Chinese J. of Chem. Eng. 8 (4) 287-293 (2000)
Determination of a Pore Structure Parameter of Porous
Media by Analysis of Percolation Network Model*
XIN Feng(##)"*, WANG Fumin(= # &), LIAO Hui()#) and LI Shaofen(# #4 $)***
School of Chemical Engineering and Technology, Tianjin University, Tianjin 200072, China
Abstract According tothe simulation of nitrogen sorption process in porous media with three-dimensional network
‘model, end the analysis for sush a prunes with percolation theory, 2 new method ls proposed to determine 8 pore
structure parameter—rmean coordination numberof pore network, which represents the conectvity among a great
umber of pores. Here the “chamber-throat™ modal and che Weibull distribution are used to describe the pore
‘Scometry and the pore sie distribution respectively. This method is based on the scaling law of percalation theory
after both effects of sorption thermodynamics and pote sie on the sorption hysteresis loops are considered. ‘The
results show that it ie an effective procedure to calculate the mean coordination number for mico- and mexo-poross
media,
Keywords coordination mimber, porous media, percolation theory, network model, sorption
1 INTRODUCTION
The aim of study on porous media is for practical
use. Previously, the complicated porous media was
treated approximately as macro-homogeneous mate-
rial, which could be described by continuum equation.
For gas-solid catalytic reaction, the complex structure
of porous catalyst cau be illustrated with the effective
parameters for diffusion and reaction by fitting the
‘experimental data, Obviously, for other reactions or
porous catalysts, the whole process mentioned above
‘must proceed once more. On the other hand, the effec-
tive parameters, such as effective diffusivity, obtained
by solving continuum equation can not directly re-
lect the properties of porous catalyst. Therefore, itis
impossible to understand the influence of microscope
structure of porous media on the mass transfer and
chemical reaction in the pores or on the pore surface
of the catalyst. Especially, for the reaction accom-
panied with changing structure of solid reactant or
catalyst, such as deactivation and regeneration of cat-
lyst or solid gasification, the continuum equation can
not describe the changes of pore size and pore struc-
ture, and how the changes influence the reaction and
transportation in catalysts
The discrete model, based on the percolation the-
‘ory, can solve above problem from investigating the
topological property of the network of pores. Consid
erable work hao beon done in Raf. (1]- ‘The most diff
cult problem is how to inspect the microscope topol-
ay of a real porous solid and establish a better dis-
crete network model". In order to characterize the
porous media, both the geometry of pores and the
Received 1990-04-16, accepted 2000-04-19
topology of the pore space are required. The former
represents geometrical sizes of pores, such as porosity;
the latter, however, represents the connectivity of a
large number of pores connected together.
‘Androutsopoulos and Manni interpreted the ex-
perimental results of mercury porosimetry with a dis-
crete network model, and regarded the connectivity
between pores as an important parameter representing
the structure of porous media, Seaton', Portsmonth
et all! and Lee et alll proposed two methods and
used them to determine the connectivity of pores from
sorption and mercury porosimetry respectively.
Space dimension, coordination number and per-
colation threshold are three main parameters in
percolation theory'"l, For three-dimensional and
two-dimensional networks, the products of thresh-
old and coordination number of bond percolation
are approximately 1.5 and 20 respectively). Some
researchers?-1I thought that when the coordination
zumber or threshold is the same, the conductivity be-
‘wcen order or disorder media ia aloo the same. Thue,
the results from the analysis of regular network can
be used in disorder system.
‘The connectivity of porous media can be obtained
from measuring the physical quantity closely related
with the coordination number, such as permeability,
diffusion flux, sorption hysteresis, mercury porosime-
try and co on. Of thom, especialy, the sorption and
porosimetry methods are easy to apply due to their
equilibrium processes. Liu et al.!1, and Sing et all!)
recognized that the extent of sorption hysteresis de-
pended strongly on the coordination number, which
+ Supported by the National Natural Science Foundation of China (No, 29776038),
++ To whom correspondence should be addressed
7 Deceased.288 Chinese J. Ch.
agreed with Mason'| and Rajiniak et afS1. ‘There-
fore, our work begins with the analysis of the sorption
hysteresis
‘The effect of the coordination number on the shape
of sorption hysteresis had been analyzed by simulating
the isotherm sorption processes in three-dimensional
network?1, The sorption hysteresis loop includes the
‘mutual result of thermodynamic properties of sorp-
tion, pore size distribution (PSD) and pore structure
properties, but it is verv difficult to isolate the con-
tribution of pore structure to the sorption hysteresis.
‘Only Seaton\4] proposed a method to get Z from the
analysis of sorption hysteresis, which needs fitting the
experimental data of sorption with generalized scaling
function, an implicit function, and completing two-
parameter evaluation of Z and L.
In this paper, a more simple method is proposed,
which can be used to calculate the mean coordination
number from sorption hysteresis with the sealing law
of percolation theory.
2 SIMULATION OF NITROGEN SORP-
TION IN 3-D NETWORK MODEL
‘The common method used for characterization of
Porous media is nitrogen sorption!"7, which includes
adsorption and desorption processes. At low temper-
ature, nitrogen adsorption in porous media includes
surface adsorption, which obeys Halsey equation
5.0
"inp.
b=
QQ)
and condensed liquid nitrogen filing the pores when
the radius is smaller than ri, which is caleulated by
the Kelvin equation for parallel plates
Wencos
~anp 0)
where bay Vas 7% @ are the characteristic parame
ters of the adsorbate. For nitrogen adsorption at
T7K, ba = 0.43nm, Vm = 34.65 x 10-%m*.mol-!,
y= 885 x 10-7 Nm“, and $= 0°.
For a given relative pressure p,, the critical radius
of a pore for adsorption is
rear tb (3)
Contrary to the adsorption is the vaporization of
liquid nitrogen and desorption of nitrogen molecule in
the pores. The vaporization obeys the Kelvin equa-
tion for cylinder pore
2m cos
a RT Inge o
‘The critical pore radius for desorption is
re=rhto 6)
December, 2000
1B, (Vol. 8, No.)
‘The ‘chamber-throat” model is used to represent
pore network, in which pore size distributions include
the pore throat radius, pore chamber radius and pore
length, which are defined as the amended Weibull
random distribution
yale
y
(6)
where rnin is the minimum pore radius.
‘The sorption process is controlled by the pore
throat radius. One of the simulating results is shown
in Fig. 1
oan}
Figure 1 Isotherms and hysteresises of the sorption
(Bc = Br = 8, =2,a¢ = 110m, ar = 85am,
eu = H0nm, rnin = 0-Snm)
6; 3
3. THEORETICAL MODEL OF CALCULAT-
ING Z FROM SORPTION HYSTERESIS
‘The nitrogen desorption in porous media can be illu
trated with the bond percolation. We define:
(a) Non-conducting bond: pore (bend) filling with
the condensed nitrogen when relative pressure is above
its condensation pressure;
(b) Conducting bond: pore (bond) containing ad-
sorbed nitrogen when relative pressure is below its
condentation pressure.
In the pore network model nitrogen in one pore
can vaporize only ifthe pore is represented by “con-
ducting bond” and it must connect directly with at
least one pore which has actually been a conducting
pore, When the pathway to outside is cloned by non-
conducting pore (bond), even if the liquid nitrogen isDetermination of Pore Stru
in a conducting pore, there is no opportunity to va-
Porize, Then the “conducting bond (pore)” in the
sense of thermodynamics is actually a nen-condueting
pore. Thus the process of desorption is restricted by
two factors. One is the thermodynamic property of
desorption process, another is the space structure of
ore network. We can describe desorption process by
following two percolation quantities:
: fraction of conducting bond, it is identified as
the number of pores that are helew their condensation
pressure divided by the total number of pores.
X: accessible fraction, which is the fraction of the
‘number of pores containing actually adsorptive nitro-
gen, divided by the total number of pores.
Although some pores are in the condition be-
low their condensation pressure, so called conducting
pores, the liquid nitrogen in these pores can not vapor
ize because there is no path to vapor phase, therefore
)
X and z, for a given pore, are related to pore
radius and its position in network, which can be re-
garded as the function of relative pressure py in the
Kelvin equation.
Percolation theory deals with the bond (pore)
number but not the pore size. The pore number must
bbe derived from the pore volume obtained from the
sorption process. Here we assume all of the pores in
any pore size have the same pore volume, Thus X
and x can be illustrated by
XB.) : Volpe) = 2(pr) : Vole) (8)
where ¥,(p.) is the nitrogen volume if the condensed
nitrogen filled in the pores vaporizes completely when
, is below their condensation pressure, and Vs(p,)
is the actually desorption nitrogen volume at relative
Pressure p.
X