www.rsc.org/pccp | Physical Chemistry Chemical Physics
Manipulation of the local density of photonic states to elucidate
fluorescent protein emission rates
Yanina Cesa, a Christian Blum, a Johanna M. van den Broek, b Allard P. Mosk, b
Willem L. Vos* bc and Vinod Subramaniam* a
Received 13th October 2008, Accepted 12th January 2009
First published as an Advance Article on the web 11th February 2009
We present experiments to determine the quantum efficiency and emission oscillator strength of
exclusively the emitting states of the widely used enhanced green fluorescent protein (EGFP). We
positioned the emitters at precisely defined distances from a mirror to control the local density of
optical states, resulting in characteristic changes in the fluorescence decay rate that we monitored by
fluorescence lifetime microscopy. To the best of our knowledge, this is the first emission lifetime control
of a biological emitter. From the oscillation of the observed emission lifetimes as a function of the
emitter to mirror distance, we determined the radiative and nonradiative decay rates of the
fluorophore. Since only the emitting species contribute to the change in emission lifetimes, the rates
determined characterize specifically the quantum efficiency and oscillator strength of the on-states of
the emitter, in contrast to other methods that do not differentiate between emitting and dark states.
The method reported is especially interesting for photophysically complex systems like fluorescent
proteins, where a range of emitting and dark forms has been observed. We have validated the analysis
method using Rhodamine 6G dye, obtaining results in very good agreement with the literature. For
EGFP we determine the quantum efficiency of the on-states to be 72%. As expected for this complex
system, our value is higher than that determined by methods that average over on- and off-states.
The discovery, development, and use of genetically-encodable
visible fluorescent proteins (VFPs) has provided revolutionary
new capabilities to visualize molecular and cellular biological
processes using fluorescence microscopy. 1–3 Despite the
widespread use of VFPs as reporters and sensors in cellular
environments the versatile photophysics of fluorescent proteins
is still subject to intense research. Understanding the
photophysics of fluorescent proteins is essential for accurate
interpretation of the biological and biochemical processes
illuminated by the fluorescent proteins as well as for the
development of biosensors based on fluorescent proteins.
A multitude of studies has addressed the spectral complexity
of fluorescent proteins. Different chromophores can form
within one protein, 4,5 intrinsic and photoinduced spectral shifts
have been observed, 6–8 and it has been demonstrated that
VFPs exhibit rapid blinking as a result of transitions between
emitting and non-emitting states. 9–11 This photophysical
complexity influences the determination of fundamental
emission properties such as the radiative and nonradiative
a Biophysical Engineering Group, Faculty of Science and Technology
and MESA+ Institute for Nanotechnology, University of Twente,
P.O. Box 217, 7500 AE Enschede, The Netherlands.
b Complex Photonic Systems (COPS), Faculty of Science and
Technology and MESA+ Institute for Nanotechnology, University
of Twente, P.O. Box 217, 7500 AE Enschede, The Netherlands.
c Photonic Bandgaps Group, Center for Nanophotonics, FOM Institute
AMOLF, Kruislaan 407, 1098 SJ Amsterdam, The Netherlands
decay rates, fluorescence quantum efficiency and oscillator
strength, since most approaches average over the different
subensembles exhibiting different photophysical properties in
the analyzed sample.
The most widely-used approach determines the quantum
efficiency of an unknown fluorophore by comparing the
wavelength integrated emission intensity of the unknown
sample to a known reference having the same absorbance at
the chosen excitation wavelength. 12 Using this value for
quantum efficiency and the measured fluorescence lifetime,
the radiative and non-radiative decay rates and emission
oscillator strengths can be calculated. However, this approach
is rigorously correct only for systems of identical emitters, and
averages over all spectral species. It does not discriminate
between (a) molecules in emitting states and those in (b)
absorbing, but non-emitting states or (c) in photoinduced dark
states created by the measurement itself. These dark states
bias the determined quantum efficiency and hence affect the
calculated values for the decay rates and oscillator strength.
Fluorescent proteins exhibit blinking attributed to spontaneous
as well as photoinduced transitions between emitting to nonemitting
states, 9–11 and exhibit intrinsic spectral fluctuations
due to chemical or environmental influences upon the
chromophore. 6,13 Consequently, the values for fluorescent
protein quantum efficiencies derived from different studies
vary considerably. A typical example is the red emitting
protein DsRed, for which values for the quantum efficiency
in the literature range from the first pioneering results reporting
23% 14 to the much higher consensus values currently, but
which still range from 68% 15,16 to 79%. 17 Another example is
This journal is c the Owner Societies 2009 Phys.Chem.Chem.Phys., 2009, 11, 2525–2531 | 2525
the green emitting protein Azami green, for which quantum
efficiencies of 67% 18 and 90% 19 were reported. It is therefore
evident that a method is needed for the determination of
quantum efficiencies that is independent of relative measurements
and the presence of non-emitting states.
We present here the first such experiments on fluorescent
proteins using a method to determine radiative and nonradiative
decay rates of an emitter, and thus quantum efficiency and
emission oscillator strength, that characterize the emitting
states of the chromophore and is not dependent on spectral
integration. This method has been applied in other
research fields to erbium ions, 20 semiconductor colloidal
nanocrystals, 21,22 silicon nanocrystals 23 and self-assembled
quantum dots. 24 We control the local density of optical states
by positioning the fluorescent protein at precisely defined
distances from a metallic mirror. From a classical point of view,
if an emitter is placed in front of a metallic mirror the field
emitted directly by the emitter interferes with the field from the
emitter’s mirror image. If the distance from the mirror is such
that the interference is constructive, the local density of states,
and hence the emission rate, will be increased. If on the other
hand, the distance is such that the interference is destructive, the
local density of states, and hence the emission rate, will be
reduced. In this way control over the distance from the emitters
to the mirror corresponds to control of the local density
of states. A classical model taking full account of material
properties of the mirror and dielectric environment was
developed by Chance et al. 25
The distance-dependent modification of the local density of
states results in a characteristic oscillation in the fluorescence
decay rate 25–28 that we monitored by fluorescence lifetime
microscopy. The results were modeled using the single mirror
model 25 yielding the radiative and non-radiative decay rates of
the emitting states, and by extension, a set of relevant
photophysical parameters of the protein including quantum
efficiency and emission oscillator strength. This approach has
the clear advantage that it is based on measurements of
emission properties reflecting purely the characteristics of the
emission transition, rather than those of the absorption as is
the case for other standard techniques to determine these
parameters. In the present approach, no prior knowledge of
absorption characteristics is required, representing a direct
determination of the emission band properties.
We first studied the well-characterized laser dye Rhodamine
6G (Fig. 1a) to validate our sample preparation and
measurement method and to measure the complete set of
photophysical parameters of this important dye. We then
demonstrated that this approach of controlling the local
density of states is suitable for the analysis of photophysically
complex systems like fluorescent proteins. We analyzed the
emission from the widely-used fluorescent protein EGFP
(Fig. 1b) to sensitively measure the decay rates, the quantum
efficiency of the on-states, and the emission oscillator strength.
To control the local density of states of the emitters, we
chose to work in a single mirror configuration. The emitters
are embedded in an isotropic medium and localized at a
Fig. 1 Structure of (a) Rhodamine 6G and (b) the green fluorescent
protein EGFP studied here. Rhodamine 6G is a common and
well-characterized laser dye with relatively simple photophysics that
was used to validate our method. EGFP is a typical representative of
the class of genetically-encodable fluorescent proteins. The emitting
chromophore forms inside the protein barrel and is completely
shielded by the protein scaffold from the external environment.
Fluorescent proteins show very complex photophysics, which makes
the determination of photophysical parameters by conventional
well-defined distance from the metallic mirror. To realize these
conditions we used the sample configuration shown in Fig. 2a.
Mirror and spacer layer fabrication
The mirrors used in this work were fabricated by multilayer
e-gun deposition on a silicon wafer. First a layer of 50 nm of
chromium was deposited to improve the adhesion of the silver
layer that acts as the mirror. The thickness of the silver layer
deposited over the chromium was 200 nm. Finally the SiO 2
layer that acts as spacer was deposited with a controlled
thickness to obtain the selected distance between the emitters
and the silver mirror. The whole deposition process was
carried out in a Balzers BAK 600 e-gun evaporation machine.
The multideposition can be done keeping the vacuum in
the chamber between different material depositions while
optimizing the conditions of the surfaces for the following
deposition. This approach helps to improve the adherence,
especially between the easy to oxidize silver layer and the SiO 2 .
The mirrors and spacer layers were characterized by
scanning electron microscopy (SEM), imaging different
positions of a cross section of each mirror-spacer. A typical
cross section SEM image of the mirror and covering SiO 2 layer
is shown in Fig. 2b. The thickness of the silver mirror and of
the SiO 2 spacer layer was found to be homogeneous over the
analyzed cross sections. The refractive index of the SiO 2
layer was measured with ellipsometry, giving a value of
n = 1.46 0.05 for the different SiO 2 spacer thicknesses.
Sample film and cover
To form a thin layer of the fluorescent molecules over the SiO 2
spacer, the emitting molecules (Rhodamine 6G from Lambda
Physics; EGFP obtained as reported in ref. 6) were embedded
in a polymer matrix to obtain a nearly isotropic orientation of
the molecules in the layer. 1% PVA was selected for this
purpose. A solution of PVA and the corresponding molecule
at low concentrations to exclude dye–dye interactions was spin
coated over the surface. From the original wafer 2 cm 2cm
samples were cut and 100 ml of the PVA solution was spin
2526 | Phys. Chem. Chem. Phys., 2009, 11, 2525–2531 This journal is c the Owner Societies 2009
etween the emitters and the mirrors, used in the analysis, is
defined by the nominal spacer thickness plus half the PVA
layer thickness, which is held constant for all samples.
Finally a thick layer of PMMA (n = 1.489) was deposited
onto the fluorescent layer to match the refractive index of the
spacer and reproduce the conditions of a single mirror. 25 A
drop of B2% PMMA in toluene solution was deposited and
dried over the sample to form a layer 41 mm.
To determine the lifetime of the respective emitter in a
homogeneous medium with the same refractive index of the
spacer used in the mirror samples, we followed the same
sample film and cover production procedure but used a quartz
coverslip as substrate (replacing the mirror) which has a
refractive index (n = 1.46).
Emission lifetime determination
Fig. 2 (a) Schematic of the sample design used to modulate the local
density of states. A thin Cr layer deposited on a silicon substrate
improves the adherence of the silver layer that acts as a mirror. A SiO 2
layer of well-defined thickness in the range 50 nm to 300 nm is
deposited on the mirror and acts as a spacer. The thin o20 nm
PVA layer containing the emitters is spincoated on top of the spacer.
Finally a thick layer of PMMA with a matching refractive index to
SiO 2 is used to create an optically isotropic environment for the
emitters. (b) Scanning electron micrograph of a crossection of the
mirror and spacer layer. The three layers (from bottom up: Cr, Ag and
SiO 2 ) are clearly distinguishable. The thicknesses of silver and SiO 2
layers are determined from these images. (c) AFM image of the spin
coated PVA layer spanning a region containing a scratch made on the
PVA layer. Height image (top) clearly shows the scratch, where PVA
was removed. From the averaged height profile (over the white area
indicated in the image) the thickness of the PVA layer can be
determined as 17 3 nm. The ‘‘hills’’ at the side of the scratch
originate from PVA being piled up by the scratching of the layer.
coated onto the sample at 6000 rpm for 10 s. The thickness of
the resulting PVA layer was characterized by atomic force
microscopy spanning a region containing a scratch made on
the layer with a Teflon tweezers, yielding a homogeneous layer
of 17 3 nm. A typical AFM image with the corresponding
averaged height profile is presented in Fig. 2c. The distance
The fluorescence decay curves were measured with a time
correlated single photon counting (TCSPC) system. The
sample was illuminated with a pulsed diode laser with a pulse
duration of 50 ps (FWHM) and a repetition rate 20 MHz
(BDL475, 469 nm, B&H, Germany). Epi illumination
(40, NA 0.6) was used, collecting the emission through the
same objective. A 15 nm band pass emission filter centered at
556 nm was used for Rhodamine 6G and a 10 nm band pass
emission filter centered at 510 nm was used for EGFP in the
detection path to limit the detection wavelength to the peak of
the emission spectrum. To eliminate any influence by scattered
or back-reflected excitation light, an additional long pass filter
(razor edge 473.0 nm, Semrock, USA) was inserted into the
detection path. The emission is focused to the active area of a
single photon avalanche diode (PDM series, MPD, Italy, peak
quantum efficiency of B50% at 500 nm) connected to the
TCSPC module (SPC-830, Becker & Hickl, Germany) to
perform time resolved fluorescence measurements. Integration
times per decay curve were chosen to yield a total of
approximately 20 000 counts per decay curve to assure
accurate determination of the characteristic lifetime (t) 29 by
the TCSPC card software (SPCimage, B&H, Germany).
3. Results and discussion
We first tested our approach by measuring a series of
Rhodamine 6G samples. A typical fluorescence decay curve
and fit for a Rhodamine 6G sample on a mirror with 150 nm
emitter–mirror distance is presented in Fig. 3. The data
obtained were fitted by a monoexponential decay function.
The quality of the fit was judged by the w 2 parameter, where a
w 2 = 1 is characteristic of an accurate fit. The residuals plotted
at the bottom of Fig. 3 show that the fitting is good in the
whole range of the decay.
Since small variations in the thickness of the layers defining
the distance to the mirror, as well as variations in the fits of the
data will give rise to changes in the recorded lifetimes, it is
necessary to collect a statistically relevant number of decay
curves for each sample. To achieve this, we collected five
separate fluorescence lifetime images at different positions of
each sample representing one emitter–mirror distance, thus
This journal is c the Owner Societies 2009 Phys.Chem.Chem.Phys., 2009, 11, 2525–2531 | 2527
to the lifetime of the emitter. The small FWHM shows the
good quality of our sample preparation and measurement.
The most frequent lifetime of 3.33 ns is the peak of the
Gaussian distribution for a Rhodamine 6G–mirror distance
of 150 nm. This value is used to calculate the decay rate for the
respective distance. For all the distances measured, the lifetime
presents a similar Gaussian distribution, with a FWHM of
0.090 0.009 ns. The peak positions vary as expected with the
emitter–mirror distance, corresponding to the expected
modulation of the decay rate.
Using the same measurement procedure described above,
we obtained an emission lifetime of 3.3 ns, corresponding to a
decay rate of 0.30 ns 1 , for Rhodamine 6G in a homogeneous
medium of refractive index of n = 1.46. This rate was then
used as a normalization factor for the analysis. The normalized
decay rates for Rhodamine 6G are plotted as a function of the
distance to the mirror in Fig. 4a. The error bars on the abscissa
(the distance to the mirror) are defined as the sum
of the uncertainties in SiO 2 and PVA layer thicknesses
(Dz = 6 nm), while the error bars for the decay rate were
calculated using the half-width of the lifetime (Dt = 0.045 ns).
In Fig. 4a we present the normalized decay rate of Rhodamine
6G versus the distance to the mirror. The results are modelled
by the single mirror configuration model. 25 Two main effects
Fig. 3 (a) Typical decay curve measured for Rhodamine 6G at a
distance of 150 nm to the mirror. The data (dark grey dots) was fitted
with a mono exponential decay (black line). The light grey curve
corresponds to the instrument response function (IRF) of the system.
The residuals are plotted at the bottom and show the very good quality
of the fit in the whole decay range. (b) Typical lifetime image, showing
the homogeneity of the sample. Each pixel represents the fitted lifetime
of a curve like the one plotted in (a). Several images were taken into
account to determine the lifetime distribution for each distance to the
mirror. (c) Lifetime distribution histogram. The histogram was built
from the 5120 constituent single pixel decay curves from 5 lifetime
images like the one plotted in (b). The distribution is very narrow and
shows a Gaussian shape with a FWHM of 0.087 ns.
accounting for any possible inhomogeneities in the emitter–
mirror distance within one sample. Each image contained
32 32 decay curves measured on an area of 16 16 mm,
see Fig. 3b. The decay times extracted from the monoexponential
fitting of the 5120 constituent single pixel decay
curves were collected in a single histogram that represents the
measured lifetime distribution for each distance emitter–
mirror, see Fig. 3c. From the distribution we can ascertain
the most frequent lifetime as well as the width of the
distribution as a measure of the quality of the lifetime
determination for each dye-mirror distance.
Fig. 3c shows a narrow distribution of measured lifetimes
that can be fitted by a Gaussian distribution. The width of the
distribution is 0.087 ns FWHM, which is very small compared
Fig. 4 (a) Normalized decay rate vs. the distance to the mirror for
Rhodamine 6G at l = 555 nm. The data are compared to theory 25 for
different orientations of the emitter dipoles with respect to the mirror
surface (n Ag (l = 555 nm) = 0.129 + i3.25; Q = 95%): parallel
(long dashed), perpendicular (short dashed) and isotropic distribution
(solid line). The experimental points are followed by the predicted
behaviour for isotropic dipole orientation for distances to the mirrors
Z 110 nm. Below 110 nm Rhodamine 6G–mirror distance the
measured decay rates are below the predicted values (grey points),
for unknown reasons. (b) Total decay rate as function of the
normalized LDOS. A linear relation between the measured decay rate
g(o,z) for Rhodamine 6G and the calculated normalized LDOS can be
observed as expected according to Fermi’s golden rule. The solid line
represents a linear fit with a nonradiative decay g nrad =0.015 0.02 ns
and a radiative decay rate g hom
rad = 0.29 0.02 ns. The resulting
quantum efficiency is Q = 95% 5%.
2528 | Phys. Chem. Chem. Phys., 2009, 11, 2525–2531 This journal is c the Owner Societies 2009
are expected. First, the decay rates oscillate with the increase
of the distance to the mirror, since the phase of the reflected
field changes with this distance. Second, the oscillation
amplitude decreases with increasing distance, since the
radiation field of the dipole decreases with the distance
between the emitter and its mirror image. Similar behaviour
is predicted for different transition dipole orientations with
respect to the mirror. Indeed a damped oscillation of the
normalized decay rate as a function of the distance to the
mirror is observed in Fig. 4a, consistent with the prediction by
theory for a single mirror configuration model.
For our preparation method of embedding emitters in a
polymer film we expect an isotropic distribution of the
emission dipole orientations with respect to the mirror.
Fig. 4 shows that for emitter–mirror distances Z 110 nm,
the oscillation in measured data indeed follows the values
predicted for isotropically distributed emitters (solid line).
Fig. 4 also depicts the normalized decay rates calculated for
the parallel (long dashed) and perpendicular (short dashed)
orientations. The calculation was performed for the
wavelength corresponding to the Rhodamine 6G emission
peak (555 nm), the refractive indexes for silver at the
corresponding wavelength (n Ag (l = 555 nm) = 0.129+i3.25 30 )
and a quantum efficiency Q = 95% (value obtained from
analysis below)). For emitter–mirror distances less than
110 nm the measured decay rates are below the predicted
values (grey data points in Fig. 4a), even considering a
deviation from the isotropic distribution orientation and all
transition moments being oriented parallel to the mirror. For
these emitter–mirror distances, we repeated the experiment
and reproduced the same decay rates. The observed deviation
from the model was only found for Rhodamine6G, and
not for the fluorescent protein EGFP (see below). While
no explanation has been found for this discrepancy yet,
these two data points were tentatively excluded for the
The measured decay rate g(w,z) is the sum of two contributions,
the non-radiative decay rate g nrad (o) and the radiative decay
rate g rad (w,z). The radiative decay rate, according to Fermi’s
golden rule, 31 is proportional to the projected local density of
optical states r(w,z) (LDOS) and depends explicitly on the
distance to the mirror z, and the transition frequency o.
Therefore the total decay rate can be expressed as
gðo; zÞ ¼g nrad ðoÞþg hom rðo; zÞ
rad ðoÞ ð1Þ
r hom ðoÞ
where g hom
rad (o)andr hom (o) are the decay rate and corresponding
LDOS respectively in a homogeneous medium of the
corresponding refractive index. The total decay rate is plotted
as a function of the normalized LDOS in Fig. 4b. The error
bars for the decay rate are based on an uncertainty in
the lifetime of half-width of the lifetime distribution
(Dt = 0.045 ns) and for the normalized LDOS are based on
the interval corresponding to the uncertainty Dz in the distance
to the mirror.
A linear relation was found between the measured decay
rate g(w,z) for Rhodamine 6G and the calculated normalized
LDOS as expected from Fermi’s golden rule. The data
were fitted by a linear function (Fig. 4b), yielding for the
non-radiative decay rate a value of g nrad = 0.015 0.02 ns 1
and for the decay rate in a homogeneous medium
rad = 0.29 0.02 ns 1 . This results in a quantum efficiency
of Q = 95% 5%. Although Rhodamine 6G was embedded
in polyvinylalcohol in our study, the determined quantum
efficiency can be compared with values determined for
Rhodamine 6G in ethanol, since ethanol and polyvinylalcohol
provide a chemically very similar environment to the emitter.
Indeed our value and the values reported in the literature for
the quantum efficiency of Rhodamine 6G in ethanol
determined with different methods, ranging from 0.94 to
0.96, 32–36 agree very well.
Inasmuch as our method only addresses emitting or ‘‘on’’
states of the fluorophore, the correspondence of our result for
Q with the literature reports reflects the relatively non-complex
photophysics of Rhodamine 6G. Although dark states have
been observed for Rhodamine 6G, these states have been
found to be linked to the triplet state and a dark state formed
through the triplet state. 37 Although the intersystem crossing
rates to the triplet state were found to vary depending on the
solvent used, 38 the probability for inter-system crossing to the
triplet state for Rhodamine 6G is generally low enough to
exclude a noticeable buildup of molecules in the triplet state,
and thus a formation of dark states, under the applied
From the homogeneous decay rate g hom
rad , the emission
oscillator strength f emiss of the transition can be calculated 39
f emiss ðoÞ ¼ 2m ee 0 pc 3
q 2 no 2 g hom
where m e and q are the electron mass and charge respectively,
e 0 is the vacuum permittivity, c is the speed of light and n is the
refractive index of the surrounding media. Here we use the
frequency corresponding to the peak of the emission (555 nm),
in the center of the detection band to calculate the effective
emission oscillator strength of the transition.
The effective emission oscillator strength for Rhodamine 6G
is thus determined to be f emiss = 0.92 0.03. To the best of our
knowledge, no value for the emission oscillator strength has
been reported for Rhodamine 6G. The absorption oscillator
strength has been reported before to be f abs =0.69. 36
Enhanced green fluorescent protein (EGFP)
We show the decay curves from the fluorescent protein EGFP
for two sample–mirror distances in Fig. 5a. The data
correspond to distances to the mirror of 170 nm and 60 nm.
The decay of EGFP emission at 60 nm distance to the mirror is
decidedly faster than the EGFP decay recorded for a
EGFP—mirror distance of 170 nm. Both decay curves were
well-fitted with a monoexponential function (t = 2.26 ns,
w 2 = 1.02 for 170 nm and t = 1.36 ns, w 2 = 1.6 for 60 nm).
Clearly, placing the EGFP molecules at different distances
from the silver mirror induced a change in the emission
lifetimes, as expected from theory. To accurately determine
the characteristic lifetime of EGFP for each distance to
the mirror, fluorescence lifetime imaging was used to
determine the most frequent lifetime and the width of
the lifetime distribution for each distance. The width of the
This journal is c the Owner Societies 2009 Phys.Chem.Chem.Phys., 2009, 11, 2525–2531 | 2529
Fig. 5 (a) Typical decay curve measured for EGFP for distances to
the mirror of 60 nm (circles) and 170 nm (triangles). Curves are
normalized to unity to clearly visualize the difference in lifetime. The
black lines correspond to mono-exponential fits. At shorter distances
to the mirror a shorter lifetime is observed (t = 2.26 ns, w 2 = 1.02 for
170 nm and t = 1.36 ns, w 2 = 1.6 for 60 nm), as predicted by the
theory. The grey circles correspond to the IRF of the system. (b)
Normalized decay rate vs. the distance to the interface for EGFP. The
measured values (circles) follow the predicted behaviour for Q = 72%
and isotropic dipole orientation (solid line), which is clearly different
from the expected modulation for Q = 100% (dashed line). (c) Total
decay rate as a function of the normalized LDOS. The linear fit (solid
line) yields values of the nonradiative decay g nrad = 0.14 0.03 ns 1
and the radiative decay rate g hom
nrad = 0.36 0.03 ns 1 . From these
decay rates the quantum efficiency is calculated to be 72% 6%.
lifetime distributions was found to be narrow (0.08 0.01 ns
FWHM) and could be fitted by Gaussian functions.
In Fig. 5b we present the observed decay rate, derived from
the most frequent lifetime observed, as a function of the
sample–mirror distance. For EGFP we expect to see the
influence of its moderate quantum efficiency of 60% reported
in the literature 40 on the modulation depth of the decay rate as
a function of distance to the mirror. A reduced quantum
efficiency results in a lower modulation depth of the decay
rate. 41 Indeed, we observe this behaviour. The normalized
decay rates show less modulation than expected for unity
quantum efficiency, but agree very well with calculations
considering a quantum efficiency of 72% (derived from
analysis below). The calculated normalized decay rate for
unity quantum efficiency is also shown in Fig. 5b to illustrate
the change in modulation depth, clearly noticeable for shorter
distances. The calculations were performed using the
wavelength corresponding to the EGFP emission peak
(512 nm), and the refractive index for silver at this wavelength
(n Ag (512 nm) = 0.129 + i3.02 30 ).
The total decay rate is plotted as function of the normalized
LDOS (Fig. 5c) and fitted with a linear function from which
the non-radiative decay rate of g nrad = 0.14 0.03 ns 1 and
the decay rate in a homogeneous medium g hom
rad = 0.36
0.03 ns 1 are derived. Using these values we calculate a
quantum efficiency of Q = 72% 6% and effective emission
oscillator strength (eqn (2)) of f emiss = 0.97 0.06.
By modulating the local density of states and observing the
change in emission lifetime, one obtains the photophysical
properties characteristic of the emitting states of the
fluorophore. Non-emitting states do not contribute to the
observed emission lifetime and are thus not accounted for.
Since conventional measurements are adversely affected by
absorbing and non-emitting as well as non-absorbing states
photoinduced during the measurement process, we expected to
find an increased value for the on state quantum efficiency.
Indeed the value for the quantum efficiency we obtain (72%) is
markedly higher than the value of Q = 60% reported
before, 16 consistent with reports that a significant fraction of
green emitting proteins can reside in dark states even at very
low excitation intensities. 42,43 The reported values for the
radiative and nonradiative decay rate for a green fluorescent
protein mutant closely related to the EGFP analyzed in this
study were smaller than the values we determine, which is a
consequence of the lower quantum efficiency used to calculate
the rates in this report. 44 The effective emission oscillator
strength of EGFP we determine is f = 0.97. To the best of
our knowledge no emission or absorbance oscillator strength
has been experimentally determined for any fluorescent
protein to date. The anionic chromophore of the green
fluorescent protein, which is also the chromophore of
EGFP studied here, has been analyzed by computational
methods. 45,46 These studies have shown that the
HOMO–LUMO coupling has the largest absorbance
oscillator strength and thus is the source of the absorption.
The calculated strong HOMO–LUMO coupling agrees well
with the effective emission oscillator strength that we
We have presented for the first time the determination of
fundamental photophysical parameters of the on states of a
fluorescent protein by systematically changing the local
density of optical states. The local density of states is changed
by controlling the spacing between the emitters and a metallic
mirror, which results in characteristic changes of the emission
lifetime. Analyzing the change of emission lifetime as a
function of the emitter distance to the mirror gives access to
the decay rates, quantum efficiency and emission oscillator
strength. Since only emitting states contribute to the emission
lifetime, only the properties of the emitting states are
determined, in contrast to other methods that average over
2530 | Phys. Chem. Chem. Phys., 2009, 11, 2525–2531 This journal is c the Owner Societies 2009
emitting and non-emitting states. The method presented does
not require measurements relative to a reference molecule. We
show that changing the local density of optical states is an
especially potent method to analyze complex photophysical
systems like fluorescent proteins, a class of molecules which
are known to exhibit significant residence times in intrinsic or
photoinduced dark states. We validated our approach with the
photophysically simple dye Rhodamine 6G, and recovered a
quantum efficiency in excellent agreement with literature. In
contrast, for the complex biological fluorescent protein EGFP,
we determined a clearly higher value for the quantum
efficiency of the emitting states exclusively (Q = 72%) than
reported values averaged over on- and off-states (Q = 60%).
The increased value agrees well with reports that a significant
fraction of green emitting proteins can reside in dark states.
Theoretical studies predict a strong HOMO–LUMO coupling,
which agrees well with our measurement.
The consistency between our measurements and previously
reported values from both experimental as well as theoretical
determinations, serve to validate the method as a new and
potentially very powerful manner to obtain fundamental
insights into the complex fluorescent protein emission
properties. We even envision the individual characterisation
of different emitting forms coexisting in many fluorescent
proteins. In this case, the decay characteristics are
multiexponential, and the influence of a modified local density
of photonic states on each individual decay component can be
We thank Niels Zijlstra and Merel Leistikow for their
contribution to the LDOS calculations and Willem Tjerkstra
for useful discussions. This work is part of the research
program of the ‘‘Stichting voor Fundamenteel Onderzoek
der Materie’’ (FOM), which is supported by the ‘‘Nederlandse
Organisatie voor Wetenschappelijk Onderzoek’’ (NWO).
W.L.V. also thanks NWO-Vici and STW/NanoNed. C.B.,
Y.C. and V.S. also acknowledge support from the MESA +
Institute for Nanotechnology.
1 M. Chalfie, Y. Tu, G. Euskirchen, W. W. Ward and D. C. Prasher,
Science, 1994, 263, 802–805.
2 J. Lippincott-Schwartz and G. H. Patterson, Science, 2003, 300,
3 J. Yu, J. Xiao, X. J. Ren, K. Q. Lao and X. S. Xie, Science, 2006,
4 V. V. Verkhusha, D. M. Chudakov, N. G. Gurskaya, S. Lukyanov
and K. A. Lukyanov, Chem. Biol., 2004, 11, 845–854.
5 S. J. Remington, Curr. Opin. Struct. Biol., 2006, 16, 714–721.
6 C. Blum, A. J. Meixner and V. Subramaniam, Biophys. J., 2004,
7 C. Blum, A. J. Meixner and V. Subramaniam, ChemPhysChem,
2008, 9, 310–315.
8 F. Malvezzi-Campeggi, M. Jahnz, K. G. Heinze, P. Dittrich and
P. Schwille, Biophys. J., 2001, 81, 1776–1785.
9 J. Hendrix, C. Flors, P. Dedecker, J. Hofkens and Y. Engelborghs,
Biophys. J., 2008, 94, 4103–4113.
10 M. F. Garcia-Parajo, J. A. Veerman, R. Bouwhuis, R. Vallee and
N. F. van Hulst, ChemPhysChem, 2001, 2, 347–360.
11 T. B. McAnaney, W. Zeng, C. F. E. Doe, N. Bhanji, S. Wakelin,
D. S. Pearson, P. Abbyad, X. H. Shi, S. G. Boxer and
C. R. Bagshaw, Biochemistry, 2005, 44, 5510–5524.
12 J. Lakowicz, Principles of Fluorescence Spectroscopy, Kluwer
Academic/Plenum Publishers, New York, 2nd edn, 1999.
13 M. Cotlet, J. Hofkens, S. Habuchi, G. Dirix, M. Van Guyse,
J. Michiels, J. Vanderleyden and F. C. De Schryver, Proc. Natl.
Acad. Sci. U. S. A., 2001, 98, 14398–14403.
14 M. V. Matz, A. F. Fradkov, Y. A. Labas, A. P. Savitsky,
A. G. Zaraisky, M. L. Markelov and S. A. Lukyanov, Nat.
Biotechnol., 1999, 17, 969–973.
15 B. J. Bevis and B. S. Glick, Nat. Biotechnol., 2002, 20, 83–87.
16 G. Patterson, R. N. Day and D. Piston, J. Cell Sci., 2001, 114,
17 R. E. Campbell, O. Tour, A. E. Palmer, P. A. Steinbach,
G. S. Baird, D. A. Zacharias and R. Y. Tsien, Proc. Natl. Acad.
Sci. U. S. A., 2002, 99, 7877–7882.
18 S. Karasawa, T. Araki, M. Yamamoto-Hino and A. Miyawaki,
J. Biol. Chem., 2003, 278, 34167–34171.
19 M. H. Dai, H. E. Fisher, J. Temirov, C. Kiss, M. E. Phipps,
P. Pavlik, J. H. Werner and A. R. M. Bradbury, Protein Eng.
Design Selection, 2007, 20, 69–79.
20 E. Snoeks, A. Lagendijk and A. Polman, Phys. Rev. Lett., 1995, 74,
21 X. Brokmann, L. Coolen, M. Dahan and J. P. Hermier, Phys. Rev.
Lett., 2004, 93, 4.
22 M. D. Leistikow, J. Johansen, A. J. Kettelarij, P. Lodahl and
W. L. Vos, Phys. Rev. B, 2009, 79, 045301; arXiv: 0808.3191.
23 R. J. Walters, J. Kalkman, A. Polman, H. A. Atwater and
M. J. A. de Dood, Phys Rev B, 2006, 73, 4.
24 J. Johansen, S. Stobbe, I. S. Nikolaev, T. Lund-Hansen,
P. T. Kristensen, J. M. Hvam, W. L. Vos and P. Lodahl, Phys.
Rev. B, 2008, 77, 4.
25 R. R. Chance, A. Prock and R. Silbey, Wiley & Sons, New York,
ed. I. Prigogine and S. A. Rice, 1978, vol. 37, pp. 1–64.
26 R. M. Amos and W. L. Barnes, Phys. Rev. B., 1997, 55, 7249–7254.
27 R. R. Chance, A. Prock and R. Silbey, J. Chem. Phys., 1974, 60,
28 S. Astilean, S. Garrett, P. Andrew and W. L. Barnes, J. Mol.
Struct. (THEOCHEM), 2003, 651, 277–283.
29 M. Maus, M. Cotlet, J. Hofkens, T. Gensch, F. C. De Schryver,
J. Schaffer and C. A. M. Seidel, Anal. Chem., 2001, 73,
30 E. D. Palik, Handbook of Optical Constants of Solids, Academic
Press, London, 1998.
31 R. Loudon, The Quantum Theory of Light, Oxford University
Press, Oxford, New York, 3rd edn, 2000.
32 K. H. Drexhage, Dye Lasers, Springer-Verlag, Berlin, New York,
33 R. F. Kubin and A. N. Fletcher, J. Lumin., 1982, 27, 455–462.
34 J. Arden, G. Deltau, V. Huth, U. Kringel, D. Peros and
K. H. Drexhage, J. Lumin., 1991, 48–9, 352–358.
35 M. Fischer and J. Georges, Chem. Phys. Lett., 1996, 260,
36 S. Delysse, J. M. Nunzi and C. Scala-Valero, Appl. Phys. B, 1998,
37 R. Zondervan, F. Kulzer, S. B. Orlinskii and M. Orrit, J. Phys.
Chem. A, 2003, 107, 6770–6776.
38 J. Widengren, U. Mets and R. Rigler, J. Phys. Chem., 1995, 99,
39 A. E. Siegman, Lasers, University Science Books, Mill Valley, CA,
40 G. H. Patterson, S. M. Knobel, W. D. Sharif, S. R. Kain and
D. W. Piston, Biophys. J., 1997, 73, 2782–2790.
41 S. Astilean and W. L. Barnes, Appl. Phys. B, 2002, 75, 591–594.
42 G. Jung, S. Mais, A. Zumbusch and C. Brauchle, J. Phys. Chem. A,
2000, 104, 873–877.
43 G. Jung and A. Zumbusch, Microsc. Res. Tech., 2006, 69, 175–185.
44 Y. X. Liu, H. R. Kim and A. A. Heikal, J. Phys. Chem. B, 2006,
45 V. Helms, C. Winstead and P. W. Langhoff, J. Mol. Struct.
(THEOCHEM), 2000, 506, 179–189.
46 A. K. Das, J. Y. Hasegawa, T. Miyahara, M. Ehara and
H. Nakatsuji, J. Comput. Chem., 2003, 24, 1421–1431.
This journal is c the Owner Societies 2009 Phys.Chem.Chem.Phys., 2009, 11, 2525–2531 | 2531