Stable Charged-Particle Acceleration and Focusing in A Laser Accelerator Using Spatial Harmonics
Stable Charged-Particle Acceleration and Focusing in A Laser Accelerator Using Spatial Harmonics
yE
0
c
2
X
n
z
o
n
z
k
n
z
n
z
e
|k
n
z
zwi
.
One nal simplifying assumption is that all of the ampli-
tudes o
n
z
are real, which can always be arranged for a
dielectric lattice having both spatial-inversion and time-
reversal symmetries. Without loss of generality, set o
n
0
1.
For an electron traveling at approximately the resonant
phase velocity, we substitute [ z
n
0
i and dene w
n
0
2rw,k
n
0
J so that the Lorentz force FeE
n
0
^ zB
may be written
E
|
x, i Re
X
q
]
|,q
xe
|qw
n
0
i
, (2)
where the sum is over all integers q and
]
y,q
yeE
0
o
n
0
q
k
n
0
q
1
n
0
n
0
q
e
|k
n
0
q
[
,
]
z,q
|eE
0
o
n
0
q
e
|k
n
0
q
[
.
Several observations are made of the Lorentz force. First,
the resonant spatial harmonic q 0 is considered. Dening
the particle phase k
n
0
[, the transverse and longitudinal
contributions are yeE
0
k
n
0
,
2
n
0
cos and eE
0
sin,
respectively. For stable longitudinal acceleration, which
requires r,2 <<0, the transverse force is defocusing.
Second, the nonresonant spatial harmonics q 0 is consid-
ered. The transverse components vanish for
n
0
n
0
q
c
2
and are otherwise proportional to k
n
0
q
2
n
0
k
n
0
.
Following the analysis of Ref. [11], the electron motion
is separated into a slowly varying secular component
X
|
i and a rapidly varying component (
|
i, x
|
i
X
|
i (
|
i, so that the secular equations of motion for
a force having the form of Eq. (2) are
m
e
12u
|,z
n
0
X
|
Re]
|,0
X
X
]
X
q0
1
2m
e
12u
],z
n
0
q
2
w
2
n
0
Re
]
|,q
x
]
]
],q
]
],q
X
.
Adopting the beam physics convention of using z as
the independent variable and introducing the notation
0
J,Jz 1,
n
0
J,Ji,
Y
00
Y
n
n
0
k
2
n
0
3
n
0
2
n
0
cos
n
2
n
0
w
2
2
2
n
0
4
n
0
w
2
n
0
B D
C E cos2
, (3)
where
n
n
0
eE
0
k
n
0
m
e
c
2
eE
0
n
0
2rm
e
c
2
e >0,
B
X
q0
o
2
n
0
q
k
2
n
0
q
q
2
1
n
0
n
0
q
2
,
C
X
q0
o
n
0
q
o
n
0
q
k
n
0
q
k
n
0
q
q
2
1
n
0
n
0
q
1
n
0
n
0
q
,
D
1
2
n
0
X
q0
o
2
n
0
q
k
2
n
0
q
1
n
0
n
0
q
q
2
,
E
1
2
n
0
X
q0
o
n
0
q
o
n
0
q
k
2
n
0
q
1
n
0
n
0
q
q
2
.
The transverse secular equation of motion, Eq. (3), yields the
criteria for simultaneous acceleration and stabilization.
The term linear in E
0
is the resonant defocusing term, and
the term quadratic in E
0
is the average effect of the rapidly
oscillating nonresonant spatial harmonics. More specically,
the terms B and C are due to the rapid transverse motion,
whereas the terms D and E are due to the rapid longitudinal
motion. The direct terms B and Dare unconditionally focus-
ing, whereas the interference terms C and E are either
focusing or defocusing, depending on the particle phase .
These interference terms result from a nonresonant spatial
harmonic driving the rapid motion, while an oppositely
traveling wave, relative to the synchronous frame, is sampled
by the averaging procedure.
Because the ponderomotive focusing terms in Eq. (3) are
proportional to 1,q
2
w
2
n
0
, the focusing strength can be sig-
nicantly increased by reducing w
n
0
2r
n
0
,J. Consider,
at rst, a photonic accelerator whose longitudinal periodicity
PRL 109, 164803 (2012)
P HYS I CAL RE VI E W L E T T E RS
week ending
19 OCTOBER 2012
164803-2
J is of the same order as the photonic lattice constant o. As
useful photonic band gaps typically lie in the frequency
range wo,2rc o, $1, one has w
n
0
$
n
0
w.
Therefore, the frequencies of the rapidly varying forces
are in the optical range, and the ponderomotive focusing
force remains correspondingly weak. If, instead, we con-
sider a photonic accelerator whose smallest repeating
length (i.e., the primitive lattice constant) along the longi-
tudinal direction is J Na, then w
n
0
$
n
0
w,N and the
ponderomotive focusing force grows correspondingly
strong. We specify that J is then primitive so that both of
the q 1 spatial harmonics do not identically vanish by
symmetry.
Breakdown ultimately limits the peak axial eld that can
be used. For the accelerator eigenmodes under considera-
tion, the peak axial eld is typically larger than the peak
dielectric eld, roughly E
peak axial
,E
peak dielect
$1.5. The
breakdown electric eld for a pulse propagating in a dielec-
tric [12] is E
peak dielect
E
th
,rce
0
p
, where E
th
is the ma-
terials damage threshold and r is the pulse length. Due to
its high dielectric coefcient and compatibility with micro-
fabrication techniques, we are strongly considering silicon
at 5 m. The damage threshold has yet to be measured in
silicon at 5 m for & ps pulses, but Soong et al. [13]
measured E
th
0.35 J,cm
2
at 2.2 m for 1 ps pulses. At
5 m, multiphoton ionization will be weaker, but tunnel
ionization will be stronger. Using this value of E
th
and
r 80 fs gives E
peak axial
6 GV,m. Other promising
materials in the infrared are sapphire and quartz, which
have more than an order-of-magnitude improvement in E
th
over silicon, at the cost of a smaller dielectric constant.
Consider the case where the only nonvanishing nonre-
sonant spatial harmonic is q 1, which provides two
simplications. First, it avoids interference terms in the
transverse secular equation of motion, so that C E 0
in Eq. (3). Second, taking the nonresonant spatial harmonic
to be slower (q >0) than the resonant spatial harmonics
avoids the mutual cancellation of the transverse electric
and magnetic forces at
n
0
n
0
q
c
2
. Comparing the
resonant defocusing forces maximum value at 0
with the nonresonant focusing forces minimum value (in
this case, the focusing force does not depend on ) yields
the transverse stability diagram shown in Fig. 1.
Having identied the transverse stability condition, we now
concentrate on longitudinal dynamics. Examining briey only
the effect of the resonant wave, we dene the momentum
deviation u
z
n
0
, where
n
0
n
0
n
0
m
e
c. Then, we
write the electrons small-amplitude longitudinal
Hamiltonian as [14]
H[, u % m
e
c
2
2
n
0
u
2
2
n
0
2
n
0
n
n
0
cossin
0
,
where we have allowed for an accelerating bucket by
setting the synchronous phase
0
<0. As this
Hamiltonian is a constant of the motion, one may deduce
the dimensions of the stable bucket in longitudinal phase
space.
To examine the dynamics in the presence of nonresonant
harmonics, the equation of motion is numerically solved as
Ju,
n
0
J
~
i
n
n
0
k
n
0
J
2
n
0
n
0
sinsin
0
X
q0
o
n
0
q
sink
n
0
q
,k
n
0
2rq
~
i
J
J
~
i
k
n
0
J
2
n
0
u,
n
0
, (4)
where
~
i w
n
0
i,2r is a dimensionless time variable nor-
malized so that
~
i 1 corresponds to a single period of the
micromotions lowest frequency components (q 1).
Two example cases,
n
0
6 and
n
0
1000, with
stable acceleration parameters, are shown in Table I.
Simultaneously integrating both the longitudinal equation
of motion, Eq. (4), and the transverse equation of motion,
m
e
y E
y
, both of these cases were veried to be stable.
Figure 2 shows longitudinal phase-space trajectories for
the
n
0
6 case. Even with such large amplitude perturb-
ing waves, a stable bucket structure is observed. Though
1 10 100 1000 10000
10
100
1000
100 MV/m
200 MV/m
400 MV/m
1 GV/m
2 GV/m
4 GV/m
FIG. 1 (color online). Stability diagram for a single nonreso-
nant spatial harmonic o
n
0
1
0. Increasing the slow modulation
length N and increasing the particle energy
n
0
improves trans-
verse stability and gives higher values of the accelerating eld E
0
for a xed net eld E
0
1 jo
n
0
1
j 6 GV,m. Locations of the
two example cases discussed in the text are shown with dot
markers. The region below the dotted line (, 0.03) is longi-
tudinally unstable. (o 1.15 m and 5 m).
TABLE I. Low- and high-energy example cases yielding
transverse and longitudinal stability with accelerating gradients
!1 GeV,m. [N 250 and E
0
1 o
n
0
1
6 GV,m].
n
0
E
0
(GV,m) o
n
0
1
E
0
(GV,m) GeV,m
6 1.414 4.586 1.0
1000 3.536 2.464 2.5
PRL 109, 164803 (2012)
P HYS I CAL RE VI E W L E T T E RS
week ending
19 OCTOBER 2012
164803-3
increasing N improves transverse stability, it also moves
the nonresonant buckets closer to the resonant buckets.
When the border regions of the two buckets overlap, the
trajectories become longitudinally unstable according to
the Chirikov criterion [15]. For instance, in Fig. 2, the
nonresonant buckets travel towards the left and are located
at u,
n
0
0.33, far enough below the resonant buckets
to not destabilize the trajectories. An approximately
equivalent longitudinal stability criterion,
q
2r
k
n
0
eE
0
o
n
0
q
2rm
3
n
0
q
2
w
2
n
0
,,
is shown in Fig. 1, where , has been determined via
comparison with numerical results.
We now give a brief example of a band-gap-conned
waveguide accelerator that provides nonresonant focusing.
It consists of a linear channel and a triangular lattice of holes
etched into silicon (see Fig. 3 and movies in Supplemental
Material [16]). The channel width is modulated over a
distance of J 8o, where o is the lattice constant. This
provides a mix of spatial harmonics, particularly a slow
sideband and a fast sideband. The beam is synchronized to
the fast sideband, labeled q 0 in Fig. 4(a). Then the q 1
spatial harmonic provides the majority of nonresonant focus-
ing, as shown by the spatial harmonic spectrum in Fig. 4(b).
To design for large N, we treat the slow modulation as a
cyclic adiabatic variation [17]. This 2D design can be gen-
eralized to a 3D design incorporating either a single nite
slab or a stack of nite slabs. In 3D, the mode is conned in
the plane by total internal reection, provided that the spatial
-0.40 -0.35 -0.30 -0.25 -0.20 -0.15 -0.10 -0.05 0.00
-0.10
-0.05
0.00
0.05
0.10
FIG. 2 (color online). Longitudinal phase-space trajectories
for a bucket having acceleration gradient 1 GeV,m being per-
turbed by a strong nonresonant wave. The blue orbits are
obtained by averaging the phase space motion over a single
period
~
i 1. The dotted line is the raw phase-space motion for
the outermost blue orbit, showing the rather violent nonresonant
longitudinal motion that provides approximately 40% of the
transverse stability in this case. The black line is the separatrix
obtained from the unperturbed Hamiltonian. [N 250,
n
0
6,
E
n
0
2
p
GV,m, E
0
1 o
n
0
1
6 GV,m,
0
r,4].
FIG. 3 (color online). Accelerator eigenmode simultaneously
providing both transverse focusing and longitudinal acceleration.
Arrows indicate electric eld, and color shading indicates mag-
netic eld in silicon. Unshaded channel in center and holes
represent vacuum. Slow modulation of the waveguide ngers
(N 8) produces two sidebands about the unmodulated wave-
guide mode. Beam is synchronized to the fast sideband. Here, we
take the design particles to be positrons, so that the synchronous
phase
0
is 3r,4. See movies in Supplemental Material [16].
(o 1.15 m and 5 m)
-0.5 0.0 0.5 1.0
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
(a)
-20 -15 -10 -5 0 5 10 15 20
0.0
0.5
1.0
1.5
2.0
2.5
3.0
(b)
FIG. 4 (color online). Spectral properties of the accelerator
eigenmode shown in Fig. 3. (a) Band diagram. Three spatial
harmonics are shown with thick lines. The fast sideband, q 0,
is the resonant spatial harmonic and is approximately speed-of-
light. The strongest nonresonant harmonic, q 1, is slightly
slower and provides the majority of the transverse focusing.
(b) Spatial harmonic spectrum. The strength of the two side-
bands, q 0 and q 2, relative to the unmodulated spatial
harmonic, q 1, can be increased by increasing the depth of the
slow modulation.
PRL 109, 164803 (2012)
P HYS I CAL RE VI E W L E T T E RS
week ending
19 OCTOBER 2012
164803-4
harmonics are mostly underneath the light cone. For trans-
verse particle connement in the x direction, the mode can be
adjusted so that it is resonantly focusing.
The approach presented here thus provides simultaneous
transverse and longitudinal stability, as shown by relevant
examples. The application of these results now takes on a
certain urgency, as it may be essential in guiding LSA
structure designs in numerous research programs, includ-
ing an all-optical free-electron laser, with both accelerator
and undulator [18] driven by lasers.
This work is supported by DARPA under Contract
No. N66001-11-1-4197, the U.S. DOE under Contracts
No. DE-FG02-07ER46272 and No. DE-FG03-92ER40693,
and the ONR under Contract No. N00014-06-1-0925. B. N.
and S. P. wish to thank DARPA MTO for support under the
SERS program, covering work on photonic accelerators.
[1] I. Blumenfeld, C. E. Clayton, F.-J. Decker, M. J. Hogan, C.
Huang, R. Ischebeck, R. Iverson, C. Joshi, T. Katsouleas,
N. Kirby et al., Nature (London) 445, 741 (2007).
[2] W. P. Leemans, B. Nagler, A. J. Gonsalves, C. Toth, K.
Nakamura, C. G. R. Geddes, E. Esarey, C. B. Schroeder,
and S. M. Hooker, Nat. Phys. 2, 696 (2006).
[3] X. E. Lin, Phys. Rev. ST Accel. Beams 4, 051301 (2001).
[4] Z. Zhang, S. G. Tantawi, and R. D. Ruth, Phys. Rev. ST
Accel. Beams 8, 071302 (2005).
[5] B. M. Cowan, Phys. Rev. ST Accel. Beams 11, 011301
(2008).
[6] J. Rosenzweig, A. Murokh, and C. Pellegrini, Phys. Rev.
Lett. 74, 2467 (1995).
[7] A. Tremaine, J. Rosenzweig, and P. Schoessow, Phys. Rev.
E 56, 7204 (1997).
[8] M. Xie, in Proceedings of the European Particle
Accelerator Conference (JACOW, Vienna, 2000), p. 895.
[9] J. Rosenzweig and L. Serani, Phys. Rev. E 49, 1599
(1994).
[10] J. D. Joannopoulos, S. G. Johnson, J. N. Winn, and
R. D. Meade, Photonic Crystals: Molding the Flow of
Light (Princeton University Press, Princeton, NJ, 2008),
2nd ed.
[11] L. D. Landau and E. M. Lifshitz, Mechanics, Course of
Theoretical Physics Vol. 1 (Pergamon Press, Oxford,
1976), 3rd ed.
[12] C. McGuinness, E. Colby, and R. L. Byer, J. Mod. Opt. 56,
2142 (2009).
[13] K. Soong, R. L. Byer, C. McGuinness, and E. Peralta, in
Proc. of 2011 Particle Accelerator Conf. (IEEE, New
York, 2011), p. 277.
[14] J. Rosenzweig, Fundamentals of Beam Physics (Oxford
University Press, New York, 2003).
[15] B. V. Chirikov, J. Nucl. Energy, Part C 1, 253 (1960).
[16] See Supplemental Material at http://link.aps.org/
supplemental/10.1103/PhysRevLett.109.164803 for mov-
ies of accelerator eigenmode.
[17] S. G. Johnson, P. Bienstman, M. A. Skorobogatiy, M.
Ibanescu, E. Lidorikis, and J. D. Joannopoulos, Phys.
Rev. E 66, 066608 (2002).
[18] T. Plettner and R. L. Byer, Phys. Rev. STAccel. Beams 11,
030704 (2008).
PRL 109, 164803 (2012)
P HYS I CAL RE VI E W L E T T E RS
week ending
19 OCTOBER 2012
164803-5