Skip to content

Fluorescence lifetime imaging with cost-efficient time-correlated photon counter

FieldValue
JournalApplied Physics Letters
Published2026-02-23
AuthorsNatakorn Sapermsap, Jakgree Preegarn, Sorawis Sangtawesin
DOI10.1063/5.0313691

A TDC7200-EVM timing module purchased for under $250 sustained 10,000 counts per second with timing uncertainty below 170 ps FWHM. Integrated into a confocal microscope, it produced fluorescence lifetime maps of reference beads and single nitrogen-vacancy centers in diamond.

The authors built the module around a Texas Instruments TDC7200-EVM board with an MSP430 microcontroller. The controller is programmable through a Python API. The chip has a nominal timing resolution of 55 ps within a 500 ns measurement range.

Characterization used a 500-MHz LVTTL generator to send two 20 ns pulses to the start and stop channels with a 50 ns delay. Event detection was linear at low repetition rates, then saturated at 10 kHz. Swabian TimeTagger and PicoQuant PicoHarp300 systems gave the expected linear response over the same range.

Jitter was defined as the standard deviation of collected times. Measurements used a 10 kHz repetition rate, with example histograms at 30 ns and 480 ns. The reported jitter fell to 25 ps at short delays and rose to about 170 ps near a 500 ns delay. The authors describe these values as smaller than the approximately 350 ps jitter of a typical avalanche photodetector. The reported system uncertainty remained below 170 ps FWHM.

Commercial time-domain timing modules often cost over $10,000. This lowers the timing hardware cost.

A 532 nm pulsed diode laser supplied excitation with a nominal pulse width below 80 ps FWHM. Detection covered NV-center fluorescence from 600 to 800 nm, collected through a 100× objective with NA 0.9.

A two-axis galvanometer scanned the sample, and a 4f relay controlled the beam path. The full system could scan 200 by 200 µm², while selected measurements used smaller fields.

A single-mode fiber with an effective core size of approximately 5 µm replaced the confocal pinhole. That size corresponded to 0.9 Airy units. Photon arrivals came from an SPCM, whose output TTL marked each detection.

The TDC operated in reverse mode. An SPCM pulse triggered the start, and the next laser synchronization signal triggered the stop. Reversing the resulting histogram recovered the fluorescence decay.

The fit-free center-of-mass method estimates lifetime from the photon distribution in each pixel. A lookup table maps the center-of-mass ratio to a lifetime ratio within the measurement window. Background photons were sampled from bins before the initial fluorescence response.

The measurement window must exclude the instrument response and cover several lifetimes. Truncation otherwise biases results toward shorter lifetimes, although iterative correction can address this bias.

The instrument response function was measured without the 600 nm long-pass filter. Neutral-density attenuation kept its count rate below 10 kHz while preserving the laser settings. The corrected lifetime was the sample center-of-mass result minus the instrument-response center-of-mass result.

Traditional single-exponential fitting is slower and requires many photons for a reliable result. The center-of-mass calculation avoids that fitting step at each pixel.

Equal proportions of 6.5 µm beads with nominal lifetimes of 1.7 ns and 5.5 ns were dried on glass. Excitation ran at 2.5 MHz, and attenuation kept the count rate below 10 kHz. The scan covered 50 by 50 µm² with 100 by 100 points and 1 s per pixel.

Each repetition used a 400 ns measurement window. The authors rebinned the histogram into 1,136 bins with 440 ps steps, equal to eight times the TDC resolution. The background window used 50 bins, or 22 ns. The post-peak measurement window used 75 bins, or 33 ns. Pixels with fewer than 200 total events were discarded.

The lifetime image showed more contrast between bead types than the intensity image, especially near edges where intensity decreased. Separate 60 s traces at four points were fitted with a single exponential for comparison.

Bead pointsNominal lifetimePoint-trace fit
A and C1.7 ns1.82 ± 0.03 ns
B and D5.5 ns4.8 ± 0.1 ns

The center-of-mass map agreed with these fitted values and previous reports. The measured 4.8 ns value differed from the nominal 5.5 ns bead value.

The sample was high-purity electronic-grade single-crystal diamond implanted with 5 keV 15N and annealed to form a single NV layer near the surface. The NV centers were approximately 5 nm deep. These shallow centers were chosen because their charge state can be unstable.

The FLIM scan covered 7 by 7 µm² with 70 by 70 pixels. Laser repetition was 10 MHz, and each pixel was acquired for 1 s. Approximately 40 NV centers were identified from background.

A 100 ns repetition window was divided into 227 bins of 440 ps. The background window contained 50 bins, or 22 ns. The post-peak measurement window contained 130 bins, or 57 ns. Pixels with fewer than 600 total events were discarded.

The measured lifetime distribution had a mean of 15.5 ns and a standard deviation of 3.6 ns. Several centers with similar fluorescence intensities had lifetimes that deviated from the roughly 15 ns average. The image therefore exposed heterogeneity that the intensity map did not show.

Several decay traces did not fit a single exponential well. The authors suggest that background fluorescence or a different emitter charge state may add components. The center-of-mass method distinguishes different average lifetimes without resolving those individual components.

The converter saturates at 10 kHz, which limits throughput relative to commercial TCSPC systems. The system remains useful when fluorescence emission is low enough to stay below that ceiling.

Center-of-mass analysis returns one average lifetime per pixel. It cannot separate overlapping decay components, and some NV-center traces require a richer model than a single exponential.

Fine sampling requirements and stage drift led the authors to use a 7 by 7 µm² field for single-center imaging. The result was a local characterization of the sampled NV population.

The authors identify single quantum emitters and experiments that must minimize photobleaching as suitable cases. They also propose the timing module and calculation method as a standalone microscope upgrade. Lifetime imaging is proposed as a rapid way to distinguish NV-center charge states.

For diamond device engineering, the data show that intensity alone cannot identify emitters with different average lifetimes. Lifetime mapping identifies heterogeneity in a shallow NV population without assigning its underlying decay components.

The paper studies single-emitter FLIM of NV centers in single-crystal diamond under 532 nm excitation. The host material family is MPCVD-grown single-crystal and polycrystalline diamond from 6CCVD, as listed in the diamond materials catalog. The engineering support page covers custom fabrication support, while 5 keV 15N implantation and annealing remain separate sample preparation steps described in the paper.

View Original Abstract

Fluorescence lifetime measurement is a versatile tool for studying the dynamics of biological, chemical, and quantum systems. However, commercially available time-correlated single-photon counting systems are typically very costly and are limited in terms of flexibility. Here, we present the use of a cost-efficient time-to-digital converter chip as a time-correlated single-photon counting module and the integration of the module into a confocal microscopy system to perform fluorescence lifetime imaging microscopy (FLIM). The system can withstand a sustained count rate of 10 000 counts per second with a timing uncertainty less than 170 ps full width at half maximum. FLIM measurement is obtained by scanning the confocal microscope over the sample and acquiring photon arrival times at each position, where the fluorescence lifetime is then computed using the fit-free center of mass method. The combined system is verified by imaging of fluorescence microbeads with known lifetimes, and we demonstrate the use of the system for imaging single-photon emitters with the nitrogen-vacancy defects in diamond. This timing module and calculation method can be used as a standalone upgrade kit for an existing microscope to enable FLIM measurement in biological studies and medical diagnosis applications.

  1. 2018 - Fluorescence spectroscopy and its applications: A review [Crossref]
  2. 2015 - High-throughput fluorescence imaging approaches for drug discovery using in vitro and in vivo three-dimensional models [Crossref]
  3. 2022 - Recent progress of fluorescent materials for fingermarks detection in forensic science and anti-counterfeiting [Crossref]
  4. 2006 - Single defect centres in diamond: A review [Crossref]
  5. 2020 - Coherent control and high-fidelity readout of chromium ions in commercial silicon carbide [Crossref]
  6. 2019 - High-fidelity spin and optical control of single silicon-vacancy centres in silicon carbide [Crossref]
  7. 2015 - Deconvolution of fluorescence lifetime imaging microscopy by a library of exponentials [Crossref]
  8. 2011 - High-speed confocal fluorescence lifetime imaging microscopy (FLIM) with the analog mean delay (AMD) method [Crossref]
  9. 2010 - A cost-efficient frequency domain fluorescence lifetime measurement system [Crossref]
  10. 2023 - Direct frequency domain fluorescence lifetime imaging using simultaneous ultraviolet and visible excitation [Crossref]