7 views

Uploaded by Calinescu Ionut-catalin

Relativistic Atomic Collisions
Excitation and ionization of one-electron atoms by relativistic projectiles
are treated simultaneously in this chapter and are not distinguished. Most applications refer to K-shell ionization which is the important process at high velocities. The theoretical methods used for relativistic collisions are very similar to those applied at nonrelativistic bombarding energies.Excitation and ionization of one-electron atoms by relativistic projectiles are treated simultaneously in this chapter and are not distinguished. Most applications refer to K-shell ionization which is the important process at high velocities. The theoretical methods used for relativistic collisions are very similar to those applied at nonrelativistic bombarding energies.

- Aieee 2002 q
- Penny EP Stuff > Eng Phys Serway 2
- Jntu Ece Syllabus
- Metal Clusters and Magic Numbers
- Tutorial 3
- advbv
- 179439_pchem3_ch22
- QMprelim_F16
- Tensor Network in Chemistry: Recent DMRG/TTNS Studies and Perspectives for Catalysis Research
- cosby1993.pdf
- n Comms 14277
- Gianni Dotto Ring Patent
- Discharge Bifurcation
- Development of Medium Pressure Laser Ionization
- 947290680[1]
- innov-x_alpha_series.pdf
- Baum-Manz
- 34072709
- 2012 james ruse phys.pdf
- notebook 4

You are on page 1of 45

Ex c i t a t i o n a nd I o n i z a t i o n

Excitation and ionization of one-electron atoms by relativistic projectiles

are t reat ed simultaneously in this chapter and are not distinguished. Most

applications refer to K-shell ionization which is the i mport ant process at

high velocities. The theoretical methods used for relativistic collisions are

very similar to those applied at nonrelativistic bombarding energies.

In Sec. 6.1 we start with an outline of first-order pert urbat i on theory,

and in Sec. 6.2, for the sake of illustration, we discuss the simple cases of

the dipole approximation and of a quasirelativistic t reat ment of the mat ri x

elements. Some of the methods used to evaluate the necessary interaction

mat ri x elements more exactly are presented in Sec. 6.3, so t hat one is in

a position to derive theoretical results and to compare t hem with the cor-

responding experimental data. In Sec. 6.4 we discuss the finite-difference

met hod for solving the time-dependent Dirac equation of the collision sys-

tern in a nonpert urbat i ve way. As an alternative nonpert urbat i ve method,

we consider two-center coupled-channel methods in Sec. 6.5, and subse-

quently specialize to single-center coupled-channel calculations.

For definiteness, we assume in this chapter t hat a single target electron

is excited or ionized by the projectile. If the active electron is carried by the

projectile, excitation can occur by the target nucleus, and one has simply to

interchange the role of target and projectile. Excitation is then influenced

by the target electrons. These and other many-electron effects are discussed

in Chap. 7.

139

140 C H A P T E R 6. E X C I T A T I O N A N D I O N I Z A T I O N

6. 1 Fi rs t - order pert urbat i on t heory

For a fast collision and for impact parameters b comparable with, or larger

t han the atomic K-shell radius, the transient perturbation of the target

atom by the projectile is small enough that the first-order time-dependent

perturbation theory is expected to be a good approximation, even for a

high-Z projectile.

6. 1. 1 General expressi ons

In first-order perturbation theory, by combining Eqs. (5.76) and (4.32), the

transition amplitude for excitation and ionization of the target electron is

written as

e ~/ZPe2 / i (Ef E i ) t / 1

Aft(b) -- i h dt e ~ - d3rT r -- Zaz) D- ~ i(rT). (6.1)

rp

Here, 7 is the Lorentz factor, Zp the projectile charge number, and ~i

and ~f are the initial and final eigenfunctions of the target electron with

energies Ei and Ef. The coordinates are explained in Fig. 3.2, and r},, given

by Eq. (5.97), is the electron-projectile distance measured in the projectile

rest system. The cross section for excitation is directly given by Eq. (5.65),

while for the differential ionization cross section we have to replace aft by

d a f i / d E f , so t hat for each specified initial and final state (see below) we

have

d a _ 27r [A~ll 2 bdb. (6.2)

d E f

In this formula, it is assumed t hat the radial continuum wave function

for each final partial wave ~f or jf, If entering in the transition amplitude

is normalized per unit energy interval, see Eq. (4.117). Since the emit-

ted electron is not represented by a plane wave but a Coulomb wave, the

approaches discussed here belong to the class of distorted-wave Born ap-

proximations (DWBA). Introducing the minimum momentum transfer hq0

through Eq. (5.105), the integration in Eq. (6.1) over the time dependence

residing in the exponential factor and in the term 1 / r ~ can be directly

carried out to yield [BAH59, AmA81]

2 qo ~ / ( X T _ b ) 2 + y~ ) i> 2 (6.3)

[A~[ - 4~ ] / f l ( l - 3a z ) e i % z T K o ( - ~

Here and in the remaining part of this chapter, we use the bracket notation

for the space integral. Furthermore, ~p is given by Eq. (5.52), and K 0

denotes the modified Bessel function. It is seen t hat the range parameter

6.1. FI R S T - O R D E R P E R T UR BA T I O N T H E O R Y 141

- 1

of the effective interaction is q0 along the path, i.e., parallel to the z-axis,

but is ~ / qo 1 perpendicular to it, i.e., in the impact paramet er plane. Hence,

as 7 increases, the integrand falls off more slowly with b t han in the beam

direction. This reflects the properties of the Li~ nard-Wiechert potential

given by Eq. (3.34) and depicted in Fig. 3.3.

For mul at i on i n t e r ms o f the Four i er t r ansfor m

For an actual evaluation of the cross section, it is more convenient to return

to the expression (6.1) and to rewrite the retarded interaction in terms of

its Fourier transform. Using the relativistically generalized Bethe integral

(5.103), the electronic transition mat ri x element in Eq. (6.1) can be cast

into the form (5.104). By decomposing q as q = (qb, qz) and performing

the time integration, one obtains from Eq. (5.106) the transition amplitude

A~(b) - i T vup Mfi(b) - -Tri u,,_ f

d2 qb

q2 + ( 1 -/ 32)q~)

and from Eq. (6.2) the cross section

C -i qb' b Mfi (q) (6.4)

~0 ~176

d ~ _ 87rr,2 qb dqb [Mfi(q)l u (6.5)

dEf [q2 + ( 1 - 32)q~]2 ,

where

Mfi(q) = (fl(1 - ~ a z ) c i q r T [i ) . (6.6)

The advantage of this formulation is t hat the complicated time depen-

dence contained in the expression (5.97) for r~, is removed and replaced

with the more explicit wave number dependence (6.5) and (6.6). The ex-

pression (6.5), valid for each target electron and for specified initial and

final states, agrees with the corresponding expression derived in a wave

picture, in which certain simplifications, described in Sec. 5.2, are adopted

[DAM78, Anh79].

Speci fi cat i on of the i ni t i al and fi nal st at es

For exci t at i on, the initial and final bound states i and f are specified by

their quant um numbers i - {hi, ~i, Pji } and f - {nf, ~f, #jf}, respectively,

see Eqs. (4.84) and (4.85). In order to obtain the complete cross section,

one averages over the initial angular moment um projections PJi and sums

over the final angular moment um projections pjf.

For i oni z at i on, Eqs. (6.2) and (6.5), the initial bound state is charac-

terized in the same way. However, for the final state of a given continuum

142 C H A P T E R 6. E X C I T A T I ON A N D I ONI Z A T I ON

energy Ef, one has to distinguish two possible situations determined by the

experimental setup.

(a) The energy Ef and the direction fr of emission of the electron is

1

detected. Then the final state f} - I Ef , fr a) with spin projection c r - +2

is a coherent superposition of partial waves ~f, pjf given by Eq. (4.124).

By inserting into (6.2) or (6.5) and summing over the (undetected) spin

projections, one obtains the angle-differential cross section.

(b) The direction of emission of the electron is not detected. Then one

integrates Eqs. (6.2) and (6.5) over the direction 1r of the emitted electron

in Eqs. (4.124) or (4.123) and sums over the spin projection a (or ms).

Owing to the orthogonality of the rotation matrices in (4.124) [or of the

spherical harmonics in (4.123)] and of the Clebsch-Gordan coefficients, one

obtains an incoherent summation over the partial waves ~f, #jf.

The resulting total energy-differential cross section for ionization is then

written explicitly as

dEf total

1 f ~ qb dqb

87r@ 2ji + 1 E E J0 [q~ + ( 1- f12)q212

~Ji ~f ' ~Jf

<Ef,~ f,#jfl(1 - fl.~)e* qrTlni,~,~ji) ,

(6.7)

where fl = v/ c.

In the following, we keep the symbols i and f for fully specified states,

i.e., [i} -- Ini, Ni, PJi ) and f} -- IEf, Nf, pjf }. Only when calculating angular

distributions, do we identify f with a coherent superposition If} - ]Ef, 1r a>

of partial waves given by Eq. (4.124).

6. 1. 2 L ong- r a nge c oupl i ngs a nd Coul omb bounda r y

c ondi t i ons

In the preceding subsection, we have assumed that the initial and final

states are undistorted target eigenstates. However, according to the discus-

sion in Sec. 5.3.3, we should use the exact solutions 1 (5.61)

9 t ) - t ) (6.8)

of the asymptotic Dirac equation (5.49), that is, we should take into account

the asymptotic distortion caused by the residual interaction - S2 Z p e 2 / R ~.

Inserting the solution (6.8) into Eq. (5.82), one finds from Eqs. (5.82)

and (5.83) that, by construction, the residual interaction term proportional

1An irrelevant constant phase factor depending on the unit of length is discarded.

6. 1. F I R S T - O R D E R P E R T UR B A T I O N T H E O R Y 143

to 1 / R ~ has to be subt ract ed from the interaction occurring in Eq. (6.1),

see Eq. (6.9) below. The reason why one usually disregards the residual

interaction t erm derives from nonrelativistic collisions, where, according

to Eq. (5.47), R- R ~, so t hat a term proportional to 1 / R has vanishing

mat ri x elements between orthogonal states.

For relativistic collisions, the situation is somewhat different. Let us ex-

amine the modified transition amplitude A e calculated with phase-distorted

states ~ satisfying Coulomb boundary conditions [ToE90b]. Since the

state-independent phase factors in Eq. (6.8) for initial and final states can-

cel each other, one has

.A~ (b)

i ~ZPe2 /

h d t e ~ ( E f - E i ) t

t J

x/ f ( l - ~ a z ) ( l

1

R,)

i) ( 6.9 )

instead of (6.1). The additional term 1/ _R I in the pert urbat i on operator

with R ~ = v/ b 2 + v2 t ~2 depends on the laboratory space coordinates t hrough

t ~ = ~ / ( t - VZ T / C 2 ) . In contrast to the nonrelativistic case, the space inte-

gral of the 1 / R ~ term does no t vanish. However, we still have to perform

the time integration. Since the integration f d 4z over the complete four-

dimensional space is Lorentz invariant, we may as well evaluate this integral

in the projectile frame. In the moving coordinate system, the term 1 / R ~

is a constant, therefore does not contribute to the mat ri x element between

orthogonal (exact) eigenstates, and hence can be discarded. We may now

transform back to the l aborat ory frame to obtain a result wi t h o ut the 1/_R ~

term, t hat is, we recover Eq. (6.1). In other words, Eqs. (6.1) and (6.9) are

equivalent, and Coulomb boundary conditions are without any influence.

It should be emphasized here t hat the equality of the two t reat ment s is

valid only in first-order pert urbat i on theory and only if we consider the

occupation amplitude for the final state at time t ~ oc.

Obviously, the same reasoning is valid for any function f ( R ~) and the

corresponding phase distortion. This means t hat the first-order transition

amplitudes for excitation and ionization (and for pair production) are in-

variant with respect to gauge transformations. This has also been confirmed

in [RuS93] by a different method.

As an illustration, we show in Fig. 6.1 the time evolution of occupa-

tion probabilities calculated from Eqs. (6.1) and (6.9), respectively, with

the modification t hat the upper limit of the time integration is taken as

t rather t han as oc. In the upper part of the figure, calculated from the

finite-time version of Eq. (6.1), we notice the effect of the long-range cou-

plings which lead to a sizable population of 2P3/2 and 2Pl/ 2 states in the

144 C H A P T E R 6. E X C I T A T I O N A N D I O N I Z A T I O N

1.0

1 0 - 1

1 0 - 2

> , 1 0 - 3

. _

. _

10- ~

o

c..

,-, lO-S

r -

0

. _

-4-

o 1.0

o 10 -1

1 0 - 2

10- 3

1 0 - 4 "

1 0 - s

1 G e V / u l s ~ l z

U g z * + U 9 1 +

u n p e r t u r b e d B o r n , - v- - - - - 2 s ~'lz / , l s ~ l z

/ / ~ - I 2 p ; , ~

/ / I r _ t ~ - z p 3 / z

+ 2 P3-/z

l s ~ / 2

b o u n d a r y - c o r r e c t e d B o r n

2s~'~z

2 p

' - 6 . 2 o . o 0 . 2 o . +

v t ( a . u . )

{ 2 l s ~ / z

2 p ~ / z

p3+, ~

Fi gur e 6.1. Time evolution of the occupation probabilities of target states in

U 92+ -~- U91+(lSl/ 2) collisions at 1 GeV/ u. Calculations are performed in the

Born approximation at b = 0.01 a.u. (about one K-shell radius). Upper part:

unperturbed initial and final states, Eq. (6.1); lower part: boundary-corrected

states, Eq. (6.9). The abscissa is the projection of the projectile-target separation

on the beam direction. The projections of the angular momenta are indicated by

1 1 3 and 3

+ , - , + for mj - - +~ , - ~ , +5, - ~ , respectively. From [ToE90b].

ent rance channel for projectile-target separations as large as 0.4 a.u. ( about

37 K-shell radii). On the ot her hand, in the lower part of the figure, the

finite-time version of Eq. (6.9) has been used. Here, as would be expect ed

from the short -range pot ent i al in Eq. (6.9), the long-range couplings have

di sappeared. For large positive times, the occupat i on probabilities in the

unper t ur bed and in the boundary-correct ed Born approxi mat i on are ex-

actly the same, thus verifying the reasoning given above. In a numerical

evaluation of the time integrals, one expects t hat the boundary-correct ed

version converges much more rapidly.

6. 2. T WO S I MP L I F I E D T R E A T ME N T S 145

6.2 T w o sim plif ied treatm ents

The t r eat ment presented so far is exact wi t hi n the first-order semiclassical

approxi mat i on of Eq. (6.1). For the sake of an i nt roduct ory illustration,

we now present two different approxi mat i ons, one valid at low values of q0,

which means small excitation energies or high projectile velocities, the other

valid for C~ZT << 1. These met hods are called the dipole approxi mat i on and

the quasirelativistic t reat ment , respectively.

6. 2 . 1 D i pol e approxi m at i on

In Bet he' s dipole approxi mat i on [Bet30], one replaces in the mat ri x element

(6.6) the factor exp(i q- r T) by 1 + i q. r T. Unless the wave number q

becomes larger t han the inverse K-shell radius a K (for K-shell excitation

or ionization) it is justified to neglect higher-order t erms in the expansion

because t he exponential decay of the bound-st at e wave function cuts off

cont ri but i ons from larger r values.

If we rewrite the Dirac mat ri x a z with the aid of the identity (5.91),

valid for exact eigenstates, and with the definition (5.105) for q0 = qz , the

dipole approxi mat i on yields

Mfi ~ <fl(1 - 13a=)eiqrT li> ~ i ( q" <flrTli> - 32qo<flzTli>), (6.10)

where each t erm obeys a different selection rule. Int roduci ng spherical

coordinates rs wi t h the magnet i c quant um number s - 0 , +1, so t hat r 0 =

z w and r +l - T~ ( x T + i YT ) , and subsequent l y defining reduced mat ri x

elements [Ros57] t hrough

/

Ji

1

( J f P j f Irs Ji Pj i > - - [

k

P ji 8

i f )

P j f

(f ]] rw I] i>, (6.11)

where Ji, P ji and jf, pjf are the angular moment a and their projections

of initial and final states, and ( . . ]-) is a Cl ebsch-Gordan coefficient, see

Eq. (4.88), we get

_qo2( 1_/ 32) 2 ( Ji 1

\

P ji 0

2

Jf

\

] [ ( f [] rT ]] i} [ 2

#j i

/

IM~I 2

and

[2 _ 1 2 ( 1

- - 2 ( qx + q~) Ji

\

Pj i + 1

P j f -- P j i '

(6.12)

j f ) 2

Pji i 1 [(f ][ rT ]] i)[2

if P j f -- P ji -~- 1.

(6.13)

146 C H A P T E R 6. E X C I T A T I O N A N D I O N I Z A T I O N

Bot h types of terms contribute to the cross section after summat i on over

final angular moment um projections, provided initial and final states have

different parity and the dipole selection rules embodied in the Clebsch-

Gordan coefficients are satisfied. In writing down Eqs. (6.12) and (6.13) we

still mai nt ai n a relativistic description of the electronic states. Expressions

(6.12) and (6.13) have different q dependencies and cannot be combined

unless the squares of the Clebsch-Gordan coefficients are equal. For the

particular transition Sl/2 --+ P l/ 2 all ( . -I. ) 2 - 1/ 3, but in general, e.g., for

sl/ 2 ~ P3/ 2, this is not true.

In order to evaluate the cross sections for each of the two cases, we

write q2 _ qg + q~, where hqb is the transverse moment um transfer. Since

1 1 dq~ we get from Eq. (6.1)

q d q - -~ dq 2 - -~ ,

do' ~ (]~ji ~ P ji ) __-- 47r@ ( Ji 1

dEf Pji 0

'N]21( f II rw II i)l 2

Jf

Pji

/

and

jr0 ~

qo2(1 - ~2)2

[q~ + (1 -/ 32)qg]2 dq~

(6.14)

do' ~ (]-~ji ~ ~ji-JI- 1) = 27rup2 ( Ji 1

dEf #ji +1

j f ) 2

~[~ji -1[- 1 I(f II rT l i)l 2

foo ~ q~ dq~ (6.15) x [q~ + (1 -/ 39.)qg]2 "

According to Eq. (6.5), the integration over q~ extends from zero to infinity.

The dipole approximation breaks down for large values of q, giving rise to

a t erm in Eq. (6.15) which diverges logarithmically. If, in the divergent

term, we introduce a cut-off qmax as the upper integration limit, we have

do' ~ (~ ji ~ ~ ji ) = 471./.]2 ( Ji 1

dEf Pji 0

and

2

Jf

\

] ] ( f ]} rT I] i) ] 2 ( 1 --/ 3 2 ) ( 6. 16)

Pji

/

do' ~ (J[~ji ~ ~[~ji -It- 1)

dEf

_ 11 Jf ) 2

P ji P ji 1

x 2 In 7 + in qmax 1

q~ + -

I(f ]l rw II i)] 2

1] . (6.17)

6 . 2 . T WO S I MP L I F I E D T R E A T ME N T S 147

In Eq. (6.17) we may neglect the t erm 7 . 2 in the second logarithm. If,

somewhat arbitrarily, we choose 2 _ 2 mE i / h 2 qmax as the nonrelativistic

inverse K-shell radius and, moreover, neglect Ef in the definition (5.105)

for q0 , we may replace 2 qmax/%2 wi t h 2 my 2 / E i .

Equat i ons (6.16) and (6.17) illustrate a very i mpor t ant feature, namel y

t hat for #j i --+ #j i transitions, the cross section approaches a const ant value 2

as the projectile energy increases, while for #ji - + #ji--t- 1 it grows propor-

tional to In 7 for large values of 7. The increase of the cross section as In 7

in the ext reme relativistic limit can be underst ood by considering the ex-

tension in the transverse direction of the LiSnard-Wiechert pot ent i al (3.34)

shown in Fig. 3.3. As a consequence, larger and larger i mpact paramet ers

contribute, and the cross section keeps increasing wi t h energy [AmA81].

This unbounded increase wi t h energy is characteristic for all per t ur bat i on

t heory approaches, which do not conserve unitarity.

6. 2 . 2 Qu as i r e l at i v i s t i c t r e a t m e n t

The cross section (6.5) is exact in first-order per t ur bat i on t heory as far as

relativistic effects are concerned. Here, we present a t r eat ment which em-

ploys semirelativistic Darwi n bound wave functions (See. 4.3.3) and semirel-

ativistic plane waves to evaluate the mat ri x element (6.6) using a convenient

coordi nat e system. The advant age of this t r eat ment is t hat the final cross

section is a sum of a nonrelativistic expression plus a relativistic modifica-

tion. For the nonrelativistic part , many t abul at i ons exist, given below, and

for the relativistic part , approxi mat e expressions are available.

For calculating the t ot al cross section, we st art from Eq. (6.7), using

an i nt egrat i on over q from q0 to infinity rat her t han ove r qb from zero

to infinity. It is advant ageous to evaluate the mat ri x element in a ~, ,),

coordi nat e system, in which the ~.-axis lies along q and the 2-axis lies in

the plane formed by q and/ 3. Then

/ 3-ct =/ 3cos A ct~ +/ 3si n A ct~. (6.18)

Here, k is the angle between q and/ 3, so t hat cos k = q z / q = q o / q , where

the z-axis points al ong/ 3. Subst i t ut i ng this expression into Eq. (6.7), the

mat ri x element of

( ~ e i q' r T = OL~ e i q z T

2The cancel l at i on bet ween 1 and ~ 2 as 3' --+ cc is a resul t of t he di pol e appr oxi mat i on

( 6. 10) . Any devi at i on from it l eads t o a finite cons t ant val ue in t he ext r eme r el at i vi st i c

l i mi t .

148 CHA P T E R 6. E X CI T A T I ON AND I ONI Z AT I ON

can be evaluated with the help of Eqs. (5.105) and (5.107) as

(f[cts eiqsTli> _ ~qo ei

--~--(f] qSTli >.

(6.19)

This and Eq. (6.18) yield

(f](1 - / 3. c~ )eiqrTli > -- (1 --/ 32 q~/q2)F(q ) _ ~ i I _ q2/q2 G(q) , (6.20)

with

F(q) = <fleiqSTli>,

G(q) = <floz:~eiq2Tli > . (6.21)

These definitions imply t hat the wave functions for the initial and final

states are specified in the x, y, z coordinate system.

Since the wave functions ]i} and If) are characterized by definite values of

the magnetic quant um number #j, the mat ri x elements F(q) and G(q) obey

different selection rules: in G, the operator c~ changes the total angular

moment um component along the 2-axis, but not in F. For any specific

combination of the magnetic quant um numbers Pji and pjf, at most one of

t hem contributes, so t hat the terms containing F and G add up incoherently

in the total cross section (6.7). This equation now becomes

/ ~ dq ( ~ 2 ( 1 - q~/q 2) 2)

do-~ - 87r@ E ~-ff IF(q)12 + (1 -/ 32 q~/q2)2 IG(q)l ,

dEf t ot a l o

(6.22)

with the underst andi ng t hat the sum is taken over all partial continuum

waves, and t hat in Eq. (6.21) the wave function If) refers only to a single

partial wave.

The t erm proportional to ~ IFI 2 is identical to t hat obtained for non-

relativistic projectiles. The t erm proportional to ~ IGI 2 is the relativistic

modification, as suggested by the factor/ 32. As/ 3 ~ 1, the first t erm be-

comes independent of/ 3 (or 7), since q0 becomes constant, see Eq. (5.105).

To obtain t he/ 3 dependence of the second term, one notes t hat the main

/3 dependence arises near q _~ q0. Hence, one can perform the integration

over q assuming G(q) ~_ G(qo) . One then finds t hat the second t erm is

proportional to

1 ln(1 /32 ] 2

~-5 - ) - 1- ~ 1n3, - 1, (6.23)

which, as/ 3 ~ 1, has the asymptotic form of Eq. (1.11).

6. 2. T WO SI MP LI FI E D T R E A T ME N T S 149

The mat ri x elements F and G have been evaluated with ls semirela-

tivistic wave functions for the electron [DaM78,Anh79]. Averaging over the

initial and summi ng over the final angular moment um projections of the

electron, and integrating over the final electron energy, one obtains the cross

section for ionizing a single K-shell electron. The expression is simplified if

one makes the substitutions

W -- 2 Ef - Ei ( 6. 24)

Z?re2/ao

and

q2a2

Q- z~ "

One then finds [MeL58, KhC69, Anh79, AnG86] a

(6.25)

(YK --

47rag 2 ZP2 4 / i m

r/K ZT in

dQ

d W mi n Q2

where

and

(IFKI ~ +

( 3ZTa) 2 (1 -- (~min/ Q)

(1 - / 32Qmi n/ Q) 2

~K - ( 9/ ~ Z T ) ~

1 2

( I G K I ~ + - ~ Q IF K I )

, (6.26)

(6.27)

Qmi n = W 2 / 4 ~] K ,

2I K xp

Wmi n = OK - - Z2e2/a O.

(6.28)

Here, IK xp is the experimental K-shell ionization energy, and for a neutral

target atom, the target charge is taken as an effective charge ZT ~ Z bar e-

0.3 according to the Slater screening rules [Sla60].

The factors IFKI ~, laKI ~ in Eq. (6.26) are squares of quasirelativistic

mat ri x elements calculated with hydrogenic Darwin and free-electron wave

functions. They are given by

[FK[ 2 -- ( 3Q+ W) QA3,

[ ( ~ K 12 __ 1 WA 2 (6.29)

4

3Since quasirelativistic wave functions are accurat e only to order C~ZT, it is in-

consistent to introduce normal i zat i on factors for these wave functions of the form

[1 + (aZT/2)2] - 1/ 2 and [1 + (kaZT/2)2] - 1/ 2, respectively [DAM78, anh79] . For this

reason, these normal i zat i on factors are omi t t ed in Eq. (6.26), as is in fact done also in

[Anh85a, AnG86].

150 C H A P T E R 6. E X C I T A T I O N A N D I O N I Z A T I O N

wi t h

An ----

27 e xp{ - 2/ k a r c t a n[2 k / ( Q + 1 - k2)]}

311 - e xp ( - 2 7r / k ) ] [( Q + 1 - k2) 2 + 4k2] n'

k - v/ W- 1. (6.30)

The i nt egrat i on over q in Eq. (6.26) has to be performed wi t h some care

(see t he discussion preceding Eq. (6.22) and in [MeL58, Anh79, AnG86] ) .

The first t er m with ]FKI 2 arises from t he longitudinal cont ri but i on present

also in nonrelativistic collisions. The t erms proport i onal to IGK[ 2 and

1 12

~Q]FK are i nt erpret ed as t ransverse and spin-flip contributions, respec-

tively [Anh79]. The t ransverse and spin-flip t erms are most i mpor t ant for

hi gh-Z t arget s, for which t he quasirelativistic approxi mat i ons act ual l y lose

their justification. Tables for t he longitudinal part of the cross section are

available [KhC69, ChM73, RIB77, BeK78, JOB79], but rat her t han using

t he projectile energy entering in t he tables, for relativistic collisions one

must use t he ion velocity entering t hrough t he par amet er ~]K given in Eq.

(6.27). 4 Est i mat es for t he t ransverse and spin-flip cont ri but i ons are given

in [Anh79, AnB87].

In Fig. 6.2, t heoret i cal ur ani um K-vacancy product i on cross sections are

shown for bare projectiles of charge Zp. According to Eq. (6.26), t he t ot al

cross section is proport i onal to Z~ . Hence, t he cross section shown has

been divided by Z~ . For 7 >- 10, t he t ransverse and spin-flip cont ri but i ons

account for more t han one-half of t he t ot al cross section. Figure 6.9 shows

t hat these cont ri but i ons bring t he t heory into close agreement wi t h t he ex-

peri ment al dat a. Al t hough, as not ed previously, t he wave functions used to

eval uat e t he cross sections are valid only for c~ZT << 1, t he agreement ap-

pears to st ret ch to C~ZT ~-- 0.7. This may be because t he collision dynami cs

explores a part of t he bound and cont i nuum wave functions which is not

sensitively dependent on c~ZT.

For general orientation, Fig. 6.3 shows reduced 1s-vacancy product i on

cross sections by bare projectiles on one-electron t arget s as a function of

In 7 for various values of ZT, cal cul at ed from Eq. (6.26). 5

4 In the theoretical expression (6.22), the projectile velocity enters through the pa-

rameter q0 in the nonrelativistic term and through q0 and/ 3 in the relativistic term.

5One-electron (ls) loss cross sections for projectiles by bare targets are given by

[AnB87]. By interchanging ZT and Zp, the 1s-ionization cross section of a one-electron

target by a bare projectile can be obtained. Figure 6.2 is taken from [Anh79] and refers

to two electrons in the K-shell, whereas Fig. 6.3 is taken from [AnB87] and refers to a

single ls electron.

6.2. T WO SI MP LI FI E D T R E A T ME N T S 151

2 5

2O

o 1 5

. J~

I 0

_ I I / / / ~ _

I 0 I 0 0 I 0 0 0

Fi gur e 6. 2. Calculated cross sections for U K-vacancy production by bare pro-

jectiles of charge Zp, divided by Z~, as a function of the Lorentz factor 0', using

Eq. (6.26). The longitudinal, transverse, and spin-flip contributions are shown,

as well as the total cross section. From [Anh79].

1 4

12

A

. . Q

,q- 10

+o

~ 8

N

out.. 6

I %

2 [

0

I I J

10 102 103

-1

3

7

10

2 0

5 O

9 2

Fi gur e 6.3. Calculated cross sections for one-electron target (ls) ionization

by bare projectiles of charge Zp, multiplied by (Z T / Z p ) 2, as a function of the

projectile Lorentz factor y. The calculated curves are based on Eq. (6.26). The

units are b/ electron. Adapt ed from [AnB87].

152 C H A P T E R 6. E X C I T A T I O N A N D I O N I Z A T I O N

6.3 Mul ti pol e expansi on

For high-Z target atoms, it is mandat ory to use exact relativistic wave func-

tions for initial and final states. The factor r s- 1 in the Dirac l s bound and

continuum wave functions [see Eqs. (4.98), (4.103), and (4.114)] contracts

the radial electronic density distribution. The resulting decrease of the ra-

dial mat ri x elements leads to smaller cross sections [AnG86] compared to

the nonrelativistic t reat ment of electrons.

For calculating exact matrix elements, one may either use a multi-

pole decomposition of the interaction in coordinate space [MeS89], or one

may decouple the angular part of the electronic and projectile coordinates

[VaB84, BEG85, Bec86] by introducing a multipole expansion for the expo-

nential exp(i q. rT) occurring in Eq. (6.4). This amounts to an expansion

in moment um space. In the following, we discuss bot h types of expansion

in turn.

6. 3 . 1 Mul t i pol e expans i on in coordi nat e s pace

Ret urni ng to the starting equation (6.1), one may t ry to decouple the time

integration associated with the nuclear motion from the integration over

the electronic coordinates by performing a multipole decomposition of the

interaction term 1 / @. Such an expansion is applicable not only in the per-

t urbat i ve formula (6.1) but also in a more rigorous coupled-channel treat-

ment, see Sec. 6.5.

For a nonrelativistic interaction 1/ Ir 1 - r21 , the st andard multipole ex-

pansion leads to a complete decoupling of the angular dependence of the

vectors r 1 and r2, i.e., a decoupling of the unit vectors r l and f2. In a rela-

tivistic collision, the Lorentz contraction acts in the z-direction, so t hat the

Cartesian coordinates are not treated on equal footing in Eq. (5.97). Some

coupling in the polar angles will therefore remain. A partial decoupling

[RuS93] can be achieved by rewriting the interaction term in the form

1 1 1

r{~ V/(XT _ b)2 + y2 + ~/ 2(z T _ v t ) 2 I P e - - P P I ' ( 6. 31)

where we have formally introduced the vectors

Pe - (Xw, YT,~ZT) with pe -- IX 2w + y2 w + 72Z2 w (6.32)

and

pp = (b, 0, ~/ vt ) with p p = v/ b 2 + ~ 2 v2 t 2 . (6.33)

6.3. MUL T I P O L E E X P A N S I O N 153

zT !

I " T

( x . ~ + y m 2 ) 1 , 2

Fi gur e 6.4. Definition of the vector Pe"

These vectors are not physical position vectors for the electron and pro-

jectile, respectively, nevertheless t hey are formally useful in part l y separat-

ing the electron mot i on from the projectile motion. In Fig. 6.4, we show the

relation of the modified vector Pe to the position vector r T = (XT, YT, ZT)"

Int roduci ng the polar angles 0 of r T and 0' of Pc, we can rewrite

p~ - - rTu(O ) (6.34)

with

u(O) - V/1 + (7 2 - 1) cos 2 O, (6.35)

so t hat

COSO t - - 7 Z T _ _ 7 C O S 0

- - p~ - - u(O) " (6.36)

We now decompose the inverse distance 1 / I p ~ - p p [ of Eq. ( 6. 31) i nt o spher-

ical harmoni cs YLM, dependi ng on the directions ~ and ~ p, in the form

1

Ip~ - pp[

EL=O 2L 4- 1 pL+l E M- - - L Y; M( R P ) YLM( R e ) ,

if Pe ~- PP,

oo 471" p L L

}-~L=O 2L + 1 [rTU(O)]L+ 1 ~-~-M=-L Y; M( P P ) YLM( P ~ ) ,

if p~ _> pp,

(6.37)

154 C H A P T E R 6. E X C I T A T I O N A N D I O N I Z A T I O N

~ p

=pp l u ( O )

Fi gur e 6.5. Illustration of the inner and outer regions in which the angular and

the radial integration decouple and of the intermediate regions in which they do

not decouple.

or, equivalently,

IPe - - P P [

47t

= E 2 L + 1 v L ( r T ' O ' p P ) Y[ M( ~ P ) YL M( ~ )

L M

(6.38)

wi t h

[rTU(O)] L

pL+l ,

vL(r~ , e, pp) - p~

if r T ~ p p / U( O ) ,

if r T ~ pp/ u(O) .

(6.39)

The distinction, which one of Pe or pp is larger, is seen to depend on t he

angle 0, unlike for t he nonrel at i vi st i c case. Since, according to Eq. (6.35),

u(O) varies bet ween 1 and 7, t he separat i on bet ween t he radial and t he

angul ar i nt egrat i on is compl et e only in t he range r w < p p / ~ / a nd r w >__ pp.

The dividing surface bet ween t he inner and t he outer region is given by t he

ellipsoid pe = rwu(O ) = pp. The inner, outer, and i nt ermedi at e regions are

i l l ust rat ed in Fig. 6.5.

The first-order t ransi t i on ampl i t ude (6.1) is now expressed as

A~ (b) - i ~/ZPe2 f

.... h d t e i ( E e - E i ) t / h Mf i ( b , t ) .

(6.40)

6.3. MUL T I P O L E E X P A N S I O N 155

Here, the mat ri x element is

47r

M~ ( b , t ) - E 2 L + 1 (fl(1 - / 3a z )

L=0

VL( r T O, pp)YL,ttjf_pj i ( ~ ) l i } Y* '~ r~J --"Ji ( P P )

(6.41)

The quantities #ji and #if are the angular moment um projections of the

initial and final states, respectively. Formula (6.40) is a possible point of

depart ure for the accurate evaluation of excitation, ionization, and pair pro-

duction cross sections using Eqs. (5.64) and (5.65) within first-order pertur-

bation theory. The mat ri x elements (6.41) can, of course, also be employed

for detailed single-center coupled-channel calculations [RuS93].

In evaluating the mat ri x elements (6.41), one has to observe t hat , owing

to (6.39), not only the radial integration over r T but also the integration

over the polar angle 0 within the interval flp/ ~/ < r T <__ pp depends on

the particular value of b and t t hrough the relation tip -- v/ b 2 + "),2v2t2 of

Eq. (6.33). Only the integral over the azimuthal angle can be carried out

easily.

6. 3 . 2 Mu l t i po l e e xpans i on in m o m e n t u m s pace

Let us now t urn to the calculation of the transition ampl i t ude in momen-

t um space which allows for a more complete separation of the integrations

associated with the electron and the projectile coordinates.

If we st art from the right-hand side of Eq. (5.104) [which leads to

Eq. (6.4)] rat her t han from the left-hand side, a decoupling is achieved

by decomposing the exponential exp( i q- r T) . In terms of a spherical repre-

sentation ~ t - q / q - - (Oq, r wher e q2 _ q~ + q02, and rT -- r T/ r T -- (0, 0),

we can write the expansion of the plane wave as

oe L

C i q' rT -- 47I E E iL JL(q r T ) Y~ M((t ) YLM(~ T ) "

L=0 M=--L

(6.42)

While in each t erm of the sum, the angular dependencies factorize, the ra-

dial parts remain coupled by the spherical Bessel function j L( q r T ) . Using

q b " b = qx b = qbsin Oq cos Cq and the integral representation of the Bessel

function JM, one may carry out the Oq integration within the angular in-

tegral on the ri ght -hand side of the mat ri x element (5.104) to get

~027r / 1 e-iqb sin O q c os (r

dC~q i d cos Oq q2(1 - 32 cos 20q)

Y;M(Oq, Oq) =

156 CHAPTER 6. EXCI TATI ON AND I ONI ZATI ON

i t _ ( ~ ) 1 ( q z )

2 ~ i - " l d q~ _ 9~qz 2 y~ , ~ rccos-,q o J M( b v / q ~ - q~ ) .

(6.43)

By substituting Eq. (6.42) into (6.4) and employing Eq. (6.43), one obtains

the transition amplitude

A~(b) 8iTrup E i L- M ~0 cc q dq

q~ _ 9~ q~,

L , M

x (f[(1 - 3Ctz)jL(qrT)YLM(~T)[i ) BLM(b , qo, q),

(6.44)

where

( rccos ~ ,04 ,

is the straight-line path factor [AmA81, VaB84, BEG85] t hat already oc-

curs in the SCA for nonrelativistic collisions [BAH59]. The Heaviside step

function O( q- q0) assures t hat q is greater than its projection q0.

The total cross section is derived either from Eq. (6.44) using the or-

thogonality relation for the Bessel functions or, more directly, from Eq.

(6.5). The final result for the ionization cross section per electron can be

written as

dcr~ _ 87c@ q dq

dEf o (q2 _/ 32q~)2 IMfi(q)l 2, (6.46)

with

(2O

Mfi(q) - 47r E iL <f](1 -/ 3az) j L( qrT) YL, , j f_, j i (~T)li}

L = 0

ar ccos- - , 0 , (6.47)

X rL,#jf_#j i q

where ]_Lji and #jf are the angular momentum projections of the initial

and final state, respectively. This formula is the starting point for the

accurate evaluation of excitation and ionization cross sections within first-

order perturbation theory [VaB84, BEG85].

Before moving on to discuss some details and generalizations, a few

comments may be appropriate.

(a) First, we observe t hat in contrast to Eq. (6.15) in the dipole ap-

proximation, there is no convergence problem at the upper limit of the

q-integration in Eq. (6.46). The cross section is well-defined without am-

biguous cut-off procedures.

6. 3. MUL T I P O L E E X P A N S I O N 157

(b) The behavior of the cross section (6.46) in the extreme relativistic

limit ~/>> 1 can be investigated by expanding the integrand around q- q0.

In this way, it is shown by Becker [Bec86] t hat for #if - #j i (and also

for IPjf - #j i I >- 2), the integral tends to a constant value as ~/ increases.

However, for #j r - #j i + 1, the integral increases with ~/ as ln~ . This

is in accordance with the results (6.17) or (6.23), derived more simply in

the dipole approximation or the quasirelativistic treatment, and with the

asymptotic behavior given in Eqs. (1.10) and (1.11).

(c) The choice of the gauge for the electromagnetic potentials is dis-

cussed in some detail in [AmA81]. Here, we have used the Lorentz con-

dition (3.3) which is Lorentz invariant and hence allows us to derive the

electromagnetic potentials A and (I) by a Lorentz transformation from the

static Coulomb potential. This renders it possible to use Eq. (6.4) to de-

rive rather simple pat h factors (6.45) which actually are the same as for

nonrelativistic projectiles. On the other hand, the Coulomb gauge defined

by V- A C~ - 0 leads to more complicated pat h factors. A comparison of

numerical results for cross sections calculated in the Lorentz and Coulomb

gauges in the plane-wave Born approximation as well as in the SCA shows

agreement with one another on the 1% level.

(d) The electronic mat ri x elements in Eq. (6.47) have to be calculated

with relativistic wave functions. The use of nonrelativistic approximations

either leads to a violation of gauge invariance (since the electromagnetic

coupling j u A P / c , Eq. (5.31) is not compatible with the SchrSdinger equa-

tion) or leads to second-order terms in the potentials and hence to a spurious

projectile charge dependence as Z~. Only by a careful analysis lAmA81] is

it possible to remove these spurious terms.

In evaluating the mat ri x elements (6.47) rigorously, it is no longer useful

to reduce the terms containing the Dirac mat ri x C~z by employing Eq. (5.90).

Rather, with the definitions (4.86) and (4.87), one may decompose the

mat ri x elements as [VaB84, BEG85]

( f l( 1 - ~C ~z) j ~( q r T ) Y Zm( ~T ) l i } - -

A1 f (1) (q) + i ~ A 2 f ( 2 ) (q ) - i ~ A 3 f ( 3 ) (q ) (6.48)

where the angular integrals are defined by

~Jf Pj"

dl - {/ ~gf I YLMI ~ ai ' )

_ _ (~ PJf ~Ji

me , ) ~_~fl YLMOz I X~i }

A3 ~ ]~jf ~Ji

- I rL. zl _ l

(6.49)

and are evaluated with st andard techniques [Ros57, Edm57]. The radial

158 C H A P T E R 6. E X C I T A T I O N A N D I O N I Z A T I O N

integrals are defined as

f0 C~

FL (1) (q) -- j L( q r ) ( g i g f -Jr- f i f f ) r 2 dr

/0

Ft 2) (q) - - JL ( q r ) g i f f r 2 d r

Fi 3)(q) -- j L (q r ) f igf r 2 dr. (6.50)

The integrals can be rewri t t en in t erms of hypergeometric functions [JaZ57,

Amu78, JaA80, TrB83, VaB84] if exact Coulomb bound-st at e wave func-

tions (4.98) and continuum partial waves (4.114) are inserted. The actual

evaluation is performed numerically [BEG85].

The calculation of the t ot al and energy-differential ionization cross sec-

tion according to Eq. (6.7) requires an averaging over all #ji and a sum-

mat i on over all values of ~f and #if. In practice, the summat i on has to be

t runcat ed at a maxi mum value ~max because of computer limitations. For

the angle-differential cross section, one has to use the wave functions (4.123)

or (4.124) with well-defined asympt ot i c moment a instead of the spherical

partial waves.

6. 3. 3 P e r t ur ba t i ve r e s ul t s a nd c ompa r i s on wi t h e xpe r -

i me nt

In an experiment, we usually have a solid or gas target, t hat is, we have

to take into account many-electron effects. Deferring a more detailed

discussion to Chap. 7, we assume for the present purpose t hat the K-

shell wave function of a complex t arget at om can be approxi mat ed by an

effective hydrogenic wave function. The generally accepted met hod uses

an effective Slater-screened t arget charge Z~ = Z bar e- 0.3 in conjunction

with experimental K-shell binding energies. These binding energies are not

consistent with the wave functions used. One may expect, however, t hat

mai nl y the inner part of the wave function is responsible for the mat ri x

elements. This part is well described by the effective nuclear charge Z~ .

The outer part is related to the binding energy but does not enter sensitively

into the results. Nevertheless, it is not clear whether the inconsistency may

lead to sizable effects in some cases. In most results to be discussed in this

chapter, it is justified to neglect the effective-charge correction.

The first step in calculating differential (in electron energy and angle)

and t ot al cross sections consists in studying the i mpact -paramet er depen-

dence of the ionization probability P (b ) . The weighted K-shell ionization

probability 27rb P (b ) usually has a maxi mum in the vicinity of the K-shell

6.3. MULTI POLE EXPANSI ON 159

radius a K. This is similar as for the excitation probabilities in Fig. 6.13 for

coupled-channel calculations.

Figures 6.6 and 6.7 show calculated double-differential ionization cross

sections d2cr/dEe df~e for fixed electron energies and fixed electron emission

angles, respectively, for the system Zp + U [MoG89]. In Fig. 6.6, we

can see t hat electron emission occurs mainly at smaller electron energies.

As Fig. 6.7 demonstrates, the angular anisotropy of the emitted electrons

becomes peaked in the forward direction as their kinetic energy increases.

Since, owing to limitations in computing power, I~1 _< 10 has been used,

the convergence of the partial-wave contributions in Eq. (4.123) becomes

unsatisfactory at backward angles, so t hat this angular range is not included

in Fig. 6.7.

In Figure 6.8, the total K-shell ionization cross section for p + U

collisions is plotted as a function of the Lorentz factor -y. As already seen

explicitly in the dipole approximation of Eq. (6.17), the cross section in-

creases monotonically with "y. The quasirelativistic calculation of Anholt

[Anh79] and the multipole expansion calculation of Becker et al. [BeG85]

show the same qualitative behavior as the collision energy increases. For

-y > 2, the absolute difference between the calculations increases slowly

with ~,. For large values of -y, Becker et al. have the problem t hat a grow-

ing number of terms contribute in the multipole expansion (calculations

have been carried up to 1 = 6); however, the authors estimate the resulting

inaccuracy to be less t han 5 %. On the other hand, differences in the wave

function will affect the matrix elements changing the magnetic quant um

number by one unit and hence will increase as ln' y for large energies. It

is therefore quite natural t hat discrepancies slowly build up as the energy

increases.

Numerical calculations have established an approximate scaling rule

lAmAS1, Bee86] for the ionization probability at high relativistic energies

(3' > 5) and fixed impact parameters. For b = 0, the ionization probability

P ( b - O) - 1.8 x lO-4Z~ (6.51)

is independent (within +10%) of the target charge up to the heaviest el-

ements. This empirical finding shows t hat for Zp _> 75 the ionization

probability exceeds unity, and hence pert urbat i on theory breaks down.

There are few experimental dat a to compare with theoretical results.

At present, only measurements for total cross sections of K-shell ioniza-

tion in various target atoms are available. In Fig. 6.9, the total K-shell

ionization cross section is shown for 4.88 GeV protons impinging on target

atoms with charge numbers ZT. For the calculations of [DAM78, Anh79],

quasirelativistic wave functions are used. For high-Z targets, the deviation

160 C H A P T E R 6. E X C I T A T I O N A N D I O N I Z A T I O N

lO

10 ! 1

u ' l

9 T 1

>~

~E

, Z3

~ 10-3

10-6

" I

I

I I i I I 1

2 S l l 2

e ( d e g )

9O

9 0 "

I

2 4. 6 8

E, (MeV)

1 0 - 3

1 0 - 6

. 1 1 0 - 6

Fi gur e 6. 6. Double-differential cross section for t he ionization of t he lSl/ 2 and

2Sl/ 2 t ar get st at es in Zp + U collisions at Elab -- 1 GeV/ u as a funct i on of t he

energy of t he ionized electron. The par amet er at each curve gives t he l abor at or y

emission angle 0. From [MoG89].

1 0 - z

~" 1 0 - 3

f -

U e l

"T

>

( l J

1 0 - ~ '

~." I0 - s

- o 0 - 6

uYl

~ 1 0 - 7

3me 2

0 2 0 / , 0 6 0 80

e { d e g)

Fi gur e 6. 7. Double-differential cross section for t ar get K-shell ionization in Zp

+ U collisions at Elab -- 1 GeV/ u as a function of t he el ect ron l abor at or y emission

angle 0 for various el ect ron energies Ee. From [MoG89].

6.3. MUL T I P O L E E X P A N S I O N 161

1/., . r~

12 " . ' "

1 0 , o . . ~ ~ S S ' ' ' ' ~

1 1 0

Y

Fi gure 6. 8. K-shell ionization cross section for p + U collisions as a function

of the Lorentz factor ~. Dashed curve: [Anh79]; solid curve: [BeG85]. In both

curves, experimental binding energies have been used.

from exact relativistic wave functions (used in [BEG85]) should be most

significant. In fact, the results for the two types of calculations differ by

only 10%. Unfortunately, the calculations using more accurat e electronic

wave functions lie farther away from the experi ment al data. Considering

the experi ment al errors, it certainly would not be justified to prefer any

of the calculations j ust on the basis of a comparison wi t h the data. For

furt her discussion see [Eic90].

In view of the scarcity of ionization dat a for relativistic at omi c projec-

tiles, it is t empt i ng to t ur n to ot her projectiles. Ionization measurement s

have been made wi t h electrons up to 23 GeV, corresponding to 3' = 4.4 x 104

[SpM94]. At these ext remel y high values of 3', the relevant i mpact parame-

ters are so large t hat t hey lie in the medi um surroundi ng the t arget at om. 6

This leads to a shielding of the projectile' s electromagnetic field [Fer40],

which causes the cross section to sat urat e [Sor87, ChE89]. Such " many-

body" effects lie outside the scope of this book. Below the onset of the

sat urat i on, though, typically for "y _< 103, one can expect t hat high-energy

electrons have ionization cross sections similar to those of protons. This is

true, as long as the t raj ect ory of the electron projectile can be approxi mat ed

by a st rai ght line for the relevant collisions. Calculations by Davidovid et

al. [DAM78], based on a relativistic first-order wave theory, show t hat even

6At these high values of -y, the de Broglie wavelength h/ (~meC ) is small enough

( ~ 10 -2 fm) t hat the concept of impact paramet er is applicable even to electrons.

162 C H A P T E R 6. E X C I T A T I O N A N D I O N I Z A T I O N

I I

2 0 \

r-,

, 4

10

0 i

7 0 80 90

ZT

Fi gur e 6. 9. K-shell ionization cross section for 4.88 GeV protons versus t he

t arget charge. Experi ment al dat a and dashed curve: Anholt et al. [Anh79], see

also Davidovid et al. [DAM78]; solid curve: Becket et al . [BEG85].

A

JQ

z 1 6 0

O

I -

U

LO

o o

1 2 0

0

k)

z

o 8 O

I - -

<

N

z

_o 4 o

_ J

. _ i

w

--r

!

,- 0

1 . 0 2 1 . 2 3 2 1 2 0 0 2 0 0 0

. . . . I . . . . . . . . I . . . . . . . . 1 . . . . . . . . I . . . . . . . . 1 . . . . . . . . I ' ' '

4 7 Ag

- - f . - -

~ , ~ ? . - i ~ ' y .."

- f . . ~ . , ~ , . . . ' " o -

~ ,;, . . "

_ - . . . . . . . . . . . . ; , . . : : : . . . . . . . . . . . . . . . . . . . . -

.

. . . . I , I l , ' " ' l . . . . - - , " " , , , I . . . . . . . . I . . . . . . . . I . . . . . . . . I , , ,

1 0 - 2 1 0 -1 1 0 ~ 1 0 1 1 0 2 1 0 3

I NCI DE NT KI NET I C ENERG Y ( Me V )

F i g u r e 6 . 1 0 . K-shell ionization cross section for Ag bombarded by electrons.

The dash-dot t ed curve gives t he result of a semiclassical calculation [Ko175], the

dashed curve a fully quant um-mechani cal calculation [Sco78], and the solid line

a calculation based on st ruct ure functions with one-phot on exchange [EsM82],

t oget her with a decomposition into longitudinal (a l) and transverse plus spin-

flip (a t) contributions. Note t hat the abscissa is not linear in ln-7. From [Gen84].

6.4. N O N P E R T UR BA T I VE A P P R O A C H E S 163

for ~ as low as 5 (2-MeV electrons), electron and proton cross sections dif-

fer by less t han 5% up to Z p = 50. Therefore, we show in Fig. 6.10 the

K-shell ionization cross section for electrons on Ag. A comparison is made

with various theories. One of these, by Eschwey and Manakos [EsM82],

allows a decomposition into longitudinal (a z) and transverse plus spin-flip

(a t ) components, shown in dot t ed lines on Fig. 6.10. The former goes to

a constant, and the latter increases as in ~, as ~ increases to high values,

illustrating the expected behavior of these cross section components. Com-

parisons of other relativistic electron dat a with Eq. (6.26) can be found in

[Anh79].

6.4 Nonpert urbat i ve approaches

The empirical scaling rule Eq. (6.51) of the ionization probability at im-

pact paramet er b - 0 shows t hat pert urbat i on theory breaks down at large

projectile charges. Already for Zp _> 75 the probability P (b - 0) >_ 1.0,

so t hat for medi um-Z projectiles, the met hod will st art to become ques-

tionable. Although this conclusion is weakened for the more i mport ant

larger impact parameters, it is nevertheless mandat ory to examine nonper-

t urbat i ve approaches. We first consider the finite difference met hod and in

Secs. 6.5 and 6.6 proceed to coupled-channel methods.

We would like to mention another relativistic nonpert urbat i ve approach,

namely the classical-trajectory Monte-Carlo ( CTMC) met hod [TeTS0,

HoH92]. The nonrelativistic CTMC version, while difficult to justify theo-

retically, is known to be successful in low-velocity collisions. The relativistic

version t reat s the nuclear as well as the electronic motion by solving the

classical relativistic equations of motion for the t hree-body system assuming

a relativistic microcanonical distribution of initial positions and moment a

and ignoring the electron spin.

6. 4 . 1 F i n i t e - d i f f e r e n c e m e t h o d s

Let us now t urn to finite-difference calculations. Here, the time-dependent

Dirac equation in its Hami l t oni an form (5.40) is solved numerically. Writ-

ing, for a small time step At,

9 (rT, t + At) = U(t , t + At ) ~ ( r T, t), (6.52)

the time evolution operator is evaluated as

( ) - 1( i 1 )

i 1At ) 1 - At H( t + At)

U(t , t + At) - 1 + ~-~ At H ( t + -2 ~-h

(6.53)

164 C H A P T E R 6. E X C I T A T I O N A N D I O N I Z A T I O N

Fi gur e 6. 11 Ti me evol ut i on of t he el ect r on densi t y di s t r i but i on in t he z T - - p-

pl ane [ p- (x 2 + y2) 1/ 2] for a U 92+ + V 91+ col l i si on at 10.0 Ge V/ u. The t ar get

r emai ns fixed at Zw = p = 0 whi l e t he pr oj ect i l e moves at b = 0 from ri ght t o

left al ong t he Zw-axis in t he posi t i ve z-di rect i on. It is l ocat ed at Zp - - 3. 1 x 103

fm for t = - 8 . 0 [in uni t s of h/ / ( m~ c 2) -- 1.28 x 10 -2~ s], at Zp = 0 for t = 0, at

Zp = 2.3 x 103 fm for t = 6, and at Zp = 4.2 x 103 fm for t - 11. The densi t i es

are gi ven in uni t s of ao 3 (a0 = 5.3 x 104 fm). Fr om [ ThB92] .

6. 4. N O N P E R T UR BA T I VE A P P R OA C HE S 165

up to second order in At. After discretizing the wave function in space as

well, one may follow its time development starting from a suitable initial

condition. The numerical technique is described in detail by Becker et al.

[BeCS3].

In order to render the numerical procedure manageable, one has to con-

fine oneself to an impact paramet er b = 0, thus enforcing cylindrical sym-

met ry about the beam direction and keeping the problem at two dimensions

in space [BEG83, ThB92]. While this choice of the impact paramet er carries

zero weight for the cross section, it is instructive to study this particular

case.

Since the t raj ect ory passes right t hrough the singular points of the

Coulomb potentials, it is necessary to replace the Coulomb potentials in

Eq. (5.40) by nonsingular potentials produced by a finite-size nucleus. The

simplest choice for the nuclear charge distribution is a homogeneously

charged sphere with radius R0, separately fixed for the target and the pro-

jectile nucleus. In other words, one has to substitute in Eq. (5.40) for bot h

collision partners

1 / r for r _> R0

1 fo(r) (6.54)

P 1( 3- r2/ R~))/ Ro for r < Ro

Figure 6.11 conveys a vivid picture of the time development of the prob-

ability distribution for U 92+ + U 91+ collisions at 10 GeV/ u [ThB92]. In

the calculations leading to these results, a grid of 600 300 meshes with

a spacing of 4.6 -2 a K = 26.5 fm (where a K is the urani um K-shell

radius) is used. The nuclear radius R0 is arbitrarily chosen to be twice

this amount in order to guarantee the stability of the numerical procedure.

The calculations are initialized at the time t = - 8 . 5 [in nat ural relativistic

units to = h/ (rnec 2) = 1.28 x 10-21s], when the projectile has a distance of

about 5.7 a K = 3.3 x 10 -~ ~ cm from the target. At to, the density of the

electron distribution is assumed to be determined by the unperturbed ls

1/2

ground state of U 91+. Figure 6.11 shows the electron density at t - - 8 . 0

and separation 5.4 a K. At the next stage, t = 0, when the nuclear charge

distributions completely overlap, the density distribution still is almost un-

altered. The main change takes place while the projectile recedes and leaves

the target at om behind, at a separation of about 4a K at t = 6 and of about

7.3a K at t = 1-1. The numerically calculated wave functions can be pro-

jected onto exact target eigenstates and thus be analyzed with respect to

their content of various excited and continuum states.

While these calculations are very instructive in permi t t i ng one to visu-

alize the collision process, they are not sufficient to calculate cross sections.

166 C H A P T E R 6. E X C I T A T I O N A N D I O N I Z A T I O N

In Sec. 6.1 it is also shown t hat it is not realistic to assume t hat in the en-

trance channel the target wave function is still unpert urbed at a projectile

separation of 5.7 a K. In reality, owing to the long range of the Coulomb in-

teraction, changes in the target wave function start to accumulate when the

projectile is still a distance of (500-2000)a K away [ToE88b], see Fig. 6.12.

Unfortunately, numerical calculations on a discretized grid within such a

large time span are presently hardly possible.

Following the arguments outlined in Sec. 6.1.2, one may expect a much

bet t er convergence if one uses a phase-distorted wave function (5.62) for the

initialization of the time evolution and, correspondingly, the short-range

interaction occurring in Eq. (6.9).

6. 5 T w o - c e n t e r c o u pl e d- c ha nne l m e t h o d s

The direct solution of the time-dependent Dirac equation (5.39, 5.40) in

the l aborat ory frame on a discrete space-time grid is still not feasible with

sufficient accuracy for a general collision geometry. However, one may solve

the problem within a limited space of basis functions. For this purpose, one

expands the wave function ff~(rT, t) in terms of a suitable set of basis states

r t). The degree of approximation then is essentially determined by

the choice and the t runcat i on of the basis set. A basis set embracing the

most i mport ant reaction channels should not only allow for excitation and

ionization but also for charge transfer.

In this section, we discuss two-center coupled-channel calculations, and

proceed to single-center coupled-channel methods in Sec. 6.6. Both methods

are based on the same idea, namely an expansion of the time-dependent

wave function in terms of atomic orbitals. In the former, more symmetrical

approach, one uses a subset of basis functions centered around the target

and another subset centered around the projectile nucleus. In the latter

method, which is computationally simpler, the basis functions are attached

only to the target nucleus, with the projectile merely acting as a source of

perturbation.

6. 5. 1 Coupl ed- channel e qua t i ons

A t reat ment of the time-dependent Dirac equation (5.39, 5.40) for relativis-

tic ion-atom collisions requires the expansion of the time-dependent wave

function ~ ( r T, t) in terms of target and projectile eigenstates Ck(rT, t) and

~Pk,(r~, t' ), respectively, as

9 (rT, t) -- E ak( t ) ~ k( rT' t) + E ak, (t ) S-l~Pk, (r~ , t' ), (6.55)

k k'

6. 5. T WO - C E N T E R C O UP L E D - C H A N N E L ME T H O D S 167

where we have referred the total wave function to the laboratory system.

The first sum represents an expansion in terms of target states; the second

sum is an expansion in terms of projectile states originally defined in the

projectile system and subsequently transformed to the laboratory system

with the aid of the operator S -1 defined in Eqs. (4.29) and (4.31).

The use of two centers for the expansion of the wave function 9 (r, t) has

three i mport ant implications: (a) In the interaction region, linear combina-

tions of atomic orbitals around the target and around the projectile nucleus

provide flexibility to approximate transient molecular orbitals which may

play a role during the collision. (b) Charge transfer is explicitly included as

a reaction channel. This not only renders charge exchange at all possible,

but also will affect other reactions by entering as an intermediate state. (c)

Specifically, ionization of an electron may occur as a direct ionization of the

target at om as well as capture into the continuum of the projectile. In this

way, one will get a better representation of the true continuum by a finite

set of basis states.

In fact, the only approximation within the met hod consists in the re-

striction to a finite set of atomic basis states. Since each of the atomic

basis sets is truncated, there is no problem with overcompleteness t hat

would arise from two complete (or nearly complete) sets. In practical cal-

culations [ToE88a, ToE88b], basis sets have been confined to 20 or 36 sub-

states, namely those of the lsl/ 2, 2Sl/ 2, 2Pl/ 2 , 2P3/2 , 3Sl/ 2, 3Pl/ 2 , and

3p3/ 2 shells. The use of exact hydrogenic eigenstates calculated from Eqs.

(4.86) and (4.98) guarantees t hat the atomic basis functions attached to

the s a me center are orthogonal in b ot h the target frame and the projectile

frame according to Eqs. (4.41) and (4.44). This would no longer be true for

variational wave functions or basis states obtained by diagonalization in o ne

of the two frames. The difficulties associated with pseudostate calculations

are discussed in Sec. 6.5.3.

In Eq. (6.55) the Lorentz transformation of the time factor [Eq. (4.45)1

included in the projectile states automatically implies the relativistic court-

t erpart s of the nonrelativistic translation factors, see Eq. (1.6).

C o up l e d e q ua t i o ns

In order to determine the time-dependent expansion coefficients ak and ak,

in Eq. (6.55), we have to insert ~ ( rT, t ) into the Dirac equation (5.39).

Observing t hat the atomic eigenfunctions satisfy

H - i h 0 ) _ _ S 2 Z p e 2

a t ~?k , ~ k , ( 6 . 5 6 )

r p

168 C H A P T E R 6. E X C I T A T I O N A N D I O N I Z A T I O N

and using Eq. (5.42), we get

0 ) ZTe2 s- l ~ ) k ,. (6.57)

H - i h - ~ s- l ~ 2 k , - -

With these two relations, the time-dependent two-center Dirac equation can

be rewritten as a set of coupled differential equations for the occupation

amplitudes ak(t ) and ak, (t ) . Combining the amplitudes into a column vec-

tor a = ( a k , ' ' ' ; ak , , ' " "), we may write the coupled equations in a compact

matrix form as

i h N 5. = V a (6.58)

or

i N -1 (6.59)

5_- ~ Va.

The overlap matrix N and the interaction matrix V are each built from four

submatrices corresponding to target-target, target-projectile, projectile-

target, and projectile-projectile transitions.

The overlap matrices are

N~k

N~k,

Ni,k

Ni' k'

' - ~ i k

= i ~ (r w) S- - l ~ k ' (r ~ p ) e i ~ E k ' (~ / h ~ 2 ) z T d3rw e- i (~Ek' - E~) t / h

v . J

= [~ ,(r}' )s-lqDk(rT)e--i~ E~ ' (' / h~ 2)zT d3rw e--i(Ek--~E~')t/h

, J

= 5i, k,. (6.60)

The last equation, valid for exact atomic eigenstates only, is derived in

Eq. (4.44). The single-center overlap matrix elements (target-target and

projectile-projectile) are time independent while the two-center matrix el-

ements (target-projectile and projectile-target) are time dependent owing

to the relative motion of the collision partners.

In analogy to the overlap, we have four types of interaction matrix

elements, namely

V~k

/ r (rT) ( - $2 ZPe2 ) . ~k (rT) d 3r T e - i (E k - E i ) t / h

rp

V~k,

/ qp~ ( rT) ( S - 1 Z T e 2 ) (v/h~2)ZT

-- ~k, ( r~ ) e i~Ek' d3rw

?~T

X e - i ( ~ / E k ' - E i ) t / h

6.5. TWO- CENTER COUPLED-CHANNEL METHODS 169

V~,k

J p ~ , ( r 1 p ) ( - S ZPe2 )

, 99k(rw) e - i ' T E i ' ( v/ h c 2 ) z w d 3 r w

r p

e - - i ( E k - - y E i , ) t / h

Vi' k'

f f ~ , (r p) ( - S -2Zwe2 )

' r (r~ ) e - i ~ / ( E k ' - E i ' ) ( v/ h c 2 ) z w d3rT

?~T

X e - i ~ / ( E k ' - E i ' ) t / h . (6.61)

One verifies t hat in the nonrelativistic limit, all matrix elements merge into

the corresponding nonrelativistic expressions including translation factors.

As long as the expansion (6.55) is not truncated, the set of coupled equa-

tions (6.59) is equivalent to the original two-center Dirac equation (5.39).

The choice and the t runcat i on of the basis set constitutes the only approxi-

mation made in solving the problem. Of course the t runcat i on implies t hat

some of the symmet ry properties of the original Dirac equation get lost in

the coupled-channel equations. Wi t h an incomplete basis set, Eq. (6.59)

is neither Lorentz invariant nor gauge invariant. An explicit numerical

example for the loss of gauge invariance is given in Sec. 6.5.4.

Initial conditions

In order to actually solve the coupled equations (6.59), one has to specify

the initial conditions. If we assume t hat the target electron is initially in

the state k0, we have

ako(t ~ - oc) = 1

ak(t - - - , - oc) = 0

ak, (t - - ~- ec) = 0

for k C ko

(6.62)

Equat i on (6.60) shows t hat an expansion in terms of atomic eigenfunc-

tions has the advantage t hat at asymptotic separations, i.e., for t ~ +oc,

all basis states are mutually orthogonal. This means t hat the expansion

coefficients ak, ak, can be interpreted unambiguously as occupation ampli-

tudes in this limit. In particular, there is no overlap with the negative-

energy continuum which might lead to spurious results if there were a finite

overlap. In the interaction region, where target and projectile states are

nonorthogonal, and where each bound state of one center has some overlap

with bound states and positive- as well as negative-energy continuum states

of the other center, an interpretation in terms of probability amplitudes is

170 C H A P T E R 6. E X C I T A T I O N A N D I O N I Z A T I O N

not needed. It suffices t hat the overlap mat ri x (6.60) is rigorously taken

into account.

C r o ss se c t i o ns

The final probabilities are

P k (b ) = ]ak(t --* ~ ) ] 2 for excitation (6.63)

and

P k , ( b ) = l a k , ( t ~ oc)l 2 for charge transfer. (6.64)

The integrated cross section for exciting state k from the initial state k0

according to Eq. (5.65) is given by

~k - 2~ P ~( V) b db ( 6.65)

while for transfer, k is replaced with k ~.

N ume r i c a l p r o c e d ur e

The main numerical task consists in evaluating the mat ri x elements defined

in Eqs. (6.60) and (6.61). In contrast to the analogous nonrelativistic

problem, there is no symmet ry about the internuclear axis because of the

Lorentz contraction in the beam direction. A fully numerical integration is

required.

The details of the numerical procedure are discussed in [ToE88b]. The

task is facilitated by a symmet ry relation

N ~ z ( t ) - ( - 1) z~ +l~ +"~ +"~ +l N ~ z ( - t ) (6.66)

for the overlap mat ri x and a corresponding relation for the potential ma-

trix elements, where a,/ ~ denote any of the labels i, k, i ~, k ~ and l~ ' z are

the orbital angular moment a for the large components of the states c~,/3,

c~,/3 1

while #j are the projections of the total angular moment a j - 1A 4- -~.

Est i mat i ng the radius of the interaction region by 4 = 4a o / ( Z w + Zp), the

limits of time integration are extended to Ivtl < 250 4, and, when needed

for convergence, to 500 4 or 1000 4. Beyond Ivtl = 20 4, the two-center

mat ri x elements are negligibly small, so t hat only single-center mat ri x ele-

ments have to be calculated. Since the calculation of the mat ri x elements

takes the overwhelming part of the computing time, it is convenient to solve

the coupled time-dependent equations simultaneously with all the different

initial conditions t hat might be of interest.

6. 5. T WO - C E N T E R C O UP L E D - C H A N N E L ME T H O D S 171

6 . 5 . 2 Ap p l i c a t i o n t o U 9 2+ -+- U 9 1+ c o l l i s i o n s

For the collision of U 92+ ions with U 91+ t arget s at 0.5 and 1.0 GeV/ u,

experimental dat a are not available but, on the other hand, the collision is

well suited to serve as a model case for which effects of relativistic electron

and relativistic projectile motion can be studied. According to Fig. 5.1,

the mat chi ng energy for lSl/ 2 electrons is 240 MeV/ u, so t hat at least for

500 MeV/ u we still have rat her "slow" relativistic collisions for which the

two-center coupled-channel met hod is most appropriate.

Figure 6.12 shows some results for the time evolution of the occupation

probabilities for t arget states (excitation) and projectile states (charge

transfer) for the impact energy of 500 MeV/ u using 36 atomic basis states

up to the 3p3/ 2 shell of t arget and projectile and assuming an impact pa-

ramet er of 0.01 a.u. This distance is approxi mat el y equal to the K-shell

radius a N and roughly corresponds to the region of maxi mum contribution

to the excitation cross section, see Fig. 6.13. The interval I vtl <<_40 aK,

displayed in Fig. 6.12, represents a narrow window out of the total time

range I vtl < ( 50 0 - 2000)aN actually t reat ed in the calculation.

In conjunction with the time developments shown, it should be noted

t hat the i nt erpret at i on of the expansion coefficients as probability ampli-

tudes near t = 0 is only approxi mat el y valid because of the nonorthogo-

nality in the reaction zone of basis states belonging to different centers.

Therefore a transient " probability" exceeding unity occurs in Fig. 6.12(a).

All figures reveal certain qualitative features: (a) The i mport ance of

multistep processes is shown by the rapid fluctuations of some of the occu-

pation probabilities. (b) Electron transfer, bot h during the collision as well

as in the exit channel, is comparable to excitation and cannot be neglected.

(c) Wi t h increasing shell size, the interaction region extends much farther

out, see Fig. 6.12(b). (d) In this t reat ment , certain target states are af-

fected by the projectile long before and long after it reaches its distance of

closest approach at t = 0. In Sec. 6.5.4, we come back to these long-range

couplings, also observed for first-order pert urbat i on theory in Fig. 6.1.

Figure 6.13 shows weighted excitation and transfer probabilities P ( b ) b

for the leading transitions as a function of the impact paramet er b. As

already apparent from Fig. 6.12, the probabilities for excitation reach out

to much larger impact paramet ers t han for transfer. A particular role is

1 1

played by the spin-flip transition ls 1/2 (~ ) --~1Sl/2 ( - ~)" Being degenerate

within the coupled-channel t reat ment , the two states are strongly coupled

even at large impact paramet ers so t hat the total cross section diverges

logarithmically. This divergence is spurious and can be remedied by in-

cluding higher-order QED corrections or hyperfine interactions removing

the degeneracy.

172 C H A P T E R 6. E X C I T A T I O N A N D I O N I Z A T I O N

1 . 0

10-1

10- 2

10- 3

10 -~,

1 0 - s

-,- 1 . 0

o 1 0 - I

..l=l

0

1 ..

10- 2

t -

O

:E. 1 0 - 3

0

~, 10- l,

o

1 0 - s

1 . 0

+ ( a )

l S l / 2 ,

U 9 2 * + U 9 1 +

2Si+/2 l s ~2

e x c i t a t i o n ,i/

3 +

U 92 ++U 91+ ( b )

2 S l / 2

2P31

4 -

]PI12

4-

/3 s~12

c h a r g e t r a n s f e r ( c )

1 0 - ' I. ls1"~2

[ 2S~/2 / S l/ 2

10 -2

10- 3

10- ~

1 0 - s

-0.1+ -0.2 0.0 0.2 0.~

vt(a.u.}

Fi gur e 6. 12. Time evolution of the occupation probabilities of target states in

excitation and of projectile states in charge transfer in U 92+ + U 91+ collisions at

500 MeV/ u laboratory energy and impact paramet er b - 0.01 a.u. ~ a K. The

abscissa plotted is the projection of the projectile-target separation on the beam

direction. The projection of the angular moment um is indicated by + , - , and + +

for pj ---- 2'1 2' 1 and ~,3 respectively. (a) Excitation from an initial lSl/ 2 state;

(b) excitation from an initial 2Sl/2 state; (c) charge transfer from an initial lSl/ 2

state. Adapted from [ToE88b].

6 . 5. T WO - C E N T E R C O UP LE D - C H A N N E L ME T H O D S 173

d

o

1.0

0.5

I 1 I I i I I

- 2 S l l 2 U 9 2 " + U 9 1 -

/ ' X 500MeV/ omu

/ \ exci tati on

lS l/2(-1/2)

.c_

C L

1.0

0.5

0.0

I I I I I I

/ " ~ t o t al

/ \ charge transf er

_ / / ~ S l j z ( 1 1 2 )

I

0.05

i mpact parameter b (a.u.)

Fi gur e 6.13. Weighted excitation and capture probabilities P(b) b for U 92+ ql_

U 91+ (lSl/ 2) collisions at 500/ u MeV laboratory energy as a function of the impact

parameter b. Only the leading probabilities into specific final states nlj(#3) are

plotted. From [ToE88b].

For detailed st at e-t o-st at e cross sections wi t hi n a 36-state fully rela-

tivistic coupled-channel t r eat ment of 500 MeV/ u U 92+ + U 91+ collisions

see [ToE88b, Eic90].

6. 5. 3 P s e udos t a t e s f or t he c ont i nuum

The coupled-channel t r eat ment discussed in the preceding subsections uses

a basis of bound states in the t arget and the projectile at om and hence is

confined to excitation and charge transfer into bound states. However, in

order to describe ionization processes on the same footing as excitation and

capture, it is mandat or y to include t arget and projectile cont i nuum states

in the basis set. It is known from nonrelativistic collisions [FrL91] t hat by

174 C H A P T E R 6. E X C I T A T I O N A N D I O N I Z A T I O N

augment i ng the atomic orbital expansion by appropriate " pseudostates"

at t ached to target and projectile, one may simulate intermediate and fi-

hal continuum states and at the same time improve the representation of

molecular states at small internuclear separations. P seudost at es are arbi-

t rary basis functions chosen so as to improve the flexibility of the basis set

in approxi mat i ng the true two-center wave function t hroughout the colli-

sion. Not being eigenstates of the unpert urbed atomic Hamiltonian, pseu-

dostates, if properly chosen, have an overlap (a) with bound highly excited

atomic orbitals not explicitly included in the basis set, (b) with atomic or

molecular continuum states, and (c) with uni t ed-at om wave functions (for

R = 0 and Z = ZT + Zp) needed to improve the representation of molecular

orbitals. This met hod is a powerful tool for simultaneously describing exci-

tation, ionization, and charge transfer in nonrelativistic collisions [FrL91].

For relativistic collisions, see [ToE89].

The use of approxi mat e wave functions such as pseudostates to describe

relativistic collisions entails two problems t hat are absent in the nonrela-

tivistic situation. One of t hem is the occurrence of negative-energy states

arising from the Dirac theory for the relativistic electron motion. The other

one is the loss of the orthogonality propert y of approxi mat e eigenfunctions

under Lorentz transformations.

Negative-energy components

We first address the problem of negative-energy components in pseu-

dostates. The Dirae equation, while formally appearing as a single-particle

equation, is fully interpretable only if considered as an equation for the

field operator, see Sees. 4.2.2 and 5.1. As a consequence, many-el ect ron-

many-posi t ron systems are equally represented by the same equation. For

most applications in atomic and molecular physics, however, we wish to

confine ourselves to a system with a fi z ed number of electrons. Therefore,

one wants to avoid " negative-energy solutions" of the Dirae equation, t hat

is solutions with eigenvalues E < - r nec 2 (defining E = rnec 2 for a free

electron at rest). Indeed, for variational problems, negative-energy solu-

tions may always occur since the Dirac Hami l t oni an is not bounded from

below. In spite of these problems, techniques have been developed to avoid

negative-energy states and to perform reliable relativistic many-electron

calculations for atoms and molecules. The most i mport ant feature com-

mon to many of the recipes appears to be t hat the ratio between upper and

lower Dirae components has to be fixed to a reasonable value and is not left

open to variation. It is thus a pragmat i c procedure to augment the set of

exact Dirae eigenstates as basis functions in a linear variational procedure

by hydrogenie wave functions in Eq. (6.55) generated by effective charges

6. 5. T WO - C E N T E R COUPLED- CHANNEL ME T HODS 175

Z* , see e.g. [TOE89, Eic90].

Loss of orthogonality

When pseudostates are included in the basis set, a difficulty arises in the

calculation of overlap and potential matrix elements t hat is peculiar to

relativistic projectile motion. As discussed in See. 4.1.4, the operator S

represents a Lorentz transformation only for rigorous eigenstates and rep-

resents an approximate Lorentz transformation for approximate eigenstates

such as those obtained by diagonalizing the single-center Hamiltonian ap-

pearing in Eqs. (5.53) and (5.54) in a basis containing pseudostates. For

example, approximate projectile eigenstates which, by construction, are or-

thogonal in the projectile system are no longer orthogonal in the target

frame, in which the collision is described. This means t hat the last of Eqs.

(6.60) is no longer valid, and instead, we have in the target system

- ~i' k' if i/, k / are bot h real states (6.67)

Ni,k, ~: ~i' k' if at least one of i ~, k ~ is a pseudostate

A similar relation holds for the Hamiltonian matrices. For example, if

the projectile Hamiltonian has been diagonalized in the projectile frame,

only the submat ri x spanned by real states remains diagonal under Lorentz

transformations.

The loss of the diagonal property of overlap and Hamiltonian matrices

under Lorentz transformations poses serious problems to calculation and in-

t erpret at i on within a framework of coupled channels involving pseudostates.

Then, no mat t er whether the projectile Hamiltonian is diagonalized in the

projectile or in the target frame, an observer in the target frame will always

find t hat for t -+ oc the coupling matrix ( H ~ - i h o Nv) in the projectile frame

is nondiagonal and hence the occupation amplitudes keep oscillating among

the various projectile basis states. Such a behavior is physically meaning-

less. Several approximate ways to treat these problems are discussed in

[TOES9]

The loss of orthogonality is generally unavoidable if basis states are used

t hat are not eigenstates of target or projectile Hamiltonian. Discretized

wave packets for the continuum, see See. 6.6.1, are expected to give rise to

similar problems.

6. 5. 4 Long- range coupl i ngs and Coul om b boundary

condi ti ons

In Figs. 6.12(a) and (b) we present calculated results for U 92+ + g 91-k col-

lisions which exhibit long-range couplings among the excited target states.

176 C H A P T E R 6. E X C I T A T I O N A N D I O N I Z A T I O N

In this subsection, we analyze this effect and show how to remove it. The

anomalous long-range couplings are of relativistic origin and are absent in

nonrelativistic coupled-channel calculations.

According to Eqs. (6.56) and (5.97), the interaction leading to target

excitation is

s2zP e2 [ 1 bxT+ 2v zT l

r~ =' YZPe2(1 - / 3az) ~oo + (R;) 3 + " " (6.68)

where R~o - (b 2 +~/2v2t2)l/2 and higher multipole terms have been omitted.

The main term in (6.68) giving rise to long-range couplings between initial

and final states is the term proportional to ~ a z / R ~ o , which is of magnetic

origin; for a more detailed analysis, see [ToE886, ToE88b].

Ba si s st a t e s sa t i sf yi ng C o ul o mb b o und a r y c o nd i t i o ns

In coupled-channel calculations, the choice of the basis states is a matter

of convenience, both for the desired interpretation of the results and for

the ease of calculations. In Eq. (6.55), we have chosen the conventional

view t hat the basis states are stationary atomic orbitals. In light of the

discussion above, it appears more appropriate to use a set of basis states

t hat satisfy Coulomb boundary conditions (5.61) and (5.62). We therefore

replace Eq. (6.55) with the expansion [ToE90b]

~ ( r T' t ) -- E a k ( t ) e - i up l n( R ' - vt ' ) ~ ) k ( r T ' t )

k

+ E a k , ( t ) e i UT l n ( R + v t ) s - l c k , ( r t p , t ' ) . (6.69)

k~

The first term represents an expansion in terms of target states asymptoti-

cally phase-distorted by the projectile, while the second sum is an expansion

in terms of projectile states asymptotically phase-distorted by the target.

When writing down the equations corresponding to (6.56) and (6.57), we

find t hat on the right-hand side, the time differentiation of the logarithmic

phase factors yields additional terms so that we get the replacement

Z p e 2 Z p e 2 Z p e 2

.._.+

' R ' r ~ r p

Z T e 2 Z T e 2 Z T e 2

- - . (6.70)

r T r T R

The phase factors themselves drop out in the one-center integrals since they

are state independent, but they remain in the two-center integrals. Wi t h

6.5. T WO - C E N T E R C O UP LE D - C H A N N E L ME T H O D S 177

these substitutions it is obvious how to modify the overlap matrix elements

(6.60) and the potential matrix elements (6.61). We emphasize t hat the

multiplication of the basis states with a phase factor and the simultaneous

replacement (6.70) of the interaction simply constitute a gauge transforma-

tion.

It becomes apparent t hat with the basis expansion of (6.70), the long

range of the Coulomb interaction is explicitly taken into account and hence

the long-range couplings discussed in See. 6.5.2 disappear. Indeed, if the

replacement (6.70) is inserted into the left-hand side of Eq. (6.68) the term

1/R~o in the square brackets responsible for these couplings is cancelled.

E l i mi nat i on of long-range couplings

The elimination of the long-range couplings by a suitable gauge transfor-

mation is illustrated in Fig. 6.14. Here, the time evolution during a U 92+

+ U 91+ collision at 1 GeV/ u is shown for the occupation probabilities of

the target states. The impact parameter is 0.01 a.u. or about one uranium

K-shell radius. Unpert urbed basis states (6.55) are used in the upper part

of the figure, while the lower part is obtained for boundary-corrected basis

states (6.69). Here, the long-range couplings visible for unpert urbed basis

1 1

states have disappeared. The 2P3/2( 89 2Pl/2(~), 3P3/ 2(~), and 3P1/2(89 )

states t hat are excited for unpert urbed states already long before the dis-

tance of closest approach is reached begin to be populated for boundary-

corrected basis states only at about vt = - 10 a K. Similarly, after the col-

lision, the boundary-corrected basis states attain their asymptotic occupa-

tion probabilities at much smaller values of vt t han the unpert urbed basis

states. Most importantly, the asymptotic values themselves are changed, in

sharp contrast to the Born approximation discussed in See. 6.1.2. A similar

behavior has been confirmed for single-center calculations [ToE90b]. Gen-

erally speaking, Fig. 6.14 explicitly illustrates t hat for an incomplete basis

set the coupled-channel equations (6.59) are not invariant with respect to

gauge transformations.

One can conclude t hat for relativistic single-center and two-center coupl-

ed-channel calculations with unpert urbed basis states, the coupling among

states is of long range and acts long before and long after the encounter

takes place, while for basis states satisfying Coulomb boundary conditions,

the couplings are localized around the distance of closest approach. How-

ever, in contrast to the Born approximation, Sec. 6.1.2, where there is no

backcoupling, the multiple couplings occurring in a coupled-channel calcu-

lation eventually lead to transition probabilities and cross sections t hat are

distinctly different for unpert urbed and boundary-corrected basis states.

This shows t hat it is mandat ory for a description of direct reactions to take

17 8 C H A P T E R 6. E X C I T A T I O N A N D I O N I Z A T I O N

1 . 0

10 -1

1 0 - 2

> , 1 0 - 3

o 10 -4

o

c. .

o . l O - S

t -

O

o 1.0

C3.

L.J

1 0 -1

1 0 - 2

1 0 - 3

1 0 - 4

1 G e V l u l s l l 2

u ? Z + + u 9 1 +

u n p e r f u r b e d

+

3 s v

2 s L , \

2 s 1+1z 2 P ; / 2

- / -

~ c l s t / z

" 3 s ~ ' /2

l s *

1 1 2

J

lo-' -o.2 o.o o.2 o'.,

b 0 u n d a r y - c 0 r r e c f e d

3 P ~ / z

, 2 ~ 3 / z - - - ~ , / R I I , ,

v t ( a . u . )

" 2 p~ '/2

F i g u r e 6. 14 . T i m e evol uti on of the occupati on probabilities of target states

in U 9 2+ + U9 1+( 1s1/ 2 ) collisions at 1 Ge V/ u l aboratory energy. Cal cul ati ons

are perf orm ed w i th 18 hydrogenic basis states at each of target and proj ectile

and f or an im pact param eter b - 0. 01a. u. T he upper set of curves represents

results obtai ned f rom unperturbed basis states ( 6. 55) , w hile the low er set of curves

corresponds to boundary-corrected basis states, Eq. ( 6. 69 ) . For the notati on, see

Fig. 6.1. From [ T oE9 0b] .

care of the long-range Coulomb interaction from the beginning. It is to be

expected t hat this al so i m proves the convergence behavior with res pect t o

the number of bas i s s t at es .

6 . 6 S i n g l e - c e n t e r c o u p l e d - c h a n n e l m e t h o d s

While two-center coupled-channel calculations have the merit of allowing

f or charge exchange, one obt ai ns great simplifications in a single-center

coupled-channel calculation, where one regards the projectile merely as the

s ource of a time-dependent electromagnetic field. In this way, one di s cards

6. 6. SI N GLE - C E N T E R C OUP LE D- C HA N N E L ME T H O D S 179

charge exchange not only as a final but also as an intermediate channel

and moreover does not permi t the formation of transient quasimolecular

states. On the other hand, by renouncing a second set of basis states and

the associated complications, one may handle larger sets centered around

the t arget and carry the calculations to higher energies.

By dropping the projectile-centered expansion in Eq. (6.55), we may

write

9 (rT, t) -- Z ak(t)~ bk(rT' t), (6.71)

k

where the Ok(rT, t) = pk( r T) e xp( - i Ekt / h) are mut ual l y orthogonal and

normalized bound-st at e and continuum ( E > rnec 2) wave functions. This

means t hat in each of Eqs. (6.60) and (6.61) only the first lines apply, in

particular one gets rid of the unpleasant t i me-dependent overlap terms.

Dropping the primed terms, the initial conditions (6.62), the expressions

for the probability (6.63), and the cross section (6.65) are valid in exactly

the same way.

6. 6. 1 Wa ve pa c ke t s f or t he c ont i nuum

The main problem of all coupled-channel calculations consists in t reat i ng

the continuum. In Sec. 6.5.3, we discuss pseudostates. Another met hod

which allows one to underst and bet t er which part of the continuum is

included is provided by the use of discretized continuum wave functions

constructed as superpositions of continuous eigenstates.

Weyl wave packets

One first segments a certain energy range of the continuum into finite in-

tervals centered around Ek and with the width AEk. By integrating the

exact continuum wave function 7) E e xp ( - i E t / h ) over the energy interval

A E k , one obtains Weyl wave packets in the form

~ Ek ( rT, t) -- 1 fEk+/XEk/2

v/ A E k JEk--/XEk/2 FE (r T ) e - i (E - E k ) t / h dE. (6.72)

These wave packets travel radially into the reaction region for t < 0 and

- - 2C4) 89 [HeG81,

outwards for t > 0 with a peak moment um Pk (E~ - rn e

MeR85]. Since the continuum wave functions are normalized on the energy

scale, see Eq. (4.117), and the energy intervals AEk belonging to different

discretized states do not overlap, the Weyl wave packets are orthonormal,

t hat is

/ ~ k ( rT' t)~ Et ( rT' t) ~k,1.

d3rT (6.73)

180 C H A P T E R 6. E X C I T A T I O N A N D I O N I Z A T I O N

go

E

0

o ~

(D

E

~ L

0 . 2

0 . 0

- 6 - 0 . 2

- 0

0

r Y

- 0 . 4

E0 = 2 . 5

lil lililil liflil:

Nl l U v :

0 2 0 4 0 6 0

R( ]d i ( ]l c o o r d i n a t e r

Fi gur e 6.15. The radial components G(r) = r g(r) and F(r ) = r f ( r ) of rel-

ativistic wave packets in relativistic units (h = c = me - 1) versus the radial

coordinate r. The s-state (a -- - 1) with the total energy E0 = 2.5 for Z = 82

is considered. The integration over continuum states is performed assuming a

width of AE = 0.3.

On t he ot her hand, t hey are no exact solutions of t he t ar get Dirac equat i on.

St a t i o na r y wave pac k e t s

If one superi mposes not t i me- dependent but st at i onar y cont i nuum solutions

of t he Dirac equat i on, one obt ai ns st at i onar y wave packets, see e.g. [BaS85],

~Ek (rT, t ) = ~Ek e x p ( - i E k t / l i ) , where

1 f E k + A E k / 2

~Ek (rw) -- v/ AE k J Ek-AEk/ 2 ~ P E( rT) dE. (6.74)

Owi ng to t he superposi t i on of adj acent cont i nuum functions, t he fall-off of

t he radi al wave funct i ons g (r ) and f ( r ) is much faster t han t he decrease 1 / r

of t he funct i ons g (r ) and f ( r ) themselves, see Fig. 6.15. In t he eval uat i on

of mat r i x el ement s cont ai ni ng t hese wave packets, it is somet i mes necessary

to t ake slowly varyi ng quant i t i es out si de of t he energy integrals.

Ga ussi a n wave pac k e t s

As is known from quant um chemistry, it is often advant ageous to expand

6.6. S I N G L E - C E N T E R C O UP L E D - C H A N N E L ME T H O D S 181

atomic or molecular orbitals in terms of Gaussian-type orbitals ( GTO)

(~z) e-~"~2 rZYl.~ (~), (6.75)

99n/rn(r) = E cu

v

where the solid harmonics rIYlm(~) are written in Cartesian coordinates,

and the Gaussian width parameters are chosen in a suitable (e.g., geometric

progression). It is the merit of this expansion t hat the single-center and

multi-center matrix elements entering in the Hamiltonian matrix can be

evaluated analytically. Collision problems, however, demand the accurate

numerical calculation of a complex auxiliary function defined as a one-

dimensional integral, which arises from the complex oscillatory factors in

Eqs. (6.60) and (6.61). It has been shown [TOE91, TOE92] t hat the method

of GTO wave packets is very powerful in nonrelativistic coupled-channel

calculations, even for energetic collisions when high-lying continuum states

are needed. The Cartesian decomposition, which comes naturally with

the GTO method, should be particularly appropriate for relativistic colli-

sions, where, because of the Lorentz contraction, the z-coordinate has to

be treated separately. Actual calculations have not yet been performed.

6. 6. 2 Nu m e r i c a l r e s ul t s

In order to solve the coupled equations (6.59), where N is the unit matrix

for a single center with orthogonal basis states, one evaluates the cou-

pling matrix elements (6.61) (first line only). Instead of a direct numerical

quadrature, one may use the multipole decomposition discussed in See. 6.3.

Adopting the multipole expansion in momentum space, See. 6.3.2, and

Weyl wave packets (6.72), Becker [Bec86] considered collisions of Zp + U

at 1 GeV/ u and b = 0. The calculated differential cross sections for K-shell

ionization are very close to the corresponding cross sections calculated in

first-order perturbation theory (see See. 6.3.3). However, for higher pro-

jectile charges (see Fig. 6.16), the calculations show t hat the cross sections

do not scale anymore with Z~ as they do in perturbation theory and hence

do not coincide in the diagram. All these calculations require considerable

computer time and cannot be easily extended to arbitrary impact parame-

ters in order to provide complete cross sections.

More extended calculations of excitation and ionization cross sections

in collisions of Zp + U 9~ at 430 MeV/ u and of Zp + X e 52+ at 82 MeV/ u

are performed by Momberger et al. [MoG90] within a one-center coupled-

channel formalism. While there is good qualitative agreement with the

experimental data, discrepancies at small values of Zp/ ZT still remain, see

Fig. 7.6 for projectile ionization. Differential ionization cross sections are

calculated in [MoG89].

182 C H A P T E R 6. E X C I T A T I O N A N D I O N I Z A T I O N

10 -31 Zp+U I

- I

10-s

e , , i

u

i =

e ~

N I O_ L , _

1.0 1.S 2 .0 2 .5

Ef (m c z )

Fi gure 6.16. Differential K-shell ionization probabilities in Zp + U collisions,

divided by Z~, at 1.0 GeV/ u and b - 0 versus the energy of the ionized elec-

tron. The curves are obtained from coupled-channel calculations by Becker et al.

[BeG87b].

N o np e r t ur b a t i ve b eh avi or at ve r y sma l l i mp a c t p a r a me t e r s

Figure 6.16 and the earlier discussions reveal t hat single-center and two-

center coupled-channel calculations yield significant deviations from the

results of pert urbat i on theory, Sec. 6.1. These deviations are expected to

be largest if the interaction is extremely strong, t hat is, for high-Z colli-

sion partners at very small impact parameters close to the limit of grazing

collisions between the nuclei.

In order to investigate this interesting behavior, Rumrich et al. [RUM91,

RuS93] have performed single-center coupled-channel calculations using sta-

tionary wave packets (6.74) for describing electronic continuum states (and

also positron states, see Sec. 10.2.3) and a multipole decomposition of the

interaction in coordinate space, see Sec. 6.3.1. Since they start from the

expansion (6.71) without boundary corrections (Sec. 6.5.4), they introduce

a damping procedure to avoid the time oscillations of the occupation ampli-

tudes at the internuclear separation where the initial conditions are speci-

fled.

In part of their calculations, they include negative-energy basis states.

The basis includes continuum states between 1.1 and 3.5 me c 2 and angular

moment um quant um numbers ~ bet ween- 2 and +2. It is found t hat for

the system Zp + P b sl+ at the bombarding energy of 1.2 GeV/ u and impact

6. 6. S I N G L E - C E N T E R C O UP L E D - C H A N N E L ME T H O D S 183

1 . 5

' 1 . 0

O

S

' - ' 0 5 ~

0.0

' I ' I ~ I '

: . . . - - ' " " ' - - . . . . . ~

-. . .

.~

' I

l , l , I l I J I i I

2 0 40 6 0 8 0 1 0 0 1 2 0

Zp

Fi gur e 6.17. lSl/ 2 and 2Sl/2 ionization rates versus Zp for the system Zp

P b 81+ at an impact parameter of b - 20 fm. The notation is: D, lSl/ 2 ionization

for 0.5 GeV/ u ; O, lSl/ 2 ionization for 1.2 GeV/ u; and / k, 2Sl/2 ionization for

1.2 GeV/ u. Full line: unitarity limit. From [RuS93].

par amet er b = 20 fro, t he ionization rat e is reduced by up to 30% if negative-

energy st at es are included.

Fi gure 6.17 shows lSl/ 2 and 2Sl/ 2 ionization rates divided by Zp 2 as a

function of the projectile charge Zp for t he i mpact par amet er b = 20 fm.

If per t ur bat i on t heory were valid one would have horizontal st rai ght lines.

The calculations show t hat t he 2Sl/ 2 ionization rat e is lower t han t he lSl/ 2

ionization rat e by a factor of 3.

In accordance wi t h t he previous remarks regardi ng t he influence of

boundar y conditions, it has been poi nt ed out by Bal t z et al. [BaR93b]

t hat t he nonper t ur bat i ve behavior seen in coupled-channel calculations is

st rongl y gauge dependent , in part i cul ar for small basis sizes. Insofar, such

results have to be viewed wi t h some caution.

In Chap. 10, in t he cont ext of el ect ron-posi t ron pair product i on, we

ret urn to the probl em of discrepancies bet ween cal cul at ed per t ur bat i ve and

nonper t ur bat i ve cross sections. In this case, the est i mat ed effects are by

far larger.

- Aieee 2002 qUploaded bysiddharth1996
- Penny EP Stuff > Eng Phys Serway 2Uploaded byapi-3724699
- Jntu Ece SyllabusUploaded byvmktpt
- Metal Clusters and Magic NumbersUploaded byyurmor
- Tutorial 3Uploaded bypradeep
- advbvUploaded bynicky_pradita
- 179439_pchem3_ch22Uploaded bySri Santika Fujianti
- QMprelim_F16Uploaded byShen
- Tensor Network in Chemistry: Recent DMRG/TTNS Studies and Perspectives for Catalysis ResearchUploaded bysouvenirsouvenir
- cosby1993.pdfUploaded byeze_firmatense
- n Comms 14277Uploaded byUtkar Sh
- Gianni Dotto Ring PatentUploaded byplan2222
- Discharge BifurcationUploaded byozlemixx
- Development of Medium Pressure Laser IonizationUploaded bymagicianchemist
- 947290680[1]Uploaded byCristhian Vizcarra
- innov-x_alpha_series.pdfUploaded byThiago
- Baum-ManzUploaded byJingrui Li
- 34072709Uploaded byAlberto Pérez
- 2012 james ruse phys.pdfUploaded byWilliam Hou
- notebook 4Uploaded byapi-340731900
- Wave Mech ProbUploaded byJosé Luis Morejón
- quiz-5.docUploaded byHằng Thanh
- The center of mass and center of charge of the electronUploaded byrgkelly62
- 2012_Physics_-_James_Ruse_Trial_with_Solutions.pdfUploaded byWilliam Hou
- Traduccion Transicion Landau-zenerUploaded byDiana Martinez
- Pt 2 Solutions Code 25-04-2013 EngUploaded byjassyj33
- Asymptotic Symmetries and Electromagnetic MemoryUploaded byFábio Duarte
- Dinesh 2008Uploaded bySiva Kumar
- Mechanism of Corona DischargeUploaded bymayuresh
- John M. Cornwall and Noah Graham- Sphalerons, knots, and dynamical compactication in Yang-Mills-Chern-Simons theoriesUploaded byPomac232

- Reference BooksUploaded byCalinescu Ionut-catalin
- PrefaceUploaded byCalinescu Ionut-catalin
- List of AbbreviationsUploaded byCalinescu Ionut-catalin
- IndexUploaded byCalinescu Ionut-catalin
- Chapter 2 - Relativistic KinematicsUploaded byCalinescu Ionut-catalin
- Chapter 4 - Relativistic Electron MotionUploaded byCalinescu Ionut-catalin
- Graphs and TablesUploaded byCalinescu Ionut-catalin
- Fundamental Constants and UnitsUploaded byCalinescu Ionut-catalin
- Chapter 7 - Ionization Many ElectronsUploaded byCalinescu Ionut-catalin
- Chapter 11 - Charge-state PreparationUploaded byCalinescu Ionut-catalin
- Chapter 13 - Cross Section DeterminationUploaded byCalinescu Ionut-catalin
- Chapter 12 - Target ArrangementsUploaded byCalinescu Ionut-catalin
- Chapter 9 - Radiative Electron CaptureUploaded byCalinescu Ionut-catalin
- Chapter 8 - Charge ExchangeUploaded byCalinescu Ionut-catalin
- Chapter 10 - Electron-positron Pair ProductionUploaded byCalinescu Ionut-catalin
- Chapter 5 - Ion-Atom CollisionsUploaded byCalinescu Ionut-catalin
- Chapter 3 - Fields of Moving ChargesUploaded byCalinescu Ionut-catalin
- Chapter 1 - IntroductionUploaded byCalinescu Ionut-catalin
- BibliographyUploaded byCalinescu Ionut-catalin

- What is..... Abel Summability?!Uploaded byAndrij Klein
- PhysRep Stochastic QuantizationUploaded byLuis Juan Utrera Molina
- complex numberUploaded byHaffiz Ating
- Rolling MagnetsUploaded byStefan Ličanin
- course.pdfUploaded byRico Pratama
- Using Ode 45Uploaded byxtreme301
- SyllabusF13_2415-2.pdfUploaded byjcastro237901
- AP AB Calculus - First Semester Review ProjectUploaded byFVCproductions
- Parte 2 Fundamentos Electromagneticos Con Matlab_Lonngren & SavovUploaded byAlbert Hernández
- MP EM Ass 16: Biot-Savart LawUploaded byBlueAstro
- Calculus BasicUploaded byajaysharma19686191
- syllabus_math3101Uploaded byDefri Ahmad Lagu Lamo
- 9780198425106Uploaded byarcanum78
- nutritionandfoodprocessingUploaded byapi-260101119
- Explorations Chapter 1Uploaded byisotherm
- Double and Triple IntegralsUploaded byxolar2002
- CIVIL Engineering jntuhUploaded bySankethPandu
- Chapter 02 (2)Uploaded byYe No
- syllABUSUploaded bynirmal_inbox
- S.Y. B.Sc. (Mathematics)-(2014-15)Uploaded byHardikChhatbar
- WillisUploaded byAbhiyan Paudel
- Syllabus 2ND SEMUploaded byrickdiesel
- B.TechUploaded bythamilarasu
- MathematicsUploaded byamjad
- harmapaUploaded bySunnyVishalSharma
- Damage-Tolerant Design of Stiffener Run-OutsUploaded byCatalin Samoil
- FSIUploaded byBSME HAI
- Acoustic BEMUploaded byTawab Abawi
- AIATS Schedule (2011, 2012)Uploaded bygoelbharat
- HydrodynamicsUploaded bydist2235