時間相関単一光子係数法を使ったピコ秒の時間分解能を持つ蛍光寿命測定法について解説します。

関係論文：Y. Nishimura, Fluorescence Lifetime Measurement and Its Applications Using Time-Correlated Single-Photon Counting, Photomedicine and Photobiology, 2024, 45, 16-23.

Yoshinobu Nishimura

8 June 2023

Introduction

Fluorescence measurement is a powerful tool for investigating the properties of excited states and their relaxation kinetics, such as phenomena in antenna pigments in photosynthesis. This chapter describes how to measure fluorescence decay, including the excitation source, detector, and apparatus needed to perform the time-correlated single photon counting (TCSPC) method. The aim is to assist readers who are interested in measuring and understanding fluorescence decay time-resolved spectra.

The comprehensive book by Prof. D.V. O'Connor and Prof. D. Phillips provides a valuable resource covering various aspects and offering useful descriptions to comprehend the entire view of the time-correlated single photon counting method [1]. However, some elements, such as the excitation source and detectors used for measurements, have become somewhat outdated. The following sections will describe recent changes in certain apparatus, as compared to what is described in the book. Nevertheless, the fundamentals of fluorescence measurement remain unchanged.

Instruments

Light sources

In recent years, the popular light source for fluorescence measurements has been a mode-locked Ti:Sa laser system equipped with a pulse selector. However, A flash lamp is also currently available for generating light pulses in the 560-700 nm range, where the Ti:Sa laser may not provide sufficient power and stability. The typical Ti:Sa laser offers a wide excitation wavelength range from 700 to 1000 nm, producing < 100 fs pulse widths. It serves as a stable oscillator, which is essential for successful TCSPC measurements, in comparison to a typical dye-laser system.

The average output power, without pulse picking, can reach 1 W if a 10 W class laser is used to pump the Ti:Sa laser. After pulse picking, the repetition rate decreases to a few MHz, and the average power reduces to 20-30 mW, which is sufficient for TCSPC measurements. This means that additional amplifier systems are not necessary. However, it is important to note that time resolution in TCSPC can be affected by laser fluctuations, such as line noise generated from the power supply and surrounding electronics. Therefore, using a simple configuration for the light source is significant. The repetition rate of the pulse train depends on the lifetime of the target sample and the processing electronics.

The available wavelength region depends on the lasing region of the Ti:Sa laser. In some cases, second harmonic generation (SHG) is used to extend the available wavelength region by producing half the fundamental wavelength. The conversion efficiency of SHG mainly relies on the type of crystal used and the peak power of the laser fundamental. If the Ti:Sa laser provides a 1 W output, third harmonic generation (THG) can also be utilized as an excitation source. However, the conversion efficiency of THG is approximately 10% of the fundamental, which is lower than that of SHG. Consequently, the output power after pulse picking is less than 1 mW. Nevertheless, this power level is sufficient for conducting TCSPC measurements, provided that long-term laser stability can be achieved during the decay measurement. Such stability can be attained in a well air-conditioned room.

If you require a wavelength range from 500 to 700 nm to excite a sample, one alternative to a flash lamp could be an optical parametric oscillator (OPO). The OPO, in combination with a Ti:Sa laser, can generate infrared femtosecond pulses ranging from 1.1 to 2.0 μm. By utilizing nonlinear crystals, the OPO enables second harmonic generation (SHG) and third harmonic generation (THG) to provide excitation wavelengths that cannot be generated by the Ti:Sa laser alone. However, a drawback of the OPO is that laser alignment becomes more challenging compared to the Ti:Sa laser, as the fundamental pulse of the OPO is in the infrared region. This means that an infrared viewer is essential for aligning mirrors.

Recently, laser diodes (LD) have gained popularity as excitation sources due to their ease of handling. Moreover, for TCSPC measurements, it is sufficient to use LD to excite samples and generate single photon events. While LD can produce excitation pulses ranging from 375 nm to 1550 nm, its available wavelengths are discrete, and there is a lack of options in the 500 nm region. Despite the inconvenience of limited wavelengths for sample excitation, LD remains an attractive choice as an excitation source. Examples of fluorescence measurements using LD will be described in a later section.

Photodetectors

Photomultiplier tubes (PMTs) play a crucial role in determining the time resolution in TCSPC, in addition to the pulse width of the excitation source and time jitter in the electronics. The statistical dispersion of the transit time spread (TTS) of photoelectrons can generally impact the width of the instrument response function (IRF) in single photon counting measurements. Conventional PMTs often exhibit low time resolution due to multiple amplification stages in the electron dynodes, which are necessary to amplify photoelectrons generated at the photocathode by incident photons. The fastest time resolution achieved with a conventional PMT, such as the R928, is reported to be less than 300 ps under specific conditions [2]. However, we do not recommend using conventional PMTs for TCSPC, despite their easy handling and cost-effectiveness. Specific PMTs designed for TCSPC applications are now commercially available.

The spectral sensitivity of a PMT depends on the photocathode material, which generates photoelectrons upon irradiation based on the photoelectric effect. Typically, alkali metals with low work functions are used for the photocathode. PMT performance depends on the radiant sensitivities of multialkali (MA), infrared-enhanced multialkali (EMA), and Ag-O-Cs (S-1) as a function of irradiation wavelength. Radiant sensitivity is calculated by dividing the photocurrent from the photocathode by the incident photon energy.

MA has a sensitivity range of 200 to 800 nm and is widely used as the photocathode material in PMTs for spectrofluorometers. Through specific activation, MA can be transformed into EMA, which extends the spectral sensitivity up to 930 nm. For detecting photons beyond 800 nm, the S-1 photocathode is frequently employed, despite its low sensitivity and considerable dark current. Cooling the photocathode using a Peltier device allows for effective reduction of dark current, enabling the detection of a substantial number of photons emitted from a sample even if it exhibits very weak emission. This is because a red-sensitive PMT emits thermionic electrons from the photocathode.

The spectral sensitivity of PMTs is also influenced by the window materials used. PMTs with MgF2 windows can operate down to 110 nm, although the use of MgF2 is challenging due to its deliquescent property. Quartz is a commonly used window material with spectral sensitivity extending to around 160 nm. It's important to note that molecular oxygen absorbs in the UV region, leading to ozone formation. UV and borosilicate glasses are the most popular window materials, and they do not encounter issues when detecting emissions in the visible region.

To achieve single photon operation, it is necessary to adjust the voltage applied to a PMT. When the applied voltage is low, no peak is observed in the pulse height distribution, as shown in Figure 1(a). As the applied voltage increases, the pulse height distribution begins to exhibit a peak, as shown in Figure 1(b), and eventually reaches a distinct peak at the optimal voltage, as seen in Figure 1(c). Optimizing the instrument response function (IRF) allows us to determine the optimal voltage for driving the MCP-PMT. It's important to note that high emission intensity can damage the photocathode of the MCP-PMT, leading to a decrease in sensitivity and unusual IRF. It is generally advisable to set the voltage below 3 kV to achieve optimal output, although the exact value may vary depending on individual cases.

Figure 1. Dependence of pulse height distribution on the voltage applied to the MCP-PMT.

Electronic devices

TCSPC measurements involve the use of multiple electronic devices to process signals from a PMT, which detects emissions from fluorescent molecules. Figure 4 depicts a block diagram of a TCSPC apparatus, consisting of a photomultiplier (PMT), a PIN photodiode (PINPD), a preamplifier (AMP), a constant fraction discriminator (CFD), a time-to-amplitude converter (TAC), an analog-to-digital converter (ADC), and a multichannel analyzer (MCA). The AMP is necessary to amplify the output of the PMT, as its amplitude may be insufficient to drive the CFD.

Figure 2. Block diagram of TCSPC measurements: PMT (photomultiplier tube), PINPD (PIN photodiode), AMP (preamplifier), CFD (constant fraction discriminator), TAC (time-to-amplitude converter), ADC (analog-to-digital converter), and MCA (multichannel analyzer).

Since there are various combinations of PMT and AMP, it is important to place an appropriate attenuator (up to a few GHz) between the PMT and AMP to avoid saturating the output of the AMP. To determine the suitable attenuator, it is convenient to have attenuators with 6 dB, 14 dB, and 20 dB settings, which correspond to multiplier factors of 1/2, 1/5, and 1/10, respectively. Trying different attenuators allows us to determine the appropriate one by examining the instrument response function (IRF). If the shape of the IRF is distorted, it is recommended to try another attenuator.

The output of the AMP consists primarily of emission signals and dark noises. Since the emission signals also exhibit a broad distribution of pulse heights, it is necessary to distinguish the desired signals from the AMP output. To achieve this, the discriminator level should be set high enough so that the counting rate of single photons decreases to 2/3 of the maximum count rate obtained at the minimum discriminator level. The excitation intensity should be adjusted so that the maximum count rate is less than 1/100 of the repetition rate of the pulsed laser to avoid distortions caused by pulse pile-up.

If the dark count rate, resulting from the dark current of the PMT, is not negligible (e.g., several hundred counts per second), the maximum count rate needs to take into account the dark count rate by subtracting it from the measured maximum count rate.

To optimize the IRF and minimize the full-width at half-maximum (FWHM), there are several tuning points within each apparatus. One such point is the zero-cross setting in the CFD. In general, the zero-cross level is highly sensitive to the FWHM of the IRF since it determines the timing of the output pulse from the CFD.

A delay line is utilized to introduce a signal delay between the CFD output and the TAC input. It is commercially available as a NIM module that comes pre-calibrated. The delay accuracy is calibrated within a range of ±100 ps, making it suitable for calibrating the time resolution of the MCA when combined with the TAC in a manufactured system.

The TAC plays a crucial role in TCSPC as it determines the time difference between the start and stop signals from the CFD output by utilizing capacitor charging. The TAC features input connectors that accept start and stop signals along with some control lines. When the START pulse is received from the corresponding CFD output, the capacitor begins to charge, and it stops charging upon receiving the STOP pulse. The time difference between the START and STOP signals is directly proportional to the voltage across the charged capacitor. However, in some cases, the linearity of this relationship is not guaranteed, and it should be checked before measuring the FWHM of the IRF. This helps identify the available channels in the MCA. To prepare for a subsequent START signal, the capacitor needs to discharge, resulting in a "dead time" of approximately 100 ns.

Commonly, the START and STOP signals are typically associated with the outputs of the PINPD and AMP, respectively, in what is referred to as the "Normal setting." The probability of detecting a PINPD signal resulting from the repetition of a laser pulse is lower compared to that of the AMP. In the "Normal setting," the AMP serves as the START signal and the PINPD as the STOP signal, which allows the TAC to start charging. However, there is an alternative configuration known as the "Reverse setting," where the AMP becomes the STOP signal and the PINPD becomes the START signal. This setting is useful to avoid time loss caused by the "dead time" of the system.

It's important to note that the "Reverse setting" may not be applicable if the time constant of the emission decay is longer than the laser repetition rate. This is due to the potential overlap of prolonged decay components from the excitation caused by the previous laser pulse. Therefore, accurately determining the repetition rate of the laser pulse using a pulse picker is crucial for achieving efficient measurement of fluorescence decay.

In certain applications, the TAC can be controlled using an additional INHIBIT signal. This feature allows for the measurement of differential decay in the presence and absence of an applied electric field, as described in [3].

Until high-performance personal computers (PCs) became popular, the Multichannel Analyzer (MCA) was commonly used as a standalone device. However, in recent years, MCAs can be implemented using PC extension boards or external boxes that run real-time programs. This advancement allows for convenient control of the MCA, along with other peripheral devices such as motor drivers, photon counters, and sample changers. The integration of these components into a PC-based system provides flexibility and ease of operation.

Measurement

Apparatus Placement

Figure 3 depicts the block diagram of fluorescence lifetime measurement using the time-correlated single photon counting method. A polarizer is employed to set the polarization angle of the excitation beam to 35.25 degrees in order to avoid photoselection of emission. The beam is then focused onto the sample cuvette using a convex lens. The emitted light from the sample is monitored at a right angle to the excitation beam to prevent detection of the excitation source. A pair of convex lenses is utilized to collect the emission and match the F value of the monochromator. A depolarizer is necessary to eliminate the polarization property of the monochromator. The desired monitor wavelength is selected using the monochromator and detected using a PMT. The output signal from the PMT is amplified by an amplifier (AMP) and processed by a constant fraction discriminator (CFD) to discriminate the emission signals from the dark noise. The START signal generated by the CFD is fed into the input of the time-to-amplitude converter (TAC). If a laser diode (LD) is employed as the laser source, the STOP signal in the "Reverse setting" is generated by the LD driver. In the case of a conventional pulse laser such as a Ti:Sa laser, the STOP signal is generated by monitoring a portion of the pulse split from the excitation pulse using a low reflection mirror. If second harmonic generation (SHG) or third harmonic generation (THG) is used to excite the sample, the fundamental pulse is suitable as the STOP signal, which should be processed by the CFD.

Figure 3. Typical setup for measuring fluorescence decay, including a laser diode and optics for collecting emission from a sample.

Measurement Conditions

When starting fluorescence decay measurements, please take note of the following points:

1. Ensure that the laser pulse remains stable throughout the decay measurement. In some cases, it may be necessary to wait a few hours to achieve a stable laser pulse.

2. Check if the absorbance at the excitation wavelength is less than 0.3. This is important to ensure homogeneous excitation of the sample in a 1 cm x 1 cm cuvette.

3. Verify the polarization of the excitation pulse. If you plan to measure anisotropic decay, make sure to use a polarizer at both the excitation and emission sides.

4. Check if the photon counting rate from the sample, relative to the repetition rate of the excitation pulse, is less than 0.01. If it exceeds 0.01, the obtained decay curve may be distorted due to pulse pile-up effects caused by multiphoton events.

Measurement of Fluorescence Decay

If you have checked the points mentioned above, the following steps can be taken:

1. Begin by measuring Rayleigh scattering at the excitation wavelength using a scattering medium, such as a sonicated phospholipid vesicle solution, instead of the sample solution. This scattering decay can serve as the Instrument Response Function (IRF), which will be used for analyzing the fluorescence decay curve through deconvolution.

2. If the peak channel of the IRF fluctuates on the multichannel analyzer (MCA), it is important to check the stability of the laser. In some cases, this fluctuation may be caused by airflow from an air conditioner. To mitigate this, consider covering both the laser and detection instruments to block the airflow.

3. Use a Delay Generator to adjust the position of the decay signal on the MCA.

4. Finally, process the fluorescence decay curve using a personal computer. For measuring time-resolved spectra, control the monochromator using a stepping motor driven by the personal computer. Accumulate photon signals at specific wavelengths for a defined period of time, and then reconstruct the time-resolved spectra on the computer.

By following these steps, you can effectively measure fluorescence decay and obtain time-resolved spectra.

Lifetime analysis

If you do not have a PC program for analyzing fluorescence decay or generating time-resolved spectra, there are commercially available software packages that you can consider. If you lack experience in measuring fluorescence decay, it is recommended to explore integrated analysis software packages such as those offered by PicoQuant GmbH or HORIBA JOBIN YVON Inc.. These packages can provide comprehensive solutions for your analysis needs.

Operation test

Once you have assembled the components for the measurement system, it is important to verify the accuracy of the measured fluorescence decay. One common approach is to compare the obtained lifetime with previously reported values to ensure consistency. Figure 6 depicts the fluorescence decay of cryptocyanine in ethanol, which was measured using the SPC630 system from Becker & Hickl GmbH and a laser diode from PicoQuant GmbH. The resulting lifetime was determined to be 75 ps, in agreement with the reported value of 75 ps [4]. Furthermore, this observation highlights that a minimum of 2000 peak counts is necessary for reliable lifetime analysis. If the peak counts are insufficient due to a very low quantum yield, caution must be exercised when interpreting the calculated lifetime.

Figure 4. The fluorescence decay of cryptocyanine in ethanol was monitored at 728 nm with excitation at 410 nm. In the plot, small dots represent the instrumental response function (IRF), while large dots represent the fluorescence decay. The solid line represents the simulated curve obtained through a least-squares fit, which yields a lifetime of 75 ps. The upper plot shows the weighted residuals calculated from the fitting data.

References

[1] O’Connor, D.V. and Phillips, D. 1984, Time-Correlated Single Photon Counting, Academic Press, London.

[2] Kinoshita, S. and Kushida T. 1983, Rev. Sci. Instrum., 53, 469.

[3] Nishimura, Y., Yamazaki, I., Yamamoto, M. and Ohta, N. 1999, Chem. Phys. Lett., 307, 8.

[4] Sundström, V. and Gillbro, T. 1981, Chem. Phys., 61, 257.