Problems with CCE simulations

1
Problems with CCE simulations

• David Pennicard
• University of Glasgow

2
Perugia radiation damage model (p-type)
Ec
Ev
3
The two acceptor levels reproduce leakage current
and Neff well
Act as electron traps when collecting charge
Ec
-
- -
Close to midgap Significant effect on space
charge and leakage
Acceptor levels Majority contain holes and have
no net charge Small proportion contain electrons
give -ve space charge Will trap excess electrons
Ev
4
The donor level has little effect on leakage and
space charge. It was added in to give correct CCE
performance
Ec
Far from midgap, so little effect on leakage and
space charge
Donor level Vast majority contain electrons and
have no net charge Very small proportion contain
hole give ve space charge Will trap excess holes
0
Ev
Act as hole traps
5
Tests of CCE in Perugia model

• With MIP signals, matches 300µm strip detector
  test results well
• This is shown in their paper, and I reproduced
  these results too
• Readout signals look reasonable

6
Tests of CCE in Perugia model

• With MIP signals, matches 300µm strip detector
  test results well
• This is shown in their paper, and I reproduced
  these results too
• Readout signals look reasonable
• BUT the trap parameters might just be fitted to
  the data from a specific device without
  necessarily giving the correct internal behaviour

7
Trapping time

• Experimentally, trapping leads to an exponential
  decay in free carrier concentrations (and hence
  the current signal)
• The trapping time is related to the concentration
  and cross-section of the relevant trap(s). For
  levels above the midgap
• Similarly, levels below the midgap will lead to a
  hole trapping time

8
Trapping time

• Experimentally, trap concentration increases with
  fluence, so the trapping times are characterised
  by
• Experimentally, ße ßh 4.010-7cm2s-1
• But, with the Perugia trap cross-sections and
  concentrations, I calculate
• ße 1.610-7cm2s-1
• ßh 3.510-8cm2s-1
• So, this would imply a factor of 4 difference in
  trapping times in the model. In principle, could
  correct this by increasing donor cross-sections
  or concentration.
• Experimentally, ß doesnt tend to be measured
  beyond 1015neq/cm2, and CCE from highly
  irradiated detectors is better than expected, so
  lower ß could be OK

9
Tests of electron hole trapping

• N-p diode
• Charge deposited at front to test hole
  collection, and back to test electron collection

N
5µm
h
e
Charge at front, most of collection signal comes
from hole drift
300µm
p-type
P
10
Tests of electron hole trapping

• These tests were done with
• No damage
• Perugia trap model, with 1015neq/cm2
• Modified Perugia trap model, with hole trap
  concentration increased by 4 to give (in
  principle) equal electron and hole trapping times
• The modified Perugia trap model gives the same IV
  characteristics and electric field distribution
  as the regular one

11
Charge deposited at front surface no traps
Brief pulse as electrons collected
Fairly uniform current until holes collected
Signal at p ohmic contact matches n (except for
sign)
12
Charge deposited at front surface with traps
Current larger at start! Scales with hole trap
conc (not simply due to altered field)
Signal decays with time (expected) but note that
charge signal is increased overall!
P signals are smaller than those at n
differences in electrode currents are not
expected in 2-terminal devices
13
Charge deposited at back surface
Brief pulse as electrons collected
Electrons drift from back surface to n Faster
collection than for holes
Fairly uniform current until holes collected
Signal at p ohmic contact matches n (except for
sign)
14
Charge deposited at back surface
Current is larger at p than n, but results here
are less unusual
15
How do the hole trap parameters affect this?

• The change in the front simulation signal
  scales with the predicted inverse trapping time
  vthNsp
• i.e. doubling the trap hole cross-section has the
  same effect as doubling the trap concentration
• altering the electron cross-section has no effect
• Changing the energy level of the hole trap has no
  effect (tested by 0.05eV)
• Putting it close to the midgap would alter Neff
• The effect on the hole collection signal is not
  significantly changed if the acceptor levels are
  removed (the change in the electric field has
  some effect)
• If the charge is deposited 25µm from the surface,
  the effect is reduced somewhat but not eliminated.

16
Behaviour within the devices

• I plotted the state of the devices a different
  points during the transient simulations, then
  viewed them with Tecplot.
• The following results use a p-type substrate and
  the Perugia trap models with increased donor
  concentration.

17
Hole collection hole current
With traps, hole current decays
Motion of holes
18
Hole collection hole recombination
Recombination matches hole transport
19
Hole collection trapped holes
Increase in trapped hole conc. after holes pass
20
Hole collection displacement current
Unequal displacement current matching unusual
readout signals
21
Electron collection electron current
Electron drift is also reasonable
Indications of correct trapping
22
Electron collection displacement current
Once again, displacement current corresponds to
odd readout current
23
Tecplot results

• The carrier transport, and the parameters
  directly associated with trapping (recombination,
  trap occupation) all seem reasonable.
• The displacement current, which reflects the
  changing electric field in the device,
  corresponds to the odd readout signals
• Possibly suggests that carrier transport and
  trapping is working OK, but the transient
  electrostatic behaviour is wrong?

24
Tests with alternative trap model

• These tests were re-done using a 2-trap model
• The model has one acceptor trap above midgap
  (which will trap electrons) and one donor trap
  level below the midgap (traps holes)
• These are at Ec-0.3eV and Ev0.3eV respectively,
  and have equal concentrations and cross-sections
• Since these are far from the midgap, there is
  little effect on Neff, and so I can do tests with
  both n- and p-type substrates

Ec
Acceptor at Ec-0.3eV
Donor at Ev0.3eV
Ev
25
Tests with alternative trap model

• Results
• The form of the graphs are the same (excessively
  high currents at the start of the simulation
  followed by a decaying signal)
• Larger currents tend to be produced when the
  charge is deposited nearer the rectifying contact
  (i.e. the n front contact if the substrate is
  p-type, the p back if the substrate is n-type).
• Given a particular substrate type and position of
  charge deposition, the current signal is larger
  at the contact near where the charge is deposited
• Given the above, the relative sizes of the
  signals are much the same for electron and hole
  transport
• A plus side the decay rates of the current
  signals are reasonable
• Example

26
Alternative trap model hole collection signal
Excessively high signal. This trap model doesnt
affect the electric field pattern, so this is
specifically a trapping effect.
But, by dividing the signal with traps by the
undamaged signal
27
Alternative trap model hole collection signal
The decay of the signal fits an exponential exp
(-t/t) t1ns, which is close to the predicted
value of 1.3ns from
Exponential curve
Simulation result
28
Alternative trap model electron signal
This is the electron collection signal from the
p contact in an n-type substrate device. The
lifetime is te0.75ns This matches the hole
lifetime, when you consider that vthh0.75vthe
Exponential curve
Simulation result
29
Further work

• In theory, readout signals can be deduced by
  finding the carrier transport then calculating
  the signal induced on the electrodes.
• Signal formation Experimentally, Ramos theorem
  is still valid even with radiation damage. But do
  these trap models alter the signal formation?
• Alternative modelling methods? Potentially,
  simulations can be done
• Suggestions?