Skip to main content

3.1.1: Time-correlated single photon counting

This electronic method is useful, when excitation is performed using high repetition rate lasers (e.g. Ti:Sapphire oscillator). After excitation of the molecule by a laser pulse, it will sit in the excited state for some time, and eventually, after time Dt, it will emit a photon. If we find a way of measuring Dt multiple times (i.e. exciting the molecule with a large number of laser pulses), the obtained waiting time values will be distributed according the probability of the radiation at particular time. This insight is employed in time-correlated single photon counting (TCSPC) experiment, the principle of which is illustrated in fig. 13. High repetition rate laser is firing ultrashort pulses into the sample. Part of the laser light is split off by a beamsplitter and directed to a fast photodiode, creating an electrical pulse. This pulse further goes into a discriminator that converts this pulse of unknown temporal shape into a digital ‘start’ signal. Start signal is used to start a time-to-voltage, or time-to-amplitude converter (TAC). TAC is a device that produces an analogue voltage signal proportional to the delay between two digital electronic pulses (‘start’ and ‘stop’). It is, in essence, just a capacitor charged by a constant current, or ramp generator. Fluorescence photon emitted by the sample is captured by a sensitive photomultiplier (PMT), the signal of which is also processed by the discriminator and directed to the TAC as a ‘stop’ signal. As a result, TAC ends up holding a voltage, proportional to the duration between ‘start’ and ‘stop’ photon, or, in other words, the period of time between excitation and emission. This voltage is stored in a multi-channel analyzer – a device consisting of a series of counters, one of each is incremented every time a voltage signal is received. The number of counter incremented corresponds to the observed voltage: e.g. if the voltage is between 0 and 0.1 V, counter No.1 is incremented, if it is between 0.1 and 0.2V – counter No.2 is incremented, and so forth. (Technically, this can be realized by converting the analog voltage into an integer number using an ADC. The obtained number is used as an address of RAM, the number at which is incremented).

            This way, a fluorescence decay histogram is recorded: the counters corresponding to often observed delay times between the excitation and emission accumulate highest numbers (they are incremented very often), whereas the counters corresponding to seldom occurring delays see just a few photons.


Fig. 1) The principle of time-correlated single-photon counting: laser pulse is split into two, one of which starts the time-to-voltage converter, whilst the other is used to excite the sample. The converter starts to accumulate the signal and proceeds until it is stopped by the signal of fluorescence photon emitted by the sample (the photon is recorded by a photon-counting photomultiplier). The voltage corresponding to the duration between the photodiode (start) and photomultiplier (stop) pulses is recorded by the multichannel analyzer. After repeating experiment multiple times, a histogram representing the fluorescence decay is measured.


TCSPC is an electronic method, and its main disadvantage is a limited time resolution. Note that it is not, in fact, limited by the response time of the detectors (photodiode, PMT): the discriminators do not register the response itself, but rather fire their electronic pulses when the signal of the detector reaches a predefined percentage of the maximum. This way picosecond and nanosecond time resolution is accessible, even if the response of the detector to a light pulse is several microseconds (the only requirement for it being the reproducibility from pulse to pulse). The measurement uncertainty comes from so called ‘jitter’, i.e. differences of signal shape from one pulse to another. In practice, TCSPC allows the time resolution of around 50 ps. Time resolution of TCSPC (and other time-resolved fluorescence experiments) is usually measured by measuring the duration of light pulse elastically scattered by the sample. Scattering is an instantaneous process and its temporal shape is determined only by the time-response of the instrument.

Another disadvantage is the fact that TCSPC is, in essence, a single-wavelength method. At each given time, fluorescence kinetics is only registered at one detection wavelength. To register the entire spectrum, one either has to measure different kinetics sequentially, or buy more detectors and all the electronic components, which is an expensive solution.

However, TCSPC has important advantages: it does not require intense excitation light. In fact, it is bad if the excitation pulse energy is too high. The underlying assumption of the method is that each excitation pulse causes only one fluorescence photon to be detected. TCSPC is also completely insensitive to the stability of the laser pulse energy (even if these energies fluctuate by the factor of two, the method still works). You can actually switch off the laser in the middle of the experiment, wait for half an hour and then continue as if nothing has happened. The measurement is cumulative, i.e. one can count photons until the desired signal-to-noise ratio is reached, and the quality of the data allows answering the question posed by the experimenter.

Compared to the other time-resolved spectroscopic methods, TCSPC is cheap. It became popular after the advent of diode lasers providing laser pulses in the 100 ps range. Such lasers cost only a few percent the price asked for a ‘real’ ultrafast system. TCSPC has also been applied in fluorescence lifetime imaging, where fluorescence recorded by a confocal microscope is also measured with time resolution. Thereby, each pixel of fluorescence image also contains a fluorescence decay information [7].