InAlAs/InGaAs/InP DHBTs with Polycrystalline InAs Extrinsic Emitter Regrowth

1
InAlAs/InGaAs/InP DHBTs with Polycrystalline
InAs Extrinsic Emitter Regrowth

• D. Scott, H. Xing, S. Krishnan, M. Urteaga, N.
  Parthasarathy and M. Rodwell
• University of California, Santa Barbara

dennis_at_umail.ucsb.edu 805-893-8044,
805-893-3262 fax
2
Advantages of InP vs. SiGe HBTs

• InP HBT Material Properties
• Available lattice-matched materials allows for
  emitter bandgap wider than base, allowing for
  higher base doping and lower base sheet
  resistance
• Electron velocities reported as high as 4?107
  cm/s
• InP HBT Processing Technology
• High topography mesa structure allows for
  small-scale integration
• Base-emitter junctions defined by etching and
  depositing a self-aligned base metal results in
  low yield and limits emitter scaling

• Si/SiGe HBT Material Properties
• Allowable lattice mismatch limits GeSi alloy
  ratio resulting in smaller emitter-base bandgap
  difference and higher base sheet resistance
• 41 lower electron velocity is seen in silicon
• Si/SiGe Processing Technology
• Planar process using silicon CMOS technology
  allows for VLSI
• Self-aligned base-emitter junctions are diffused,
  extrinsic base and emitter wider than the active
  junction allows for high degree of scaling

3
Evolution Cbc Reduction in III-V HBTs
Emitter
Emitter
Collector
Collector
Base
Base
Subcollector
Subcollector
S.I. Substrate
S.I. Substrate
Mesa HBT
Cbc Reduction HBT
Emitter
Emitter
Collector
Base
Subcollector
Collector
S.I. Substrate
Transferred Substrate HBT
Highly Scaled HBT
4
UCSB Highly Scaled HBT

• UCSB has demonstrated laterally
• scaled HBTs with emitters written
• by e-beam lithography.
• These HBTs show problems with
• High emitter resistance, Rex
• Low yield
• These devices demonstrated lower than predicted
  values of f? despite aggressive thinning of the
  epitaxial layers.

5
Si/SiGe HBT Process Advantages

• Highly scaled
• very narrow active junction areas
• very low device parasitics
• high speed
• Low emitter resistance using wide n polysilicon
  contact
• Low base resistance using large extrinsic
  polysilicon contact
• High-yield, planar processing
• high levels of integration
• LSI and VLSI capabilities

Published Si/SiGe HBT f? as high as 210
GHz InP-based HBT f? as high as 341 GHz
6
Polycrystalline n InAs

• Polycrystalline InAs grown on SiNx Hall
  measurements as high as
• Doping 1.3 ?1019 cm-3, Mobility 620
  cm2/Vs
• Results in doping-mobility product of 8?1021 (V
  s cm)-1

• Compare these numbers to InGaAs lattice matched
  to InP
• Doping 1.0 ?1019 cm-3, Mobility 2200
  cm2/Vs
• Results in doping-mobility product of 22?1021 (V
  s cm)-1

Polycrystalline InAs has potential as an
extrinsic emitter contact!
7
Base-Collector Template for Regrown Emitter HBT
Base-collector template prior to regrowth
Base-collector template as-grown
8
Regrown Emitter Fabrication Process
Regrowth
Emitter/cap etch
Metalization
Base/collector etch
9
Large-area Small-emitter HBTs
10
First Attempt Results
Regrown area
SiNx
Regrown area very rough
Transistor action!!
11
Growth and Process Improvements
Regrown area
Regrown area
SiNx
SiNx
First attempt at the base- emitter junction
without RHEED or pyrometer
Second attempt with improved pre-regrowth
processing and RHEED/pyrometer features added to
the wafer
12
Growth and Process Improvements
First attempt at the base- emitter junction
without RHEED or pyrometer
Second attempt with improved pre-regrowth
processing and RHEED/pyrometer features added to
the wafer
13
Base-emitter Regrowth SEM Detail
14
Base-emitter Regrowth SEM
2 µm emitter regrowth 30K magnification
1 µm emitter regrowth 55K magnification
15
Second Attempt DC Results

• Unintended InAlAs Layer (gt50Å)
• wide-bandgap layer acts as a current block from
  emitter to base
• reduces common-emitter gain
• may account for the dip in common-emitter curves

Common-emitter gain, ß gt 15
16
Base-emitter Current Leakage
Evidence of resistance seen in the base-emitter
diode
Evidence of base-emitter leakage seen in Gummel
17
Third Attempt DC Results
Third Attempt DC Results
Common-emitter gain, ß gt 20
18
Base-collector Grade Design Error
InP collector
InP collector
InGaAs base
InGaAs base
Base-collector band diagram with the incorrect
base-collector grade. This mistake may account
for the oscillations seen in the HBT I-V curve.
Base-collector band diagram with the corrected
base-collector grade. A thin, heavily-doped layer
was inserted between the grade and collector to
pull the conduction band down at the
grade-collector junction.
19
Regrowth with Buried Base Contact
20
InP HBTs with polycrystalline InAs extrinsic
emitter regrowth

• Objective
• Emulate high-yield 0.2 um SiGe emitter process
• Polycrystalline extrinsic emitter ? wide contact
  for low resistance
• Future Work
• RF devices need to be designed and demonstrated
• GaAsSb based DHBTs should be demonstrated
• Higher scaling in the regrown emitters needs to
  be examined
• Growth Related Work
• A low-resistance p-type polycrystalline contact
  needs to be verified
• Regrowth of the base will need to be explored to
  obtain a fully planar HBT completely analogous to
  the Si/SiGe HBT

21
InP HBTs with polycrystalline InAs extrinsic
emitter regrowth
Objective Emulate high-yield 0.2 um SiGe
emitter process Polycrystalline extrinsic
emitter ? wide contact for low resistance Future
Work (short-term) Improve DC
characteristics. Improve base capping layer
to lower extrinsic base resistance GaAsSb
base layers for higher carbon incorporation
Deep submicron scaling of regrown emitter.
RF device demonstration Future work
(long-term) full SiGe-like process flow for
submicron InP HBT regrown emitter,
regrown extrinsic base over buried dielectric
spacer for Ccb reduction
22
Future Work

• DC Device Work
• DC characteristics should be demonstrated without
  the design errors
• Improvements will be made to the base capping
  layer to lower extrinsic base resistance
• GaAsSb based DHBTs should be demonstrated
• Higher scaling in the regrown emitters needs to
  be examined
• RF Device Work
• RF devices need to be designed and demonstrated
• Growth Related Work
• A low-resistance p-type polycrystalline contact
  needs to be verified
• Regrowth of the base will need to be explored to
  obtain a fully planar HBT completely analogous to
  the Si/SiGe HBT