Langmuir 1992,8, 577-580
577
Adsorption Measurements of Argon, Neon, Krypton, Nitrogen, and Methane on Activated Carbon up to 650 MPa P. Malbrunot, D. Vidal, and J. Vermesse Laboratoire d'lngbnierie des Matbriaux et des Hautes Pressions, CNRS, Centre Universitaire Paris-Nord, Avenue Jean Baptiste Clgment, 93430 Villetaneuse, France
R. Chahine and T. K. Bose" Groupe de Recherche sur les Diblectriques, Universitb du Qubbec d Trois-Rivikres, C.P. 500, Trois-Rivikres, Qubbec, Canada G9A 5H7 Received July 29,1991.I n Final Form: October 21, 1991 The physisorption of argon, krypton, neon, nitrogen, and methane on GAC activated carbon has been measured in the above critical region by a dielectric method. The measurements were done at room temperature and at pressures up to 650 MPa corresponding to reduced densities of up to 3.25. With the exception of nitrogen, all the measured excess adsorption isotherms show a similar behavior. They exhibit a maximum followed by a downward straight line intercepting the bulk density axis at around the liquid density of the adsorbate in the normal liquid range. The surface excess adsorption isotherms are well represented over the entire density range by Fischer's three-parameter integral equation. The results are also characterized in terms of the critical parameters and the reduced variables of the adsorbates. Introduction Most of the past experimental and theoretical work on physical adsorption of gases on solid surfaces has been carried out in the subcritical region of the adsorbate. However, many important industrial applications involve the adsorption of gases a t temperatures and pressures above the critical point of the adsorbate, and yet the availability of adsorption data and models at realistic process conditions is very limited. Such processes are in the fields of coal mining; separation and purification of hydrogen, light hydrocarbons, and several other gases; storage and transport of fuel gases like natural gas in microporous activated carbons; and catalytic reactions. Adsorption data and models are important tools for selection, design, and simulation of these processes. This paper presents the results of an experimental study of the excess adsorption of argon, neon, krypton, methane, and nitrogen on a microporous activated carbon (GAC250) a t room temperature and pressures up to 650 MPa, corresponding to reduced temperatures TIT, up to 6 and reduced densities pIpc up to 3.25. In sections I and I1 we describe, respectively, the experimental method and the sample preparation. In section 111we interpret our results on the basis of existing phenomenological models and theories on virial expansions and fluid-solid interactions.
I. Experiments
capacitance, CE,and consequently to an equilibrium density, pi, determined by the Clausius-Mossotti function (CM): (1)
where e (=CE/Co) is the measured dielectric constant and A,, B,, and C, are, respectively, the first, second, and third dielectric virial coefficients of the adsorbate. Depending on the pressure range of the experiment and the type of the measured gas, contribution to CM from B, and higher Coefficients can be more or less significant. After reaching equilibrium, the gas in the capacitance cell is expanded into the sample cell containing the adsorbent and the new equilibrium density, Pb, is again determined by eq 1. The excess amount of adsorption, at T and Pb (or T and the corresponding pressure P),is then determined by ns = n - pb(v- vs)
(2)
where n = (pi - Pb)Vc is the total amount of gas in the volume V occupied by the adsorbent and VS is the skeleton volume of the solid adsorbent (pores excluded). Vs is determined by volumetric means using a reference gas for which na is zero by definition. The reference gas is usually helium at low pressure. The adsorption isotherm is constructed point by point by maintaining a balance of alternate fill ups of the capacitancecell and expansions into the sample cell. For an adsorbent having a total surface area Ast, the excess adsorption per unit area rs is given by
The experimental method used in this study is similar to the standard volumetric method for measuring adsorption but uses the dielectric technique for the accurate measurement of the density of gases at high pressure.' It consists of a capacitance cell of volume V, and geometric capacitance COconnected by a valve to a sample vessel of volume V filled with the adsorbent. The whole assembly is maintained at a constant temperature,T. First, the two cells are evacuated. The capacitance cell is then filled with the adsorptive gas (adsorbate) up to an equilibrium
The experimental apparatus used in this study and the procedure for adsorption measurements up to 200 MPa have been described in detail in a previous paper.2 However, with the extension of the measurement range up to 650 MPa, the use of A, alone to calculate the density was no longer valid and higher order dielectric virial coefficients were taken into account in eq
(1) Bose, T. K.; Chahine, R.; Marchildon, L.; St-Arnaud, J. M. Reo. Sei. Instrum. 1987, 58, 2279.
(2) Vidal, D.; Malbrunot, P.; Guengant, L.; Vermesse, J.; Bose, T. K.; Chahine, R. Reo. Sci. Instrum. 1990,61, 1314.
ns rs = Ast
0743-7463/92/2408-0577$03.00/00 1992 American Chemical Society
(3)
Malbrunot et al.
578 Langmuir, Vol. 8, No. 2, 1992
Table I. Excess Adsorption of Argon, Neon, Krypton, Methane, and Nitrogen on GAC-250 as a Function of the Density at 25 OC
Kr
Ar Pb,
na,
mol/L
mg/g
1.019 66 1.53 1.045 320 3.59 1.062821 4.95 1.118 088 9.14 1.243 210 18.11 1.296 772 21.77 1.350492 25.34 1.434 684 30.77 1.489 201 34.21 1.524303 36.39 1.562135 38.73 1.585909 40.18
140.1 180.6 187.2 173.6 96.9 64.7 36.5 -3.7 -27.4 -41.6 -58.9 -66.5
a
Pb,
t
mol/L
1.012889 0.68 371.3 2.28 489.8 1.043 596 1.073 174 3.79 500.8 5.69 475.7 1.111 495 1.248 291 12.12 317.8 1.345 613 16.40 202.2 114.3 1.432317 20.05 41.9 1.515 959 23.45 1.621 674 27.58 -38.1 1.686 563 30.04 -81.1 1.730094 31.67 -116.5 1.770025 33.14 -142.5
mol/L
n', mg/g
1.006014 2.01 1.009 476 3.16 1.012 786 4.26 1.022996 7.64 1.049 958 16.49 1.082 695 27.09 1.122475 39.76 1.150846 48.68 1.170 712 54.87 1.186 228 59.68 1.198 546 63.48
6.5 9.8 12.3 19.6 26.4 17.6 -6.4 -29.2 -45.9 -59.5 -69.3
Pb,
z
1. These coefficients were obtained for each gas by fitting eq 1 to measured values of the dielectric constant as a function of the density. Three coefficients were used in the case of neon, four in the case of argon and methane, and five in the case of krypton and nitrogen. The accuracy of the density measurements3 is about 0.05%. Also,in additiontothe correctionsof high-pressure effects on the volume of cells and sample holder: the compressibility of the solid adsorbent was taken into account and VSwas
corrected using carbon compressibility data.' This correction factor amounts to 2% of the skeleton volume at the maximum pressure. The maximum overall uncertainty of the excess adsorption results presented in this paper varies between 1% and 5% depending on the gas studied. However,the reproducibility of the results is better than 2%.
11. Sample Preparation Coal-derived activated carbon GAC 250, supplied by Soci6t6 CECA (France),was used in all the measurements. It is a microporous adsorbent with a specific area of 1030 m2/g and a cumulative pore volume of about 0.55 cm3/g. The measured samples,all taken from the same lot, were outgassed under highvacuum conditions at 400 O C for 12 h and kept in sealed glass capsules2under a residual pressure of less than lWa Pa. The helium volume and the porosity of the adsorbent samples were determined before and after the high-pressure measurements, and no difference was found. Gas adsorbates were supplied by L'Air Liquide with the following purities: >99.9% for argon, neon, and nitrogen, >99.95% for krypton; and >99.995% for methane.
111. Results and Discussion Adsorption measurements of argon, krypton, neon, methane, and nitrogen at 25 "C were carried out using the dielectric method. The results of the measurements are listed in Table I where the high ends of the measured densities correspond to pressures varying between 600 and 650 MPa with the exception of krypton where the highest density corresponds to 513 MPa. All the measured isotherms exhibit a maximum followed by a linear decrease to negative values. The position and the amplitude of the observed maximas are in agreement with published re~ults.~-llHowever, our results and those of Moffet and Weale5are the only ones which clearly show negative excess (3) Vermesse, J.; Vidal, D.; Malbrunot, P. Int. J. Thermophys., submitted for publication. (4) Birch, F. In Handbook of Physical Constants; Clark,S. P., Ed.; The Geological Society of America, 1966;Section 7, memoir 97. (5) Moffat, D. H.; Weale, K. E. Fuel 1955,34,469. (6)von Antropoff, A. Kolloid-Z. 1955, 143, 98. (7) Palvelev, V. T. Bull. Acad. USSR, Cl. Sci. Technol. 1945, 578. (8) Palvelev, V. T. Dokl. Akad. Nauk SSSR 1948,62,779. (9) van der Sommen, J.; Zwietering,P.;Eillebrecht, B. J. M.;Krevelen, D. W. Fuel 1955, 34, 444.
Nz
CHI
Ne na, mg/g
Pbt
ne,
t
mol/L
mg/g
1.060060 1.231354 1.367 516 1.487 423 1.578 249 1.647 124 1.698912 1.741 919 1.776 266 1.793419
2.99 10.88 16.60 21.34 24.76 27.28 29.14 30.66 31.86 32.45
83.5 57.9 32.5 13.4 1.5 -6.6 -13.0 -18.3 -20.9 -22.2
Pb,
c
n',
mol/L ma/a
1.021 680 1.63 1.046 266 3.46 1.061 033 4.56 1.117 941 8.69 1.222 123 15.81 1.257 306 18.08 1.298 597 20.69 1.385 061 25.97 1.434 588 28.96 1.471 087 31.12 1.505 910 33.34 1.527368 34.70
91.2 109.1 110.9 96.9 41.5 27.2 13.0 -8.0 -11.2 -14.9 -15.2 -14.5
adsorption in the high-pressure zone. This could be explained by a possible change in the pore space under high external pressure.12 An early phenomenological modeling of adsorption behavior at high pressure was given in the case of methane adsorption on ~ o a l . ~ JInl this two-phase model, the adsorbed moleculesare considered to be in a condensed phase of density pa. Hence, if a surface saturation is reached at high pressures so that pa became constant, the maximum of the isotherm should be followed by a downward straight line intercepting the bulk density axis at pa which is found to be in the order of magnitude of liquid or solid density of the adsorbate.12 Our results seem to be in concordance with this model; the intercepts of the isotherms with the density axis do indeed occur slightly ahead of the corresponding liquid densities of the adsorbates in the normal liquid range with the exception of neon and nitrogen. For neon, the intercept is far ahead of its normal liquid density. For nitrogen, there is a leveling of the isotherm below the density axis. This kind of behavior was also observed in the case of Nz and CO adsorption on alumina at high pressures and different temperatures.13J4 On the basis of the alumina results and other considerations, a different model has been propo~ed.'~J~ In this model, the adsorbed gas behaves as a mobile twodimensional layer with restricted freedom for translational movements. At high pressures the adsorbed N2 molecules undergo greater entropy loss than other molecules, suggesting new restrictions on the rotational freedom of Nz. The onset of lateral interactions and the consequent restricted mobility are explained as being due to the overcrowdingof adsorbed moleculeson the relatively more active parts of the surface rather than being distributed uniformly on the entire surface.12 In terms of gas-solid interactions, the behavior of the excess adsorption isotherm a t high pressures has been attributed9 to high repulsive interactions between the molecules that outweigh their attractive interactions with the adsorbent. For a system of hard spheres interacting with an adsorbing wall at temperatures well above the (10) Specovius, J.; Findenegg, G. H. Ber. Bunsen-Ges. Phys. Chem. 1978, 82, 174. (11) Findenegg, G. H.; Korner, B.; Fischer, J., B o b , M. Ger. Chem. Eng. 1983,6, 80. (12) Menon, P. G. Chem. Reu. 1968,68, 277. (13) Michels, A.; Menon, P. G.; ten Seldam, C. A. R e d . Trau. Chim. Pays-Bas 1961,80, 483. (14)Menon, P. G. J. Am. Chem. SOC.1965,87,3057. (15) de Boer, K. H.; Menon, P. G. Pr0c.K. Ned. Akad. Wet. 1962,B65, 17. (16)Menon, P. G.; Ramamurthy, P. Kolloid 2.Z . Polym. 1965, 206, 159.
Langmuir, Vol. 8, No. 2, 1992 579
Adsorption Measurements of Ar, Ne, Kr, N , and CH4
Ar T'- 2 4 9
Kr 1.79
CH,
Ne 8.84
2.1
i
N2 3.14
OC
L 02
0.0
41
t '
0
'
10
I
20
'
I
"
I
30
0,O
'
I
02
1
l
04
1
l
0,B
0,8
P (moldl)
~
1,O
(
~
12
I
~
1,4
P*
Figure 1. Surface excess adsorption isotherms as a function of the bulk density of the adsorbates. Points are experimental results, and lines are generated by eq 6. The fit parameters for each gas are listed in Table 11.
Figure 2. Surface excess adsorption behavior in terms of the reduced variables using the fluid-fluid parameters of the adsorbate listed in Table 111.
Table 11. Fit Parameters of Equation 6 for GAC-250at
Table 111. Parameters of the Lennard-Jones Fluid-Fluid Potentialsz1
25
adsorbate argon krypton methane neon nitrogen
B, nm 5.13 16.63 21.34 0.17 4.51
O C
-k, nm (L/mol) 0.167 0.626 0.829 0.0046 0.207
4,L/mol
0.76 2.18 3.01 0.018 0.71
critical temperature, it has been shown1' that the surface excess adsorption defined in eq 3, rS,can be represented at low density by a virial expansion
rS= Bp + Cp2 + Dp3+ ...
(4)
where B, C, and D are the gas-solid virial coefficients. The expansion starts approximately like a geometrical series with alternating signs. It was suggested18that Fscan then be approximated over a wider density range by the sum of an infinite geometrical series
=Bp 1
+ qp
with q = -C/B. In order to take account of the maximum of the isotherm, eq 5 was extended in an empirical way to -BP-kP 1 +qp
2
which is consideredl9 as a Langmuir-type isotherm. Equation 6 was applied1° to adsorption isotherms of argon and methane on Graphon up to 15 MPa and at different temperatures. The fit was within experimental accuracy in the case of argon, but serious deviations occurred for the methane isotherms as these extended to densities beyond the maximum in surface excess.18 Even though eq 6 did not fit well the earlier results beyond the maximum, we find a good agreement of our results with the proposed model over the entire density range for all gases except nitrogen. We used a least-squares method to fit the data, and the results of the fitting are shown in Figure 1. The deduced parameters for each gas are given in Table 11. The representation of our experimental results in terms of the reduced parameters ( p l p c and TIT,) of the absor(17) Fischer, J. Mol. Phys. 1977, 34, 1237. (18) Fischer, J.; Specovius, J.; Findenegg,G. H. (!hem.-1ng.-Tech.1978,
-.
50.
(19) Findenegg, G. H. Proceedings of the Engineering Foundation Conference, 1983; Mayers, A. L., Belfort, G., Eds.; p 207.
adsorbate argon krypton methane neon nitrogen
u,nm
0.341 0.368 0.379 0.276 0.37
elk, K 119.8 166.7 142.1 33.7 95.0
P.at 25 "C
2.49 1.79 2.10 8.84 3.14
bates exhibits the following characteristics: The amplitude of the isotherm and the slope of the linear section vary inversely with the temperature. The maximum of the isotherm flattens out and is shifted to higher densities with increasing temperatures. It is located between 0 . 2 ~ ~ and 0 . 7 ~ The ~ . intercept of the downward line with the density axis varies inversely with temperature, and it is located between 1 . 5 and ~ ~2.3~~. The above described behavior could be related to the pore geometry of the adsorbent as it has been shown in a recent theoretical investigation of adsorption at supercritical temperatures in slitlike carbon micropores.20 In this investigation, the nonlocal density functional theory and grand canonical Monte Carlo simulations were used to study the effects of high temperatures and pore width on the isotherms. Lennard-Jones (LJ)pair potentials were used to calculate the fluid-solid and the fluid-fluid interactions,and all variables were reduced using the fluidfluid parameters of the adsorbate. Figure 2 shows our results in terms of the reduced variables Fa*,p*, and P: Fa* = r&u2
p* = pLu3
T* = kT/c
(7)
where L is Avogadro's constant and u and t are, respectively, the fluid-fluid distance and energy parameters in the LJ pair potential. The values used21 in this representation are listed in Table 111. Although a direct comparison of our results with the predictions of the simulation for methane20 is rather difficult to make, the latter extended only to p* = 0.3, there is nevertheless a good agreement in the amplitude and the position of the maximum (for slit widths >20u). For the carbon used in our measurements and at room temperature, the maxima of argon, krypton, methane, and nitrogen are located around p* = 0.1 and the maximum for neon is around p* = 0.2. (20) Tan,2.;Gubbins, K. E. J. Phys. Chem. 1990,94,6061. (21) Le Neindre, B. In High Pressure Chemistry and Biochemistry; van Eldik, R., Jonas, J., Eds.;D. Reidel Publishing,Co.: Dordrecht,1987; p 51.
l
~
580 Langmuir, Vol. 8, No. 2, 1992
IV. Conclusions We have presented results of the adsorption of spherical and nonspherical gases on activated microporous carbon at room temperature and over a very wide density range. With the exception of nitrogen, all the measured excess adsorption isotherms show a similar behavior. The surface excess adsorption isotherms are well represented over the entire densityrange by Fischer’s three-parameter empirical equation. The results were also characterized in terms of the critical parameters and the reduced variables of the
Malbrunot et al.
adsorbates. The results for methane are in qualitative agreement with simulation results based on the nonlocal mean-field density functional theory; a more rigorous comparison was difficult to make because of the limited density range of the simulation and the lack of experimental results in the high-density region. We hope that our present results will help the extension of the theoretical models to higher densities. Registry No. C, 7440-44-0;Ar, 7440-37-1; Ne, 7440-01-9;Kr, 7439-90-9; Nz,7727-37-9; CHI, 74-82-8.
