Tailoring Light-Matter Interaction with a Nanoscale Plasmon Resonator
Nathalie P. de Leon,1,2 Brendan J. Shields,2 Chun L. Yu,1 Dirk Englund,2 Alexey V.
Akimov,2,3 Mikhail D. Lukin,2* and Hongkun Park1,2*
1Department of Chemistry and Chemical Biology, and 2Department of Physics, Harvard
University, Cambridge, MA 02138, 3Lebedev Institute of Physics, Moscow, Russia
*To whom correspondence should be addressed. Email: firstname.lastname@example.org,
(Received January 12, 2012)
We propose and demonstrate a new approach for achieving strong light-matter
interactions with quantum emitters. Our approach makes use of a plasmon
resonator composed of defect-free, highly crystalline silver nanowires surrounded
by patterned dielectric distributed Bragg reflectors (DBRs). These resonators have
an effective mode volume (Veff) two orders of magnitude below the diffraction limit
and quality factor (Q) approaching 100, enabling enhancement of spontaneous
emission rates by a factor exceeding 75 at the cavity resonance. We also show that
these resonators can be used to convert a broadband quantum emitter to a
narrowband single-photon source with color-selective emission enhancement.
PACS numbers 73.20.Mf, 42.50.Pq, 81.07.Gf, 81.07.Bc
Techniques for controlling light-matter interactions in engineered electromagnetic
environments are now being actively explored. Understanding these interactions is not
only of fundamental importance but also of interest for applications ranging from optical
sensing and metrology to information processing, communication, and quantum science
[1-3]. To enhance the coupling between an optical emitter and a desired mode of the
radiation field, two approaches can be used . One strategy is to increase the lifetime of
the confined optical excitation in high-Q dielectric resonators, such as whispering gallery
structures, micropillars, and photonic crystals [3, 5]. Another strategy is to reduce the
effective mode volume (Veff) of confined radiation [6-9], as is currently explored using
plasmonic nanostructures capable of confining light to dimensions well below the
Plasmonic resonators, which combine the benefits of both strategies, have potential
for engineering light-matter interaction at nanoscales and achieving large coupling
between an emitter and the radiation field [2, 10]. However, experimental realization of
these structures has remained an outstanding challenge [2, 11, 12]. Prior attempts to
control and engineer surface plasmon polariton (SPP) propagation have relied on
patterning of metal films using techniques such as focused ion beam and reactive ion
etching [13, 14]. Unfortunately, these patterning methods introduce defects that act as
scattering centers. Moreover, standard metal deposition techniques typically generate
polycrystalline films with short SPP propagation lengths. Consequently, plasmonic
cavities demonstrated thus far typically had large mode volumes to minimize absorption
and scattering in the metal [15-17].
In this Letter, we propose and experimentally demonstrate the new strategy to realize
a plasmonic resonator with an exceptionally small mode volume and a moderate Q that
can drastically modify the interaction between a quantum emitter and SPPs. Our new
approach, illustrated in Figure 1, takes advantage of chemically synthesized silver
nanowires for tight confinement and reduced group velocity of optical radiation.
Owing to their high crystallinity, these NWs support propagation of surface plasmon
polaritons (SPPs) over several micrometers in the visible range . We define plasmon
distributed Bragg reflectors (DBRs) and cavities by patterning polymethylmethacrylate
(PMMA), a low-index dielectric. Unlike metal/air structures that have been previously
employed to manipulate SPP propagation [13-17], patterned PMMA does not suffer from
high scattering losses because the relatively low index contrast (compared to metal/air
boundaries) reduces susceptibility to lithographic imperfections.
Thin silver NWs surrounded by air support a fundamental SPP mode whose spatial
extent is on the order of the wire radius . The effective refractive index for this mode
increases when the surrounding medium changes from air to PMMA (
air > 1),
providing the index contrast needed to define DBRs for SPPs. FDTD simulations show
that a quarter-wave stack composed of several PMMA slabs reflects the incoming SPPs
with > 90% efficiency (Fig. 1(c)). The stopband (the wavelength range over which the
quarter-wave stack acts as an efficient plasmon mirror) can be tuned over the entire
visible range by varying the stack period, with a bandwidth exceeding 100 nm. A sharp
resonant feature appears within the stopband when two DBRs are placed together to
define a plasmon cavity. Specifically, for a 100-nm diameter NW with DBRs composed
of 6 PMMA slabs, a plasmon cavity with Veff ~ 0.04 (!/n)3 and Q ~ 100 can be achieved
(Fig. 1(d), see also Supplementary Material).
When a quantum emitter is placed within this plasmon cavity, its spontaneous
emission rate can be dramatically modified. The Purcell factor (F), which is defined as
the ratio of the spontaneous emission rates within the cavity (") and in free space ("0),
scales as Q/Veff (see Supplemental Material). Equivalently, F can also be expressed as
F0!, where F0 and ! are the Purcell factor of a bare silver NW and the cavity
contribution, respectively. We note that ! is also the finesse of the cavity. For a plasmon
cavity with a 100-nm diameter silver NW (Fig. 1(d)), F0 ranges from 1 to 10 depending
on emitter placement , and ! ~ 20. F can therefore be as high as 200 when a quantum
emitter is placed at the peak electric field of the cavity mode.
Our plasmon resonator architecture offers distinct advantages over other photonic and
plasmonic cavity structures [3, 15, 19, 20]. First, the Purcell enhancement achievable in
our architecture improves dramatically as the device dimension is pushed well below the
diffraction limit . As the NW radius (R) decreases, Veff " R3, while Q can be kept
constant by choosing the cavity length to be the half the SPP wavelength (see
Supplemental Material). Therefore, the Purcell enhancement also scales as 1/R3,
indicating that extraordinarily strong coupling can be achieved for small diameter NW
devices. This is in stark contrast to dielectric photonic waveguides, in which the field
confinement decreases exponentially when the structure dimensions shrink below the
diffraction limit [21, 22].
In addition, due to its ultrasmall mode volume, our resonator can be used to direct the
emission of a broadband quantum emitter into a single cavity mode whose resonant
wavelength is selected by the cavity design. When a broadband emitter (e.g. a solid state
emitter with a broad phonon sideband) is coupled to a cavity with a much narrower
resonance, the total Purcell factor becomes independent of Q and increases only when Veff
is decreased (see Supplemental Material). Therefore, in this broadband emitter regime,
ultra-small mode volume plasmon resonators provide the only means to achieve efficient
single photon sources in which the color of emission can be selected, and the rate into
that mode can be enhanced over emission into free space.
We realize the plasmon resonator experimentally by first spin-coating PMMA on a
Si/SiO2 wafer, drop-casting silver NWs, and then spin-coating another layer of PMMA.
Electron beam lithography and subsequent development yield suspended NWs in periodic
PMMA slabs (Fig. 1(a)) [19, 23]. These plasmon DBRs and cavities were characterized
via transmission measurements of single nanostructures as a function of wavelength. Due
to the wavevector mismatch between propagating SPPs and free space photons, SPPs
couple to the far field only at defects or wire ends. We obtain a transmission spectrum by
focusing a supercontinuum laser to a diffraction-limited spot at one end of the NW, and
recording the scattered intensity at the other end as a function of wavelength using a
charge-coupled device (CCD) (Fig. 2(a)).
The data in Fig. 2(b) clearly show that the plasmon DBR exhibits a stopband in
transmission, as predicted by the simulations. By comparing the transmission intensity
just inside and outside of the stopband (over which range the propagation losses,
collection efficiency, and in/out-coupling efficiencies should be constant, see
Supplemental Material), we estimate the reflectivity of the DBR to be 90-95%. The drop-
offs in intensity at shorter and longer wavelengths are due to material absorption and the
near-IR cutoff of our optics, respectively.
Transmission spectra of cavities show a peak in the middle of the stopband. In the
device shown in Fig. 2(c), the peak is at 638 nm and the full width at half maximum is 11
nm, corresponding to a Q of 58. The highest Q observed to date in our plasmon cavities is
94, close to the theoretically simulated maximum value. The transmission intensity on
resonance is attenuated compared to that outside of the stopband due to higher absorption
losses caused by the longer effective path length on resonance, leff. From the measured
value of Q, we determine that leff is ~5 µm (see Supplemental Material). This value is
comparable to the SPP propagation length in bare silver NWs , and indicates that
losses in our resonators are dominated by material absorption.
We next demonstrate the utility of our cavities to control both the color and rate of
spontaneous emission of solid-state optical emitters. First, we show emission
modification in an ensemble of CdSe quantum dots. These quantum dots were coupled to
the devices by mixing them homogeneously in PMMA before fabrication. A fluorescence
spectrum of the quantum dots coupled to the NW was obtained by exciting one NW end
with green light (! = 532 nm) and collecting quantum dot fluorescence at the other NW
end. Fig. 3(a) clearly shows that the plasmon-coupled emission is narrowed and shifted
toward the resonance peak when compared to the fluorescence spectrum of uncoupled
The fluorescence lifetime is also modified by the plasmon cavity. The emission from
uncoupled quantum dots exhibits a single-exponential decay, characterized by a lifetime
(#free) of 16 ns ± 3 ns. In contrast, the emission from quantum dots coupled to the NW
exhibits a multiexponential decay because the quantum dots are distributed throughout
the PMMA, and the detected emission originates from an ensemble of quantum dots
along the NW. Notably, the initial slope of this decay yields the shortest lifetime (#coupled)
of less than 250 ps (Fig. 3(b)). This value suggests that the largest effective Purcell factor
Feff = #free/#coupled is >75, despite the detuning of the cavity resonance relative to the peak
of quantum dot emission. It is difficult to disentangle effects of nonradiative decay from
lifetime data alone, but we note that quantum dots coupled to a silver NW in unpatterned
PMMA show a multiexponential decay with initial slope corresponds to a lifetime of 4
ns. This evidence indicates that the much shorter lifetime observed in the cavity-coupled
decay is due to radiative emission enhancement rather than nonradiative decay (see
We next demonstrate control over emission properties of individual diamond nitrogen
vacancy (NV) centers using plasmon resonators. Diamond nanocrystals were coupled to
silver NWs by co-depositing them during fabrication. Approximately 10% of
nanocrystals exhibited stable, broad fluorescence characteristic of NV centers (Fig. 4(b))
. Once a single NV coupled to the wire was identified, the resonator structure was
defined by electron beam lithography around the NV center.
Fig. 4(a) shows scanning confocal microscope images of a resonator device with a
coupled NV. The top panel shows a reflection image of the device, and the middle panel
shows the fluorescence image recorded as the laser is scanned over the device. When the
NV is excited (circled), using an independently scanning collection channel (bottom
panel) we observe fluorescence from three locations: one corresponding to direct
emission from the NV, and two corresponding to the ends of the NW. The direct
emission from the NV exhibits strongly anti-bunched autocorrelation (inset of Fig. 4(b)),
indicating that it originates from a single NV center. Photon-correlation measurements
between the NV fluorescence spot and the NW end also show strong anti-bunching (inset
of Fig. 4(c)), showing that the emission from the wire end also originates from the NV [6,
9]. Assuming an SPP out-coupling efficiency of ~5% for these NWs [6, 24], we estimate
that 50-60% of the emission couples into SPPs for a typical device.
The NV fluorescence changes drastically when it is coupled to the plasmon cavity.
Before resonator fabrication, the plasmon-coupled NV fluorescence exhibits a broad
spectrum spanning a range of 630-740 nm (small, superimposed Fabry-Perot oscillations
originate from scattering by the NW ends: see Supplemental Material) [9, 16]. After
resonator fabrication, the plasmon-coupled NV spectrum exhibits a peak on resonance
with the cavity mode and suppressed fluorescence within the stopband (Fig. 4(c)). The
peak position can be placed anywhere across the NV fluorescence spectral range by
changing the cavity design, thereby enabling the selection of the wavelength of single
photon emission. The plasmon-coupled fluorescence intensity outside of the stopband (#
> 720 nm), Iout, is essentially unaltered by the cavity, and gives the baseline SPP-coupled
fluorescence. Notably, the fluorescence intensity on resonance (# = 637 nm), Ires, is
higher than Iout. In contrast, the transmission spectrum measured by launching SPPs at the
wire end with the supercontinuum laser shows that the transmitted intensity on resonance,
Tres, is lower than that outside of the stopband, Tout.
Comparison of these intensities gives the radiative Purcell enhancement due to the
cavity on resonance ! = (Ires/Iout)(Tout/Tres); for the device shown, we find that ! is 11 ±
3. This resonant enhancement is in addition to a broadband enhancement, F0, due to the
bare silver NW, which is estimated to be in a range of 1.5-2.5 for NWs of these
dimensions [6, 9]. Combining these factors together (F = F0!), we estimate the overall
Purcell enhancement to be as high as 35 at the resonance peak, a value that exceeds the
largest Purcell enhancement reported to date for NV centers coupled to dielectric cavities
[26, 27]. We note that the observed Purcell enhancement is still lower than the theoretical
maximum value expected for these devices, most likely because the NV center is not
located optimally within the cavity.
The strong emitter-cavity coupling observed in the present study can be improved in
several ways. Precise placement of NV centers at the peak electric field of the cavity
mode would ensure maximum Purcell enhancement for a given device. Furthermore,
because the Purcell enhancement scales as 1/R3, it can be made substantially higher using
thinner wires [7, 8]. In the present study, we were limited to larger (~100 nm) diameter
NWs because we relied on far-field excitation and detection of SPPs. While out-coupling
to the far field is less efficient in thinner wires, efficient coupling to thinner NWs can be
accomplished with near-field techniques such as evanescently-coupled optical fibers 
and electrical detection [24, 29]. Other resonator geometries, such as those that make use
of recently developed hyperbolic metamaterials can potentially be used to further
enhance the coupling .
The realization of nanoscale plasmon resonators with exceptionally small mode
volumes and high quality factors opens new possibilities for integrated plasmonic
systems, novel realization of nanoscale lasers and spasers , sub-diffraction sensing,
and optical interfacing of solid-state qubits. For instance, color-selective single photon
sources could have applications in quantum cryptography, and these resonators can be
used to direct NV emission into the zero-phonon line for coherent optical manipulation, a
crucial requirement for the realization of such applications as single photon transistors
. Other possibilities include high spatial resolution imaging and enhanced coupling of
individual molecules. Furthermore, the use of patterned, low-loss dielectrics for
controlling SPP propagation in nanoscale plasmonic structures can potentially be
extended towards other applications such as plasmonic circuit elements [33, 34], out-
coupling gratings , and meta-materials [35, 36].
We acknowledge J. T. Robinson, A. L. Falk, F. Koppens, and J. D. Thompson for
helpful discussions. We also gratefully acknowledge support from NSF, DARPA, the
Packard Foundation, and the NSF and NDSEG graduate research fellowships (NdL).
 S. Noda, M. Fujita, and T. Asano, Nature Photonics 1, 449 (2007).
 D. K. Gramotnev and S. I. Bozhevolnyi, Nature Photonics 4, 83 (2010).
 K. Vahala, Nature 424, 839 (2003).
 E. M. Purcell, Physical Review 69, 681 (1946).
 D. Englund, D. Fattal, E. Waks, G. Solomon, B. Zhang, T. Nakaoka, Y. Arakawa, Y. Yamamoto,
and J. Vuckovic, Physical Review Letters 95, 013904 (2005).
 A. V. Akimov, A. Mukherjee, C. L. Yu, D. E. Chang, A. S. Zibrov, P. R. Hemmer, H. Park, and
M. D. Lukin, Nature 450, 402 (2007).
 D. E. Chang, A. S. Sorensen, P. R. Hemmer, and M. D. Lukin, Physical Review Letters 97,
 D. E. Chang, A. S. Sorensen, P. R. Hemmer, and M. D. Lukin, Physical Review B 76, 035420
 R. Kolesov, B. Grotz, G. Balasubramanian, R. J. Stohr, A. A. L. Nicolet, P. R. Hemmer, F.
Jelezko, and J. Wrachtrup, Nature Physics 5, 470 (2009).
 S. A. Maier, Optical and Quantum Electronics 38, 257 (2006).
 W. L. Barnes, A. Dereux, and T. W. Ebbesen, Nature 424, 824 (2003).
 E. Ozbay, Science 311, 189 (2006).
 B. Wang and G. P. Wang, Applied Physics Letters 87, 013107 (2005).
 M. Sandtke and L. Kuipers, Nature Photonics 1, 573 (2007).
 B. Min, E. Ostby, V. J. Sorger, E. Ulin-Avila, L. Yang, X. Zhang, and K. Vahala, Nature 457, 455
 H. Ditlbacher, A. Hohenau, D. Wagner, U. Kreibig, M. Rogers, F. Hofer, F. R. Aussenegg, and J.
R. Krenn, Physical Review Letters 95, 257403 (2005).
 V. J. Sorger, R. F. Oulton, J. Yao, G. Bartal, and X. Zhang, Nano Letters 9, 3489 (2009).
 B. Wiley, Y. Sun, and Y. Xia, Langmuir 21, 8077 (2005).
 Y. Zhang and M. Loncar, Optics Express 16, 17400 (2008).
 M. Boroditsky, R. Vrijen, T. F. Krauss, R. Coccioli, R. Bhat, and E. Yablonovitch, Journal of
Lightwave Technology 17, 2096 (1999).
 L. Tong, R. Gattass, J. Ashcom, S. He, J. Lou, M. Shen, I. Maxwell, and E. Mazur, Nature 426,
 F. Le Kien, S. D. Gupta, V. I. Balykin, and K. Hakuta, Physical Review A 72, 032509 (2005).
 C. J. Barrelet, A. B. Greytak, and C. M. Lieber, Nano Letters 4, 1981 (2004).
 A. L. Falk, K. F. H. L., C. L. Yu, K. Kang, N. d. L. Snapp, A. V. Akimov, M.-H. Jo, M. D. Lukin,
and H. Park, Nature Physics 5, 475 (2009).
 F. Jelezko, C. Tietz, A. Gruber, I. Popa, A. Nizovtsev, S. Kilin, and J. Wrachtrup, Single
Molecules 2, 255 (2001).
 T. van der Sar, J. Hagemeier, W. Pfaff, E. C. Heeres, T. H. Oosterkamp, D. Bouwmeester, and R.
Hanson, arXiv:1008.4097v1 [quant-ph] (2010).
 D. Englund, B. Shields, K. Rivoire, F. Hatami, J. Vuckovic, H. Park, and M. D. Lukin, Nano
 M. Hochberg, T. Baehr-Jones, C. Walker, and A. Scherer, Optics Express 12, 5481 (2004).
 P. Neutens, P. Van Dorpe, I. De Vlaminck, L. Lagae, and G. Borghs, Nature Photonics 3, 283
 Z. Jacob, J.-Y. Kim, G. V. Naik, A. Boltasseva, E. E. Narimanov, and V. M. Shalaev, Appl. Phys.
B 100, 215 (2010).
 R. F. Oulton, V. J. Sorger, T. Zentgraf, R.-M. Ma, C. Gladden, L. Dai, G. Bartal, and X. Zhang,
Nature 461, 629 (2009).
 D. E. Chang, A. S. Sorensen, E. A. Demler, and M. D. Lukin, Nature Physics 3, 807 (2007).
 D. Pacifici, H. J. Lezec, and H. A. Atwater, Nature Photonics 1, 402 (2007).
 N. Engheta, A. Salandrino, and A. Alu, Physical Review Letters 95, 095504 (2005).
 D. R. Smith, J. B. Pendry, and M. C. K. Wiltshire, Science 305, 788 (2004).
 V. Shalaev, Science 322, 384 (2008).
FIG. 1 (color online). (a) Schematic of device concept (top) and SEM image using an
Inlens detector (bottom) of the DBR resonator fabricated around a silver NW. The
sample is tilted at 30° from two different orientations to demonstrate that the NW is
suspended from the substrate in PMMA. Scale bar = 1 µm. (b) Simulated electric field
intensity of a plasmonic cavity at wavelengths inside the stopband (top left), outside the
stopband (bottom left) and on resonance (right). (c) Simulated transmission spectrum of a
DBR consisting of a 100 nm silver NW and PMMA slabs with a period of 200 nm. A
quarter-wave stack composed of 6 PMMA slabs reflects the incoming SPPs with a
reflectivity exceeding 90%. (d) Simulated transmission spectrum of plasmon cavity
composed of two DBRs. The nominal cavity length corresponds to !SPP, and the
linewidth gives a Q of 100.
FIG. 2 (color online). (a) CCD image of SPP propagation overlaid on an SEM image of a
device. (b) Transmission spectrum of plasmonic DBR. The minimum intensity in the
stopband indicates a reflectivity of 90-95%, and sideband oscillations can be seen outside
the stopband. (c) Transmission spectrum of plasmon resonator. The peak in the stopband
at 638 nm has a width of 11 nm, corresponding to a Q of 58. SEM images of both devices
are shown in the insets, with scale bars that correspond to 1 µm.
FIG. 3 (color online). (a) Fluorescence spectrum of CdSe quantum dots in the substrate
(blue, dashed) and coupled to the plasmon resonator (red). The transmission spectrum of
the device is overlaid (gray, crosses). The fluorescence of coupled quantum dots is
shifted and narrowed by the cavity resonance. (b) Lifetime measurements obtained by
excitation with a pulsed laser of coupled (red) and uncoupled (blue, dashed) quantum
dots. Uncoupled quantum dots have a lifetime of 16 ns, while coupled quantum dots
show multi-exponential behavior, with the initial slope corresponding to a lifetime of less
than 250 ps.
FIG. 4 (color online). (a) Scanning confocal microscope images of the device. The top
panel shows an image of the reflected green laser light. The middle panel shows
fluorescence in the red as the laser is scanned over the area. When the laser is focused
onto the NV in the cavity (indicated with a white circle), an independent collection
channel is scanned over the area to detect fluorescence (bottom). The bright spot in the
center results from direct excitation and detection of the NV fluorescence. Two additional
spots can be seen from the ends of the wire, where SPPs excited by the NV scatter into
free space. Scale bar = 1 µm. (b) The fluorescence spectrum from the center of the NW
shows broadband emission characteristic of NVs. The inset shows autocorrelation of the
fluorescence, which is strongly anti-bunched, indicating that emission results from a
single NV center. (c) The fluorescence spectrum from the end of the NW (red) shows
significant modification, which corresponds to the transmission spectrum of the device
(black). Cross-correlation between the SPP-coupled emission and the emission collected
directly from the NV indicates that the fluorescence at the end of the NW originates from
the same NV center (inset).