Professional Documents
Culture Documents
Abstract
The DubininRadushkevich (DR) equation is widely used for description of adsorption in microporous materials,
especially those of a carbonaceous origin. The equation has a semi-empirical origin and is based on the assumptions of a
change in the potential energy between the gas and adsorbed phases and a characteristic energy of a given solid. This
equation yields a macroscopic behaviour of adsorption loading for a given pressure. In this paper, we apply a theory
developed in our group to investigate the underlying mechanism of adsorption as an alternative to the macroscopic
description using the DR equation. Using this approach, we are able to establish a detailed picture of the adsorption in the
whole range of the micropore system. This is different from the DR equation, which provides an overall description of the
process. 2001 Elsevier Science Ltd. All rights reserved.
0008-6223 / 01 / $ see front matter 2001 Elsevier Science Ltd. All rights reserved.
PII: S0008-6223( 00 )00265-7
1328 C. Nguyen, D.D. Do / Carbon 39 (2001) 1327 1336
Nomenclature
C affinity coefficient of the BET equation
Cm micropore capacity
E potential of gas phase molecule
P relative pressure
Q heat of adsorption at zero loading
R gas constant
T temperature
W volume adsorbed
W0 micropore volume
p pressure
p0 vapour pressure
DH isosteric heat of adsorption
DHv ap vaporisation heat
r pore size (radius or half width)
characteristic energy derived from DR equation
s liquid surface tension
r liquid density
gm liquid molar volume
s subscript referring to solid slab surface
p subscript referring to micropore
some light on the working mechanism behind the DR which the characteristic energy and the micropore volume
equation. can be obtained.
As an illustration of the applicability of the DR equa-
tion, Fig. 1 shows the characteristic curves of nitrogen and
2. Description of the DR equation and our model of benzene adsorption onto two activated carbon (AC) sam-
adsorption ples, Ajax AC and ACF. The adsorption temperature for
nitrogen is 77 K, whereas that for benzene is 303 K. The
2.1. The DR equation equation of the straight lines describing the linear section
of the plots is used to calculate the micropore volume and
In its basic form, the DR equation can be written as: the characteristic energy, from which an effective pore size
can be estimated. Results of such calculations are shown in
W/W0 5 exp[2(A /E)2 ] or W/W0 Table 1. Similar results would be obtained from a direct
5 exp[2(RT ln (x) /E)2 ] (1) fitting of the DR equation to the experimental data using a
Table 1
Parameters calculated from nitrogen and benzene data using the DR equation
Temp. (K) r (cc / mol) Cms (mmol / g) W0 (cc / g) E (kJ / mol) r (nm)
ACF
Nitrogen 77 34.68 6.95 0.24 6.9 0.66
Benzene 303 89.80 2.53 0.23 18.1 0.66
Ajax
Nitrogen 77 34.68 13.9 0.48 7.0 0.65
Benzene 303 89.80 5.34 0.47 17.8 0.67
non-linear optimisation routine. As seen in Table 1, the coefficient is a function of the difference between the heats
results calculated from the adsorption data of different of adsorption. This difference is contributed mainly by the
probe molecules are consistent with each other. This is difference between the adsorption potential energy in a
considered to be another advantage of the DR equation. pore and that on a surface, since other energy contributing
Table 1 also indicates that although the mean micropore factors such as the vibrational and rotational energies, are
sizes are about the same, the micropore volume of Ajax assumed to be similar for those two situations. Even when
AC is about two times larger than that of the ACF sample. there is a difference between the sum of these energies of
the fluid and adsorbed phases, this difference is expected to
2.2. Potential enhancing adsorption model be smaller than the change in the potential energy. With
this assumption, we shall use the potential energy at zero
loading as that in the calculation of the difference in heats
2.2.1. Adsorption mechanism
of adsorption in Eq. (4) (Q p (r)). The potential energies (in
In our approach [7], we adopt a common mechanism for
an absolute sense) at zero loading corresponding to the
the adsorption process occurring in carbonaceous pores as
well depth and at the center of the pore are plotted in Fig.
a sequence of surface layering followed by condensation
2 as a function of pore size for benzene adsorption in
(or pore filling), irrespective of their sizes. The principal
carbon pores. These energies are determined from the
feature in our approach is the allowance for the enhance-
pair-wise summation of a benzene molecule with all atoms
ment in adsorption induced by the overlapping of the
on the walls of the carbon pore. The carbon pore wall is
potential fields of opposite pore walls. This enhancement
assumed to have a structure similar to the stack of graphite
not only affects the adsorbed phase, but also the gas phase
layers [8].
molecules occluded within the pore. Because of this
Based on the concepts presented above, adsorption in
enhancement effect on the occluded molecules, the fluid
pores can be pictured as a process whereby gas phase
phase pressure within the pore, pp , is different from the
molecules are drawn into the pore interior. Once inside,
bulk pressure p. According to the Boltzmann distribution
they are further pressed against each other as a result of the
of molecules, the pore pressure can be estimated as
overlapped potential forces. If the enhanced pressure is
follows:
S
pp (r) 5 p exp
2 kE gp (r)l
]]]
RT
D (3)
SQ p (r) 2 Q s
Cp (r) 5 Cs exp ]]] .
RT
D (4)
Fig. 2. Potential energy of adsorption used in the DR equation
Eq. (4) means that the enhanced adsorption affinity and our method.
1330 C. Nguyen, D.D. Do / Carbon 39 (2001) 1327 1336
beyond the corresponding vapor pressure, the adsorbed with x p (r) 5 pp (r) /p0 . The liquidsolid contact angle is
phase turns into liquid (albeit compressed liquid). It is, assumed to be zero for simplicity. The mathematical
therefore, understood that in some narrow pores, due to the interpretation of the arguments in the previous section can
very large enhanced pressures, the adsorbed phase exists as then be written as follows.
liquid even at low bulk pressures. This liquid filling The thickness of the adsorbed layer a
process progresses to pores larger in size as the bulk
(0 # a # r)
pressure increases. This is a picture of the micropore filling
process in detail, in contrast to the macroscopic description r if pp $ p0
available from the Dubinin theory [1,2], and to some
extent even the HK approach [9]. Parallel to this, in larger
pores, adsorption occurs as a surface layering process, and
5
5 t if pp , p0 and a 2 t . r k
r if pp , p0 and a 2 t # r k
(7)
this layer will grow with pressure until a stage is reached The adsorption isotherm can thus be generated for pores of
when the enhanced pressure reaches the vapor pressure any size. Examples of these single pore isotherms are
value or the conditions for pore filling are satisfied (for shown in Fig. 3 for nitrogen adsorption at 77 K in slit-like
mesopores this is termed capillary condensation). carbonaceous pores of different size.
These single pore isotherms are used as the local
2.2.2. Single pore isotherms isotherms in our technique to calculate the pore size
The statistical thickness of the adsorbed layer t is distribution (PSD) of the subject solid. The optimisation
evaluated using the BET equation [10], which takes the procedure involved is presented elsewhere [11].
following form when variables are allowed to bear depen-
dence on the pore size: 2.2.3. PSD of the ACF sample from nitrogen adsorption
at 77 K
x p (r)Cp (r)
t(r) 5 t m ]]]]]]]] (5) The model is tested against adsorption data of nitrogen
(1 2 x p (r))(1 2 x(r) 1 x p (r)Cp (r)) at 77 K for a microporous carbon sample (Takeda ACF).
Results of the model fitting and the extracted PSD of the
with t m being the monolayer thickness. The threshold half
sample are shown in Fig. 4. As seen, the fitting is excellent
width r k when the pore filling by pore condensation occurs
at both low and high pressures. The extracted PSD is
as
presented in the form of a histogram. As shown, the
sm g sample is a microporous material with a majority of pores
r k 5 ]]] (6)
RT ln (x p (r)) falling into the micropore range (by IUPAC definition
Fig. 3. Nitrogen single-pore isotherm at 77 K for carbon slit-like pores of half widths of 1.46, 2.54, 3.26, 5.06, 8.66, 17.65 (top) and 0.17,
0.24, 0.38, 0.74, 1.82 nm (bottom); (reading from left to right).
C. Nguyen, D.D. Do / Carbon 39 (2001) 1327 1336 1331
Fig. 4. Nitrogen isotherm fitting and the resulting PSD of sample ACF.
2r , 2 nm). As presented above, the DR equation gave for traditional HK method, where adsorption in a pore is zero
this sample a mean pore half width of 0.66 nm (or a pore until a critical pressure, when the pore volume is instanta-
width of 1.32 nm). neously filled by the adsorbate liquid. According to the HK
mechanism, pores are empty or completely filled with the
liquid adsorbate, i.e. there is no half full status for pores.
The overall adsorption on the ACF is a summation of
3. Volume filling of micropore theory versus adsorption in pores of all sizes. The contributions of each
sequential adsorption individual pore size, which appear in the PSD shown in
Fig. 4, are presented in Fig. 6. Table 2 summarises the
3.1. Progress of adsorption contributions of adsorption in different pore ranges at three
levels of reduced pressure, 6.3310 25 , 2.98310 22 , and
Strictly speaking, the fractional loading in the DR 0.39. In the case of the ACF, we divide the pore spectrum
equation defined as u 5 W( p) /W0 accounts for adsorption into 24 sub-ranges, each characterised by a nominal pore
in micropores only. This means that the contribution of size. The division allows more sub-ranges to be in the
larger pores to the uptake must be removed from the micropore region to reflect the microporous nature of the
experimental data before they are used with the DR ACF. In the table, the half width, pore volume and
equation. Furthermore, the DR equation provides only the adsorption data relating to micropores are shown in italics
macroscopic information. For example, when the change in for easier reference. Furthermore, to distinguish pores
the free energy, A 5 RT ln (x), is equal to the characteristic already filled to their maximum capacity, the amounts
energy of the system, the fractional loading is 0.37, adsorbed in these pores are printed bold. Let us first study
meaning 37% of the micropore volume is occupied by the the adsorption at the low relative pressure of 6.3310 25 .
adsorbate molecules. Since all carbonaceous solids have a Column 3 of Table 2 shows that adsorption occurs mainly
distribution of pore sizes, the DR theory does not provide in the smallest micropores. The first three pore sub-ranges
information on the individual fractional filling for each are filled completely with adsorbate (bold italic) while
pore size. One would expect small pores to be filled more larger pores are layered with adsorbate molecules. The
than the larger ones. Such information is important in the overall capacity at this pressure is 3.04 mmol / g, compared
better understanding of the variation of adsorption affinity to 3.06 mmol / g calculated using the DR equation. Here we
and heat of adsorption with loading. see that besides the overall capacity, our approach provides
Our approach models adsorption as a process occurring a detailed dispersion of this adsorption of 3.04 mmol / g. At
in all pores simultaneously, with the amount adsorbed in a a relative pressure of 2.98310 22 , all pores having half
pore being dependent on the pore size. This mechanism is width less than 0.588 nm are filled. This pore filling
schematically illustrated in Fig. 5. As illustrated in the progresses to pores of larger size when the pressure
figure, a fractional loading means that all pores narrower increases. For example, at a reduced pressure of 0.39, the
than a certain size are filled with adsorbate molecules, threshold pore size for complete filling is 1.266 nm, which
while in larger pores, adsorption layer is built up on the is in the mesopore range. The dominance of adsorption in
pore walls, and the thickness of the adsorbed layer is a micropores, especially at low pressures, is obvious from
function of the pore size. This is different from Dubinins Table 2. When the relative pressure increases from 6.33
theory, and also from the mechanism assumed in the 10 25 to 2.98310 22 to 0.39, the contribution of micropore
1332 C. Nguyen, D.D. Do / Carbon 39 (2001) 1327 1336
Fig. 6. Nitrogen isotherms in different pore sub-ranges (cf. Table 2) of the ACF sample.
C. Nguyen, D.D. Do / Carbon 39 (2001) 1327 1336 1333
Table 2
Nitrogen adsorption onto ACF at 77 K at three relative pressures (per gram of sample)
Half Fraction x 5 6.3 3 10 25 x 5 2.98 3 10 22 x 5 0.39
width volume
Loading Energy Loading Energy Loading Energy
nm cc
mmol % J % mmol % J % mmol % J %
0.286 0.004 0.1256 4.1 1.32 4.6 0.1256 2.0 1.32 2.3 0.1256 1.6 1.32 2.0
0.320 0.047 1.3611 44.7 13.10 46.1 1.3611 21.3 13.10 23.3 1.3611 17.8 13.10 19.7
0.357 0.041 1.1914 39.1 10.84 38.1 1.1914 18.7 10.84 19.3 1.1914 15.5 10.84 16.3
0.397 0.029 0.2410 7.9 2.11 7.4 0.8267 13.0 7.24 12.9 0.8267 10.8 7.24 10.9
0.439 0.009 0.0367 1.2 0.31 1.1 0.2653 4.2 2.27 4.0 0.2653 3.5 2.27 3.4
0.485 0.007 0.0144 0.5 0.12 0.4 0.2069 3.2 1.74 3.1 0.2069 2.7 1.74 2.6
0.534 0.021 0.0241 0.8 0.20 0.7 0.6021 9.4 4.99 8.9 0.6021 7.9 4.99 7.5
0.588 0.030 0.0219 0.7 0.18 0.6 0.8551 13.4 7.03 12.5 0.8551 11.2 7.03 10.6
0.645 0.026 0.0136 0.4 0.11 0.4 0.3667 5.7 3.00 5.3 0.7624 9.9 6.23 9.4
0.706 0.018 0.0069 0.2 0.06 0.2 0.2249 3.5 1.83 3.3 0.5144 6.7 4.19 6.3
0.772 0.006 0.0018 0.1 0.01 0.1 0.0674 1.1 0.55 1.0 0.1688 2.2 1.37 2.1
0.843 0.005 0.0012 0.0 0.01 0.0 0.0490 0.8 0.40 0.7 0.1340 1.7 1.09 1.6
0.919 0.005 0.0010 0.0 0.01 0.0 0.0465 0.7 0.38 0.7 0.1382 1.8 1.12 1.7
1.001 0.006 0.0011 0.0 0.01 0.0 0.0516 0.8 0.42 0.7 0.1663 2.2 1.34 2.0
1.090 0.004 0.0007 0.0 0.0328 0.5 0.1141 1.5
1.266 0.002 0.0003 0.0 0.0159 0.2 0.0637 0.8
1.683 0.002 0.0002 0.0 0.0085 0.1 0.0155 0.2
2.219 0.002 0.0001 0.0 0.0066 0.1 0.0118 0.2
2.912 0.005 0.0003 0.0 0.0155 0.2 0.0278 0.4
3.805 0.006 0.0002 0.0 0.0144 0.2 0.0258 0.3
4.958 0.006 0.0002 0.0 0.0117 0.2 0.0208 0.3
6.444 0.009 0.0002 0.0 0.0134 0.2 0.0240 0.3
8.361 0.014 0.0003 0.0 0.0161 0.3 0.0288 0.4
9.214 0.007 0.0001 0.0 0.0071 0.1 0.0126 0.2
Micropore 0.254 3.042 99.9 28.39 6.240 97.8 55.09 7.318 95.5 63.87
Overall 0.308 3.044 100 6.382 100 7.663 100
adsorption decreases slightly from 99.9 to 97.8 and then 3.2. Potential energies of adsorption
95.5%.
Table 3 presents the results of the nitrogen adsorption The only parameter concerning the solid properties
onto the ACF sample at three pressure levels calculated appearing in the DR equation is the characteristic energy E
using the DR and our method. The amount adsorbed (or E0 ). By investigating a large number of systems, the
calculated using the DR equation is by and large equal to dependence of the characteristic energy on the size of the
the sum of the amounts adsorbed in all pores using our carbon slit pore is found by Dubinin and co-workers
technique. The difference is more significant at the higher [12,13], and independently by Stoeckli [14] to take the
end of the pressure range, where adsorption in mesopores following approximate form:
occurs in addition to that in micropores. This is because k
the DR equation, unable to deal with mesopores, underpre- r5] (8)
E0
dicts the data at high relative pressures, while our tech-
nique describes well the data at all pressure levels. with k being a constant (12 kJ nm / mol). This equation is
Table 3
Nitrogen adsorption onto ACF sample at three relative pressures calculated using our method and the DR equation
Relative Overall capacity Micropore capacity
pressure (mmol / g) (mmol / g)
Experimental Our method DR eq. Our method
25
6.3310 3.03 3.04 3.12 3.04
2.98310 22 6.36 6.38 6.25 6.24
0.39 7.63 7.66 6.9 7.31
1334 C. Nguyen, D.D. Do / Carbon 39 (2001) 1327 1336
parison between the methods and / or a justification of the overall nitrogen isotherms for such a hypothetical carbon
suitability can be made based on experimental measure- sample. It is clear that the overall isotherm of this sample
ments of the isosteric heats, but these results are not does not support the concept of an adsorption characteristic
available at this stage. curve. Thus, for the DR equation to be applicable, the
porous solid must have a distribution of micropores, and
3.4. PSD and the applicability of the DR equation the distribution must not be very skewed. This explains
why the DR equation is not applicable to all microporous
In this section we will try to address the question materials.
regarding the application range of the DR equation. To A more general equation of the volume filling of
increase the smoothness (resolution) of the simulated micropores theory, the DubininAstakhov equation [17],
isotherm, we sub-divide the pore spectrum (half widths can also be explained similarly by considering the iso-
from 0.286 to 9.214 nm) of the ACF sample into 44 therms presented in the form of a log (Cm ) versus log n
sub-ranges. Fig. 8 shows the isotherms of nitrogen ad- (1 /x) plot.
sorption in these sub-ranges. Here the isotherms are shown
for a relative pressure interval ranging from 1310 24 to
1310 21 , which is the pressure interval where the DR
equation is normally investigated. As seen, none of the 4. Conclusions
single pore isotherms can be considered to be a straight
line. However, by adding up the individual isotherms, we Several techniques for analysis of adsorption in micro-
get the overall isotherm shown as the dotted line in Fig. 8, pores are presented. The DR equation as well as our
which is practically a straight line. This line is seen to fit technique are capable of describing the adsorption data.
the experimental data very well. A linear approximation of The former is by and large a means to fit the experimental
the overall isotherm plotted in this mode is an explanation data, while the later can be used to extract information
for the existence of a DR characteristic curve. regarding the PSD. It is found that the overall adsorption
It is then understood that the DR equation is not behaviour as expressed by the DR equation can be derived
applicable for adsorbents having a single pore size, and by combining processes occurring in individual pore
that a distribution of pore sizes is a prerequisite for the DR ranges using our method of analysis. The method also
equation. It is also possible to prove that samples having allows the assessment of details such as how much of the
very skewed PSDs may not comply with the DR equation. adsorption capacity and the heat released is contributed by
Fig. 9 shows the PSD together with the individual and the individual pore ranges.
Fig. 8. Nitrogen isotherms plotted in a log (Cm ) versus log 2 (1 /x) mode (single pore isotherms, solid lines; overall isotherm, dotted line;
experimental, solid circle).
1336 C. Nguyen, D.D. Do / Carbon 39 (2001) 1327 1336
Fig. 9. An arbitrary PSD (left) and the nitrogen isotherms in pore sub-ranges (right). The overall isotherm (dotted line) does not comply
with the DR equation principle.