Download as pdf or txt
Download as pdf or txt
You are on page 1of 17

pubs.acs.

org/jced Article

Use of a New Size-Weighted Combining Rule to Predict Adsorption


in Siliceous Zeolites
Published as part of the Journal of Chemical & Engineering Data Hans Hasse special issue.
L. L. Romanielo, M. G. M. V. Pereira, S. Arvelos, and E. J. Maginn*
Cite This: https://dx.doi.org/10.1021/acs.jced.9b00621 Read Online

ACCESS Metrics & More Article Recommendations *


sı Supporting Information
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
Downloaded via WESTERN UNIV on March 26, 2020 at 20:17:39 (UTC).

ABSTRACT: The results of extensive grand canonical Monte Carlo simulations of adsorption in siliceous zeolites are reported and
used to assess the accuracy of a size-weighted (SW) combining rule for Lennard-Jones cross-parameters. Adsorption isotherms,
Henry’s Law constants, and isosteric heats were computed for 45 different systems consisting of unary, binary, ternary, and
quaternary mixtures of light hydrocarbons in silicalite, pure methanol, and carbon dioxide in silicalite and light hydrocarbons in
siliceous forms of the zeolites TON, LTA, and CHA. A comparison is made with the conventional Lorentz−Berthelot (LB)
combining rule as well as other force fields and experimental data. In addition, the influence of intramolecular scaling methods and
long-range tail corrections on computed properties was evaluated. Overall, the SW combining rule gives similar, and sometimes
superior, results when compared to the LB combining rule as well as force fields tuned to experimental data.

1. INTRODUCTION and small molecules; many of these were parametrized to


Adsorption-based separation processes that utilize zeolites as obtain accurate liquid properties or to match pure component
the adsorbent are used widely in the petrochemical industry. vapor−liquid phase equilibria. Most represent nonbonded

ÄÅ É
ÅÅi y12 i y6ÑÑÑ
Examples include the separation of olefins from paraffins and interactions with the Lennard-Jones (LJ) 12-6 potential

ÅÅjj σi zz j z Ñ
Uvdw = 4εiÅÅÅjj zz − jjj zzz ÑÑÑÑ
the separation of xylene isomers. Zeolites are also used as

ÅÅjk ri z{ j ri z ÑÑ
σ
ÅÇ k { ÑÖ
catalysts, and here too, the adsorption behavior of the zeolite i
is crucial. The experimental determination of adsorption iso-
therms for pure components and mixtures is extremely time- (1)
consuming and expensive.1 As a result, molecular simulation is
where σi and εi are the size and energy parameters for atom i.
an attractive means for estimating adsorption isotherms and
Many of the force fields developed for hydrocarbon adsorption
serves as a useful complement to experimental measurements.
in zeolites1,2,4−7 follow the Kiselev model,8 which assumes that
Simulations are especially attractive as a rapid screening tool in
the silicon atoms of the zeolite do not interact with the
the search for new adsorbents. They also provide a detailed
adsorbates because silicon has a small polarizability and tends
molecular-level picture of the adsorption process, which gives
insight into how materials might be modified to achieve a given
level of performance. Among the different molecular simula- Received: June 28, 2019
tion methods, grand canonical Monte Carlo (GCMC) has been Accepted: March 12, 2020
widely used to study gas adsorption in many types of zeolites.1−7
A key input to these simulations is the force field, which
describes intramolecular and intermolecular energetic interactions.
A number of force fields have been developed for hydrocarbons

© XXXX American Chemical Society https://dx.doi.org/10.1021/acs.jced.9b00621


A J. Chem. Eng. Data XXXX, XXX, XXX−XXX
Journal of Chemical & Engineering Data pubs.acs.org/jced Article

Table 1. Structures and Unit Cell Parameters of Zeolites Used


unit cell length unit cell angle
framework (with ref) a (Å) b (Å) c (Å) α (deg) β (deg) γ (deg) supercell
MFI 54
20.070 19.920 13.420 90 90 90 3 × 3 × 3
TON55 14.105 17.842 5.256 90 90 90 4 × 2 × 6
CHA56 13.675 13.675 14.767 90 90 120 4 × 4 × 4
LTA57 11.919 11.919 11.919 90 90 90 4 × 4 × 4

to be shielded by the oxygen atoms of the lattice. This reduces Other combining rules that make unlike LJ energy parame-
the number of interactions that must be computed between ters functions of both the energy and size parameters have been
framework atoms and the adsorbate, thus speeding up the proposed.12,15 Based on these works and also on the results
calculations. reported by Romanielo et al.16 that indicate that the relative
For pure and mixed-gas adsorption simulations in zeolites, molecular size has a mean effect over azeotrope-like behavior,
one also needs to know the “cross-parameters” (εij and σij) that here, we investigate the performance of a new “size-weighted”
describe the intermolecular interactions between the unlike (SW) combining rule to evaluate the LJ energy well depth
atoms, in order to account for the adsorbate−adsorbate and parameter for all unlike atoms. For the LJ energy param-
adsorbate−adsorbent interactions. Unfortunately, there is no eters, this combining rule takes the form:
foolproof way of obtaining the cross-parameters. The Lorentz− σi σj
Berthelot (LB) combining rule is often used ε
σj i
+ σ εj
i
εij =
εij = εi ·εj (2) 2 (4)

σi + σj while the LJ size cross-parameters σij are given as the arith-


σij = metic mean, according to eq 3. The SW combining rule does
2 (3) not rely on any custom fitting and only requires pure LJ param-
but it has been claimed that the LB combining rule over- eters. Note that SW reduces to arithmetic mean in two dif-
estimates the magnitude of the cross-energy-parameter.9 Other ferent limiting conditions: (a) when the parameters related to
combining rules have also been proposed, most notably the the atom size are the same (σi = σj), and (b) when the product
geometric mean rule, in which the LJ energy parameter is com- between the two LJ parameters are the same (εi σi = εj σj).
puted in the same manner as in the LB combining rule (eq 2), We carried out a series of adsorption calculations in pure silica
but a geometric mean is also used for the LJ size parameter σij. zeolites to evaluate the accuracy of the SW combining rule
Note that these simple combining rules can lead to large errors relative to the LB combining rule. We also evaluated the accu-
in predicting the properties of mixtures. Studies involving fluid racy of the SW combining rule by comparing with results from
mixtures10−12 have indicated that these rules often fail to previous simulations and from our own calculations using
describe mixture thermodynamic properties, even for simple other force fields.
systems. Waldman et al.12 reported errors of over 50% in mixed-
rare-gas systems containing very different well depths. As a 2. METHODOLOGY
result, “custom” cross-parameters tuned to experimental data 2.1. Models. We applied the LB and SW combining rules
have been proposed. For example, Dubbeldam et al.6 proposed to the zeolite model first proposed by June et al.17 (the “June”
a set of cross-parameters to predict adsorption of alkanes in force field), which follows the Kiselev approach8 of neglecting
zeolites. It was observed that the energy well depth parameters interactions between the adsorbate and silicon atoms. GCMC
for all unlike atoms were larger than the values given by the simulations were carried out on unary, binary, ternary, and
geometric and LB combining rules. In some cases, these quaternary mixtures of n-alkanes and also CO2, methanol, and
parameters were adjusted to match experimental data such as 2-methylpentane in silicalite. Adsorption simulations of pure
adsorption capacity, adsorption heat, and Henry’s Law con- components (methane, C1; ethane, C2; propane, C3; and car-
stant.4 Such an approach can lead to more accurate simula- bon dioxide, CO2) were also performed on other zeolite types
tions, but it obviously limits the predictive ability of the force (CHA, TON, and LTA). All the zeolite structures used were
field. It is desirable to have an accurate combining rule that the purely siliceous form and were treated as rigid. Alkane
only requires pure component parameters. adsorbate molecules were all represented with the TraPPE
Bai et al.13 proposed a new force field for all-silica zeolites force field.18 The TraPPE force field parameters proposed by
called TraPPE-zeo, in which the parameters related to the Potoff et al.19 for CO2 and by Chen et al.20 for methanol were
dispersion and electrostatic forces for the oxygen and silicon used. The simulation results of pure component adsorption
atoms of the zeolite were adjusted to match experimental were compared to experimental data as well as previously
adsorption data for n-heptane, CO2, and ethanol in silicalite reported simulations. For the binary, ternary, and quaternary
(the all-silica form of MFI) and propane in theta-1 (the all- mixtures, in addition to the June force field with combining
silica form of TON). Given these zeolite parameters, LJ rules LB and SW, simulations were also performed using the
interactions between all framework atoms and other adsorbate force fields proposed by Bai et al.13 and Du et al.1 The same
atoms were assumed to follow the LB combining rule, which combining rule was used to account for interactions between
leads to a more predictive force field. The authors tested all unlike atoms.
TraPPE-zeo by computing the adsorption and diffusion The relevant unit cell parameters and supercell sizes used in
behavior of several other adsorbates and found that the force the simulation box for all zeolites in this work are listed in
field performed well. The force field has only recently been Table 1. We note that there are different crystal structures
used to predict mixed-gas adsorption.14 reported in the literature for different zeolites, and the choice
B https://dx.doi.org/10.1021/acs.jced.9b00621
J. Chem. Eng. Data XXXX, XXX, XXX−XXX
Journal of Chemical & Engineering Data pubs.acs.org/jced Article

Table 2. Lennard-Jones Parameters and Partial Charges of


the Atoms Used in This Work
atom type ε/kB (K) σ (Å) q (e) ref
Oz 89.6 2.806 −0.750a 17,13a
Si 0.0 0.00 1.500a 17,13a
CH4 148.0 3.73 18
CH3 98.0 3.75 18
CH2 46.0 3.95 18
CH 10.0 4.68 18
C(CO2) 27.0 2.80 0.700 19
O(CO2) 79.0 3.05 −0.35 19
CH3(OH) 98.0 3.75 0.265 20
O(H) 93.0 3.02 −0.700 20
H(O) 0.0 0.00 0.435 20
a
Charges were taken from Bai et al.13 and used just for adsorption
simulations of CO2 and methanol

of a given structure can have small differences on adsorption


characteristics. We have chosen to use the structure of MFI
reported by Olson et al.54 and the idealized structures given by
the International Zeolite Association for TON, CHA, and
LTA.55−57 For alkane simulations, no electrostatic interactions
were considered between the sorbates and the zeolite frame-
work. For CO2 and methanol, partial charges of −0.75e and +1.5e
were placed on the oxygen and silicon atoms of the zeolites, while
only the oxygen atoms carried LJ interactions with the sorbates.
The LJ parameters and partial charges of each atom used in
this work are presented in Table 2.
2.2. Simulation Details. It is known that the simulation of
long chain alkanes using Monte Carlo requires efficient algo-
rithms for proper sampling, especially in regions close to satu-
ration. In this work, we used the open source package Cas-
sandra V1.2, which utilizes a configurational-bias algorithm
to efficiently sample alkane insertions and conformational
changes. Details of the algorithms and acceptance rules of each
move are presented and discussed elsewhere.21
Pure gas adsorption isotherms are conventionally described
in terms of the adsorbed amount (loading) at a given tempera- Figure 2. Experimental isotherm adsorption data reported in the litera-
ture as a function of the pressure in the gas phase. At equilibrium, ture for ethane31−33 (top) and n-hexane30,34,35 (bottom) in silicalite.

Figure 1. Results of GCMC simulation in the fluid phase at 300 K. (a) Pressure vs chemical potential (μ) and (b) pressure vs density (ρ). The
colors blue and red indicate the components n-butane and n-hexane, respectively. Squares and crosses are simulations that apply and do not apply
an intramolecular scaling factor, respectively. Circles are the NIST density reference data.

C https://dx.doi.org/10.1021/acs.jced.9b00621
J. Chem. Eng. Data XXXX, XXX, XXX−XXX
Journal of Chemical & Engineering Data pubs.acs.org/jced Article

Figure 3. Adsorption isotherms for methane (a and b) and ethane (c and d) in silicalite. The closed circles are experimental data from refs 31
and 32, The open squares and crosses are GCMC simulations of this work using combining rules SW and LB, respectively. Simulations applying tail
correction are in blue, while simulations without tail corrections are in red.

the chemical potential of species i (μi) must be the same in between nearby bonded atoms. To investigate the impact of
both phases at a given temperature and pressure. Therefore, to intramolecular scaling factors on adsorption results, GCMC
obtain the adsorption isotherm using GCMC, it is necessary to simulations were carried out using two popular scaling
establish the relationship between chemical potential and procedures: (a) considering there is no LJ interaction between
pressure. Many works1,2,5,6,22,23 use a technique to initially atoms separated by up to 2 bonds, whereas a scaling factor of
determine the reference chemical potential of the fluid in the 0.5 was applied to the LJ interaction for atoms separated by 3
ideal gas state, in which only intramolecular interactions are bonds and full LJ interactions were used for atoms separated
calculated to subsequently correlate the fugacity to a prese- by 4 or more bonds (0 0 0.5 1.0), and (b) considering full
lected equation of state. In the present work, a different meth- LJ interactions only for atoms separated by 4 or more bonds
odology was adopted. GCMC simulations in the fluid phase (0 0 0 1.0). In the gas phase, a 14 Å cutoff radius was used with
were performed with set chemical potentials (shifted by a analytic tail corrections.
species-specific constant). The pressure, gas phase density, The adsorption isotherms were obtained via GCMC simula-
internal energy, and enthalpy were then computed as a func- tions in which the box has the shape and size of the supercell
tion of the shifted chemical potential. The adequacy between used. The use of analytic tail corrections in the adsorbed phase
the chemical potential and the pressure obtained was observed has been questioned because tail corrections assume that the
in relation to the predicted fluid density. radial distribution function is unity beyond the cutoff, which is
Different force fields (and authors) use of a variety of intra- not the case in a periodic system like a zeolite. To investigate
molecular “scaling factors” that reduce the interaction energy the effect of tail corrections on computed results, GCMC
D https://dx.doi.org/10.1021/acs.jced.9b00621
J. Chem. Eng. Data XXXX, XXX, XXX−XXX
Journal of Chemical & Engineering Data pubs.acs.org/jced Article

Figure 4. Adsorption isotherms of propane (a and b) and CO2 (c and d) in silicalite. The open squares and crosses are GCMC simulations of this
work using combining rules SW and LB, respectively. Simulations applying tail corrections are blue, while simulations without tail corrections are
red. Experimental data from refs 30, 32, and 33.

simulations were performed in the adsorbed phase using a 13 Å been taken for ensuring that the pressure is low enough that
cutoff with and without applying analytic tail corrections to the the coverage fraction tends to zero but the average number of
Lennard-Jones potential. Translation, insertion and deletion, molecules adsorbed in the supercell is larger than unity. Simu-
rotation, and regrowth moves (when needed) were used with lations in this region must guarantee a linear isotherm.
equal probability. The number of GCMC steps varied depending The isosteric heat (qst) was calculated by the approach given
on the difficulty of the simulation. For simple hydrocarbon by Vuong and Monson,24 in which the isosteric heat is given
isotherms such as pure methane, 2 × 106 steps were used to
ij ∂U (a) yz
by eq 6:

qst = H (b) − jjjj (a) zzzz


achieve equilibration and an additional 2 × 106 steps were

k ∂N {T , V (a)
performed for the production run, from which the average
values were obtained. However, for larger hydrocarbons such
(6)
as pure n-butane and n-hexane, 5 × 106 steps were used for
both equilibration and production runs. (b)
where H is the bulk enthalpy, taken from the GCMC simu-
From very low-pressure pure component simulations, the values lation of the gas phase. The derivative term is obtained from
for the Henry’s Law constant were evaluated directly from eq 5: the GCMC simulations in the adsorbed phase at the same
μ,P,T condition of the gas phase via the following expression:
ij ∂U yz
⟨N ⟩
jj zz
kH = lim
k ∂N {T , V̲
P→0 P (5) ⟨UN ⟩ − ⟨U ⟩⟨N ⟩
=
⟨N 2⟩ − ⟨N ⟩2 (7)
where ⟨N⟩ is the average number of molecules adsorbed in the
zeolite per unit cell. It is important to notice that some care has where U is the internal energy of the adsorbed phase (a).
E https://dx.doi.org/10.1021/acs.jced.9b00621
J. Chem. Eng. Data XXXX, XXX, XXX−XXX
Journal of Chemical & Engineering Data pubs.acs.org/jced Article

Figure 5. Adsorption isotherms of n-butane in silicalite. (a) Effect of intramolecular scaling factor and (b) effect of tail correction. The open squares
and crosses are GCMC simulations of this work using combining rules SW and LB, respectively. The colors indicate: blue (use of intramolecular
scaling factor and tail correction), red (with tail correction and no intramolecular scaling factor), and green (without tail correction or
intramolecular scaling factor).

The GCMC simulations for the mixtures followed the same potential used as input to the GCMC simulations (μ) and the
methodology used for the pure component simulations, pressure (P), and to evaluate the gas phase enthalpy (H(b)).
although the number of GCMC steps required to obtain reli- As can be seen in Figure 1, the way in which intramolecular
able statistics was much larger. For example, the quaternary scaling is applied shifts the relationship between the pressure
mixture required 1 × 108 steps for both the equilibration and and the scaled chemical potential, and the shift is larger for the
production runs. longer chain. However, the intramolecular scaling has virtually
Although practical applications of adsorption always involve no effect on the relationship between the pressure and gas
mixtures, the number of studies of multicomponent adsorption phase density. For clarity, Figure 1 only shows the results for
(experimental and theoretical/simulation) in silicalite is signifi- n-butane (C4) and n-hexane (C6) at 300 K, but similar behavior
cantly lower than studies involving pure components. To com- is observed for other systems and temperatures. The results
pare mixed-gas adsorption results obtained by the use of indicate that the TraPPE model does an excellent job matching
combining rules LB and SW in conjunction with the June force the vapor phase PVT behavior of these alkanes, consistent with
field with other force fields, we also ran GCMC simulations previous studies.18−20,26−28 The results at the other tempera-
using different sets of parameters such as those proposed by tures and for other molecules were equally accurate. Numerical
Du et al.1 and by Bai et al.13 Note that while Krishna et al.25 data are available in the Supporting Information.
applied the set of parameters proposed by Du et al.1 to study 3.2. Adsorption in MFI-Type Zeolite. As already pointed
multicomponent adsorption of alkanes in silicalite, these out by Bai et al.,13 experimental isotherm data reported by dif-
authors only evaluated the effect of the change of pressure ferent groups can vary significantly, which might be related to
with equimolar mixtures. They did not investigate different differences in methods of synthesis and pretreatment and also
compositions at fixed pressure nor were any comparisons with measurement techniques. Therefore, it is important to con-
experimental data made. sider the uncertainty of experimental data when making com-
Since experimental and simulation data are not all at the parisons with simulation results. Figure 2 shows experimental
same state point, we fit pure component experimental iso- adsorption data for ethane and n-hexane in silicalite at various
therms to the dual site Langmuir (DSL) isotherm model and temperatures. Results from different authors at nearly the same
made quantitative comparisons between simulation results and temperature show significant differences. For example, in the case
the DSL isotherms. For gas mixtures, polynomial expressions
of ethane at 0.1 bar, there is a deviation of almost 2 molecules/
were used to fit the experimental data of each component and
unit cell between experimental data sets that differ by only 8 °C,
a comparison was made between the simulations and the
whereas in the case of n-hexane at temperatures between 353 and
polynomial fit. Details of the fitting procedures are available in
374 K, this deviation is up to 3 molecules/unit cell. Also, the
the Supporting Information. Adsorption loadings were deter-
expected inflection in the n-hexane adsorption isotherm,
mined by performing three independent simulations and com-
puting the average; uncertainties were taken as the standard attributed by Smit29 to the “commensurate freezing” effect
deviation. All results are tabulated in the Supporting Information. related to the similar size of the silicalite zigzag channel is not
observed in all reported data. Only at 373 K in the data
reported by Zhu et al.30 is there a slight inflection at a loading
3. RESULTS of 4 molecules/unit cell. This variation in experimental data
3.1. GCMC Simulation of the Gas Phase. GCMC simu- means that it is not always possible to unambiguously assess the
lations of the gas phase were performed to validate the force relative accuracy of different force fields. We therefore stress that
field, to obtain the relation between the shifted chemical comparing the accuracy of simulations and experiments requires
F https://dx.doi.org/10.1021/acs.jced.9b00621
J. Chem. Eng. Data XXXX, XXX, XXX−XXX
Journal of Chemical & Engineering Data pubs.acs.org/jced Article

Figure 6. Adsorption isotherms of n-pentane in silicalite. (a) Effect of intramolecular scaling factor and (b) effect of tail correction. The open
squares and crosses are GCMC simulations of this work using combining rules SW and LB, respectively. The colors indicate: blue (use of
intramolecular scaling factor and tail correction), red (with tail correction and no intramolecular scaling factor), and green (without tail correction
or intramolecular scaling factor). Simulations are at 303 and 343 K.

Figure 7. Adsorption isotherms of n-hexane in silicalite. (a) Effect of intramolecular scaling factor and (b) effect of tail correction. The open
squares and crosses are GCMC simulations of this work using combining rules SW and LB, respectively. The colors indicate: blue (use of
intramolecular scaling factor and tail correction), red (with tail correction and no intramolecular scaling factor), and green (without tail correction
or intramolecular scaling factor). Simulations are at 303 and 373 K.

a holistic view that takes into account a wide range of data, Abdul-Rehman et al.32 The results indicate that the use of tail
which we have attempted to do here by studying many dif- corrections results in an increase in the adsorbed amount at a
ferent systems and comparing average absolute deviations. Also, given pressure and temperature.
some input variables of molecular simulation, such as the use of Figure 4 presents the results of GCMC simulation of propane
tail corrections in accounting for long-range intermolecular LJ at 300 and 373 K and CO2 at 277 and 353 K. For the C1, C2,
interactions and scaling factors for intramolecular LJ interac- and C3, the major relative deviations between simulations with
tions, can lead to significant differences in computed results and without tail corrections were observed at lower temperature
and are examined in this work. and pressure. The relative mean deviation between them was
Pure Components. The results for GCMC simulations around 12%, which is significantly greater than the uncer-
performed for methane and ethane at 300 and 373 K using the tainties of the computed loadings. For CO2, however, where
June force field with combining rules SW and LB with and the electrostatic contribution to the intermolecular interactions
without tail correction are presented in Figure 3. The results is much greater than the van der Waals contributions, the
are compared to experimental data reported by Zhu et al.31 and relative mean deviation was only about 7%. There are also
G https://dx.doi.org/10.1021/acs.jced.9b00621
J. Chem. Eng. Data XXXX, XXX, XXX−XXX
Journal of Chemical & Engineering Data pubs.acs.org/jced Article

Figure 8. Adsorption isotherms of 2-methyl-pentane in silicalite. (a) Effect of intramolecular scaling factor and (b) effect of tail correction.
Experimental data from ref 38. The open squares and crosses are GCMC simulations of this work using combining rules SW and LB, respectively.
The colors indicate: blue (use of intramolecular scaling factor and tail correction), red (with tail correction and no intramolecular scaling factor),
and green (without tail correction or intramolecular scaling factor).

Table 3. Absolute Mean Deviation (AMD) and Relative Mean Deviation (RMD) of GCMC Simulations Based on Dual Site
Langmuir Fit from Experimental Data for Adsorption in Silicalite
AMD (molecules/uc) RMD (%)
this work this work
adsorbate T (K) (with ref) LB SW previous reported simulation (with ref) LB SW
CH4 C1 30032 1.58 0.17 0.991 0.9126 0.8834 47.5 5.8
CH4 C1 33831 0.81 0.67 34.7 28.4
CH4 C1 37331 0.69 0.40 34.7 19.3
C2H6 C2 30032 1.69 0.86 0.981 0.9026 1.0034 36.0 21.4
C2H6 C2 33831 075 0.42 20.7 6.6
C2H6 C2 37331 0.54 0.5 16.6 9.9
C3H8 C3 30032 0.70 0.52 0.181 0.1926 0.2434 11.8 7.2
C3H8 C3 33831 0.41 0.38 10.5 5.6
C3H8 C3 37331 0.30 0.50 12.8 9.6
C4H10 C4 30032 0.25 0.53 0.591 0.5726 0.4234 14.7 15.0
C4H10 C4 33831 0.29 0.36 6.6 6.7
C4H10 C4 37331 0.21 0.49 10.0 11.6
C5H12 C5 30334 0.38 0.70 1.005 1.047 12.9 18.3
C5H12 C5 32334 0.35 0.65 8.7 12.5
C5H12 C5 34334 0.70 0.83 25.9 21.5
C6H14 C6 30334 0.77 1.01 1.1805 1.0060 0.907 18.9 24.6
C6H14 C6 30330 0.25 0.28 3.3 3.7
C6H14 C6 34334 0.78 0.97 19.4 21.4
C6H14 C6 37335 0.78 1.14 19.0 26.2
C6H14 C6 37330 2.29 2.73 56.5 69.5
CO2 CO2 27733 1.12 1.03 1.1012 25.2 18.0
CO2 CO2 30537 1.02 0.28 0.3312 41.7 7.8
CO2 CO2 30833 0.90 0.77 24.8 17.4
CO2 CO2 35333 0.98 0.53 0.94 24.0 14.6
CH4O C1OH 30339 2.09 1.43 1.10 33.5 26.6
C6H14 2mC5 42338 1.28 0.79 0.546 1.0212 65.2 46.5
total average 0.84 0.73 24.4 18.3

differences between the results obtained with the SW and LB used. Figure 5b presents the effect of tail correction, when the
combining rules, with the SW rule giving higher adsorption intramolecular scaling factor is not applied. As already
amounts. mentioned, the use of an intramolecular scaling factor in the
Figure 5a presents the effect of the intramolecular scaling gas phase promotes a shift in the correlation of chemical poten-
factor on the adsorption of n-butane when tail corrections are tial versus pressure at a given temperature. Also, when it was
H https://dx.doi.org/10.1021/acs.jced.9b00621
J. Chem. Eng. Data XXXX, XXX, XXX−XXX
Journal of Chemical & Engineering Data pubs.acs.org/jced Article

applied to GCMC simulations of the adsorbed phase, the where NP is the number of (simulated) points inside the
scaling factor promoted an increase in loading, especially at low experimental range and xcal i and xexp
i are the calculated and
and medium pressure. For n-butane, it is possible to observe experimental values, respectively.
that the LB rule performed better when the tail correction and The average values of Henry’s Law constants taken from
intramolecular scaling factor were used. However, the SW rule three independent simulations and isosteric heats (qst) of
presented better performance when the tail correction and n-alkanes in silicalite using LB and SW are compared with the
intramolecular scaling factor were not used. experimental data and previous simulations in Tables 4 and 5,
The results of GCMC simulations performed for adsorption
of n-pentane at 303 and 343 K and n-hexane at 303 and 373 K Table 4. Values of Henry’s Law Constant for Adsorption of
are shown in Figures 6 and 7, respectively. Considering both n-Alkanes at 300 K in Silicalitea
temperatures, the LB rule presented a better performance
kH (molecules/uc/bar)
when the intramolecular scaling factor was not used but the tail
correction was. However, the best performance of the SW rule this work
was observed when intramolecular scaling factor and tail LB SW reference data
correction were not used. methane 2.7 6.17 2.83;1 4.39;5 6.7732
Figure 8 shows the results for the branched alkane (2mC5) ethane 46.2 69.20 79.75;1 69.18;5 180.1732
at 423 K. For this component, the best performance for both propane 3.54 × 102 5.94 × 102 7.20 × 102;1 5.77 × 102;5
combining rules was obtained when the tail correction and 17.42 × 10232
intramolecular scaling factor were used. butane 4.83 × 103 8.56 × 103 9.79 × 103;1 7.65 × 103;5
9.50 × 103;32 1.24 × 10458
For the results presented so far, it is clear that input GCMC
pentane 5.0 × 104 8.93 × 104 7.05 × 104;5 1.1 × 105;32
variables such as the intramolecular scaling factor and tail 4.06 × 10558
corrections can have a significant impact on the adsorption hexane 1.6 × 105 2.89 × 105 1.7 × 104;32 1.19 × 106;5
amount, which makes it difficult to compare different models/ 1.37 × 10758
force fields and combining rules. However, in general, we a
Reference data and simulated included.
observed that when using the June force field with the SW
combining rule, the best results were obtained by not applying Table 5. Values of Isosteric Heat of Adsorption of n-Alkanes
the intramolecular scaling factor and tail corrections, with one at 300 K in Silicalitea
exception: the branched alkane, for which the best results were qst (kJ/mol)
obtained when applying both. Interestingly, the best results for
this work
the June force field with the LB combining rule were obtained
using tail corrections for all but using intramolecular scaling LB SW reference data
factor only for n-butane and 2-methyl-pentane (2mC5). There- methane 17.93 19.69 20.0; 19.48; 19.72;6 20.39;32 19.059
1 5

fore, in order to not present a biased comparison, we evaluated ethane 28.57 28.05 30.6;1 29.87;5 32.81;6 32.78;32 32.059
the absolute and relative mean deviation of the GCMC simu- propane 38.13 38.92 39.0;1 38.31;5 39.32;6 39.85;32 42.559
lations performed using the best scenario for the LB rule. Tail butane 42.7 47.84 50.0;1 49.35;5 48.46;6 48.27;32 48.7;31 48.658
corrections were applied for all components. The intramolecular pentane 57.34 58.15 59.09;5 57.6;6 41.8;32 59.458
scaling factor was used only for n-butane and 2-methyl-pentane. hexane 70 73.09 69.48;5 68.05;6 70.5;32 70.358
a
In Table 3, the deviations of some previous literature simu- Reference data and simulated included.
lation results are presented against experimental data. It is
important to notice that most of these reported simulations
used some kind of parametrization against experimental data.
Du et al.,1 Vlugt,5 and Krishna et al.23 used tuned LJ parame-
ters of cross-interactions of Oz-CH4, Oz-CH3, and Oz-CH2.
Pascual et al.7 used a set of butane isotherms to optimize their
zeolite force field. The values proposed by them (σO = 3.0 Å
and εO/kB = 93.53 K) are slightly larger than those proposed
by June et al.17 (σO = 2.806 Å and εO/kB = 89.6 K). Lu et al.36
used the set of tuned parameters proposed by Du et al.1 to
account interactions between Oz-CH4, Oz-CH3, and Oz-CH2
and applied the Jorgensen combining rule for parameters
between the following atoms: CH4−CH3, CH4−CH2, and
CH3−CH2. Therefore, their results are very similar to those
reported by Du et al.,1 who used LB for those interactions.
In the Supporting Information, numerical results and plots of
our GCMC simulations used to determine the absolute mean
deviation (AMD) along with the experimental data and previous
simulations are reported. The AMD was computed over the Figure 9. Adsorption isotherm of methanol in a MFI-type zeolite. The
range where experimental data were taken from the following closed circles are the experimental data from ref 39. The filled crosses
expression and open squares are GCMC simulations of this work using
combining rules LB and SW, respectively. The open crosses denote
NP
∑i = 1 xical − xiexp previous simulations reported by Bai et al.13 The continuous lines
AMD = represent the fit of experimental data using the dual site Langmuir
NP (8) (DSL) model.

I https://dx.doi.org/10.1021/acs.jced.9b00621
J. Chem. Eng. Data XXXX, XXX, XXX−XXX
Journal of Chemical & Engineering Data pubs.acs.org/jced Article

Figure 10. Adsorption of the binary mixtures in silicalite at 300 K and 3.45 bar: (a) C1(1) + C2(2), (b) C1(1) + C3(2), (c) C1(1) + C4(2),
and (d) C2(1) + C3(2). The colors indicate the components: (1) blue and (2) red. Closed circles are the experimental data reported by Abdul-
Rehman et al.32 The squares are GCMC simulations using the SW combining rule, and the crosses represent the LB combining rule. The open
squares and thick crosses represent results using the intramolecular scaling factor and tail corrections, while filled squares and thin crosses
represents results that did not apply the intramolecular scaling factor and tail corrections. The continuous lines are polynomial fits of the
experimental data.

respectively. These results were obtained from GCMC simula- combining rules for the adsorption of methanol in silicalite. All
tions where the scaling factor (0 0 0.5 1) and tail correction simulations underestimate methanol loading in the very low-
were applied. The uncertainty obtained from the standard pressure range (<0.004 bar). Between 0.004 and 0.01 bar, the
deviation of three independent simulations was less than 1%. simulations using SW agree well with experimental data. How-
However, there is significant variability among the different ever, for pressures greater than 0.01 bar, the LB and SW com-
experimental data sets, as well as between previous simulations, bining rules overestimate methanol loading while TraPPE-zeo
especially for the Henry’s Law constant, which makes compari- agrees very well with the experimental data.
son difficult. The high variability of the simulations results can Binary Mixtures. The effect of applying the intramolecular
arise from many aspects, including the GCMC inputs adopted, scaling factor and tail corrections on adsorption results was
such as applying or not applying the intramolecular scaling investigated for binary mixtures. Figure 10 presents the binary
factor and tail corrections. For instance Bai et al.13 used the tail mixture adsorption isotherms of C1(1) + C2(2), C1(1) +
correction on their GEMC simulation, while Du et al1 and C3(2), C1(1) + C4(2), and C2(1) + C3(2) in silicalite at 300 K
Vlugt et al.5 did not. and 3.45 bar. The results show that the effect of tail correc-
Figure 9 shows a comparison between experimental iso- tions is in general more significant for mixtures with C2. It is
therms, previously reported simulations using TraPPE-zeo,13 in agreement with the results for adsorption of pure compo-
and those using the June force field with LB and SW nents (Figures 3 and 4). At 300 K and 3.45 bar, C2 presents
J https://dx.doi.org/10.1021/acs.jced.9b00621
J. Chem. Eng. Data XXXX, XXX, XXX−XXX
Journal of Chemical & Engineering Data pubs.acs.org/jced Article

Figure 11. Adsorption of the binary mixtures in silicalite at 300 K and 3.45 bar: (a) C1(1) + C2(2), (b) C1(1) + C3(2), (c) C1(1) + C4(2), and
(d) C2(1) + C3(2). The colors indicate the components: (1) blue and (2) red. Closed circles are the experimental data from ref 32. The
filled crosses and open squares are GCMC simulations with combining rules SW and LB, respectively. The open circles and open crosses are
GCMC simulations using the force fields proposed by Du et al.1 and TraPPE-zeo.13 The continuous lines are polynomial fits of the experimental
data.

major sensitivity to the tail correction for both combining data reported by Abdul-Rehman et al.32 and Schuring et al.40
rules. and also with previous simulations of Dubbeldam et al.6 The
A comparison of predicted n-alkane adsorption of mixtures effect of application of the intramolecular scaling factor
was also made between simulations using the June force field (0 0 0.5 1) was very small on the adsorption of C1 + C4,
with the LB and SW combining rules, simulations using the while it was more significant on the adsorption of 2mC5 +
TraPPE-zeo12 force field with the LB combining rule, simula- nC6. This is consistent with what was observed for the adsorp-
tions using the “tuned” parameters proposed by Du et al.,1 and tion of pure components; the effect of scaling the intra-
experimental data reported by Abdul-Rehman et al.32 For this molecular factor is more significant at a low pressure and for
comparison, all simulations were carried out by applying tail longer alkane chains.
corrections. The results are presented in Figure 11. All of the Adsorption of Ternary and Quaternary Mixtures. GCMC
simulations show the same qualitative behavior, with the simulations using the June force field with combining rules LB
“tuned” parameters arguably performing the worst for methane and SW as well as the TraPPE-zeo force field13 and customized
(C1). The results expressed by the absolute mean deviation force field proposed by Du et al.1 were performed for adsorp-
(AMD) observed for each component in the mixtures studied tion of the following ternary gas mixtures: C1 + C2 + C3, C1 +
are presented in Table 6. C2 + C4, and C1 + C3 + C4, and the quaternary mixture:
The effect of using the intramolecular scaling factor was C1 + C2 + C3 + C4, in silicate at 300 K and 3.45 bar. All these
evaluated for the binary mixtures of C1 + C4 and 2mC5 + simulations were performed without intramolecular scaling
nC6. The results are shown in Figure 12, with experimental factors but with tail corrections. The results are presented in
K https://dx.doi.org/10.1021/acs.jced.9b00621
J. Chem. Eng. Data XXXX, XXX, XXX−XXX
Journal of Chemical & Engineering Data pubs.acs.org/jced Article

Table 6. Absolute Mean Deviation (AMD) of GCMC Simulations Based on Polynomial Fits from Experimental Data for
Adsorption of n-Alkanes Mixtures in Silicalite
absolute mean deviation (AMD) (molecules/uc)
force field and combining rule used in this work
June17 + LB June17 + SW TraPPE-zeo13 + LB Du et al.1
system C1 C2 C3 C4 C1 C2 C3 C4 C1 C2 C3 C4 C1 C2 C3 C4
C1C2 0.27 0.52 0.29 0.39 0.29 0.87 0.59 0.81
C1C3 0.33 0.87 0.19 0.65 0.34 0.87 0.29 1.18
C1C4 0.49 0.30 0.35 0.33 0.46 0.39 0.48 0.45
C2C3 0.28 0.53 0.09 0.43 0.19 0.33 0.32 0.38
C1C2C3 0.13 0.16 0.47 0.51 0.13 0.40 0.10 0.32 0.65 0.10 0.13 0.17
C1C2C4 0.22 0.06 0.73 0.10 0.18 0.63 0.28 0.21 0.72 0.23 0.12 0.51
C1C3C4 0.10 0.29 0.60 0.19 0.42 0.56 0.07 0.23 0.55 0.13 1.02 1.57
C1C2C3C4 0.16 0.30 0.14 1.26 0.07 0.18 0.30 1.31 0.20 0.34 0.08 1.30 0.17 0.23 0.42 1.80
AAMD 0.24 0.26 0.46 0.74 0.24 0.19 0.44 0.71 0.26 0.39 0.43 0.74 0.28 0.32 0.63 1.08
OAAMD 0.43 0.40 0.45 0.50

Figure 12. Adsorption of binary mixtures in silicalite: (a) C1 + C4 at 300 K and 3.45 bar and (b) 2mC5(1) + nC6(2) at 433 K and 0.066 bar.
The colors black and red denote the components 1 and 2, respectively. The closed circles are experimental data from refs 32 and 40. The
squares and crosses are GCMC of this work. The squares are results using the SW combining rule, and the crosses are for the LB combining rule.
The empty squares and thick crosses are results that applied the intramolecular scaling factor and tail correction, while the filled squares and
thin crosses are results did not apply these factors. The stars are previous GCMC simulations.6 The continuous lines are fits of the experimental
data.

Figure 13 and are compared with experimental data reported reported by Hampson and Rees,41 Savitz et al.,42 and Song
by Abdul-Rehman et al.32 While there are small variations et al.43 as well as previous simulations using the TraPPE-zeo
among the four different models tested, the June force field force field.13 Table 7 shows the average absolute mean deviation
with both combing rules (SW and LB) performed very well, (AAMD) and relative mean deviation (RMD). At 298 K, the
especially if you consider that no parametrization has been TraPPE-zeo force field matches perfectly the experimental data
made. The average of absolute mean deviation (AAMD) is pre- reported by Hampson and Rees41 but gives a slightly lower
sented in Table 6 adsorption at 373 K when compared to the data of Song
3.3. Adsorption in LTA-, CHA-, and TON-Type Zeolites. et al.43 The SW combining rule shows better agreement with
To test the accuracy of the SW combining rule for different the experiment than LB for all components at both tempera-
zeolites, GCMC simulations were performed to predict the tures tested. Additional experimental data for propane adsorp-
adsorption of light alkanes in zeolite types LTA, CHA, and tion in a TON-type zeolite would be desirable in order to
TON as well as CO2 in LTA and CHA. clarify the isotherm-type behavior, since the data reported by
TON. GCMC simulations were performed using the June Hampson and Rees41 and Song et al.43 present a considerable
force field with combining rules LB and SW to predict adsorp- level of disagreement.
tion isotherms of methane (C1), ethane (C2), and propane LTA. Vujić and Lyubartsev44 investigated the effect of blocking
(C3) in a TON-type zeolite at two temperatures. The results sodalite cages, reporting only a small difference between the
shown in Figure 14 are compared to the experimental data computed adsorption loading using free and blocked sodalite
L https://dx.doi.org/10.1021/acs.jced.9b00621
J. Chem. Eng. Data XXXX, XXX, XXX−XXX
Journal of Chemical & Engineering Data pubs.acs.org/jced Article

Figure 13. Adsorption of light n-alkane mixtures in silicalite at 300 K and 3.45 bar: (a) C1(1) + C2(2) + C3(3), (b) C1(1) + C2(2) + C4(3),
(c) C1(1) + C3(2) + C4(3), and (d) C1(1) + C2(2) + C3(3) + C4(4). The colors indicate the components: (1) black, (2) red, (3) blue, and
(4) magenta. Closed circles are the experimental data reported by Abdul-Rehman et al.32 The filled crosses and open squares are
GCMC simulations using the June force field with combining rules LB and SW, respectively. The open circles and open crosses are GCMC
simulations using the force fields proposed by Du et al.1 and TraPPE-zeo,13 respectively. The continuous lines are the polynomial fits of the
experimental data.

cages. Therefore, in this work, no blocking has been made. methods of sample preparation, and the measurement techniques
Figure 15 presents the results of our GCMC simulations per- employed. However, the values predicted for ethane adsorption
formed to obtain CO2 isotherm in a LTA-type zeolite along at 303 K using an all-silica zeolite structure are between the
with results reported Vujić and Lyubartsev44 and also experi- values reported by Nam et al.47 and Mofahari and Saleki.48 The
mental data reported by Palomino et al.45 use of combining rule SW led to a smaller average absolute mean
The results of GCMC simulations for adsorption of methane, deviation and average relative mean deviation for C1, C2, C3,
ethane, propane, and CO2 in a LTA-type zeolite are shown in and CO2 adsorption in LTA compared to those of the LB rule
Figure 16. The results are compared to the experimental data (see Table 7).
reported by Palomino et al.,45 Loughlin et al.,46 Nam et al.,47 CHA. The adsorption isotherms of methane, ethane, and
Mofarahi and Saleki,48 and Grande et al.49 in 5A zeolite. There CO2 in CHA were computed by GCMC simulations using the
are large differences in the adsorbed amounts of ethane and combining rules LB and SW with the June force field. Also, for
propane reported by different experimental groups. These dis- this zeolite, no blocking of pockets/inaccessible parts were
crepancies are likely due to differences in the Si/Al ratio, made. The results are presented in Figure 17, along with the
M https://dx.doi.org/10.1021/acs.jced.9b00621
J. Chem. Eng. Data XXXX, XXX, XXX−XXX
Journal of Chemical & Engineering Data pubs.acs.org/jced Article

Figure 14. Adsorption isotherms of methane (C1), ethane (C2), and propane (C3) in TON at: (a) 298 and (b) 309 K for C1 and C2 and 373 K
for C3. The colors red, black, and blue represent the components C1, C2, and C3, respectively. Experimental data from refs 41−43. The filled
crosses and open squares are GCMC simulations using the June force field with combining rules LB and SW, respectively. Open crosses are
previous Gibbs ensemble Monte Carlo simulations using the force field TraPPE-zeo.13 The lines are fits of the experimental data using the dual site
Langmuir (DSL) model.

Table 7. Absolute (AMD) and Relative Mean Deviations (RMD) of GCMC Simulations Based on Dual Site Langmuir Fits
from Experimental Data for Adsorption of Methane, Ethane, Propane, and CO2 in LTA-, CHA-, and TON-Type Zeolites
absolute mean deviation (AMD) relative mean deviation (RMD)
adsorbate (molecule/uc) (%)
zeolite T (K) (with ref) LB SW LB SW
CH4 C1 LTA 30046 0.276 0.297 16.87 13.79
CH4 C1 CHA 30350 0.177 0.027 36.94 6.67
CH4 C1 TON 29843 0.060 0.043 48.74 21.58
CH4 C1 TON 30942 0.195 0.050 64.53 17.36
C2H6 C2 LTA 30347 0.658 0.669 53.04 55.72
C2H6 C2 LTA 30348 0.739 0.691 23.79 21.56
C2H6 C2 CHA 30350 1.206 0.749 52.82 37.73
C2H6 C2 CHA 30352 0.873 0.540 61.00 42.98
C2H6 C2 TON 29841 0.212 0.144 65.12 28.96
C2H6 C2 TON 30942 0.216 0.137 50.08 36.94
C3H8 C3 LTA 30046 1.020 1.107 35.64 37.07
C3H8 C3 TON 29846 0.390 0.293 43.84 32.84
C3H8 C3 TON 37343 0.327 0.174 40.62 23.16
CO2 CO2 LTA 30345 0.483 0.329 24.80 15.58
CO2 CO2 CHA 30351 0.909 0.509 44.89 29.93
total average 0.516 0.384 44.18 28.12

experimental data reported by Pham et al.,50 Pham and against the experimental data as well as simulation results
Lobo,51 Olson et al.,52 and Pourmadhi and Maghsoudi.53 The obtained using a conventional Lorentz−Berthelot (LB) com-
calculated values of AMD and RMD, based on the fit of the bining rule and previously investigated force fields that seek to
experimental data by the DSL model, are shown in Table 7. improve upon the LB combining rule. A total of 45 different
A very good prediction of methane in CHA was achieved with systems were examined, including unary, binary, ternary, and
the use of SW. The results using SW are less accurate for CO2 quaternary mixtures of light hydrocarbons in silicalite, pure
and C2 isotherms but still have lower deviations from the methanol, and CO2 in silicalite and CO2 and light hydrocarbons
experiment than those observed with the use of the LB com- in zeolites having the TON, LTA, and CHA structures.
bining rule. Considering the totality of the investigated systems, the results
indicate that the use of the SW combining rule leads to a slight
4. CONCLUSIONS but significant improvement of the predicted adsorption
The accuracy of a new predictive “size-weighted” (SW) com- behavior when compared to the classical LB combining
bining rule was evaluated by conducting GCMC simulations in rule using the same base zeolite force field. Interestingly, the
different siliceous zeolites and comparing the simulation results SW combining rule gives similar and sometimes superior
N https://dx.doi.org/10.1021/acs.jced.9b00621
J. Chem. Eng. Data XXXX, XXX, XXX−XXX
Journal of Chemical & Engineering Data pubs.acs.org/jced Article

Figure 17. Adsorption isotherms of methane, ethane, and CO2 in


CHA. The colors red, black, and magenta indicate the components
methane, ethane, and CO2, respectively. Circles are the experimental
Figure 15. Adsorption isotherms of CO2 in LTA at 303 K. data from refs 50−53. The filled crosses and open squares are GCMC
Experimental data from ref 45. Previous GCMC simulations from simulations run at 303 K using June et al.17 force field with combining
ref 44. The red squares and crosses are GCMC simulations using the rules LB and SW, respectively. The lines are fits of the experimental
June force field with combining rules SW and LB, respectively. data using the dual site Langmuir (DSL) model.

tail correction should be applied, while for LB, both input


variables should be used.


*
ASSOCIATED CONTENT
sı Supporting Information
The Supporting Information is available free of charge at
https://pubs.acs.org/doi/10.1021/acs.jced.9b00621.
Tables of summaries of setup and results for GCMC
simulations in gas and adsorbed phases and DSL fitted
parameters, figures of GCMC simulation results, adsorp-
tion isotherms, and charts of total number of butane
molecules, and details of isotherm fitting procedure (PDF)

■ AUTHOR INFORMATION
Corresponding Author
E. J. Maginn − Department of Chemical and Biomolecular
Figure 16. Adsorption isotherms of methane, ethane, propane, and Engineering, University of Notre Dame, Notre Dame, Indiana
CO2 in LTA. The colors red, black, blue, and magenta indicate the 46556, United States; orcid.org/0000-0002-6309-1347;
components methane, ethane, propane, and CO2, respectively. Circles Email: ed@nd.edu
are the experimental data from refs 45−49. The filled crosses and
open squares are GCMC simulations using June et al.17 force field Authors
with combining rules LB and SW, respectively. The simulations were L. L. Romanielo − School of Chemical Engineering,
performed at 300 K for C1 and C3 and at 303 K for C2 and CO2. The Universidade Federal de Uberlândia, 38400-902 Uberlândia,
lines are fits of the experimental data using the dual site Langmuir MG, Brazil; orcid.org/0000-0003-2653-5145
(DSL) model. M. G. M. V. Pereira − School of Chemical Engineering,
Universidade Federal de Uberlândia, 38400-902 Uberlândia,
results when of to compared those other force fields that were MG, Brazil
tuned to experimental data. This suggests that the purely S. Arvelos − School of Chemical Engineering, Universidade
predictive SW combining rule is a reasonable choice to use Federal de Uberlândia, 38400-902 Uberlândia, MG, Brazil
for predicting adsorption in zeolites when experimental data Complete contact information is available at:
are not available. However, it is important to notice that the https://pubs.acs.org/10.1021/acs.jced.9b00621
input GCMC variables such as the intramolecular scaling
factor and tail corrections play a significant role on the Notes
The authors declare no competing financial interest.


prediction of the adsorption amount. Overall, the June force
field17 for zeolites with the TraPPE force field to sorbates
present a reasonable prediction of adsorption isotherms ACKNOWLEDGMENTS
using both combining rules SW and LB. However, on using Support for the development of Cassandra was provided by the
the SW rule, neither the intramolecular scaling factor nor the National Science Foundation under an SSE grant (ACI-1339785)
O https://dx.doi.org/10.1021/acs.jced.9b00621
J. Chem. Eng. Data XXXX, XXX, XXX−XXX
Journal of Chemical & Engineering Data pubs.acs.org/jced Article

and the Air Force Office of Scientific research under AFOSR (19) Potoff, J. J.; Siepmann, J. I. Vapor-Liquid Equilibria of Mixtures
Award No. FA9550-14-1-0306. M.G.M.V.P. and L.L.R. were Containing Alkanes, Carbon Dioxide, and Nitrogen. AIChE J. 2001,
supported by National Council for Scientific and Techno- 47, 1676−1682.
logical Development (CNPq) of Brazilian federal government. (20) Chen, B.; Potoff, J. J.; Siepmann, J. I. Monte Carlo Calculations
for Alcohols and Their Mixtures with Alkanes. J. Phys. Chem. B 2001,
S.A. was supported by the Minas Gerais Research Foundation
105, 3093−3104.
(FAPEMIG). (21) Shah, J. K.; Marin-Rimoldi, E.; Mullen, R. G.; Keene, B. P.;

■ REFERENCES
(1) Du, Z. Z.; Manos, G.; Vlugt, T.J.H.T; Smit, B. Molecular
Khan, S.; Paluch, A. S.; Rai, N.; Romanielo, L. L.; Rosch, T. W.; Yoo,
B.; Maginn, E. J. Cassandra: An Open Source Monte Carlo Package
for Molecular Simulation. J. Comput. Chem. 2017, 38, 1727−1739.
(22) Smit, B. Simulating the Adsorption Isotherms of Methane,
Simulation of Adsorption of Short Linear Alkanes and Their Mixtures
Ethane, and Propane in the Zeolite Silicalite. J. Phys. Chem. 1995, 99,
in Silicalite. AIChE J. 1998, 44, 1756−1764.
(2) Smit, B.; Siepmann, J. I. Computer Simulations of the Energetics 5597−5603.
and Siting of N-Alkanes in Zeolites. J. Phys. Chem. 1994, 98, 8442− (23) Krishna, R.; Paschek, D. Molecular Simulations of Adsorption
8452. and Siting of Light Alkanes in Silicalite-1. Phys. Chem. Chem. Phys.
(3) Tan, S.; Prasetyo, L.; Zeng, Y.; Do, D. D.; Nicholson, D. On the 2001, 3, 453−462.
Consistency of NVT, NPT, μVT and Gibbs Ensembles in the (24) Vuong, T.; Monson, P. A. Monte Carlo Simulation Studies of
Framework of Kinetic Monte Carlo - Fluid Phase Equilibria and Heats of Adsorption in Heterogeneous Solids. Langmuir 1996, 12,
Adsorption of Pure Component Systems. Chem. Eng. J. 2017, 316, 5425−5432.
243−254. (25) Krishna, R.; Smit, B.; Calero, S. Entropy Effects during Sorption
(4) Smit, B.; Maesen, T. L. M. Molecular Simulations of Zeolites: of Alkanes in Zeolites. Chem. Soc. Rev. 2002, 31, 185−194.
Adsorption, Diffusion, and Shape Selectivity. Chem. Rev. 2008, 108, (26) Aimoli, C. G.; Maginn, E. J.; Abreu, C. R. A. Force Field
4125−4184. Comparison and Thermodynamic Property Calculation of Super-
(5) Vlugt, T. J. H.; Krishna, R.; Smit, B. Molecular Simulations of critical CO2 and CH4 using Molecular Dynamics Simulation. Fluid
Adsorption Isotherms for Linear and Branched Alkanes and Their Phase Equilib. 2014, 368, 80−90.
Mixtures in Silicalite. J. Phys. Chem. B 1999, 103, 1102−1118. (27) Santos, M. S.; Franco, L.; Castier, M.; Economou, I. G.
(6) Dubbeldam, D.; Calero, S.; Vlugt, T. J. H.; Krishna, R.; Maesen, Molecular dynamics simulation of n-alkanes and CO2 confined by
T. L. M.; Smit, B. United Atom Force Field for Alkanes in calcite nanopores. Energy Fuels 2018, 32 (2), 1934−1941.
Nanoporous Materials. J. Phys. Chem. B 2004, 108, 12301−12313. (28) Martin, M. G.; Siepmann, J. I. Novel Configurational-bias
(7) Pascual, P.; Ungerer, P.; Tavitian, B.; Pernot, P.; Boutin, A. Monte Carlo Method for Branched molecules. Transferable Potentials
Development of a Transferable Guest-Host Force Field for for Phase Equilibria. 2. United-atom Description of -Branched-
Adsorption of Hydrocarbons in Zeolites. Phys. Chem. Chem. Phys. Alkanes. J. Phys. Chem. B 1999, 103, 4508−4517.
2003, 5, 3684. (29) Smit, B.; Maesen, T. L. M. Commensurate “freezing” of alkanes
(8) Bezus, A. G.; Kiselev, A. V.; Lopatkin, A. A.; Du, P. Q. Molecular in channels of a zeolite. Nature 1995, 374, 42−44.
Statistical Calculation of the Thermodynamic Adsorption Character- (30) Zhu, W.; Kapteijn, F.; van der Linden; Moulijn, J. A.
istics of Zeolites Using the Atom-Atom Approximation. Part 3. Equilibrium Adsorption of Linear and branched C6 Alkanes on
Adsorption of Hydrocarbons. J. Chem. Soc., Faraday Trans. 2 1978, Silicalite-1 Studied by the Tapered Element Oscillating Microbalance.
74, 367−379. Phys. Chem. Chem. Phys. 2001, 3, 1755−1761.
(9) Maitland, G. C.; Rigby, M.; Smith, E. B.; Wakeham, W. A. (31) Zhu, W.; Kapteijn, F.; Moulijn, J. a. Adsorption of Light
Intermolecular Forces -Their origin and determination; Clarendon Press: Alkanes on Silicalite-1: Reconciliation of Experimental Data and
Oxford, England, 1981. Molecular Simulations. Phys. Chem. Chem. Phys. 2000, 2, 1989−1995.
(10) Delhommelle, J.; Millié, P. Inadequacy of the Lorentz-Berthelot (32) Abdul-Rehman, H. B.; Hasanain, M. A.; Loughlin, K. F.
Combining Rules for Accurate Predictions of Equilibrium Properties Quaternary, Ternary, Binary, and Pure Component Sorption on
by Molecular Simulation. Mol. Phys. 2001, 99, 619−625. Zeolites. 1. Light Alkanes on Linde S-115 Silicalite at Moderate to
(11) Boda, D.; Henderson, D. The Effects of Deviations from High Pressure. Ind. Eng. Chem. Res. 1990, 29, 1525−1535.
Lorentz-Berthelot Rules on the Properties of a Simple Mixture. Mol. (33) Sun, M. S.; Shah, D. B.; Xu, H. H.; Talu, O. Adsorption
Phys. 2008, 106, 2367−2370. Equilibria of C1 to C4 alkanes, CO2, and SF6 on Silicalite. J. Phys.
(12) Waldman, M.; Hagler, A. T. New Combining Rules for Rare Chem. B 1998, 102, 1466−1473.
Gas van Der Waals Parameters. J. Comput. Chem. 1993, 14, 1077− (34) Sun, M. S.; Talu, O.; Shah, D. B. Adsorption Equilibria of C5-
1084. C10 Normal Alkanes in Silicalite Crystals. J. Phys. Chem. 1996, 100,
(13) Bai, P.; Tsapatsis, M.; Siepmann, J. I. TraPPE-Zeo: Trans- 17276−17280.
ferable Potentials for Phase Equilibria Force Field for All-Silica (35) Song, L.; Rees, L. V. C. Adsorption and Transport of n-Hexane
Zeolites. J. Phys. Chem. C 2013, 117, 24375−24387. in Silicalite-1 by the Frequency Response Technique. J. Chem. Soc.,
(14) Shah, M. S.; Tsapatsis, M.; Siepmann, J. I. Monte Carlo Faraday Trans. 1997, 93, 649−657.
Simulations Probing the Adsorptive Separation of Hydrogen SulFide/ (36) Lu, L.; Wang, Q.; Liu, Y. Adsorption and Separation of Ternary
Methane Mixtures Using All-Silica Zeolites. Langmuir 2015, 31, and Quaternary Mixtures of Short Linear Alkanes in Zeolites by
12268−12278. Molecular Simulation. Langmuir 2003, 19 (25), 10617−10623.
(15) Kong, C. L. Combining rules for intermolecular potential (37) Choudhary, V. R.; Mayadevi, S. Adsorption of Methane,
parameters. II. Rules for the Lennard-Jones (12−6) potential and the Ethane, Ethylene, and Carbon Dioxide on Silicalite-I. Zeolites 1996,
Morse potential. J. Chem. Phys. 1973, 59, 2464. 17, 501−507.
(16) Romanielo, L. L.; Arvelos, S.; Tavares, F. W.; Rajagopal, K. A (38) Cavalcante, C. L.; Ruthven, D. M. Adsorption of Branched and
Modified Multi-Site Occupancy Model: Evaluation of Azeotropelike Cyclic Paraffins in Silicalite. 1. Equilibrium. Ind. Eng. Chem. Res. 1995,
Behavior in Adsorption. Adsorption 2015, 21, 3−16. 34, 177−183.
(17) June, R. L.; Bell, A. T.; Theodorou, D. N. Prediction of Low (39) Dubinin, M. M.; Rakhmatkariev, G. U.; Isirikyan, A. A.
Occupancy Sorption of Alkanes in Silicalite. J. Phys. Chem. 1990, 94, Differential Heats of Adsorption and Adsorption-Isotherms of
1508−1516. Alcohools on Silicalite. Bull. Acad. Sci. USSR, Div. Chem. Sci. (Engl.
(18) Martin, M. G.; Siepmann, J. I. Transferable Potentials for Phase Transl.) 1989, 38, 1950−1953.
Equilibria. 1. United-Atom Description of n -Alkanes. J. Phys. Chem. B (40) Schuring, D.; Koriabkina, A. O.; De Jong, A. M.; Smit, B.; Van
1998, 102, 2569−2577. Santen, R. A. Adsorption and Diffusion of n-Hexane/2-Methyl-

P https://dx.doi.org/10.1021/acs.jced.9b00621
J. Chem. Eng. Data XXXX, XXX, XXX−XXX
Journal of Chemical & Engineering Data pubs.acs.org/jced Article

pentane Mixtures in Zeolite Silicalite: Experiments and Modeling. J.


Phys. Chem. B 2001, 105, 7690−7698.
(41) Hampson, J. A.; Ress, L. V. C. Adsorption of Lower
Hydrocarbons in Zeolite NaY and Theta-1. Comparison of Low
and High Pressure Isotherm Data. In Stud. Surf. Sci. Catal.; Hattori,
T., Yashima, T., Eds.; Elsevier: Amsterdam, 1994, 83, 197−208.
(42) Savitz, S. S.; Siperstein, F.; Gorte, R. J.; Myers, A. L.
Calorimetric Study of Adsorption of Alkanes in High-Silica Zeolites. J.
Phys. Chem. B 1998, 102, 6865−6872.
(43) Song, L.; Sun, Z.; Duan, L.; Gui, J.; MacDougall, G. S.
Adsorption and Diffusion Properties of Hydrocarbons in Zeolites.
Microporous Mesoporous Mater. 2007, 104, 115−128.
(44) Vujić, B.; Lyubartsev, A. P. Transferable force-field for
modeling CO2, N2, Ar and O2 in all silica and Na+ exchanged
zeolite. Modell. Simul. Mater. Sci. Eng. 2016, 24, 1−26.
(45) Palomino, M.; Corma, A.; Rey, F.; Valencia, S. New Insights on
CO2-Methane Separation Using LTA zeolites with Different Si/Al
Ratios and a First Comparison with MOFs. Langmuir 2010, 26 (3),
1910−1917.
(46) Loughlin, K. F.; Hasanain, M. A.; Abdul-Rehman, H. B.
Quaternary, Ternary, Binary, and Pure Component Sorption on
Zeolites. 2. Light Alkanes on Linde 5A and 13X Zeolites at Moderate
to High Pressures. Ind. Eng. Chem. Res. 1990, 29, 1535−1546.
(47) Nam, G. M.; Jeong, B.; Kang, S.; Lee, B.; Choi, D. Equilibrium
Isotherms of CH4, C2H6, C2H4, N2, and H2 on Zeolite 5A Using a
Static Volumetric Method. J. Chem. Eng. Data 2005, 50, 72−76.
(48) Mofarahi, M.; Saleki, S. M. Pure and Binary Adsorption
Isotherms of Ethylene and Ethane on Zeolite 5A. Adsorption 2013, 19,
101−110.
(49) Grande, C. A.; Gigola, C.; Rodrigues, A. E. Adsorption of
Propane and Propylene in Pellets and Crystals of 5A Zeolite. Ind. Eng.
Chem. Res. 2002, 41, 85−92.
(50) Pham, T. D.; Lobo, R. F. Adsorption Equilibria of CO2 and
Small Hydrocarbons in AEI-, CHA-, STT-, and RRO-Type Siliceous
Zeolites. Microporous Mesoporous Mater. 2016, 236, 100−108.
(51) Pham, T. D.; Xiong, R. X.; Sandler, S. I.; Lobo, R. F.
Experimental and Computational Studies on Adsorption of CO2 and
N2 on Pure Silica Zeolites. Microporous Mesoporous Mater. 2014, 185,
157−166.
(52) Olson, D. H.; Camblor, M. A.; Villaescusa, L. A.; Huehl, G. H.
Light Hydrocarbon Sorption Properties of Pure Silica Si-CHA and
ITQ-3 and High ZSM-58. Microporous Mesoporous Mater. 2004, 67,
27−33.
(53) Pourmahdi, Z.; Maghsoudi, H. Adsorption Isotherms of
Carbon Dioxide and Methane on CHA-Type Zeolite Synthesized in
Fluoride Medium. Adsorption 2017, 23, 799−807.
(54) Olson, D. H.; Kokotailo, G. T.; Lawton, G. T.; Meier, W. M.
Crystal Structure and Structure-Related Properties of ZSM-5. J. Phys.
Chem. 1981, 85 (15), 2238−2243.
(55) Database of Zeolites Structures, Framework Type TON.
http://america.iza-structure.org/IZA-SC/framework.php?STC=TON
(accessed on 2018-03-26).
(56) Database of Zeolites Structures, Framework Type CHA.
http://america.iza-structure.org/IZA-SC/framework.php?STC=CHA
(accessed on 2018-03-26).
(57) Database of Zeolites Structures, Framework Type LTA. http://
america.iza-structure.org/IZA-SC/framework.php?STC=LTA (ac-
cessed on 2018-03-26).
(58) Maginn, E. J.; Bell, A. T.; Theodorou, D. N. Sorption
Thermodynamics, Siting, and Conformation of Long n-Alkanes in
Silicalite As Predicted by Configurational-Bias Monte Carlo
Integration. J. Phys. Chem. 1995, 99, 2057−2079.
(59) Zhu, W.; Kapteijn, F.; Moulijn, J. A. Equilibrium Adsorption of
Light Alkanes in Silicalite-1 by the Inertial Microbalance Technique.
Adsorption 2000, 6, 159−167.

Q https://dx.doi.org/10.1021/acs.jced.9b00621
J. Chem. Eng. Data XXXX, XXX, XXX−XXX

You might also like