Realization of a time-correlated photon counting technique for fluorescence analysis

: An ultralow level light detection module, the time-correlated photon counter, is proposed and evaluated for ﬂuorescence analysis. The time-correlated photon counter employs a silicon photomultiplier as a photon counting sensor in conjunction with a Poisson statistics algorithm and a double time windows technique, and therefore it can accurately count the photon number. The time-correlated photon counter is compatible with the time-correlated single photon counting technique and can record the arrival time of very faint light signals. This low-cost and compact instrument was used to analyze the intensity and lifetime of ﬂuorescein isothiocyanate; a limit of detection of 16 pg/ml with a large linear dynamic range from 2.86 pg/ml to 0.5 µg/ml was obtained, and the lifetime of ﬂuorescein isothiocyanate was measured to be 3.758 ns, which agrees well with the results of a sophisticated commercial ﬂuorescence analysis instrument. The time-correlated photon counter may be useful in applications such as point-of-care testing.


Introduction
Fluorescence analysis is a highly sensitive trace analysis method used in biology, medicine, chemistry, environmental detection, etc. Photomultiplier tubes (PMTs) or single photon avalanche diodes (SPADs) are usually the detectors in fluorescence analysis instruments. However, PMTs suffer from high bias voltage and are fragile and expensive, while SPADs suffer from having no photon-number discrimination and a small photosensitive area [1].
A new type of ultralow level light detector with photon number resolving capability, silicon photomultipliers (SiPMs) have been developed and widely applied in recent years [2][3][4][5]. SiPMs consist of multiple pixels of avalanche photodiodes (APDs) working in Geiger mode and feature perfect photon number and time resolution. In other words, SiPMs can distinguish "Yes" or "No" photons, and also count the photon number in a light pulse. Additionally, SiPMs can measure the arrive time of photons with a time resolution of ∼53 ps [6].
To improve SiPMs for ultralow level light detection, many efforts have been made to overcome the main flaw of SiPMs, which is the high dark count rate (DCR). M.F. Santangelo adopts the net photo current method by subtracting the dark current from the photo current, which requires a precise measurement of the weak current (approximately hundreds of nanoamperometers) [7][8][9]. A photon counting method is also an option for chemiluminescence detection [10]. Another method is a gated charge integration method by which one obtains the charge of the pulse to determine the light intensity [11]. A similar method, which measures the pulse height of SiPMs, is also considered [12]. The time-correlated single photon counting (TCSPC) technique is adopted for measuring the intensity of Raman scattering by recording the photon numbers in a very narrow time gate using SiPMs [13]. However, the TCSPC technique has a low detection efficiency. The improved technique, time-correlated photon counting, which takes advantage of the photon number resolving capability of SiPMs, achieves a higher efficiency [14]. All the abovementioned applications using SiPMs suffer from correlated noise (after pulsing and crosstalk) and DCR fluctuations [15].
To simplify and miniaturize the time-correlated photon counting technique and to eliminate the influence of the correlated noise and DCR fluctuations of SiPMs, the authors reported an electronic module based on a Poisson statistics algorithm (PSA) and double time windows (DTW) with SiPMs [16]. In this study, we made an ultralow level light detection module, time-correlated photon counter (TCPC), which is compatible with the TCSPC technique; therefore, this module can be applied to the measurement of a fluorescence lifetime (FLT) [17]. Both the fluorescence intensity and lifetime measurements for fluorescein isothiocyanate (FITC) solution using the TCPC module were performed, and the FLT results were compared with those of a sophisticated commercial fluorescence spectrometer.

TCPC architecture
The schematic diagram of a TCPC module based on SiPMs is shown in Fig. 1. The TCPC module is composed of a SiPM, negative temperature coefficient thermistor (NTCT), thermoelectric cooler (TEC), amplifier, comparator, bias source, field programmable gate array (FPGA) and microprogrammed control unit (MCU). The SiPM detector in TO encapsulation is assembled with the NTCT and TEC. A temperature adjustment program based on a proportion integration differentiation (PID) algorithm is achieved in the MCU, which can read the NTCT to determine the temperature and control the TEC to cool the SiPM. The SiPM is supplied by the bias source and works in Geiger mode. The output signal from the SiPM is first amplified by a wide bandwidth amplifier (with a gain of 40 dB and bandwidth of 330 MHz) and is converted to a digital signal by a fast comparator. Using the FPGA, a counter is designed to implement PSA for precisely counting pulses, and a time-to-digital converter (TDC) is designed to implement DTW for accurately recording the arrival time of each pulse and synchronous trigger signal. The MCU is a master controller to coordinate the TEC, NTCT, bias source, comparator, and FPGA to all work together. All the data the FPGA acquired are transmitted to the MCU through the inter-integrated circuit (IIC) protocol, and then, they are transmitted to a controller to be processed through the universal asynchronous receiver transmitter (UART). According to the PSA method, the minimum detectable photoelectron number µ min and the maximum detectable photoelectron number µ max are determined by the noise equivalent power and availability for the zero photoelectron events number, respectively. The specific formulas are as follows [16].
where ∆t is the width of the time gate, and f is the repetition rate of the excitation light.

Fluorescence analysis
A schematic diagram of the optical apparatus for fluorescence analysis is shown in Fig. 2.
An orthogonal optical arrangement is adopted, and the excitation light is orthogonal to the fluorescence emission light. The excitation light, filtered with an excitation filter, is focused by a convergent lens to enhance the light intensity and to excite more fluorescence from the specimen. Two bandpass emission filters are used to prevent scattered excitation light from entering the TCPC. The pinhole is conjugated with the specimen to further block the scattered light. The synchronous trigger signal of the light source is connected to the TCPC as an external trigger. Finally, all the data are transmitted to a computer via UART. The fluorescence intensity can be represented by the mean photoelectron number (MPEN). The TCSPC technique is adopted for measuring the FLT. The fluorescence decay is R(t). Theoretically, R(t) can be fit by the multiexponential equation given by Eq. (3).
where t is the time, and α i and τ i are the amplitude and lifetime of i'th component, respectively.
The measured decay F(t) is a convolution of the real decay R(t) with the instrument response function h(t).
Therefore, the instrument response function (IRF) need to be confirmed in advance. The real fluorescence decay can be obtained by reconvolution or deconvolution method [18]. To accurately evaluate the fitted lifetime results, the reduced chi-squared value χ 2 is adopted. The closer the χ 2 is to 1, the more credible the fitted FLT is. When χ 2 is between 0.8 and 1.3, the fitted FLT result is considered to be "ideal" [19].

Experiments
The TCPC module based on the EQR SiPM (silicon photomultiplier with epitaxial quenching resistor) developed in NDL (Novel Device Laboratory) has been fabricated and demonstrated in the fluorescence analysis experiment.

SiPM detector
The EQR SiPM detector uses an epitaxial silicon layer as the quenching resistor, which has advantages, including high photo detection efficiency, while maintaining a large dynamic range, excellent time resolution and low temperature coefficient, etc. [6].
The EQR SiPM employed in this experiment has an active area of 1 mm 2 and a cell pitch of 12.5 µm. The EQR SiPM features a low breakdown voltage of 20 V, a gain on the order of 4.5×10 5 and a photon detection efficiency equal to 33.6% (at a wavelength of 420 nm and an overvoltage of 5 V) [6].
Usually, the DCR of SiPMs significantly reduces with decreasing temperature. As shown in Fig. 3, regarding the EQR SiPM at an operating voltage of 25 V, although the DCR at room temperature (20°C) is approximately 414 kHz, the DCR at -20°C is approximately 30 kHz. According to Eq. (1), the µ min is proportional to the square root of the DCR; thus, the µ min can be reduced 3.7 times. It would be helpful to optimize the limit of detection (LOD) in these fluorescence intensity measurements. In an FLT measurement, decreasing the DCR also means reducing the background counts and results in a better efficiency. Therefore, lower operating temperature of the SiPM will be beneficial for the measurement of fluorescence intensity and lifetime.  Figure 2 shows a photo of the TCPC module, which is assembled with two circuit boards. The module is a compact structure with size of 48×48×30 mm 3 . The first board includes the SiPM, TEC, NTCT, amplifier, TEC controller, fast comparator and MCU. The TO shell of the SiPM attaches to the black aluminum block as closely as possible in order to improve the heating efficiency. The gap between the TO shell and the aluminum block is filled with heat-conducting silicone grease. The second board includes the FPGA and bias source. The TCPC is controlled by a computer via LabVIEW (National Instruments) software.

Fluorescence analysis apparatus
The fluorescence analysis apparatus is made by the 3D printing technology. FITC (purity ≥ 90%, Sigma-Aldrich) is used as the specimen. The FITC solution placed in a transparent cuvette was first dissolved in dimethyl sulfoxide (DMSO, purity > 99.5%, Xilong Scientific), and then diluted in deionized water. A supercontinuum source, which has a repetition rate of 500 kHz and a pulse width of 100 ps, is used as the light source. A neutral density filter is utilized to adjust the excitation light intensity. In addition, there is one bandpass filter (475/35 nm, Thorlabs) for the excitation light, and there are two bandpass emission filters (520/10 nm and 530/43 nm, Thorlabs) for the fluorescent light.
Either in the measurement of the intensity or the lifetime, all parameters can be set via LabVIEW software. The operating voltage of the EQR SiPM was set to 25 V, and its temperature was controlled at -20°C through the TEC and NTCT. The threshold must exceed the electrical noise and was set as 28 mV. The Poisson counts number was set to 1 million for a good signal-to-noise ratio. Considering the FLT of FTIC is approximately 4 ns [20], the time gate width was set to 20 ns for collecting as many fluorescence photons as possible.
In the fluorescent intensity measurements, different FITC solution specimens with varying concentrations from 2.86 pg/ml to 0.5 µg/ml are analyzed. The apparatus can't completely block the scattered light from reaching the detector. A bandpass filter with larger optical density may reduce the background of the scattered light by rejecting more unwanted wavelength from the excitation light. Therefore, to obtain a larger dynamic range, the MPEN of each specimen is obtained by subtracting that of background which only contains the deionized water. The background specimen will be first measured and the results will be inputted to the LabView program to perform an automatic background subtraction in the next measurements of fluorescence specimens. Each specimen is measured repeatedly at least 3 times. According to Eqs. (1) and (2), for DCR = 30 kHz, ∆t = 20 ns and f = 500 kHz, the µ min and µ max can be estimated as 4.9×10 −5 and 13.1, respectively. This estimation predicts a good lowest detectable level and a large linear dynamic range (LDR). In the fluorescent lifetime measurements, the photon rate is recommended to remain at less than 5% of the excitation rate to overcome the pile-up effect, which means the MPEN should be lower than 0.05 [21]. On the other hand, the acquisition time becomes very long for levels of MPEN that are too low, thus the MPEN needs to be properly considered by adjusting the intensity of the excitation light or changing the concentration of the specimen. For characterizing the IRF, an instrument response standard solution is made using sodium iodide (NaI, purity 99%, Macklin) as a quencher [22], which combines 19 g NaI in 13 ml FITC solution with a concentration of 4 µg/ml. The IRF for this experiment is obtained as 272 ps.

Fluorescence intensity
The MPEN at different FITC concentrations is shown in Fig. 4, with varying concentrations from 2.86 pg/ml to 0.5 µg/ml. The R-square value of the linear fit is 0.997, which indicates a good linear fit. The background, including only deionized water, is measured with a MPEN 0 of 6.57×10 −5 and standard deviation (STD) of 1.51×10 −5 . The LOD is defined as the concentration whose MPEN equals the MPEN 0 plus 3 times the STD of the buffer. As a result, the LOD is 16 pg/ml, and the LDR is approximately 5 orders of magnitude, ranging from 2.86 pg/ml to 0.5 µg/ml. Reducing the DCR by effectively cooling the SiPM improves the minimum detectable photoelectron number. Combined with the PSA and DTW methods, the TCPC provided satisfactory results, including a low LOD and large LDR. Additionally, better performance is expectable for the TCPC employing a promoted SiPM.
For comparison, some fluorescence detection results using different SiPMs as detectors are shown in Table 1. The TCPC based on SiPMs can obtain satisfactory results as those measurements using precise but bulky instruments.

Fluorescence lifetime
As shown in Fig. 5, the left figure was obtained by the TCPC, and the right figure was obtained by the FLS980 (Edinburgh Instruments Ltd). The lifetime results were processed using the FAST   The data acquisition time for the TCPC and the FLS980 are about 20 min and 24 min, respectively. And it's obvious to find that the data acquisition efficiency of the TCPC is lower than the FLS980. This is because of the slow transmission speed of the UART, the low repetition of the supercontinuum source and a higher DCR of the SiPM. In future work, we plan to adopt the USB 2.0 protocol with a maximum speed of 480 Mbps, increase the repetition rate of the excitation light, and reduce the DCR of the SiPM to achieve a higher data acquisition efficiency.

Conclusion
A TCPC module based on PSA and DTW with SiPMs is evaluated for the time-correlated photon counting technique and fluorescence analysis. Owing to the excellent photon counting performance and precise time correlated features of EQR SiPMs, a low LOD of 16 pg/ml and a large LDR of approximately 5 orders of magnitude are verified for measuring fluorescence intensity. Additionally, the FLT is determined to be 3.758 ns, which was confirmed by a commercial instrument. The TCPC module is proved to be standalone, portable, cost-effective and highly sensitive, and it is compatible with the TCSPC technology and, thus, can be used in time of flight (ToF) applications, including FLT measurements. The TCPC module can be used in point-of-care testing (POCT), such as polymerase chain reaction (PCR), microfluidics, quantitative interpretation of test strips, light detection and ranging (LiDAR), radiation monitoring, and so on.

Funding
National Natural Science Foundation of China (11875089, 61534005).

Disclosures
The authors declare no conflicts of interest.