7114 J. Phys. Chem.

1995, 99, 7114-7119

Thermodynamic Stability of Hydrates for Ethane, Ethylene, and Carbon Dioxide

Bjarn Kvamme”
Telemark Institute of Technology, Kjllnes, N-3914 Porsgrunn, Norway

Hideki Tanaka
Department of Polymer Chemistry, Graduate School of Engineering, Kyoto University,
Sakyo-ku, Kyoto 606-01, Japan
Received: November 1, 1994; In Final Form: January 27, 1995@

An extension of the theory of van der Waals and Platteeuw was recently proposed by Tanaka and co-workers.
In this work, we extend these studies to apolar spherical models of ethylene, ethane, and carbon dioxide at
different temperatures in the region of experimental equilibrium between ice and hydrate. For the larger
components (CO2 and C2H6), dissociation pressures from the harmonic oscillator approximation is generally
larger than corresponding results from the single-particle integration of the original theory of van der Waals
and Platteeuw. Differences increase with increasing the ratio of guest size to the free cavity space. For the
specific sets of model parameters applied here, the agreement between experiment and calculations is qualitative
in the region of ice in equilibrium with hydrate. In the liquid water region estimated dissociation pressures
for ethylene and carbon dioxide are in better agreement with experiments for the harmonic oscillator approach
and underpredicted from the single-particle integration. Extension to mixtures are discussed, and some examples
are presented. The results for mixtures generally reflect the same type of behavior as that observed for pure
components.

1. Introduction oscillator approach will account for guest-water interactions

Recently Tanaka and c o - ~ o r k e r s l -examined
~ the thermo-
dynamic stability of several hydrates of structures I and 11. They
observed that for spherical guest models the potential energy
had only one minimum. The free energy of inclusion of a guest
in a particular cavity could then be modeled as the sum of this
energy minimum and the free energy of a sum of harmonic
oscillators representing the displacements from this minimum.
This approach makes it thus possible to divide the total change
in free energy of cage occupancy into the effect of guest-water
interactions, the entropy due to combinations of occupation
pattem, and the effect of guest movements in the cavity on the
water molecules. This coupling of the guest was not included
in the original theory of van der Waals and Platteuw? which
contains only the effect of guest-water interactions. In
summary it was found in these studies that when the ratio of
the guest size to the size of the free space in the cavity is small
to moderate the effects of the guest on the lattice vibrations is
small and the assumption of free guest-movement in the theory
of van der Waals and Platteeuw4 is reasonable. On the other
hand, when the ratio gets closer to unity, there is a significant
effect of the guest on the vibrational frequencies of the lattice
and thus the chemical potential of the water molecules in the
lattice. Similar observations has been made by other author^.^^^
Recent experimental measurements’ also verify that the TIP4P
model is able to represent these vibrational shifts accurately.
A hybrid model was therefore suggested by Tanaka and
K i y ~ h a r ain
, ~which the ratio of guest-to-cavity size determines
which approach should be used for a specific guest in a specific
cavity. For small-to-moderate ratios the single-particle integra-
tion may be more accurate. And when the size of the guest
approches the size of the free space in the cavity the harmonic
@ Abstract published in Advance ACS Abstracts, April 1, 1995.
0022-365419512099-7114$09.0010
as well as effects of guest movements on the chemical potential
of water.
In this work we extend these studies to simplified models of
some larger molecules filling the large cavity of structure I. As
in previous studies guest molecules are still modeled as spherical
Lennard-Jones 12-6 particles. Chosen parameters correspond
to previously published parameters applied to represent proper-
ties of methane, ethane, ethylene, and carbon dioxide. Calcula-
tions have been performed for several temperatures in the region
of experimentally measured equilibrium between hydrate and
ice, Le., up to a temperature of 273.15 K.
A brief summary of the theory is given in section 2. The
results obtained from the harmonic oscillator approach and the
single-particle integrations are summarized in section 3. Cal-
culated dissociation pressures for hydrate in equilibrium with
ice are presented in section 4 and compared to experimental
data. In section 5 we extrapolate these data to the region of
hydrate in equilibrium with liquid water and compare derived
results with experimental data. In section 6 we present some
calculations for binary mixtures. Our conclusions are given in
section 7.
2. Theory
A brief summary of the theory is given below. More
extensive discussion of the theory is given A
straightforward extension of eqs 6 and 8 in Tanaka and
Kiyohara3 leads to the following grand canonical partition
function for a general multicomponent hydrate:
p=l r-1
where the sums runs over different guest-types p in small
cavities and different guest types r in large cavities. N is the
0 1995 American Chemical Society
Thermodynamic Stability of Hydrates J. Phys. Chem., Vol. 99, No. 18, 1995 7115

number of cavities of type small (s) or large (1) occupied by TABLE 1: Lennard-Jones Parameters for Different Guest
guest p and r, respectively. Equation 1 assumes that the Molecules
occupancy of each cavity is independent of the other cavities.
This is not exact, but a large portion of the cavities will normally a(& 3.405 4.047 3.758 4.232 4.418 4.486
be filled in a naturally occurring hydrate. The change in aPr dkg (K) 119.8 231.0 148.6 205.0 230.0 189.0
and up, with limited variations of filling fractions between 0.8
and 1.0 may also be small for apolar guest molecules.2 In the TABLE 2: Coefficients in Eq 8 for Ice and Empty Clathrate
results presented here ups and up, are evaluated for complete of Structure I and II‘
filling. For guest molecules with a small-to-moderate ratio of ice I I1
guest size to free cavity space, u is evaluated according to the ko -87.1349 -87.4146 -88.2306
equation of van der Waals and Platteeuw: kl.10-3 26.7350 27.3750 27.8398
kr 10-5 -74.1636 -75.6181 -76.8049
a = exp@(p + kT In b) (1) (2) k3*10-*
k440-9
10.5173
-76.185 1
10.7057
-77.4836
10.8682
-78.6373
k5.1O-I’ 22.3667 22.7365 23.0709
where b is given by:
Energy in units of kJ/mol and temperature in kelvin.
b = (mkT/2d2)3’2J Vcagcexp[-pw(v)] dv (3) TABLE 3: Coefficients in Eq 8 for the Free Energy of
Guest Inclusion in Structure Io
where m is the mass of the guest k is Boltzmann’s constant, h methane xenon
is Planck’s constant divided by 2n, w(v) is the energy of large small large small
interaction between the guest and the water molecules constitut- ko -114.2304 -91.5112 -84.4709 -72.7673
ing the lattice of the cavity as a function of the three-dimensional k,.10-3 43.4398 35.1972 34.1924 28.8028
volumetric coordinate v. Equation 3 is applicable to a spherical k2*lOw5 -108.3548 -89.4591 -85.7005 -73.2875
guest model. For larger ratios, Le., cavities where the size of k3*10-* 14.6687 12.2205 11.6288 10.0111
the guest approaches the size of the free volume, a is evaluated -103.5370 -86.7137 -82.1932 -71.0235
k5.10-“ 29.9008 25.1289 23.7582 20.5783
by the harmonic oscillator approximation and is expressed as
Values for methane are estimated from the single-particle integra-
a = e x p v ( p - Ag - u)] (4) tion (eqs 2 and 3) and the values for xenon are calculated according to
the harmonic oscillator approximation (eqs 4 and 5). Energy in units
where u is the minimum of the energy of guest-lattice of kUmol and temperature in kelvin.
interaction and Ag is the free energy of vibration of the guest
the hydrate is obtainable by direct diffentiation of the logarithm
in the cavity. The free energy of individual oscillators are thus
given by:
of eq 1 with respect to the number of water molecules:
i=2 -
, p=ii
g = kTJln(gho) h(w) dw (5) pw = p i - k T x v j ln(1 +cap) (9)
j= 1 p=l
where h(w) is the density of state normalized to the (mean)
number of degrees of freedom per molecule for each system.
where the outer summation is over the two types of cavities, vj
Ag in eq 4 is the sum of the individual oscillators representative
is the number of cavity of type j per water molecules and the
for movements of the specific guest in its host cavity. The
inner summation runs over all guest types p in each of the two
partial derivative of the logarithm of eq 1 with respect to the
cavities indicated by the subscript j .
chemical potential of the different guests give the average
occupation number of each particular guest p in a small cavity
3. Simulated Results and Data Correlation
as:
TIP4P8 has been used for water in all calculations presented
here. Guest-guest and guest-water interactions are calculated
from the Lennard-Jones 12-6 potential with parameters for the
different guest moleculesg listed in Table 1. Lorentz-Berthelot
(7) mixing rules are applied for cross interactions. Lennard-Jones
parameters for the short-range part of the TIP4P potential are 0
+ c
p=l
= 3.154 8, and clkg = 78.02 K. Chemical potentials of the
guests in the gas phase are calculated from the virial equation
cutoff after the second term. Fitted coefficients in eq 8 for
Calculations of kT In b (in eq 2) and Ag +
u (in eq 4) have chemical potential of ice and empty hydrates are given in Table
been performed at temperatures in the range from T = 123.15 2 for temperatures in the region from 123.15 to 273.15 K.
to 273.15 K at intervals of 5 K. Details on simulation Similar coefficients for the free energy of inclusion is given in
procedures and other numerical details concerning the evalua- Tables 3-5 for components studied in this work as well as some
tions of eqs 3 and 5 are given The results are components reported in a previous study.’O All calculated results
fitted to a polyonomial in inverse temperature: are reproduced with eq 8 within the fifth or sixth digit and are
i= 5 thus considered to be well within the expected numerical errors
involved in the normal-mode analysis or the single-particle
i=O
integration, respectively.
At 273.15 K the calculated dissociation-pressures for carbon
where superscript “fitted” indicates a fitted polynomial of free dioxide agrees to within the fourth digit when comparing the
energy or chemical potential. The form of eq 8 is also used originally calculated results with results using correlated data
for empty clathrates and ice. The free energy of the water in from eq 8.
7116 J. Phys. Chem., Vol. 99, No. 18, 1995
TABLE 4: Coefficients in Eq 8 for the Free Energy of
Guest Inclusion in Structure I1 for Argon" ~~~ ~
large small
ko -113.5524 -108.1709
k,.10-3 53.7144 54.6157
k2.10-j -139.7383 - 152.5330
k3*10-' 19.7217 22.6634
k4.10-9 -144.5817 -172.8917
kj.lO-" 43.1898 53.3152
a Values are estimated from the single particle integration (equations
(2) and (3)). Energy in units of kJ/mole and temperature in Kelvin.
TABLE 5: Coefficients in Eq 8 for the Free Energy of
Guest Inclusion in Structure I"
ko -94.3200 -90.7898 -89.5173 -90.8131 -90.6393 -92.7517
k ~ * l O - ~ 35.9880 33.5906 32.2242 32.2038 34.0915 34.4641
k2.10-j -91.3558 -84.7612 -82.3276 -81.6181 -86.8068 -87.0646
k3*10-8 12.4725 11.5352 11.2727 11.1298 11.8680 11.8553
-88.4739 -81.6643 -80.0956 -78.8840 -84.2523 -83.9571
ks*lO-" 25.6337 23.6295 23.2309 22.8417 24.4230 24.2979
a All guests are only filling the large cavity. Notation VW in
paretheses indicate values evalated from eqs 2 and 3 and the remaining
values are evaluated by eqs 4 and 5. Energy in units of kJ/mol and
temperature in kelvin.
4. Hydrates in Equilibrium with Ice
Calculated dissociation pressures for methane, ethylene,
ethane, and carbon dioxide are listed in Table 6. Calculated
dissosiation pressures for ethylene are also compared to
experimental valuesll in Table 7. The purity of the ethylene
gas is not given. However, just to indicate possible shifts in
dissociation pressures due to impurities, we also tabulate
calculated pressures for gas mixtures containing 1 and 5 mol
% methane, respectively. For the pure ethylene calculations
there are no significant differences in the estimated filling
fraction between results from the single-particle integration and
the harmonic oscillator approximations, respectively. The
estimated filling-fractions decreases linearly from 0.943 at
269.15 K to 0.939 at 273.15 K. At 1% methane there is a
similar decrease from 0.868 to 0.860 for the harmonic oscillator
approximation and 0.865 to 0.857 for the single-particle
integration. There are corresponding increases in the fraction
of methane in the large cavities from 0.069 to 0.073 and from
0.071 to 0.075, respectively. The changes in filling character-
istics from 1% methane to 5% methane is negligible within the
expected accuracy of the calculations. The filling fractions of
methane in the small cavities decrease from 0.269 to 0.265 for
the 5% methane mixture in going from 269.15 to 273.15 K for
the harmonic oscillator approximation. The corresponding range
of changes for the single-particle integration are 0.275 to 0.272.
Differences between 1% and 5% methane are small. Calculated
dissociation pressures for ethane is tabulated in Table 8 together
with experimental data.12 The indicated purity of the experi-
metal ethane was 98.3%. Calculated dissociation pressures for
1.7% methane is listed for comparison. The filling character-
istics are very similar to the case of ethylene discussed above.
In Table 9 we tabulate some comparisons for methane with
different sources of experimental data. The filling fractions vary
almost linearly with temperature between 0.794 and 0.761 for
the small cavity and between 0.918 and 0.904 for the large
cavity. We have also tabulated a set of calculations with a vinal
equation extended to the third coefficient.
The large cavity of structure I is the most asymmetric of the
cavities in clathrates of type I and 11. Nevertheless, the average
spherical diameter of this cavity is 8.66 A. With the TIPS
parameters above and the Lennard-Jones parameters from Table
Kvamme and Tanaka
1, this gives an average free path length of 1.274 A (before the
guest-host interactions turns to repulsive) for ethylene, 1.088
A for the model of ethane and 1.020 for carbon dioxide. In
view of this shorter "free" distance for ethane we would expect
the harmonic oscillator approximation to be more appropriate
for ethane than ethylene and also more important in terms of
possible interactions with lattice vibrations.
By comparing Tables 7 and 8 we see that the differences
between results from the two methods are larger for ethane than
for ethylene. The harmonic oscillator approximation is that
closest to the experimental pressures although the calculations
are systematically shifted when compared to the experimental
values. The differences between the two methods for ethylene
are probably within experimental accuracy. The model for
carbon dioxide is larger than the ethane model and the
differences in estimated dissociation pressures are more pro-
nounced (see Table 6). This increase in pressure for these large
components are expected and in accordance with the effect of
interference between guest movements and vibrational modes
of the lattice.
Thermal expansion of the hydrate structure was not included
in this study. It is known that the cell size for ethylene oxide
hydrate may increase from 11.87 at 80 K to 12.03 at
250 K, but the thermodynamic consequences are not yet verified
in detail. Tanaka and Kiyohara3 reports dissociation pressures
for xenon at 273.15 K to be 1.5 and 1.8 bar for the single-
cavity integration and harmonic oscillator approach respectively
when the cavity size is 12.03 A. The corresponding numbers
with a cell size of 11.84 A was found to 1.4 and 1.5 bar,
respectively. For guest molecules with a larger guest-to-cavity
size ratio we expect a reduced cavity size to result in increased
differences between dissociation pressures calculated from the
two methods, as also discussed above.
The use of nonspherical models in the harmonic oscillator
approach increases the numerical complexity significantly. And
realistic models for carbon dioxide should also contain effects
of a significant permanent quadupole moment.15 In addition it
is not clear that the polarizability in carbon dioxide and ethylene
can be neglected in realistic simulations of these systems. In
summary the calculations presented here represent more a
comparison of two procedures in a fashion similar to the way
the procedures would normally be applied in the industrial scene
as a io01 for estimating dissosiation pressures of natural gas
hydrates.
The Lennard-Jones parameters are not unique, and it would
perhaps be too optimistic to believe that a specific choice of
these parameters should give a perfect match of the experimental
dissociation pressures. There is also a systematic shift of
approximately 7 bar on the values for methane when compared
to experimental results12J6 (Table 9). Implementation of the
third virial coefficient shifts the calculated dissociation pressures
in the correct direction but the effect is, as expected, small.
Impurities of a component filling the large cavities could have
a significant impact on the dissociation pressures. The experi-
mental datal6 are based on high-purity methane (more than
99.9%12 methane or 99.7%16 methane). In the data of Frost
and Deaton,16 C02 is indicated to be the dominant component
in the remaining 0.3%. Even such small amounts of carbon
dioxide wilI result in a significant reduction of the theoretical
dissociation pressures according to the adsorbtion theory of van
der Waals and Platteeuw4 as given by eq 9. Equations 2 and 4
require only that there is enough COz to fill the large cavities
of the structure and as such the amount of C02 enters only into
the calculated dissociation pressures through the guest chemical
potential, as calculated via the virial equation. Without specific
Thermodynamic Stability of Hydrates J. Phys. Chem., Vol. 99, No. 18, 1995 7117

TABLE 6: Estimated Dissociation Pressures for Ethylene, Ethane, Carbon Dioxide, and Methane as a Function of
Temperature
T (K) czH4 CZH4(VW) czH6 CzHdvW) coz COZ(VW) CH4(VW)
123.15 0.000 17 0.000 16 0.OOO 02 0.OOO 01 0.OOO 34 O.Oo0 18 0.029 77
128.15 0.000 35 0.000 32 0.OOO 04 o.Oo0 02 0.OOO 70 0.OOO 38 0.048 68
133.15 0.000 69 0.OOO 62 0.OOO 08 0.OOO 05 0.001 38 0.000 76 0.077 07
138.15 0.001 28 0.001 16 0.OOO 17 o.Oo0 10 0.002 61 0.001 44 0.118 15
143.15 0.002 26 0.002 07 o.Oo0 34 0.OOO 20 0.004 70 0.002 60 0.175 76
148.15 0.003 86 0.003 55 o.Oo0 64 0.000 38 0.008 13 0.004 5 1 0.254 44
153.15 0.006 36 0.005 88 0.001 15 0.OOO 70 0.013 55 0.007 55 0.359 48
158.15 0.010 14 0.009 41 0.002 01 0.001 22 0.021 87 0.012 24 0.496 94
163.15 0.015 72 0.014 66 0.003 37 0.002 05 0.034 26 0.019 26 0.673 65
168.15 0.023 72 0.022 23 0.005 48 0.003 35 0.052 27 0.029 49 0.897 17
173.15 0.034 97 0.032 93 0.008 67 0.005 32 0.077 83 0.044 09 1.17575
178.15 0.050 44 0.047 72 0.013 37 0.008 24 0.113 33 0.064 44 1.51823
183.15 0.071 30 0.067 79 0.020 13 0.012 46 0.161 69 0.092 29 1.93394
188.15 0.098 95 0.094 54 0.029 66 0.018 43 0.226 38 0.129 69 2.432 64
193.15 0.135 00 0.129 61 0.042 82 0.026 71 0.311 47 0.179 08 3.024 32
198.15 0.181 28 0.174 89 0.060 67 0.037 99 0.421 70 0.243 29 3.719 14
203.15 0.239 84 0.232 53 0.084 47 0.053 10 0.562 44 0.325 57 4.527 31
208.15 0.312 99 0.304 94 0.115 72 0.073 03 0.739 79 0.429 59 5.458 98
213.15 0.403 25 0.394 82 0.156 16 0.098 93 0.960 55 0.559 48 6.524 14
218.15 0.513 37 0.505 12 0.207 78 0.132 12 1.232 30 0.719 83 7.732 59
223.15 0.646 31 0.639 09 0.272 86 0.174 14 1.563 33 0.915 68 9.093 81
228.15 0.805 28 0.800 24 0.353 96 0.226 72 1.96276 1.15257 10.61699
233.15 0.993 66 0.992 35 0.453 96 0.291 80 2.440 52 1.436 52 12.31096
238.15 1.21506 1.219 51 0.576 03 0.371 56 3.007 38 1.77403 14.184 16
243.15 1.4733 1 1.48607 0.723 69 0.468 40 3.675 01 2.172 16 16.24471
248.15 1.77241 1.796 66 0.900 80 0.584 98 4.456 03 2.638 48 18.50034
253.15 2.116 61 2.156 23 1.11157 0.724 20 5.364 09 3.181 10 20.958 45
258.15 2.510 33 2.570 01 1.360 58 0.889 24 6.413 93 3.808 76 23.626 11
263.15 2.958 23 3.043 55 1.65281 1.08356 7.621 49 4.530 79 26.510 12
268.15 3.465 21 3.582 74 1.99364 1.310 90 9.004 05 5.357 18 29.617 01
273.15 4.036 36 4.193 78 2.388 90 1.575 32 10.580 36 6.298 63 32.953 06
(I Notations as in Table 5. Temperatures in kelvin and pressure in bar.
TABLE 7: Estimated Dissociation Pressures for Ethylene TABLE 9: Estimated Dissociation Pressures for Methane
According to Either Harmonic Oscillator Approximation or According to the Single-Particle Inteeratiod
Single-Particle Integration (vWyl
pure ethylene 1% methane 1 5% methane
259.04 16.48 24.12 23.82
T ( K ) Pcnp P P(vW) P P(vW) P P(VW) 262.37 17.93 26.05 25.72
269.15 4.64 3.57 3.70 3.07 3.17 3.06 3.16 264.21 19.03 27.15 26.81
270.15 4.80 3.69 3.82 3.17 3.27 3.16 3.26 266.48 20.82 28.55 28.19
271.15 5.00 3.80 3.94 3.27 3.37 3.26 3.36 268.59 22.00 29.90 29.52
272.15 5.18 3.92 4.07 3.37 3.47 3.36 3.47 270.93 23.92 31.44 31.05
273.15 5.47 4.04 4.19 3.47 3.58 3.46 3.57 270.04 23.44 30.85 30.46
271.71 24.82 31.97 31.57
Experimental data from Diepen and Scheffer." Temperatures in
(I The fiist experimental point is collected from Roberts, Brown-
Kelvin and pressure in bar.
scombe and HoweI6 and the remaining points are from Frost and
TABLE 8: Estimated Dissociation Pressures for Ethane Deaton.I2 Temperatures in kelvin and pressure in bar. The last column
According to Either Harmonic Oscillator Approximation or is pressures estimated with the virial coefficient added to the equation
Single-Particle Integration (vw)" of state for calculation of guest chemical potential.
pure ethane 1.7% methane
provided that the melting point of TIP4P is known rigorously.
T(K) Pem P P(VW) P P(VW) This requires the calculation of enthalpy and specific heat
263.54 3.13 1.68 1.10 1.54 1.04 capacity for liquid water at the model melting point and also
266.48 3.57 1.87 1.23 1.72 1.16 the heat capacity for at least one temperature above this
269.26 4.05 2.08 1.37 1.91 1.29 temperature to account for the temperature dependency. From
272.04 4.57 2.30 1.51 2.11 1.43
a theoretical point of view this is desirable to do since it would
Experimental data from Frost and Deaton.lz Temperaturesin kelvin give a complete and consistent model system comparison
and pressure in bar. The amount of impurity in the applied ethane is between the two procedures. The neccesary work in order to
represented as methane. do this correctly, including an accurate calculation of the melting
point of TIP4P is outside the frames of the project reported in
details on the experiment there is no point in presenting any this paper but it is something that we will consider for future
numerical results. In an earlier work of Deaton and Frost17 no calculations.
significant changes were observed experimentally by the
In this work we extend the model results to the liquid region
removal of the carbon dioxide from the methane.
using experimental data of real liquid water. The enthalpies of
ice, empty clathrates, and guest inclusion are obtained by
5. Hydrates in Equilibrium with Liquid Water
numerical differentiation of the corresponding chemical poten-
The extension to the region of equilibrium between hydrates tials. Experimental enthalpies of melting and residual heat
and liquid water could be made within the model system capacity of liquid water is applied in the extension to liquid
7118 J. Phys. Chem., Vol. 99, No. 18, 1995 Kvamme and Tanaka

TABLE 10: Coefficients in Eq 8 for Liquid Water and TABLE 13: Estimated Dissociation Pressures for Ethane
Empty Clathrates of Structure I and I1 for the Range of According to Either Harmonic Oscillator Approximation or
Temperatures from 273.15 to 323.15 Ka Single-Particle Integration (vWY
liquid I n P P(VW
ko - 109.0343 -90.4838 -92.7208 T (K) Pexp B B+C B BSC
k1-10-3 37.7476 27.4822 29.1830 273.71 5.10
kz*10-6 -8.1591 -6.1197 -6.5852 2.54 2.54 1.67 1.67
~ 0 - 7 63.0339 47.8027 52.1977 274.82 5.79 2.87 2.87 1.89 1.90
275.37 6.62 3.06 3.06 2.01 2.02
a Energy in units of kJ/mol and temperature in kelvin. 277.59 8.14 3.90 3.90 2.57 2.57
278.70 9.31 4.41 4.40 2.90 2.90
TABLE 11: Coefficients in Eq 8 for Free Energy of Guest 279.26 10.07 4.68 4.67 3.08 3.08
Inclusion in Structure I for the Range of Temperatures from 279.82 10.82 4.98 4.97 3.27 3.27
273.15 to 323.15 Ka 280.37 11.65 5.28 5.27 3.47 3.47
280.93 12.55 5.61 5.59 3.69 3.69
cz& cz&(Vw) czH6 CzHdVW) coz COdVw) 281.48 13.44 5.95 5.93 3.91 3.91
ko -96.0017 -92.7571 -92.1975 -93.1063 -91.4060 -95.1423 282.04 14.48 6.32 6.29 4.15 4.15
k ~ . l O - ~34.5132 32.6162 31.9286 31.6200 31.9655 33.7798 282.59 15.58 6.71 6.67 4.40 4.40
k ~ * l O --7.0194
~ -6.6424 -6.6280 -6.4987 -6.4582 -6.9125 283.15 16.89 7.12 7.08 4.67 4.66
k3*10-' 52.3969 49.8144 50.3513 49.0811 47.8260 52.0478 284.26 19.86 8.02 7.96 5.25 5.24
Energy in units of kJ/mol and temperature in kelvin. 285.37 23.06 9.03 8.94 5.90 5.88
286.48 27.30 10.17 10.03 6.62 6.60
TABLE 12: Estimated Dissociation Pressures for Ethylene Experimentaldata from Frost and Deaton.I2 Temperaturesin kelvin
According to Either Harmonic Oscillator Approximation or and pressure in bar. Notations as in Table 12.
Single-Particle Integration (vWY
TABLE 14: Estimated Dissociation Pressures for Methane
According to the Single-Particle Integrationa

273.15 5.47 4.04 4.04 4.19 4.19

273.35 5.54 4.11 4.11 4.28 4.27 273.71 27.65 34.53 34.06
273.55 5.64 4.20 4.20 4.37 4.37 274.26 29.03 36.24 35.69
273.75 5.83 4.29 4.29 4.46 4.46 275.37 32.41 39.92 39.20
273.95 5.88 4.39 4.39 4.56 4.56 275.93 34.23 41.91 41.09
274.15 6.14 4.48 4.48 4.66 4.66 273.59 28.41 34.17 33.71
274.35 6.25 4.58 4.58 4.76 4.76 275.82 35.03 41.51 40.71
274.55 6.42 4.68 4.67 4.87 4.86 274.26 29.04 36.24 35.69
274.75 6.67 4.78 4.78 4.97 4.97 a Temperatures in Kelvin and pressure in bar. The last column is
275.15 6.94 4.99 4.98 5.19 5.19 pressures estimated with the virial-coeffeisient added to the equation
275.45 7.13 5.15 5.14 5.36 5.36 of state for calculation of guest chemical potential. The first four points
275.75 7.63 5.31 5.31 5.54 5.53 in this region was collected from Frost & Deaton.lz The next two points
276.32 7.59 5.64 5.64 5.88 5.87 are from Snell et.al.'* and the last point is from Jhaveri and Robinson.19
277.55 8.82 6.42 6.41 6.70 6.69
278.55 10.07 7.13 7.11 7.45 7.43
279.85 11.67 8.16 8.13 8.54 8.50
281.85 14.66 10.03 9.96 10.52 10.44 Calculations for methane are compared to experimental
284.15 19.34 12.70 12.56 13.37 13.20 data12J6 in Table 14. The observed shifts in dissociation
285.85 24.33 15.11 14.87 15.95 15.66 pressures is similar to that observed for the lower temperatures
287.35 29.21 17.63 17.24 18.66 18.19 (Table 9).
288.45 33.51 19.75 19.21 20.94 20.28
Dissociation pressures for carbon dioxide are compared in
Experimental data from Diepen and Scheffer." Temperatures in Table 15. The differences between the two approaches are
kelvin and pressure in bar. Calculated results for temperatures above significant at the harmonic oscillator approach, giving results
273.15 K are exptrapolated as indicated in text. Columns below symbol
B are calculated using only second virial-coefficient while B C + that are closer to the experimental
indicate that both second and third vinal coefficients are included. Considering the numerical uncertainties involved in the
double derivative used to obtain the heat capacities, it is not
water, from the enthalpy of ice at 273.15 K. For empty clear that the differences in the ethylene example above is very
clathrates and guest inclusion the chemical potentials are significant.
extended using a constant residual heat capacity estimated at The most important result of this section and the previous
273.15 K. Factors in eq 8 for the range of temperature 273.15 section is the effect of guest size on the dissociation pressures.
to 323 K are listed in Table 10 for the estimated residual For guest-to-cavity size ratios that are close to unity the effect
chemical potentials of liquid water and empty clathrates. of guest vibrations on the lattice vibrations reduces the stability
Corresponding factors for ethylene, ethane, carbon dioxide, and of the hydrate and increases the corresponding dissociation
methane are listed in Table 11. pressure. The differences between the two procedures for the
A comparison of estimated dissociation pressures for ethylene models of ethane and carbon dioxide are significant and indicate
with experimental data" is given in Table 12. Results are that the hybrid approach might be more appropriate for systems
generally in good agreement with experimental data and the containing components with a guestkavity size ratio close to
harmonic oscillator results are in better agreement with experi- unity.
ment than corresponding results from the single particle
integration.
6. Hydrates for Mixtures
For ethane the differences between the two procedures are
larger than in the ethylene example when compared to experi- Of the four components methane, ethylene, ethane, and carbon
mental data1* (Table 13), as also seen for the results at lower dioxide, most experimental mixtures are mixtures of the other
temperature (Table 8). components with methane. And according to the results from
Thermodynamic Stability of Hydrates J. Phys. Chem., Vol. 99,No. 18, 1995 7119
TABLE 15: Estimated Dissociation Pressures for Carbon size and energy parameters involved. Work is yet in progress
Dioxide According to Either the Harmonic Oscillator in this direction.21
Approach or the Single-Particle Integration (Denoted by
vW) As Estimated Using Either Only the Second Virial 7. Conclusions
Coeffisient (B) or the Second and Third (B 0" + We have calculated dissociation pressures for spherical
models of ethylene, ethane, and carbon dioxide at temperatures
from 123.15 to 273.15 K according to the single-particle
integration and the harmonic oscillator approximation suggested
273.71 13.24 11.24 11.09 6.68 6.66
274.26 14.11 11.97 11.78 7.10 7.07 by Tanaka and co-w~rkers.l-~Comparisons with the limited
275.37 16.13 13.57 13.29 8.02 7.98 amount of experimental data in this region of temperatures
276.48 18.48 15.39 14.99 9.06 8.99 cannot provide solid evidence of which method is the best. In
277.59 20.87 17.46 16.89 10.22 10.13 addition it may not be correct to interpret deviations from
278.70 24.20 19.83 19.02 11.53 11.40 experimentally measured data as a measure for the quality of
279.26 27.85 21.15 20.19 12.26 12.09
279.82 32.13 22.57 21.42 13.03 12.82 simulated results for model systems. In the models for carbon
280.93 32.13 25.72 24.07 14.70 14.41 dioxide, ethane, and ethylene carbon dioxide is the largest and
ethylene is the smallest. This is also reflected in calculated
Experimental data from Frost and Deaton.I2 Temperaturesin kelvin
and pressure in bar. dissociation pressures from the two procedures. Pressures
calculated from the harmonic oscillator approximation is gener-
TABLE 16: Estimated Dissociation Pressures for Mixtures ally higher than corresponding results calculated from the single-
of Ethane and Methane" particle integration for the largest components. For ethylene
calculated dissociation pressures are almost indistinguishable
274.80 0.564 9.45 6.23 4.17 within the expected experimental accuracies. For the applied
277.60 0.564 12.89 8.62 5.72 sets of Lennard-Jones parameters the calculated pressures are
280.40 0.564 17.58 12.03 7.85 in only qualitative agreement for both methods. We have also
283.20 0.564 24.34 17.18 10.84 proposed an extension of the theory to conditions of hydrate in
274.80 0.904 15.24 19.37 14.22 equilibrium with liquid water. At present this extension is made
277.60 0.904 20.96 27.63 19.82 using experimental data for real liquid water. Calculated
280.40 0.904 28.89 42.72 28.45
dissociation pressures from the harmonic oscillator approach
XCQ denotes mole fraction methane in the gas mixture. Experi- in this extended region are generally in better agreement with
mental data of Frost and Deaton.I2 Virial equation with only second experimental data for the systems investigated in this study.
virial coefficient. Methane calculated by single-particle integration and
ethane calculated by both procedures with VWin parenthesis indicating However, results are systematically shifted similarly to the
the latter procedure. Temperatures in kelvin and pressure in bar. results for the ice region. We have also examined a few results
for mixtures. For the methane/ethane mixtures the shifts in
the two preceding sections we might expect a shift in estimated estimated pressures are of the same order of magnitude as for
pressures for the higher temperatures where the effect of the corresponding pure components.
methane on the stability increases (see also Tables 9 and 14). References and Notes
This can also be seen from Table 16 where we have listed some (1) Tanaka, H . Chem. Phys. Lett. 1993, 202, 345.
results for methane and ethane together with experimental data (2) Tanaka, H.; Kiyohara, K. J. Chem. Phys. 1993a, 98, 4098.
(3) Tanaka, H.; Kiyohara, K. J. Chem. Phys. 1993b, 98, 8110.
of Deaton and Frost.'* At the lowest temperatures pressures (4) van der Waals, J. H.; Platteeuw, J. C. Adv. Chem. Phys. 1959, 2,
are underestimated, similar to the behavior of pure ethane 1.
(Tables 8 and 13). At the higher temperatures estimated ( 5 ) Hwang, M. Ph.D. Thesis, University of Pittsburgh, 1989.
(6) Hwang, M. J.; Holder, G. D.; Zele, S. R. Fluid Phase Equilib. 1993,
pressures are overestimated with respect to the experimental 83.
values. For a methane and carbon dioxide mixture containing (7) Tse, J. S.; Powell, B. M.; Sears, V. F.; Handa, Y. P. Chem. Phys.
66 mol % methane at 277.0 K the experimental pressure*O is Lett. 1993, 215, 383.
(8) Jorgensen, W. L.; Chandrasekhar, J.; Madura, J. D.; Impey, R. W.;
28.4 bar while the estimated result for the single-particle Klein, M. L. J. Chem. Phys. 1983, 79, 926.
integration is 22.66 bar at this point. The corresponding result (9) Hirschfelder, J. 0.;Curtiss, C. F.; Bird, R. B. Molecular Theory of
from the harmonic oscillator approach is 37.92 bar. The free Gases and Liquids; Wiley: New York, 1954.
(10) Tanaka, H.; Nakanishi, K. Mol. Simul. 1994, 12, 317.
energy of methane is calculated by single-particle integration (11) Diepen, G. A. M.; Scheffer, F. E. C. Reel. Trav. Chim. 1950, 69,
in both cases. 593.
There are some other sources of experimental data for (12) Frost, E. M.; Deaton, W. M. Oil Gas J. 1946, 45, 170.
(13) Hollander, F.; Jeffrey, G. A. J. Chem. Phys. 1977, 66, 4699.
mixtures of these components but unfortunately mainly with (14) McMullan, R. K.; Jeffrey, G. A. J. Chem. Phys. 1965, 42, 2725.
results presented graphically. (15) Kvamme, B.; F@rrisdahl,0. K. Fluid Phase Equilib. 1993,83,427.
(16) Roberts, 0.L.; Brownscombe, E. R.; Howe, L. S. Oil Gas J. 1940,
It follows from this section as well as the two previous 39, 37.
sections that for ethane and methane the parameters may not (17) Deaton, W. M.; Frost, E. M. Proc. Natural Gas Dept., Am. Gas
be appropriate,and it would thus be interesting to examine other Assoc. 1940, 122.
(18) Jhaveri, J.; Robinson, D. B. Can. J. Chem. Eng. 1965, 43, 75.
established parameters for these components. (19) Snell, E. L.; Otto, F. D.; Robinson, D. B. AlChE J. 1961, 7,482-
For industrial purposes it would be inconvenient to perform 485.
(20) Unruh, C. H.; Katz, D. L. Trans. AlME 1949, 186, 83.
a normal-mode analysis for every component and every tem- (21) Kvamme, B. An extended absorbtion theory for calculation of
perature that might be of interest. In this respect it is interesting hydrate equilibrium. For presentation at ISOPE-95, den Haag, June 11-
to see if we could collect the information contained in the 16, 1995.
simulated data into a convenient interpolation correlation of the JF942946Z