Good morning everyone. I’m Yasushi Takahashi in the Institute for Solid
State Physics, university of Tokyo. We carry out the research in
collaboration with Lucent Technology. In this presentation, I would like
to speak about absolute value of absorption coefficient for single
quantum wire. First I will explain the background and motivation to
measure the absorption spectra for quantum wires. Next, I will talk about
sample structure and fabrication process of single quantum wire. Next,
about transmission experiment to measure the absorption spectra of
quantum wire, and then, I will speak about experimental results of
absorption spectra at 5 K and room temperature. Lastly, I will summarize
my presentation.
In quantum wires, large exciton absorption at the band edge have been
expected due to enhanced density of states and enhanced excitonic effects
in 1D systems. These physisc have intensively studied in V-groove
quantum wire structure or T-shaped quantum wire structure after seminal
papers by Kapon or Wegscheider and coworkers. We have studied
T-shaped quantum wires. However, there are no reports of absolute value
of absorption coefficient in quantum wires. This is mainly because the
volume of quantum wires is too small to measure the absorption spectra
using ordinary transmission measurements with perpendicular
configuration of wires and transmission light. And so we used waveguide
transmission measurement with parallel configuration of wires and
transmission light.
This viewgraph shows the sample structure of single wire embedded in
an optical waveguide. Percentages mean Al content of AlGaAs. T-wire is
formed at this T-shaped intersection of two quantum wells. We call this
well, stem well and this well, arm well. T-wire size is 14 x 6 nm. These
contour curves show constant probability for electrons confined in a
T-wire. This structure is fabricated by the cleaved edge overgrowth
method, which consists of two step MBE growth separated by in situ
cleavage. In the first MBE growth, these layers, 50% AlGaAs optical
cladding layer, stem well, and optical cladding layer were made at 600
centigrade. After the cleavage along [110] surface, arm well, and the
other layer to form optical waveguide are made. This second MBE
growth was executed at 490 centigrade. After the growth of an arm well,
annealing at 600 centigrade for 10 minutes is added. This process
improves quality of T-wire significantly. In this picture, the light colored
layers have higher refractive index. Therefore, this region acts as optical
waveguide, where T-wire is embedded. We calculated optical
confinement factor, which is overlap ratio between the lowest optical
mode and single T-wire is very small. It is smaller than multiple quantum
well lasers by the factor of 400. This disadvantage may become critical
drawback for optical devices. In this point, it is important to perform the
direct absorption measurement for quantum wire device. The cavity
length is 500 um. Two cavity mirrors are left uncoated.
This presents PLE spectra of T-wires at 5 K. PLE measurement is useful
to investigate spectral shape of absorption, but isn’t useful to measure the
absorption coefficient. As you can see, the lowest energy peak
corresponds to 1D exciton ground state, and this flat absorption is due to
1D continuum states. The purpose of my sutdy is to measure the absolute
value of this peak.
This figure shows the configuration of waveguide transmission
experiment. The key point in this experiment is very simple, to make the
transmission light interact 500 um long T-wire. This method compensate
for small volume of T-wire. When Input power I0, transmitted light
intensity I is attenuated by exp(-L) in single pass. L is cavity length of
500 um and the modal absorption coefficient alpha is our goal. Tunable
Cw TiS laser was used as a transmission light source. Laser was focused
to about 1 um diameter spot by this objective lens, and the transmitted
light through the waveguide was collected by this objective lens. In order
to eliminate the background light, we used single core with confocal
microscope configuration. We measured incident intensity I0 removing
the sample and transmitted intensity inserting the sample. We obtained
transmittance I/I0. Two polarization of transmitted light were used. Arm
polarization is parallel to the arm well and stem polarization is parallel to
the stem well.
This viewgraph shows the transmittance spectrum at 5 K for arm
polarization. X-axis is photon energy, y-axis is transmittance in log scale.
Strong attenuation is observed in this energy. Below this energy, clear
and flat Fabry Perot oscillations are observed due to reflection. The
transmittance include reflection is given by this equation. We can
translate this spectrum into the absorption spectrum using this formula. 
and is unknown value.  is modal absorption coefficient and  is
coupling efficiency between input laser and optical waveguide. Next, I
will explain Analysis of the Fabry Perot fringes enables estimation of the
coupling efficiency.
This shows the magnification of Fabry Perot fringes. There is no
absorption by T-wire but intrinsic waveguide loss exists. We can derive
the loss using Hakki & Paoli method, which is independent of coupling
efficiency and so we can introduce the waveguide loss of 3.5 cm-1.
Substituting this value into this formula, we can derive coupling
efficiency of 0.24. In this way, we can correctly derive the absorption
spectrum from the transmittance spectrum.
This graph is absorption spectrum of a single wire at 5 K for arm
polarization. Longitudinal axis is modal absorption. Firstly, We obtained
the same spectral shape as PLE spectrum. Therefore, this peak is due to
1D exciton ground state, and this flat structure is due to 1D continuum
states, and increasing absorption above this energy is due to the arm well.
In this peak, a full width at half maximum is 1.6 meV. This indicates the
high uniformity of our quantum wire in the whole cavity region.
Secondly, the absorption coefficient of 80 cm-1 gives transmittance of
1.8 % by a single pass of the 500 um waveguide. In other words, 98% of
transmission light is absorbed by a 500 um long T-wire. This
demonstrates a strong excitonic absorption of the single T-wire in spite of
the small optical confinement factor. Third point, the absorption of 1D
continuum is much smaller than 1D exciton peak, and the inverse square
root singularity at the band edge is absent as theoretically predicted.
Next, we tried to measure the absorption spectra at room temperature
because room temperature excitonic absorption value is very important
for practical optical devices. However, we were not able to measure it for
single wire device. Thus, we used 20 wire device.
This is the structure of 20 multiple T-wires. The size of T-wire is the
same as single wire device. 20 T-wires are embedded in an optical
waveguide in order to increase the optical confinement factor. That value
is 9.3 times larger than that of single T-wire device. Coupling efficiency
in transmission measurement also increases to 0.4 due to larger
waveguide size. We measured room temperature absorption spectra using
this device.
This is the result at 297 K. The solid curve represents absorption for arm
polarization. Strong and sharp absorption peak is observed. We believe
this is the first observation of room temperature 1D excitonic absorption.
The absolute value of 1D exciton peak is 160 cm-1. This value
corresponds to transmittance of 2.8x10-4 by a single pass of the 500 um
waveguide. This demonstrates that the quantum wires in an optical
waveguide have a strong excitonic absorption even at room temperature.
Therefore, quantum wires are applicable to optical devices. A full width
at half maximum in this peak is estimated as 7.2 meV. The dotted curve
represents absorption for stem polarization. Stronger absorption for arm
polarization indicates that T-wires have polarization dependence similar
to the arm quantum well. This characteristic originates from the strong
quantum confinement in the direction of [110].
This graph shows absorption spectra for various temperatures. As
increasing temperature, 1D exciton peak at 5 K changes continuously to
this room temperature peak. Therefore, this experiment proves this
absorption peak is due to 1D exciton.
In summary, we measured absorption spectra for T-shaped quantum wires
using straightforward waveguide transmission measurement. The modal
absorption coefficient of 1D exciton for single T-wire is 80 cm-1 at 5 K.
Room temperature 1D exciton absorption was observed. Absorption
coefficient for 20 T-wires is 160 cm-1. T-wire exciton has large
polarization anisotropy. I hope these results support applications of
quantum wires to practical optical devices.
Thank you for your attention.