Transcript Device Research Conference, June 19, 2012 Regrown Ohmic Contacts to InxGa1-xAs Approaching the Quantum Conductivity Limit Jeremy J.

Device Research Conference, June 19, 2012
Regrown Ohmic Contacts to
InxGa1-xAs Approaching the
Quantum Conductivity Limit
Jeremy J. M. Law,a,b Andy D . Carter,a Sanghoon
Lee,a Arthur C. Gossard,a,b and Mark J. W. Rodwella
a)
Department of Electrical and Computer Engineering, University of California, Santa Barbara
b)
Materials Department, University of California, Santa Barbara
Outline
• Motivation
• Ballistic FET Current and TLM Quantum Conductance
• Process Flow
• Sample Structures
− Regrowth TLM (RGTLM)
− Transmission Line Measurement (TLM)
• Results
− Metal-semiconductor (TLM)
− Metal-semiconductor and semiconductor-channel (RGTLM)
• Theory Comparison
• Conclusion
2
Motivation
Unmetalized
Source/Drain
Ungated Channel
Metal–Source/Drain
Source/Drain–Channel
• Two interfaces of interest
– Metal–regrowth interface
– Regrowth–channel interface
• Sheet resistance of regrowth
• Sheet resistance of ungated region
• Must ascertain contribution to overall access resistance from all
of above
3
FET Ballistic Current = TLM Quantum Conductance
• Fundamental limits to contact resistance to a two-dimensional channel?
• Quantum limited contact resistance1, 2 equivalent to ballistic transconductance
(
charge density µ Ef - Ec
(
velocity µ Ef - Ec
(
µ (V
)
4
3 2
- Ef .d - Ec
3 2
gs
1
th
current µ Ef ,d + qdV - Ec
12
) (
- V ) - (V
current µ Ef ,s - Ec
(
)
1
gd
- Vth
)
)
(
µ dV × Ef ,s - Ec
3 2
3 2
)
) - (E
3 2
12
(
)
conductivity µ ( carrier density )
µ dV × carrier density
P. M. Solomon et al., IEDM Tech. Dig., 1989, p. 405; 2 J Guo et al., IEEE Elec. Dev. Lett. 33, 525 (2012).
12
12
f ,d
- Ec
)
3 2
Regrowth TLM (RGTLM) Process Flow
• Understand source (regrowth) to channel interface
• Rudimentary process flow
• Approximates FET structure and process flow
– Independent of high-k properties
• Four-point Kelvin measurement
5
Epi growth
Dummy Pillar Definition
Regrowth
Planarization
Isolation
S/D Metalization
TLM Process Flow
• Understand metal to source (regrowth) interface
• Rudimentary process flow
• Can be done on same die as RGTLM
• Four-point Kelvin measurement
6
Epi growth
Regrowth
Isolation
S/D Metalization
Sample Structures: TLM
InAs RG on d–doped 25 nm
In0.53Ga0.47As channel
InAs RG on d–doped 15 nm InAs
channel
7
InAs RG on 100 nm n+
In0.53Ga0.47As channel
In0.53Ga0.47As → InAs RG on 100
nm n+ In0.53Ga0.47As channel
TLM Results
InAs RG on d–doped
25 nm In0.53Ga0.47As
channel
Slope: 23.8 W; Intercept/2: 2.1 W–mm
InAs RG on d–doped 15
nm InAs channel
8
Slope: 19.3 W; Intercept/2: 3.0 W–mm
InAs RG on 100 nm
n+ In0.53Ga0.47As
channel
Slope: 7.4 W; Intercept/2: 4.6 W–mm
In0.53Ga0.47As → InAs RG
on 100 nm
n+ In0.53Ga0.47As
channel
Slope: 11.3 W; Intercept/2: 3.0 W–mm
Sample Structures: RGTLM
9
InAs RG on d–doped 25 nm
In0.53Ga0.47As channel
InAs RG on 100 nm n+
In0.53Ga0.47As channel
InAs RG on d–doped 15 nm InAs
channel
In0.53Ga0.47As → InAs RG on 100
nm n+ In0.53Ga0.47As channel
Regrowth TLM Results
InAs RG on d–doped
25 nm In0.53Ga0.47As
channel
Slope: 540 W; Intercept/2: 120.8 W–mm
InAs RG on d–
doped 15 nm InAs
channel
10
Slope: 269 W; Intercept/2: 68.2 W–mm
InAs RG on 100 nm
n+ In0.53Ga0.47As
channel
Slope: 32 W; Intercept/2: 55.6 W–mm
In0.53Ga0.47As → InAs
RG on 100 nm n+
In0.53Ga0.47As
channel
Slope: 15 W; Intercept/2: 12.7 W–mm
Results Summary
• Contact resistance to thin channels (small ns) limited by quantum conductance
• Low contact resistance of 12.7 W–mm (11.1 W–mm2)
• Contact resistance low ns channels 136.4 W–mm close to theoretical 80 W–mm
N+ Regrowth
Composition
InAs
InAs
InAs
In0.53Ga0.47As → InAs
Thickness
60 nm
60 nm
60 nm
60 nm
Doping
5-10×1019 cm-3 5-10×1019 cm-3 5-10×1019 cm-3 5-10×1019 cm-3
Sheet Resistivity
23.8 W
7.4 W
19.3 W
11.3 W
Composition
In0.53Ga0.47As
In0.53Ga0.47As
InAs
In0.53Ga0.47As
Thickness
25 nm
100 nm
15 nm
100 nm
Doping
9×1012 cm-2
3-5×1019 cm-3
9×1012 cm-2
3-5×1019 cm-3
Sheet Resistivity
540 W
32 W
269 W
15 W
120.8 W-mm
55.6 W-mm
68.2 W-mm
12.7 W-mm
4.6 W-mm
1.5 W-mm2
3.0 W-mm
0.4 W-mm2
3.0 W-mm
0.8 W-mm2
Channel
Access Resistivity
Metal/Regrowth Contact Resistivity 2.1 W-mm
0.2 W-mm2
Conclusion
• Ballistic FET current equivalent to quantum conductance of TLM
• Should not add to FET contact resistance
• Material independent, i.e. true for all semiconductor materials
• Metal–regrowth contact resistance is small portion of overall Rc
– ~ 3.0 W–mm (1.0 W–mm2)
• Regrown ohmic contacts (136 W–mm) within a factor of 2 of
theoretical 80 W–mm
• 12.7 W–mm (11.1 W–mm2) is true measure of interface properties
– This includes regrowth to channel and metal to regrowth
12
Backup slides
MBE Regrowth by Migration Enhance Epitaxy (MEE)
InAs Quasi MEE
In, As, and Si shutters open
As shutter open
InGaAs Quasi MEE
In, Ga, As, and Si shutters open
14
As shutter open
MBE Regrowth: Close to 2-D Quantum conductivity Limit:
Unidirecti onal 2D density of states : cdos,1  q 2 gm* / 2 2
Charge density in left - moving states :  s1  cdos,1V f 1
Red States : charge moving in  x direction; left to right
Leftward - moving Fermi Velocity : E f 1  qV f 1  m*v 2f 1 / 2  v f 1  2qV f 1 / m*
Blue States : charge moving in - x direction; right to left
Mean leftward electron v elocity : v1  (4 / 3 )v f 1  (4 / 3 )  2qV f 1 / m
Energies taken relative to conduction band minimum in 2 - D channel.
*
Leftward current : J1   s1 v1  cdos,1V f 1 (4 / 3 )  2qV f 1 / m*

 4  2q
Total current : J  cdos,1    
 V f31/ 2 - V f32/ 2
*
3

m
 

 4  2q 3 1 / 2
Conductivity G  J / V f  cdos,1    
 V f
*
 3  m 2
q 2 21/ 2 1/ 2
Gvalley   3 / 2  ns ,valley including spin degeneracy.
 
Total conductivi ty found by summing over valle ys and vertical eigenstates
UCSB regrowth resistance measuremen ts are being limited by this effect
15