A high energy resolution gamma-ray TES calorimeter with 0.5 ms response time

T. Oshima, Y. Yamakawa, H. Kurabayashi, A. Hoshino,Y. Ishisaki, T. Ohashi,

(Department of Physics, Tokyo Metropolitan University )K. Misutda (ISAS/JAXA)

K. Tanaka (SII Nano-Technology)

Applications & requirements• Astrophysics

– 44Ti （68.9/78.3 keV, T1/2~ 59 yr） γ-ray emission from supernovaremnants

• direct information on the radioactive isotopes produced by the explosions• not obscured by galactic dust

⇒ provides a stringent test of theories of supernova nuclear reactions. Fly γ-ray calorimeters on balloons!• Material analysis (EDXRD)

– γ-ray diffraction spectroscopy of crystals under ultra highpressure(~MPa) → Poster by A. Hoshino

– interesting structures ~80keV– time evolution ⇒ fast response time

• Our GOAL–Bandwidth ~ 100 keV–Energy resolution ΔE ~ a few times 10 eV–Fast response time τ << 1ms–High Efficiency @ ~80keV

Ti/Au TES

Sn (Z=50)（t0.3mm）

Energy (keV)

Qua

ntum

effi

cien

cy

Our first trial (SII-115)

TES: Ti/Au(40/70nm)X-ray absorber:Au(500nm)Membrane: SiN bridgeTc: 151 mK

Pulse shape analysisshows no significantfluctuation of τrise .Let’s try a faster one!

500 µm

300 µm

SII-115

Si

Au

Sn absorber

a bSiN membraneNbTES

cross section

epoxy

ΔEFWHM =138 eV @60keVτrise~ 110µs

τfall ~ 2.0 ms

b

a15µm spacer

Sn

0.79mm0.87mm

Ca =11.2 pJ/K

ΔEbaseline= 81 eV

Energy(keV)

Cou

nts/

bin

Energy resolution:Response:

Trade off• Trade off between the signal loss and the position

dependence.– τfall↓( ≅ G↑) ⇒ τfall / τrise ↓ ⇒ ΔTTES ↓ ⇒ signal loss– τrise↓( = Ga ↑) ⇒ Position dependence in the absorber↑

• How fast can we make?⇒ SPICE simulation• Boundary conditions:

–Limited Ga : @150mK (@100mK)• Stycast 2850FT ~ 100nW/K (30nW/K) for r =100µm, h=20µm• Kapitza ~ 170nW/K (50nW/K) per r =100µm boundary (2 boundaries)• electron-phonon ~ 450nW/K (85nW/K)⇒ Ga ~ 50nW/K (~20nW/K)

–Saturation at the transition edge:• lower limit to Ca

Ca , Ta

Ga

C, T

Tb

G

Our Second trial (SII-155)

TES: Ti/Au(40/120nm)X-ray absorber: Au(500nm)Membrane: SiN squareTc: 151 mKC = 2.0 pJ/KG = 3.5 nW/KΔE = 12 eV @ 5.9 keV

500 µm

300 µm

SII-155

Au

( )

Saturation

approx.

Ca ~ 7pJ/K

⇒ size ~ 700µm

Temperature change of the TES

Ca (pJ/K)

ΔT

(mK)

10nW/K

20nW/K

50nW/K Ga =100nW/K

E =100keV, C =2pJ/K

Ca , Ta

Ga

C, T=153mK

3.5nW/KTb=132mK

Thermal model

G

60 keV

100 keV

T (mK)

R (m

Ω)

ΔTmax < 1mK

SPICE Simulation: Noise spectrumEquivalent circuit

Rs

RTES

LinC

Ca1/Ga

1/G

PETF

thermal electrical

Ca , Ta

Ga

C, T=153mK

3.5nW/KTb=132mK

Thermal model

G

Frequency (Hz)

Johnson noise of– Shunt resistor Rs =3mΩ– TES RTES = 25.58mΩPhonon noise at– G = 3.5nW/K– Ga = 10nW/K– Readout (SQUID) noise 20pA/√Hz– Total

Current noise spectrum

(LLNL: Miyazaki et al.)

I n (A

/Hz1

/2 )

SPICE Simulation: Position dependence (1)• FEM-like model with SPICE

– Absorber: 3D network of C and G 7 x 7 x 3 elements (Please imagine a 3D picture!)– stycast: 1 element– TES: 1 element

• generate pulses using a modeled R-T curve• optimal filtering with simulated noise spectrum⇒ pulse height ⇒ ΔEFWHM

• Material Parameters of Sn:– κSn = 1.24T 3.1 W/K/m (foil)– cSn = 0.054 pJ/K

• Scan parameters: Ga , absorber area α

T (K)

R (Ω

)

modeled R-T curve

Operatingpoint

Thermal model

stycast

TESG

Ga/2

Ga/2

SPICE Simulation: Position dependence (2)

Make Ga as large as possible! 700µm 25 eV && τfall = 300µs1050µm 36 eV && τfall = 600µs achievable.

Ga (nW/K)

ΔES/N (eV)

ΔEposition (eV)

τfall (ms)

ΔEtotal (eV)

size = 700µm, 1050µm

Our Second trial (SII-155)

Stycast 2850FT

300µm

TES: Ti/Au(40/120nm)X-ray absorber: Au(500nm)Membrane: SiN squareTc: 151 mKC = 2.0 pJ/KG = 3.5 nW/KΔE = 12 eV @ 5.9 keV

Ca = 6.9 pJ/K

500 µm

300 µm

SII-155

630 µm

670 µm

Sn

Surface roughness of Sn: σ ~ 2.5 µm σ ~ 0.6 µmPolished

with aluminapowder(~3 µm)

SII-155+Sn results (1)

good linearity2.1%@60keV

±10eV

Comparison withthe SPICE simulation

Measured 60keV Simulation: Ga=40nW/K 30nW/K 20nW/K 10nW/KΔ

I (µA

)

Time (ms)

Ga ~ 30nW/K

Time (ms)

ΔI (µA

)

averaged pulses of various EE=60keVτrise = 46 µsτfall = 520 µs

Operating point(Vb =1.25µV, α ~200)

60 keV

100 keV

T (mK)

R (m

Ω)

SII-155+Sn results (2)

Np-L

AmSn-escapeAm

Sn-escapeAm

Np-LβNp-Lα

Np-Lγ

241Am5367sec, 20580 events

ΔEbaseline=37.9±0.7 eV

SII-155+Sn results (2)

Np-L

AmSn-escapeAm

Sn-escapeAm

Np-LβNp-Lα

Np-Lγ

241Am5367sec, 20580 events

ΔEFWHM =38.4±0.9 eV@60 keV

ΔEbaseline=37.9±0.7 eV

Energy resolution of the nuclear γ-ray lines（26, 60 keV）agrees with ΔEbaseline⇒ no degradation by the position dependence

SPICE simulation: Ideal 100mK device

Ca (pJ/K)

ΔT

(mK)

10nW/K

20nW/K

50nW/K Ga=100nW/K

E=100keV, G=1nW/K

15.15.014.2970301140012.76.810.842040310509.73.59.1160405700

ΔEtot(eV)ΔEpos(eV)ΔES/N(eV)τfall (µs)Ga(nW/K)G(nW/K)Size(µm)

Results for E=60keV signalScan parameters: Ga = 10 - 50nW/K, G = 1 - 5nW/KHard to list all. Results for some parameters are shown.

Ca ~ 4pJ/K⇒ size ~ 1000µm

Saturation:

α

model R-T curve

T (K)

R (Ω

)

2mK

T = 98.7mK R = 0.3 Ω α ~ 50

Summary• Designed a fast response γ-ray TES calorimeterusing SPICE simulation.

• 150mK device with 700µm x 700µm x 300µm tinabsorber demonstrated

– τfall = 520 µs, τrise = 46µs– ΔE= 38.4±0.9 eV eV@60keV, ΔEbaseline=37.9±0.7 eV– No degradation of ΔE due to the position dependence

seen as the simulation tells.• SPICE simulation for 100mK device shows

– τfall < 500 µs– ΔE < 15 eVis achievable with 1mm2 x 300um absorber

SPICE Simulation: non-linearityα

T (K)

R (Ω

)

modeled R-T curve

Operatingpoint

・Generate simulated pulses using a modeled R-T curve・Apply optimal filtering⇒ pulse height

Equivalent circuit

Rs =3mΩRTES

LinC

Ca1/Ga

1/G

ETF

thermal electrical

Ca , Ta

Ga

C, T=153mK

3.5nW/KTb=132mK

Thermal model

G

Ga (nW/K)

non-

linea

rity

(%) size=500µm (3.8pJ/K)

750µm (8.6pJ/K) 1000µm (15pJ/K) 1250µm (24pJ/K) 1500µm (34pJ/K)