Time-Correlated Single Photon Counting

TCSPC, photon counting, time-correlated single photon counting, detection of individual photons, single-photon detectors (SPD), photosensors

Time-correlated single photon counting (TCSPC) is a widely used technique for the time-resolved measurement of fluorescence. The particularly high time resolution that is achieved today is a special feature of this technology. In other areas, too, this method achieves the technically feasible state of the art - e.g. individual photons are recognized. This measurement method is used as an essential part of imaging techniques such as FLIM (fluorescence lifetime microscopy). Time-Correlated Single Photon Counting is a particularly non-invasive measurement technique because the light pulses required for sample excitation have low energy. The aforementioned single-photon detection also contributes here.

In detail, TCSPC is based on the detection of the arrival times of the photons on a detector. The time of emission of the laser pulse is usually taken as the reference signal, which thus reflects the time of excitation of the sample. The sample is excited with the laser pulse at a high repetition rate. The high repetition rate is of great advantage in order to obtain a statistically significant course of the de-excitation behavior of the sample (in the shortest possible time) from the single-photon measurement. Fluctuations in the source intensity and the resulting measurement errors are also compensated for by the enormous frequency of measurements. To measure the photon emission times, TDC systems are used on the one hand and fast ADCs in the Gsps/s range on the other.

Time-Correlated Single Photon Counting
Analysis of decay data making use of TCSPC.

In the time-correlated single photon counting, individual photons are recorded which are emitted after a periodic excitation of an examination object. This is used to determine the time course of the de-excitation behavior of the sample. In order to obtain the entire course, many such single photon recordings are carried out and viewed over a certain period of time. In this way, changes in light intensity that occur very quickly can also be measured. The aim of TCSPC is to investigate photon emission processes that typically take place in the picosecond range.


The main application is in the field of fluorescence lifetime measurements (photoluminescence decay measurements), in which the emission times of the photons are determined relative to the impact of excitation light pulses. Today, TCSPC technology surpasses all popular techniques in terms of sensitivity, dynamic range, and precision.


The already mentioned time delay between excitation and photon emission (here called emission time) is measured with an extremely high time resolution. The excitation is carried out by a short laser pulse. The light intensity is chosen so that many of these pulses are emitted without excitation - as a rule, this is more than 95% of all pulses. Accordingly, the pulse repetition rate is much higher than the photon exchange rate. This is to ensure that, if possible, only one occurring photon was generated within the measured signal period, and that photons that originate from a previous pulse are not also recorded. In order to obtain statistically significant results, the individual measurements are repeated several million times.


The individual measurements are entered in a histogram which shows how often each individual emission time occurs. If necessary, this measuring principle can also be used for imaging methods, such as FLIM.

These are the advantages of TCSPC measurements:

  • The light pulses required for sample excitation have a low intensity so that the sample is damaged as little as possible and nonlinear effects are avoided.
  • TCSPC has a high time resolution.
  • TCSPC offers a wide dynamic range.
  • Fluctuations in the pulse amplitude of the excitation light source only have a minor effect on the results of a TCSPC measurement.
  • The demands on the detectors are lower than for analog measurements since TCSPC is carried out in the so-called "counting range" of the detectors.
  • In contrast to streak cameras, which also offer a high time resolution and fast acquisition of the entire spectrum, TCSPC systems are significantly cheaper and only require a femtosecond laser for excitation in a few cases.
  • TCSPC can be combined with a scanning technique and used for fluorescence lifetime imaging (FLIM).

With many detectors, measurement errors caused by different pulse amplitudes are avoided by a constant fraction discriminator. This does not evaluate the temporal position using a fixed threshold value, but rather determines the exact point in time at which the pulse amplitude reaches a certain proportion of the pulse height. Thus, pulses of different amplitudes still start (or stop) the "clock" at the same time. Pulse height fluctuations either due to light source instabilities or due to the intrinsic pulse height distribution of photomultipliers are therefore largely irrelevant.

Applications for TCSPC:

TCSPC is mainly used in single-molecule detection and spectroscopy as well as in fluorescence correlation spectroscopy. In addition to the investigation of fluorescence lifetimes, time-correlating single photon counting is used for fiber optic research in order to investigate the time course of pulsed or modulated light (see also optical time-domain reflectometry, quantum research and quantum key distribution).

Structure of a TCSPC decay measurement

Typical structure of a TCSPC measurement

Laser sources, sometimes also femtosecond lasers, serve as the excitation source. With a high repetition rate (usually in the range of 1 - 100 MHz), a laser driver sends a start signal to a computer at the same time as the pump pulse for the laser source, which triggers the time measurement. When choosing the laser sources, the excitation energies of the materials to be examined and the expected fluorescence lifetime play a decisive role.

The laser pulses themselves are usually shorter than 0.1 ns and are directed onto the sample via a neutral density filter. The filter suppresses stray light that could be falsely detected by the detector and attenuates the excitation pulse accordingly. If a photon is emitted from the sample within a pulse cycle and hits the detector, this triggers the stop signal for this time measurement.
In front of the detector, there is a so-called cutoff filter, i.e. a “long pass filter”, which limits the measurement to the relevant spectral range. In some measurement setups, the light emitted by the sample molecule is also passed through a monochromator in order to carry out a wavelength-selective measurement.

The synchronization of the measurement signals and their digitization is implemented with the help of fast, high-precision data acquisition technology, which delivers histograms for the recorded time data as a result - this is the core competence of cronologic.



The following photon detectors are used for TCSPC measurements:

  • PMT: (Photomultiplier Tube) - the classic way, cheap, high gain, not good for fast detection.
  • MCP: Microchannel plates are very popular because they are more sensitive than PMTs and have faster response times.
  • APD / SPADs (Avalanche Photodiodes): The semiconductor equivalent of the PMT offers very small FWHM (30-400ps) and consequently enables the detection of very fast photoluminescence decays.
  • Hybrid PMTs combine the classic PMT with an avalanche diode, offering a low propagation time spread and advantages in terms of counting efficiency. These detectors are practically free from post pulses. Much of the gain is achieved in a single step. Therefore they deliver single-photon pulses with a narrow amplitude distribution. With such detectors, it is much easier to differentiate between accumulated photons. The small-amplitude fluctuation eliminates the influence of the CFD circuit on the time jitter.

What should you watch out for when evaluating TCSPC measurements?

Special attention should be paid to possible measurement errors due to the effect of the so-called "pulse pile-up" during a TCSPC measurement:
If two or more photons are emitted in a very short time interval, it can happen, for example, due to dead times during data acquisition, that individual photons are not recorded. The resulting measurement error can even systematically falsify the histogram obtained since the “photon arriving later” was always not recorded. In principle, the problem of the “pile-up” of photons can be countered by keeping the excitation intensity (and thus also the percentage of photons that come from the sample) very small. So it happens that TCSPC measuring devices are designed in such a way that only a fraction, often only one, of 50 excitation pulses also generates a response photon.

The individual counts are recorded in time bins with the help of their timestamps generated by the TDC and displayed accordingly.

How exactly is data recorded during the TCSPC measurement?

In order to register the arrival time of the fluorescence photons relative to the time of laser excitation, electronics with the highest possible resolution are used. The time resolution of TCSPC is primarily determined by the time resolution of the detector. With the classic TCSPC devices, the data was recorded with the help of a TAC (Time to Amplitude Converter) and a transient recorder (Analog to Digital Converter, ADC).

The emitted light signal and the reference light signal are processed by a constant fraction discriminator (CFD). After running through the CFD, the reference pulse activates the TAC. The TAC internally generates a voltage ramp on a capacitor, whereby the voltage rise is stopped as soon as the sample photon is detected. Thus, the voltage present in the TAC is greater, the later the photon is emitted. The voltage reached at the time of the stop signal is then evaluated by an ADC and converted into a digital time stamp. In the reverse TAC mode, the roles of the excitation pulse and the detected photon are reversed. In this case, the excitation pulse stops the voltage increase in the TAC.
The timestamp can be processed further in a multi-channel analyzer (MCA) in order to obtain a data output. This classic measurement method has a comparatively long dead time and only supplies the waveform of the optical signal.

Newer TCSPC systems, therefore, use the fast and efficient data acquisition of modern TDCs (Time to Digital Converter/time interval analyzer), which convert the arrival time of the measurement signals directly into digital time stamps. Due to the short dead times of modern TDCs, it is sometimes possible to record several photons per cycle. If necessary, additional information can be recorded for each photon, e.g. the wavelength, the polarization, its position within an image area, the time from the stimulation of the sample, the excitation wavelength, etc. This results in many other options for displaying the measurement data obtained.

The limits of TCSPC measurements

With a TCSPC measurement, the resolution is limited by various factors, the sum of which is referred to as the IRF (Instrument Response Function):

  • The finite width of the laser pulse (largest factor)
  • finite rise/fall in electrical signals
  • The time resolution of the detectors
  • Jitter in electronics

The IRF time sum is usually determined directly during the measurement by directing the laser (for the purpose of creating a prompt) at a non-emitting scattering solution for each decay measurement of the sample. For capturing the prompt, the slot width of the monochromator used must be retained. In this way, the previously obtained measurement data on the fluorescence lifetime can be related to the inaccuracies in relation to the resolution. If the recorded time data for the fluorescence lifetime is in a different order of magnitude than the IRF time sum, then one relies more heavily on the data obtained than if they are in a similar range. When evaluating the decay data, in addition to considering the course of the decay itself, but also its standard deviation is considered in order to ensure that these are distributed as randomly as possible over the entire measurement.