HAL Id: jpa-00210277
https://hal.archives-ouvertes.fr/jpa-00210277
Submitted on 1 Jan 1986
HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
Absorption or emission during a collision : a test case H+ 2
C. Stehlé, N. Feautrier
To cite this version:
C. Stehlé, N. Feautrier. Absorption or emission during a collision : a test case H+ 2. Journal de Physique, 1986, 47 (6), pp.1015-1020. �10.1051/jphys:019860047060101500�. �jpa-00210277�
Absorption
oremission during a collision :
a test case H+2
C. Stehlé and N. Feautrier
Département d’Astrophysique Fondamentale,
Observatoire de Paris, Section de Meudon, 5, place Jules Janssen, 92190 Meudon, France (Reçu le 27 decembre 1985, accepté le 18 février 1986)
Résumé. 2014 La théorie couplée des collisions en présence d’un champ de rayonnement peut être utilisée pour décrire les processus d’absorption ou d’émission pendant une collision. Adoptant un formalisme semi-classique
de collision, nous pouvons étudier en fonction du temps l’ évolution d’un complexe collisionnel. Le cas de l’hydro- gène atomique perturbé par des protons est particulièrement intéressant car le problème collisionnel est soluble exactement pour un grand domaine de distances interatomiques. L’étude faite dans le cas particulier de la raie Lyman 03B1 permet de déterminer le temps caractéristique de l’absorption ou de l’émission. Les limites « impact »
et « quasistatiques » sont ensuite analysées.
Abstract 2014 The close coupled theory of scattering in a radiation field may be used to describe absorption or
emission processes occurring during a collision. Using a semi-classical description of the collision we study the
evolution of the collisional complex. The case of atomic hydrogen perturbed by protons is particularly interesting
since the exact solution of the dynamical problem is known for a large domain of internuclear distances. In the
particular case of the Ly 03B1 line we determine the characteristic time of the absorption or the emission. The « impact »
and « quasistatic » limits are then analysed
Classification
Physics Abstracts
32.70 - 32.90
1. Introduction.
The influence of collisions on atomic radiative pro-
cesses includes a large amount of various
phenomena :
line
broadening,
redistribution oflight,
collisioninduced
absorption
or emission, radiativecharge
transfer... The
major
part of these processes may beinterpreted
in term of radiative interactionsduring
a collision.
Recently,
much attention has beenpaid
to these radiative collisional reactions and it has been shown that the collisional formalism is well
adapted
for
expressing
theproperties
of the radiation asfunctions of the radiative S-matrix elements and cross
sections
[[1-3]
and references listed in[3]].
In the caseof line
broadening
the samedescription
is available outside the line core,corresponding
to the «oneperturber » approximation
[4] whichrequires
thecondition
ð’tc.ðro >
1 whereAT,,
denotes the interval time between two collisions and Aco thedetuning
fromthe line centre.’ The radiative process is then connected with the
dynamics
of thebinary
system(atom
+perturber)
on time scales ðw - 1 smaller thanAT,
[5].Our purpose here is to
study
the evolution of the collisional complex in order to determinequantita- tively
the characteristic time of theabsorption
or theemission. We
adopt
a semi-classicaldescription
of thecollision which makes the
physical interpretation
easier.
We focus our attention on the Ly a line
wings
ofhydrogen
atomsperturbed by
protons but similar discussion may beperformed
for othercolliding
partners(A. Spielfiedel,
1985, aparaitre).
Hereafterwe shall assume a
complete degeneracy
of the H levelswhich is
justified
forlarge detunings
and we makethe usual no
quenching approximation.
In thedipolar approximation
for the electrostatic interaction, thesolution of the
dynamical problem
is exact[6]
withoutthe traditional adiabatic
assumption
or any numerical computation. It follows that thestudy
of the timedependence
of the radiative process is easy. In addition thelong
range of H - H+ interaction(linear
Starkeffect) yields
the collision duration T, verylarge
sothat the
probability
ofabsorption
or emissionduring
the collision is
particularly important leading
topossible
consequences for redistribution oflight [7].
After a short
description
of the method, we recallbriefly
the determination of exact wave functions of anhydrogen
atom perturbedby charged particles
andwe
give
anexpression
for the « cumulated »absorption
or emission
probability
at time T, i.e. theprobability
that a radiative transition occurs in the time interval between the
beginning
of the collisions and the time TArticle published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:019860047060101500
1016
when the collision is not
necessarily
achieved At the end of the collision (T -+oo)
thisprobability
is relatedto the total
absorption
or emission rate. Hence ourpurpose is to
analyse
which part of the collision hasgiven the
major
contribution to this rate. This allowsa rough determination of AT, the duration of the interaction between the radiation field and the col-
liding
partners which cannot exceed the collisional duration. Thelarge
variations of AT with thedetuning
are
responsible
of the different behaviours of theabsorption
or the emission rate : the twolimiting
cases(small
andlarge detunings)
areanalysed
2. Method.
Following
Yakovlenko[1],
weadopt
a moleculardescription
of the collision and treat the radiative collisional reaction as a transition between twoadiabatic states
Ii> and If>
of thecomplete (colliding particles
A and P + monochromatic fieldg)
system.For
simplicity,
we assume that P is structureless but the same method can beapplied
in thegeneral
case.The method is
given
here forabsorption
but similarconclusions would be drawn for emission. This two state model
gives
theprincipal
features of the process,particularly
in theH’
case as can be seen in the next section. The reaction can be describedby :
The total Hamiltonian can be written :
where
HA
is the Hamiltonian of A,UAP
the interaction between A and P,H.
the Hamiltonian of the free radiation held andV AP-g
the interaction between the «quasimolecule »
AP and the field In the absenceof the radiative field, we define the adiabatic
quasi-
molecular states
(A
+P);
and(A
+P)f.
These stateshave
energies Ei(R) = hcoi
+U;(R)
andEf(R) = nWf
+Uf(R) respectively, hcoi
andhcof being
theenergies
of the involved atomic states for A andU;(R)
andUf(R)
the molecularpotentials
(i I UAP i) and f I UAp I f >. Ui
andUf
are functions of time tthrought
the relationR(t).
Denoting by ) I n. >
theeigenfunctions
of the freefield with energy nO) hco, we shall take the
products
ofwave
junctions I i > =- (A
+P);
*nw ) and I f > =-
I (A
+P)f
*n. - I >
as a basis for thecomplete
system. The total wave function isexpanded
in termsof this basis :
where
6;(t)
=Ei(t)
+ nhw andSf(/)
=Ef(t)
+(n - 1)
1iroare the
energies of I i > and I f >.
The time
dependent Schrodinger equation gives
thefollowing
system ofequations
for theamplitudes ai(t)
andaf(t) :
with the initial conditions
a;( - oo)
= 1 andaf( - oo)
= 0.In the
dipolar approximation, V(t)
can beexpressed
in terms of the
dipole
momentd(t)
of the quasimole-cule at the distance
R(t)
and theamplitude
&w of theradiative field
V(t) = -
1i - 1 dE..
In the weak field limit, the system(3)
is solved in the first order pertur- bationapproximation leading
to thefollowing
expres- sion for theprobability I aif(7) I’
to be in stateI f)
at time T :
Substituting
theexpression
ofBf
and8;,
we obtainthe usual
expression :
When T goes to
infinity, 1 aif(T) 12 gives
theprobability
of
absorption
for agiven impact
parameter p and agiven
relativevelocity
v.This very
simple
adiabatic model breaks downgenerally
due to strong non adiabatic effectsduring
the collision. The
colliding
system H + H+ forms anexception
aslong
as the electrostatic interaction poten- tial is dominatedby
itsdipolar
partcorresponding
to the
perturbing
electric microfield inplasma physics.
Considering
anhydrogen
atomperturbed by
a staticelectric field, the
eigenvectors belonging
to a fixedprincipal
quantum number n are the well known Stark statesnil i2 >.
These statesquantified along
the field axis are
analogous
to adiabatic molecularstates for atom-atom systems. When the
perturbed
proton movesalong
itstrajectory (assumed rectilinear),
the electric field rotates
leading
to a rotationalcoupling
among the adiabatic Stark states. As was first
pointed
out
by
Lisitsa and Sholin[6]
the symmetryproperties
of the
hydrogen
atomcorresponding
to the rotation group04 [8]
can be used to obtain exact wave functionsunn’n" and exact energy levels
Enn’n"( t)
of the atom in arotating
electric field This exactdiagonalization
ofthe Hamiltonian in the
rotating
system is a conse-quence of the same
dependence
on R of the electro- static interactionpotential
and the rotationalcoupling.
The construction of the wave functions Unn’n" is
explain-
ed in details in
[6]
and wegive
hereonly
the methodusing
the same notations.The result is more
easily
obtained in therotating
coordinate system : at each moment t of the collision,
we define the x axis directed along the intemuclear axis and the z axis normal to the collision plane.
The transformation from the laboratory to the
rotating
axis at time t
corresponds
to successive rotationsgiven by
the three Eulerangles (00, 00, V/0) corresponding
to the initial orientation of the intemuclear axis and the
angle 1/I(t)
of rotation around the z axisduring
the collision
(4/(0)
=0).
For agiven impact
parameterp and
velocity
v, theeigenfunctions
unn,n" are obtained from theparabolic
wave functions unÏ1Î2(corresponding
to the x axis of
quantization) by
means ofsimple
rotations :
Pt
andP2
aregiven by
the relations :a =
3/2
ne2ao/1i
measures thestrength
of the electro- static interactionpotential
for anhydrogen
atom instate n. For the energy
Enn’n,,(t)
we obtain :where a
= 1 + b
andhwn
is the energy of theatomic state n.
Using the j I nn’n" >
diabatic states asa basis for the radiative
problem,
one can see imme-diately
that the transition results in a sum of two state processes describedby :
This
clearly
shows that the method described above may be usedIn the H + H + case, it is
interesting
to remark thatthe matrix elements of the
dipole
moment operator d inthe I nn’n" >
basis areindependent
of time t, so thatthe time
dependence
of the radiativecoupling V(t)
comes from the rotation of d relative to the radiative field
C..
The tensorialcomponents dk
of thedipole
moment are known in the molecule fixed frame and the electric field is defined in the
laboratory
system.Denoting by
ep thepolarization
vector of the radiative field and ek the unit vectors in therotating
frame,V(t)
isgiven by :
The time
dependence
of Y is hencegiven by
the termexp[ip(t)]. Using expression (8)
forEf
andEi,
the
probability 1 af(1) p
to be in stateI f) =- I nf nf’ ? >
at time T is obtained from
(4)
after anangular
average relative to the orientation(00, 00, 1/10)
of the collision axis : -A summation relative to p has to be
performed
in the caseof unpolarized
radiation fields.In the case of the Lyman a transition,
expression (10)
takes the form :where nf and nf
take the values ±1 j2.
The time
dependence of 1 aif(T) 12
isjust given by :
with s = ± 1, ±
(o-f
±1).
Achange
ofsign
for bothAco and s leads to the same result for
A(T),
thus wediscuss
only
the case ð.úJ > 0. At the end of the collision when T tends toinfinity, A(oo) corresponds
to thequantity
2nA([6] expression (24))
obtainedby
Lisitsaand Sholin
using
an autocorrelation function method The SIN part of theexponential
in(12)
does notcontribute to
A(oo)
so that theonly interesting
part of the timedependence
of the radiative transition isgiven by :
1018
P(T)
can also be studied in function of Ror 0
usingthe
following
relations : andThis collisional
description
of theabsorption
process allows a very simpleinterpretation
of thequasistatic
limit : when the two
potential
curves&i(R)
and&f(R)
intersect at
Ro
the transitionprobability
may bedominated
by
the contribution of aregion
close toRo.
Neglecting
the timedependence
of thedipole
matrix element, theequality Uf - U;
= h Am atRo
cor-responds
to thestationary phase
in(5).
A calculation of thisintegral by
the saddlepoint
method leads to theusual Landau formula
[9]. By integration
relativeto p and
averaging
over the Maxwellian distribution of velocities, we obtain thequasistatic
formula. Astudy
of the evolution of the system with the time T allows toprecise
thevalidity
of this limit in the caste oflarge detunings
and also theimpact
limit for smalldetunings.
3. Results.
3.1 LARGE DETUNINGS 4m. - We are
only
concernedhere with the values of s
leading
tostationary
phase (or FranckCondon) points
for agiven
An, the oppo- sitesign
valuesgiving exponentially decreasing
con-tributions to the transition
probability.
Figure
la shows the characteristic features of P in function of Rduring
the collision,Ro
denotes the Franck Condonpoint
At first, it is clear that the transition ismainly
localized nearRo.
Large variationsof the
phase 4>( t)
of the COS term for other values of R leads to oscillations in P, but these oscillations do not contribute on an average to the finalprobability.
Fig. la. - Variations of the transition rate P(7J in units
of P(oo) vs. the intemuclear distance R (atomic units).
8 = 1 + erf = 56.1 ; the stationary phase point is R0 ~ 31.4 ua.
The arrows refer to increasing times along the trajectory.
(t2 = (R2 - p2)
x 12 484 where t, R, p are expressed ina.u.).
Fig. 1 b. - Same as figure 1 a, but relative to the transition rate Ps(t)/Ps( (0).
It is
interesting
to compare this behaviour with thestationary
phase results in order to obtain thevalidity
conditions of it The transition is assumed to be localized at
Ro corresponding
to time to andangle 1/J( to).
At
Ro :
Retaining
the two lowest order terms of theexpansion
of
0(t)
near + to, we obtain thefollowing approximate expression :
The variations of the exact and
approximate phase 4>8(t)
in function of t are shown infigure
2.They
areFig. 2. - Time variations of the phase 0(t) = wt - et/J( t) along the trajectory s = 1 + Qf = 56.1. Units are rds for the
phase and (103 ua) for the time. The broken curve shows the series expansion «Ps(t) in the vicinity of the stationary phase points ± to = ± 344.8 au
both
symmetric
withregard
to t = 0, very similar in thevicinity
of t =:t to; but they clearly
differ for other values of t.Assuming
that the two contributions at t = ± to do not interfere, the use ofexpression (15)
for
os(t)
leads to thestationary
phaseexpression PS(T)
forP(T).
When T tends toinfinity, Ps(7)
is
given by
a verysimple expression :
The variation of PS in function of R
(Fig.
1 b) presents the samegeneral
behaviour as P,particularly
in thevicinity
ofRo.
But, asexpected
from the variation ofos(t)
with t, theperiods
of oscillations are very dif- ferent Nevertheless, this is of littleimportance
sincethe final result
P( oo)
is not affectedby
these oscillations.The variations
of P(oo)
andPS(oo)
in function of theimpact
parameter p axecompared
infigure
3 for agiven
Am andvelocity
v. The agreement isquite good
as long as the
impact
parameter p is lower thanRo.
For p >
Ro,
the Franck Condonpoint
is located near or inside the classical forbiddenregion
and the saddlepoint
method becomes invalidOne can see that the
validity
conditions of the quasi-static limit is
directly
givenby
the use of the saddlepoint
method forcalculating
theintegral
in(3).
At first, if AT represents the time interval near ± togiving
the main contribution to theabsorption,
thecorresponding
variation ofos(t)
in(15)
has to be of theorder of n
leading
to theequality :
Fig. 3. - Variations of P(0) = 0.25 P(oo) vs. the impact parameter p (full curve) compared with the result of the saddle point method
(expression
(16), brokencurve)
andwith the mean value of this expression where the oscillations of SIN2 are weighted to 1/2 (dotted curve); 8 = 1 + a,.
Units are (10- 27 s - 1) for P and atomic units for p.
This characteristic time of the radiative process
varying
as Aw- I becomes very small for large detun-ings
and this allows aninterpretation
in terms of«
quasistatic
limit ».Obviously,
theperturber
movesalong
itstrajectory
and the expression « static » has not themeaning
of « static ions ».For the use of the saddle
point
method, the contri- bution of values of t outside the interval AT has to be very small,corresponding
to many oscillations ofØ(1)
as well for T -> 0 as for T --> oo. This first condi- tion is realized ifThen the rotation of the
dipole during
the collision becomesnegligible
and thestationary
phasepoint Ro given by
the relation Aco -- 3Ii /me Rg
isindependent
of p and v. The second condition expresses the neces-
sity
for thestationary
phaseregion
to besufficiently
distant from the
turning point
(t = 0).Putting AT 4 to
and
assuming
the relation(18)
realized, we obtain :This relation is
closely
related to the usual condition for thequasistatic approximation
in the linewings
theories
[10, 11] :
the time of interest åW-l has to be smallcompared
to the collision duration p/vRo/v.
At the
stationary
phasepoint
Aw = 3n /me Ro leading
to the
inequality
3 hAco/m,, v’ >>
1.In the limit of this
binary
model, thequasistatic
limit is obtained for protons as soon as
I AA
I > 1 Aat E = 1 eV. For electrons in the same conditions, the
criterion would
give
therelation I A A I >>
100 A whichis out of the
validity
limit of the use of classical tra-jectories
anddipolar potential.
If the
preceding
conditions are realized, it ispossible
to
integrate analytically
theprobability PS( (0)
over p(with
the mean value 0.5 for the SIN2term).
Afteraveraging
over the Maxwellian distribution of the kinetic energy of relative motion, we obtain the usual Holtsmarkexpression
for them-dependent
rate oftransition
which leads to a
profile varying
as N I1m-SI2.3.2 SMALL DETUNINGS 4b£V. - The
typical
variationof the
probability
P in function of R for I1À = 10-3 Agiven
infigure
4 shows that the transition may occurthroughout
the collision and is not localized at any distance. Thephase stationary points
tend toinfinity.
For a
dipolar potential,
the collision time is infinite but theDebye screening
at distance pD due to othercharge
particles
diminishes the potential range so that the collision duration isgiven by
A T a iPDIV-
The contribution of
large impact
parameters becomespredominant corresponding
to s = ± 1 in(12)
so thatwe
point
out theimportance
of the rotation of the1020
Fig.
4. - Same as figure 1a with e = 1 + af = 28.56.dipole (p-term
in(11))
whichgives
themajor
contri-bution. If we introduce the time of interest
Aco- 1,
this time is very
large compared
to r, for small values of Acoaccording
to the relationAw. ’rc
1. Inparti-
cular, when thisinequality
is realized for the half widthAW1/2
we obtain the usualvalidity
conditionof the
impact
limit :AW1/2.PD/V
1. This condition is verified for protons for N 1014 cm- 3 at 104 K.In the
impact
limit, thepredominance
of distantcollisions allows a second order
perturbational approach previously
described forHn
lines[12]
andleading
to for the Lyman a linewhere 0 denotes the temperature and N the proton
density.
4. Conclusion.
We have
presented
in this paper a method for thestudy
of thedynamics
of radiative collisional reactions.Using
theparticular
caseof H + H+, we
have shownthat the characteristic time for the
absorption
or theemission may be related to ð.w- 1 : this
quantitative study
allows aphysical interpretation
of the so called«time of interest ». Furthermore we have
proved
thatthe
quasistatic
limit is valid for protonsfor I A A
> 1 Aand is never valid for electrons. This condition is more
drastic than it is
usually
asserted [7]. Weemphasize
that this limit is
independent
of theperturber density
as
long
as simultaneous strong collisions do not occur(N
1014cm-3).
Some conclusionsgiven
here may be extended to morecomplicated
systems.References
[1] YAKOVLENKO, S. I., Sov. J. Quant. Electron. 8 (1978) 151.
[2] JULIENNE, P. S., Phys. Rev. A 26 (1982) 3299.
[3] MIES, F. H., Theoretical Chemistry : Advances and
Perspectives, edited by D. Henderson (Academic Press, New York) 1981, vol. 6B, p. 12.
[4] BARANGER, M., Atomic and Molecular Processes edited
by D. R. Bates (Academic Press, New York) 1962.
[5] REYNAUD, S. and COHEN-TANNOUDJI, C., J. Physique
43 (1982) 1021.
[6] LISITSA, V. S. and SHOLIN, G. V., Soviet Physics
J. E. T. P. 34 (1972) 484.
[7] YELNIK, J. B., BURNETT, K., COOPER, J., BALLAGH,
R. J. and VOSLAMBER, D., Astrophys. J. 248 (1981)
705.
[8] HUGHES, J. W. B., Proc. Phys. Soc. (London) 91 (1967)
810.
[9] LANDAU, L. D. and LIFSHITZ, E. M., Quantum Mecha-
nics : Non Relativistic Theory (3rd ed., Pergamon Press, Oxford) 1977.
[10] HOLSTEIN, T., Phys. Rev. 79 (1950) 744.
[11] GRIEM, H. R., KOLB, A. C. and SHEN, K. Y., Phys.
Rev. 116 (1959) 4.
[12] STEHLÉ C. and FEAUTRIER, N., J. Phys. B : At. Mol.
Phys. 17 (1984) 1477.