Você está na página 1de 34

Chapt e r 9

Ra di a t i v e El e c t r on
Ca pt ur e
In Chap. 8, we discussed Coulomb or nonradiative electron capture. In
the nonrelativistic energy regime, the cross section for this process falls off
asymptotically with increasing projectile velocity v as v -12 (or as v -11 in
second order). This rapid decrease is caused mainly by the requirement
t hat a given moment um component in the initial electronic wave function
has to find its counterpart in the final moment um wave function displaced
by the moment um mev of an electron traveling with the speed of the pro-
jectile. If, however, the electron transfer is accompanied by the emission of
electromagnetic radiation, the emitted photon acts as a third body carrying
away energy and moment um released by the formation of the final bound
state. Hence, the condition of moment um matching will be relaxed so t hat
the cross section for radiative electron capture (REC) falls off as v -5 for
high nonrelativistic projectile velocities.
For free electrons, capture cannot take place at all without the emission
of photons owing to energy and moment um conservation. This means,
qualitatively, t hat electrons loosely bound in low-Z target atoms or in outer
shells are more likely to be captured with photon emission t han without.
Hence for low-Z target atoms at high projectile energies (e.g., for ZT -- 1
with Ep _> 10 MeV/ u, [SpiT9]), the REC cross section exceeds the cross
section for Coulomb capture. From this point of view, the REC mechanism
deserves particular attention.
Starting in Sec. 9.1 with a preliminary pictorial description of REC
within the impulse approximation, we t urn in Sec. 9.2 to the direct per-
t urbat i ve formulation. An insight into qualitative features, in particular
247
248 CHAPTER 9. RADI ATI VE ELECTRON CAPTURE
the angular distribution, is obtained from a nonrelativistic t reat ment in
See. 9.3. As a consequence of these considerations, we argue t hat presently
the best description of REC is provided by t reat i ng it essentially as the
inverse of the photoelectric effect. Therefore, in See. 9.4, we discuss several
ways to obtain photoelectric cross sections so t hat in See. 9.5 we can use
these results to derive the desired REC cross sections. Finally, in See. 9.6,
we present further theoretical and experimental results.
We use full units in the maj or part of this chapter, except for the more
technical sections 9.4.3 and 9.5.3, where we adopt relativistic (natural) units
h = me = c = 1, see Appendix.
9. 1 Ge n e r a l f e a t u r e s
A loosely bound target electron may be considered as approxi mat el y free in
a high-energy collision. In this limit, REC is identical with radiative recom-
bination (RR), in which an electron initially moving with the velocity - v
in the projectile frame is capt ured into a bound st at e with the simultaneous
emission of a photon of energy ha/ and wave number k ~. Experimentally,
the process of REC has been first identified by Raisbeck and Yiou [RaY71],
by Schnopper et al. [SCB72], and by Kienle et al. [KiK73], and since t hat
t i me has been extensively studied bot h experimentally and theoretically.
In the following, we st art with a brief outline of the t reat ment . We
assume t hat the cross section aRR for radiative recombination is known
since it is essentially the inverse of the atomic photoelectric effect. It is t hen
nat ural to refer all moment a to the projectile frame. If the t arget electron
has the moment um q with respect to the t arget nucleus and the projectile
moment um is characterized by the Lorentz factor ~,, its moment um in the
projectile frame q~ is obtained from Eq. (2.29). If the moment um 3'mev
of an electron traveling with the relative speed of the target towards the
projectile is large compared to the electron moment um q in a low-Z t arget
atom, one may use the impulse approximation [K1J75] to write the double-
differential REC cross section in the projectile frame (primed coordinates)
a s
d2 CrREC - - / d 3 dCrRa (q') [@i (q)]2 6 (?tW' + E[ - E[) (9 1
)
df~' d(ha/ ) - q df~' ' "
where @i ( q) is the Fourier transform of the initial electronic t arget wave
function. The delta function expresses the energy conservation between
the final electronic energy E[ and the photon energy hcJ' in the projectile
frame on the one hand and the initial electronic energy E{ (also in the
projectile frame) on the other hand. According to Eq. (2.29), these energies
9.1. G ENERAL F EATURES 249
are wri t t en as
E( - " ) / ( Ei - Vqz)
- ~me c2 - ~[ ~i l - ~yvqz ( 9. 2 )
and
E~ - mec 2 -l e}l . (9.3)
H ere I ~il ~ nd I~1 ~ re t he b i ndi ng energ i es of i ni t i al t ~ rg et ~ nd fin~ l proj ect i l e
states. Inserting the expressions (9.2) and ( 9. 3) i nt o Eq. (9.1), we have
d2Cr a EC
d~' d( ~' )
daRa(q, )
d 3 q df2~
J dcraa(q, )
d a q df ~
I@i(q)125(hw ' + E~ - ")/Ei -1-- ")'Vqz )
I@i(q)125( hw' - 141 + ~l ~[ - Te + ~Vqz ) ,
(9.4)
where Te = mec2(') ' - 1) is the kinetic energy of an electron traveling with
the same speed as the projectile. Since the initial moment um di st ri but i on
of the t arget electron is peaked around q ~ 0 and since we usually consider
ZT < Zp, the recombi nat i on cross section varies slowly and can be t aken
outside the integral over the transverse moment um. In this approxi mat i on,
we get
dCrRE C _ _ 1 ( df f RR) , f f i( qz) , (9.5)
df t ~ d( hw' ) "/v df t ~ qlz=3,(_rneV+qz)
where
f
,.7i(qz) - J d2qbl@i(q)l 2 (9.6)
is the Compt on profile. The integration extends over the transverse mo-
ment um qb- In Eq. (9.5), the longitudinal moment um q~ is fixed by w ~, so
t hat the shape of the phot on spect rum
h~' - 14t- ~t~il + To - ZVqz
(9.7)
is det ermi ned by the spread of the longitudinal moment um qz around its
mean value qz - O, which yields the peak energy
~ ; - ~ E ~ - E~ - I ~ f l - ~/[(~i[-n t- re- (9.8)
The t erm ~/vqz in Eq. (9.7) describes the Doppler broadening t hrough the
Compt on profile.
However, since the cross section aaR decreases with increasing x-ray
energy, see See. 9.5, the actually measured peak is usually shifted to lower
energy.
250 CHAPTER 9. RADI ATI VE ELECTRON CAPTURE
In the l aborat ory system, the phot on energy hw depends on the emission
angle 0 wi t h respect to the beam direction via
h(.U - - IC} I - - 3` l ~i I + r e - - 3` Vq z
3' (1 -- 3 COS 0) ' ( 9. 9)
which follows from Eq. (2.51). Similarly, if Eqs. (9.4) or (9.5) lead to an
angular di st ri but i on in the projectile system as a function of the emission
angle 0' wi t h respect to the beam axis, we obt ai n the angular di st ri but i on
in the l aborat ory system by the t ransformat i ons
cos 0 ~ = cos 0 - 3 and dft ~ = 1 (9.10)
1 - 3 cos 0 dft 3`2 (1 - 3 cos 0) 2,
given by Eqs. (2.53) and (2.54), respectively.
9. 2 Th e t r a n s i t i o n a mp l i t u d e for RE C
While Sec. 9.1 describes an intuitively appealing approach based upon the
impulse approxi mat i on, a complete theoretical description should st art from
the general t r eat ment of See. 5.4.2. We set out from the formulation of the
pert urbat i ve t ransi t i on amplitude. In cont rast to the cases t reat ed there,
the per t ur bat i on giving rise to REC is the electromagnetic interaction of
the active electron wi t h the phot on field. After analyzing this met hod in
See. 9.3, we t urn to a different description in See. 9.4.
As in the preceding chapters, we use a straight-line version of the i mpact
paramet er met hod to describe the projectile motion, see Fig. 3.2. It is
convenient to describe the radi at i ng system in its own rest system; therefore
we refer all coordinates to the projectile frame. In order to ensure the
transverse polarization of the photons, we work in the radi at i on gauge (also
called Coul omb gauge) for the electromagnetic vector pot ent i al A ~ with
V ~. A ~ - 0. We t hen can write the interaction operat or as - e 3 ` " A~ =
e"/. A ~ = e @a . A ~, where the vector pot ent i al appropri at e for the emission
of a phot on with a wave vector k ~, energy hw ~ = hc U and unit polarization
vector fl~ wi t h ( ~- 1, 2) i s given by [Sak67]
1
Vw' uae . (9.11)
Here, V is the normal i zat i on volume, for which periodic boundary condi-
tions are required. If all integrations are carried out in the projectile system,
9.2. THE TRANS I TI ON AM PLI TUDE F OR REC 251
the i mpact -paramet er-dependent transition amplitude between specific ini-
tial and final atomic states is written as
A(~ ~ ) - i f d r ' / ' d 3' - - ' ' t ' ' .
h rp ~f ( r p, ) ( eU~ . Ax ) S~i ( r T, t ) (9.12)
As long as exact wave functions are used for the motion of a single electron
in the field of the residual target atom and the projectile, this expression is
exact within the one-photon approximation, i.e., within first-order pertur-
bation theory in the photon field.
Since, in practice, it is not possible to evaluate Eq. (9.12) with exact
solutions of an effective three-body problem, approximate wave functions
have to be introduced. For the time being, it is instructive to adopt the
Born approximation, t hat is, to use undistorted eigenstates ~i and ~} of
target and projectile with eigenenergies (in their rest system) Ei and El.
The implications of this drastic simplification are discussed following Eq.
(9.19). Inserting explicitly the time dependence of the wave functions (4.74)
and writing E i - - h0 2i and E[ - hw~, we get
A ( ) ) ( b ) :
1
h Va; ' dr' 3 , ' ?(rp)e uvft
^t --i(k' ' -uv't')
X ~' U Ae .rp ~i (rT)e--zwit.
(9.13)
After introducing the Fourier transforms q~i(q) - ~i( q b , qz) and qS~(p) -
~ ( PD, Pz ) of the initial and final states, respectively, and transforming the
target-frame coordinates rT, t to the projectile system by the Lorentz trans-
formation (3.31), we are able to perform the space and time integrations to
obtain
1
-7 Vw' q pf ( qb- hk~b, q - - hk'z)
^ ! l
X a - U A S @i(q) 6( hoot + Ef -- ~/E i -t- " ) ' Vq z ) e i q b ' b / h,
(9.14)
where q_ = 7qz - ~ / ( v/ c 2 ) Ei 9 The delta function establishes the relation
(9.8) for the photon energy and allows one to eliminate qz so t hat one is
left with the two-dimensional integral
, , t ) ei q b . b / h (9.15)
a . u~ S ~ i ( q b , q+
252 CHAPTER 9. RADI ATI VE ELECTRON CAPTURE
^ !
Fi gur e 9.1. O rientation of the wave vector k p and the polarization vectors u 1
^ t
and u2.
wi t h
q+ = [ei - ( E; + h~ ' ) / . y ] / v
q_ = [E i /~- (El + hwl)]/v. (9.16)
For Coul omb capture, the definitions (9.16) reduce to the definitions (8.6).
Wi t h phot on emission, the resonance condition is q+ = 0. From the t ran-
sition ampl i t ude (9.15), the st at e-t o-st at e cross section is obt ai ned as
d2o -
d a ' d ( h~ ' )
v(~')~ j
(27rhc)3 ~~1 deb ]A(~) (b)]2"
~2
(9.17)
When inserting Eq. (9.15) into (9.17), the integration over the i mpact
par amet er plane can be performed to yield the cross section in the projectile
frame as
d2a
da ' d( h~' )
e 2 1 f I
2- -
--1 12.
c~. u~ S ~i(qb, q+) (9.18)
If the initial and final electron polarizations are not detected, one has to
average over initial and to sum over final electron subst at es in order to get
the cross section per electron. Let 0 / be the emission angle of the phot on
9.2. THE TRANS I TI ON AM PLI TUDE F OR REC 253
wi t h respect to the beam direction in the projectile system. After the b-
i nt egrat i on has been carried out, it is possible and convenient to choose a
coordi nat e syst em in which the wave vector k ~ of the phot on and the linear
pol ari zat i on vector fl~ lie in the x- z plane, and the ot her polarization vector
^ t
u 2 perpendi cul ar to it, t hat is in the y direction, see Fig. 9.1. Explicitly
wri t t en in component s, we t hen have
k ' - ( k' sin 0 ', O, k ' cos 0 ')
fi~l - (cos 0 ', 0 , - sin 0' )
^' - ( 0 1 0).
U 2 , ,
(9.19)
The angular di st ri but i on in the l aborat ory system is obt ai ned from Eq.
(9.18) wi t h the aid of Eqs. (9.9) and (9.10).
We now t ur n to a discussion of the limitations i nt roduced by the use
of approxi mat e wave functions. The first point is the role played by the
current operat or - ec c ~ entering in the formulation given above. In gen-
eral, in a charge transfer collision, one will t ry to use approxi mat i ons in
which the initial st at e is an eigenstate of the channel Hami l t oni an Hi, see
Eq. (5.53), while the final st at e is an eigenstate of anot her Hami l t oni an
Hf r Hi, see Eq. (5.54). If this is indeed the case, the operat or - e c ~
will no longer satisfy a continuity equat i on like Eq. (4.17) t oget her wi t h
the charge density and hence will not qualify as a current operat or j to be
used wi t h the electromagnetic interaction j . A/ c . In fact, one may con-
vince oneself t hat a source t erm resulting from (Hi - Hf) appears in the
continuity equation. Since in suggestive approxi mat i ons, e.g., those suc-
cessfully used in Coul omb capture, see Chap. 8, the Hami l t oni ans Hi and
Hf belong to different centers and hence are drastically different, there is
no current operat or consistent wi t h such an approach. As a consequence, it
is not possible to define a Born approxi mat i on or distorted-wave Born ap-
proxi mat i on for REC, see, e.g., [HiW87, HiW89], consistent wi t h the usual
definition j - - ec c ~ for the current operat or.
In the absence of an exact t hree-body wave function, a consistent de-
scription is obt ai ned if bot h initial and final states are const ruct ed as eigen-
states of either Hi or Hr . Since usually in REC one considers systems wi t h
ZT << Zp, the only practical way to t reat the process is to consider ini-
tial and final states as cont i nuum and bound states of the pr oj ectil e. The
t arget at om is t aken into account only insofar as it generates an initial elec-
tronic moment um distribution. This is the impulse approxi mat i on in the
projectile frame discussed in Section 9.1.
254 CHAPTER 9. RADI ATI VE ELECTRON CAPTURE
9 . 3 A n o n r e l a t i v i s t i c t r e a t me n t
For qualitative purposes, it is instructive to specialize the transition ampli-
tude (9.15) to a nonrelativistic collision. In view of the foregoing discussion
we choose the projectile system as a reference frame. Since, furthermore,
it is also known [K1J75, BrD74] t hat the plane-wave impulse approxima-
tion or, equivalently, the Born approximation leads to acceptable results in
the projectile frame, we use unpert urbed nonrelativistic hydrogenic-bound
states as initial and final states. Assuming spinless particles for the mo-
ment, we replace c~ by vd / c = Pel/(meC).
For high projectile velocities, REC into projectile ls states will be dom-
inant, and, although this limitation is easily removed, we also assume ls
target states. Wi t h the ls hydrogenic moment um wave function being given
by [BeS57]
2V/-2 F)5/2
9 '~z (9.20)
ls(q)- +
where Qz - mee2Z/ h, the l s-l s REC cross section per electron (9.18) in
the projectile frame can be written as
d~' d(hw') he
(meV)2 ~ d2qb [@f(p)(p. uA)@i(q)l 2, (9.21)
where
q = (qb, q+)
p = (qb, q _ ) - hk ' (9.22)
and, in analogy to Eq. (8.7),
q+ - [ e i - (e~ + hw')]/ v +l~meV. (9.23)
The energy conservation expressed by Eq. (9.23) is a result of retaining
the full space-time dependence e xp( - i k ~. r ~ + iw~t~), sometimes denoted as
retardation, of the photon wave function.
Kleber and Jakubassa [K1J75] show t hat the Born approximation
Eq. (9.21) in the impact paramet er picture is identical to the plane-wave
impulse approximation Eq. (9.1).
Angular distribution
Let us now concentrate on the angular distribution of the emitted photons,
assuming t hat the photon polarization is not detected. While the momen-
t um wave function ~i of the initial state enters only insofar as it cuts off the
9.3. A NONRELATI VI S TI C TREATM ENT 255
1.2
I i I i i I i
_
dale)
\ d a l g O o l . s i n Z e
- " ~
- \
1 . 1 - \ /
_ \ /
\
- " . . . ./ /
. . . . . .
/
- t
f
- I
t
t
0 . 9 0 5 ; a , I I , I I , I
3 6 0 9 0 1 2 0 1 5 0 8 0
0
Fi gur e 9.2. Calculated angular distributions of K REC photons in 197-MeV/u
Xe + Be collisions divided by sin 2 0 and normalized at 90 ~ Dashed curve: Sauter
formula (9.32), see Sec. 9.4.2; solid curve: nonrelativistic Born approximation
calculation, Eq. (9.24); dash-dot curve: relativistic approximation calculation for
non-spin-flip transitions only, Eq. (9.18). See also Fig. 9.8.
i nt egrat i on over qb at large values of %, the emission angle 0 ~ enters directly
into gSf when p is inserted from Eq. (9.22). If X denotes the azi mut hal angle
of qb, i.e., the angle between qb and k~ in the i mpact paramet er plane, we
may write
E q3f(p)(p- fi~,)@~(q)12 oc
),
(qb COS X COS 0' - - q_ sin 0')9 4- q~ sin 2 X
(q~ - 2 hq b k ' cos X si n0' + h2 k '2 + q2 _ 2 hq _ k ' cos0' + Q 2p) 4
1
(q2 4- q_2_ 4- Q 2 T)4" ( 9. 24)
In order to illustrate the mai n features of this expression, it is useful
to i nt roduce some approxi mat i ons. The mai n cont ri but i on to the integral
in Eq. (9.21) arises for q+ = 0, in agreement wi t h the resonance condition
(9.8). The wi dt h A(hc~') ~ vQz w of the resonance is det ermi ned by the
Doppler spread in q+ permi t t ed by the t arget charge ZT. In the limit of
a vanishing t arget charge ZT or of ci ~ 0, the cont ri but i ons of transverse
moment a qb are strongly peaked around qb = 0. At the resonance, we t hen
t 1
have cf + ha l - -~rnev 2 and q_ ~ - me v . Since furt hermore ha/ << 2rnec 2
256 CHAPTER 9. RADI ATI VE ELECTRON CAPTURE
we can write the photon angular distribution in the projectile system as
do" sin 2 0 t
dft' e< (1 +/ 3 cos Or) 4" (9.25)
Using Eq. (9.10) to transform Eq. (9.25) into the laboratory system, one
finds at the resonance frequency
do
df~ e( sin 2 0. (9.26)
Hence the cross section vanishes in the forward and backward directions.
Indeed, since a transverse photon carries the angular moment um -1-1 in
its direction of propagation, transitions into atomic s-states of a spinless
electron moving along the z-axis are forbidden if the photon is emitted
along the positive or negative z-axis [FaM59]. In a relativistic description
(See. 9.5.3) where the electron spin is taken into account, forward and
backward emission is consistent with angular moment um conservation.
In an exact nonrelativistic calculation, if the integration over qb is ac-
tually carried out, one obtains an angular distribution very close to Eq.
(9.26), see Fig. 9.2. In order to derive the result (9.25), it is essential to
take into account the ret ardat i on of the photon momentum, i.e., not to use
the dipole approximation replacing e xp( - i k ~- if) ~ 1. This would lead to a
photon angular distribution in the projectile f r ame proportional to sin 2 0 ~.
As Eq. (9.26) shows, the forward-backward asymmet ry (9. 25)i n the pro-
jectile frame caused by the retardation is fully compensated by the Lorentz
transformation to the laboratory system. This was first pointed out by
Spindler [Spi79, SpB79].
The angular distribution (9.26) is in good agreement with experimental
dat a for intermediate projectile charges, see, e.g., Fig. 9.8, even at relativis-
tic velocities when the approximations leading to this result are no longer
justified.
The role of electron spin
So far, in Eq. (9.21), we have disregarded the electron spin. If one takes
into account the interaction of the electron magnetic moment with the
magnetic field of the photon, the situation changes. The conservation of the
angular moment um along the polar axis no longer prohibits photon emission
in the forward or backward direction, because a spin-flip transition can
compensate for the angular moment um carried away by the photon. This
means t hat in a relativistic description REC into 1=0 states at f orward or
backward angles can occur only by spin-f lip processes mediated by magnetic
9.4. THE PHOTOELECTRI C CROS S S ECTI ON 257
inter actions. Therefore, angular distributions may change significantly, see,
e.g., Fig. 9.10.
In leading order, spin-flip transitions are medi at ed by the magnetic
dipole t erm in the multipole expansion [Mos66] of the electromagnetic in-
teraction, which does not act on the space part of the wave functions.
Therefore, if the magnetic dipole interaction is evaluated between exact
bound and continuum radial wave functions, its mat ri x element vanishes
owing to orthogonality so t hat only the next order in the expansion of the
photon plane wave can give a finite contribution. O n the other hand, if the
magnetic dipole interaction is evaluated with appr oximate wave functions,
as provided by the Born or distorted-wave Born approximation, there will
be, in general, a finite overlap giving rise to spurious contributions. The
argument shows t hat spin-flip transitions are extremely sensitive to the use
of accurate wave functions.
We may i mmedi at el y draw a conclusion regarding relativistic collisions.
Since the st art i ng equation (9.12) for relativistic REC is based on the Dirac
c t operator which embraces bot h non-spin-flip as well as spin-flip transi-
tions, any Born t ype approxi mat i on will lead to incorrect spin-flip contri-
butions to the angular distribution.
O n the other hand, if one confines oneself to non-spin-flip transitions,
a relativistic Born t reat ment leads to results t hat are similar to those of a
nonrelativistic theory, disregarding the spin degree of freedom. Figure 9.2
shows the deviation of various angular distributions from a pure sin20 de-
pendence in the l aborat ory system. Besides relativistic and nonrelativistic
cross sections integrated over the REC resonance, the figure also includes
the differential cross section derived from the Sauter theory [Sau31] of the
relativistic photoeffect, see Sec. 9.4.2. In all cases, the depart ure from a
sin 2 0 behavior around 90 o is very small at this energy. The cross sections
vanish at forward and backward angles as is required by angular moment um
conservation.
Much more t han for non-spin-flip transitions, is it crucial for spin-flip
transitions and hence for spin-averaged relativistic angular distributions to
describe the initial electronic continuum st at e (with respect to the projec-
tile) by an exact Coulomb wave function or at least by a wave function t hat
is orthogonal to the final bound state.
9 . 4 Th e p h o t o e l e c t r i c c r o s s s e c t i o n
In order to avoid the problems discussed in Sec. 9.3, it is reasonable to
describe radiative electron capt ure as the modified inverse of the atomic
photoelectric effect: a continuum electron is capt ured by the projectile
258 CHAPTER 9. RADI ATI VE ELECTRON CAPTURE
with the simultaneous emission of a photon. As is discussed preliminarily
in Sec. 9.1, the original t hree-body problem is t hen reduced to a t wo-body
problem, and the t arget merely enters by providing an initial moment um
di st ri but i on of the cont i nuum electrons as seen from the projectile nucleus.
The initial moment um spread of the electrons bound in the t arget at om
leads to a broadeni ng of the phot on resonance.
As a first step in deriving REC cross sections, we discuss the photo-
electric cross section. For the photoelectric effect, approxi mat e formulas
for the differential cross section are available [PrL64a,b, PrR73], most l y
for K-electron emission. When a phot on with energy hw interacts wi t h a
hydrogenic t arget at om of charge Z, an electron with t ot al energy E (in-
cluding the rest mass) may be emitted. If I%1 is the binding energy of the
initially bound electron, the conservation of energy requires t hat
E - hc~ + m~c 2 - I %1 ,
(9.27)
so t hat there exists a one-to-one relation between E and hw. Before dis-
cussing the general procedure for calculating the exact photoelectric cross
section in Sec. 9.4.3, we first consider a nonrelativistic approxi mat i on in
Sec. 9.4.1 and in Sec. 9.4.2 a relativistic approach valid for low-Z atoms.
9. 4. 1 The St obbe f or mul a
Assumi ng hw << me c2 and a Z << 1, it is appropri at e to use a nonrelativistic
1 2
description. Let Ekin -- -~rneV be the kinetic energy of an electron ejected
from the K-shell. Then Eq. (9.27) is replaced wi t h
Eki n --- ~t(M- ICbl, ( 9. 28)
where
I%1
= 1 + = 1 + u 2 (9.29)
Eki n Eki n
is expressed by the Sommerfeld paramet er u = Ze2 / hv, see Eq. (5.52).
Using exact nonrelativistic hydrogenic initial and final states, disregarding
ret ardat i on effects, and adopt i ng the dipole approxi mat i on, Stobbe [Sto30]
has obt ai ned the photoelectric cross section per K-shell electron in the
form [BeS57, Som39]
287c2Ct meC2 ( P2 )3 e-4uarctan(1/u)
O'ph = 3 hw /~ 1 + u 2 1 - e -2~" ' (9.30)
where ~c - h/ mec is the Compt on wavelength of the electron.
9.4. THE PHOTOELECTRI C CROS S S ECTI ON 259
The corresponding differential cross section is
d(~ph : 3
df~p h O'ph ~ sin 20. (9.31)
The Stobbe cross section proves to be quite useful in REC calculations up
to projectile energies of a few hundred MeV/u, corresponding to electron
kinetic energies ( 7- 1)me c2, well below the electron rest energy.
9 . 4 . 2 Th e S a u t e r f o r mu l a
Sauter [Sau31] has derived an analytic expression for the cross section for
the relativistic photoeffect assuming/3 ~ 1 and disregarding terms of the
order a2Z2 in an c~Z expansion of the matrix element. For a discussion
of the approximations, see Fano et al. [FaM59] and Prat t et al. [PrR73].
The Sauter formula has been used to derive the corresponding REC cross
sections and to compare them with experimental data [RAY71, AnM84].
According to the Sauter formula [Sau31, Nag60, PrR73], the differential
photoionization cross section per K-shell electron for unpolarized photons
and electrons is
dCrph
df~ph
)5 ;~3
-- O~ 6 Z 5 me c2
ha~ )~ 7
sin 20 [
x (1 - ~e os 0) 4 1 +
7(7 - 1)(72 - 2)(1 - ~cos0)] . (9.32)
By integrating over the solid angle, one obtains the total cross section per
K-electron as
Crph
x/ 3 3 7a { 47( 7- 2) [
~+ 1
7+1
1 1+; ] }
- 2/3,)--------- 5 in 1 - . (9.33)
The Sauter formulas (9.32) and (9.33) have been improved by Gavrila
[Gav59, Gav60], by Nagel [Nag60, HUN67], and by Prat t [PrR73], who
included higher-order terms in aZ.
9 . 4 . 3 Ex a c t r e l a t i v i s t i c c a l c u l a t i o n s
The exact evaluation of the relativistic photoelectric cross section requires
a partial-wave expansion of the Coulomb continuum function. This means
260 CHAPTER 9. RADI ATI VE ELECTRON CAPTURE
t hat closed-form expressions can no longer be derived, and one has to re-
sort to numerical methods. Detailed formulations and t abul at i ons of cross
sections can be found in [Nag60, A1J65, HUN67, PrR73, Sco73]. However,
the existing t abul at i ons are not always sufficient for use in connection with
REC, which requires differential cross section on a sufficiently dense mesh.
Since for this purpose independent comput er codes have to be developed
[ICS94], we here present an outline of the formulation, assuming unpolarized
photons and a polarization-insensitive detector for the emi t t ed electrons.
For a given photon energy hc~ and binding energy lebl of the electron
in the bound initial st at e b, the final electron energy E (including the rest
mass) is determined by Eq. (9.27). In t reat i ng the electron spin, one may
either adopt a quantization in a fixed space direction or, alternatively, in
the direction of the electron moment um. Each of these approaches has
certain advantages.
Quantiz ation of the electron spin in the z - dir ec tion
We first consider the formally simpler representation, in which the electron
spin is quantized with respect to an arbi t rary z-direction. This is possible
if the electron spin is not measured and sometimes allows for a numerically
faster evaluation.
The differential cross section for a single electron is [PrR73] 1
daph me c2
_= o~ E E IMp, b( ms )~+' #b)12 (9.34)
d ~ 4 hw 2jb + 1 ' '
ptb ms
where we have averaged over the (2jb + 1) angular moment um projections
1
#b in the bound state and have summed over the spin components ms - +~
of the emi t t ed electron. Furthermore, we have averaged over the circular
polarizations )% = 1 and A_ = - 1 of the incoming photon. 2 Because of
the summat i on over all other angular moment um projections, #b, ms, this
is equivalent to taking one photon polarization, e.g., A+ = 1 only. The
quant i t y c~ denotes the fine-structure constant.
In analogy to Eq. (9.12), the transition mat ri x element (adopting rela-
tivistic units h = c = me = 1 in this subsection from here on) is
Mp,b(ms, )~+, #b) -- /
~ptp,ms (r) a . fi+ @ik.r ~]jb,lab (r) d3r,
( 9.35 )
1Equation (4.2.22) of [PrR73] for the normalization of the radial wave functions
contains a misprint. The expressions [(E i 1)1/2/2E] should be replaced with [(E
1)/2E] 1/2. In this case, agreement with [HUN67, A1J65] is obtained. The expression
(9.34) is consistent with t he corrected normalization (9.37).
2In conjunction wi t h an angular moment um representation, it is convenient to sum
over t he circular rat her t han over the linear polarizations A1,2.
9.4. THE PHOTOELECTRI C CROS S S ECTI ON 261
where, according to Eq. (4.123), but wi t h a different normal i zat i on,
1
1
Vp,m~(r) -- 47~ Ei l e - i A"
~,tt # -- ms ms
i f ~ ( r ) X'-~
J )
p
(9.36)
is t he part i al -wave expansi on of t he wave function describing t he relativistic
electron emi t t ed wi t h asympt ot i c moment um p (p2 = E 2 - 1, Pz = pcos0)
and spin proj ect i on ms wi t h respect to t he z-axis. The summat i on in
1
Eq. (9.36) ext ends over t he relativistic quant um numbers ~ - :i:(j + ~) and
t he angul ar moment um proj ect i ons p. Associ at ed wi t h g are t he orbital
angul ar moment a 1 = 1A and 1 ~ = 1B of t he upper and lower component ,
respectively, given in Eq. (4.85). The X~ are t he normal i zed t wo-component
spi n-angul ar functions (4.87), and t he g~( r) and f ~ ( r ) are t he (real) radial
cont i nuum wave functions (4.114).
The Coul omb wave functions are normal i zed on t he energy scale and
t hei r asympt ot i c behavi or is given by Eq. (4.118), t hat is
/
1 E
+
1
t / cos (pr + a~ )
r V 7rp
/
t / E- 1 si n(pr + a~). (9.37)
1
f~ -~ r V ~P
In t he limit of a point nucleus, t he Coul omb phase shift (4.120) is
a~ -- A~ + r/ln(2pr), (9.38)
where
At ~ z
1 1
arg - arg F(s + iT) - ~s
s + i ~
a Z E
= ~ , s - V/~ ~ - ( ~Z) ~. (9.39)
P
Since angul ar di st ri but i ons are sensitive to phases, one has to make sure
t hat t he phase conventions Eqs. (9.37) and (9.38) (or anot her set of consis-
t ent phase conventions) are si mul t aneousl y satisfied. 3
3In [ICS94] the generalization for an extended nucleus given by Miiller et al. [MuR73]
has been used.
262 CHAPTER 9. RADI ATI VE ELECTRON CAPTURE
Returning to Eq. (9.35), the exact hydrogenic bound-state wave func-
tion is identified with Eq. (4.86). If we now insert the expansion (6.42) of
the photon plane wave and perform the spinor algebra, we arrive at integrals
over three spherical harmonics which can be expressed by Clebsch-Gordan
coefficients (4.88). Collecting the results, we obtain
O O
Mp, b( ms, ~+, #b) -- 47rV~ E E iL+l-1 eiA'~
L=0
( 1 )
1 2 J F + ( L , n ) Y1 +1
x (0, 0) (9.40)
Pb n t- 1 -- ms ms Pb + 1 ~b , , b --'~s ,
where
F + (L n)
~t b '
[ /o
AL( j b Pb j ; l'bl) g~ ( r ) j L( k r ) f b ( r ) r 2 d r
- - AL( j b Pb j ; l b l ' ) f ~ ( r ) j L( k r ) g b ( r ) r 2 d r , (9.41)
and
AL( j b # b j ; / 1 1 2 ) - (2L + 1)V/2/1+ 1 ( /1 1
2/2 + 1 ~b -t-
12
x 1
#b -t- ~
( l l L
0 0
1 )(
-~ j l l
1 1
1 )
-~ jb
1
[2 #b
1
# b + ~
(9.42)
It is not difficult to carry the analytical reduction further [PrR73,HuN67] by
squaring the matrix element (9.40) and by expanding the resulting products
of spherical harmonics. In this way, additional summations are introduced.
However, for numerical evaluation, it is often convenient to take directly
the modulus square of the matrix element (9.40) and to introduce it into
Eq. (9.34).
Hel i c i t y r ep r es en t a t i on
An alternative description, convenient for calculating total cross sections
(9.46) and necessary for angular correlations, see, e.g., [Eic94], employs the
helicity representation (4.124) of the electron continuum wave function,
9.4 . THE PHOTOELECTRIC CROSS S ECTION 263
in which the electron spin is quantized along the asymptotic momentum.
Adopting a different normalization than in Eq. (4.124),
1
Cp,~(r) - ~i ' e -izx~ V/47r(2/+ 1) 1
~,# 0 O"
if ~( r ) x~_~
J )
O"
(9.43)
is the partial-wave expansion with the spin projection a with respect to p.
The rotation matrix D~( 2 ~ 15) carries each partial wave from an original
quantization along the z-axis (expressed by the Clebsch-Gordan coefficient)
into a quantization axis along p.
The differential cross section is then obtained as an expansion in terms
of Legendre polynomials P~(cos0) as
dO ' ph __ OL 7/~e C2 ~ E B~ P~(cos 0). (9.44)
df~ 4hw 2jb + 1
12
with the expansion coefficients
S/]
-- 87r ~' ~( - 1) ~b-1/2 ~ E( - 1 ) (L-1-L+l)/2 cos(A~ - A~)
ttb L~ ~
x ( 2/ + 1)(2j + 1)(21+ 1)(2~+ 1) 0 0
( ){ 1}
J 2 v 1 j
X
pb+l --pb--1 0 j 1 V
rL(L, U (Z,
0
(9.45)
where F+( L, n ) is given by Eq. (9.41). The summations occurring in
Eq. (9.45) are limited by the selection rules embodied in the Clebsch-
Gordan coefficients and 6j-symbols {.-.}. In particular, if the electron is
initially bound in a ls 1/2 state, a rather small number of terms contributes.
Nevertheless, the double sums occurring in Eq. (9.45) usually require longer
computing times than needed for squaring the undeveloped matrix element
(9.40). For numerically calculating the differential cross section, the latter
is therefore a more practical starting point.
Most of the published calculations have been performed for the K-shell
photoeffect, taking into account screening corrections which are needed for
neutral atoms. However, if one wants to use existing tabulations to derive
264 CHAPTER 9. RADI ATI VE ELECTRON CAPTURE
REC cross sections for bare projectiles, one has to compensate for these
undesired screening corrections [MeA85].
The total cross section
The total photoelectric cross section is obtained by integrating Eq. (9.44)
over the photon angles. Then only the term with u = 0 contributes in the
sum, and the total cross section per electron is
O~Tr me c2 )~
B0 (9.46)
~ph : h~d 2jb + 1
with the positive-definite expression
ttb t~
E iL F~+ (L, ~)
L
(9.47)
Ra d ia tive cor r ections
Radiative corrections to the atomic photoeffect [BOG82, McG77] can be
expressed by a corrective factor [1 (~/~)5] multiplying the basic differ-
ential cross section (9.34) or (9.44). The quantity 5, calculated to lowest
order in a and aZ, increases with the photon energy and the electron ejec-
tion angle. The correction term (c~/~)5 is always negative and of the order
0.2% for 100-keV, 1.5% for I-MeV, and 5% for 5-MeV photon energy. For
the REC experiments discussed in Sec. 9.6, the radiative correction is well
within the experimental uncertainty. Since for large photon energies the
correction becomes quite appreciable, it would be interesting to increase
the experimental accuracy to a point t hat allows for a test of these results.
9. 5 F r o m t he a t o mi c p h o t o e f f e c t t o RE C
9. 5. 1 Ra di a t i ve r e c o mb i n a t i o n
O nce the cross s ect i on O ph(~' , 0' ) of the photoelectric effect in the pro-
jectile frame (we reintroduce primed quantities) has been computed, it is
a simple mat t er to write down the cross section a RR( EI , 0 t) for radiative
recombination. This is the cross section for capturing a free electron of
energy E' into a bound atomic state (binding energy I%1) with the simul-
taneous emission of a photon of energy J . Since according to Eq. (9.27),
hut = Et - me c2 + ]s we may use E ~ as well as J - ktc to characterize
9.5. F ROM THE ATOM I C PHOTOEF F ECT TO REC 265
the process. We now choose the direction of the incoming electron as the z
direction, so t hat 0' is the emission angle of the photon.
By the principle of detailed balance, the cross section for radiative re-
combi nat i on into a vacant shell is wri t t en as
d 2 O'
~a a ( E ' , )
d E' d ~; h
( w' / c ) 2 d2(Yph(E ', 0' )
= (2jb + 1) p,2 d E' drY'el
(2jb + l ) [" hW' "~2 1 d20 ph( E' , O ')
~, 2 / 3272 d E' d~tel
. (9.48)
Here, O-ph is the photoelectric cross section per electron and the factor
(2jb + 1) converts the averaging of Eq. (9.34) over bound-st at e angular
moment um projections into a summat i on over a filled shell while the factor
( w' / c ) 2 / p '2 replaces the phase space factor of the outgoing electron by t hat
of the emi t t ed photon.
In applications to relativistic i on-at om collisions, the z-direction is usu-
ally defined as the direction of the projectile motion. This is opposite to
the direction of the electron moment um as seen from the projectile. Hence
for REC, the angle 0' of the photoelectric effect or of the radiative recom-
bi nat i on has to replaced by 7r - 0', or cos 0' is replaced by - c os 0'.
The t ot al cross section for radiative recombi nat i on into an empt y pro-
jectile subshell Jb is obt ai ned from Eqs. (9.48) and (9.46) in the form
e xa c t CtTl" ~tC0 g Uo
O'RR -- /~2~2 ?TteC2
(9.49)
where B0 is given by Eq. (9.47).
The Stobbe cross section
In the nonrelativistic limit, the factor in Eq. (9.48) is ( ha/ / r nec2) 2 (1//3272)
= ( hw/ 2rnec2) ( 1 +u2), so t hat the St obbe cross section (9.30) for the photo-
electric effect can be rewri t t en for radiative electron capt ure into an empt y
1
K-shell ( J b - 5) as
( /23 ) 2 -4uarctan(1/u)
Stobbe __ 287r20~ g e (9.50)
aRR - - 3 l +u 2 1- - e - 2~ '
where ~: -- h/ mec is the Compt on wavelength of the electron and u -
Ze2 / hv is the Sommerfeld paramet er. The const ant s in front of the u-
dependent t erms make up a factor of 9164.7 barn.
As ment i oned above, the St obbe cross section proves to be quite useful
to est i mat e REC into the K-shell up to projectile energies of a few hundred
266 CHAPTER 9. RADI ATI VE ELECTRON CAPTURE
MeV/ u, corresponding to electron kinetic energies ( ~/ - 1 ) mec 2 well below
the electron rest energy. The angular distribution is discussed in Sec. 9.4.1.
The S a ut er cross sec tion
For relativistic electron energies, one may apply the approximate Sauter
formula (9.33). Inserting into Eq. (9.48), one obtains the cross section for
radiative recombination into an empt y K-shell in the form
o. Sa ut e r 47rc~6Z5 (rneC2) 3
RR -- ha2 g
x/77 g+ 1- ~l n . (9.51)
7 + 1 2/372 1 - / 3
The angular distribution is given by Eq. (9.32). We notice t hat for
moderately relativistic collisions (7 -< 2), the leading term in Eq. (9.32)
has the same sin20 angular dependence in the laboratory system as the
nonrelativistic REC expression of Eq. (9.25). However, with increasing
values of 7, the second term in (9.32) becomes more i mport ant and leads to
a shift of the maxi mum in the angular distribution to forward angles. In the
extreme relativistic energy regime, the Sauter angular distribution (9.32)
behaves as cos 2 ~ except in the very forward direction. This dependence,
in its effect on REC, is illustrated in Figure 9.3 for various values of 7.
In the exact forward direction 0 = 0 as well as in the backward direction
0 = ~, the Sauter cross section vanishes, although, with the inclusion of the
electron spin, the conservation of angular moment um does not enforce such
a behavior. It has been conjectured [Nag60] t hat this curious feature of the
differential cross section calculated in first order of aZ, observed already
by Fano et al. [FaM59], may be accidental for the Coulomb potential.
Higher-order terms in a Z yield a finite contribution at 0 = 0 and at 0 = 7r
[SAW55, Nag60]. Exactly calculated differential cross sections are discussed
in Sec. 9.6.3.
For the exact relativistic cross section, Sec. 9.4.3, no closed formula
can be given. Numerically evaluated total cross sections are presented in
Sec. 9.6.2 and are compared with the Stobbe cross sections.
9. 5. 2 The mome nt um di s t r i but i on i n t he t a r ge t
In relativistic electron capture, the velocity of the captured electron is es-
sentially determined by the velocity - v of the target atom with respect to
the projectile. However, an electron bound in the target (which is assumed
to have a low charge ZT << Zp) has a moment um distribution p(q), so t hat
there is a moment um spread around the moment um - P0 = - Tmev.
9.5. F ROM THE ATOM I C PHOTOEF F ECT TO REC 267
I I I I I I I I I I I I j
J
1. 0
0. 8
k 0.6
N
~ 0. 4
~ _
O.2
3 0 ~ 6 0 ~ 9 0 ~ 1 2 0 ~ 1 5 0 ~ 180 ~
p h o f o n a n g l e
Fi gur e 9.3. Normalized Sauter angular distribution (9.32) for the emission of
REC photons integrated over the resonance and transformed to the laboratory
system. The curves are labeled by the projectile energy Tp in GeV/u. The
angular distributions for Tp < 1 GeV/u are almost indistinguishable from the
curve for 1 GeV/u.
Hydrogenic wave f un c tion s
If we assume a spherically averaged di st ri but i on for a single electron (av-
eraging over a compl et e principal shell n), t he nonrelativistic hydrogenic
moment um di st ri but i on is given by
1 8Q5n
Pn(q2) -- --~ E E 1~Sntm(q)12 = (9.52)
71-2 (q2 -t- Q 2) 4,
l rn
where ~Snl,~(q) is t he Fourier t ransform of t he electronic wave function in
t he st at e {nl r n} and
me (?2 ZT
Qn - ~ . (9.53)
n h
In order to approxi mat e t he full moment um di st ri but i on p(q2) in a mul-
tielectron t arget at om, one has to sum (9.52) over all t arget electrons,
inserting t he appropri at e effective charges in Eq. (9.53)
The moment um di st ri but i on (9.52) has a maxi mum at q = 0 and a
wi dt h det ermi ned by Qn: t he di st ri but i on becomes more and more peaked
around q = 0, t he lower ZT, and t he higher n.
268 CHAPTER 9. RADI ATI VE ELECTRON CAPTURE
Using Eq. (9.52), the Compt on profile (9.6) can be wri t t en in the form
f f n( qz ) - 27r pn(q2)qdq - 37r (q2 + Q 2n)3. (9. 54)
z
Roothaan-Hartree-Fock wave f unctions
For multielectron atoms, the hydrogenic density distribution may not be
accurate enough. More realistic moment um distributions can be obtained
from Root haan-Hart ree-Fock (RHF) wave functions t abul at ed by Clementi
and Roetti [C1R74]. In these calculations, single-electron orbitals i of sym-
met ry l, m are represented as linear combinations of Slater-type orbitals
(STO ) in the form
7~i,lm(r) -- E Ci,lnRln(?~)Ylm(r)' (9.55)
n
where n - {n, p} refers to the integer nodal quant um number n as well
as to the label p distinguishing basis functions with the same n, l m, while
Rln (r) is the STO radial wave function
Rln(? ~) -- Nlnrn-le-r (9.56)
with the normalization factor Nl n , the expansion coefficients Ci,zn, and the
optimized values ~n being t abul at ed in [C1R74] for all atoms and specified
principal quant um numbers i.
From Eqs. ( 9. 55) and (6.42), we obtain the Fourier transform of the
electron wave function as
~bi,zm(q) - (27rh) -3/2 47ri 1Yzm(0) ~ CilnXln(q),
n
where
Iln(q) -- Nln rn-le-r dr
(9.57)
(9.58)
l l n - -
Non n! ( q)
q (~2 _jr_ q2)(n+1)/2 sin (n + 1)arct an
Nln ( n - 1 ) ' ( q )
q2 ( ~ + q2)n/2 sin nar ct an
Win n' ( q )
q (r _~_ q2)(n-[-1)/2 COS (Tt -Jr- 1) a r c t a n ~nn "
(9.59)
0n - -
and j l ( qr) is the spherical Bessel function.
For convenience, we give the integrals (9.58) for s- and p-states, which
are derived in the same way as Eq. (8.13).
9.5 9 F ROM THE ATOM I C PHOTOEF F ECT TO REC 269
Integrals for higher angular momenta are similar. The spherically averaged
moment um density for a single electron in an i, 1 shell is
1 , 12
p(q2) = 21+l El ~i ~' c ( q)
m
2 Ci In I l n 9 (9.60)
71" n
Using the tabulated values, it is not difficult to calculate the momentum
distribution. In analogy to Eq. (9.54), one may also compute the Compton
profile by numerical integration.
9. 5. 3 Convol ut i on wi t h t he t a r ge t mome nt um de ns i t y
Since, with a probability given by p(q), the electron momentum within the
target has a component qb transverse to the beam direction, the momen-
t um seen from the projectile deviates slightly from the negative z direction
(which is defined by the beam) and the corresponding angular distribution
will be tilted by this amount. For example, for the rather low-energy colli-
sion of 197 MeV/ u Xe 54+ on Be, see Fig. 9.5, one estimates a deviation from
the negative z-direction by arct an(Q 1/ / ~) ~ 4.3 ~ while for higher energies
the angle will be less and for higher target charges greater. In general, the
effect on the angular distribution will be small, but for completeness, we
give a formulation which includes this effect.
For a given electron momentum q in the target, the Lorentz transformed
momentum q~ in the projectile forms the axis of an axially symmetric pho-
ton angular distribution. If we denote the polar and azimuthal angles of
the photon with respect to q~ as and r a given photon direction 0 ~ (we
choose the x-axis such that the photon azimuthal angle r = 0) in the pro-
jectile frame will receive contributions from various axes r (0q, Cq) and,
associated with them, from corresponding angles 0~ and r The angles are
illustrated in Fig. 9.4. Let us consider the spherical triangle composed of
the pole P0 = (0 = 0), of the point P~ = ( 0 ' , 0' = 0) giving the direction of
the emitted photon, and of the point Pq - (Oq, Cq) giving the direction of
9 m 0! !
the electron moment um q~ The sides in this triangle are So , s.~ - Oq,
!
and Sq - 0 ~. Knowing the angle Cq at the pole, we can express the opposite
_ ~
side So by the law of cosines as
O' ' 0' ' 0' ' . (9.61) cos --cosO qcos +si nO qsi n cOSOu
In this way, we may eliminate the angle O' by which the photon is emitted
with respect to the direction Cl'.
270 CHAPTER 9. RADI ATI VE ELECTRON CAPTURE
Z
P0
1
e
o
Fi gur e 9.4. The angles characterizing the photon direction and the electron
momentum in the projectile system.
We t hus get a generalization of Eq. (9.1) which in d oRR/ d f t ~ takes into
account t he deviations of t he electron moment a q~ from t he negative beam
axis and integrates over their contributions. Adopt i ng relativistic units,
h = c = me = 1, in this subsection, t he differential REC cross section in
t he projectile frame is
d2aREC(a/, 0' ) 1 d2aaEC(W ', 0 t) 1
da/ df F 27r dw ~ d cos 0 ~ 27r
f d2qb f d q z / dcosO ~
d 2aRa(q' , )
d~' d cos 0'
0 COS ~t
0 cos 0 ~
p(q2) 5(w' + E[ - ")'Ei -+- 7vq cos Oq)
x 5(cos 0' ' 0' ' 0' ' . - cO S0 qCO S - s i n0qs i n cOSCq) (9.62)
Here, r is measured wi t h respect to r chosen to define the x axis.
In order to eval uat e Eq. (9.62), we appl y Eq. (2.29) to derive t he
Lorentz- t ransformed vector q~ from q. For the angles, we get
, _q
sin Oq - qt sin Oq
, 7__ V/1 +q2
cOS0q = q, (q cos Oq - v ),
(9.63)
9.6. THEORETI CAL AND EX PERI M ENTAL RES ULTS 271
where
t _ ( . U '
q cos Oq - - qz -- w~ = qo, (9.64)
7v
with the resonance photon energy c~{) - 7 Ei - E} given by Eq. (9.8). Cor-
respondingly, the kinetic energy of the target electron, which, within the
impulse approximation, is assumed to be free, is fully determined by the
transverse momentum qb, SO t hat
E- v/q~ + q02 + 1. (9.65)
From the Lorentz transformation (2.29), we obtain
q, 2_ E, 2- 1_ 72 ( v/ q ~ + qg +1
w ~ "7- ~, ) 2 _ 1. ( 9 . 6 6 )
With these relations and (9.61), we are now able to evaluate Eq. (9.62).
The REC cross section is obtained as a two-dimensional integral over the
cross section for radiative recombination
d2 aREC 0')
dw' df F
1 d 2 oal ~ ( q ' 0' )
27r3'v d2qb dw' d cos
0 COS 0'
0 cos 0'
p(q~ + q~). ( 9 . 6 7 )
In the limiting c a s e qb << q' ~ v, SO t hat the axis of the angular distri-
, , _ q '
bution coincides with the z-axis, we have sin0q - 0 , cO S0q - ( Tv / ) -
- , - , 0' 0' 1 0c os 0' / 0c os 0' - 1 and cos - c os . We then get
d2aREC (W ', 0 ' ) 1 d 2 a Ra ( q ' , 7r - - 0 ' ) / d2
dw' d e' = 27c' ) ' v --~w~-dcossO ' qbP(q~ + qg), ( 9 . 6 8 )
which is identical to the usual result, Eqs. (9.5) and (9.6). The cross section
has a peak at q= 0 or
CU; -- ~E i - E; ( 9. 69)
with a Doppler width given by Q~ as defined in Eq. (9.53). The target
momentum density enters the cross section only via the Compton profile.
9 . 6 Th e o r e t i c a l a n d e x p e r i me n t a l r e s u l t s
9. 6 . 1 Ra di a t i ve r e c o mb i n a t i o n
The general dependence of REC cross sections on the projectile charge and
energy is best represented if we consider radiative recombination in which
272 CHAPTER 9. RADI ATI VE ELECTRON CAPTURE
. J 3
c-
O
0
0
0
I , _
o
n , "
n , "
1 03
1 02
101
100
10-I
1 0 -2
1 0 -3
' ' . . . . I ' . . . . . . . I '
Z =l O 0
2~.-~: ...
80 ~ . . . .
6 0 ~ . , , , ~ . ~ _
" " ' " " ' " ' " " " " . , , , . , . . . , . , . . . . .
4o . . . . . . . . . .
\
I I I I ' ' 1 , i i , , i i i I !
1 0 -1 1 0 0
En e rg y ( Ge V/u)
Fi gur e 9.5. Energy and charge dependence of the cross section for radiative re-
combination (in barns) with capture into the projectile K-shell. Solid lines: exact
(relativistic) calculated cross sections; short-dashed lines: Stobbe cross sections,
Eq. (9.50), assuming the correct relativistic relation between projectile energy
and velocity; long-dashed lines: Eq. (9.50), however assuming the nonrelativistic
1
relation E- -~ M v 2 between projectile energy and velocity.
the electron is assumed to be at rest in the initial state. In this case, no
t arget paramet ers enter. The exact total cross sections are given for the
K-shell by the solid lines in Fig. 9.5 and for the L-shells in Fig. 9.6.
A curious feature was noted in [StK92] by comparing t abul at ed exact
photoionization cross sections with the simple result (9.50) of the nonrela-
tivistic dipole approximation. Although these t reat ment s are widely differ-
ent, the total cross sections for radiative recombination agree rat her well for
energies below 1 GeV/ u, a result t hat was confirmed in systematic relativis-
tic calculations [ICS94]. Figure 9.5 presents a comparison of the exact cross
section calculated from Eq. (9.49) by including all partial waves I~1 __ 20
(solid line) with the cross section calculated according to the nonrelativistic
Stobbe formula (9.50) for radiative electron capture into an empt y K-shell.
The short-dashed lines are comput ed by assuming the correct relation be-
9.6. THEORETI CAL AND EX PERI M ENTAL RES ULTS 273
10 2
101
~ - 10 0
t -
O
. = , =
10- 1
03
03
03
o 10- 2
o
rr"
n"
10 . 3
10 - 4
~-' ' . . . . . I . . . . . . . . I '
Z =l O 0
" 80
r 6 0 ~,
=
4 0 " ' " ' ,
,.. ",~,. - , ~
\ -
%
10 . 5 10- ~ 100
Energy ( GeWu)
Fi gure 9.6. Energy and charge dependence of the cross section for radiative
recombination (in barns) with capture into the projectile L-shell. All results are
obtained from exact relativistic calculations. Solid lines: capture into the L 1-shell;
long-dashed lines: capture into the L2-shell; and short-dashed lines: capture into
the L3-shell.
tween the velocity v and the projectile energy, while the long-dashed line
1 2
uses the relation Ekin -- -~Mpv , as if the projectile speed were nonrela-
tivistic. For the latter case, the results of the nonrelativistic dipole approx-
imation do not deviate much from those of the exact calculations, even for
relativistic energies (up to 1 GeV/ u), where the energy-velocity relation
is far from the nonrelativistic limit. This astonishing agreement must be
regarded as fortuitous.
A similar equivalence cannot hold for the differential cross section, which
nonrelativistically would be represented by a strict sin 2 0 distribution (9.31),
while in the exact cross sections, considerable deviations from this form
occur. O ne reason is t hat spin-flip contributions to the cross section yield
nonzero values at forward and/ or backward angles, see Fig. 9.10.
In analogy to Fig. 9.5, Fig. 9.6 shows the exact relativistic RR cross
sections for capture into the L_l , L2- , and L3-shel l s corresponding to final
274 CHAPTER 9. RADI ATI VE ELECTRON CAPTURE
2sl/2, 2pl/2 , and 2p3/2 states. Here, capture into the Ll-shell dominates
for all charges and energies.
9. 6 . 2 Tot a l REC c r os s s e c t i ons
In order to present t ot al REC cross sections in a systematic way, it is use-
ful to render different collision systems comparable (a) by normalizing the
cross sections per target electron, (b) in the case of incompletely stripped
projectiles, by considering cross sections per projectile vacancy in the shell
considered, and (c) by displaying cross sections as a function of the "adia-
baticity paramet er" ~], which decides whether the collision is fast (~ > 1) or
slow (~ < 1) in the sense of Sec. 5.4.1. According to Eq. (5.63), we define
-- Ekin ( Me V/ u)
m~c2(~, 1) ~ 40.31 . (9.70)
~/ -- Ebind Z~
The last approximation assumes a nonrelativistic total binding energy and
a nonrelativistic energy-velocity relation.
Experi ment al cross sections normalized per t arget electron according to
(a) may be compared with theoretical cross sections for radiative recom-
bination, see Eq. (9.49). In this way, t arget properties do not enter since
the electrons are assumed to be at rest with respect to the t arget nucleus,
which is a good approxi mat i on for low-Z targets. Figure 9.7 shows t hat
the available experimental total REC cross sections per t arget electron as
a function of the adiabaticity paramet er r/ fall approxi mat el y on a single
universal curve [StK92]. This behavior finds its count erpart in the exact
theoretical total cross sections for radiative recombination. It t urns out
t hat for calculating the total cross section, it is sufficient to add up the
contributions from the K-, L-, and M-shells [ICS93, StK95]. While for the
K- and L-shells, the curves for the different projectile charge numbers are
well separated as a function of the collision energy, as shown in Figs. 9.5
and 9.6, and similarly for the M-shells, t hey form a close-lying bundle as a
function of the paramet er ~ in the representation of Fig. 9.7. The theoret-
ical curves, which all range from 100 MeV/ u to 1.0 GeV/ u, are in rat her
good agreement with the experimental dat a points. Wi t h one exception,
only experimental dat a for projectile energies larger t han 140 MeV/ u are
included. However, there exist many more dat a points at lower energies
which follow the bundle of theoretical curves to higher values of r/.
We conclude t hat total REC cross sections can be reliably est i mat ed by
theoretical cross sections for radiative recombination, which are domi nat ed
within each principal shell by capture into s-states. Now it is a general
observation [RoG94] t hat nonrelativistic dipole total cross sections (but
9.6. THEORETI CAL AND EX PERI M ENTAL RES ULTS 275
1 0 3 . . . . . . ~ . . . . . . . . ~ ,
. . o
12:
0
o
( D
oo
O
k _
( D
10 2
101
1 0 o
%
8 O
" ~ " ' ~ " 6 0
0 X e "' -
9 Au , , ' , .
a Pb "9
9 Bi
@ U
10-1 , , , , , , ,
I O 0
I I i I I I I I I
101
Ad i ( ]be t i c i t y p( ] rume t e r
40
x
x
x
x
x
Fi gur e 9.7. Experimental total REC cross sections for capture into bare pro-
jectiles per target electron in barn as a function of the adiabaticity paramet er r/
defined in Eq. (9.70). The projectiles are indicated explicitly; the target atoms
range from Be to A1. In deriving total cross sections, realistic theoretical an-
gular distributions are taken into account. From [StK95]. The curves represent
exact theoretical cross sections for radiative recombination into the K + L + M
shells calculated according to Eq. (9.49) [ICS94]. The numbers give the projectile
charges.
not angul ar di st ri but i ons) for t he phot oel ect r i c effect of s-st at es (and ot her
processes involving s-st at es) are valid far beyond t hei r ant i ci pat ed energy
limits, see also Fig. 9.5. Ther ef or e it is not surpri si ng t hat exact t heor et i cal
REC cross sect i ons are well appr oxi mat ed by consi st ent l y nonrel at i vi st i c
St obbe cross sections. If one i nt r oduces t he l at t er in Fig. 9.7, t he velocities
and charge number s ent er only in t he combi nat i on u o( 1/ x/ ~, see Eq. (9.50),
so t hat t he bundl e of exact t heor et i cal curves cont r act s to a single uni versal
c u r v e .
276 CHAPTER 9 . RADI ATI VE ELECTRON CAPTURE
10
,.s 8
r
.g 6
0 0
4
o
0
2
0
0 3 0 6 0 9 0 1 2 0 1 5 0 1 8 0
Angle ( d e g )
Fi gur e 9.8. Differential REC cross section in barn/sr for 197 MeV Xe 54+ on Be
as a function of the laboratory angle. Measured values are from [AnA84]. The
solid curve represents calculated results [ICS94] multiplied with a normalization
factor of 0.77 in order to fit the data. Hartree-Fock wave functions for the target
atom [C1R74] are used.
9 . 6 . 3 Di f f e r e n t i a l RE C c r o s s s e c t i o n s
As discussed in Sec. 9.3, in the theoretical description of nonrelativistic
REC, two ingredients combine in such a way t hat the resulting phot on an-
gular di st ri but i on is simply proport i onal to sin 2 0 in the l aborat ory system.
The two ingredients are (a) the inclusion of the full ret ardat i on, t hereby
keeping the complete exponential of the phot on wave function in Eq. (9.11),
and (b) the t ransformat i on (9.10) of the angles from the projectile system
to the l aborat ory system.
As an example, Fig. 9.8 shows measured and calculated REC cross sec-
tions for 197 MeV/ u Xe 54+ projectiles on Be targets. The calculated cross
section is normalized to fit the data. The normal i zat i on factor is 0.77,
which may be within the syst emat i c errors of the experiment. Aside from
the overall normalization, the experi ment al angular dependence is well re-
produced by the theory. For the moment um di st ri but i on within the t arget
at om, Hartree-Fock wave functions [C1R74] have been used. The calculated
t ot al K- REC cross section is 81.85 b, compared to an i ndependent l y mea-
9.6. T HE ORE T I CAL AND E XPE RI ME NT AL RE SUL TS 2 7 7
5,, ,
4
,,o, n- 1 L - s h
O80 100
L - s hell
K - s hell
' 2 0 ' ,' 8o
P hot on ener gy ( k eV )
F i g u r e 9. 9. Cal cul ated REC phot on spect rum for 2 95 Me V/ u U 9 2 + o n N atoms
for the l ab oratory phot on ang le of 1320 . Solid line: cal cul ated wi th approxi-
mat e Roothaan- H artree- Fock wave functi ons for the targ et atom; dashed line:
cal cul ated wi t h a hydrog eni c moment um di stri b uti on. From [ICS94] .
i i I i I
~ - 5 0 K - s h e l l t o t a l
.13
C
.s 4 0
( 9
O0
3 0
O
t.ul
a: : 2 0
; : 5" ~ 10 " , j . , . . . . . - .......
0 . . . . . . . . . . . . ~
0 3 0 6 0 9 0 1 2 0 1 5 0 1 8 0
Angl e ( d e g )
F i g u r e 9. 10 , Cal cul ated differential K- and L- shell REC cross secti ons for 2 95
Me V/ u U 92+ on N atoms as a functi on of the l ab oratory ang le. A pproxi mat e RH F
wave functi ons have b een used. For K- REC, the contri b uti ons wi t hout spin- flip
and wi th spin- flip are shown separately. A dopt ed from [ICS94] .
278 CHAPTER 9. RADI ATI VE ELECTRON CAPTURE
sured total capture cross section (which is mainly REC, see Fig. 8.9) of
(72 + 3) b.
In Fig. 9.9, we show the calculated photon spect rum at the fixed ob-
servation angle of 132 o in the l aborat ory system for the collision of U 92+
+ N at 295 MeV/ u. The maxi ma occur at the photon resonance hwD,
see Eq. (9.8), corresponding to zero longitudinal moment um qz = 0 of the
electron in the t arget atom. The widths of the peaks for K-REC and L-
REC are caused by the electronic moment um spread in the target. The
dashed lines assume a hydrogenic moment um distribution (9.52), while the
solid lines are calculated from approximate Root haan-Hart ree-Fock wave
functions according to Eq. (9.60). A comparison of hydrogenic and RHF
results reveals t hat , as expected, screening effects reduce the moment um
spread, in particular of outer electrons, so t hat photon spectra get more
sharply peaked. However, the area under the peak is not much affected,
leaving angular distributions (Figs 9.8 and 9.10) less sensitive to electron
screening.
Figure 9.10 shows the differential REC cross section for capture into
the K- and L-shells of the projectile assuming a RHF moment um distri-
bution in the target. O ne notices strong deviations from a simple sin20
distribution predicted by the nonrelativistic theory and observed in Xe 54+
+ Be collisions, see Fig. 9.8. This difference must be at t ri but ed mainly
to the higher projectile charge of U 92+ as compared to Xe 54+. Relativis-
tic effects become more evident if one splits the K-REC cross section in
Fig. 9.10 into a non-spin-flip part and a spin-flip contribution. The non-
spin-flip part still follows approximately a sin 2 0 distribution, in accordance
with Fig. 9.2, while the spin- flip distribution is peaked at forward angles.
As discussed in Sec. 9.3, the spin degree of freedom is required to conserve
angular moment um at forward (or backward angles). More details can be
found in a systematic discussion of angular distributions [EiI95].
In Fig. 9.11, the subshell contributions are displayed separately for 90
MeV/ u U 9~ collisions with carbon foils. At this energy, it is experimen-
tally possible to resolve the contribution from the L3-shell (P3/2 states, lower
curve) from the combined contributions of the L1- and L2-shells (2Sl/2 and
2Pl/2 states, respectively, upper curves). The theoretical angular distri-
butions deviate markedly from symmet ry about 90 ~ with the shift of the
maxi mum depending on the subshell. As one can see from Fig. 9.11(a), the
forward peaking for REC into the into the j = 1/2 sublevels can be ex-
plained by transitions to the 2sl/2 state (dashed line), which can be part l y
at t ri but ed to spin-flip transitions in analogy to Fig. 9.10. At backward an-
gles, the behavior is mainly determined by REC into the 2Pl/2 state (dotted
line). The theoretical angular distribution for REC into the 2P3/2 state, see
9.6. THEORETI CAL AND EX PERI M ENTAL RES ULTS 279
60.0
=
5 0 . 0
L_
9 , ~ 4.0 .0
no~ 30 . 0
n o 2 0 . 0
10.0
0"0 f
2 5 .O ,t
2 0 . 0
t . .
~ 15 .0
CX
n o
" o 1 0 . 0
n o
' I ' I ' I ' J
I
o o ~ . . . . . .
. f l . . . . . " ~%%
I %%
b )
' 0 r t
0 . 0 , ' , I , I ,
4 5 9 0 1 3 5 1 8 0
t ab or at or y angt e Ot.b ( d eg)
Fi gur e 9. 11. Differential L-shell REC cross sections for 89 MeV/ u U 9~ on C
atoms as a function of the laboratory photon angle. (a) REC into the j - 1/2
levels. Experimental data: diamonds; theoretical cross sections [ICS94]" dashed
line: 2Sl/2, dotted line" 2pl/2 , solid line: sum of both. (b) REC into the 2p3/2
level. All dat a points are normalized by the common factor 0.65. From [StG94].
280 CHAPTER 9. RADI ATI VE ELECTRON CAPTURE
Fig. 9.11(b), is enhanced at backward angles similarly as the 2Pl/2 curve,
which is in accord with the experimental data. All data points are normal-
ized by a common factor of 0.65, which is still within the total absolute
uncertainty of the measurements [StG94].
The experimental results are in excellent agreement with theoretical
predictions using an exact relativistic description. O ne may conclude t hat
for high projectile charges, subshell-resolved photon angular distributions
probe details of the bound-state projectile wave functions and, at forward
(or backward) angles, the contribution of the electron spin caused by mag-
netic interactions. Accurate relativistic calculations are needed for reliable
predictions.

Você também pode gostar