Generation of sub-50-fs vacuum ultraviolet pulses
by four-wave mixing in argon
Marcus Beutler,* Masood Ghotbi, Frank Noack, and Ingolf Volker Hertel
Max-Born-Institute for Nonlinear Optics and Ultrafast Spectroscopy, Max-Born-Strasse 2a, 12489 Berlin, Germany
* Corresponding author: email@example.com
Received December 21, 2009; accepted March 22, 2010;
posted April 1, 2010 (Doc. ID 121762); published April 30, 2010
We report on the generation of femtosecond pulses at 160 nm with energies up to 240 nJ at 1 kHz repetition
rate and sub-50-fs pulse duration. This pulse energy is a 1-order-of-magnitude improvement compared with
previous sub-100-fs sources in this wavelength range. The pulses are generated by four-wave difference-
frequency mixing process between the fundamental of a Ti:sapphire laser and its third harmonic in argon.
Pulse duration measurements are achieved by pump–probe ionization of Xe gas providing the cross corre-
lation between the fifth harmonic and the fundamental. © 2010 Optical Society of America
OCIS codes: 190.7110, 190.4223, 190.4380.
Conclusive single-photon pump–probe experiments
of small molecules such as water, small molecular
clusters, and biomolecules require the use of UV and
vacuum UV (VUV) photons [1–3]. Energetic photons,
especially at high repetition rates, are particularly
useful for pump–probe photoelectron spectroscopy.
There are nonlinear crystals that provide phase
matching for frequency conversion into the wave-
length range below 200 nm. Unfortunately, for many
of the commercially available crystals, the short
wavelength cut-off is given by their transparency
range. While sum-frequency generation (SFG) in
beta-barium borate (BBO) is possible down to about
189 nm, lithium triborate LiB3O5 (LBO), LiB4O7
(LB4), and KB5O8·4H2O (KB5) have been success-
fully applied for SFG down to 172.7 nm, 170.3 nm,
and 166 nm, respectively . In KBBF  wave-
lengths as short as 156 nm have been generated as
the fifth harmonic of a Ti:sapphire laser. Also non-
phase-matched second- harmonic generation (SHG)
down to 125 nm, but with pulse energies below the
picojoule range, has been presented in SrB4O7(SBO)
. Apart from the lack of phase matching, group ve-
locity mismatch (GVM) becomes an important issue
in generation of optical pulses in the deep UV spec-
tral range, limiting the available shortest pulse dura-
tion. To our knowledge, pulses shorter than ?100 fs
generated in nonlinear crystals have not been re-
ported for the VUV spectral range.
Using noble gases as nonlinear medium in a third-
or higher-order process is another approach to gener-
ate deep UV pulses. Earlier attempts were always
based on femtosecond laser seeded excimer amplifier
modules providing ultrashort pump pulses with sev-
eral milijoule pulse energy at 248 or 193 nm on the
other hand, limiting the repetition rate to about
10 Hz. Wittmann et al.  generated 300 fs pulses at
frequency mixing (FWDFM) of pulses from ArF exci-
mer amplifier in argon. Generation of some different
FWDFM of ultrashort 248 nm pulses with a dye laser
. In all of these cases phase matching arises from
the anomalous dispersion of wavelengths shorter
than a resonant absorption line. This leads to a con-
version efficiency of about 1%. However, the band-
width is limited, and further shortening of the output
pulses cannot be expected.
Whereas for four-wave mixing (FWM) under near
resonant conditions phase matching is easy to
achieve , in collinear nonresonant FWM additional
dispersion is needed as, for example, by hollow
waveguides (HWGs). Misoguti et al.  have shown
that dispersion of HWGs allows phase-matched
FWM also in the VUV spectral range. More recently
Tzankov et al.  presented energetic pulses in the
100 nJ range with durations of about 140 fs by mix-
ing the third harmonic (TH) and fundamental fre-
quency (FF) of a 1 kHz Ti:sapphire laser in an HWG.
FWM is not limited to the use of HWGs. In 
FWM process is performed in a femtosecond white-
light filament in argon yielding a tunable source of
few-cycle pulses in the visible wavelength range. Also
in a filament generated in neon Suzuki et al. pro-
duced 12 fs pulses at the TH of Ti:sapphire laser .
This was achieved by FWM between the FF and SH
of the laser.
State-of-the-art solid-state laser systems now offer
sub-50-fs pulses with several milijoules of energy at a
repetition rate in the kilohertz range with an almost
perfect beam profile. These properties allow using
higher-order processes for frequency conversion into
the VUV. Using a high-harmonic generation (HHG)
scheme, Kosma et al.  generated 11 fs pulses at
the fifth harmonic of a 1 kHz Ti:sapphire laser, but
the pulse energy was limited to only a few nano-
In this Letter we report the generation of femtosec-
ond VUV pulses by FWM of FF and TH of a 1 kHz
Ti:sapphire laser. The pump laser system provides up
to 3 mJ pulses at 800 nm with a duration of ?43 fs. A
schematic of the setup is shown in Fig 1. Using beam
splitter BS1, a small fraction of ?40 ?J is separated
as the probe for pulse measurement. A fraction of
66% of the pulse energy reflected by BS2 is used to
generate the TH in a tripling stage by SHG in a first
nonlinear crystal and successive combination with
the remaining FF in the second crystal for THG. SH
May 1, 2010 / Vol. 35, No. 9 / OPTICS LETTERS
0146-9592/10/091491-3/$15.00© 2010 Optical Society of America
is generated in a 0.5-mm-long BBO crystal cut at ?
=29° for Type I phase matching. A birefringent cal-
cite plate compensates for GVM between SH and re-
maining FF. A half-wave plate rotates the polariza-
tion of the FF before the THG in the second crystal.
THG is achieved by collinear SFG between the FF
and the SH in a 0.3-mm-long Type I BBO cut at ?
=44°. This setup generates up to 200 ?J at 266 nm
with a duration of ?100 fs measured by self-
The generated TH is collinearly combined with the
rest of the fundamental pulse transmitted by BS2
??800 ?J? and focused into an argon-filled gas cell.
The FF is focused with an f=100 cm BK7 lens, while
the focusing of the TH is accomplished independently
by a f=50 cm MgF2lens. Each of the two beams alone
generated a weakly glowing column in the cell. When
the beams overlapped spatially and temporally in the
gas cell, the length and brightness of the column sig-
nificantly increased. Under this condition, the gen-
eration of radiation at 160 nm with power of several
100 nJ energy can be observed, which is collinear
with the two generating beams. A typical spectrum,
acquired by a McPherson 0.2 m monochromator
optimized CCD camera (Andor D0420-BN-995), is
shown in Fig. 2. It has an FWHM of 1.12 nm, which
would correspond to a transform-limited pulse dura-
tion of 34 fs when Gaussian pulses are assumed.
The efficiency of the VUV generation process
strongly depends on the argon pressure inside the
chamber. We obtain the highest output pulse energy
of up to 240 nJ at a pressure of ?28 mbar with a fast
drop at lower pressures, while decreasing more
slowly toward higher pressures, as can be seen in
Fig. 3. Because phase matching is not possible, the
process can be optimized only by taking advantage of
the phase shift of Gaussian beams close to the focal
region (Gouy phase shift). Bjorklund  has shown
that the highest efficiency should be reached for
about ??b?k?opt??2, where b is the confocal parameter
−?k3?+k3?−k??. Using the Sellmeier coefficients for
argon from  and the confocal parameter of
?3.3 cm, calculated from the measured beam waist
radius of 65 ?m of the FF beam, we obtain ??b?k?opt?
=1.8 for our measured optimum pressure of 28 mbar,
which is in good agreement with the theory. Figure 3
also shows the theoretical curve for pressure depen-
dence as defined in  with the experimental values
b=3.3 cm, L=30 cm, f=15 cm, where L is the length
of the cell and f is the focus position relative to the
entrance window. The theoretical curve reproduces
the asymmetric shape and also the pressure for
maximum conversion efficiency quite well.
To separate the VUV pulses from the FF and TH
2n/?0and ?k is the phase mismatch ?k=k5?
Fig. 1. (Color online) Experimental setup for VUV genera-
tion by FWM. BS1, beam splitter (R?2%); BS2, beam split-
ter (R=66%); DL1, delay line for FF-TH temporal overlap;
DL2, delay line for pulse duration measurement; L1, lens
for FF (f=100 cm); CP, calcite compensation plate; HWP,
half-wave plate; L2, lens for TH (f=50 cm); M1, VUV colli-
mation mirror (r=150 cm); M2, VUV focusing mirror
(r=100 cm); M3, flat mirror for collinear combination with
probe beam; SM, VUV spectrometer; M4, movable mirror
for coupling into SM; W, MgF2window; TOF, time-of-flight
mass spectrometer; PM, VUV power meter (Startech
ated by FWM (black squares). The FWHM of a Gaussian fit
is 1.12 nm (solid curve), corresponding to a transform-
limited pulse duration of 34 fs when Gaussian pulses are
(Color online) Typical spectrum of the VUV gener-
Fig. 3. (Color online) Measured VUV energy depending on
the argon pressure (black squares) and the theoretical
curve simulated by using our experimental values (solid
curve). The optimum VUV energy is obtained at a pressure
of 28 mbar.
OPTICS LETTERS / Vol. 35, No. 9 / May 1, 2010
we used special dielectric high-reflecting VUV mir- Download full-text
rors (Layertec GmbH) that are also used for collima-
tion and focusing into our experimental chamber (see
Fig 1). The argon filled FWM chamber is isolated
from the experimental chamber by a MgF2window.
To keep the dispersion as small as possible the thick-
ness of the window is only 0.1 mm.
To determine the pulse duration we performed a
cross-correlation measurement in a xenon-filled time-
of-flight (TOF) mass spectrometer. The Xe pressure
was about 10−5mbar. The ionization potential of Xe
is 12.15 eV, and thus absorption of one VUV photon
?7.72 eV? and three FF photons ?1.55 eV? is necessary
for ionization. The small part of the FF reflected by
BS1 ??40 ?J? is focused collinearly with the VUV
pulse into the interaction region of the TOF. The de-
lay between FF and VUV is controlled by the motor-
ized delay line DL2. The ions are accelerated by a
voltage of 300 V and detected by a microsphere plate.
The measured ion rate served as a cross-correlation
In Fig 4 the cross-correlation trace between the FF
and the shortest VUV pulses is shown. The Gaussian
fit has a FWHM of 49 fs. Owing to the linear depen-
dence on the VUV intensity and cubic dependence on
the FF intensity using the relation ?vuv=??corr
a VUV pulse duration of 42 fs is obtained. This dura-
tion is similar to the FF duration but ?25% longer
than the transform limited pulse duration of 34 fs
supported by the VUV spectrum, as seen in Fig 2.
In comparison with near-resonant mixing pro-
cesses, where conversion efficiencies of about 1% is
possible, the achieved conversion efficiency of 0.15%
is considerably smaller. However, taking into account
the repetition rate of 1 kHz, we obtained a higher av-
erage power and with shorter output pulses. In fact,
the present conversion efficiency is similar to that in
HWG schemes [10,11].
In conclusion, we have demonstrated the genera-
tion of sub-50-fs pulses at the fifth harmonic of a
1 kHz Ti:sapphire laser system with energies up to
240 nJ. Since all processes are off-resonant, it should
be possible to produce even shorter pulses by exploit-
ing nonlinear pulse compression techniques prior to
the FWM process. Also, a tunable source in the VUV
spectral range by mixing the TH with the output of
an optical parametric amplifier can be realized.
1. I. V. Hertel and W. Radloff, Rep. Prog. Phys. 69, 1897
2. K. Kosma, S. A. Trushin, W. Fuss, and W. E. Schmid, J.
Phys. Chem. A 112, 7514 (2008).
3. S. A. Trushin, W. E. Schmid, and W. Fuß, Chem. Phys.
Lett. 468, 9 (2008).
4. V. Petrov, F. Rotermund, F. Noack, J. Ringling, O. Kit-
telmann, and R. Komatsu, IEEE J. Sel. Top. Quantum
Electron. 5, 1532 (1999).
5. T. Kanai, T. Kanda, T. Sekikawa, S. Watanabe, T. To-
gashi, C. Chen, C. Zhang, Z. Xu, and J. Wang, J. Opt.
Soc. Am. B 21, 370 (2004).
6. V. Petrov, F. Noack, D. Shen, F. Pan, G. Shen, X. Wang,
R. Komatsu, and V. Alex, Opt. Lett. 29, 373 (2004).
7. M. Wittmann, M. T. Wick, O. Steinkellner, P. Farma-
nara, V. Stert, W. Radloff, G. Korn, and I. V. Hertel,
Opt. Commun. 173, 323 (2000).
8. S. P. Le Blanc, Z. Qi, and R. Sauerbrey, Appl. Phys. B
61, 439 (1995).
9. J. F. Reintjes, Nonlinear Optical Parametric Processes
in Liquids and Gases (Academic, 1984).
10. L. Misoguti, S. Backus, C. G. Durfee, R. Bartels, M. M.
Murnane, and H. C. Kapteyn, Phys. Rev. Lett. 87,
11. P. Tzankov, O. Steinkellner, J. Zheng, M. Mero, W.
Freyer, A. Husakou, I. Babushkin, J. Herrmann, and F.
Noack, Opt. Express 15, 6389 (2007).
12. F. Théberge, N. Aközbek, W. Liu, A. Becker, and S. L.
Chin, Phys. Rev. Lett. 97, 023904 (2006).
13. T. Fuji, T. Horio, and T. Suzuki, Opt. Lett. 32, 2481
14. K. Kosma, S. A. Trushin, W. E. Schmid, and W. Fuß,
Opt. Lett. 33, 723 (2008).
15. G. C. Bjorklund, IEEE J. Quantum Electron. 11, 287
16. A. Börzsönyi, Z. Heiner, M. P. Kalashnikov, A. P.
Kovács, and K. Osvay, Appl. Opt. 47, 4856 (2008).
Fig. 4. (Color online) Cross-correlation trace of fundamen-
tal and VUV pulse (black squares) in Xe with the Gaussian
fit (solid curve) of 49 fs (FWHM).
May 1, 2010 / Vol. 35, No. 9 / OPTICS LETTERS