STRUCTURAL AND BIAS VOLTAGE EFFECTS ON THE FREQUENCY RESPONSE OF InGaAs InP PHOTODIODES

1
STRUCTURAL AND BIAS VOLTAGE EFFECTS ON THE
FREQUENCY RESPONSEOF InGaAs/ InP PHOTODIODES

• Jorge Manuel Torres Pereira
• Institute of Telecommunications/DEEC - IST
• Lisbon, PORTUGAL

CSNDSP06

19-21 July 2006

Patras, Greece
2

• SUMMARY
• Introduction
• The Device Structure
• Simulation Model
• Results
• Conclusions

CSNDSP06
19-21
July 2006

Patras, Greece
3
INTRODUCTION p-i-n Photodiodes () easy to
fabricate () good frequency response () low
noise ( - ) low sensitivity
Compound semiconductor materials III-V e.g.
InGaAs, InP,...
CSNDSP06

19-21 July 2006

Patras, Greece
4
THE DEVICE STRUCTURE
Schematic of the p-i-n structure used in this
work. The figure shows the light incident from
the left.
CSNDSP06

19-21 July 2006

Patras, Greece
5
SIMULATION MODEL
The continuity equations
CSNDSP06

19-21 July 2006

Patras, Greece
6
In the absorption region
a absorption coefficient f1 amplitude of the
sinusoidal optical flux q magnitude of the
electron charge
CSNDSP06

19-21 July 2006

Patras, Greece
7
Solving the continuity equations for each layer,
the electron and hole current densities at the
boundary on the right may be related to the
current densities at the boundary on the left by
Ti the current transfer matrix current
source vector
CSNDSP06

19-21 July 2006

Patras, Greece
8
The total current density, for a structure with N
layers, may be written as
where
which may be written as
CSNDSP06

19-21 July 2006

Patras, Greece
9
For the union of two layers the quantities T,
and D follow a simple set of rules
CSNDSP06

19-21 July 2006

Patras, Greece
10
The output current densities Jp(la,w) and Jn(0,w)
may be related to the input current densities
Jp(0,w) and Jn(la,w) through quantities which may
be expressed in terms of T, and D.
In this case the total current density is
obtained by
with
CSNDSP06

19-21 July 2006

Patras, Greece
11
The Ti, values for the
ith layer
CSNDSP06

19-21 July 2006

Patras, Greece
12
The electric field profile in the absorption
region
where Ud is called the punchthrough voltage,
given by
CSNDSP06

19-21 July 2006

Patras, Greece
13
The drift velocity as a function of the electric
field
CSNDSP06

19-21 July 2006

Patras, Greece
14
Empirical expressions for v(E)
mn 1.05 m2V-1s-1 mp 0.042m2V-1s-1 vnl
6x104 m/s vpl 4.8x104 m/s b 7.4x10-15
(m/V)2.5 g 2.5
CSNDSP06

19-21 July 2006

Patras, Greece
15
Table I Material parameters
CSNDSP06

19-21 July 2006

Patras, Greece
16
RESULTS
.
Simulated frequency response of p-i-n structures
with different absorption region widths and bias
voltages
CSNDSP06

19-21 July 2006

Patras, Greece
17
.
The 3-dB bandwidth as a function of the bias
voltage for several values of the absorption
region width
CSNDSP06

19-21 July 2006

Patras, Greece
18
The 3-dB bandwidth as a function of the bias
voltage for several values of the residual donor
concentration
CSNDSP06

19-21 July 2006

Patras, Greece
19
The 3-dB bandwidth as a function of the
absorption region width for several values of
the bias voltage
CSNDSP06

19-21 July 2006

Patras, Greece
20
The 3-dB bandwidth as a function of the
absorption region width for several values of
the residual donor concentration
CSNDSP06

19-21 July 2006

Patras, Greece
21
CONCLUSIONS

• The results from the simulations seem to agree
  quite well with those from the experiment.
• The frequency response is mainly dependent on the
  absorption region width, with shorter devices
  having a wider bandwidth.
• The 3-dB bandwidth is seen to be dependent on
  bias and doping concentration only for devices
  with large absorption region widths. In this case
  the bandwidth increases when the bias voltage
  increases and the doping concentration decreases.

CSNDSP06

19-21 July 2006

Patras, Greece
22
THANK YOU FOR YOUR ATTENTION
CSNDSP06

19-21 July 2006

Patras, Greece