PHYSICAL REVIEW D

VOLUME 31, NUMBER 8

15 APRIL 1985

Differential cross section of electron-positron

bremsstrahlung

Eberhard Haug
Lehrstuhl fu'r Theoretische Astrophysik, Uniuersita't Tubingen,
Auf der Morgenstelle l2, D 7400 -Tiibingen, Federal Republic of Germany
(Received 29 August 1984)

A formula is given for the doubly differential cross section of electron-positron bremsstrahlung
which is exact in lowest-order perturbation theory. Angular distributions and energy spectra of the
emitted photon are computed in the center-of-mass system and in the laboratory system. The results
are compared with available approximations and with the cross sections of electron-electron and
electron-proton bremsstrahlung.

I.

INTRODUCTION

coming particles. This property is essential for the astroThe angle-independent

physical applications.
photon
spectrum can be easily computed by numerical integration. The results are compared with various approximations and with the corresponding processes of electronproton (ep) and electron-electron (e e ) bremsstrahlung.

The calculation of the fully differential cross section for

in collisions between
the production of bremsstrahlung
free electrons and positrons is a straightforward application of quantum
electrodynamics.
However, even in
lowest-order perturbation theory eight Feynman diagrams
contribute to the matrix element, four of them representing scattering graphs and four representing virtual annihilation graphs. ' Therefore the evaluation of the traces is
very laborious and the resulting cross-section formula is
extremely lengthy. It is most simply derived from the
corresponding expression for electron-electron bremsstrahlung ' by means of the well-known substitution law. "
Stimulated by experiments
with colliding electronpositron beams of high energy, most calculations of the
and spectrum of electron-positron
angular distribution
(e e+) bremsstrahlung were performed at ultrarelativistic
' where
energies'
only two of the eight Feynman diagrams give an appreciable contribution. Besides, it can be
shown that all the interference terms of the matrix element can be neglected within the high-energy approximation. The resulting cross sections were given either in the
center-of-mass system or in the laboratory system, i.e., the
rest system of one of the incident particles.
Recent interest in the process of e e+ bremsstrahlung
arose from the study of hot astrophysical plasmas which
are expected to exist in active galactic nuclei and in
' At semirelativistic
gamma-ray bursters.
temperatures electron-positron pairs are created through photonand particle-particle
photon, photon-particle,
interactions. ' For a full understanding of all the physical processes occurring in hot astrophysical plasmas it is indispensable to know the cross sections of the contributing
reactions in a wide energy range. So far, the lack of the
e e+ bremsstrahlung spectrum has been most serious. '
In the present paper a manageable formula for the cross
section of e e+ bremsstrahlung differential in photon energy and angles is given. It is obtained by integrating
analytically the fully differential cross section over the angles of the outgoing positron, without any approximations. Since the formula is expressed as a function of invariant products, it can be specialized to any frame of
reference, e.g., the c.m. or the rest system of one of the in-

lung.

31

2120

II. CROSS SECTIONS

In the elementary process of e e+

bremsstrahlung
p) and a
positron with four-momentum
q=(e+, q) collide under
the emission
of a photon with four-momentum
k=(k, k). ' The outgoing particles have the fourmomenta p'=(e', p') and q'=(e'+, q'). The differential
cross section for unpolarized particles is given by'

(Fig. 1) an electron with four-momentum

aro

dO

p'
q'
5"(p+q
k)
[(pq)

1/2
1]'~

=(e,

d p' d q' d k
e'I

t
e'+

= ,

is the fine-structure
constant
and
is the classical electron radius. A is the absolute square of the matrix element summed over the spin
directions of the electrons and positrons and the polarization directions of the photons; it is a complicated function
of invariant products between the four-momenta of the
where

o.

ro=e Imc

"

FIG. 1. Elementary process of electron-positron

1985

bremsstrah-

The American Physical Society

DIFFERENTIAL CROSS SECTION OF ELECTRON-POSITRON. . .

31

particles and photon which is too lengthy to be reproduced here. A was derived by means of the substitution
rule from the traces evaluated by Anders' for the corresponding process of e e bremsstrahlung.
By squaring the conservation law,

p+q =p +q +k

The choice of the signs in (11) depends on the frame of

reference considered. With the aid of (3) D can be expressed by the three invariant products (pq), (pk), and
(qk). Using (12) the fully differential cross section is
given by

= (pq) (pk) (qk)

d0=

[(pq)
)&d

(3)

5(6+ + e

1]'i

of

aid

=e'+q'de'+dQq,

and

the

(5)

relations
dkdQk, and

d'q'=

q' dq' d Q

E'

k)de+ =

e'-q' e'+(p'q')

e' q'
I

e'+(p'q')

e'

the fully differential cross section becomes

0 2

kq'

In order to express energy and momentum of the outgoing positron by the momenta p, q, and k, Eq. (2) is multiplied by the four-vector q' yielding

q'. (p+q

k) =(p'q')+

Bq'=D
Ce+

(9)

with the notations

CD+BR',
B2

BD+CR
B2

C2

With

W=(D +B

the cross section takes the form

0'ro 2 kq

2~

(pq)

=r,

(pk)

=~i,

(qk)

e'

C)' = , e'+(p'q')
q

(12)

&'

[(pq)

1]'i

=~p,

p'= (p'+ q')'= 2[(p'q')+ 1]

=2(r ~i i~q 1 1)
the e e+ bremsstrahlung
cross-section differential with
respect to the photon energy and photon angles can be
written as

0
k

pro

is the unit vector in the direction of the outgoq' = 1, the solution of (9) is
ing positron. Using e'+
C2

(15)

This equation is the most convenient starting point for the

integration over the solid angle Qq since only the function
A depends on the angles of the final positron, and this expression can be integrated exactly. As the momenta p, q,
and k form a triangle in the system S' [cf. Eq. (14)], the
p' is quite arbiorientation of the vectors p' and q'=
4m is allowed kinematitrary. So the full solid angle Qq
cally, i.e., the limits of integration are independent of the
mornenta p, q, and k.
The laborious but elementary integration yields a cross
section which again can be expressed in covariant form,
namely, by the three invariant products (pq), (pk), and
(qk). Introducing the notations

(10)

q '= q'/q'

S'

d 0
dk dQkdQq

o(k, 8) =

B=q'. (p+q k), C=e +e+ k,

D=(p'q')+1

(14)

1,

In the system

0
dk dQkdQq

(13)

8'=2@'p'=2@'q' .

k=k

e
e+

5(e+ +E'

kq'
W

p'+a'=p+a-j =0
e'
e'+
= e' = ,' (e+ + e k) = ,
'
(p'q') =2e'

e' e'+k

=[(p+q q' k)'+1]'" .

the

1]'"

It is easy to see that the function W takes a very simple

form if one specializes to the center-of-mass system S' of
the outgoing electron and positron where

e+ e k )

k,

q'd

where e' is now defined by

With

~' [(pq)'

dk d Qd Q,

can be derived. Using Eqs. (2) and (3), only 5 of the 10

cairo
The integrapossible invariant products are independent.
tion of (1) over d p' is easily performed by means of the 5
function resulting in

e'

0 2

d 0.

(2)

the relation,

(p'q')

2121

2m'

p(r

JadQ

1)'i

..

(19)

4)'
The expression for (1/m)(p
A d Q is given in
It is not possible to further integrate
the Appendix.
o.(k, 8) over the photon angles analytically. However, the
numerical computation of the cross section
d2
(20)
k

EBERHARD HAUG

2122

is quite easy in the c.m. and in the laboratory systems:

Because of the rotational symmetry around the momenturn vector p or q, respectively, the cross section is not
dependent on the azimuth angle; so the solid angle of the
2~ sin0d 0 and
outgoing photon is simply given by d Ak
one has to perform only a single integration over the angle

0.
In the
momenta
physical
numerical

general case of arbitrary directions of the initial

p and q, as, for instance, in a thermal astroplasma, the computation of o(k) requires two

integrations.

31

finite cross section at the short-wavelength

limit. This
fact is well known in the theory of electron-nucleus bremsstrahlung.
Generally, the factor (23) is dependent on the momenta
p, q, and k, i.e., it is different for various photon angles.
In the c.m. system, however, p is independent of the photon angle so that F + is only a function of the photon
energy k for given momenta p and q.
In the following sections the Coulomb factor is not taken into account because its effect is not significant at the
energies considered.

III. COULOMB CORRECTION

IV. RESULTS

The cross sections derived in the preceding section are

exact within lowest-order perturbation theory. Whereas
radiative corrections are assumed to be small in the energy
region considered, the Coulomb correction may be significant at low energies. Correct results can be expected only
if the conditions

The formula (A 1) can easily be programmed for the

of the doubly differential cross section
computation
o(k, 8). One should, however, pay attention to the fact
that roundoff errors may occur in the calculation of cross
sections at very high energies which may even lead to negative values of o.(k, 8). In these cases it is necessary to
employ variables with double precision. The following results for o.(k, 8) and o(k) are given in the c.m. system and
in the laboratory system where one of the initial particles
is at rest.

= 1,

a'=, 1

CX

(21)

are satisfied, where

(P

1 )1/2

A. Angular distributions
1]1/2

t)2

(p'q')

(22)

4)1/2

p' 2

Fe

a'

(23)

2'lM

This factor is always larger than 1 as a consequence of the

Coulomb attraction between electron and positron. That
is, the true values of the cross sections are always higher
than those given by the formulas of Sec. II. Due to the
small factor o.
a and a', however,
37 in the quantities
F + is approximately equal to 1, especially at high energies (p, p'= 1). An important
exception is the shortlimit given by p~2. Here a
wavelength
and
Oc
F + moo. As can be seen from Eqs. (19) and (Al),
cr(k, 8) tends to zero for p~2. By applying the factor
F + the quantity (p 4)'/ cancels out resulting in a

=,

'~

c.m.

system

The c.m. system is defined by

p+q=p'+q'+k=0,

are the relative velocities (in units of the speed of light) of

the electrons and positrons in the initial and final state,
respectively. For the brernsstrahlung
process in the field
of a nucleus, the cross section in the Born approximation
can be corrected by a simple factor derived by Elwert'
for nonrelativistic energies. This factor which is given by
the ratio of probabilities for finding the final and initial
electron, respectively, at the position of the nucleus, has
been shown to yield accurate results in the full energy
range for nuclei with low atomic numbers Z. ' ' In the
case of electron-positron bremsstrahlung
a corresponding
correction factor can be obtained by calculating the ratio
of probabilities of finding the two initial and the two final
particles, respectively, at the same position. It has the
form

in the

e+ e

e .
=

(24)

The invariants ~ and p have the form

r=2e 1,

=4e(e k),

(25)

i.e., they are independent of angles. Because of the symmetry of Eq. (Al) with respect to the products
a1

(pk) =k (e p cos8),

~2

= (qk) = k (e+p cos8),

(26)

where 0 is the photon angle relative to p, the photon distributions

in the c.m. system are symmetric
about
0 = m /2, and all angles 0 are allowed kinematically.
Figure 2 shows the cross section o(k, 8) as a function of
8 for the kinetic energy F.= (e 1)mc = 10 keV of th
e initial particles and for various photon energies hv=mc k.
At these nonrelativistic energies the photon angular distributions form smooth curves. The minimum at 0=~/2
for low photon energies changes into a maximum at
higher values of hv. The corresponding cross sections for
e e bremsstrahlung
are smaller by a factor of 20 to 30
resulting from their quadrupole nature. The maximum
bremsstrahlung
are given in
values of o (k, 8) for e e
Fig. 2 by marks at the vertical axes.
In Fig. 3 are plotted the angular distributions for mildly
relativistic particle energies, E=300 keV, and various
photon energies. Due to the relativistic beaming most of
the photons are emitted near 0 = 0 and 0= ~. The angular
distributions are similar to the corresponding curves for
e e bremsstrahlung,
however, the latter cross sections
are still smaller, in particular for low photon energies.

DIFFERENTIAL CROSS SECTION OF ELECTRON-POSITRON. . .

31
40

F =10 keV

20100

energies (e&~1) the cross sections for

bremsstrahlung
are virtually equal. For
E=50 MeV (cf. Fig. 3 of Ref. 3) the relative differences
are of the order of 10 for the important angles around 0
and rr. Here the relativistic beaming is most pronounced,
the width at half maximum
of the curves being
8&&2-1/(2e). Around 8=m/2 the cross section has decreased by many orders of magnitude.
An excellent approximation for the cross section o(k, 8)
in the c.m. system at ultrarelativistic energies which holds
has been deboth for e e+ and e e bremsstrahlung
rived by Baier, Fadin, and Khoze (BFK) who systematically expanded all quantities in powers of 1/e and retained the leading term of the expansion. The equality of
the cross sections for the two processes follows from the
fact that at these high energies the exchange-type Feynman graphs for e e collisi. ons make the same contribution as the diagrams of direct type, whereas the contribution of the annihilation graphs in the case of e e+ collisions can be neglected. The agreement between the BFK
formula and the present results is better than 0. 1% for
E=SO MeV and the important photon angles. The only
distinct differences occur at photon energies very close to
the high-frequency limit p =4 or k =p /e at large angles
8. Here the cross section for e e+ bremsstrahlung is
higher than that given by the BFK formula, in contrast to
the e e case where it is lower. The inaccuracy of the
approximation
is, however, not significant because the
contribution to the total cross section o(k) from these
large angles is fully negligible.

At relativistic

30

-e e-

1.5-

hv

~10 keV

1.0
Vl

0.5Sh

e e-

0
0.5

0.4
0.3

0.2
0.1-

-e-e-

0 Oo

30o

60o

90o

120o

150o

180o

FIG. 2. Differential cross section o.(k, o) of e e+ bremsstrahlung in the c.m. system for E=10 keV and various photon
energies h v. The maximum cross sections of e e bremsstrahlung are given by marks on the ordinates.

e+

and e e

B.

Angular distributions

In the laboratory system one of the initial particles is at

rest. Choosing q=0, the quantities ~ and p have the
for m

mb

sr

r=e,
00 ke
10

=2[v +1 k(e +1
p cos8)] .

(27)

Here 8 is the photon angle relative to the direction of the

incoming electron. The maximum photon energy occurs
for p =4 and is given by

k,(e,8) =

6
keV

in the laboratory system

10 keV

~(k, g)
keV

2123

sr
4

The absolute maximum

direction 0=0:

k,(e

0 ee-

)=

(28)

+1 p cosO

of k is reached in the forward

(29)

p+1

Photons with energies (e 1)/(e +p+1) &k

(e )
can be emitted only into a cone with half apex angle 8o

&k,

6
10 b
keV. sr

given by
0

00

300

600

90'

1204

150o

g 180o

FIG. 3. Differential cross section o(k, O) of e e+ bremsstrahlung in the c.m. system for E=300 keV and various photon energies hv. The maximum cross sections of e e bremsstrahlung are given by marks on the ordinates.

= (e +1)k (e 1)

(30)
pk
Figure 4 shows the cross sections o.(k, 8) for the kinetic
l)mc =10 keV of the incoming elecenergy E =(e
tron for various photon energies. For comparison the angular distributions of electron-proton (ep) bremsstrahlung
are depicted. The ep cross sections are smaller at low
cosOp

2]24

B~RHARD HAU
3
0(k, g)

1p keV

=3pp

keV

mb

key sr
b
keV

s'

300
I

8-

80

10 b
key. sr

mb

key sr

gp4p-

3Po

60'

3Po

60'

120'

150'

120'

150'

6-

ep

20
I

900

00

30

1200

150'

mb
key. sr

'b

gpo

2-

10
key. sr
1.5

p. 5

g00

300

crross section
the laboratory s y stem for

F&~ 4

Dy ffeerentia

strahlung
p &, comp
photpn en g'
n protonn bremsstra
t h]ung

'th

75P

(k, e) of

' e+ brems-

"

E = 10 k ey
the

2p

and var'
crosss section of

eg~.

wherea~ t ey exceed the e e + CI OSS SeCphoton

ergreS, w
tions at h 1g herp h oton energies.
l tted the a g ular
istrjbut&ons for
In FIg
are po
= 300 keQ,' aga ' in comparisoon wit' h the cross sect1on
hlung. Wi'th Increasing p hoton en+ b
the
a
o
owed
sstrahlung
angular
reg
ergy
is more a nd more re stricted due to the kinemat1ca
le o tE
characterized b y t e . miting ang
(30)]. T
'
shapes of thee e e + ang ular distri butjons are similar to
those
e bremsstra
trahlung (cf. Fi1g . 4o f R f 3), how'
e case by
herinthe
e -e+
ever, the va lues of ~ (k g) areghe
.
factor of a bo3.' At low and edjum Photon energies
the e e + cross sect, 1onss exceed the ep cross se ctions (e"
cept or the small conntribution at
g les 6'& g ) whereas
h ep
e bremsstra ahlung domhigher p hoton en ergies the
inates
riesthe e e+
e bremsstra.
e
ward direct o
1 to the e e
tion is&u y
c
hoton energies and ang 1es the relative
Hence the r
roximation
mu
su ffi clen
angles ~ an
ave
s 1n the e e
case.
limit,
g
omentum transfer
f too the I eco in g artlc' l e is very low at
t e
g + $ /p
the cross sect1ons
e e+
roxirnately equa
e and e e brems stra hlung are approxirn

10o

po

3po

20

0
g 40

FK". 5.
e+ bremsrenti
ross
o section ~ k,' g) o
strahlung in
e la ora tory system for E 300 ke g and var&pus photonn energ&es
en
~ith t
compared w
pss sectipn o
].ectron-proto pn bremss trahlung
ra
e maxim m cross see(ep
'
tions pf
rem ssa
trahlung are g en by mar kss pn the prdinates.

ec

tot
erg ies.

22

oss sect on for eP

at very high en-

C. Photon spectra
ra in tthee c.m. system
lculated by num
numerical
h
s the spectraa for kinetic

ke)i

oe

integra-

300 keV. At low

&1, the short-wave en

lung spectrum Is g'

=2
( & )

=2E .
or h +max-

kmax=p
max

dr p ole approx1rna
oxirnation

onre

e formula
rahlung in t h e

16~ro

or t e

Garibyan
ss s

c.

p+p k'~
( k)

')

(31)

ssion agreees
. 6, this expre ss

quite
lt fo k' etic energies
th.
f h d
ewp ercent. In
ia energies s E, however, the apthe region o
of interme diate
es'
as
o
be use dasaroug
ximation (31) may only

..,.
resen t r

r
For ex treme-relativis

fa

'

sstthee expressionn'

DIFFERENTIAL CROS
ROSS SECTION OF EL

31
500

ON-POSITR

2l25

ion of the two curvess is sshifted

i te to lower valuues of

max

4QO

'c energies
a
b derived from Eq.. (31) taking m o
or

300

ula can

(k)

The argumentnt o
of the

has to be dependent
200

If

b't

the initial mome

menta p an
possi'ble photon energy is

100

(pq)

0keV

80

'

E 40

50

100
E=

60-

150

R(k)

200 keV

=" "'1 1+

this is equivalent

0),
q

20

FIG.
E'ER(k)
. s stem for various
nonrelativistic approxima

= garo

g ( k)
'
ini

of e

400

hy

pk

500

e + bremsstrahlung

'

in the

ith the

'

3'

ek

'tic

b written as

x ) '~
(1

(34)

ln 4e

In the c.m. system

'
to Eq.. (31), an d in the laboratory y

64 ~ro

300

200

100 keV

to

300 keV

40

'

""

cross secti'on for e e + bremss ""hlu"

20

(33)
)

'stic limit this express~onn rreduces

f io of

~b

hi h

~++~ p+0

60

th

2e k

p+p-4k)'
p

(35)

( 4k'

'

kinetic energie s E &20 keV Eq.. (35) is a good ape t for the neig o
prox
ofh
h
e ends a
'
'
i e on the choiceeof
little
xw
b e ak
o xwhichcan
=4k/

1.5

ko(k), b

(32)
'

for the cross sectionns

h
bremsstrahlung
as 1on

e e

of both
d

0.5

E ~100 keV

0.2-

appreciably

th

h'

hf

do

which is vali'd for all p h oton

udin

the

ko(k), b

ar

11
k 8)th
or t
tio
e t
'
e g ies
approxima ation down to ener
'b
rtalso the sharp drop of 0 k) nnear the shorwavelength limitit see Fig.
ig. 7 of Re f. 3) in contrast to Eq..

(32).

0.1

so kiy

70

80
k o'(k), rr) b

60
40

D. Photon spectra
(q

=0)

in the
t e laboratory
a
system

20

n s ectra in the laboratory

'
for kinetic
etic eelectron
ec
energies

and

lotted.

Th e ltt
a
b t th

e e cro0
hi g herp
speectrum is harder t an

oon

o
'

d
d th
t e ep

50

100

150

e+ bremsstrahlung

r various electronn en
section of electron-proton

bremsstra

200 keV

h~

ung

'

in the

EBERHARD HAUG

or x =2hv/E . For intermediate energies E the formula (35) can still serve as a rough approximation as can
100 keV.
be seen in Fig. 7 for E
For the energies of Fig. 7 the cross sections of e e
are considerably lower than the e e+
bremsstrahlung
cross sections, the ratios being about 150 for E =10
keV, 10 to 15 for E =100 keV, and 3 to 5 for E =300

laboratory system in the whole range between nonrelativistic and extreme-relativistic energies. From the comparison with available approximation formulas and with
the cross sections of other bremsstrahlung
processes the
following conclusions can be drawn.
are con(i) The cross sections of e e+ bremsstrahlung
siderably higher than the e e cross sections up to enerenergies the
gies of a few MeV. At extreme-relativistic
cross sections for the two processes coincide except for a
very small region near the high-energy end of the spectrum.
(ii) In the laboratory system the cross sections of e e+
are of the same order of magniand ep bremsstrahlung
tude. The ep spectra are harder than the e e+ spectra.
(iii) The analytical formulas for the cross sections in the
limits were found
nonrelativistic and extreme-relativistic
to be good approximations, in particular in the c.m. system. At intermediate energies, between =20 keV and
=10 MeV, there is no-alternative to the formula derived
in this paper.
(iv) In hot astrophysical plasmas with a positron comwill give a significant
ponent the e e+ bremsstrahlung
contribution to the total x-ray emission.

keV.
the
In the extreme-relativistic
energy region e
cross sections for e e+ and e e bremsstrahlung coincide. The approximation formula of Baier, Fadin, and
Khoze for the laboratory system is not as accurate as the
expression for the c.m. system since the reciprocal expansion parameter (pq) is equal to e in the laboratory system
(q=0) but it is =2@ in the c.m. system. As a consequence the agreement between the present results and the
BFK formula is better than 1% only at energies E
beyond 1000 MeV (see Fig. 10 of Ref. 3).

V. CONCLUSIONS

spectra for e

Angular distributions and photon

bremsstrahlung were computed in the

c.m.

31

e+

system and the

APPENDIX

The integration of the fully differential cross section over the angles of the outgoing positron results in the formula:
3
4K)

, (p'+3)

'T+

(p

Xg
2K)

K2P

K)K2P

v+1 K2
2K, K

X4

4)

K)p

HX5

(KI+K2)

K)

K27

2+ 2/c )

K)K2

4z
2

2K2 7

2. 5p
2(r~ 1)'

&+ 1

4z2

+3

K)

K2

2v+ 3
2

K)(H 1)

K)K2P

4K)

K2

Kl+Kp

(K) ~Kg)

SC2

+ 2(r~ 1)

12K'

K2P

1
&2
+
+ 3~
2K&(r+ 1)
2K(Kp

p-L)
W')

Kl

1
~X2 w+ 12(r+1)

4) ~3K, ~6
(3Kz'~2K,

2(r~ 1)

5z 11~j ~ 1

13%+7
2K2

2p

7K~

+4K' 10&+2+
1

K2

( 11K

27p

+ 2p

K)K2

K(

(Kq

Kg&+Kg+4)

)r+ 4r+ 9K) 8 ) 2+

8

8~
K)K2

..

DIFFERENTIAL CROSS SECTION OF ELECTRON-POSITRON.

+4p

1K1

fey

3K' + '7

Kl(r+ 1)
K]

+
+PL, 3 .

1
8

Kl+K2

KIKp

+,
2

1)

2KIKp

,K1K2

4p

K2 2(r+ 1)

zIP

3r(Kl+ K2) 2

K] +Kg

2K

2
2K2r
+K2 2
p 1)

3K I K2r+ K 1 K2)
2K]Kp

2K) 1 +

ICI

3+ (Kl+K2)

x&~z

K]

1
+ g+1

+K2

(Kl+K2)

r(p 2 2)+ KI p +2+2 Kp (2r

2

1
(4r 3
+ K(+Kg

+ p'12

4+]+2

4+1+2

2+
(Kl+K2)
2

(Kl+K2)

2K1

K 1+2

(p'+6),

' +

27

+ IC2)

(Kl

+K2)

+2

1
1
+ (r+1)'
2
KI +Kg

pI

IC1(IC1+IC2)

7+1

+ K1K2(r

27-3

4r 4
r K2+ 2)

1
(3K2r+ K1K2
+ 2(K1+K2)

(K, +K2)(r+1)

(Kl

Kp

3K2r+2K2 ]+
2(r+1

[(r ,K2)p +4r

r+2 +

2127

+ 2m~

K)+Kg

Kp+ 1
KI

1) + 4r
K2(r+

4r

Kp

2r

4r-Kp

KI

. + IK1~K2I

(Al)

with the notations

L] 1n

p(r

4)'/
1)+(p
Wl

K2

2~I

r' 1+(p/2)[(p 4)(r

K] +Kg

Xl

8'I

PR
Xs P
W,
X4 =

8I

(p

4)-]/2 + (K2 r+

~1

~I +

'

(K2

Xs=,(r 1)
Kl

K2

++1
Z~,

'

1)]' '

4)+p(p

/
4)'
W2

1]',

X2

2p

K, '(r'

I.3 1n + (

4K1(K1+K2)

8'1 [(r
K2) +2Kl

1),
p~ I

(p

1)
p(p

4) 1/'2

K]V

W2

Kl+ K2

= IK2[K2p

(p

4)1/2

4)+8K1(K1+K2)]I

2I.
+ (r K2r+ K 1) (r 1)g '/2
1

(A2)

1 2
P(P'2 4) '/'

r+ 1)(p' 4)'
p(p'

1n 1+ K2p

I-&

'

4)'/' 1
Kl

S + [(K,r+ 1) R ' 2K, 'S]

8']

3S

(r2 1)

R =(r+1)(Ic2
Kl)
K2(K1+K2), S=2K,K2r

+ 2(r+ 1) p'K12( 3S
IC1

K,

K2

1)

2I.4
(r2

1)

I/2

2128

EBERHARD HAUG

The complicated function A in the integrand was derived from the corresponding expression for e e bremsrule.
The
strahlung
by means of the substitution
correctness of A was verified by a comparison of e e
cross sections with an independent calcubremsstrahlung
The two results agreed exlation of Mack and Mitter.
The correctness of the formula (A 1 ) was
cellently.
checked as follows.
(1) All terms of (Al) including those which are obtained
by interchanging sc& and ~2 have been calculated separately. Thus all the errors violating symmetry could be
discovered easily.

G. Altarelli and F. Buccella, Nuovo Cimento 34, 1337 (1964).

I. Hodes, Ph. D. thesis, University of Chicago, 1953.
E. Haug, Z. Naturforsch. 30a, 1099 (1975).
~J. M. Jauch and F. Rohrlich, The Theory of Photons and Elec
trons, 2nd edition (Springer, New York, 1976).
5S. M. Swanson, Phys. Rev. 154, 1601 (1967).
V. N. Baier, V. S. Fadin, and V. A. Khoze, Zh. Eksp. Teor.
Fiz. 51, 1135 (1966) [Sov. Phys. JETP 24, 760 (1967)].
7V. N. Baier, V. S. Fadin, and V. A. Khoze, Zh. Eksp. Teor. .
Fiz. 53, 2194 (1967) [Sov. Phys. JETP 26, 1238 (1968)].
8V. N. Baier, V. S. Fadin, V. A. Khoze, and E. A. Kuraev,
Phys. Rep. 78, 293 (1981).
F. A. Berends, R. Kleiss, P. de Causmaecker, R. Gastmans,
and Tai Tsun Wu, Nucl. Phys. 8206, 61 (1982).
F. A. Berends and R. Kleiss, Nucl. Phys. B228, 537 (1983).
'G. Cavallo and M. J. Rees, Mon. Not. R. Astron. Soc. 183,

359 (1978).
A. P. Lightman, R. Giacconi, and H. Tananbaum,
J. 224, 375 (1978).

Astrophys.

31

(2) The cross sections computed with the aid of (Al) in

the c.m. and laboratory systems were compared with the
results from numerical integration of the triply differential cross section over dQ& yielding full agreement within
the accuracy of the numerical integration in all energy regions.
An additional test for the correctness of (A 1) provides
the fact that the resulting cross sections in the c.m. and
laboratory systems agree with the available approximations both in the nonrelativistic
and in the extremerelativistic limits.

B. McBreen, Astron. Astrophys. 71, L19 (1979).

J. Rees, in Plasma Astrophysics, edited by T. D. Guyenne
and G. Levy (European Space Agency, Paris, 1981), p. 297.
R. Svensson, Astrophys. J. 258, 335 (1982}.
S. Stepney and P. W. Guilbert, Mon. Not. R. Astron. Soc.

~4M.

204, 1269 (1983).

Relativistic units mc and mc for energy and momentum are
used. The metric is such that the product of two four-vectors
a.b.
a = (ao, a) and b = (bo, b) is defined by (ab) = aobo
T. Anders, Ph. D. thesis, Universitat Freiburg, Germany, 1961;
Nucl. Phys. 59, 127 (1964).
G. Elwert, Ann. Phys. (Leipzig) 34, 178 (1939).
G. Elwert and E. Haug, Phys. Rev. 183, 90 (1969).
'R. H. Pratt and H. K. Tseng, Phys. Rev. A 11, 1797 (1975}.
J. Joseph and F. Rohrlich, Rev. Mod. Phys. 30, 354 (1958).
G. M. Garibyan, Zh. Eksp. Teor. Fiz. 24, 617 (1953).
2~A. I. Akhiezer and V. B. Berestetskii, Quantum
Electro
dynamics (Interscience, New York, 1965).
D. Mack and H. Mitter, Phys. Lett. 44A, 71 (1973).