Invited Paper Ultra high-throughput single molecule spectroscopy with a 1024 pixel SPAD Ryan A. Colyera*, Giuseppe Scaliaa, Federica A. Villab, Fabrizio Guerrierib, Simone Tisac, Franco Zappab, Sergio Covab, Shimon Weissa, Xavier Michaleta* a b Department of Chemistry & Biochemistry, UCLA, Los Angeles, CA Dipartimento di Elettronica ed Informazione, Politecnico di Milano, Milano, Italy c Micro Photon Devices, Bolzano, Italy ABSTRACT Single-molecule spectroscopy is a powerful approach to measuring molecular properties such as size, brightness, conformation, and binding constants. Due to the low concentrations in the single-molecule regime, measurements with good statistical accuracy require long acquisition times. Previously we showed a factor of 8 improvement in acquisition speed using a custom-CMOS 8x1 SPAD array. Here we present preliminary results with a 64X improvement in throughput obtained using a liquid crystal on silicon spatial light modulator (LCOS-SLM) and a novel standard CMOS 1024 pixel SPAD array, opening the way to truly high-throughput single-molecule spectroscopy. Keywords: single-molecule, photon counting, fluorescence, FCS, LCOS, CMOS, SPAD array, high-throughput. 1. INTRODUCTION Single–Molecule Fluorescence Spectroscopy (SMFS) methods have found application in scientific domains as diverse as super-resolution imaging, structural biochemistry, and single-protein tracking in live cells, yielding insights into outstanding fundamental biological questions [1]. Since they have to operate at the single-molecule level (i.e, low concentrations), they generally require a long acquisition time (several minutes) to have adequate statistical accuracy. Therefore, an increased throughput in SMFS is desirable mainly for two reasons: i) many different reactions can be monitored together at the same time thanks to the multi-spot geometry; ii) fast evolving dynamic systems can be observed by acquiring the same kind of data from different locations and pooling them together to obtain good statistical accuracy before the dynamic system has changed [2]. In order to increase the single molecule throughput, data should be collected in parallel by means of a multi-spot excitation, fast multi-pixel detection, algorithms to handle many channels and suitable software to process all the data. A novel approach to High-Throughput Fluorescence Correlation Spectroscopy (HT-FCS) has been developed in a confocal geometry, in which multiple microscopic volumes in a solution are simultaneously illuminated with a tightly focused laser beam. Each spot of the excitation pattern is mapped in a pixel of a photon detector array. HT-FCS has been presented previously by us using 8 excitation spots with an 8x1 photon detector array [3-4]. In this article we present a description of the developed methods, and we present data obtained with a 32x32 photo detector array, showing a remarkable improvement in terms of throughput and parallelization. The experimental results show a factor of 64 improvement in FCS throughput, demonstrated with both beads and freely diffusing Cy3B in sucrose. * ryancolyer@yahoo.com, michalet@chem.ucla.edu, Phone: 1-310-794-6693, Fax: 1-310-267-4672 Single Molecule Spectroscopy and Imaging IV, edited by Jörg Enderlein, Zygmunt K. Gryczynski, Rainer Erdmann, Proc. of SPIE Vol. 7905, 790503 · © 2011 SPIE · CCC code: 1605-7422/11/$18 · doi: 10.1117/12.874238 Proc. of SPIE Vol. 7905 790503-1 2. METHODS AND EXPERIMENTAL SETUP Fluorescence Correlation Spectroscopy is a technique that analyzes the fluctuations in fluorescence intensity recorded from a sample, due to changes in the number of particles entering or leaving the focal volume. FCS is generally performed at nanomolar concentrations, which provides a good compromise between obtaining enough signal during the finite time of the measurement and being able to observe separate bursts of fluorescence light above the noise [5]. In order to estimate the diffusion constant (D) and the concentration (C) of the sample, the intensity time trace and the Auto Correlation Function (ACF) of the luminescence signal are computed. From the ACF of the time trace it is possible to detect the bursts and compute the transit time (d) of the molecule in the excitation volume. The transit time is related to the diffusion coefficient (D) by means of the following formula: 4 , (1) where is the beam waist (-1/e to 1/e) perpendicular to the optical axis (xy plane) of the Point Spread Function , (PSF) of the excitation volume, s indicates a specific sample and k the single channel (excitation spot and corresponding detector pixel). In order to compute the sample concentration, the ACF of the measured signal is fitted with the following formula: 1 is the ACF and channel: 1 is a parameter proportional to the concentration ( (2) ), that depends on the sample and on the (3) is the excitation volume and is a parameter that depends on the ratio between the background intensity and where the total intensity of the time trace per each channel. Since it is difficult to measure the parameters and , , they are estimated using a reference sample with known D0 and C0, fitting and from the experimental data. In fact we have: (4) , 4 (5) In HT-FCS, the ACF curves computed from different pixels can be merged together in order to obtain a good statistic in a short time. However, the parameters extracted from ACFs of different pixels (i.e., different channel k) are usually very spread due to the non-uniformity in the excitation spot size, in the alignment, and in the pixel performance. Therefore, before merging all the data together, a calibration of the ACFs is necessary to rescale the curves More details about this procedure and its validation are reported in [3]. In the present approach, we exploited a Liquid Crystal on Silicon (LCOS) to generate a pattern of excitation spots and a Single Photon Avalanche Diode array (SPADa) to detect the fluorescence light of the emission path. A schematic of the overall setup is illustrated in figure 1. The LCOS pattern is imaged into the objective by means of a recollimating lens and focused on the sample plane by the microscope objective (UPlan Apo, Olympus, Center Valley, PA; 60X, NA = 1.2) generating diffraction limited spots. The pindot just before the objective is a spot of metal, which blocks all the unmodulated light. The pattern re-emitted by the sample is magnified in order to ensure a perfect alignment with the detector. Proc. of SPIE Vol. 7905 790503-2 Figuure 1. Schematic of the experimeental setup usingg an LCOS-SLM M. Blue lines reprresent the excitaation light path. Greeen lines represen nt the emission liight path towardds the detector (S SPAD array). A first f array of spoots is generated in ann intermediate image i plane in front f of the LCO OS. A recollimaating lens sends this pattern to the t back of an objeective lens, which h focuses it into the sample. CS: Coverslip, Obj: objective lens, FM: flippable mirror. m Top left insett: The LCOS paattern degrees off freedom are coontrollable by software. The patttern pitch and nuumber of spots can also be adjusted. 2.1 Liquid Crystal C on Siliicon (LCOS) The LCOS S (X10468-01, Hamamatsu, Bridgewater; B N is a phase modulator NJ) m that relay on a pixel-by-pixel bassis [6]. It is commonlyy used in a special-frequen s ncy domain annd requires a iterative andd time-consum ming computatiion. Our approach usees the LCOS in n the real-spacee domain to generate a real-sppace array of spots s at an interrmediate focal plane in front of the LCOS, L as show wn in Figure 2. According to the Huygens-F Fresnel principple light rays prropagate outwaard in all directions, thhus with a pro oper phase shifft pattern it is possible to buuild a fixed paattern exploitinng the construcctive and destructive innterference [7]]. The applicattion of this maathematical method in a micrroscopy setup has h been demoonstrated for the first time in [3]. Figgure 2. Schemattic of the Huygeens-Fresnel princciple: rays interfeere constructively at a focal point when w they have thee sam me total phase delaay. This method can be replicaated in order to create a multi--spot excitationn pattern with adjustable spott number, size,, e since it is i distance and angle. It makees the alignmennt procedure off the excitationn spots with thee array pixels easier, possible to rootate the opticaal pattern and move m single sppots without touuching the phyysical setup. In order to managge the LCOS patterrn generation, custom c C and LabVIEW L softw ware has been developed. Proc. of SPIE Vol. 7905 790503-3 2.1 Singlee-Photon Avala anche Diode array a What is required r in parrallel HT-FCS is an array off detection channnels providinng high sensitivvity together with w high frame rate. For both requ uirements Single Photon Avalanche A Dioddes (SPADs) are ideal candidates to reaach high performances in compact microelectroni m ics chips. The convenience of o fluorescencee microscopy with SPAD arrrays has been alreadyy demonstrated [3,8,9]. A SPAD is essentially a p-n junction biased above breakdown forr exploiting ann operation moode also calledd Geigermode, becauuse of the analo ogy with gas counters c of ionnizing radiationn. One digital pulse is produuced for every detected photon. A neearby circuit prrovides the avaalanche quenchhing and rechaarge mechanism ms. In fact, upoon photon detection, an avalanche caarrier multipliccation process is triggered within w the SPAD D and the queenching circuitrry stops the avvalanche current by loowering the rev verse voltage below breakdow wn. This prevennts the destruction of the devvice, while the recharge circuitry reseets the SPAD for the next detection cycle by raising thee bias voltage again to the operating o level. SPADs show single photon sensitiivity and no reeadout noise, since they proovide a digital pulse signal only o when a photon p is detected or when w a carrier is i thermally geenerated, thus leading to a so called dark-count [10]. Silicon SPADs can be integrated i withh the front-end circuitry in CM MOS technoloogy and single--chip monolithic arrays of “smart pixxels” have alreaady been manuufactured [11,112]. The meassurements repo orted in this paper have beenn obtained by means of a 322 x 32 SPAD array developeed in the SPADlabs att Politecnico dii Milano, in a standard s high-vvoltage 0.35µm m CMOS technnology. The arrray operates inn photoncounting moode, i.e. pulses arriving withiin a set integraation time are summed up annd the digital output is thereefore the number of phhotons counted d during consecutive time fraames. Every pixxel comprises a 20 µm-diam meter SPAD, a VariableV Load Quenchhing Circuit [1 13], an 8 bit coounter and a laatch memory. The T pitch betw ween pixels is 100µm and thee overall chip dimensiions are 3.5x3.5 mm. Figure 3 shows the laayout of the enttire array and a scheme of a single pixel. Thanks T to the compact active area of the entire chip, this array is particularly p ideeal for confocal microscopy and a spectroscoppy, since it avoids thee need to sepaarately align a pinhole array.. The small acctive area relattive to the spoot spacing intrrinsically provides thiss clipping of ou ut-of-focus lighht. Figgure 3. 32 x 32 pixel Single Photon Avalancche Diodes arraay. The inset shhows a schemaatic of the pixeel coomprehensive off a 20 µm-diameeter SPAD, a Varriable-Load Queenching Circuit [8], an 8 bit couunter and a latchh meemory. Measuredd Photo Detection Efficiencyy (PDE) tops 43% at 5V exxcess-bias, whiile Dark-Count Rate (DCR) at room temperature is less than 5k kcps (counts peer second) for 70% of the pixxels and afterppulsing probabbility is about 3%, 3 with 300ns dead time t (time afteer each detectiion during whhich the detectoor is gated-offf). All pixels operate o in paraallel in a global shutteer mode and readout r is perfformed with noo blind-time between b conseccutive frames. In fact at eacch frame photon counnts are stored into the latches (one for eaach pixel) whiile photon couunters are reseet and readoutt can be performed during d the inteegration time of the follow wing frame. Thhe maximum frame-rate cann be set by the t user, depending on the system clock of the readout r board: with a conveenient 100MHzz system clockk we achievedd a freerunning speeed of 100,000 frames/s f from the whole 10224 pixels [14]. Even higher frame fr rates cann be easily achiieved by addressing annd reading outt smaller sub-ppixel sets of thhe array chip, with w an absoluute maximum of o 100,000,0000 frame/s from a singlee pixel, i.e. a photon p countinng time-taggingg resolution off 10ns. In this paper p we reporrt measuremennts on an 8x8 portion of o the 32x32 deetector. Proc. of SPIE Vol. 7905 790503-4 This arraay detector fulffills the main requirements r o HT-FCS: larrge number of pixels, high seensitivity (dow of wn to the single-photonn level), very high acquisitioon speed (i.e. either high fraame-rates or very v short integgration time-sllots) and small active area [2]. nd postprocesssing 2.1 Data traansmission an In order to readout the array chip, we developed a compact detecction module employing e an Opal Kelly XE EM3010 compact boaard featuring a Xilinx Spartaan-3 FPGA, 322 MB 16-bit wide w SDRAM and an on-boaard PLL. The USB U 2.0 interface proovides fast com mmunication frrom FPGA to the remote PC C [15]. We devveloped also a user friendly interface i with LabView w, in order to configure c the array a settings and a download the t photon couunts from the FPGA, F and alsoo another tool for multti-channel data analysis and plotting. p Figuree 4 shows a scrreenshot of the two tools. Figuure 4. Screensho ot of the two cusstom tools develloped with LabV VIEW: on the left the SPAD arrray configurationn and setup alignmentt program; on thee right the multi--channel data annalysis program. The param meters used to configure the array chip are the integrationn time (durationn of each fram me), the dead tim me (time after each deetection, during g which the dettector is gated--off in order too reduce afterpuulsing probabillity), the addreess of the sub-array reggion-of-interesst so that data readout can obtain o only thhe desired pixeels, the numbeer of bits-per-ppixel for accumulatingg counts (in facct in many appplication with low photon ratee, such as FCS, it is preferablle not to read all a 8 bits, in order to achieve a higher frame rate) annd the durationn of the overaall measuremennt. Photon couunt data is dow wnloaded from the FPG GA to the PC at a maximum m transfer rate of o 22 MByte/ss (USB 2) and stored for posst-processing. It I is also possible to plot p in real-tim me the counts of o each pixel byy means of inttensity maps, to t see frames acquired a by thee overall array or withh time traces, an nd to focus on and compare the t photon-couunting time deppendence of inddividual pixelss. During image displaay the averagee DCR can be b automaticallly subtracted for each pixeel, which is coonvenient for visually emphasizing the desired signal s during alignment. In order to verify the correctt data transmisssion, crosschecks are performed and a both actu ual and expeccted durations of each meaasurement aree compared inn order to veerify the completenesss of the inform mation retrievedd by and from the t FPGA. The secoond LabView program, p createed for post-proocessing of thee acquired datta, makes it poossible to repreesent the photon counnts in time-dep pendent wavefform traces wiith adjustable time binning, and to compuute the autocoorrelation function (AC CF) [16,17]. It also a enables enntering diffusioon models for fitting f of all chhannels indepenndently or com mbined in order to extraact the parameters of interestt ( and ) [3]. [ Since a callibration process is required in i order to com mpute the diffusion coeefficient and th he dye concenntration, a featuure has been developed d to fully f automate the processingg. Pixels with high DCR D (higher th han about 20 kcps) could have h too low of a signal-too-noise ratio to t provide meeaningful information. Moreover, sin nce in the calibbration proceddure some averrage operationns among all pixels are madee, a very noisy pixel would w negativeely affect the reesults of the enntire array. Theerefore, a darkk count measurre is performedd and the tool avoids computing c the ACF and perfforming the callibration proceedure on those pixels that shoow a DCR higgher than an adjustablee threshold leveel. Proc. of SPIE Vol. 7905 790503-5 With these two developed programs HT-FCS data acquisition and analysis is an automated, straight-forward, and user-friendly process. 3. RESULTS 3.1 Validation of the calibration process To evaluate the capability of our system for performing an appropriate calibration we used 100nm fluorescent beads in H2O. Figure 5 shows ACFs for the beads, where the data are acquired simultaneously from 64 different channels (using an 8x8 sub-array of the SPAD Array). The comparison between row curves and calibrated curves shows that the calibration process correctly adjusts the curves so that every channel is yielding a similar measurement. Figure 5. Left: Raw ACF curves of 100nm beads in H2O acquired from 8x8 channel. Right: Calibrated ACF curves, where the channel dependence is removed and all the curves collapse to show the diffusion behavior of the sample. After this calibration it is possible to merge the curves obtaining an ACF curve 64 time faster than using one single channel. Therefore we increased the throughput by 64 times. 3.2 Single molecule measurements To evaluate the capability of our system for performing single molecule measurements with single fluorophores, we performed a number of measurements of Cyanine 3B (Cy3B) under various conditions. Figure 6 shows raw and calibrated ACFs for Cy3B 5nM in 200mM NaCl buffer with 40% w/w of sucrose. The data were acquired simultaneously from 64 different channels. Though the correlation amplitude, dampened by the background, is a little lower then with beads, after calibration and fitting the curves still give reliable diffusion times and concentrations. Proc. of SPIE Vol. 7905 790503-6 F Figure 10. Exam mple ACFs for 5nnM Cy3B in 40% % sucrose beforee (left) and after (right) calibratioon. Increasinng the number of o excitation sppots also causees an increase inn the backgrouund due to out--of-focus light. This optical cross-talk interferess with the abilitty to trivially scale s to higher numbers of chhannels, such ass using the fulll 32x32 array. To solvve this problem m the geometryy of the excitatiion pattern muust be changed, such as by siggnificantly incrreasing the spot sepaaration in the saample plane inn order to avoidd cross contribuutions from othher spots. This introduces chaallenges in identifyingg the optimal excitation e geom metry for good signal strengthh while also fittting all 32x32 spots within thhe field of view, and will be the sub bject of future work. w 4. DISCUSSION D N AND CON NCLUSIONS S The feasibbility of singlee molecule HT T-FCS has beenn demonstratedd with a sub-aarray (8x8 pixeels) of a 32x32 SPAD array, using Cy3B C at low co oncentration. New N measurem ments with larger sub-arrays are a in progress.. A highly flexible spot generation appproach using an a LCOS has been developped. It allows rapid changess to spot position, spaacing, and ang gle, and it is very v conveniennt for the com mplicated task of multi-pixeel alignment. A 32x32 photon-counting SPAD arrray has been em mployed for thee first time in HT-FCS, H which has led to a 64X 6 increase inn singlehroughput. Neew user-frienddly LabVIEW programs haave been deveeloped for auutomated molecule sppectroscopy th acquisition and a analysis off the substantiaal quantity of data d generated with multi-chaannel FCS. Futture developments will focus on impproving the illu umination conttrast of the exccitation geomettry when scalinng to larger numbers of pixells so that HT-FCS andd other high-thrroughput singlee-molecule speectroscopy techhniques can maake use of the full f 32x32 SPA AD array 5. ACKNO OWLEDGM MENTS w supported in i part by NIH H grants R01 GM084327 G andd R01 EB0063553, and by NS SF grant DBI-00552099. This work was We thank Drr. Ted Laurencee and Dr. Michhael Culbertsonn for providingg C code to impplement multittau algorithms for time stamped and binned data, respectively. REF FERENCES [1] [2] [3] [4] i (Topics in Flluorescence Sppectroscopy, Voolume Thoompson, N.L., "Fluorescence Correlation Sppectroscopy", in 1), J.R. J Lakowicz,, Plenum Presss, New York, (11991). X. Michalet, M R. A..Colyer,, G. Sccalia, T. Kim, M. M Levi, D. Ahharoni, A. Chenng, F. Guerrierri, G. Arisaka, J. J Milllauda, I. Rech, D. Resnati, S.. Marangoni, A. A Gulinatti, M. Ghioni, S. Tisa, F. Zappa, S. S Cova, S. Weiss, "Higgh-throughput single-molecuule fluorescencee spectroscopyy using parallell detection," Prroceeding SPIE E, Vol 7608, 2010. R. A. A Colyer, G. Scalia, S I. Rech, A. Gulinatti, M. M Ghioni, S. Cova, C S. Weiss,, X. Michalet, "High-throughput FCS S using an LCO OS spatial lightt modulator annd an 8x1 SPAD D array," Biom medical Optics Express E 1(5):1408-31, 2010. R.A A. Colyer, G. Sccalia, T. Kim, I. I Rech, D. Ressnati, S. Maranngoni, M. Ghiooni, S. Cova, S.. Weiss, X. Micchalet "Higgh-throughput multispot singgle-molecule sppectroscopy," Proceedings P off SPIE, 2010. 7571: 7 p. 75710G G. Proc. of SPIE Vol. 7905 790503-7 [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] I. Rech, D.R., S. Marangoni, M. Ghioni, S. Cova, "Compact-eight channel photon counting module with monolithic array detector," Proc. of SPIE 6771(13), 1-9 (2007). H.T. Dai, Y.J. Liu, X. Wang, J. Liu, "Characteristics of LCoS Phase-only spatial light modulator and its applications," Optics Communications, 2004. 238: p. 269-276. X.W. Wang, H. Dai, and K. Xu, "Tunable reflective lens array based on liquid crystal on silicon," Optics Express, 2005. 13(2): p. 352-357. S. Marangoni, I. Rech, M. Ghioni, P. MacCagnani, M. Chiari, M. Cretich, et al., "A 6 × 8 photon-counting array detector system for fast and sensitive analysis of protein microarrays," Sensors and Actuators, B: Chemical,2010 149(2), 420-426. E. P. Dupont, E. Labonne, C. Vandevyver, U. Lehmann, E. Charbon, M.A. Gijs, "Monolithic Silicon Chip for Immunofluorescence Detection on Single Magnetic Beads," Anal. Chem. 2010, 82 (1), 49–52. F. Zappa, S. Tisa, A. Tosi, S. Cova, "Principles and features of Single Photon Avalanche Diode Arrays," Sensors & Actuactors. A: Physical, Vol. 140, Issue 1, 103-112 (2007). F. Zappa, A. Tosi, A. Dalla Mora, F. Guerrieri, and S. Tisa “Single-photon avalanche diode arrays and CMOS microelectronics for counting, timing, and imaging quantum events,” Proc. of SPIE Vol. 7608, 2010. E. Charbon, “Single-photon Imaging in CMOS,” Proc. of SPIE Vol. 7780, 2010). S.Tisa, F. Guerrieri, F. Zappa, “Variable-Load Quenching Circuit for Single-Photon Avalanche Diodes”, Optic Express, Vol. 16 Issue 3, pp.2232-2244, January 2008. Guerrieri F, Tisa S, Tosi A, Zappa F, “Two-Dimensional SPAD-Imaging Camera for Photon Counting”, IEEE Photonic Journal, October 2010. www.opalkelly.com/products/xem3010 (User Manual) T.A. Laurence, S.F., T. Huser, "Fast, flexible algorithm for calculating photon correlations," Optics Letters 31(6), 829-831 (2006). M.J. Culbertson, D.L.B., "A distributed algorithm for multi-tau autocorrelation," Review of Scientific Instruments 78(4), 044102-1-6 (2007). Proc. of SPIE Vol. 7905 790503-8