assume that this was fixed however, but I haven't tested it. 

[Ben Buse]

Hi John,

I've just stumbled across a database of experimental data from literature which David Joy compiled for SE and BSE yields (whilst looking for something else - as is the way)

http://web.utk.edu/~srcutk/htm/interact.htm

For Ag values are 0.394 (1); 0.415(2), 0.423(3),0.409(35)

For Fe values are 0.278(1) and 0.298 (13)

numbers in brackets are the references in the report

Which are closest to PENEPMA and Win X-ray. I was using casino V 2.48

Ben

[hdemers]

Hi Ben,

For BSE simulation, you should use CASINO v3 and not version 2. Also you need to specify which elastic cross section models you use. In CASINO v3, from a limited testing, I found Mott-Browning model to closer match with the Joy database, but the ELSEPA model should be better. PENELOPE use ELSEPA, but I am not sure which settings they have use to generate the cross section, so it may explain the difference between CASINO ELSEPA and PENELOPE ELSEPA.

Regards,
Hendrix

[hdemers]

Here is the post with the Au BSE coefficients calculated for different densities attached:

http://probesoftware.com/smf/index.php?topic=126.msg508#msg508

The reason I mention the Au BSE calculations on different densities is because some time ago I found that putting in different densities in Casino caused it to produce significantly different BSE values. I pointed this out to Raynald Gauvin and he went "sacre bleu"!     

I assume that this was fixed however, but I haven't tested it.

Hi John,
It have to be an old version of casino. I just try it with casino v2 and v3 and not difference with different densities:

Casino v3 20 kV Au
rho = 1:      0.477801 (0.0015)
rho = 19.3: 0.478413 (0.0015)
Casino v2 20 kV Au
rho = 1:      0.449739 (0.0015)
rho = 19.3: 0.449722 (0.0015)

In estimate 3 sigma variation is in ().

Regards,
Hendrix

[Probeman]

Here is the post with the Au BSE coefficients calculated for different densities attached:

http://probesoftware.com/smf/index.php?topic=126.msg508#msg508

The reason I mention the Au BSE calculations on different densities is because some time ago I found that putting in different densities in Casino caused it to produce significantly different BSE values. I pointed this out to Raynald Gauvin and he went "sacre bleu"!     

I assume that this was fixed however, but I haven't tested it.

Hi John,
It have to be an old version of casino. I just try it with casino v2 and v3 and not difference with different densities:

Casino v3 20 kV Au
rho = 1:      0.477801 (0.0015)
rho = 19.3: 0.478413 (0.0015)
Casino v2 20 kV Au
rho = 1:      0.449739 (0.0015)
rho = 19.3: 0.449722 (0.0015)

In estimate 3 sigma variation is in ().

Regards,
Hendrix

Hi Hendrix,
Thanks for testing.  Yes, I assumed it was fixed many years ago when I showed it to Raynauld.  I think it was v. 1.x of Casino actually now that I think of it.

Many people seem to think intuitively that density should affect backscatter in bulk samples, but the physics says no!
john

[Ben Buse]

Hi Ben,

For BSE simulation, you should use CASINO v3 and not version 2. Also you need to specify which elastic cross section models you use. In CASINO v3, from a limited testing, I found Mott-Browning model to closer match with the Joy database, but the ELSEPA model should be better. PENELOPE use ELSEPA, but I am not sure which settings they have use to generate the cross section, so it may explain the difference between CASINO ELSEPA and PENELOPE ELSEPA.

Regards,
Hendrix

Thanks that's very helpful

Ben