2308IEEE PHOTONICS TECHNOLOGY LETTERS, VOL. 16, NO. 10, OCTOBER 2004
First Integrated Combiner Based on Self-Switching
in Quantum Dots
E. A. Patent, J. J. G. M. van der Tol, P. R. A. Binetti, Q. Gong, Y. S. Oei, R. Nötzel, J. E. M. Haverkort,
P. J. van Veldhoven, J. H. Wolter, and M. K. Smit
(QDs). The switching is studied in an integrated optical combiner
circuit. This is a Mach–Zehnder interferometer with an unequal
power distribution over the branches. The refractive index is in-
nonlinearities are measured. This device is aimed to avoid an un-
wanted 3-dB loss, fundamental in passive optical splitters.
Index Terms—Integrated combiner, nonlinearities, quantum
branches , as is shown in Fig. 1. When the light signal enters
into one of the input ports, it is unevenly distributed over the
0.85 inourcase.Furthermore, thereisa fundamental
difference between the signals at the output ports of the cou-
pler. If this phase difference can be compensated by the induced
nals from the branches are in phase and constructively interfere
at theoutput port.The optimum recombination is thenobtained.
In this way, the transmission can be improved by more then
2 dB with respect to a passive 3-dB splitter.
For the combiner function, the required effect is self-phase
modulation. In this effect, an optical signal changes the refrac-
tive index for its own propagation, i.e., in the nonlinear phase
terials, however, the all-optical nonlinearities are requiring too
high input power levels, making them unsuitable for use in pho-
tonic integrated circuits. Semiconductor optical amplifiers have
shown their applicability ,  as nonlinear phase shifters, but
they require an external current source that makes the combiner
function active. Since the combiner is a passive function, the
preference is to have purely passive nonlinear all-optical effects
HE DEVICE discussed in this letter is a Mach–Zehnder
interferometer (MZI) with nonlinear phase shifters in the
Manuscript received February 17, 2004; revised June 9, 2004. This work was
supported by the Dutch Technology Foundation STW.
E. A.Patent,J.J. G.M.vanderTol,P. R.A.Binetti,Y.S.Oei, andM.K.Smit
are with Eindhoven University of Technology, eiTT/COBRA Inter-University
Research Institute, Opto-Electronic Devices Group, 5600MB Eindhoven, The
Netherlands (e-mail: email@example.com).
Q. Gong, R. Nötzel, J. E. M. Haverkort, P. J. van Veldhoven, and J. H. Wolter
are with Eindhoven University of Technology, eiTT/COBRA Inter-University
Research Institute, Semiconductor Physics Group, 5600MB Eindhoven, The
Digital Object Identifier 10.1109/LPT.2004.833952
Nonlinear phase shifter.
Schematic layout of the low-loss optical combiner circuit. NL PS:
Quantum dots (QDs) in semiconductors are a candidate for
passive nonlinear phase shifters. They are expected to provide
improved all-optical nonlinearities due to their delta-function
density of states. This results in sharp excitonic absorption
peaks with considerably larger peak absorption then in a bulk
or quantum-well structures. When the QD is filled with a single
electron-hole pair, the ground state becomes transparent, while
two electron-hole pairs within a single dot already generate
optical gain. This behavior is expected to result in a very
small switching energy in QDs , which can result in a large
refractive index change.
II. DESIGN AND FABRICATION
The QD sample was grown by chemical beam epitaxy on a
(100) oriented InP substrate. The QDs were prepared with the
at 500 C on top of a lattice-matched Ga In
. A thin GaAs layer was inserted un-
derneath the QDs allowing reproducible tuning of the emission
wavelength in the 1.55- m range .
/cm . The single QD layer is embedded in a
(1.25) InGaAsP waveguide core (film layer),
which is covered with an InP cladding of 1.3 m.
For the waveguide masking, a SiN -layer with a thickness of
50 nm was used. The ridge waveguides were etched employing
an optimized CH –H reactive ion etching technique alternated
with an O descumming process to remove polymer deposi-
tions. The etching depth into the film layer is 100 nm (Fig. 2).
The measured photoluminescence (PL) spectrum at room
temperature is presented in Fig. 3.
It can be seen that the peak corresponding to the InGaAsP
layer is at 1.29
m, whereas the emission from InAs QDs is
centered at around 1.52
m. The broad peak of the PL of the
QDs reflects the variation in dots size, not the behavior of indi-
vidual dots, which should show a very sharp emission.
For couplers, multimode interference (MMI) components are
used. They are polarization insensitive and compact. The length
1041-1135/04$20.00 © 2004 IEEE
PATENT et al.: FIRST INTEGRATED COMBINER BASED ON SELF-SWITCHING IN QDs2309
Fig. 2. Waveguide profile in the QD sample.
Fig. 3. PL spectrum at the room temperature of the QD sample.
of the multimode waveguide
of the half beat length
the coupling ratio
. In our experiment,
10 m. The 3-dB MMI output splitter has a length of
. In the experiment,
10 m. All the waveguides in the circuit have a width of 3 m
and the bends have a radius of 500
width of 2 m to increase the photon density and, thus, the all-
ponents were fabricated in the same layer stack. Because the
nonlinear properties are accompanied by an absorption effect,
this increases the total loss of the device.
can be described in terms
. The unbalanced coupler with
(Fig. 1) has a length of
m with a width of
m with a width of
m. Phase shifters have a
III. MEASUREMENT RESULTS
Prior to the characterization of the combiner circuit, mea-
surements of test structures processed in the same run for
on-chip measurements were performed. Waveguide propaga-
tion losses were measured using the Fabry–Pérot technique
. In these measurements, a distributed feedback laser oper-
ating at 1534.8 nm is scanned in wavelength by changing its
temperature. The resonances in the transmitted light through a
waveguide are analyzed to obtain a value of loss in decibel per
centimeters. For the transverse-magnetic (TM)-polarized light,
3- m-wide waveguides show losses of 5.5–7 dB/cm. Losses
for transverse-electric-polarized light are very high as the
material is polarization dependent due to strain created in QDs
during growth, and also due to their shape. In this letter, only
TM-polarized light will be considered. The measurement setup
transmission in the interferometric direction, black line, in the splitter direction.
(a) Ports 1–3. (b) Ports 2–3.
Transmission signals versus input power. Gray line represents the
suffers from polarization instability because of some length of
standard single-mode fibers used. Therefore, detection of the
input and output power was done within a short time interval
(less then half a minute) in which changes in polarization in the
measurement setup could be neglected. The light signal is cou-
pled in and out of the chip by two microscope objectives. The
light detection system contains a photodetector and a lock-in
amplifier with a spatial filter in order to filter out the scattering
light and light propagating through the slab waveguide (film
Transmission measurements show a bleaching effect, pos-
sibly caused by free carriers. As the measurements are per-
formed at a wavelength of 1550 nm, which is still within the
PL peak of the QDs (see Fig. 3), the all-optical refractive index
change cannot be expected without a bleaching component .
Higher input signal intensities lead to an increase of the wave-
guide transmission due to depletion of free carriers. This issue,
however, needs further investigation.
The combiner circuit was characterized as follows. First,
transmission signals from Ports 1 to 3 and from Ports 2 to 3
(Fig. 1) were measured as a function of input power. In this
direction, the circuit should show switching behavior due to
unequal distribution of light over the MZI branches. Next,
transmissions from Port 3 to Ports 1 and 2 were measured.
In this direction, the device behaves like a splitter because of
equal distribution of power over the MZI branches. Thus, no
switching behavior should be seen. By comparing the signals
from two opposite directions, all unwanted effects (like, e.g.,
2310IEEE PHOTONICS TECHNOLOGY LETTERS, VOL. 16, NO. 10, OCTOBER 2004 Download full-text
directions. (a) Ports 1–3. (b) Ports 2–3.
Ratios of the transmission signals propagating in the opposite
bleaching) can be excluded. The measurement results are
presented in Fig. 4.
As can be seen from Fig. 4(a) and (b), transmission curves
and are different. This can be explained by the ex-
citation of the first-order mode in the 3-dB output splitter, re-
ported previously in , or by a linear phase shift between the
branches due to fabrication imperfections, etc. In the “splitter”
direction, i.e., injection through Port 3, the curves show no sign
of interference, but only the bleaching also obtained for straight
waveguides. In the other, “interferometric” direction, however,
can only be attributed to a nonlinear switching.
In order to single out the interferometric effect, transmission
curves measured in opposite directions are divided (Fig. 5).
Improvement of the transmission in the interferometric di-
rection with respect to the splitter direction can be seen from
the graphs presented above. These curves represent ratios of
the transmission signals propagating in opposite directions. The
sults in a horizontal line, again illustrating the presence of an
optical power-induced interference effect.
The maximum improvement of the transmission versus input
power is 1.76 dB. From Fig. 5, it is seen that a maximum in the
transmission is just reached with the available power. This im-
plies a phase shift between the branches of more than
These results are obtained for the power levels of a few milli-
watts. This is based on an estimate of coupling losses (3–4 dB)
and propagation losses in the chip (3–5 dB).
In this letter, we have shown a device that can avoid a funda-
mental 3-dB loss of a passive optical splitter–combiner. The op-
tical combiner is an MZI with unequal distribution of light over
the branches. These branches contain nonlinear phase shifters
where an induced
-phase shift is required. We have shown
such a device in the QD material system and obtained an im-
provement of up to 1.76 dB (with respect to a 3-dB splitter)
in transmission due to all-optical switching. The mechanism
of this switching is not yet fully clear. Previous results show
that nonlinear effects in InGaAsP or quantum wells (such as
the wetting layer ion the QD-sample) are much too small to ex-
plain the observed effects. Therefore, we attribute the nonlinear
switching to theQDs. Measurements usingsignals at two wave-
lengths—pump and probe—are prepared to study these mecha-
nisms of the nonlinear effects in QDs.
 J. J. G. M. van der Tol, H. de Waardt, and Y. Liu, “A Mach–Zehnder
interferometer-based low-loss combiner,” IEEE Photon. Technol. Lett.,
vol. 13, pp. 1197–1199, Nov. 2001.
 E. A. Patent, J. J. G. M. van der Tol, and N. Calabretta, “A pattern effect
Brussel, Dec. 3, 2001, pp. 233–236.
 J. E. M. Haverkort, R. Prasanth, A. Deepthy, E. W. Bogaart, J. J. G. M.
van der Tol, E. A. Patent, G. Zhao, Q. Gong, P. J. van Veldhoven, R.
Nötzel, and J. H. Wolter, “Wavelength insensitive all-optical switching
in quantum dots,” in Proc. 2003 Annu. Symp. IEEE/LEOS Benelux
Chapter, Enschede, The Netherlands, Nov. 20–21, 2003, pp. 193–196.
 R. Prasanth, “Photonic switching in III/V nanostructures,” Ph.D. disser-
tation, Eindhoven Univ. of Technology, The Netherlands, 2004.
 Q. Gong, R. Nötzel, P. J. Veldhoven, T. J. Eijkemans, and J. H. Wolter,
“Wavelength tuning of InAs quantum dots grown on InP (100) by chem-
ical-beam epitaxy,” Appl. Phys. Lett., vol. 84, no. 275, pp. 275–277,
 M. Bachmann, P. A. Besse, and H. Melchior, “Overlapping-image
multimode interference couplers with a reduced number of self-images
for uniform and nonuniform power splitting,” Appl. Opt., vol. 34, pp.
 H. P. Zappe, Introduction to Semiconductor Integrated Optics.
wood, MA: Artech House, 1995.
 C. Vázquez, C. Aramburu, M. Galarza, and M. López-Amo, “Experi-
mental assessment of access guide first-order mode effect on multimode
interference couplers,” Opt. Eng., vol. 40, no. 7, pp. 1160–1162, July