Principles of Fluorescence Spectroscopy

112 TIME-DOMAIN LIFETIME MEASUREMENTS
The deflected beam is captured by a prism, which deflects
the beam out of the laser cavity (Figure 4.17). This procedure
is called cavity dumping.
For cavity dumping, the AO crystal is pulsed at the
desired repetition rate. For instance, for a 1-MHz repetition
rate, the RF pulses are sent to the cavity dumping crystal at
1 MHz, which selects one pulse in 80 to be dumped from
the dye laser cavity. The RF pulse width is narrow enough
to extract a single optical pulse from the dye laser. The
arrival times of the acoustooptic and laser pulses have to be
matched. The timing of a cavity dumper is typically
obtained by dividing the frequency of the mode locker by
factors of two, to obtain progressively lower repetition
rates. A somewhat confusing terminology is the use of
"continuous wave" (CW) to describe the 80-MHz output of
a dye laser. This term refers to continuous operation of the
cavity dumper, resulting in a continuous train of pulses at
80 MHz.
A valuable aspect of cavity dumping is that it does not
typically decrease the average power from the dye laser, at
least within the 1–4-MHz range typical of TCSPC. To be
specific, if the 80-MHz output of the dye laser is 100 mW,
the output of 4 MHz will also be close to 100 mW. When
optical power is not being dumped from the dye laser, the
power builds up within the cavity. The individual cavitydumped
pulses become more intense, which turns out to be
valuable for frequency doubling the output of the dye laser.
A final problem with the R6G dye laser output is its
long wavelength from 570 to 610 nm. While shorter wavelength
dyes are available, these will typically require a
shorter wavelength pump laser. Argon ion lasers have been
mode locked at shorter wavelengths, but this is generally
difficult. For instance, there are only a few reports of using
a mode-locked argon ion laser at 351 nm as an excitation
source for TCSPC. 41 Even after this is accomplished, the
wavelength is too long for excitation of protein fluorescence.
Fortunately, there is a relatively easy way to convert
the long-wavelength pulses to shorter wavelength pulses,
which is frequency doubling or second harmonic generation.
The cavity-dumped dye laser pulses are quite intense.
When focused into an appropriate crystal one obtains photons
of twice the energy, or half the wavelength. This
process is inefficient, so only a small fraction of the 600-nm
light is converted to 300 nm. Hence careful separation of
the long-wavelength fundamental and short-wavelength
second harmonic is needed. The important point is that frequency
doubling provides ps pulses, at any desired repetition
rate, with output from 285 to 305 nm when using an
Figure 4.18. Output power of commonly used laser dyes.
R6G dye laser. These wavelengths are ideal for excitation of
intrinsic protein fluorescence.
A convenient feature of dye lasers is the tunable wavelength.
The range of useful wavelengths is typically near the
emission maximum of the laser dye. Tuning curves of typical
dyes are shown in Figure 4.18. Most of these dye lasers
are used after frequency doubling. We use R6G for excitation
of intrinsic protein fluorescence, and 4-(dicyanomethylene)-2-methyl-6-(4-dimethylaminostyryl)-4H-pyran
(DCM) and Pyridine 2 (Py2) for excitation of extrinsic
probes. Excitation of tyrosine requires output of shorter
wavelengths than what is available from R6G. Rhodamine
560 and rhodamine 575 were found suitable for tyrosine
excitation using an argon ion or Nd:YAG laser pump
source, respectively. 42
4.4.4. Flashlamps
Prior to the introduction of ps lasers, most TCSPC systems
used coaxial flashlamps. A wide range of wavelengths is
available, depending on the gas within the flashlamp. These
devices typically provide excitation pulses near 2 ns wide,
with much less power than that available from a laser
source. Flashlamp sources became available in the
1960s, 43–45 but their use in TCSPC did not become widespread
until the mid-1970s. 46–49 Because of the lower repetition
rate and intensity of the flashlamps, long data acquisition
times were necessary. This often resulted in difficulties
when fitting the data because the time profile of the
lamps changed during data acquisition. 50–51 While these
problems still occur, the present lamps are more stable, provide
higher repetition rates to 50 kHz, and can provide
pulse widths near 1 ns. 52-55
Figure 4.19 shows a typical coaxial flashlamp. Earlier
flashlamps were free running, meaning that the spark