Phase-modulation fluorometry using a frequency-doubled pulsed laser diode light source

Abstract

We used the Hamamatsu model PLP-01 picosecond light pulser as a 413-nm excitation light source for frequency-domain fluorescence measurements. In comparison with sync-pumped/cavity dumped/frequency-doubled dye lasers, the 413-nm PLP-01 shows a longer FWHM (40 ps), a similar pulse repetition rate (up to 10 MHz), much less output power at a fixed wavelength (0.44 mW peak, 220 nW maximum average power), but is less expensive, small-sized, and easy to handle. Using the PLP-01, we have been able to perform fluorescence measurements up to an upper modulation frequency of about 2000 MHz, and to resolve mixtures of fluorophores exhibiting different lifetimes. During our tests, we observed remarkable and lasting (2 h) time drifts between the optical output and the electrical trigger input or output. At present, work is in progress at Hamamatsu to eliminate these drifts.

1. INTRODUCTION

Synchronously-pumped and cavity dumped dye lasers are now in common use as excitation light sources in time-domain as well as in frequency-domain fluorescence spectroscopy. Owing to their high peak power, these lasers allow efficient frequency doubling over extended wavelength ranges, yielding tunable short-wavelength excitation radiation. Unfortunately, these laser systems are rather expensive, require considerable lab space, and are difficult to align and maintain. For specific applications, frequency-doubled pulsed semiconductor laser diodes can be considered as an alternative light source. They offer pulse durations less than 50 ps, pulse repetition frequencies in the MHz range, and wavelengths as short as 410 nm. At present, however, the available output power is much lower than in the case of a common dye laser. Hamamatsu is producing a 410-nm model (C3551–02 controller/LDH-041-CS laser diode head) which shows a constant peak power of 0.44 mW, a pulse duration of 40 ps, and a maximum repetition frequency of 10 MHz. This corresponds to an average power of only 220 nW at the 10 MHz repetition rate, or to 3.5*10⁴ emitted excitation photons per laser pulse. Taking into account all optical losses, the question arises if this low power is still adequate to perform useful fluorescence lifetime measurements on samples with moderate quantum efficiencies. In this paper, we report on test measurements using frequency-domain methods. This means, we consider the PLP-01 picosecond light pulser to be a laser light source which is intensity-modulated at the pulse repetition frequency, and at all harmonics of that frequency^1–3. Owing to the short pulse duration of 40 ps, efficient intensity modulation is available up to about 11 GHz. By measuring the phase shift of the fluorescence radiation relative to the laser radiation, the fluorescence decay can be investigated.

2. EXPERIMENTAL

In fig. 1, our test set-up is shown schematically. The scattering or fluorescent samples are arranged closely to the LDH-041-CS laser diode head in order to capture as much excitation light as possible. A Coherent series 700 sync-pumped, cavity dumped, and frequency doubled dye laser is used to drive an Antel model AR-S3 picosecond photodetector. The photodetector output signal triggers the PLP-01 light pulser, and is used as RF signal for a reference mixer. A Gigatronics model 905 frequency synthesizer, phase-locked to the dye laser, generates the LO signals for this reference mixer and a signal mixer. The LO frequencies differ from the harmonics of the laser pulse repetition frequency by 25 Hz. The dye laser can also be directed towards the sample compartment to compare the semiconductor laser emission with the well-known dye laser radiation. A Hamamatsu model R2566U MCP photomultiplier is used to measure scattered radiation as well as fluorescence light. The photomultiplier output signal is down-mixed to 25 Hz in the signal mixer, and relative phase measurements are accomplished using a standard SLM phase meter. The set-up contains also a DC signal channel between the photomultiplier and the SLM module which has been described earlier⁴. The 25-Hz AC signal and the DC signal are processed within the SLM module to obtain the modulation information, AC/DC.

Fig. 1.

Experimental set-up used to measure AC/DC, phase-modulation data, and the drift of the optical output pulse relative to the trigger input.

In order to test the suitability of the 413-nm PLP-01 for phase-modulation fluorometry, we measured at first the modulation degree (AC/DC) for scattered light up to 3800 MHz modulation frequency. In a second step, we repeated this measurement using the dye laser as a light source. By dividing the PLP-01 AC/DC values by the corresponding dye laser values, we obtained the harmonic content of the PLP-01 picosecond light pulser relative to the dye laser. Owing to this procedure, most of the potential apparatus artifacts are eliminated. The results shown in fig. 2 indicate that, besides the short 40 ps pulse, the PLP-01 seems to have a longer emission component with a characteristic time constant of about 500 ps. This 500 ps component contributes to the harmonic content at all frequencies below 1000 MHz. At higher frequencies, the 500 ps component contributes only to the measured DC signal. Therefore, the resultant harmonic content for the PLP-01 is lower than for the sync-pumped, cavity dumped, and frequency-doubled dye laser.

Fig. 2.

Harmonic content of the Hamamatsu 413-nm PLP-01 picosecond light pulser relative to the Coherent series 700 dye laser.

During our tests, we observed a remarkable and lasting time drift between the PLP-01 optical output pulse and the electrical trigger input pulse. Typical drift plots are shown in fig. 3, and in fig. 4. For the drift measurements, we used the experimental set-up of fig. 1 with a scattering sample. Any drift within the dye laser would shift the LO signals of both mixers, and would be of no influence. A drift within the MCP-photomultiplier, however, would have the same effect as a drift within the PLP-01. Using scattered dye laser radiation, we have verified that there is no measurable photomultiplier drift under our laboratory conditions. The time drift, Δt, has been calculated from the measured phase drift, ΔΘ, by the equation Δt = ΔΘ/2πf, where f is the modulation frequency. The accuracy of our phase measurements was better than 1 degree. This corresponds to a timing accuracy for the drift measurements of better than 1.8 ps, assuming a modulation frequency of 1500 MHz. According to fig. 3, we had to wait for a warm-up time of about 3 h prior to any fluorescence lifetime measurement. From fig. 4 it can be seen that the time drift shows a very similar behavior on different days, and that the calculated time drift values are independent of the modulation frequency used.

Fig. 3.

Drift of the PLP-01 optical output pulse relative to the electrical trigger input pulse. The time drift has been calculated using phase shift data measured at 1533.18 MHz modulation frequency. At zero time, the PLP-01 was switched on. The pulse repetition rate was 3.795 MHz.

Fig. 4.

Drift of the PLP-01 optical output pulse relative to the electrical trigger input pulse. The time drift has been calculated using phase shift data. Curves A and B correspond to individual measurements taken on consecutive days, using 345.345 MHz and 1533.18 MHz modulation frequency. At zero time, the PLP-01 was switched on. The pulse repetition rate was 3.795 MHz.

The C3551–02 controller of the PLP-01 system is equipped with an electrical trigger output. We expected this signal to be drift-free. Therefore, we examined in an additional experiment if there is any drift between the PLP-01 optical output and the available electrical trigger output. Unfortunately, the trigger output pulse of the PLP-01 has a FWHM of more than 10 ns, and shows relatively long rise and decay times, even if subnanosecond trigger input pulses are applied. A long and smooth trigger output pulse is rather inconvenient, however, for applications of the PLP-01 in phase-modulation fluorometry. In order to take advantage of the high-frequency content of the PLP-01 optical pulses, a high-frequency reference signal is needed. Preferably, this should be a subnanosecond pulse or a subnanosecond-risetime pulse at the same repetition frequency as the optical output. On sync-pumped, cavity dumped dye lasers, this reference signal can be generated easily by splitting off part of the laser output to drive a high-speed photodiode. The frequency-doubled output power of the PLP-01 is too low, however, to use this method. Instead, we have triggered the PLP-01 in this drift measuring test at a 10 MHz frequency-synthesizer clock frequency. In this case, the PLP-01 trigger output has also this 10 MHz frequency, which can be used to phase-lock a second frequency-synthesizer. Any drift of the trigger output signal of the PLP-01 would be transferred to this second frequency-synthesizer whose output signal frequency can be set to any harmonic of the laser pulse repetition frequency.

In fig. 6, the time drift of the PLP-01 optical output relative to the electrical trigger output is shown. Again, the time drift has been calculated using phase shift data. Unfortunately, this drift can also not be neglected, neither in phase-modulation fluorometry, nor in time-correlated single-photon counting-based fluorometry. In our opinion, a high-speed pin photodiode, integrated into the laser head by the manufacturer, could be used to generate a short and drift-free trigger output pulse. To drive the photodiode, scattered or reflected NIR could be used. If this should be impossible for some reason, an optical NIR reference output might be helpful.

Fig. 6.

Drift of the PLP-01 optical output pulse relative to the electrical trigger output pulse. The time drift has been calculated using phase shift data measured at 490 MHz modulation frequency. Curves A and B correspond to individual measurements taken on consecutive days. At zero time, the PLP-01 was switched on. The pulse repetition rate was 10 MHz.

3. OPTICAL DELAY MEASUREMENTS

We used known optical delays to simulate very short fluorescence lifetimes. In these tests, the signal phase including the optical delay was measured relative to the reference signal phase. In a second step, the signal phase without the optical delay was measured again. In a third step, the phase difference between the phase readings of step 1 and step 2 was calculated. This procedure corresponds completely to the procedure used in fluorescence lifetime measurements. As expected, the modulation of the delayed light relative to that of the undelayed light did not change with frequency, and was very close to 100%. In fig. 7, the measured phase shift versus modulation frequency is shown for optical delays of 16 ps and 82 ps. For fluorophores with 16 ps and experiment can be considered to be a very good simulation, whereas the 82 ps experiment is still a reasonable one. As expected, the delay-generated phase shifts versus frequency show a linear relationship, and the measured values agree with the theoretical values.

Fig. 7.

Test of the phase measuring accuracy by means of known optical delays of 16 ps and 82 ps.

4. FLUORESCENCE LIFETIME MEASUREMENTS

In order to test the PLP-01 in actual decay time measurements we selected four fluorophores with lifetimes ranging from 0.5 to 6.4 ns, and which are known to be standards for single-exponential decays. These fluorophores were chromatographically pure and we used spectroscopic grade toluene and dimethylformamide as solvents. The solutions were purged by nitrogen before the measurements. The emission was observed through a Corning 3–73 band-pass filter which absorbs the emission and/or scattered light below 420 nm.

Frequency responses for perylene in toluene and 4-dimethylamino-4’-methoxystilbene (DMS) are shown in fig. 8. These measurements were performed after the PLP-01 was on for 3 hours, to allow time for stabilization of the drift. The experimental data are well matched by solid lines representing the best single-exponential fits, and deviations (lower panel) are random.

Fig. 8.

Phase and modulation data for perylene in toluene (o, χ_R² = 1.2), and DMS in DMF (o, χ_R² = 10.9). The solid lines shoe the best single-exponential fits to the data.

The apparent degree of random error (0.5° in phase, and 0.01 in modulation) is only slightly larger than that found for measurements with the higher intensity dye lasers, which is typically 0.3° in phase and 0.007 in modulation.

We also examined the PLP-01 for use in the resolution of a multi-exponential. For this purpose, we chose two fluorophores, diphenylanthracene (DPA) and 2,5-bis[5-tert-Butylbenzoxazolyl (2)]thiophene (BBOT), each of which shows single-exponential decays times of 6.40 ns and 1.32 ns, respectively (fig. 9, top). Frequency-domain data for the mixture is shown in fig. 9, bottom. It is instructive to examine data for a mixture of fluorophores. An example is shown in fig. 9, bottom, which shows the frequency response for a mixture of DPA (6.40 ns) and BBOT (1.32 ns).

Fig. 9.

Top: Phase and modulation data for DPA (o) and BBOT (o) in toluene. The solid lines show the best single-exponential fit to the data. Bottom: Phase and modulation data for the mixture DPA-BBOT. The dashed line shows the best single-exponential fit to the data, and the solid line represents the double-exponential fit.

The presence of multiple decay times is immediately evident from an attempt to fit the data to a single decay time. The best fit, shown as the dashed line, is obviously inadequate. In contrast, the data can be explained by a curve with decay times of 6.84 ns and 1.37 ns and near equal amplitudes (solid line). The expected amplitudes are f₁ = 0.5 and f₂= 0.5, where f_i = α_iτ_i, / Σα_iτ_i.

5. CONCLUSION

In conclusion, using the Hamamatsu 413-nm model PLP-01, we have been able to perform phase-modulation fluorescence measurements up to an upper modulation frequency of about 2000 MHz, and to resolve mixtures of fluorophores exhibiting different lifetimes. We observed remarkable and lasting time drifts between the PLP-01 optical output and the electrical trigger input and output, which results in an inconvenient long warm-up time of about 3 h. We propose to integrate a high-speed photodiode into the laser head in order to generate a drift-free subnanosecond trigger output pulse, or to establish an optical trigger output. At present, work is in progress at Hamamatsu to eliminate the drifts, and to shorten the trigger output pulse.

Fig. 5.

Experimental set-up used to measure the drift of the optical output pulse relative to the electrical trigger output. Any drift of the PLP-01 trigger output is transferred to the 490 MHz output signal of the model 905 synthesizer.