Proceedings of 2005 Particle Accelerator Conference, Knoxville, Tennessee

PRODUCTION OF TERAHERTZ SEED RADIATION FOR FEL/IFEL

MICROBUNCHERS FOR SECOND GENERATION PLASMA BEATWAVE
EXPERIMENTS AT NEPTUNE
J. E. Ralph, S. Ya. Tochitsky, C. Sung, C. Joshi, J. Rosenzweig
University of California at Los Angeles, Los Angeles, California
Abstract short bunches of electrons in the range of 100 fs that have
To achieve phase locked injection of short electron a direct correlation with the laser beatwave.
bunches in a plasma beatwave accelerator, the Neptune Simulations show that this method of microbunching
Laboratory will utilize microbunching in an FEL or IFEL requires a 10 MW pulse of 343 µm radiation as the pump
system. These systems require terahertz (THz) seed for the IFEL [7].
radiation on the order of 10 kW for the FEL and 10 MW A complimentary approach which is being pursued at
for the IFEL bunchers. We report results of experiments Neptune is the use of a free electron laser [7] with similar
on THz generation using nonlinear frequency mixing of bunching and synchronization. This approach has the
CO2 laser lines in GaAs. A two-wavelength laser beam advantage of decreasing the power requirements of the
was split and sent onto a 2.5 cm long GaAs crystal cut for THz radiation to approximately 10 kW. However the
noncollinear phase matching. Low power measurements electron beam current needed is in the range of 100 A for
achieved ~1 W of 340 µm radiation using 200 ns CO2 the FEL where as only test particles are needed for the
pump pulses with wavelengths 10.3µm and 10.6µm. We IFEL. Both approaches have the advantage of using the
also demonstrated tunability of difference frequency same undulator and require a THz source.
radiation, producing 240µm by mixing two different CO2
laser lines. By going to shorter laser pulses and higher BACKGROUND
intensities, we were able to increase the conversion According to theory, the coupled wave equations for
efficiency while decreasing the surface damage threshold. nonlinear optical processes can result in a new source
Using 200ps pulses we produced ~2 MW of 340 µm leading to difference frequency generation (DFG) where,
radiation. Future studies in this area will focus on ω1 − ω 2 = ω3 (1)
developing large diameter Quasi-Phase matched
structures for production of high power THz radiation Here ω1, ω2 and ω3 correspond to 10.3µm, 10.6 µm and
using collinear two frequency radiation. 340 µm respectively. To maximize the nonlinear effect
over a length much longer than the length by which the
INTRODUCTION waves become π/2 out of phase, phase matching is
One of the primary goals of the Neptune Laboratory at required.
r r r r
UCLA is the phase locked injection of relativistic k1 − k 2 − k 3 = ∆k (2)
electrons from the Neptune linac into a laser driven r
relativistic plasma beatwave[1]. Here the electric field of Phase matching occurs when the phase mismatch ∆k of
an amplitude modulated high intensity laser pulse the wave vectors goes to 0. Since GaAs is an isotropic
generates a space charge wave by longitudinally bunching crystal, with refractive indices different for k3, k2 and k1 ,
the plasma. The resultant longitudinal field can be used to therefore noncollinear phase matching was required for
accelerate electrons at a very high gradient up to 1 GeV/m matching in a bulk crystal.
for a plasma density of 1016 cm-1 [2,3].
At Neptune, the beat-wave generated by 10.3 µm and
10.6 µm produces a relativistic plasma wave with a
wavelength equal to 343 µm. The Neptune laser system is
currently capable of producing a 200 Joule 200ps laser
pulse on two wavelengths [4].
Measurements of accelerated electron spectra from both Figure 1: Noncollinear Phase Matching
resonant [3] and non-resonant [5] plasma beatwaves have
produced a 100% energy spread since the electron bunch Generation of THz radiation by mixing two CO2 laser
is much longer than the accelerating region of the plasma lines noncollinearly in GaAs was first demonstrated using
wave. two CO2 lasers in a cryogenically cooled GaAs sample in
One approach to phase locking microbunches is to drive experiments at MIT [8]. The anomalous dispersion of the
an inverse free electron laser (IFEL) as a buncher with crystal allowed for noncollinear phase-matching. GaAs
electromagnetic radiation at the difference frequency of presents a good choice for frequency mixing due to its
the two laser frequencies [6]. This approach offers very high damage threshold, high nonlinear coefficient of 50
pm/V as well its availability in large aperture high quality

c
0-7803-8859-3/05/$20.00 2005 IEEE 2780
 Proceedings of 2005 Particle Accelerator Conference, Knoxville, Tennessee

crystals. In this paper, we describe the results of our shown in figure 2. The profile of the beam was elliptical
study of generation of THz radiation using noncollinear as can be expected from noncollinear mixing.
phasematching in GaAs at room temperature.

MEASUREMENTS
Low Power Measurements
The goal of the low power measurements was to study
the phase matching conditions for noncollinear
configuration at room temperature in addition to
developing a THz seed radiation source for future FEL
microbuncher experiments.
The nonlinear crystal used in all experiments is a semi-
insulating large aperture GaAs crystal (2.0 cm × 4.0 cm ×
2.5 cm) shown below in Figure 2. To decrease THz
absorption, the sample used was Chromium doped which
reduces free carrier concentration; the crystal has a
specific resistivity greater than 3×108 ohm cm. To
maximize the nonlinearity, the crystal was cut so that the Figure 2: Beam profile measurements of the THz beam
polarization of the incident light is parallel to the [111]
axis by cutting a surface along (110) plane. Figure 3 is a plot of the angle scan measurements. The
For low power, we used a 1 Hz, two wavelength CO2 width of the angle scan corresponds to an effective
125 kw laser with 250ns FWHM pulses for pumping. The interaction length between 8 mm and 10 mm as shown.
beam was focused to 1 mm at the crystal. Note that for noncollinear configuration, the interaction
lenth is limited by the length over which the beams are
overlapped. The full width half maximum of the angle is
0.12 degrees. The optical setup used has an angle error of
0.011 degrees.
Measured Power
Calculated Power for 8mm interaction
Calculated Power for 10mm interaction

Data 1
1.2

Figure 1: Large aperture GaAs crystal 1

For noncollinear geometry, the phase matching angle is 0.8

found according to equation 3 below.
Power [arbitrary units]

0.6

k 2 + k 22 − k 32
θ phasematch = cos −1
( 1 ) (3)
2 k1k 2 0.4

Here the angle is the internal phasematching angle. The 0.2

external angle is found by applying Snell’s law. An
external phasematching angle of 2.24 degrees was
0
calculated. We measured a maximum in signal at the 2.2 2.25 2.3 2.35 2.4 2.45 2.5

angle 2.32 degrees. The wavelength of the newborn External Angle [degrees]
radiation was verified with a narrowband filter centered at
Figure 3: Angle Scan Data vs. Calculations
350µm with a FWHM of 2 cm-1. All THz signal
measurements were made using a Golay cell. This
The newborn beam was coupled to a 5 mm radius
measurement produced 2 watts of difference frequency
copper waveguide by through a horn. No attenuation
radiation at 343µm in a 250ns pulse. Additional losses were measured over a length of 1 meter. Total
measurements mixing a different pair of lines produced losses were dominated by the coupling efficiency of the
newborn radiation at 240 µm with roughly the same horn. This demonstrates the feasability of guiding over an
efficiency at room temperature. extended distance which is required for the meter or
Measurements of the profile were taken 50 mm from longer FEL undulator.
the back surface of the GaAs crystal using a Golay cell
mounted on two translation stages and fitted with a 2.5 High Power Measurements
mm iris. The beam profile of the newborn radiation is The goal of high power measurements was to
demonstrate the feasibility of generating high power THz

2781 c
0-7803-8859-3/05/$20.00 2005 IEEE
 Proceedings of 2005 Particle Accelerator Conference, Knoxville, Tennessee

radiation which can be used to seed an IFEL contained both 10.6 µm and 10.3 µm components,
microbuncher. however only one wavelength in each arms was used for
the DFG process of interest. The pump beam was a
collimated beam as summarized in Table 1 below. We see
an increase in conversion efficiency as is implied by
equation 2 due to the increase in pump intensity.

Table 1: Pump Specifications

Pulse Intensity Pulse Spot Size THz
Repetition Length Power
rate
1 Hz 5mW 200ns 1mm 2W
1 shot / 5 ~1GW/c 250ps 20 mm × 2 MW
minutes m2 30 mm

For the beam size we had, the effective interaction

length of the two beams was the full length of the crystal.
Extending this length has the effect of increasing
Figure 4: Schematic of High Power Optical Setup
conversion efficiency, however it decreases the angle
tolerances for noncollinear geometry. The full width half
The optical setup shown in figure 4 used for these
maximum of the angle scan for high power is only .06
measurements was designed to provide greater than 0.01
degrees. By going to a large aperture crystal of shorter
degree angle accuracy and is similar to the setup used for
length, similar high power can be achieved with about the
low power measurements. The two wavelength pulse was
same tolerance as in the low power case.
split using a 60 degree NaCl beam splitter since at high
power using a grating is difficult due to damage threshold.
By varying the distance d between the two irises, very CONCLUSIONS AND FUTURE PLANS
good precision was achieved. Since the angle θ is equal FOR THZ GENERATION AT NEPTUNE
to tan-1(d / pathlength), having a pathlength longer than 3 The measurements outlined above clearly show the
m resulted in very good precision. Additionally, to potential for achieving the THz power requirements of
achieve maximum interaction between the short pulses, both the IFEL and FEL bunching schemes. By increasing
the path length difference between the arms was adjusted the pump and increasing the effective length, a 1 Hz 10
to less than 5 mm. kW FEL seed source will be possible. One method for
The Neptune laser system can generate high power THz increasing the interaction length in the crystal for both
radiation owing to three main factors. First, the Neptune low power and high power is to use a quasi-phase
CO2 laser produces short pulses on the order of 200 ps. matched structure which will allow collinear phase
Measurements performed at Neptune demonstrated a matching over an extended length. This possibility is
surface damage threshold for GaAs of 30MW/cm2 for being explored.
long 200ns pulses. However by going to 200ps pulses, Work supported by the U.S. Department of Energy
the damage threshold increased to roughly 2GW/cm2. under Contract No. DE-FG03-92ER40727.
Second, the laser system produces a beam up to 6
inches in diameter. This entire beam can be used for REFERENCES
difference frequency generation. Third, due to the fact that
very large high quality GaAs crystals are readily available [1] S. Ya Tochitsky, P. Musumecci, C. E. Clayton, C.
from the semiconductor industry, the ability to produce Pellegrini, J. B. Rozenzweig and C. Joshi, Advances
very high power THz radiation is a matter of scaling. Accelerator Concepts 2002, AIP Conference
Limitations of power are imposed only by the damage Proceedings 647, Melville, New York, pp.786-795
threshold of GaAs. Power scaling is governed by the [2] C. Joshi et. al., Nature, 311, p.529(1984)
equation 4 assuming all beams are the same size. [3] S. Ya. Tochitsky et. al., Phys. Rev. Lett. 92, 9(2004)
2 [4] S. Ya. Tochitsky et. al., Opt. Lett. 24,1717(1999)
ω3 d eff [5] C. V. Filip et. al. , Phys. Rev. E. 69, (2004)
P344 = P10.60 P10.28leff2 (4) [6] P. Musumecci et. al., Proceedings of the International
c 3ε 0 n1n2 n3 Conference on Lasers 2001, edited by V. J.
This equation assumes no absorption and perfect phase Corcoran, STS Press, McLean, VA, 2002, pp.41-48
matching. Using the crystal described above, we [7] C. Sung et. al., (This Proceedings)
measured 2 MW of THz radiation which agrees within a [8] R. L. Aggraval, B. Lax and G. Favrot, Appl. Phys.
factor of 2 with calculations after accounting for Fresnel Lett., Vol. 22, No 7, 1973
losses on all surfaces and 6 cm2 pump. Each arm

c
0-7803-8859-3/05/$20.00 2005 IEEE 2782