1: \documentclass[twocolumn,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}

2: %\documentclass[preprint,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}

3: 

4: % Some other (several out of many) possibilities

5: %\documentclass[preprint,aps]{revtex4}

6: %\documentclass[preprint,aps,draft]{revtex4}

7: %\documentclass[prb]{revtex4}% Physical Review B

8: 

9: \usepackage{graphicx}% Include figure files

10: \usepackage{dcolumn}% Align table columns on decimal point

11: \usepackage{bm}% bold math

12: 

13: \begin{document}

14: \def\a{\alpha}

15: \def\b{\beta}

16: \def\g{\gamma}

17: \def\G{\Gamma}

18: \def\d{\delta}

19: \def\D{\Delta}

20: \def\e{\epsilon}

21: \def\k{\kappa}

22: \def\l{\lambda}

23: \def\s{\sigma}

24: \def\t{\tau}

25: \def\om{\omega}

26: \def\Om{\Omega}

27: \def\lg{\langle}

28: \def\rg{\rangle}

29: \newcommand{\be} {\begin{equation}}

30: \newcommand{\ee} {\end{equation}}

31: \newcommand{\Be} {\begin{eqnarray}}

32: \newcommand{\Ee} {\end{eqnarray}}

33: 

34: 

35: \preprint{APS/123}

36: \title{Rotational Correlation Functions of Single Molecules}

37: 

38: \author{G. Hinze}

39:  \email{hinze@mail.uni-mainz.de}

40:  \homepage{http://www.uni-mainz.de/FB/Chemie/Basche/}

41: \author{G. Diezemann}

42: \author{Th.{ }Basch\'{e}   }

43: \affiliation{Institut f\"ur  Physikalische Chemie,

44:  Johannes Gutenberg-Universit\"at, D-55099 Mainz}

45: \date{\today}

46: 

47: 

48: \begin{abstract}

49: Single molecule rotational correlation functions are analyzed for

50: several reorientation geometries. Even for the simplest model of

51: isotropic rotational diffusion our findings predict

52: non-exponential correlation functions to be observed by

53: polarization sensitive single molecule fluorescence microscopy.

54: This may have a deep impact on interpreting the results of

55: molecular reorientation measurements in heterogeneous

56: environments.

57: \end{abstract}

58: 

59: 

60: 

61: \pacs{33.15.Vb, 87.64.Ni, 61.43.FS, 67.40.Fd}

62: %7.79.-v     Scanning probe microscopes and components (see also 68.37.-d in surfaces and interfaces)

63: %61.43.Fs    Glasses

64: %67.40.Fd    Dynamics of relaxation phenomena

65: %87.64.Ni    Optical absorption, magnetic circular dichroism, and fluorescence spectroscopy

66: %33.15.Vb    Correlation times in molecular dynamics

67: 

68: 

69: \maketitle

70: 

71: Reorientation of small molecules or segments of macromolecules

72: undergoing conformational changes are elementary processes and

73: often of crucial importance for the properties of materials. In

74: most of the experimental techniques probing rotational dynamics,

75: ensembles of molecules have been monitored

76: \cite{berne1990,boettcher1992,boehmer2001}. Actually, many

77: chemical and biological systems are heterogeneous, a fact that

78: e.g. is reflected in their dynamical properties. In the context of

79: the non-exponential decay of bulk correlations in supercooled

80: liquids this has been termed dynamic heterogeneity. Proteins,

81: polymers, colloids and coarsening systems share this feature of

82: non-exponential relaxation.

83: %

84: In the past, bulk techniques have been developed that allow a

85: characterization of heterogeneous rotational dynamics in complex

86: systems in terms of a distribution of reorientational rates

87: fluctuating in time\cite{richert2002,sillescu2002}. There are,

88: however, hints that even without (bulk) heterogeneity rotational

89: correlations functions (RCFs) show deviations from exponential

90: decay on a microscopic scale\cite{dreizehn1998}. Such a behavior

91: has also been found in numerical calculations on spin glasses

92: \cite{Castillo03,Cugliandolo03}.

93: 

94: In order to get a deeper insight into the nature of the rotational

95: dynamics and its relation to structural properties of

96: heterogeneous materials, it is highly desirable to conduct

97: experiments providing local (as opposed to averaged bulk)

98: information. It has been shown by numerous examples that a single

99: molecule is an exquisitely sensitive probe of dynamical processes

100: in its local environment\cite{basche1997,science1999}. Along these

101: lines, temporal fluctuations of excited state

102: lifetimes\cite{vallee2003}, excitation\cite{ambrose1991} and

103: emission spectra\cite{lu1997} and electronic coupling

104: strengths\cite{lippitz2004} have been studied by single molecule

105: spectroscopy and analyzed in terms of fluctuating environments.

106: Here we focus on the rotational dynamics of single fluorophores

107: investigated by polarization resolved

108: microscopy\cite{ha1996,ha1999,harms1999,vdbout1,vdbout2,vdbout3}.

109: In order to extract quantitative and reliable information from

110: such measurements, a detailed analysis of local RCFs is of utmost

111: importance. It is the purpose of the present letter to provide

112: such an analysis for the simple model of rotational diffusion of

113: single fluorophores.

114: 

115: Fluorophores preferentially absorb photons whose electric field

116: vectors are aligned parallel to the transition dipoles of the

117: molecule \cite{lakowicz1999}. These dipoles have a well defined

118: orientation with respect to the molecular axes. Similarly,

119: emission also occurs with light polarized along a fixed axis. The

120: experimental situation we will analyze in the remainder of this

121: paper is depicted in Fig. 1 together with the definition of the

122: axes system and the angles used.

123: %

124: The orthogonal polarization resolved emission signals $I_p$ and

125: $I_s$ depend on both, absorption and emission efficiency. The

126: contributions of $I_p$ and $I_s$ to the signals collected in

127: fluorescence correlation spectroscopy have been analyzed in

128: detail\cite{pecora75,pecora76}.

129: %

130: 

131: {A customary measure of dipole orientation can be obtained by

132: calculating the reduced linear dichroism $d$ from $I_p$ and $I_s$

133: via }

134: %For this more general case a meaningful measure of dipole

135: %orientation can be obtained by calculating the reduced linear

136: %dichroism $d$ from $I_p$ and $I_s$ via

137: \begin{equation}

138:    d=\frac{I_p-I_s}{I_p+I_s}

139: \end{equation}

140: %

141: {This frequently used

142: quantity\cite{harms1999,vdbout1,vdbout2,vdbout3,hochstrasser2003}

143: fluctuates in the course of time due to molecular rotational

144: dynamics.}

145: %

146: The normalization ensures that $d$ depends solely on the

147: orientation of the emission dipole and not on the absorption

148: efficiency.

149: % eingefügt 28.5.04 nach Report

150: {Hence we represent the transition dipoles not by vectors but by

151: their orientation using angles $\theta, \phi$.

152: %

153: We want to point out that the analogy of $d$ to what is often

154: called polarization\cite{lakowicz1999} could lead to the wrong

155: assumption to use the anisotropy values instead. However, in

156: contrast to fluorescence depolarization methods this would not

157: make any sense here.}

158: %

159: Subsequently the auto-correlation function of $d(t)$ is

160: calculated,

161: \begin{equation}

162:    C_d(t)=\left< d(t+\tau) d(\tau) \right>

163: \end{equation}

164: In the present letter we will show that $C_d(t)$ is a non-trivial

165: correlation function which nevertheless gives valuable information

166: about the time scale and the geometric aspects of molecular

167: reorientations.

168: % eingefügt 27.5.2004 nach report

169: {It should be noted that in principle the fraction $I_p/I_s$ leads

170: to the azimuth $\phi$, however, an extremely well S/N ratio would

171: be needed for reliable values.}

172: %

173: \begin{figure}

174: % Requires \usepackage{graphicx}

175: \includegraphics[width=7.5cm,height=3cm]{fig1}  %[width=4cm,height=4cm]

176: \caption{Schematic experimental setup: The polarization of

177:   light emitted by a single fluorophore depends on the orientation of the transition

178:   dipoles with respect to the detection system. The two

179:   orthogonally polarized components of the signal are separated by a polarizing

180:   cube beam splitter.

181:   }

182:   \label{111}

183: \end{figure}

184: 

185: Well known bulk techniques allowing to monitor molecular

186: rotational dynamics are provided by dynamic light scattering

187: \cite{berne1990}, dielectric spectroscopy \cite{boettcher1992} or

188: nuclear magnetic resonance (NMR) \cite{boehmer2001}. Due to the

189: different ranks $l$ of the respective interaction tensors distinct

190: RCFs

191: %

192: \be\label{Cl.t} C_l(t)=\lg

193: P_l(\cos{(\theta(t))}P_l(\cos{(\theta(0))}\rg \ee

194: %

195: of the relevant Legendre polynomial are obtained. Here, $\theta$

196: is defined in the same way as in Fig.1. For instance, the

197: interaction of a permanent electric dipole moment with an applied

198: electric field transforms in the same way as the first order

199: Legendre polynomial $P_1(\cos{\theta})=\cos{\theta}$ and therefore

200: $C_1(t)$ is observed in dielectric spectroscopy. NMR and light

201: scattering are $l=2$ techniques and yield $C_2(t)$. Below it will

202: be shown that $C_d(t)$, apart from being a local correlation

203: function, cannot be related to one of the $C_l(t)$ in a simple

204: manner.

205: %

206: %

207: %

208: %\subsection*{Theory}

209: 

210: %

211: The theoretical description of reorientational dynamics in

212: molecular liquids is a complicated many particle problem. For a

213: qualitative understanding of the behavior of rotational

214: correlation functions, however, a treatment in terms of rotational

215: Brownian motion is usually sufficient. In the present letter we

216: restrict ourselves to the simple model of anisotropic rotational

217: diffusion, thus neglecting inertial effects completely. This

218: approximation is well suited for the systems we have in mind,

219: namely biological systems or supercooled liquids and glasses,

220: where the time scale of the relevant relaxational modes is slow

221: compared to typical vibrational frequencies.

222: 

223: In all following considerations we assume the transition dipole

224: matrix elements to be stationary quantities. In this case we have

225: $C_d(t)=\lg d(t)d(0)\rg$, with

226: \begin{equation}\label{def_d}

227:    d(\theta,\phi)={\cos^2{\theta}-\sin^2{\theta}\cos^2{\phi}\over

228:    \cos^2{\theta}+\sin^2{\theta}\cos^2{\phi}} .

229: \end{equation}

230: Note that $d(\theta,\phi)$ cannot be expressed in terms of the

231: spherical harmonics $Y_{2m}(\theta,\phi)$ and this is why $C_d(t)$

232: cannot be related to the corresponding RCF $C_2(t)$. This is to be

233: contrasted to the optical anisotropy, the correlation function of

234: which is directly proportional to $C_2(t)$.

235: 

236: In order to keep the treatment general, in the following we treat

237: auto-correlation functions of the form

238: %

239: \be \label{Cx.t} C_X(t)=\lg X(\Om_{PL}(t))X(\Om_{PL}(0))\rg. \ee

240: %

241: Here, $X$ denotes some arbitrary function of the molecular

242: orientation $\Om_{PL}$ of the principal axis system (P) of the

243: relevant interaction tensor in the laboratory fixed frame (L). In

244: the present context we are primarily interested in the case of

245: fluorescence, $X\!=\!d$, in which case the z-axis of the P-system

246: coincides with the direction of the transition dipole. However,

247: other choices like e.g. $X\!=\!e^{i{\bf q}{\bf r}}$ relevant for

248: incoherent neutron scattering can be analyzed in an identical way.

249: If the molecular orientation $\Om(t)$, given in terms of Eulerian

250: angles\cite{rose57}, is modelled as a stochastic process, $C_X(t)$

251: can be written in the form\cite{vkamp81}:

252: %

253: \be\label{Cx.stoch} C_X(t)={1\over

254: 8\pi^2}\int\!d\Om\int\!d\Om_0X(\Om)X(\Om_0)P(\Om,t|\Om_0) \ee

255: %

256: where $P(\Om,t|\Om_0)$ denotes the conditional probability to find

257: $\Om$ at time $t$ given $\Om_0$ at time $t\!=\!0$. Throughout this

258: letter we restrict ourselves to the model of anisotropic

259: rotational diffusion of symmetric top molecules, in which case one

260: has

261: %

262: \begin{equation}\label{P.Dlmn}

263: \begin{array}{c}

264:   P(\Om,t|\Om_0)=\sum_{l,m,n}\left(\frac{2l+1}{8\pi^2}\right)

265:              D_{mn}^{(l)*}(\Om_0)D_{mn}^{(l)}(\Om)  \\ \\

266:    e^{-\left[l(l+1)D_x+m^2(D_z-D_x)\right]t} \\

267: \end{array}

268: \end{equation}

269: %

270: Here, the $D_{nm}^{(l)}(\Om)$ are Wigner rotation matrix

271: elements\cite{rose57}. The rotational diffusion coefficients $D_y$

272: and $D_x$ are equal, but different from $D_z$. The limit of

273: isotropic rotational diffusion is recovered for $D_x\!\!=\!\!D_z$.

274: It is important to point out that $\Om$ in eq.(\ref{P.Dlmn})

275: denotes the orientation of the diffusion tensor (D) in the

276: L-system, $\Om\!\equiv\!\Om_{DL}$.

277: 

278: In order to proceed in the calculation of $C_X(t)$, we expand the

279: quantity $X(\Om_{PL})$ in the expression for the correlation

280: function, eq.(\ref{Cx.t}), in terms of $D_{nm}^{(l)}$,

281: $X(\Om_{PL}(t))=\sum_{l,mn}X_{l;mn}D_{nm}^{(l)}(\Om_{PL}(t))$ with

282: $X_{l;mn}={2l+1\over 8\pi^2}\int\!d\Om D_{nm}^{(l)*}(\Om)X(\Om)$.

283: Next, $D_{nm}^{(l)}(\Om_{PL}(t))$ is expressed in terms of the

284: relevant $\Om_{DL}(t)$ via $\Om_{PL}(t)\!=\!\Om_{PD}+\Om_{DL}(t)$

285: using $D_{nm}^{(l)}(\Om_{PL}(t))\!= \!\sum_\mu

286: D_{n\mu}^{(l)}(\Om_{PD})D_{\mu m}^{(l)}(\Om_{DL}(t))$. Here,

287: $\Om_{PD}\!=\!(\a,\theta_{PD},\pi-\phi_{PD})$ is a shape and

288: symmetry dependent molecular quantity. For most relevant cases

289: $X(\Om_{PD})$ has axial symmetry, which allows to replace the

290: $D_{0m}^{(l)}$ by spherical harmonics $Y_{lm}$. Performing the

291: calculation using eqns.(\ref{Cx.stoch}) and (\ref{P.Dlmn}) yields:

292: %

293: \be\label{Cx.Dlmn} C_X(t)=\sum_l A_l C_l(t) .\ee

294: %

295: With $A_l={1\over 4\pi}

296: \sum_m\left|\int_0^{2\pi}\!d\phi\int_0^\pi\!d\theta\sin{\theta}

297: X(\theta,\phi)Y_{lm}(\theta,\phi)\right|^2$ and

298: %

299: \be\label{Fl.def}

300: C_l(t)=\sum_m\left|D_{0m}^{(l)}(\Om_{PD})\right|^2

301:      e^{-\left[l(l+1)D_x+m^2(D_z-D_x)\right]t} ,

302: \ee

303: %

304: the RCFs eq.(\ref{Cx.Dlmn}) are calculated numerically.

305: Typically, it is sufficient to use $l$ values up to $l\!=\!20$.

306: Thus, $C_d(t)$ is given by a weighted sum of a large number of

307: RCFs. This fact has not been noticed in the literature before to

308: the best of our knowledge. As already mentioned above, for the

309: optical anisotropy one has $\lg r(t)r(0)\rg\!=\!C_2(t)$ which

310: decays as a superposition of at most three exponentials.

311: %

312: %

313: %

314: %\subsection*{Isotropic rotational diffusion}

315: 

316: %

317: For the case of isotropic rotational diffusion, i.e.

318: $D=D_x=D_y=D_z$, eq.(\ref{Cx.Dlmn}) simplifies to

319: \begin{equation} \label{isotrop}

320:    C_X(t)=\sum_l A_l e^{-l(l+1)Dt}

321: \end{equation}

322: For symmetry reasons, the odd components vanish.

323: % Änderung nach Report 27.5.2004

324: {A similar expression has been reported recently

325: \cite{hochstrasser2003}.}

326: %

327: In table~\ref{tab1} the pre-factors are listed up to $l\!=\!20$.

328: %

329: \begin{table}[h]

330: \begin{tabular}{|c|c||c|c|}

331:   \hline

332:   $l$ & $A_l$ & $l$ & $A_l$ \\

333:   \hline

334:   2 & 0.83501300 & 12 & 0.00417137 \\

335:   4 & 0.10020500 & 14 & 0.00266071 \\

336:   6 & 0.03101390 & 16 & 0.00180131 \\

337:   8 & 0.01352020 & 18 & 0.00127513 \\

338:  10 & 0.00708445 & 20 & 0.00093625 \\

339:   \hline

340: \end{tabular}

341: \caption{Pre-factors $A_l$ from eq.(\ref{isotrop}) for isotropic

342: rotational diffusion} \label{tab1}

343: \end{table}

344: %

345: The resulting correlation function significantly deviates from an

346: exponential decay as shown in Fig.2. From fitting the computed

347: data with a stretched exponential function

348: $f(t)=e^{-\left(t/\tau_c\right)^{\beta}}$ we obtain

349: $\tau_c=0.87/6D$ and $\beta=0.871$. Therefore, even for this

350: simple model we find intrinsic non-exponential relaxation. The

351: optical anisotropy decays exponentially, $\lg

352: r(t)r(0)\rg\!=\!e^{-6Dt}$.

353: %

354: \begin{figure}

355:   \includegraphics[width=6cm,height=5cm]{fig2}\\

356:   \caption{Rotational correlation function $C_d(t)$ for isotropic

357:      diffusion (solid line). For comparison the exponential

358:      correlation function $l=2$ (eq.\ref{isotrop}:

359:      $A_2=1, A_{l\neq 2}=0$) is plotted as dotted line}

360: \label{121}

361: \end{figure}

362: %

363: %

364: %

365: %\subsection*{Anisotropic rotational diffusion}

366: 

367: %

368: The model of stochastic isotropic reorientations presents a

369: drastic approximation as most molecules do not exhibit a spherical

370: shape. Thus, in many cases of practical interest one is confronted

371: with asymmetric top molecules. As noted above, however, in the

372: present letter we restrict ourselves to the case of symmetric tops

373: for simplicity. This means that we consider an ellipsoidal

374: diffusion tensor with three components $D_x=D_y$ and $D_z$, see

375: Fig. \ref{figaniso}a. For standard $l=1,2$ methods the RCFs

376: resulting for this model consist of two and three exponentials,

377: respectively, with time constants depending on the degree of

378: anisotropy $\d=D_z/D_x$. It is evident from eq.(\ref{Fl.def}) that

379: for $D_x\neq D_z$ the degree of non-exponentiality is more

380: pronounced due to the fact that for each $l$ now one has $l+1$

381: exponentially decaying functions with different weights. However,

382: the orientation of the transition dipoles with respect to $D_z$

383: determines the degree of this additional non-exponentiality. Only

384: for $D_z\!\parallel\!\vec{\mu}$ the same stretching exponent as

385: for isotropic reorientation is obtained.

386: %

387: \begin{figure}

388:   \includegraphics[width=7cm,height=9cm]{fig3}\\

389:   \caption{(a) Stretching exponent $\beta$ depending on the anisotropy

390:   $\delta=D_z/D_x$ ($D_x=D_y$). $\theta_{PD}$ denotes the angle between

391:   the z-axis of the diffusion tensor and the transition dipole $\vec{\mu}$.

392:   (b) Correlation times normalized by $\tau_0=\tau(D_z\!\parallel\!\vec{\mu})$. $\tau$

393:   and $\beta$ were obtained by fitting $C(t)$ with

394:   $exp(-(t/\tau)^{\beta})$

395:   }

396: \label{figaniso}

397: \end{figure}

398: %

399: %

400: %

401: %\subsection*{Rotational jump processes}

402: 

403: %

404: An entirely different approach to the calculation of RCFs via

405: rotational random walk simulations has proved to be successful in

406: the context of supercooled liquids\cite{gerald98}. Here, molecular

407: reorientation has been modelled by rotational jump processes.

408: Starting from an arbitrary orientation, after a certain waiting

409: time drawn from Poisson statistics the next orientation is chosen

410: at random with the only restriction that starting and end position

411: differ by an angle $\gamma$. In the limit of $\gamma \rightarrow

412: 0$ isotropic rotational diffusion is obtained with results

413: identical to those obtained with the analytic approach.

414: %

415: \begin{figure}

416:   \includegraphics[width=6cm,height=7cm]{fig4}\\

417:   \caption{Influence of the reorientation mechanism on the

418:   stretching ($\circ$) and the time scale ($\bullet$) of the correlation

419:   function, calculated for rotational jump processes with varying

420:   jump angles $\gamma$. For comparison the correlation times were

421:   normalized by the pure $l=2$ correlation time $\tau_2$.

422:   }

423: \label{figrandomjump}

424: \end{figure}

425: %

426: In Fig.~\ref{figrandomjump} the influence of the jump angle

427: $\gamma$ on the correlation function $C_d(t)$ defined by

428: eqns.(\ref{def_d}) and (\ref{Cx.t}) is plotted. With increasing

429: $\gamma$ the correlation decays become more exponential, which

430: essentially originates from an convergence of the different $l$

431: contributions to the RCF.

432: %

433: %

434: %

435: %\subsection*{Internal rotations}

436: 

437: %

438: So far, we have assumed that the principal axes system (P)

439: relevant for experimental observables has a fixed orientation

440: $\Om_{PD}$ relative to the 'diffusion tensor' system (D), and only

441: the orientation of the latter with respect to the laboratory

442: system (L) is time dependent: $\Om(t)=\Om_{DL}(t)$. In principle,

443: one can treat the case of internal rotations by assuming a

444: composite Markov process~\cite{vkamp81}

445: $\{\Om_{PD}(t),\Om_{DL}(t)\}$ and using an appropriate master

446: equation for the composite process. If the internal motion is

447: independent of the tumbling motion of the whole molecule one can

448: factorize the corresponding probability functions and averages.

449: This case is of particular relevance for the conformational

450: motions of polymers or proteins in solution and has been treated

451: extensively in the literature to which we refer\cite{WS78}.

452: %

453: %

454: 

455: {In the present analysis we have neglected the influence of the

456: numerical aperture (NA) on the optical path. Using very high NA

457: objectives (NA $>1$) out-of-plane contributions of the transition

458: (emission) dipole to the fluorescence signal have to be

459: considered\cite{ha1999,vdbout1}. However, it was pointed out that

460: even for NA$=1.25$ the dichroism signal is only slightly

461: influenced by this effect\cite{vdbout1}. Additionally, if one were

462: to measure rotational dynamics over an extended temperature range,

463: it is experimentally more convenient to use low NA ($<1$)

464: objectives. Here still an influence of the NA is expected, which,

465: however, becomes smaller the smaller the NA is. Nevertheless, the

466: accurate incorporation of the NA into our theoretical description

467: has still to be worked out. We also would like to mention that

468: methods have been developed which utilize the longitudinal field

469: component of high NA objectives to extract the full three

470: dimensional orientation of transition dipoles\cite{3D1,3D2}.

471: Although quite powerful, such measurements require a very good S/N

472: ratio for subsequent data analysis and are limited with regard to

473: time resolution. }

474: 

475: 

476: 

477: 

478: 

479: %

480: %\subsection*{Conclusions}

481: 

482: %

483: While in the past only bulk experimental techniques were available

484: for studying molecular dynamics, recent experimental progress has

485: enabled to study dynamical processes at the single molecule level.

486: In particular, fluorescence microscopy has allowed to track

487: molecular orientation in time. Here the merits of single molecule

488: microscopy emerge in probing rotational dynamics in heterogeneous

489: environments in a direct way.

490: %

491: We found that even for the model of isotropic rotational diffusion

492: the obtained correlation function decays non-exponentially.

493: Motional anisotropy increases the deviation from exponentiality.

494: %

495: If the RCFs calculated for the simple model of anisotropic

496: rotational diffusion are parametrized by a stretched exponential,

497: we find stretching parameters $\beta \geq 0.85$ for physically

498: reasonable values of the anisotropy $\delta$ between $0.3$ and

499: $3$. From this we conclude that only $\beta$ values smaller than

500: roughly $0.85$ can be taken as an indication for intrinsically

501: non-exponential dynamics. Our findings suggest a careful

502: interpretation of single molecule rotational correlation functions

503: as obtained by polarization resolved microscopy.

504: 

505: 

506: \begin{thebibliography}{9}

507: \bibitem{berne1990} % 1

508:   B. J. Berne, R. Pecora, {\it Dynamical Light Scattering} (R. E.

509:   Krieger, Malabar, Florida, 1990)

510: \bibitem{boettcher1992} % 2

511:   C. J. F. B\"ottcher, P. Bordewijk, {\it Theory of Electric

512:   Polaraizaton} (Elsevier, Amsterdam, 1992)

513: \bibitem{boehmer2001} % 3

514:   R. B\"ohmer, G. Diezemann, G. Hinze, E. R\"ossler, Prog. NMR

515:   Spectr. {\bf 39} 191 (2001)

516: \bibitem{richert2002} % 4

517:    R. B\"ohmer, Curr. Opin. Solid State Mater. Sci. {\bf 3} 378

518:    (1998); H. Sillescu, J. Non-Cryst. Solids {\bf 243} 81 (1999);

519:    M.D. Ediger, Annu. Rev. Phys. Chem. {\bf 51} 99 (2000)

520: \bibitem{sillescu2002} % 5

521:   H. Sillescu, R. B\"ohmer, G. Diezemann, G. Hinze, J. Non-Cryst.

522:   Solids {\bf 307-310} 16 (2002)

523: \bibitem{dreizehn1998} % 6

524:   R. B\"ohmer, R. V. Chamberlin, G. Diezemann, B. Geil, A. Heuer,

525:   G. Hinze, S. C. Kuebler, R. Richert, B. Schiener, H. Sillescu,

526:   H. W. Spiess, U. Tracht, M. Wilhelm, J. Non-Cryst. Solids {\bf

527:   235-237}, 1 (1998)

528: \bibitem{Castillo03} % 7

529:   H. E. Castillo, C. Chamon, L. F. Cugliandolo, J. L. Iguain, M. P.

530:   Kennett, Phys. Rev. B {\bf 68} 134442 (2003)

531: \bibitem{Cugliandolo03} % 8

532:   L. F. Cugliandolo, private communications

533: \bibitem{basche1997} % 9

534:   T. Basch$\acute{e}$, W. E. Moerner, M. Orrit, U. P. Wild, {\it Single

535:   Molecule Optical detection, Imaging and Spectroscopy} (VCH, Weinheim, 1997)

536: \bibitem{science1999} % 10

537:   {\it Frontiers in Chemistry: Single Molecules}, Science {\bf

538:   283} 1670 (1999)

539: \bibitem{vallee2003} % 11

540:   R. A. Vallee, N. Tomczak, L. Kuipers, G. J. Vansco, N. F. van

541:   Hulst, Phys. Rev. Lett. {\bf 91} 038301 (2003)

542: \bibitem{ambrose1991} % 12

543:   W. P. Ambrose, W. E. Moerner, Nature (London)  {\bf 349} 225

544:   (1991)

545: \bibitem{lu1997} % 13

546:   H. P. Lu, X. S. Xie, Nature (London) {\bf 385} 143 (1997)

547: \bibitem{lippitz2004}% 14

548:   M. Lippitz, C. G. H\"ubner, Th. Christ, H. Eichner, P. Bordart,

549:   A. Hermann, K. M\"ullen, Th. Basch\'{e}, Phys. Rev. Lett. {\bf

550:   92} 103001 (2004)

551: \bibitem{ha1996} % 15

552:   T. Ha, Th. Enderle, D. S. Chemla, P. R. Selvin, S. Weiss, Phys.

553:   Rev. Lett. {\bf 77} 3979 (1996)

554: \bibitem{ha1999} % 16

555:   T. Ha, T. A. Laurence, D. S. Chemla, S. Weiss, J. Phys. Chem. B

556:   {\bf 103} 6839 (1999)

557: \bibitem{harms1999} % 17

558:   G. S. Harms, M. Sonnleitner, G. J. Schütz, H. J. Gruber, Th.

559:   Schmidt, Biophys. J. {\bf 77} 2864 (1999)

560: \bibitem{vdbout1} % 18

561:   L. A. Deschenes, D. A. Vanden Bout, Science {\bf 292} 255 (2001)

562: \bibitem{vdbout2} % 19

563:   L. A. Deschenes, D. A. Vanden Bout, J. Chem. Phys. {\bf 116} 5850 (2002)

564: \bibitem{vdbout3} % 20

565:   L. A. Deschenes, D. A. Vanden Bout, J. Phys. Chem. {\bf 106} 11438 (2002)

566: \bibitem{lakowicz1999} % 21

567:   J. R. Lakowicz, {\it Principles of Fluorescence Spectroscopy} (Kluwer Academic, New York, 1999)

568: \bibitem{pecora75} % 22

569:   S. R. Arag\'on, R. Pecora, Biopolymers {\bf 14} 119 (1975)

570: \bibitem{pecora76} % 23

571:   S. R. Arag\'on, R. Pecora, J. Chem. Phys. {\bf 64} 1791 (1976)

572: \bibitem{hochstrasser2003} % 24

573:   E. Mei, J. Tang, J. M. Vanderkooi, R. M. Hochstrasser, J. Am.

574:   Chem. Soc. {\bf 125} 2730 (2003)

575: \bibitem{rose57} % 25

576:   M.E. Rose, {\it Elementary Theory of Angular Momentum}, Wiley, New

577:   York (1957)

578: \bibitem{vkamp81} % 26

579:   N.G. van Kampen: {\it Stochastic Processes in Physics and

580:   Chemistry}, North-Holland, Amsterdam, New York, Oxford (1981)

581: \bibitem{gerald98} %27

582:   G. Hinze, Phys. Rev. E {\bf 57} 2010 (1998)

583: \bibitem{WS78} %28

584:   R.J. Witteborg and A. Szabo, J. Chem. Phys. {\bf 69} 1722 (1978)

585: \bibitem{3D1} % 31

586:   B. Sick, B. Hecht, L. Novotny, Phys. Rev. Lett. {\bf 85} 4482

587:   (2000)

588: \bibitem{3D2} % 32

589:   J. T. Fourkas, Opt. Lett. {\bf 26} 211 (2001)

590: 

591: \end{thebibliography}

592: 

593: \end{document}

594: