You are on page 1of 6

12 March 1999

Chemical Physics Letters 302 1999. 4954

Quasiclassical molecular dynamic calculations of vibrationally


and rotationally state selected dissociation cross-sections:
N q N2 , j / 3N
F. Esposito, M. Capitelli

Centro di Studio per la Chimica dei Plasmi del CNR, Dipartimento di Chimica, Uniersita degli Studi, ia Orabona 4, 70126 Bari Italy
Received 22 December 1998; in final form 22 December 1998

Abstract
Complete sets of dissociation cross-sections for the process N q N2 , j . 3N have been calculated by using a
quasiclassical trajectory method over a potential energy surface developed by Lagana et al. J. Phys. Chem. 91 1987. 312..
The results have been then used to calculate the global dissociation rate of the process for thermal conditions. The results
show an Arrhenius behaviour and are in good agreement with experimental values. q 1999 Elsevier Science B.V. All rights
reserved.

1. Introduction
Non-equilibrium vibrational kinetics is a topics of
large interest due to its interconnection with plasma
chemistry, plasma physics, laser chemistry and dynamics of expanding flows w1,2x.
Significant efforts have been paid in the last years
to refine input data and of to rationalise whole
kinetics. Concerning the refinement, several calculations have been performed for VV vibration
vibration. and VT vibrationtranslation. rates involving moleculemolecule and moleculeatom
closed-shell. interactions. Systematic studies of the
VT processes involving open-shell atoms interacting with vibrationally excited molecules have been
carried out by Lagana et al. w3,4x. However, little
attention has been paid to dissociation processes,

Corresponding author. Fax: q39-080-5442024

although they play an important role in quasi-equilibrium and non-equilibrium vibrational kinetics.
In this Letter we present quasiclassical dynamic
calculations for the process:
N q N2 , j . 3N ,

1.

i.e., for the interaction of nitrogen atoms with vibrationally and rotationally excited N2 molecules and
j are, respectively, vibrational and rotational quantum numbers.. For the calculation, we make use of
the same potential energy surface PES. obtained
and then used by Lagana et al. w3,4x to get an
extended set of VT rates, i.e., for the process
N q N2 , j . N q N2 X , all jX . .

2.

The goal of our calculations is to construct a


complete set of cross-sections that is from each
possible , j . state compatible with the used PES.
for process 1.. These cross-sections can be used to
calculate rate coefficients by averaging on proper

0009-2614r99r$ - see front matter q 1999 Elsevier Science B.V. All rights reserved.
PII: S 0 0 0 9 - 2 6 1 4 9 9 . 0 0 0 9 9 - 8

50

F. Esposito, M. Capitellir Chemical Physics Letters 302 (1999) 4954

translational and rotational distribution functions.


Rate coefficients reported here have been obtained
by using Boltzmann distributions not only for translational and rotational degrees of freedom but also
for vibrational states. As a final result we obtain a
thermal dissociation rate which can be compared
with the experimental data. This comparison is rather
encouraging thus confirming the judicious construction of the PES by Lagana et al., and the accuracy of
quasiclassical method used in the present work.

2. Method of calculation
As a first step we calculated the allowed maximum value jmax of the rotational quantum number
for which the asymptotic diatomic channels of the
LEPS PES of Lagana w3x have still a relative minimum at finite internuclear distance. Then we calculated the maximum vibrational state compatible with
each value of rotational quantum number by using
the WKB method w5x, obtaining in this way the
complete set of possible initial states for molecular
nitrogen, including quasibound states classically
bound but subject to quantum tunnelling.. Collision
partners initially in a quasibound state are of particular importance for modelling the recombination process w6x.
A quasiclassical trajectory method code has been
written and tested, initially following the standard
procedure w5,710x. A RungeKutta fourth-order integrator was used with variable timestep size con and a velocity
trolled by a space interval of 0.025 A

interval of 0.01 Arfs. The method of controlling


step size is that of Ref. w11x, taking the minimum
step suggested by control routine for each component in phase space. A total energy error check
6 = 10y4 eV. and a 1 in 10 back-integration check
of trajectory accuracy were performed. A stratified
was adopted.
sampling of impact parameter up to 4 A
Trajectory results were weighted using the method
used in Ref. w5x and discussed extensively in Ref.
w12x. The numerical code is described in detail in
Ref. w13x, where some statistical experiments and
comparisons are presented using different weighting
functions. The final form of our code reflects the
logical need of uniforming the quasiclassical treat-

ment pseudo-quantization. of reagents and products.


In standard QCT method, reagents are prepared by
selecting only integer values of rovibrational actions,
while products, obviously non-quantized being the
dynamics purely classical, are analysed essentially
by nearest-integer or some weighting procedure w5x.
On the contrary, in our code reagents are prepared by
assigning rovibrational actions of the molecule in
continuous intervals around quantal values of interest, then weighting both reagents and products in the
same way. This procedure is slightly more demanding in computational terms because internal molecular energy, necessary in the preparation of reagents,
cannot be tabulated as in the standard procedure
because of continuous values of rovibrational actions. To increase the accuracy we chose not to
interpolate this initial energy by the WKB method.
for each trajectory but to calculate it explicitly.
A hypothesis of crucial importance for reducing
computational cost has been adopted in this work. In
trajectory calculations, only 1 in 25 integer values of
rotational quantum number is considered, in the interval 0250, for all possible vibrational quantum
number values. The hypothesis is that the logarithm
of rate coefficient results can be well approximated
by interpolating or fitting with a simple function,
the same for all the states. on j values. Really, we
performed some experiments in this sense, with a
very good result for dissociation rate coefficients: in
this case, a parabolic fitting of logarithm of rate
coefficients works very well see Fig. 1., while
dependence from j is quite low for rate coefficients

Fig. 1. Values of rate coefficients plotted as a function of rotational quantum number for different vibrational levels full lines
represent an interpolation curve..

F. Esposito, M. Capitellir Chemical Physics Letters 302 (1999) 4954

relative to process 2.. We adopted in this work only


linear interpolation of all rate coefficients up to the
higher j value of our module 25 selection, but we
plan to fit them with polynomials of second for
dissociation. and third for energy exchange. degree
up to the real maximum j value compatible with
each vibrational state w9,10x. However, neglecting
rate coefficients of highest j values at most the last
25 levels on a total number of 261. does not introduce generally large errors, because their weight in
Boltzmann distributions of interest is quite limited.
This point can change in non-equilibrium distributions.
Particular attention has been devoted to threshold
energy problems affecting cross-sections. Instead of
choosing discrete values, we adopted a continuous
variation of translational energy between 10y3 and 6
eV a procedure cited in Ref. w14x.
We obtained a highly noisy kind of cross-section
by using a large number of bins 300. on energy
axis; then we treated the results with a smoothing
procedure essentially the windowed median described in Ref. w15x. applied three times, with five
bins of radius half-width of smoothing., reduced
in proximity of limits of energy interval to the
distance from extrema.
This procedure seems to produce quite realistic
plots of cross-sections, particularly in the threshold
energy region, with the exception of the extrema of
energy axis where smoothing has a limited action.
We do not follow the suggestion from Ref. w15x of
using asymmetrical intervals of smoothing with a
constant width at the extrema because in this context
this procedure can really introduce large errors for
low-energy threshold cross-sections.

3. Results
Cross-sections for process 1. are reported in Fig.
2a,b as function of translational energy for selected
rovibrational states. In particular Fig. 2a shows the
effect of for j s 0, while in Fig. 2b the effect of
varying j with s 40 is reported. In both cases, as
expected, we observe a large increase of cross-section values with and j, and a corresponding decrease of the threshold energy.

51

Fig. 2. a,b: Cross-sections as a function of the translational energy


Etr for different values js 0, a.. and for different J values
s 40, b...

In Fig. 3 we show rate coefficients, labelled with


Eint , the initial state internal energies of which are
close within 0.01 eV. to y4, y2 and 0 eV under
the dissociation limit 9.91 eV.. Rate coefficients are
plotted as a function of quantum number j on the
same axis is decreasing in order to keep constant
the internal energy of initial state.. As it is apparent
from Fig. 3, up to about j s 75 rate coefficient
values are approximately invariant, especially at low
energies.
Beyond j s 100, rate coefficient values invariably
tend to decrease, especially for the higher energies.
We also considered selection of states based on
energy difference Ediff . between the top of the
rotational barrier and initial state energy, in order to
avoid the inhibition of dissociation due to rotational
barrier. In this manner rate coefficients labelled
Ediff ., invariably decreasing with j, are higher than
those in the preceding case but for the lowest inter-

F. Esposito, M. Capitellir Chemical Physics Letters 302 (1999) 4954

52

It is also possible to average the results of 4. on a


Boltzmann distribution for the vibrational levels at
the temperature T s T :
k diss T . s Qy1

exp yE , js0rkT . k diss . .

5.

Fig. 3. Rate coefficient as a function of rotational quantum


number at fixed internal energy Eint is the energy of the initial
state measured from the dissociation limit; Ediff is Eint but
measured from the top of the rotational barrier..

nal energy value. In this last case no variation is


visible at all. The consequence of this observation is
that rovibrational states of similar internal energy do
not show similar values of dissociation rate coefficients, especially for high values of internal energy.
Rate coefficients of process 1. are obtained by
averaging cross-sections using a Maxwell velocity
distribution function:
k diss , j . s k 3 T 3 pmr8 .

y1 r2

=Hd E E exp yErkTtr . s , j E . .

3.

Vibration-dependent rate coefficients are obtained


by averaging on a Boltzmann distribution function of
rotational states:
k diss . s Qy1
r

Fig. 5a reports k diss as a function of 1rT. The plots


show an Arrhenius behaviour, with an activation
energy of 9.36 eV in the interval 1000020000 K.
The Arrhenius behaviour is justified by the use of
equilibrium distributions. In Fig. 5a experimental
data w16x referred to thermal conditions are shown.
The agreement between calculated and measured
values is good, especially at lower temperatures. The
larger deviation at high temperatures could be due to
deviations from Boltzmann vibrational distributions
because of the dissociation reaction at intermediate
and high vibrational levels.
In Fig. 5a we also reported dissociation rates for
process 1. obtained by Doroshenko et al. w17x based
on a Russian compilation as well as the results
obtained by the ladder climbing model w18x. This
model introduces a pseudo-level above the last bound
level of the diatomic molecule in order to take into
account dissociation. We can note that the ladder
climbing model underestimates the rate coefficients
with respect to the experimental data, while the
opposite is true for the results of Doroshenko et al.
The good agreement between the present results
and experimental data is an indirect proof of the

g j 2 j q 1.
j

=exp yE , jrkTrot . k diss , j . ,

4.

with g j being nuclear degeneracy 2 for even and 1


for odd rotational states., and Q r being rotational
partition function.
Energies of rotational states for each vibrational
quantum number are directly calculated from the
diatomic potential adopted in this work and inserted
in expression 4..
Rate coefficients for selected values of are
reported in Fig. 4 as a function of the roto-translational temperature T s Ttr s Trot . The figures shows a
strong increase of the rate coefficients with the temperature and with the initial vibrational state.

Fig. 4. Rate coefficients averaged on a Boltzmann distribution


over rotational states as a function of temperature for different
values.

F. Esposito, M. Capitellir Chemical Physics Letters 302 (1999) 4954

53

These results are in good agreement with experimental value 5 = 10y1 3 . w19x.

4. Conclusions
In this Letter we have presented some results
obtained by means of the quasiclassical trajectory
method for the calculation of a complete set of
cross-sections for dissociation process of rovibrationally excited molecular nitrogen induced by collision with nitrogen atoms. Good agreement has been
found between the present results averaged on a
Boltzmann vibrational distribution and experimental
data for both dissociation and exchange reactions.
This agreement is an indication of the reliability of
the PES proposed by Lagana et al. and of molecular
dynamics method used in this work.
Work is now in progress to insert either the
calculated rates in vibrational kinetic codes describing equilibrium and non-equilibrium conditions, and
the cross-sections in direct Monte Carlo simulations.

Fig. 5. Global dissociation a. and exchange b. rate constants as a


function of 1r T for the different values see text..

reliability of the PES proposed by Lagana et al. In


addition, we also calculated cross-sections and thermal rate coefficients for the exchange process:
N q N2 N2 q N .

Acknowledgements
This work has been partially supported by ESA
Contract ESA No. 12736r97rNLrPA Advanced
Prediction Methods for Plume Flow. and by MURST
under Project No. 9703109065006.. The authors
thank A. Lagana for useful advice.

6.

In Fig. 5b our thermal rate coefficient linearly interpolated is reported as a function of 1rT and compared with experimental value w19x at 3400 K and
previous theoretical calculations w20x.
The agreement is very satisfactory, confirming
not only the good choice of PES by Lagana but also
globally the reliability of the molecular dynamics
method used in this work. Thermal rate coefficient
value is slightly smaller 6.8 = 10y13 cm 3rs
molecule. than the Lagana et al. value 7.1 = 10y1 3 .
obtained with the standard QCT method and averaging over initial conditions without first calculating
cross-sections, and to reduced dimensionality quantal
result RIOS. of Lagana et al. 5.6 = 10y1 3 . w19x.

References
w1x M. Capitelli Ed.., Non-Equilibrium Vibrational Kinetics,
Topics in Current Physics, vol. 59, Springer, Berlin, 1986.
w2x M. Capitelli Ed.., Molecular Physics and Hypersonic Flows,
NATO ASI Series C482, Kluwer, Dordrecht, 1996.
w3x A. Lagana,
E. Garcia, L. Ciccarelli, J. Phys. Chem. 91 1987.
312.
w4x A. Lagana,
E. Garcia, J. Phys. Chem. 98 1994. 502.
w5x N.C. Blais, D.G. Truhlar, J. Chem. Phys. 65 1976. 5335.
w6x R.E. Roberts, R.B. Bernstein, C.F. Curtiss, J. Chem. Phys. 50
1969. 5163.
w7x M. Karplus, R.N. Porter, R.D. Sharma, J. Chem. Phys. 43
1965. 3259.
w8x R.N. Porter, L.M. Raff, in: W.H. Miller Ed.., Dynamics of
Molecular Collisions, Part B, Chap. 1, Plenum Press, New
York, 1976, pp. 152.

54

F. Esposito, M. Capitellir Chemical Physics Letters 302 (1999) 4954

w9x M.D. Pattengill, Rotational Excitation III: Classical Trajectory Methods, in: R.B. Bernstein Ed.., AtomMolecule
Collision Theory, Plenum, New York, 1979.
w10x D.G. Truhlar, J.T. Muckerman, Reactive Scattering CrossSections III: Quasiclassical and Semiclassical Methods, in:
R.B. Bernstein Ed.., AtomMolecule Collision Theory,
Plenum, New York, 1979.
w11x W.H. Press, B.P. Flannery, S.A. Teukolsky, V.T. Vetterling,
Numerical Recipes, Chap. 15, Integration of Ordinary Differential Equations, Section 2, Adaptive Stepsize Control for
RungeKutta, Cambridge University Press, 1986.
w12x D.G. Truhlar, B.P. Reid, D.E. Zurawsky, J.C. Gray, J. Chem.
Phys 85 1981. 786.
w13x F. Esposito, Ph.D. Thesis, University of Bari in preparation..
w14x F.J. Aoiz, L. Banares,
V.J. Herrero, V.S. Rabanos,
I. Tanarro,

J. Chem. Phys. 101 1997. 6165.

w15x W.H. Press, B.P. Flannery, S.A. Teukolsky, V.T. Vetterling,


Numerical Recipes, Chap. 13, Statistical Description of Data,
Section 9, Smoothing of Data, Cambridge University Press,
1986.
w16x Experimental results from J.P. Appleton, S. Byron, R.K.
Hanson, D. Baganoff are collected, in: C. Park Ed.., Two
Temperature Interpretation of Dissociation Rate Data for N2
and O 2 , AIAA-88-0458.
w17x V.M. Doroshenko, N.N. Koudryavtsev, S.S. Novikiv, V.V.
Smetanin, High Temp. 28 1990. 82.
w18x I. Armenise, M. Capitelli, C. Gorse, J. Thermophys. Heat
Transfer 12 1998. 45.
w19x A.N. Wright, C.A. Winkler, Active Nitrogen, Academic
Press, New York, 1968.
w20x M. Capitelli Ed.., Molecular Physics and Hypersonic Flows,
NATO ASI Series C482, Kluwer, Dordrecht, 1996, p. 35.

You might also like