Device Research Conference 2011
1.0-THz fmax InP DHBTs in a refractory
emitter and self-aligned base process for
reduced base access resistance
Vibhor Jain, Johann C. Rode, Han-Wei Chiang, Ashish Baraskar,
Evan Lobisser, Brian J Thibeault, Mark Rodwell
ECE Department, University of California, Santa Barbara, CA 93106-9560
Miguel Urteaga
Teledyne Scientific & Imaging, Thousand Oaks, CA 91360
D Loubychev, A Snyder, Y Wu, J M Fastenau, W K Liu
IQE Inc., 119 Technology Drive, Bethlehem, PA 18015
vibhor@ece.ucsb.edu, 805-893-3273
1
Outline
• Need for high speed HBTs
• Fabrication
– Challenges
– Process Development
• DHBT
– Epitaxial Design
– Results
• Summary
2
Why THz Transistors?
Transistor Power Gain (dB)
40
Digital Logic for
Optical fiber circuits
35
30
25
THz amplifiers for
imaging, sensing,
communications
20
15
10
5
0 9
10
10
10
11
10
Freq (Hz)
12
10
High gain at microwave frequencies  precision analog design,
high resolution ADC & DAC, high performance receivers
3
HBT process requirements
• Refractory emitter contact and metal stack
– To sustain high current density operation
• Low stress emitters
– For high yield
• Low base access resistance
– For improved device fmax
• Thin emitter semiconductor
– To enable a wet etched emitter process for reliability and scalability
4
Fabrication Challenges – Stable refractory emitters
Emitter yield drops during base contact, subsequent lift-off steps
Fallen emitters
High stress in emitter metal stack
Poor metal adhesion to InGaAs
Need for low stress, high yield emitters
5
Fabrication Challenges – Base-Emitter Short
Undercut in thick emitter semiconductor
Helps in Self Aligned Base Liftoff
InP Wet Etch
Slow etch plane
Fast etch plane
For controlled semiconductor undercut
 Thin semiconductor
To prevent base – emitter short
 Vertical emitter profile and line of sight metal deposition
 Shadowing effect due to high emitter aspect ratio
6
Fabrication Challenges – Base Access Resistance
f max
f

8RbbCcb
We
Wgap  contact
We
Wbc
Rbb   sh,e 
  sh,bc 
  sh,gap 

12 Le
6 Le
2 Le Acontacts
Wbc
 sh,g ap   sh,e ,  sh,b c
• Surface Depletion
• Process Damage
 Need for very small Wgap
Wgap
• Small undercut in InP emitter
• Self-aligned base contact
7
Composite Emitter Metal Stack
W emitter
TiW
W
Erik Lind
TiW emitter
Evan Lobisser
• W/TiW metal stack
• Low stress
• Refractory metal emitters
• Vertical dry etch profile
8
FIB/TEM by E Lobisser
Vertical Emitter
Base
Metal
SiNx
TiW
Vertical etch profile
BCB
100nm
W
Low stress
High emitter yield
Scalable emitter process
9
FIB/TEM by E Lobisser
Narrow Emitter Undercut
Dual SiN
sidewall
Controlled
InP undercut
Mo contact
InGaAs cap
InP emitter
50nm
Narrow BE gap
10
Epitaxial Design
Material
Doping (cm-3)
Description
10
In0.53Ga0.47As
81019 : Si
Emitter Cap
20
InP
51019 : Si
Emitter
15
InP
21018 : Si
Emitter
30
InGaAs
9-51019 : C
Base
13.5
In0.53Ga0.47 As
51016 : Si
Setback
16.5
InGaAs / InAlAs
51016 : Si
B-C Grade
3
InP
3.6 1018 : Si
Pulse doping
67
InP
51016 : Si
Collector
7.5
InP
11019 : Si
Sub Collector
5
In0.53Ga0.47 As
41019 : Si
Sub Collector
300
InP
21019 : Si
Sub Collector
Substrate
SI : InP
1.5
1
0.5
Energy (eV)
T(nm)
0
-0.5
-1
-1.5
Emitter
Base
-2
Collector
-2.5
0
50
100
Distance (nm)
150
200
Vbe = 1 V, Vcb = 0.7 V, Je = 24 mA/mm2
Thin emitter semiconductor
 Enables wet etching
11
Results - DC Measurements
Common emitter I-V
15
I
b,step
@Peak f,fmax
Je = 20.4 mA/mm2
P = 33.5 mW/mm2
= 200 mA
10
e
2
J (mA/mm )
2
P = 20 mW/mm
30
2
P = 30 mW/mm
25
2
A = 0.22 x 2.7 mm
je
20
5
2
V
BVceo = 3.7 V @ Je = 0.1
β = 17
mA/cm2
4
5
10
10
-5
Solid line: V
Dashed: V
cb
cb
= 0.7V
20
= 0V
15
n = 1.19
c
10
n = 1.87
10
-7
b
10
-9
I
b

5
I
c
0
0.2 0.4 0.6 0.8
Vbe (V)
Gummel plot

ce
3
(V)
-3
b
1
-1
c
0
I , I (A)
0
BV
10
0
1
12
TEM – Wide, misaligned base mesa
FIB/TEM by H. Chiang
Misalignment
0.5 mm
• 220 nm emitter-base junction
50 nm
• 1.1 mm wide base-collector mesa
Small EB gap
13
RF Data
35
Gains (dB)
25
H
20
15
10
f
U
30
Ic = 12.1 mA
= 1.0 THz
max
Je = 20.4 mA/mm2
Vcb = 0.7 V
21
P = 33.5 mW/mm2
A = 0.22 x 2.7 mm
W-Band
measurements to
verify f/fmax
je
5
0 9
10
2
f = 480 GHz

10
10
11
10
freq (Hz)
10
12
14
Base Post Cap
Ccb , po st 
 0 r  Apo st
Tc
Ccb,post does not scale with Le
 Adversely effects fmax as Le ↓
 Need to minimize the Ccb,post value
15
Base Post Cap
Ccb , po st 
 0 r  Apo st
Tc
Ccb,post does not scale with Le
 Adversely effects fmax as Le ↓
 Need to minimize the Ccb,post value
Undercut below base post
y = 1.09x - 0.02
4
C
cb
(fF)
6
No contribution of Base post to Ccb
2
0
0
1
2
3
L (mm)
e
4
5
16
Base Metal Resistance
Rbb,metal
Le
  sh,metal 
6Wbc
BP
Ib
Base
• Rbb,metal increases with emitter length
 fmax decreases with increase in emitter length
17
Base Metal Resistance
Rbb,metal
Le
  sh,metal 
6Wbc
BP
Base
Ib
• Rbb,metal increases with emitter length
 fmax decreases with increase in emitter length
400
(GHz)
1000
L = 3mm
max
e
L = 4mm
300
e
f

f (GHz)
500
L = 5mm
e
200
0
5
10 15 20
2
J (mA/mm )
e
25
800
600
L = 3mm
400
L = 4mm
200
L = 5mm
e
e
e
0
5
10 15 20
2
J (mA/mm )
e
25
18
Parameter Extraction
f
cb
max
1000
= 0.7V
600
f

500

f (GHz)
400
600
V
cb
0
300
200
5 10 15 20 25 30
Je (mA/mm2)
6
V
cb
5
4
=0V
V
cb
C
cb
200
= 0V
(fF)
fmax (GHz)
V
3
V
= 0.7 V
5
10 15 20 25 30
cb
Jkirk = 23 mA/mm2 (@Vcb = 0.7V)
2
0
= 0.5 V
2
J (mA/mm )
e
19
S(1,1)
Equivalent Circuit
Ccb,x = 2.72 fF
Rex ~ 4.2 Wmm2
Rcb = 27 kW
Rbb = 27 W
Ccb,i = 0.52 fF
Rc = 3.4 W
Base
Col
Rbe = 86 W
S21/10
Ccg = 3.2 fF
Cje + Cdiff = 8.8 + 59.2 fF
S12x5
Rex = 7 W
gmVbee-j
0.234Vbee(-j0.14ps)
Emitter
--- : Measured
x : Simulated
S11
S22
freq (1.000GHz to 67.00GHz)
Hybrid- equivalent circuit from measured RF data
20
Summary
• Demonstrated DHBTs with peak f / fmax = 480/1000 GHz
• Small Wgap for reduced base access resistance  High fmax
• Undercut below the base post to reduce Ccb
• Narrow sidewalls, smaller base mesa and better base ohmics
needed to enable higher bandwidth devices
21
Thank You
Questions?
This work was supported by the DARPA THETA program under HR0011-09-C-006.
A portion of this work was done in the UCSB nanofabrication facility, part of NSF funded NNIN network and MRL
Central Facilities supported by the MRSEC Program of the NSF under award No. MR05-20415
22