VOLUME 52, NUMBER 8 PHYSICAL REVIEW LETTERS 20 FEBRUARY 1984

Free-Free Transitions in Intense High-Frequency Laser Fields

M. Gavrila and J. Z. Kamihski"'
Eundamenteel Onderzoek der Materie-Instituut voor Atoom —en Molecuulfysiea,
1098 SJ Amsterdam, The Netherlands
(Beceived 27 December 1983)

A nonperturbative theory is developed for free-free transitions of electrons colliding

with atoms in intense high-frequency laser fields. The energy of the incoming electrons
is assumed to be smaller than that of the photons. Bather simple expressions are de-
rived for the multiphoton absorption amplitudes. Large deviations from the laser-free
case are shown to appear in elastic scattering. The theory applies at already existing
laser frequencies but extends beyond, to the extreme ultraviolet range.
PACS numbers: 32.80.Kf, 32.90.+a
Studies of multiphoton free-free transitions of tion (Kramers and Henneberger') to the Schro-
electron-atom systems in intense laser fields dinger equation gives'
have been done predominantly at low frequencies,
stimulated by the theory of Kroll. and Watson. " [2P'+V(r+ a(t))]4=i(stI'/st),
where'
(&)

However, lasers of increasingly high frequencies

have become available, some of them operated B(t) = —c ' f, A( t') dt' = 5, sin&dt,
at high intensities. ' No theory exists so far for ao = —a/(dc. (2)
this case. We shall present here a method for This shoul. d be solved by imposing the boundary
treating f ree-free transitions specif ically adapted conditions of our problem: an incoming current
for the high-frequency, high-intensity regime. of particles of energy E = p'/2, and radially out-
The method is nonperturbative with respect to going currents of scattered particl. es of energies
the coupl. ing of the electron to the radiation field. and momenta
A full. y realistic description of the target atom,
E„=E+n&u, E„=p„'/2, n=0, +]., +2, . . . . (8)
taking into account its internal. degrees of free-
dom, is prohibitively difficul. t. %'e shall repre- Equation (I) has periodic time-dependent coef-
sent the atom by a potential. model, the merits ficients. As usual, we seek a quasiperiodic so-
and limitations of which are well understood. ' lution of the form
Besides, it is a logical. starting point towards a q(~r t) 8 -istic+ q (~r) e -tnmt
more complete solution of the problem. The po-
tential. wil. l. be taken to be of the central. self-con- Then, we Fourier analyze the potential:
sistent type: Coulomb-like at the origin [V(r) V(r+a(t))=g„'" „V„(a„r)e '" '.
= —Z/x], short range or ionic [V(r) = —Z'/r] at
l. arge distances, but unspecified otherwise. By some algebraic manipulations the coefficients
The l. aser field will be represented by a mon- can be written as
ochromatic pl. ane wave, linearly pol. arized, in V (a, ' r) = (i"/m) f V( r+ a u) T (u)(l - u') 't'du
the dipole approximation. The plane-wave as-
sumption is not critical, as the extension to a
laser pulse can subsequently be made. ' Linear where T„(u) are Chebyshev polynomials.
polarization is assumed in view of simplifying Insertion of Eqs. (4) and (5) into Eq. (1) leads to
the algebra, and the dipole approximation is jus-
a system of coupled differential equations for the
tified in the frequency range we are interested in
components g„(r), which we write
(from the visible to the extreme ultraviolet).
Consequently, we take the electrodynamic poten-
tials of the wave in the form A = a cos~t (with%,
[2P'+V, —(E+ntd)]g„= — p V„'

real) and p =0.'

Application of the space translation transforma- The boundary conditions require that our solutions
$„(a„td; r) behave asymptotically as follows:
4c(a&&, &; r) —exP(i[P r + y, ln(Pr —P r )]) + f, (a„td; r) exP[i(Pr —y, ln2Pr)] /y, (8)
(„(aot &o; r) —f„(ao, &u; x) exp[i(p„t —y„ ln2p„r)]/r (n c 0), (9)

Q& 1984 The American Physical Society 613

VOLUME 52, I@UMBER 8 PHYSICAL REVIEW LETTERS 20 FEBRUARY 1984

with y„= -Z'/p„(for a short-range potential Z' By use of the Green's operator G(Q) associated
=0). Equation (8) contains the elastic scattering to it, where 0 is the energy parameter, Eq. (7)
amplitude f, (u„~; r), and Eq. (9) that for absorp- may be formally solved as
tion/emission f„(o.„~;&"). The associated scat- '" .-G'"(E.
4. = 0-, ~. ) Z V. &. -. .(»)
tering cross sections are" (m &n)

da„/dQ Here gz'" is the (a, -dependent) solution of the

equation"
=(P. IP)lf. (c., ~; &)I' (s =o, +I, +2, . . .). (1o) = Egp,
F(p (13)
We shall now describe a method for handling
satisfying the boundary condition Eq. (8) with an
the system Eq. (7). The left-hand side contains
amplitude f, ('&(a„r). It then follows from Eq.
the Hamiltonian
(12) that the g„satisfy the boundary condition re-
H = 2P +Vo(ao; r). (11) quired by Eqs. (8) and (9) with the following ex-
pression for the scattering amplitudes:
f.( ., .
;&)=f,'"( ;~)~„,-(1-~„,) —&(-, ( 'I
V„I g„-"')

(14)
&m~~) &m ~~~)
Besides &t&z'", Eq. (14) also contains (z ' ', which is an incoming-wave solution of Eq. (13), as well
as the unknown set of components („,(r) satisfying Eq. (12).
By repeated insertion of Eq. (12) into Eq. (14) an expansion can be derived for f„.
Obviously, the
iteration will have practical significance only if the successive terms decrease sufficiently rapidly.
Since this mill not be true in general, it is important to establish the conditions under mhich the first
nonvanishing term of Eq. (14) represents a good approximation. For (a) &u» ~E, (a, )~, where Eo(n, )
is the ground-state energy of the Hamiltonian Eq. (11)"; (b) o,'ru» 1; (c) ~»E, it was possible to ex-
tract the exact form of the dominant contribution to the last term of Eq. (14) (denoted below by f "')
for an ambit any potential. of the type discussed before. ln the case of elastic scattering we find
- -
Be f, "&(&')
Imf, '"(r)
Z'
6a (u' [ Po
&„( 0 ) ~ 1,&(- ) + ~
&

&' 0 P,
&
&„( 0)
&+&(
)]
&' 0»
—(inc, '&u)'+ O(lna, '&u)
Ina '(u+ O((n '(o)')
where po= Pr" is the fina/ momentum [see Eq. (3)
for n = 0], and the corrective terms 0 also de- V, (no& 7) is simply the potential created by a lin-
pend on cr„E, 0. Thus at given a„" and suffi- ear "cha.rge" distribution of density m '(1-u') 'i'
ciently high &u [obeying conditions (a), (b), and (c) extending from —a, (u =1) to a(u =-1), the po-
and the dipole-approximation assumption] it is tential. generated by the unit of "charge" being
possible to satisfy the inequality (d) ~f, "&(a„&ul V(r) This b.ehavior appears natural due to the
E, 8)~«( f, ' (e0&„E, O)~. Whereas this holds in rapid oscillations of the center of force in Eq.
general over m ide ranges of parameter values, it (1) 11
should nevertheless be checked for each case For the absorption amplitude (n) 0) we get to
separately because f,'0& may become exceptional- lowest order from Eq. (14)
ly small for certain angles (e.g. , see Fig. 2).
Conditions (a)-(d) together ensure the dominance (~. ~' ~) = —2, &0-, „' 'I V. (P'")
f. I (18)
of the first term in Eq. (14). However, this may
hold under wider conditions than we were able to Thus, the Fourier component V„acts in our re-
prove. gime as a transition operator between scattering
Thus, the elastic amplitude f, reduces, to low- states gq'" of the dressed potential V, . The am-
est order (in the sense discussed above), to fo"&, plitude f„edpend osn ~ via the final momentum
which is that calculated from the time-independ- p„[see Eq. (3)]. Because &u was assumed to be
ent Schrodinger equation Eq. (13). This shows large at given a„" f„ill
al, l. wbe small with re-
that in the high-frequency, high-intensity regime spect to f, . (This contrasts with the low-fre-
the incoming electron feels only the static dis- quency case where many f„may be larger than
torted potential V, (801 r), the "dressed" potential f,.) Note that the condition (c) above precludes
associated to V(r). From Eq. (7) it follows that free-free emission (n( 0).
614
VOLUME 52, NUMBER 8 PHYSICAL REVIEW LETTERS 20 FEBRUARY 1984

Since the original potential V(r) is spherically symmetric, V, (o„r) and all V„(cT„r) have axial sym-
metry around a, (we have assumed linear polarization). For example, in the case of pure Coulomb
potential V(&) = —Z/&, Eq. (6) yields
V(n r)= —(2Z/tt)(x, r ) '~'K(2 '~'(1 r"-i )' '}
where r, = r e, and K is a complete elliptic integral
of the first kind. V, has a logarithmic singularity
'
along the distribution of charges, and r, ' singul. arities at its end points (all weaker than the original
Coulomb singularity). The dressed Coulomb potential Eq. (17) is represented graphically in Fig. l.
The axial symmetry of the elastic scattering problem Eg. (13) makes it resemble the electron-dia-
tomic-molecule and the proton-deformed-nucleus scattering (both in the static approximation with the
internal degrees of freedom neglected). The powerful computer programs developed for these cases
can be adapted to our needs. '4 However, for the time being, one may get an idea of the magnitude of

"
the effects involved by considering a simplified problem, in which the dressed potential V, (a» r) is
simulated by its spherical average V, (r).
We shall illustrate the simplified version on the elastic scattering from a pure Coulomb potential.
In this case, from Eq. (17),

V.(~) =
—(Z/maop) 2arcsinp —pin 1,[1 „,i»
y+ ~y1
p2]t /2
p2I1/2 p (1,
(Z/~ p) p)]
where p =r/ao. Since V, (r) can be regarded as a
!
Coulomb potential modified by a short-range dis- tortion, the scattering amplitude can be calculat-
ed by the two-potential formalism. The total, "
modified amplitude f,(i) is then the sum of the
Coulomb amplitude f'(r) and of a short-range
contribution f'(x), which can be obtained from a
phase-shift calculation. In Fig. 2 we give the ra-

2, 0

1.8-

I . 2-
O
I, O-
O
0.8-
0.6—

2 0.4-
2
FIG. 1. Values of the dressed Coulomb potential p& 0.2-
in a plane passing through the symmetry axis defined
by G, o. In this plane, taken as the horizontal base of 0.. 0 I I I I I I I I I I I I I I I I I

0 20 40 60 80 I 00 I 20 140 160 I 80
the figure, distances are measured in units of ~().
Along the vertical we represent (- c.oVO/Z) in atomic
SCATTERING ANGLE

units, up to the value 10. The saddle of the figure re-
flects the rise in pp near the line of singularities (ex-
tending from —Pyo to Pyo) and their increasing strength
towards the end points. At radial distances larger than
eo in the horizontal plane, the distortion of the Coulomb
potential fades away (the level lines become circular).
FIG. 2. Ratio of differential cross section for elastic
scattering from the averaged dressed Coulomb potential
Eq. (18), to the Rutherford cross section. Nuclear
charge Z=1; electron energy J" =0.05 By; 0.0 ——1, solid
line, and ~0 —3, dashed line [for the definition of ~0
see Eq. (2) and Ref. 9) .

615
VOLUME 52, NUMBER 8 PHYSICAL
tio of the modified to the original (Rutherford)
cross sections for two cases within the validity
of our theory, and accessible experimental. ly. '
It is apparent that the modified cross section
(for V, ) can differ considerably from that of the
laser-free case. '"
We conclude that electron-atom free-free tran-
sitions in the intense, high-frequency laser fields
now becoming available for experiments reveal
new features, contrasting with those known at
low frequencies. '~ A detailed numerical inves-
tigation is in progress (see Ref. 14). We are ap-
plying simil. ar methods to multiphoton ionization.
The authors wish to acknowledge the numerical
assistance of Dr. M. J. Offerhaus. One of them
J.
(J.K.) would like to thank Professor Kistemak-
er and Professor J. Los for their hospitality at
the Fundamenteel Onderzoek der Materie-Insti-
tute for Atomic and Molecul. ar Physics, Amster-
dam. This work was sponsored by the foundation
Fundamenteel Onderzoek der Materie with finan-
cial support from Nederlandse Organisatie Voor
Zuiver-Wetenschappel. ijk Onderzoek, the Nether-
lands Organization for the Advancement of Pure
Research.
~~~permanent address:1nstitute for Theoretical
Physics, Warsaw University, Hola 69, 00-681 Warsaw,
Poland.
N. M. Kroll and K. M. Watson, Phys. Bev. A 8, 804
(1973).
For a background of the problem, see M. J. Mittle-
man, Theory of Jasex-Atom Interactions (Plenum,
New York, 1982); L. Bosenberg, Adv. At. Mol. Phys.
18, 1 (1982); M, Gavrila and M. van der Wiel, Com-
ments At. Mol. Phys. 8, 1 (1978).
3T. S. Luk, H. Pummer, K. Boyer, M. Shahidi,
H. Egger, and C. K. Rhodes, Phys. Rev. Lett. 51, 110
(1983).
4For the case of radiation-free scattering, see
Ch. Joachain, Quantum Collisions Theory (North-Hol-
616
REVIEW LETTERS 20 FEBRUARY 1984
land, Amsterdam, 1975), Chap. 20. With the laser on,
the extra difficulties stemming from the decay of the
atom by multiphoton ionization are attenuated in our
regime, since in the limit of high frequencies the ioni-
zation is quenched.
5H. Kriiger and Ch. Jung, Phys. Bev. A 17, 1706
(1978) .
6Note that our premises are the same as those of
Kroll and Watson, Ref. 1.
YH. A. Kramers, Collected Scientific Papers (North-
Holland, Amsterdam, 1956), p. 262; W. C. Henneber-
ger, Phys. Bev. Lett. 21, 838 (1968).
We are using atomic units throughout.
op can be expressed in atomic units as op=(I/
I()) ~~, whereI is the time-averaged beam inten-
sity and Ip —3.51 x 10'p W/cm . The values of np of
present interest lie somewhere between 0 and 100.
For &u =0.23 a. u. and I=10 W/cm, which correspond
to the conditions of Ref. 3 (under somewhat stronger
focusing), ~p —-3.1.
For a laser pulse with slowly varying amplitude on
the atomic scale, Eq. (10) should be time averaged ap-
propriately; see Bef. 5.
"The energy eigenvalue equation Eqs. (11) and (13)
was considered before for bound states in a high-
frequency laser field by J. I. Gersten and M. H. Mittle-
man, J. Phys. B 9, 2561 (1976).
' It follows from Eq. (6) that by inc~easing +p pp be-
comes sha11ower, and therefore )Ep(c, p) ( decreases in
general from its unperturbed value at ~p =0 This
trend is confirmed by the calculation of Gersten and
Mittleman, Bef. 11, Fig. I.
~3The low-frequency result of Kroll and Watson is
derived under the same constraint (see Bef. 1, and
also Mittleman, Ref. 2, Sec. 6.4).
'4A computation along these lines for several model
potentials is now being carried out in collaboration
with Dr. J. van de Bee.
'~This should give the right order of magnitude for
elastic scattering, especially when the de Broglie wave-
length of the electron is larger than the extension of
the "charge" distribution (p~p~ 1). On the other hand,
interesting features related to the dependence on the
direction of the polarization vector will be washed out.
See the book of Joachain, Bef. 2, Chap. 17.
The ratio is equal to 1 for vanishing 0 because f~
tends to infinity, whereas f' stays finite.