The reversibility of the adsorption of methane–methyl mercaptan mixtures in nanoporous carbon

CARBON

5 0 ( 2 0 1 2 ) 2 2 5 –2 3 4

Available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/carbon

The reversibility of the adsorption of methane–methyl mercaptan mixtures in nanoporous carbon Monika Golebiowska a, Michael Roth b, Lucyna Firlej Carlos Wexler a,*

a,c

, Bogdan Kuchta

a,d

,

a

Department of Physics and Astronomy, University of Missouri, Columbia, MO 65211, USA Department of Physics, University of Northern Iowa, Cedar Falls, IA 50614, USA c Laboratoire Charles Coulomb UMR 5221, Universite´ Montpellier 2, F-34095 Montpellier, France d Laboratoire Chimie Provence, Universite´ Aix-Marseille 1, Marseille, France b

A R T I C L E I N F O

A B S T R A C T

Article history:

The results of extensive molecular dynamics simulations and theoretical considerations of

Received 21 January 2011

the adsorption of methane–methyl mercaptan mixtures in slit-shaped carbon nanopores

Accepted 17 August 2011

are presented. We observe significant mobility of both methane and mercaptan molecules

Available online 24 August 2011

within the pore volume, between pores, and between adsorbed and gas phases for a wide range of temperatures and pressures. Although mercaptans adsorb preferentially relative to methane, the process remains reversible, provided non-oxidizing conditions are maintained. A mercaptan/methane ratio of the order of 200 ppm in the adsorbed phase is sufficient for the gas phase to have a mercaptan concentration above the human threshold for detection. The reversibility of the adsorption process and low concentration of mercaptans makes it unlikely that these would be harmful for adsorbed natural gas storage systems.  2011 Elsevier Ltd. All rights reserved.

1.

Introduction

Natural gas (NG) consists mainly of methane, which has hydrogen to carbon ratio higher than any other molecule used as a primary fuel. Thus NG has the highest energy density per unit mass and the lowest carbon dioxide emission per unit energy, making it an effective and relatively clean alternative for energy applications, especially when considering its lower cost compared to gasoline.1 Unfortunately, the low density of methane compared to liquid fuels makes its storage more difficult, requiring compression at very high pressures (p P 250 bar  3600 psi compressed NG, CNG) or liquefaction at very low temperatures (T  162 C, liquefied NG, LNG). Either alternative increases significantly the cost of operation both in terms of production (energy costs, equipment, safety) and storage (bulky and/or cryogenics tanks). In particular, for

the case of vehicular use, the CNG requires heavy bulky tanks, significantly reducing the available cargo/passenger space. A promising alternative is to store the fuel as adsorbed NG (ANG). This is possible through physisorption of a gas into a suitable porous solid, designed to hold the fuel at relatively low pressures (e.g., 35 bar  500 psi). Highly porous media have sufficient volumetric storage ability to store NG at densities comparable to a high-pressure CNG tank [1–3]. The lower operational pressure allows thinner tank walls and allows a conformable shape [1–3], resulting in potentially significant cost savings, improvements in safety and reduced cargo volume loss. Carbon-based nanoporous materials appear to be one of the most attractive candidates for room temperature ANG storage for various reasons: (i) low-cost, no toxicity and high availability; (ii) low weight; and (iii) the fact that isosteric

* Corresponding author. E-mail address: [email protected] (C. Wexler). 1 See: http://www.iangv.org/natural-gas-vehicles/natural-gas.html. 0008-6223/$ - see front matter  2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.carbon.2011.08.039

226

CARBON

5 0 ( 2 0 1 2 ) 2 2 5 –2 3 4

z-direction [nm]

z-direction [nm]

heats of adsorption for methane in activated carbon is 18– 20 kJ mol1 (2200–2400 K) [2,4], very close to the ‘‘Optimum Conditions for Adsorptive Storage’’ of Bhatia and Myers [5]. Some of the best performing carbons have been manufactured from organic wastes (corn cob, see Ref. [3], and a broad range of experimental methods have been applied to characterize such materials (see [6] and references therein). Furthermore, an understanding of the adsorption mechanism and kinetics at the microscopic level comes from numerical simulations. Monte Carlo [7,8] and molecular dynamics [9] methods have been applied to analyze a wide range of aspects: isotherms of adsorption [10], interaction energies [11,12], methods of adsorbent structure approximations [13,14] and migration of gas molecules into and out of the pores [15]. In the present paper, our simulations consider only the idealized slit-pore model, which captures most of the physics of the adsorption/desorption process. However, it is important to remember, that this model lacks the more complex aspects of the real structures, for example pore mouths included in more realistic models, and introducing disorder which may be needed to understand possible barriers at the pore entry [16–18]. In contrast to the progress in understanding methane adsorption, very little is known about how other important components of NG act in an ANG setup. Since methane (and other alkanes in NG) is flammable and potentially explosive, significant safety considerations must be satisfied for common use. The current procedure is to add some odorants to methane, which is odorless. Typically about 200 ppb of mercaptans are added. Such concentration enables detection of gas in air when the concentration of NG reaches 1/5th of the lower explosive limit (5% by volume in air at 200C, see, e.g., [19]). To the best of our knowledge there are no studies of the mechanism of adsorption and kinetics of mercaptans in the context of ANG systems. Since the reversibility of mer-

captan adsorption has not been established, most current studies of ANG systems are focused on completely removing mercaptans from gas mixtures (e.g., by increasing mercaptan captivation in chemically modified carbons [20]). It seems likely that a significant deployment of ANG systems would benefit by the addition of the simple safety mechanism provided by easy human detection of NG leaks. Therefore, in this paper we present a detailed computational study of the behavior of methane–mercaptan mixtures in nanoporous carbons. Our results indicate that although mercaptans adsorb preferentially as compared to methane, they still are able to migrate within sub-nm pores, and that their adsorption is reversible. We estimate that a modest increase in the concentration of mercaptans in NG (prior to adsorption) will be sufficient to have in the desorbed phase the amount of mercaptans above the human detection threshold. Our conclusions further indicate that mercaptans should not contaminate/clogg the adsorbents, thus their use in ANG systems should be possible.

2.

Simulation setup

Our simple model of activated carbon consisted of graphene-based slit pores to simulate the process of adsorption. Molecular dynamics (MD) simulations in the canonical ensemble (NVT) have been performed. A parallel MD code, NAMD, [21] was applied, with time step equal 1 fs. The simulation box of dimensions 14.0 nm · 10.0 nm · 2.10 nm is shown on Fig. 1. Three parallel graphene layers of dimensions 10.0 nm · 10.0 nm were placed in the center of the box. The separation between sheets (graphene center to center distance) is equal 0.7 nm. The periodic boundary conditions were used in all directions, yielding slits which are infinite in the y-direction, but of finite extent in the xdirection. Such configuration was deliberately chosen for

2.1

0.0 0.0

2.0

12.0 14.0 x-direction [nm]

0.0

2.0

12.0 14.0 x-direction [nm]

2.1

0.0

Fig. 1 – Side view of the 14.0 · 10.0 · 2.10 nm3 simulation box containing three graphene sheets (blue circles), methane (small green circles) and methyl mercaptan (larger, blue and yellow circles). The graphene sheets are 10.0 · 10.0 nm2, leaving space for a gas phase in equilibrium with the adsorbed phase. Top: the initial placement of gas molecules beyond slit volume. Bottom: stabilized system. Periodic boundary conditions were used in all directions. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

0.0

1.0

227

5 0 (2 0 1 2) 2 2 5–23 4

2.0 x-direction [nm]

Distribution of CH4 molecules

CARBON

0.0

0.5

1.0

1.5

2.0

x-direction [nm]

Fig. 2 – (a) Gas distribution on the edge of a typical slit. (b) Corresponding density profile of molecules in x-direction (outside the slit, x < 2 nm).

H atom S atom

-CH 3 superatom Fig. 3 – OPLS–UA representation of methane (left) and methyl mercaptan (right).

three reasons. First, it allows observing and quantifying the adsorption at the pore edges which is totally missed in the simulation with infinite pore walls. Second, the free box space (outside the pore) allows the molecules’ migration between different pores, a process which cannot be observed if the pore wall surface is infinite. Lastly, a pore width of 0.7 nm maximizes the adsorption potential depth, thus making it the most critical in terms of studies of adsorption reversibility. The systems were allowed to stabilize for 5 ns before the production runs began (checked for equilibration by verifying energy stability in a function of time). The properties of the system were obtained from the last 0.5 ns of the simulations. Since MD operates in the canonical ensemble the pressure of the gas was determined a posteriori for each run after equilibration is achieved. The simulations procedure was as follows: The initial configuration (Fig. 1a) consisted of a random placement of methane and mercaptan molecules in the volume outside the slits. Depending on pressure and temperature, the analyzed systems contained 1836–2550 (T = 195 K, dry ice temperature), 1326–2040 (T = 298 K, room temperature), and 1224–1938 (T = 320 K) gas molecules in a constant proportion (98% methane, 2% methyl mercaptan). At energetic and conformational equilibrium gas molecules adsorbed within slits and on the pores edges, while some remained in gas phase in the volume outside the slits. This is shown schematically in Figs. 1 and 2 (projection along the y direction). The amount adsorbed was calculated for each frame of the simulations. The adsorption isotherms were calculated and averaged over the frames of the production runs. The analysis of methane and mercaptan trajectories and gas-slit interaction energies were also obtained from the last 0.5 ns of simulations. Pressures of the gas mixture were obtained using the ideal gas law from the

density of molecules in the gas phase (Fig. 2b), in the regions x < 1 nm and x > 13 nm (Figs. 1b and 2).

3.

Interaction parameters/choice of force field

The interaction between methyl mercaptan, methane and carbon atoms in the graphene layers has been modeled using OPLS force field parameters [22]. The United Atom (UA) approach was used. It enabled 5-fold reduction of the simulation time with respect to time- and resource consuming All Atom (AA) simulations. In OPLS–UA representation methane molecule is described as a single super-atom. Methyl mercaptan consists of three units: a –CH3 super-atom, sulfur atom and explicit hydrogen atom in the thiol group (see Fig. 3). Nonbonded solid–fluid and fluid–fluid interactions were modeled by a Lennard–Jones (12-6) potential (calculated with a 1.5 nm cutoff), and partial charging of individual atoms/super-atoms was included (but not polarization): "   6 # X X qi qj rij 12 rij  ; þ Uðrij Þ ¼ 4eij rij rij 4pe 0 rij i j>1 where rij, eij, rij, qi are the interparticle distances, well depths, collision radii and individual charges, respectively; the set of these non-bonded parameters is given in Table 1. In order to verify the choice of force field parameters, test simulations of gas mixtures without graphene were first conducted. Two molar fractions of mercaptans were considered: 0.065 and 0.019 and systems were equilibrated at the three temperatures: 195 K, 298 K and 320 K. At room temperature and 320 K the mixtures of gases are homogeneous at both mercaptan concentrations. At 195 K the odorant molecules aggregate at and higher mercaptan concentration, but self-

Table 1 – The OPLS–UA force field parameters used in MD simulations. These parameters are comparable with previously used in the literature [8,11,12,23]. United atom CH4 super-atom S (–SH) H (–SH) –CH3 super-atom C (graphene)

r [nm]

ek1 [K]

q [e]

0.373 0.355 0.000 0.378 0.340

148.05 125.89 0.00 104.20 28.00

0.00 0.45 0.27 0.18 0.00

300 250 200 150 100 in slit on edges total

50 0 0

50

100

150

200

250

Mass adsorbed [g/kg C]

Mass adsorbed [g/kg C]

CARBON

5 0 ( 2 0 1 2 ) 2 2 5 –2 3 4

300 250 200 150 100 in slit on edges total

50 0 0

50

100

150

200

Mass adsorbed [g/kg C]

228

300

in slit on edges total

250 200 150 100 50 0 0

250

50

p [bar]

p [bar]

100

150

200

250

p [bar]

Fig. 4 – Adsorption isotherms of methane–mercaptan mixtures in 0.7 nm slits at temperatures 195 K, 298 K and 320 K (left to right).

clustering is absent when the mercaptan content. These results are consistent with the fact that methyl mercaptan is liquid below 279 K.

4.

Results and discussion

4.1.

Adsorption isotherms

Fig. 4 shows the adsorption of methane/methyl mercaptan mixture in slit-shaped graphitic pores 0.7 nm wide, at 195 K, 298 K and 320 K. In all cases we observe significant adsorption not only in the space between the graphene sheets (black curves), but also on the edges of the pores (red curves). The significance of edge-adsorption is evident, giving about 1/3 of the total mass adsorbed (blue curves) for the pore geometry under consideration.

4.2. Adsorption energy: maximum value and fluctuations, mobility In this section we present the analysis of solid–fluid interaction energies as functions of simulation time (1 frame = 100 fs) for methane and methyl mercaptan (separately), at 195 K, 298 K and 320 K and various pressures. The values of the strongest interaction energies (averaged over the number of molecules) are presented in Table 2. For each case the strongest interaction energy is observed when gas molecule is present between two graphitic sheets. Fluctuations of the interaction energies are coupled with the ability of both methane and mercaptan to migrate between the gas phase and inner volume of the slit (Figs. 5 and 6). We observed that in the relatively narrow pores considered (a 0.7 nm carbon–carbon pore leaves only a 0.36 nm opening due to the carbon width), the methyl mercaptans penetrate the pore structures by orienting the methyl–sulfur–hydrogen

plane parallel to the slit opening. In addition, the stronger methyl mercaptan interaction with the graphene (nearly 2 times stronger compared to methane–graphene), further hinders their motion. Given the small number of mercaptans (and, in consequence, poor statistics for that component of the mixture), it is difficult to obtain reliable adsorption/ desorption isotherms for this component. We thus focused on performing an energetic and dynamical analysis of the gas mixture. The time evolution of the interaction energy between graphene walls and representative methane and methyl mercaptan molecules is shown in Figs. 5 and 6, respectively, at lowest and highest pressures achieved at each simulation temperature: 6.5–206 bar at 195 K, 14.1–167 bar at 298 K, and 16.2–215 bar at 320 K. In all cases methane molecules are quite mobile, and they diffuse more as the gas pressure increases due to saturation of the deepest adsorption sites. The methyl mercaptan–substrate interaction energies also fluctuate significantly throughout simulation time, even at temperature as low as 195 K (Fig. 6). Despite high substrate– adsorbate interaction energy, mercaptan molecules remain mobile and change their positions with respect to the slit. More significant energy fluctuations appear at room and higher temperature.

4.3. Diffusion in adsorbed methane–methyl mercaptan mixtures The mercaptan–graphene interaction energy for some representative molecules correlates to the changes in the center of mass positions for the odorants in the z direction.

4.3.1.

Graphene surface

We first analyzed trajectories of molecules in a reference adsorbing system, consisting of a single graphene plane. In

Table 2 – Energies of the strongest gas–substrate interactions. 195 K P [bar]

Methane Energy [K]

Methyl mercaptan Energy [K]

298 K P [bar]

Methane Energy [K]

Methyl mercaptan Energy [K]

320 K P [bar]

Methane Energy [K]

Methyl mercaptan Energy [K]

6.5 21.5 48.1 113.0

2367 2264 2178 1856

4056 4052 3993 3939

21.0 48.4 99.5 167.0

2299 2300 2155 1917

4101 4096 4071 3982

16.2 33.7 71.5 215.0

2422 2442 2421 2419

4147 4147 4147 4146

T = 298 K

T = 320 K

p = 6.5 bar

p = 14.1 bar

p = 16.2 bar

0

0

-500

-500

-500

-1000 -1500 -2000

Energy [K]

0

-1000 -1500 -2000 -2500

-2500 0

0

0 -500

Energy [K]

Energy [K]

-1000 -1500 -2000 0

200 400 600 800 1000

-1000 -1500 -2000 -2500

-2500

0

200 400 600 800 1000 Time step

p = 215 bar

-500

-500

-2000

-2000

200 400 600 800 1000 Time step

0

-1500

-1500

p = 167 bar

0

-1000

-1000

-2500 0

200 400 600 800 1000 Time step

p = 206 bar

Energy [K]

229

5 0 (2 0 1 2) 2 2 5–23 4

T = 195 K

Energy [K]

Energy [K]

CARBON

200 400 600 800 1000

Time step

0

Time step

200 400 600 800 1000

Time step

Fig. 5 – Graphene–methane interaction energy vs. time step for various representative molecules. The variations in energy correspond to migrations between slits and edges, and to the gas phase.

T = 195 K

T = 298 K p = 14.1 bar

p = 16.2 bar

0

0

-1000

-1000

-1000

-2000 -3000

Energy [K]

0

Energy [K]

Energy [K]

p = 6.5 bar

-2000 -3000

0

200 400 600 800 1000

-2000 -3000 -4000

-4000

-4000

0

Time step

p = 206 bar

0

200 400 600 800 1000 Time step

p = 167 bar 0

-1000

-1000

-1000

-3000

Energy [K]

0

-2000

-2000 -3000

0

200 400 600 800 1000

-2000 -3000 -4000

-4000

-4000

200 400 600 800 1000 Time step

p = 215 bar

0

Energy [K]

Energy [K]

T = 320 K

0

200 400 600 800 1000

Time step

Time step

0

200 400 600 800 1000

Time step

Fig. 6 – Graphene–mercaptan interaction energy vs. time step for various representative molecules.

this case the simulation cell, similar to that depicted in Fig. 1, consisted of two graphene plates separated by 10.0 nm, therefore considered as isolated. Fig. 7 shows typical trajectories of mercaptan adsorbed in such system. At low temperatures

and low pressures mercaptan molecules tend to freely migrate over the surface of the adsorbent. The migration is significantly reduced at higher pressures due to the saturation of substrate with methane molecules yielding hindered motion

230

CARBON

5 0 ( 2 0 1 2 ) 2 2 5 –2 3 4

of the mercaptans. At room temperature we observe significant in-plane motion even at surface saturation. At higher pressures, a significant number of adsorption/desorption events occur, and some mercaptans are present in the gas phase.

4.3.2.

0.7 nm slit-shaped pores

10.0

y -direction [nm]

z -direction [nm]

The geometry of narrow, 0.7 nm wide pores provides the most severe conditions (geometric and energetic) for adsorption and generates an extreme adsorption scenario: highest adsorbed phase density and highest probability for mercaptans to be trapped inside the pores. Fig. 8 shows the trajectories of

mercaptan molecules in the simulation box, at moderate pressures and two temperatures: 195 K and 298 K. At 195 K mercaptan molecules diffuse into the pores but their motion remains constrained to a limited fraction of the pore volume. At 298 K mercaptan mobility is significantly higher; the adsorbed molecules rapidly move inside the slit and are able to probe a wide area between the slits walls. In consequence they can also desorb from one pore into the gas phase and then adsorb into another one (Fig. 8, lower left panel), i.e., three-dimensional movement is possible. Fig. 9 shows the radial distribution functions for mercaptan molecules in a 0.7 nm slit pore at 195 K, 298 K and 320 K

0.0

10.0

0.0 0

0.5 Time [ns]

0.0

10.0 x -direction [nm]

10.0

y -direction [nm]

z -direction [nm]

T = 195 K and p = 1.2 bar 10.0

0.0

0.0 0

0.5 Time [ns]

0.0

10.0 x -direction [nm]

10.0

y -direction [nm]

z -direction [nm]

T = 195 K and p = 24.8 bar

0.0

10.0

0.0 0

0.5 Time [ns]

0.0

10.0 x -direction [nm]

10.0

y -direction [nm]

z -direction [nm]

T = 298 K and p = 3.6 bar

0.0

10.0

0.0 0

0.5 Time [ns]

0.0

10.0 x -direction [nm]

T = 298 K and p = 69.7 bar Fig. 7 – Trajectories of a single methyl mercaptan on graphite surface.

CARBON

5 0 (2 0 1 2) 2 2 5–23 4

at various pressures. At 195 K the presence of multiple peaks in the distribution function show tendency of the mercaptans to aggregate, contrary to what happens in the 2% methane– mercaptan mixtures in absence of the adsorbent (gas phase). At higher temperatures, the relatively less structured distribution function is indicative of absence of aggregation. The dynamical behavior of methane and methyl mercaptan inside the graphitic slits was further analyzed via the determination of the molecules’ mean square displacement (MSD). We considered both MSDs in two dimensions (2D, the in-plane motion within a slit), and three dimensions (3D, including migration between pores). In all cases analyzed, MSDs grow linearly in time within the margin of error, indicating a normal diffusion regime. Fig. 10 shows in detail the 2D MSDs for both methane and mercaptans at various temperature and pressure values. Methane’s migration inside of the slit volume decreases with increasing pressure. At 195 K mercaptan migration inside the slit is the strongest at lowest gas pressure. With increasing temperature the mobility of mercaptans depends on the amount of gas adsorbed. However, the differences are small at and above the critical temperature. Finally, at 320 K no correlation is seen between mercaptans’ mobility and the system pressure change. Given the linear MSD vs. t observed in all cases, we calculated 2D and 3D self-diffusion coefficients of methane and methyl mercaptan: D ¼ lim t!1

1 ðMSDÞ; 2dt

where d is the dimensionality of the system (2 or 3). The resulting diffusion constants are shown in Fig. 11. Except for the low-temperature mercaptans, the relative size of the diffusion constants are consistent with the larger mass and size

231

of the mercaptan molecule; comparison between the 3D and 2D diffusion constants reveals that diffusion between pores is common for methane and rarer for mercaptan.

4.4. Required enhancement of mercaptan concentration in natural gas For human detection of natural gas leaks, the concentration of mercaptans in the gas phase should be above ca. 200 ppb (see discussion in Section 1). Since mercaptans bind to graphite more strongly than methane (Table 2), it is necessary to increase their concentration in the adsorbed phase so that any amount desorbed contains at least the required concentration. Here we present a qualitative assessment of mercaptan concentrations for detectability; therefore, because of important safety considerations, experiments will inevitably have to yield quantitative detectability results. To evaluate the required mercaptan concentration we have assumed that: (i) Interaction between adsorbed molecules will be neglected (a decent, though not quantitatively correct, assumption for supercritical adsorption); (ii) The molar volume of the adsorbed phase is much smaller than that of the gas phase (reasonable except at pressures near saturation); and (iii) The ideal gas law applies for the gas phase. With the simplifying assumptions above, integration of the Clausius–Clayperon equation [21] yields, for each species: ln P ¼ 

DHads þ C; kT

where P is the partial gas pressure, DHads is its isosteric heat of adsorption, and C is a species-dependent constant. The ratio of concentration in the adsorbed phase for each species is related to that in the gas phase by

Fig. 8 – Typical trajectories of single methyl mercaptan in 0.7 nm slit at 195 K and 298 K. Left: side view, right: top view. The upper panel depicts an adsorption event and limited in-plane diffusion. The lower panel shows various adsorption/ desorption events and both in-plane diffusion and out-of-plane movement.

232

CARBON

30

p = 6.5 bar p = 21.50 bar p = 48.11 bar p = 113.3 bar

600 400 200 0

40

Distribution of CH3SH

800

Distribution of CH3SH

1000

Distribution of CH3SH

5 0 ( 2 0 1 2 ) 2 2 5 –2 3 4

p = 21.0 bar p = 48.4 bar p = 99.5 bar p = 167 bar

20

10

0

p = 16.2 bar p = 33.7 bar p = 71.6 bar p = 134 bar p = 215 bar

30 20 10 0

0.4 0.6 0.8 1.0 1.2 1.4

0.4 0.6 0.8 1.0 1.2 1.4

r [nm]

0.4 0.6 0.8 1.0 1.2 1.4

r [nm]

r [nm]

Fig. 9 – Radial distribution functions of mercaptans inside 0.7 nm slit at 195 K, 298 K and 320 K. Note the considerable difference between the y-axis scales.

10000 8000

MSD

2000

MSD

12000

CH3SH

p = 6.5 bar 21.5 bar 48.1 bar 113 bar

2500

1500

12000

CH4 CH3SH p = 14.1 bar 33.0 bar 69.7 bar 135 bar

6000

6000

4000

4000

500

2000

2000

0.1

0.2

0.3

0.4

0 0.0

0.5

0.1

0.2

Time [ns]

0.3

0.4

CH3SH

p = 16.2 bar 33.6 bar 74.1 bar 134 bar 215 bar

8000

1000

0 0.0

CH4

10000

MSD

3000 CH4

0 0.0

0.5

0.1

Time [ns]

0.2

0.3

0.4

0.5

Time [ns]

Fig. 10 – Two-dimensional mean square displacement as a function of simulation time for methane and methyl mercaptan at 195 K, 298 K and 320 K.

2 1 0

20

CH3SH 2D

20

0

50

100

150

200

250

p [bar]

CH3SH 2D CH3SH 3D

2

CH3SH 3D

15

CH4 3D

15

8

3

CH4 3D

25

8

4

CH3SH 3D

25

D [10 m /s]

CH3SH 2D

CH4 2D

CH4 2D

D [10 m2 /s]

5 2

CH4 3D

8

D [10 m /s]

CH4 2D

6

10

10 5

5

0

0 0

50

100

150

p [bar]

200

250

0

50

100

150

200

250

p [bar]

Fig. 11 – Diffusion coefficients vs. pressure at 195 K, 298 K and 320 K. 

qthiol

qmethane

ads



qthiol

gas

ads eð DHads thiol  DHmethane Þ=kT qmethane  gas qthiol ads eð Eads ¼A thiol  Emethane Þ=kT qmethane

¼

ð1Þ ð2Þ

where A is a constant of order one (we have used the fact that the difference between isosteric heats and binding energies of adsorption are quite similar for both species [25]). In order for the mercaptan concentration in the gas phase to be above a certain desired value, that in the adsorbed phase should be enhanced by the reciprocal of the exponential factor in the equation above. Considering the case in which the adsorption energies of methane and methyl mercaptan differ the most (Table 2), to achieve the requisite 200 ppb mercaptan concentration in the gas phase, the adsorbed phase must have a mercaptan concentration of ca. 200 ppm at 298 K, and 40 ppm at 320 K. Therefore, it would be unlikely that the presence of such

small amounts of the mercatan additive in ANG could cause major problems, contaminating or clogging the adsorbents. In consequence, the use of mercaptans/natural gas mixtures in ANG systems should be possible, without significant cost or detriment to the system performance, while maintaining a safe odorant concentration in the gas phase.

5.

Summary and conclusions

The goal of this study was to analyze if the use of odorants in ANG systems, such as mercaptans, could cause significant problems to the adsorbant due to irreversible adsorption and/or pore clogging. We focused on adsorption of gas mixtures in semi-finite, 0.7 nm wide pores in a wide range of pressures and three reasonable operational temperatures 195 K (dry ice cooled systems), 298 K and 320 K (room temperature and above). The geometry of the pore was chosen to

CARBON

5 0 (2 0 1 2) 2 2 5–23 4

model the extreme situation, with strongest adsorption potentials (due to the superposition of the van der Waals contribitions from both pore walls) and molecular motion that is strongly hindered (steric constrains in narrow pores). We believe that this geometry represents the most irreversible conditions for physisorption in ANG systems, in particular of mercaptans, due to their high binding energies to the substrate. Note that under certain conditions, mercaptans may be oxidized in activated carbon (e.g., to dimethyl disulfide) [24,26], which could lead to significant irreversible adsorption. Within the scope of our simulations it is not possible to study such chemical reactions, but under the non-oxidizing conditions of our study, these reactions should not be present. This raises an important issue: activated carbons to be used for ANG tanks should have minimal presence of heteroatomic functional groups [17]. Our analysis shows that even in constrained geometries the adsorption of methane and odorant molecules remains reversible. Methyl mercaptan is able to migrate within the pore volume, desorb and/or migrate between pores. Even though mercaptans bind to the adsorbent’s surface more strongly than methane, only a relatively modest increase of mercaptan concentration in the adsorbed phase is necessary to keep its concentration in the gas phase above human detection threshold. The estimated concentration levels in the adsorbed phase are in the parts per million (vs. parts per billion in the gas phase). It appears to be unlikely that at suggested odorants’ concentrations pore clogging would be significant. From this perspective, a safe ANG tank should be able to operate continuously for numerous charge–discharge (adsorption–desorption) cycles before any additional processing is needed (such as moderate heating of the ANG tank while connected to a vacuum pump). Our results suggest that odorant molecules, insofar as reversibility of the adsorption/desorption process is concerned, may be used in ANG systems for enhanced security without major increases in cost and without adsorbent poisoning. As with any theoretical/computational results, these call for experimental validation.

[4]

[5] [6]

[7]

[8]

[9]

[10]

[11]

[12] [13]

[14]

[15] [16]

[17]

[18]

Acknowledgements We acknowledge Peter Pfeifer and Galen Suppes for helpful discussions. This work was supported in part by the California Energy Commission under Contract No. 500-08-022.

[19] [20]

[21] R E F E R E N C E S

[22] [1] Dai XD, Liu XM, Qian L, Qiao K, Yan ZF. Storage of natural gas by adsorption on activated carbon. Energy Fuels 2008;22(5):3420–3. [2] Matrangaa KR, Myers AL, Glandt ED. Storage of natural gas by adsorption on activated carbon. Chem Eng Sci 1992;47(7):1569–79. [3] Pfeifer P, Burress JW, Wood MB, Lapilli CM, Barker CM, Pobst JS, et al. High-surface-area biocarbons for reversible on-

[23]

[24]

233

board storage of natural gas and hydrogen. Mater Res Soc Symp Proc 2008;1041:1041R02. He Y, Seaton NA. Monte Carlo simulation and pore-size distribution analysis of the isosteric heat of adsorption of methane in activated carbon. Langmuir 2005;21:8297–301. Bhatia SK, Myers AL. Optimum conditions for adsorptive storage. Langmuir 2006;22:1688–700. Tafipolsky M, Amirjalayer S, Schmid R. Atomistic theoretical models for nanoporous hybrid materials. Micropor Mesopor Mater 2009;124:110–6. Gusev VY, O’Brien JA. A self-consistent method for characterization of activated carbons using supercritical adsorption and grand canonical Monte Carlo simulations. Langmuir 1997;13:2815–21. Ayappa KG, Ghatak C. The structure of frozen phases in slit nanopores: a grand canonical Monte Carlo study. J Chem Phys 2002;117(11):5373–83. El-Sheikha SM, Barakat K, Salem NM. Phase transitions of methane using molecular dynamics simulations. J Chem Phys 2006;124(123517):1–9. Sweatman MB, Quirke N. Characterization of porous materials by gas adsorption at ambient temperatures and high pressure. J Phys Chem B 2001;105:1403–11. Sweatman MB, Quirke N. Gas adsorption in active carbons and the slit-pore model 1: pure gas adsorption. J Phys Chem B 2005;109:10381–8. Albesa AG, Llanos JL, Vicente J. Comparative study of methane adsorption on graphite. Langmuir 2009;24:3836–40. Ravikovitch PI, Vishnyakov A, Russo R, Neimark AV. Unified approach to pore size characterization of microporous carbonaceous materials from N2, Ar, and CO2 adsorption isotherms. Langmuir 2000;16:2311–20. Davies GM, Seaton NA. The effect of the choice of pore model on the characterization of the internal structure of microporous carbons using pore size distribution. Carbon 1998;36(10):1473–90. Nicholson D. Simulation studies of methane transport in model graphite micropores. Carbon 1998;36(10):1511–23. Haldoupis E, Nair S, Sholl DS. Calculation of diffusion limitations in metal organic framework materials: a tool for identifying materials for kinetic separations. J Am Chem Soc 2010;132:7528–39. Nguyen TX, Bhatia SK. Kinetic restriction of simple gases in porous carbons: transition state theory study. Langmuir 2008;24:146–54. Rios RVRA, Silvestre-Albero J, Sepulveda-Escribano A, Molina-Sabio M, Rodrguez-Reinoso F. Kinetic restrictions in the characterization of narrow microporosity in carbon materials. J Phys Chem C 2007;111:3803–5. United States Federal Register 49 CFR 192.625(a), Canadian Standards Organization Z662-99 Sect. 4.17.1. Bashkova S, Bagreev A, Bandosz TJ. Effect of surface characteristics on adsorption of methyl mercaptan on activated carbons. Ind Eng Chem Res 2002;41(17):4346–52. Phillips JC, Braun R, Wang W, Gumbart J, Tajkhorshid E, Villa E, et al. Scalable molecular dynamics with NAMD. J Comput Chem 2005;26:1781–802. Jorgensen WL. OPLS force fields. In: Schlyer PvR, editor. Encyclopedia of computational chemistry, vol 3. New York: Wiley; 1998. Nguyen TX, Bhatia SK, Nicholson D. Close packed transitions in slit-shaped pores: density functional theory study of methane adsorption capacity in carbon. J Chem Phys 2002;11(23):10827–36. Bashkova S, Bagreev A, Bandosz TJ. Adsorption of methyl mercaptan on activated carbons. Environ Sci Technol 2002;36:2777–82.

234

CARBON

[25] Sircar S, Mohr R, Ristic C, Rao MB. Isosteric heat of adsorption: theory and experiment. J Phys Chem B 1999;103:6539–46.

5 0 ( 2 0 1 2 ) 2 2 5 –2 3 4

[26] Bashkova S, Bagreev A, Bandosz TJ. Adsorption/oxidation of CH3SH on activated carbons containing nitrogen. Langmuir 2003;19:6115–21.