Dispersion-based pulse shaping for multiplexed
two-photon fluorescence microscopy
Guillaume Labroille,1Rajesh S. Pillai,1Xavier Solinas,1Caroline Boudoux,1,2
Nicolas Olivier,1Emmanuel Beaurepaire,1and Manuel Joffre1,*
1Laboratoire d’Optique et Biosciences, Ecole Polytechnique, CNRS, and INSERM U696, 91128 Palaiseau, France
2Centre d’Optique, Photonique et Laser, Engineering Physics Department, Ecole Polytechnique de Montreal,
P.O. Box 6879, Station Centre-Ville, Montreal, Quebec, Canada
*Corresponding author: email@example.com
Received June 29, 2010; revised August 27, 2010; accepted August 30, 2010;
posted September 20, 2010 (Doc. ID 130844); published October 12, 2010
We demonstrate selective two-photon excited fluorescence microscopy with shaped pulses produced with a simple
yet efficient scheme based on dispersive optical components. The pulse train from a broadband oscillator is split
into two subtrains that are sent through different amounts of glass. Beam recombination results in pulse-shape
switching at a rate of 150 MHz. Time-resolved photon counting detection then provides two simultaneous images
resulting from selective two-photon excitation, as demonstrated in a live embryo. Although less versatile than
programmable pulse-shaping devices, this novel arrangement significantly improves the performance of selective
microscopy using broadband shaped pulses while simplifying the experimental setup.
180.4315, 320.5540, 170.3880.
© 2010 Optical Society of
Spectral phase shaping of broadband femtosecond
pulses has been shown to be an effective approach for
selective two-photon excited fluorescence (2PEF) micro-
scopy [1–5]. Similar to coherent control of two-photon
absorption [6,7], selective fluorophore excitation using
shaped broadband pulses results from narrowing the la-
ser two-photon spectrum. This is achieved using destruc-
tive interference between undesired combinations of
pulse spectral components. As compared to the use of
a tunable femtosecond laser with a narrower spectral
width, one advantage of broadband pulse shaping lies
in the ability for fast switching between different pulse
shapes, in turn allowing multiplexed addressing of sev-
eral chromophores. This point is particularly attractive
for imaging biological systems [3–5].
In this Letter, we show that the phase-shaping needed
for selective 2PEF microscopy does not necessarily re-
quire the use of a programmable pulse shaper as in pre-
vious studies but can be achieved using dispersive glass
components placed before the microscope. Apart from
the advantages of simplicity and efficiency, this simple
shaping scheme turns out to be particularly well suited
for use in combination with an efficient multiplexing
scheme previously demonstrated in the context of multi-
Stokes Raman scattering (CARS) microscopy [11,12]. We
oping embryos with a 5 μs pixel acquisition time and a
switching rate of 150 MHz between the two pulse shapes.
Two-photon excitation efficiency of a chromophore is
governed by the overlap between the absorption spec-
trum of the chromophore and the two-photon spectrum
of the pulse (see, e.g., [4,7]). The two-photon spectrum
describes the frequencies that the pulse can produce
by two-photon excitation and is given by the spectrum
of the laser field squared in time domain [4,7]. A strategy
for achieving selective two-photon excitation with a
broadband pulse is to use a spectral phase profile
φðωÞ that is antisymmetric with respect to a particular
frequency ω1. The associated group delay τ1ðωÞ ¼
dφ1=dω is then symmetric with respect to ω1. Frequen-
cies ω and 2ω1− ω, symmetric around ω1, therefore
correspond to the same group delay and can always
mix to contribute to frequency 2ω1in the two-photon
spectrum. In contrast, the production of other frequen-
cies in the two-photon spectrum is strongly inhibited
due to the nonsimultaneity of the required frequencies.
Various spectral phase profiles suitable for selective
two-photon excitation have been previously proposed
or demonstrated, including sinusoids [3,6,7,13], quasi-
random binary phase masks , pseudorandom Galois
fields , or third-order spectral phase [4,7]. We use
here this latter configuration, where the spectral phase
6φ000ðω − ω1Þ3;
and the corresponding group delay is a parabolic
2φ000ðω − ω1Þ2:
The resulting two-photon spectrum is centered at fre-
quency 2ω1, with a spectral width decreasing when jφ000j
is increased [4,7]. As shown in this Letter, one advantage
of such a third-order spectral phase is that it can be pro-
duced by using simple dispersive optical elements in-
stead of a more complex programmable pulse shaper.
Indeed, it is well known that a prism compressor pro-
duces not only a second-order spectral phase but also
a significant amount of third-order spectral phase .
After compensation between the second-order terms of
opposite signs resulting respectively from the prism com-
pressor and the microscope objective, the net spectral
phase is mostly a third-order function of frequency.
3444 OPTICS LETTERS / Vol. 35, No. 20 / October 15, 2010
© 2010 Optical Society of America
As shown in Fig. 1, a prism compressor produces a
group delay decreasing with frequency but with a strong
curvature (a). The remaining microscope optics produce
a group delay that exhibits an almost linear increase with
frequency (b). Assuming a transform-limited incident
laser pulse (constant group delay), the resulting group
delay (a þ b) is then a parabolic function with a negative
curvature, corresponding to Eq. (2). The center fre-
quency of this parabola, ω1, can be easily shifted toward
higher frequencies by introducing an additional amount
of linear dispersion, e.g., with a piece of glass inserted in
the optical path (c). Indeed, let us call φ00
order phase introduced by the glass. Neglecting higher-
order terms that are negligible with respect to those of
the prism compressor, the group delay introduced by
the glass simply reads φ00
center frequency. The total group delay then becomes
glassðω − ω0Þ, where ω0is the laser
τ2ðωÞ ¼ τ1ðωÞ þ φ00
glassðω − ω0Þ;
ω − ω1þφ00
The new parabola τ2ðωÞ is thus centered on frequency
ω2¼ ω1− φ00
φ000< 0. The final step of rapid switching between
τ1ðωÞ and τ2ðωÞ is achieved by time multiplexing [8–12],
as discussed in the following.
The experimental setup is shown in Fig. 2. The 15 fs
pulses produced at a 75 MHz repetition rate by a broad-
band titanium:sapphire oscillator (Synergy PRO, Femto-
lasers, Austria) are negatively chirped in a prism-pair
compressor (N-SF14, 40:5 cm between prisms). The
beam is then split in two parts using a half-wave plate
and a polarizing beam splitter. The horizontally polarized
part of the beam goes straight into the microscope objec-
tive (40×, 0.8 NA, φ00≈ 1970 fs2, φ000≈ 1300 fs3). In the
other path (vertically polarized), the pulses are dispersed
glass=φ000, a value greater than ω1 because
(φ00≈ 4300 fs2, φ000≈ 3250 fs3). This alternate beam is then
recombined with the perpendicularly polarized main
beam using a second polarizing beam splitter. The alter-
nate path is adjusted to be 2 m longer than the direct
path, so that the corresponding 75 MHz pulse train is ex-
actly interlaced with the 75 MHz pulse train of the main
path. This arrangement results in a 150 MHz pulse train
with a different pulse shape every other pulse, which cor-
responds to a 150 MHz switching rate. The 2PEF is then
acquired using a photon-counting scheme synchronized
with the 150 MHz pulse train, with two different counters
for even and odd pulses [8–10].
Figure 3(c) shows the two-photon spectra associated
with the two pulse shapes, measured with a BBO crystal.
Although thetwo pulses havenearly identical one-photon
spectra, their two-photon spectra are strongly focused to-
ward shorter or longer wavelengths, as desired. Accord-
ing to the spacing between the two peaks and to Eq. (4),
the corresponding third-order spectral phase of the pulse
is roughly −29000 fs3. The small oscillatory feature ob-
served on the high-frequency wing of each spectrum can
tral phase (third-order group delay). Despite thissmall re-
sidual term, Fig. 3 demonstrates that suitable two-photon
spectra can be produced at the focus of the microscope
would allow a more accurate shaping. Although the
throughput of our pulse-shaping scheme was not optimal
compressor,wemanaged toobtain shaped pulses with an
average power of about 40 mW at the focus of the micro-
scope objective. It was then possible to acquire images of
biological samples with pixel rates of 120 kHz.
Simultaneous dual-excitation imaging capability is de-
monstrated by observing a developing drosophila em-
bryo, as shown in Fig. 3. The pulses with blueshifted
two-photon spectra used for obtaining Fig. 3(a) selec-
tively excite the endogenous fluorophores in the yolk
of the embryo, whereas the pulses with redshifted two-
photon spectra provide preferential excitation of green
fluorescent protein (GFP)-labeled nuclei [Fig. 3(b)].
Figure 3(d) is a composite image combining images (a)
and (b). The fact that the colors are not pure green and
prism compressor, (b) the optics of the microscope, and (c) the
optics of the microscope plus an additional dispersive glass
placed in the beam path, and total group delay without
(a þ b) and with (a þ c) the dispersive glass. The additional dis-
persion results in a shift of the extremum (diamond).
Schematic plot of the group delay introduced by (a) the
izing beam splitter; GM, galvanometer-mounted mirrors; Trep,
laser period (1=Trep≈ 75 MHz); c, speed of light; cTrep≈ 4 m.
Experimental setup: HWP, half-wave plate; PBS, polar-
October 15, 2010 / Vol. 35, No. 20 / OPTICS LETTERS3445
blue indicates that there is some excitation crosstalk
between the two channels. This can be attributed to
several factors, including the overlap between the 2PEF
excitation spectra of the two fluorophores, the excited
state lifetimes of the fluorophores in the range of several
nanoseconds, and the response time of the photomulti-
are truly simultaneous, the contrast can be improved by
linear combinations using prior measurement of the rela-
tive excitation efficiency of the two chromophores [3,4].
Higher contrast would also be obtained in the case of
faster signals (e.g., harmonic generation) or lower laser
To summarize, we have demonstrated a new pulse-
shaping scheme for in vivo selective two-photon excited
fluorescence microscopy based on time multiplexing and
dispersive optical elements. Although the purpose of this
shaping approach is clearly not to generally compete
with more versatile conventional programmable pulse
shapers, we have shown that it was compatible with
the delivery of broadband pulses at the focus of a micro-
scope objective with a spectral phase suitable for selec-
tive 2PEF microscopy. Furthermore, the fact that the
second pulse shape was obtained from the first one by
a simple additional dispersion enabled time multiplexing,
allowing a 150 MHz switching rate between the two
shapes. As compared to previous implementations of
2PEF microscopy with shaped broadband pulses, the set-
up demonstrated here is thus simpler and benefits from a
much higher switching rate. Finally, we note that this
pulse-shaping scheme could be directly incorporated
in multiphoton microscopes employing even broader
bandwidth laser sources.
We thank Emmanuel Farge for providing us with the
transgenic eGFP labeled Drosophila strain. We thank
Delphine Débarre and Pierre Mahou for fruitful com-
ments. This work was supported by Agence Nationale
de l’Essonne (Astre 2006), and Fondation Louis D. de
l’Institut de France.
1. I. Pastirk, J. D. Cruz, K. Walowicz, V. Lozovoy, and M.
Dantus, Opt. Express 11, 1695 (2003).
2. J. D. Cruz, I. Pastirk, M. Comstock, V. Lozovoy, and M.
Dantus, Proc. Natl. Acad. Sci. U.S.A. 101, 16996 (2004).
3. J. P. Ogilvie, D. Debarre, X. Solinas, J.-L. Martin, E.
Beaurepaire, and M. Joffre, Opt. Express 14, 759 (2006).
4. R. S. Pillai, C. Boudoux, G. Labroille, N. Olivier, I. Veilleux,
E. Farge, M. Joffre, and E. Beaurepaire, Opt. Express 17,
5. K. Isobe, A. Suda, M. Tanaka, F. Kannari, H. Kawano, H.
Mizuno, A. Miyawaki, and K. Midorikawa, Opt. Express
17, 13737 (2009).
6. D. Meshulach and Y. Silberberg, Nature 396, 239 (1998).
7. V. Lozovoy, I. Pastirk, K. Walowicz, and M. Dantus, J. Chem.
Phys. 118, 3187 (2003).
8. W. Amir, R. Carriles, E. Hoover, T. Planchon, C. Durfee, and
J. Squier, Opt. Lett. 32, 1731 (2007).
9. R. Carriles, K. Sheetz, E. Hoover, J. Squier, and V. Barzda,
Opt. Express 16, 10364 (2008).
10. E. Chandler, E. Hoover, J. Field, K. Sheetz, W. Amir, R.
Carriles, S. Ding, and J. Squier, Appl. Opt. 48, 2067 (2009).
11. I. Rocha-Mendoza, W. Langbein, P. Watson, and P. Borri,
Opt. Lett. 34, 2258 (2009).
12. W. Langbein, I. Rocha-Mendoza, and P. Borri, Appl. Phys.
Lett. 95, 081109 (2009).
13. J. D. Cruz, I. Pastirk, V. Lozovoy, K. Walowicz, and M.
Dantus, J. Phys. Chem. A 108, 53 (2004).
14. M. Comstock, V. Lozovoy, I. Pastirk, and M. Dantus, Opt.
Express 12, 1061 (2004).
15. V. Lozovoy, B. Xu, J. Shane, and M. Dantus, Phys. Rev. A 74,
16. R. Fork, O. Martinez, and J. Gordon, Opt. Lett. 9, 150 (1984).
obtained with blueshifted pulses preferentially exciting (a) blue
endogenous fluorescence and (b) redshifted pulses preferen-
tially exciting GFP. (c) Corresponding two-photon spectra mea-
sured with a BBO crystal for the blueshifted and redshifted
pulses. (d) Combination of (a) and (b) according to the color
scale. Refer to movie file (Media 1).
Imaging with shaped pulses. (a), (b) Drosophila images
3446 OPTICS LETTERS / Vol. 35, No. 20 / October 15, 2010