HAL Id: jpa-00208122
https://hal.archives-ouvertes.fr/jpa-00208122
Submitted on 1 Jan 1974
HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
Hartree-Fock evaluations of specific-mass isotope shifts
J. Bauche
To cite this version:
J. Bauche. Hartree-Fock evaluations of specific-mass isotope shifts. Journal de Physique, 1974, 35
(1), pp.19-26. �10.1051/jphys:0197400350101900�. �jpa-00208122�
HARTREE-FOCK EVALUATIONS
OF SPECIFIC-MASS ISOTOPE SHIFTS
J. BAUCHE
Laboratoire Aimé
Cotton,
CNRSII,
Bâtiment505,
91405Orsay,
France(Reçu
le 23juillet 1973)
Résumé. - A la suite d’un travail
précédent
sur la série 3d, desdéplacements spécifiques
demasse sont
calculés,
par la méthode de Hartree-Fock nonrelativiste,
dans des atomes des séries4d, 4f,
5d et5f,
pour les raies de résonance des alcalins et dansquelques
spectres à deux et trois électrons. Dans les atomes lourds,quand
ondispose
de résultatsexpérimentaux
sur les spectres X d’atomesmuoniques,
on compare directementexpérience
et résultats de Hartree-Fock pour les raiesoptiques
àgrand déplacement spécifique.
Dans les autres cas, en utilisant desdiagrammes
de
King,
on obtient les valeursexpérimentales
de certaines combinaisons linéaires dedéplacements spécifiques
de deuxraies,
que l’on compare aux résultats de Hartree-Fock.Dans toutes les séries
longues nlN,
valeursexpérimentales
etthéoriques
diffèrent auplus
d’unfacteur deux. Le résultat essentiel de la
théorie,
à savoir laprédiction
que les transitions corres-pondant
au saut d’un électron nl ont de loin lesplus grands déplacements
de masse, est ainsi confirmé.Abstract. -
Following
aprevious
paper on the 3dseries, specific-mass isotope
shifts are comput- ed, by means of the non-relativistic Hartree-Fockmethod,
in atoms of the 4d, 4f, 5d and 5fseries,
in the alkalis and in a few two- and three-electron spectra. In
heavy
atoms, whenisotopic
measure-ments of muonic X-ray transition lines are
available,
directcomparisons
withexperiment
arepresented
foroptical
transitions withlarge specific
shifts. In the other cases,through
the use ofKing diagrams,
thecomparison
is made between the Hartree-Fock(HF)
andexperimental
valuesof some known linear combination of the
specific
shifts of two lines.In all nlN
long
series, the HF andexperimental
values agree within a factor of two. Thisgives
confidence in the main HF result, i. e. the
prediction
that the lines with thejump
of an nl electron have by far thelargest
mass shifts.Classification
Physics Abstracts
5.231 1
1. Introduction. - Measurements on the
isotope
shifts of samarium
by Striganov,
Katulin and Eliseev[1 ]
and their
interpretation by King [2]
renewed tenyears ago the interest in the atomic mass
isotope
shift.
Through
these results it was known thatspecific
mass
isotope
shift may be ofappreciable
influenceeven in the
spectra
of atoms with atomic number Zlarger
than 60.Consequently,
forpeople
interestedin field
isotope shift,
the ab initio evaluation of thespecific
massisotope
shift is ofimportance,
becauseits value can then be subtracted from the
experimental
result to isolate the field shift contribution.
Except
for verylight
atoms,specific
mass shiftshave
generally
been evaluatedthrough
the use ofthe Hartree-Fock
(HF)
method. In1970,
Bauche and Crubellier[3]
havepublished
extensive results obtained in that way in the arcspectra
of the first series of transition metals(the
3dseries).
In thisseries,
thefield shift is of
relatively
smallimportance
and canbe considered as a correction. So the
comparisons
between
experiment
andtheory
for thespecific
shiftare safe when the
isotope
shifts arelarge.
For thetransitions with the
largest shifts,
it appears thatexperimental
and theoretical values of thespecific
shifts agree to about 30
%.
Encouraged by
thisresult,
we haveproceeded
toelements with
larger
values of Z. This paperpresents, first,
results for the first series of rare earths(the
4fseries).
Then the other threelong
series(4d,
5d and5f)
are
studied,
each seriesthrough
theexample
of oneelement.
Eventually
wepresent
evaluations for the first resonance lines of the alkalis and for miscella-neous one-, two- and three-electron
spectra.
Whereverpossible, comparisons
betweenexperimental
and theo-retical results are
given. Unfortunately,
it is notknown in which way the HF results are affected
by
the
relativity phenomenon.
2.
Principles.
- We first recall thegeneral
proce- dureapplied.
The totalspecific
shift 6 of a stateis,
to first order of
perturbation,
theexpectation
value of thespecific
shiftoperator
Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:0197400350101900
20
where pi is the momentum of electron i and M the
mass of the nucleus.
Knowing
the valuesof ù
forlevels A and B and nuclear masses
Mi
andM2 (M2
>MI),
thespecific
shift DQ for transition B ---> A(with
the usualsign conventions)
iswhere
ul,(A)
=6(A, M2) - u(A, Ml)
is thespecific
shift for level A.
From now on we deal with atomic wavefunctions of the
configurational angular type (e.
g. the Hartree- Fock(HF) wavefunctions).
Aprevious
paper[3]
recalls the way to build the formal
expression
of rin terms of the radial
integrals
For a state in pure
LS coupling,
thisformal
expres- sion does notdepend
on thequantum
numbers J andMJ
of the state[4]. Furthermore,
it does notdepend
on itsintermediate-coupling expansion
in the(pure) configurations
where noG’
Slaterintegral
ispresent
in the formalexpressions
for the relative electrostaticenergies
of the terms[5]. Examples
ofthis last case are
dN s2, dN+ 1
s,fN S2, pN and,
of course, the so-called one-electronconfigurations.
For any
monoconfigurational
atomic LSstate,
the HF methodyields
radial functionsR(n, 1)
for thevarious subshells. From these
functions,
it isstraight-
forward to
compute
the Jintegrals
and then -u. Wehave used the very convenient HF
computer
code of Froese[6],
whichyields
the radial functions in nume-rical form.
Apart
fromsimple
numericaldata,
therequired input
consists in the formalexpansion
ofthe electrostatic energy, in terms of Slater
integrals,
for the considered atomic LS term.
3. HF results. - All the results
presented
hereafterrefer to atomic pure LS terms.
Tables
I,
II and III containrespectively
the results obtained in the arcspectra
of the 4fseries,
ofmolyb-
denum
(4d series),
osmium(5d)
andplutonium (5f)
and of the alkalis and miscellaneous one-, two- and three-electron
spectra.
In eachtable,
the first three columns indicaterespectively
theelements,
the chosenpairs
ofisotopes (mass
numbersA1
andA2,
withAA
= A 2 - A 1 = 2)
and for each element the LS termsconsidered ;
the last three columns containrespectively,
for each LS term, thecomputed
valueof the
quantity k
= -(M/m) u
in atomic energy units(notation
from Vinti[4] ;
m is the electronic mass ; k is apurely
electronicquantity,
notdepending
onnuclear
masses),
the shifted value of k when that of the first LS term considered for each element isarbitrarily
chosen to be zero, andeventually
thespecific
shiftQ12
in mK(1
mK =10-3 cm-l)
for thechosen
pair
ofisotopes (it
can be shown that this lastfigure
is theproduct
of thepreceding
oneby
thenumber 239
078/A1 A2).
In each
spectrum,
we have selected forcomputation
low
configurations
which arespectroscopically impor-
tant. In
configurations
with more than one LS term,we have
generally
selected forcomputation
at leastthe lowest term, or at least a term with the maximum
spin.
For
deciding
whether thespecific
shift in a line is«
large »
or « small », its value is oftencompared
tothe normal-mass shift value. In the tables
given,
thiscomparison
is mosteasily
obtainedby noting
thatthe normal mass shift in any line at = 5 000
Á
isequivalent
to a contribution close to 0.1 to the Akquantity
of the line(Ak being
the difference between the k values of the levelsinvolved).
3.1 THE 4f SERIES
(TABLE 1).
- In the results forthe 4f
series,
amajor phenomenon
appears : in eachspectrum,
the value of thespecific
shiftdepends essentially
on the number of 4f electronspresent
in theconfiguration.
Thisphenomenon
isquite analogous
to that observed on the results for the 3d series
[3].
The
larger
the number of 4felectrons,
thelarger
thespecific
shift of the term ; moreover, the increase inspecific
shift per 4f electron has almost the same value(30 mK,
for AA =2)
in any arcspectrum
of the series. In contrast, if twoconfigurations
differonly
in the number of6s, 6p and/or
5delectrons,
their
specific
shiftsonly
differby
a fewmK,
at most.3.2 THE
4d,
5d AND 5f SERIES(TABLE II).
- Foreach of the
long periods
other than 3d and4f,
welimit ourselves to the
study
of one element near themiddle of the series.
From the results of table II it appears at once that in these three
nlN
seriesagain
the Hartree-Fock value of thespecific
shift of aconfiguration depends
almostonly
on its number of nl electrons.3. 3 THE ALKALIS. MISCELLANEOUS SIMPLE SPECTRA
(TABLE III).
- In thealkalis,
wegive only
results for the first resonance lines np ---> ns. As from sodium to cesium there is alarge
variation in atomicweight,
it is better to look at the values of Ak
(fifth column)
than of 4W
(sixth column).
Then it is clear that thepredicted specific
shift is notlarges 1 Ak is
less than0.1,
as it is in the lowest np --+ ns transitions of thelong
series of tables 1 and II.In Table
III,
results on some miscellaneous confi-gurations
are alsopresented.
3.4 DETAILED CONTRIBUTIONS OF THE DIFFERENT SUBSHELLS
(TABLE IV).
- Togive
an idea of the way in which the different subshells of theconfigurations
contribute to the k values of tables 1 to
III,
wepresent
in tableIV,
for a fewtypical
cases, theseparate
contributions for the interactions inside the core(denoted core-core),
between the core and the differentexternal subshells
SI, S2,
S3(core-S1, S2, S3)
andbetween the external subshells
(external-external).
On this
table,
one can observe forexample,
in anynlN
séries : ,- the
major
influence of the nlelectrons, ,
- the
screening
effects on the external non-nl elec-trons,
- the relaxation of the core and the
self-screening
effect in the
nlN subshell,
which are essentialpheno-
mena for
quenching partially
the effect of the nl-elec- tronjump.
TABLE 1
TABLE II
22
TABLE III
TABLE IV
4.
Description
of thecomparisons
withexperi-
ment. - 4.1 PRINCIPLES. - In the very
light
ele-ments
(Z 20),
theexperimental isotope
shifts arepractically
pure mass shifts. In the elements of the 3d series[3],
the fieldisotope
shift is much smaller than thelarge
mass shifts due to thejump
of a 3d elec-tron ; therefore it can be subtracted in an
approximate
way to isolate the mass effect with a
good
accuracy.But now, in the
4d, 4f,
5d and 5fseries,
the fieldisotope
shift iscertainly
the dominant contributionto the
experimental isotope
shift in mostoptical lines.
So we consider
that,
in themedium-weight
andheavy elements,
theonly
safe way to extract some information on thespecific
shift from theexperimen-
tal results is the use of
King diagrams.
We refer tothe paper
by King [7]
for thepractical
way to use this well-known method.4.1.1
King diagrams
withoptical lines.,- Firstly,
we are interested in
King diagrams
built with results obtainedsolely
onoptical
lines. In these cases, eachKing
line built from the measurements onoptical
lines a and b
yields
the numerical values of the linear combination(notations
as in[8] ; (0
is theintercept
of theline),
where
Ka (resp. Kb)
is theproduct by Al A2/(A2 - A1) (with A 2 > A 1)
of the mass shift betweenisotopes
1and 2 in line a
(resp. b)
and where p is theslope
of theKing
line. In thefollowing,
we use aquantity
whichwe denote
Sp(a, b),
deduced from(o(a, b) by
subtract-ing
theeasily-calculated
contribution of the Bohrmass
isotope
shift. In terms of thespecific
shiftsAu
in the
lines,
we have ’To compare
experiment
andtheory,
we choose tocompare
experimental
values ofSp(a, b) (obtained
from
King lines)
with thecorresponding
HF values(which, actually, depend
on theslope p
of theKing line,
anexperimental quantity).
4.1.2
King diagrams
withoptical
and muonic lines.-
Secondly,
we can useKing diagrams
built withresults obtained on both
optical
and muonicX-ray
lines. A
King
line of thistype yields practically
anumerical value of the mass shift
Ka
in theoptical
line
[8].
In thefollowing,
this allows directcompari-
sons between theoretical and
experimental specific
shifts in Nd and Mo.
However,
due to the difficulties in the evaluation of thenuclear-polarization
correc-tion to the muonic shifts and in view of other pro- blems
[9],
thequestion
as to whether the relativeisotope
shifts in fieldoptical
and muonic lines areclose
enough
is notyet
settled.4.1.3
Close-configuration mixing effects.
- Theformula
given
above forSp(a, b)
leads to astraight-
forward HF evaluation in the cases where the a and b lines occur between pure LS terms of pure
configu-
rations.
But,
in thecomplex spectra, close-configu-
ration
mixing
and intermediatecoupling
areimpor-
tant
phenomena.
Both can be studied at the sametime
through
aparametric study,
when the number of measured levels is sufficient. It has been observedon a few cases
[10]
thatconfiguration mixing
hasby
far the
strongest
effects.Therefore,
for theexperimen-
tal
comparison,
we tend to select linescorresponding
to levels for which the
configurational
identification issupported theoretically by
someconfiguration- mixing study.
For a level whose wavefunction is known to be
a Ycl + PtJ’C2 (tpCi
is anintermediate-coupling
wave-function inside
configuration CJ ; a2
+p2
=1),
weconsider that a
satisfactory approximation
to thetheoretical value of the
quantity k
isThis is an
application
of the« sharing
rule »[10]
which has been known for a
long
time[11].
To
conclude,
the HF values of thespecific
shiftsof the levels and of
(,)(a, b)
arecomputed,
in thefollowing,
from the results in tables 1 andII,
theintermediate-coupling expansions
of the relevantlevels
(when available)
and theslope p
of theKing-line (when necessary).
Theintermediate-coupling effects,
linked to the variation of the
specific
shift from term to term in agiven configuration,
can beneglected
ina first
approximation (see 5.1).
4.2 THE 4f SERIES. - For the 4f
series,
all compa- risons betweentheory
andexperiment
aregathered
in table V. Lines « a » of this table
correspond roughly
TABLE V
24
to the
jump
of a 4f electron andoptical
lines « b » to thejump
of a6p
electron.4.2.1 Cerium. - For the case of
cerium,
we referto a very recent paper
by Champeau [12].
In thispaper, a
King
line is built from transitionsin Ce 1
(line a)
and4f2 6p -> 4f2
6s in Ce II(line b).
The HF result obtained
by Champeau
forAu
in thefirst transition
(38.3 mK)
is almostequal
to the value37 mK which we deduce from table 1 for the transi- tion
4f2
5d 6s5K --->
4f 5d6s2 3H.
His final compa- rison betweentheory
andexperiment
for the value ofSp(a, b)
isreported
in table V on thepresent
paper.4.2.2 Samarium. - For the case of
samarium, Hansen,
Steudel and Walther[13]
have drawnKing
lines with transitions 03BB = 5 088
Á (line a)
and 5 252and 5 271
Á (two examples
of lineb),
all of thetype
(4f6
6s6p
+4f5
5d6s2)
-+4f6 6s2 7F.
For our pur- pose, weprefer
toexchange
rôlesof 03BE and (
in theirgraph,
i. e. to take thegraph symmetrical
of theirgraph
withrespect
to the firstdiagonal.
The wave-function
expansions
of the upperlevels, resulting essentially
from themixing
of4f6(7F)
6s6p
with4f5
5d6s2,
are known from the workby Carlier,
Blaise and
Schweighofer [14].
4.2.3
Dysprosium.
- Dekker et al.[15]
measuredisotope
shifts in theDy
1spectrum,
on six différentlines,
between sevenisotopes
and with an accuracy around 0.5 mK. Five of these lines are suitable for ourstudy : they
link theground
level of theDy
1spectrum, 4fi° 5s2 5I8 (for
whichclose-configuration mixing
iscertainly small),
with three levelsbelonging
to4f 9(6H
+6F)
5d6s2
for about 96%
and two otherto
4f10(5I)
6s6p
for about 98%.
These identifications have been obtainedrecently by Wyart [16] through
a
parametric study
of themixing
of the twoquoted subconfigurations.
We choosearbitrarily
the line03BB = 5 652
Á (line a)
among the first three lines and 03BB = 5 547Á (line b)
among the last two, because the other linesbring
almostexactly
the same infor-mation.
4.2.4
Neodymium. -
Bruch et al.[8] published
a
King diagram
forneodymium
withoptical
andmuonic
X-ray
lines.Among
the twooptical lines, only
the line 03BB, = 5 621À
was identifiedby Wyart [16],
who ascribed it to the
[4f3(4I) ,
5d6S2] ’H 3 ---> 4f4 6s2 5 14
transition. The
experimental
value for the mass shift is - 23.8mK,
deduced fromfigure
1 of reference[8].
Because the
mixing
of4f3
5d6s2
with4f4
6s6p
isunknown,
thisfigure
is analgebraic
upper bound to the true value of the mass shift for thetransition.
4. 3 THE
4d,
5d AND 5f SERIES. - 4. 3. 1Molybde-
num. - In the
spectrum
of MoI,
recent results havebeen obtained
by Aufmuth, Wilde,
Behrens andClieves
[17].
Combined in aKing diagram
with themeasurements on muonic
X-ray
linesby Macagno
and Chasman and coworkers
[18], they yield
the spe- cific shifts of the transitions betweenseptets
of4d5 5s, 4d5 5p
and4d4
5s5p
and betweenquintets
of4d5 5s, 4d5 5p
and4d4 Ss2.
For the shifts of all fourconfigu-
rations to be linked
together,
we can assume that theexperimental
difference between thespecific
shifts of4d5
5s7S,
and’S2
for theisotopic pair
96-98 is close to the HF value(see
tableII).
Indeed those levels areprobably
rather pure, because no other7S
or’S
term exists in theperturbing
evenconfigurations 4d4 5s2
and
4d6.
For the odd terms4d5 5p’P
and4d4
5s5p7p,
there remains the
problem
of theirmixing through
theelectrostatic-interaction
operator
G.Using
the for-mal
expression
of theoff-diagonal
element of theoperator G,
the values ofR’(4d 5p,
5s5p)
andR’(4d 5p, 5p 5s)
in theneighbouring
Tc 1spectrum
and theexperimental energies
of the terms, we findthat the
experimental
term4d5 Sp’P
contains about 7%
of the term4d4
5s5p 7 P.
This allows us to deduce from the measurements the «experimental » specific isotope
shifts of the pure terms,by
use of thesharing
rule.
Eventually
wepresent figure
1 for thecomparison
between
theory
andexperiment.
FIG. 1. - Comparison between experiment and theory in Mo 1 (values in mK).
4.3.2 Osmium. - In Os
1, optical configurations
built on
5d5, 5d6
and5d 7
cores appear. Their Hartree- Fockspecific
shifts areroughly
in arithmetical pro-gression.
For thecomparison
withexperiment,
werefer to a recent
study by Champeau
and Miladi[19]
on the
nearby
elementtungsten.
These authors have measured the shifts in the lines 03BB = 4 982A (line a ; 5d4
6s6p -+ 5d4 6s2)
and = 4 269À (line b ; 5d3 6s2 6p -+ 5d5 6s)
forisotopes 180, 182,
184 and186.
They
deduce from thecorresponding King-
graph
theexperimental
valueSp(a, b) = - 70 + 34 K,
to be
compared
to the Hartree-Fock value - 96 K.4.3.3 Plutonium. - Plutonium is the heaviest ele- ment where mass
isotope
shift has ever been observed.The results of Tomkins and Gerstenkorn
[20]
for thelines 03BB = 5
712,4 Á (line a ; 5f5
6d7s2 -+ 5f6 7s2)
and03BB = 4
735,4 Á (line b ; 5f 6 ,7s 7p
--+5f6 7s2)
can beused to build a
King graph,
whichyields
theexperi-
mental value
Sp(a, b) = -
497 ± 120 K. Themixing
of the relevant odd
configurations
has never beenstudied,
but the values of the shiftsthemselves,
essen-tially
due to fieldisotope effect,
indicate that the upper level of line b isappreciably
mixed with5f5
6d7s2.
The HF value ofSp(a, b), -
287 K(obtain-
ed from the results of table II and the
assumption that,
aseverywhere
else in this paper, the np i ns tran- sition has a very smallspecific shift),
is analgebraic
lower bound to a correct theoretical
evaluation, namely
an evaluationtaking
themixing
into account.4.3.4 Two-electron spectra. - A direct
comparison
with
experiment
can begiven
for thespecific
shift A-uin the transitions 3s
3p 1,3’P --> 3S2 IS
ofmagnesium
and 4s
4p 1P --> 4s2 1S
of calcium(Table VI).
TABLE VI
5. Discussion and conclusion. - 5.1 VARIATION OF Q INSIDE A PURE CONFIGURATION. - The influence of
configuration mixing
on thespecific
shift has been studied above(§ 4.1.3).
We consider now the
intermediate-coupling pheno-
menon. This
phenomenon
would have no influence if the LS terms of the consideredconfiguration
hadthe same
specific
shift. So thequestion
is : does the HFspecific
shiftchange appreciably
from one LSterm to the other in a
given configuration ?
A
part
of the answer isgiven
inparagraph 2,
where a difference is noted between the
configurations
where the formal
expression
of à does notdepend
on the LS term and those where it does.
Now,
the HF method is not a central-field method.For two LS terms of the same
configuration
withdifferent formal
expressions
of the electrostatic energy, the two sets of HF radial functions are notexactly
the same ; so, even when the formal expres- sions of (i areidentical,
their numerical HF valuesgenerally
differ. In the strict central-fieldscheme,
these
changes
can be viewed upon asfar-configuration mixing
effects. A numericalexample
can be foundin Mo 1
4d55s (Table II) :
the theoreticalspecific
shifts in the terms
’S
and5S
differby
3 mK forAA =
2,
which is small withrespect
to the effect of thejump
of a 4delectron,
but notnegligible.
As concerns
configurations
withG1 integrals
foropen-shell electrons,
wepresent
four terms in NdI,
all in the
coupling (4f3 4I,
5d6s2).
Due to the 4f-5dinteraction
[23],
the formalexpressions of u
are alldifferent. But the
changes
in the HF radial functions cancel almostexactly
the effect of the formalchanges
on the numerical value of 7 : the
resulting
numericalspecific
shifts are almost identical in thisparticular
case.
Eventually,
toget
an idea of the first-order varia- tion of à inside aconfiguration,
i. e. the variation due to thechanges
in its formalexpression,
we canlook at the squares of the J
integrals,
which arelisted in table VII. In the 4f
series, only
the valuesTABLE VII
for samarium are
given,
because those for the other lanthanides are almost the same(but
forcerium,
where
J’(4f, 5d)
is muchsmaller).
We also knowthat,
from one term to another in agiven configuration,
the
changes
in the coefficients of theseJ2 quantities
in the formal
expansions
of k are at most of the orderof 1. So we conclude that the variation of the HF
specific
shift from term to term in a purecomplex configuration
of thelong-period
elements is at mostof the order of the normal mass effect in an
optical
line of the
spectrum.
5.2 COMPARISON BETWEEN THEORY AND EXPERI- MENT. - Some
comparisons
withexperiment (for Nd,
Os, Mg
and Ca above and in[3])
have beenpresented
directly
onspecific-shift
values. All the others above(rare earths)
concernquantities Sp(a, b). However,
one sees that there are
practically only
twotypes
ofspecific-shift
values in the HF results for theinteresting
pure
configurations
of each rare-earthspectrum :
the difference between these shifts is characteristic of thejump
of an f electron. We deducethat,
whatever the(accounted for) configuration mixing,
the tests onSp(a, b)
can be looked upon as tests on the value relative to an f-electronjump.
26
Then we come to the
following
conclusions. In anyspectrum
of thelong
series nl(1
= 2 or3),
for a tran-sition where an nl electron
jumps,
the Hartree-Fock methodyields
an ab initio evaluation of thespecific
mass shift which is correct in
sign
andapproaches
theexperimental
value inside a factor of two. Such transi- tions are those for which the theoretical value of thespecific
mass shift is around ten times or more the normal mass shift.More
precisely
the HF value appears to be close to 1.5 or 2 times theexperimental
value in thespectra
of the3d, 4f,
4d and 5d series.Figure 1,
in which wepresent
thecomparison
formolybdenum,
offers astriking similarity
with theanalogous figures
in the1970 paper
[3]
for the cases of Ni 1 and Cu I. Even-tually,
in theexample
wepresent
forplutonium,
theexperimental
value islarger
than the theoretical one.In the tables of this paper there also appear results for another
important type
of transitions : for the lowest np --+ nsjumps
of anyspectrum (except maybe
in the very
light elements),
thespecific
mass shift ispredicted
to be at mostequal
inmagnitude
to thenormal mass shift.
Experimental
evidence for this isgiven only
formagnesium
and calcium(Table VI),
but with poor numerical
agreement
betweenexperi-
ment and
theory.
However it must be noted that allour
comparisons
betweenexperiment
andtheory
in along
nlseries,
but those forneodymium
andmolybde-
num, deal with a
couple
ofoptical transitions,
onewith
practically
thejump
of an nl electron and theother with a np -+ ns
jump
of interest in this discus- sion. Thesecomparisons
must therefore beconsidered,
in
principle,
as tests for the combined assertion that thespecific
shift islarge
andpositive
for thejump
ofan nl electron to another subshell and that it is small for the lowest p --> s
jump.
Much work will be needed in the future to obtain accurate ab initio evaluations of the
specific
mass shiftin
medium-weight
andheavy
atoms. Moreprecisely,
progress is necessary in the
computation
of bothcorrelation effects and relativistic effects. For correla-
tion,
a recent paperby
Labarthe[24] already
showsthat it is not sufficient to
compute
the crossed second- order effects of 03A3(the specific shift)
and G(the
elec-trostatic
interaction).
As concernsrelativity,
no resultis
yet
available on its influence on thespecific
shift inheavy
atoms.Acknowledgments.
- Wegratefully acknowledge
the use of the HF numerical
computer
program of C. Froese-Fischer forobtaining
the radial electronic functions which are the basis of this work. Thanksare also due to the
experimental
group at Hannoverfor communication of their
results,
and to R.-J.Champeau
and W. H.King
forhelpful
sugges-tions.
References
[1] STRIGANOV, A. R., KATULIN, V. A. and ELISEEV, V. V., Optics and Spectroscopy 12 (1962) 91.
[2] KING, W. H., J. Opt. Soc.
Am.
53 (1963) 638.[3] BAUCHE, J. and CRUBELLIER, A., J. Physique 31 (1970) 429.
[4] VINTI, J. P., Phys. Rev. 56 (1939) 1120.
[5] STONE, A. P., Proc. Phys. Soc. 74 (1959) 424.
[6] FROESE, C., Comput. Phys. Commun. 1 (1970) 151.
[7] KING, W. H., Proc. R. Soc. A 280 (1964) 430.
[8] BRUCH, R., HEILIG, K., KALETTA, D., STEUDEL, A. and WENDLANDT, D., Colloque de Spectroscopie Atomique, Caen, 1968, J. Physique, 30 (1969) Suppl. C.
[9] STACEY, D. N., Proc. 2nd Int. Conf. on Atomic Physics (Plenum, New York) 1971.
[10] BAUCHE, J., Physica 44 (1969) 291.
[11] JONES, E. G., Proc. Phys. Soc. A 45 (1933) 501.
[12] CHAMPEAU, R.-J., Physica 62 (1972) 209.
[13] HANSEN, J. E., STEUDEL, A. and WALTHER, H., Z. Phys.
203 (1967) 296.
[14] CARLIER, A., BLAISE, J. and SCHWEIGHOFER, M.-G., J. Phy- sique 29 (1968) 729.
[15] DEKKER, J. W. M., KLINKENBERG, P. F. A. and LANGKEM- PER, J. F., Physica 39 (1968) 393.
[16] WYART, J.-F., Doctoral Thesis, Orsay (1973).
[17] AUFMUTH, P., Diplomarbeit, Hannover (1970) ; WILDE, D., Diplomarbeit, Hannover (1970) ; BEHRENS, H.-O., Diplomarbeit, Hannover (1971) ; CLIEVES, H.-P., Diplomarbeit, Hannover (1972).
[18] MACAGNO, E. R., Thesis, New York (1968) ;
CHASMAN, C., RISTINEN, R. A., COHEN, R. C., DEVONS, S.
and NISSIM-SABAT, C., Phys. Rev. Lett. 14 (1965) 181.
[19] CHAMPEAU, R.-J. and MILADI, M., to appear in J. Physique (1974) n° 2.
[20] TOMKINS, F. and GERSTENKORN, S., C. R. Hebd. Séan.
Acad. Sci. 265 (1967) 1311.
[21] CRAWFORD, M. F., KELLY, F. M., SCHAWLOW, A. L. and GRAY W. M., Phys. Rev. 76 (1949) 1527.
[22] EPSTEIN, G. L. and DAVIS, S. P., Phys. Rev. A 4 (1971) 464.
[23] JUDD, B. R., Phys. Rev. 125 (1962) 613 ; see p. 623.
[24] LABARTHE, J.-J., J. Phys. B : Atom. Molec. Phys. 6 (1973)
1761.