Langmuir 1994,10, 4250-4252
4250
Adsorption of Water Vapor on Nonporous Carbon S. S. Barton,* M. J. B. Evans, and J. A. F. MacDonald Chemistry and Chemical Engineering, Royal Military College of Canada, Kingston, Ontario, Canada Received June 6,1994. In Final Form: August 18,1994@ The surface of the nonporous carbon Cabot Black Pearls 570 was modified by oxidation in aqueous nitric acid. In an attempt to produce a series of carbons having a range of oxidation levels, the time of contact with the acid solutionwas varied from 1to 8 h. The uasrecieved”and oxidized samples were characterized by water vapor adsorptionat 298 K, nitrogen adsorption at 77 K, and temperature-programmed desorption and enthalpy of immersion in water measurements at 298 K. The water vapor isotherms were analyzed by fitting to a modified DArcy-Watt equation and the relationships between the parameters thus obtained, and the surface oxide concentration and immersional enthalpies were examined.
Introduction As part of an on-going study of the effects of surface oxides on the adsorption of polar molecules by carbon, the nonporous Black Pearls 570 (Cabot Corp.) was oxidized, using aqueous nitric acid. A series of samples was produced by varying the time of contact, with boiling acid solution, from 1to 8 h. I t should be pointed out that, in spite of care being taken, this treatment is still essentially uncontrolled. The “asreceived” and oxidized samples were characterized by water vapor adsorption, temperatureprogrammed desorption measurements, and enthalpy of immersion calorimetry. An equation, derived by DArcy and Watt’ for application to the adsorption of water on heterogeneous adsorbents, was used to analyze the water adsorption isotherms. This equation is based on a statistical thermodynamic treatment of adsorption on a multisite substrate and can be expressed as
a=xi=l
SFih
skh +1 + Kih 1 - kh
(1)
where a = mass absorbed by 1g of adsorbent, Si = mass absorbed on primary sites of type i , s = mass absorbed on secondary sites, Ki = a constant which is a measure of the attraction of the ith site for the adsorbate, k = a constant which is a measure of the attraction of the secondary site for the adsorbate, h = plpo = relative pressure. The first term in eq 1represents the sum of independent, Langmuir-type adsorptions on n types of primary sites. The second term describes multilayer adsorption on the occupied primary sites. It is implied that there is only one type of secondary site on which, for the adsorbate water, two- or three-dimensional hydrogen-bonded clusters can begin to build up even before the primary sites are all occupied. In this work, eq 1 is considered to be simply a fourconstant equation:
a=- SKh 1+Kh
+-1 skh -kh
(la)
by assuming that there is only one type of surface site for primary adsorption as well as only one type of site for secondary adsorption. The first term, the Langmuir-type contribution, gives the adsorption on strongly bonding sites, and the second Brunauer Type I11 contribution Abstract publishedinAdvanceACSAbstracts,October 1,1994. (1) D’Arcy, R.L.;Watt, I. C. Trans. Faraday SOC. 1970, 66, 1236.
@
describes adsorption on the weaker secondary sites.4 Similar considerations, but based on kinetics, underlie the formulations of Dubinin and Serpinski2 and Barton et al.3 which purport to describe water adsorption on porous carbons. The second term in eq l a is formally identical to the equation proposed by Dubinin and Serpenski (DS-1),2in which s corresponds to the concentration of active surface sites on a porous carbon. Indeed, it is an easy matter to formulate eq l a from simple kinetic or material balance equations describingthe primary and secondary adsorptions. A serious flaw in this simple treatment is the implication that all the primary adsorbed water molecules should be available to act as secondary sites for the adsorption of a second water molecule. That this appears not to be the case will be evident later. Values of S, K, s, and k are obtained by a nonlinear curve fitting routine so that the observed isotherm may, in effect, be separated into two component isotherms. To examine the credibility of the above procedure, it is important to show that there are rational correlations between the results of the isotherm analysis and independent observations made by temperature-programmed desorption of surface oxides and immersion calorimetry.
Experimental Section Cabot Black Pearls 570 was oxidized in the followingmanner. Twenty gram samples were wetted with 100 mL of water in 1 L round-bottomedflasks which were cooled in an ice bath as 300 mL of 8 M nitric acid was slowly added. m e r being gradually warmed to room temperature with gentle stirring, the flasks were fitted with condensers and warmed t o a slow boil, causing the evolution of brown fumes. Four samples were prepared by boiling for various periods of time, followed by cooling to room temperature. The oxidized carbons were separated from the nitric acid solution and washed with copious amounts of water
before being placed in Soxhlet extractors, where continuous washing was carried out for 1 week. The samples were finally dried, in air at 100 “C. The “asreceived”carbon was dried at 100 “C but not subjected to any other treatment. A McBain-Bakr, quartz spiral balance, encased in a thermostatted cabinet, was used to measure the water adsorption isotherms after the carbon samples had been degassed at 110 “C t o O.OS 0.06 0.04
In Figure 1are shown the linear relationships observed between the total oxygen concentration [OYmmobg-l (carbon) and the correspondingcarbon dioxide and carbon monoxide contributions to the oxygen concentration. The observed linearities suggest that during the oxidation procedure, since the relative amounts ofthe surface oxides which desorb as carbon dioxide and carbon monoxide remain reasonably constant, only the amount and not the structure of the surface oxide is changed with increasing oxidation. The ratio [COY[CO,l is approximately 4. The water vapor adsorption isotherms obtained are shown in Figure 2. The isotherm for the “as received” material is a Brunauer Type I11 or a Type (a),according to the classification system for adsorption of water vapor on nonporous carbons proposed by Carrott.‘j Such isotherm shapes indicate a low surface polarity. The isotherms for the oxidized samples, on the other hand, show a n initial rise at low relative pressures, followed by a slow increase in adsorption with increasing relative pressure, and finally a substantial increase in uptake as the relative pressure approaches 1. This isotherm behavior is classified by Carrott as Type (d) (or Brunauer Type 11)and is indicative of a carbon black ofintermediate to high polarity. The software package Sigma Plot 4.1 was used to fit the experimental isotherm data to eq la. The nonlinear curve fitting routine uses the Marquardt-Levenberg algorithm to determine those parameters which minimize the sum of the squares of the differences between equation values and t h e y data values. The parameter values for eq l a , found in this way, are given in Table 1, and the fitted curves calculated using 1992, 30,201.
0.6
Figure 2. Water vapor adsorption isotherms: (A)as received, (‘11 1 h oxidation; (0)2 h oxidation; (0)4 h oxidation; (0)8 h oxidation.
0.02
P.J. M. Carbon
0.4 h
Results and Discussion
(6) Carrott,
0.2
t
0.00 0.0
................. ........
0.2
0.4
0.6
..............
0.8
1.0
h
Figure 3. Isotherm component contributions to 4 h oxidation: (- - -) Langmuir; (-) Type 111; ( 0 )data. Table 1 S sample (gg-l) AR 0.0000 1 h o x 0.0379 2 h o x 0.0431 4 h o x 0.0500 8 h o x 0.0534
S
K 0.0000 2.843 2.013 3.545 5.827
(a-9 1.15 x 3.5 x 4.9 x 3.4 x 7.8 x
0.835 1.000 0.982 1.000 0.975
a0
ao+
(gg-1)
s (gg-1)
0.0000 0.0280 0.0289 0.0390 0.0456
0.0115 0.0315 0.0338 0.0424 0.0534
them are shown in Figure 2. In Figure 3, as a typical example, are shown the adsorption data for the 4 h oxidized sample as well as the two components of the fitted curve. It is apparent from Table 1 that, for the “as received” material, the Langmuir-type contribution is zero. The meaning attributed to the s value of the Type I11 portion is that it represents the amount adsorbed when all the weak (or secondary) sites on the surface are occupied. That is, s is analogous to the monolayer capacity of the Brunauer Type I11 isotherm. Subsequent, weak, multilayer, hydrogen-bonded adsorption then takes place as the relative pressure increases. It is assumed here that only the coverage of the weak (or secondary) s sites is sufficiently energetic to contribute significantly to the enthalpy of immersion. The assumption finds support in the work of Morimoto and M i ~ v awho , ~ observed, from (7) Miura, IC;Morimoto, T.Langmuir 1994, 10,807.
4252 Langmuir, Vol. 10,No. 11, 1994
Barton et al. Table 2
bet area Ahi [OI sample (m2.g-') (J-g-l) (mmo1.g-l) AR 100 7.9 0.545 16.2 1 hox 123 1.74 17.9 1.95 2 hox 123 4 hox 126 22.6 2.24 30.5 8hox 149 3.41
-0
1
2
3
4
(0) or [&+s]/mmol/g (sample)
Figure 4. Correlation of isotherm parameters, temperature desorption results, and enthalpy of immersion: (D) [O]mmoVg (sample); (box) [ao + SI mmoVg (sample); ( x ) literature (see text). the temperature variation of the Type I11 water isotherms on oxidized graphite, that the isosteric heat of adsorption decreased, with increasing HzOcoverage, from about 70 kJ-mol-l to the latent heat of condensation of water (44 kJ-mol) at the Type I11 monolayer coverage. For the oxidized carbons, however, there are contributions from both the Langmuir and Type I11 components. When the affinity parameter Kis large, then the primary sites can become saturated at vapor pressures below the saturation vapor pressure. That is to say, the Langmuir contribution should display a plateau. For smaller K values, as is the case here, the potential number ofprimary sites are only partially covered, below saturation vapor pressure, and a plateau region is not observed. However, this number of primary sites occupied a t saturation vapor pressure may be calculated from the Langmuir contribution by settingh = 1. These calculatedaovalues are given in Table 1. From this table, it is apparent that the s sites are only about 14%of the number at a0 sites. That is to say, only this percentage of a0 sites are involved in the initiation of Type I11 multilayer formation at low relative pressures. The reasons for this state of affairs are not immediately evident. For what follows, it is assumed that the sum a0 s gives that amount of water adsorbed at h = 1 with sufficient energy to contribute significantly to the enthalpy of immersion. In order to test the correctness ofthe isotherm analysis, it is necessary to demonstrate that there is a reasonable correlation between the isotherm parameters and the results of the temperature-programmed desorption experiments and the calorimetry measurements. To this end, in Figure 4 is shown the relationship between the concentration of surface oxide [OYmmolg-l (sample) a n d or [a0 sYmmo1.g-l (sample) with the enthalpy of immersion. There seems t o be a strong correlation between these three quantities, in that the plot approximates fairly well to a single straight line. This single linear correlation implies that there is a 1:l correlation between the quantities [Ol and [a0 +SI. In addition, from
+
+
equivalent monolayers [a,
+ SI
0.40 0.900 0.966 1.18 1.26
a0
0.40 0.80 0.826 1.08 1.07
the intercept on the enthalpy axis an areal enthalpy of immersion can be calculated (assuming the AR BET area of 100 m2 g-l; see Table 1). This value is 0.018 f .012 J-m-2 and within experimental error is very close to the areal immersional enthalpy for Graphon (0.031 J*m-2),8 Spheron 6 (0.032 J*m-2),9and graphite (0.032 J*m-2),10 carbons for which the surface oxide concentration is very small or zero. It may be concluded from Figure 4 that there is a strong correlation between the surface oxide concentration and the energetic active site concentration and that both these quantities correlate well with the immersional enthalpy. The observed linearity implies either that the oxide-active sites are essentially monoenergetic or that the proportions of sites of differing energies remain reasonably constant with the extent of oxidation. Finally, an estimate ofthe number of layers of adsorbed water which results from primary and secondary interactions of high energy can be made in the following way. With 0.105 nm2as the area of an adsorbed water molecule, monolayer coverage would correspond to an uptake of 15.8 pmol*m-2of surface.6 Through the use of this figure, and the BET surface area, the number oflayers corresponding to the total number of sites [a0+ SI and only the primary sites may be calculated. The results are shown in Table 2. It appears that the total number of sites corresponds to more than monolayer coverage ( ~ 1 . 2 layers) 6 for the highly oxidized carbons, whereas the primary a0 (or s for AR) sites yield a value which exceeds monolayer coverage only slightly, indicating approximately 1.08layers ofclosepacked water molecules on the highly oxidized carbon surface. This result implies that the surface of the 4 h oxidized carbon is completely covered by chemisorbed oxygen. In conclusion, it may be stated that the reasonableness of the isotherm parameters for the D'Arcy-Watt type equation appears to find some support from the independent measures of surface oxide content and interaction enthalpy. It may be speculated from this study that, since the chemisorbed oxygen on the carbon surface seems to show an approximately 1:1 interaction with the physisorbed water molecules,the water-oxide interaction is essentially independent of the actual structure (phenolic, carboxylic, carbonylic, lactonic, etc.) in which the oxygen atom is incorporated. ( 8 ) Healey, F. H.; Yu,Y.F.;Chessick, J.J.J . Phys. Chem. 1955,59,
339. ...
(9)Wade, W. H. J. Colloid Sci. 1969,31, 111. (10)Good, R.J.;Girifalco, L. A.; Kraus, G. J.Phys. Chem. 1968,62, 1418.