1:
2: %\documentclass[aps,prl,twocolumn,groupedaddress]{revtex4}
3: \documentclass[aps,prb,preprint,superscriptaddress]{revtex4}
4: %\documentclass[aps,prb,twocolumn,superscriptaddress]{revtex4}
5:
6: \bibliographystyle{apsrev}
7: \usepackage[dvips]{graphicx}
8: %\usepackage[dvips]{color}
9:
10: \begin{document}
11:
12: \title{Statistical analysis of time-resolved emission from ensembles of semiconductor quantum dots: interpretation of exponential decay models}
13:
14: \author{A. F. van Driel}
15: %\homepage[]{Your web page}
16: %\thanks{}
17: %\altaffiliation{}
18: \affiliation{Debye Institute, Utrecht University, P.O. Box 80 000,
19: 3508 TA Utrecht, The Netherlands}
20:
21: \author{I. S. Nikolaev}
22: \affiliation{Center for Nanophotonics, FOM Institute for Atomic
23: and Molecular Physics (AMOLF), 1098 SJ Amsterdam, The
24: Netherlands}\affiliation{Complex Photonic Systems (COPS),
25: Department of Science and Technology and MESA+ Research Institute,
26: University of Twente, 7500 AE Enschede, The Netherlands}
27: \author{P. Vergeer}
28: \affiliation{Debye Institute, Utrecht University, P.O. Box 80 000,
29: 3508 TA Utrecht, The Netherlands}
30: \author{P. Lodahl}
31: \affiliation{Complex Photonic Systems (COPS), Department of
32: Science and Technology and MESA+ Research Institute, University of
33: Twente, 7500 AE Enschede, The
34: Netherlands}\affiliation{COM$\cdot$DTU Department of
35: Communications, Optics, and Materials, Nano$\cdot$DTU, Technical
36: University of Denmark, Denmark}
37: \author{D. Vanmaekelbergh}
38: \affiliation{Debye Institute, Utrecht University, P.O. Box 80 000,
39: 3508 TA Utrecht, The Netherlands}
40: \author{W. L. Vos}\email[Electronic address: ]{w.l.vos@utwente.nl}
41: \homepage[Webpage: ]{www.photonicbandgaps.com}
42: %\altaffiliation{}
43: \affiliation{Center for Nanophotonics, FOM Institute for Atomic
44: and Molecular Physics (AMOLF), 1098 SJ Amsterdam, The
45: Netherlands}\affiliation{Complex Photonic Systems (COPS),
46: Department of Science and Technology and MESA+ Research Institute,
47: University of Twente, 7500 AE Enschede, The Netherlands}
48:
49: \date{Prepared for Phys. Rev. B in October, 2006.}
50:
51: \begin{abstract}
52: We present a statistical analysis of time-resolved spontaneous
53: emission decay curves from ensembles of emitters, such as
54: semiconductor quantum dots, with the aim to interpret ubiquitous
55: non-single-exponential decay. Contrary to what is widely assumed,
56: the density of excited emitters and the intensity in an emission
57: decay curve are not proportional, but the density is a
58: time-integral of the intensity. The integral relation is crucial
59: to correctly interpret non-single-exponential decay. We derive the
60: proper normalization for both a discrete, and a continuous
61: distribution of rates, where every decay component is multiplied
62: with its radiative decay rate. A central result of our paper is
63: the derivation of the emission decay curve in case that both
64: radiative and non-radiative decays are independently distributed.
65: In this case, the well-known emission quantum efficiency can not
66: be expressed by a single number anymore, but it is also
67: distributed. We derive a practical description of
68: non-single-exponential emission decay curves in terms of a single
69: distribution of decay rates; the resulting distribution is
70: identified as the distribution of total decay rates weighted with
71: the radiative rates. We apply our analysis to recent examples of
72: colloidal quantum dot emission in suspensions and in photonic
73: crystals, and we find that this important class of emitters is
74: well described by a log-normal distribution of decay rates with a
75: narrow and a broad distribution, respectively. Finally, we briefly
76: discuss the Kohlrausch stretched-exponential model, and find that
77: its normalization is ill-defined for emitters with a realistic
78: quantum efficiency of less than 100 $\%$.
79: \end{abstract}
80:
81: % insert suggested PACS numbers in braces on next line
82: \pacs{}
83: % insert suggested keywords - APS authors don't need to do this
84: \keywords{}
85:
86: %\maketitle must follow title, authors, abstract, \pacs, and \keywords
87: \maketitle
88: % body of paper here - Use proper section commands
89: % References should be done using the \cite, \ref, and \label commands
90:
91: \section{Introduction}
92: Understanding the decay dynamics of excited states in emitters
93: such as semiconductor quantum dots is of key importance for
94: getting insight in many physical, chemical and biological
95: processes. For example, in biophysics the influence of F\"orster
96: resonance energy transfer on the decay dynamics of donor molecules
97: is studied to quantify molecular
98: dynamics\cite{Lakowicz,2005Medintz}. In cavity quantum
99: electrodynamics, modification of the density of optical modes
100: (DOS) is quantified by measuring the decay dynamics of light
101: sources. According to Fermi's 'Golden Rule' the radiative decay
102: rate is proportional to the DOS at the location of the
103: emitter\cite{Loudon}. Nanocrystalline quantum
104: dots\cite{2002Crooker,2002Zhang,2005Medintz},
105: atoms\cite{1970Drexhage,1997Amos} and dye
106: molecules\cite{2002Danz,2002Astilean} are used as light sources in
107: a wide variety of systems. Examples of such systems are many
108: different kinds of photonic materials, including metallic and
109: dielectric
110: mirrors\cite{1970Drexhage,1997Amos,2002Danz,2002Astilean,2002Zhang},
111: cavities\cite{2001Bayer}, metallic
112: films\cite{2005Biteen,2005Song}, two\cite{2005Fujita,2005Kress}-,
113: and three-dimensional\cite{2004Lodahl} photonic crystals.
114:
115: \begin{figure}
116: \includegraphics[width=0.8\columnwidth]{figure1.eps}
117: \caption{\label{Scheme} Schematic of the relation between decay of
118: an excited state $X^{*}$ to the ground state $X$ and experimental
119: observable parameters. The density of emitters in the excited
120: state is equal to $c(t)$ which can be probed by transient
121: absorption. The emitted light intensity as a function of time
122: $f(t)$ is recorded in luminescence decay measurements. In
123: photothermal measurements the released heat $(g-f)(t)$ after
124: photoexcitation is detected. $g(t)$ describes the total decay,
125: i.e., the sum of the radiative and the non-radiative decay. }
126: \end{figure}
127:
128: Figure \ref{Scheme} shows how observable parameters are related to
129: the decay of an excited state $X^*$ to the ground state $X$. In
130: photoluminescence lifetime measurements the decay of the number of
131: excited emitters is probed by recording a photoluminescence decay
132: curve ($f(t$)). The number of excited emitters $c(t)$ can be
133: probed directly by transient absorption
134: measurements\cite{1995Foggi,1998Klimov,2000Neuwahl} and
135: non-radiative decay ($g-f)(t)$ can be recorded with photothermal
136: techniques\cite{Rosencwaig,2003Grinberg} (see Fig. \ref{Scheme}).
137: $g(t)$ is here defined as the \emph{total} intensity, i.e., the
138: sum of the radiative and non-radiative processes. In this paper we
139: discuss photoluminescence lifetime measurements, which are
140: generally recorded by
141: time-correlated-single-photon-counting\cite{Lakowicz}. The decay
142: curve $f(t)$ consists of a histogram of the distribution of
143: arrival times of single photons after many excitation-detection
144: cycles\cite{Lakowicz}. The histogram is modelled with a
145: decay-function from which the decay time of the process is
146: deduced.
147:
148: In the simplest case when the system is characterized by a single
149: decay rate $\Gamma$, the decay curve is described by a
150: single-exponential function. However, in many cases the decay is
151: much more complex and strongly differs from single-exponential
152: decay\cite{1986James,1990Siemiarczuk,1990Brochon,1995Foggi,2002Crooker,2003Wlodarczyk,2004Wuister,2004Lodahl}.
153: This usually means that the decay is characterized by a
154: distribution of rates instead of a single rate\footnote{In the
155: case of strong coupling in cavity quantum electrodynamics, the
156: decay of even a single emitter is not single-exponential.
157: Experimental situations where this may be encountered are emitters
158: in a high-finesse cavity, or van Hove singularities in the LDOS of
159: a photonic crystal.}. For example, ensembles of quantum dots in
160: photonic crystals experience the spatial and orientational
161: variations of the projected LDOS explaining the
162: non-single-exponential character of the decay\cite{2006Nikolaev}.
163: It is a general problem to describe such relaxation processes
164: which do not follow a simple single-exponential decay. Sometimes
165: double- and triple-exponential models are justified on the basis
166: of prior knowledge of the emitters\cite{Lakowicz}. However, in
167: many cases no particular multi-exponential model can be
168: anticipated on the basis of physical knowledge of the system
169: studied and a decision is made on basis of quality-of-fit.
170:
171: Besides multi-exponential models, the stretched-exponential model
172: or Kohlrausch function\cite{1854Kohlrauscha} is frequently
173: applied. The stretched-exponential function has been applied to
174: model diffusion processes\cite{2001Deschenes}, dielectric
175: relaxation\cite{1980Lindsey}, capacitor
176: discharge\cite{1854Kohlrauschb}, optical Kerr effect
177: experiments\cite{2004Torre} and luminescence
178: decay\cite{2002Schlegel,2003Chen,2004Fisher}. The physical origin
179: of the apparent stretched-exponential decay in many processes
180: remains a source of intense
181: debate\cite{1985Huber,1991Alvarez,2002Lee}.
182:
183: Surprisingly, in spite of the rich variety of examples where
184: non-single-exponential decay appears, there is no profound
185: analysis of the models available in the literature. Therefore, we
186: present in this paper a statistical analysis of time-resolved
187: spontaneous emission decay curves from ensembles of emitters with
188: the aim to interpret ubiquitous non-single-exponential decay.
189: Contrary to what is widely assumed, the density of excited
190: emitters $c(t)$ and the intensity in an emission decay curve
191: ($f(t)$ or $g(t)$) are not proportional, but the density is a
192: time-integral of the intensity. The integral relation is crucial
193: to correctly interpret non-single-exponential decay. We derive the
194: proper normalization for both a discrete, and a continuous
195: distribution of rates, where every decay component is multiplied
196: with its radiative decay rate. A central result of our paper is
197: the derivation of the emission decay curve $f(t)$ in case that
198: both radiative and non-radiative decays are independently
199: distributed. In this most general case, the well-known emission
200: quantum efficiency is also distributed. Distributed radiative
201: decay is encountered in photonic media\cite{2006Nikolaev}, while
202: distributed non-radiative decay has been reported for colloidal
203: quantum dots\cite{2002Schlegel,2004Fisher} and powders doped with
204: rare-earths\cite{2005Vergeer}. We derive a practical description
205: of non-single-exponential emission decay curves in terms of the
206: distribution of total decay rates weighted with the radiative
207: rates. Analyzing decay curves in terms of distributions of decay
208: rates has the advantage that information on physically
209: interpretable rates is readily available, as opposed to previously
210: reported analysis in terms of lifetimes. We apply our analysis to
211: recent examples of colloidal quantum dot emission in suspensions
212: and in photonic crystals. We find excellent agreement with a
213: log-normal distribution of decay rates for such quantum dots. In
214: the final Section, we discuss the Kohlrausch stretched-exponential
215: model, and find that its normalization is ill-defined for emitters
216: with a realistic quantum efficiency of less than 100 $\%$.
217:
218: \section{Decay models}
219: \subsection{Relation between the concentration of emitters and the decay curve}
220: A decay curve is the probability density of emission which is
221: therefore modelled with a so-called probability density
222: function\cite{Dougherty}. This function tends to zero in the limit
223: $t\rightarrow\infty$. The decay of the fraction of excited
224: emitters $\frac{c(t^{'})}{c(0)}$ at time $t'$ is described with a
225: reliability function or cumulative distribution function
226: $\left(1-\frac{c(t^{'})}{c(0)}\right)$\cite{Dougherty}. Here
227: $c(0)$ is the concentration of excited emitters at $t'=0$. The
228: reliability function tends to one in the limit $t^{'}\rightarrow
229: \infty$ and to zero in the limit $t^{'}\rightarrow 0$. The
230: fraction of excited emitters and the decay curve, i.e., the
231: reliability function and the probability density
232: function\cite{Dougherty}, are related as follows:
233: \begin{equation}
234: \label{master}\int_{0}^{t^{'}} g(t)dt=1-\frac{c(t^{'})}{c(0)}
235: \end{equation}
236: Physically this equation means that the decrease of the
237: concentration of excited emitters at time $t^{'}$ is equal to the
238: integral of all previous decay events. Or equivalently: the total
239: intensity $g(t)$ is proportional to the time-derivative of the
240: fraction of excited emitters. As an illustration, Fig.
241: \ref{StrExpFg} shows a non-single-exponential decay function
242: simultaneously with the corresponding decay curve. It is clear
243: that both curves are strongly different. In many reports, however,
244: the distinction between the reliability function and the
245: probability density function is neglected: the intensity of the
246: decay curve $g(t)$ is taken to be directly proportional to the
247: fraction of excited emitters $\frac{c(t^{'})}{c(0)}$. This
248: proportionality only holds for single-exponential decay and not
249: for non-single-exponential decay, which has important consequences
250: for the interpretation of non-single-exponential decay.
251: \begin{figure}
252: \includegraphics[width=0.8\columnwidth]{figure2.eps}
253: \caption{\label{StrExpFg}(color online) Plot of a
254: non-single-exponential decay of the fraction $\frac{c(t)}{c(0)}$
255: (black solid curve, left axis) and the corresponding total
256: intensity decay curve $g(t)$ (red dashed curve, right axis). The
257: curves that describe the fraction of excited emitters and the
258: corresponding intensity decay curve are strongly different. In
259: this example, $\frac{c(t)}{c(0)}$ is the Kohlrausch
260: stretched-exponential decay of the fraction (Eq. \ref{str1}, black
261: solid curve) and $g(t)$ the corresponding decay curve (Eq.
262: \ref{str2}, red dashed curve). We have taken $\beta$=0.5 and
263: $\Gamma_{str}=1$.}
264: \end{figure}
265:
266: \subsection{Single-exponential decay}
267: In this section, we will illustrate some concepts with the
268: well-known single-exponential model. We will also indicate which
269: features of single-exponential decay will break down in the
270: general case of non-single-exponential decay. It is well known
271: that in case of first-order kinetics the rate of decrease of the
272: concentration is constant in time:
273: \begin{equation}
274: \frac{d\,c(t^{'})}{dt^{'}}=-\Gamma c(t^{'})
275: \end{equation}
276: where $\Gamma$ is the decay rate of the process. As a consequence,
277: the concentration $c(t^{'})$ decreases single-exponentially in
278: time:
279: \begin{equation}\label{fracsinexp}
280: \frac{c(t^{'})}{c(0)}=\exp(-\Gamma t^{'})
281: \end{equation}
282: \begin{figure}
283: \includegraphics[width=0.8\columnwidth]{figure3.eps}
284: \caption{\label{SinExp}(color online) Luminescence decay curve of
285: emission from a dilute suspension of CdSe quantum dots (open dots,
286: left axis). Data were collected at the red side of the emission
287: maximum of the suspension, at $\lambda$ = 650 $\pm$ 5 $nm$.
288: Single-exponential modelling (red dashed curve, right axis) yields
289: a decay time of 39.0$\pm2.8ns$ and a $\chi_{r}^{2}$ of 1.12. The
290: average photon arrival time $<t>$, calculated with Eq.
291: \ref{sinexp2}, is 39.1 $ns$.}
292: \end{figure}
293: The mathematical expression for the luminescence decay curve can
294: be obtained by inserting Eq. \ref{fracsinexp} into Eq.
295: \ref{master}, where $\Gamma$ is identified with the total decay
296: rate $\Gamma_{tot}$, resulting in:
297: \begin{equation}
298: \label{master2} g(t)=\Gamma_{rad}\exp(-\Gamma_{tot}
299: t)+\Gamma_{nrad}\exp(-\Gamma_{tot} t)
300: \end{equation}
301: where $\Gamma_{rad}$ is the radiative decay rate, $\Gamma_{nrad}$
302: is the nonradiative decay rate and $\Gamma_{tot}$ is the total
303: decay rate with $\Gamma_{tot}$ = $\Gamma_{rad}$ + $\Gamma_{nrad}$.
304: In a luminescence decay measurement the recorded signal is
305: proportional to the first term of $g(t)$ only which is $f(t)$:
306: \begin{equation}
307: \label{sinexp1} f(t)=\alpha\Gamma_{rad}\exp(-\Gamma_{tot} t)
308: \end{equation}
309: and therefore a single-exponential luminescence decay process is
310: modelled with Eq. \ref{sinexp1}. The pre-exponential factor
311: $\alpha$ is usually taken as adjustable parameter, and it is
312: related to several experimental parameters, i.e., the number of
313: excitation-emission cycles in the experiment, the
314: photon-collection efficiency and the concentration of the emitter.
315: Henceforth $\alpha$ will be omitted in our analysis. A comparison
316: between Eqs. \ref{sinexp1} and \ref{fracsinexp} shows that in the
317: case of pure single-exponential decay neglect of the distinction
318: between the reliability function (Eq. \ref{fracsinexp}) and the
319: probability density function (Eq. \ref{sinexp1}) has no important
320: consequences, since both the fraction and the decay curve are
321: single-exponential. As Fig. \ref{StrExpFg} shows, this neglect
322: breaks down in the case of non-single-exponential decay.
323:
324: Figure \ref{SinExp} shows a luminescence decay curve of a dilute
325: suspension of CdSe quantum dots in chloroform at a wavelength of
326: $\lambda$ = 650 $\pm$ 5 $nm$\cite{2005vanDriel}, with the number
327: of counts on the ordinate and the time on the abscissa. Clearly,
328: the data agree well with single-exponential decay as indicated by
329: the quality-of-fit $\chi_{r}^{2}$ of 1.12, close to the ideal
330: value of 1. This means that all individual quantum dots that emit
331: light in this particular wavelength-range do so with the same rate
332: of $\frac{1}{39.0}$ $ns^{-1}$. It appears that the rate of
333: emission strongly depends on the emission frequency and that it is
334: determined by the properties of the bulk semiconductor
335: crystal\cite{2005vanDriel}.
336:
337: Since $f(t)$ as given by Eq. \ref{sinexp1} is a probability
338: density function, the probability of emission in a certain
339: time-interval can be deduced by integration. The total probability
340: for emission at all times between $t=0$ and $t\rightarrow\infty$
341: is given by
342: \begin{equation}
343: \label{norml} \int_{0}^\infty f(t)dt=\int_{0}^\infty \Gamma_{rad}
344: \exp(-\Gamma_{tot} t)dt=\frac{\Gamma_{rad}}{\Gamma_{tot}}
345: \end{equation}
346: which is equal to the luminescence quantum efficiency. The
347: luminescence quantum efficiency is defined as the probability of
348: emission after excitation\cite{Lakowicz}. The correct recovery of
349: this result in Eq. \ref{norml} shows that Eq. \ref{sinexp1} is
350: properly normalized.
351:
352: The average arrival time of the emitted photons or the average
353: decay time can be calculated by taking the first moment of Eq.
354: \ref{sinexp1}:
355: \begin{equation}
356: \label{sinexp2} <t>=\tau_{av}=\frac{\int_{0}^\infty
357: f(t)tdt}{\int_{0}^\infty f(t)dt}= \frac{1}{\Gamma_{tot}}
358: \end{equation}
359: Only in the case of single-exponential decay the average decay
360: time $<t>$ is equal to the inverse of the total decay rate
361: $\Gamma_{tot}$. The average arrival time for the data in Fig.
362: \ref{SinExp} was $<t>=39.1$ ns, very close to the value of 39.0
363: $\pm2.8ns$ obtained from single-exponential modelling, which
364: further confirms the single-exponential character of the decay of
365: quantum dots in suspension.
366:
367:
368: \subsection{Discrete distribution of decay rates}
369: In contrast to the example shown in Fig. \ref{SinExp}, there are
370: many cases in which decay curves cannot be modelled with a
371: single-exponential function. As an example, Fig. \ref{MultiExp}
372: shows a strongly non-single-exponential decay curve of spontaneous
373: emission from CdSe quantum dots in an inverse opal photonic
374: crystal\cite{2004Lodahl,2006Nikolaev}.
375: \begin{figure}
376: \includegraphics[width=0.8\columnwidth]{figure4.eps}
377: \caption{\label{MultiExp}(color online) Luminescence decay curve
378: of emission from CdSe quantum dots in a titania inverse opal
379: photonic crystal (dots, left axis). The lattice parameter of the
380: titania inverse opal was 340 nm and the emission wavelength
381: $\lambda=595 nm$. (a) A log-normal distribution of rates (Eq.
382: \ref{lognor} and \ref{lognor2}, red dashed curve, right axis)
383: models the data extremely well ($\chi_{r}^{2}$=1.17). The
384: $\Gamma_{mf}$ is 91.7 $\mu s^{-1}$ ($\frac{1}{\Gamma_{mf}}=10.9$
385: $ns$) and the width of the distribution $\Delta\Gamma$ is 0.57
386: $ns^{-1}$. (b) In contrast, a Kohlrausch stretched-exponential
387: model (red dashed curve, right axis) does not fit the data
388: ($\chi_{r}^2=60.7$). The stretched-exponential curve corresponds
389: to $\Gamma_{str}$=96.2 $\mu s^{-1}$ ($\frac{1}{\Gamma_{str}}$=10.4
390: ns), an average decay time $<t>$ of 31.1 $ns$, and a $\beta$-value
391: of 0.42.}
392: \end{figure}
393: If a non-single-exponential decay curve is modelled with a sum of
394: single-exponentials, the decay curve has the following form:
395: \begin{equation}
396: \label{multi1}f(t)=\frac{1}{c(0)}\sum_{i=1}^{n}c_{i}\Gamma_{rad,i}\exp(-\Gamma_{tot,i}t)
397: \end{equation}
398: where $n$ is the number of different emitters (or alternatively
399: the number of different environments of single
400: emitters\cite{2006Nikolaev}), $c_{i}$ is the concentration of
401: emitters that has a radiative decay rate $\Gamma_{rad,i}$, and
402: $c(0)$ is the concentration of excited emitters at $t=0$, i.e.,
403: the sum of all concentrations $c_{i}$. When the different
404: fractions (or environments) are distributed in a particular way, a
405: distribution function $\rho(\Gamma_{tot})$ may be used. Such a
406: function describes the distribution or concentration of the
407: emitters over the emission decay rates at time $t=0$. The fraction
408: of emitters with a total decay rate $\Gamma_{tot,i}$ is equal to
409: \begin{eqnarray}
410: \label{distrib}\frac{c_{i}}{c(0)}&=&\frac{1}{c(0)}\frac{(c(\Gamma_{tot,i-1})+c(\Gamma_{tot,i+1}))}{2}
411: \\&=&\frac{1}{2}\int_{\Gamma_{tot,i-1}}^{\Gamma_{tot,i+1}}\rho(\Gamma_{tot})d\Gamma_{tot}\nonumber
412: \\ &=&\rho(\Gamma_{tot,i})\Delta\Gamma_{tot}\nonumber
413: \end{eqnarray}
414: where $\rho(\Gamma_{tot,i})$ expresses the distribution of the
415: various components $i$ over the rates $\Gamma_{tot,i}$ and has
416: units of inverse rate $s$. $\Delta\Gamma_{tot}$ is the separation
417: between the various components $i$ in the sum.
418: %Since the
419: %distribution function is normalized, the sum of all fractions is
420: %equal to 1:
421: %\begin{equation}
422: %\label{ro}\sum_{i=1}^{n}\frac{c_{i}}{c(0)}=1
423: %\end{equation}
424: %Taking a distribution function into account,
425: The decay curve now has the following mathematical form:
426: \begin{equation}
427: \label{multi}f(t)=\sum_{i=1}^{n}\Delta\Gamma_{tot}\rho(\Gamma_{tot,i})\Gamma_{rad,i}\exp(-\Gamma_{tot,i}t)
428: \end{equation}
429: It is important to note that in Eq. \ref{multi} every component in
430: the sum is correctly normalized since every component is
431: multiplied with its radiative decay rate $\Gamma_{rad,i}$.
432:
433: \subsection{Continuous distribution of decay rates}
434: For infinitesimal values of $\Delta\Gamma_{tot}$, Eq. \ref{multi}
435: can be written as an integral:
436: \begin{equation}
437: \label{multiint}f(t)=\int_{0}^{\infty}\Gamma_{rad}(\Gamma_{tot})\rho(\Gamma_{tot})\exp(-\Gamma_{tot}t)d\Gamma_{tot}
438: \end{equation}
439: In the case of single-exponential decay the distribution function
440: is strongly peaked around a central $\Gamma_{tot}$-value, i.e.,
441: the distribution function is a Dirac delta function. Inserting a
442: Dirac delta function into Eq. \ref{multiint} recovers Eq.
443: \ref{sinexp1}:
444: \begin{eqnarray}
445: f(t)&=&\int_{0}^{\infty}\Gamma_{rad'}\,\delta(\Gamma_{tot}
446: -\Gamma_{tot'})\exp(-\Gamma_{tot}t)d\Gamma_{tot} \nonumber \\
447: \label{delta1}&=&\Gamma_{rad'}\exp(-\Gamma_{tot'}t)
448: \end{eqnarray}
449: This result confirms that the generalization to Eq. \ref{multiint}
450: is correct since it yields the correctly normalized
451: single-exponential functions.
452:
453: In Eq. \ref{multiint} it is tacitly assumed that for every
454: $\Gamma_{tot}$ there is one $\Gamma_{rad}$: the function
455: $\Gamma_{rad}(\Gamma_{tot}$) relates each $\Gamma_{tot}$ to
456: exactly one $\Gamma_{rad}$. In general both $\Gamma_{tot}$ and
457: $\Gamma_{rad}$ vary independently, and Eq. \ref{multiint} is
458: generalized to
459: \begin{eqnarray}
460: &f(t)&\label{multiint3}\\
461: &=&\int_{0}^{\infty}\left[\int_{0}^{\Gamma_{tot}}d\Gamma_{rad}
462: \,\,\rho_{\Gamma_{tot}}(\Gamma_{rad})\,\,\Gamma_{rad}\right]\rho(\Gamma_{tot})\exp(-\Gamma_{tot}t)d\Gamma_{tot}
463: \nonumber
464: \end{eqnarray}
465: where $\rho_{\Gamma_{tot}}(\Gamma_{rad})$ is the normalized
466: distribution of $\Gamma_{rad}$ at constant $\Gamma_{tot}$. For
467: every $\Gamma_{tot}$ the integration is performed over all
468: radiative rates; a distribution of $\Gamma_{rad}$ is taken into
469: account for every $\Gamma_{tot}$. Eq. \ref{multiint3} is the most
470: general expression of a luminescence decay curve and a central
471: result of our paper. From this equation every decay curve with a
472: particular distribution of rates can be recovered. An example
473: described by Eq. \ref{multiint3} is an ensemble of quantum dots in
474: a photonic crystal. In photonic crystals the local density of
475: optical states (LDOS) varies with the location in the crystal and
476: the distribution of dipole orientations of the
477: emitters\cite{1996Sprik}. Therefore, an ensemble of emitters with
478: a certain frequency emit light with a distribution of radiative
479: rates $\Gamma_{rad}$. In addition, when an ensemble of emitters
480: has a distributed $\Gamma_{tot}$ and a single radiative rate
481: $\Gamma_{rad}$, i.e., $\rho_{\Gamma_{tot}}(\Gamma_{rad})$ is a
482: delta-function, then Eq.~\ref{multiint3} reduces to Eq.~
483: \ref{multiint}. Even though the non-radiative rates may still be
484: distributed, Eq.~\ref{multiint} suffices to describe the decay
485: curve since for every $\Gamma_{tot}$ there is only one
486: $\Gamma_{rad}$. Such a situation appears, for example, with
487: powders doped with rare earth ions\cite{2005Vergeer} and with
488: polymer films doped with quantum
489: dots\cite{2002Schlegel,2004Fisher}.
490:
491: Interestingly, an ensemble of emitters with a distribution of
492: rates $\Gamma_{tot}$ is \emph{not} completely characterized by a
493: single value of the quantum efficiency (as opposed to
494: Eq.~\ref{norml} for single-exponential decay). In such an
495: ensemble, the quantum efficiency is distributed, since each
496: $\Gamma_{tot}$ is associated with a distribution of radiative
497: rates $\Gamma_{rad}$. The average quantum efficiency $<QE>$ can be
498: calculated by integrating Eq. \ref{multiint3} for all times:
499: \begin{eqnarray}
500: <QE>&=&\int_{0}^{\infty}f(t)dt\label{multiint5}\\
501: &=&\int_{0}^{\infty}\int_{0}^{\infty}\left[\int_{0}^{\Gamma_{tot}}d\Gamma_{rad}
502: \,\,\rho_{\Gamma_{tot}}(\Gamma_{rad})\,\,\Gamma_{rad}\right]\rho(\Gamma_{tot})\exp(-\Gamma_{tot}t)d\Gamma_{tot}dt
503: \nonumber
504: \end{eqnarray}
505:
506: \begin{figure}
507: \includegraphics[width=0.8\columnwidth]{figure5.eps}
508: \caption{\label{Distrib}Log-normal distribution of $\Gamma$. This
509: distribution was modelled to the data of Fig. \ref{MultiExp}
510: (curve a, quantum dots in photonic crystal) and Fig. \ref{SinExp}
511: (curve b, quantum dots in a diluted suspension), with
512: $\Gamma_{mf}$ and $\Delta\Gamma$ as adjustable parameters. For (a)
513: $\Gamma_{mf}$ is 91.7 $\mu s^{-1}$ ($\frac{1}{\Gamma_{mf}}=10.9$
514: $ns$) and the width of the distribution $\Delta\Gamma$ was 0.57
515: $ns^{-1}$ and for (b) $\Gamma_{mf}$ was 25.8 $\mu s^{-1}$
516: ($\frac{1}{\Gamma_{mf}}=38.8$ $ns$) and the width of the
517: distribution $\Delta\Gamma$ was 0.079 $ns^{-1}$.}
518: \end{figure}
519:
520: Most often, detailed information on the relation between
521: $\Gamma_{tot}$ and $\Gamma_{rad}$ is not available. Then,
522: modelling directly with a distribution of decay rates is
523: applied\cite{1985Jamesc,1986James,1990Brochon,2002Lee,2003Wlodarczyk}.
524: This approach has a major advantage over modelling with a
525: stretched-exponential function, where it is complicated to deduce
526: the distribution of decay rates (see below). A function of the
527: following form is used to model the non-single-exponential decay
528: curve:
529: \begin{equation}
530: \label{multi2}
531: f(t)=\int_{0}^{\infty}\sigma(\Gamma_{tot})\exp(-\Gamma_{tot}
532: t)d\Gamma_{tot}
533: \end{equation}
534: In Eq. \ref{multi2} the various components are not separately
535: normalized as in Eq. \ref{multiint3}. Modelling with Eq.
536: \ref{multi2} boils down to using an infinite series of
537: single-exponentials which are expressed with only a few free
538: parameters. The form of the distribution can usually not be
539: predicted and a decision is made on basis of quality-of-fit. While
540: a good fit does not prove that the chosen distribution is unique,
541: it does extract direct physical information from the
542: non-single-exponential decay on an ensemble of emitters and their
543: environment\cite{2006Nikolaev}.
544:
545: It is widely assumed that $\sigma(\Gamma)$ is equal to the
546: distribution of total rates
547: \cite{1986James,1990Brochon,2001Lee,2003Wlodarczyk,2005Berberan}.
548: A comparison with Eq. \ref{multiint3} shows that this is not true
549: and reveals that $\sigma(\Gamma)$ contains information about both
550: the radiative and non-radiative rates:
551: \begin{equation}\label{multiint4}
552: \sigma(\Gamma_{tot})=\rho(\Gamma_{tot})\int_{0}^{\Gamma_{tot}}\rho_{\Gamma_{tot}}(\Gamma_{rad})\,\,\Gamma_{rad}d\Gamma_{rad}
553: \end{equation}
554: Thus $\sigma(\Gamma)$ is the distribution of total decay rates
555: weighted by the radiative rates. This conclusion demonstrates the
556: practical use of Eq. \ref{multiint3}: the equation allows us to
557: completely interpret the distribution of rates found by modelling
558: with Eq. \ref{multi2}. Such a complete interpretation has not been
559: reported before.
560:
561: \subsection{Log-normal distribution of decay rates}
562: Distribution functions that can be used for $\sigma(\Gamma)$ are
563: (sums of) normal, Lorentzian, and log-normal distribution
564: functions. In Fig. \ref{MultiExp}(a) the luminescence decay curve
565: of quantum dots is successfully modelled with Eq. \ref{multi2},
566: with a log-normal distribution of the rate $\Gamma$
567: \begin{equation}\label{lognor}
568: \sigma(\Gamma)= A\exp\left[-(\frac{\ln \Gamma -
569: \ln\Gamma_{mf}}{\gamma})^2\right]
570: \end{equation}
571: where $A$ is the normalization constant, $\Gamma_{mf}$ is the most
572: frequent rate constant (see Fig. \ref{Distrib}). $\gamma$ is
573: related to the width of the distribution:
574: \begin{equation}\label{lognor2}
575: \Delta\Gamma=2\Gamma_{mf}\sinh(\gamma)
576: \end{equation}
577: where $\Delta\Gamma$ is equal to the width of the distribution at
578: $\frac{1}{e}$. The most frequent rate constant $\Gamma_{mf}$ and
579: $\gamma$ are adjustable parameters, only one extra adjustable
580: parameter compared to a single-exponential model. Clearly, this
581: model (Eq. \ref{multi2} and \ref{lognor}) describes our
582: non-single-exponential experimental data extremely well. The
583: $\chi_{r}^{2}$ was 1.17, $\Gamma_{mf}$ was 91.7 $\mu s^{-1}$
584: ($\frac{1}{\Gamma_{mf}}=10.9$ $ns$) and the width of the
585: distribution $\Delta\Gamma$ was 0.57 $ns^{-1}$. In addition to
586: $\Gamma_{mf}$ and $\Delta\Gamma$, an average decay rate can be
587: deduced from the log-normal distribution in Fig. \ref{Distrib} .
588: However, this average is biased since the various components are
589: weighted with their quantum efficiency, as shown in Eq.
590: \ref{multiint4}.
591:
592: Modelling with a log-normal distribution of decay rates yields
593: direct and clear physical parameters, for instance the shape and
594: width of the decay rate distribution. The log-normal function is
595: plotted in Fig. \ref{Distrib} (curve a). The broad distribution of
596: rates demonstrates the strongly non-single-exponential character
597: of the decay curve. In Ref. \onlinecite{2006Nikolaev} we were able
598: to relate the width of this broad distribution to the spatial and
599: orientational variations of the LDOS in inverse-opal photonic
600: crystals.
601:
602: The log-normal model was also modelled to the decay curve from
603: quantum dots in suspension (Fig. \ref{SinExp}). The distribution
604: is plotted in Fig. \ref{Distrib} (curve b). $\Gamma_{mf}$ was 25.8
605: $\mu s^{-1}$ ($\frac{1}{\Gamma_{mf}}=38.8$ $ns$), close to the
606: lifetime deduced from the single-exponential modelling of 39.0
607: $\pm2.8ns$. The narrow width of the distribution $\Delta\Gamma$ of
608: 0.079 $ns^{-1}$ is in agreement with the single-exponential
609: character of the decay curve.
610:
611: \subsection{Stretched-exponential decay}
612: Besides the multi-exponential models discussed in Sections C-E,
613: the Kohlrausch stretched-exponential decay
614: model\cite{1854Kohlrauscha,1980Lindsey} is widely applied to model
615: non-single-exponential decay curves. The fraction of excited
616: emitters, i.e., the reliability function, of the Kohlrausch
617: stretched-exponential model is equal to:
618: \begin{equation}
619: \label{str1}\frac{c(t^{'})}{c(0)}=\exp(-(\Gamma_{str}
620: t^{'})^{\beta})
621: \end{equation}
622: where $\beta$ is the stretch parameter, which varies between 0 and
623: 1, and $\Gamma_{str}$ the total decay rate in case of
624: stretched-exponential decay. The stretch parameter $\beta$
625: qualitatively expresses the underlying distribution of rates: a
626: small $\beta$ means that the distribution of rates is broad and
627: $\beta$ close to 1 implies a narrow distribution. The recovery of
628: the distribution of rates in case of stretched-exponential decay
629: is mathematically complicated\footnote{In case of the
630: stretched-exponential model the distribution of the rates is
631: unknown and is generally deduced by solving the following equation
632: \cite{1985Jamesc,1986James,2001Lee,2002Schlegel,2005Berberan}:
633: $\frac{\beta}{t}(\Gamma_{str} t)^{\beta}\exp(-(\Gamma_{str}
634: t)^{\beta})=\int_{0}^{\infty}\sigma(\Gamma)\exp(-\Gamma t)d\Gamma$
635: where $\sigma(\Gamma)$ is the distribution function of total decay
636: rate weighted by $\Gamma_{rad}$. To deduce $\sigma(\Gamma)$ an
637: inverse Laplace transform is applied. For $\beta\neq0.5$ and
638: $\beta\neq1$ there is no analytical solution of this equation and
639: therefore it is difficult to deduce the distribution function.
640: This difficulty can be circumvented by modelling directly with a
641: known distribution function, as is shown in this paper.} and only
642: feasible for specific
643: $\beta$'s\cite{1980Lindsey,1985Huber,1990Siemiarczuk,1991Alvarez}.
644:
645: The decay curve corresponding to a Kohlrausch
646: stretched-exponential decay of the fraction
647: $\frac{c(t^{'})}{c(0)}$ can be deduced using Eq. \ref{str1} and
648: Eq. \ref{master}, and results in:
649: \begin{equation}
650: \label{str2} g(t)=\frac{\beta}{t}(\Gamma_{str}
651: t)^{\beta}\exp(-(\Gamma_{str} t)^{\beta})
652: \end{equation}
653: The normalization of Eq. \ref{str2} can, in analogy with Eq.
654: \ref{norml}, be deduced by integration for all times between $t=0$
655: and $t\rightarrow\infty$, which yields 1. Therefore, an important
656: consequence is that Eq. \ref{str2} is correctly normalized
657: \emph{only} for emitters with a quantum yield of 1
658: ($\Gamma_{rad}=\Gamma_{tot}$ and $f(t)=g(t)$). It is not clear how
659: normalization should be done in realistic cases with quantum yield
660: $<100\%$. To the best of our knowledge, this problem has been
661: overlooked in the literature.
662:
663: The main advantage of modelling with a Kohlrausch
664: stretched-exponential function is that the average decay time
665: $<t>$ can readily be calculated. The average decay time is equal
666: to\cite{1980Lindsey}:
667: \begin{equation}
668: \label{str3} <t>=\tau_{av}=\frac{\int_{0}^\infty
669: g(t)tdt}{\int_{0}^\infty
670: g(t)dt}=\frac{1}{\Gamma_{str}\beta}\overline{\Gamma}[\frac{1}{\beta}]
671: \end{equation}
672: where $\overline{\Gamma}$ is the mathematical Gamma-function. For
673: the single-exponential limit of $\beta\rightarrow1$ Eq.~\ref{str3}
674: reduces to Eq.~\ref{sinexp2}. Note again that in this average the
675: various contributions are weighted with their quantum efficiency
676: and that the average decay time $<t>$ differs from
677: $\frac{1}{\Gamma{str}}$ (see Section E). Indeed, for the data in
678: Fig. \ref{MultiExp}(b) Eq.~\ref{str3} yielded an average decay
679: time of 31.1 $ns$ , strongly different from the
680: $\frac{1}{\Gamma_{str}}$-value of 10.4 $ns$, in contrast to the
681: result (Eq. \ref{sinexp1}) for single-exponential decay.
682:
683: It is important to note that Eq.~\ref{str3} is the average decay
684: time $<t>$ corresponding to the decay curve given by
685: Eq.~\ref{str2}. In, for instance,
686: Refs.~\onlinecite{2002Schlegel,2006Kalkman} the average time given
687: by Eq.~\ref{str3} has erroneously been associated to fluorescence
688: decay described by Eq.~\ref{str1}.
689:
690: In contrast to the single-exponential model, the reliability
691: function and the probability density function of a
692: stretched-exponential do \emph{not} have the same form (see Fig.
693: \ref{StrExpFg}); the probability density function contains a
694: time-dependent pre-factor. Therefore, the relation between the
695: reliability function and the probability density function (Eq.
696: \ref{master}) has important consequences. For a $\beta$-value of
697: 0.5 the average decay time of the reliability function (Eq.
698: \ref{str1}) and of the probability density function (Eq.
699: \ref{str2}) differ by more than a factor of ten. Thus it is
700: important to take into consideration whether Eq. \ref{str1} or Eq.
701: \ref{str2} is used to describe the experimental photoluminescence
702: decay curve. This is important since in many reports
703: \cite{2001Lee,2002Lee,2002Schlegel,2003Chen,2004Fisher,2005Berberan},
704: the luminescence decay curve in modelled with the time dependence
705: of Eq. \ref{str1}. We remark that while Eq. \ref{str1} can be used
706: to account for the deviation from single-exponential decay, it
707: does not represent the true Kohlrausch function, but is simply an
708: alternative model. We argue that using the Kohlrausch
709: stretched-exponential as a reliability function to model the
710: fraction $\frac{c(t^{'})}{c(0)}$\footnote{We remark that in Refs.
711: \onlinecite{1854Kohlrauscha,1980Lindsey} capacitor discharge and
712: dielectric relaxation are studied, which are, in contrast to a
713: fluorescence decay curve, indeed described by a reliability
714: function.} implies that the proper probability density function,
715: i.e., Eq. \ref{str2}, must be used to model a luminescence decay
716: curve. Fig. \ref{MultiExp}(b) shows the modelling of experimental
717: data with Eq. \ref{str2}, with $\Gamma_{str}$ and $\beta$ as
718: adjustable parameters. The $\beta$-value was 0.42 and
719: $\Gamma_{str}$ was 96.2 $\mu s^{-1}$
720: ($\frac{1}{\Gamma_{str}}=10.4$ $ns$). Modelling with
721: stretched-exponential is obviously more satisfactory than
722: single-exponential, but here fails at long times, reflected by the
723: high $\chi_{r}^2$-value of 60.7.
724:
725: %In case of the stretched exponential model the distribution of the
726: %rates is unknown and is generally deduced by solving the following
727: %equation
728: %\cite{1985Jamesc,1986James,2001Lee,2002Schlegel,2005Berberan}:
729: %\begin{equation}
730: %\label{str4}\frac{\beta}{t}(\Gamma_{str}
731: %t)^{\beta}\exp(-(\Gamma_{str}
732: %t)^{\beta})=\int_{0}^{\infty}\sigma(\Gamma)\exp(-\Gamma t)d\Gamma
733: %\end{equation}
734: %where $\sigma(\Gamma)$ is the distribution function. To deduce
735: %$\sigma(\Gamma)$ an inverse Laplace transform is applied. For
736: %$\beta\neq0.5$ and $\beta\neq1$ there is no analytical solution of
737: %Eq. \ref{str4} and for this reason it is difficult to deduce the
738: %distribution function
739: %\cite{1980Lindsey,1990Siemiarczuk,1991Alvarez}. This difficulty
740: %can be circumvented by modelling directly with a known
741: %distribution function, as is shown below. It should be noted that
742: %$\sigma(\Gamma)$ does not represent the distribution of total
743: %decay rates, but the distribution of total decay rates weighted by
744: %the radiative rate (see Eq. \ref{multiint4}).
745:
746: \section{Conclusions}
747: We have presented a statistical analysis of time-resolved
748: spontaneous emission decay curves from ensembles of emitters, in
749: particular colloidal quantum dots, with the aim to interpret
750: ubiquitous non-single-exponential decay. Contrary to what is
751: widely assumed, the density of excited emitters $c(t)$ and the
752: intensity in an emission decay curve ($f(t)$ or $g(t)$) are not
753: proportional, but the density is a time-integral of the intensity.
754: The integral relation is crucial to correctly interpret
755: non-single-exponential decay. We have derived the proper
756: normalization for both a discrete, and a continuous distribution
757: of rates, where every decay component is multiplied with its
758: radiative decay rate. A central result of our paper is the
759: derivation of the emission decay curve $f(t)$ in case that both
760: radiative and non-radiative decays are independently distributed
761: (Eq. \ref{multiint3}). In this case, the well-known emission
762: quantum efficiency can not be expressed by a single number
763: anymore, but it is also distributed. We derive a practical
764: description of non-single-exponential emission decay curves in
765: terms of a distribution of total decay rates weighted with the
766: radiative rates. Analyzing decay curves in terms of decay rate
767: distributions opposes to the usual and widely reported analysis in
768: terms of distributed lifetimes. We apply our analysis to recent
769: examples of colloidal quantum dot emission in suspensions and in
770: photonic crystals, and we find that this important class of
771: emitters is well described by a log-normal distribution of decay
772: rates with a narrow and a broad distribution, respectively.
773: Finally, we briefly discuss the Kohlrausch stretched-exponential
774: model; we deduce the average decay time and we find that its
775: normalization is ill-defined for emitters with a realistic quantum
776: efficiency of less than 100 $\%$.
777:
778: \section{Acknowledgments}
779: This work is part of the research program of both the "Stichting
780: voor Fundamenteel Onderzoek der Materie (FOM)", and "Chemische
781: Wetenschappen", which are financially supported by the
782: "Nederlandse Organisatie voor Wetenschappelijk Onderzoek (NWO)".
783:
784:
785: \begin{thebibliography}{44}
786: \expandafter\ifx\csname
787: natexlab\endcsname\relax\def\natexlab#1{#1}\fi
788: \expandafter\ifx\csname bibnamefont\endcsname\relax
789: \def\bibnamefont#1{#1}\fi
790: \expandafter\ifx\csname bibfnamefont\endcsname\relax
791: \def\bibfnamefont#1{#1}\fi
792: \expandafter\ifx\csname citenamefont\endcsname\relax
793: \def\citenamefont#1{#1}\fi
794: \expandafter\ifx\csname url\endcsname\relax
795: \def\url#1{\texttt{#1}}\fi
796: \expandafter\ifx\csname
797: urlprefix\endcsname\relax\def\urlprefix{URL }\fi
798: \providecommand{\bibinfo}[2]{#2}
799: \providecommand{\eprint}[2][]{\url{#2}}
800:
801: \bibitem[{\citenamefont{Lakowicz}(1999)}]{Lakowicz}
802: \bibinfo{author}{\bibfnamefont{J.~R.} \bibnamefont{Lakowicz}},
803: \emph{\bibinfo{title}{Principles of Fluorescence Spectroscopy}}
804: (\bibinfo{publisher}{Kluwer Academic/Plenum Publishers},
805: \bibinfo{address}{New York, Boston, Dordrecht, London, Moscow},
806: \bibinfo{year}{1999}), \bibinfo{edition}{2nd} ed.
807:
808: \bibitem[{\citenamefont{Medintz et~al.}(2005)\citenamefont{Medintz, Uyeda,
809: Goldman, and Mattoussi}}]{2005Medintz}
810: \bibinfo{author}{\bibfnamefont{I.~L.} \bibnamefont{Medintz}},
811: \bibinfo{author}{\bibfnamefont{H.~T.} \bibnamefont{Uyeda}},
812: \bibinfo{author}{\bibfnamefont{E.~R.} \bibnamefont{Goldman}},
813: \bibnamefont{and}
814: \bibinfo{author}{\bibfnamefont{H.}~\bibnamefont{Mattoussi}},
815: \bibinfo{journal}{Nature Materials} \textbf{\bibinfo{volume}{4}},
816: \bibinfo{pages}{435} (\bibinfo{year}{2005}).
817:
818: \bibitem[{\citenamefont{Loudon}(2001)}]{Loudon}
819: \bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Loudon}},
820: \emph{\bibinfo{title}{The quantum theory of light}}, Oxford science
821: publications (\bibinfo{publisher}{Oxford University Press},
822: \bibinfo{address}{Oxford,}, \bibinfo{year}{2001}), \bibinfo{edition}{3rd} ed.
823:
824: \bibitem[{\citenamefont{Crooker et~al.}(2002)\citenamefont{Crooker,
825: Hollingsworth, Tretiak, and Klimov}}]{2002Crooker}
826: \bibinfo{author}{\bibfnamefont{S.~A.} \bibnamefont{Crooker}},
827: \bibinfo{author}{\bibfnamefont{J.~A.} \bibnamefont{Hollingsworth}},
828: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Tretiak}}, \bibnamefont{and}
829: \bibinfo{author}{\bibfnamefont{V.~I.} \bibnamefont{Klimov}},
830: \bibinfo{journal}{Physical Review Letters} \textbf{\bibinfo{volume}{89}},
831: \bibinfo{pages}{186802} (\bibinfo{year}{2002}).
832:
833: \bibitem[{\citenamefont{Zhang et~al.}(2002)\citenamefont{Zhang, Wang, and
834: Xiao}}]{2002Zhang}
835: \bibinfo{author}{\bibfnamefont{J.~Y.} \bibnamefont{Zhang}},
836: \bibinfo{author}{\bibfnamefont{X.~Y.} \bibnamefont{Wang}}, \bibnamefont{and}
837: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Xiao}},
838: \bibinfo{journal}{Optics Letters} \textbf{\bibinfo{volume}{27}},
839: \bibinfo{pages}{1253} (\bibinfo{year}{2002}).
840:
841: \bibitem[{\citenamefont{Drexhage}(1970)}]{1970Drexhage}
842: \bibinfo{author}{\bibfnamefont{K.~H.} \bibnamefont{Drexhage}},
843: \bibinfo{journal}{Journal of Luminescence} \textbf{\bibinfo{volume}{1,2}},
844: \bibinfo{pages}{693} (\bibinfo{year}{1970}).
845:
846: \bibitem[{\citenamefont{Amos and Barnes}(1997)}]{1997Amos}
847: \bibinfo{author}{\bibfnamefont{R.~M.} \bibnamefont{Amos}} \bibnamefont{and}
848: \bibinfo{author}{\bibfnamefont{W.~L.} \bibnamefont{Barnes}},
849: \bibinfo{journal}{Physical Review B} \textbf{\bibinfo{volume}{55}},
850: \bibinfo{pages}{7249} (\bibinfo{year}{1997}).
851:
852: \bibitem[{\citenamefont{Danz et~al.}(2002)\citenamefont{Danz, Heber, and
853: Brauer}}]{2002Danz}
854: \bibinfo{author}{\bibfnamefont{N.}~\bibnamefont{Danz}},
855: \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Heber}}, \bibnamefont{and}
856: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Brauer}},
857: \bibinfo{journal}{Physical Review A} \textbf{\bibinfo{volume}{66}},
858: \bibinfo{pages}{063809} (\bibinfo{year}{2002}).
859:
860: \bibitem[{\citenamefont{Astilean and Barnes}(2002)}]{2002Astilean}
861: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Astilean}} \bibnamefont{and}
862: \bibinfo{author}{\bibfnamefont{W.~L.} \bibnamefont{Barnes}},
863: \bibinfo{journal}{Applied Physics B-Lasers and Optics}
864: \textbf{\bibinfo{volume}{75}}, \bibinfo{pages}{591} (\bibinfo{year}{2002}).
865:
866: \bibitem[{\citenamefont{Bayer et~al.}(2001)\citenamefont{Bayer, Reinecke,
867: Weidner, Larionov, McDonald, and Forchel}}]{2001Bayer}
868: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Bayer}},
869: \bibinfo{author}{\bibfnamefont{T.~L.} \bibnamefont{Reinecke}},
870: \bibinfo{author}{\bibfnamefont{F.}~\bibnamefont{Weidner}},
871: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Larionov}},
872: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{McDonald}}, \bibnamefont{and}
873: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Forchel}},
874: \bibinfo{journal}{Physical Review Letters} \textbf{\bibinfo{volume}{86}},
875: \bibinfo{pages}{3168} (\bibinfo{year}{2001}).
876:
877: \bibitem[{\citenamefont{Biteen et~al.}(2005)\citenamefont{Biteen, Pacifici,
878: Lewis, and Atwater}}]{2005Biteen}
879: \bibinfo{author}{\bibfnamefont{J.~S.} \bibnamefont{Biteen}},
880: \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{Pacifici}},
881: \bibinfo{author}{\bibfnamefont{N.~S.} \bibnamefont{Lewis}}, \bibnamefont{and}
882: \bibinfo{author}{\bibfnamefont{H.~A.} \bibnamefont{Atwater}},
883: \bibinfo{journal}{Nano Letters} \textbf{\bibinfo{volume}{5}},
884: \bibinfo{pages}{1768} (\bibinfo{year}{2005}).
885:
886: \bibitem[{\citenamefont{Song et~al.}(2005)\citenamefont{Song, Atay, Shi, Urabe,
887: and Nurmikko}}]{2005Song}
888: \bibinfo{author}{\bibfnamefont{J.~H.} \bibnamefont{Song}},
889: \bibinfo{author}{\bibfnamefont{T.}~\bibnamefont{Atay}},
890: \bibinfo{author}{\bibfnamefont{S.~F.} \bibnamefont{Shi}},
891: \bibinfo{author}{\bibfnamefont{H.}~\bibnamefont{Urabe}}, \bibnamefont{and}
892: \bibinfo{author}{\bibfnamefont{A.~V.} \bibnamefont{Nurmikko}},
893: \bibinfo{journal}{Nano Letters} \textbf{\bibinfo{volume}{5}},
894: \bibinfo{pages}{1557} (\bibinfo{year}{2005}).
895:
896: \bibitem[{\citenamefont{Fujita et~al.}(2005)\citenamefont{Fujita, Takahashi,
897: Tanaka, Asano, and Noda}}]{2005Fujita}
898: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Fujita}},
899: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Takahashi}},
900: \bibinfo{author}{\bibfnamefont{Y.}~\bibnamefont{Tanaka}},
901: \bibinfo{author}{\bibfnamefont{T.}~\bibnamefont{Asano}}, \bibnamefont{and}
902: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Noda}},
903: \bibinfo{journal}{Science} \textbf{\bibinfo{volume}{308}},
904: \bibinfo{pages}{1296} (\bibinfo{year}{2005}).
905:
906: \bibitem[{\citenamefont{Kress et~al.}(2005)\citenamefont{Kress, Hofbauer,
907: Reinelt, Kaniber, Krenner, Meyer, Bohm, and Finley}}]{2005Kress}
908: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Kress}},
909: \bibinfo{author}{\bibfnamefont{F.}~\bibnamefont{Hofbauer}},
910: \bibinfo{author}{\bibfnamefont{N.}~\bibnamefont{Reinelt}},
911: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Kaniber}},
912: \bibinfo{author}{\bibfnamefont{H.~J.} \bibnamefont{Krenner}},
913: \bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Meyer}},
914: \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Bohm}}, \bibnamefont{and}
915: \bibinfo{author}{\bibfnamefont{J.~J.} \bibnamefont{Finley}},
916: \bibinfo{journal}{Physical Review B} \textbf{\bibinfo{volume}{71}},
917: \bibinfo{pages}{241304} (\bibinfo{year}{2005}).
918:
919: \bibitem[{\citenamefont{Lodahl et~al.}(2004)\citenamefont{Lodahl, van Driel,
920: Nikolaev, Irman, Overgaag, Vanmaekelbergh, and Vos}}]{2004Lodahl}
921: \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Lodahl}},
922: \bibinfo{author}{\bibfnamefont{A.~F.} \bibnamefont{van Driel}},
923: \bibinfo{author}{\bibfnamefont{I.~S.} \bibnamefont{Nikolaev}},
924: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Irman}},
925: \bibinfo{author}{\bibfnamefont{K.}~\bibnamefont{Overgaag}},
926: \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{Vanmaekelbergh}},
927: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{W.~L.} \bibnamefont{Vos}},
928: \bibinfo{journal}{Nature} \textbf{\bibinfo{volume}{430}},
929: \bibinfo{pages}{654} (\bibinfo{year}{2004}).
930:
931: \bibitem[{\citenamefont{Foggi et~al.}(1995)\citenamefont{Foggi, Pettini, Santa,
932: Righini, and Califano}}]{1995Foggi}
933: \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Foggi}},
934: \bibinfo{author}{\bibfnamefont{L.}~\bibnamefont{Pettini}},
935: \bibinfo{author}{\bibfnamefont{I.}~\bibnamefont{Santa}},
936: \bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Righini}}, \bibnamefont{and}
937: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Califano}},
938: \bibinfo{journal}{Journal of Physical Chemistry}
939: \textbf{\bibinfo{volume}{99}}, \bibinfo{pages}{7439} (\bibinfo{year}{1995}).
940:
941: \bibitem[{\citenamefont{Klimov and McBranch}(1998)}]{1998Klimov}
942: \bibinfo{author}{\bibfnamefont{V.~I.} \bibnamefont{Klimov}} \bibnamefont{and}
943: \bibinfo{author}{\bibfnamefont{D.~W.} \bibnamefont{McBranch}},
944: \bibinfo{journal}{Physical Review Letters} \textbf{\bibinfo{volume}{80}},
945: \bibinfo{pages}{4028} (\bibinfo{year}{1998}).
946:
947: \bibitem[{\citenamefont{Neuwahl et~al.}(2000)\citenamefont{Neuwahl, Foggi, and
948: Brown}}]{2000Neuwahl}
949: \bibinfo{author}{\bibfnamefont{F.~V.~R.} \bibnamefont{Neuwahl}},
950: \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Foggi}}, \bibnamefont{and}
951: \bibinfo{author}{\bibfnamefont{R.~G.} \bibnamefont{Brown}},
952: \bibinfo{journal}{Chemical Physics Letters} \textbf{\bibinfo{volume}{319}},
953: \bibinfo{pages}{157} (\bibinfo{year}{2000}).
954:
955: \bibitem[{\citenamefont{Rosencwaig}(1980)}]{Rosencwaig}
956: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Rosencwaig}},
957: \emph{\bibinfo{title}{Photoacoustics and Photoacoustic Spectroscopy}}
958: (\bibinfo{publisher}{John Wiley \& Sons}, \bibinfo{address}{New York},
959: \bibinfo{year}{1980}).
960:
961: \bibitem[{\citenamefont{Grinberg et~al.}(2003)\citenamefont{Grinberg, Sikorska,
962: and Sliwinski}}]{2003Grinberg}
963: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Grinberg}},
964: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Sikorska}}, \bibnamefont{and}
965: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Sliwinski}},
966: \bibinfo{journal}{Physical Review B} \textbf{\bibinfo{volume}{67}},
967: \bibinfo{pages}{045114} (\bibinfo{year}{2003}).
968:
969: \bibitem[{\citenamefont{James and Ware}(1986)}]{1986James}
970: \bibinfo{author}{\bibfnamefont{D.~R.} \bibnamefont{James}} \bibnamefont{and}
971: \bibinfo{author}{\bibfnamefont{W.~R.} \bibnamefont{Ware}},
972: \bibinfo{journal}{Chemical Physics Letters} \textbf{\bibinfo{volume}{126}},
973: \bibinfo{pages}{7} (\bibinfo{year}{1986}).
974:
975: \bibitem[{\citenamefont{Siemiarczuk et~al.}(1990)\citenamefont{Siemiarczuk,
976: Wagner, and Ware}}]{1990Siemiarczuk}
977: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Siemiarczuk}},
978: \bibinfo{author}{\bibfnamefont{B.~D.} \bibnamefont{Wagner}},
979: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{W.~R.} \bibnamefont{Ware}},
980: \bibinfo{journal}{Journal of Physical Chemistry}
981: \textbf{\bibinfo{volume}{94}}, \bibinfo{pages}{1661} (\bibinfo{year}{1990}).
982:
983: \bibitem[{\citenamefont{Brochon et~al.}(1990)\citenamefont{Brochon, Livesey,
984: Pouget, and Valeur}}]{1990Brochon}
985: \bibinfo{author}{\bibfnamefont{J.~C.} \bibnamefont{Brochon}},
986: \bibinfo{author}{\bibfnamefont{A.~K.} \bibnamefont{Livesey}},
987: \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Pouget}}, \bibnamefont{and}
988: \bibinfo{author}{\bibfnamefont{B.}~\bibnamefont{Valeur}},
989: \bibinfo{journal}{Chemical Physics Letters} \textbf{\bibinfo{volume}{174}},
990: \bibinfo{pages}{517} (\bibinfo{year}{1990}).
991:
992: \bibitem[{\citenamefont{Wlodarczyk and Kierdaszuk}(2003)}]{2003Wlodarczyk}
993: \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Wlodarczyk}} \bibnamefont{and}
994: \bibinfo{author}{\bibfnamefont{B.}~\bibnamefont{Kierdaszuk}},
995: \bibinfo{journal}{Biophysical Journal} \textbf{\bibinfo{volume}{85}},
996: \bibinfo{pages}{589} (\bibinfo{year}{2003}).
997:
998: \bibitem[{\citenamefont{Wuister et~al.}(2004)\citenamefont{Wuister, van
999: Houselt, Donega, Vanmaekelbergh, and Meijerink}}]{2004Wuister}
1000: \bibinfo{author}{\bibfnamefont{S.~F.} \bibnamefont{Wuister}},
1001: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{van Houselt}},
1002: \bibinfo{author}{\bibfnamefont{C.~D.~M.} \bibnamefont{Donega}},
1003: \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{Vanmaekelbergh}},
1004: \bibnamefont{and}
1005: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Meijerink}},
1006: \bibinfo{journal}{Angewandte Chemie-International Edition}
1007: \textbf{\bibinfo{volume}{43}}, \bibinfo{pages}{3029} (\bibinfo{year}{2004}).
1008:
1009: \bibitem[{\citenamefont{Nikolaev et~al.}(2005)\citenamefont{Nikolaev, Lodahl,
1010: van Driel, and Vos}}]{2006Nikolaev}
1011: \bibinfo{author}{\bibfnamefont{I.~S.} \bibnamefont{Nikolaev}},
1012: \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Lodahl}},
1013: \bibinfo{author}{\bibfnamefont{A.~F.} \bibnamefont{van Driel}},
1014: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{W.~L.} \bibnamefont{Vos}},
1015: \bibinfo{journal}{http://arxiv.org/abs/physics/0511133}
1016: (\bibinfo{year}{2005}).
1017:
1018: \bibitem[{\citenamefont{Kohlrausch}(1854{\natexlab{a}})}]{1854Kohlrauscha}
1019: \bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Kohlrausch}},
1020: \bibinfo{journal}{Annalen der Physik} \textbf{\bibinfo{volume}{91}},
1021: \bibinfo{pages}{179} (\bibinfo{year}{1854}{\natexlab{a}}).
1022:
1023: \bibitem[{\citenamefont{Deschenes and Vanden~Bout}(2001)}]{2001Deschenes}
1024: \bibinfo{author}{\bibfnamefont{L.~A.} \bibnamefont{Deschenes}}
1025: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{D.~A.}
1026: \bibnamefont{Vanden~Bout}}, \bibinfo{journal}{Science}
1027: \textbf{\bibinfo{volume}{292}}, \bibinfo{pages}{255} (\bibinfo{year}{2001}).
1028:
1029: \bibitem[{\citenamefont{Lindsey and Patterson}(1980)}]{1980Lindsey}
1030: \bibinfo{author}{\bibfnamefont{C.~P.} \bibnamefont{Lindsey}} \bibnamefont{and}
1031: \bibinfo{author}{\bibfnamefont{G.~D.} \bibnamefont{Patterson}},
1032: \bibinfo{journal}{Journal of Chemical Physics} \textbf{\bibinfo{volume}{73}},
1033: \bibinfo{pages}{3348} (\bibinfo{year}{1980}).
1034:
1035: \bibitem[{\citenamefont{Kohlrausch}(1854{\natexlab{b}})}]{1854Kohlrauschb}
1036: \bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Kohlrausch}},
1037: \bibinfo{journal}{Annalen der Physik} \textbf{\bibinfo{volume}{91}},
1038: \bibinfo{pages}{56} (\bibinfo{year}{1854}{\natexlab{b}}).
1039:
1040: \bibitem[{\citenamefont{Torre et~al.}(2004)\citenamefont{Torre, Bartolini, and
1041: Righini}}]{2004Torre}
1042: \bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Torre}},
1043: \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Bartolini}},
1044: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Righini}},
1045: \bibinfo{journal}{Nature} \textbf{\bibinfo{volume}{428}},
1046: \bibinfo{pages}{296} (\bibinfo{year}{2004}).
1047:
1048: \bibitem[{\citenamefont{Schlegel et~al.}(2002)\citenamefont{Schlegel,
1049: Bohnenberger, Potapova, and Mews}}]{2002Schlegel}
1050: \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Schlegel}},
1051: \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Bohnenberger}},
1052: \bibinfo{author}{\bibfnamefont{I.}~\bibnamefont{Potapova}}, \bibnamefont{and}
1053: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Mews}},
1054: \bibinfo{journal}{Physical Review Letters} \textbf{\bibinfo{volume}{88}},
1055: \bibinfo{pages}{137401} (\bibinfo{year}{2002}).
1056:
1057: \bibitem[{\citenamefont{Chen}(2003)}]{2003Chen}
1058: \bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Chen}},
1059: \bibinfo{journal}{Journal of Luminescence} \textbf{\bibinfo{volume}{102}},
1060: \bibinfo{pages}{510} (\bibinfo{year}{2003}).
1061:
1062: \bibitem[{\citenamefont{Fisher et~al.}(2004)\citenamefont{Fisher, Eisler,
1063: Stott, and Bawendi}}]{2004Fisher}
1064: \bibinfo{author}{\bibfnamefont{B.~R.} \bibnamefont{Fisher}},
1065: \bibinfo{author}{\bibfnamefont{H.~J.} \bibnamefont{Eisler}},
1066: \bibinfo{author}{\bibfnamefont{N.~E.} \bibnamefont{Stott}}, \bibnamefont{and}
1067: \bibinfo{author}{\bibfnamefont{M.~G.} \bibnamefont{Bawendi}},
1068: \bibinfo{journal}{Journal of Physical Chemistry B}
1069: \textbf{\bibinfo{volume}{108}}, \bibinfo{pages}{143} (\bibinfo{year}{2004}).
1070:
1071: \bibitem[{\citenamefont{Huber}(1985)}]{1985Huber}
1072: \bibinfo{author}{\bibfnamefont{D.~L.} \bibnamefont{Huber}},
1073: \bibinfo{journal}{Physical Review B} \textbf{\bibinfo{volume}{31}},
1074: \bibinfo{pages}{6070} (\bibinfo{year}{1985}).
1075:
1076: \bibitem[{\citenamefont{Alvarez et~al.}(1991)\citenamefont{Alvarez, Alegria,
1077: and Colmenero}}]{1991Alvarez}
1078: \bibinfo{author}{\bibfnamefont{F.}~\bibnamefont{Alvarez}},
1079: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Alegria}}, \bibnamefont{and}
1080: \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Colmenero}},
1081: \bibinfo{journal}{Physical Review B} \textbf{\bibinfo{volume}{44}},
1082: \bibinfo{pages}{7306} (\bibinfo{year}{1991}).
1083:
1084: \bibitem[{\citenamefont{Lee et~al.}(2002)\citenamefont{Lee, Kim, Tang, and
1085: Hochstrasser}}]{2002Lee}
1086: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Lee}},
1087: \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Kim}},
1088: \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Tang}}, \bibnamefont{and}
1089: \bibinfo{author}{\bibfnamefont{R.~M.} \bibnamefont{Hochstrasser}},
1090: \bibinfo{journal}{Chemical Physics Letters} \textbf{\bibinfo{volume}{359}},
1091: \bibinfo{pages}{412} (\bibinfo{year}{2002}).
1092:
1093: \bibitem[{\citenamefont{Vergeer et~al.}(2005)\citenamefont{Vergeer, Vlugt, Kox,
1094: den Hertog, van~der Eerden, and Meijerink}}]{2005Vergeer}
1095: \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Vergeer}},
1096: \bibinfo{author}{\bibfnamefont{T.~J.~H.} \bibnamefont{Vlugt}},
1097: \bibinfo{author}{\bibfnamefont{M.~H.~F.} \bibnamefont{Kox}},
1098: \bibinfo{author}{\bibfnamefont{M.~I.} \bibnamefont{den Hertog}},
1099: \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{van~der Eerden}},
1100: \bibnamefont{and}
1101: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Meijerink}},
1102: \bibinfo{journal}{Physical Review B} \textbf{\bibinfo{volume}{71}},
1103: \bibinfo{pages}{014119} (\bibinfo{year}{2005}).
1104:
1105: \bibitem[{\citenamefont{Dougherty}(1990)}]{Dougherty}
1106: \bibinfo{author}{\bibfnamefont{E.~R.} \bibnamefont{Dougherty}},
1107: \emph{\bibinfo{title}{Probability and Statistics for the engineering,
1108: computing and physical sciences}} (\bibinfo{publisher}{Prentice-Hall
1109: International, Inc.}, \bibinfo{address}{Englewood, New Jersey},
1110: \bibinfo{year}{1990}).
1111:
1112: \bibitem[{\citenamefont{van Driel et~al.}(2005)\citenamefont{van Driel, Allan,
1113: Delerue, Lodahl, Vos, and Vanmaekelbergh}}]{2005vanDriel}
1114: \bibinfo{author}{\bibfnamefont{A.~F.} \bibnamefont{van Driel}},
1115: \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Allan}},
1116: \bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Delerue}},
1117: \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Lodahl}},
1118: \bibinfo{author}{\bibfnamefont{W.~L.} \bibnamefont{Vos}}, \bibnamefont{and}
1119: \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{Vanmaekelbergh}},
1120: \bibinfo{journal}{Physical Review Letters} \textbf{\bibinfo{volume}{95}},
1121: \bibinfo{pages}{236804} (\bibinfo{year}{2005}).
1122:
1123: \bibitem[{\citenamefont{Sprik et~al.}(1996)\citenamefont{Sprik, van Tiggelen,
1124: and Lagendijk}}]{1996Sprik}
1125: \bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Sprik}},
1126: \bibinfo{author}{\bibfnamefont{B.~A.} \bibnamefont{van Tiggelen}},
1127: \bibnamefont{and}
1128: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Lagendijk}},
1129: \bibinfo{journal}{Europhysics Letters} \textbf{\bibinfo{volume}{35}},
1130: \bibinfo{pages}{265} (\bibinfo{year}{1996}).
1131:
1132: \bibitem[{\citenamefont{James et~al.}(1985)\citenamefont{James, Liu, De~Mayo,
1133: and Ware}}]{1985Jamesc}
1134: \bibinfo{author}{\bibfnamefont{D.~R.} \bibnamefont{James}},
1135: \bibinfo{author}{\bibfnamefont{Y.-S.} \bibnamefont{Liu}},
1136: \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{De~Mayo}}, \bibnamefont{and}
1137: \bibinfo{author}{\bibfnamefont{W.~R.} \bibnamefont{Ware}},
1138: \bibinfo{journal}{Chemical Physics Letters} \textbf{\bibinfo{volume}{120}},
1139: \bibinfo{pages}{460} (\bibinfo{year}{1985}).
1140:
1141: \bibitem[{\citenamefont{Benny~Lee et~al.}(2001)\citenamefont{Benny~Lee, Siegel,
1142: Webb, Leveque-Fort, Cole, Jones, Dowling, Lever, and French}}]{2001Lee}
1143: \bibinfo{author}{\bibfnamefont{K.~C.} \bibnamefont{Benny~Lee}},
1144: \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Siegel}},
1145: \bibinfo{author}{\bibfnamefont{S.~E.~D.} \bibnamefont{Webb}},
1146: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Leveque-Fort}},
1147: \bibinfo{author}{\bibfnamefont{M.~J.} \bibnamefont{Cole}},
1148: \bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Jones}},
1149: \bibinfo{author}{\bibfnamefont{K.}~\bibnamefont{Dowling}},
1150: \bibinfo{author}{\bibfnamefont{M.~J.} \bibnamefont{Lever}}, \bibnamefont{and}
1151: \bibinfo{author}{\bibfnamefont{P.~M.~W.} \bibnamefont{French}},
1152: \bibinfo{journal}{Biophysical Journal} \textbf{\bibinfo{volume}{81}},
1153: \bibinfo{pages}{1265} (\bibinfo{year}{2001}).
1154:
1155: \bibitem[{\citenamefont{Berberan-Santos
1156: et~al.}(2005)\citenamefont{Berberan-Santos, Bodunov, and
1157: Valeur}}]{2005Berberan}
1158: \bibinfo{author}{\bibfnamefont{M.~N.} \bibnamefont{Berberan-Santos}},
1159: \bibinfo{author}{\bibfnamefont{E.~N.} \bibnamefont{Bodunov}},
1160: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{B.}~\bibnamefont{Valeur}},
1161: \bibinfo{journal}{Chemical Physics} \textbf{\bibinfo{volume}{315}},
1162: \bibinfo{pages}{171} (\bibinfo{year}{2005}).
1163:
1164: \bibitem[{\citenamefont{Kalkman et~al.}(2006)\citenamefont{Kalkman,
1165: Gersen, Kuipers, and Polman}}]{2006Kalkman}
1166: \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Kalkman}},
1167: \bibinfo{author}{\bibfnamefont{H.}~\bibnamefont{Gersen}},
1168: \bibinfo{author}{\bibfnamefont{L.}~\bibnamefont{Kuipers}}, \bibnamefont{and}
1169: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Polman}},
1170: \bibinfo{journal}{Physical Review B} \textbf{\bibinfo{volume}{73}},
1171: \bibinfo{pages}{075317} (\bibinfo{year}{2006}).
1172:
1173: \end{thebibliography}
1174:
1175: \end{document}
1176: