Vous êtes sur la page 1sur 5

Chemical Physics Letters 619 (2015) 158162

Contents lists available at ScienceDirect

Chemical Physics Letters


journal homepage: www.elsevier.com/locate/cplett

Lithium(I) in liquid ammonia: A quantum mechanical charge eld


(QMCF) molecular dynamics simulation study
Niko Prasetyo a , Lorenz R. Canaval c , Karna Wijaya a,b , Ria Armunanto a,b,
a

Austrian-Indonesian Centre (AIC) for Computational Chemistry, Gadjah Mada University, Sekip Utara, Yogyakarta 55281, Indonesia
Department of Chemistry, Faculty of Mathematics and Natural Sciences, Gadjah Mada University, Sekip Utara, Yogyakarta 55281, Indonesia
Theoretical Chemistry Division, Institute of General, Inorganic and Theoretical Chemistry, University of Innsbruck, Innrain 80-82,
A-6020 Innsbruck, Austria
b
c

a r t i c l e

i n f o

Article history:
Received 7 November 2014
In nal form 28 November 2014
Available online 4 December 2014

a b s t r a c t
The solvation of Li(I) in liquid ammonia has been investigated by an ab initio quantum mechanical chargeeld molecular dynamics (QMCF-MD) simulation. Being the rst simulation of a metal cation in liquid
ammonia employing this methodology, the work yields a wide range of accurate structural and dynamical
data. Li(I) is tetrahedrally coordinated by four ammonia molecules in the rst solvation shell at a distance
Two ligand exchange attempts have been observed within 12 ps of simulation time. The second
of 2.075 A.
solvation shell shows a more labile structure with numerous successful exchanges. The results are in
excellent agreement with experiments.
2014 Elsevier B.V. All rights reserved.

1. Introduction
Lithium is a metal of high interest in industry because of its
applications in modern technologies such as lithium batteries,
which are an essential part of mobile electronic devices [1]. In
academia, lithium is often used as a starting point to assess new
theoretical methods in computational chemistry because of its simplicity.
The solvation of lithium in liquid ammonia has been subject
of standard computational chemistry methods such as ab initio
HartreeFock [2] and MllerPlesset-2 (MP2) [3], ab initio self
consistent eld for molecular interactions (SCF-MI) [4] and density functional theory (DFT) calculations [5] Various simulation
methods such as molecular dynamics simulations [3,6], a quantum
mechanical molecular mechanical molecular dynamics (QM/MMMD) simulation [7] and a Monte Carlo simulation using a three
body potential [8] have previously been reported. Experimentally,
neutron diffraction [9,10] and X-ray diffraction [11], infrared spectroscopy [12] and inelastic X-ray scattering [13] have been used to
investigate the solvation properties of lithium in liquid ammonia.
Ab initio calculations in gas phase cannot yield any information
about dynamical properties and solvent inuence of the system
at hand. Both the pair potential approach and the Monte Carlo

simulation showed results differing from experimental data, and


QM/MM simulations have not offered details of dynamical properties such as mean residence times of ligands, ion-ammonia
vibrational spectra and detailed insights of the second solvation
shell.
Nowadays, computer resources are sufcient to perform simulations with a large quantum mechanical region to analyze the
structural and picosecond dynamical properties of a metal ion plus
its solvation layers accurately. The quantum mechanical chargeeld molecular dynamics (QMCF-MD) method has proven to be an
accurate method to study highly charged ions [14,15], heavy metal
ions [16,17], oxo-anions [18] small molecules [1922] and metalorganic [23,24] compounds in aqueous solution. To the best of our
knowledge, no application of the QMCF-MD method on metals in
liquid ammonia has been reported so far. In this letter, we present
the rst application of the QMCF framework to investigate Li(I) in
liquid ammonia. From the simulation data, structural properties
such as radial distribution functions (RDFs), angular distribution
function (ADFs) and coordination number distribution (CNDs) are
evaluated along with mean residence times (MRTs) of ligands and
ion-ammonia vibrational frequencies.

2. Method
Corresponding author at: Austrian-Indonesian Centre (AIC) for Computational
Chemistry, Gadjah Mada University, Sekip Utara, Yogyakarta 55281, Indonesia.
E-mail address: ria.armunanto@ugm.ac.id (R. Armunanto).
http://dx.doi.org/10.1016/j.cplett.2014.11.066
0009-2614/ 2014 Elsevier B.V. All rights reserved.

The QMCF-MD methodology [25] is an enhancement of conventional QM/MM approaches [2629], which does not require
any potential functions for any species in the system investigated

N. Prasetyo et al. / Chemical Physics Letters 619 (2015) 158162

except of the solventsolvent interactions. The scheme allows to


neglect non-Coulomb interactions between the core zone and the
MM region. The forces in the different zones are calculated according to Eqs. (1)(3). This is the main advantage of the QMCF ansatz,
because the construction of such potentials is a very difcult and
time consuming labor, sometimes being impossible. The QM region
in the QMCF method is enlarged and consists of two subregions: the
core zone and the layer zone. The core zone contains the solute and
the rst solvation shell, whereas the layer region contains further
solvent molecules [30], which results in a higher computational
effort.
FJcore = FJQM
layer

FJ


M

qMM
qQM
I
J

 qMM
qQM
I
J

N1 +N2

rIJ2

I=1

1+2

rIJ2

I=1

N2


  
+ 1 rIJ 3
2 1

FIJnC

rc

+ FIJnC

(2)

(3)

I=1

/ J
I=
where FJcore corresponds to the quantum mechanical forces acting
layer

on a particle J in the core region. FJ

corresponds to the quantum

mechanical forces acting on a particle J in the layer region and FJMM


corresponds to the forces acting on a particle J in the MM region. To
ensure a smooth transition of the particles between the QM and the
MM region, a small smoothing layer is dened (Eq. (4)), S(r) being
a smoothing function as described in literature [25].
layer

FJsmooth = S(r)(FJ

= FJMM ) + FJMM

(4)

The MRTs of the ligands belonging to the solvation shells of lithium


were evaluated by the direct method (Eq. (5)) [31].
 0.5 =

tsimulation Nav
0.5
Nex

(5)

tsimulation represents the total simulation time, Nav is the average


0.5
coordination number of a particle in the respective shell and Nex
is the number of exchange events persisting 0.5 ps. A sustainability
coefcient can be dened as
Sex =

0.5
Nex

(6)

0.0
Nex

0.0 is the number of all transitions through a shell boundwhere Nex


ary. The inverse of sustainability coefcient (Rex ) indicates the
number exchange attempts required for a successful exchange.
The ion-ammonia vibrational spectrum was evaluated using
velocity auto-correlation functions (VACFs) C(t) dened as

Nt N
C(t) =

Nt N

Table 1
Characteristic values of LiN and LiH radial distribution functions obtained from
QMCF-MD simulation.a

Atom pair
N
Li
Li
H

rM1

rm1

N(m1)

rM2

rm2

N(m2)

2.075
2.625

3.08
3.28

4.05
12.82

4.82
4.88

6.60
6.40

25
75

a
rMi and rmi are the distances in of the ith maxima and minima observed in the
corresponding RDF. N(mi) is the average coordination number of the ith shell.

2.1. Simulation protocol

(1)

= FJQM +

FJMM =

159

vJ (tI )vJ (tI + t)

Nt
J

(7)

vJ (tI )vJ (tI )

N is the number of particles, Nt is the number of time origins,


tI and vJ denotes a certain velocity component of particle J. The
power spectrum was obtained by a subsequent Fourier transformation. Ion-ammonia vibrational frequencies were computed using
the approximative normal coordinate analysis [32]. Frequencies
obtained via HartreeFock calculations are often scaled by 0.89
in order to correct the missing electron correlation and the inuence of vacuum environment [33,34]. However, in this study the
values were not scaled because our simulation included explicit
solvent that provided a realistic non-vacuum environment, and the
inuence of electron correlation is expected to be small.

For the QMCF-MD simulation, a cubic box with side length


of 20.7 A containing one Li(I) ion and 215 ammonia molecules
was used. Periodic boundary conditions, the minimum image
convention and an NVT ensemble were applied. The simulation
temperature was kept constant at 235.15 K using the Berendsen
weak coupling algorithm [35] with a relaxation time of 0.1 ps. The
density of the system was kept constant at 0.690 g/cm3 , corresponding to the density of pure liquid ammonia at 235.15 K. A
exible model of ammonia including intra- and inter-molecular
potentials was used in the MM region [6]. The radii of quantum

mechanical core zone and layer zone were set to 3.4 and 6.6 A,
respectively. A smoothing zone of 0.2 A was applied at the border of the QM and the MM region. All atoms in the QM region
(core zone and layer zone) were treated at the HartreeFock level
of theory using the well established DZP Dunning [36] basis sets. To
integrate the equation of motion, a second-order Adams-Bashforth
predictor-corrector algorithm was used with a time step of 0.2 fs.
the
To correct the cutoff of the long-range interaction above 10.3 A,
reaction eld approach was used [37]. The system was heated to
700 K for 2 ps and then re-equilibrated at 235.15 K for 2 ps. Sampling was done every fth step during a simulation time of 12 ps.
All QM calculations within the simulation were executed with the
TURBOMOLE 5.9 package [38,39]. The popular VMD package was
employed to visually analyze the trajectory [40].
3. Result and discussion
The Li+ N and Li+ H RDFs are displayed in Figure 1a and structural characteristics are listed in Table 1. There are two solvation
shells, the rst one ranging from 1.78 to 3.08 A with an average
coordination number of 4.05. This coordination number is in excellent agreement with data from experiments employing neutron
diffraction [9,10] and X-ray diffraction [11]. The rst LiN RDF
peak located at 2.075 A is also in excellent agreement with experimental values of 2.01 and 2.06 A from X-ray [11] and neutron
diffraction measurements [9,10], respectively (see Table 2). In the
work at hand, the hydrogen bonds between rst and second shell
are described by quantum mechanics, whereas an earlier QM/MM
simulation only included the rst solvation shell in the QM treatment [7], resulting in a slightly enlarged ion-ammonia distance.
Earlier Monte Carlo and pair potential simulations not including
three-body potentials report a wrong coordination number of six
[6,8].
the maximum LiN
The second shell ranges from 3.08 to 6.60 A,
This shell contains approxprobability being located at 4.82 A.
imately 25 ammonia molecules. This value is larger than the
coordination number of the second hydration shell of lithium(I)
in water [1]. Hayama et al. [11] reported a value of 30 for the coordination number and a second shell LiN RDF maximum at 5.50 A
employing X-ray measurements. The weakly pronounced non-zero
valley of the Li+ N RDF between the rst and second shell indicates
few ligand exchange events between these shells. The non-zero
Li+ N RDF value after 6.5 A indicates that the second solvation

160

N. Prasetyo et al. / Chemical Physics Letters 619 (2015) 158162

Figure 1. (a) LiN (solid line) and LiH (dashed line) radial distribution functions including the corresponding integration information, and (b) angular distribution function
NLiN in the rst solvation shell of the Li+ ion.

shell is more labile, and a number of ligand exchange processes


are observed between the second solvation shell and the bulk.
The coordination numbers obtained from the RDFs are average
values. A detailed analysis of the coordination number distribution for the rst and second shell is displayed in Figure 3.
Occurring in 95% of the congurations found during the simulation, the preferred coordination number is four for the rst
solvation shell, indicating a rather stable structure. The probability of a vefold conguration was found to be 5%. For the second
shell, a broader coordination number distribution ranging from 21
to 28 was observed with a maximum at 25. The probability of
each coordination number is less than 30%, which indicates that
the second solvation shell is less stable allowing frequent ligand
exchanges events with the bulk environment (see Figure 4). A
snapshot from the simulation trajectory depicting the tetrahedrally
arranged ammonia molecules of the rst solvation shell is provided
in Figure 3b.
To elucidate the structure of rst shell solvation, an NLiN
ADF analysis was performed (see Figure 1b). A single peak ranging
from 80 to 140 with a maximum at 107.5 indicates a exible
tetrahedral arrangement. The absence of other peaks excludes the
possibility of a four-fold planar coordination geometry. A slightly
pronounced shoulder near 80 represents the small probability of
a [Li(NH3 )5 ]+ conguration. Neutron diffraction [9,10] and X-ray
diffraction [11] experiments conrm that the conguration of the
rst shell solvation is tetrahedral, which was also found employing
a QM/MM simulation [7].

To evaluate the picosecond dynamics of the system, MRT values


for each solvation shell were calculated using the direct method
[31]. No MRT value for the ammonia ligands in the rst shell can
be provided, as no successful exchange events were observed during the simulation. However, two ligand exchange attempts were
found at 2.65 and at 10.95 ps simulation time (Figures 2b and 3a,
marks tA1 and tA2 , respectively). For the second shell, the MRT value
amounts to 2.8 ps. This is slightly higher than the value for liquid
ammonia reported from a QM/MM simulation (1.62 ps) [41] and for
liquid ammonia that contains one NH4 + ion (1.82 ps) [42]. This indicates that Li+ has a stronger interaction with ammonia compared
to NH4 + . A number of 107 successful exchange events were found
for the second shell. The Rex value of rst solvation shell is zero
because no successful ligand exchanges were found. For the second solvation shell, this number amounts to 5.8. This value shows
the strong inuence of Li(I) on liquid ammonia up to the second
solvation shell.
Figure 2b depicts the evolution of the root-mean-squaredeviation (RMSD) of lithium and its rst shell ligands from an
ideal tetrahedral arrangement. A peak at 2.65 ps (mark tA1 ) is
found, which represents the rst ligand exchange attempt (see
also Figure 3a). Along with an increase of the coordination number
from 4 to 5, which indicates an associative exchange mechanism,
a change from the tetrahedral geometry (Figure 3b) to a trigonal bipyramidal ligand arrangement (Figure 3c) is observed. This
geometry is also found for the second exchange attempt at 10.95 ps
(mark tA2 ), however, less perfect due to the events less pronounced

Table 2
Structural properties of rst solvation shell of Li+ : Li+ N distance (d) in and coordination number (CN).
Method

CN

Geometry

System

QMCF-MD simulation

2.075

Tetrahedral

QM/MM simulation

2.15

Tetrahedral

Monte Carlo simulation

Octahedral

Pair potential simulation

2.29

Octahedral

Ab initio SCF-MI calculation


Classical MD
DFT calculation
X-ray diffraction
ND
ND

2.105
2.15
2.01
2.06

4
4
4
4
4
4

Tetrahedral
Tetrahedral
Tetrahedral
Tetrahedral
Tetrahedral
Tetrahedral

Liquid ammonia: 215 ammonia molecules, 1 Li+


QM treatment: rst and second solvation shell
Liquid ammonia: 215 ammonia molecules, 1 Li+
QM treatment: rst solvation shell only
Liquid ammonia: 201 ammonia molecules, 1 Li+
pair potential plus three-body interactions
Liquid ammonia: 215 ammonia molecules, 1 Li+
pairwise interactions
Gas phase Liammonia clusters; HF/6-31+G*
Liquid ammonia; 239.8 K
Gas phase Li(NH3 )4 + -cluster
Liquid ammonia; 200 K; 22 MPMa lithium
Li 21 MPMa in liquid ammonia
Liquid ammonia; 230 K; 21 MPMa lithium

Mole percent metal.

Refs.
ion; 235.15 K;

This work

ion; 235 K;

[7]

ion; 277 K;

[8]

ion; 235 K;

[6]
[4]
[3]
[2]
[11]
[9]
[10]

N. Prasetyo et al. / Chemical Physics Letters 619 (2015) 158162

161

Figure 2. (a) Coordination number distribution of the rst and second solvation shell, and (b) evolution of the root-mean-square-deviation of the rst shell coordination
geometry from an ideal tetrahedral arrangement. Marks tA1 and tA2 indicate the two observed ligand exchange attempts during the simulation time of 12 ps.

Figure 4. Power spectrum of Li+ in liquid ammonia obtained via Fourier transformed
velocity auto-correlation functions from the QMCF-MD simulation. For comparison,
experimentally obtained values are depicted as single.

Figure 3. (a) Evolution of selected Li(I)N distances during the simulation. The bor Marks tA1 and tA2 indicate
derline of the rst and second shell is shown at 3 08 A.
the two observed ligand exchange attempts during the simulation time of 12 ps. A
detailed plot is given for the rst exchange attempt. Snapshots obtained from the
QMCF-MD simulation depicting, (b) an ideal tetrahedral rst shell arrangement, and
(c) a trigonal bipyramidal coordination geometry observed at the exchange attempt
at tA1 .

strength. The RMSD plot shows a second peak located at 9.8 ps,
which corresponds to a strongly squeezed tetrahedral coordination.
The power spectrum of lithium(I) in liquid ammonia is displayed
in Figure 4. The maximum is located at 358 cm1 and a smaller
peak is found at 550 cm1 . This corresponds to force constants of
35 Nm1 and 82.8 Nm1 , respectively. These values are in very
good agreement with the experimentally observed frequencies of
361 and 561 cm1 (35.6 and 86.0 Nm1 ) for the Li(I)N vibrational stretching of [Li(NH3 )n ]+ species in liquid ammonia [43], and

the IR absorptions of 7 LiNH3 (320 cm1 ), LiNH2 D (335 cm1 ) for


lithiumammonia complexes in solid argon [44]. Deviations of the
results obtained from the QMCF-MD simulation from the experimental values may be attributed to the different concentrations and
isotopes, effects of counter ions and different experimental setups
of the systems investigated.
To conclude, a rather stable structure with a coordination number of 4 was observed for lithium(I) in liquid ammonia. In contrast,
lithium(I) in aqueous solution posses a more exible rst hydration
shell [1], the average coordination number being 4.5 and the MRT
value amounting to 2.56 ps [45].
4. Conclusions
The solvation structure and dynamics of Li(I) in liquid ammonia
have been studied using an ab initio QMCF-MD simulation. The
12 ps long simulation shows that the Li(I) ion is coordinated to
four ammonia molecules with a LiN distance of 2.075 A and a
tetrahedral conguration. Structural properties such as RDF, ADF
and CND are in good agreement with experimental data. The rst

162

N. Prasetyo et al. / Chemical Physics Letters 619 (2015) 158162

solvation shell is rather stable, no successful ligand exchange


events occur, compared to the second shell, which is more labile
and shows an average coordination number of 25. The QMCF-MD
simulation has yielded results for structural and dynamics of the
solvated lithium(I) ion, which are in excellent agreement with
available experimental data.
Acknowledgements
Financial support for this work from a MSc grant of the Indonesia
Endowment Fund for Education Ministry of Finance Republik
Indonesia (LPDP Kementrian Keuangan RI) for Niko Prasetyo (S557/LPDP/2013) is gratefully acknowledged. Financial support
from a PhD grant of the Leopold-Franzens-University of Innsbruck
(Rector Univ. Prof. Dr. Dr.hc.mult. Tilmann D. Mrk) for Lorenz R.
Canaval is gratefully acknowledged. This work was supported by
the Austrian Ministry of Science BMWF as part of the UniInfrastrukturprogramm of the Focal Point Scientic Computing at the
University of Innsbruck.
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]

H.H. Loefer, B.M. Rode, J. Chem. Phys. 117 (2002) 110.


E. Zurek, P.P. Edwards, R. Hoffmann, Angew. Chem. Int. Ed. Engl. 48 (2009) 8198.
E.A.E Orabi, G. Lamoureux, J. Chem. Theory Comput. 9 (2013) 2324.
A. Famulari, F. Moroni, M. Raimondi, T. Thorsteinsson, J. Mol. Struct. THEOCHEM
549 (2001) 85.
U. Pinsook, R.H. Scheicher, R. Ahuja, S. Hannongbua, J. Phys. Chem. A 112 (2008)
5323.
S.V. Hannongbua, T. Ishida, E. Spohr, K.Z. Heinzinger, Naturforsch 582 (1988)
572.
T. Kerdcharoen, K. Liedl, B. Rode, Chem. Phys. 211 (1996) 313.
S. Hannongbua, Chem. Phys. Lett. 288 (1998) 663.
J.C. Wasse, S. Hayama, S. Masmanidis, S.L. Stebbings, N.T. Skipper, J. Chem. Phys.
118 (2003) 7486.
H. Thompson, J.C. Wasse, N.T. Skipper, S. Hayama, D.T. Bowron, A.K. Soper, J.
Am. Chem. Soc. 125 (2003) 2572.
S. Hayama, N.T.N. Skipper, J.C. Wasse, H. Thompson, J. Chem. Phys. 116 (2002)
2991.
T.E. Salter, V.A. Mikhailov, C.J. Evans, A.M. Ellis, J. Chem. Phys. 125 (2006) 34302.

[13] C.A. Burns, G. Vank, H. Sinn, A. Alatas, E.E. Alp, A. Said, J. Chem. Phys. 124 (2006)
024720.
[14] T.S. Hofer, A.K.H. Weiss, B.R. Randolf, B.M. Rode, Chem. Phys. Lett. 512 (2011)
139.
[15] O.M.D. Lutz, T.S. Hofer, B.R. Randolf, A.K.H. Weiss, B.M. Rode, Inorg. Chem. 51
(2012) 6746.
[16] L.R. Canaval, A.K.H. Weiss, B.M. Rode, Comput. Theory Chem. 1022 (2013) 94.
[17] A.O. Tirler, A.K.H. Weiss, T.S. Hofer, J. Phys. Chem. B 117 (2013) 16174.
[18] A.B. Pribil, T.S. Hofer, V. Vchirawongkwin, B.R. Randolf, B.M. Rode, Chem. Phys.
346 (2008) 182.
[19] O.M.D. Lutz, C.B. Messner, T.S. Hofer, L.R. Canaval, G.K. Bonn, C.W. Huck, Chem.
Phys. 435 (2014) 21.
[20] S.T. Moin, L.H.V. Lim, T.S. Hofer, B.R. Randolf, B.M. Rode, Inorg. Chem. 50 (2011)
3379.
[21] S.T. Moin, A.K.H. Weiss, B.M. Rode, Comput. Theory Chem. 1034 (2014) 85.
[22] A.K.H. Weiss, T.S. Hofer, RSC Adv. 3 (2013) 1606.
[23] O. Lutz, C. Messner, T.S. Hofer, M. Gltzle, C.W. Huck, G.K. Bonn, B.M. Rode, J.
Phys. Chem. Lett. 4 (2013) 1502.
[24] A. Bhattacharjee, A.K.H. Weiss, V. Artero, M.J. Field, T.S. Hofer, J. Phys. Chem. B
118 (2014) 5551.
[25] B.M. Rode, T.S. Hofer, B.R. Randolf, C.F. Schwenk, D. Xenides, V. Vchirawongkwin, Theor. Chem. Acc. 115 (2005) 77.
[26] M.J. Field, P.A. Bash, M. Karplus, J. Comput. Chem. 11 (1990) 700.
[27] D. Bakowies, W. Thiel, J. Phys. Chem. 3654 (1996) 10580.
[28] T.S. Hofer, Pure Appl. Chem. 86 (2014) 105.
[29] A. Warshel, M. Levitt, J. Mol. Biol. 103 (1976) 227.
[30] A. Khan, A.K.H. Weiss, R. Uddin, B.R. Randolf, B.M. Rode, T.S. Hofer, J. Phys. Chem.
A 116 (2012) 8008.
[31] T.S. Hofer, H.T. Tran, C.F. Schwenk, B.M. Rode, J. Comput. Chem. 25 (2004) 211.
[32] P. Bopp, Chem. Phys. 106 (1986) 205.
[33] A. Scott, L. Radom, J. Phys. Chem. 3654 (1996) 16502.
[34] D.J. DeFrees, A.D. McLean, J. Chem. Phys. 82 (1985) 333.
[35] H.J.C. Berendsen, J.P.M. Postma, W.F. van Gunsteren, A. DiNola, J.R. Haak, J.
Chem. Phys. 81 (1984) 3684.
[36] T.H. Dunning, J. Chem. Phys. 53 (1970) 2823.
[37] D.J. Adams, E.M. Adams, G.J. Hills, Mol. Phys. 38 (1979) 387.
[38] R. Ahlrichs, M. Br, M. Hser, H. Horn, C. Klmel, Chem. Phys. Lett. 162 (1989)
165.
[39] S. Brode, H. Horn, M. Ehrig, D. Moldrup, J.E. Rice, R.J. Ahlrichs, Comput. Chem.
14 (1993) 1142.
[40] W. Humphrey, A. Dalke, K. Schulten, J. Mol. Graph. 14 (1996) 33.
[41] A. Tongraar, T. Kerdcharoen, S. Hannongbua, J. Phys. Chem. A 110 (2006) 4924.
[42] A. Tongraar, S. Hannongbua, J. Phys. Chem. B 112 (2008) 885.
[43] D.J. Gardiner, J. Chem. Phys. 59 (1973) 175.
[44] A. Loutellier, L. Manceron, J.P. Perchard, Chem. Phys. 146 (1990) 179.
[45] G.E. Bene, T.S. Hofer, B.R. Randolf, B.M. Rode, Chem. Phys. Lett. 521 (2012) 74.

Vous aimerez peut-être aussi