Cosmic Ray Background in TES array

tobascothwackUrban and Civil

Nov 15, 2013 (3 years and 6 months ago)

69 views

Piet de
Korte


Febr
. 2011

Cosmic Ray Background in TES array

SAFARI DETECTOR SELECTION REVIEW 21
-
23 JUNE 2010

2

Cosmic Ray background

Cosmic Ray Energy Deposition

Minimum Ionizing Particles in Si deposit 1.66 MeV/g/cm
2




152
keV

in 0.4 mm (wafer thickness)


27
keV

in 70
μ
m Si (beam thickness
)


0.4
keV

in 1
μ
m Si
(pixel thickness)

Cosmic Ray Count rate


3.14 c/cm
2
(thin surface)


0.4 c/beam (in detector plane)


0.07 c/beam (perp. to det. plane
)


3 10
-
4

c/s/pixel

Simulations


Full Si
-
wafer + 2 beams of the array


Cu
-
layers


Cu
-
surface contact to bath


Measured values for Heat capacity and heat conductance


Heat conduction by electrons and phonons


Bias power loading

Requirement



Thermal noise < 1
uK
/
√Hz



So for an exponential decay heat pulse (
Perveval,s

theorem)
amplitude x
root(time constant) (A.

τ
)

< approximate 10
-
6

K/
√Hz


Cosmic Ray Hits in pixels



Energy dump about 400
eV

in 1um thick Si
3
N
4



Pixel size 130 x 70 um
2

for M
-
band


Count rate 3 10
-
4

c/s


Recovery time 40
ms


Dead time 12us/s/pixel.

Pulses have a steep rise time of about
30us,
a saturated
maximum
(4fW)
and therefore easy to recognize before data decimation and
filtering.
In
the telemetry data they will be spikes covering only
1
-

2
samples. In total we might have to delete a 40
ms

period after the
onset

1
10
6


1
10
5


1
10
4


1
10
3


0.01
0.1
1

10
7


0
1
10
7


2
10
7


3
10
7


4
10
7


400 eV Cosmic Ray in SAFARI pixel
Time(s)
Amplitude(A)
SAFARI DETECTOR SELECTION REVIEW 21
-
23 JUNE 2010

4

Cosmic Ray background (Hits in beams)

152 keV hit (top)
Cu temperature
0,05
0,055
0,06
0,065
0,07
0,075
0,08
0,085
0,09
0,095
1E-09
1E-08
1E-07
1E-06
1E-05
1E-04
1E-03
1E-02
1E-01
Time [s]
Temperature [K]
Center
+ 1 pixel
+ 2 pixels
+4 pixels
+8 pixels
edge
Temperature increase vs time in beams

Influence Cosmic Ray Hit in beam:



The slow tails of all hits modeled (up to 8 pixels away from hit)
generate signal above (


200 x) the noise limit



Dead time for rejection equals 0.4 c/s x 3
0
ms

=
1.2%


Cosmic Ray beam hits and pixel signals

1
10
6


1
10
5


1
10
4


1
10
3


0.01
0.1
5

10
3


0
5
10
3


0.01
0.015
0.02
Safari pixel response to nearby Cosmic Ray beam hit
Time(s)
Normalized amplitude
Pixel response to cosmic ray hit of a beam. This is calculated for
the most nearby pixel. Due to the time
contants

the amplitude of
the response is only about 0.1
fW

saturation energy.

Countrate

: 0.4 c/beam

Deadtime

: 10ms/s/beam

SAFARI DETECTOR SELECTION REVIEW 21
-
23 JUNE 2010

6

Cosmic Ray background (Hits in bulk substrate)


Temperature profiles are modeled for
the outermost pixels (
-

1 mm)

on
each Si
-
beam


Au wire bonds for thermal anchoring of the Cu
-
coated Si
-
wafer to bath
reduces the approximately exponential return to normal of the slow
pulse part to about
τ

< 1
ms.

(about 10
ms

without wire
-
bonds)


So on the basis of Perceval’s theorem we can accept
dT
-
pulses with an
amplitude of about 30
uK

before crossing the noise requirement


The highest
dT

for the slow component (Cosmic Ray incident at end of
beam: 1mm from pixel) is 40
uK
, slightly larger than this value


The maximum fast component (same position) in this case is 20
mK

with
2 us decay time, 30x above the noise threshold.


At 2 mm distance the fast component is 0.5
mK
, and contributes just the
excepted noise level.


So for an area of about 3 mm radius the background has to be rejected,
leading to about 1c/s or 1%
deadtime
.


Outside this area the signals drop fast and don’t contribute significantly to
the noise level


Conclusion: Frame hits generate temperature noise below the
1uK/
√Hz and only the ones closer than 3 mm from the end of the
beams have to be rejected, generating about 1%
dead time

Developments required:



Optimization of thermal anchoring wafer to bath



Measurements on dead time for beams and thermal noise from
frame

SAFARI DETECTOR SELECTION REVIEW 21
-
23 JUNE 2010

7

Cosmic Ray background (Hits in bulk substrate)


Temperature profiles are modeled for
the outermost pixels (
-

1 mm)

on
each Si
-
beam


Au wire bonds for thermal anchoring of the Cu
-
coated Si
-
wafer to bath
reduces the approximately exponential return to normal of the slow
pulse part to about
τ

< 1 ms. (about 10 ms without wire
-
bonds)


So on the basis of Perceval’s theorem we can accept dT
-
pulses with an
amplitude of about 30 uK before crossing the noise requirement


The highest dT for the slow component (Cosmic Ray incident at end of
beam: 1mm from pixel) is 40 uK, slightly larger than this value


The maximum fast component (same position) in this case is 20 mK with
2 us decay time, 30x above the noise threshold.


At 2 mm distance the fast component is 0.5 mK, and contributes just the
excepted noise level.


So for an area of about 3 mm radius the background has to be rejected,
leading to about 1c/s or 1% deadtime.


Outside this area the signals drop fast and don’t contribute significantly to
the noise level


Conclusion: Frame hits generate temperature noise below the
1uK/
√Hz and only the ones closer than 3 mm from the end of the
beams have to be rejected, generating about 1% deadtime

Developments required:



Optimization of thermal anchoring wafer to bath



Measurements on dead time for beams and thermal
noise from frame

0,0

0.5,0

1,0

2,0

4,0

4,6

8,0

12,0

12,4

Distances x, y (mm) from end of beam


X parallel to beam

Temperature excursions due to Cosmic Ray hits of wafer near
end of beams

Impact Cosmic rays on spectrum

150
200
250
300
1
10
5


1
10
4


1
10
3


fts+detector
40Hz detector and 0.1 fW flat spectrum + detector noise+ 80 beam hits
wavenumber (cm-1)
Power/bin (fW)
150
200
250
300
1
10
5


1
10
4


1
10
3


fts+detector
40Hz detector and 0.1 fW flat spectrum + detector noise
wavenumber (cm-1)
Power/bin (fW)
On the left a flat spectrum of total 0.1
fW

measured with FTS and with
3 10
-
19 W/√Hz detector noise.

On the right the same measurement but randomly during the scan 80
beam hits have been added to the data.

Intensity is in
fW
/bin. The number of bins is 512 between 160 and
290 cm
-
1

Cosmic Ray Hits on Spectrum

150
200
250
300
1
10
5


1
10
4


1
10
3


0.01
fts+detector
40Hz detector and 0.1 fW flat spectrum + detector noise + 1 pixel hit
wavenumber (cm-1)
Power/bin (fW)
This shows the impact of 1 pixel hit. This hit obviously has to
be removed from the data

Discussion

1)
Dead time for Cosmic Ray events doesn't seem to be a
problem (0.001% for single pixel hits, 1.2% for beam hits, 1%
for frame hits near beams)

2)
Hits in beams and in the wafer near the end of the beams will
in general result in rather small signals (< 0.02 of saturation)
with long rise times (1
ms
), and decayed
a
fter about 30
ms

3)
Search for hits in individual pixels seems relatively easy. On
the ground this means removal of spikes of 1 or 2 samples
long (total about 40
ms
)

4)
Search for the small coincident signals of various pixels along
one beam could be more difficult,
and probably has to be done
on board in full bandwidth data!

5)
Crude assessment of the impact of these hit on the FTS spectra
is done. Removal is needed to maintain the low noise floor of
the instrument.

6)
Heat sinking of beams and wafers is crucial for success