PH YS ICAL REVI EW VOLUME 77, NUMBER 1 JANUARY 1, 1950

The Influence of Nuclear Structure on the Hyperfine Structure of Heavy Elements

AAGE BOHR
Department of Physics, Columbia University, Negro Fork, Ne2o Fork*
AND

V. F. WEIssKQPP
Department of Physics and Laboratory for NucLear Science and Engineering,
Massachusetts Institute of Technology, Cambrljge, Massachusetts
{Received September 27, 1949)

The influence on the h. f.s. of the 6nite size of the nucleus is considered and the effect is calculated for
simple models of the nuclear magnetism. It is pointed out that the distribution of magnetic dipole density
over the nuclear volume may vary greatly from nucleus to nucleus depending on the relative contributions
of spin and orbital magnetic moments to the total nuclear moment. On this basis an attempt is made to
interpret the observed discrepancy between the h. f.s. ratio of the Rb isotopes and the ratio of the magnetic
moments as determined by the magnetic resonance method. A study of such anomalies may give some
information regarding the structure of nuclear moments, in particular, regarding the nuclear gl. -factor.

r. INTRODUCTION tion, the electron density varies approximately as

1— ZR'/apRp, where Rp is the nuclear radius.
RECENT accurate determination' of the nuclear
moments of the Rb isotopes by the magnetic In a model in which the nuclear magnetic moment is
resonance method has indicated that the ratio of the considered as a smeared-out dipole distribution, the
h. f.s. splittings in Rb'~ and Rb'7, measured previously h. f.s. would thus be expected to differ from the value
with great precision, ' does not agree exactly with the calculated for a point dipole at the nuclear center by a
value calculated from the ratio of the moments, if the factor 1+a, where
nuclei are considered as point dipoles. The h. f.s. ratio p = —(ZRp/ap) (R'/Rp') A, . (1)
is found to be larger by 0.33 percent, while the experi-
mental uncertainty involved in the comparison is For heavy atoms, relativity becomes of importance and
judged to be about 0.05 percent. its main effect in the present connection is to increase
It has been pointed out by Bitter' that anomalies the absolute magnitude of the electron density at the
of this order may be expected if the nuclear magnetic nucleus by a factor of about (ap/2ZRp)'&' », where
moments are represented of
p= (1 Z'n') &— and n is the 6ne structure constant. The
by some distribution
magnetism over the nuclear volume rather than by a
total h. f.s. is increased in corresponding measure and,
thus,
point dipole. Since effects of this type might offer
information regarding the structure of nuclear moments, p= —(ZRp/up)(u /2ZR )'&' &'(R'/R ') —
(2)
we shall attempt a somewhat more detailed analysis of
The assumption of a uniform distribution of the
the problem.
nuclear charge restricts our considerations to nuclei
containing a large number of protons. For the lightest
II. QUALITATIVE CONSIDERATIONS
elements the dependence of the electronic wave func-
In the non-relativistic approximation, the h. f.s. of an tion on the position of the individual protons must be
s-state is proportional to the average electron density taken into account. 4
at the location of the nuclear magnetic moment. It is If the magnetic moment is uniformly distributed
here assumed that the moment distribution is spheri- over the nucleus, we have (R'/RpP) A, =-,P and, by putting
cally symmetric, but even large angular asymmetries "
R0=1.5X10 cm A&, values for e are obtained which
have only a minor effect. At a small distance R from a for light elements are of the order of 0.01 of a percent
central charge Ze, the electron density is propor- and which become quite appreciable, of the order of
tional to 1 — 2ZR/ap where ap is the radius of the hydro- several percent, for the heaviest elements.
gen atom. Inside the nucleus itself, however, the wave For p, d. states the h. f.s. anomaly is negligible in
function decreases somewhat more slowly with dis- the non-relativistic approximation, in which the corre-
tance. Thus, in a nucleus of uniform charge distribu- sponding wave functions vanish at the center. In heavy
elements, however, due to relativity effects, the value
* On leave from Institute for Theoretical Physics, Copenhagen, of p may become appreciable for pt-states. For these
Denmark. states the small components have the character of
' F. Bitter, Phys. Rev. 75, 1326 {1949).
' S. s-wave functions and determine the density at the
Millmann and P. Kusch, Phys. Rev. 58, 438 {1940).
3 F. Bitter, Phys. Rev. 76, 150 (1949); see also H. Kopfermann, center. The value of e will therefore be of the order of
Eernmomente {Akademische Verlagsgesellschaft, Leipzig, 1940),
p. 17. ' A. Bohr. Phys. Rev. 73, 1109 {1948).
94
 H YPERF I NE STRUCTURE OF HEAVY ELEMENTS
Z'o. ' of that for an S~-state. For an electron of higher angular momentum s. Denoting the spin g-factor by
total angular momentum than ~k, the inHuence on the g„ the vector potential produced by the nucleus may be
h. f.s. of the finite size of the nucleus may be disregarded. written
In the particular case of Rb, one obtains from (2) a 1
value for e of about 0.4 percent. Whereas effects of this A, (r) = —~" drgw(R)g, sXV, (3)
~

/r —Rf& ~,
magnitude are thus to be expected in the h. f.s. of each where
isotope, it might at first appear surprising that the
ratio of the h. f.s. shows an anomaly of the same order. t
draw(R) = 1.
In fact, it becomes necessary to assume an essentially
different distribution of magnetism in the two nuclei. The magnetic interaction of the nucleus with an
It appears that a natural explanation of such varia- atomic electron is given by eaA(r) and, considering the
tions is possible if one considers the nuclear magnetic interval rules, it is necessary only to evaluate the
moment as composed of two intrinsically different diagonal matrix element of this operator for the state
parts, a spin moment and an orbital moment. In fact, in which the electronic, as well as the nuclear, angular
the latter part, originating from currents in the nucleus, momentum has its maximum component in the s
will in general be equivalent to a magnetic dipole direction. Denoting the electron part of the wave
distribution which increases toward the center, and function by f, one finds that,
should therefore produce smaller anomalies in the h. f.s.
than the former, the spin part. For example, a rotating —
charged sphere is equivalent to a magnetization in-
lV = dr, f*e nA(r) P
creasing toward the center and would give rise to a
value of e only about half that corresponding to a = &2e~ dr„F(r)G(r) (AXr, ), (—
4)
sphere of uniform magnetization. Depending on the
proportion of spin magnetic moment and orbital
magnetic moment in the nucleus, considerable varia-
where Ii and 6 represent the two radial wave functions.
tions in the h. f.s. anomalies may thus occur. Indeed, a
The upper and lower sign refer to sg and pi states,

nucleus in which the spin moment is directed oppositely respectively.

to that of the total nuclear magnetic moment might Introducing (3) into (4), it is convenient to write
even have a larger h. f.s. than that corresponding to a 16m
point dipole. W. = & eg, ~~ drsw(R)q, (5)
In the case of Rb, the values of the angular momenta 3
and the magnetic moments actually indicate a very
different alignment of spin and orbital momentum in the If the distribution w(R) is spherically symmetric, q,
two isotopes, and therefore an essentially different dis- takes the simple form
tribution of magnetism. Indeed, while the magnetic
moment of Rb" (I=-,', +=2.75) appears primarily due q, =s, " FGdr.
to the spin moment of the odd proton, the moment of R
Rb" (I=5/2, @=1.35) would seem to be largely of
orbital type with only a small, and perhaps even nega- In the case of angular asymmetries, one must add to (6)
tive, spin contribution. On the basis of such considera-
tions, it thus appears possible to understand the com-
paratively large decrease in the h. f.s. of Rb" relative
's,
—,
3XZ
+-,
3I'Z
's„+-, 's,
3Z' —R'
&o
r
~ r'
FGdr, (7)—
to that of Rb".
It is of interest that no effect of comparable magni- but since the effect of such asymmetries for not too
tude has been found in the cases of Ga' or Tl' where the extreme models is of only minor inHuence, we shall in
two isotopes have the same spins and comparable the following use the simple expression (6) for q, .
moments and, therefore, presumably, similar distribu- Defining
tions of magnetism.
FGdr/ I FGdr, (8)
III. H.F.S. DUE TO SPIN MAGNETIC MOMENT 0 ~0
In the more quantitative considerations, we shall it follows that W, is decreased by the relative amount
treat separately the cases of spin and orbital moment.
(ag)4lf as a consequence of the deviation of the nuclear
In the former case we represent the nucleus by a dis-
magnetization from a point dipole.
tribution of magnetic moment, given by a density
function w(R), and having the direction of the spin IV. H. F.S. DUE TO ORBITAL MOMENT
' R. B. Pound, Phys. Rev. 73, 1112 (1948). It may erst be noted that the magnetic 6eld of a
' H. Poss, Phys. Rev. 75, 8$ (1949). current density distribution i, for which div i=0, is
 A. BOB R AN D V. F. WEISSKOPF
TABLE I. Values of b for sy- and py-states. powers of x, one finds, for an s;-state

10 0.08 percent
b(s))
G= k~ 1 —3—y'x'+ —
1
y'x"+ ~
= Ze'/hc
y—
20 0.20 8 10
30 0.38 0.02 percent (14)
40 0.68 0.05
50 1.12 0.13
60 1.71 0.29 I5 4O) 1O
70 2.52 0.60
80 3.58 1.16
and, for a pt-state
90 4.80 2. 11
3
F= k 1 ——p'x'+ —
1
y'x'+
equivalent to that produced by a magnetic dipole i 8 10 )
density p given by i= — t," rot@, and thus the problem
of an orbital moment is reducible to that considered
G= —-', kyx~ 1+—
( 4Ro~&
in the preceding paragraph.
3 y5
Still, in many cases, it is more convenient to con-
sider directly the interaction of a moving nuclear (1 9 i x-"+—
+—
1
particle with the atomic electron. Denoting by Ze and &' &'-x"+ (15)
M the charge and mass of the particle, and its wave 45 40 ) 10
I
)),
function by q(R), we may use the expression
where k is a constant depending on the normalization
Ze f of the entire wave function. These approximate expres-
gc(r) = dry(p~(R) Py(R) (9) sions have an accuracy of about one percent, even for
fr —Rf
I

Mc Z-values corresponding to the heaviest elements.

The integrals in the numerators of (8) and (12) may
and, introducing in (4) one finds t.hat Wr, can be ex- now be expressed in terms of powers of E.. The main
pressed in the form term is proportional to R and since the higher terms
never amount to more than ten percent of the leading
16'
iVL= + eg&) drew(R)qL, (10) term, one may conveniently replace these terms by
3 appropriate multiples of R'. If values are chosen inter-
mediate between those corresponding to a uniform
where w(R) = q(R) -'and
~ ~
gr, =Ze/2Mc. Furthermore, distribution w(R) and those corresponding to a surface
distribution, the error involved will, for the most plausi-
pR
—Ii Gdr
oct
ble models, only amount to a few percent. In this
q'I, = I.g I I'Gdr+ l

manner, one obtains

~. R

where L is the orbital angular momentum of the FGdr=0. 23k'R, y(1 —0.2y')
particle. Thus, if
R
f
1 ——
r3
g
~FGdr/
(ao
FGd. , (12) X —(- 1) ]
( 1+1.44Rome)
J
[ (16)
~

R') &g

pR g
the contribution to the h. f.s. of the orbital angular FG ) 1 — —dr = 0.62
p8
'I FGdr,
momentum is decreased by a rels, tive amount (eI.)A, . R'i )
J
V. EVALUATION OF ELECTRONIC %'AVE FUNCTIONS where the factor in square brackets is to be included
for p;-states only.
In order to calculate a, and ~1., it is necessary to Outside the nucleus the electron moves in a Coulomb
evaluate the electronic wave functions in the interior field screened at larger distances by the other atomic
of the nucleus, in which region they deviate signifi- electrons. However, we need consider only the un-
cantly from the wave functions corresponding to a screened part of the field, since the h. f.s. interaction
point nucleus. Representing the nucleus by a homo- takes place primarily in this central part of the atom.
geneously charged sphere of radius Ro, the potential The integral in the denominators of (8) and (12) may
in this region is given by
thus be expressed in terms of the well-known solutions
V= (-', ——,'x')Ze/Ro = r/Ro(1.
x— (13) to the wave equation for an unscreened Coulomb field,
normalized relatively to the wave functions in the
Solving the radial wave equations by an expansion in interior of the nucleus by the boundary conditions at
 HYPERFINE STRUCTURE OF HEAVY ELEMENTS
the nuclear surface. ' One inds approximately It is, of course, diKcult to estimate the mean values
of (R/Ro)' entering into the expressions for (x, )A, and

,'7-
5 )2ROZy '&'-» (Kr, )A, . It, may be noted that for a uniform distribution
~Gdr= I(2P —1) lI' ', but one might
4p mc(u, ) c(4u' —1) over the sphere, the mean value equals —,
mell imagine a tendency of the unpaired particles which
, (1 1~2)2 (s)) (17) contribute to the nuclear moment to stay near the
surface, in which case the mean value would be some-
X~'
( —1) (1 —
what larger. Since, however, it cannot exceed unity,
o 727')' (P:)
37' one may tentatively assume a value for (R'/R02)A, of
about ~5. On this assumption one obtains, from (18)
—
where p= (1 y')&. The use of the wave functions for and (21),
an unscreened field in the evaluation of (17) is always
well justified for s-states, and is also valid for p~-states gegc ( 1 1
in heavy atoms. ~=0.3b (22)
It follows that the values of ~, and x~ are independent g, (1)
gr Egr— gr(2))
of k and, consequently, of the particular st- or pt-state
but it need not be emphasized that the approximation
under consideration. One may conveniently write
may be rather crude.
a, = bR'/R '
(18) VII. DISCUSSION AND COMPARISON WITH
Kr,
——0.62bR'/R02
EXPERIMENTAL DATA
the coefficients only on Z and Ro.
where
For 80=1.5&10 " b depend
cm&(A&, the values of b for s~- and
The expression (22) for 6 involves the spin and
orbital g-factors, the values of which must be expected
pt-states are given in Table I.
to depend on the type of nuclei. The large majority of
VI. COMBINED EFFECT OF SPIN AND nuclei having angular momenta contain an odd number
ORBITAL MOMENT of nucleons. In this case it is most often assumed that
If the total nuclear magnetic moment is composed the spin g-factor equals that of the odd particle, i.e. ,
of a spin part and an orbital part, the relative change in g, = g (proton) for Z odd and g, = g (neutron) for Z even.
the total h. f.s. splitting may be written The choice of gl, , however, is more uncertain and we
shall consider two possible assumptions regarding the
&s Avns &L Avnr y (19) origin of the orbital momentum.
where n, and nJ. represent the fractions of the nuclear On the one hand, we may assume with Margenau
moment due to spin moment and orbital moment, re- and %igners that the nuclear matter as a whole is in-
spectively. These quantities may be expressed in terms volved in the orbital momentum. This leads in a plausible
of the g-factors: manner to gr, =Z/A, in units of e/2Mc, M being the
nucleon mass.
g g~
—gI- On the other hand, Schmidt' has tried to account for
ns= nl =1 ns~ (20) nuclear moments by ascribing the orbital momentum
gr gs —gr. to the motion of the odd particle in the nucleus. This
where g~ is the total nuclear g-factor. leads to gr, = 1 for Z odd and g1, =0 for Z even. It may
The experiments determine most easily the diBerence be added that these g-values do not necessarily imply a
of the e-values for two isotopes. Of course, the values of single particle model of the nuclear moment. Any model
(K ) and (xr, ) may vary somewhat from isotope to
A A
in which, for Z odd, only protons, and for Z even, only
isotope, but, it appears that larger eGects may be neutrons, contribute to the orbital momentum leads to
expected due to difkrences in n, and nl, . If we neglect the same g-values and is therefore equivalent for our
the fiuctuations in (x,)A„and (~c) A„and if, moreover, the purpose.
values of g, and gJ. are the same for the two isotopes, as The two assumptions regarding gi, lead to appreciably
seems a plausible assumption for isotopes diGering by diGerent values for 6, and the phenomenon in question
an even number of neutrons, one obtains the expression: might thus oGer some evidence regarding the nature of
the orbital momentum.
6=—e(i) —e(2) The value of d has as yet been measured only for the
= ((~.)A —(«) )
ggl I
1
—1 (21) Rb isotopes (Z=37, A =85 and 87), for which has been
g, —
A (
)
gr &gr(1) gr(2)) found 6= 0.33+0.05 percent for the ground state, which
is an s~ term. Expression (22) gives 6 = 0. 11 percent for
for the inhuence of the 6nite size of the nucleus on the
gi. =Z/A=0. 43 and 6=0.29 percent for gi, =1. The
h. f.s. ratio of isotopes 1 and 2.
~See G. Racah, Nuovo Cimento 8, 178 (1931); also J. E. ' H. Margenau and E. Wigner, Phys, Rev. 58, 103 (1940).
Rosenthal and G. Breit, Phys. Rev. 41, 459 (1932). 9 Th. Schmidt, Zeits. f. Physik 106, 358 (1937}.
 JENTSCHKE
latter value is in good agreement with the empirical Measurements of the h. f.s. anomalies in other ele-
value while the estimate for gz, =Z/A is too small by a ments would be desirable. There exist a number of odd
factor three. Although the approximations involved in elements (e.g. , Sb, Eu, Ir) having isotopes with widely
this estimate are somewhat crude, it seems diKcult to different gg-values and for which the value of 6 ac-
obtain a suKciently large value of 6 for gl. =Z/A, cording to (22) should be appreciable. In the case of
even under rather extreme assumptions regarding the even elements with isotopes of odd atomic number (22)
values of (N, )A„and (zr, )A„ in the two isotopes. gives a vanishing 6 for the Schmidt model, whereas for
In this connection it may be mentioned that the gl, =Z/A, values of 6 of more than one percent would
Schmidt model is also favored by other evidence re- be expected in several cases (e.g. , Yb and Hg).
garding nuclear magnetic moments. In fact, as is well A number of elements (e.g. , Na and K) have isotopes
known, the gr-values of all odd nuclei fall within the of odd numbers of protons and neutrons whose spin
limits given by the Schmidt model, whereas the and h. f.s. have been measured. These nuclei are of
gI-values of a number of nuclei with high spin fall out- special interest for the problem of nuclear structure,
side the limits predicted by the model in which gr, = Z/A. and a study of their h. f.s. anomalies might give some
Moreover, reference may be made to the recent suc- indication regarding the composition of their moments.
cesses achieved by the individual particle model of This work was assisted by the Ernest Kempton
nuclear structure in accounting for the angular momenta Adams Fund for Physical Research of Columbia Uni-
of nuclei. versity and by the Joint Program of the ONR and AEC.

P IS YSI CAL R EVI EW VOLUME 77, NUMBER 1 JANUARY 1, 1950

On the a —&- and a —X-Branching of the Heaviest Natural and Arti6cial

Radioactive Substances*
%. JENTSCHKE
(Received June 24, 1949)

Based on simple empirical regularities in e-decay properties of the heavy elements, the following sub-
stances which show P-emission or E-capture should also be a-radioactive with e— P- or a —E-branching ratios
of at least 10 'o: Ms Thy, Ms Tht, AcK; seRa~, 9&Pa233, »Np234, »Np'I', »Np23e»Npms»Np23a 94Pu"' »Am'42.

I. INTRODUCTION the newly established 4n+1-series s3Bi"' suffers an

N addition to the long known dual disintegration of a—P-decay. '
- the C-bodies of the three natural radioactive disin-
II. EMPIRICAL REGULARITIES OF THE 6-DISINTE-
tegration series, the existence of a number of other GRATION OF THE HEAVIEST NUCLEI
double disintegrations has been established. M. Percy'
succeeded in proving that actinium emits a weak
In the range of O.-instability of the heaviest nuclei all
elements emitting P-rays or showing K-capture must
O.-radiation. This disintegration creates element 87
also be o.-radioactive. Based on simple empirical regu-
(francium). B. Karlik's and T. Bernert's' proof of the
larities, therefore, branching ratios for some of these
dual disintegration of RaA made it possible to establish
radioactive elements which have not as yet been
the occurrence in the natural disintegration series of
measured are stated below. However, only dual disin-
element 85 (astasinm) through the isotope 8+t"'.** "
tegrations with branching-ratios of 10 or higher will
Prior to this, element 85 has already been arti6cially
obtained and chemically investigated through reaction
be mentioned.
»Bi'"(a, 2N)»At'" by D. R. Corson, K. R. McKenzie, If the disintegration energies of the natural o.-radio-
and E. Segre. ' This element shows 0.—E-branching. In
active elements are plotted in a diagram as function of
their mass numbers~ (Fig. 1), a family of curves is
* This paper was written in August, 1947. formed by lines connecting nuclei of the same atomic
' M. Percy, J. de phys. et rad. 10, 435 (1939). M. Percy and
Lecoin, J. de phys. et rad. 10, 439 (1939). 4Hagemann, Katzin, Studier, Ghiorso, and Seaborg, Phys.
~ B. Karlik and T. Bernert, Zeits. f. Physik
123, 51 (1944). Rev, 72, 252 (1947); English, Cranshaw, Demers, Harvey,
**In the case of ThA and AcA, also, long range a-particles had Hincks, Jelley, and May, Phys. Rev. 72, 253 (1947).
'
been observed. The simplest explanation would be a dual decay J. Schintlmeister, Oesterreichische Chemiker Zeitung Nr. 17,
of ThA and AcA. But applying energy-conservation rules it 1938, Nr. 9/12, 1943. Regularities in Geiger Nuttal diagrams
follows that there exist considerable difBculties in accepting this referring to the individual values of Z were pointed out by
interpretation. Berthelot in the J. de phys. et rad. , serie VIII, 3, 17 (1942) and
' Corson, McKenzie, and Segrh, Phys. Rev. 57, 1087 {1940). by N. Feather, Proc. Roy. Soc. Edinburgh, 62, 211 (1946).