Elsevier

Carbon

Volume 49, Issue 1, January 2011, Pages 117-125
Carbon

An experimental evaluation and molecular simulation of high temperature gas adsorption on nanoporous carbon

https://doi.org/10.1016/j.carbon.2010.08.050Get rights and content

Abstract

A combination of experiments and molecular simulations has been used to further understand the contribution of gas adsorption to the carbon dioxide (CO2) selectivity of nanoporous carbon (NPC) membranes as a function of temperature and under mixed gas conditions. Whilst there have been various publications on the adsorption of gases onto carbon materials, this study aims to benchmark a simulation model with experimental results using pure gases. The simulation model is then used to predict mixed gas behaviour. These mixed gas results can be used in the assessment of NPC membranes as a suitable technology for both carbon dioxide separations from air-blown syngas and from natural gas. The gas adsorption experiments and molecular simulations have confirmed that CO2 is more readily adsorbed on nanoporous carbon than methane (CH4) and nitrogen (N2). Increasing the temperature reduces the extent of adsorption and the CO2 selectivity. However, the difference between the CO2 and N2 heats of adsorption is significant resulting in good CO2/N2 separation even at higher temperatures.

Research highlights

Gas adsorption of CO2, N2 and CH4 on nanoporous carbon determined experimentally. ► Experimental results used to validate molecular simulations. ► Simulations extended to mixed gas adsorption systems. ► Increasing the temperature reduces carbon dioxide sorption and the CO2 selectivity. ► However, good CO2/N2 separation is observed even at higher temperatures.

Introduction

Nanoporous carbon (NPC) membranes manufactured from the pyrolysis of polymers have attracted significant research interest over the two last decades due to the high selectivities reported [1], [2], [3], [4], [5], [6]. The suitability of NPC membranes for separation of carbon dioxide (CO2) from nitrogen (N2) in syngas, produced during the air-blown gasification of coal is of particular interest [7]. Also, separation of carbon dioxide from methane (CH4) is a common operation in the natural gas industry that might be enhanced through the use of such membranes.

An ideal NPC membrane will have a narrow pore size distribution, which is centred at a size just larger than that of the desired penetrant thus facilitating gas separation via molecular sieving. In this study the desired penetrant is CO2, which has a kinetic diameter of 3.3 Å, and as such a pore size distribution centred around 4–5 Å would result in reasonable molecular sieving for this molecule. However, in a real manufactured NPC membrane there are often some pores present in the 5–10 Å range [8], which allow both larger and more polar gas molecules to adsorb onto the pore walls and move through the membrane via the surface diffusion mechanism [9]. The adsorptive behaviour of the polar CO2 molecules on nanoporous carbon is therefore an important consideration in further understanding the mechanism of CO2 transport through NPC membranes. The objective of this study is to determine the CO2 adsorptive selectivity over CH4 and N2 of nanoporous carbon with increasing temperature in both pure gas and gas mixtures using a combination of molecular simulations and gas adsorption experiments.

The adsorption behaviour of gases on nanoporous carbon generally falls outside the simplest model for adsorption, Henry’s Law, especially for the more adsorptive gases like CO2 [10]. Type 1 adsorption, which occurs when the pores are extremely small, is more common for materials like nanoporous carbon and can be fitted well using the Langmuir model [11]. The Langmuir model assumes that the adsorption energy is constant and independent of surface coverage, and that the maximum adsorption (Vm) occurs when the surface is covered by a monolayer of adsorbate. The amount of gas adsorbed (V) at a particular pressure (p) is described by the Langmuir equation [12]:V=Vmbp1+bp

The parameter b determined at various temperatures, can then be used to calculate the heat of adsorption (ΔHads) from an Arrhenius-type relationship [10]:b=b0expΔHadsRTSubstituting Eq. 2 into Eq. 3 gives the following form of the Langmuir model as a function of temperatureV=Vmb0expΔHabsRTp1+b0expΔHabsRTpThis model assumes that Vm is independent of temperature.

For a mixed gas with components A and B, the fractional adsorption (θ) of one component (A) onto the substrate surface from the bulk phase can be described using the dual site or extended Langmuir model [13], [14].θA=bApA1+bApA+bBpB

However, the extended Langmuir isotherm is most accurate when the bulk concentrations of the two components are similar [13], which is not the case here.

Another approach is to use the ideal adsorbed solution theory (IAST), which predicts mixture adsorption behaviour from the shape of the pure gas adsorption isotherms [15]. Chen and Sholl [16] have examined the accuracy of the IAST method for the adsorption of CO2/CH4 on carbon nanotubes using transition matrix Monte Carlo simulations. They found the IAST method based upon pure gas isotherms fitted with the dual-site Langmuir model to be accurate over a large pressure range. Ritter and Yang [17] have also examined the accuracy of several adsorption models for the competitive adsorption of CO2/CH4 on activated carbon and have found IAST to provide a better fit to the experimental data than the dual-site Langmuir isotherm. More recently, Campo et al. [18] have demonstrated that the SIPS equation, an extended version of the Langmuir isotherm (with an additional parameter) provides a good fit for CO2 adsorption.

Molecular simulation can be used to predict the adsorption of gases on a substrate (e.g. nanoporous carbon), and is a very useful computational tool for understanding the performance and transport mechanisms behind the experimental observations.

In particular, grand canonical Monte Carlo (GCMC) simulation can be used for calculating equilibrium properties such as adsorption isotherms [19], [20], [21]. NPC is commonly modelled using the coordinates of a model porous media, such as C168 Schwarzite (illustrated in Fig. 1, porosity of 0.67 and density of 1.294 g/cm3, compared with NPC, porosity of 0.73 and density of 1.6 g/cm3 [7], [8]). Although the structure of an NPC and the C168 schwarzite differ, the carbon surface curvatures are similar, so that segments of the C168 Schwarzite may provide a similar environment for adsorbates as found experimentally in an NPC. Klauda et al. [22] have studied the effect of the curved carbon surface on gas adsorption using molecular modelling of planar graphite, curved surfaces of C60 fullerene and C168 Schwarzite. The interaction potentials between nitrogen and the C60 and C168 surfaces were found to be different than those for the planar graphite, resulting in higher adsorption energies on the curved surfaces. However, the curved surface only had a small effect on the interaction potentials between oxygen and the carbon surfaces. The authors suggested that this difference in adsorption energies on the curved carbon surface might explain the high O2/N2 selectivity observed experimentally in NPC membranes by Shiflett and Foley [5], [23].

Using grand canonical Monte Carlo (GCMC) simulation, Jiang et al. [19] modelled the adsorption of oxygen (O2) and nitrogen (N2) in C168 Schwarzite. They found that at low pressure the adsorption of N2 is favoured, but as the pressure is increased the adsorption of O2 dominates due to its smaller size. The simulation isotherms were fitted well to the two-site Langmuir model. In a later publication, Jiang and Sandler [24] examined the mixed gas adsorption of O2 and N2 using both GCMC and Gibbs ensemble Monte Carlo simulation.

Arora and Sandler [25], [26] argued that the O2/N2 adsorption selectivity of the curved C168 Schwarzite surface predicted by molecular modelling could not explain the experimental results observed by Shiflett and Foley [5], [23], and it is also necessary to understand the diffusion of these gases through nanoporous carbon. Arora and Sandler [26] concluded that using an ab initio potential they developed for describing the interaction between the gas molecules and the carbon surface better explains the high selectivity seen experimentally. To understand the effect of pore size of nanoporous carbon on diffusion, they [25] subsequently performed molecular dynamics simulations using carbon nanotubes with constrictions that represented the pore size of nanoporous carbon and allowed for molecular sieving. If the appropriate-sized constriction was present, O2 passed through the nanoporous carbon more readily than N2 resulting in a greater overall O2/N2 selectivity.

With the purpose of studying the suitability of nanoporous carbons for separating CO2 from N2 in a typical power station flue gas, Jiang and Sandler [21] have used a combination of quantum mechanics (to obtain the ab initio potential) and GCMC to predict the adsorption. They claimed that the ab initio potential provides a more accurate adsorbate-adsorbent interaction potential for nanoporous carbon, with an adsorption selectivity of CO2/N2 of about 160 being observed for a binary mixture (21:79 mol% of CO2:N2) at ∼100 kPa and 27 °C.

In our previous experimental work [7], mixed gas permeance measurements for the CO2/CH4 gas pair on nanoporous carbon at elevated temperatures revealed a reduction in selectivity with increasing operating temperature. This was a result of the larger increase in the CH4 permeance with increasing temperature due to an increase in its rate of diffusion. The smaller change in the CO2 permeance was thought to be related to loss of the adsorption component of the CO2 permeance at higher temperatures. This result agrees with other published experimental work that revealed a small change in the CO2 permeance leading to a reduction in CO2/CH4 selectivity with increased operating temperature [1], [27], [28], [29]. Molecular simulations of the adsorption of a CO2/CH4 gas mixture in carbon nanopores by Sedigh et al. [29] showed that CO2 is indeed adsorbed more strongly than CH4 onto the pore wall and that the adsorption decreases with increasing temperature.

In the present work, experimental measurements of the adsorption of pure gases (CO2, CH4 and N2) on nanoporous carbon at a range of temperatures are used to validate a molecular simulation model. This model is then used to predict the adsorption behaviour under mixed gas conditions and at elevated temperatures.

Section snippets

Experimental adsorption

Samples of bulk nanoporous carbon were made by the pyrolysis of polyfurfuryl alcohol in a crucible at 550 °C with a soak time of 2 h under an ultra high purity argon purge of 200 ml/min. The adsorption isotherms of the pure gases (CO2, N2 and CH4) on the nanoporous carbon at temperatures up to 200 °C were measured using a gravimetric high pressure (GHP-300) analyser fitted with a Cahn microbalance (VTI Corporation, USA) shown in Fig. 2. To obtain the adsorption isotherms the GHP-300 was operated in

Pure gas adsorption

The measured and simulated adsorption isotherms of pure CO2, CH4 and N2 on the nanoporous carbon are presented in Fig. 3, Fig. 4, Fig. 5, respectively. The adsorption behaviour for all three gases follows a Type 1 isotherm with very little hysteresis as is expected for a material with nano-sized pores [11]. Adsorption of CO2 is highest on nanoporous carbon followed by CH4 and then N2. Additionally, as the operating temperature of the system is increased, the loading decreases more prominently

Conclusions

The experimental and simulated isotherms for the adsorption of pure gases on NPC show that the value of the heats of adsorption are in order of CO2  CH4 > N2. Also, the simulation model provides a reasonably good match to the experimental data. The deviations between the simulation and the experimental data may be caused by larger amphorous pores that exist in laboratory nanoporous carbon that are difficult to model.

Simulations of the adsorption of CO2:CH4 (10:90 vol.%) and CO2:N2 (10:90 vol.%)

Acknowledgements

The authors would like to acknowledge the funding provided by the Australian Government through its CRC Program to support this research. Infrastructure support from the Particulate Fluids Processing Centre, a special research centre of the Australian Research Council is also gratefully acknowledged.

References (39)

  • G. Arora et al.

    Nanoporous carbon membranes for separation of nitrogen and oxygen: insight from molecular simulations

    Fluid Phase Equilib

    (2007)
  • A.B. Fuertes et al.

    Preparation of supported carbon molecular sieve membranes

    J Membr Sci

    (1998)
  • J. Gilron et al.

    Knudsen diffusion in microporous carbon membranes with molecular sieving character

    J Membr Sci

    (2002)
  • A. Hirotani et al.

    Grand canonical Monte Carlo simulation of the adsorption of CO2 on silicalite and NaZSM-5

    Appl Surf Sci

    (1997)
  • I. Menendez et al.

    Aging of carbon membranes under different environments

    Carbon

    (2001)
  • S. Lagorsse et al.

    Carbon molecular sieve membranes. Sorption, kinetic and structural characterization

    J Membr Sci

    (2004)
  • M.B. Shiflett et al.

    Ultrasonic deposition of high selectivity nanoporous carbon membranes

    Science

    (1999)
  • M.B. Shiflett et al.

    Characterization of supported nanoporous carbon membranes

    Adv Mater

    (2000)
  • S. Brunauer et al.

    Adsorption of gases in multimolecular layers

    J Am Chem Soc

    (1938)
  • Cited by (0)

    View full text