126
OPTICS LETTERS / Vol. 19, No. 2 / January
15, 1994
Generation of 21-fs millijoule-energy pulses
by use of Ti:sapphire
Jianping Zhou, Chung-Po Huang, Chengyu Shi, Margaret M. Murnane, and Henry C. Kapteyn
Department of Physics, Washington State University, Pullman, Washington 99164-2814
Received August 24, 1993
We have demonstrated the generation of 21 + 2 fs duration pulses, with an energy of 0.5 mJ, a bandwidth of
44 nm, and a repetition rate of 10 Hz, using a chirped-pulse amplification scheme in Ti:sapphire. We use 11-fs,
5-nJ pulses from a Ti:sapphire oscillator as the input for an eight-pass Ti:sapphire amplifier. A broadband,
low-dispersion chirped-pulse stretcher-and-amplifier design was used for propagation and amplification of the
ultrashort pulses. By the addition of a prism pair to the stretcher, both the second- and third-order dispersions
are minimized, permitting us to generate near-transform-limited amplified pulses at a duration near the
theoretical limit imposed by gain narrowing in Ti:sapphire.
High-power, ultrashort laser pulses are currently
used for studies of ultrafast x-ray generation, ' 2
short-wavelength lasers,3 5 extreme-ultravolet harmonic generation,6 -8 and multiphoton ionization.
The extremely large bandwidth and high gain
of Ti:sapphire9 has made this material one of
the most promising for the generation of ultrashort optical pulses. Pulses as short as 11 fs
(Refs. 10-13) were generated directly from a selfmode-locked Ti:sapphire laser by optimization of
intracavity dispersion compensation in the oscillator.
The high saturation fluence and damage threshold
of Ti:sapphire make it attractive for ultrashortpulse amplifiers also. However, aberrations and
phase distortions in the amplification schemes
employed to date have limited high-energy pulse
durations to >60 fs.14"5 In this Letter we discuss
the design considerations necessary to propagate
and amplify 20-fs-duration pulses to the millijoule
level.'6 Experimentally we have generated 0.5-mJ,
21-fs pulses with a near-transform-limited bandwidth
of 44 nm, using a simple amplifier system. This
bandwidth is also near the theoretical limit set by
gain narrowing in Ti:sapphire. Our approach should
scale to higher pulse energies in the near future.
In the design of an amplifier capable of generating
ultrashort pulses, the same considerations that apply
to optimization of the oscillator also apply to the amplifier. Thus we minimize higher-order dispersion
by minimizing total material path length. The amplifier uses a chirped-pulse amplification scheme,'7"- 9
whereby the broad-bandwidth pulse from the laser
oscillator passes through a diffraction-grating setup
to stretch the pulse in time. This pulse is then
amplified to high energy while a modest intensity is
maintained within the amplifier crystal. Following
amplification, we can recompress the stretched pulse
by passing it through a grating pair.
Figure 1 shows the system configuration. The
input to the amplifier is from a self-mode-locked
Ti:sapphire oscillator that produces transformlimited 11-fs-duration pulses with a repetition rate
of 83 MHz, a center wavelength of 796 nm, and
0146-9592/94/020126-03$6.00/0
a FWHM bandwidth of 62 nm. The initial pulse
energy is 5 nJ. Single pulses are selected from the
mode-locked pulse train at 10 Hz by a KD*P Pockels
cell and crossed polarizers (Corning Polarcor). This
Pockels cell is placed immediately outside the
oscillator in order to suppress amplified stimulated
emission (ASE) that can originate from the amplifier,
reflect from the oscillator output coupler, and
be reamplified to high energy. Next, the beam
passes through the pulse stretcher, which uses allreflective optics to minimize material path length and
chromatic aberration. The stretcher uses a single
grating, a single parabolic mirror of focal length 30 in.
(76.2 cm) to form an achromatic telescope with no
spherical aberration, and two flat mirrors. We use
an Al-coated 300-groove/mm grating, 1 in. x 2 in.
(2.54 cm X 5.08 cm), blazed for 750 nm with a
reflectivity of 74% at 800 nm, and placed 49 cm from
the parabolic mirror. Reflections from the grating
are at Littrow angle and disperse the beam in the
horizontal plane, with slight beam deviations vertically so that the input and output beams are slightly
displaced. The total bandwidth acceptance of the
stretcher is approximately 140 nm, and the 11-fs
pulses are stretched to approximately 20 ps. An
identical set of gratings is used in a standard,
Fig. 1. Diagram of laser system. PC, Pockels cell; ML,
mode-locked.
© 1994 Optical Society of America
January
double-pass compressor design. The single-grating
Littrow stretcher design maintains alignment while
the separation and/or angle of the grating is adjusted.
When combined with the relatively modest stretching
factor (-2000), optimization of the pulse duration is
straightforward and repeatable.
The amplifier uses an 8-mm-long highly doped
(0.23%, Union Carbide), Brewster-cut Ti:sapphire
rod. It is pumped longitudinally with 90 mJ of
532-nm light from a doubled Q-switched Nd:YAG
laser (Continuum YG-681C). The pump beam is
imaged into the amplifier with a 1-m focal-length lens
that focuses -20 cm in front of the crystal. A highefficiency (HEPA) air filter is used to exclude dust
from the region of the focus to prevent breakdown. The amplifier rod is at the focus of four
1-mi radius-of-curvature dielectric mirrors, which
form an eight-pass figure-eight amplifier.2 0" 2 2 The
amplified beam is near focus as it passes through
the crystal. In order to suppress ASE the beam
focuses through a four-hole array placed near the
crystal, and a saturable absorber (RG-850, Schott
Glass Technologies) is used before the final pass
through the amplifier. This filter eliminates most
of the ASE while transmitting approximately 95% of
the signal spectrum. Since all these elements are
incorporated within the eight-pass configuration the
setup is simple and compact.
In order to obtain zero second- and third-order
dispersion2 3 in the amplifier system simultaneously,
we insert a pair of SF-18 prisms in a 29-cm double
pass configuration between the stretcher and the
amplifier. We verified third-order compensation
by performing an autocorrelation of the input cw
mode-locked beam after it propagated through
the entire system. The pulse duration could be
recompressed to 18 fs by use of the prism-grating
15, 1994 / Vol. 19, No. 2 / OPTICS LETTERS
127
shift in the center position on the CCD readout. As
in the case of third-order compensation, we verified
this calibration by using the single-shot correlator
with cw-mode-lockedpulses of known duration, which
yield a readily observable signal by use of -300 mW
of 15-fs-duration pulses. The estimated calibration
accuracy is 10%, and measurements were taken with
different parts of the amplified beam to eliminate the
possibility of any profile effects.
Figure 3 shows the effects of spectral narrowing by
propagation through the amplifier system. Curve
(a) is the spectrum through the system without amplification. Clipping at the long-wavelength end of
the spectrum (which is due to the stretcher aperture)
is visible, as is narrowing to 58 nm, which is due to
the gratings. Curve (b) is the amplified spectrum,
which narrows to 44 nm and red shifts 9 nm as
a result of the gain profile of Ti:sapphire and the
gain saturation of the chirped pulse. The Fourier
transform of this spectrum corresponds to an 18-fs
pulse. Theoretical simulations of gain narrowing
yield results consistent with these data. Assuming
an infinite input bandwidth results in a bandwidth of
48 nm. Thus our amplifier generates a bandwidth
near the theoretical gain-narrowing limit.
Figure 4 shows the profile of the amplified and recompressed beam. There is no evidence of self-phase
modulation or self-focusing of the beam. A pinholetransmission measurement, made by focusing the
A
?:
._--\212 fs
configuration but could be compressed to only 24 fs
without the prisms. In both cases the stretcher
grating angle and the prism position were adjusted to optimize pulse duration. The beam does
not compress back to 11 fs because of spectral
modulation and slight clipping by the diffraction gratings. In a simplified setup, using 8 cm
of Ti:sapphire instead of the full amplifier, we verified
that the compressed pulse had no pedestal <10-3
below the peak intensity.
On amplification the pulse experiences a gain
of nearly 107, and the output of the multipass
amplifier is -2 mJ. The double-pass grating compressor pair has a 25% overall transmission, which
reduces the compressed pulse energy to 0.5 mJ
(Au-coated 300-groove/mm
gratings
with 82% effi-
ciency, which are available, should nearly double
the throughput to - 1 mJ). Figure 2 shows the
single-shot autocorrelation trace of the attenuated
compressed pulse, indicating a duration of 21 fs
(assuming a sech2 pulse shape). Repeated measurements were in the range of 18-22 fs. The
autocorrelation measurement used a 300-,tm-thick
KD*P crystal and a crossing angle between the two
beams of -2°. We calibrated the path delay by
moving one of the two arms of the correlator with
an encoded motor micrometer and by observing the
-60
-40
-20
0
20
40
60
Time (fs)
Fig. 2. Single-shot autocorrelation trace of a sample of
the 0.5-mJ amplified and recompressed pulse. Assuming
a sech2 pulse shape, we see that the pulse duration is
21 ± 2 fs.
I-
(a)j
I
,
750
800
850
Wavelength (nm)
900
Fig. 3. Spectra of the laser system output: (a) unamplified, with FWHM 58 nm [curve (a)] and amplified, in
which gain narrowing reduces the spectrum to 44 nm
FWHM [curve (b)].
128
OPTICS LETTERS / Vol. 19, No. 2 / January
I
15, 1994
The authors acknowledge generous support from
Union Carbide Crystal Products and NewportKlinger. The authors also acknowledge the generous
assistance of George Venikouas and Milan Kokta
of Union Carbide, and John McIntosh and Chris
Baldwin at Washington State University. M. Murnane acknowledges support from a Sloan Foundation
Fellowship.
N<
References
..~~~~~0
-0.4
-0.2
0.0
0.2
Position (cm)
0.4
Fig. 4. Beam profile of the amplified pulse. The beam
size is 2.15 mm FWHM in the x direction and 1.95 mm
in the y direction.
attenuated beam with a 1-m radius-of-curvature
mirror at near-normal incidence, demonstrated a
transmission of 65% through a 100-,um-diameter
pinhole. This corresponds to a waist size of 66 /um.
From beam-size measurements, an ideal Gaussian
beam waist at this point should be 45-65 /.Lm;
thus the beam retains <1.5 times diffraction-limited
focus. Extrapolating our beam-size measurements
back into the amplifier crystal, we estimate the
maximum beam fluence there to be -0.9 J cm-2 ;
thus, although we stretch the pulse to only 20
ps, we achieve saturation in the amplifier without
significant nonlinear distortion. Although the peak
fluence in the crystal is 5 X 1010W cm-2 , the limited
material path length (-1-3 mm) at this fluence
limits the B integral to less than one.
The ASE level of the resulting beam is 1-5% of
total beam energy, depending on the alignment of
the system, which we measured by disabling the
Pockels cell and measuring the output energy. However, most of this ASE occurs as a postpulse 150 ns
after the short pulse. The ASE intensity was also
observed to decrease at optimum alignment when
the pulse was being amplified. A second Pockels cell
after the amplifier should reduce the ASE level substantially. The rms energy fluctuation of the amplified pulse is -8%, consistent with modest saturation
of the gain.
In conclusion, we have used the technique of
chirped-pulse amplification in a Ti:A12 03 -based
amplifier system to produce pulses of 21 + 2 fs
with an energy of 0.5 mJ. Dispersion and nonlinear
effects are minimized by use of a short amplifier
crystal and reflective optics wherever possible.
These techniques should be generally applicable to
amplifier systems that use laser-pumped gain media,
to pulse energies of >1 J. The development of such
systems has significant implications for high-field
physics and for coherent and incoherent x-ray and
extreme-ultraviolet radiation generation.
This research was supported by the National Science Foundation, donors of the Petroleum Research
Fund, administered by the American Chemical
Society, and the Washington Technology Centers.
1. M. M. Murnane, H. C. Kapteyn, M. D. Rosen, and
R. W. Falcone, Science 251, 531 (1991).
2. M. M. Murnane, H. C. Kapteyn, S. P. Gordon, J. Bokor,
E. N. Glytsis, and R. W. Falcone, Appl. Phys. Lett. 62,
1068 (1993).
3. H. C. Kapteyn, Appl. Opt. 31, 4931 (1992).
4. N. H. Burnett, and P. B. Corkum, J. Opt. Soc. Am. B
6, 1195 (1989).
5. D. C. Eder, P. Amendt, and S. C. Wilks, Phys. Rev. A
45, 6761 (1992).
6. J. K. Krane, M. D. Perry, S. Herman, and R. W.
Falcone, Opt. Lett. 17, 1256 (1992).
7. A. L'Huillier and P. Balcou, Phys. Rev. Lett. 70, 774
(1993).
8. J. J. Macklin, J. D. Kmetec, and C. L. Gordon III, Phys.
Rev. Lett. 70, 766 (1993).
9. P. F. Moulton, J. Opt. Soc. Am. B 3, 125 (1986).
10. C. P. Huang, M. T. Asaki, S. Backus, M. M. Murnane,
H. C. Kapteyn, and H. Nathel, Opt. Lett. 17, 1289
(1992).
11. M. T. Asaki, C. P. Huang, D. Garvey, J. Zhou,
H. C. Kapteyn, and M. M. Murnane, Opt. Lett. 18,
977 (1993).
12. F. Krausz, M. E. Fermann, T. Brabec, P. F. Curley,
M. Hofer, M. H. Ober, C. Spielmann, E. Wintner, and
A. J. Schmidt, IEEE J. Quantum Electron. 28, 2097
(1992).
13. B. Proctor and F. Wise, Appl. Phys. Lett. 62, 470
(1993).
14. A. Sullivan, H. Hamster, H. C. Kapteyn, S. Gordon,
W. White, H. Nathel, R. J. Blair, and R. W. Falcone,
Opt. Lett. 16, 1406 (1991).
15. R. Shepherd, Lawrence Livermore National Laboratory, Livermore, Calif. 94550 (personal communication, 1993).
16. We note that other methods for compensating highorder dispersion in amplification systems have been
proposed. See, for example, W. E. White, F. G. Patterson, R. L. Combs, D. F. Price, and R. L. Shepherd,
Opt. Lett. 18, 1343 (1993); B. E. Lemoff and C. P. J.
Barty, Opt. Lett. 18, 1651 (1993).
17. D. Strickland and G. Mourou, Opt. Commun. 56, 219
(1985).
18. M. Pessot, P. Maine, and G. Mourou, Opt. Commun.
62, 419 (1987).
19. 0. E. Martinez, IEEE J. Quantum Electron. QE-23,
1385 (1987).
20. R. L. Fork, H. Avramopoulos, H. L. Fragnito, P. C.
Becker, K. L. Schehrer, and C. Hirlimann, Opt. Lett.
14, 1068 (1989).
21. P. Georges, F. Estable, F. Salin, J. P. Poizat, P. Grangier, and A. Brun, Opt. Lett. 16, 144 (1991).
22. C. Le Blanc, G. Grillon, J. P. Chambaret, A. Migus,
and A. Antonetti, Opt. Lett. 18, 140 (1993).
23. R. L. Fork, C. H. B. Cruz, P. C. Becker, and C. V.
Shank, Opt. Lett. 12, 483 (1987).