Quantum engineered field-effect transistors for warm, wide-bandwidth,

near quantum-limitedTerahertz heterodyne receivers

Mark Sherwin

UCSB

Physics Department

and

Institute for Terahertz Science and Technology

Herschel Space Observatory

Bilpratt et. al., A&A 518 L1 (2010)

Heterodyne Instrument for Far Infrared (HIFI)

Th. De Grauw, et. al., A&A 518 L6 (2010)

HIFI spectrum of Orion hot core

Th. De Grauw, et. al., A&A 518 L6 (2010)

35Cl/37Cl ratio in dense molecular clouds

J. Cernicharo et. al., A&A L115 (2010)

H2O in cometary atmosphere

3x105 km

P. Hartogh et. al., A&A 518, L150 (2010)

Coherent detection (e. g., heterodyne mixing)

Frequency

Sig

nal

fLO= 1 THz to 5 THz

fsigfIF = 1 - 10 GHz

• Receiver Noise:

Top = TA+ TM + TAmp / M

• Integration Time:

(TR)2

Mixer

TM

M

PLO

fLO

TAmpfIF

IF AmpBackend

SpectrometerfIF = | fLO - fsig|

HETERODYNE RECEIVERf / f = 107 - 108

Quantum Limit

(double sideband)TM≥hf/2kB

TQ= 25K @ 1 THz

TA

DSB system noise temperature on HIFI

Th. De Grauw, et. al., A&A 518 L6 (2010) Operating temperature 1.7K

Heterodyne receivers for future missions

• Higher frequency (HD @ 2.7 THz, OI@4.7 THz)

• Lower noise

• High temperature operation (40-100K)

• Wider bandwidth (>15 GHz vs. 4 GHz for phonon-cooled HEB)

• Low LO power

• Arrays

Superconducting hot-electron bolometers

Superconducting hot-electron bolometers

Theory, proposed mixers

Proposed mixers

• Mixer noise temperature (classical theory): 200K DSB

• Operating temperature: Top=30-100K

• LO frequency: 1.5-5 THz

• IF bandwidth: >15 GHz

• LO power requirement: – ~1 µW@ Top=30K

– ~100µW@ Top =100K

• Planar, suitable for arrays

Quantum-engineered FET as hot-electron bolometer

SourceDrain

IF

THz

Front gate (to antenna)

Back gate (to antenna)

absorberdetector

“Tunable antenna-coupled intersubband terahertz (TACIT) sensor”

Relevant physics in quantum wells

• Absorption: intersubband transition in quantum well

• Detection: temperature-dependent mobility of 2-D electron gas

IF

THz

absorber

IF

THzdetector

Example: intersubband absorption in square quantum well

Intersubband absorption in square well

0 20 40 60 80Position (nm)

Ene

rgy

(meV

)250

150

50

0

100

200

Al0.3Ga0.7AsGaAs

in-plane wavevector k

Ene

rgy

Absorption vs. dc electric field, constant Ns

2.4 THz 4.8 THz

Absorption peak vs. dc electric field, charge density

Experiment:Williams et. al., PRL, 2001)

Theory:Ullrich and Vignale Ibid.

ns=1010 cm-2

ns=13x1010 cm-2

Time-dependent Local density approximationAll parameters from experiment

Frequency (T

Hz)2.4

4.8

Intersubband absorption below 1 THz

Detection: temperature-dependent mobility

IF

THzdetection

• Mobility of 2-D electron gases vs. temperature

• Mobility determined by electron temperature Te.

Detection

• Hot-electron bolometric

IF

THzdetection

1R||

dR||

dT

⎛

⎝ ⎜

⎞

⎠ ⎟ ≡γ =0.01−0.03K−1@50K

ρ=dVdP

=IR||γT1

CV

=IR||γT1

1NSAkB

Speed

• Phonon cooling

• Diffusion cooling– 20 GHz IF bandwidth demonstrated in 4µm long millimeter-wave mixer. (M.

Lee et. al., APL 2001)

J. N. Heyman et. al., PRL 74, 2682 (1995)

109

1010

1011

0 0.04 0.08 0.12

1/T (K-1)

50 25 10

T (K)IF

bandwidth (G

Hz)

0.16

1.6

16

Coupling efficiency: impedance of active region

LV(t)

Area ACharge Q+ + + + + + + + + + + + + + + + + + + + +

- - - - - - - - - - - - - - - - - - - - - - - - - - - - -2-DEG=

Dipole sheeton springs

On resonance

€

C =ε 0εA

L

R =q2

ε 2ε 02m *

Ns f12n(T)

2πf( )2A2Γ

R can be 20 in future designs

C can be tuned out byrf embedding circuit design

Source-Load coupling can be>90%!

The math

€

Z ω( ) =Vω

Iω

€

Z(ω*) =Vω*

Iω*

=L

iω*εε0A+

NSe2 f12n(T )ε 2ε0

2Am *ω*2 2Γ=

1iω*C

+ Rzz

€

= 1iωεε0A

L −χ 2D ω( )

εε0

⎛

⎝ ⎜

⎞

⎠ ⎟€

Vω = EωL − P2D /εε0€

χ2D ω( ) =NSe2 f12n(T )

m *1

ω*2 −ω2 + i2ωΓ

€

P2D(ω) = χ 2D(ω)Eω

M. Sherwin et. al., Proceedings of 2002 Monterey Submm workshop

LV(t)

Area ACharge Q+ + + + + + + + + + + + + + + + + + + + +

- - - - - - - - - - - - - - - - - - - - - - - - - - - - -

Can couple efficiently to ~10,000 electrons!

LV(t)

Area ACharge Q+ + + + + + + + + + + + + + + + + + + + +

- - - - - - - - - - - - - - - - - - - - - - - - - - - - -2-DEG=

Dipole sheeton springs

On resonance

49 Ohms*

18 Ohms*

*on resonance for high-quality 40 nm square well, Te=50 K, f=2.3 THz, A=5µm2,L=0.15 µm

Theory for NEP, TM

• NEP (direct detector)

– =insertion loss, =1/R(dR/dT), NS= sheet density, A=active region area

• Double sideband noise temperature (mixer)€

NEP = α −12kB

NS A

τ

4

γ 2+ Te

2

€

TM = α −1 4

γ 2Te

+ Te

⎛

⎝ ⎜

⎞

⎠ ⎟

M. Sherwin et. al., Proceedings of 2002 Monterey Submm workshop

Johnson noise Thermal conductionnoise

Performance limits for TACIT mixer

10,000 electrons106 cm2V-s @ low temp.Bulk LO phonon scattering

First-generation devices

• Twin slot coupled to coplanar waveguide

• Demonstrated direct detection, electric field tuning

• Difficult fabrication with very low yield

• Quantum well far from optimal

• Microwave embedding circuit far from optimal

• Collaboration with W. R. McGrath, Paolo Focardi

Completed TACIT detector

Source

Drain

Front gate bias line

Back gate bias line

10nm

130

nm

10.8 nm

197

nm

500

nm

1000

nm

200

nm

70 n

m9 nm

70 n

m

3 nm

0.5

mm

subs

trat

e

buff

er

etch

sto

p

cap Si d

elta

Dop

ing

(101

2 c

m-2)

Coupled quantum wellsba

rrie

rs

cap

GaAs Al0.3Ga0.7As

Sample structure design and growth

Intersubband absorption characterization

THz in (FTIR) THz out

bolometer

1.6 THzDesign frequency

theory

Electrons in quantum well

Fabrication of 1st generationTACIT detector

Epoxy Bond and Stop-Etch (EBASE), Sandia

1. Process front side

2. Epoxy bond to GaAs wafer

3. Etch away substrate

4. Process back side, etch vias

Experimental Set Up

IF

THz

GaAs Host Substrate

Silicon LensUCSB Free Electron Laser at 40-80 cm-1

Sample in a Helium Flow Cryostat, T=20-100K

Current amplifierSources and measures channel

oscilloscope

0 V

Gate Voltage

0

5

10

15

-5 0 5 10 15

Signal /mV

Time/ microseconds

5s, 1mW

0

10

20

30

40

50

-5 0 5 10 15

Signal /mV

Time/ microseconds

Response to fast THz pulses

1.5 ns 3.5 ns

Tunability with gate voltage

300

400

500

600

700

800

900

1000

0.25

0.26

0.27

0.28

0.29

0.3

0.31

0.32

0.33-1 -0.5 0 0.5 1 1.5

signal/ nanoAmps

signal/milliVolts

front gate - back gate Voltage

Sample ATb=100KTe~120Kf=1.53 THzFEL

Sample BTb=77KTe~100Kf=1.6 THzMoleculargas laser

Pho

tocu

rren

t (n

A)

Photovoltage (m

V)

Charge density held constant.

Origin of “double peak”

2000 2100 2200 2300 2400 2500-50

0

50

100

150

200

250

300field = -0.2 mV/Angstrom

Energy / meV

2000 2100 2200 2300 2400 2500

field = 0.2 mV/Angstrom

1.4

1.5

1.6

1.7

1.8

1.9

2

2.1

45

50

55

60

65

70

Intersubband Resonance

Frequency /THz

Intersubband Resonance

Frequency /cm

-1

-0.2 -0.1 0 0.1 0.2

Electric field (mV/Angstrom)

Inte

rsub

band

ab

sorp

tion

fre

quen

cy (

TH

z)

New design*

• Easier fabrication• Eliminate some parasitics• Can rapidly iterate to optimize

Inspired by Chris McKenney’s thesis, Cleland group.

TACIT mixer development strategy

• Microwave engineering• MBE growth of wafers with high mobility,

narrow intersubband linewidth.• Develop fabrication process• Test and iterate

40

TACIT specs to enable new missions

• High-temperature operation• Wide bandwidth• High frequency• Low noise

41

Herschel cryogenics

• Herschel– Large dewar drives up mission costs– Duration limited to 3.5 years

42

Long-duration mission with cryocooler

• Atmospheric Infrared Sounder (AIRS)– Detectors @ 58K– Long-lifetime cryocooler– Launched 2002

43

Potential platforms for TACIT mixers

• Explorer-class missions– Astrophysics– Planetary science

• Long-duration ballooning• SOFIA

44

Summary and conclusions

• THz heterodyne spectroscopy: important science• TACIT mixers offer improvements over state of art

• Timely to develop TACIT mixers in concert with new mission concepts

45

Superconducting HEB actual

TACIT mixer theory

Operating temp. 2K 30-100K

Bandwidth 4 GHz >15 GHz

Noise temperature @ 2.5 THz

1000K 200K

Acknowledgments

• W. R. McGrath (JPL): Antenna design• P. Focardi (JPL): Antenna impedance, mode matching

theory• G. B. Serapiglia (UCSB->law school): processing,

characterization, experiments• Sangwoo Kim (UCSB-> Tanner Research Labs): room-

temperature devices.• M. Hanson (UCSB): sample growth• A. C. Gossard (UCSB): sample growthFunding: NASA, NSF

46

Insertion losses for this device and experiment

• Matching antenna pattern– Only 4% coupled into antenna mode*– Device fried before could be improved– Mode matching can be increased to >90%+

Antennapattern

Illumination pattern

*Computed following Goldsmith, “Quasioptical systems”

+Focardi, McGrath and Neto, IEEE MTT 2004.

Zload=(1.7-i39)

Pin

€

PLoad

PIn

=1−ZS − ZL

*

ZS + ZL

2Zsource=(20-i40)

• Matching antenna and load impedances- Only 2.5% coupled from antenna into load

Responsivity at Tbath=80K

• Roptical=V/Pin=0.1 V/W

• Rinternal=Roptical/(total insertion loss)=107±75 V/W

• Relectrical

• Theory: Rinternal =(1100±400) V/W€

Relectrical =

V

I−

dV

dI

⎛

⎝ ⎜

⎞

⎠ ⎟

2V= (400 ±150)V /W

Electrical transport data

This sample

106 mobility2-DEG

Electrical transport data