Optical Wavelength-division Multiplexing/Demultiplexing Devices
M. Vieira1,2, P. Louro1,2, M A Vieira1,3, A. Fantoni1, M. Fernandes1, M. Barata1,2
1
Electronics Telecommunications and Computer Dept, ISEL, Lisbon, Portugal.
2
CTS-UNINOVA, Lisbon, Portugal. 3CML-Traffic Department, Lisbon, Portugal
[email protected]
Abstract
Results on the use of a double a-SiC:H p-i-n heterostructure for signal multiplexing and demultiplexing
applications in the visible range, are presented. Pulsed monochromatic beams together (multiplexing
mode), or a single polychromatic beam (demultiplexing mode) impinge in the device and are
absorbed, accordingly to their wavelength.
Red, green and blue pulsed input channels are transmitted together, each one with a specific
transmission rate. The combined optical signal is analyzed by reading out, under different applied
voltages, the generated photocurrent. Results show that in the multiplexing mode the output signal is
balanced by the wavelength and transmission rate of each input channel, keeping the memory of the
incoming optical carriers. In the demultiplexing mode the photocurrent is controlled by the applied
voltage allowing regaining the transmitted information. An electrical model gives insight into the
device operation.
Keywords: Optical devices, a-SiC heterostructures, optical communication, multiplexing and
demultiplexing applications over POF
1.
Introduction
The standard communication over polymer
optical fibers (POF) uses only one single channel
[1]. To increase bandwidth for this technology the
only possibility is to increase the data rate, which
lowers the signal-to-noise ratio and therefore can
only be improved in small limitations. To open up
this communication bottleneck Wavelength
Division Multiplexing (WDM) can be employed.
WDM enables the use of a significant portion of
the available fiber bandwidth by allowing many
independent
signals
to
be
transmitted
simultaneously, with each signal located at a
different wavelength. Routing and detection of
these
signals
can
be
accomplished
independently, with the wavelength determining
the communication path by acting as the
signature of the origin, destination or routing.
Components are therefore required to be
wavelength selective. Although they are well
known for infrared telecommunication, they must
be completely renewed because of the different
transmission windows. Only the visible spectrum
can be applied when using POF for
communication. So, the conception of new
devices for signal (de)multiplexing in the visible
spectrum is a demand in this field [2, 3]
This paper presents results on the
applicability of a multilayered a-SiC:H
279
heterostructures
as
an
electrically
programmable
optical
filters
for
WDM
transmission
over
POF.
In-house
communication, car network and traffic control
applications are foreseen due to the low cost
associated to the amorphous a-SiC:H and POF
technologies.
2.
2.1
Experimental details
Double-junction WDM device
configuration.
The device is a double heterostructure in a
glass/ITO/a-SiC:H (p-i-n)/ a-SiC:H(-p)/a-Si:H(i’)/a-SiC:H(-n)/ITO configuration produced by
Plasma Enhanced Chemical Vapor Deposition as
depicted in Figure 1. On the top of the figure it is
displayed the recombination profiles (straight
lines) under red (λR =650 nm) green (λG =550 nm)
and blue (λB =450 nm) optical bias and different
electrical bias (-6V<V<0V). The generation
profiles are also shown (symbols). We used a
device simulation program ASCA-2D [4] to
analyze the profiles in the investigated structure.
The thickness and the absorption coefficient of
the front photodiode are optimized for blue
collection and red transmittance, and the
thickness of the back one adjusted to achieve full
absorption in the greenish region and high
collection in the red spectral one. As a result,
both front and back diodes act as optical filters
confining, respectively, the blue and the red
optical carriers, while the green ones are
absorbed across both [5].
21
10
λL=650 nm (-6<V<+1 )
λL=450 nm
λG=550 nm
20
10
-3 -1
R/G (cm s )
19
0V
10
18
10
-6 V
17
10
0V
16
10
-6V
same optical components; and the electrical
components are typically less expensive than
the optical ones.
2.3 Optical characterization
The devices were characterized at 1 kHz
through spectral response (400-800nm) and
photocurrent-voltage
(-10V
<V
<+3V)
measurements. In Figure 2a it is displayed the
measured spectral photocurrent and in Figure 2b
the ac current-voltage characteristics under
illumination are shown. In this last measurement
three modulated monochromatic lights: R
(λR=650 nm); G (λG=550 nm) and B (λB=450 nm),
and their polychromatic combinations; R&G
(Yellow); R&B (Magenta); G&B (Cyan) and
R&G&B (White) illuminated separately the device
and the photocurrent was measured as a function
of the applied voltage.
15
10
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
Position (µm)
InOx
Photocurrent (µA)
I
(a-Si:H)
1000 nm
N
(a-SiC:H)
Front
diode
NP
-
Substrate
(a-SiC:H)
InOx
n (a-SiC:H)
I
(a-SiC:H)
200 nm
(a-SiC:H)
P
1.0
Back
diode
Electrical bias
+2 V
0V
-2V
-5 V
-8 V
-10V
2 kHz
0.8
0.6
0.4
0.2
0.0
400
Figure 1 a-SiC:H WDM device. a) Recombination profiles
(straight lines) under red (λR =650 nm) green (λG =550 nm)
and blue (λB =450 nm) optical bias and different applied
voltages (-6V<V<0V). The generation profiles are also shown
(symbols). b) Device configuration.
500
600
700
800
Wavelength (nm)
2.2 WDM working principle
Monochromatic beams together or a single
polychromatic beam (mixture of different
wavelength) impinge in the device and are
absorbed, accordingly to their wavelength (Figure
1), giving rise to a time and wavelength
dependent electrical field modulation across it [6].
In the multiplexing mode the device faces the
modulated
light
incoming
together
(monochromatic input channels), each one with
a specific transmission rate, and the electronic
signal is readout, under reverse bias. In the
multiplexing mode a polychromatic pulsed light
is projected onto the device and the readout
performed by shifting between forward and
reverse bias.
By reading out the photocurrent generated
by the incoming photons, the input information is
electrically multiplexed or demultiplexed and can
be transmitted again. Here, the (de)multiplexing
of the channels is accomplished electronically,
not optically. This approach offers advantages in
terms of cost since several channels share the
280
Photocurrent (A)
a)
1x10
-6
1x10
-6
8x10
-7
6x10
-7
4x10
-7
2x10
-7
0
-10
Yellow (R&G)
Magenta(R&B)
Cyan (G&B)
White (B&G&R)
Red (R)
Blue (B)
Green (G)
-8
-6
-4
-2
Voltage (V)
0
2
b)
Figure 2 a) Spectral photocurrent under different applied
voltages. b) ac IV characteristics under red (R); green (G)
and blue (B), yellow(R&G); magenta (R&B); cyan (G&B) and
white (R&G&B).
Results show that, as the applied voltage
changes from forward to reverse the blue/green
spectral collection is enlarged while the red one
remains constant (Figure 2a). The photocurrent
(Figure 2b) under red modulated light is
independent on the applied voltage while under
blue, green or combined irradiations, it increases
under reverse bias.
If the blue spectral component is present
(B&R, B&G), a sharp increase with the reverse
bias is observed. Under positive bias the blue
signal becomes negligible and the R&B, the G&B
and the R&G&B multiplexed signals overlap,
respectively with the R, the G and the R&G
signals. This behavior illustrates, under forward
bias, the low sensitivity to the blue component of
the multiplexed signal. It is interesting to notice
that under reverse bias the green signal has a
blue-like behavior, while under forward bias its
behavior is red-like confirming the green photons
absorption across both front and back diodes
(Figure 1b).
3.
Results and discussion
3.1 Influence of the frequency
In Figure 3a it is displayed, under reverse
bias (-5 V), the spectral photocurrent measured
at different light modulation frequencies. In
Figure 3b the multiplexed signals (solid lines)
obtained under different values of the
modulation light frequencies (1k Hz, 10 kHz and
100 kHz), at –5 V are shown. The superimposed
red and green dashed lines correspond to the
different input channels, respectively λL=650
and λL=550 nm, and are displayed just to
illustrate the different ON-OFF states of the light
bias.
3,5x10
f=15Hz
0.30
2000 Hz
1000 Hz
400 Hz
200 Hz
75 Hz
15 Hz
V=-5 V
-8
3,0x10
Photocurrent (A)
0.35
Photocurrent (µA)
-8
-8
2,5x10
-8
2,0x10
Results show that, under reverse bias and
low frequencies (f<400Hz), the spectral
response in the range of 550-650 nm (Figure
3a) increases with the frequency of the light that
impinges the device, suggesting different
capacitive effects during the device operation. In
the remaining regions it is independent on the
light modulation frequency.
The multiplexed signal (Figure 3b) shows that
in each cycle it is observed for the different
frequencies, the presence of four levels. The
highest occurs when both red and green
channels are ON (R&G) and the lower when
both are OFF (dark). The green level (G)
appears if the red channel is OFF and is lower
then the red level (R) that occurs when the
green channel is OFF. This behavior is
observed even for high frequencies although the
measured current magnitude is reduced.
In
Figure 4
the
ac
current-voltage
characteristics under different wavelengths: 650
nm (R); 450 nm (B); 650 nm & 450 nm (R&B),
and frequency regimes are displayed.
0.25
f=1.5 KHz
0.20
R&B
R
B
B
R
R&B
0.15
0.10
0.05
0.00
-4
-2
0
2
Voltage(V)
-8
1,5x10
Figure 4 – ac IV characteristics under R(λL=650 nm);
B(λL=450 nm), R(λL=650 nm) & B(λL=450 nm) modulated
light and different light frequencies (15Hz;1.5KHz)
-8
1,0x10
-9
5,0x10
0,0
400
500
600
700
Results show that in both regimes, under red
modulated light, the collection efficiency remains
independent on the applied voltage being higher
at high frequencies, as expected from Figure 3.
Under blue irradiation the collection do not
depend on the frequency regime and slowly
increases as the applied voltage changes from
forward to reverse.
800
Wavelength (nm)
a)
0,5
λL=650 nm
Photocurrent (A)
0,4
λL=550 nm
R&G
100 kHz
10 kHz
1 kHz
0,3
R
3.2 Bias sensitive
multiplexing
G
0,2
0,1
0,0
-1
0
Time (ms)
1
division
The effect of the applied voltage on the
output transient multiplexed signal is analyzed.
To readout the combined spectra, the generated
transient photocurrent due to the simultaneous
effect of two (λR=650 nm, λB=450 nm) and three
(λR=650 nm, λG=550 nm, λB=450 nm) pulsed
monochromatic channels was measured, under
different applied voltages.
Dark
-2
wavelength
2
b)
Figure 3 a) Spectral photocurrent under different light
modulation frequencies at -5 V. b) Wavelength division
multiplexing (solid lines) at – 5 V under different values of
the modulation light frequency. The red and the green
dashed lines guide the eyes into the input channels.
281
Figure 3b the ratios between the three
frequencies were always one half.
In Figure 4c de dependence of each pulsed
single channel with the applied voltage is also
displayed. As expected from Figure 2 the red
signal remains constant while the blue and the
green decrease as the voltage changes from
negative to positive. The lower decrease in the
green channel when compared with the blue
one can be ascribed to the decrease of the
photocurrent in the back diode under positive
bias (green red-like behavior).
Data show that the multiplexed signal depends
on the applied voltage and on the wavelength and
transmission rate of the each input channel.
Under reverse bias, there are always four
(Figure 4a) or eight (Figure 4b) separate levels
depending on the number of input channels. The
highest level appears when all the channels are
ON and the lowest if they are OFF. Furthermore,
the levels ascribed to the mixture of two input
channels (R&B, R&G, G&B) are higher than the
ones due to the presence of only one (R, G, B).
The step among them depends on the applied
voltage and channel wavelength. As expected
from Figure 2 and Figure 3, as the reverse bias
increases the signal exhibits a sharp increase if
the blue component is present. Under forward
bias the blue signal goes down to zero and the
red and/or green ones remain constants.
1.2
Red
1.0
Blue
Photocurrent (µA)
0.8
0.6
0.4
-5 V
0.2
V
0.0
+2 V
0.5
1.0
1.5
2.0
2.5
3.0
Time (ms)
a)
2.0
Green
1.8
Red
Blue
1.5
Photocurrent (µA)
1.3
-10V
1.0
0.8
-5V
0.5
0V
+3V
0.3
0.0
-2
-1
0
1
2
Time (ms)
b)
3.3 Bias sensitive
demultiplexing
Photocurrent (a.u.)
0.4
Red (+1V)
Green(+1V)
Blue (+1V)
Red (-10V)
Green(-10V)
Blue (-10V)
0.3
0.2
0.1
0.0
0.0
0.5
1.0
1.5
Time (ms)
c)
Figure 4 -Transient multiplexed signals at different applied
voltages and input wavelengths: a) R&B (λR,B=650nm, 450
nm). The highest frequency of the input signal is 1.5 kHz. b)
R&G&B (λR,G, B=650 nm, 550 nm, 450 nm). The highest
frequency of the input signal is 1 kHz c) dependence of the
colour channel with the applied voltage
The results are displayed, respectively, in
Figure 4a and in Figure 4b. The input
wavelength channels are superimposed in the
top of the figures to guide the eyes. The
reference level was assumed to be the signal
when all the input channels were OFF (dark
level). In Figure 3a the red frequency was 1.5
KHz and the blue one half of this value while in
282
wavelength
division
Different wavelengths which are jointly
transmitted must be separated to regain all the
information. These separators are called
demultiplexers. A chromatic time dependent
wavelength combination of red and blue (Figure
5a) or red, green and blue (Figure 5b) with
different transmission rates, were shining on the
device. The generated photocurrent was
measured under negative and positive bias to
readout the combined spectra.
If only two R and B channels are involved (four
levels; Figure 5a), under forward bias, the blue
component of the combined spectra falls into the
dark level, tuning the red input channel. Thus, by
switching between reverse and forward bias the
red and the blue channels were recovered and
the transmitted information regained.
If three R, G and B input channels with
different transmission rates are being used (eight
levels; Figure 5b), under reverse bias, the levels
can be grouped into four main thresholds,
ascribed respectively to the simultaneous overlap
of three (R&G&B), two (R&B, R&G, B&G), one (R,
G, B) and none (dark) input channels. Since
under forward bias, the blue component of the
multiplexed signal approaches the dark level
(Figure 3) the R, the G and the R&G components
are tuned. By comparing the multiplexed signals
λR=650 nm
3.0
λB=450 nm
Multiplexed signal (a.u.)
2.0
Magenta
junctions. A voltage source, V, has been applied
giving rise to a current I. Two ac current sources
with different frequencies, I1 and I2, are used to
simulate the input blue and red channels,
respectively. The green channel is simulated by
two ac current sources with the same
frequencies, I3 and I4, since the green light is
absorbed across both front and back intrinsic
layers (see Figure 1).
10uA 12uA
C
I
I
1000pF
V=-5 V
Q1
Red
1.0
I
V
I
I
0
Blue
0.0
Dark
V=+2 V
-1.0
-2.0
Input signals
under forward and reverse bias and using a
simple algorithm that takes into account the
different sub-level behaviors under reverse and
forward bias (Figure 2b) it is possible to split the
red from the green component and to decode
their RGB transmitted information. The digital
wavelength division demultiplex signals are
displayed on the top of both figures.
20uA
0.5
1.0
1.5
2.0
2.5
3.0
Q2
Time (ms)
8uA
I
2000pF
I
C
a)
λG=550 nm
λR=650 nm
Multiplexed signal (a.u)
3
λB=450 nm
White
2
Cyan
V=-10V
Dark
V=+3V
-1
8
-2
-1
0
1
2
3
I(I1)
7
Time (ms)
Figure 5 Transient multiplexed signals under negative and
positive bias. a) Polychromatic red and blue time dependent
mixture. b) Polychromatic red, green and blue time
dependent mixture. The digital wavelength demultiplexed
signal is displayed on the top of both figures.
I(I3);I(I4)
6
Photocurrent (µA)
b)
4.
a)
In Figure 7 it is shown the input channels, I(I1),
I(I2), I(I3) and I(I4); and the simulated multiplexed
signal, under forward and reverse bias. A good
agreement between experimental and simulated
results is achieved (Figure 3b).
Magenta
Red
0
I
Figure 6 a) Equivalent electrical circuit of the pinpin
photodiode..
Yellow
Green
Blue
1
I
Input signals
4
I(I2)
4
3
I
2
1
-1
0.5
Based on the experimental results (Figure 2,
Figure 3 and Figure 4) and device configuration
(Figure 1) an electrical model was developed and
supported by a SPICE simulation.
As displayed in Figure 6, the device is
considered to be as two phototransistors
connected back to back, modeling respectively
the a-SiC:H p-i-n-p and a-Si:H n-p-i-n sequences.
One transistor, Q1, is pnp type and the second,
Q2, npn. In order to simulate the n-p internal
junction, the collector and bases of both
transistors are shared. Capacitors, C1 and C2,
are used to simulate the transient capacitance
due to the minority carrier trapped in both p-i-n
283
Blue
-10V
+0.6V
0
Electrical WDM simulation
Red
Green
5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
Time (ms)
Figure 7 Signals obtained using SPICE simulation under red
(I2), blue (I1) and green (I3, I4) pulsed lights and different
applied voltages.
Results show that, if the device is biased
negatively, Q1 and Q2 are in their reverse active
regions. The p-n internal junction is forwardbiased and the external voltage drops across both
front and back reverse-biased junctions, mainly at
the front one due to its higher resistivity. So, the
external current, I, depends not only on the
balance between blue, green and red
photocurrents (I1, I2, I3, I4)) [7] but also on the
end of each half-cycle of each modulated current.
Here, the movement of charge carriers with an
increase/decrease in the irradiation, results in a
charging or a displacement current similar to the
current that charges the capacitors C1 and C2 in
opposite ways.
Results reveal that the device acts as a
charge integrator, keeping the memory of the
input channels. Under reverse bias, when all the
channels are simultaneously ON (Figure 4b), I1
and I3 flow across Q1 collector towards the base
of Q2 and together with I2 and I4 give rise to the
highest signal. When all the channels are
simultaneously OFF (Figure 4b), the current is
limited by the leakage current of both active
junctions (dark level). If only the blue channel is
ON the carriers generated by the blue photons
are injected into the base of Q1. C1 charges
positively and C2 negatively as a reaction to the
decrease in the red and green irradiations. The
opposite occurs if only the red channel is ON.
Nevertheless, if the only channel ON is the green
one, both front and back contribution must be
considered and the photocurrent is a balance
between the blue- and the red-like contributions
(Figure 1). If only two input channels are ON
(R&B, R&G, G&B) both front and back
generations are taken into account. So, once
triggered (R&B&G), the device continues to
conduct until the current through it drops below a
certain threshold value, such as at the end of a
half-cycle, keeping the information of the
wavelength (R, G, B) and transmission rate
(frequency) of the impinging light.
When a positive voltage is applied to turn the
internal junction from ON to OFF, the junction
capacitance across the internal n-p junction is
charged. The charging current flows through the
emitter of the two transistors. The device behaves
essentially as a npn phototransistor with the pnp
transistor acting like a emitter-follower with a very
small gain. So, under lower positive voltages, the
only carriers collected come from the red and/or
green channels enabling the demultiplexing of the
previous multiplexed signal (Figure6).
Comparing both the experimental and the
simulated results it is observed that, under
negative applied voltages, the multiplexed signal
keeps the memory of the single input channels.
By switching between forward and reverse bias
the red, green and the blue channels were
recovered.
284
5.
Conclusions
Results on the applicability of a multilayered
double p-i-n a-SiC:H/a-Si:H heterostructures, as
WDM devices in the visible spectrum, were
presented.
Three modulated input channels were
transmitted together, each one located at different
wavelength and frequencies. The combined
optical signal was analyzed by reading out the
photocurrent generated across the device.
Results show that by switching between
positive and negative voltages the input channels
can be recovered. A physical model supported by
an electrical simulation gives insight into the
device operation.
References
[1]
W.
Daum,
J.
Krauser,
P.E.
Zamzow,
O.Ziemann.“Polimer Optical Fibers for Data Communication“
Springer-Verlag, 2002.
[2]
S. Randel, A.M.J. Koonen, S.C.J. Lee, F. Breyer,
M. Garcia Larrode, J. Yang, A. Ng'Oma, G.J Rijckenberg,
H.P.A. Boom. “Advanced modulation techniques for polymer
optical fiber transmission”. proc. ECOC 07 (Th 4.1.4). Berlin,
Germany (2007) 1-4.
[3]
M. Haupt, C. Reinboth and U. H. P. Fischer.
“Realization of an Economical Polymer Optical Fiber
Demultiplexer”, Photonics and Microsystems, 2006
International Students and Young Scientists Workshop,
Wroclaw, 2006.
[4]
A. Fantoni, M. Vieira, R. Martins, Mathematics and
Computers in Simulation, Vol. 49. (1999) 381-401.
[5]
P. Louro, M. Vieira, Yu. Vygranenko, A. Fantoni,
M. Fernandes, G. Lavareda, N. Carvalho, Mat. Res. Soc.
Symp. Proc., 989 (2007) A12.04.
[6]
M. Vieira, M. Fernandes, J. Martins, P. Louro, R.
Schwarz, and M. Schubert, IEEE Sensor Journal, 1 (2001)
158-167.
[7]
Edward S. Yang, “Microlectronic devices”, Cap. 5,
Department of Electrical Engineering, Colombia University,
Mc Graw-Hill, Inc. 1988.