Electron capture processes in collisions between Mg (62S 1/2) atoms and Na+(1S0) ions in the 0.10-4.00 keV energy range

Article abstract of DOI:10.1016/j.cplett.2003.09.161

The fluorescence of excited species formed in collisions between neutral magnesium atoms and sodium ions has been studied experimentally using a crossed ion-atom beam technique. The only emission detected corresponded to the Na(3²P_3/2,1/2)→Na(3²S_1/2) decay of neutral Na atoms formed by electron capture. Its strength has permitted to resolve it into its J=1/2 and J=3/2 states, and for the latter the M_j magnetic states have also been separated, cross-section dependence on collision energy or branching ratio being given for all of them.

Full text of DOI:10.1016/j.cplett.2003.09.161

Chemical Physics Letters 382 (2003) 106–111
www.elsevier.com/locate/cplett
Electron capture processes in collisions between Mg (62S1=2)
atoms and Naþ(1S0) ions in the 0.10–4.00 keV energy range
*
ꢀ
J. de Andres, M. Sabido, J.M. Lucas, J. Sogas, A. Aguilar
ꢀ
Departament de Quꢀımica Fꢀısica, CeRQT/Parc cientꢀıfic de Barcelona, Universitat de Barcelona,
C/Martꢀı i Franqueꢁs, 1, 08028 Barcelona, Spain
Received 18 September 2003
Published online: 6 November 2003
Abstract
The fluorescence of excited species formed in collisions between neutral magnesium atoms and sodium ions has been
studied experimentally using a crossed ion–atom beam technique. The only emission detected corresponded to the
Na(32P3=2;1=2) ! Na(32S1=2) decay of neutral Na atoms formed by electron capture. Its strength has permitted to resolve
it into its J ¼ 1=2 and J ¼ 3=2 states, and for the latter the Mj magnetic states have also been separated, cross-section
dependence on collision energy or branching ratio being given for all of them.
ꢀ 2003 Elsevier B.V. All rights reserved.
1. Introduction
are relatively easy to generate, as is also the mea-
sure of the fluorescent emission from the decay of
electronically excited atoms, but also because their
main dynamical behavior can be studied per-
forming relatively simple model calculations. In
this field, since the pioneering experimental work
on alkali ion–alkali atom systems done by Perel
et al. ([2] and references cited there), interesting
studied have been performed by Aquilanti et al.
([3] and references cited there), Tolk et al. [4],
Andersen et al. [5], and Brunetti et al. [6], covering
many symmetrical and asymmetrical alkali
ion–atom pairs. More elaborated experiments
involving alkali atoms excited by polarized lasers
and devoted to observe the specific alignment
effects on electron transfer collisions having been
Collisions between ions and neutrals (atoms or
molecules) play an important role in different areas
of the Physics and Chemistry of low density plas-
mas and ionized gases [1]. Relevant examples of
ion-neutral processes are those taking place in in-
terstellar clouds, planetary ionospheres, gas dis-
charge and in plasma deposition of thin metal
films. Ion–atom collisions are a current research
topic in the field of Chemical Dynamics using
molecular beams techniques, and among these the
case where both atoms are alkali in nature has
attracted much attention. This is due to a series of
circumstances: both ion and neutral atom beams
* Corresponding author. Fax: +34-93-402-1231.
E-mail address: antonio@qf.ub.es (A. Aguilar).
€
done by Bahring et al. [7] and more recently by
Thomsen et al. [8].
0009-2614/$ - see front matter ꢀ 2003 Elsevier B.V. All rights reserved.
doi:10.1016/j.cplett.2003.09.161

J. de Andrꢀes et al. / Chemical Physics Letters 382 (2003) 106–111
107
Taking into account the relatively simple elec-
tronic structure of the alkali atoms, where essen-
tially only the single ns valence electron becomes
excited in the energy range currently under study,
worthwhile results on theoretical calculations can
be obtained even using relatively simple models
which take into account the intrinsic non-adiabatic
nature of the electronic excitation or electron
transfer processes. Thus, Aquilanti [9] improved
earlier studies of Perel et al. [2] and Melius et al.
[10], while McMillan [11] and Shingal et al. [12]
performed semiclassical studies. Recently Na-
kamura et al. [13] reported on a complete picture
of the two-state curve crossing problems which can
be applied to elastic and inelastic scattering as well
as to perturbed bound state problems.
states. Taking this fact into account, it can be
expected that spin–orbit interaction will play an
important role in the interpretation of experi-
mental results. In this Letter we report on the
first results, obtained for the (Mg–Na)þ colli-
sional system, which seems to offer the possibility
of a direct application of the results obtained to
the design or modification of light sources, for
instance. This Letter is organized in the following
way: A brief description of the experimental
apparatus is contained in Section 2, while Section
3 is devoted to present experimental results.
A discussion of the experimental results forms
Section 4.
A systematic study of collisions involving al-
kali-ions and alkali-atoms ([14] and references
cited there) has been performed in the last decade
by our research group from both experimental
and theoretical points of view. One-electron ab
initio calculations using non-empirical relativistic
pseudopotentials to model the core electrons have
been successfully applied to several alkali ion–
atom pairs allowing an interpretation of the most
relevant experimental results obtained. Since in-
formation on collisions involving atoms with
more than a single valence electron is rather
scarce (see below), we have recently begun to
employ alkali-earth targets, starting with magne-
sium atoms. In the range of relatively low colli-
sion energies (under 5.0 keV) in which we
operate, only one of the two valence electrons of
this atom can be expected to become excited, just
as it happens on alkali–alkali collisions, but the
presence of the second electron puts these systems
into a higher complexity level, from the viewpoint
of their theoretical interpretation at least. In
particular, asymptotic excited channels can now
correlate neutral Mg atoms in both singlet and/or
triplet states, while charge transfer processes lead
to the formation of fragments in doublet states.
Thus, in principle, the system can evolve from the
2. Experimental setup
A detailed description of the experimental
crossed beam setup has been published already
in [14]. In this apparatus a beam of alkali atoms
(Naþ in the present case) is generated by a
thermionic source and accelerated by a 0.10–4.00
keV electric field, focused by an einzel lens sys-
tem and later collimated. The ion beam crosses
at a right angle the thermal neutral beam of
ground-state Mg atoms generated by heating in
an oven. The fluorescence produced by decay of
the excited electronic states produced in the
collision is collected by an optical system placed
perpendicularly to the collision plane over the
scattering center, analyzed by a 50 cm mono-
chromator and read by a photomultiplier. In
order to optimize data collection, both signal
intensity and ion current values are automatically
scanned at 10 readings per second and stored
in a multichannel system which furthermore
ensures an optimal correlation between both sets
of data.
Background pressure is usually maintained in
the range of 10ꢀ7–10ꢀ6 mbar during experiments.
In order to report absolute cross-section values,
experiments on the Mg + eꢀ system were done to
calibrate our experimental system [15]. To this
end cross-sections for the inelastic process
Mg(1S0) + eꢀ ! Mg(31P) + eꢀ were measured in
arbitrary units and then compared with previously
reported data [16] in absolute units, normalizing
1
singlet entrance channel (Mg(3s2, S0) + Naþ(1S0)
in this paper), correlating with a singlet quasi-
molecular state, and leads to the formation of
asymptotic singlet, doublet or triplet atomic
states through singlet and triplet quasimolecular

108
J. de Andrꢀes et al. / Chemical Physics Letters 382 (2003) 106–111
them at 1400 eV collision energy in the laboratory
frame of reference.
transitions ought to be observed too, but by
comparing these results with some preliminary
ones obtained for Mg–Rbþ and Mg–Csþ it is
possible to see that there is a tendency in Mg–Xþ
collisions for electron transfer processes to become
more efficient when ion mass increases. On the
other hand, while there are not many experimental
data about ion–atom inelastic collisions involving
closed-shell alkali-earth targets atoms: Aquilanti
et al. report on inelastic collisions in Hg–Naþ
[20,21] and Cd–Naþ [22] system. However, taking
into account the different electronic structure na-
ture of the closed shell Hg and Cd target atoms,
different behavior can be expected when compar-
ing with those obtained for the Mg–Na+ system.
Thus, in the same energy range we have studied,
the Hg + Naþ collision leads to the excitation of
mercury to both Hg(63P1) and Hg(63D1;2), and the
electron capture channel leading to Na(32P) was
also measurable. In the case of Cd–Naþ system
only the direct excitation channels leading to
Cd(1P1) and Cd(3P1) could be measured.
It is also worth noting in Fig. 1 that excitation
function shape differs considerably from equiva-
lent ones in alkali–alkali collisions [14], lacking a
maximum and having instead a noticeable inflec-
tion at some 1700 eV collision energy and leveling
out at some 2600 eV, both energies in the labora-
tory frame of reference. When comparing it with
the equivalent one for Hg–Naþ it can be appreci-
ated how the shape is rather similar. Moreover,
both dependences tend to level out at high energies
at similar cross-section values (some 4.5 · 10ꢀ16
cm2 for (Hg–Na)þ [21] and about 2.5 · 10ꢀ16 cm2
for (Mg–Na)þ).
3. Experimental results
3.1. Total cross-sections
In the collision between Mg and Naþ no
wavelengths attributable to direct magnesium ex-
citation were observed in the whole energy range
studied, and among the possible neutral sodium
emissions only the one corresponding to the decay
of the first excited state formed by electron transfer
from Mg could be measured:
þ
þ
1
1
Mgð31SÞ þ Na ð S0Þ ! Nað32P1=2;3=2Þ þ Mg ð S0Þ
Nað32P1=2;3=2Þ ! Nað32S1=2
Þ
k ¼ 5889–5895A
For this transition, total excitation cross-section
dependence on collision energy is given in Fig. 1.
Such a result seems somewhat unusual since for
this transition signal strength is high enough
(experimental threshold is well below 500 eV in
the laboratory frame of reference) that, at least
in alkali–alkali collisions other, higher energy
3.2.
State-to-state
cross-sections
for
the
Na(32P1=2 ; 3=2) ! Na(32S1=2) decay
For this transition, the relatively strong signal
obtained has permitted to increase the resolution to
allow separation of the two J ¼ 1=2 and J ¼ 3=2
states and thus the study of their relative popula-
tion dependence on collision energy, the corre-
sponding branching ratio being shown in Fig. 2. It
can be readily appreciated how in the energy range
studied all values lie below the statistical value of
two, meaning that the J ¼ 1=2 state is relatively
Fig. 1. Total emission cross-section vs. collision energy (in the
laboratory frame of reference) for the Na(32P1=2;3=2) !
Na(32S1=2) decay in Mg–Na+ collisions. Full line represents
fitting by a LZS rotationally coupled model, dotted line the
same by radially coupled one.

J. de Andrꢀes et al. / Chemical Physics Letters 382 (2003) 106–111
109
effect was observed for this spectral line, which can
only arise from the non-statistical population of its
magnetic sublevels. However, a lesser effect was
also observed for the Na(32P1=2) ! Na(32S1=2) de-
cay which, being of the 1/2–1/2 type, should not to
show any polarization at all [14], but this can be
justified considering that many components of the
optical systems (monochromator, lenses and oth-
ers) may act as polarizers distorting the fluorescent
emission. This effect was exploited to quantify this
distortion, using this information to correct the
Na(32P3=2) ! Na(32S1=2) data and to obtain its true
polarization fraction values (as usually defined
[17]) as well as their energy dependence, which is
shown in Table 1. It can be seen that there is a
marked negative dependence on collision energy
and that the fraction inverts at some 2400 eV.
When measuring de-excitation cross-sections,
these must be corrected for possible anisotropic
effects in order to obtain the true excitation cross-
section. This can be done by means of the polari-
zation fraction [17] and such corrected values are
represented along with the originals in Fig. 3. The
very low strength intensity measured when the
polarizer is in place made it impossible to measure
polarization fractions at the lower collision energy
values, where it can be readily appreciated that, for
all points for which both values could be deter-
mined, divergence between corrected and original
data is negligible.
Fig. 2. Branching ratio r(Na(32P3=2))/r(Na(32P1=2) vs. collision
energy (in the laboratory frame of reference) for the
Na(32P1=2;3=2) ! Na(32S1=2) decay in Mg–Naþ collisions.
more populated than was to be expected according
to a statistical behavior. There is a marked energy
dependence (the most directly comparable alkali–
alkali collision in terms of mass, Na + Naþ, shows a
negligible dependence for the same spectral line
[14]) and signs of an oscillatory structure which,
however, is far too small when compared to error
bars to be interpreted.
3.3. Polarization of the Na(32P3=2) ! Na(32S1=2
)
Polarization measurements contain also infor-
mation about magnetic sublevel population, and it
is possible to obtain magnetic cross-section values
by using angular momenta algebra (see [18] for
fluorescent emission
When a linear polarizing filter was installed in
the systemÕs light path, a marked polarization
Table 1
Polarization fractions as a function of collision energy for the Na(32P3=2) ! Na(32S1=2) decay in Mg–Naþ collisions
E (eV)
Polarization fraction
E (eV)
Polarization fraction
E (eV)
Polarization fraction
1000
1100
1200
1300
1400
1500
1600
1700
1800
)0.24
)0.23
)0.07
)0.29
0.05
1900
2000
2100
2200
2300
2400
2500
2600
2700
0.10
)0.10
0.01
2800
2900
3000
3100
3200
3300
3400
3500
0.05
0.10
0.12
0.09
0.11
0.11
0.14
0.16
0.02
0.00
0.07
0.09
0.01
)0.08
0.05
0.10
0.04
)0.02

110
J. de Andrꢀes et al. / Chemical Physics Letters 382 (2003) 106–111
at higher collision energies. Unfortunately, a
comparison with the corresponding one for the
charge transfer processes in Hg–Naþ and Cd–Naþ
[20–22] is not possible because authors report only
on the magnetic cross-sections energy dependence
for neutral target excitations.
4. Discussion
A first approximation to an interpretation of
the experimental results can be obtained by ap-
plying to them a simple two-state model such as
that of Landau–Zener–Stuckelberg (LZS) [14]
even when, as is the case here, collision processes
involve a manifold of states which correlate with
different excited channel products. In Fig. 1 are
represented the results of such a fitting: It can be
readily appreciated that, at low energies, a rota-
tionally coupled LZS model [9] manages to fit
quite well the experimental results. This could
mean that the process is essentially controlled by a
R–P type coupling, the calculated threshold en-
ergy of 137.7 eV lying far outside the range of the
experimental setup. The proportionality constant
4.12 · 10ꢀ18 eVꢀ1=2 cm2 is of the same order of
magnitude as those observed in alkali–alkali col-
lisions in which this spectral line is also observed
[14].
Fig. 3. Emission (circles) and excitation (triangles) cross-sec-
tions as a function of collision energy.
At the unusually low energy of 1700 eV in the
laboratory frame (when compared with alkali–al-
kali collisions) the fitting becomes invalid and it is
necessary to use a LZS radially coupled model in
order to obtain a good fit for the remaining ex-
perimental data. Moreover, the transition zone
coincides with the minimum already mentioned,
although calculated threshold energy has a much
lower value of only 1492.5 eV, while the propor-
tionality constant has a value of 1.92 · 10ꢀ14
eVꢀ1=2 cm2, well within usual alkali–alkali values
for this process. The presence of both kinds of
coupling in the energy range considered may mean
either that the electron capture process involves
essentially both R–R coupling and a R–P one, or
that the global behavior of this system is such that
a simple two-state model is unable to describe it
properly. Therefore it may be necessary to take
into account the full manifold of states involved
Fig. 4. Magnetic cross-sections dependence on collision energy
for Na(32P3=2) formation in Mg–Naþ collisions.
details). Magnetic cross-sections for both absolute
values of the Mj components (jMjj) are represented
in Fig. 4. It is interesting that, like similar pro-
cesses in Rb–Rbþ, Rb–Naþ and Na–Rbþ colli-
sions, their energy dependence is positive and that
population of the jMjj ¼ 3=2 state is favoured

J. de Andrꢀes et al. / Chemical Physics Letters 382 (2003) 106–111
111
References
and the corresponding couplings between them in
order to get a meaningful interpretation of the
experimental results.
[1] D.L. Lambert, Phys. Scr. T47 (1993) 186.
[2] H. Daley, J. Perel, Phys. Rev. A 4 (1971) 162.
[3] V. Aquilanti, P. Casavecchia, G. Grossi, J. Chem. Phys. 71
To the best of our knowledge the most elabo-
rated theoretical study on the (Mg–Na)þ system
[19], reports on the potential energy curves for five
of six low-lying asymptotic states at different levels
of ab initio calculations as well as the radial and
rotational couplings between some of them. Ac-
cording to such results, we should have been able
to observe fluorescent emissions from excited
magnesium atoms in our experiments, but as it has
been already said, this was not the case. It seems,
therefore that more theoretical effort needs to be
devoted to the study of the ion–atom systems in-
volving more than one valence electron, including
dynamical calculations. This is especially true in
the case of the (Mg–Na)þ system, for which high-
quality ab initio calculations and dynamical stud-
ies are being developed by our research group.
(1979) 3547.
[4] N.H. Tolk, C.W. White, S.H. Neff, W. Lichten, Phys. Rev.
Lett. 11 (1973) 671.
[5] J.Ø. Olsen, N. Andersen, T. Andersen, J. Phys. B 10 (1977)
1723.
[6] B. Brunetti, F. Elisei, C. Vergari, Nuovo Cimento 63 (1981)
98.
€
[7] A. Bharing, I.V. Hertel, E. Meyer, H. Schmidt, Phys. Rev.
Lett. 53 (1984) 1433.
[8] W. Thomsen, N. Andersen, E.E.B. Campbell, I.V. Hertel,
S.E. Nielsen, J. Phys. B: At. Mol. Opt. Phys. 31 (1998) 3429.
[9] V. Aquilanti, Z. Physik. Chem. 90 (1974) 1.
[10] C.F. Melius, W.A. Goddard III, Phys. Rev. Lett. 29 (1972)
975.
[11] W.L. McMillan, Phys. Rev. A 4 (1971) 69.
[12] R. Shingal, J.C. Noble, B.H. Brandsen, D.R. Flower,
J. Phys. B 19 (1986) 3951.
[13] H. Nakamura, C. Zhu, J. Chem. Phys. 101 (1994) 10630.
[14] J. Sogas, M.E. Aricha, J. de Andrs, M. Albert, J.M. Lucas,
A. Aguilar, Phys. Chem. Chem. Phys. 3 (2001) 3638.
[15] A. Salichs, Master Thesis, Universitat de Barcelona, 2000.
[16] D. Leep, A. Gallagher, Phys. Rev. A 13 (1976) 751.
[17] N. Bendali, D.H. Duong, P. Juncar, J.M.St. Jalm, J.L.
Vialle, J. Phys. B 19 (1986) 233.
Acknowledgements
This work has been supported by the Spanish
DGICYT project PB97-0919, the Spanish Minis-
try of Science and Technology project BQU2001-
3018 and the Generalitat de Catalunya-DURSI
project 2000SGR 00016. M.S. thanks to the
Universitat de Barcelona for a research grant. J. de
A. wishes to thank also the personnel of Optica
Cottet for their help in solving a technical
problem.
[18] S.T. Chen, A.C. Gallagher, Phys. Rev. A 17 (1978) 571.
[19] J.P. Bruna, S.D. Peyerimhoff, in: I. Prigogine, S.A. Rice
(Eds.), Advances in Chemical Physics, vol. LXVII, John
Wiley & Sons, Chichester, 1987, p. 1.
[20] V. Aquilanti, P. Casavecchia, Chem. Phys. Lett. 47 (1977)
288.
[21] V. Aquilanti, P. Casavecchia, G. Grossi, J. Chem. Phys. 68
(1978) 1499.
[22] V. Aquilanti, P. Casavecchia, G. Grossi, J. Chem. Phys. 71
(1979) 3546.