Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

Volume 177, number I CHEMICAL PHYSICS LETTERS 8 February 1991

Adsorption of xenon in zeolite Y: a molecular dynamics study *


S. Yashonath
Solid State andStructural Chemistry Unit, Indian Institute of Science, Bangalore, 560 012, India

Received 13 March 1990; in final form I9 November 1990

A molecular dynamics calculation of a realistic model of xenon adsorprion in sodium-Y zeolite is reported. The equilibrium
properties such as the energy distribution function, the centre-of-cage-centre-of-mass radial distribution function, and the cage-
occupancy distribution function are obtained. The results are in reasonable agreement with those of earlier calculations. The
dynamics is significantly slower than that observed for other small organic molecules. Xenon exhibits anisotropic motion in the
large cavity.

1. Introduction and computational details. The results and discus-


sion are presented in section 3.

Adsorption in zeolites has attracted much atten-


tion in recent times. The problem is complicated by
2. The details of calculation
the large number of factors influencing the adsorp-
tion in the channels and cages of the zeolites. Among The crystal structure of faujasite has been deter-
these, geometrical as well as the chemical state of the mined by several workers [ 6-8 1. The space group is
zeolite determined by the %/Al ratio and the nature
Fd3m. In the present calculation, the atomic coor-
of the extraframework cations play important roles dinates and the lattice parameter a=24.85 8, re-
in determining the adsorption characteristics. Ad-
ported by Fitch et al. have been used [ 81. The zeo-
sorption of xenon in faujasites has been studied by
lite lattice was kept rigid and only the xenon
Rowlinson and co-workers by direct calculation of coordinates were included in the integration. The
partition functions, and by a grand canonical Monte
number of extraframework sodium cations were 48
Carlo method [ 1,2]. These results have yielded in- per unit cell corresponding to a Si/AI ratio of 3.0. Of
teresting insights into the structure of the adsorbed the 48 cations, 16 are in Si sites and the remaining
fluid, distribution of occupancies and thermody-
32 occupy S,, sites [ 4,8]. In the present calculation,
namic behaviour. Recently, some work on adsorp-
we have adopted the model of Kiselev and co-work-
tion of small organic molecules in faujasites has also ers [ 91 for the zeolite-xenon interaction. The inter-
been reported [ 3-51.
action between the adsorbed xenon and the zeolite
Here, we report a molecular dynamics calculation is confined to the oxygens and the sodiums. This is
of xenon adsorbed in faujasite. Our interest is to study a reasonable assumption in view of the fact that the
the equilibrium and the dynamical behaviour of the
silicon and the aluminium are buried within the ox-
adsorbed xenon in faujasite and to compare our re-
ygen lattice, thus preventing the close approach of
sults with those of earlier calculations. In section 2,
the xenon to these atoms. The xenon-(oxygen, so-
we discuss the details of the intermolecular potential
dium) interaction was taken to be of the Lennard-
Jones form [ 91,

* Contribution
istry Unit.
No. 691 from Solid State and Structural Chem-
@ik=-A$+$,i=Xe,k=O,Na.

54 0009-2614/91/$ 03.50 0 1991 - Elsevier Science Publishers B.V. (North-Holland)


Volume 177,number 1 CHEMICALPHYSICS LETTERS 8 February 199I

Table I and the calculated value of 17.25 kJ/mol obtained


Potential parameters for xenon-Nay zeolite at T= 375 K
by Kiselev and Du [ 91 in view of the fact that the
Atom A ( IO'kJ mol-’ A6) B ( lo6 kJ mol-’ A’*) Si/Al ratio employed by us here is higher and qs,de-
creases with an increase in the Si/Al ratio [ 13 1. Part
0 Na 0 Na of the discrepancy is due to the neglect of the po-
Xe 8.2793 Il.1345 7.9079
larization interaction. It is difftcult to compare our
2.9143
results to those of Woods et al. due to the spherical
potential approximation employed by them which
The dispersion parameters Aik were obtained from does not correspond to any specific Si/Al ratio.
the Kirkwood-Muller expression and the repulsion Fig. 1 shows the interaction-energy distribution
parameters Blk by imposition of the constraint that function, N(U). The curve shows a main peak near
the force should be zero at the equilibrium distance. - 12 kJ/mol and a shoulder near - 15 kJ/mol. The
The details are given in ref. [ lo]. The polarization shoulder arises from the xenon near the minimum
energy is taken into account in the present calcula- energy or the adsorption site. The xenon atoms which
tion since this is a many-body term and would result are not near the adsorption site contribute to the
in a considerable increase in the computational time. higher-energy region comprising the main peak and
The xenon-xenon interaction was modelled by a the high-energy tail. This may be compared with the
Lennard-Jones potential, adsorption of methane in faujasite which showed a
characteristic bimodal distribution. A similar bi-
(2) modal distribution was also observed in the case of
benzene [ 51. In contrast to these, the present results
with a=4.10 8, and e/k=221 K [lo]. show an almost unimodal distribution with only a
Molecular dynamics calculations were performed barely visible shoulder which seems to suggest the
in the microcanonical ensemble with periodic absence of two or more markedly distinct regions in
boundary conditions. The system consisted of one the supercage for the case of xenon. Therefore, it is
unit cell of faujasite consisting of 384 oxygen, 48 so- expected that a spherically averaged potential should
dium and eight xenon atoms. The xenon atoms were be more appropriate for xenon adsorbed in faujasite
uniformly distributed at the start of the simulation than for methane and benzene for which there ap-
in the large cavities or the supercages. A time step of pear to exist several significantly different regions in
1.0 fs was employed for the integration. Equilibra- the supercage. The higher-energy region, namely the
tion was performed over a period of about 20 ps and region above - 8 kJ/mol, was initially thought to be
averages accumulated over a fairly long period of 0.1
ns. A shifted potential with a potential truncation of
IO A has been employed. Integration was carried out
by means of a simple Verlet scheme [ 121. c

3. Results and discussion O.lO-


2
z
The average temperature of the system is 375.9 K.
The total interaction energy of a xenon atom is
( U) = ( U,) + ( Ugh), where ( V,) is the guest-
guest and ( Ugh) is the guest-host contribution. We
0.00
have obtained a value of - 12.33 kJ/mol for ( U) . - 20.0 -12 0 -40
The guest-guest contribution to the total interaction U.kJ/mol
energy was found to be small. The isosteric heat of Fig. 1. Energy distribution function for xenon in zeolite NaY at
adsorption, qa, is 15.46 kJ/mol. This compares rea- a concentration of one atom per large cavity and a temperattire
sonably with the experimentalvalue of 180.0 kJ/mol of 375 K.

55
Volume177,number I CHEMICALPHYSICSLETTERS 8 February1991

due to xenon migrating from one large cavity to an- peak in the cot-corn rdf. Thus, if P(r, 0, 4) is the
other. Closer examination, however, revealed that probability of finding a xenon atom at a point given
this is not so. The region with U> - 8 kJ/mol seems by (r, 8, $), where r, 0, 9 are the polar coordinates
to be due to xenon in the proximity of the cage centre. of points within the large cavity, then it is evident
The distribution of the methane in the supercage that P( r, 19,4) varies as a function of i3and # as well
is shown in fig. 2. Here, the centre-of-cage-centre-of- as of r. This suggests that the potential function has
mass (cot-corn) radial distribution function (rdf ) strong angular dependence. It is not known if these
is plotted as a function of the distance of methane anisotropies are also tetrahedral in symmetry reflect-
from the centre of the supercage. The cot-corn rdf ing the cage symmetry, since the potential felt by the
shows a main peak near 4 A and a smaller one near guest atom arises not only from just the atoms com-
1 A. The main peak in the rdf suggests that the xenon prising the inner walls of the large cavity but also from
atoms spend a large fraction of their time near the those beyond it. The tail near 6 A arises from the dif-
walls of the supercage. This is similar to the behav- fusion of xenon from one large cavity to another. This
iour shown by methane and other molecules [ 5,141. tail is likely to become increasingly prominent with
This type of cot-corn rdf seems to be a characteristic increase in temperature.
feature of X,Y-type of zeolites. The shape of the coc- Fig. 3 shows the xenon-xenon radial distribution
corn rdf and the position of the peaks are strongly function. The main peak near 4.5 A corresponding
influenced by two factors: One is the nature of the to the minimum in the Xe-Xe interaction potential
interaction and the other is the geometry of the cage. (0=4.10 A) suggests the existence of a significant
The comparison with the results of Rowlinson and population of dimer, even at this low concentration
co-workers [ 21, who employed a spherically aver- of one xenon per supercage. Such dimer formation
aged potential, is worthwhile. The present results has been observed for a one-site model of methane
show a broader spread of the main peak extending adsorbed in ZSM-5 [ 31. For a noble gas fluid, the
to much smaller r values and a small peak near 1 A second peak in the g(r) should appear at approxi-
absent in the earlier study [2]. This means that the mately $ r, where rl is the position of the first peak.
molecules are, in fact, able to stray significantly more It is interesting to note the second peak in the xe-
often towards the centre of the cage than suggested non-xenon rdf appears only beyond I 1 A. It is there-
by a spherically averaged potential. This view is in fore, apparent that the zeolite influences the fluid
accordance with the fact that the sites in the large structure strongly, at least at low loadings.
cavity are localized regions near the cage wall, and The cage distribution function is shown in fig. 4.
it is only in these regions that molecules spend a fairly It is seen that the fraction of cages with zero-occu-
large amount of time, thus contributing to the main pancy is 0.33 which is less than the fraction of cages
with an occupancy of one. This is in contrast to a
6
value of 0.37 obtained by Rowlinson and co-workers
/f-l

/ \ I

Fig. 2. Centre-of-cage-centre-of-mass radial distribution func- Fig. 3. Xenon-xenon radial distribution function for xenon ad-
tion for xenon in sodium-Y zeolite at a temperature of 375 K. sorbed in sodium-Y zeolite at 375 K.

56
Volume 177,number I CHEMICALPHYSICS LETTERS 8 February 1991

ment with the energy distribution function shown in

o.4, fig. I. The calculated diffusion coefficient was also


found to be much smaller than that of methane in
Nay.

-03r
I I
Fig. 5 shows the velocity autocorrelation function
5 for the motion of xenon in Nay. The striking feature
IL 0.2
of the velocity autocorrelation function here is the
slow decay of the correlation as compared with those
reported in the literature for other atoms or mole-
0.1
cules such as methane adsorbed in faujasite [ 3,5,14].
This is primarily due to the exceptionally high mass
-1:,:
012345 of xenon, We have also calculated the velocity au-
N tocorrelation function for the velocity component
Fig. 4. Distribution of occupancies, F(N), for the large cavity at parallel and perpendicular to the surface of the zeo-
a temperature of 375 K. lite cage,

for both zero- and one-occupancy. However, the dif-


ference is not surprising in view of the relatively large
size of the supercage as a consequence of using the
spherically averaged potential. The distribution
function shows that there exist cages with a high oc- c,(~)=(~,(O)~fi,(t)) 3 (3)
cupancy of rive, even at the low loading of one xe-
where i(t) is the unit vector from the centre of the
non/cage.
cage to the centre-of-mass of the xenon atom. Thus,
ii,,(t) is the velocity component parallel to the sur-
face whereas z?*(t) is the velocity component per-
4. Dynamics
pendicular to the surface. The behaviour of the C,,(I)
and C, ( f ) is shown in the inset of fig. 5. We are not
The cage residence time, site residence time and
aware of any experimental studies related to the dy-
the site-site migration time have been calculated. The
cage residence time, z,, was determined using a dis-
tance criterion to decide whether a given molecule
belonged to a specific cage. A molecule within a dis-
tance of 5.9 A of a cage is said to reside in the cage
in question. The value reported here, 9.9 ps, is found
to be about an order of magnitude larger than that
found for methane near 300 K [ 14 1. The difference
can be attributed to the large mass of xenon com-
pared to that of methane. This decreases the mobil-

_o.toO
ity of xenon significantly compared to that of meth-
ane. This is also supported by the velocity
autocorrelation function discussed below. In the cal-
culation of the site residence time (r$), an energy cut-
0.0 2 4 6 8
off of - 13 kJ/mol has been employed. The calcu-
t,ps
lated site residence time is 0.59 ps, which is again
Fig. 5. Xenon velocity autocorrelation function for xenonin zeo-
significantly greater than the value obtained for
lite NaY at 375 K. The inset shows (a) thevelocity autocorrela-
methane. The site-site migration time (r,) is the tion function for the velocity component perpendicular to the
average time the particle spends between two visits zeolite surface, and (b) the velocity component parallel to the
to a site. It is found to be 1.3 ps, which is in agree- zeolite surface.

57
Volume 177,number I CHEMICALPHYSICSLETTERS 8 February 1991

namics of xenon in faujasite. The velocity autocor- References


relation function, C, (l), crosses the x-axis below 1
ps, thereby indicating the high-frequency motion [ 1 ] J.S. Rowlinson, Proc. Roy. Sot. 402 (I 985) 67.
[ 21 G.B. Woods,A.Z. Panagiotopoulosand J.S. Rowlinson,Mol.
being performed by the xenon atom in the direction
Phys. 63 (1988) 49.
perpendicular to the surface. In contrast, along the [3] P. Demo&, E.S. Fois, G.B. Suffritti and S. Quatieri,
direction parallel to the surface the motion is of rel- preprint.
atively lower frequency, as indicated by C,,(t). The [4] S. Yashonath, J.M. Thomas, A.K. Nowak and A.K.
calculated diffusion coefficient at this loading and Cheetham, Nature 331 (1988) 601.
[ 51 P. Demontis, S. Yashonath and M.L. Klein, J. Phys. Chem.
temperature was found to be 2.4~ low8 m2 s-‘. It
93 (1969) 5016.
should be possible to measure the diffusion coeffl- [6] D.H. Olsen, J. Phys. Chem. 72 (1968) 4366.
cient by ‘29Xe pulsed-field-gradient NMR. [ 71 G.R. Eulenberger,D.P. Schoemaker and J.G. Keil, J. Phys.
We have described a molecular dynamics calcu- Chem.71 (1967) 1812.
[8] A.N. Fitch, H. Jobic and A. Renouprez, J. Phys. Chem. 90
lation of a realistic model of xenon adsorbed in zeo-
(1986) 1311.
lite NaY. The results are generally in good agreement [9] A.V. Kiselev and P.Q. Du, J. Chem. Sot. Faraday Trans II
with the grand canonical and partition function cal- 77 (1981) I.
culations of Rowlinson and co-workers using a [lo] A.G. Bezus, A.V. Kiselev, A.A. Lopatkin and P.Q. Du, J.
spherical potential but differ in some details. The de- Chem. Sot. Faraday Trans. II 74 (1978) 367.
[ 1I ] J.O. Hirschfelder, C.F. Curtiss and R.B. Bird, Molecular
cay of the correlation is slow as compared to other
theory of gases and liquids (Wiley. New York, 1964).
organic molecules adsorbed in faujasite. We are pres- [ 121G. Ciccotti and J.P. Ryckeart,Comput. Phys. Rept. 4 (1986)
ently investigating several other small molecules ad- 345.
sorbed in faujasite. [ 131H. Stach, U. Lohse, H. Thamm and W. Schirmer, Zeolites
6 (1986) 74.
[ 141S. Yashonath, P. Demontis and M.L. Klein, Chem. Phys.
Letters I53 (1988) 551.

You might also like