Towards Solid State
Photomultiplier
High Speed Avalanche Photodiodes
QD based Photodiodes
Dr Chee Hing Tan (陳志興）
Reader in Optoelectronic Sensors
Contents
Introduction
Basics of avalanche photodiodes
Electron-APD and nm scale avalanche regions
High speed avalanche photodiodes
Quantum dot based photodiodes.
© The University of Sheffield
Taiwan talk April 2012
Introduction
© The University of Sheffield
Taiwan talk April 2012
The Department of Electronic
and Electrical Engineering
The University of
Sheffield
• Origins date back to Firth College (1879) + Sheffield
Technical School (1884) + Sheffield School of Medicine
(1828)
• University College of Sheffield in 1897
• University of Sheffield – granted Charter in 1905
© The University of Sheffield
Taiwan talk April 2012
Overview of EEE department (Top 5)
• ~31 members academic staff
• ~50 research staff
• Student community comprises ~500
undergraduate and postgraduate taught
and research students
• Research and teaching activities are spread
across 3 research groups
• Communications
• Electrical Machines & Drives
• Semiconductor Materials & Devices
© The University of Sheffield
Taiwan talk April 2012
EPSRC National Centre for III-V
Technologies
Epitaxial Growth Capabilities
MOVPE Ga,In,Al,As,P
MOVPE Ga,In,Al,As,(P)
MOVPE Ga,In,Al,N
MBE 2
MBE 1
Ga,In,Al
As.P,N
Ga,In,Al
As,Sb, Bi
© The University of Sheffield
Taiwan talk April 2012
Low noise APDs
High speed
APDs
CHT
MWIR and LWIR
JPD&CHT
Chee Hing Tan
Jo Shien Ng
John Paul David
Novel avalanche
materials
JPD & JSN
© The University of Sheffield
Single photon
X-ray, UV
JSN, CHT & JPD
Taiwan talk April 2012
Physics of avalanche photodiodes
© The University of Sheffield
Taiwan talk April 2012
PMT and MCP
www.hamamatsu.co.uk
200nm-1700nm
4-25m
MCP
PMT
• High gain
• Fast response
•  3/8” to 20”
• Sensitive to vibration,
magnetic field,
temperature
•Higher cost
www.tectra-gmbh.com
• High gain
• Fast response
• Immune to magnetic
fields and vibration
• Lower power
consumption
• Compact
• High cost
D* >1x1015cmHz1/2/W but not suitable for imaging arrays
© The University of Sheffield
Taiwan talk April 2012
Impact ionisation
Holes and electrons ii
Significant gain fluctuation
Stochastic, Significant noise
and limited bandwidth
Electrons (holes) only ii
Minimal gain fluctuation
Deterministic, Low noise and high
bandwidth
h
p
drift
avalanche
M=1
absorption avalanche
M=24
absorption avalanche
n
Geiger
mode
10-2
Photocurrent
Dark current
10-3
absorption avalanche
10-4
M=24
absorption avalanche
Current (A)
M=6
10-5
10-6
10-7
10-8
10-9
10-10
M=24
absorption avalanche
© The University of Sheffield
M=24
absorption avalanche
10-11
0
5
10
15
20
Reverse bias (V)
Taiwan talk April 2012
25
Gain and excess noise for k=0 and
k=1
Excess noise vs. Gain is a very
good indicator of k value
Gain vs. Reverse bias is a
reasonable indicator of
k = / value
Measured noise
F
Ideal shot noise
F  kM  (2  1 / M )(1  k )
150
M  exp( w )
10
Gain
Excess noise factor
100
50
1
M
1  w
8
F M
6
4
F  2 1/ M
2
0
10
20
30
Reverse bias (V)
40
50
5
10
15
Gain
© The University of Sheffield
Taiwan talk April 2012
20
Current technologies
k =0, 0.1,……1.0
Excess noise factor
10
InGaAs/InP (1.1-1.65m)
-Mature IR APDs/SPADs
-Poor performance relative
to Si
8
Si (0.4-1.0m)
-Excellent, mature and
affordable APDs/SPADs
6 Increasing k
4
CdxHg1-xTe(1.1-4.5m)
-Excellent IR APDs
-Limited availability
2
5
10
15
20
Gain
To achieve a solid state PMT, k=0 is the most promising approach
CdHgTe (or CMT) APD is the leading technology. What next??????
© The University of Sheffield
Taiwan talk April 2012
Single photon detection
J.Beck et al., IEEE LEOS Newsletter, Oct. 2006
Baker et al. SPIE Vol.
6940
3D imaging, SPAD with raster scan,Buller et al. Heriot-Watt
© The University of Sheffield
Taiwan talk April 2012
Research: Can InAs work as a
Solid State PMT ???
© The University of Sheffield
Taiwan talk April 2012
InAs electron-APDs
InAs (eV)
Cd0.3Hg0.7Te (eV)
Eg (eV)
0.35
0.29
EGX (eV)
1.39 (3.91Eg)
3.18 (10.97Eg)
EGL (eV)
0.98(2.8Eg)
1.97(6.79Eg)
Cd0.3Hg0.7Te
© The University of Sheffield
Taiwan talk April 2012
Initial challenges in InAs
Almost no prior work due to preconceptions
•
Unavoidable surface leakage
•
High tunnelling current
High background doping
•
SIMs for InAs12
1e-3
1e-4
1e-5
1e-6
1e+21
1e+6
1e+20
1e+5
1e+19
1e+4
1e+18
1e+3
Si
Be
As
Al
1e+17
1e+16
1e+2
1e+1
1e+15
1e+14
0
2
4
6
8
10
1e+0
0
12
2
Reverse voltage (V)
2
Reverse leakage current density (A/cm )
2
200m radius
100m radius
Tunnelling
Multiplied
tunnelling
1
0.1
0.01
W=1m
0
2
4
6
8
10
Reverse bias voltage (V)
© The University of Sheffield
4
6
distance (m)
100
10
count (/s)
atom concentration (cm-3)
Leakage
current
(A) 1e-2
Reverse leakage current density (A/cm )
•
12
1000
100
10
50m radius
25m radius
Tunnelling
Multiplied
tunnelling
1
0.1
0.01
W=2m
0
5
10
15
20
Reverse bias voltage (V)
Taiwan talk April 2012
Dark current
proportional to ni
100
2
Jsurface
10-1
proportional to ni
10-2
Jbulk
10-3
10-4
10-5
EA=0.18eV
10-6
EA=0.36eV
10-7
Experimental
background limited data
10-8
2
4
6
8
10
1000/T (K-1)
12
14
10-4
10-5
10-6
10-7
10-8
10-9
10-10
10-11
10-12
10-13
10-14
10-15
10-16
10-17
10-3
Total reverse current (A)
Bulk Current Density (A/cm2)
101
Surface Current per unit length (A/cm)
10-2
10-4
10-5
10-6
77K
10-7
100K
125K
150K
175K
200K
220K
250K
290K
10-8
10-9
10-10
10-11
10-12
0
Bulk at high T
Surface leakage at low T
© The University of Sheffield
2
4
6
8
10
12
14
16
Reverse Bias Voltage (V)
 Dark current for 110µmdiameter SU-8 passivated APD
Taiwan talk April 2012
Responsivity / leakage
current comparison in pin
0.9
Responsivity (A/W)
0.8
1012
1011
0.7
0.6
0.5
0.4
Sheffield InAs 300K
Sheffield InAs 77K
Judson InAs
Hamamatsu InAs
0.3
0.2
1010
0.1
1.4
4
6
8
10
-1
1000/T (K )
12
High Detectivity InAs PD,
Ready to move towards Solid
State PMT
1.8
2.0
10
Sheffield InAs
Hamamatsu InAs
Judson InAs
2
108
2
1.6
Wavelength ( m)
Calculated detectivity
Judson InAs
Hamamatsu InAs
109
Current density (A/cm )
Detectivity (cmHz1/2/W)
1013
14
1
0.1
0.01
0.0
0.5
1.0
1.5
2.0
Reverse voltage (V)
Taiwan talk April 2012
2.5
3.0
InAs electron-APDs
10
3.5 m
2 m
1 m
Gain
8
6
4
10
2
9
p
i
n
8
0
0
2
4
6
8
7
Multiplication
Reverse Bias Voltage (V)
1000
V vs M (75um cap A)
V vs M (50um cap A)
V vs M (50um cap A)
M
100
6
5
n
4
i
3
p
2
10
1
0
1
0
5
10
15
20
25
V
30
0
2
4
6
8
10
Reverse bias voltage (V)
No measurable hole impact ionisation, confirming
electron only avalanche process
© The University of Sheffield
Taiwan talk April 2012
Avalanche Noise
4.0
InAs
AlInAs
CMT
(SWIR)
Excess noise factor, F
Excess Noise Factor, F
3.5
2.0
3.0
2.5
2.0
1.8
1.6
1.4
1.2
1.5
1.0
0
1.0
5
10
15
20
25
30
Multiplication, M
CMT data from Beck et al. Proc. SPIE Vol. 5564, p. 44-53
5
10
15
Multiplication factor, M
20
25
F is independent of temperature
AlInAs data from Goh et al. Journal of Quantum
Electronics
VISION: InAs single photon avalanche photodiode
for visible to SWIR wavelengths up to 3 m
© The University of Sheffield
Taiwan talk April 2012
SWIR Geiger mode Arrays/FPAs
Princeton
Lightwave: 32x32
GmAPD camera
10-2
-4
10
Current (A)
Rothman et. al, Detectors for Astronomy Workshop 2009
Photocurrent
Dark current
10-3
FLIR (Core by Indigo)
but low bias < 5V
but InGaAs/InP works
at >20V
10-5
-6
10
10-7
10-8
10-9
10-10
10-11
0
5
10
15
20
Reverse bias (V)
25
VISION: Work towards low voltage APD FPA
(Solid State PMT FPA??)
© The University of Sheffield
Taiwan talk April 2012
InAs electron-APDs
Bandwidth limited by the transit
time, not the avalanche process.
This means unlimited gainbandwidth product.
3dB bandwidth (GHz)
5
4
580GHz
Gain-bandwidth
product
3
2
1
1
10
100
Multiplication factor
InAs low noise high APDs for free space
communication (>40 Gb/s) and LIDAR at 1.55-3m
© The University of Sheffield
Taiwan talk April 2012
InAs X-ray APDs
Fano limited energy resolution
Improved FWHM for X-ray detection
© The University of Sheffield
Taiwan talk April 2012
InAsBi at Sheffield
X R D s p e c tra o f In A s B i o n In A s
ST B0H 3
ST B0H 5
ST B0H 6*
1 07
I n te n s ity ( c p s )
1 06
1 05
1 04
[ B i] = 4 . 2 %
1 03
[ B i] = 1 . 7 %
1 02
-4 0 0 0
[ B i] = 0 . 4 %
-3 5 0 0
-3 0 0 0
-2 5 0 0
-2 0 0 0
-1 5 0 0
-1 0 0 0
O M E G A _2T H E T A
-5 0 0
-42 meV/% Bi for InAsSb
-46 meV/% Bi for InAsSbBi
-55 meV/% Bi for InAsBi
© The University of Sheffield
0
500
1000
1500
( a r c se c )
Bi based IR
e-APDs?
Taiwan talk April 2012
Exploring the nonlocal effects in
nm scale avalanche region for
Solid State PMT
5 nm 50 nm
500 nm
© The University of Sheffield
1m
Taiwan talk April 2012
Avalanche statistics at high field
SMC E=600kV/cm
600kV/cm
number of ii events
800
2.5e+19
5.0e+18
1.0e+19
1.5e+19
2.0e+19
2.5e+19
p(x,t)
2.0e+19
1.5e+19
600
600kV/cm
300kV/cm
400
200
1.0e+19
SMC E=300kV/cm
5e-12
5.0e+18
0.0
4e-12
0.2
0.4
3e-12
1.5e-7
x(m)
5.0e-8
0.0
1.0
t(s
)
300kV/cm
1e-12
1.0e-7
0.8
x/w
2.0e-7
0
Monte Carlo
simulation in
GaAs
2.5e+17
0
1e+17
2e+17
3e+17
3e+17
2.0e+17
p(x,t)
1.5e+17
1.0e+17
5e-11
5.0e+16
4e-11
)
3e-11
3.0e-6
© The University of Sheffield
2.5e-6
2e-11
2.0e-6
1.5e-6
t(s
2.5e-7
2e-12
0.6
1e-11
1.0e-6
x(m)
5.0e-7
0.0
0
Taiwan talk April 2012
Excess noise in thin avalanche
regions
15
4
3
k=0.2
GaAs
2
10
k=2.4
InP
Fujitsu SAM APD
k=1
10
decreasing w
5
k=0.4
Excess Noise Factor, F
w = 1m
w = 0.5m
w = 0.3m
w = 0.2m
w = 0.1m
w = 0.049m
Excess noise factor, F
Excess Noise Factor, F
5
8
Si
k=0.5
6
k=0.3
4
k=0.2
2
2
k=0
1
1
2
3
4
k=1
blue: Me
red:Mmix
4
6
8
10
Multiplication, M
5 6 7 8 910
Multiplication, Me
1
10
Multiplication, M
Excess noise corresponds to k~0.1-0.2 when w <100 nm.
Even smaller w is required to reduce excess noise towards unity. Is
wide bandgap materials the answer?
© The University of Sheffield
Taiwan talk April 2012
Excess noise in wide bandgap
materials
Al Ga As
0.8
Al0.6Ga0.4As
0.2
6
Fh(0.1m n+-i-p+)
Fe (0.1m p+-i-n+)
Fe (0.05m p+-i-n+)
Fe (0.025m p+-i-n+)
excess noise factor, Fe
5
4
k=0.1
3
2
Fe(26nm)
1
2
4
6
8
10
12
14
multiplication factor, Me
Noise comparable to Si when w = 25
nm in wide indirect bandgap materials
(similar trends for x= 0.7 and 0.9)
© The University of Sheffield
Taiwan talk April 2012
Novel AlAsSb APDs
InP
5.0
F (M~10)
4.5
4.0
InAlAs
3.5
3.0
2.5
2.0
0.0
0.5
1.0
1.5
2.0
Avalanche Width (mm)
2.5
Excess Noise Factor
7
AlAsSb APD
Hamamatsu Si APD
Fujitsu InGaAs/InP APD
6
5
4
3
2
1
2
4
6
8
10
12
Gain
200nm AlAsSb has extremely low noise
Can we achieve even lower noise using thinner
avalanche region?
© The University of Sheffield
Taiwan talk April 2012
14
Extremely low temperature coefficient
of breakdown voltage
30
10-4
25
Cbd (mV/K)
Measured current (A)
10-5
10-6
10-7
10-8
10-9
10-10
20
15
10
5
10-11
0
10-12
0
5
10
15
Reverse bias (V)
Material
AlAsSb
Si
GaAs
Al0.6Ga0.4As
GaInP
InP
InAlAs
w
(nm)
80
70
100
89
-80
80
© The University of Sheffield
20
Cbd
(mV/K)
0.95
1.53
1.67
3.18
-3.9
2.2
0.0
w
(nm)
230
290
270
-200
200
200
Cbd
(mV/K)
1.47
4.38
8.89
-3.68
6.0
4.1
0.2
0.4
0.6
0.8
1.0
Avalanche region thickness ( m)
Taiwan talk April 2012
High speed avalanche photodiodes
© The University of Sheffield
Taiwan talk April 2012
Current materials for 1.55m
APDs
Gain_bandwidth Product (GHz)
400
300
InAlAs APDs
InAlAs APDs Sheffield
InP APDs
Ge on Si APD
200
100
0
1980 1985 1990 1995 2000 2005 2010 2015
Year
InAlAs and InP have reached their limits ???
InAs requires very small area and have some issues with
surface leakage.
Waveguide APD with gain ~3 works at 40Gb/s.
© The University of Sheffield
Taiwan talk April 2012
APD design
M=10
Avalanche
multiplication
Drift time
100
GBP (GHz)
absorption
M=10
Absorption width(nm)
>200GHz
>1THz
>500GHz
100
10
1
40
60
10
100
1000
20
Bandwidth (GHz)
1000
80
Multiplication thickness (nm)
© The University of Sheffield
100
GBP1_20nm
GBP2_40nm
GBP3_60nm
GBP4_80nm
GBP5_100nm
GBP6_150nm
GBP7_200nm
GBP0_no absorption
BW1_20nm
BW2_40nm
BW3_60nm
BW4_80nm
BW5_100nm
BW6_150nm
BW7_200nm
BW0_no absorption
10
Avalanche region thickness(nm)
100
200GHz
Wm<100nm Wa<1000n
m
500GHz
Wm<45nm
Wa<500nm
1THz
Wm<20nm
Wa<100nm
For bandwidth >100GHz wa <150
and wm<20 nm
Taiwan talk April 2012
Wide band APDs
Absolute impedance
1000
Lm
wridge
Zm
d
Zt
p
i
n
Gl
Zb
wgap
Impedance ( )
800
wp
Ci
L=15
L=30
L=55
50
600
400
200
0
10
20
30
40
Frequency (GHz)
Bias= 17.0 V
Relative response (dB)
2
L = 30 m
L = 55 m
XPDV diode
0
-2
-4
-6
10
© The University of Sheffield
20
30
Taiwan talk
April 2012
Frequency
(GHz)
40
Wide band APDs
Absorption =1200nm
2D Graph 1
-50
16.5v vs Col 3
17V vs Col 8
17.5v vs Col 13
18V vs Col 18
18.5v vs Col 23
19V vs Col 28
19.5v vs Col 33
20v vs Col 38
22V vs Col 43
24v vs Col 48
26V vs Col 53
28V vs Col 58
30V vs Col 63
31V vs Col 68
32V vs Col 73
33V vs Col 78
Noise floor vs Col 82
Response (dBm)
-60
-70
-80
-90
-100
-110
-120
10
20
30
40
Frequency (GHz)
24.8 V
24 V
23.6 V
23 V
22 V
21 V
20 V
19.5 V
19 V
18.5 V
18 V
Noise floor
XPDV diode
-50
Response (dBm)
-60
-70
-80
-90
-100
-110
-120
10
20
Frequency (GHz)
24/04/2012© The University of Sheffield
30
40
Absorption =400nm
36
QD detectors for imaging:
How to achieve low cost and
multifunction??
1550oC
www.kentoptronics.com
© The University of Sheffield
150oC
Taiwan talk April 2012
QD on Si for low cost FPA
UoSHEFF
UCL
QDIP on Silicon is promising!
© The University of Sheffield
Taiwan talk April 2012
VISION: Reconfigurable IR sensors
Can be dynamically configured into
High resolution
multicolour FPA
Multispectral : Row-by-row
Multispectral : Full FPA
Future Intelligent
Sensors?
No moving parts
Multispectral: Pixel-by-pixel
© The University of Sheffield
No cooled optical filters needed
Taiwan talk April 2012
Algorithmic spectrometer
20nm
0
2
4
6
8
10
( m)
Optimised
Algorithim
0
2
4
6
8
10
8
10
( m)
Quantum Dot Infrared Photodiodes
(QDIPs)
MWIR LWIR
0
2
4
6
( m)
© The University of Sheffield
Taiwan talk April 2012
Multicolor using a single
detector
1.0
Projection
algorithm
Responsivity (A/W)
0.9
0.8
0.7
0.6
0.5
0.4
0.3
1.2
0.2
0.1
1.0
4
6
8
10
Wavelength (m)
Multispectral
imaging without
dispersive optics?
12
Responsivity (A/W)
0.0
0.8
0.6
0.4
0.2
0.0
-0.2
4
6
8
10
12
Wavelength (m)
© The University of Sheffield
Taiwan talk April 2012
0.1
Power per unit Area (W/m 2)
Power per unit Area (W/m 2)
Spectroscopy without
dispersive optics ??
Power on Device
Theoretical Reconstruction - 0.25um FWHM
0.01
4
5
6
7
8
9
10
11
12
13
Wavelength (m)
0.1
Power on Device
Algorithm Reconstruction - 0.25um FWHM
Algorithm Reconstruction - 0.5um FWHM
0.01
4
5
6
7
8
9
10
11
12
Wavelength (m)
Vines et al. Versatile spectral imaging with an algorithmbased spectrometer using highly tuneable quantum dot
infrared photodetectors. IEEE Journal of Quantum
Electronics, 47(2), 190-197.
Absorption feature as small
as 300nm measured
© The University of Sheffield
Taiwan talk April 2012
13
Thank you
To
Discover
And
Understand.
Seeing photons: Progress and Limits of
Visible and Infrared Sensor Arrays
(http://www.nap.edu/catalog.php?record_id=12896)
p.62
“ultralarge-format arrays; mosaic tiling
technologies; pixel size reduction; smarter pixels and
on-focal plane processing; improved threedimensional (3-D) integration and hybridization;
higher-operating-temperature devices; multicolor
pixels; improved short-wavelength infrared (SWIR)
arrays; photon counting technologies and lower
readout noise; curved focal surfaces; lower-power
operation; radiation hardening; cost reduction; and
improved cooler technology.”
24/04/2012© The University of Sheffield
45
Novel Applications
120oC
http://www.sensorsinc.com/swirconops.html
24/04/2012© The University of Sheffield
46
Detectivity
Bulk semiconductors
InSb, InAs, HgCdTe
Conduction band
Valence band
Detectivity = Responsivity x Area0.5
Noise current
Impurity doped semiconductors
Ge:Zn, Si:As
Conduction band
Impurity band
Valence band
Conduction band
Impurity band
Bulk is the leading technology in
terms of D* What next????
Valence band
24/04/2012© The University of Sheffield
47
LWIR photodetector comparison
A Rogalski
Photonic Bandgap
h
p
i
n
24/04/2012© The University of Sheffield
49
Photonic Bandgap
J. Rosenberg, physics.optics,16July2009
Posani et al., APL 88,151104,2006
VISION: Multispectral and polarimetric sensing
24/04/2012© The University of Sheffield
50
BULK N and Bi detectors
51
InAsBi
InAsN
24/04/2012© The University of Sheffield
MWIR Type II InAs/GaSb
EMRSDTC
Room temperature Type II FPA on GaAs
24/04/2012© The University of Sheffield
52
InGaAs
InGaAs has large / at
low fields.
What about InAs???
© The University of Sheffield
Taiwan talk April 2012