HAL Id: jpa-00206339
https://hal.archives-ouvertes.fr/jpa-00206339
Submitted on 1 Jan 1965
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, estdestinée 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.
Raman scattering of light from H- centers in CaF2
M. Ashkin
To cite this version:
M. Ashkin. Raman scattering of light from H- centers in CaF2. Journal de Physique, 1965, 26 (11),
pp.709-716. �10.1051/jphys:019650026011070900�. �jpa-00206339�
RAMAN SCATTERING OF LIGHT FROM H2014 CENTERS IN
CaF2 By
M.ASHKIN,
Westinghouse Research Laboratories, Pittsburgh, Pa.
Résumé. 2014 L’auteur a étudié la diffusion Raman de la lumière par les vibrations localisées de monocristaux de CaF2, contenant de faibles concentrations d’ions H- substitués aux F-. Il donne les règles de sélection. En considérant le défaut comme une particule vibrant dans un puits anhar- monique, il calcule les énergies de l’état fondamental, et des premier et deuxième harmoniques du
mode localisé. A l’aide de ce modèle, il calcule les intensités pour les transitions de l’état fonda- mental aux composantes Raman actives de ces niveaux. Ces intensités contiennent les coefficients de développement de la polarisabilité, qui sont conservés comme paramètres, à l’exception du
coefficient du premier ordre Pxy,z, qui a été calculé par la méthode des variations. La valeur obtenue est Pxy,z = 0,163 Å2. On ne peut donner de résultats expérimentaux concluants, à cause de diffi-
cultés dans la préparation des échantillons : ceux qui contenaient une concentration suffisante en
ions H- donnaient une large bande d’émission dans la région où l’on prévoyait la raie Raman.
Abstract. - The Raman scattering of light from the localized vibrations of single crystals of CaF2 containing low concentrations of H- substituted for fluorine ions has been studied. Selec- tion rules are given. For a model of the defect as a particle vibrating in an anharmonic well, the energies of the ground state, fundamental, and the first and second harmonic of the localized mode are calculated. Intensities are calculated with this model for transitions from the ground
state to the Raman active components of these levels. These intensities contain the expansion
coefficients of the polarizability which are left as parameters with the exception of the first order coefficient Pxy,z. This coefficient has been calculated variationally ; the calculated value is
Pxy,z = .163 Å2. No definite experimental results can be reported because of trouble with
sample preparation. Samples with a sufficient concentration of H- show a broad emission band in the region where the Raman line is expected.
PHYSIQUE 26, 1965,
1. Introduction. - Raman
scattering
from a -crystal
is the inelasticscattering
oflight by
thelattice vibration of the
crystal.
The initial and final states differby
the lattice modes excited.These modes are restricted in a
perfect crystal by
two
general
considerations if one isusing
radiationwhich is
essentially
of zero wave number. In the firstplace
the translationsymmetry
of theperfect
lattice and energy conservation of the
scattering
process
requires
the sum of the wave vectors of theparticipating
modes to beeffectively
zero. Thesecond consideration concerns the
symmetry
of theallowed modes that can
participate
in the Raman effect. In the case of the first order Raman effecta mode must transform like a second rank tensor under the
operation
of thecrystal point
group in order to be Raman active.When defects are
present
in acrystal
the Ramanspectra
is altered. The removal of lattice trans- lationalsymmetry
breaks the zero wave number selection rule which canchange
linespectra
toband
spectra
and introduce additional bands whichwere not
present
in the absence of defects. The defect may, in some case, also alterdrastically
someof the
modes
of theperfect crystal.
The case of alight
mass defect is anexample
where modes fromtop
of a lattice bandsplit
off to form discrete(possible degenerate)
levels. If these new levelsare Raman active then
they
show up as new lines in thespectra.
Similar considerations hold for the infrared
absorption
oflight by crystals
contain defects.An
impurity-induced
first order electric momentwas invoked
by
Lax and Burstein[1]
to accountfor
part
of theabsorption
in diamond.Schafer [2],
and Fritz
[3]
have observed localized vibrations in the infraredspectra
of alkali-halidecrystals
con-taining
substitutionalhydrogen
ions.Hayes
etal.
[4]
haveobserved,
in the infraredspectra
of substitutional H- and D- inCaF2,
the funda-mental and the second and third harmonic of the localized vibration. In this last
system
the infra-red active fundamental and overtones of the loca- lized vibration are also Raman active but there are
additional overtones which
though
infrared inac- tive are Raman active.Stekhanov and
Eliashberg [5]
have studied the Ramanspectra
of KCIcrystals containing I, Br,
and Li. These defectsystems give
rise toresonance modes
which,
with theirovertones,
appear in the observed
spectra.
Thefrequency
ofthese resonance modes lie within the bands of lattice
frequencies
of theperfect crystal.
Experiments
to measure the Ramanscattering
of
light
from the localized vibrations of H- substi-tuting
for some of the F inGaF2
are in progress[6].
These
experiments
use a He-Ne laser as alight
source. Difficulties in
preparing samples
haveoccurred and
samples
with a sufficient concen-tration of H" have so far had a broad emission
Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:019650026011070900
710
band in the
region
where the Raman lines due to the localized mode areexpected.
Direct obser- vation of thespectra
has notyielded
any conclu- sive evidence of the localized mode. The first derivative of the Ramanspectra
hasyielded
resultsof a very
preliminary
nature and theexperiments
need to be carried further before any definitive conclusions will be drawn.
Therefore,
most ofthis paper will deal with the theoretical
aspects
ofthe
problem.
The fundamental and harmonics of the localized mode will be classifiedaccording
totheir transformation
properties
under thepoint
group at the F site and the selection rules which follow are stated. When the
crystal
isvibrating
in the localized
mode,
a model of the defect as aparticle vibrating
in an anharmonic well will be used to calculate the location of the overtones and the Raman intensities for agiven experimental
situation. This model was used
by
Sennett[7]
to discuss the infrared
experiments
ofHayes
etal.
[4].
This latterexperiment provides enough
information to determine the
parameters
of themodel : the vibrational
frequency
of the harmonic localized mode and the three anharmonic force constants. The electronicpolarizability
in theRaman case is introduced
phenomenologically
andis
expanded
to second order in thedisplacements
of the defect atom with some of the coefficients left
as
parameters.
The first orderpolarizability
hasbeen calculated
variationally
for apoint-ion-model
of the lattice.
II. Selection rules. -- If a
point
defect is intro-duced
substitutionally
into aperfect crystal
andthe force constants of the
perturbed crystal
aredifferent from those of the
perfect crystal only
inthe
vicinity
of the defect then in the harmonicapproximation
the Hamiltonian of thecrystal
withdefect can be
diagonalized exactly [8].
When themass of the defect atom is much less than the mass
of the substituted host atom, there may occur vibrational modes of the
lattice,
the localizedmodes,
withamplitudes highly
localized about the defect site andfrequencies
outside the bands of theperfect crystai.
When
anharmonicity
is included in thedescrip-
tion of the
lattice,
the modes of the harmonic lat- tice arecoupled together.
Thiscoupling
of theharmonic modes leads in a familiar way to finite
widths,
shifts of levels andpartial
removal ofdege-
neracies of overtone and combination levels.
The
description
of the defectcrystal vibrating
in the localized modes as a
particle
in an anhar-monic well is
essentially equivalent
toretaining
theterms of the Hamiltonian
expressed
in the normal modes ot the harmonic defectcrystal
which containonly
localized modeoperators.
Sennett[7]
hasconsidered the effects of
coupling
to the bandmodes. His calculations were
necessarily
some-what
qualitative
due to the lack of inforrnation about thepertinent
atomic force constants fo-these interactions. We will not consider this cour
pling.
Theapproximation
ofusing
the mass of H-as the
equivalent
mass for theparticle
in the well is also made.In this
description
the harmonic oscillator(h. o.)
states for the
particle
will be used as basis statesand the
anharmonicity
treatedby perturbation theory.
We use the notation of
Wilson, Decius,
andCross
[9]
tor the irreduciblerepresentations
of thesymmetry
groups. The fluorine inCaF2
is at a siteof Td
symmetry.
The H- defect will be assumed to possess thissymmetry, i.e.,
under theoperations
of the group
Td,
the wellpotential
transformsaccording
toA 1 (the
invariantrepresentation).
The wave functions of the anharmonic well can be
expressed
as a linear combination of h. o. func- tions. Since the anharmonicpotential
canonly
mix functions of like
symmetry,
the full wavefunction is constructed from h. o. functions of the
same
symmetry.
Forsymmetry
considerations it is sufficient todiscuss
thesymmetry
of the unper- turbed h. o. function for each level.The
symmetry
of the localized modes(1. m.)
are characterizedby
thedisplacements
of the defect atom and therefore the 1. m. transformsaccording
to
F2
thepolar
vectorrepresentation
of Td. Thereare three
degenerate modes,
which can be viewedas vibrations along the three
mutually
perpen- dicular x, y, z-axis. A normalized h. o. wave function for agiven
vibrational level will be deno- tedby nx ny nz
> where nx ny nz are the number ofquanta
in the modevibrating along
the x, y,and z-axis
respectively.
The nth level of theharmonic oscillator has n = nx + nv + n,,.
The
intensity
of the Raman scatteredlight
per unit solidangle
per scatterer can be written[9.0].
were
Qo
is thefrequency
of the incidentlight,
m =
00
+ Q is thefrequency
of the scatteredlight,
nk is the unit vector in the direction of the electric vector of onelinearly polarized
compo-nent of the scattered
radiation,
and E+ and E- =(E+)*
are thecomponents
of the incident electric field when the field is writtenThe coefficient icxy,(3À is
given
bywhere .En is the energy of a vibrational
state In
>of the
system,
Z is the vibrationalpartition
func-tion,
andP-1
is thetemperature
times Boltzmann’s constant.Only
transitions from theground
stateof the oscillator will be considered and the thermal average will be
omitted, Pap
is theoperator
for theelectronic
polarizability
tensor.Before
discussing specific
levels thegeneral
selec-tion rules which follow from
symmetry
conside-rations will be stated. P transforms like a second rank tensor under the
operations
of the group Tdand P therefore transforms like
A 1
-f- E +F 2,
ormore
specifically
thefollowing assignment
can bemade :
Recalling
that theground
state n = 0belongs
to
A 1,
the allowed Raman transitions from theground
state areTransitions to
A 2
andFlare
forbidden.To conclude this section we
give
thedecompo-
sition of the levels n =
0, 1, 2, 3,
into the irredu-cible
representations
of the groupTd,
the norma-lized wave functions that
belong
to the represen- tations which are Rarnanactive,
and list thedege-
neracy of the h. o. level :
n = 0
(ground state), A 1, non-degenerate
A1 : ’¥(A1) = 1000 >
n = 1
(fundamental), F2,
3-folddegenerate
The
superscript
on the lv’s labels the irreduciblerepresentation
to which itbelongs,
thesubscript
labels the row of the
representation
to which Tbelongs.
For n =3,
the twoF2 representations
are
distinguished by primes.
III.
Energy
shifts. --- In the absence of anhar-monicity
the levels of the nowisotropic
three di-mensional harmonic oscillator well are
degenerate.
The
degeneracies
for n =0, 1, 2,
3 have, beenlisted
previously.
When theanharmonicity
isincluded the
degeneracy
ispartially removed,
states
belonging
to different irreducible represen-tations, apart
from accidentaldegeneracy,
havedifferent
energies
and when an irreducible repre- sentation occurs more than once for agiv en n
setsof
partners
have differentenergies.
The Hamiltonian for a
particle
in an anhar-monic well
including
terms of fourth order in thedisplacements
iswhere
where p is the momentum and x, y, z are the dis-
placements along the x,
y,z-axis, respectively,
ofa
particle
of mass m,and coo
is thefrequency
ofthe t,hree
degenerate
harmonic localized modes.V3
andV4
are constructed from all the terms cubic andquadric
in thedisplacements, respectively,
which
belong
toA1
in Td. Thecoefficients, L, M1,
and
M2
are related to the Fourier coefficients of the atomic force constantsappropriate
to interactionsonly involving
the localized mode. Values forL, M 1, M 2’
and coo were obtainedby
Sennett from acomparison
of the observed energy levels in the infraredspectrum
with theoreticalexpressions
forthe
F2
levels. Our theoreticalexpressions
differin one case from Sennett’s results but the calcu- lated values of
L, M1, M2,
and wo are insensitive to this difference and we use Sennett’s values.The
change
in t,he energy of astate In
> be-longing
an irreduciblerepresentation
which occursonce in the
decomposition
of a level is712
where
E:::)
is theunperturbed
energy of the statem
> and Em is theperturbed
energy toÜ(V4)
=0 ( V3).
WheIl an irreducible represen- tation occurs more than once, as for the n =.3F2 levels,
it is necessary to usedegenerate pertur-
bation
theory.
The
perturbed n =1, 2,
and 3 levels are consi- dered and the calculation of energysplitting
arecarried to second order in the third order
potential
and to first order in the fourth order anharmonic
potential.
The
energies
a.reexpressed
in terms ofÀ ==
A/2mcoo
the mean squareparticle displace-
ment. The calculated energy shifts are
where
The numerical values are based on the
following
values for the model
parameters given by
Sennett
[7 j :
Figure 1
shows the level scheme for the Raman active n ==0, 1, 2, 3
levels.FIG. 1. - Level scheme for Raman active levels.
IV. Intensities of the Raman scattered
light.
-Equations (2.1)
and(2.2)
express theintensity
ofthe Raman scattered
light
in terms of matrix ele-ments of the
polarizability operator
between diffe- rent vibrational states. Intensities will be obtain- ed for transitions between theground
state andstates which reduce to one of the n ---
1, 2,
or3 harmonic oscillator states in the absence of
anharmonicity.
In thesecalculations.,
wavefunc-tons correct to second order in the cubic
anharmonicity [O(V’)
=O(V4)]
will be used and terms linear andquadratic
in thedisplacement
of the defect will be included in an
expansion
of the
polarizability.
The calculations will be limited to thespecial geometry
ofexperiments
inprogress with a He-Ne laser.
This
expansion
will be written wherePg§
is the staticpolarizability,
audits the «-Cartesian
component
of thedispla-
cement of the defect. P1°> is
responsible
forRayleigh scattering,
P(l) and P(2)gives
the domi-nant contribution to the first and second order Raman
effect, respectively.
The
expansion
coefficients inEq. (4.2)
and(4.3) display
the sitesymmetry
of the defect and it is therefore convenient to let each ofthe x,
y, z-axis beparallel
to the two-fold axes of theCaF2 crystal.
Under a
symmetry operation
of the groupTd,
theexpansion
coefficientsobey
thefollowing
tensortransformation laws :
where s is the matrix of the
symmetry operation
in
equation.
When weapply
theseexpressions
wefind,
among others(cx, P
== x, y,z)
The
remaining
non-zeroPao,g
are obtained froniE cf. (4.7)
with the use ofThese relations show that
are the
independent
second orderexpansion
coef-ficients and
are the
independent components
of oc(3,y8.It is convenient to introduce the
following
ope- ratorsand combinations of coefficients
and to set
In terms of this
notation,
we may writeThe
operator A1
connects theA1 component
ofthc same or different states.
&,L(82)
connects theA1 colnponent
of a state to the first(second)
row ofthe E
component
of astate ; P(2)
connects theA 1 component
of a state to the third row of theF2 component
of a state.In the
present experiments
with a He-Nelaser,
in some cases the
CaF2 sample
is inside the lasercavity
andprecautions
are taken not todestroy
thelaser action. This amounts to
making
theangle
pp between the inward normal of the entrance face of the
sample [100]
and the direction of propaga-tion of the incident
light equal
to the brewsterangle.
ForCaF2
pp ss 56°. In thepresent
expe- riments the normal andpropagation
direction arein the
(001) plane
and the scatteredlight
is viewedalong
the[001]
direction. Theangle
between therefracted incident beam and
[100;
?,, ss 34°. Infigure
2 the relevant directions are shown.FIG. 2. - Sample orientation relative to incident light.
The
polarisation
directions areparallel
to the(001)
face of thecrystal
and we can thereforewrite.
The
intensity
scattered into unit solidangle
inthe direction of the z-axis for the
geometry
offigure
2 iswhere
Under proper conditions the
separate
contri-butions from each
polarization
direction i =1,
or 2 could be measured. In the
present experi-
ments the detector
accepts light
ofarbitrary pola-
rization directions in the
xy-plane
andEq. (4.27)
will be summed over all these
possible states.
Wefind
,o-
714
The calculation of 3zr,>z and Jxv,xy are
straight-
forward and we obtain the
following
results :The
quantities a7 and aT
areHere
,ZV-:r
andBT-
enter in the proper zeroth order wavef unctions for thedegenerate n
==3, F2
levels(3Td row)
and are obtained from the
following expressions :
The numerical values are based on values for the
model
parameters
found in Sennett’s thesis[7]
andgiven
inEq. (3.6).
V.
Polarizability.
- The coefficientsPaev.z,
...,Paev,aev
have been left asparameters
which can bedetermined from
intensity
measurements. Wenow calculate
variationally
the first order coef- ficientPrv,z
for apoint-ion-lattice
model. Alldynamical
effects due to the motion of the ion will beneglected
and thepolarizability
will beobtained from the energy of an H- ion substituted for a F- ion
displaced
an infinitesimal distance from itsequlibrium position
in a uniform electric field. Theremaining
atoms of theCaF2
lattice aretreated as
point-ions
and remain at theirequili-
brium
positions.
To obtain
Pzv,z
we can let thedisplacement
u ofthe
proton
of H- bealong
the z-axis(x,
y, and zare two-fold axes of the
crystal)
and the electric field F be in the
xy-plane
If u and F are treated as
small,
then theenergyE
of the
system
can beexpand
in powers of u and Fas follows :
Prv,z is identified as the coefficient of
(Fx
Fvu) /2
inthis
expansion.
The energy
(5.3)
is obtainedby minimizing
theexpectation
value of the Hamiltonian withrespect
to variations in the
parameters
of somesuitably
chosen wavefunction. A similar calculation
by Gourary [11]
of the energy and wavefunctions of U-centers in the alkali halides without the dis-placement
and the electric field gave results whichwere in fair
agreement
withexperiment.
The calculation is
performed
insteps
tosimplify
the work. Atomic units
(a. u.)
will be used and the H anxiltoni an will be writtenw h e r,-,
and
The
origin
of the coordinatesystem
has beenchosen as the
equilibrium position
of H-. The715 kinetic energy of the
proton
of H- has beenneglec-
ted and the sum in
(5.7)
is over allpoints
of theCaF 21
latticeexcept
theorigin
with Z = - 1 forF and + 2 for Ca. The interaction of the
proton
with the electric field F.u in the
present problem
vanishes
by Eqs. (5.1)
and(5.2).
The
quantity
is first minimized. The variational function
Uo
ischosen as
, ""r , 9.
where a
and P
are the variationalparameters
and
No
is a normalization factor. This functionmultiplied by (1.
+cr12)
was usedby
Chan-drasekhar
[12, 13]
for the free ion(in
freespace)
andas it stands
by Gourary
in the calculation cited[11].
Omitting
this correlation term crl2gives
a valueof the ionization
potential
for the free ion which is smaller than the measured value. This ionizationpotential
is smallcompared
with the total energy of both the free ion and the ion in thecrystal.
In view of this and of the
degree
of accuracy desired in thiscalculation,
we haveneglected
thecorrelation term.
From the form of
Ho(u)
ofEq. (5.5),
we can setu = 0 to obtain a
and P
from(5.8).
Then thesolution of the
equations
gives
the wavefunction’Y¿.
The lattice sums which occur in
Eq. (5.8)
can bereduced to sums over a
Cscl-type
lattice of latticeparameter a j2
where a is the latticeparameter
of
CaF2.
__We also use the
Madelung
constant for CsCIThe minimization is
performed
on acomputer
and we find
-
With these values
The wavefunction
Uo
has the same character asthat of a free
ion ;
onetightly
bound site andone
loosely
bound site. Like the alkali halide case, the free ion is "larger "
than the ion in acrystal ;
a = 1.. 03925and p
= 0.28309 for the free ion. The values of aand P given
inEqs. (5.13)
and
(5.14)
will be fixed in theremaining parts
ofthe calculation.
We next include
simple
tetrahedral terms in the wavefunction. It is useful to definefor an
operator
0 and a variational function X.For the variational function we take
and minimize
H,(O)
>T. withrespect
to w andT.
The final
step
in the variationalprocedure
is torninimize H >- with
where
with
y, g theonly
variationalparameters.
Theform
(5.18)
issuggested by perturbation theory
with H’ the
perturbation.
We make thefollowing approximation
inevaluating
H >it :With this
approximation
thequantity
to mi-nimize is
..
where
We include the terms
explicitly appearing
inEq. {5.3)
inevaluating (5.20).
We will not write out
explicitly
the terms ofHO(O)
>’Yo withEq. (5.16)
andEq. (5.20)
withEqs. (5.18)
and(5.19).
Beforepresenting
theresults of the calculation we remark that the work
was shortened
by
use of the "gradient
formulas[14]
and tables of3-j symbols [16].
When the minimization is
performed
we findthat (1) and T N 10-4 and the tetrahedral terms in the wavefunction appear to be
negligible.
Thesetwo
parameters
will setequal
to zero in the latterparts
of the calculation.716
For small u and F we ca.n write
Eq. (5.20)
inthe form
where
A,
..., G are known coefficients which welist later and E(O)’ is
independent
of F. Thevalues of y
and u
that minimize E areThe order of these
quantities
are consistent withour calculation of E. When these
quantities
aresubstituted in
Eq. {5.24}
we findand
The results of a numerical calculation
give
Therefore,
Schwartz
[16]
calculated the staticpolarizability
for H- in free space and obtained P(O) = 26.8
A3.
From the
compressing
of the wavefunction weexpect
and indeed find the value of P(O) in thecrystal
to be smaller than the free space value.When better
samples
areobtained,
it may bepossible
to compare the calculated value ofPxv,x
with a value obtained from absolute
intensity
measurements
[8].
Acknowledgments.
- The author has benefited from many discussions on theexperimental aspects
of this
problem
with Dr. D. ’w. Feldman and Dr. J. H.Parker,
Jr. Dr. J.Murphy
has assisted in the firstpart
of this pa.per. The numerical work wasably performed by
Miss BrendaKagle.
Discussion
M. COWLEY. - The lack of tetrahedral distortion in your wave function appears
surprising
in view ofWilliss measurement of the
anisotropic Debye-
Waller factor of
CaF2.
M. RUSSELL. - At
Royal
RadarEstablishment,
we have looked for the H- localised mode in
CaF2
at 77 oK and 20 oK. It is known that the line width is
considerably
smaller at lowtemperatures
than at room
temperature.
We have observed afluorescence in our
crystal
whichprevented
any observation of the localised mode.REFERENCES [1] LAX (M.) and BURSTEIN (E.), Phys. Rev., 1955, 97, 39.
[2] SCHAFER (G.), Phys. Chem. Solids, 1960, 12, 233.
[3] FRITZ (B.), Phys. Chem. Solids, 1962, 23, 375.
[4] HAYES (W.), JONES (G. D.), ELLIOTT (R. J.) and
SENNETT (C. T.), Lattice Dynamics, Proc. Int.
Conf. held at Copenhagen, ed. by R. F. Wallis, Pergamon Press, New York, 1965, p. 475.
[5] STEKHANOV (A. I.) and ELIASHBERG (M. B.), Soviet Physics, Solid State, 1964, 5, 2185. Soviet Physics,
Solid State, 1965, 6, 2718.
[6] These experiments are being performed by D. W. FELD-
MAN and J. H. PARKER at the Westinghouse Research Laboratories.
[7] SENNETT (C. T.), Some Problems in the Theory of Solids, Thesis, U. of Oxford, unpublished, 1964.
[8] MARADUDIN (A. A.), MONTROLL (E. W.) and WEISS (G. H.), Theory of Lattice Dynamics in the Har-
monic Approximation, Academic Press, New York,
1963.
[9] WILSON (E. B., Jr.), DECIUS (J. C.) and CROSS (P. C.),
Molecular Vibrations, McGraw-Hill, New York,
1955.
[10] BORN (M.) and HUANG (K.), Dynamical Theory of Crystal Lattices, Oxford, University Press, 1954.
[11] GOURARY (B. S.), Phys. Rev., 1958, 112, 337.
[12] CHANDRASEKHAR (S.), Astrophys. J., 1944, 100, 176.
[13] BETHE (H. A.) and SALPETER (E. E.), Handbuch der
Physik, Vol. XXXV, Springer-Verlag, Berlin. 1957.
[14] ROSE (M. E.), Elementary Theory of Angular Momen- tum, John Wiley, New York, 1957.
[15] ROTENBERG (M.), BIVINS (R.), METROPOLIS (N.) and
WOOTEN (J. K., Jr.), The 3-j and 6-j Symbols,
The Technology Press, Cambridge, Mass., 1959.
[16] SCHWARTZ (C.), Phys. Rev., 1961, 123. 1700.