HAL Id: jpa-00208236
https://hal.archives-ouvertes.fr/jpa-00208236
Submitted on 1 Jan 1975
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.
Relative intensities and polarizations in H2 rotational lines excited by electron impact
P. Baltayan, O. Nedelec
To cite this version:
P. Baltayan, O. Nedelec. Relative intensities and polarizations in H2 rotational lines excited by electron impact. Journal de Physique, 1975, 36 (2), pp.125-133. �10.1051/jphys:01975003602012500�.
�jpa-00208236�
RELATIVE INTENSITIES AND POLARIZATIONS
IN H2 ROTATIONAL LINES EXCITED BY ELECTRON IMPACT
P. BALTAYAN and O. NEDELEC Laboratoire de
Spectrométrie Physique (*)
Université
Scientifique
et Médicale deGrenoble,
BP53,
38041Grenoble-Cedex,
France(Reçu
le 22juillet 1974)
Résumé. 2014 Les valeurs relatives des intensités et des polarisations sur les raies de rotation de H2
d’une même bande excitée par impact électronique sont calculées dans
l’approximation
des atomesréunis par addition de vecteurs. Elles
correspondent
à une valeur quelconque de la vitesse des élec- trons au-delà du seuil. Les expériences faites sur les bandes 3p303A0u
~ 2s303A3+g
sont en accord raison- nable avec la théorie. Les effets de cascade sont discutés.Abstract. 2014 The relative values of the intensities and
polarizations
in H2 rotational lines in aband excited by electron impact are calculated in the united atom
approximation
by vector addition.They are valid at any electron energy beyond threshold.
Experiments
made on the 3p303A0u ~
2s303A3+g
bands show reasonable agreement with the
theory.
Cascade effects are discussed.Classification
Physics Abstracts
5.482
1. Introduction. - If an atomic vapour is excited
electronically,
it is well known that the emittedlight
is
polarized.
It has then beenpossible
to extend tomany excited states the
level-crossing
and doubleresonance
experiments
donepreviously by optical
excitation. As such
experiments
were extended tomolecules
by optical
and electronic excitation in many laboratories[1],
weattempted
to calculatethe
polarization
of molecular linesby
electronimpact
in a few
particular
cases[2].
In this paper, the calcula- tions areapplied
toH2
andcompared
to the measure-ments.
The
polarization
of the emittedlight
is due to theprivileged
direction of the incident electrons. In atoms, and in the united-atomapproximation
formolecules,
the orbital
angular
momentum L transferredby
the electrons is
perpendicular
to the electron beam at threshold[3]
and hascylindrical
symmetry around the beambeyond
threshold. Calculations of thepolarization
at threshold have been made in atoms[4]
using simple
vectôr additioninvolving
the variousangular
momenta in theground
and the excited states and theangular
momenta L and Sgiven by
the electrons. We have extended these calculations to molecules in the
Born-Oppenheimer
and united- atomapproximations. Among
the lines in aband,
the electronic and vibrational part of the wave func- tion of the involved levels are the same : the variations in the intensities andpolarizations
of the emittedlight depend
on the values of the rotational momen- tum and it ispossible
to obtain a relation valid for any energy of the incident electrons.In helium
°[5J,
absolute values of thepolarization
as a function of the electron
speed
have been measured andcompared
to thetheory
at threshold.Agreement
has been found for a few lines excited
by
a mono-energetic
electron beam(AE -
0.02eV).
No rela-tion is known between the
polarization
at thresholdand at a
given
energy. InH2,
suchprecise
measure-ments are made much more difficult
by
the weakintensities and the small
separations
of the molecular lines.However,
thepolarization
has been measuredas a function of the electron energy
[6]
in aH2 triplet
line. The curve is very similar to the
corresponding
curve obtained for atoms with an energy
spread
ofabout 1 eV : the
polarization
is maximum for electronenergies
about two times the threshold value.In our
experiments
forH2,
we chose the energy of the electrons whichyielded
the maximum value of thepolarization (35 eV)
and the measurements were restricted to the relativepolarizations
of therotational lines in a few intense bands.
AÏ
thespeed homogeneity
and thedirectivity
of the beam haveno
explicit importance
we used a beam current anda gas pressure
high enough
to obtain agood signal-to-
noise
ratio,
but lowenough
to avoid collisiondepola-
rization
[7, 8].
It is well known that
complications
in the excita- tion process occur near threshold due to thetemporary
attachment of the incident electron. Thisphenome-
(*) Associé au C.N.R.S.
Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:01975003602012500
126
non is called core or Fechbach resonances
[9].
Atthirty
eV inH2,
where the ionization energy is 15.4eV,
no excited state is obtained
through
resonances.As in
He,
resonances may occuronly
viaH2 triply
excited states which
give
rise toauto-ionizing
levels.The absence of such resonances
permits
asimple description
of the collision.The conditions for the
validity
of thetheory
areconsidered and numerical calculations are
perform-
ed
(§ 2).
Thepolarization
andintensity
measure-ments are described and the cascade effects are also estimated
(§ 3).
2.
Theory. -Z .1
CONDITIONS OF VALIDITY. - The calculations aresimply
vector additions if the follow-ing
conditions are fulfilled :- The molecule is described in the
Born-Oppen-
heimer
approximation :
theelectronic, vibrational,
and rotational parts of the wave function are inde-
pendent [10].
- The united-atom
approximation
is valid : the orbitalangular
momentum L is agood
quantum number.- L = 0 in the
ground
state : themagnitude
of Lto be added to excite a
given
electronic state isuniquely
defined.
- The collision process is instantaneous : the nuclei of the molecule are fixed
(Franck-Condon principle),
and L in the excited state is thatgiven by
the incident electron.
- The addition of orbital and
spin angular
momenta may be considered
independently
and thespins
of the incident electrons areisotropic [4].
- The
diffusing potential
hasspherical
symmetry : the electronscattering
isindependent
of the direc-tion of the molecular axis.
- The difference in energy
required
to excitethe rotational levels in a band is small
compared
tothe
spread
in energy of the incident electron beam : this energy difference has no influence on the relative cross-sections.The
validity
of these conditions are discussed inour
experimental
case in 2.2 and thephysical meaning
of the calculations is
explained
in 2.3.2.2 THE EXPERIMENTAL CONDITIONS. - We have chosen the more intense and
separable
lines in theH2
spectra : the Fulcher
bands, 3p ’Hu --+
2s3Eg
tran-sition,
excited from the 1 s’,E ground
state(Fig. 1).
The
Born-Oppenheimer approximation
is not gene-rally
valid forH2
due to thestrong
vibration-rotation interaction.However,
the variation of the Franck Condon factors with N isnegligible
when v = 0 forone of the levels of the transition or when the two
potential
energy curves may besuperposed by
atranslation in energy. If these conditions are
fulfilled,
the calculations for the excitation or the radiativedecay
aresimplified
as the relative transitionprobabi-
lities between different rotational states may be expres-
FIG. 1. -
Possible
transitions for electronic excitationand radiative emission 3p 3IIu -> 2s
3 E g+.
o and p mean respec-tively ortho and para.
sed
using Wigner’s 3 - j
and6 - j symbols.
Thisis the case for our
experiment :
in theground
state,only
v = 0 ispopulated;
in the FulcherQ bands,
the calculated Franck Condon factors show a variation of less than 1
%,
but their variation is greater for the R and P lines which arise from levelsperturbed by
3p 3Eû (11,12].
The united atom
approximation
is valid for thehighest
excited states ofH2,
which tend towards theRydberg
series[10, 13].
For n =3,
the internuclear distance is small(1 A) compared
to the diameter of the classical orbit(13 A according
to Slater[14]
and Messiah
[15]).
Thus we take the orbitalangular
momentum L of the molecular excited state as a
good
quantum number with theorigin
at the centerof the molecule.
The united-atom
approximation
also supposes that the electrons are diffusedby
aspherical potential, independent
of the direction of the molecular axis.Calculation of inelastic cross-sections for a few eV electrons in
H+ [16]
have shown that theanisotropy
due to the molecular axis is
negligible compared
tothe
polarization
inducedby
the incidentélectron.
In our
experimental
case, it ispossible
to showby simple
considerations that thishypothesis
may be reasonable. In theH2 molecule,
theprobability density
distribution of the electrons in theground
state has
spherical
symmetrybeyond
0.8A
from thecenter
[10
p.352].
For 35 eV incident energy and 13 eV energyloss,
theprobability density
distribu-tions of incident and emergent waves
[15]
such asAI. = 1 are the most
important
at the limit of thespherical
part of the molecule.The collision
time,
about10-16
s( 17],
is the shortest of all the considered times. Theperiod
of vibration is greater than 4 xlO-15
s which is the value for thehighest
considered v(v
=3).
Theperiod
of rota-tion is greater than 5 x
10- 14
s which is also the value for thehighest
considered N(N
=4).
Theperiod
of L due to the nuclear electric field may bededuced from the energy
separation
between3p 3HU
and
3p 3 Eu+ :
10- Il s.Although
a few of theseapproximations
appear to beapproximately valid,
it seemsinteresting
toapply
thesimple theory
as a firstapproximation
tocalculate the relative intensities and
polarizations
in
H2
lines. It allows the calculations to be doneusing only
vector addition and not molecular wave-func- tions. Itsvalidity
will bejustified a posteriori by
theexperimental
results.2. 3 PHYSICAL MEANING. - We shall
emphasize
thefact that the calculations
give
the relativepolariza-
tions and intensities at any electron
velocity.
The different
angular
momenta of the moleculeare denoted as usual
[10]
except that the notation for the rotationalangular
momentum ischanged
from K to 1V to conform to more recent usage.
The orbital
angular
momentum Lgiven
from theelectron beam to the molecule may be defined in the united-atom
approximation :
L = 1 for a s - ptransition,
L = 2 for a s - d transition. If the quan- tization axis oz is takenparallel
to the electronbeam,
the excited state hascylindrical
symmetry around oz and thedensity
matrix isdiagonal.
Atthreshold,
L is
perpendicular
to the beam :ML
= 0. Thedensity
matrix p of the excited state is at threshold :
T ML
is an irreducible tensor operator[ 15] ;
N’and N"denote rotational levels in the excited and in the
ground
staterespectively.
Beyond threshold,
for agiven speed
for an incidentelectron,
we are unable topredict
theprobability 1
aLM,,l’
for an orbitalangular momentum [ lez )
to be added to the molecule. The
density
matrixmay be
expressed
as a function of aLML :with
We assume that the
ground
state isisotropic :
and we express the matrix elements
[18]
in the bcoupling
case.The component of the
density
matrix on thetensorial
basis N’TÔ
definedby
Omont[19]
are[2, 7] :
with
where Vi,
andF.
arerespectively
the vibrational and the electronic wave functions.Q = 0
since oz istaken
parallel
to the beam. We know thatas
they correspond
toopposite
deviations of the incidentelectron,
and the non-zero componentsN’ Il correspond
to even values of x[20].
To describe the
linearly polarized light
emittedafter electron
impact,
we have to consider x = 0(population)
and x = 2(alignment).
The relative values of thepopulations
may be calculated from(3).
The
alignments
in the various rotational states aremaximum at threshold where
Lpô- 2
is maximum.Regardless
of the electron energy, their absolute value cannot becalculated,
but their ratios aregiven by (3) :
the relative valuesof
thealignment
in the’ rotational levels in a band are
independent of
thespeed of
the incident electrons.If a level N’ is excited from many levels N" in the
ground
state, the relativepopulations
oralign-
ments are obtained from a sum over N"
weighted by
the relative
populations PN" :
The
population
of N’ excited from N" isgiven by :
The
degree
ofalignment
on a line N’ ---> N is definedas :
where
III
andIl
are the intensities of thelight
propa-gating
in a directionperpendicular
to the electronbeam,
withpolarization parallel
orpërpendicular
to the electron beam. It may be obtained from the
weighted
sum :The
tN’->N(N")
may be found in[2].
We now consider the
experimental
case of an inci-dent electron beam
having
anangular spread.
Letoz’ be the direction of an ensemble of incident elec-
128
trons
making
anangle
8 with respect to oz. The obtai- nedalignment along
oz isgiven by :
where
rôo-(B)
=2(3 cos2
9 -1)
is a rotation matrixelement
[15, 20]
and(N’p0x=2)oz’
isgiven by (3).
Theresulting alignment
will be obtainedby
theweighted
sum over the various 0
angles.
A lack ofdirectivity
of the electron beam therefore decreases the
magni-
tude of the
alignment
but does notchange
the relative values of thealignments
in the rotational lines of aband.
2.4 CALCULATIONS. - After this discussion of the
physical problems applied
tosinglet
states, wegive
the formulae in a form suitable for numerical calcula- tions when electronic and nuclear
spins
are present for the case of bcoupling.
We first consider the case where J and F are
good
quantum numbers. We then consider the
special
case of the ortho
3p 3 nu
levels ofH2
where J is not agood
quantum number.Finally,
wegive
the calculated values of thedegree of alignment
of theP, Q,
R Fulcherlines.
2.4.1 Natural width «
hyperfine structure «fine
structure. -
Using
thehypothesis
of Percival and Seaton[4],
it can be assumed that there is anequal probability
forpopulating
all thespin
states. Whenan
isotropic
electronicspin
isadded,
one obtains[19, 7, Ô 1
- -. -. -
The addition of an
isotropic
nuclearspin gives, assuming
that J’ is agood
quantum number :The observed
intensity I(eu, À)
derived from thed’Yakonov
[21] theory
andexpressed
in the nota-tion of Omont
[19],
when the fine andhyperfine
structures are unresolved
by
theoptical
system, is :where
cP:(eu) [21]
are determinedby
thepolarization
e used to observe thelight
emittedalong
the direction u.We have assumed that the
hyperfine
structure is muchgreater
than the natural width of theemitting
level. Theintensities
expressed
in that waygive
the same results as the Hônl-London formulae[10].
The
degree of alignment t
measured on a lineoriginating
from the rotational level N’(N’ -
Ntransition)
is :For lines
originating
from levels N" which are excited fromonly
one level N" of theground
state(e.
g. theQ
branches of the Fulcherbands) :
The case with zero
spin (S
= 0 or I =0)
iseasily
deduced from(11)
and(12).
(1) Formula (5) of reference [2], with S" = 0, gives the same
result as (8) using the relation :
where S, L, V, V’ are integers.
2.4.2 Natural
width « hyperfine
structure =fine
structure. - The measurements made
by
Freundand Miller
[22]
have shown that fororthohydrogen,
in the
3p 3 nu level,
thehyperfine splittings
are compa- rable to the fine structure. That is to say, J’ is nolonger
agood
quantumnumber,
and the wavefunc- tion for a state of totalangular
momentum F’is :
where the sum is over all
Jl
values for asingle
rotatio-nal level N’. The
wavefunction (J’) F’MF >
is aneigenfunction
of the total hamiltonianwhere
Jefs
andJehfs
denoterespectively
the fineand
hyperfine
structure hamiltonians. The matrixelements of
Jefs
andJehfs
with the functions1 yN’ SJ’ I ;
F’MF >
have been calculatedby
Fon-tana
[23],
Chiu[24]
andby
Jette and Cahill[25].
Wehave
diagonalized
the matrix of X, and calculated the coefficientsC(F’
, for N’ =1, 2, 3
and 4[7],
using
the fine andhyperfine
structure constants measuredby
Freund andMiller,
and calculations of Lombardi[26].
(1)
The constants areexpressed
inMc/s.
(2) d
has been calculatedby
Lombardi.c is deduced from c - 3 d measured
by
Freund and Miller.When J’ is not a
good
quantumnumber, @)px
andI(e., Â)
areexpressed
in terms ofC(j,)j,
and(12)
becomes :The numerical values of t calculated with these formulae are
given
in table 1.z
TABLE 1
3.
Experiments.
- 3 .1 POLARIZATION MEASURE- MENTS. - We shall compare the relative values of thedegree
ofalignment
measured on the Fulcher bands with the values obtained from ourtheory
andgiven in §
2.4.In our
experiment,
the molecularhydrogen
gas is excited in a triodeglass
cell. Wemeasûred
thedegree
of
polarization
p of a line P =IL
and then calcu-Il
II+Il
1 - I.l
2 Plated the
degree
g ofalignment
=1"" + 2 11- = 3 - P . /jj +211 i 3-P .
A
simple
theoretical relation between the values of t at threshold and at agiven
energy is notknown,
but
in § 2. 3,
we have shown that the relative values of t for the lines of a band at threshold areequal
tothe relative values of t at any energy whatsoever.
3.1.1 The
experimental
set-up. - The cells werecontinuously pumped
while molecularhydrogen
cir-culated at the
working
pressure.High purity (99,999
5%) H2
gas was used. The pressure in the cellwas measured with a Pirani gauge which was cali- brated
against
a MacLeod gauge. Before use, the cells were baked at 300 °C over aperiod
of 24 hours.In the
triode,
the electrodes were made ofmolyb-
denium which absorbs
considerably
lessH2
thandoes the tantalum
generally
used.The
grid-cathode voltage
was 35 V and the currentwas 20 mA. The energy
spread
due to the cathod emission is about 1 V which isgreater
than the diffe-rences between the threshold
energies
used to excitethe rotational levels in a
band,
asrequired
for thevalidity
of thetheory.
The pressure was 4 x10 - 2
torr.130
It was low
enough
for thedepolarization by
collisionto be at the limit of the
experimental uncertainty
for all the studied lines
[7, 8],
buthigh enough
toobtain a
reasonably good signal-to-noise
ratio. From the inelastic collision cross sections[30]
we estimatethat the chance for an electron to
experience
an energy loss is about 1%. However,
we recall that thespeed homogeneity
of the electron beam has noexplicit importance (§ 2.3).
The emitted
light
wasanalysed by
a Jobin-Yvontype HRS 1 monochromator
having
adispersion
of 12
Â/mm.
In order to separate thelines,
we useda resolution of 1
Á,
as shown in the spectrum offigure
2. Thelight
was detectedby
an EMI 9558QB photomultiplier
cooled to - 50 OC.FIG. 2. - The Fulcher bands. Triode excitation.
3.1.2 Polarisation measurements. - The
polari-
zation measurements were made
by
a modulationtechnique.
The studiedlight passed through
a linearpolarizer rotating
at afrequency f,
beforeentering
the
grating
monochromator. The instrumentalpola-
rization
mainly
due to the monochromator is elimi- nated with a set ofplane-parallel glass plates placed
before the entrance slit of the monochromator.
Then the
photomultiplier signal
is the sum of atime-independent
currentio proportional
toIli
+Il,
and of a component
i2 f
at afrequency 2 f,
propor- tional toIII
-Il.
With a systemregulating
thephotomultiplier
current[27], io
iskept
constant.The relative values of the
polarization
of the lines ofa band are then
given by
the relativeamplitude
ofi2 f
measured for each line with a lock-in detection.To
improve
thesignal-to-noise ratio,
thesignal
wasaccumulated in a
computer (T 2000)
over more thantwo hours : one data was taken every second and the
sum of 8000 such values was used for each line.
During
thistime,
the current and the pressure werekept
constant in the cell. For each line we have taken the mean value of about seven suchsignal
averages.In our
experiments,
the earth’smagnetic
fieldwas
compensated.
The width of the Hanle curvesobserved on these lines
[7, 8]
show that thedepolari-
zation
by
this field would be a few per cent.For the 2 -+ 2 Fulcher
band,
thewavelengths
ofthe considered
Q
branch lines(Q(2), Q(2), Q(3), Q(4))
and R branch lines
(R(0), R(l), R(2), R(3))
are within25
A
and 50A respectively.
Thespectral
response of thephotomultiplier
can then be assumed constant in each branch[7].
3.1.3 Results. - Absolute
polarization
measure-ments have been made
by comparing
thesignal
obtained with the studied line to the
signal given by
a
completely polarized light. They
were all foundto be
positive
and less than 12% :
The lines of the
Q
branchesoriginate
from levels which are excited fromonly
one rotational level of theground
state(Fig. 1).
Thedegrees
ofalignment t
of the
Q
branches are thenindependent
of the relativeFiG. 3. - Comparison of observed and calculated values of the relative degrees of alignment of Fulcher lines.
Calculated : - without cascade correction. - - - with cascade correction from dl, dH, dA (from left to right respectively).
Observed : D v = 2 ; 0 v = 3. The vertical lines are the estimated
populations
in theground
state, i.e. of the temperature.We have measured P for the lines
Q 1, Q 2, Q 3, Q
4of the 2 - 2
band,
and for the linesQ 1, Q
2 ofthe 3 --> 3
band,
which are well-resolved with ourmonochromator. Our results are
given
infigure
3a.Agreement
isgood
within the limits ofexperimental
error. The effect of cascades is discussed
in §
3.3.The lines of the R and P branches
originate
fromlevels which are excited from two rotational levels of the
ground
state(Fig. 1).
Thedegree t
ofalignment
of these lines
depends
then on the relativepopulation
of the rotational levels of the
ground
state, i.e. on thetemperature.
The calculations show that t is not very sensitive to thetemperature :
forT =
(750
±200) K ,
At/t
is less than 4%.
Thechange
of the relative values of t for the lines of a band is therefore much smaller than theprecision
of our measurements. The tempe-rature of the gas between the
grid
and theplate
in atriode
glass
cellanalogous
to ours, has been estimated in ourlaboratory [8]
to be 800 K. Ourintensity
measurements
yield
a temperature of about 750 K(see Fig.
2 andFig. 4).
FIG. 4. - Comparison of observed and calculated values of the relative intensities in Q branches.
Calculated : -. Observed : D, 0 , d, correspond to v = 2, 3, 4 respectively. (a) Calculated at various températures. (b) This work, 77 K, 5 x 10-3 torr, HF discharge, Fulcher bands. (c) This work, 700-800 K, 1.5 x 10-2 torr, triode, Fulcher bands. (d) Ref.
[29], 311 K, 5 x 10-2 torr, 15 keV, Werner bands.
For the R
branches,
we have measured P for the lines R0,
R1,
R2,
R 3 of the 2 --. 2band,
whichare well resolved
by
our monochromator. Our resultsare
given
infigure
3b. The agreement is better than would beexpected
when one considers the fact that the R branchesoriginate
from levels of il + type which areperturbed by
the 3s3 E g+
level[10].
For the P
branches,
thepolarizations
are smalland the lines are not well resolved
by
the monochro-mator. Thus we believe that our
experimental
resultsare not
significant
and we do not quote them here.3.2 INTENSITY MEASUREMENTS. -
Intensity
measu-rements at very low temperature may be a sensitive test for our
theory
because the band headonly
isexpected
to be present.Conversely,
relative intensities may be used to estimate the temperature of the gas.We compare the intensities of
Q
lines ofpn --+
sEtransitions.
The
populations
in theground
state must be cal- culatedseparately
for the ortho and para modifica-tions,
as in ourexperiments,
even at 77K,
the ortho Hpara transformation does not have sufficient time to occur
[10].
Theproportions
of ortho and para remain 3 : 1. Thepopulation
in theground
staterotational level N of each ortho or para
species
isproportional
to(2
N +1) exp(- BN(N
+1) hc/kT),
where the rotational constant B = 60.8 cm -1. To compare the intensities of all the lines in a
band,
wecalculate their relative intensities for
equal
pressure of ortho and para, and theexperimental
intensities of the ortho lines are dividedby
three.The
populations
in the excited state may be calcula- ted from formula(5).
InH2,
A Il = 0 in theground
state, L = 1 and A’ = 1 for
3p 3ilu.
For theQ
bran-ches of the Fulcher
bands,
N’ = N"(Fig. 1).
In thiscase, formula
(5)
shows that thepopulations
of thelevels N’ are
proportional
to thepopulation
ofN" = N’ in the
ground
state. As the N’ levels radiateonly
to N = N’(Fig. 1),
theintensity
of eachQ(N)
line is
proportional
to thepopulation
of the rotational level N" = N of theground
state.3 . 2 .1 HF
discharge (Fig. 4b).
- In order to main-tain the walls at a well defined low temperature, an electrodeless
discharge
wasplaced
inliquid nitrogen,
in a
glass
Dewar which was fitted with an observa- tion window. Theapplied
electric field was about 100V/cm
at 100MHz,
whichcorresponds
to amaximum electron energy of 40 eV.
When
decreasing
the power and the pressure, the relative intensities ofQ(l), Q(2), Q(3)
lines tend toa limit. At 77
K, Q(2)
andQ(3)
are notpredicted
butdo occur. We can conclude that the united-atom
approximation
is notprecise enough
toexplain
theexperimental
intensities. Thediscrepancy
is however mostlikely
due to cascades. It can be used to estimatethe contribution of cascades in
populating
the3p 3 nu
level. This estimation is an upper value as the rota- tional temperature in the
discharge
canonly
behigher
than 77 K. Its small
importance
agrees with the conclusion of reference[8]
who detected no effectsof cascades in
magnetic
resonanceexperiments.
3.2.2 Triode
(Fig. 4c). - We
report the relative intensities of theQ
lines in the triode where thepolarization
measurements have been made. The best fit for the temperaturegives
750K, although
the temperature is
probably
nothomogeneous.
3. 2. 3
High energy electrons (Fig. 4d ).
-We report
the results of reference
[29],
on2p 1 il u --+ 1s1Fg+ .
The authors
verify
that the relative intensities of theQ
branches agree with theoptical
selection rules attheir measured temperature 311 K. The agreement is better than in
figure
4b : the cross-section of then = 2 level is much greater than for the upper levels which may cascade onto it.
132
3.3 CASCADE EFFECTS. - We attempt to estimate the
importance
of cascades and their effects on the measureddegrees
ofalignment.
We can use our
intensity
measurements at 77 K(see Fig.
4 and§ 3.2)
to estimate the contribution of thé cascades topopulate
the3p 3 llu level, assuming
that the
Q(2)
andQ(3)
lines are dueonly
to cascades.A p level can be
populated
from s and dlevels,
and alsofrom
f levels(4f
-2p
transitions have beenobserved).
The contribution of d levels isprobably
the most
important,
because the 4d -2p
lines aremuch more intense that the s - p or the f -> p lines
[28].
We calculate(see § 2.4)
the relativepopula-
tions of the first rotational levels of each d
3 EII d,
and the
resulting
relativepopulation
after theirdecay
on
3p 3II u.
These relativepopulations
arenearly independent
of thehigher
level(dl, dll,
ord4)
and are
proportional
to0.35, 0.40,
0.25 forQ(l), Q(2), Q(3) respectively. By comparing
the observed intensities ofQ(2)
andQ(3)
with that ofQ(l),
we finda cascade contribution of 17
%
to the totalpopula-
tion of
3p 3 nu.
The
degree
ofalignment
of the linecorresponding
to the transition N’ - N, for
singlet
levels is :For
singlet levels,
the influence of cascades canbe calculated with formula
(16)
whichgives
theratio of the
degree
ofalignment
on two consecutive transitions of a cascadeN1 --> N2 --> N3 [20] :
In the case of « b »
coupling,
thedegree
ofalignment
for a transition between twotriplet
levels(N’ --> N)
can be
expressed using
tsgiven by ( 15) :
-
parahydrogen :
-
orthohydrogen :
Formulae
(17)
and(18)
allow one to deduce thecascade effects for
triplet
levels from that ofsinglet
levels. The ratios
tT/ts
can be obtained for the different lines from the values calculated for the direct excita- tion of plllu (using
formula(15))
and of p3llu
(table 1, § 2.4).
The calculated relative values of the
degree
ofalignment
forQ
and R branches of the Fulcherbands, taking
into account the direct excitation of3p 3llu
and the cascades fromd3IllJ,
are shownin
figures
3a and 3b. The influence of each leveldl, dll,
dJ has been consideredseparately, assuming
that it
gives
one third of the cascade.The influence of cascades as estimated above is smaller than the
experimental
error,except
for theQ(4)
line which is the least intense. We recall that theimportance
of the cascade isprobably exaggerated.
4. Conclusion. - The agreement between the
experiment
and thetheory
isremarkably good
inspite
of the manyapproximations
in thetheory
andthe presence of a
light
cascade in theexperiment.
The relative values of the
polarization
show very different features between the fourQ
lines and the four R lines with variations as great as seven. Thediscrepancy
is however within theexperimental
errorestimated to be about 15
%.
More refined measure-ments with a resolution in the electron energy of a
few
10 - 2
V atenergies
less than 1 eV above threshold would avoid the effect ofcascades,
but the resonanceswould
probably
make theexperimental
results lesssignificant
with respect to thesimple theory.
Acknowledgments.
- We are indebted to J. C. Jolivetwho
helped
us toimprove
theexperimental
set up and made the first measurements and to J. M. Caste-jon
who made theglass
cells.References [1] NEDELEC, O., BALTAYAN, P., J. Phys. B : At. Mol. Phys.
3 (1970) 1646.
[2] BALTAYAN, P., NEDELEC, O., J. Phys. B : At. Mol. Phys.
4 (1971) 1332.
[3] LAMB, W. E., Phys. Rev. 105 (1957) 559.
[4] PERCIVAL, I. C., SEATON, M. J., Phil. Trans. Roy. Soc. London
A 251 (1958) 113.
[5] MC FARLAND, R. H., SOLTYSIK, E. A., Phys. Rev. 127 (1962) 290.
[6] JETTE, A. N., CAHILL, P., SCHWARTZ, R., J. Chem. Phys.
19 (1967) 283.
[7] BALTAYAN, P., Thèse, Grenoble, 1973.
[8] MARECHAL, M. A., Thèse, Grenoble, 1973.
MARECHAL, M. A., JOST, R., LOMBARDI, M., Phys. Rev.
A 732 (1972) 740.
[9] SCHULZ, G. J., Rev. Mod. Phys. 45 (1973) 378.
[10] HERZBERG, G., Spectra of diatomic molecules (Van Nostrand Co, New York) 1950.
[11] VILLAREJO, D., STOCKBAUER, R., INGHRAM, M. G., J. Chem.
Phys. 50 (1969) 1754.
[12] HALMAN, M., LAULICHT, I., J. Q. R. S. T. 8 (1968) 935.
[13] MULLIKEN, R. S., CHRISTY, A., Phys. Rev. 38 (1931) 87.
[14] SLATER, J. C., Phys. Rev. 36 (1930) 57.
[15] MESSIAH, A., Mécanique Quantique (Dunod, Paris) 1964.
[16] TEMKIN, A., VASAVADA, K. V., CHANG, E. S., SILVER, A.
Phys. Rev. 186 (1969) 57.
[17] SMITH, F. T., Phys. Rev. 118 (1960) 349.
[18] RUBIN, P. J., Opt. Spectr. 20 (1966) 325.
[19] OMONT, A., J. Physique 26 (1965) 26.
OMONT, A., Thèse, Paris, 1967.
OMONT, A., MEUNIER, J., Phys. Rev. 169 (1968) 92.
[20] NEDELEC, O., J. Physique 27 (1967) 660.
NEDELEC, O., Thèse, Grenoble, 1966.
[21] D’YAKONOV, M. I., Sov. Phys. J. E. T. P. 20 (1965) 1484.
D’YAKONOV, M. I., Opt. Spectr. 19 (1965) 372.
[22] MILLER, T. A., FREUND, R. S., J. Chem. Phys. 56 (1972) 3165.
[23] FONTANA, R. P., Phys. Rev. 125 (1962) 220.
[24] CHIU, L. Y. C., J. Chem. Phys. 40 (1964) 2276.
[25] JETTE, A. N., CAHILL, P., Phys. Rev. 160 (1967) 35.
[26] LOMBARDI, M., J. Chem. Phys. 58 (1973) 797.
[27] EDEL, P., Thèse de 3e cycle, Grenoble, 1972.
[28] DIEKE, G. H., The hydrogen molecule wavelength tables (Wiley- Interscience, New York) 1972.
[29] REICH, H. J., SCHMORANZER, H., Z. Phys. 186 (1965) 305.
[30] CORRIGAN, S. J. B., J. Chem. Phys. 43 (1965) 4381.
CARTWRIGHT, D. C., KUPPERMANN, A., Phys. Rev. 163 (1967)
86.
KHARE, S. P., Phys. Rev. 157 (1967) 107.
PROK, G. M., MONNIN, C. F., HETTEL, H. J., J. Q. R. S. T.
9 (1969) 361.