Chapter 6:

Optical Receivers and

Transceivers
 Photodetectors

A photo detector (optical receiver) converts

an optical signal into an electrical signal

Basic requirements
• Sensitivity at the required wavelength
• Efficient conversion of photons to electrons
• Small area for low capacitance and a fast response
• Low noise
• Sufficient area for efficient coupling to optical fiber
• High reliability
• Low cost
Copyright VPIsystems. All rights reserved. 1
 Direct bandgap

hv>Eg
 Bandgaps
• Silicon,
Indirect, shorter wavelengths, good for APD
• Germanium
Indirect, both shorter & longer, good for APD
• Gallium Arsenide (GaAs)
Direct bandgap, matched to InP, not good for APD
• InGaAs
Tunable bandgap, longer wavelength, matched to
InP
• InGaAsP
similar to InGaAs
 Photodetection Process

• Absorption of photons and generation of free

electrons and holes

• Transport of electrons and holes

• Amplification (optional)
 Basic Structure:
PN Photodiode
Electron p-type n-type
semiconductor semiconductor
Ec
Eg - Ve + Ve
hf > Eg p n
Ev
Hole
Electrons are attracted
Energy

to +ve contact
Detector
Current Detector
(Holes) Current
(Electrons)
Holes are attracted
to -ve contact
Distance
depletion
region
Copyright VPIsystems. All rights reserved. 1
 The Probability of
Photon Absorption
Depends on
• Thickness of intrinsic region, d
• The material’s absorption coefficient, 
• The wavelength of the incident light, λ
• No. of photons lost on their way to the intrinsic region
• Surface reflection
d

Block
Input photons/s Output photons/s
of
A material Aexp(-d)

Absorbed photons/s
A(1-exp(-d))

Copyright VPIsystems. All rights reserved. 1

 Reverse Bias and
Photocurrent
Reverse biasing the photodetector
• Increases the electric field in the depletion region
• Decreases its capacitance (increase speed)
• Increases its sensitivity and frequency response

photocurrent

Field strength

Reverse p
bias
Depletion
Region
load
n
Distance

Copyright VPIsystems. All rights reserved. 1

 Absorption
Absorption coefficient depends on wavelength,
and the material (bandgaps again)

105
Optical absorption coefficient

104
Si GaAs InGaAsP Ge
Band-
 (cm-1)

103 edges

102

10
0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
Wavelength (m)

Copyright VPIsystems. All rights reserved. 1

 Dark Current
• Dark current (flows even when there is no light)
• As the incident light increases…
the photocurrent increases linearly
Photocurrent

Increasing incident
light power

Dark current
(zero light power)
Reverse bias (V)

Copyright VPIsystems. All rights reserved. 1

 PIN Photodetector
Adding an “intrinsic” region between P and N
• increases depletion region width
• increases absorption of incident light
• increases the quantum efficiency of the photo detector

Anti-reflection coating
Field strength

P-type

Distance
Metal
Depletion
contacts Intrinsic
Region

n-type

Copyright VPIsystems. All rights reserved. 1

 Speed of PIN Photodetector
Speed - maximum detectable modulation rate
• i.e. the electrical frequency response

Speed is limited by the:

• Carrier velocity in depletion region
• Diffusion time of carriers outside the depletion region
• Time constant of the p-n junction capacitance
• Photodetector load resistance

capacitance
p i n
- Ve diffusion diffusion + Ve
velocity

Copyright VPIsystems. All rights reserved. 1

 Avalanche Photodetector (APD)
Avalanche region – higher internal electric field
• Accelerates carriers – more kinetic energy
• High energy collision frees bound electrons
• Freed electrons can collide - free more bound electrons
• Results in current gain (avalanche multiplication)

Field strength
n+ Avalanche
Gain in this region p Region

Absorption in this region

Intrinsic
Region
intrinsic

p+
Copyright VPIsystems. All rights reserved. 1
 Avalanche photodiodes (APD)
d

• Multiplication of carriers: IM  M  I p
• Responsivity:
R  R0 M
Gain vs. reverse bias voltage
 Quantum Efficiency

The quantum efficiency () is the probability that an

incident photon will produce an electron-hole pair

Direct bandgap
 = electron flux / photon flux

electrons per second

= hv>Eg
photons per second

   1  R p  1  eW 

Rp: reflection coeff  Collection coeff

W: absorption width
Copyright VPIsystems. All rights reserved. 1
 Responsivity
The responsivity, R0 (A/W), is the photocurrent produced per
unit of incident optical power

R0 = Ip/Pi = q/hf [A/W]

with Ip = (electron flux) • q [A]
Pi = (photon flux) • hf [W]
  m

1.24
The responsivity of a device relates to its design:
R = (1 - r) exp(-Dc) [1 - exp(-d)] (q/hc)

Facet Contact Intrinsic Wavelength

factor absorption absorption factor
Copyright VPIsystems. All rights reserved. 1
 Responsivity & Wavelength
Dependence
Different materials suit different wavelengths

R (A/W) 100% quantum efficiency

1.0

InGaAs
0.5 Ge
Si

0.6 0.8 1.0 1.2 1.4 1.6 1.8

Wavelength (m)

Copyright VPIsystems. All rights reserved. 1

• Anti-reflection coating
ncoating  nair nsemi
1 
Wcoating 
4 ncoating

• Bandwidth or response time

   RC
2
  transit
2

1
Bandwidth 

 Comparison between
APD & PIN
An APD:
• has gain, while a PIN does not
• can detect a weaker signal than a PIN
• requires a higher bias voltage than a PIN
• is noisier than a PIN
• is more sensitive to variations in temperature and
bias voltage than a PIN
• is more expensive than a PIN

Copyright VPIsystems. All rights reserved. 1

 Noise Sources in
Photodetectors
Electronic Shot Noise
• associated with the quantum nature of the light
• each incident photon produces an electron’s worth of current.

The total shot noise associated with a photocurrent current I

flowing through a potential barrier is:

<i2shot> = 2qIB
observation
Electron charge
bandwidth

Copyright VPIsystems. All rights reserved. 1

 Noise Sources in
Photodetectors
Thermal Noise
• is the result of thermally induced random fluctuations in the
charge carriers in a resistance
• occurs even when no voltage is applied across the resistance

Mean square thermal noise current is given by:

Boltzmann’s temperature
constant
bandwidth
4kTB
<i2ther> =
R
resistance

Copyright VPIsystems. All rights reserved. 1

 Noise Sources in
Photodetectors
Dark Current Noise Id
• flows in the photo detector even in the absence of light.
• caused by current leakage paths in the photo detector
• and thermal excitation of carriers across the p-n junction

Id gives rise to an additional shot noise current with a

mean-square value of:

<i2dark> = 2qId B

Copyright VPIsystems. All rights reserved. 1

 Noise Sources in
Photodetectors
APD Excess Noise
• is present in avalanche photodiodes because the avalanche
multiplication is essentially a random process

This causes the shot noise of the photodiode to be

multiplied by:

Noise Multiplication factor = M2+x

(0.1 < x < 1.0)
where M is the avalanche gain

Copyright VPIsystems. All rights reserved. 1

 Noise Sources in
Photodetectors
Optical Excess Noise
• can be broadly defined as any noise that appears along with the
received signal, other than quantum shot-noise

Most common types:

• Laser Intensity Noise
• Modal Noise
• Mode Partition Noise
• Amplified Spontaneous Emission (ASE) Noise in optical amplifiers

Copyright VPIsystems. All rights reserved. 1

Equivalent Circuit of a Photodiode
 Signal-to-Noise Ratio (SNR)
Electrical SNR (assuming shot noise dominates)

Pin isignal + ishot

<i 2shot> = 2qIB = 2qR0PinB

i 2signal = (RoPin)2

(RoPin)2 Pin
SNR = =
2qRoPinB 2hfB
q
Ro= , Responsivity of the photodetector
hf
Copyright VPIsystems. All rights reserved. 1
 SNR in Optical Systems
Mixing products in the detected photocurrent:

id  Ro| Es + EASE |2 = Ro( |Es|2+ |EASE|2 + 2Re[EsEASE] )

= RoPs+ RoPASE + RoPASE-ASE + RoPs-ASE

 id = is + iASE + isp-sp + is-sp

P
P
Signal id
ASE

Bandwidth Bo Responsivity Ro f
Optical Spectrum Detector Electrical Spectrum
Copyright VPIsystems. All rights reserved. 1
 Total SNR of Optical Systems
The SNR at receiver with responsivity Ro, and
electrical bandwidth B

is2
SNR =
(i2s-sp + i2sp-sp + i2shot + i2ther)B

includes contributions from id, iASE,

optical excess noise induced shot noise

Copyright VPIsystems. All rights reserved. 1

Optical receiver
Front-end
Low-impedance Front-end
Noise-bandwidth tradeoff
High-impedance front-end
Trans-impedance front-end
Equivalent circuit for Trans-impedance front-end
Noise in Trans-impedance front-end
Transfer function for Trans-impedance front-end

f
Example of Trans-impedance front-end
Optical receiver
EEquilization or electrical filtering
q
u
l
i
z
e
r
Main amplifier with AGC
Automatic gain control
Decision circuit

Rising or falling time of the clock control the sampling instants

At the sampling instant, compare the instantaneous voltage of
input waveform with the threshold bit 1 or 0
Signal regeneration– 3R

2R 1R

3R
re-amplification
Decision circuit
Optical receiver specifications
Optical receiver sensitivity:
The minimum average optical power required
to achieve BER=10-9 or 10-12
incorrect bits received
Bit error ratio (BER): BER 
total bits sent
Calculation of BER
Decision circuit compares decision current I with the threshold ID
If I(t)>ID, bit 1; I(t)<ID, bit 0

For bit 0, an error occurs if I>ID; and for bit 1, if I<ID

 p1P 0 | 1  p0P 1 | 0
Ne
BER 
Nt

p1  p0  0.5 Probabilities of bits “1” and “0”

BER  0.5P 0 | 1  P 1 | 0

ID
 I  I1 2   I1  I D 
P 0 | 1 
1 1
 exp
2  1   2 1 
dI  erfc  

 2 1 
2
2
 
 I  I 0 2   I D  I0 
P 1 | 0 
1 1

2  0 I D
exp
2 0 
dI  erfc  

 2 0 
2
 2

1  I1  I D   I D  I 0 
  erfc  
BER  erfc    2 
4   2 1   0  

By optimizing ID, BER becomes minimum.

I1  I D I D  I0
Q  Q factor
1 0
If thermal noise dominated,
 0 I1   1 I 0 I1  I 0
ID  ID 
1  0 2
 1  I1  I D   I D  I 0 
  erfc  
BER  erfc    
4   2 1   2 0  

Using optimized ID

1  Q 
BER  erfc  
2  2
I1  I 0
Q
1  0

 
If Q>3
1 exp  Q 2 / 2
BER 
2 Q
 Minimum average received power in an APD
P(t)
P1 NRZ

P0 t
PIN: M=1
I1  MRP1  2 MR P FA=1

Average power: P  P1  P0  P1 Receiver sensitivity

2 2
Be
RMS noise for bit 1: 1  Be s2   T2

for bit 0:  0  T
4 k BT
  2qM FA R  2 PB  Bn  
2
2 2
T Fn Bn
s
e RL
Optically pre-amplified Rx
Current/voltage
Optical signals Optical
filter Rx

Optical amplifier

2
I (t )  R GEs  E ASE  is  iT

 R GPs  GEs E ASE  GE ASE Es  E ASE   is  iT
  2
 

I  R  GPs  PASE  PASE  EASE

2
 Ssp Bo

Ssp  h nsp G 1

Variance of the current:
For bit 1
2  I2  I 2
 SA ,1  4  RGP1   RS sp Be 
2
  T2   s2   SA
2
  AA
2

 AA
2
 2R2 Ssp2  2Bo  Be  Be
 2
SA  4  RGPs   RS sp Be  12  2SA,1  2AA

 AA
2
 2R2 Ssp2  2Bo  Be  Be
For bit 0
I 1 I 0
Q  SA ,0  4  RGP0   RS sp Be 
2

1  0
 AA
2
 2R2 Ssp2  2Bo  Be  Be
1
BER  erfc Q
2
 2  02  2SA,0  2AA
 B0
Be
Be

B0
Useful expressions:
• Tx has infinite Q factor (i.e. noise free Tx)
Then Q factor at Rx:

2OSNR Bo Q 2  OSNR
Bo
Q
1  1  4OSNR Be Be

Ps I
Optical signal-to-noise ratio (OSNR): OSNR   s
PASE I ASE
Be  Be 
OSNR  Q 1  Q
Bo  Bo 
OSNR dB 
 1 
QdB   1   10  OSNRdB 

20

 2 
• Tx has a limited Q factor

Then, Q factor at Rx:

1 1 1
 2
 2
QR QTx Q Rx

Q Rx given as above

• pulse distortion is not considered in the above equations.

• Upper bound estimation of Q factor or BER