Professional Documents
Culture Documents
,2U
q
, I
the pon-
deromotive energy. The recent two experiments by Blaga
et al. [10] and Quan et al. [11] on the photoionization of
atoms and molecules in strong midinfrared laser elds
reveal a previously unexpected characteristic spikelike
low-energy structure (LES) in the energy distribution of
electrons emitted along the laser polarization direction, see
Fig. 1(a). These observations manifest a striking contrast to
the prediction of the SFA or the SFA with Coulomb cor-
rections [12] and point to a lack of complete understanding
of strong eld physics. Varying the laser polarization from
linear to circular, LES is signicantly reduced. The latter
indicates that forward rescattering [13] is playing an es-
sential role in this process which has been pointed out by
Blaga et al. [10] and Faisal [14]. However, the mechanism
which creates the LES remains obscure. Many questions
remain unanswered. How exactly does the LES arise? Why
does it have a peaked structure? Why is the effect of
rescattering more pronounced in midinfrared laser elds?
In this Letter, we investigate in detail and identify the
mechanism of LES. We employ the classical-trajectory
Monte Carlo (CTMC) method with tunneling and the
Coulomb eld of the atomic core fully taken into account.
In addition, we provide a qualitative theoretical estimation
for the Coulomb eld effects: initial Coulomb focusing
(CF), multiple forward scattering, and asymptotic CF. We
quantify their relative role in the electron dynamics and
conclude that (1) the behavior of the transverse (with
respect to the laser polarization direction) momentum
change of the electron due to Coulomb eld effects with
respect to the ionization phase is the key for understanding
of the LES, and (2) at midinfrared wavelengths, multi-
ple scattering of the ionized electron plays a decisive
FIG. 1 (color online). Photoelectron spectra (PES): (a) The
experimental result (squares) for a Xenon atom in a laser
eld with peak intensity I
0
3.2 10
13
W,cm
2
and wave-
length 2.3 m [10] as well as the corresponding
CTMC simulation (circles). (b) CTMC simulations for a hydro-
gen atom with 2 m, I
0
9.0 10
13
W,cm
2
: exact
(circles); with totally neglecting the Coulomb potential (NCP)
(stars); NCP only for the electron longitudinal momenta P
k
(squares) and NCP only for the transverse momenta P
?
(triangles). The high-energy limit of LES dened by the break
in slope is indicated with an arrow.
PRL 105, 113003 (2010)
P HYS I CAL RE VI E W L E T T E RS
week ending
10 SEPTEMBER 2010
0031-9007,10,105(11),113003(4) 113003-1 2010 The American Physical Society
nonperturbative role. In particular, the transverse momen-
tum change of the electron due to multiple scattering
distorts the electron phase space to create peaks at low
electron energies, while the longitudinal momentum
change due to initial CF shifts the peak energy to higher
energies. We investigate the scaling of the LES parameters.
In the (1 regime, the electron oscillation amplitude
in the laser eld n E
0
,w
2
exceeds the distance of the
electron from the atomic core at the tunnel exit z
0
I
,E
0
: n,z
0
$2,
2
)1 and the transversal distance
traveled by the electron during one laser period x
0
$
2r
E
0
p
,2I
1,4
w: n,x
0
$1,2r
2I
,w
q
)1, where
E
0
and w are the laser eld amplitude and frequency,
respectively (atomic units are used throughout).
Therefore, in this regime, the electron travels far from
the core during its oscillation in the laser eld, and the
Coulomb eld distorts the electron trajectory only at posi-
tions very close to the core. This happens immediately
after ionization, corresponding to the initial CF, and
when the electron revisits the atomic core and rescatters.
The number of scatterings is large for midinfrared
wavelengths and low-energy photoelectrons [15]: N
s
$
n,x
0
$10 at the parameters in [10]. The third Coulomb
effect is the asymptotic CF when the electron momentumis
disturbed by the Coulomb eld after the laser pulse is
switched off. This usually plays an important role for
low-energy photoelectrons [7,16,17] but is not essential
for the LES as shown below.
In our 3D CTMC simulation, the ionized electron wave
packet is formed according to the Ammosov-Delone-
Krainov (ADK) ionization rate [18] and further propagates
classically. The electrons are born at the tunnel exit with
the following conditions: (i) Along the laser polarization
direction, the initial position z
|
is derived fromthe effective
potential theory [19] and the initial momentum
|k
0
[20]. (ii) The transversal coordinates are x
|
y
|
0. The
transverse momentum
|?
follows the corresponding ADK
distribution [18]. The transverse momentum components
are
|x
|?
sin and
|y
|?
cos, where is the
azimuthal angle randomly distributed within an interval of
(0, 2r). The positions and momenta of electrons after the
laser pulse are used to calculate the nal asymptotic
momenta [17] at the detector. Only electrons emitted along
the laser polarization direction within an angle of 2.5
|
, the phase
space of electrons contributing to the LES (per unit
|
) has
local maxima. This can be tested via articially discarding
the electrons between certain characteristic phases, see
Fig. 2(c). The latter shows that the occurrence of the high-
est part of LES is, in fact, related to the electrons with
ionization phases within an interval (1.64, 1.77) exactly
corresponding to the slope change between B and D. The
behavior of
|?
versus the ionization phase, in fact, reects
the one of the transverse momentum change due to the
Coulomb eld (u
?
), see Fig. 4(a) below. Then, one needs
to explain u
?
nonmonotonic behavior with respect to
|
.
Third, we investigate the transverse and longitudinal
momentum disturbance of the electron (u
?
and u
k
)
due to multiple scattering, initial and asymptotic CF for
electrons with the maximal probability for each phase
[circles in Fig. 2(a)]. (i) The transverse momentum change
due to Coulomb potential Vr at the sth scattering can be
estimated as u
s
?
%
R
r
?
VriJi $
s
,r
3
s
ui
s
, where
r
s
is the distance from the core at the scattering moment,
s
is the one in the transversal plane, and ui
s
is the scattering
duration. When the electronvelocity
s
at scattering is large,
ui
s
$2r
s
,
s
. In the opposite case, ui
s
$2
2r
s
,jE
s
j
p
is
determined by the laser eld E
s
at scattering.
Accordingly, u
s
?
2
s
,r
2
s
s
, if
2
s
)r
s
jE
s
j, or
u
s
?
2
3,2
s
,
r
5
s
jE
s
j
q
, otherwise. u
s
?
is sensitive
to r
s
and
s
. Their values for the different scattering events
are shown in Figs. 3(c) and 3(d). As for the formula of the
u
s
k
,
s
in the expression for u
s
?
should be substituted by
the absolute value of the scattering coordinate jz
s
j along the
laser polarization direction. (ii) The momentum change due
to initial and asymptotic CF are estimated numerically.
In Fig. 3(e), the relative contribution of multiple scat-
tering (up to fourth scattering) and initial CF to u
?
is
shown calculated using the above formulas. At >1.77
(line D), mainly rst scattering and initial CF plays a role.
With a little shift of phase towards 1.68 (line C), the second
scattering enhances rapidly and the rst scattering weak-
ens. The signicant contribution fromthe second scattering
results in the slope change between lines C and D in
Fig. 2(b). Decreasing the phase further, the third scattering
becomes comparable with initial CF, the rst and second
scattering. The competition between rstthird scattering
and initial CF results in the at slope between lines Band C
[see Fig. 2(a)]. The increasing contribution of the rst
third scattering as well as the emerging contribution from
the fourth scattering are responsible for the further slope
increase after the point B. Summing up our estimations for
u
?
and u
k
due to scattering, initial and asymptotic CF,
the exact momentum change is reproduced, see Fig. 4. The
multiple scattering is crucial for the transverse momentum
change. It determines the u
?
|
behavior which, in turn,
determines the ionization phases corresponding to the LES
peaks, and the shape of LES. The initial CF plays a less
important role for the transverse momentum change, but a
very signicant one for the longitudinal momentum
change. The latter shifts the LES peak, determining the
position of LES. These points are consistent with the
conclusions from Fig. 1(b). The asymptotic CF is only
important at near zero energies and has little impact
on LES.
Finally, we investigate the dependence of LES on the
laser intensity and wavelength, see Fig. 5. Two parameters
characterize the LES, the rst is E
H
, the high-energy limit
of LES [10], as labeled in Fig. 1(b). The second is the
contrast ratio (R), the ratio between the integral yield of
photoelectrons in the energy interval (0, E
H
) and the total
ionization yield for electrons emitted along the laser
polarization direction. R characterizes the relative height
and E
H
the width of the LES, respectively. E
H
increases
1.2 1.4 1.6 1.8 2.0
0.00
0.05
0.10
0.15
0.20
1.2 1.4 1.6 1.8 2.0
0.15
0.20
0.25
0.30
0.35
i
|
P
|
Exact
ICF+ACF
ICF+FS
ACF
(a)
B A D C
Exact
ICF
ICF+FS
ICF+FS+ACF
i
(b)
P
|
|
B A D C
FIG. 4 (color online). The estimation of (a) juP
?
j and (b) uP
k
described in the text. The contributions of asymptotic CF, the
combined contribution of asymptotic and initial CF, and the one
of asymptotic CF and scattering are shown. The laser and atom
parameters are the same as in Fig. 1(b).
1.2 1.4 1.6 1.8 2.0
0.00
0.01
0.02
0.03
0.04
0.05
1.2 1.4 1.6 1.8 2.0
0
50
100
0
1
2
3
0
100
200
300
400
-0.15
-0.10
-0.05
0.00
0 3 6 9 12 15 18 21 24 27 30
i
D C B
A
|
P
|
(e)
1
2
3
4
ICF
D C B A
i
r
s
(c)
1
2
3
4
D C B A
P
s
(d)
1
2
3
4
P
r
(b)
(a)
0.00
0.05
0.10
0.15
0.20
0.25
P
|
|
-0.05
0.00
0.05
E
(
t
)
FIG. 3 (color online). (a) The transverse uP
?
(solid) and
longitudinal uP
k
(dashed) momentum changes due to the
Coulomb eld. (b) The distance r (solid) from the core and
electric eld (dashed) vs the laser phase for one specic trajec-
tory (
|
1.72). (c) The position r
s
and (d) the momentum
s
at
the sth scattering versus the ionization phase. (e) The estimated
uP
?
due to different number of scatterings and initial CF.
The laser and atom parameters are the same as in Fig. 1(b).
PRL 105, 113003 (2010)
P HYS I CAL RE VI E W L E T T E RS
week ending
10 SEPTEMBER 2010
113003-3
approximately linearly with 1,, slightly deviating from
this scaling due to the increasing contribution of the initial
CF to the u
k
. The R decreases monotonically with in-
creasing intensity. With increasing wavelength, R rst
keeps constant and then decreases slowly. The E
H
and R
behavior can be explained qualitatively by inspecting the
phase space, see Figs. 5(d) and 5(e). In Figs. 5(d) and 5(e),
the curves in phase space rotate counterclockwise resulting
in the reduction of electrons within the LES. The reason is
the competition between the initial CF and scattering. When
xing the Keldysh parameter [see Figs. 5(c) and 5(f)], the
curves in phase space shift parallel to larger phase which
yields to the R decrease with decreasing the wavelength.
This partly explains why the LES has not been experimen-
tally observed with near infrared lasers. Another reason is
the relative suppression of multiple scattering at near infra-
red wavelength due to quantum effects [21].
In conclusion, the experimental results have been
successfully reproduced via the semiclassical calculations
and the origin of LES is claried. At midinfrared wave-
lengths the multiple rescattering of ionized electron plays a
signicant nonperturbative role which is the main factor in
creating the peaks in LES.
We gratefully acknowledge C. I. Blaga and L. F.
DiMauro for sharing the experimental data and C. H.
Keitel for valuable discussions.
Note added in proof.The recent work by Yan et al.
[23] provides a complementary interpretation of the LES
via specic quantum orbits.
*Chengpu.Liu@mpi-hd.mpg.de
k.hatsagortsyan@mpi-hd.mpg.de
[1] W. Becker et al., Adv. At. Mol. Opt. Phys. 48, 36 (2000).
[2] P. B. Corkum, Phys. Rev. Lett. 71, 1994 (1993); K. J.
Schafer et al., ibid. 70, 1599 (1993).
[3] L. V. Keldysh, Sov. Phys. JETP 20, 1945 (1964); F. H. M.
Faisal, J. Phys. B 6, L89 (1973); H. R. Reiss, Phys. Rev. A
22, 1786 (1980).
[4] S. V. Popruzhenko, G. G. Paulus , and D. Bauer, Phys. Rev.
A 77, 053409 (2008); I. A. Burenkov et al., Laser Phys.
Lett. 7, 409 (2010).
[5] R. Moshammer et al., Phys. Rev. Lett. 91, 113002 (2003);
A. Rudenko et al., J. Phys. B 37, L407 (2004).
[6] A successful quantum mechanical explanation is given by
F. H. M. Faisal and G. Schlegel, J. Phys. B 38, L223
(2005).
[7] T. Nubbemeyer et al., Phys. Rev. Lett. 101, 233001
(2008).
[8] E. S. Shuman, R. R. Jones, and T. F. Gallagher, Phys. Rev.
Lett. 101, 263001 (2008).
[9] C. Erny et al., Appl. Phys. B 96, 257 (2009).
[10] C. I. Blaga et al., Nature Phys. 5, 335 (2009); F. Catoire
et al., Laser Phys. 19, 1574 (2009).
[11] W. Quan et al., Phys. Rev. Lett. 103, 093001 (2009).
[12] G. Duchateau et al., Phys. Rev. A 63, 053411 (2001).
[13] The backwards scattering would contribute to the plateau
of the PES, see G. G. Paulus et al., Phys. Rev. Lett. 72,
2851 (1994).
[14] F. H. M. Faisal, Nature Phys. 5, 319 (2009).
[15] The number of recollision can be dened as N
s
$
wl
m
,2r
J
, where l
m
is the maximal drift distance
and
J
the drift velocity. During the drift in the laser
polarization direction
J
$
2s
p
with the electron energy
s and l
m
$n because the electron passes the origin only
when the drift distance is less than n. Because of the
electron drift in the transversal direction,
J
$
?
.
Accordingly, N
s
$nw,2r minf1,
?
, 1,
2s
p
g
minfn,x
0
, 1,2r
2U
,s
q
g.
[16] K. I. Dimitriou et al., Phys. Rev. A 70, 061401(R)
(2004).
[17] N. I. Shvetsov-Shilovski et al., Laser Phys. 19, 1550 (2009).
[18] A. M. Perelomov, V. S. Popov, and V. M. Terenev, Sov.
Phys. JETP 23, 924 (1966); M. V. Ammosov, N. B. Delone,
and V. P. Krainov, ibid. 64, 1191 (1986); N. B. Delone and
V. P. Krainov, J. Opt. Soc. Am. B 8, 1207 (1991).
[19] L. D. Landau and E. M. Lifshitz, Quantum Mechanics
(Pergamon, Oxford, 1977), p. 293.
[20] T. Brabec, M. Y. Ivanov, and P. B. Corkum, Phys. Rev. A
54, R2551 (1996).
[21] The contribution to LES of electrons with chaotic dynam-
ics increases with decreasing laser wavelength. However,
at rather small laser wavelength, when the impact parame-
ter at rst scattering becomes comparable with the elec-
tron de Broglie wavelength:
?
,w & 1,
2s
p
, the classical
description fails. Then, the diffraction of the electron wave
packet makes multiple scattering inefcient [see also J.
Tate et al., Phys. Rev. Lett. 98, 013901 (2007)] which
suppresses the LES. For a LES energy s % 3 eV,
?
0.1, the above condition is & 1 m.
[22] G. L. Yudin and M. Y. Ivanov, Phys. Rev. A 63, 033404
(2001); D. Comtois et al., J. Phys. B 38, 1923 (2005); C.
Huang et al., Opt. Express 18, 14293 (2010).
[23] Tian-Min Yan et al., arXiv:1008.3144v1.
FIG. 5 (color online). The dependence of E
H
(circles) and R
(squares) on (a) laser intensity I
0
at 2 m, (b) wavelength
at I
0
9.0 10
13
W,cm
2
, and (c) wavelength at 0.534.
The solid line is a t to all of the LES data. The triangles and
dashed lines are our estimations. In (d), (e), and (f) the electron
distribution in phase space for different laser intensities and
wavelengths are displayed.
PRL 105, 113003 (2010)
P HYS I CAL RE VI E W L E T T E RS
week ending
10 SEPTEMBER 2010
113003-4