background limited antenna coupled MKID arrays for ground based imaging

Jochem Baselmans, Andrey Baryshev, Stephen Yates, Akira Endo, Lorenza Ferrari, Pascale Diener, Jan-Joost Lankwarden, Pieter de

Visser, Reinier Janssen, Henk Hoevers, Teun Klapwijk

1

2

Outline

• A-MKID • Photon noise limited peformance • Hybrid antenna coupled MKID • Test chip noise performance – effect of Al and interfaces • A-MKID system tests

A-MKID

3

• 2 color imaging instrument for APEX, 870 µm and 350 µm • 15 arcmin FOV, 1 Fl pixel spacing • 2 separate arrays of 4 sub-arrays in the FP, polarizer to select band band 870 µm 350 µm # pixels 20.000 pixels 3.200 pixels goal sensitivities 34 mJy s0.5

59 mJy s0.5

goal pixel NEP 2.7e-15 W/Hz0.5 1.4e-14 W/Hz0.5

A-MKID - readout

4

• Phase readout due to low Q of the devices (20.000) due to loading • MPIfR developped readout (full IQ)

• Bandwidth 1.25 GHz • 32768 bins (76 kHz resolution) • 8 ENOB (expected)

A-MKID - readout

5

• Phase readout due to low Q of the devices (20.000) due to loading • MPIfR developped readout (full IQ)

• Bandwidth 2.5 GHz • 32768 bins (76 kHz resolution) • 8 ENOB (expected) • based upon E2V 10AQ 190 ADC, 4x1.25 Gsample/sec • and analog devices AD9739 DAC

Photon noise limited performance

6

• Random arrival rate of photons creates a white noise in quasiparticle number rolled of at the lifetime

• On top of this there is recombination noise from the random recombination of excess quasiparticles

• If all quasiparticles are due to photon absorption:

Quasiparticle recombination

Photon arrival

KID responsivity

Noise PSD

Photon noise limited performance

7

• Photon noise level ∝ n and independent of radiation power!

• Photon noise level not much above the G-R noise

• Not easy to reach • Carefull material selection required

• Photon noise limited KIDs have always a contribution due to qp

recombination • Sphoton + SG-R = Sphoton (1+0.85) at 350 GHz for Al

Quasiparticle

recombination Photon arrival

KID responsivity

Noise PSD

Hybrid antenna coupled MKID

8

Hybrid antenna coupled MKID NbTiN as MKID material

9

• Lower frequencie noise • 1.2 THz gap frequency • Very good readout power handling

Rami Barends, Ph.D. Thesis (Deft University) 2009

Pi=-40 dBm Pi=-30 dBm

NbTiN on Si Pi=-40 dBm

Hybrid antenna coupled MKID Al to detect quasiparticles

10

• Al on saphire (see P. de Visser) has shown generation-recombination noise

• Using Al to absorb the radiation should allow us to reach the photon noise limit

See Visser et al., Phys. Rev. Lett. 106, 167004 (2011)

Hybrid antenna coupled MKID Antenna – lens to absorb radiation

11

• Lens array • creates space for KIDs • High filling fraction • Focussing of radiation to antenna

• Si lens array + Parylene-C l/4 AR coating

CST modelling + measurements

12

-50

0

50100 -100

-50

0

50

100

-30

-25

-20

-15

-10

-5

0

151210

36

22151411

Tone number16

1716819

2320

-30

-25

-20

-15

-10

-5

0

Fabrication – Hybrid KIDs on Si

13

Dry etch 300 nm NbTiN 13.5 sccm SF6 20 sccm O2 65 ° slopes

50 nm Al sputter depo Wet etch of excess Al or Lift-off process

250 nm Al sputter depo On sacrifical resist layer Wet etch to define bridges

Fabrication: Al lift-off vs wet ecthing

Lift-off process (untill recent) Hard to reproduce Wet etch process

Result Hybrid KIDs with lift-off process

• Photon noise limited performance • High optical efficiency • But…. 1/f noise

• Use amplitude readout -> too stringent requirements digital electronics

• Reduce phase noise at low frequencies

15

See Yates et al. http://arxiv.org/abs/1107.4330

Test chip

16

• CPW resonators 3-2-3 mu wide • Change the length of the Al section • Also HW devices (without interface)

Green = substrate Blue = Al on substrate Red = Al on NbTiN White = NbTiN

1 min Ar+ RF cleaning prior to Al deposition, definition with lift-off • No effect of interfaces • Frequency noise determined soleley by Al

• Widening NbTiN section has no influence • Noise scales with Al length

• Noise spectra with much Al have 1/F below 100 Hz • F-0.7 for 100% Al on Si resonators

Test chip performance Lift- off devices

17

0 10 20 30 40 50 60 70 80 90 100 110-196

-194

-192

-190

-188

-186

-184

-182

Al fraction

SF/F

2 at

Pi=

-40d

Bm

and

1 k

Hz

Quarter waveHalf waveexpected level Gao et al8x9 array

Much wider resonators

F-0.5

F-1

At optimum power

H15 batch made with lift-off

Expected for NbTiN

Same – no interface effect

0 10 20 30 40 50 60 70 80 90 100-195

-194

-193

-192

-191

-190

-189

-188

-187

-186

-185

Al fraction

SF/F

2 at

Pi=

-40

dBm

and

1 k

Hz

[dB

k/H

z]

QWHWJiansong line

No Ar+ etch prior to Al deposition, but buffered HF dip, definition wet etch • Frequency noise only weak function of Al length • Noise level higher for pure NbTiN resonators • Hybrid frequency noise low, especially at low F: – 10 dB at 1 Hz

18

H20 batch made with lift-off

F-0.5

F-1

10

110

210

310

410

5-200

-195

-190

-185

-180

-175

-170

-165

SF/F

02

F [Hz]

80% KID 13100% KID 15100% HW KID 19NbTiN KID1

20% KID 340% KID 760% KID 9100% Opt 1

At optimum power Expected for NbTiN

40% Al H15

40% Al H20

Test chip performance wet – etch devices

19

Detector array

BP filter @ 0.3 K

LP filter 0.4 THz @ 1K

LP filter 0.6 THz @ 4K

LP filter 0.8 THz @ 77K

goretex @ 300K

System test for A-MKID test camera He7 test cryostat

100

102

104

10-15

10-14

10-13

Phase NEP at 1kHz, 1.8e-015 W/Hz1/2

f (Hz)

NE

Psy

s W

/Hz1/

2

20

100

101

10-15

10-14

10-13

f(Hz)

Pha

se S

yste

m N

EP

W/H

z1/2

System test for A-MKID test camera H20 with etching full system measurement

• 1/f significantly reduced • 1/f noise is now from the system and can be (partly) removed

Drift due to system

Array NEP Correlated noise removed

BLIP

amplitude

phase

F spacing

• 4±2 MHz dF • Q=20.000 • Only for large arrays!

21

4.8 5 5.2 5.4

-20

-15

-10

-5

0

F [GHz]

S21

[dB

]

4.93 4.94 4.95 4.96

-15

-10

-5

0

F [GHz]

S21

[dB

]

Conclusions

22

• We are developping large arrays for the A-MKID camera and NIKA • Noise properties differ from 1 layer KIDs, especially at low F • NbTiN-Al interface does not play a role

• Devices are photon noise limited at high modulation frequencies • Approach BLIP in band of operation

Results H10 with lift off measured with cryogenic BB

23

• Photon noise limite performance • 1/F from the device dominates below 10 Hz