Nuclear PhysicsA263 (1976) 75--85; (~) North-Holland Publishing Co.

, Amsterdam
N o t to be reproduced by photoprint or microfilm without written permission from the publisher

O N A B S O L U T E V A L U E S OF ~-DECAY RATES
T. FLIESSBACHt
Lawrence Berkeley Laboratory, Berkeley, Calif., USA
and
H. J. MANG
Technische UniversitiitMiinchen, 8046 Garchiny, Germany
Received 8 August 1975
Abstract: A possible way to remove the discrepancy between calculated and measured ~-widths is
discussed. The decay rates of 212Po and 21°po are computed with the help of shell-model
wave functions for parent and daughter nucleus.

1. Introduction

The discrepancy between calculated and measured ~-widths on tb.e one hand and
the rather good agreement for relative decay rates on the other has long been a cum-
bersome fact for theorists *). In the following we suggest how to possibly remove
this discrepancy. As a first step we offer a thorough criticism of those basic assump-
tions which underly all existing theories of ~-decay.
This is done in sect. 2 where also a general formula for the decay constant is
rederived. In sect. 3 the computational techniques which are used are described,
and sect. 4 contains an application to the decay of Z*2po and 2*°po.

2. A formula for the decay constant

Starting from the Schr6dinger equation

ih a = hie>, (2.1)
~t
with the initial condition
I~k>,=o = [q~a+4), (2.2)
where [~a+47 is the wave function of the parent nucleus, one can obtain the decay
constant in the following way: Write

l~b> = a(t)l~a+4) + f b(E', t)lq, r,>dE', (2,3)

* On leave from Techn. Univ. Miinchen, 8046 Garching, Germany

75
76 T. FLIESSBACH AND H. J. MANG

with
(~EI~e,) = 5(E--E'), (2.4a)
<~elHl~g,> - - E,~(E-E'). (2.4b)
The ket ICe) is of course not an exact eigenstate of the Hamiltonian, but the set [~n)
diagonalizes H in a subspace of the Hilbert space. The wave functions ICe) used
here are given in eq. (2.7) below. Upon inserting I•) ofeq. (2.3) into the Schr6dinger
equation, one obtains with the help of well-known techniques 2)

la(t)l 2 = e -x', Eo = (~a+4lHl~a+4>,

2 -- 2= [(~a+4lH_Eol~go>12" (2.5)
h
The essential assumption which enters into this derivation is that the width F = h2
is small compared to the kinetic energy Ex of the emitted ~-particle. This assump-
tion, however, is not problematic at all.
The trouble starts as soon as the matrix element (~A+41H-EoI~Eo> is evaluated
with model wave functions for I~a+4) and I~Eo). The wave function of the parent
nucleus [~A+4) is then taken as a generalized shell-model wave function 1) and
I~Eo) is written in the more general case as
[~eo) = E d [ U~,(R~,)X~,~a), (2.6)
V

where Iz~) is the intrinsic wave function of the ~-particle and 1 ~ ) is the wave
function of the daughter nucleus. The function U~,,(R~,)describes the relative motion
of the two nuclei. The index v labels the states of the daughter nucleus that can
be reached in the decay. The symbol d stands for antisymmetrization. The proper
coupling of angular momenta is understood.
For the sake of simplicity only the case of a single 0 ÷ to 0 + transition will be
considered in this paper. In this case the wave function I ~ o ) h a s the simpler
form
I ~ o ) = d l Uv.~,(R~,)Z~~a). (2.7)
Before going on, it should be noted that when I~eo) is introduced into eq. (2.5),
the antisymmetrization operator ~¢ can, of course, be replaced by a factor [-(z)g)-]~.
Obviously the approximate expressions for ICeo) are good approximations only
for not too small values of R~ because all closed channels are neglected in
eqs. (2.6) and (2.7). Being more specific, this means that R~ should certainly be larger
than the rms radius of the daughter nucleus 1). Such a restriction poses no serious
problem provided [~A+4) is eigenfunction of a Hamiltonian H 0 where
nol~a+4> = Eol~a+4>, (2.8)
which coincides with the Hamiltonian H for radii smaller than a radius Ro which
in turn is large enough so that I~o> is still a good approximation at Ro in the sense
 g-DECAY RATES 77

just described. Then values of R~ which are smaller than R 0 give no contribution to
the matrix element ( ~ a +41H- Eol~Eo). If, moreover, the relation

HlCno> = Eol~Eo>, (2.9)

with
H = ~
Hiatr+ a _ _ _ A. + W,a(R~),
Hintr
2M.
is valid for values of R~ larger than R 0, then the matrix element can be brought into
the form

(~a+4IH-E°I~E°) = '~d OR ~ /R=Ro dO, (2.10)

N Z

That such a radius can be found, is the basis of the theory. Attempts have been
made to avoid the introduction of such a radius 3), but then the question arises how
to compute I~Eo) for small values of R.
In the derivation presented here, the next step is the calculation of U~,(R). It can
definitely not be calculated from a Schr6dinger equation of the type

- A+V(R)-E Ug(R) = O, (2.11)

2M~,
where V ( R ) is an ordinary energy independent potential. Clearly eq. (2.11) has as
the consequence that
(U~[ U~,) = 6 ( E - E ' ) , (2.12)
and this equation contradicts eq. (2.4a), as can be shown in the following way:

1~) = .I d 3R Un(R)I,~b(R- R~)X~~A). (2.13)

We define 4)

(..e~c~(R-Rc,)X~,dpal,.~C~(R'-R~,))C~,qba) = 6 ( R - R ' ) - K ( R , R'), (2.14)

and use in the following the short-hand notation

(1 -/~) U(R) = f (6(R - R') - K(R, R')) U(R')d3R '.

,J

The normalization of UE is then

(V~ll-1?,lU~,> = ~(E-E'). (2.15)

This normalization of UE deviates from the normalization to a di-function (eq. (2.12))
and this deviation has a drastic effect on the final result.
78 T. FLIESSBACI-/AND H. J. MANG

In principle, UE is determined by eq. (2.4b) together with the normalization

eq. (2.4a) just discussed. Eq. (2.4b), however, leads to a Schr6dinger equation for
UE with a non-hermitian Hamiltonian which contradicts eq. (2.11). Therefore, one
cannot use a real local potential to calculate UE in a consistent theory. One way to
circumvent such difficulties is to introduce a new function 4)t

= (1- (2.16)

The wave function I~E) can be written as

ICE) = f d aR f2r(R)[~¢(l - g ) * - ~'5(R- R,,)X~,~a). (2.17)

It is now consistent to approximate f2g by a solution of a Schr6dinger equation

with a real local potential. This can be seen immediately for the normalization
(eq. (2.4a)), because the states [ d ( 1 - / ~ ) * - * r S ( R - R ~ ) z ~ # a ) are exactly normalized
to 5(R-R'). The Schr6dinger equation for f2E is discussed in detail in ref. 5). The
matrix element (#A+4[H-Eo[#Eo) can now be written

(~a+,,lH_Eolq, r.o) = h2R2f [g2 OGN df2E,,GN) (2.18)

2M~ d ~ n, ~ dR R= Rodr2'
with
GN(R) = (1 - g)-~G(R). (2.19)

In practice this means that one must substitute GN for G in the usual a-decay theory.
At the same time, UE has to be replaced by f2E. In practice this means no change
at all, because Ur has been calculated always as if it were f2E.

3. Calculation of the operator ( l - K ) -~

The essential problem now is the calculation of the operator ( 1 - / ~ ) - ~ . The
treatment will be restricted to states I~A+4) and I~A) with angular momentum zero,
as already indicated. The radial wave functions of the single-particle states are taken
as oscillator functions.
Two possibilities have been considered to calculate eigenvalues and eigenfunctions
of/~:
(a) The state I d S ( R - R ~ ) x ~ ) is expanded in a series of oscillator wave functions.
The coefficients of the expansion are denoted by gk(R) where k stands for the set of
quantum numbers which determine a four-particle state uniquely:

k = {(n~ l~ J l , n212 J2 ;Jp), (n3 13 J2, n, I, j4; jN)}.

¢ If the operator K has eigenvaluesequal to 1, we restrict the operator to the space orthogonal to
the corresponding eigenfunctions.
 o~-DECAY R A T E S 79

The operator/~ is then given by

K(II, 11') = ~_, gk(ll)gk(R'). (3.1)

k

The sum runs over all configurations k which contain at least one occupied level
of I~A>. Therefore the sum is not finite and R could not be calculated with the
required accuracy within a reasonable finite configuration space. For the
simplified model where the oscillator constants of [X~> and kbA> are set equal, (3.1)
is a direct expansion in eigenfunctions o f / ( and can be easily evaluated. Two examples
are given in ref. s).
(b) The operator R is written as a matrix in the space of functionsfi(R) defined by
fi(R) = n exp ( - 2fl(R- D,)2). (3.2)

The matrix elements are easily calculated because

4
f~(R,)x, = ]--I q h s ( r j - D , ) x ° ( 1, 2)X°( 3, 4), (3.3)
j=l

where qbls are harmonic oscillator functions of a 1S state and go° are singlet spin
functions for protons and neutrons respectively.
Because of (3.3), a matrix element of 1 - R turns out to be just the overlap of
two Slater determinants

<fill-I~lfj> = <dX~,(D,)~aldT.~,(Dj)¢a> with x~(D)) = fj(R~)x~. (3.4)

Furthermore the angular integration can be carried out analytically and one is left
with functions hi(R).
The only problem with this procedure is whether a choice of points D i can be
made so that the space {hi} is sufficient to describe/~. Because of the non-orthogonality
of the basis the distance between neighbouring points D i and Di+ 1 cannot be chosen
smaller than a certain AD which is determined by the number of relevant digits in the
numerical calculation. With sixteen relevant digits and for equidistant points D i,
D can be choosen as small as 0.4 fm. On the other hand with D = 0.4 fm the set of
functions {hi) can be used to describe smoothly varying functions and oscillating
functions with a maximum wave number kmax = 2rc/2AD ~ 8 fm-1. Obviously/~
is non-zero only if the wave function of the 0~-particle and the wave function of the
daughter nucleus 1@4> overlap. Therefore the spatial position of the ~-particle
must be in the region of non-zero I4',t> and it must not move too fast. This means that
the important regions are given by R, < 1.2A ~ fm and k~ < 4kF = 5.4 fm-1
Therefore the space described above is sufficient to determine K with reasonable
accuracy. Numerically the convergence of the results could be shown for AD tending
towards 0.4 fm. The relative error in the result for K is then 10 -4 or smaller.
It is clear from this discusssion that for sufficiently large R the function g2r of
eq. (2.16) must tend towards liE. Therefore the question is now, whether R 0 can be
80 T. FLIESSBACH A N D H. J. M A N G

chosen in this "asymptotic" region and at the same time be small enough to allow
the approximation of I~A+4) by a shell-model wave function. The answer to this
question was implicitely given by many authors (including one of us H.J.M.) in
many papers and was yes. In the following section we shall demonstrate that in the
light of the numerical calculations for the decays of 212Po and 21°Po the answer
must be changed to no.

4. Applications and conclusions

The two 0t-transitions
210~ 206 4
8¢V0126 --* s 2 P b 1 2 ¢ q - 2 H e 2 ,
212~ 208 4
aerO~28 -+ a2Pb,26+2He2,

were chosen to test the theory, because these are the cases in which one can hope
to approximate the wave functions reasonably well by simple shell-model wave
functions. The configuration of the protons in Po was taken as (lh~)2+ and the con-
figuration of the neutrons was (2g~)02+ for 212p0 and (3P~)o 2 for 2°6pb.
The oscillator constants are ~ = 0.17 fm -2 for 1~,4+4~ and I~a) in the region
around 2°apb, and fl = 0.47 fm -2 for [Z~).
In figs. 1 and 2 the reduced amplitudes G(R) and GN(R) are shown. In a very simple
physical picture one writes the decay constant as

2 = p=vP, (4.1)

where p~ is the at-particle density, v the velocity of the ~-particle, and P the barrier
penetrability. Here we identify p~ with R2[GNI 2 while earlier p~ was identified with
R21GI 2. That the latter identification is incorrect has been already indicated and is
discussed in detail in ref. s). Of course, we shall show that eq. (2.5) together with
eq. (2.18) leads to an expression for the decay constant which is essentially the
same as eq. (4.1). As a first result we read offfigs. 1 and 2: On the basis of pure shell-
model wave functions the ~-particle density p~ = R E ] G N [ 2 is larger than usually
thought (when p~ = R21G[2) by a factor of 300 to 400 on the average. The factor is
approximately 100 at the nuclear surface (R ~ 7 fm) where p= has its maximum.
The final problem to be discussed is the choice of a radius Ro. When such a
choice has been made the decay constant can be calculated. We first note that in
the surface region of the nucleus GN and also G, decrease like e x p ( - ? R 2 ) . This
means that the effective potential felt by the ~-particle when emerging from the nucleus
rises drastically in the nuclear surface. Such an effective potential can be calculated
from G~

- E . -- RG,. (4.2/
 I t J I | J ~ J I I

----R-G.IO 2i2po ~ a +2°ePb

(~hg/z)o2 (3p1,,2)~
R.G N A 21°Po ~ ~ +2°6pb - - - - RG. IO (Ih912)(~ (29112)(~

L=0 RGN L=O

T
E
% 'E R,
% C~

0 5 0 5
R (fm) R (fro)
Fig. 1. The functions G and GN ('or 21°Po --> z°6Pb+c¢. Fig. 2. The functions G and GN for 212Po --~ z°aPb+c(.
82 T. FLIESSBACH AND H. J. MANG

i I I I I I I
30 R GN(212po) Igo's best fit
(Woods -Saxon )
Igo's exponenti¢ll Veff(GN)
potentiol

20

/'\.
>
®I0
. -- //

-I0
1 I I I I I
6 T 8 I0 II 12 13 14
R (fro)
Fig. 3. The effective potential Veff(GN) as defined in eq. (4.2) is compared to two of Igo's potentials
which fit the scattering data on 2°Spb.

A n example is given in fig. 3. This potential rises rapidly f r o m RA+ ,. F o r a realistic

s-nucleus potential, however, such a rapid rise occurs only at R = R a + R which is
substantially larger than R a + , as can be seen in fig. 3.
We cannot but conclude that R0, the channel radius, must be chosen smaller than
7.0 fm in the cases considered here. F r o m 6.2 fm to 7.0 fm eq. (4.2) yields an
effective potential which is approximately equal - 20 MeV. Below 6.2 fm Veff looses
its meaning at the zeros o f GN where it becomes singular• The reason is that GN
is not really solution o f a Schr6dinger equation with a local potential. Because f2E,
on the other hand, is taken as the solution o f just such an equation it can only be
joined to GN in the region mentioned above, i.e. for values o f R 0 between 6.2 and
7.0 fm. Then the value o f the decay constant will be independent o f the exact
choice o f Ro within these limits.
O n the basis o f these assumptions we can n o w proceed to a calculation o f
the absolute value o f the decay constant. According to eqs. (2.5) and (2.18) the decay
constant 2 is given by t
2 -- 2zr h 2 R 2 GN t912~ OGN f2 E . (4.3)
h 2M~ t3R tgR ~=Ro
* The integration on angles has been performed in eq. (4.3), and hence Gn and "Qea depend
only on R[see ref. 5)].
 or-DECAY RATES 83

As usual we define a reduced width by

2 = -1p62, (4.4)
h
where P is the barrier penetrability. The WKB approximation for P yields

P = exp -2 ,
(4.5)

q= ~ VR-E ,

where R1 and R 2 a r e inner and outer turning point respectively.

For values of R < R 1 the WKB approximation gives t

ROE(R) = (2~h2 p) ½ 1 i6) (4.6)

with
K(R) = [2M~
m h2 ( E - V ( R )
)]~, (4.7)

where V(R) is the a-nucleus potential. Upon inserting (4.6) in (4.3) one obtains for the
reduced width

62- h2 ll-iR°K(R°)12 R~°GN+ R~° dGN2=Ro'R (4.8)

2M~ R 2 2R o K(Ro) 1 - iR o K ~R
Because RoK(Ro) >> 1, formula (4.8) can be further simplified.

,:o
2M~ Ro2 ~ - (R° GN)2+ - - " (4.9)
K2(Ro) \ aR l J
Finally we must check whether the reduced width is indeed approximately constant
for 6.2 f m < Ro < 7 fm and how it agrees with experimental data. In an exact
theory, of course, the result of the calculation should be completely independent of
the choice of R0. Here we can only hope to have this independence in the limited
region indicated above, because only there GN and OE can be considered as solutions
of a Schr6dinger equation with the same potential. Unfortunately this potential
is not determined by scattering data, because they are insensitive to the value of the
potential for distances smaller than 8.5 fro. We therefore decided to use the value
which follows from eq. (4.2) and the calculated GN. This prescription gives K(Ro) = 2
fm-1, and with this value our results are indeed nearly constant for a channel radius
R o in the range 6.2 fm < Ro -< 7.0 fro, the variation being less than 10%.
t Normalization corresponds to J(Ro)/j(oo)= P and Jf2E*(R)~E,(R)R2dR = tS(E--E') where j is
the current.
 84 T. FLIESSBACH AND H. J. MANG

Besides eq. (4.9) we also used a formula which expresses the reduced width as
a product of a single-particle width 62p. [ref. 6)] and a spectroscopic factor S.
In our case S is given by

S = J daR [GN(R)I2. (4.10)

(In earlier calculations G was used instead of GN.) The width t52.p,is defined in
ref. 6). We evaluated the expression given there but inserted a single-particle wave
function (single-~) which had the same number of nodes as GN (or G). The number of
nodes is ten for 2t°po and eleven for 21Zpo.

TABLE 1
6 2 in units of 10 -2 MeV

~2 from
Mang's formula ~2 = ~52s.p.S c52= F,xp/P

Calculated from GN
2t2po 0.7 1.6 2 . 9 . . . 15.6
2t°po 0.8 1.2 0.4... 4

Calculated from G
212p0 0.006 0.004 2 . 9 . . . 15.6
2t°po 0.004 0.0035 0.4... 4

In order to compute a reduced width from the measured lifetime we calculated

penetrabilities from two different ~-nucleus potentials. One was Igo's best fit 1)
to scattering of oc-particles on lead and the other Igo's exponential potential 1).
Experimental and calculated results are compared in table 1. Clearly we can reproduce
the oc-width of 210po even without configuration mixing. On the other hand it comes
as no surprise that the width of 212p0 cannot be reproduced without admixing
other shell-model configurations. From the fact tbat 212p0 is just doubly magic 2°spb
plus an ~-particle, one would expect that quite a substantial admixing of bigher con-
figurations is needed to explain the ~-width. This problem is under investigation.

We acknowledge useful discussions with J. O. Rasmussen and N. K. Glendenning.

One of us (T.F.) thanks the "Deutsche Forschungsgemeinschaft" for a grant and the
Nuclear Chemistry Division of the Lawrence Berkeley Laboratory for the hospitality
extended to him.

References
1) H . J . Mang, Ann. Rev. Nucl. Sci. 14 (1964) 1
2) H. Casimir, Physica 1 (1934) 193
 g-DECAY RATES 85

3) V. I. Furman, S. Holan, S. G. Kadmensky and G. Stratan, Nucl. Phys. A2,26 (1974) 131;
K. Harada and E. A. Rauscher, Phys. Rev. 169 (1968) 818;
S. G. Kadmensky and V. E. Kalechits, Yad. Fiz. 12 (1970) 70
4) H. Feshbach, Documents on modern physics: Reaction dynamics (Gordon and Breach, New
York, 1973)
5) T. Fliessbach, Z. Phys. A272 (1975) 39
6) R. G. Thomas, Prog. Theor. Phys. 12 (1954) 253
7) G. Igo, Phys. Roy. 115 (1959) 1665