- Email: [email protected]
The adsorption of water by oxidised microporous carbon
Carbon Printed
Vol. 25. No. 1, pp. 81-83, in Great Bntain.
0008-6223187 0 1987 Per@xwm
1987
$3.00 + .@I Journals Ltd.
THE ADSORPTION OF WATER BY OXIDISED MICROPOROUS CARBON Department
MICHAEL J. B. EVANS of Chemistry and Chemical Engineering, Royal Military College of Canada, Kingston, Ontario, Canada K7K 5KO (Received 15 May 1986)
Key Words-Microporous
carbon, water adsorption, BPL carbon.
relative pressure p/p,,, and a, is the adsorption on primary sites. The term (1 - ka) accounts for the decrease in adsorption centers as pores become filled, k being a constant. The equation may also be expressed in quadratic form
The purpose of this paper is to reexamine the analysis of water vapor adsorption isotherms measured as part of a study of the effects of oxidation on the adsorptive properties of BPL activated carbon (Calgon Corporation)[l]. The semiempirical isotherm of Dubinin and Serpinsky(21 was used in the original analysis to estimate the number of high-energy sites which act as primary adsorption sites for water on the predominantly hydrophobic carbon surface. The model proposed by Dubinin and Serpinsky assumes that water molecules adsorbed on these primary sites act as sites for further adsorption. Thus if a surface contains relatively few primary sites, an isotherm more or less corresponding to the type-5 isotherm of the Brunauer classification[3] would be expected, whereas if a significant number of primary sites exists, the water vapor adsorption isotherm would have an initial Langmuir-type section followed by a rapid increase in adsorption with increasing relative pressure and finally a flatter portion which is a consequence of the limited adsorption space available in a porous material. Experimental isotherms measured on BPL carbon at 25°C using a McBain helical quartz spring balance and a quartz bourdon pressure gauge (Texas Instruments Inc., Model 144) do indeed show this type of behavior[l]. Oxidation of the carbon by nitric acid modifies the isotherm at both low and high relative pressures. Typical isotherms are shown in Fig. 1. The changes at low relative pressure are believed to stem mainly from the introduction of oxygen-containing surface complexes into the porous carbon, while changes at high relative pressure are presumably a result of the removal of carbon by oxidation which modifies the pore structure and hence the sorptive capacity of the carbon. Mercury porosimetry shows that BPL carbon has considerable macro- and mesoporosity as well as microporosity[4]. The Dubinin-Serpinsky (DS) equation can be written in the form
a(p,/p)
~(a,, + a)(1 -
(2)
so that a graph of the left-hand side of this equation against the amount adsorbed, a, should be an in-
verted parabola if the equation is obeyed. As can be seen from Fig. 2, experimental data can be fitted to a parabola reasonably well over the upper 75% of the adsorption range, but agreement is poor at low amounts adsorbed. Since the basis of the Dubinin and Serpinski model is the existence of primary sites which are responsible for adsorption at low relative pressures, it is to be expected that their equation would fit experimental data in this region at least as well, if not better than, data for higher relative pressures. Clearly, this is not the case for water on BPL carbon. The validity of equating the value of a, with the amount adsorbed on primary sites is therefore open to question. In order to demonstrate the applicability of their equation to the adsorption of water on activated carbon, Dubinin and Serpinsky published two isotherms which it was claimed were well described by eqn (1)[2]. Unfortunately, no points appear to have been measured below a relative pressure of approximately 0.12, and neither isotherm features an initial Langmuir-type region. The data are not presented in sufficient detail to determine how well their equation fits the experimental isotherms, nor whether agreement is poor at low amounts adsorbed as is found with the isotherms presented here. An equation derived by D’Arcy and Watt[4] for analysis of isotherms for the absorption of water by biopolymers and related macromolecules is based on assumptions that closely resemble those used in deriving the DS equation. In D’Arcy and Watt’s model it is assumed that sorption of water takes place on strongly binding primary sites, which are the hydrophilic groups occuring in proteins and similar molecules, and to a lesser extent on weakly binding sites
a PIP0 =
= ca, + c(1 - ka,)a - ckd,
ka) ’
where a represents the amount of water adsorbed at 81
MICHAEL J. B. EVANS
82
on low energy sites and a third component isotherm which accounts for secondary absorption. In cases where only one type of primary sites is assumed to exist the equation of the isotherm can be expressed as
0,3.g-’ a/g
KK’WPO)
a=1 -
0.2-
“.*
0
“.’
.zo, I
0
x
8
0.2
. Y
xx
x
0.4 Relative
x
0.6 Pressure,
0.8 p/p.
I.0
Fig. 1. Adsorption isotherms for water on BPL carbon at 25°C: x , unoxidised; 0, 2-h oxidation; 0,7-h oxidation.
elsewhere on the polymer chains. Absorbed molecules then act as secondary sites for further absorption. Using such a model D’Arcy and Watt derived an equation which is a composite of one or more Langmuir isotherms describing absorption on high energy sites, of which there may be several types, a linear or Henry’s law isotherm describing absorption
K(plp,)
kk’(P/PO)
+ CWPO) +
1 - k(PlPo)’
where K’ and k’ are related to the numbers of primary and secondary sites, and C is the Henry’s law constant for absorption on low-energy sites. The constant K is a function of the primary site-sorbate interaction energy, and k is a function of the sorbatesorbate interaction energy. A weighted nonlinear least squares regression analysis was employed to fit eqn (3) to sorption isotherms for water on a variety of macromolecules. D’Arcy and Watt found that a unique set of five parameters could be calculated for each material, and illustrated several isotherms which showed excellent agreement between the experimental data and the fitted equation. Application of eqn (3) to the analysis of adsorption isotherms for water on BPL carbon shows that agreement between the equation and experimental data is far better than is obtained with the DS equation at low relative pressure. Portions of the experimental
I
I
I
(b)
0.4 a+, 7
0.3
0-I
o-2 a/cvf
o-3
(3)
0-I
o-2 a/s a”
Fig. 2. Dubinin-Serpinsky fit [eqn (3)]: (a) 2-h oxidation; (b) 7-h oxidation.
o-3
Adsorption of water
I
I
I
I
a-
03 t
83
I
(a)
(bj
I
J
1.2 -
1.1 -
0
I.5 Relative
0.1
Pressure,
0.2
I 0.3
1 o-4 1 5
p/p,
Fig. 3. Comparison between experimental and fitted isotherms: broken line, DS equation solid line, D’Arcy and Watt equation [eqn(3)]; (a) 2-h oxidation. (b) 7-h oxidation. are shown in Fig. 3 with fitted D’Arcy and Watt isotherms. The values of a, obtained using the DS equation correspond to 2.91 x lO*Oand 0.66 x lo*” primary sites per gram of carbon, for the samples oxidised for 2 and 7 h, respectively, whereas the K' values correspond to 5.48 x lO*O and 3.05 x lO*Oprimary sites per gram of carbon, respectively. It is planned to measure the quantity of oxygen evolved from oxidised BPL carbon by temperatureprogrammed desorption[6] in order to see whether the number of primary sites determined using the D' Arty and Watt equation can be correlated with the amount of oxygen present in surface groups. Such information may also lead to a better understanding of the relationship between heats of immersion and the concentration and nature of oxygen-containing surface groups[ 1,7,8].
isotherms
[eqn
(2)]:
REFERENCES
1. S. S. Barton, M. J. B. Evans, J. Holland and J. E. Koresh, Carbon 22,265 (1984). 2. M. M. Dubinin and V. V. Serpinsky, Carbon 19, 402 (1981). 3. S. Brunauer, L. S. Deming, W. S. Deming and E. Teller, .I. Am. Chem. Sot. 62.1723 (1940). 4. S. S. Barton, J. R. Dacey and M.‘J. B. Evans. Colloid Polym. Sci. 260, 726 (1982).
5. R. L. D’Arcy and I. C. Watt, Trans. Faraday Sot. 66, 1236 (1970).
6. N. R. Laine, F. J. Vastola and P. L. Walker, 1. Phys. Chem. 67,203O (1963). 7. H. E Stoeckli and F. Kraehenbuhl, Carbon 19, 353 (1981). 8. H. F. Stoeckli, F. Kraehenbuhl and D. Morel, Carbon 21, 589 (1983).
DISCUSSION H. A. Resing-I suggest that theoretical adsorption isotherms should not be linearized for curve fitting. There are adequate non-linear curve fitting routines available for most computers. Furthermore such linearization changes the weights of data points in a way that is not obvious. D-R and BET plots, and others, require a recalibration of the observer’s intellect to take into account the warping of the raw data, thus hindering communication between advocates of the various isotherms and with the world in general. The isotherm (experimentally observed) is the reality, and we want to see how the theory passes through these data points. M. J. B. Evans-I quite agree with Dr. Resing. Linearization often gives a misleading picture since it tends to overemphasise certain data points at the expense of others. Rarely do we see a comparison between experimental data and the fitted isotherm, regardless of whether a linear or a non-linear fitting method is employed. Error bars are seldom shown in figures in many journals and details of any error analysis are often barely adequate, if they are given at all. As a result it is impossible to gauge the reliability of the data or the excellence of the fit in many papers appearing in the literature.
Vol. 25. No. 1, pp. 81-83, in Great Bntain.
0008-6223187 0 1987 Per@xwm
1987
$3.00 + .@I Journals Ltd.
THE ADSORPTION OF WATER BY OXIDISED MICROPOROUS CARBON Department
MICHAEL J. B. EVANS of Chemistry and Chemical Engineering, Royal Military College of Canada, Kingston, Ontario, Canada K7K 5KO (Received 15 May 1986)
Key Words-Microporous
carbon, water adsorption, BPL carbon.
relative pressure p/p,,, and a, is the adsorption on primary sites. The term (1 - ka) accounts for the decrease in adsorption centers as pores become filled, k being a constant. The equation may also be expressed in quadratic form
The purpose of this paper is to reexamine the analysis of water vapor adsorption isotherms measured as part of a study of the effects of oxidation on the adsorptive properties of BPL activated carbon (Calgon Corporation)[l]. The semiempirical isotherm of Dubinin and Serpinsky(21 was used in the original analysis to estimate the number of high-energy sites which act as primary adsorption sites for water on the predominantly hydrophobic carbon surface. The model proposed by Dubinin and Serpinsky assumes that water molecules adsorbed on these primary sites act as sites for further adsorption. Thus if a surface contains relatively few primary sites, an isotherm more or less corresponding to the type-5 isotherm of the Brunauer classification[3] would be expected, whereas if a significant number of primary sites exists, the water vapor adsorption isotherm would have an initial Langmuir-type section followed by a rapid increase in adsorption with increasing relative pressure and finally a flatter portion which is a consequence of the limited adsorption space available in a porous material. Experimental isotherms measured on BPL carbon at 25°C using a McBain helical quartz spring balance and a quartz bourdon pressure gauge (Texas Instruments Inc., Model 144) do indeed show this type of behavior[l]. Oxidation of the carbon by nitric acid modifies the isotherm at both low and high relative pressures. Typical isotherms are shown in Fig. 1. The changes at low relative pressure are believed to stem mainly from the introduction of oxygen-containing surface complexes into the porous carbon, while changes at high relative pressure are presumably a result of the removal of carbon by oxidation which modifies the pore structure and hence the sorptive capacity of the carbon. Mercury porosimetry shows that BPL carbon has considerable macro- and mesoporosity as well as microporosity[4]. The Dubinin-Serpinsky (DS) equation can be written in the form
a(p,/p)
~(a,, + a)(1 -
(2)
so that a graph of the left-hand side of this equation against the amount adsorbed, a, should be an in-
verted parabola if the equation is obeyed. As can be seen from Fig. 2, experimental data can be fitted to a parabola reasonably well over the upper 75% of the adsorption range, but agreement is poor at low amounts adsorbed. Since the basis of the Dubinin and Serpinski model is the existence of primary sites which are responsible for adsorption at low relative pressures, it is to be expected that their equation would fit experimental data in this region at least as well, if not better than, data for higher relative pressures. Clearly, this is not the case for water on BPL carbon. The validity of equating the value of a, with the amount adsorbed on primary sites is therefore open to question. In order to demonstrate the applicability of their equation to the adsorption of water on activated carbon, Dubinin and Serpinsky published two isotherms which it was claimed were well described by eqn (1)[2]. Unfortunately, no points appear to have been measured below a relative pressure of approximately 0.12, and neither isotherm features an initial Langmuir-type region. The data are not presented in sufficient detail to determine how well their equation fits the experimental isotherms, nor whether agreement is poor at low amounts adsorbed as is found with the isotherms presented here. An equation derived by D’Arcy and Watt[4] for analysis of isotherms for the absorption of water by biopolymers and related macromolecules is based on assumptions that closely resemble those used in deriving the DS equation. In D’Arcy and Watt’s model it is assumed that sorption of water takes place on strongly binding primary sites, which are the hydrophilic groups occuring in proteins and similar molecules, and to a lesser extent on weakly binding sites
a PIP0 =
= ca, + c(1 - ka,)a - ckd,
ka) ’
where a represents the amount of water adsorbed at 81
MICHAEL J. B. EVANS
82
on low energy sites and a third component isotherm which accounts for secondary absorption. In cases where only one type of primary sites is assumed to exist the equation of the isotherm can be expressed as
0,3.g-’ a/g
KK’WPO)
a=1 -
0.2-
“.*
0
“.’
.zo, I
0
x
8
0.2
. Y
xx
x
0.4 Relative
x
0.6 Pressure,
0.8 p/p.
I.0
Fig. 1. Adsorption isotherms for water on BPL carbon at 25°C: x , unoxidised; 0, 2-h oxidation; 0,7-h oxidation.
elsewhere on the polymer chains. Absorbed molecules then act as secondary sites for further absorption. Using such a model D’Arcy and Watt derived an equation which is a composite of one or more Langmuir isotherms describing absorption on high energy sites, of which there may be several types, a linear or Henry’s law isotherm describing absorption
K(plp,)
kk’(P/PO)
+ CWPO) +
1 - k(PlPo)’
where K’ and k’ are related to the numbers of primary and secondary sites, and C is the Henry’s law constant for absorption on low-energy sites. The constant K is a function of the primary site-sorbate interaction energy, and k is a function of the sorbatesorbate interaction energy. A weighted nonlinear least squares regression analysis was employed to fit eqn (3) to sorption isotherms for water on a variety of macromolecules. D’Arcy and Watt found that a unique set of five parameters could be calculated for each material, and illustrated several isotherms which showed excellent agreement between the experimental data and the fitted equation. Application of eqn (3) to the analysis of adsorption isotherms for water on BPL carbon shows that agreement between the equation and experimental data is far better than is obtained with the DS equation at low relative pressure. Portions of the experimental
I
I
I
(b)
0.4 a+, 7
0.3
0-I
o-2 a/cvf
o-3
(3)
0-I
o-2 a/s a”
Fig. 2. Dubinin-Serpinsky fit [eqn (3)]: (a) 2-h oxidation; (b) 7-h oxidation.
o-3
Adsorption of water
I
I
I
I
a-
03 t
83
I
(a)
(bj
I
J
1.2 -
1.1 -
0
I.5 Relative
0.1
Pressure,
0.2
I 0.3
1 o-4 1 5
p/p,
Fig. 3. Comparison between experimental and fitted isotherms: broken line, DS equation solid line, D’Arcy and Watt equation [eqn(3)]; (a) 2-h oxidation. (b) 7-h oxidation. are shown in Fig. 3 with fitted D’Arcy and Watt isotherms. The values of a, obtained using the DS equation correspond to 2.91 x lO*Oand 0.66 x lo*” primary sites per gram of carbon, for the samples oxidised for 2 and 7 h, respectively, whereas the K' values correspond to 5.48 x lO*O and 3.05 x lO*Oprimary sites per gram of carbon, respectively. It is planned to measure the quantity of oxygen evolved from oxidised BPL carbon by temperatureprogrammed desorption[6] in order to see whether the number of primary sites determined using the D' Arty and Watt equation can be correlated with the amount of oxygen present in surface groups. Such information may also lead to a better understanding of the relationship between heats of immersion and the concentration and nature of oxygen-containing surface groups[ 1,7,8].
isotherms
[eqn
(2)]:
REFERENCES
1. S. S. Barton, M. J. B. Evans, J. Holland and J. E. Koresh, Carbon 22,265 (1984). 2. M. M. Dubinin and V. V. Serpinsky, Carbon 19, 402 (1981). 3. S. Brunauer, L. S. Deming, W. S. Deming and E. Teller, .I. Am. Chem. Sot. 62.1723 (1940). 4. S. S. Barton, J. R. Dacey and M.‘J. B. Evans. Colloid Polym. Sci. 260, 726 (1982).
5. R. L. D’Arcy and I. C. Watt, Trans. Faraday Sot. 66, 1236 (1970).
6. N. R. Laine, F. J. Vastola and P. L. Walker, 1. Phys. Chem. 67,203O (1963). 7. H. E Stoeckli and F. Kraehenbuhl, Carbon 19, 353 (1981). 8. H. F. Stoeckli, F. Kraehenbuhl and D. Morel, Carbon 21, 589 (1983).
DISCUSSION H. A. Resing-I suggest that theoretical adsorption isotherms should not be linearized for curve fitting. There are adequate non-linear curve fitting routines available for most computers. Furthermore such linearization changes the weights of data points in a way that is not obvious. D-R and BET plots, and others, require a recalibration of the observer’s intellect to take into account the warping of the raw data, thus hindering communication between advocates of the various isotherms and with the world in general. The isotherm (experimentally observed) is the reality, and we want to see how the theory passes through these data points. M. J. B. Evans-I quite agree with Dr. Resing. Linearization often gives a misleading picture since it tends to overemphasise certain data points at the expense of others. Rarely do we see a comparison between experimental data and the fitted isotherm, regardless of whether a linear or a non-linear fitting method is employed. Error bars are seldom shown in figures in many journals and details of any error analysis are often barely adequate, if they are given at all. As a result it is impossible to gauge the reliability of the data or the excellence of the fit in many papers appearing in the literature.