Recent progress with TES microcalorimeters and signal multiplexing
J. UllomNIST
NASA GSFC
SRON
J. BeallR. DorieseW. DuncanL. FerreiraG. HiltonR. HoranskyK. IrwinB. Mates
G. O’NeilN. MillerC. ReintsemaD. SchmidtL. ValeY. XuB. Zink
Transition-edge sensor (TES) calorimetry
Tem
pera
ture
Time
C
EC
G
energy (x-ray)
ConductanceG ThermalC
HeatCapacity
temperature response
0.02
0.04
0.06
095.8 96 96.2
Temperature (mK)
Res
ista
nce
()
I
V SQUIDcurrent amp
thermometer
TES issues• single pixel performance
energy resolution
capture efficiency speed
• multipixel arraysease of fabrication, homogeneity
• stability of operation
• readout of arrays
• expected noise sources:- fluctuations in thermal impedances- Johnson noise
• unexpected noise source:- behaves like white voltage noise
Obstacle to better TES resolution: unexplained noise
L/R roll-off
Johnson noise
phonon noise
unexplained noise
Different TES geometries
additional normal metal features
definition: = (T/R) dR/dT perpendicular bars reduce
Noise vs. geometry: unexplained noise and correlated
• low designs have little unexplained noise
• perpendicular normal features reduce noise and
all data at 60% RN
7
6
5
4
3
2
1
0
Unexplained Noise/Johnson Noise
10008006004002000
standard
parallel&perp
sparsepartial perp
densepartial perp
wedge
dense parallel
islands
#2standard
densefull perp
≈≈ 450 450 ≈≈ 450 450 ≈≈ 500 500 ≈≈ 150 150
≈≈ 4400 ≈≈ 40 40 ≈≈ 1515
SRON parameter study
normal islands and bars …
noise measurements to follow
Design strategy: match E-max to 5.9 keV, lower
= 45C = 0.9 pJ/KM = 1.2-1.4
1.5 m Bi50% at 6 keV261 s
400 m
2.0
1.5
1.0
0.5
0.0
C (pJ/K)
100806040200
25%RN
A:3.2eVB:2.9eV
C:2.4eV
E-max10.2keV
E-max20.6keV
sllerE-max
lrerE-max
X-ray absorbers
simplest absorber = material stacked on TES
need machined collimator to shield streets
what is fill fraction ?
for NxN array, max wires in 1 street = N (near center)
demonstrated: 2 wires & spaces in 3.5 m for N=30, min street width ~ 55 m with litho development, ~ 25 m ?
fill fraction = 67% [83%] for 250 m pixels = 86% [93%] for 700 m pixels
= 90% [95%] for 1 mm pixels
demonstrated: 2.4 eV in 250 m device 2.9 eV in 400 m devicepredict 4.5 eV in 680 m, 6.0 eV in 830 m
element
d for 95% QE at 6 keV
C/m2 [10-19 J/K] size for C = 2.5 pJ/K
Au 3.5 m 240 (at 100 mK) 320 m x 320 m
Bi 6.3 m 5.8 (?) 2080 m x 2080 m
Cu 29.5 m 2900 90 m x 90 m
Sn 7.8 m 1.3 4390 m x 4390 mHgTe, …
BiTES
SiNx
Si
~ 4 eV
(L pix+Lstreet)2
L pix2
X-ray absorbers - mushrooms
very high fill fractionoverhang can shield streetsmore challenging design, fabrication
absorber
SiNx
Si
TES
normal metal - Au
normal metal - heat pipe
Bi
GSFC - 4 m electroplated Au~2.5 eV at 6 keVTc = 65 mK, = 7 ms
GSFC
BiCu, ~4.5 eV at 6 keV
NIST
TES for 100 keV: attach bulk absorber
1 mmSn aborber: QE = 20% at 100 keV
Mo/Cu thermometer
now 27 eV at 103 keV
E = 4.9 eV;Number of counts = 255534;
5825 5850 5875 5900 5925Energy(eV)
0
5000
10000
15000
stnuo C
16 hour acquisition, no gain correction E 10% worse than in short record
3/4 hour acquisition, no gain correctionno detectable drift
TES stability
stable long-term operation possible …
SRON NIST/CSTL
… but cannot yet be taken for granted. Requires close attention to stray RF power, stray magnetic fields, and temperature stability. Also, some dependence on device andbias point. These dependencies not yet understood.
Measure many TESs in multiplexed test setup
6.25 mm
interfacechip
32:1 multiplexerchip
8 x 8 sensorarray
individualsensor
128-pixel MUX facility complete
four 32-channel SQUID
MUX chips16 x 16 x-ray array will be tested at the end of June
presently: -ray cal
Arrays lots of data: multiplexed R(T) curves
• variation in transition shape variation in response• we can already engineer the transition width; soon we will engineer the transition smoothness
sensor± (s)
open loop BW (MHz)
E (eV) single muxed
pixels per column
pixels in square array
1000 3.5 3.6 4.0 196 38416
300 12 1.8 2.0 95 9025
50 12 4.5 5.0 32 1024
Future mux performance
presently, cryogenic BW ~1.5 MHz and electronics BW 3 MHz (designed in 1999)
we will increase system BW to 3.5 MHz …- minor adjustments
and then to 12 MHz.- cold series array, electronics redesign
simulated MUX performance:
but …NeXT ?
New SQUIDS!
gradiometric summing coils
gradiometric SQUIDs
asymmetric V-
• mutual inductance optimized for x-ray measurements
• asymmetric V- greater dynamic range & linearity
• gradiometric design less magnetic shielding & crosstalk
(will help system engineering)
• 100 mK testing in June; production run scheduled for July
optimism is in order - TES calorimeters continue to improve
• energy resolutions < 3 eV at 5.9 keV, 27 eV at 103 keV
• very promising results in complex absorber structures - mushrooms: ~ 2.5 eV - attached bulk absorbers: 27 eV (-ray)
• array fabrication feasible- some work ahead to improve homogeneity
• lengthy, stable spectra feasible- some work ahead to make routine
• time-domain SQUID mux works well
> 196 NeXT-like pixels [1 ms] in 1 channel in 2007 ?
> 32 fast pixels [50 s] in 1 channel also very feasible
Conclusions
a TES option forNeXT could be VERY large
does the science casejustify a -ray array ?