1120 OPTICS LETTERS / Vol. 29, No.

10 / May 15, 2004

Terahertz-pulse generation by photoionization of air

with laser pulses composed
of both fundamental and second-harmonic waves
Markus Kress, Torsten Löff ler, Susanne Eden, Mark Thomson, and Hartmut G. Roskos
Physikalisches Institut der Johann Wolfgang Goethe-Universität Frankfurt am Main, Robert-Mayer-Strasse 2-4,
D-60054 Frankfurt am Main, Germany

Received November 18, 2003

Intense radiation in the terahertz (THz) frequency range can be generated by focusing of an ultrashort laser
pulse composed of both a fundamental wave and its second-harmonic field into air, as reported previously
by Cook et al. [Opt. Lett. 25, 1210 (2000)]. We identify a threshold for THz generation that proves that
generation of a plasma is required and that the nonlinearity of air is insufficient to explain our measurements.
An additional THz field component generated in the type I b-barium borate crystal used for second-harmonic
generation has to be considered if one is to avoid misinterpretation of this kind of experiment. We conclude
with a comparison that shows that the plasma emitter is competitive with other state-of-the-art THz emitters.
© 2004 Optical Society of America
OCIS codes: 190.4380, 260.3090, 320.2250, 320.7120, 350.5440.

Cook and co-workers1,2 showed that eff icient tera- sin4 共a兲 dependence of the SHG eff iciency (in the ab-
hertz (THz) pulse generation is achieved when sence of saturation) is depicted by the solid curve in
laser pulses composed of a superposition of both Fig. 1(a).3
fundamental and second-harmonic (SH) spectral In the course of our investigation we found that
components are focused into air.1,2 The process of THz radiation is also generated in the BBO crystal
THz generation was initially attributed to four-wave itself. Figure 1(c) plots the net THz signal generated
rectification mediated by the third-order nonlinearity in the BBO crystal as a function of a. We measured
of air,1 but later Cook et al. found indications for a this value by suppressing THz generation at the beam
plasma-driven process.2 To elucidate the mechanism focus by blocking the optical beam behind the BBO
for THz generation further, we extended the experi- crystal with a THz-transparent polyethylene foil.
ments of Cook et al., and here we report conclusive THz-generation in the BBO crystal was identif ied
evidence that plasma generation is essential for THz as x 共2兲 -based optical rectification of the fundamental
emission from the focal region. wave as observed in other nonlinear crystals such as
In our experiments we employed a 1-kHz ZnTe.4 Model calculations5,6 based on the nonlinear
Ti:sapphire laser system (Clark CPA 2001), which dielectric tensor for BBO describe the rotation-angle
provided 150-fs laser pulses at 775 nm, with a dependence of the THz signal’s amplitude well [solid
maximum pulse energy of 860 mJ. Following the
approach of Cook et al., we focused the pulses through
a 100-mm-thick type I b-barium borate (BBO) crystal
that had been phase matched for SH generation (SHG)
and placed at a variable distance from the beam
waist in ambient air, where the THz radiation was
generated. Because of the different phase velocities
of the v and 2v f ields in air, the phase between the
two components at focus depends on the distance
between the BBO crystal and focus. We performed
the measurements with a relative phase optimized
for THz generation either by using an optimized
BBO-to-focus distance or by inserting and tilting
a thin quartz plate between the BBO crystal and
focus.1 The emitted THz radiation was collimated by
an off-axis paraboloid 共 f 苷 120 mm兲 and then focused
with a second one 共 f 苷 50 mm兲 onto a 1-mm-thick
ZnTe crystal for electro-optical detection.
We varied the angle between the polarizations of the
v and 2v beams by rotating the BBO crystal about the
beam axis. Figure 1(a) shows the dependence of SHG
efficiency on rotation angle a, def ined as the angle
between the fundamental polarization and the extra- Fig. 1. (a) SHG efficiency and detected THz signal from
ordinary axis of the BBO crystal (and hence the po- (b) the focus and (c) the BBO crystal as a function of crystal
larization of the 2v f ield). The theoretically expected rotation angle.

0146-9592/04/101120-03$15.00/0 © 2004 Optical Society of America

 May 15, 2004 / Vol. 29, No. 10 / OPTICS LETTERS 1121

curve in Fig. 1(c)]. The slight deviations that occur in
the regions of high SHG conversion eff iciency indicate
additional THz generation from the 2v radiation,
which was not included in the model calculations.
Figure 1(b) displays the net amplitude of the THz ra-
diation emitted from the focal region. The THz signal
from the BBO crystal, which unavoidably overlaps the
detected THz signal from the focus, was numerically
subtracted from the data. The THz signal from the
focus exhibits a different a dependence from the radi-
ation from the BBO crystal and vanishes at each mul-
tiple of 90±. This result can be understood by the fact
that, at 0± (180±), the SHG efficiency is zero, whereas
at 90± (270±) the polarizations of the optical f ields are
perpendicular to each other, in both cases no mixing
of the v wave with the 2v wave is possible. From
0± to 90± the THz emission exhibits a maximum at
an angle a 艐 55±. From an analysis7 that takes into
account the symmetry properties of the third-order
nonlinear susceptibility tensor of isotropic media, and
because the THz detector is sensitive only to THz radi-
ation with the same polarization as the probe beam, it
can be seen that the THz signal should be proportional
to the projection of the electric f ield vector of the 2v
wave onto that of the v wave. The resultant cos共a兲 fit
(weighted by the field strengths, as discussed below) is
illustrated by the solid curve in Fig. 1(b).
Although the maximum peak amplitude of the de-
tected THz radiation from the focus is .5 times larger
than that from the BBO crystal, the contribution of the
latter is generally not negligible, especially in light of
the distinct rotation-angle dependence of the two THz
signals. One observes that the THz signal from the
BBO vanishes at certain rotation angles, e.g., a 艐 37±.
Four-wave rectification predicts the following depen-
dence of the maximal THz amplitude on relative phase
w between the v and 2v waves7:
We measured the dependence of the THz signal as
a function of J0 (the fundamental pulse energy before
the BBO crystal), as shown by the f illed squares in
Fig. 3(a) for a BBO rotation angle of a 苷 37± and a fo-
cal length of f 苷 100 mm. Figures 3(b) and 3(c) show
expanded views of the data in the lower pulse-energy
range for f 苷 100 mm and f 苷 200 mm, respectively.
Using the relation E2v ~ Ev 2 ~ J0 in relation (1), one
expects a quadratic dependence, i.e., EdetTHz ~ x 共3兲 J0 2 .
The solid curves in Figs. 3(a) –3(c) display quadratic
fits to the mid-energy range of the experimental
data (for a focal length of 100 mm, this is the regime
J0 苷 100 250 mJ). At laser pulse energies below
100 mJ, the measured amplitudes lie below the fit
curves, and the amplitude values become unmeasur-
ably small below 35 mJ (100-mm focal length) and
70 mJ (200 mm). These energy values correspond
to intensities of ⬃1.5 3 1014 W兾cm2 , the ionization
threshold of air.7,8
In previous experiments with a dc bias applied to the
focal region we also found a threshold behavior that
we attributed to the onset of plasma formation.9,10 To
relate these results to the present ones, we modified
the setup described above by applying an additional dc

EdetTHz ~ ppl ~ x 共3兲 Ev 2 E2v sin共w兲 . (1)

Fig. 2. THz amplitude versus BBO-to-focus distance (with

One can change the phase by varying distance d focal length f 苷 200 mm).
between BBO crystal and focus to produce a phase
shift w 苷 v兾c共2n2v 2 nv 兲d, where nv and n2v are
the refractive indices of air at the fundamental and
the SH frequencies, respectively. To test the phase
dependence of relation (1) we measured the THz
amplitude versus distance d at a laser pulse energy
of 190 mJ (Fig. 2, filled squares). One can clearly
observe phase-change-induced oscillations as expected
from theory (solid curve). The linear decrease of the
envelope, incorporated ad hoc into the model calcula-
tion, arises from the change of the SHG conversion
efficiency because of the change in the beam spot size
in the BBO crystal.
In attempting to model the emission process in ac-
cordance with relation (1) under the assumption that
the nonlinearity arises purely from air, we failed to re-
produce the measured signal amplitude by 2 orders of
magnitude [using the value of x 共3兲 from the literature
data for electrically biased SHG in air7]. This failure
led us to test whether the observed THz emission is Fig. 3. THz amplitude versus laser pulse energy for vari-
linked to the formation of a plasma in the laser focus. ous sources (for details, see text).
1122 OPTICS LETTERS / Vol. 29, No. 10 / May 15, 2004
Fig. 4. Comparison of tabletop THz emitters. You et al.
refers to Ref. 12; Hamster et al., to Ref. 14.
bias field of 32 kV兾cm to the focal position. By chang-
ing the polarity of the bias we could separate the THz
signal generated by the static-field asymmetry (ex-
ternal-bias method) from the signal discussed above.
Figure 3 (circles) shows that the amplitude of the
external-bias THz signal scales linearly with J0 .
The linear-fit curves to the data cut the J0 axis at the
same values as observed for SH biasing. This is a
clear indication that in both cases plasma formation
is a prerequisite for generation of the observed THz
radiation.
The threshold behavior is in contrast to the obser-
vations of Ref. 1. A possible explanation for the dis-
crepancy is that the authors of Ref. 1 did not take
the additional THz signal component from the BBO
crystal into account. Figure 3(d) shows the J0 depen-
dence of this component measured for a crystal rotation
angle of a 苷 0±. The dependence was found to be
linear and without threshold [as expected for a x 共2兲 pro-
cess], and the THz amplitude at low J0 is of the same
order of magnitude as that of the signal from the focus.
Apparently, the threshold behavior of the signal from
the focus may well be masked by the signal from the
BBO crystal if the latter is not eliminated experimen-
tally or numerically.
Returning to the THz signal from the focus, we note
that for pulse energies greater than 250 mJ the THz
amplitude in Fig. 3(a) falls below the fitted quadratic
curve. At these high pulse energies, defocusing of the
laser beam by the plasma11 becomes important and
reduces the effective peak intensity. In addition, at
larger plasma volumes, phase mismatch and THz ab-
sorption effects are more likely to become significant.
Finally, we compare the plasma THz source with the
most powerful tabletop state-of-the-art THz emitters
(Fig. 4). To our knowledge, the strongest reported
THz emitter in this class is a system that employs a
large-aperture GaAs antenna, with a THz amplitude
of as much as 150 kV兾cm.12 Such a device employs
pulsed biasing (up to 10.7 kV兾cm) to prevent electrical
breakdown, which gives rise to high levels of ambient
noise that limit the usefulness of the technique.
With 1-kV兾cm dc bias we reach a THz amplitude of
10.5 kV兾cm (filled circles), although even this emitter
has recharging-related noise issues. In addition, all
semiconductor emitters exhibit saturation within this
range of pulse energies. A saturation limit is not
yet reached for the plasma sources discussed in this
paper. The SH-biased plasma emitter at present
reaches 6.8 kV兾cm [Fig. 4, squares; see also Fig. 3(a)]
and clearly is a competitive THz source, especially
as it is free from recharging noise. In contrast, the
external-bias plasma source in atmospheric-pressure
conditions (open diamonds) is inferior in terms of both
efficiency and noise performance.9,10
THz emission can also be obtained from laser-
generated plasmas without any form of bias but
relying on ponderomotive laser forces.13,14 The THz
amplitudes reached, however, are too low for prac-
tical purposes, if gas targets are used (triangles in
Fig. 4; amplitudes estimated from the experimental
parameters provided in Ref. 14). With a solid-state
target, the THz pulse energy can be raised by 3 orders
of magnitude (data not shown)13 but at the price of
significant x-ray emission.
This research was funded by the Gesellschaft
für Schwerionenforschung under contract OF-ROS.
M. Kress’s e-mail address is m.kress@physik.uni-
frankfurt.de.
References
1. D. J. Cook and R. M. Hochstrasser, Opt. Lett. 25, 1210
(2000).
2. D. J. Cook, J.-X. Chen, and R. M. Hochstrasser, in
Ultrafast Phenomena XII, T. Elsaesser S. Mukamel,
M. M. Murnane, and N. F. Scherer, eds. (Springer-
Verlag, Berlin, 2000), p. 197.
3. T. Löff ler, S. Eden, M. Thomson, and H. G. Roskos,
in Ultrafast Phenomena XIII, R. D. Miller M. M.
Murnane, N. F. Scherer, and A. M. Weiner, eds.
(Springer-Verlag, Berlin, 2002), p. 274.
4. B. B. Hu and X.-C. Zhang, Appl. Phys. Lett. 56, 506
(1990).
5. V. Dmitriev, G. Gurzadyan, and D. Nikogosyan, Hand-
book of Nonlinear Optical Crystals (Springer-Verlag,
Berlin, 1997).
6. A. Yariv, Quantum Electronics (Wiley, New York, 1989).
7. T. Löff ler, “Erzeugung intensiver Pulse im Tera-
hertzfrequenzbereich mittels laser-generierter Plas-
men,” Ph.D. dissertation (Universität Frankfurt,
Frankfurt, Germany, 2003), http://www.rz.uni-
frankfurt.de/piweb/femto/index.html.
8. S. Augst, D. D. Meyerhofer, D. Strickland, and S. L.
Chin, J. Opt. Soc. Am. B 8, 858 (1991).
9. T. Löff ler, F. Jacob, and H. G. Roskos, Appl. Phys. Lett.
77, 453 (2000).
10. T. Löff ler and H. G. Roskos, J. Appl. Phys. 91, 2611
(2002).
11. M. Mlejnek, E. M. Wright, and J. V. Moloney, Phys.
Rev. E 58, 4903 (1999).
12. D. You, R. R. Jones, and P. H. Bucksbaum, Opt. Lett.
18, 290 (1993).
13. H. Hamster, A. Sullivan, S. Gordon, and R. W. Falcone,
Phys. Rev. E 49, 671 (1994).
14. H. Hamster, “Generation of subpicosecond tera-
hertz radiation by laser-produced plasmas,” Ph.D.
dissertation (University of California, Berkeley,
Berkeley, Calif., 1993).