sPHENIX EMCal discussion

[John Haggerty <haggerty AT bnl.gov>]

Craig and Justin,

On 1/17/20 1:22 PM, woody wrote:
> Dear All,
>    We had a lively discussion at today's EMCAL meeting about how the 
> radiation damage in the SiPMs will affect the noise and our ability to 
> see MIPs in the EMCAL after we start running. While our paper on 
> radiation damage in the SiPMs doesn't answer this question directly, it 
> shows the effect we expect to see in the SiPMs currents as a function of 
> radiation dose, which I think we have now taken as a given. There are 
> also lots of other slides and talks about this subject, but we have not 
> really faced how this will affect the EMCAL calibration and performance, 
> especially after some period of running. I think some simulation of this 
> would be extremely helpful and I would therefore encourage anyone who is 
> interested to join in the discussion and help us understand how we will 
> eventually deal with this problem.

For the purposes of the simulation, I think the best information we have 
about how the detector will behave as it becomes radiation damaged comes 
from the SiPM boards that we sent to Lowell to be radiation damaged in 
their reactor:

> https://www.uml.edu/research/radlab/neutron-facilities.aspx

I like that exposure best because a) it's our exact devices, b) the 
doses were chosen to reflect our best estimate of the dose we expect in 
sPHENIX based on Jin's radiation maps, and c) Lowell is in the business 
of radiation exposure, so I think we can trust their dosimetry.  The 
data we have is limited but consistent; Eric, Sean, and Craig selected 3 
SiPM boards (== 4 towers == 16 SiPM's) and dosed each to about 1E10, 
1E11, and 1E12 1 MeV equivalent n/cm^2.  Some of what we know from that 
is attached, which is the pedestal widths at the 3 doses taken through 
the EMCAL electronics.  I think that data can be  codified into the 
simulation, at least for a first pass.

There are some complications we have to account for to put this 
carefully into our simulations.  [The first is that I should make sure 
that what I'm showing you is in a normal gain channel and not the 
switchable high gain channel, but that should be just checking the 
bookkeeping.]  You need the energy scale in the channel to use it in the 
Monte Carlo, and there is an additional effect due to the fact that the 
SiPM's are drawing current which effectively reduces the bias applied 
unless corrected for; I tried to do that and check it with the LED's, 
but I had problems with the procedure, which is also a long story, but 
boils down to the difficulty of making the LED's reproducible over time 
and temperature and removal and insertion of the daughterboards. The 
reduced gain due to current draw is not a negligible effect, and it's 
part of what we have to keep under control to keep the EMCAL calibrated 
in sPHENIX.

I think we can come up with some estimates for everything I've 
described, and that can be the first pass at adding it to the 
simulation.  I think we should actually do a second round of this using 
some sector 0 SiPM's where we dose more SiPM boards and do the 
calibration and correction more carefully based on what we learned from 
the first round.  There is yet one more complication I should mention, 
which are the dose estimates--Jin did a typically great job in his 
estimates, but estimates are, well, estimates and maybe we should make 
some measurements at higher doses to be prepared for the worst.

Finally, if anyone read this far, I think we do not want to divorce the 
calibration from the detector group, which is why I encouraged Justin to 
address the EWMCAL meeting.  We have spent a loooong time developing the 
calorimeter concepts into a working detector, and the detector groups 
are the repository of knowledge about how the detectors are built, how 
they work, and what we know about their characteristics.

--
John Haggerty
email: haggerty AT bnl.gov
cell: 631 741 3358

Attachment: radped.pdf
Description: Adobe PDF document