Multi-fold Enhancement of Quantum Dot Luminescence in a Plasmonic Metamaterial
K. Tanaka,
1, 2, ∗
E. Plum,
1, †
J. Y. Ou,
1
T. Uchino,
3
and N. I. Zheludev
1, ‡
1
Optoelectronics Research Centre, University of Southampton, SO17 1BJ, UK
2
Sony Corporation, Shinagawa-ku, Tokyo, 141-0001, Japan
3
School of Electronics and Computer Science, University of Southampton, Southampton SO17 1BJ, UK
(Dated: August 27, 2010)
We report that hybridizing semiconductor quantum dots with plasmonic metamaterial leads to a
multi-fold intensity increase and narrowing of their photoluminescence spectrum. The luminescence
enhancement is a clear manifestation of the cavity quantum electrodynamics Purcell effect that can
be controlled by the metamaterial’s design. This observation is an essential step towards understand-
ing loss compensation in metamaterials with gain media and for developing metamaterial-enhanced
gain media.
Control of Joule losses is a key challenge for plasmonic
and metamaterial technologies. Losses hamper the de-
velopment of negative index media for super-resolution
and optical cloaking devices, and plasmonic data pro-
cessing circuits. Lowering losses is also crucially impor-
tant for the performance of spectral filters, delay lines
and, in fact, practically any other metamaterial and plas-
monic applications [1]. Although using superconducting
metamaterials can largely eliminate losses in THz and
microwave metamaterials [2], Joule losses at optical fre-
quencies are unavoidable. Recent works report compen-
sation of losses with gain in metamaterials aggregated
with semiconductor quantum dots (QDs) [3] and organic
dyes [4] embedded into the metal nanostructures. Para-
metric metamaterials gain systems are also under inves-
tigation in theory [5–7]. Another grand goal of active
metamaterials research is to improve laser gain media
and to develop a ‘lasing spaser’ device: a ‘flat’ laser with
emission fueled by plasmonic excitations in an array of
coherently emitting meta-molecules [8]. An essential part
of this development shall be the study of luminescence of
active material hybridized with plasmonic nanostructures
that could support collective, coherent plasmonic excita-
tions in the lasing spaser. Here we report the first study
of photoluminescence of semiconductor QDs hybridized
with asymmetric split-ring plasmonic metamaterial. This
type of metamaterial supports a closed-mode Fano-type
excitation which has the key characteristics required for
the lasing spaser application: the mode is formed by
collective interactions between individual meta-molecules
that shall ensure coherent laser action [9]. In this letter,
we experimentally demonstrate that the photolumines-
cence properties of QDs can be greatly enhanced by the
plasmonic metamaterial.
Figure 1(a) schematically illustrates a plasmonic meta-
material combined with QDs. The metamaterials stud-
ied here consist of periodic arrays of asymmetrically split
ring slits (negative structure), which have been success-
fully applied to switching, nonlinear and sensor applica-
tions [10]. The metamaterial arrays with a total size of
40 × 40 µm each were fabricated by focused ion beam
milling in a 50nm-thick gold film on a glass substrate
1000 1200 1400 1600 1800 2000
0
20
40
60
80
100
(b)
(a)
Glass Substrate
Gold Film
(50 nm)
QD/PMMA
Layer
(180 nm)
Pump Laser
(λ=532nm)
Photoluminescence
T
R
A
Wavelength (nm)
T, R, A (%)
w
a
D
g
t
X
Y
FIG. 1: (color online). (a) Schematic of a plasmonic metama-
terial functionalized with quantum dots (QDs). (b) Measured
sp ectra of transmission T , reflection R, and absorption A for a
QD-coated metamaterial with a unit cell size of D = 545 nm.
The incident wave is polarized parallel to the y-axis. The in-
sets show a sketch of QDs in the resonant mode volume as
seen from the substrate side and a scanning electron micro-
graph of the unit cell (without QDs). Feature sizes: unit cell
D = 545 nm, horizontal slit a = 470 nm, top vertical slit and
gap t = g = 170 nm and slit width w = 65 nm.
[see inset of Fig. 1(b)].
In order to systematically investigate the correlation
between QD photoluminescence spectrum and the spec-
tral position of the Fano plasmonic metamaterial reso-
nance, we manufactured five metamaterial arrays with
different unit cell sizes ranging from D = 545 nm to
645 nm, slit width w = 65 nm and a fixed ratio of
t/g = 1. We used lead sulfide (PbS) semiconductor
2
quantum dots from Evident Technologies with a lumi-
nescence peak around 1300 nm and mean core diameter
of 4.6 nm. These QDs were dispersed in polymethyl-
methacrylate (PMMA) and the QD/PMMA solution was
then spin-coated onto the metamaterial arrays forming a
180 nm thick layer. We estimate the QD area density on
the array to be 1.6 × 10
5
µm
−2
and thus approximately
4000 quantum dots per meta-molecule are trapped in
the groves of the structure. Spectra (transmission, re-
flection, and absorption) and photoluminescence charac-
teristics of the metamaterials with QDs were measured
using a microspectrophotometer. In the photolumines-
cence measurements, the QDs were optically pumped by
a frequency doubled CW YAG laser (λ = 532 nm) from
the substrate side through the metamaterial array [see
Fig. 1(a)]. The YAG laser was focused to a (∼ 100 µm)
spot with intensity 35 W/cm
2
by the microscope objec-
tive (N.A.= 0.28). Photoluminescence emitted from the
QDs was collected by the same objective with a polarizer
so that only a selected polarization component of pho-
toluminescence could be detected. Fig. 1(b) shows the
spectral characteristics of the QD/PMMA-coated meta-
material with D = 545 nm in the absence of the pump.
The absorption spectrum shows a narrow resonance peak
around 1300 nm (Q-factor ' 11).
The photoluminescence of the metamaterial function-
alized with QDs is presented in Fig. 2. Here we measure
the y-polarization, for which the metamaterial’s Fano-
mode can be excited, see Fig. 1(b). The photolumines-
cence spectrum of the QD/PMMA layer on the glass sub-
strate (i.e. without metamaterial) is shown at the top of
Fig. 2(a) and peaks at λ
0
= 1280 nm which is indicated
by the shaded area.
The presence of the plasmonic metamaterial drasti-
cally changes the QD photoluminescence characteristics:
it leads to a multi-fold intensity enhancement as well
as spectral narrowing of the photoluminescence peak.
For instance, for D = 545 nm the photoluminescence
peak intensity is enhanced by a factor of 8, while the
full width at half maximum (FWHM) of the photolu-
minescence peak is decreased to approximately 100 nm
compared to 176 nm without metamaterial. Here the
metamaterial’s absorption resonance wavelength λ
abs
al-
most perfectly matches the QD emission wavelength λ
0
(see red dashed absorption spectrum). This suggests
that the drastic photoluminescence enhancement results
from interaction between the excited-state gain medium
(QDs) and the surface plasmon resonance [11–14], and,
in particular, can be understood in terms of the cavity
quantum electrodynamics (QED) Purcell effect [15, 16]
as discussed later. Such coupling between QD-excitons
and metamaterial surface plasmons must be sensitive to
a mismatch ∆λ = λ
abs
− λ
0
. Indeed, when the meta-
material resonance is red-shifted, by increasing the unit
cell size from D = 545 nm to 645 nm, the photolumines-
cence spectrum is weakened, broadened and distorted.
1000 1150 1300 1450 1600
0 50 100 150 200
0
2
4
6
8
10
0 50 100 150 200
100
130
160
190
220
250
Wavelength (nm)
Photoluminescence intensity (arb. units)
D=545nm
D=575nm
D=595nm
D=645nm
D=620nm
without
metamaterial
FWHM (nm)
∆λ (nm)
Enhancement
FWHMEnhancement
(a)
(b)
with
metamaterial
FIG. 2: (color online). Photoluminescence controlled by plas-
monic metamaterials. (a) Photoluminescence spectra of the
QDs without (top) and with metamaterial layer. A best-fit
Gaussian curve is plotted with each spectrum (solid line). The
dotted line indicates the metamaterial absorption spectrum.
(b) Intensity enhancement and full width at half maximum
(FWHM) of the QD-metamaterial photoluminescence spec-
tra. Enhancement is normalized to the photoluminescence
p eak intensity measured without a metamaterial layer. For
comparison, reference values measured without a metamate-
rial layer are indicated by arrows.
In all cases the photoluminescence peak is shifted from
its original position λ
0
towards the respective metama-
3
terial’s absorption resonance λ
abs
. For a large mismatch
∆λ > 150 nm (i.e. D ≥ 620 nm), the photolumines-
cence spectrum becomes non-Gaussian and appears to
develop two peaks close to λ
0
and λ
abs
, respectively. In-
triguingly, the observed photoluminescence red-shift b e-
comes quite large, reaching almost 200 nm. The inten-
sity enhancement and FWHM of the photoluminescence
spectra are summarized in Fig. 2(b) as a function of the
mismatch ∆λ. Narrow photoluminescence spectra with
greatly enhanced intensity are observed when the QD-
luminescence matches the metamaterial resonance wave-
length (i.e. small ∆λ). On the other hand, for a large
mismatch (∆λ > 150 nm) the photoluminescence spec-
trum becomes even broader than it is without the meta-
material layer. Here we note that all metamaterial sam-
ples had almost identical transmission levels at the pump
wavelength (532 nm), and therefore there was no signifi-
cant difference in pump power reaching the active layer.
One might attempt to explain such broadening as a re-
sult of filtering the QD luminescence spectrum through
the metamaterial, thus assuming no plasmon-exciton in-
teraction. However, such a simple explanation can be
ruled out, as it does not explain any photoluminescence
enhancement resulting from the presence of the meta-
material. Furthermore, the convoluted spectrum of QD
luminescence and metamaterial transmission does not
agree well with the photoluminescence measurements.
The metamaterials studied here have profound
polarization-dependent properties. While a strong plas-
monic Fano-resonance is excited by y-polarization [see
Fig. 1(b)], this resonance vanishes for the orthogonal po-
larization [3]. This polarization dependence may be ex-
pected to affect the interaction with the isotropic QDs,
and hence the polarization dependence of the photolu-
minescence was measured as illustrated by Fig. 3 for the
metamaterial with D = 545 nm. By changing the polar-
ization state from y to x, the absorption spectrum be-
comes featureless around λ
0
[Fig. 3(b)]. The correspond-
ing photoluminescence drastically degrades [Fig. 3(a)],
providing additional evidence that the photolumines-
cence spectrum is controlled by the plasmonic resonance.
We argue that the observed enhancement of photolu-
minescence can be understood in terms of the cavity QED
Purcell effect [15, 16]. Indeed, the spontaneous emission
decay rate is proportional to the density of photon states
that the photonic environment offers for spontaneous de-
cay. Thus the internal dynamics of a quantum system
are controlled by a photonic environment that is resonant
with radiative transitions of the source. Enhancement of
the radiation rates has been seen in various systems in-
cluding QDs in nanocavities and photonic crystals. In
our experiments the ensemble of QDs with its exciton
emission line is placed at a resonant plasmonic metama-
terial. The metamaterial creates an environment equiv-
alent to a microcavity with a quality factor Q, and a
mode confined in an ultrasmall volume V that enhances
1000 1150 1300 1450 1600
0
15
30
45
60
Absorption (%)
Wavelength (nm)
Photoluminescence intensity
(arb. units)
(a)
(b)
X
Y
XY
Y
X
XY
FIG. 3: (color online). Polarization dependence of (a) photo-
luminescence and (b) absorption, measured for the metama-
terial with D = 545 nm. Polarizations x and y are introduced
in Fig. 1, while xy is the intermediate polarization.
the density of photon states leading to the Purcell factor
enhancement of luminescence: F
p
=
3
4π
2
¡
λ
n
¢
3
Q
V
. Here n
is the refractive index of the medium and λ is the wave-
length.
For the sake of rough estimate, the mode volume is
calculated by V = 2(a + t)wh (where a = 470 nm, t =
170 nm, w = 65 nm, and h = 50 nm), i.e. the mode is
assumed to be confined in the slits of the metamaterial
metal film of thickness h. With λ = 1300 nm, Q = 11,
and n = 1.48, this gives the following value for the Purcell
factor F
p
= 136. This is of the same order of magnitude
as the experimentally observed enhancement of overall
luminescence µ = 8, corrected for the fraction of QDs in
the slits f = V/(D
2
p), where p = 180 nm is the thickness
of the QD/PMMA layer: F
p,exp
= µ/f = 103. Here
we note that the general Purcell enhancement formula
used in our calculations only gives approximate values
for luminescence enhancement in plasmonic systems [17].
Peculiarity of our experimental conditions in compar-
ison with numerous reports on the Purcell factor en-
hancement of luminescence of individual QDs is in a
large number of QDs located within the mode volume
(∼ 4000): the exciton line is inhomogeneously broadened
due to a natural variation of the QD sizes [Fig. 4(a)].
Here, detuning of the plasmon resonance from the cen-
tre of the exciton emission line leads to the Purcell en-
hancement being applied to the wing of the emission
line as it is clearly manifested by the transformation
of the photoluminescence spectrum presented in Fig. 2.
The photoluminescence spectra resulting from this Pur-
cell enhancement may be expected to be proportional
4
1150 1300 1450 1600
1150 1300 1450 16001000 1150 1300 1450 1600
Wavelength (nm)
D=575nm D=595nm
D=645nm
D=620nm
D=545nm (b) (c) (d)
(e)
(f)
Intensity (arb. units)
(a)
Quantum dots
Metamaterial
Inhomogeneously
broadened
exciton ensemble
Homogeneously
broadened
plasmon resonance
Luminescence
of the coupled system
Exciton-Plasmon
coupling
e-h pairs
FIG. 4: (color online). Nature of photoluminescence change
in the plasmonic metamaterial. (a) Energy diagram of the
QD-metamaterial coupled system. (b)-(f) Comparison of
the measured photoluminescence (data points) with χ
A
(λ) =
P L
0
(λ) × A(λ) (lines) for metamaterials with different unit
cell sizes ranging from D = 545 nm to 645 nm. Here, P L
0
(λ)
is normal QD photoluminescence without metamaterial struc-
ture, and A(λ) is the metamaterial’s absorption spectrum.
to χ
A
(λ) = P L
0
(λ) × A(λ), where P L
0
(λ) is the nor-
mal QD photoluminescence spectrum without metama-
terial structure and A(λ) is the absorption spectrum of
the metamaterial array (a measure of the local density
of states). As illustrated by Fig. 4(b)-(f), χ
A
(λ) (lines)
is in excellent agreement with the measured photolumi-
nescence (data points) in all cases, providing further evi-
dence for the Purcell effect and plasmon-exciton coupling
in the QD-metamaterial system.
We argue that in a coupled QD-plasmonic metamate-
rial system the resonant enhancement of luminescence
can be exploited for increasing optical gain and thus
for the development of compact, low-threshold lasing de-
vices. At the same time it is not clear yet what effect the
profound Purcell enhancement of luminescence has on
the metamaterial’s Joule losses as it reduces the fraction
of energy that is transferred to the plasmonic system.
In summary, we have experimentally demonstrated
multi-fold enhancement and substantial spectral narrow-
ing of photoluminescence from semiconductor quantum
dots resulting from resonant coupling to a plasmonic
metamaterial. We have shown that the intensity en-
hancement and spectral width of the photoluminescence
in the combined system are controlled by the spectral
overlap of the emission peak of free QDs and the metama-
terial’s plasmonic resonance, and thus this effect is linked
to exciton-plasmon coupling between QDs and metama-
terial. The observed photoluminescence enhancement
provides the first and clear demonstration of the cavity
QED Purcell effect in metamaterials.
Financial support of the Engineering and Physical Sci-
ences Research Council, UK is acknowledged.
∗
Electronic address: KenjiD.Tanaka@jp.sony.com
†
Electronic address: erp@orc.soton.ac.uk
‡
Electronic address: niz@orc.soton.ac.uk,www.
metamaterials.org.uk
[1] N. I. Zheludev, Science 328, 582 (2010).
[2] V. Fedotov, A. Tsiatmas, J. H. Shi, R. Buckingham,
P. de Groot, Y. Chen, S. Wang, and N. I. Zheludev, Opt.
Express 18, 9015 (2010).
[3] E. Plum, V. A. Fedotov, P. Kuo, D. P. Tsai, and N. I.
Zheludev, Opt. Express 17, 8548 (2009).
[4] S. Xiao, V. P. Drachev, A. V. Kildishev, X. Ni, U. K.
Chettiar, H.-K. Yuan, and V. M. Shalaev, Nature 466 ,
735 (2010).
[5] M. Wegener, J. L. Garc´ıa-Pomar, C. M. Soukoulis,
N. Meinzer, M. Ruther, and S. Linden, Opt. Express 16,
19785 (2008).
[6] A. Fang, T. Koschny, M. Wegener, and C. M. Soukoulis,
Phys. Rev. B 79, 241104 (2009).
[7] Y. Sivan, S. Xiao, U. K. Chettiar, A. V. Kildishev, and
V. M. Shalaev, Opt. Express 17, 24060 (2010).
[8] N. I. Zheludev, S. L. Prosvirnin, N. Papasimakis, and
V. A. Fedotov, Natur Photon. 2, 351 (2008).
[9] V. A. Fedotov, M. Rose, S. L. Prosvirnin, N. Papasi-
makis, and N. I. Zheludev, Phys. Rev. Lett. 99, 147401
(2007).
[10] B. Luk‘yanchuk, N. I. Zheludev, S. A. Maier, N. J. Halas,
P. Nordlander, H.Giessen, and C. Chong, Nature Mater.
9, 707 (2010).
[11] J. R. Lakowicz, Anal. Biochem. 298, 1 (2001).
[12] K. Aslan, I. Gryczynski, J. Malicka, E. Matveeva, J. R.
Lakowicz, and C. D. Geddes, Curr. Opin. Biotechnol. 16,
55 (2005).
[13] D. J. Bergman and M. I. Stockman, Phys. Rev. Lett. 90,
027402 (2003).
[14] M. I. Stockman, Nature photon. 2, 327 (2008).
[15] K. J. Vahara, Nautre 424, 839 (2004).
[16] L. Novotny and B. Hecht, Principles of Nano-Optics
(Cambridge University Press, Cambridge, 2006).
[17] A. Koenderink, arXiv.org p. 1007.1074v1 (2010).
