Subdiffraction Photon Guidance by
Chia-Jean Wang,†Ludan Huang,‡Babak A. Parviz,†and Lih Y. Lin*,†
Department of Electrical Engineering, Department of Physics,
UniVersity of Washington, Seattle, Washington 98195
Received August 21, 2006; Revised Manuscript Received September 26, 2006
We report on a waveguide composed of a cascade of gain-enabled quantum dots with subwavelength dimensions. Fabrication is demonstrated
through DNA-mediated self-assembly and a two-layer molecular self-assembly process that enables rapid prototyping. The device, which is
identified with fluorescence microscopy and tested by optical near field detection, allows optical signal transfer at a well-defined wavelength
in flexible routing geometry, such as straight paths and 90° bends. The structure serves as a critical building block for nanophotonic systems
with high integration density.
Realization of ultrahigh-density photonic integrated circuits
requires the ability to guide light at subdiffraction limit
dimensions without extensive loss due to sharp waveguide
bending. To fully capture the inherent advantages of large
bandwidth, capacity, and high-speed modulation in optical
communication at the submicrometer scale, structures such
as the silicon-on-insulator high index contrast,1one-
dimensional negative dielectric,2photonic crystal,3,4nano-
wire, and nanoribbon5waveguides provide a wealth of
options. However, the same techniques may still be subjected
to the diffraction limit, may be difficult to integrate with
optical and electronic components on chip, and cannot
maintain low-loss guiding at sharp junctions. Alternatively,
metal nanoparticle arrays, which are made of Ag, Au, or Al
assembled through sputtering, ion implantation, thermal
evaporation, and lift-off,6-9present a nanoscale method of
guiding energy via plasmon excitation. Still, optical energy
transmission through these metal nanoparticles is subject to
resistive heating and loss and necessitates conversion to
translate plasmon energy back to optical energy. In keeping
with the low dimensionality of nanoparticles and harnessing
a gain-enabled material, we elaborate upon the principle of
operation, methods for fabrication, and demonstration of a
quantum-dot waveguide including both straight and corner
formations. The device expands the toolbox of nanophotonic
components to further allow energy transmission over sharp
bends and through confined spaces.
The quantum dot (QD), which exhibits three-dimensional
(3D) electron-hole (e--h+) pair confinement leading to
sharp emission peaks largely tuned by particle size and
composition, have enabled applications spanning detection
of biomolecules10,11to lasers with lower threshold current
density compared to those made from bulk material.12-14In
colloidal form, QDs may be modified with various surface
chemistries useful for self-assembly processes requiring
precise attachment schemes. Furthermore, in an appropriate
semiconductor system, the particle reveals a gain mechanism
that uses a pump source to create population inversion of
excited electrons to generate net emission over absorption,
similar to optical amplifying action.15Consequently, the
propagating optical signal may not only experience less
transmission loss but increase in strength as well. Capitalizing
on the specified advantages, the quantum-dot waveguide
consists of an array of densely packed quantum dots anchored
to a substrate via self-assembly. Operation of the device,
depicted in Figure 1A, requires a pump laser of photon
energy greater or equal to the separation of the second
electron and hole states to be placed overhead to excite
electron-hole pairs and enable gain. For a three-state system,
the e--h+pairs quickly relax to the first conduction- and
valence-band states, where without further manipulation they
will likely recombine for spontaneous emission of pho-
tons.16,17However, aligning a signal laser of energy equiva-
lent to separation between the first electron and hole state
to the input edge of the waveguide, photons generated by
stimulated emission may cascade through the device creating
an amplified output, moderated by the interdot coupling
efficiency and the gain available in each dot. The FDTD
simulation of the Poynting vector distribution, indicating
photon transmission behavior in a QD cascade waveguide,
is provided in Figure 1B.18The parameters match test
conditions such that we use 10-nm-diameter quantum dots,
which approximate the size of the core/shell nanoparticles,
and an input signal with wavelength at 633 nm and
polarization in the y direction. Upon absorption of both pump
* Corresponding author. E-mail: firstname.lastname@example.org.
†Department of Electrical Engineering.
‡Department of Physics.
Vol. 6, No. 11
10.1021/nl061958g CCC: $33.50
Published on Web 10/24/2006
© 2006 American Chemical Society
and signal light, the accepting QDs produce energy as
photons via stimulated emission, with the majority of energy
aimed in the propagation direction to be absorbed by the
neighboring QD causing a cascade of energy transmission.
Hence, the Poynting vector magnitude illustrates that the
power per unit area lies in the path of forward propagation.
Compared to other possible analogous designs where QDs
are formed through lattice mismatch between epitaxially
grown layers19and waveguides are defined by etching the
dielectric layer, the proposed design achieves simpler
fabrication and eliminates the minimum width constraint to
achieve guided mode approximated by the transmission
wavelength from conventional dielectric waveguiding. To
note, prior modeling of the gain response for 5 × 5 × 5
nm3CdSe/ZnS core/shell quantum dots, with an additional
5 nm shell thickness, indicates that there is net gain occurring
between 0.1 and 1 nW optical pump power, with absorption
dominating at the low end and gain saturation at the higher
end.17In addition, implementing an ABCD transfer matrix
technique allows estimation of the relative output intensity
with respect to the photon coupling coefficient between two
adjacent QDs and the gain parameter inherent in each. As
an example, the simulation results for a waveguide composed
of five quantum dots in a 1D array (Figure 1C) indicate a
tradeoff between the gain and coupling coefficient where a
smaller value for one may compensate a higher value for
the other to attain high transmission efficiency.
Fabrication of our nanophotonic waveguides has been
accomplished through both DNA-mediated self-assembly and
a simpler two layer molecular self-assembly procedure
relying on specific covalent bonds. The DNA process enables
programmable deposition of quantum dots to serve as a
template for simultaneous assembly of multiple QD type
waveguides. The procedure, shown in Figure 2A, begins with
spin-coating polymethylmethacrylate (PMMA) on a silicon/
silicon dioxide substrate and using electron-beam lithography
(EBL) to pattern trenches on the surface of PMMA, thus
defining the global shape of the waveguide. We increase the
number of exposed hydroxyl groups at the bottom of these
trenches by a short exposure to oxygen plasma. The -OH
groups act as an anchor for gas-phase self-assembly of
3-mercaptopropyltrimethoxysilane (MPTMS) molecules that
provide exposed thiol (SH) termini for the subsequent
binding step. The introduction of acrylamide-modified single-
stranded DNA (ss-DNA) to the thiol groups enables specific
binding of the ss-DNA to the bottom of the PMMA trenches,
which is then free to bind to the exact complimentary chain
creating a highly selective binding location. Prior to the DNA
hybridization step, immersion of the sample in buffered
acrylic acid passivates the nonreacted MPTMS20and then a
Figure 1. QD nanophotonic waveguide. (A) Operation: a pump
source excites electrons to the second energy state in the conduction
band of the quantum dot to generate gain, and a signal light
corresponding to the first excited state is placed at the waveguide
edge to cause stimulated emission of photons that propagate
downstream to produce an output signal mediated by the inter-dot
coupling. (B) Poynting vector distribution of the quantum-dot
cascade. (C) Calculated relative output intensity as a function of
gain per quantum dot for a 1D array waveguide using varied
coupling coefficients, η.
Figure 2. DNA-mediated self-assembly of QD waveguides. (A)
EBL pattern PMMA-coated substrate, treat with O2plasma; deposit
MPTMS monolayer; covalently bind with 5′-acrydite-DNA; hybrid-
ize with biotin-modified cDNA; bind streptavidin-QDs to biotin-
cDNA sites, remove PMMA with dichloromethane. (B) XPS results
confirming MPTMS deposition through sulfur content on substrate.
(C) Fluorescence micrograph of 5 µm × 1 µm, 2 µm × 1 µm, and
1 × 1 µm2patterns and (D) 1-µm-width waveguide.
Nano Lett., Vol. 6, No. 11, 2006
droplet of the custom-synthesized biotinylated complemen-
tary DNA (cDNA) solution is placed onto the surface. Next,
we introduce 655 nm emission quantum dots conjugated to
streptavidin, a protein that preferentially attaches to biotin.
In the final step, a dichloromethane rinse dissolves the
PMMA to leave the QD pattern secured to the substrate by
linking chemistries. To check the deposition of the self-
assembled molecular (SAM) layers, we compared exposed
to control samples using X-ray photoelectron spectroscopy
(XPS). Only the surfaces treated with MPTMS exhibit a
sulfur signature (in Figure 2B), with the average percent
compositions at 58.6% oxygen, 28.3% silicon, 12.4% carbon,
and 0.7% sulfur. Alternatively, the untreated samples yielded
an average of 63.0% oxygen, 29.6% silicon, and 7.4%
carbon. Further assembly steps using DNA are confirmed
with fluorescence observation and examples of the final QD
fluorescence patterns, in Figure 2C and D, include an array
of rectangles ranging in size from 5 µm × 1 µm down to 1
× 1 µm2, and a 1-µm-wide waveguide.
The second technique diverges from the DNA-based
procedure after the generation of Si-OH groups on the
surface. Instead, we deposit 3′-aminopropyltriethoxylsilane
(APTES) by immersion in solution to create a surface
expressing amines for further chemical reactions.21A droplet
of carboxylated quantum dots suspended in 1x phosphate
buffer solution (PBS) mixed with 1-ethyl-3-(3-dimethylami-
nopropyl)-carbodiimide (EDC), an amine-carboxyl coupling
reagent, is applied to the sample as the second layer. After
waiting for a minimum of 1 h, during which the QDs bind
to the sample, excess material is rinsed off with 1x PBS
buffer and the PMMA is also removed with dichloromethane
to reveal the waveguides. As confirmation of process, XPS
in Figure 3B reveals APTES deposition such that the
resulting percent compositions are 1.2% nitrogen, 29.8%
silicon, 54.5% oxygen, and 14.4% carbon, whereas the
control gave an average of 60.6% oxygen, 32.3% silicon,
and 7.1% carbon. The ratio of the nitrogen peak to the other
dominant elements is appropriate to coverage by APTES,
and lack of nitrogen in controls provides further verification.
Fluorescence micrographs and AFM images, in Figure 3C
and D, show attachment of 655 nm emission quantum dots
aligned in the 100- and 500-nm-width lines. An additional
atomic force micrograph is shown in Figure 3C to provide
more details on the QD shape and distribution within the
waveguide. To note, the 100 nm waveguides are diffraction-
limited in terms of optically detecting discontinuities.
However, the topographical profile demonstrates a continu-
ous structure. It is noted that advanced optical lithography
can be used for defining the waveguide patterns in both
fabrication methods for mass production, if the resolution is
sufficient. In either case, the waveguide is defined by the
design file and EBL where nominal tolerance is within 5%
of the specified linewidths. Because the APTES method is
subject to fewer intermediate stages, the packing is denser
where QDs are separated by about 5-10 nm where that for
DNA may range up to an order of magnitude more. More
investigation is required to extract the coupling coefficients
in the 2D waveguide as for our case because of the
complexity caused by variation of separation between QDs
at a range of directions in the forward path.
The waveguiding behavior of the quantum-dot cascades
was determined by aligning a pair of tapered fiber probes to
the device. Coarse adjustment was done with the aid of gold
alignment markers, and fluorescence imaging of the waveguide
under pump light illumination allowed us to precisely identify
the device and further tune the probe position. The designated
input fiber is coupled to a 639 nm signal laser, which falls
within the gain peak centered at 655 nm, and the output taper
connects to a photodetector with a nanowatt detection limit.
The pump light, with 405 nm wavelength, is collimated and
oriented overhead to illuminate the device and excite e--h+
pairs to create gain. With a 20x objective and the collimator
in place, we calculated a maximum pump-power distribution
of 1.4 nW/QD, corresponding to 4 mW over a 20-µm-
diameter illumination area, which is within the gain regime
determined by the theoretical model.17Subsequently, ramping
the pump level up and down while logging the photodetector
Figure 3. Two-layer self-assembly of QD waveguides. (A) EBL
pattern PMMA-coated substrate, treat with O2 plasma; deposit
APTES monolayer; covalently bind carboxylated QDs to the amine
terminal group, remove PMMA with dichloromethane. (B) XPS
confirming APTES deposition through nitrogen content. Fluores-
cence and corresponding atomic force micrograph of (C) 500 nm
with inset depicting granular QD shape and (D) 100 nm wide
waveguides in (i) single and pair formations spaced (ii) 200 nm
and (iii) 500 nm apart. Scale bar is 1 µm in length.
Nano Lett., Vol. 6, No. 11, 20062551
measurements over time, with the signal laser on and off
and the fibers aligned on and then off the waveguide, for a
total of four cycles allowed us to determine the effect of the
waveguide in transmitting light.
Each combination of pump and signal parameters is polled
for every 5 seconds over a 10-min duration with the
collection of values averaged for further analysis. On the
whole, the standard deviation is 2 orders of magnitude less
than the mean. The effect of fluorescence from the QDs is
removed from the total level by subtracting the signal-off-
waveguide from signal-on-waveguide readings at each pump
power. Then, the resulting on-waveguide reading may be
divided by the off-waveguide value to derive the effect of
the quantum-dot cascades such that a parallel trend where
increased pump power produces higher waveguide-to-
substrate ratios indicates transmission and guiding of light
by the waveguide. The first 5 min of data are removed to
account for signal laser stabilization and the outliers, which
are defined as being more than one standard deviation from
the mean of the raw data, are filtered from the data set. The
results for straight 10-µm-long and 10 µm × 10 µm corner
500-nm-wide waveguides made by the two-layer self-
assembly fabrication process are provided in Figure 4A-D,
respectively. Although the width dimension at 500 nm is
much larger than the QD diameter, signal transmission does
not occur as through a regular dielectric waveguide but rather
by a sequence of stimulated emission through the collection
of 3D confined nanocrystals. The behavior follows the
expected pattern such that the transmitted optical power is
greater with the waveguide than without the waveguide given
sufficient pump power. The net signal on the waveguide is
consistently higher than that on the substrate and, in addition,
the ratio increases with the pump power in both cases,
implying a gain mechanism. Otherwise, the QD would have
high absorption, and the measured output power would be
lower than that without the waveguide, as in the lower pump
power regions of Figure 4A-D. It is expected that the ratio
for the cornered waveguide is lower than that for the straight
waveguide because of its longer length and 90° bend.
Compared to prior calculation,17the QD exhibits gain starting
around 0.02 nW pump power per QD, which is indicated in
the sharp rise in signal above threshold. The similar trend
of increased ratio between the cornered waveguide and the
straight waveguide shows the promise of extremely flexible
routing using the proposed self-assembled colloidal quantum-
dot waveguides. The versatility of the design is one main
advantage compared to conventional dielectric waveguides
besides subdiffraction waveguiding, and is important for
realizing extremely efficient and compact photonic integrated
In summary, we have introduced two different self-
assembly fabrication methods for colloidal quantum-dot
waveguides, where the DNA-mediated process provides an
element of programmability and the two-layer procedure
allows for rapid prototyping and optimization. Testing of the
APTES-based devices shows that similar behaviors and
relative waveguiding intensities exist between straight and
corner patterns with subdiffraction limit dimensions where
the transmitted signal levels generally increase with pump
power. Both techniques may be extended toward deposition
Figure 4. Measured transmission results from 500-nm-width waveguides demonstrating higher signal on the waveguide than on substrate.
Straight (10 µm) waveguide results: (A) net signal power and (B) resulting relative power ratio. Corner (10 × 10 µm2) waveguide: (C)
net signal power and (D) resulting relative power ratio. The inset of D is the fluorescence micrograph of the corner waveguide; the scale
bar is 5 µm in length.
Nano Lett., Vol. 6, No. 11, 2006
of multiple quantum-dot-type structures to reduce crosstalk Download full-text
between waveguides.22Since gain in quantum dots is subject
to pump and signal wavelengths, for two structures each
tuned to transmit light at different wavelengths, overlap in
signal transmission from one waveguide to another is highly
suppressed. Investigation of such crosstalk contribution is
an area of continuing work and will further illuminate the
effectiveness of the device as a gain-enabled and wavelength-
selective nanophotonic waveguide.
Acknowledgment. We gratefully acknowledge funding
from the National Science Foundation (NSF) ADVANCE
program, NSF award no. 0524648, and the UW Royalty
Research Fund. C.-J.W. thanks the NSF Graduate Fellowship
Program for financial support. Work was performed in part
at the University of Washington Nanotech User Facility
(NTUF), a member of the National Nanotechnology Infra-
structure Network (NNIN), which is supported by the
National Science Foundation.
(1) Barrios, C. A.; Almeida, V. R.; Panepucci, R. P.; Lipson, M. IEEE
J. LightwaVe Technol. 2003, 21, 2332.
(2) Takahara, J.; Yamagishi, S.; Taki, H.; Morimoto, A.; Kobayashi, T.
Opt. Lett. 1997, 22, 475.
(3) Bogaerts, W.; Baets, R.; Dumon, P.; Wiaux, V.; Beckx, S.; Taillaert,
D.; Luyssaert, B.; Van Campenhout, J.; Bienstman, P.; Van Thourhout,
D. IEEE J. LightwaVe Technol. 2005, 23, 401.
(4) Johnson, S. G.; Villeneuve, P. R.; Fan, S.; Joannopoulos, J. D. Phys.
ReV. B. 2000, 62, 8212.
(5) Law, M.; Sirbuly, D. J.; Johnson, J. C.; Goldberger, J.; Saykally, R.
J.; Yang, P. Science 2004, 305, 1269.
(6) Brongersma, M. L.; Hartman, J. W.; Atwater, H. A. Phys. ReV. B
2000, 62, R16356.
(7) Maier, S. A.; Barclay, P. E.; Johnson, T. J.; Friedman, M. D.; Painter,
O. Appl. Phys. Lett. 2004, 4, 3990.
(8) Maier, S. A.; Kik, P. G.; Atwater, H. A.; Meltzer, S.; Harel, E.; Koel,
B. E.; Requicha, A. A. G. Nat. Mater. 2003, 2, 229.
(9) Yatsui, T.; Nomura, W.; Ohtsu, M. Nano Lett. 2005, 5, 2548.
(10) Chan, W. C. W.; Nie, S. Science 1998, 281, 2016.
(11) Bruchez, M., Jr.; Moronne, M.; Gin, P.; Weiss, S.; Alivisatos, A. P.
Science 1998, 281, 2013.
(12) Ledentsov, N. N.; Grundmann, M.; Heinrichsdorff, F.; Bimberg, D.;
Ustinov, V. M.; Zhukov, A. E.; Maximov, M. V.; Alferov, Z. I.;
Lott, J. A. IEEE J. Sel. Top. Quantum Electron. 2000, 6, 439.
(13) Ustinov, V. M.; Zhukov, A. E.; Egorov, A. Y.; Maleev, N. A.
Quantum Dot Lasers; Oxford University Press: Oxford, 2003.
(14) Cao, H.; Xu, J. Y.; Xiang, W. H.; Ma, Y.; Chang, S.-H.; Ho, S. T.;
Solomon, G. S. Appl. Phys. Lett. 2000, 76, 3519.
(15) Asada, M.; Miyamoto, Y.; Suematsu, Y. IEEE J. Quantum Electron.
1986, 22, 1915.
(16) Klimov, V. I.; Mikhailovsky, A. A.; Xu, S.; Malko, A.; Hollingsworth,
J. A.; Leatherdale, C. A.; Eisler, H.-J.; Bawendi, M. G. Science 2000,
(17) Wang, C.-J.; Lin, L. Y.; Parviz, B. A. IEEE J. Sel. Top. Quantum
Electron. 2005, 11, 500.
(18) Huang, L.; Wang, C.-J.; Lin, L. Y. Opt. Lett., submitted for
(19) Park, G.; Shchekin, O. B.; Huffaker, D. L.; Deppe, D. G. IEEE
Photonics Technol. Lett. 2000, 13, 230.
(20) Hu, W.; Sarveswaran, K.; Lieberman, M.; Bernstein, G. H. IEEE
Trans. Nanotechnol. 2005, 4, 312.
(21) Demers, L. M.; Ginger, D. S.; Park, S.-J.; Li, Z.; Chung, S.-W.;
Mirkin, C. A. Science 2002, 296, 1836.
(22) Wang, C.-J.; Lin, L. Y.; Parviz, B. A. Conference on Lasers and
Electro-Optics (CLEO), 2006, QML7.
Nano Lett., Vol. 6, No. 11, 2006 2553