Transcript Reverse Breakdown Voltage Studies

Reverse Breakdown Voltage Studies
• Use Minglee’s leakage current as a function of reverse
bias voltage measurements to look for evidence of ESD
at array level.
• Post-hoc analysis of arrays with dead TX and reverse
leakage data available gives no hints.
• Statistical analysis of full data set from Minglee
– excludes files with strange leakage behaviour, one or more
channels with leakage current> 1 nA for V<10V).
– This leakage is not an artefact of the test system but is not
correlated with ESD.
– These devices were excluded from production to be safe.
1
Leakage vs Reverse Bias Voltage
Define breakdown when
leakage current > 1000 pA
Determine bias voltage at
which
(a) >=1 channel in breakdown:
Vmin
(b) All 12 channels in
breakdown: Vmax
Plot Vmin and Difference
DV= Vmax - Vmin
2
Vmin
Tail on low side for
Vmin, correlated with
large DV, typically
one channel with
early breakdown.
Tail on high side
for DV
All arrays with
channels with
DV>0.1 were
excluded from
production TXs
DV
3
Taiwan QA
• Removed all TXs with strange data ie high
leakage (>1 nA for low voltage).
• Removed all obvious suspicious TX with
DV>= 0.1 V.
• Allowed through 5 intermediate cases with
0.06 < DV < 0.1
• None of these 5 have died.
4
Profile plot of
<Leakage current> vs channel#
For lowest voltage at which at
least one channel in breakdown
(I> 1000 pA).
Why are edge channels worse
(particularly channel 11) ?
Not an artefact of test set-up
Must be a handling issue or
effect of wafer dicing.
5