approach with <strong>temperature</strong> <strong>measurements</strong> based on fluorescencelifetime changes of Rhodam<strong>in</strong>e B and with <strong>measurements</strong> basedon diffusion changes of thermosensitive particles. In contrast tothe last two methods, 2fFCS is generally applicable and does notdepend on the availability of specific materials.Temperature dependent <strong>measurements</strong> require a precise<strong>temperature</strong> control <strong>in</strong> the sample environment. The technicalimplementations of such a sample environment e.g. <strong>in</strong> a microfluidicchamber 22 or a sealed microscope sample cell is far beyondthe scope of this contribution. Recently, we have reported ona <strong>temperature</strong> controlled and sealed sample cell for fluorescencemicroscopes. 23 As already mentioned above, the opticalrequirements of the 2fFCS technique are comparable to otheroptical techniques and thus 2fFCS is well suited for LOC devices.Methods and materialsTheoretical backgroundHere, we briefly recall the theoretical background of 2fFCS. 24 In2fFCS, two overlapp<strong>in</strong>g detection volumes are generated, whichare identical <strong>in</strong> shape but shifted to each other perpendicular tothe optical axis. This is achieved by us<strong>in</strong>g one (or two) pulsedlasers <strong>in</strong> such a way that the polarization of each laser pulse isturned by 90 with respect to the preced<strong>in</strong>g pulse. When send<strong>in</strong>gthis excitation light through a Nomarski prism as used <strong>in</strong>conventional differential <strong>in</strong>terference contrast (DIC) microscopy,each laser pulse is slightly deflected accord<strong>in</strong>g to itspolarization. After focus<strong>in</strong>g the light through a water-immersionobjective with high numerical aperture, this generates two fociwith small lateral shift between them. To dist<strong>in</strong>guish whichfluorescence photon was generated by which laser pulse (i.e. <strong>in</strong>which focus), one measures the photon detection times withpicosecond accuracy us<strong>in</strong>g time-correlated s<strong>in</strong>gle-photon count<strong>in</strong>g(TCSPC) and associates each detected photon with the latestpreced<strong>in</strong>g excitation pulse. This allows for calculat<strong>in</strong>g the autocorrelationfunction (ACF) for each focus as well as the crosscorrelationfunction (CCF) between both foci. Because thelateral shift between both foci is precisely known (it only dependson the properties of the Nomarksi prism), a global analysis ofboth the ACFs and CCF allows for determ<strong>in</strong><strong>in</strong>g an absolutevalue of the diffusion coefficient of fluorescent molecules orparticles <strong>in</strong> solution.As was shown <strong>in</strong> detail <strong>in</strong> ref. 24, adequate model functions forthe diffusion-related part of the ACF or CCF (neglect<strong>in</strong>g fora moment any photophysics-related fluorescence fluctuations)are given by eqn (1):~gðt; d; vÞ ¼ c rffiffiffiffiffið ðpkðz 1 Þkðz 2 Þdz 1 dz 24 Dt 8Dt þ w 2 ðz 1 Þþw 2 ðz 2 Þ" # (1)ðz 2 z 1 Þ 22d 2exp4Dt 8Dt þ w 2 ðz 1 Þþw 2 ðz 2 Þwhere t is the lag-time of correlation, d is the lateral distancebetween the detection volumes, 3 1 and 3 2 are factors proportionalto overall excitation <strong>in</strong>tensity and detection efficiency <strong>in</strong> eachfocus, c is the concentration of fluorescent molecules or particles,and D their diffusion coefficient. Here, the functions k(z) andw(z) are given by eqn (2)–(4):andwithð a drrkðzÞ ¼2R 2 ðzÞ exp0" # 2 1=2lex zwðzÞ ¼ w 0 1 þpw 2 0 n (2)2r 2¼ 1R 2 ðzÞexp2a 2R 2 ðzÞ(3)" #21=2lem zRðzÞ ¼R 0 1 þpR 2 0 n (4)where l ex and l em are excitation and centre emission wavelengths,n is the sample refractive <strong>in</strong>dex, a is the confocal p<strong>in</strong>holeradius, and w 0 and R 0 are fit parameters. For calculat<strong>in</strong>g theACF of each focus, one has to set, <strong>in</strong> eqn (1), d to zero, and toreplace 3 1,2 by either 3 12 or 3 22 , respectively. The <strong>in</strong>tegration <strong>in</strong>eqn (1) has to be performed numerically. Fitt<strong>in</strong>g of experimentaldata is done globally for both the ACFs and the CCF, where onehas fit parameters 3 1 $c 1/2 , 3 2 $c 1/2 , w 0 , R 0 , and D.As was recently shown <strong>in</strong> ref. 25, the absolute fitted values ofw 0 and R 0 can become rather arbitrary when work<strong>in</strong>g underoptical conditions with strong aberration. However, as was alsoshown <strong>in</strong> ref. 25, the fitted value of diffusion coefficient is stillremarkably exact. In other words, the absolute values of w 0 andR 0 , are not significant for the calculation of the diffusion coefficientand therefore they are not reported.The distance d between detection volumes is a setup constantand has to be determ<strong>in</strong>ed only once for a given <strong>measurements</strong>ystem. This can be done for example by determ<strong>in</strong><strong>in</strong>g the diffusioncoefficient of fluorescently labelled polymer particles withknown size. This calibration procedure has been described <strong>in</strong>detail <strong>in</strong> ref. 26.More technical details concern<strong>in</strong>g the setup can be found <strong>in</strong>ref. 24.ExperimentsDynamic light scatter<strong>in</strong>g (DLS) <strong>measurements</strong> were performedon a standard ALV 5000 system, equipped with a laser of 633 nmwavelength. Scatter<strong>in</strong>g <strong>in</strong>tensity was detected at angles of 60 ,90 , and 120 , respectively, and the hydrodynamic radius wascalculated with a second order cumulant fit us<strong>in</strong>g the Stokes–E<strong>in</strong>ste<strong>in</strong> relation. The measurement system was equipped witha <strong>temperature</strong> controlled water bath giv<strong>in</strong>g a precision <strong>in</strong> sample<strong>temperature</strong> stabilization of 0.2 K.The 2fFCS measurement system was based on a Micro-Time200 Fluorescence Spectroscopy and Microscopy System(MT200, PicoQuant GmbH, Berl<strong>in</strong>, Germany) as described <strong>in</strong>ref. 26 and 27 with a dual-focus modification as presented <strong>in</strong> ref.24 and Fig. 1. The setup is equipped with two identical 470 nmlasers (LDH-P-C-470B), two identical 635 nm lasers (LDH-P-635), as well as one 532 nm laser (PicoTA530N), whose beam issplit <strong>in</strong>to two mutually time-delayed pulse tra<strong>in</strong>s. All lasers arel<strong>in</strong>early polarized <strong>in</strong> such a way that, after comb<strong>in</strong><strong>in</strong>g the light ofall lasers, one obta<strong>in</strong>s for each wavelength pulse tra<strong>in</strong>s withalternately switch<strong>in</strong>g polarization. Pulse width is 50 ps, andThis journal is ª The Royal Society of Chemistry 2009 Lab Chip, 2009, 9, 1248–1253 | 1249
Fig. 1Pr<strong>in</strong>ciple of a confocal two focus experiment.impact of a sample refractive <strong>in</strong>dex of 1.52 when focus<strong>in</strong>g at 10mm above a chip’s bottom glass surface, or the impact of a samplerefractive <strong>in</strong>dex of 1.4 when focus<strong>in</strong>g at 27 mm).A data acquisition time of 30 m<strong>in</strong>utes per <strong>temperature</strong> wasemployed both for the lifetime and diffusion <strong>measurements</strong>.Temperature is controlled by a custom-made <strong>temperature</strong>regulation 23 with an absolute <strong>temperature</strong> accuracy of 0.05 K(with<strong>in</strong> the detection volume). The achievable <strong>temperature</strong> rangefor 2fFCS and lifetime <strong>measurements</strong> is 5 to 65 C. Dur<strong>in</strong>g<strong>measurements</strong>, the sample chamber was sealed to prevent solventevaporation and convection.pulse repetition rate is either 20 or 40 MHz, adjustable to thespecific fluorescence lifetime of the measured fluorophore. 28 Thelasers are coupled <strong>in</strong>to a polarization ma<strong>in</strong>ta<strong>in</strong><strong>in</strong>g s<strong>in</strong>gle modefibre for optical clean<strong>in</strong>g, and, after re-collimation, reflectedtowards the microscope’s objective (UPLAPO 60x W, 1.2 N.A.,Olympus Europa (Hamburg, Germany) by a major triple-banddichroic (z470/532/638rpc, AHF-Analysentechnik, T€ub<strong>in</strong>gen,Germany). Before enter<strong>in</strong>g the objective, the laser beam is passedthrough a Nomarski prism (U-DICTHC, Olympus Europe,Hamburg, Germany) to deflect pulses accord<strong>in</strong>g to their polarization<strong>in</strong>to two slightly <strong>in</strong>cl<strong>in</strong>ed directions. After focus<strong>in</strong>gthrough the objective, this generates two overlapp<strong>in</strong>g foci.Fluorescent light is collected by the same objective and focusedonto a s<strong>in</strong>gle circular aperture (diameter 200 mm). After pass<strong>in</strong>gseveral emission filters (HQ505/30m for l ex ¼ 470 nm, HQ580/70m for l ex ¼ 532 nm, and HQ687/70m for l ex ¼ 635 nm, allpurchased from AHF-Analysentechnik, T€ub<strong>in</strong>gen, Germany),the light is split by a non-polariz<strong>in</strong>g beam splitter cube andfocused onto two s<strong>in</strong>gle photon avalanche diodes (SPAD, PDMseries, Micro Photon Devices, Bolzano, Italy). A dedicateds<strong>in</strong>gle-photon count<strong>in</strong>g electronics (PicoHarp 300, PicoQuantCompany, Berl<strong>in</strong>, Germany) is used to record s<strong>in</strong>gle-photondetection events with 4 ps temporal resolution (time tagged timeresolved or TTTR detection mode 29 ). Us<strong>in</strong>g the picoseconddetection times of the fluorescence photons, one determ<strong>in</strong>es bywhich laser pulse and thus <strong>in</strong> which focus the photon wasgenerated. 24 Us<strong>in</strong>g this <strong>in</strong>formation, ACFs and CCF arecomputed with a custom-built MatLab rout<strong>in</strong>e. 30 When calculat<strong>in</strong>gall correlation functions, only photons from differentSPAD detectors are correlated to elim<strong>in</strong>ate afterpuls<strong>in</strong>g and deadtime effects of the detectors.The typical spatial resolution <strong>in</strong> the direction of the opticalaxis is below one micron; the resolution <strong>in</strong> the perpendicularplane is given by the diffraction limit and is typically halfa micron.The robustness of 2fFCS aga<strong>in</strong>st refractive <strong>in</strong>dex mismatchwas experimentally demonstrated <strong>in</strong> ref. 24, where the diffusionof the dye Atto655 <strong>in</strong> aqueous solutions of vary<strong>in</strong>g concentrationsof guanid<strong>in</strong>ium hydrochloride up to 6M had beenmeasured, not f<strong>in</strong>d<strong>in</strong>g any deviation of the determ<strong>in</strong>ed diffusioncoefficient from its theoretically expected value. In ref. 24, anextensive theoretical study of the performance of 2fFCS underdifferent optical and photophysical conditions was presented,show<strong>in</strong>g the exceptional robustness of 2fFCS even underrefractive <strong>in</strong>dex mismatch correspond<strong>in</strong>g to the <strong>in</strong>troduction ofan additional slab of glass of ten micrometer thickness betweenthe objective and the sample (correspond<strong>in</strong>g, for example, to theMaterialsAll samples are prepared us<strong>in</strong>g LiChrosolv water for chromatography(No. 115333, Merck KGaA, Darmstadt, Germany).For 2fFCS experiments we used Atto655-maleimid (No. AD655-4, ATTO-TEC, Siegen, Germany). Rhodam<strong>in</strong>e B (Bio-Chemika, No. 83689) was obta<strong>in</strong>ed from Sigma-Aldrich (Seelze,Germany). 2fFCS s<strong>in</strong>gle molecule experiments were carried outwith nanomolar dye concentrations. Fluorescent labelled poly-nisopropylacrylamide(PNIPAM Rh B ) microgel was synthesizedfollow<strong>in</strong>g standard protocols as published <strong>in</strong> ref. 31–36 us<strong>in</strong>ga mixture of unlabelled and labelled monomers with a molarratio of approx. 1 : 0.016. Labelled monomers (methacryloxyethyl-thio-carbamoyl-rhodam<strong>in</strong>eB, No. 23591) were purchasedfrom Polysciences. (400 Valley Road, Warr<strong>in</strong>gton, PA). DLS and2fFCS experiments are performed at same sample concentrationsof 0.05 wt% microgel solution.We studied the effect of bleach<strong>in</strong>g on 2fFCS <strong>measurements</strong> <strong>in</strong>ref. 24 and ref. 37, while <strong>in</strong>vestigat<strong>in</strong>g the robustness of 2fFCSwith respect to optical saturation of fluorescence. When theexcitation <strong>in</strong>tensity is becom<strong>in</strong>g so large that bleach<strong>in</strong>g starts toimpact the measurement, it leads to an apparent <strong>in</strong>crease <strong>in</strong> thediffusion coefficient. 2fFCS cannot compensate for that andconsequently one has to employ excitation <strong>in</strong>tensities wherebleach<strong>in</strong>g does not yet affect the measurement. In this study2fFCS <strong>measurements</strong> were performed at different excitation<strong>in</strong>tensities and it was ensured that the experimental results were<strong>in</strong>dependent of excitation <strong>in</strong>tensity.Results and discussionLifetime <strong>measurements</strong> on Rhodam<strong>in</strong>e BWe first measured the lifetime dependence of Rhodam<strong>in</strong>e B <strong>in</strong>aqueous solution and <strong>in</strong> methanol. In Fig. 2, a comparison ofs<strong>in</strong>gle molecule lifetime data of Rhodam<strong>in</strong>e B (crosses) with<strong>liter</strong>ature values from ref. 11 and 14 (dots and circles) is presented.We could perfectly reproduce the published values forwater.However, <strong>in</strong> contrast to the report <strong>in</strong> ref. 11, we f<strong>in</strong>d a different<strong>temperature</strong> dependence of the Rhodam<strong>in</strong>e B lifetime <strong>in</strong> methanol(stars). Tak<strong>in</strong>g <strong>in</strong>to account that the lifetime change iscaused by a different rotational flexibility of the dye moleculewhich is aga<strong>in</strong> dependent on the solvent’s viscosity, a different<strong>temperature</strong> dependence of the fluorescence lifetime <strong>in</strong> water andmethanol, respectively, can be expected. This illustrates a disadvantageof the method: besides be<strong>in</strong>g dependent on a specific dye(Rhodam<strong>in</strong>e B), the exact <strong>temperature</strong> behaviour of the lifetime1250 | Lab Chip, 2009, 9, 1248–1253 This journal is ª The Royal Society of Chemistry 2009