Felipe Jiménez-Ángeles1, Abbas Firoozabadi1,2. 1. Reservoir Engineering Research Institute, Palo Alto, California 94301, United States. 2. Department of Chemical and Environmental Engineering, Yale University, New Haven, Connecticut 06510, United States.
Abstract
Adsorption of functional molecules on the surface of hydrates is key in the understanding of hydrate inhibitors. We investigate the adsorption of a hydrocarbon chain, nonionic and ionic surfactants, and ions at the hydrate-aqueous interface. Our results suggest a strong connection between the water ordering around solutes in bulk and the affinity for the hydrates surface. We distinguish two types of water ordering around solutes: (i) hydrophobic hydration where water molecules form a hydrogen bond network similar to clathrate hydrates, and (ii) ionic hydration where water molecules align according to the polarity of an ionic group. The nonionic surfactant and the hydrocarbon chain induce hydrophobic hydration and are favorably adsorbed on the hydrate surface. Adsorption of ions and the ionic headgroups on the hydrate surface is not favorable because ionic hydration and the hydrogen bond structure of hydrates are incompatible. The nonionic surfactant is adsorbed by the headgroup and tail while adsorption of the ionic surfactants is not favorable through the head. Water ordering is analyzed using the hydrogen bond and tetrahedral density profiles as a function of the distance to the chemical groups. The adsorption of solutes is studied through the free energy profiles as a function of the distance to the hydrate surface. Salt lowers the melting temperature of hydrates, disrupts hydrophobic hydration, reduces the solubility of solutes in the aqueous solution, and increases the propensity of solutes to be adsorbed on hydrate surfaces. Our studies are performed by the unbiased and steered molecular dynamics simulations. The results are in line with experiments on the effect of salt and alkanes in hydrate antiagglomeration.
Adsorption of functional molecules on the surface of hydrates is key in the understanding of hydrate inhibitors. We investigate the adsorption of a hydrocarbon chain, nonionic and ionic surfactants, and ions at the hydrate-aqueous interface. Our results suggest a strong connection between the water ordering around solutes in bulk and the affinity for the hydrates surface. We distinguish two types of water ordering around solutes: (i) hydrophobic hydration where water molecules form a hydrogen bond network similar to clathrate hydrates, and (ii) ionic hydration where water molecules align according to the polarity of an ionic group. The nonionic surfactant and the hydrocarbon chain induce hydrophobic hydration and are favorably adsorbed on the hydrate surface. Adsorption of ions and the ionic headgroups on the hydrate surface is not favorable because ionic hydration and the hydrogen bond structure of hydrates are incompatible. The nonionic surfactant is adsorbed by the headgroup and tail while adsorption of the ionic surfactants is not favorable through the head. Water ordering is analyzed using the hydrogen bond and tetrahedral density profiles as a function of the distance to the chemical groups. The adsorption of solutes is studied through the free energy profiles as a function of the distance to the hydrate surface. Salt lowers the melting temperature of hydrates, disrupts hydrophobic hydration, reduces the solubility of solutes in the aqueous solution, and increases the propensity of solutes to be adsorbed on hydrate surfaces. Our studies are performed by the unbiased and steered molecular dynamics simulations. The results are in line with experiments on the effect of salt and alkanes in hydrate antiagglomeration.
Clathrate hydrates are crystalline structures
of hydrogen-bonded
water molecules where guest molecules, such as methane, propane, carbon
dioxide, and nitrogen, are encaged. Ions are not found as guests in
hydrates. The cages in hydrates are designated as 456, where l, m, and n represent the number of square, pentagonal, and hexagonal faces.
The lattice structure of clathrate hydrates is defined by the geometry
and the number of cages in the unit cell. The unit cell of structure
I (sI) is formed by two 612 and six 51262 cages; the unit cell of structure II (sII) is formed by 16
512 cages and eight 51264 cages,
and the unit cell of structure H (sH) is formed by two 512 cages, two 435663 cages, and one
51268 cage.[1,2]Natural
gas hydrates are important in energy production both as
a methane resource stored in hydrates at the ocean floor, and as a
safety and environmental concern during hydrocarbon production. Hydrates
may form swiftly during hydrocarbon transport and obstruct the flowlines.
Knowledge of nucleation and growth is essential for the exploitation
and utilization of hydrates. A number of molecular simulation studies
have been conducted to reveal the early stages of hydrates nucleation
of clathrate hydrates.[3−14] The use of surfactants in the hydrate research fields is of broad
interest in flow-assurance and gas storage.[15−23]In the energy industry hydrate mitigation is often by means
of
chemical additives. Hydrate inhibitors are classified as thermodynamic
inhibitors (TIs) and low dosage inhibitors (LDIs).[24,25] Thermodynamic inhibitors (such as methanol and glycol) and ions[26,27] shift the hydrate formation conditions to lower temperature and
higher pressure. Low dosage hydrate inhibitors are classified as kinetic
inhibitors and agglomeration inhibitors.[28] Hydrate kinetic inhibitors (HKIs) delay the onset of nucleation
or slow the hydrates growth rate. Common HKIs are soluble polymers
of amide groups such as N-vinylpirrolidone, N-vinylcaprolactam, and N-vinylacetamide.[29] Amphiphilic molecules[1,25,28,30,31] are used as hydrate agglomeration inhibitors (HAIs)
which prevent the coalescence of small hydrate crystallites to ensure
fluid flow as a slurry. The nucleation rate of clathrate hydrates,
however, may increase in the presence of some anionic surfactants
such as sodium dodecyl sulfate.[15−20] Antifreeze proteins are suggested as a green alternative of gas
hydrate inhibition.[32−35]Understanding the adsorption of surfactants on hydrate surfaces
is essential in flow-assurance and gas storage. To a great extent,
the efficiency of low dosage hydrate inhibitors (kinetic and antiagglomerants)
is determined by the adsorption on the hydrates surface.[25,28,29,36,37] Hydrogen bonding[38,39] and structural fit[36] are suggested as
the origin of binding between the hydrate surface and the inhibitors.
Molecular simulations are employed to shed light into the adsorption
mechanisms of hydrate inhibitors.[29,38−46] According to Anderson et al.[29] two molecular
characteristics lead to a strong binding of inhibitors on the hydrate
surface: a charge distribution on the edge of the inhibitor similar
to water molecules and the size congruence between the inhibitor group
and the open cavities at the hydrate surface. Yagasaki et al.[44] found no hydrogen bonding between the kinetic
inhibitor vinylpirrolidone and the hydrate surface in opposition to
the conventional picture. According to recent MD simulations, the
adsorption of a quaternary ammonium surfactant at the surface of sII
hydrates is stronger from a liquid hydrocarbon phase than from an
aqueous phase.[46.5]The water structure
around hydrophobic and amphiphilic molecules
(known as hydrophobic hydration) has some similarities with the structure
of hydrates and has been observed in different studies.[47−54] The idea of hydrophobic hydration can be traced back to a thermodynamic
model of dissolution where it is proposed that water molecules form
transient clathrate-like clusters around hydrophobic molecules.[55] Recent spectroscopic experiments confirm the
picture of hydrophobic hydration by measuring hydrogen bond enhancement
of water near hydrophobic solutes.[54] It
is suggested that hydrophobic hydration plays a fundamental role in
hydrophobic forces.[56−58] Here we will show that hydrophobic hydration around
solutes is related to the affinity of chemical groups for the surface
of hydrates.Recently we have performed an experimental investigation
of salt
and alkane hydrocarbon concentration on hydrate antiagglomeration.[59,60] The surfactant molecule M1 which will be described later was used
an antiagglomerant. Salt enhances hydrate antiagglomerations. Normal
alkanes (normal decane) also enhance hydrate antiagglomeration. In
the hydrate literature the effect of hydrocarbons is often expressed
in terms of the ratio of water to hydrocarbon liquids as water cut.
It is well-known that at high water cuts hydrate antiagglomeration
is facilitated. There has been no theoretical explanation for the
effect of hydrocarbons except some general bulk solubility effects.
The interaction of ionic and nonionic surfactants with hydrates is
not understood. The main objective of this work is to examine the
mechanism of adsorption of solute molecules at the hydrate–aqueous
interface and the effect of NaCl in the process. The solutes are n-decane, a nonionic surfactant (cocamidopropyl dimethylamine),
a cationic surfactant (didodecyl dimethylammonium chloride), and an
anionic surfactant (sodium dodecyl sulfate). First, we analyze the
structure of water molecules around solutes in an aqueous phase; the
structure of water around methane is also investigated. Next, we study
the adsorption of solute molecules at the hydrate–aqueous phase
interface. The effect of NaCl is investigated at the concentrations
of 0, 4, and 10 NaCl wt %. A key objective of this work is a molecular
explanation of the salt effect and normal alkane effect on hydrate
antiagglomeration. The remainder of the paper is organized as follows:
simulation models and methods; the results split in two subsections,
solutes in the bulk and adsorption of solutes on methane hydrate surface;
discussion and conclusions drawn at the end of the paper. The SI contains a discussion of the effect of salt
in hydrate three-phase equilibrium, additional results and details
of the calculation methods, verification of the steered molecular
dynamics technique, and tables.Figure shows n-decane and the
surfactants investigated in this work.
Molecule M1 has been used as an effective antiagglomerant.[31,61,62] Molecule M2 is an ionic quaternary
ammonium surfactant (QAS) which is effective in hydrate antiagglomeration
where there is a significant amount of liquid hydrocarbons.[62] Molecule M3 is sodium dodecyl sulfate (SDS)
which is known to increase the nucleation rate of clathrate hydrates.[15−18]
Figure 1
(a)
Molecular structure of the solute molecules: n-decane
(n-C10); molecule M1, cocamidopropyl
dimethylamine; molecule M2, didodecyl dimethylammonium chloride; molecule
M3, sodium dodecyl sulfate. Indexes a, b, and c designate atoms in the amine, amide, and
ammonium groups, respectively. The partial charges are provided in Tables S1–S3 in the SI. (b) Snapshot of
the simulation setup containing a hydrate slab next to an aqueous
solution containing one surfactant molecule M1. The crystal slab is
made of 2208 water molecules and 384 methane molecules; the aqueous
phase contains one solute molecule (n-C10, M1, M2, or M3), 5700 water molecules, and Ns NaCl ionic pairs; Ns = 0, 72,
and 180 at 0, 4, and 10 NaCl wt %, respectively. The hydrogen bonds
in the hydrate slab are represented as black sticks and as thin black
lines in the aqueous phase. The color key for various species is at
the bottom.
(a)
Molecular structure of the solute molecules: n-decane
(n-C10); molecule M1, cocamidopropyl
dimethylamine; molecule M2, didodecyl dimethylammonium chloride; molecule
M3, sodium dodecyl sulfate. Indexes a, b, and c designate atoms in the amine, amide, and
ammonium groups, respectively. The partial charges are provided in Tables S1–S3 in the SI. (b) Snapshot of
the simulation setup containing a hydrate slab next to an aqueous
solution containing one surfactant molecule M1. The crystal slab is
made of 2208 water molecules and 384 methane molecules; the aqueous
phase contains one solute molecule (n-C10, M1, M2, or M3), 5700 water molecules, and Ns NaCl ionic pairs; Ns = 0, 72,
and 180 at 0, 4, and 10 NaCl wt %, respectively. The hydrogen bonds
in the hydrate slab are represented as black sticks and as thin black
lines in the aqueous phase. The color key for various species is at
the bottom.
Results
Solutes in
the Bulk
We analyze the water structure
around hydrocarbon molecules, methane, and n-C10 and surfactant molecules at different salt concentrations.
The structural changes of water around methane are not captured in
the radial distribution function (see Figure S5 in the SI); only slight changes are observed by increasing the salt
concentration[63,64] from 0 to 10 NaCl wt %. The changes
occur in the hydrogen bond and tetrahedral structure of water molecules
surrounding the solute. Water molecules form instantaneous hydrate-like
structures around methane, namely, pentagonal and hexagonal rings
(Figure a). At 10
NaCl wt % the ions disrupt some of the hydrogen bonds around the methane
molecule (see Figure b).
Figure 2
Structure of water molecules around the solute molecules. Each
row contains two instantaneous snapshots showing the water molecules
at a distance of r < 0.52 nm from the solute molecules
at 0 and 10 NaCl wt %; the number of hydrogen bonds per water molecule NH(r) and the density of tetrahedral
water γ(r) at a distance r from a selected site. (a–d) Methane, (e–h) n-decane, and NH(r) and γ(r) are from the fifth carbon atom;
(i–l) molecule M1 and NH(r) and γ(r) are from the CH3 methyl head groups; (m–p) molecule M2 and NH(r) and γ(r)
are from the CH3 methyl head groups; (q–t) molecule
M3 and NH(r) and γ(r) are from the oxygen atoms of the SDS headgroup. The dashed
lines in the snapshots represent the hydrogen bonds; the solid blue
line in part a highlights a tetrahedral structure; T = 277 K, and P = 300 bar.
Structure of water molecules around the solute molecules. Each
row contains two instantaneous snapshots showing the water molecules
at a distance of r < 0.52 nm from the solute molecules
at 0 and 10 NaCl wt %; the number of hydrogen bonds per water molecule NH(r) and the density of tetrahedral
water γ(r) at a distance r from a selected site. (a–d) Methane, (e–h) n-decane, and NH(r) and γ(r) are from the fifth carbon atom;
(i–l) molecule M1 and NH(r) and γ(r) are from the CH3 methyl head groups; (m–p) molecule M2 and NH(r) and γ(r)
are from the CH3 methyl head groups; (q–t) molecule
M3 and NH(r) and γ(r) are from the oxygen atoms of the SDS headgroup. The dashed
lines in the snapshots represent the hydrogen bonds; the solid blue
line in part a highlights a tetrahedral structure; T = 277 K, and P = 300 bar.Hydrophobic solutes induce temporary hydrogen bond structures
resembling
those in hydrates known as hydrophobic hydration. Ionic groups tend
to align the dipole moment of water. We study the structural changes
of water by means of the hydrogen bond and tetrahedral structures.
A hydrogen bond forms when the center–center distance of two
wateroxygen atoms is less than 0.35 nm, and the two oxygen atoms
make an angle less than 25° with a hydrogen atom belonging to
one of the water molecules. In the particular case of molecule M3
(SDS), we consider the oxygen headgroup atoms as hydrogen bond acceptors.
We define NH(r) as the
average number of hydrogen bonds per water molecule at a distance r from a site i. A water molecule in a
perfect tetrahedron forms four hydrogen bonds; the angle formed between
two hydrogen bonds is about 109.5°. Water molecules in tetrahedrons
are found in ice and hydrates.[65] A measure
of the deviation of tetrahedrality is given by the order parameter q = ∑ ∑ (|cos θ| cos θ + 1/9)2, where n is the number
of nearest neighbors of water molecule i, and θ is the angle between the oxygen atoms
of water molecules j, i, and k. A water molecule is considered tetrahedral if it has q < 0.4 and is coordinated with at least three water
molecules. This classification was proposed by Baez and Clancy[66] and has been used in a number of studies.[12,13] We define γ(r) as the average number density
of tetrahedral water molecules at a distance r from
a site i of a molecule. The averages are taken over
5 × 103 independent configurations from over 20 ns
of MD simulation.Figure c portrays
the average number of hydrogen bonds of a water molecule at a distance r from the center of a methane molecule. At zero salt concentration
the maximum number of hydrogen bonds is around 3.85 at r ≈ 0.48 nm. Lower values of NH are observed as the salt concentration increases. Our prediction
of the number of hydrogen bonds as a function of the distance to the
methane molecule is in agreement with the calculations reported by
Grdadolnik et al.[54] in a recent study of
the hydrogen bond strength of water near small hydrophobic solutes.
The increase in both salt concentration and temperature decreases
the number of hydrogen bonds. The number density of tetrahedral water
γ reflects the icelike structure of water. Figure d shows γ as a function
of the distance to a methane molecule. The density of tetrahedral
water is significantly higher close to the methane molecule than in
the bulk; at zero salt concentration the density of tetrahedral water
is nearly 3 times higher than in the bulk. As the salt concentration
increases the number of tetrahedral water molecules close to methane
molecule decreases but always is at least 2 times higher than in the
bulk.Ions stay away from methane molecules[63,67] due to dielectric changes of the medium. Ions stay favorably hydrated
in the liquid phase. Bringing an ion close to a methane molecule implies
an energy penalty of dehydration and competition with hydrogen bond
structures. The free energy of transferring methane from the gas phase
into the aqueous phase ΔG is given by ΔG = −kBT ln P(0)
where P(0)
is the probability of finding a cavity of volume v0 free of solvent molecules. Our results confirm the view
of dissolution of methane in water by formation of transient hydrate-like
structures.[55] The water-free cavity is
stabilized by the clathrate-like structures. Ions destabilize the
hydrogen bond structures; hence, the probability to form a water-free
cavity in the aqueous phase decreases in the presence of NaCl. In
other words, the free energy of placing a methane molecule into the
aqueous phase increases by increasing the salt concentration.[64]Figure e,f shows
instantaneous structures of water molecules around n-decane at 0 and 10 NaCl wt %, respectively. Water molecules form
instantaneous pentagonal and hexagonal rings mainly at the end methyl
groups of the n-decane molecule while the hydrogen
bond network is disconnected in the methylene group at the middle
of the molecule. Non-hydrogen-bonded water molecules form dangling
OH groups which are seen as defects in water’s hydrogen-bonding
network.[50,51] In our MD simulations, dangling OH groups
around n-decane are mainly observed at the middle
of the molecule. At 10 NaCl wt %, ions disrupt the hydrate-like structure
around n-decane (see Figure f). Figure g,h shows, respectively, the average number of hydrogen
bonds per water molecule and the tetrahedral density from a CH2 methylene group at the middle of the n-decane
molecule. NH(r) and γ(r) functions calculated from the CH3 end methyl
group are provided in the SI (Figure S6). The number density of tetrahedral water around the CH3 groups is lower than around methane because the hydrocarbon chain
obstructs the connectivity among the water molecules. The obstruction
is pronounced at the middle of the n-decane molecule,
and the number density of tetrahedral water is significantly reduced;
at zero salt concentration the maximum tetrahedral density is 29,
22, and 17 molecules/nm3 in methane, CH3, and
the middle CH2 of n-decane, respectively.
The number of hydrogen bonds per water molecules is similar to methane
in both CH2 and CH3 groups of n-decane (see Figure S6 in the SI).Figure i,j is snapshots
of water molecules around molecule M1 at 0 and 10 NaCl wt %, respectively.
We see instantaneous hydrate-like rings around the surfactant head
and tail. The hydrate-like structures are disrupted by the ions at
10 NaCl wt % (see Figure j). The average number of hydrogen bonds of a water molecule
at a distance r from the CH3 groups in
the head is shown in Figure k, and the number density of tetrahedral water is presented
in Figure l. The number
of hydrogen bonds below r ≈ 0.38 nm is lower
than in the bulk because water molecules cannot access this region.
The maximum number of hydrogen bonds and the number of tetrahedral
water is located at r ≈ 0.4 nm. Our results
imply that a shell of hydrate-like water is formed around the head
of molecule M1. Similarly to methane and n-decane,
the number of hydrogen bonds per water molecule is reduced by NaClsalt. The hydrate-like structure of water is higher around the surfactant
head than in the bulk; the density of tetrahedral water is less affected
by NaCl ions close to the surfactant head than in the bulk. Around
the tail of molecule M1 the water molecules behave similarly as in
the n-decane molecule; that is, more tetrahedral
structures form around the tail end while in the middle of the tail
the tetrahedral structure is reduced. We observe dangling OH groups
around the amide group in the middle of molecule M1 due to the electrostatic
attraction between the amideoxygen atom and the waterhydrogen atoms.Hydration around ionic groups (ionic hydration) is different from
hydrophobic hydration. Usually the dipole moment of water tends to
be aligned by the electrostatic interaction with ionic groups. Instantaneous
arrangements of water molecules around molecule M2 are shown in Figure m,n at 0 and 10 NaCl
wt %, respectively. Typically, the water molecules are oriented with
the oxygen atom pointing toward the cationic headgroup. At zero salt
concentration the hydrogen bond network is open around the surfactant
head. At 10 NaCl wt % ions are close to the surfactant head, and no
closed hydrogen bond network is formed. At the two tail ends the hydrate-like
structure is similar to n-decane. The average number
of hydrogen bonds of a water molecule is below the bulk value for r ≲ 0.6 (Figure o). The number density of tetrahedral water is much
lower around the head of M2 than around the head of M1 (see Figure l,p). The non-hydrogen-bonded
water molecules form dangling OH groups pointing outward from the
ionic headgroup due to the electrostatic repulsion between the hydrogen
atoms and cationic ammonium group.A water molecule close to
the SDS headgroup is usually found with
a hydrogen atom pointing toward the oxygen atom of SDS (see Figure q,r). Hence, we consider
the oxygen atoms of SDS as hydrogen bond acceptors. From this consideration
we find significant differences between the water structures around
the headgroup of molecule M3 and the other molecules. In particular,
we see a maximum of NH at r ≈ 0.3 nm for SDS (M3, see Figure s) while the other molecules have a decreasing
behavior at similar distances (Figure o); for 0 NaCl wt % NH =
4 at r ≈ 0.3 nm which indicates perfect tetrahedral
coordination of a water molecule. For M3 γ has a small peak
at r ≈ 0.28 nm (Figure t) which is not observed for the other molecules
(Figure d,h,l,p).
Our results indicate that the anionic surfactant headgroup induces
the formation of hydrogen bonds and tetrahedral water which may be
related to the increase of hydrate nucleation rate.[15−18] The water structure around the
SDS headgroup is different from hydrophobic hydration. As usual, NaCl
decreases the formation of hydrogen bonds and tetrahedral structures.In Raman scattering measurements of the hydrophobic hydration shell
around hydrophobic solutes,[49] it is found
that chains longer than 1 nm have weaker hydrogen bonds than bulk
water, and the structure is less ordered. Our MD simulation results
for n-decane show that the hydration layer is broken,
and tetrahedrality is significantly reduced at the middle of the chain.
A defective hydrogen bond network with a variable number of dangling
OH groups (non-hydrogen-bonded water molecules) is observed in linear
alcohols and other amphiphilic solutes;[50,51] the number
of dangling OHs is significantly lower for ammonium ions than for
alcohols. We observe dangling OH groups in the hydrate-like network
around surfactants and hydrophobic solutes. In our MD simulation results,
a significant reduction of the hydrogen bond network and dangling
OH groups are observed around the ammonium group in agreement with
the experimental measurements.[50,51]
Adsorption of Solutes on
Methane Hydrate Surface
Now
we investigate the adsorption of n-decane, a nonionic
surfactant M1, a cationic surfactant M2, and an anionic surfactant
M3 on the hydrate surface. We observe the segments of the solute molecules
preferentially adsorbed induce hydrate-like structures around them
(hydrophobic hydration). The preferential adsorption of hydrophobic
molecules is related to the effect of an oil phase in hydrate antiagglomeration.
The addition of NaCl salt enhances the adsorption of the surfactants
on the hydrate surface. The enhancement of surfactant adsorption from
the salt effect is in line with experimental observations.[59,60] The molecular mechanisms will become clear in the following investigations.The free energy profiles of adsorption of n-decane
at 0, 4, and 10 NaCl wt % are shown in Figure a–c, respectively. The error bars
are computed from three independent simulations. Figure d shows the configuration of n-decane when the steered methyl group is at z ≈ 2.5 nm, and the free methyl group is adsorbed on the hydrate
surface; Figure e
shows the steered methyl group at z ≈ 1.7
nm adsorbed on the hydrate surface. There is a decrease of the free
energy of adsorption of n-decane as the salt concentration
increases. At 0 NaCl wt % (Figure a) there is a free energy drop of about −1 kJ/mol
when the steered methyl group is at z ≈ 2.5
nm, and the free methyl group is adsorbed on the hydrate surface.
A minimum of about −4.5 kJ/mol depth at z ≈
1.7 nm is observed when the pulled methyl group is adsorbed on the
hydrate surface. A steep increase of the free energy profile is observed
for z ≲ 1.7 nm implying that further penetration
of n-decane into the hydrate solid structure is obstructed
by the strongly hydrogen-bonded water molecules that require a high
energy cost to be removed. At 4 NaCl wt %, the adsorption of the free
methyl group gives a free energy drop of about −3 kJ/mol at z ≈ 2.2 nm; the depth of the minimum at z ≈ 1.7 is about −4.5 kJ/mol nm. At 10 NaCl wt % there
is a free energy drop ∼−4 kJ/mol when the free methyl
group is adsorbed into a hydrate cavity, and the steered methyl group
is at z ≈ 2.5 nm (see Figure d). The free energy minimum is about −7
kJ/mol when the pulled CH3 group is adsorbed on the hydrate
surface (see Figure d). Interestingly an instantaneous hydrate-like structure is formed
around the n-decane molecule.
Figure 3
Adsorption of n-decane molecule on the surface
of methane hydrate (sI). Potential of mean force W(z) as a function of the distance between the steered
methyl group and the hydrate slab; the origin is set inside the hydrate
slab (see Figure b).
The NaCl salt concentration is (a) 0, (b) 4, and (c) 10 NaCl wt %.
Snapshots of n-decane molecule at 10 NaCl wt % when
the pulled methyl group (ice-blue particle) is at (d) z ≈ 2.5 nm and (e) z ≈ 1.7 nm. The
red thin lines show hydrogen bonding around n-decane
molecule. The spring constant is k = 2000 kJ/(mol
nm2); the pulling velocity is 5 × 10–3 nm/ns, P = 300 bar, and T = 277
K.
Adsorption of n-decane molecule on the surface
of methane hydrate (sI). Potential of mean force W(z) as a function of the distance between the steered
methyl group and the hydrate slab; the origin is set inside the hydrate
slab (see Figure b).
The NaCl salt concentration is (a) 0, (b) 4, and (c) 10 NaCl wt %.
Snapshots of n-decane molecule at 10 NaCl wt % when
the pulled methyl group (ice-blue particle) is at (d) z ≈ 2.5 nm and (e) z ≈ 1.7 nm. The
red thin lines show hydrogen bonding around n-decane
molecule. The spring constant is k = 2000 kJ/(mol
nm2); the pulling velocity is 5 × 10–3 nm/ns, P = 300 bar, and T = 277
K.The adsorption of molecule M1
on the hydrate surface is analyzed
in Figure . The free
energy profiles at 0, 4, and 10 NaCl wt % are shown in Figure a–c, respectively. Figure d is a snapshot when
the headgroup atom is at z ≈ 2.6 nm, and the
surfactant tail is adsorbed on the hydrate surface; Figure e shows the surfactant headgroup
adsorbed on the hydrate surface at z ≈ 1.9
nm. A decrease of the free energy of adsorption of molecule M1 on
the hydrate surface is observed as the salt concentration increases.
At zero salt concentration, there is a free energy drop of about −2
kJ/mol when the tail is adsorbed on the hydrate surface, and the headgroup
is away from the hydrate surface at z ≈ 2.6
nm. From about z ≈ 2 nm the free energy profile
increases. At 4 NaCl wt %, the free energy profile decreases from
about z ≈ 2.8 nm when the surfactant tail
is adsorbed on the hydrate surface and reaches a minimum of −4
kJ/mol at z ≈ 1.9 nm, representing the adsorption
of the surfactant head. There is no significant effect of NaCl from
0 to 4 NaCl wt %. Similarly, the free energy profile at 10 NaCl wt
% shows a decrease from z ≈ 2.8 nm (see Figure d) reaching a minimum
of about −11 kJ/mol at z ≈ 1.9 nm when
the surfactant head is adsorbed on the hydrate surface (see Figure e). The surfactant
is found in three main favorable positions: In one configuration the
tail end is adsorbed on the hydrate surface, and the head is in the
aqueous phase (see Figure d); in another configuration the head is adsorbed on the hydrate
surface, and the tail is in the aqueous phase (not shown). In the
third configuration both the tail and the head are adsorbed on the
hydrate surface (see Figure e). The first configuration is preferred at zero salt concentration
while the third configuration is the most favorable at salt concentrations
of 4 and 10 NaCl wt %. We highlight the instantaneous hydrate-like
structure around the head when it is adsorbed (see Figure e). The effect of salt in improving
the efficiency of molecule M1 as hydrate antiagglomerant, in our simulations,
is in line with our recent measurements.[59,60] Note that there is no alkane in the system. Molecule M1 becomes
effective at higher salt concentration in full agreement with experimental
observation. Surfactant M1 is the only molecule reported in the literature
that has this feature.
Figure 4
Adsorption of molecule M1 on the surface of methane hydrate
(sI).
Potential of mean force W(z) as
a function of the distance between the pulled headgroup atom N and the hydrate slab; the origin is set
inside the hydrate slab. The NaCl salt concentration is (a) 0, (b)
4, and (c) 10 NaCl wt %. Snapshots showing configurations of molecule
M1 at 10 NaCl wt % when the N headgroup
atom is at (d) z ≈ 2.6 nm and at (e) at z ≈ 1.9 nm. The red thin lines show hydrogen bonding
around molecule M1. The spring constant is k = 2000
kJ/(mol nm2); the pulling velocity is 5 × 10–3 nm/ns, P = 300 bar, and T = 277
K.
Adsorption of molecule M1 on the surface of methane hydrate
(sI).
Potential of mean force W(z) as
a function of the distance between the pulled headgroup atom N and the hydrate slab; the origin is set
inside the hydrate slab. The NaCl salt concentration is (a) 0, (b)
4, and (c) 10 NaCl wt %. Snapshots showing configurations of molecule
M1 at 10 NaCl wt % when the N headgroup
atom is at (d) z ≈ 2.6 nm and at (e) at z ≈ 1.9 nm. The red thin lines show hydrogen bonding
around molecule M1. The spring constant is k = 2000
kJ/(mol nm2); the pulling velocity is 5 × 10–3 nm/ns, P = 300 bar, and T = 277
K.The adsorption of molecule M2
at 0, 4, and 10 NaCl wt % is examined
in Figure . At the
three salt concentrations the adsorption of the cationic surfactant
head is unfavorable. At 0 and 4 NaCl wt %, the free energy profile
increases from about z ≈ 3 nm as the surfactant
headgroup approaches to the hydrate surface. At 10 NaCl wt %, there
is a free energy minimum of −2.5 kJ/mol related to the adsorption
of the tail on the hydrate surface when the steered headgroup atom
is at z ≈ 2.7 nm (see Figure d); from this point the free energy increases
when the ionic head moves toward the hydrate surface. The free energy
increase is related to the unfavorable formation of hydrate-like structures
around the ionic headgroup, implying that the hydration layer around
the ionic headgroup is incompatible with the hydrate structure. This
is related to the difference of dielectric constant between the aqueous
phase and the clathrate hydrate solid; ionic groups stay more favorably
in the liquid phase due to the higher dielectric constant. In Figure e we see that the
hydrogen bond network closes around the methane molecules of the hydrate
crystal but is open around the surfactant head. The closed cavities
prevent a further approach of the surfactant molecule.
Figure 5
Adsorption of molecule
M2 on the surface of methane hydrate (sI).
Potential of mean force W(z) as
a function of the distance between the pulled headgroup atom N and the hydrate slab; the origin is set
inside the hydrate slab. The NaCl salt concentration is (a) 0, (b)
4, and (c) 10 NaCl wt %. Snapshots showing configurations of molecule
M2 at 10 NaCl wt % when the N headgroup
atom is at (d) z = 2.5 nm, and (e) z = 2.1 nm. The red thin lines show hydrogen bonding around molecule
M2. The spring constant is k = 2000 kJ/(mol nm2); the pulling velocity is 5 × 10–3 nm/ns, P = 300 bar, and T = 277
K.
Adsorption of molecule
M2 on the surface of methane hydrate (sI).
Potential of mean force W(z) as
a function of the distance between the pulled headgroup atom N and the hydrate slab; the origin is set
inside the hydrate slab. The NaCl salt concentration is (a) 0, (b)
4, and (c) 10 NaCl wt %. Snapshots showing configurations of molecule
M2 at 10 NaCl wt % when the N headgroup
atom is at (d) z = 2.5 nm, and (e) z = 2.1 nm. The red thin lines show hydrogen bonding around molecule
M2. The spring constant is k = 2000 kJ/(mol nm2); the pulling velocity is 5 × 10–3 nm/ns, P = 300 bar, and T = 277
K.The adsorption of molecule M3
on the hydrate surface is unfavorable
even though tetrahedral structures may form around the SDS headgroup.
The same as in surfactant M2, the free energy profile increases as
the surfactant headgroup approaches to the hydrate surface at 0 and
10 NaCl wt % (see Figure a,b). The small free energy minimum corresponds to the adsorption
of the tail on the hydrate surface when the steered headgroup atom
is between 2.5 nm ≲ z ≲ 3 nm (see Figure c). The minimum at z ≈ 2 nm corresponds to a configuration like that
in Figure d and is
related to a partial coupling between the surfactant headgroup and
the hydrate structure from the tetrahedral structures (see the discussion
of Figure q–t).
The NaCl in the aqueous phase does not appreciably affect the adsorption
of SDS.
Figure 6
Adsorption of molecule M3 on the surface of methane hydrate (sI).
Potential of mean force W(z) as
a function of the distance between the pulled headgroup atom S and
the hydrate slab; the origin is set inside the hydrate slab. The NaCl
salt concentration is (a) 0 and (b) 10 NaCl wt %. Snapshots showing
configurations of molecule M3 at 10 NaCl wt % when the S headgroup
atom is at (c) z = 2.5 nm, and (d) z = 2 nm. The spring constant is k = 2000 kJ/(mol
nm2); the pulling velocity is 5 × 10–3 nm/ns, P = 300 bar, and T = 277
K.
Adsorption of molecule M3 on the surface of methane hydrate (sI).
Potential of mean force W(z) as
a function of the distance between the pulled headgroup atom S and
the hydrate slab; the origin is set inside the hydrate slab. The NaClsalt concentration is (a) 0 and (b) 10 NaCl wt %. Snapshots showing
configurations of molecule M3 at 10 NaCl wt % when the S headgroup
atom is at (c) z = 2.5 nm, and (d) z = 2 nm. The spring constant is k = 2000 kJ/(mol
nm2); the pulling velocity is 5 × 10–3 nm/ns, P = 300 bar, and T = 277
K.The hydrate structure may significantly
affect the adsorption of
surfactants due to interactions of the surfactant with the guest molecules
and the geometry of the cages. We conducted simulations of the adsorption
of molecules M1 and M2 in the presence of structure II of propane
hydrates, and we found that the adsorption of the nonionic surfactant
molecule M1 on structure II is more favorable than on structure I
at similar conditions; the free energy minimum by adsorption of M1
headgroup on structure sII is about −7 kJ/mol (see Figure S7a in the SI) whereas it is about −2
kJ/mol on structure sI (see Figure a). The adsorption of ionic headgroups on clathrate
hydrates structure II is unfavorable similarly to methane hydrates
(see Figure S7b in the SI).To confirm
our prediction from the potential of mean force calculations,
we conducted MD simulations leaving the molecules free to allow them
to look for their favorable location. We see that n-decane is adsorbed on the hydrate surface with one methyl group
adsorbed into a cavity from the hydrate surface; the residence time
is more than 40 ns in a 90 ns MD simulation (see Figure S8 in the SI). Molecule M1 is adsorbed on the hydrate
surface through both the tail end and the headgroup (see Figure S9 in the SI); the residence time is more
than 50 ns. In a 130 ns MD simulation molecule M2 is not adsorbed
on the hydrate surface; most of the time it is excluded from the liquid
layer next to the hydrate surface when it comes close to the hydrate
surface (see Figure S10 in the SI). The
molecular simulation results are in line with experimental data that
ionic antiagglomerants are not effective where water cut is high.[31]
Discussion
The preference of ionic
species to stay in the aqueous phase rather
than in the hydrate surface can be explained in terms of the dielectric
constant difference between the two media; the dielectric constant
is about 80 for the aqueous phase and about 50 for clathrate hydrates
sI.[1,68] In the aqueous phase, the dipole moment
of water molecules is favorably aligned by the interaction with ions.
In the solid clathrate, the hydrogen bonding tends to maintain a water
molecule at fixed position and orientation. Ions stay in the liquid
phase because they are more favorably hydrated than in the solid clathrate
where there is a strong resistance to change the water orientation
due to hydrogen bonding. The ions and clathrate hydrates are incompatible.
The rejection of ions from the hydrate surface and the disliking between
ions and hydrocarbons are understood in similar terms. In the continuous
theory of dielectrics,[69−71] a charge q near the interface between
two media experiences a force from an image charge q′ = q(ϵ1 – ϵ2)/(ϵ2 + ϵ1); ϵ1 and ϵ2 are the charges of the media 1 and
2, respectively. When ϵ1 > ϵ2 the
sign of the image charge q′ is the same as
the real charge q; hence, the force from the interface
on q is repulsive. Our MD simulation captures the
repulsive effect by the dielectric discontinuity as predicted in the
classical theory of dielectrics.The dissolution of a hydrocarbon
chain in water requires a free
energy cost to break hydrogen bonds and create a cavity free of water
molecules.[72] Hydrophobic hydration is an
indication of the disliking of liquid water molecules
and hydrocarbon chains; the hydrocarbon interaction with a hydrogen
bond surface is more favorable than with liquid water. The segregation
of hydrophobic molecules toward the hydrate surface minimizes the
volume where hydrogen bonds are disrupted. Hence, hydrophobicity results
in adsorption of a hydrocarbon chain at the hydrate surface.[73,74] The favorable adsorption of hydrophobic groups on the hydrates surface
has been indicated in molecular simulation studies.[40] Pure hydrocarbon chains may not be classified as hydrate
inhibitors because their solubility in aqueous phase is extremely
low. In hydrate antiagglomeration the oil phase serves as the medium
where antiagglomeration takes place. Our results for n-decane indicate that hydrocarbons may have an active role in hydrate
antiagglomeration. This suggestion is in line with recent experimental
results.[59,60]The primary role of the headgroup
is to aid dissolution of the
surfactant in aqueous solution. In solution the surfactants may be
adsorbed on the hydrates surface. Based on the potential of mean force
the nonionic surfactant M1 is favorably adsorbed on the hydrate surface
by the tail and the head. In rocking cell experiments molecule M1
has been demonstrated to be an effective antiagglomerant[31,61,62] in the absence of liquid hydrocarbons.[62] The surfactant M2 may be an effective hydrate
antiagglomerant when there is a liquid hydrocarbon phase;[62] it is not effective when there are no substantial
amounts of hydrocarbons in the system. Quaternary ammonium surfactants
(QASs) have been developed and patented as antiagglomerants and are
effective in the presence of liquid hydrocarbons.[30] The structure of QAS consists of a central ammonium group
bonded to four n-alkyl chains that may be of different
lengths. Our results suggest that the QAS can be attached to the hydrate
surface through the hydrocarbon chains but unlikely by direct adsorption
of the ionic headgroup. Effective antiagglomerant QASs consist of
two or more n-butyl chains[25,30] that may attach to the hydrate surface. The ions in the aqueous
phase create more hydrophobic repulsion of the hydrocarbon chains
in the solution toward the hydrate surface. In experiments, the performance
of hydrate antiagglomerants is improved with NaCl[30,59,60] as mentioned above.
Conclusions
The
main conclusions from this work are as follows:Hydrophobic molecules induce instantaneous structuring
of water molecules in the bulk aqueous phase similar to clathrate
hydrates which is referred to as hydrophobic hydration. We quantify
hydrophobic hydration of methane, n-decane, and a
nonionic and an ionic surfactant by the number of hydrogen bonds and
tetrahedrality of water molecules.Hydrophobic
hydration is enhanced at the end of hydrocarbon
chains and around the nonionic amide surfactant headgroup. Hydrophobic
hydration is reduced at the middle of hydrocarbon chains and by NaCl
ions.The structure of water around the
ionic headgroups is
different from hydrophobic hydration. A significant reduction of hydrogen
bonds and tetrahedral structures is observed around the cationic ammonium
headgroup. On the contrary, the anionic headgroup of the sodium dodecyl
surfactant promotes the formation of tetrahedral water structures
which may be related to the increase of the hydrate nucleation rate.Hydrophobic hydration is directly related
to the affinity
of the chemical groups for the surface of hydrates; the hydrocarbon
chain end and the nonionic amide group are preferentially adsorbed
on the hydrate surface while adsorption of ions and the ammonium group
is unfavorable. Hydrophobic groups are hydrophobically hydrated on
the hydrate surface.The underlying cause
for ionic groups to be repelled
from hydrophobic molecules and from the hydrate surface toward the
liquid phase is due to the difference of dielectric constants. Ionic
groups are more favorably hydrated in the aqueous phase, and they
are not part of the clathrate hydrate solid.The adsorption of a solute molecule on the hydrate surface
is enhanced by NaCl ions in the aqueous phase. The affinity of ions
for water molecules and the dislike for hydrophobic groups drive the
surfactant molecules to the hydrate–water interface.The adsorption of nonionic groups on clathrate
hydrates
is more favorable on structure II than on structure I. The adsorption
of ionic surfactants is unfavorable in both structures.
Models and Methods
We represent hydrocarbon chains
like n-decane
and the surfactants’ tail using the TraPPE united atom model
of liquid hydrocarbons.[75] Molecule M1 is
cocamidopropyl dimethylamine, a nonionic surfactant which consists
of a dimethyl amine in the head, an amide in the middle, and a hydrocarbon
tail. We take into account the full atomic structure of the amine
and amide groups using the OPLS-AA force field.[76,77] Molecule M2, didodecyl dimethylammonium chloride, is a quaternary
ammonium surfactant, containing an ammonium headgroup bonded to two
methyl and two dodecyl groups. The full atomic structure of the ammonium
headgroup is taken into account using the OPLS-AA force field.[78] The partial charges and Lennard-Jones parameters
of molecules M1, M2, and M3 are provided in Tables S1–S3 in the SI, respectively.Water molecules
are simulated using the TIP4P-ice model which describes
accurately the three-phase equilibrium of methane hydrates.[79,80] Methane molecules are described as spherical particles interacting
via a Lennard-Jones potential; the parameters are σ = 0.373
nm and ϵ = 1.23 kJ/mol.[81] The interaction
between dissimilar atoms is computed using the Lorentz–Berthelot
combining rules. We verified the validity of the force field parameters
and combination rules by computing the free energy of solution of
the trimethylamine molecule [N(CH3)3] and the
enthalpy of hydration of the tetramethylammonium ion [N(CH3)4+]. Both
values are in agreement with experimental measurements (see the SI).To simulate the adsorption of solutes
on the hydrate surface we
employ the setup shown in Figure b consisting of a hydrate slab next to an aqueous solution.
The hydrate slab is made of 4 × 4 × 3 unit cells of methane
hydrate crystal (2204 water molecules and 384 methane molecules).
The focus of our work is on structure I (sI) of clathrate hydrates,
and some additional studies are performed on structure II (sII). Unless
it is explicitly stated the hydrate structure is sI. The aqueous phase
is made of 5700 water molecules and contains 20 methane molecules;
72 and 180 ionic NaCl pairs are added in the aqueous phase at 4 and
10 NaCl wt %, respectively. The hydrate slab dimensions are 4.8 nm
× 4.8 nm × 3.6 nm along x, y, and z, respectively. The simulation box dimensions
are 4.8 nm × 4.8 nm along the x and y directions, respectively, and the box length along the z direction is around 12.2 nm. Adsorption on structure II
of clathrate hydrates is simulated for methane and molecules M1 and
M2. Adsorption of methane is simulated to validate the steered molecular
dynamics method using a setup similar to that employed by Yagasaki
et al.[44] of methane hydrates sII; the stable
phase of methane hydrates is sI. The hydrate slab for methane adsorption
consists of 2 × 2 × 2 sII unit cells and is made of 1088
water molecules and 192 methane molecules; the aqueous phase is made
of 2162 water molecules. The adsorption of molecules M1 and M2 on
clathrate hydrate sII is simulated using a hydrate slab consisting
of 3 × 3 × 2 unit cells and constituted by 2448 water molecules
and 144 propane molecules; sII is the stable phase of propane hydrates.
The aqueous phase is made of 6688 water molecules. The study of the
solute molecules in the bulk aqueous phase is performed without the
hydrate slab.
Steered Molecular Dynamics Simulations
We employ steered
molecular dynamics to investigate the adsorption of solute molecules
on the hydrate surface. The technique consists of connecting a dummy
atom to target atom of the solute molecule and then moving the dummy
atom at constant velocity v; the connection is through
a harmonic potential with a spring constant k. The
force exerted by the molecules in the medium is the negative of the
pulling force. The mean force F̅ is the average force over 20 ns of pulling. The
work to transfer the molecule between two points is given byW(z) is
the potential of mean force as a function of the distance to the hydrate
slab z; z0 is a reference
position of the atom in the solute molecule hooked to the dummy atom.
At constant pressure and temperature the work is the Gibbs free energy
change by moving the molecule, W = ΔG.In statistical mechanics of homogeneous fluids
the potential of mean force between two particles W(r) is related to the pair correlation function g(2)(r) by[82,83]where β = 1/(kBT); kB is the
Boltzmann constant, and T the absolute temperature.
In theory of statistical mechanics of inhomogeneous fluids W(z) is related to the density profile
ρ(z) by[84]where ρb is the reference
density (bulk) of an inhomogeneous fluid.Steered molecular
dynamics (SMD) simulations are performed in the
following way: (1) A 0.5 ns run is carried out to equilibrate the
system at the prescribed pressure and temperature; in this step we
use the Berendsen[85] thermostat (τ = 0.1 ps) and barostat (τ = 0.5 ps). The molecule is initially placed about
3 nm away from the hydrate surface. (2) Steered molecular dynamics
simulations are performed by pulling one atom of the solute molecule
along the z direction. For n-decane
the pulling is applied on one methyl group (end group); for molecules
M1, M2, and M3 the pulling is applied on the N, N, and S headgroup atoms, respectively.
Other atoms are not steered or constrained, and steering is not applied
in the x and y directions. We use
a spring constant of k = 2000 kJ/(mol nm2), and the pulling is performed at a velocity of v = 5 × 10–3 nm/ns
along the z direction. Steered molecular dynamics
gives similar results of the potential of mean force as the umbrella
sampling method (see Figures S11 and S12 in the SI). The simulation time is about 600 ns per run, and the
CPU time is about 160 ns/day using four P-100 GPU cards from Nvidia.
The force running average F̅(z) is calculated over intervals of 20 ns;
within this time frame the molecule explores a large number of orientations
and positions in the x and y directions
(see Figure S13 in the SI). Each plot of
the potential of mean force and the error bars are calculated from
at least three independent simulation runs. The temperature and pressure
are controlled using the Nosé–Hoover[86,87] thermostat (τ = 2 ps) and the
Parrinello–Rahman barostat (τ = 4 ps), respectively. Simulations are performed at T = 277 K and P = 300 bar. A time-step of 2 fs is
used to integrate Newton’s equation of motion. Short-range
interactions are truncated at 1.2 nm, and long-range electrostatic
interactions are computed using the smooth particle mesh Ewald summation.
Three-dimensional-periodic boundary conditions are applied. The simulations
are performed using the open source code Gromacs.[88−90]
Authors: P N Perera; K R Fega; C Lawrence; E J Sundstrom; J Tomlinson-Phillips; Dor Ben-Amotz Journal: Proc Natl Acad Sci U S A Date: 2009-07-20 Impact factor: 11.205
Authors: Konrad Meister; Simona Strazdaite; Arthur L DeVries; Stephan Lotze; Luuk L C Olijve; Ilja K Voets; Huib J Bakker Journal: Proc Natl Acad Sci U S A Date: 2014-12-02 Impact factor: 11.205
Authors: Baofu Qiao; Felipe Jiménez-Ángeles; Trung Dac Nguyen; Monica Olvera de la Cruz Journal: Proc Natl Acad Sci U S A Date: 2019-09-09 Impact factor: 11.205