2922

OPTICS LETTERS / Vol. 35, No. 17 / September 1, 2010

Pressure-controlled phase matching to

third harmonic in Ar-filled
hollow-core photonic crystal fiber
J. Nold,1,2 P. Hlzer,1,3 N. Y. Joly,1,2 G. K. L. Wong,1 A. Nazarkin,1 A. Podlipensky,1
M. Scharrer,1 and P. St.J. Russell1,2
1

Max Planck Institute for the Science of Light, Gnther-Scharowsky-Strasse 1, 91058 Erlangen, Germany
2

Department of Physics, University of Erlangen-Nuremberg, Erlangen, Germany

School of Advanced Optical Technologies, University of Erlangen-Nuremberg, Erlangen, Germany

Received June 18, 2010; revised July 15, 2010; accepted July 19, 2010;
posted August 4, 2010 (Doc. ID 130373); published August 25, 2010

We report tunable third-harmonic generation (THG) in an Ar-filled hollow-core photonic crystal fiber, pumped by
broadband < 2 J, 30 fs pulses from an amplified Ti:sapphire laser system. The overall dispersion is precisely controlled by balancing the negative dielectric susceptibility of the waveguide against the positive susceptibility of the
gas. We demonstrate THG to a higher-order guided mode and show that the phase-matched UV wavelength is
tunable by adjusting the gas pressure. 2010 Optical Society of America
OCIS codes: 060.5295, 060.4370, 190.2620, 190.4223, 190.4370, 320.7110.

Many applications in solid-state physics, surface science,

biophysics, and chemistry require short pulses in the UV
spectral region [14]. Such pulses can be produced by
third-harmonic generation (THG) in a conventional gas
cell, where a pump pulse is strongly focused into the
gas, resulting in THG over the Rayleigh length. The conversion efficiency is limited by the short interaction
length, poor phase-matching, and by self-focusing and ionization within the interaction region [5]. In a gas-filled
hollow-core photonic crystal fiber (PCF), on the other
hand, tight field confinement, excellent transverse beam
quality, and almost unlimited interaction lengths greatly
enhance the nonlinear gaslight interaction [6]. Furthermore, the long path lengths allow one to operate at lower
intensities, thus avoiding self-focusing and ionization.
These features have resulted in applications such as ultralow-threshold stimulated Raman scattering [7] and the
transmission of high-power optical solitons [8]. For efficient THG, however, phase matching is a key factor as
yet unexplored in the context of gas-filled hollow-core
PCF. In this Letter, we report phase-matched THG from
800 to 266 nm in an Ar-filled kagom PCF by controlling
the pressure. Although phase matching can also be
achieved in capillaries [9], the leakage loss scales as
the inverse cube of the bore radius [10], reaching impractical levels (366 dB=m at 800 nm wavelength) for radii
equal to that of a typical hollow PCF core.
In our experiments, we used amplified 30 fs pump
pulses centered at 800 nm with a repetition rate of
1 kHz (i.e., 1 J pulses yield 1 mW average power). The
kagom PCF had a core diameter of 29:6 m {Fig. 1(a)
[11]}, and the pitch of the cladding lattice was 13:4 m,
with glass webs of thickness 250 nm. Its transmission
window extended from 750 nm to 1:2 m, with a loss
of 1:1 dB=m and a dispersion of 1:8 ps:nm1 km1 at the
pump wavelength [Fig. 1(b)]. A sample length of 20:5 cm
was used in all the experiments, both ends being
mounted in gas cells. It was filled by first evacuating both
the cells and fiber and then flooding one of the cells with
Ar. The pressure was adjustable from 0 to 10 bars. Diagnostic tools included a spectrometer and a mode profiler,
0146-9592/10/172922-03$15.00/0

both sensitive to the UV. Conversion efficiencies were

measured with a calibrated UV-sensitive power meter,
after spatially separating the pump and third harmonic
with a prism.
The effective refractive indices of the modes of the
Ar-filled kagom PCF can be approximated by the modes
of a hollow dielectric waveguide having the same core
area [12]:
s
u2
nmn n2Ar 2mn2
k a
s
p u2
p
u2
1 2mn2 1
mn
;
p0 k a
2p0 2k2 a2
1
where p is the gas pressure, p0 is atmospheric pressure,
k 2= is the vacuum wave vector, the vacuum
wavelength, a is the effective radius of the fiber core
[11], umn is the nth zero of a mth order Bessel function

Fig. 1. Experimental details. (a) Scanning electron micrograph of the kagom PCF structure. (b) Measured loss and dispersion spectra in the range 700 to 1300 nm. (c) Schematic of
the apparatus. The gray-shaded boxes are gas cells, and the
dotted rectangles are flip mirrors. The PCF is 20:5 cm long.
2010 Optical Society of America

September 1, 2010 / Vol. 35, No. 17 / OPTICS LETTERS

2923

of the first kind (m is the azimuthal and n the radial order), and is related to the wavelength dependence of
the dielectric constant of Ar:



T0
B1
B2

;
T 1 1 =2 1 2 =2

with T 0 273 K, B1 2:033229 104 , B2 3:445831

104 , 1 14:36 nm, 2 89:81 nm, and the wavelength
is in units of nanometers [13]. Note that Eq. (1) is valid
only if umn =ka2 1a good approximation for the
PCF used.
Phase matching between pump and third-harmonic
modes will occur when
n11 ; p nmn =3; p 0:

In Fig. 2 we plot solutions of Eq. (3) for different UV

mode orders. Within the ranges of wavelength and pressure considered, phase matching turns out to be possible
only if the third harmonic is in the HE13 mode, as seen in
Fig. 3, where combinations of pressure and wavelength
satisfying Eq. (3) are plotted against each other for
different core diameters. Note that the waveguide dispersion in the kagom PCF is stronger than in the large-core
capillaries commonly used in high-harmonic generation.
This means that higher gas pressures are required to
achieve phase matching, which has the beneficial effect
of increasing the nonlinearity. Figure 2 also shows that
phase matching to other higher-order modes in the UV
is possible at pressures higher than those accessible in
the present experiments.
The measured UV spectra are also shown in Fig. 3 as a
function of Ar pressure for two different pulse energies
(0:7 J and 1:3 J). For the PCF used in the experiment
(core radius 14:9 m), the model is in very good agreement with the observed third-harmonic wavelength [the
solid curve in Figs. 3(a) and 3(b)]. Conversion to the third
harmonic peaks at a pressure of 5:1 bars. Moreover, as
required by the phase-matching conditions, it is generated in the HE13 mode, as seen in the near-field image

Fig. 3. Third-harmonic spectrum as a function of Ar pressure

for two different pump pulse energies: (a) 0:7 J and (b) 1:3 J.
Also shown are theoretical phase-matching curves for different
core radii, the solid curve corresponding to a core radius of
14:9 m. (c) Measured and (d) calculated near-field mode profiles of the third harmonic at the fiber end face.

of the UV mode profile [Fig. 3(c)]. We also compare this

profile with that calculated by the finite-element solutions of the Maxwell equations for the actual fiber structure {Fig. 3(d) [14]}. Note that for lower pulse energies
[Fig. 3(a)], the optimum phase-matching wavelength,
within the bandwidth of the pump pulse, gradually shifts
to a longer wavelength as the pressure is increased.
Assuming the pump field (complex amplitude E 1 ) to be
in the HE11 mode and the third-harmonic field (amplitude
E 3 ) in a higher-order mode, the evolution of the harmonic
field for small conversion efficiencies is governed by
=z D3 E 3 i 3 E 31 iE 3 2i XP E 3 jE 1 j2 3 E 3 ;
4
where 3k1 SP jE 1 j2 k3 is the wave-vector
mismatch between pump and third harmonic, including
the effects of self-phase (SP) modulation at the pump
frequency. The nonlinear coupling constant 3 is proportional to G x 3; ; ; , where 3 is the third-order
nonlinear susceptibility and G is the modal overlap
defined by
R
3 ; 31 ; dd
R 2
G
;
5
3 ; dd
and 1 and 3 are the TE field distributions of the pump
and third-harmonic modes. In Eq. (4), XP is the crossphase (XP) modulation constant, 3 is the linear loss
coefficient at the third-harmonic frequency, and D3 is
an operator taking into account linear dispersion at
the harmonic frequency. If L 1 over the fiber length
L, ignoring the effects of SP and XP modulation, the harmonic spectrum can be written:

Fig. 2. Refractive index mismatch between pump (in the fundamental HE11 mode) and several different higher-order modes
at the third-harmonic frequency, for different values of gas pressure. The profile of jEj2 for each UV mode is also plotted, the
numbers indicating the pressure in bars. The step in pressure
between adjacent curves is always equal to 1 bar.

jF 3 j2 4 23 jIj2 23 2
1 ;

1
where v1
3 v1 is the group velocity walk-off rate and
is the deviation from the
R center-band third-harmonic
frequency. I 21 E 31 ei d is the Fourier
spectrum of the cube of the pump pulse field at the fiber

2924

OPTICS LETTERS / Vol. 35, No. 17 / September 1, 2010

loss. From Eq. (6), taking 3 7:8 1027 m2 =V2 =bar

[5], we estimate a conversion efficiency of 2:5 103 %
for a pump pulse energy of 1 J, which is in good agreement with the experimentally measured value in Fig. 4. If
the deleterious effects of group velocity walk-off and UV
attenuation could be eliminated, the conversion efficiency is predicted to be 5%.
In conclusion, the ability to achieve phase matching,
along with long interaction lengths, tight mode confinement, and high pulse energies (20 J can be routinely
launched without damage), seems likely to lead to enhanced gas-based nonlinear optical devices with many
potential applications.
Fig. 4. Measured third-harmonic power as a function of average pump power at a pressure of 5:1 bars. The full curve is
based on Eq. (6). The arrow marks the efficiency at a power
of 1 mW (pulse energy 1 J).

input. From Eq. (6), it follows that the harmonic spectrum

is modulated by a Lorentzian whose width and position
are determined by the phase mismatch, group velocity
walk-off, and attenuation. At higher-power levels far from
resonance, the nonlinear refractive index increase of the
gas is twice smaller at (SP modulation) than at 3 (XP
3
3
modulation), i.e., XP jE 1 j2 2 SP jE 1 j2 [1,15]. This means
that the Kerr nonlinearity will increase the phase mismatch between the fields, shifting the peak of the harmonic spectrum (for all pressures) toward a lower frequency,
as seen in Fig. 3. We attribute the observed reduced bandwidth of the frequency-upconverted spectra (at a fixed gas
pressure) to group velocity walk-off, estimated from Eq.
(1) to be 50 fs=cm.
Finally, we measured the conversion efficiency at a
pressure of 5:1 bars. As expected from a direct frequency
tripling process based on third-order nonlinearity, there
is a cubic relationship between the pump and thirdharmonic power [Fig. 4]. The rather low value of conversion efficiency (2:5 103 %) results from poor modal
overlap between pump and third harmonic (overlap integral G 0:023), UV attenuation, and group velocity walkoff (50 fs=cm). Cutback measurements at 260 nm yielded
3 dB=m for the fundamental mode, which agrees fairly
well with simple estimates based on the Fresnel reflection at the core boundary; the higher-order mode (with
a steeper incident angle on the core boundary) has an
estimated loss of 39 dB=m. The steady-state UV pulse
shape is governed by a leading edge that sits on the pump
pulse and a trailing edge that decays exponentially due to

Contact details for the authors are available at the

group website www.pcfiber.com.
References and Notes
1. C. G. Durfee, III, S. Backus, H. C. Kapteyn, and
M. M. Murnane, Opt. Lett. 24, 697 (1999).
2. N. A. Anderson, C. G. Durfee, III, M. M. Murnane, H. C.
Kapteyn, and R. J. Sension, Chem. Phys. Lett. 323,
365 (2000).
3. L. Berg and S. Skupin, Opt. Lett. 33, 750 (2008).
4. S. Ullrich, T. Schults, M. Z. Zgierski, and A. Stolow, J. Am.
Chem. Soc. 126, 2262 (2004).
5. J. F. Reintjes, Nonlinear Parametric Processes in Liquids
and Gases (Academic, 1984).
6. P. St.J. Russell, IEEE J. Lightwave Technol. 24, 4729 (2006).
7. F. Benabid, J. C. Knight, G. Antonopoulos, and P. St.J.
Russell, Science 298, 399 (2002).
8. D. G. Ouzounov, F. R. Ahmad, D. Mller, N. Venkataraman,
M. T. Gallagher, M. G. Thomas, J. Silcox, K. W. Koch, and
A. L. Gaeta, Science 301, 1702 (2003).
9. C. G. Durfee, III, S. Backus, M. M. Murnane, and H. C.
Kapteyn, Opt. Lett. 22, 1565 (1997).
10. M. J. Renn, R. Pastel, and H. J. Lewandowski, Phys. Rev.
Lett. 82, 1574 (1999).
11. The effective core diameter corresponds to a circle with the
same area as the hexagonal core. In the present case, the
distance between two diametrically opposite faces of
the hexagonal core is 27:9 m, yielding to an effective
circular core diameter of 29:6 m.
12. E. A. J. Marcatili and R. A. Schmeltzer, Bell Syst. Tech. J. 43,
1783 (1964).
13. A. Brzsnyi, Z. Heiner, M. P. Kalashnikov, A. P. Kovcs,
and K. Osvay, Appl. Opt. 47, 4856 (2008).
14. Code provided by JCMwave GmbH (www.jcmwave.com).
15. R. R. Alfano, P. L. Baldeck, P. P. Ho, and G. P. Agrawal, J.
Opt. Soc. Am. B 6, 824 (1989).