iii-v and group iv epitaxy for low energy optoelectronics · 2018-10-31 · iii-v and group iv...
TRANSCRIPT
A Science & Technology Center
III-V and Group IV Epitaxy for Low Energy Optoelectronics
Christopher Heidelberger1, Seth Fortuna2, Ming Wu2, Eugene A. Fitzgerald1
1Department of Materials Science and Engineering, Massachusetts Institute of Technology
2Department of Electrical Engineering and Computer Sciences, University of California, Berkeley
September, 2017
A Science & Technology Center
Outline
Materials for optical linkHBT integration for high performance
communicationLikely application for these technologies
E3S Seminar Page 24/13/2017
A Science & Technology Center
Active region p-doping analysis from UC Berkeley
E3S Seminar Page 34/13/2017
Assumptions:(1) Surface recombination
velocity = 3e4 cm/s(2) Auger recombination
coefficient = 2e-29 cm-6s-1(3) Sp. Em. Enhancement = 640,
Antenna Efficiency = 45%Q-factor = 37
(4) Bulk SRH ignored(5) Injection carrier density =
1.5e18cm-3
Target
A Science & Technology Center
Growth of undoped LED structure
E3S Seminar Page 44/13/2017
InP (substrate)
InP bottom contact 300 nm
InGaAsP barrier
InGaAs QWInGaAsP barrier
InGaAs QWInGaAsP barrier
InGaAs QW
InGaAsP barrier
60 nm
6 nm10 nm6 nm10 nm6 nm
40 nm
InP top contact 30 nm
InGaAs top contact 40 nm
LED structure schematic: XTEM
HT = 200 kV, (110) on pole
A Science & Technology Center
Luminescence of undoped LED structure
E3S Seminar Page 54/13/2017
>10X increase in quantum efficiency
MQW (MIT)
Previousdesign (outsidevendor)
1450 1500 1550 1600 1650
(nm)
0
0.2
0.4
0.6
0.8
1
Inte
nsity
(arb
.)
PL of blanket film
1550 nm target
Quantum efficiency (large-area LEDs, UC Berkeley)
A Science & Technology Center
LED structure with Zn p-type doping
E3S Seminar Page 64/13/2017
undoped active region:
p+ active region (Zn doping ~ 5 x 1018):
InGaAs contact
InP contact
InGaAs/InGaAsPMQW
InP contact
InGaAs/InGaAsPMQW
DIC Image
PL of Blanket Films
1450 1500 1550 1600 1650
(nm)
0
0.2
0.4
0.6
0.8
1
1.2
1.4
Inte
nsity
(arb
.)
undoped
5 x 10 18 cm -3 Zn doping x100
poor EL and IV characteristics for fabricated LED
A Science & Technology Center
Lattice matched growth
7
InP substrate (S-I)
InP regrowth, UID20 nm
InGaAs, C-doped200 nm
InP cladding, UID10 nm
600 C
500 C
550 C
Growth temperature
V/III ratio = 4CBrCl3 ratio = 9%
p = 3.2 x 1018 cm3
qx (nm-1)
qz (n
m-1
)
4.815 4.82 4.825 4.83 4.8356.74
6.76
6.78
6.8
6.82
6.84
6.86
6.88
InP substrate
InGaAs
fully strained
In fraction = 52.2%
A Science & Technology Center
Post-growth N2 anneal: effect on PL intensity and lifetime
Decrease in intensity/lifetime after post-growth anneal increased Auger recombination due to higher hole concentration
Laser power = 500uWWavelength = 1000nm20X objective
without anneal
with anneal 𝜏𝜏 ≈ 140ps
𝜏𝜏 ≈ 700ps
Time-resolved (UC Berkeley)
without anneal, 3.2 x 1018 cm-2
with anneal, 1 x 1019 cm-3
1400 1500 1600 1700 1800 1900
(nm)
10 -2
10 -1
10 0
Inte
nsity
(arb
. uni
ts)
UID In0.53
GaAs control
In0.52
GaAs; 3.2x10 18 cm -3 ; no anneal
In0.52
GaAs; 1.0x10 19 cm -3; 500 C, N2, 10 min anneal
A Science & Technology Center
InGaAs LED structures: conclusions
Demonstrated undoped InGaAs/InGaAsP LED structure with good morphology and high quantum efficiency
Grew C-doped InGaAs with p > 1 x 1019 cm-3 and recombination limited by Auger recombination
Future work Grow InGaAsP cladding layers at similar growth conditions as C-doped
InGaAs Grow and characterize InGaAs/InGaAsP LED structure with high p-type
doping Fabricate and test antenna-enhanced LED structure with heavily-
doped active region (UC Berkeley)
E3S Seminar Page 94/13/2017
A Science & Technology Center
GaAsP/InGaP HBT on Si
heterojunction bipolar transistor (HBT): high frequency (& independent of litho) high power efficiency
GaAsP/InGaP
A Science & Technology Center
EBIC measurements of GaAsP HBTs
GaAsP subcollector (n+)
GaAsP collector (n)
GaAsP base (p+)
InGaP emitter (n+)
GaAsP contact (n++)
Current Amplifier
Signal Conditioning
Computer/
Display
electron beam
ohmic contact
e-h pairs generated
𝑉𝑉𝐴𝐴ΔSiGe graded buffer (n+)
Si substrate (n+)
A Science & Technology Center
Experimental loop
12
Epitaxial Growth
(MOCVD)
Fabricate HBTsEBIC + Electrical
Measurements
Design Structure
0.4 0.6 0.8 1 1.2 1.4
VBE (V)
10 -12
10 -10
10 -8
10 -6
10 -4
10 -2
I C, I
B (
A)
IC
IB
A Science & Technology Center
Iterative approach to improved defect density
C base doping
Zn base doping
improved InGaP
calibration
thick subcollector
layer
Misfit Dislocation
Density
Modification
1900 cm-1
EBIC
900 cm-1 600 cm-1 < 1 cm-1
30 µm
A Science & Technology Center
High current gain demonstrated on Si substrate
GaAsP subcollector (n+)
GaAsP collector (n)
GaAsP base (p+)
InGaP emitter (n+)
GaAsP contact (n++)
ΔSiGe graded buffer (n+)
Si substrate (n+)
C base doping
Zn base doping
improved InGaP
calibration
thick subcollector layer
Modification
0.5 1 1.5 2 2.5
VBE (V)
10 -12
10 -10
10 -8
10 -6
10 -4
10 -2
I B,
IC
(A
)
IC
IB
60 µm diameter HBT
𝛽𝛽 = 156
= 3.7 × 106 cm-2
< 1 cm-1
30 µm
C. Heidelberger and E. Fitzgerald, J. Appl. Phys., under review
A Science & Technology Center
• combined models for threading and misfit dislocations
• misfit dislocations have pronounced effect on current gain, even at low densities
C. Heidelberger and E. Fitzgerald, J. Appl. Phys., under review
“Current Gain Map”
A Science & Technology Center
Applications/Market: Is Computing a Target Market? 28nm likely to be ‘last profitable node’ for non-monopolistic market
participants? De-integration occurring for performance and market reasons Computing cores will be mature for a decade or more Data center ‘farms’ or even ‘cities’
• Key necessary innovations obscured by monopolistic corporate structure• University research likely irrelevant to key problems without an informing paradigm
Communication and power management take front-seat• Within-data-center more important for silicon ICs because of market size• i.e. telecom thinking must leave silicon photonics (i.e. = telecom photonics)
What to do in next decade or two? Disruptive* path suggests key silicon technology (with new added value in
materials) to appear in communication, lighting, and power management ICs; new era of ASICs• Nothing to do with data centers, computing, etc. initially• Grow in other markets for decades, then spill back into computing
– Real 5G wireless– Visible integrated optoelectronics
• Old silicon nodes (and 200mm) are playground and commercialization for innovation
*Actual Christensen definition, not Valley term