SciELO - Scientific Electronic Library Online

vol.39 número2Change detection in time series using the maximal overlap discrete wavelet transformAnalytical/numerical solution for confined jet diffusion flame (sandia flame c) índice de autoresíndice de materiabúsqueda de artículos
Home Pagelista alfabética de revistas  

Servicios Personalizados




  • No hay articulos citadosCitado por SciELO

Links relacionados

  • No hay articulos similaresSimilares en SciELO


Latin American applied research

versión impresa ISSN 0327-0793

Lat. Am. appl. res. vol.39 no.2 Bahía Blanca abr./jun. 2009



Equilibrium adsorption of binary mixtures of light hydrocarbons in activated carbon

M.R. Amora JR., D.V. Canabrava, M. Bastos-Neto, A.E.B. Torres, C.L. Cavalcante JR. and D.C.S. Azevedo

Grupo de Pesquisa em Separações por Adsorção (GPSA), Departamento de Engenharia Quimica, Universidade Federal do Ceará, Campus do Pici, Bl. 709 - 60455-760, Fortaleza, CE, Brazil,

Abstract - Several studies have been reported with the purpose of storing NG in porous media by adsorption. Experimental data and mathematical models that describe the adsorption phenomena of pure methane, the main component of NG (about 90%) already exist, however they do not describe satisfactorily the behavior of NG storage cycles, because of the presence of several other hydrocarbons like ethane, propane, butane and heavier alkanes. This study presents measurements of experimental data for the adsorption equilibrium of binary mixtures of low-molecular-weight alkanes and compares them with mathematical models for multicomponent adsorption prediction based on each pure component adsorption data. Experiments were performed with the binaries methane-ethane, methane-propane and methane-butane. Experimental results were compared with predictions from the Extended Langmuir model and Ideal Adsorbed Solution (IAS) model.

Keywords - Storage; Natural Gas; Activated Carbon; Binary Adsorption.


Environmental issues associated to the use of fossil fuels are increasingly more evident, namely those related to global warming, acid rain and suspended solid particles released in the air.

Due to natural gas (NG) cleaner combustion, its use has been intensely encouraged around the world. However, a viable large-scale system of distribution and usage depends greatly on the storage technology. As an alternative to natural gas compression at high pressures (ca. 200 bar), storage by adsorption in porous media at moderate pressures (35-40 bar) emerges as a way to decrease operational costs and transport NG to remote regions not served by gas pipelines.

Activated carbon (AC) has been reported as the best adsorbent for NG storage purposes, due to its high specific surface area and high micropore volume. Also, because of its lack of affinity for water, the use of activated carbon prevents significant capacity loss due to adsorbed water.

Experimental and simulated data from our group for pure methane adsorption in activated carbon have been reported (Azevedo et al., 2007; Bastos-Neto et al., 2005a,b; 2007). Nevertheless in some cases those data may not represent satisfactorily a real NG storage system because NG is a mixture of hydrocarbons (methane being the main component) containing also ethane, propane, butane, heavier alkanes, nitrogen and carbon dioxide. As a result, a more precise model for NG storage systems should consider the effects of different compositions.

The objective of this study is to compare experimental binary adsorption data of light alkanes that constitutes NG (C1, C2 and C3) with isotherm models for multicomponent systems determined from pure component adsorption data to compare those models and analyze the effects of each adsorbed component on one another.


A. Pure Component Data

The adsorption equilibrium isotherms of pure components (C1, C2 and C3) were measured gravimetrically using a high-precision magnetic suspension balance (Rubotherm, Germany). The carbon sample WV1050 (Mead Westvaco) was initially regenerated under vacuum (10-4 Torr) at 100°C overnight (at least 8 hours). After that, the experiment temperature was set and then pure gas was introduced in the sample cell up to a given pressure. Each assigned pressure corresponds to an experimental equilibrium point. Pressure values ranged from 0 to 70 bar for methane, from 0 to 40 bar for ethane and from 0 to 15 bar for propane. The measured weight increase data were handled as described by Bastos-Neto et al. (2005b) in order to obtain absolute adsorption isotherms, including the appropriate corrections due to buoyancy effects.

B. Binary Data

The apparatus used on the binary equilibrium measurements consists of a 30-mL steel cylinder (miniprototype) filled with activated carbon WV1050, a gas chromatograph (Varian-Model CP3800) for gas phase composition analysis, and an analytical balance (± 0.5 mg) for mass difference measurement. The experimental setup is illustrated in Fig. 1.

Figure 1: Experimental system for binary adsorption measurements.

Several experiments were carried out at room temperature (25°C) varying binary mixture compositions (C1-C2 and C1-C3) according to the following steps:

1. The vessel was filled with a known quantity of activated carbon and submitted to vacuum (10-4 Torr) overnight (at least 8 hours);

2. The pressure regulator valve for C2 or C3 bottle was set up to a constant value (usually between 3 and 10 bar) and the heavier gas (C2 or C3) was fed into the vessel until that pressure was reached. After adsorption equilibrium had been reached (at least one hour), the amount of gas that had entered the vessel was obtained by mass difference;

3. The pressure regulator valve for the methane bottle was set to 35 bar and connected to the vessel. When a new equilibrium stage was attained (after at least one hour), the mass difference was measured and the mass of methane that had entered the vessel was obtained;

4. A sample of the gas phase was analyzed by gas chromatography and the adsorbed phase concentrations were calculated using the amounts of each gas (previously obtained) and adsorbent present in the vessel in equilibrium


A. Extended Langmuir

Langmuir theory for pure component may be extended to multicomponent adsorption under the same assumptions (Ahmadpour et al., 1998). For multicomponent systems the amount adsorbed by component i can be written as:


The advantage of this model is its mathematical simplicity. However for multilayer adsorption and high coverage this model did not fit experimental data well.

B. Ideal Adsorbed Solution (IAS)

This model is extensively used in multicomponent equilibrium adsorption studies (Ruthven, 1984). It does not require data for mixtures and is thermodynamically coherent. The calculation of an integral equation to obtain the reduced spreading pressure in needed.

Pure component data were fit by O'Brien-Myers equation (O'Brien and Myers, 1984):


The binary data were fit by a procedure described by Do (1998).


A. Pure Component Data

For the sake of comparison, pure component isotherms were plotted together in a relative pressure scale, P/Po (Fig. 2). Because methane is a supercritical fluid in the experimental conditions, the Dubinin equation (Do, 1998) was used to estimate a pseudo vapor pressure. One can note that higher adsorption capacities were found for heavier components, which may cause adsorbent saturation and consequently capacity loss even at lower pressures for long-term cyclic processes

Figure 2: Pure component adsorption isotherms in relative scale (P/Po). Po at 25°C: propane: 7.6 bar / ethane: 50.8 bar / methane (pseudo): 112.5 bar.

The parameters obtained for Langmuir (Eq. 1) and O'Brien-Myers (Eq. 2) models from those pure component experimental data fit are shown in Tables 1 and 2, respectively.

Table 1:. Monocomponent Langmuir model parameters.

Table 2: Monocomponent O'Brian-Myers equation parameters.

B. Binary Data

Tables 3 and 4 show the measured adsorbed amounts for each component in mixtures C1-C2 and C1-C3, respectively, for 35 bar and 25°C.

Table 3: Measured adsorbed amounts for mixture C1-C2.

Table 4: Measured adsorbed amounts for mixture C1-C3.

C. C1-C2 isotherms

In Fig. 3 the experimental data, Langmuir fit and IAS fit are plotted for C1-C2 mixture. Note that both models are qualitatively well predictive, since higher C2 adsorbed amounts are observed compared to C1. In addition, the IAS model matched experimental points better, which was probably expected since it is more consistent with the system (alkanes on AC).

Figure 3: Models and Experimental Data for C1-C2.

The discrepancies between model and experiment may be attributed essentially to experimental inaccuracy possibly due to gas leaks and equilibrium disturbance during the GC sampling.

D. C1-C3 isotherms

Figure 4 shows a similar plot for the mixture C1-C3. The comments of the previous plot are valid for the qualitative model fittings. On the other hand, the profiles for C1-C3 present little disparity.

Figure 4: Models and Experimental Data for C1-C3.

For a short pressure increment, methane is almost not adsorbed while propane is highly accumulated in the carbon pores until its saturation.

Figures 5 and 6 illustrate the total adsorbed amount of the mixture related to methane rate for C1-C2 and C1-C3 respectively. Experimental points were restricted to the low-mass regions due to the balance capacity.

Figure 5: Total absorbed amount for C1-C2.

Figure 6: Total absorbed amount for C1-C3.

For both cases, when the methane concentration increases there is a mass capacity loss. For C1-C2 mixture this capacity loss is more gradual than for C1-C3 mixture. On the other hand, the lower the methane fraction the quicker the adsorbent will become saturated on heavier components in long-term use.


The adsorption data of pure components were obtained through gravimetry, a widely-used method with high accuracy granted by the magnetic suspension microbal-ance (uncertainty <0.002%). However, only through gravimetry it is not possible to determine the adsorption equilibrium of mixtures. For that reason, a method based on gravimetry coupled with chromatography was used to acquire the binary data.

The adsorbed amounts for the binary cases were cal-culated by mass balances since we know the total amount that is fed and the gas phase concentration. In this step the mass of each component in the gas phase is calculated based on the concentration, pressure and temperature of the system. These estimates may be sub-ject to uncertainties because the pressure is measured with the aid of a manometer, which is not as accurate as the pressure sensor used for the pure component data acquisition in the Rubotherm suspension microbalance. Two multicomponent models widely used in the litera-ture (Langmuir and IAS) were applied to evaluate the experimental results for our binary data.

As expected, the heavier NG components were more easily adsorbed in the AC sample, which caused meth-ane to be proportionally less adsorbed. The mixture composition strongly affected the total adsorbed amount, especially for methane molar/mass fractions over 80%, which correspond approximately to natural gas composition.

Better matching was observed for C1-C3 rather than for C1-C2 when comparing model predictions. IAS model was more accurate than Langmuir model for all predictions. Despite the potential experimental inaccu-racies for the binary data measurements, we observed satisfactory fits between our experimental results and the models that were tested


bmodel parameter, bar-1
Ppressure, bar
P0vapor pressure, bar
qabsorbed amount, g/g
qsabsorbed amount at saturation, g/g
σmodel parameter

The authors acknowledge the financial support received from PETROBRAS, FUNCAP, FINEP, PRH-ANP, and CNPq.


1. Ahmadpour A., K. Wang and D.D. Do, "Comparison of Models on the Prediction of Binary Equilibrium Data of Activated Carbons," AIChE Journal 44, 740-752 (1998).         [ Links ]

2. Azevedo, D.C.S., J.C.S. Araujo, M. Bastos-Neto, A.E.B. Torres, E.F. Jaguaribe and C.L. Cavalcante Jr,, "Microporous activated carbon prepared from coconut shells using chemical activation with Zinc Chloride." Microporous and Mesoporous Materials. 100, 361 - 364 (2007).         [ Links ]

3. Bastos-Neto M., A.E.B. Torres, D.C.S. Azevedo and C.L. Cavalcante Jr., "A Theoretical and Experimental Study of Charge and Discharge Cycles in a Storage Vessel for Adsorbed Natural Gas," Adsorption-Journal of the International Adsorption Society 11, 147-157 (2005a).         [ Links ]

4. Bastos-Neto M., A.E.B. Torres, D.C.S. Azevedo and C.L. Cavalcante Jr., "Methane Adsorption Storage Using Microporous Carbons Obtained From Coconut Shells." Adsorption-Journal of the International Adsorption Society 11, 911-915 (2005b).         [ Links ]

5. Bastos-Neto, M, D.V. Canabrava, A.E.B. Torres, E. Rodriguez-Castellon, A. Jimenez-Lopez, D.C.S. Azevedo and C.L. Cavalcante Jr., "Effects of textural and surface characteristics of microporous activated carbons on the methane adsorption capacity at high pressures." Applied Surface Science. 253, 5721 - 5725 (2007).         [ Links ]

6. Do, D.D., Adsorption Analysis: Equilibria and Kinetics, Series on Chemical Engineering, Imperial College Press, London, 2 (1998).         [ Links ]

7. O'Brien, J.A. and A.L. Myers, "Physical Adsorption of Gases on Heterogeneous Surfaces - Series Expansion of Isotherms Using Central Moments of the Adsorption Energy Distribution," Journal of Chem. Soc., Faraday Trans., 80, 1467 (1984).         [ Links ]

8. Ruthven, D.M., Principles of Adsorption and Adsorption Processes, Wiley, New York (1984).         [ Links ]

Received: March 14, 2008.
Accepted: August 28, 2008.
Recommended by Subject Editor: Orlando Alfano.

Creative Commons License Todo el contenido de esta revista, excepto dónde está identificado, está bajo una Licencia Creative Commons