PicoQuant Support Knowledge Base - What is the IRF and how to measure it in fluorescence spectroscopy?

What is the IRF?

In fluorescence lifetime measurements, the Instrument Response Function (IRF) is a measure of the instrument’s accuracy in determining fluorescence lifetimes. It represents the total response time of the measurement system, which includes all components that contribute to registering a fluorescence excitation–emission process:

the excitation source (typically a pulsed laser, LED or lamp),
detection optics and opto-mechanical components (grating/monochromator, filters, etc.),
opto-electronics (photomultiplier tubes, photodiodes, microchannel plates, single-photon avalanche diodes, superconducting nanowire single-photon detectors, etc.), and
detection electronics (CFD, TDC, ADC, TAC, etc.).

Ideally, if none of these components added any timing uncertainty, the IRF would be infinitely narrow (mathematically, a Dirac delta function). In reality, every component has some timing uncertainty, and the overall IRF is the convolution of all of them.

Why and when is it important to measure the IRF?

The measured TRPL histogram is a convolution of the IRF and the actual fluorescence decay. To obtain the lifetime from the measured histogram, it has to be deconvoluted, or, in practice, fitted by reconvolution. With the IRF known, the actual lifetime can be determined.

Measuring the IRF is essential when the lifetime, or one of the lifetime components of a multi-exponential sample, is ≤ 10 times the IRF width.
For samples with longer lifetimes, a tail fit (without a measured IRF) is generally sufficient. This also applies to phosphorescence and delayed fluorescence measurements.

What determines the IRF width?

In a fluorescence spectrometer, the IRF width depends mainly on four factors:

Pulse width of the laser or LED (τ_p)
Dispersion in the grating of the monochromator (τ_g)
Temporal uncertainty of the detector (τ_d)
Timing uncertainty of the TCSPC electronics (τ_e)

These contributions combine into the IRF. A good approximation of the total IRF width is:

IRF ≈ √( τ_p² + τ_g² + τ_d² + τ_e² )

Consequences:

The IRF is dominated by the slowest component (the one with the largest timing uncertainty). Improving a component that is already much faster than the others brings almost no benefit.
Ideal combinations are therefore components with similar timing uncertainties.
A short IRF means that short lifetimes can be resolved.

How to measure the IRF in a spectrometer

The IRF is recorded by measuring a sample that responds instantaneously, i.e. a scattering sample instead of a fluorescent one.

Option 1: Elastic (Rayleigh) scattering

Place a scattering solution, such as Ludox (colloidal silica), in the sample holder instead of the fluorescent sample.
Set the emission monochromator to the excitation wavelength, and remove any filters that block the excitation light.
Keep all other settings (excitation source and its settings, detector, time resolution) the same as for the sample measurement.
Optimize the count rate to avoid pile-up. The default IRF optimization condition in EasyTau is a count rate of ~0.1 % of the repetition rate.
Record the IRF and use it in the reconvolution fit of the sample decay.

Option 2: Raman scattering

Raman scattering is also an instantaneous process and can therefore be used to measure the IRF. Measure the pure solvent (e.g. water) and set the emission monochromator to the solvent’s Raman band instead of the excitation wavelength. For water, the strong O–H stretching band is shifted by ~3400 cm^-1 from the excitation, e.g. 405 nm excitation gives a Raman band at ~470 nm. Since the Raman band is red-shifted, it lies closer to the emission wavelength of the sample, and emission filters that block the excitation light can often stay in place. Note that the Raman signal is weak, so longer acquisition times are needed, and the sample’s fluorescence must not overlap with the Raman band (measure the pure solvent only). In EasyTau, if the solvent data is included in the 'solvent database', then one can select this automatically from the software, without manually setting the monochromator at the Raman wavelength.

Note on the detection wavelength

When using elastic scattering, the monochromator is set to the excitation wavelength, which differs from the PL wavelength of the sample. Strictly speaking, the change in dispersion changes τ_g and therefore also the IRF shape. This effect may not be negligible if the Stokes shift (PL wavelength − excitation wavelength) is large. In such cases, it is preferable to measure the IRF closer to the emission wavelength, using a quenched dye or Raman scattering instead of Ludox.

For a step-by-step description on how to measure IRF for microscopy applications, see the TCSPC Wiki: How to measure the instrument response function (IRF).