Search the CONFOCAL archive at
http://listserv.acsu.buffalo.edu/cgi-bin/wa?S1=confocal

Arthur,

The idea of pulse stretching is an interesting one, and the experimental
implications are different depending on your coupling strategy.  As a
general rule, fiber coupling using femtosecond pulses is too complicated to
be practical.  That's not to say that it can't be done, it's just more
difficult than it's worth.  In this special case, the concept of 'pre-
chirping' is not straight forward because the peak-power levels are so high
that non-linear effects begin to show up that aren't easily compensated
for.  One can avoid these effects by keeping the average power low but then
you run into trouble with coupling efficiency into the fiber, and getting
enough energy to your sample (which is why you wanted to fiber couple to
begin with)  Another disadvantage of fiber coupling is that re-optimizing
the 'pre-chirp' fiber apparatus is usually necessary following a wavelength
change, making tuning the system a chore.

Direct coupled systems, as you point out, will also exhibit some pulse-
stretching but not nearly as much.  Exactly how much stretching depends
upon how much, and what type of material is used in the various optical
components making up the beam path.  If these parameters are known, then
one can very accurately calculate the resultant pulse width.  Trouble is
that microscope objective designs are proprietary and this information is
not readily available.  However, some investigators have measured the pulse
width through different microscopes so we can guess at typical values.
From the data presented here on the listserver from Amplitude systems:

Input pulse: 100 fs
Output pulse: 200 fs

Input pulse: 200 fs
Output pulse: 220 fs

The calculated dispersion assuming 800 nm is 5000 fs^2.  A number typical
of most microscope set-ups.  This number changes with wavelength - as the
wavelength gets shorter, the dispersion goes up.  (If interested, I can
provide you with an excel spreadsheet to compute these for yourself)  The
real question is how does this effect the 2-photon absorption efficiency?
One can also calculate this difference in terms of fluorescence generated
for a given pulse width.  For the first case, the resulting pulse is 200 fs
leading to a fluorescence intensity of 75 arb. units.  For the second,
where the resulting pulse is 220 fs the intensity is 60 arb. units.
(Calculations outlined by Peter T.C. So, in Methods of Cellular Imaging, A.
Periasamy, p. 147) So it is not correct to assume that 'less pulse-
strecthing means more signal'  There are more factors at play that
determine the optimum parameters.  In fact, even with highly dispersive
microscope systems, pulse-stretching can be compensated for by only small
increases in the average power.

Feel free to contact me with any questions:
Ian Read
Spectra-Physics
650-966-5346