Scoring the spectrum of deposited energy in a thin silicon sensor, radiative corrections and high energy simulations

[Jaap1957]

Hi,

I want to simulate energy deposition in a thin (15mu) SI slab from a 0.3MeV electron beam.
I found already many answers in the EGS manual and in egsinp files like
EGS_HOME\sprrznrc\sprrznrc_template.egsinp

A few remaining questions:

1 default setting for Electron Impact Ionization is Off.
would it be a good idea to put one of the "on" options here?

Further, when comparing the output of egs_phd for
ge-example and icrm-hpge-example

running ge-example i get
***************** Warning:
You are trying to use radiative Compton corrections
without having included rad_compton1.mortran
Turning radiative Compton corrections OFF ...

2 Should i include radiative Compton corrections for a 0.3MeV electron beam ?
2A the makefile for the pp applications is quite different from the one where the hint points to.
How to include the rad_compton1.mortran functionality in a egs pp application?

3 Bound Compton scattering norej, simple or On: Which one to choose?

Thanks !

[rtownson]

Yes, it's a good idea to turn EII on. I think that radiative Compton corrections will have a small impact at 0.3 MeV but it's good practice to check.

See here for some instructions:
https://www.reddit.com/r/EGSnrc/comments/ecuql7/radiative_compton_corrections_on/

Bound Compton scattering On is a good choice.

[ftessier]

Reminder: radiative Compton corrections requires compilations of extra source files, i.e., just turning "On" the option is not sufficient. See the EGSnrc manual page 46.

[ftessier]

Oh, I see Reid has already pointed to instructions on reddit! 👍🏻

[Jaap1957]

Thanks Mr Townson, that solves the program side of the problem.
I simulated with 1 layer of 32 mu Si , 2 MeV e- pencil beam, 1e6 histories.
That sitaution is one of the entries in the table at pg 23 of
Straggling in thin silicon detectors, Hans Bichsel, Reviews of Modern Physics, Vo!. 60, No. 3, July 1988
He gives : 7.1 keV deposited, 5.5 keV width of pulse height.

EGSnrc produces absorbed dose per e- = 9.9 keV
The pulse width spectrum , obtained with egs_phd gives
avg = 9.4 std = 4.4
Here the pulse height spectrum i get:

1 Why does the pulse height spectrum cut off at 8.5 keV ?

2 Why does EGSnrc give 20% more deposited energy than literature ?

[ftessier]

Thank you for the input file. I am able to reproduce your data. The 1 keV edge is the ae threshold artifact (secondary electrons are never created below ae, hence the depletion), and as expected it moves in stride with the value of ae:

The first peak is located at the minimum energy it takes for the electron to cross the slab (with only sub-threshold events and no deflections, i.e., straight CSDA); this is why it increases with thickness (or more exactly mass thickness as seen in the tungsten data on the right). Keep in mind that what is scored here is the total energy deposition for the entire shower (one incident electron), not the energy of every single energy deposition event.

The edge at 2 keV in silicon, I verified quickly that it depends on Z, so it appears "real" (although perhaps it could be an artefact in the data tabulations?):

[ftessier]

Out of curiosity, here is the spectrum of the individual energy deposition events, for silicon slabs of 150 nm and 1500 nm. As you can see, the energy deposition physics remains largely the same, however the first peak due to CSDA electrons going straight through is displaced to the right, and is much less pronounced, as expected (electrons are less likely to go straight trough for a thicker sample). The ae artifact peak is clearly visible at 0.001 MeV here.

[Jaap1957]

To finish this exercise, here's the table of Hans Bichsel, where i added my first steps on MC ground.
see the attached excel file.
Bichsel.xlsx

The maximum correlates the best with the value of the literature.

Indeed this is old literature already ...
What kind of new experimental or MC data would you suggest to look at in the field of :

• 100 - 300 keV
• Si layers 15 - 100 mu
• Conversion to e-hole pairs in EPI
  ?

[ftessier]

Thank you for sharing your results. Question: did you extend the density effect corrections above 1 GeV?

Fair agreement (is it within uncertainties?). As @mainegra pointed out, we have not looked at high-energy simulations with EGSnrc for a long while... But typically in that range all approximations improve. In fact, most EGSnrc transport settings probably don't matter at that point!

For the parameters you suggest (what is EPI?), I would suggest to look at electron microscopy (EM); transmission EM if you can reduce the silicon thickness.

[Jaap1957]

The distribution of EGSnrc contains density files up to 1E4 MeV
HEN_HOUSE\pegs4\density_corrections\elements\silicon.density
has as last enrties:
6.00E+03,1.431E+01 7.00E+03,1.462E+01 8.00E+03,1.488E+01 9.00E+03,1.512E+01
1.00E+04,1.533E+01

so it ends with 1e4 = 10GeV.

[Jaap1957]

Another question:
Now I want to work at 2GeV.

When i try to set UP and or UE > 999 MeV, they fall to the defaults without error mention.
In the input file I set

:start media definition: # Only 1 media definition block allowed

UE = 1000.0 # maximum energy for electrons (kinetic+0.511)
UP = 1000.0 # maximum energy for photons (kinetic)
AE = 0.512 # lowest energy for electrons (kinetic+0.511)
AP = 0.001 # lowest energy for photons (kinetic)

which results in output:
....
PEGSLESS INPUT. CALCULATING ELECTRON CROSS-SECTIONS.
density_file silicon

Medium data:

AE = 5.12000E-01 MeV, UE = 5.05110E+01 MeV
AP = 1.00000E-03 MeV, UP = 5.00000E+01 MeV

When I set
UE = 999.0 # maximum energy for electrons (kinetic+0.511)
UP = 999.0 # maximum energy for photons (kinetic)

output reads
Medium data:

AE = 5.12000E-01 MeV, UE = 9.99000E+02 MeV
AP = 1.00000E-03 MeV, UP = 9.99000E+02 MeV

so it happens above 999

Where can I override the 999 limit?

[mainegra]

BTW, All the other cross section data files in EGSnrc seem to go all the way to 100 GeV or more ...

[ftessier]

Ah ok thanks for the answer @mainegra!

[mainegra]

@ftessier as I said, the limit in pegsless is because it relies on the use of density correction files to define media!!!

[ftessier]

Yes, understood, we were typing at the same time, you answered just before I asked; very efficient support! 😄

[Jaap1957]

These maximum values are set in the get_media_inputs.mortran source file starting on line 814. @blakewalters do you know why these maximum values are set so "low", and if it is safe to increase them?

Again many thanks Mr Tessier !
I have changed the 999 entries in
\HEN_HOUSE\src\get_media_inputs.mortran
at line 838 and 847 to 20200 (20GeV).
now the simulation runs fine, and i can start re calculating these old data ...

[Jaap1957]

To summarize the remakrs of Mr Townson and Mr Tessier:

To add radiative compton corrections, follow the instructions here:
https://www.reddit.com/r/EGSnrc/comments/ecuql7/radiative_compton_corrections_on/
The result is a Makefile like the attached file (remove the .txt) (this one is in egs_phd):
Makefile.txt

For simulations above 1 GeV, change the 999 entries in\HEN_HOUSE\src\get_media_inputs.mortran at lines 838 and 847 to, for example, 10100 (10 GeV).

The pegs4 tables at HEN_HOUSE\pegs4\density_corrections\... reach up to 10GeV.

[ftessier]

Note that documentation about turning on radiative Compton corrections are also found in the EGSnrc manual, Section 2.2.2.iii (page 45).

[ftessier]

Further discussion in #831.