High Performance, Low Cost…

Fluorescence Lifetime Measurements

When excited by pulses of light at the appropriate wavelength, certain molecules in solution will fluoresce. The release of absorbed energy is not instantaneous but occurs over a certain time period. The fluorescence decay curve of a sample is a measure of the intensity of light released vs the time after the initial excitation. For many molecules this decay process is exponential in nature, and the fluorescence lifetime is defined as the time taken for the fluorescence intensity to fall to 1/e of its initial intensity. The fluorescence also occurs at the same and at slightly longer wavelengths than the exciting light, due to various photophysical processes taking place in the molecules that we will not go into here; for more information search on the term “Jablonski diagram”.

Fluorescence lifetimes can be extremely short. The molecule quinine found in tonic water has a fluorescence lifetime of only 19.4 nsec. Quinine absorbs near-UV light at around 370 nm and fluoresces in the blue region of the spectrum, which one can easily observe by placing some tonic water in a glass and shining in the light from a 370 nm LED.

To actually measure a fluorescence lifetime as short as 20 ns, the exciting light source must be pulsed and the duration of these pulses needs to be much shorter than 20 ns and ideally, just a few ns. The instrument we have developed here employs a dedicated nanosecond LED pulse generator and a detection system that uses a technique know as photon counting to measure these very short fluorescence lifetimes.

The instrument has an ACAM TDC-GP2 time-to-digital converter (TDC) chip interfaced to a P8X32A-Q44 Propeller™ chip. The Propeller chip provides a trigger, or start pulse, to generate a nanosecond duration LED flash, and on each flash a Hamamatsu R7155 photomultiplier tube detects the arrival of a fluorescence photon and generates a stop pulse. The TDC accurately measures the time interval between start and stop pulse pairs and returns a corresponding digital code that the Propeller uses as an index to increment a point in a buffer that represents a histogram of start-stop times. By repeating this process at a data rate of around 105 LED flashes per second, a fluorescence decay curve is built up in the Propeller’s buffer memory.

After a sufficient number of LED flashes, the fluorescence decay curve is transferred back to a host PC via the on-board USB interface (FT245RL). Exactly the same method works whether we are measuring the pulse profile of an LED or a fluorescence decay curve.

When using this photon counting technique care must be taken to avoid the phenomenon know as “pulse pile-up”. The system has an inherent dead time after the first photon is detected; if additional photon events generate detector pulses in the one LED flash they will not be recorded by the system and the measured decay will be biased to shorter times than the true decay.

Fluorescence lifetime measurements can also be useful analytically. For example the fluorescence of many ruthenium (Ru) complexes depends on the concentration of dissolved oxygen [O2] in solution. In an aqueous solution of the species [Ru(bpy)3]2+ that has been purged with N2 the fluorescence lifetime is ~ 650 ns, but this shortens dramatically in an unpurged solution to just a few hundred ns and the fluorescence also weakens considerably in intensity. This is due to a process known as fluorescence quenching that is described by the Stern-Volmer equation.

Propeller Time-To-Digital Converter

The PCB shown at right has an ACAM TDC-GP2 time-to-digital converter chip interfaced to a P8X32A-Q44 Propeller™ chip. The TDC-GP2 measures the time interval between a start-stop pulse pair with a timing resolution of ~50 ps.

During operation, a start pulse generated on-board triggers a nanosecond duration LED flash off-board (see below for details). A fluorescence photon detected by a pulse counting PMT then results in a stop pulse.

On receipt of each start-stop pulse pair, the resulting measured time interval is used to index a single memory location in a 15000 point on-board memory buffer - implementing a time interval analyser. Each channel in this buffer represents a uniques start-stop delay in the range 0-1450 ns. A counter in that memory location is incremented so that after a large number of start-stop events have been recorded, the pulse profile of the LED is built up.

Propeller TDC-GP2 Connectors

Two mini-DIN6 connectors on the TDC-GP2 PCB give access to various signals as follows:

J3 mini-DIN6

Pin 1 - External start
The on-board time-to-digital converter measures the time interval between a start pulse and a stop pulse. An external start pulse can be applied at this pin. An on-board start pulse that is used to trigger an external nanosecond LED pulse generator is generated on pin P18 of the P8X32A-Q44. The signal that is routed to the start input (pin 31) of the TDC-GP2 chip can come directly from P18 or it can be sourced from the external start input by installing jumper JP1 – when this jumper is in the 1-2 position the input comes from P18; in the 2-3 position the input is sourced from the external start.

Pin 2 - External stop
The on-board time-to-digital converter will measure the time interval between a start pulse and a stop pulse. An external stop pulse can be applied at this pin. For nanosecond lifetime measurements the stop pulse will come from the Hamamatsu pulse counting photomultiplier (model # H7155). In this case a 50Ω resistor should be connected between this pin and ground.

To allow testing of the system, pin P19 on the P8X32A-Q44 can also generate a stop pulse having a well defined time delay relative to the start pulse on P18. A jumper placed on JP2 selects the input source to the stop input (pin 27) of the TDC-GP2 chip – in the 1-2 position the source becomes P19, while in the 2-3 position the source derives from the external stop.

Pin 3 - Trigger pulse (from Propeller pin P24)
The Propeller chip can generate a pulse on this pin to trigger an external pulse generator, or an oscilloscope.

Pin 4 - PMT bias control voltage
A 0-1000 mV DAC output is available on this pin. This can be used to set the -HV bias on an analog PMT with an integral -HV supply. In that case the PMT’s output can be connected to an oscilloscope.

Pin 5 - +5V output for a Hamamatsu H7155 pulse counting PMT.
Power provided on this pin is switched on via a solid state relay, just prior to trigger pulses that initiate nanosecond LED pulses. At the end of a measurement sequence the 5V supply is switched off to protect the PMT.

Pin 6 - Ground

J4 mini-DIN6

Pin 1 - Flash trigger
An on-board opto-isolator allows triggering of a camera flash lamp via a hot shoe connector. One of the pins of the hot shoe is connected here, while the other is grounded at pin 2.

Pin 2 - P8X32A-Q44, pin P25 - generic I/O pin
An uncommitted pin on the P8X32A-Q44 chip is wired to this pin for general purpose use.

Pin 3 - P8X32A-Q44 P26: TSL235 light-to-frequency converter input
For measurements of light intensity on the tens of microsecond timescale and longer, the output of a TSL235 light-to-frequency converter can be connected here (if the TSL235 is run from a +5V supply a 1kΩ series resistor must be placed between the TSL235’s output pin and this input pin.

Pin 4 - P8X32A-Q44, pin P27 - generic I/O pin
An uncommitted pin on the P8X32A-Q44 chip is wired to this pin for general purpose use.

Pin 5 - 5V

Pin 6 - Ground

Nanosecond LED Pulse Generation

Various methods for generating extremely short duration pulses from standard LED’s have been described in the scientific literature. A number of designs have appeared that employ very fast avalanche transistor pulse generators - but achieving avalanche breakdown requires high voltages (typically 50-250V) depending on the transistor.

The PCB shown at right is based on the circuit described in the paper “A simple sub-nanosecond ultraviolet light pulse generator with high repetition rate and peak power”, P. H. Binh, V. D. Trong, P. Renucci, and X. Marie, Rev. Sci. Instrum. 84 (083102) 2013.

This generator is based on the use of a step recovery diode, short-circuited transmission line, and current-shaping circuit and has the advantage of operating at low voltages.

3.3V logic level trigger pulses from the photon counting board (i.e. start pulses) are boosted to 7V by an on-board IXSYS IX4427 MOSFET driver that drives the input of the generator described in the cited publication.


This is the vi that controls the time-to-digital converter PCB.

Here, the vi is measuring the pulse profile of a 370 nm LED. The user enters the # of desired pulse pairs and as the vi executes, the cumulative profile is displayed in yellow, while the instantaneous profile from the last set of pulse pairs appears in the green trace at left. 16 million LED flashes were produced in this experiment.

The vi incorporates a binning mode; a binning parameter of 3 means that 15k raw data points have been reduced to 5k points in the final display.

The pile-up % indicates the % of LED flashes that resulted in receipt of a stop pulse from the detector.

Measured LED pulse profiles

The plot at left shows measured LED pulse profiles for blue (435 nm) and near UV (370 nm) LED’s using the nanosecond pulse generator described previously. The shorter wavelength LED produced a narrower pulse - full widths at half maxima (FWHM’s) here are ~2.4 ns and ~4 ns, respectively.

The offset of ~50 nsec in these traces can be ignored.

Additional Capabilities

The Propeller TDC-GP2 board can also be used to measure transient optical signals by connecting a TSL235 light-to-frequency converter. In addition, an external trigger is provided to trigger external events, such as firing a photoflash to generate millisecond duration light pulses. These features allow investigation of things such as

• Measuring the lamp profile of a Xe camera flash
• Determining a human heart rate via oximetry using a red LED.

The results page shows an example of each of these experiments.

In summary, the TDC-GP2 PCB can be used to study optical processes occurring over an extremely broad timescale - varying from a few tens of nanoseconds to many seconds.

XMOS Startkit Photon Counting Shield

A plug-in shield for the XMOS Startkit has also been developed and this is shown in the image opposite. The PCB is functionally similar to the Propeller implementation described above.

The screw terminal headers provide (i) LED/LD trigger pulses, (ii) receive externally generated START/STOP pulses, and (iii) provide switched 5V power under relay control to a Hamamatsu photon counting head.

Jumpers are placed on headers JP1-JP3 to control the various possible operating modes.

The three pin header in the upper left corner of the PCB connects to an FT232R USB-to-serial converter cable, allowing data to be transferred to a host PC, running LabVIEWTM.

A Nanosecond LED Pulse Generator Using an Avalanche Transistor

The PCB shown in the photo below takes input trigger pulses at connector J2 and applies them to the base pin of a 2N2369A transistor (Q3) wired in an avalanche circuit. When avalanche breakdown occurs, the energy stored in a capacitor connected to the 2N2369A’s collector pin is very rapidly dumped through the emitter circuit to ground. As a result, an LED connected in the emitter circuit (with a current limiting series resistor) at J3 generates an ultra-short light pulse.

In order to observe avalanching behaviour with the 2N2369A transistor being used here one requires a bias voltage of ~ 40 - 50V. To generate this voltage the PCB also has a two transistor multivibrator circuit (Q1, Q2) wired to the secondary of an audio output transformer having a 10:1 turns ratio, thereby giving a 10x voltage gain.

A low voltage input from a DC plugpack connected at J1 and an LM317 voltage regulator and potentiometer R2 provide the adjustable HV bias for the avalanche circuit. This bias connects via a 100k charging resistor to the 2N2369A’s collector (which has the energy storage capacitor to ground).

In practice, a continuous train of short trigger pulses is applied at J2 and the bias HV is gradually increased until the LED lights up. Note that if the HV bias is set too low the LED will not light up; if the bias is set too high avalanching will occur spontaneously without the presence of trigger pulses. When the HV bias is correctly set the LED will visibly glow but no output will be observed when the trigger signal is disconnected.

Optical Pulse Profiles vs Collector Capacitance, C

The collector pin of the 2N2369A transistor rises up to the HV bias voltage with a time constant determined by the charging characteristic of the RC circuit. The voltage then drops rapidly back to ground when avalanching occurs. This cycle is repeated in response to each trigger pulse.

It is a worthwhile exercise to investigate how the generated optical pulse profiles depend on the capacitance between the 2N2369A’s collector and ground. Intuitively, one expects that larger capacitance values with produce longer and brighter pulses, while smaller values will produce less bright but narrower pulses.

In the PCB photo above, the capacitors in question are C11 and C12. Small ceramic capacitors typically have a working voltage rating of only 50V and consequently two were wired in series on the PCB to provide a safety margin.

The plot below left shows measured pulse profiles when the net capacitance is varied from 39 pF, to 10 pF and finally 3.3 pF. In these experiments the LED was a 450 nm device and the measurements were performed using the XMOS photon counting shield shown above.

From this data one sees that the optical pulses are slightly asymmetric (having an extended tail) and that using a lower capacitance does indeed result in shorter pulse widths. It is important to note that these runs are not all of the same duration; the LED was visibly much brighter using the larger capacitor values. To observe successful avalanching it was necessary to increase the bias voltage for the smaller capacitance values.

The green trace at lower right shows the result of a second experiment in which a total of ~108 LED flashes were generated. The point spacing in this trace is roughly 120 psec and while the pulse profile roughly approximates a gaussian envelope, some fine structure is evident including a prominent narrow spike in the middle of the pulse (also observed in the first experiment).

Using this LED the FWHM is seen to be approximately 4 ns with a 3.3 pF capacitor to ground at the 2N2369A collector.
The pulse profile shown opposite was measured in the identical setup but now using a Roithner LaserTechnik 370 nm LED. The FWHM here is ~ 1.2 nsec, which represents a 3-fold improvement on the earlier result obtained using the 450 nm LED.

These results demonstrate that there is considerable scope for further investigation in the quest to push down to even narrower pulse widths. Key factors to explore will be LED type, and further optimization of the circuit layout/component values.

Importantly, the timing resolution of the existing pulse counting system has been shown to be quite capable of measuring pulse widths in the sub-nanosecond regime.

TDC7200 XMOS Shield

The TDC7200 is another time-to-digital converter IC that measures the time interval between a start pulse and up to 5 stop pulses.

Intended primarily for ultrasonic sensing applications, the time interval measurement can be made in one of two modes, the first providing a range from 12 ns to 500 ns, and the second from 250 ns to 8 ms. Time resolution is specified as 55 ps.

Here, an XMOS Startkit shield incorporating the TDC7200 is being evaluated for possible use in fluorescence lifetime instrumentation similar to what we have already described above for the ACAM TDC-GP2 chip.

TDC7200 Performance

The figure opposite shows some preliminary results obtained using the TDC7200. Here, the XMOS processor generates a START pulse, followed by a STOP pulse that is controlled by a variable delay, set from a LabVIEWTM front panel.

The delay is entered as a timer tick count, with each tick corresponding to a duration of 10 ns. An excellent regression line is obtained, with the slope of 176.7 indicating that the timing resolution per TDC count is 56.6 ps. This is in excellent agreement with the value specified in the manufacturer’s datasheet.