Generation of single optical plasmons in metallic
nanowires coupled to quantum dots
A. V. Akimov1,4*, A. Mukherjee1*, C. L. Yu2*, D. E. Chang1, A. S. Zibrov1,4, P. R. Hemmer3, H. Park1,2& M. D. Lukin1
optical emitters is an outstanding problem in quantum science
and engineering. It is of interest for ultimate control over light
quanta1, as well as for potential applications such as efficient
photon collection2, single-photon switching3and transistors4, and
long-range optical coupling of quantum bits5,6. Recently, sub-
stantial advances have been made towards these goals, based on
modifying photon fields around an emitter using high-finesse
approach for engineering photon–emitter interactions4,9via sub-
wavelength confinement of optical fields near metallic nano-
structures10–13. When a single CdSe quantum dot is optically
excited in close proximity to a silver nanowire, emission from the
quantum dot couples directly to guided surface plasmons in the
correlations between the emission from the quantum dot and the
ends of the nanowire demonstrate that the latter stems from the
generation of single, quantized plasmons. Results from a large
number of devices show that efficient coupling is accompanied
by more than 2.5-fold enhancement of the quantum dot spontan-
eous emission, in good agreement with theoretical predictions.
Surface plasmons, or surface plasmon polaritons, are propagating
excitations of charge-density waves and their associated electromag-
netic fields on the surface of a conductor10. Much like the optical
modes of a conventional dielectric fibre, a broad continuum of sur-
face plasmon modes can be confined on a cylindrical metallic wire
and guided along the wire axis12,13(Fig. 1a). However, as opposed to
dielectric waveguides14,31, thethin wires can maintain propagation of
*These authors contributed equally to this work.
1Department of Physics,2Department of Chemistry and Chemical Biology, Harvard University, Cambridge, Massachusetts 02138, USA.3Department of Electrical and Computer
Engineering, Texas A&M University, College Station, Texas 77843, USA.4P.N. Lebedev Physical Institute RAS, Leninskiy prospect 53, Moscow, 119991, Russia.
50 nm wire
100 nm wire
Scattered intensity (a.u.)
80 100 120
Distance from wire (nm)
Enhancement factor, P
Figure 1 | Radiativecoupling of quantum dots toconducting nanowires. a, A
guided surface plasmons of the nanowire with respective rates Crad, Cpl.
b, Theoretical dependence of the enhancement factor P (solid line) and
efficiency of emission into surface plasmons (dashed line) on distance of the
emitter from the nanowire edge. The red (blue) curve corresponds to a wire
diameter of 100nm (50nm). c, Simulations of the electric field amplitude
(arbitraryunits) emitted bya dipole(blue filled circle) positioned 25nm from
one end of a conducting nanowire (whose surface is outlined) 3mm in length
and 50nm in diameter. The vertical scale (r) is enlarged compared to the
horizontal (z) to clearly show the near field of the surface plasmons. Upon
back-reflection. h, Emission angle. d, Amplitude of the Poynting vector of the
light scattered from the far end of the nanowire, as a function of h (see c), for
wires of diameter 100nm (red curve), 50nm (blue) and 25nm (green).
Vol 450|15 November 2007|doi:10.1038/nature06230
surface plasmon modes localized transversely to dimensions com-
parable to the wire diameter d, even when it is much smaller than the
optical wavelength l. This subwavelength localization is accompan-
ied by a dramatic concentration of optical fields10,11. In addition, the
surface plasmon modes propagate with greatly reduced velocities
because they involve the motion of charge-density waves9,15,16.
The emission properties of a nanoscale optical emitter can be
significantly modified by the proximity of a nanowire that supports
surface plasmons. In principle, three distinct decay channels exist.
First,directopticalemission intofree-space modesispossible,witha
rate modified from that of an isolated quantum dot owing to the
damped non-radiatively owing to ohmic losses in the conductor17.
Last, and most importantly, the tight field confinement and reduced
velocity of surface plasmons can cause the nanowire to capture the
majority of spontaneous radiation into the guided surface plasmon
modes9, much like a lens with extraordinarily high numerical aper-
ture. For an optical emitter placed within the evanescent surface
plasmon mode tail, the spontaneous emission rate into the surface
plasmons9is proportional to (l/d)3. In contrast, the free-space
emission rate can be enhanced by at most a factor of four, whereas
non-radiative dampingbecomessignificant onlyforverysmall wire–
emitter separation9. Thus, for an optimally placed emitter, the spon-
taneous emission rate Cplinto surface plasmons can far exceed the
radiative andnon-radiative rates (Cradand Cnrd, respectively), which
results in highly efficient coupling to surface plasmons and enhance-
emitter (C0). This enhancement can be characterized by a Purcell
factor, P5Ctotal/C0, which for thin wires is predicted to be large9.
We emphasize that this strong coupling is caused by the geometrical
effect of tight confinement of the surface plasmons, and occurs far
away from the plasmon resonance frequency of nanowires18. It does
not involve an optical cavity2,3,5–8, and can be achieved simulta-
neously over a broad continuum of optical frequencies.
silver nanowires comprise a simple experimental system to investi-
gate the emitter–surface plasmon coupling. As illustrated in Fig. 1a,
the spontaneous emission of a quantum dot is split between photon
emission into free space, which can be detected by an optical micro-
scope, and the excitation of surface plasmons (Cnrdis negligible for
our chosen parameters, as described below). During propagation
along the smooth nanowire, surface plasmons do not couple to the
observable far-field modes of the surrounding dielectric. However,
much like a conventional antenna, an abrupt end of the wire can
scatter surface plasmons radiatively into far-field modes, thus facil-
itating their detection using an optical microscope. A simulation of
this effect is shown in Fig. 1c, where a quantum dot is placed 25nm
cently away from the nanowire edge, substantial emission into free
space results from surface plasmon scattering at the far end of the
wire. Silver nanowires were prepared using a solution-phase polyol
Information). The samples were created by spinning quantum dots
onto a glass substrate, covering them with an ,30-nm layer of poly
(methylmethacrylate) (PMMA; see Supplementary Information for
detailed analysis of the PMMA layer), and then depositing dry wires
on top. Finally, the sample was overcoated with a thick layer of
PMMA. Scanning electron microscopy images revealed that the dia-
meters of the silver nanowires were 102624 nm (Supplementary
Information). The closest allowed distance between the quantum
dots and nanowires is determined by the thickness of the PMMA
layer and the quantum dot shell radius (,5nm), and is ,35nm
(Methods and Supplementary Information). The experimental set-
up for studying the quantum dot–nanowire system (Fig. 2a) is based
on a modified confocal microscope with three scanning channels.
One channel (I) was used for imaging nanowires, and the second
channel (II) was used for imaging quantum dots. The third channel
(III), which can independently image any diffraction-limited spot
within the field of view of the objective lens, was used to detect the
scattered surface plasmons from the nanowire ends.
In general, the coupling between an optical emitter and single
surface plasmons should be stronger for thinner wires9(Fig. 1b).
However, for thinner wires, the out-coupling efficiency of surface
plasmons to the far-field at the wire end decreases owing to a large
wavevector mismatch. In this case, significant surface plasmon
QDs in buffer
Ch ICh IICh III
Figure 2 | Experimental set-up. a,Three-channelconfocal microscope with
532nm laser excitation source. b, Layout of sample containing quantum
dots and nanowires. c, Left, channel I: nanowire image. Middle, channel II:
image of quantum dots. The red circle denotes the position of the coupled
quantum dot, and the same point is also denoted in the leftmost image.
Right, channel III: the excitation laser was focused on the quantum dot (red
while two smaller spots correspond to surface plasmons scattered from the
nanowire ends. The blue circle indicates the farthest end of the nanowire,
used for photon cross-correlation measurements.
NATURE|Vol 450|15 November 2007
formation within the nanowire12(Fig. 1c) and eventual energy loss
due to heating (ohmic losses). The effect of nanowire diameter on
out-coupling efficiency is illustrated inFig.1d, where the intensity of
the scattered radiation from the wire end is plotted for different wire
diameters. For a 25-nm nanowire, hardly any scattering is seen from
plasmons, but the scattering is significant for a 100-nm wire (this
was verified experimentally by exciting surface plasmons directly
with a laser focused at one wire end; Supplementary Information).
Nanowires with d<100nm exhibit both reasonable emitter–surface
plasmon couplings and surface plasmon to far-field scattering, and
thus were chosen for the experiments. The large bandwidth of the
surface plasmon–emitter coupling enables us to perform the experi-
ments at room temperature, where a single quantum dot spectral
width exceeds 15nm (Supplementary Information).
Figure 2c presents an experimental demonstration of directed
emission of a quantum dot into surface plasmons. The leftmost
of quantum dots detected at 655nm with channel II. These two
images were used to determine the positions of the nanowire and
quantum dot relative to each other. Owing to the resolution limit of
our optical system, the actual distance between a quantum dot
and the nanowire could not be determined, and only quantum
dots that appear directly on top of a nanowire were chosen for
experiment. The rightmost panel shows a coupled wire–dot system
imaged with channel III. When the proximal quantum dot (circled
in red) was excited by the laser, the nanowire ends literally lit up.
The large spot around the red circle corresponds to emission from
the quantum dot itself, whereas the two other points coincide with
the wire ends. Significantly, a high degree of correlation was seen
dot and the end of the coupled wire (Fig. 3a). These observations
indicate that the source of the fluorescence from the wire end is the
Photon coincidence measurements1of the quantum dots (Fig. 3b)
demonstrate that these quantum dots can only emit a single photon
at a time. In these measurements, the free-space fluorescence from
the quantum dot was equally split into two channels using a beam
splitter and detected by avalanche photo-diodes. The coincidences
between two channels were recorded as a function of time delay t. If
the quantum dot emits only one photon at a time it can only be
recorded at one of the channels, and therefore zero coincidences
are expected between the two channels at t50, as seen in Fig. 3b.
The slight offset from zero can be attributed to stray light, dark
counts of the detectors and the resolution limit of the electronics
The light emission at the nanowire end is a result of single, quan-
measurements between the free-space fluorescence of the quantum
dot and emission from the wire end. This near-zero coincidence is a
consequence of the fact that the single photon emitted from a
modes, but never both simultaneously.
Data presented in Fig. 3, along with measured count rates, can be
used to quantify the coupling strength of the quantum dot to the
surface plasmons. As this coupling creates a new decay channel for
the quantum dot, its decay rate is expected to increase. To study
this enhancement, observed coincidence data were fitted to a simple
two-level model of quantum dot emission21(Fig. 3b; see also
pumping rate R from the ground to an excited state of a quantum
dot and a decay rate Ctotalback to the ground state. In this model,
the temporal width of the anti-bunching dip is given by
to the incident power. Therefore, by extracting Dt from coincidence
to R50, Ctotalcan be obtained (Fig. 4a).
The natural lifetimes of individual dots (20–30ns) vary owing to
the heterogeneity in their structures. However, a comparison of the
lifetime distributions of 30 coupled and 100 uncoupled quantum
dots (Fig. 4b) clearly demonstrates that statistically the lifetime
(decay rate) of the exciton in coupled quantum dots is shortened
(enhanced). The average lifetime of the coupled (uncoupled)
quantum dots was found to be 1364ns (2265ns). At the same
time, the distribution for coupled quantum dots has a larger
weight towards shorter lifetimes. Specifically, certain coupled and
uncoupled quantum dots exhibited lifetimes as short as 6ns and
15ns, respectively, indicating that P.2.5 is achieved for some
coupled quantum dot–nanowire systems. The apparent efficiency
the ratio of photon counts (n) obtained directly from the dot and
from the wire ends, gm<nends/(ndot1nends), and is found to be
,27% for the best coupled quantum dot–nanowire system (Fig. 4c
and Supplementary Information). This value does not account for
the surface plasmons that are dissipated before they reach the wire
ends. Correcting for the measured average absorption lengths in our
nanowires allows us to deduce that the actual efficiency approaches
Þ, where the excitation rate R is proportional
Time, t (s) Time delay, (ns)
Time delay, (ns)
Intensity (103 counts s–1)
In 2/(R + tot)
Figure 3 | Demonstration of single surface plasmon generation. a, Time
trace of fluorescence counts (red curve) from a coupled quantum dot and
dueto quantumdotblinking19. b,Second-ordercorrelationfunctionG(2)(t)
of quantum dot fluorescence. The number of coincidences at t50 goes
almost to zero, confirming that the quantum dot is a single-photon source.
The width of the dip depends on Ctotaland the pumping rate R as shown.
c, Second-order cross-correlation function between fluorescence of the
quantum dot and scattering from the nanowire end, obtained by
coincidences between channel II (quantum dot) and channel III (wire end).
The black and red traces in b, c indicate experimental data and best fits,
NATURE|Vol 450|15 November 2007
very efficient coupling to surface plasmons. We note that this coupl-
ing efficiency significantly exceeds that recently observed between
atoms and dielectric nanofibres14,31.
The broadband nature of the strong coupling is demonstrated by
comparing the optical spectra associated with emission from the
quantum dot and from the wire end. For individual dots randomly
drawn from an inhomogeneous ensemble with l5655615nm, we
ical ,15-nm-wide spectra. This is consistent with the ability of
metallic wires to guide a broad range of optical frequencies22and
with theoretical predictions (Supplementary Information) that
strong coupling can be obtained for a broad continuum of frequen-
cies away from the peak of the observed plasmon resonances18.
Further insight into the quantum dot–surface plasmon coupling
can be obtained by comparing our experimental observations with
detailed electrodynamic calculations9. Our model of quantum dot
emission near a silver nanowire embedded in a dielectric medium
includes losses as well as multiple surface plasmon modes. Figure 1b
shows the total spontaneous emission rates and the efficiency
g5Cpl/Ctotalfor single surface plasmon generation as a function of
quantum dot distance from the wire (d550 and 100nm). Here the
oriented, because this direction is expected to yield the dominant
contribution to enhancement. For quantum dots positioned 35nm
from the wire and for a 100nm wire, the calculation yields a Purcell
factor P<3.7. The lower enhancement observed experimentally
can be attributed to the contributions from other polarizations and
this distance of separation, the non-radiative decay rate (Cnrd,
0.05C0) is predicted to be negligible (Supplementary Fig. 10). In
addition to enhanced emission into surface plasmon modes, our
theory also predicts a moderate increase in the radiative emission
to a metallic surface17. For 100nm wires and 35nm nanowire–
quantum dot distances, the surface plasmon generation efficiency g
is theoretically estimated to be ,50%, which is consistent with our
Further comparison with theoretical predictions is obtained by
repeating our observations with thicker PMMA layers (Fig. 4c, d).
These measurements demonstrate that both enhancement and
estimated coupling efficiency rapidly decrease as the minimum
quantum dot–nanowire spacing increases, and become very small
for PMMA thicknesses above 100nm. These observations are also
in good agreement with the above theoretical predictions. The large
variances in the Purcell factors obtained for different devices are due
primarily to variations in the distance between quantum dots and
nanowires beyond the minimum allowed distance set by the PMMA
been explored in a variety of fascinating systems, from transmission
through subwavelength structures11to biomedical devices10and
proposals for realizing ‘perfect’ lenses and invisibility cloaks10.
Enhancement of fluorescence23,24, polarization-dependent coupl-
ing25,26and normal mode splitting27,28near subwavelength structures
have also recently been observed. The present work extends these
developments in two principal directions. First, we have shown
experimentally and theoretically that the present approach results
simultaneously in significant enhancement of surface plasmon emis-
sion and efficient collection into guided modes propagating along a
individual emitters and individual, quantized surface plasmons. It
thus bridges the fields of nanoscale plasmonics and quantum optics,
and opens up the possibility of using quantum optical techniques to
achieve new levels of control over the interaction of single surface
plasmons and to realize novel quantum plasmonic devices.
In the current set-up, the benefits of using smaller wires must be
balanced against poor out-coupling to free-space modes. However,
this trade-off can be circumvented by using optimized geometries
and evanescent out-coupling to mode-matched optical fibres9,16,23.
used, for example, for efficient single-photon sources, high resolu-
ing5. Furthermore, in such systems an individual emitter can be
made optically opaque to single incident surface plasmons, which
can be used to produce large optical nonlinearities for realization of
single-photon transistors4. Beyond these specific applications, the
ability to create and control individual quanta of current oscillating
at optical frequencies and accompanied by guided radiation with
subwavelength localization opens up intriguing new possibilities at
the interface of optics and electronics.
Samples were prepared by spin-coating a solution of chemically synthesized
CdSe quantum dots (mixed with Na2B4O7and cysteine) onto a plasma-cleaned
glass slide at 3,000r.p.m. for 60s under a nitrogen atmosphere. Three minutes
top at 6,000r.p.m. for 60s. The quantum dots used do not dissolve in toluene
and are unperturbed during the spin-coating process (experimentally, we find
that the arrangement of quantum dots on the surface remains unchanged). A
stamp with the modified silver nanowires was placed on top of the slide and
pressed for a few seconds. The stamp was left there for 20min and then gently
on top at 1,000r.p.m. for 60s (Fig. 2b).
focused onto the sample using a Nikon CFI Plan Fluor 1003 oil immersion
objective NA 1.3, while a mirror mounted on a galvanometer is used to scan the
incoming beam. Channel II acts as a confocal microscope and is used to image
single quantum dots, via fluorescence at 655nm. Channel I is combined with
channel II using a 90:10 beam splitter that directs part of the reflected laser light
towards a detector and can be used to image the silver nanowires. Channel III is
PMMA thickness (nm)
Excitation power (µW)
PMMA thickness (nm)
Efficiency , m (%)
Γtotal + R (ns–1)
Γtotal normalized distribution
Figure 4 | Characterization of quantum dot–nanowire coupling. a, The
linear dependence of G(2)width on laser power (black filled circles) is
extrapolated to zero power (red filled circle), yielding Ctotalof a quantum
dot. b, Normalized histograms of quantum dot lifetimes. The black (grey)
bars denote the distribution of uncoupled (coupled) quantum dots.
Overlapping parts of the histograms are indicated by outlinedand vertically
stacked bars. c, Average Purcell enhancement, P, versus PMMA thickness.
Red line, average value of P. Height and width of grey bars indicate the
standard deviations of P and PMMA thickness, respectively. d, Measured
maximum and average efficiencies of emission into the surface plasmons
versus PMMA thickness. Black (red) filled triangles, average (maximum)
apparent coupling efficiencies gm, without compensating for surface
plasmon losses. Red filled diamonds, maximum actual efficiency g, after
compensating for dissipation. Error bars in a, c, d indicate 61s.d.
NATURE|Vol 450|15 November 2007
imaging system. It also includes a galvanometer which allows us to image any
diffraction limitedspot within the fieldof viewto detect fluorescence at 655nm.
Additional details of our experimental set-up are provided in Supplementary
Received 10 April; accepted 4 September 2007.
1.Haroche, S. & Raimond, J.-M. Exploring the Quantum: Atoms, Cavities, Photons.
(Oxford Univ. Press, New York, 2006).
two-dimensional photonic crystal. Phys. Rev. Lett. 95, 013904 (2005).
Nature 436, 87–90 (2005).
Chang, D. E., Sørensen, A. S., Demler, E. A. & Lukin, M. D. A single-photon
transistor using nano-scale surface plasmons. Nature Phys. advance online
publication doi:10.1038/nphys708 (26 August 2007).
Cirac, J. I., Zoller, P., Kimble, H. J. & Mabuchi, H. Quantum state transfer and
entanglement distribution among distant nodes in a quantum network. Phys. Rev.
Lett. 78, 3221–3224 (1997).
cavity QED. Phys. Rev. Lett. 83, 4204–4207 (1999).
Hennessy, K. et al. Quantum nature of a strongly coupled single quantum
dot–cavity system. Nature 445, 896–899 (2007).
Wilk, T., Webster, S. C., Kuhn, A. & Rempe, G. Single-atom single-photon
quantum interface. Science 317, 488–490 (2007).
Chang, D. E., Sørensen, A. S., Hemmer, P. R. & Lukin, M. D. Quantum optics with
surface plasmons. Phys. Rev. Lett. 97, 053002 (2006).
10. Atwater, H. A. The promise of plasmonics. Sci. Am. 296, 56–63 (2007).
11. Genet, C. & Ebbesen, T. W. Light in tiny holes. Nature 445, 39–46 (2007).
in silver nanowires. Nano Lett. 6, 1822–1826 (2006).
13. Ditlbacher, H. et al. Silver nanowires as surface plasmon resonators. Phys. Rev.
Lett. 95, 257403 (2005).
14. Nayak, K. P. et al. Optical nanofiber as an efficient tool for manipulating and
probing atomic fluorescence. Opt. Express 15, 5431–5438 (2007).
15. Takahara, J., Yamagishi, S., Taki, H., Morimoto, A. & Kobayashi, T. Guiding of a
one-dimensional optical beam with nanometer diameter. Opt. Lett. 22, 475–477
16. Chang, D. E., Sørensen, A. S., Hemmer, P. R. & Lukin, M. D. Strong coupling of
single emitters to surface plasmons. Phys. Rev. B 76, 035420 (2007).
17. Chance, R. R., Prock, A. & Silbey, R. Molecular fluorescence and energy transfer
near interfaces. Adv. Chem. Phys. 37, 1–65 (1978).
processing. Nano Lett. 2, 165–168 (2002).
intensity time trace of collections of CdSe nanocrystal quantum dots based on
single quantum dot fluorescence blinking statistics. Phys. Rev. E 73, 011106
using surface-enhanced Raman spectroscopy. Nano Lett. 3, 1229–1233 (2003).
21. Lounis, B., Bechtel, H. A., Gerion, D., Alivisatos, P. & Moerner, W. E. Photon
antibunching in single CdSe/ZnS quantum dot fluorescence. Chem. Phys. Lett.
329, 399–404 (2000).
22. Dickson, R. M. & Lyon, L. A. Unidirectional plasmon propagation in metallic
nanowires. J. Phys. Chem. B 104, 6095–6098 (2000).
23. Hochberg, M., Baehr-Jones, T., Walker, C. & Scherer, A. Integrated plasmon and
dielectric waveguides. Opt. Express 12, 5481–5486 (2004).
24. Biteen, J. S., Lewis, N. S. & Atwater, H. A. Spectral tuning of plasmon-enhanced
silicon quantum dot luminescence. Appl. Phys. Lett. 88, 131109 (2006).
25. Zhang, J., Ye, Y.-H., Wang, X., Rochon, P. & Xiao, M. Coupling between
72, 201306(R) (2005).
26. Mertens, H., Biteen, J. S., Atwater, H. A. & Polman, A. Polarization-selective
plasmon-enhanced silicon quantum-dot luminescence. Nano Lett. 6, 2622–2625
27. Bellessa, J., Bonnand, C. & Plenet, J. C. Strong coupling between surface
plasmons and excitons in an organic semiconductor. Phys. Rev. Lett. 93, 036404
28. Dintinger, J., Klein, S., Bustos, F., Barnes, W. L. & Ebbesen, T. W. Strong coupling
between surface plasmon-polaritons and organic molecules in subwavelength
hole arrays. Phys. Rev. B 71, 035424 (2005).
29. Klimov, V. V., Ducloy, M. & Letokhov, V. S. A model of an apertureless scanning
microscope with a prolate nanospheroid as a tip and an excited molecule as an
object. Chem. Phys. Lett. 358, 192–198 (2002).
30. Smolyaninov, I. I., Elliott, J., Zayats, A. & Davis, C. C. Far-field optical microscopy
plasmon polaritons. Phys. Rev. Lett. 94, 057401 (2005).
31. Sague ´, G., Vetsch, E., Alt, W., Meschede, D. & Rauschenbeutel, A. Cold atom
physics using ultra-thin optical fibers: light-induced dipole forces and surface
interactions. Preprint at Æhttp://arxiv.org/quant-ph/0701167æ (2007).
Supplementary Information is linked to the online version of the paper at
Acknowledgements We acknowledge discussions with M. Loncar, J. Doyle,
A. Sørensen and M.-H. Yoon, and support from the NSF, DARPA, Harvard-MIT
CUA, Harvard CNS, the DTO, the Packard Foundation and Samsung Electronics.
Author Information Reprints and permissions information is available at
www.nature.com/reprints. Correspondence and requests for materials should be
addressed to M.D.L. (firstname.lastname@example.org) and H.P.
NATURE|Vol 450|15 November 2007