InSb-Based Heterostructures for Electronic Device Applications

Michael Santos Homer L. Dodge Department of Physics & Astronomy

University of OklahomaNorman, OK 73019 USA

Funded by NSF DMR-0808086 and DMR-0520550

Collaborators at OklahomaMBE Growth and Characterization: Madhavie Edirisooriya, Tetsuya

Mishima, Chomani Gaspe, Mukul Debnath

Transmission Electron Microscopy: Tetsuya Mishima

Cyclotron Resonance: James Coker, Ryan Doezema

Ballistic Transport: Ruwan Dedigama, Sheena Murphy

Bandgap versus Lattice Constant for III-V Materials

electron m* g-factor E(k)GaAs 0.067mo -0.5 least non-parabolic

InAs 0.023mo -15 more non-parabolic

InSb 0.014mo -51 most non-parabolic

R. Chau, S. Datta, A. Mujumdar, Technical Digest, CSICS 2005, pp. 17-20.

QW Bandgap3.4 eV1.4 eV

1.1-1.23 eV0.8 eV

0.55-0.75 eV0.36/0.8 eV

0.36 eV0.18 eV

Achievements• Fabricated by Intel from InSb/AlxIn1-xSb

material grown by QinetiQ• For ultra high speed, very low power

digital logic applications• Takes advantage of high mobility and

high saturation velocity for electrons in InSb quantum wells

InSb Quantum Well Field-Effect Transistors

Challenges• Stable and reliable gate dielectrics

for III-V semiconductors• Integration of III-V devices on Si

substrates• p-channel III-V FET for CMOS

configurations• New logic/memory circuits and

architecture (?)

Al0.10In0.90Sb

Al0.10In0.90Sb

Al0.20In0.80Sb

Al0.20In0.80Sb

Al0.20In0.80Sb

InSb well

Buffer Layers

GaAs (001) substrate

30 nm

20 nmSi δ-doping

Si δ-doping

InSb Quantum Wells for Mesoscopic Applicationsδ-doping

24 meV

0 meV

83 meV72 meV

δ-doping185 meV

141 meV

• 25 nm well • Al0.15In0.85Sb barriers

10 1000.0

4.0x104

8.0x104

1.2x105

1.6x105

4.0x1011

8.0x1011

1.2x1012

1.6x1012

Elec

tron

Mob

ility

(cm

2 /Vs)

Temperature (K)

Ele

ctro

n de

nsity

(/cm

2 )

Grown by Molecular Beam Epitaxy

Outline1. Introduction

2. Transmission Electron Microscopy• Effect of Crystalline Defects on Electron Mobility• Effect of Buffer Layer on Defect Densities

3. Transport Experiments on 2D Electron Systems• Quantum Hall Effect• Quantized Conductance in Wires• Magnetic Focusing

4. Realization of 2D Hole Systems• Transport properties• Cyclotron resonance

5. Summary

Defect Evaluation by Cross-sectional TEM (X-TEM)

54.7°

(111) glide plane

1μm

· · · Matrix spots· · · Twin spots

000

002

220 180°

Diffraction pattern

InSb

9%AlInSb

9%AlInSb

GaAs(001)

Threading dislocation (TD)

QW

SLS

Micro-twin (MT)

220 Dark-field (DF) X-TEM

T.D. Mishima, M.B. Santos et al., J. Cryst. Growth 251, 551 (2003).

AlxIn1-xSb (x ×100)%AlInSb

Formation of Misfit Dislocations

360nm18%AlInSbInterlayer

9%AlInSb

9%AlInSb

31°

Front view

Misfit dislocations

200nmSide view

Threading dislocations can change into a misfit dislocation at interfaces.

1µmFront view

200nm InSbInterlayer

AlSb

9%AlInSbmatrix

9%AlInSbmatrix

InSb QW

GaAs(001)

Lattice mismatchwith 9%AlInSb

InSb

18%AlInSb

+ 0.5%

− 0.5%

T.D. Mishima, M. Edirisooriya, and M.B. Santos, Appl. Phys. Lett. 88, 191908 (2006).T.D. Mishima and M.B. Santos, J. Vac. Sci. Technol. B22, 1472 (2004).

Dislocation Filtering Mechanism

[1] J.W. Matthews and A.E. Blakeslee, J. Cryst. Growth 27, 118 (1974).[2] N.A. El-Masry et al., J. Appl. Phys. 64, 3672 (1988): Observed in strained-layer superlattice (SLS).[3] T.D. Mishima et al., to be published.

50nm

24%AlInSb

12%AlInSb

Dislocation filtering [2]

Full bending

Looping

Reaction

220 DF X-TEM [3]

Lateral movement of TDs→ Defect filtering

Pre-existingthreading dislocation

Matrix(Substrate)

Interlayer(Epi-layer) hc

h < hc h = hc h > hc

Critical thickness Misfit dislocationBending of pre-existing threading dislocations [1]

Effect of Interlayers on Threading Dislocations (TDs)

10 2 3Thickness (µm)

TD d

ensi

ty (/

cm2 )

5×108

1×109

2×109

5×109

1×1010

1st

2nd

3rd

GaAs

12%AlInSb

GaAs

12%AlInSb

240nm160nm 1st

2nd

3rd

75%

50%

Triple-interlayersample Non-interlayer

samples

Total 1.6µm

24%AlInSb12%AlInSb

GaAs

400nm

T.D. Mishima et al., Appl. Phys. Lett. 88, 191908 (2006).

Non-interlayerInterlayer

AlxIn1-xSb/AlyIn1-ySb interlayers can be used to improve the performance of any InSb- (InAsSb)-based devices

in which AlxIn1-xSb is used as a buffer, insulating, or barrier layer material.

Effect of Micro-twin on InSb Quantum Well

wθθ

θm ×+

=cos2sin

sin3

50nm

QW

< ~ 2°

~14 nm0.16 nm 2°

20nm

m

9%AlInSb

mL’

L9%AlInSb

w

20nm

InSbQW

wθθ

θL ×+

=cos2sin

cos3'wθθ

L ×+

=cos2sin

3

θ=15.8°(001)

InSbQW

InSbQW

θ

[1] T.D. Mishima et al., J. Cryst. Growth 251, 551 (2003).[2] T.D. Mishima et al., Phys. Stat. Sol. (c) 5, 2775 (2008).[3] T.D. Mishima et al., J. Vac. Sci. Technol. B23, 1171 (2005).

m w

ww L’L

Micro-twin

002 DF X-TEMStep Terrace

2°-vicinal interface

Bending of QW

HR-TEM

002 DF X-TEM [1] Geometrical relation [2]

MT

002MT DF X-TEM [3]

MT MT

35°-tilted 002 DF X-TEM [2]

Electron Mobility and Structural Defects

4×104

3×104

2×104

0

3.6×109

0

1.2×1046×103

1.8×109

10 1000.0

4.0x104

8.0x104

1.2x105

1.6x105

4.0x1011

8.0x1011

1.2x1012

1.6x1012

Elec

tron

Mob

ility

(cm

2 /Vs)

Temperature (K)

Ele

ctro

n de

nsity

(/cm

2 )

Hal

l Mob

ility

(cm

2 /Vs)

Car

rier D

ensi

ty(/c

m2 )

T.D. Mishima et al., Appl. Phys. Lett. 91, 062106 (2007).

Electron Mobility in InSb QW @RT30,000 41,000 cm2/Vs (37%↑)

(n = 5.5×1011/cm2)

Temperature (K)

MT density980 /cm

7,600 /cmOn-axis GaAs(001)[ tQW = 20nm ]

2° off-cut GaAs(001)[ tQW = 20nm ]

2° off-cut GaAs(001)[ tQW = 25nm ]

Summary of Defect Studies

• Mobility partially limited by threading dislocations and micro-twin defects

• Threading dislocations reduced by interlayers in the buffer layer

• Micro-twin density reduced by growth on off-axis substrates

Outline1. Introduction

2. Transmission Electron Microscopy• Effect of Crystalline Defects on Electron Mobility• Effect of Buffer Layer on Defect Densities

3. Transport Experiments on 2D Electron Systems• Quantum Hall Effect• Quantized Conductance in Wires• Magnetic Focusing

4. Realization of 2D Hole Systems• Transport properties• Cyclotron resonance

5. Summary

0.0 0.5 1.0 1.5 2.0

Rxx

( Ω)

B (T)0 5 10 15

Rxy

(h/e

2 )

Rxx

(kΩ

)

0

1

2

ν=1

ν=2

ν=2/3

0

1

(b)

50

0

T = 1.4K(a)ν=5

ν=9

Quantum Hall Effect in InSb

( )

me

meB

BgnEE

c

Bcini

2

B*

*2

12

1,

==

±++=

⊥ µω

µω

quantized orbits spin splittingsubband energy

• Strong Integer QHE with Zeeman splitting resolved at low B

• Landau and Zeeman splittings equal at tilt of 65° for ν=2 (m*g=0.84)

• Fractional QHE not observed

Vg (V)

-0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8G

(e2 /h

)

0.0T0.8T1.7T2.1T2.8T3.4T4.2T5.6T7.0T

0

0

0

0

012

23

4

54

6

6

6

10

2 2

Quantum Point Contacts

Increase B Increase confinementLevel depopulation

( )

*22

020

2*

21

22

21

,

,,

2

meBmM

BgMk

nEE

cc

By

mnm

=+==

±+++=

ωωωωωω

µω

VS,D

D

G G

S

IS,D

VGVG

Spin Orbit Effects

η αGaAs 27.6 eV A3 5.2 e A2

InAs 27.2 eV A3 117 e A2

InSb 760 eV A3 523 e A2

R. Winkler, Springer Tracts in Modern Physics 191 (2004)

Bulk Inversion Asymmetry (Dresselhaus splitting)

Structural Inversion Asymmetry(Rashba splitting)

“Transverse electron focusing in systems with spin orbit coupling,” Usaj and Balseiro, PRB 70, 041301 (2004)

“Spin separation in cyclotron motion,” Rokhinson et al., PRB 93, 146601 (2004)

“Spin-polarized reflection of electrons in a two-dimensional electron system,” Chen et al., APL 86, 032113 (2005)

//0

2//

2

2kE

mkE zα±=

( ) ( )

+−±= 5

//23

//2

2//

2

2sin21

2 //kOkkk

mkE z φη

Spin-dependent trajectories expected:

Deduced values:Dresselhaus parameter, η=7.6×10-28 eV mRashba parameter, α=1.0×10-11eV m

Magnetic Focusing Data

2 3 4 612

3

5

2 3 4 612

3

5

injector

gategate

collector

Vc

Iinj

-

+

Vg Vg

--

++injector

gategate

collector

Vc

Iinj

-

+

Vg Vg

--

+injector

gategate

collector

Vc

Iinj

-

+

Vg Vg

--

++

• Shubnikov de Haas oscillations ns=3.62x1011cm-2

• Measured period of ≈0.2T implies L≈1.0µm

rc Brcrc Bsc

fc n

eBv

r πω

2==

A.R. Dedigama et al., Physica E 34, 647 (2006).

Summary of Ballistic Transport Experiments

• Ballistic transport observed at T~200K in 0.5µm devices

• Quantized conductance observed in point contacts

• Magnetic focusing features observed

• Goal: Spin-dependent transport devices

Outline1. Introduction

2. Transmission Electron Microscopy• Effect of Crystalline Defects on Electron Mobility• Effect of Buffer Layer on Defect Densities

3. Transport Experiments on 2D Electron Systems• Quantum Hall Effect• Quantized Conductance in Wires• Magnetic Focusing

4. Realization of 2D Hole Systems• Transport properties• Cyclotron resonance

5. Summary

heavy

light

(b)

(a)

(in plane)

The Physics of Low-Dimensional Electron Systems by John H. Davies (1998).

• In-plane hole masses are anisotropic• Degeneracy of “light” and “heavy” holes lifted by strain and confinement• “Light” holes have heavy in-plane mass, “heavy” holes have light in-plane mass• Anticrossing expected between “light” and “heavy” hole bands

Bulk

QuantumWell

Two-Dimensional Hole Systems in III-V Materials

p-FET

n-FET

p-type InSb Quantum Well

µH at 300K Reference

In0.2Ga0.8As 260 cm2/Vs R.T. Hsu et al., Appl. Phys. Lett. 66, 2864 (1995).

In0.53Ga0.48As 265 cm2/Vs Y-J. Chen and D. Pavlidis, IEEE Trans. Elec. Dev. 39, 466 (1992).

In0.82Ga0.18As 295 cm2/Vs A.M. Kusters et al., IEEE Trans. Elec. Dev. 40, 2164 (1993).

InSb 700 cm2/Vs M. Edirisooriya et al., J. Crystal Growth (in press).

In0.4Ga0.6Sb 1500 cm2/Vs B.R. Bennett et al., Appl. Phys. Lett. 91, 042104 (2007).

Ge 3100 cm2/Vs M. Myronov, Appl. Phys. Lett. 91, 082108 (2007).

Al0.10In0.90Sb

Al0.10In0.90Sb

Al0.20In0.80Sb

Al0.20In0.80Sb

Al0.20In0.80Sb

15nm InSb well

Buffer Layer

GaAs (001) substrate

30 nm

20 nm

Be δ-doping

Be δ-doping

First realization of remotely-doped p-type InSb QWs.

t1964.2K

p=2.2x1011cm-2

B (T)

0 2 4 6 8R

xy (k

Ω)

0

5

10

15

20

25

30

0

0.4

0.6

0.8

1.0

0.2

Rxx (kΩ

)

1.2

ν=2

34

7

p-type InSb Quantum Well at Low Temperature

Mobility in p-type QW is ~3 (~55) times smaller at 25K (300K) than in n-type QW with same layer structure.

Temperature (K)

0 50 100 150 200 250 300

Mob

ility

(cm

2 /V. s

)

1000

10000

100000DensityMobility

1

0.1

10

100

Density (10

11cm-2)

OU Wafer t1942D hole system

µh= 700 cm2/Vs @300K (T250)µh= 29,900 cm2/Vs @77K (T241)µh= 55,600 cm2/Vs @20K (T241)

5

B (T)3 4 5

k (c

m-1

)

150

200

250

InSb phonons

ν=2ν=3

S499n=2.2x1011cm-2

n=0

n=1n=2

B (T)0 5 10 15

E (m

eV)

0

50

100

Spin Resolved Cyclotron Resonance in n-type InSb QW

S842n=4.2x1011cm-2

( )

me

meB

BgnEE

c

Bcini

2

B*

*2

12

1,

==

±++=

⊥ µω

µω

quantized orbits spin splittingsubband energy

ω (meV)

55 60 65

T(B

)/T(0

)

0.8

1.0

B=13.25T

Modeling of Landau Levels in InSb QW

B (T)

0 2 4 6 8

E (e

V)

-0.04

-0.02

0.00

0.02

H1

H2

H3

L1

• Narrow Gap ⇒ conduction band/valence band mixing ⇒8 band k•p effective mass approximation (Pidgeon-Brown + kz + confinement).

• Superlattice/MQW effects using Finite Difference method.• Band gap discontinuities & strain effects included.• Magneto-Optics using Fermi’s Golden Rule.• Material Parameters from experimental measurements.

B (T)

0 2 4 6 8

E (e

V)

0.30

0.35

0.40

C1

C2

t212b (p=3.5x1011cm-2)

Frequency (cm-1)

20 40 60 80 100 120

T(B

)/T(0

)

1.0

1.5

2.0

2.5

3.0

2.0T

7.0T

6.0T

5.0T

4.0T

3.0T

6.5T

5.5T

4.5T

3.5T

2.5T

t212b (p=3.5x1011cm-2)

Frequency (cm-1)

100 150 200 250

T(B

)/T(0

)

1.0

1.2

1.4

1.6

1.8

2.0

InSb

pho

nons

GaA

s ph

onon

s

7.0T

8.0T

9.0T

10.0T

11.0T

12.0T

13.0T

14.0T

9.5T

10.5T

11.5T

Spin Resolved Cyclotron Resonance in p-type InSb QW

p~3.5x1011cm-2

CR across wide range of B

t212p=3.5x1011cm-2

B (T)

0 2 4 6 8 10 12 14

Ene

rgy

(eV

)

0.00

0.01

0.02

0.03 H10d-H11d H10u-H11u H11d-H12d H11u-H12u

t212p=3.5x1011cm-2

B (T)

0 2 4 6 8 10 12 14

Ene

rgy

(eV

)

0.00

0.01

0.02

0.03 H10d-H11d H10u-H11u H11d-H12d H11u-H12u

CR Positions (experiment)

• Good agreement with theory…

• …except for one branch.

CR in p-type InSb Quantum Well

CR Positions (theory)

t212p=3.5x1011cm-2

B (T)

0 2 4 6 8 10 12 14

Ene

rgy

(eV

)

-0.03

-0.02

-0.01

0.00

0.01

0.02

0.03

EF

H10d H10u H11d H11u H12d H12u H13d H13u

Fan Diagram

B (T)

0 2 4 6 8 10 12 14 16 18

ν (c

m-1

)

0

50

100

150

200

250

300

0.05mo

0.10mo

t196bt196b

t196bp~2x1011cm-2

B (T)

0 2 4 6 8 10 12 14 16 18

ν (c

m-1

)

0

50

100

150

200

250

300

0.05mo

0.10mo

t194ct194c

t194cp~3x1011cm-2

B (T)

0 2 4 6 8 10 12 14 16 18

ν (c

m-1

)

0

50

100

150

200

250

300

0.05mo

0.10mo

t198bt198bt198b

t198bp~5x1011cm-2

Dependence on Hole Density

Effective mass increases with increasing hole density.

SummaryMolecular Beam Epitaxy of InSb-based heterostructures• Mobility partially limited by micro-twins and dislocations• Defect densities depends on buffer layer composition• Room temperature mobility as high as 41,000 cm2/Vs

Electron transport through InSb mesoscopic structures• Ballistic transport observed at T~200K in 0.5µm devices • Quantized conductance observed in point contacts• Magnetic focusing features observed

2D Hole systems in InSb quantum wells• Mobility lower than for 2D electron systems• Cyclotron resonance shows m*≥0.04me• Good prototype system for p-type III-V quantum wells