Backscattering Brief Report
Problems with Scintillator from Last Time
When we ran last time, two concerns were brought up with our scintillator
data:
- scintillator quality was bad: It was an old piece of NE102 found in
our shop, which obviously suffered from microfractures or "crazing".
- scintillator charging not adequately addressed: The NE102 was coated
with a thin layer of graphite, which mitigated charging. However, the
effects of the graphite could never be quantified very well (except from
back of the envelope arguments that it should have no effect).
Resolution of those Problems from Michael Betancourt's work this summer
- scintillator quality was improved: Eljen scintillator (EJ204) was used.
Both sides of the scintillator presented a smooth face, as opposed to the
crazing observed for the previous samples.
- scintillator charging was addressed: An evaporated Al coating of
1 Ohm/square (500 Angstroms) was used to mitigate charging, as opposed to
the graphite coating of the previous targets. It was additionally
confirmed explicitly that the scintillator NIBF data using the current
integration method varied as expected with the various bias voltages.
This was not done in detail for the data collected in summer 2002.
Additional systematic effects
- scintillator aging in electron beam: It was found that running
higher energy beam at currents in excess of 1 nA tended to increase
the NIBF by up to 20-30 percent over the course of several hours. The
increase in NIBF seemed to be correlated with increasing energy, as
well as with increasing current. The damage manifested itself in a
visible reduction in light output from the affected part of the
target, which would not heal with time (over the course of several
days, when kept under vacuum). When an affected target was removed
from the backscattering chamber, an obvious brown mark could be seen,
which appeared to be at a few millimeters depth into the scintillator.
No obvious change in the mirror interface of the Al coating was
observed. This potential systematic was characterized by running at
various energies and currents over long periods of time. For
production data-taking, virgin scintillator was run at currents below
1 nA and data at 120 keV was taken last in order to ensure good data
quality. The systematic uncertainty was previously uncontrolled, and
is likely responsible for the global 20% in Mike's data compared with
the previous data.
- smaller "duckbill": The so-called "duckbill" is a piece of
conducting target rod which penetrates inside the top of the grid
which we use to characterize the effects of secondary electrons up to
+/- 100 Volts. In our previous article, this is referred to as the
"target rod" correction. The duckbill must be held at the potential
of the chamber, because of poor isolation between the chamber and
target rod. Some backscattered electrons hit the duckbill and are
correctly included in the backscatter count by summing with the
chamber. However, sometimes low-energy secondary electrons are
created when a backscattered electron strikes the duckbill. For this
reason, a residual dependence on target voltage is observed, even when
the grid is biased at -100 V, which should prevent most low-energy
secondaries from getting through. Indeed, in the previous data, this
was accounted for by correcting "grid on" data with "grid off" data,
applying a "duckbill fraction" correction, which successfully flattens
out the residual dependence on target voltage. However, the "duckbill
fraction" which successfully accounts for a perfectly flat dependence
did not agree within 50% with the geometrical solid angle one would
expect. Thus a systematic uncertainty equal to the range of NIBF's
sampled by varying the duckbill fraction with the 50% uncertainty was
assigned. This resulted in a systematic uncertainty of 5% from this
effect. For Mike's data, the duckbill was remachined from scratch.
This reduced the size of the duckbill, and corresponding a reduction of the
correction to 3% was achieved.
Final Plots
The plots from Mike's and my analysis, along with G4 and Penelope curves
provided by Junhua and Seth are included below (data and MC compilation
by Junhua):
Summary and Plans
With these new plots in hand, we intend to write a brief report for
Phys. Rev. C updating our previous work with these new results on
scintillator.
The main points above and the final plots would be included in this
report. I draw your attention to the plot comp_PEN_scint.eps which previously
contained a rescaling factor 1.4 and now contains a rescaling factor 1.2.
This factor caused us considerable consternation last time, and I feel
it is now well in hand. (I'd also note that it didn't change much
outside the 12% systematic uncertainty in normalization we were assigning
last time, and would again assign this time.)
One additional issue which we have not addressed since last time is the
behavior of Penelope at low energy (although this is addressed in detail
in Seth's thesis in particular for Be). I'm not sure if we should consider
having a bit closer look in particular for scintillator which contains
hydrogen. A concern previously was the tendency of secondary electron
corrections for lighter materials to be very large in Penelope, but virtually
non-existent in G4.
Assuming we do not do more study of Penelope/G4 discrepancies,
I think a reasonable schedule would be:
- 1st Draft of new article: Monday Nov. 29, 2004.
- comments due: Friday Dec. 3, 2004.
- 2nd Draft: Monday Dec. 6, 2004.
- submit to Phys. Rev. C: before Dec. 25, 2004.
I would generate the first draft based on our previous article and
Mike's SURF report.
Any comments on the work of this summery or on the figures (before I
spend a lot of time generating the first draft) would be greatly appreciated.
Jeff Martin
Nov. 22, 2004.