J.

%'.

OTZ AN

G.

M I

SSON

approximately 0.25. On the other ha, nd, for 660-kev photons, the total Compton cross section appears to be independent of the electron binding energy. It is interesting to note that at 660 kev, the small- and largeangle behavior of d|7~ has a compensating effect in which the total cross-section ratio, ore/a. r, is approximately equal to unity for both tin and gold.

ACKNOWLEDGMENTS

We wish to thank Professor Mare Ross for helpful discussions. Also one of us (JWM) wishes to thank Professor Mario Ageno for his suggestions during the course of the work and for his kind hospitality in making available the facilities of the Physics Laboratory of the Istituto Superiore di Sanita.

PHYSI CAL REVIEW

VOLUME

124, NUMBER

DECEMBER 1, 1961

Electron Scattering from Hydrogen*

CHARLEs SGHwARTz

Department

of Physics, U'nzoerszty of Calzfornza, Berkeley, Calzfornza (Received June 12, 1961)

Kohn's variational principle has been used to calculate S-wave elastic scattering of electrons from atomic hydrogen, using up to 50 trial functions of the type introduced by Hylleraas to describe the bound states of two-electron atoms. The phase shifts calculated at several energies up to 10 ev appear to have converged well, leaving residual uncertainties mostly less than one thousandth of a radian. Taking extra pains to include the effect of the long-range force at zero energy, we have also determined very accurate values for the scattering lengths.

INTRODUCTION

HE scattering of electrons from hydrogen atoms has been the subject of a great many calculations since it presents what is probably the simplest nontrivial real problem in scattering theory. We undertook a program of computing definitive values of the S-wave elastic phase shifts for this system, making no approximations other than those imposed by the 6nite speed and capacity of modern computing machines. Our use of the variational method' for this scattering problem completes, in a sense, the famous work on the bound states of two-electron atoms begun more than thirty years ago by Hylleraas. Probably the most interesting, and quite unexpected, result of this program has been the realization of the extraordinary nature of the convergence of the "stationary" phase shift. It has been recognized for some time' that, in contrast with bound-state problems, the addition of more variational parameters in a scattering calculation does not necessarily lead to a better answer. This behavior is blamed on the nonexistence of any minimum (or maximum) principle. The error in a variational calculation may be represented by

where 6 is the unknown error in the trial wave function. Only for systems where one knows the (finite) number of eigenvalues of II below the value E can one possibly state that the expression (1) must be negative. ' For scattering at any 6nite energy it is clearly impossible to make any such statement. We have discussed elsewhere4 how, by taking a great deal of numerical data, one can draw smooth curves and see an eGectively regular convergence for the general scattering problem at any energy. This paper will present the results of this treatment for the e-II problem. We use Kohn's variational principle,

ftan8/k7= tan8/k+ (2nz/ks)

where
(2rtz/Pt')

H)fdridrs, f(E

(2)

V 1'+ V s' (E H) = k' 1+ (2/rrs), + (2/ri)+ (2/rs)

with lengths in units of )tt'/ztze'; and our trial wave function is (for singlet or triplet states) p+g,

f=

io= (1&P)s)2e "zLsinkri/kri

5 (E H) Adr

(gl
t, m, n&0

+tan8 coskri/kri(1 e (a/2) (rz+rz)r, t

t"ts)"')7/4zrv2,

(3a)

using only a small number of parameters has recently been reported by Y. Hara, T. Ohmura, and T. Yamanouchi, Progr. Theoret. Phys. (Kyoto) 25, 46t (1961). ~ See, for example, H. S. W. Massey in Irundbgch der Physik edited by S. Flugge (Springer-Verlag, Berlin, 1956), Vol. 36.

*Supported in part by the Advanced Research Projects Administration through the U. S. Once of Naval Research. ' A calculation identical to ours but limited to zero energy and

X (ri

rs "&ri "rs )/4)rv2,

(3b)

and T. F. O' Malley, Phys. Rev. 119, 164 (1960), have in this manner established minimum principles for scattering at zero energy. These authors have recently extended their method to positive energies, but only by mutilating the potentials. 4 C. Schwartz, Ann. Phys, (to be published).

' L. Rosenberg, L. Spruch,

I-,

LEcTRON

CATTERI

NG
lb. O--

F ROM

H
0.003
roclloh
,

where Ei2 exc an coordinates ri an d x and ~is the exchanges ' h' h is variable scale parameter wnicu so im important in studyactual ua variation of parameters ing the convergence. . Th e ac yields the standard linear problem

-- 1. 94

1.92
1,90

15.5

003

(i, j=1, 2,
' Mz i2

15.0

d I0h

1.88

/If),
14.5
14.0
Ian
k

1.86
'I.84

(2'/8 )

Xz(E H)Xjdridrs)
Q

1.82

tan8

E, = (2ns/Ii')
'

X (Li'

r, dr2) H) pd
inear

13.5
I F80

13,0

le78

tion of tanb is also a trivial

12.5

1.76
1.74

12.0

1.72

number

E of param ameters.

The numbers

11.5

=ox
I
I

S
k
OA

=0
=

0.7
I I

1. 70
I

d 50 in the following results identi tif y the e o order of ' in includin al/ terms in (3b) through total or th e t rii 1 t t t d powers, (1+m+I), from 2 to 7 fo 1 to 6 for the singlet state, respectively.
RESULTS

0.6

1,0

1,4

1,8

2.2

1.0

1.4

1.8

2.2

(a)

(b)

hase shift at k=0.4; typical well-behaved results using only five values. es. b h S' shift at k=0.7; e phase glet ing p esult with wild values at ~=1. 4. typica l resu
//:

Fi ure 1 shows the stationary values o ~,tan6

fina 1 results

(see Ta bl e

is derived from such curves

w

for the tnp e 2 ev). This is our most thoroug y s u ie entin curves shown representing some 150 points in t is som so a reat deal of structure is no s o
erin p roblem see re ere of the scattering ' cer certainl y do exist or a . are so narrow as a unc ion ut the larger matrices, but ey ar

answer and assigned our uncertainty

ic s ou
e

ht that a reasonable measure o

characteristic

4). Similar singularities

0
K)

a
h

~~

say as a unc ion

and then come up again rom

leadtoa

asses through

rossovere overestimate of the accuracy; zero, an i s m

2.90 2.89-

Most of our results were obtaine rom on y 2 shows a typical case. Figure
/

lI:.
Il
Il

Il:

I
I
I

I
/

2.88
2.87

.
s

neighbors,

~=1. .4, g ave results noticeably "out of line" wit (ir= for the an siz

its

(1;
)l

t3

-tan3
k

E I. esu s o varia e-IX ariational ona calculations of S-wave e 7 scattering. The e numbers in parentheses give e u the last digit quoted.

2.85

S=O
Ltans/kg
k

5=1
(radians)
I

tans/kg

S (radians)

2.84
2.83

0.1
S =1
4

=0.4
32 34
I

0, 2 0.3 0.4

0.5

0.4

0.8

).2

l.b

20

24

28

ri let bases h'ft t k=0.4: the most thoroughly studies case, over. 150 po' oints represented.

ita

0.6 0.7 0.8 0.866

2.553 (1) 2.0673 (9) 1.6964(5) +15.87 (4) 1.4146 {4) 5.17 (2) 1.202 (1) 2.845 (8) 1.041(1) 1.917 (5) 0.930(1) 1.530(5) 0.886(1)
~ ~

6.68 (2) 9.23 (2) 26.4(1)

+150(20)

2.056(4) 2.260 (3) 2.492 (4) 2.833 (2) 3.384(3) 4.40(1) 6.74(2) 17.2 (6)

2.9388(4) 2.71/1(5) 2.4996(8) 2.2938(4) 2. 1046(4) 1.9329(8) 1.7797 (6) 1.643 {3) 1.563 (1)

CHARLES SCH WARTZ

7,1
7+0

r
h 13 x 13

atom is

7x

3x3

t at
2.20

(A' ~ 2rit rit) (n/r ), and it has been ' in such a circum cums t ance the h as m a zero energyis
1

6,9-

2.1 6
2.12
2.08 2.04
Ot

6.8 "

(n/2r'). (a/r)

6.7
6.6

6.S6,4-

2, 00

The slow ow convergence noted above for k=0 b, bl, arne d on the dificult cu y of expanding a ' ies o as slowl y as1 r' r in terms of r"e ". 't may be inferred th at the long-range At finite energies rgies it
is not hard to show that for the go c arge rom a neutral spin-zero s atomic oun -state system with groundt wave function h as t h e followin ex

1.96
'1.92

6.3 6.2

i-

1. 88

6.1
6.0
$,9
I

1.84

of th

coor oodiinate t

rp

of the scattering particle.

S=O
I I

I(

0
0.4

s =1

1.80

8
ro

0!
2mo'

0.4

0.8

1.2

2.0 2,

0.8

1.2

1.6

2,0

+
ro'W
ro'

'), +0(r(i

(a)

(b)

where W&') is the e po ol anzed target wave ve function,

(Ep

Fio. 3(a) and (h'i. ( ). irst zero-energy results.

Hp)e '

=DC(" .

(6)

data taken nearby, at ~=1.33 and 1.45 s i h ese proceedings one sh Quid dism 1 1 h in an use only the smoothest re
en

and

e ipo e moment o erator is i its s static polarizability,

R(g&(oi

D. gy(t)

(&)

a=1.8 and 2.4, but it

e next

rom
is seen.
'

ehavior

Thus an im proved choice for the function (3a) is

225

men s given iven in reference Bavin g sa satisfied the requirements n ee at =0 a strict minim

f]

cipie 1 for the scattering length ' Figure 3(a) and (b) s h ow our results w ooth b o1as; b ut it is also ere is very poor. A ces o successive minima of these curves sug1 gests a convergence like

cos
where the

(r, +r22/2)
12

r' ~1

f3
'

4mv2,

(8)

larities

+ 0: at r

s are functions
d

used to s

=595

pe

1 e o h e much bett ter results described f in the next section. Another conver gence enc problem arises at the her end A 'f th' of rgy region. Abov art of e asymptotic -' =( . a erm e' ", where k'=~k' ) &. is a small imaginar y numb um er and the o , h ch ou g find it di io wi 1 6 dimcult to repro 0 c ions bl'd' ~ we were unable to make k 0866 ductions from d t'for th's'n let s s more approachable, gy

e 1)! (n

ldp 'Iir 1

' ' ""'"""

t't"t

r o settin u al this improved ve p was very reat. . , an t e scale p e er rc in g was varied parameter n y. ero-e ese new ero-energy results, shown i i a much im rove rate approxxima t e1y equal to .1 E4 ur 6 1 p't'd 17686~0.0002 iscussion4 of these computations i t was
5L.
Spruch,

T. I'.
t

O'Malle a ey, and a

L. Rose berg,
"p

' We have
-RANGE FORCE PROBLEM OF THE LONG-RA
At

made a sim simplified analysis of this

l

'o p

fth

se investigations

It is well known that, in the adiabatic a r from a h y d rogen the pote tial seen n a at lar d arge distances

eingth

1 we used ony

resu s were essentially

no better than in the

gorato

t t

ELECTRON SCATTERI NG FROM

6.20
S =
1

808

It = 0 (improvedj

TAsLz
kap

II. e+-H phase

shifts.

=- 1

0 3X3 0 7x7 6.16 g 13 x 13

1.800

i 5px5p
6.12

~ 22x 22 ~ 34x 34
1.794

8 (radians)

0. 1 0.2 0.4 0.3 0.5 0.6 0.7 0.151 0.188 0.168 0.120 0.062 0.007 0.054

6.08

1.786

6.04

1.778

numerical accuracy at each step. This is a very interesting phenomenon for which we do not have any general theoretical understanding, but it appears to be a rather general property of variational calculations. The computations reported here were carried out on the IBM 704 facility of the University of California Computation Center.

'

6.00

/
~Z~g
i

QX
e'

P ~o ~$

77P

POSITRON SCATTERING
A few simple modiications of the programs allowed us to calculate elastic S-wave phase shifts for the scattering of positrons by atomic hydrogen. Since without the space symmetry we now need more terms in (3b) for each total power, (l+nt+tt); the results do not converge as rapidly as for e . The results, shown in Table II, have probable errors of about ~0.001 radian. For the scattering length we find the upper 2.10; and from the apparent rate of conbound u+~ vergence we believe that a+ will not be as little' as

5.96
i

o, = 5.96540.003
't

a, = 1.768640.0002
i
I

1.762

0.4

0.8

1.2

1.6

2.0

0.4

P.B

1.2

1.6

2.0

(a)

(b)

FIG. 4 (a) and (b). Improved zero-energy results.

mentioned that numerical inaccuracies Lround-off errors accumulated in solving Eq. (4)] could be a serious problem. In particular, we could not determine the stationary value from our 50+50 matrix at zero energy (the old way) to any better than about 1%%A. However, with this improved calculation the numerical uncertainties were much reduced. There thus seems to be a correlation between good convergence and good

2. 11.
8 This behavior was also noted at the end of Appendix 2 of reference 4. 8 Compare with previous results of L. Spruch and L. Rosenberg, Phys. Rev. 117, 143 (1960).

PH

YSI CAL REVIEW

VOI UME

124, NUMBER

DECEM BER 1, iggg

Continuous
Department

Photoelectric Absorption Cross Section of Helium*

D. J. BAKER, JR., D. E. BEDO,
AND

D. H.

TGMBQULIAN

of Physics and Laboratory of Atomic and Solid State Physics, Cornell University, Ithaca, Zero borh (Received Juiy 28, 1961)

The continuous photoelectric absorption cross section of helium has been measured in the spectral region extending from 180 to 600 A with greater accuracy and the observations are found to agree with the calculations of Huang and Stewart and Wilkinson. A grazing incidence spectrometer with a photomultiplier was used for a single measurement at 180 A while the remaining measurements were carried out in a normal incidence spectrometer utilizing photographic techniques. Whereas in previous experiments the absorbing gas sample was allowed to 611 the entire spectrometer chamber, in the current measurements the gas was confined to a small cell provided with suKciently transparent windows. The use of an absoprtion cell reduces contamination and facilitates the measurement of gas pressures. The results indicate that the cross section varies from a value of 0.98&0.04 Mb at 180 A to a value of 7.7~0.3 Mb at the absorption edge located at
504 A.

recent success in the development of which are transparent to extreme ultraviolet radiation, we have carried out measurements of the continuous photoelectric absorption cross section of * Supported in part by the Office of Naval Research.
windows

~OLLOKING

I. INTRODUCTION

helium. The use of such windows in the construction of a gas cell permits coninement of the gas sample to a restricted volume. This technique has certain advantages over earlier methods'' which used the entire

' P. Lee and G. L. Weissler, Phys. Rev. 99, 540 (1955}. N. Axelrod and M. P. Givens, Phys. Rev. 115, 97 (1959).