Ammonia clathrate hydrates as new solid phases forTitan, Enceladus, and other planetary systemsKyuchul Shina, Rajnish Kumarb, Konstantin A. Udachina, Saman Alavia, and John A. Ripmeestera,1

aSteacie Institute for Molecular Sciences, National Research Council of Canada, Ottawa, ON, Canada K1A 0R6; and bNational Chemical Laboratory,Council of Scientific and Industrial Research, Pune 411008, India

Edited by Mark H. Thiemens, University of California San Diego, La Jolla, CA, and approved July 26, 2012 (received for review April 12, 2012)

There is interest in the role of ammonia on Saturn’s moons Titanand Enceladus as the presence of water, methane, and ammoniaunder temperature and pressure conditions of the surface and in-terior make these moons rich environments for the study of phasesformed by these materials. Ammonia is known to form solid hemi-,mono-, and dihydrate crystal phases under conditions consistentwith the surface of Titan and Enceladus, but has also been assigneda role as water-ice antifreeze and methane hydrate inhibitor whichis thought to contribute to the outgassing of methane clathratehydrates into these moons’ atmospheres. Here we show, throughdirect synthesis from solution and vapor deposition experimentsunder conditions consistent with extraterrestrial planetary atmo-spheres, that ammonia forms clathrate hydrates and participatessynergistically in clathrate hydrate formation in the presence ofmethane gas at low temperatures. The binary structure II tetra-hydrofuran + ammonia, structure I ammonia, and binary structureI ammonia + methane clathrate hydrate phases synthesized havebeen characterized by X-ray diffraction, molecular dynamics simu-lation, and Raman spectroscopy methods.

ice ∣ single crystal X-ray diffraction ∣ hydrogen bonding ∣hydrate inhibitors ∣ ethane

Ammonia has long been seen as a key species in extraterres-trial space, both interstellar and on outer planets, moons,

and comets and the interplay of ammonia, methane, and waterhas been the subject of a considerable number of studies andspeculation (1–8). The main role assigned to ammonia has beenthat of an antifreeze for ice and clathrate hydrate formation,modifying the stability region of the solid ice and methane clath-rate hydrate phases as a thermodynamic inhibitor (2, 5, 9). How-ever, ammonia is a methane-sized molecule, thus based on sizealone it has the potential for being a suitable guest for clathratehydrate cages. Issues that may have prevented ammonia frombeing considered as a suitable clathrate guest molecule includethe notion that guest species need to be hydrophobic in order tobe incorporated into clathrates, and the observation of a numberof stoichiometric nonclathrate phases of ammonia and waterobtained upon cooling aqueous ammonia solutions (2). Previousexperimental work on the water-ammonia and water-methane-ammonia systems had not shown evidence for the enclathrationof ammonia (5, 10–19). Close inspection, however, shows that thelow pressure ammonia dihydrate (18) and the high pressure phaseII of ammonia monohydrate (19) have structural features in com-mon with canonical clathrate and semiclathrate structures.

Recent structural analysis and molecular simulations haveshown that some guest molecules which form strong hydrogenbonds with the water framework of the clathrate hydrate latticemay nonetheless produce stable phases (20–23). It is thereforereasonable to consider that ammonia has potential as a clathrateguest molecule. Furthermore, previous experimental and com-putational studies have shown that in the gas phase, water formsdodecahedral cages around ammonia and ammonium ions.These structures may be indicative of a geometric tendency to-wards clathrate hydrate structures for the ammonia and ammo-nium molecules (24–27).

ResultsAs a first step in this study we show that in the presence of otherclathrate hydrate forming substances, ammonia indeed can beincorporated in cages within the clathrate lattice. Incorporationof ammonia in a clathrate hydrate phase was accomplished byfreezing a solution of the well known clathrate hydrate formertetrahydrofuran (THF) and approximately 5% aqueous ammoniawith an approximate THF:water mole ratio of 1∶17 at −10 °C.Crystals formed were harvested after a few days and those suita-ble for diffraction were identified and mounted on a diffract-ometer at low temperatures. The above experimental conditionscan be compared with those in the study of Dong et al. (15, 16)who used THF:water ratios of about 1∶63 with up to 5% ammo-nia to study methane hydrate formation in the presence ofammonia. In their experiments, Dong et al. (15, 16) did not usesufficient THF to lead to the direct formation of the binaryTHFþNH3 clathrate hydrate and ammonia hydrate formationwas not observed.

Structural data were recorded at 100 K (see Materials andMethods). The structure was shown to be the usual cubic structureII (sII) clathrate hydrate (28), space group Fd-3m with a unit celledge of 1.71413(6) nm. The THF molecules occupy the largecages, as expected, and 39% of the small cages are filled with am-monia molecules (Fig. 1). This measured small cage occupancy islikely to be a function of the starting composition. Whether smallcages could be completely filled with a more ammonia-rich liquidis uncertain and should be studied separately. One of the watermolecules in the small cages which encapsulate the ammoniaguests has moved out of its normal position, 0.118 nm inward intothe small cage thus forming H2O⋯H-NH2 hydrogen bonds(0.267–0.272 nm) with NH3 (as suggested by hydrogen positionson the ammonia guest). The formation of HOH⋯NH3 hydrogenbonding could also lead to this water displacement. This displace-ment breaks the hydrogen bond of this water molecule withanother water from an adjacent large (or small) cage (O⋯O dis-tance 0.393 nm), thus distorting both large and small cages. Theequivalent O⋯O distance in the undistorted edges of the waterpolyhedra is 0.279 nm. It should be noted that the minimum heavyatom N⋯O distance in the clathrate hydrate is smaller than in theammonia hydrates, where the more uniform local environment ofthe NH3 molecule results in N⋯O distances of 0.305 to 0.330 nm.

Because structural evidence for the mixed THFþNH3 sIIhydrate showed that it is possible to incorporate ammonia into

Author Contributions: R.K., K.S., and J.A.R. performed the vapor deposition experiments;K.S., K.A.U., and J.A.R. did the powder and single-crystal XRD experiments; S.A., K.S., andJ.A.R. performed the molecular dynamics simulations; and J.A.R., S.A., and K.S. wrote themanuscript.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Data deposition: The single-crystal XRD structure of the binary tetrahydrofuran andammonia structure II clathrate hydrate has been deposited in the Cambridge StructuralDatabase, Cambridge Crystallographic Data Centre, Cambridge CB2 1EZ, United Kingdom(CSD reference no. 864781).1To whom correspondence should be addressed. E-mail: john.ripmeester@nrc-cnrc.gc.ca.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1205820109/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1205820109 PNAS ∣ September 11, 2012 ∣ vol. 109 ∣ no. 37 ∣ 14785–14790

CHEM

ISTR

Y

Dow

nloa

ded

by g

uest

on

Dec

embe

r 14

, 202

0

clathrate cages, attempts were made to form clathrate hydrateswith only ammonia as guest. Pure ammonia clathrate hydratesynthesis cannot be done by simply cooling aqueous ammoniasolutions, as a variety of stoichiometric hydrates of ammonia areknown to form preferentially (2, 11, 18, 19). Other ways of form-ing clathrate hydrates include vapor deposition of water at lowtemperatures to yield amorphous ice, followed by exposure of theice to a pressure of guest gas and annealing (29), or vapor code-position of water and the potential guest material at low tempera-tures, again, followed by annealing (23, 30, 31). It has been shownthat below approximately 140 K ice surfaces are relatively inertunless strong hydrogen-bond donors or acceptors are present. Forexample, molecules normally hydrolyzed by water such as formal-dehyde will not do so when in contact with ices at low tempera-ture and will template clathrate hydrate lattices when amorphousice is annealed (30).

Following the latter procedure, described in detail in Materialsand Methods, vapor codeposits of ammonia and water were pre-pared at T < 20 K. The codeposited crystals were harvested at77 K and kept in liquid nitrogen until appropriate measurementscould be made. A portion of each amorphous product wasmounted in a low temperature cell on a powder X-ray diffract-ometer and the samples annealed stepwise, increasing tempera-ture from approximately 100 K. Fig. 2 shows the transformationof the amorphous deposit upon annealing. Fig. 2A shows the pow-der X-ray diffraction (PXRD) pattern for amorphous ice withtraces of ices Ic and Ih at 100 K. At 140 K, the amorphous ice haslargely transformed to ices Ic and Ih plus a number of othercrystalline phases (the Bragg positions for Ic are omitted). Thesereflections continue to develop with increase of temperature to150 K. The inset (Fig. 2B) shows a vertical expansion of the 150 Kpattern. For each pattern the profiles are matched to those ofknown crystalline phases, including the stoichiometric ammoniahydrates (hemi-, mono-, and dihydrates), hexagonal ice Ih, andclathrate hydrates. Eight reflections (shown in Fig. 2) can be as-signed to the cubic structure I (sI) ammonia clathrate hydrate andindexed to be consistent with sI clathrate hydrate, space groupPm-3n, with unit cell edge 1.1818(2) nm at 150 K.

The interactions of ammonia with methane clathrate hydrateare also considered to be important in determining the phaseequilibria on Titan and Enceladus. Several vapor deposits ofmethane, ammonia, and water were prepared as described inMaterials and Methods and annealing of the vapor codeposits wasfollowed by PXRD. Fig. 3A shows the PXRD pattern at 112 K,indicating the material to be mainly amorphous except for somecrystalline Ih. At 143 K (Fig. 3A, second trace) most of the pat-tern for the amorphous deposit has transformed into ice Ic plusother crystalline phases that upon closer inspection suggested thepresence of both cubic sI and sII clathrate hydrates. At 160 K, thispattern is well developed and a vertically expanded detail is givenin Fig. 3B. Profile matching for each powder pattern shows that agood fit is obtained by considering the presence of ices Ic and Ih,

as well as cubic sI [Pm-3n, a ¼ 1.1847ð8Þ nm] and cubic sII clath-rate hydrates [Fd-3m, a ¼ 1.7161ð9Þ nm]. To observe both com-mon cubic clathrate hydrate structures is surprising. Structure Iis the usual clathrate that forms for methane at low pressuresalthough metastable sII methane hydrate has been observed aswell (30). A blank run for a vapor codeposit of water and methanebut without ammonia did not reveal any clathrate hydrate undersimilar conditions of annealing (Fig. S1). Further annealing ofthe three-component codeposit at 180 K showed (Fig. 3A) thatthe sII clathrate hydrate pattern had largely disappeared, andreflections for the stoichiometric ammonia hydrates were absent.The transformation from mixed sI and sII hydrates to sI hydratewas confirmed by observing methane C-H symmetric stretchfrequencies around 2;905 cm−1 by Raman spectroscopy (32)(Fig. S2). At low temperature, the deposited methane peakappeared at 2;908 cm−1, which is close to the Raman shift ofdissolved methane in water (33). This peak splits into a doubletat 160 K; one peak at 2;903 cm−1 representing the methane inthe large cages and the second at 2;914 cm−1 representingmethane in the small cages. As temperature increases, the ratioof methane in the large cages to that of the guest in the smallcages increases, which implies the transformation to sI phase(six large cages to two small cages) which has a greater proportionof large cages than sII (8 large cages to 16 small cages). Similarenhanced processes of CO2 clathrate hydrate formation on ice

Fig. 1. The large and small cages of the cubic structure II THF-ammoniabinary clathrate hydrate from single crystal X-ray diffraction. The ammoniaguest has moved awater molecule out of its normal position by pulling it intothe small cavity by forming a H2N-H⋯OH2 or H3N⋯HOH hydrogen bond.

Fig. 2. PXRD patterns of NH3 and H2O co-deposition sample (A) recorded at100, 140, and 150 K; (B) expansion of the pattern at 150 K. Lattice parameters(150 K) are: a ¼ 0.4527ð1Þ nm, b ¼ 0.5586ð1Þ nm, c ¼ 0.9767ð3Þ nm for am-monia monohydrate (space group: P212121); a ¼ 0.4496ð1Þ nm and c ¼0.7339ð1Þ nm for ice Ih (space group P63∕mmc); a ¼ 1.1818ð2Þ nm for sIammonia clathrate hydrate (space group Pm-3n); a ¼ 0.8404ð2Þ nm, b ¼0.8441ð3Þ nm, and c ¼ 5.345ð1Þ nm for ammonia hemihydrate (space groupPbnm); and a ¼ 0.7125ð1Þ nm for ammonia dihydrate (space group P213).

14786 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1205820109 Shin et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 14

, 202

0

nanoparticles at low temperatures in the presence of hydrogenbonding ether guest molecules have been observed (34).

To study molecular level structure and dynamics of ammoniacontaining clathrate hydrates, we performed molecular dynamicssimulations of the THF (in large cages) + NH3 (in 39% of smallcages) binary sII clathrate hydrate from 100 to 240 K. We alsosimulated the pure NH3 sI and sII clathrate hydrates and binaryCH4 (large cages) + NH3 (small cages) sII clathrate hydratesfor temperatures in the range of 100 to 240 K. The moleculardynamics simulation methodology is described in detail in Mate-rials and Methods. For the duration of the simulations, all theclathrates are mechanically stable at simulation temperatures

up to 200 K and did not show signs of decomposition and cagecollapse.

In the molecular simulations of the sI and sII pure ammoniahydrate phases, we observe hydrogen bonding between theammonia guest molecules and the water molecules in both thesmall and large clathrate hydrate cages. Cases where ammoniais the hydrogen donor (H2N-H⋯OH2) and hydrogen acceptor(H3N⋯H-OH) are both common but the latter is more frequent(Figs. S3 and S4). On the other hand, in the binary sII hydratephases with THF or methane in the large cages, ammonia showedeither no hydrogen bonding or only a small probability for hy-drogen bond formation with water molecules of the small cages(see Figs. S5 and S6). From integrating the hydrogen bondingpeak in the radial distribution functions (RDF), we calculatedthe probability of hydrogen bonding, p, in the pure NH3 sI, sII,and binary NH3 þ CH4 sII clathrate hydrates for ammoniamolecules and the results are given in Table 1. Peaks in the RDFwith H2N-H⋯OH2 or H3N⋯H-OH distances less than 0.25 nmwere assigned to hydrogen bonding. The areas underneath thehydrogen bond peaks in the RDF give a measure of the hydrogenbond probability (see Materials and Methods). The probability ofhydrogen bonding increases with temperature, which shows thatthe ammonia-water hydrogen bonding in the clathrate hydrates isendothermic. The temperature dependence of the hydrogenbonding probability is evaluated using the van’t Hoff plot oflnK [where K ¼ p∕ð1 − pÞ] as a function of 1∕T (Fig. S7).The slopes of the linear plots are used to determine the enthal-pies of hydrogen bond formation for ammonia in the differentenvironments which are also given in Table 1. The formation ofhydrogen bonds in the sII clathrate hydrate is less endothermicand the more facile formation of these bonds may destabilizethe sII clathrate hydrate water lattice to a greater extent thanthe sI case.

Sample hydrogen bonded configurations of the ammoniaguests in the large and small cages of the sI and sII clathratehydrates are shown in Fig. 4. Hydrogen bond configurationswith ammonia acting as the proton donor and the proton accep-tor (often simultaneously) are observed. The ammonia guest re-mains mostly in the cage but in some configurations the ammoniamolecule displaces one of the water molecules from a lattice site.The water molecules in the hexagonal faces of the large clathratehydrate cages are less hindered and may also have less stablewater-water hydrogen bonds which require less energy to breakcompared to the water molecules forming hydrogen bonds inpentagonal faces. Snapshots from the simulations show that theammonia molecules preferentially hydrogen bond with the watermolecules in the hexagonal faces. Ammonia-water hydrogenbonding severely distorts the small and large cage structures. Attemperatures below 200 K, the hydrate phase maintains itsmechanical stability without framework collapse despite the am-monia-water hydrogen bonds and the defects they induce in thehydrate lattice hydrogen-bonding network. As we only performeda molecular dynamics simulation and not a full free energy cal-culation comparing the thermodynamic stability of the clathratehydrate phase to the products, we cannot comment on the ther-modynamic stability of the ammonia clathrate hydrate phase. Bymechanical stability we imply that the hydrate structure did not

Fig. 3. PXRD patterns of NH3, CH4, and H2O co-deposition sample (A) re-corded at 112, 143, 160, and 180 K; (B) expansion of the pattern at 160 K.Vertical thin red lines and arrows indicate the peaks from sII hydrate and bluelines and arrows indicate the peaks from sI hydrate. Lattice parameter(160 K): a ¼ 0.4491ð3Þ nm and c ¼ 0.7333ð5Þ nm for Ih (space groupP63∕mmc), a ¼ 0.6361ð3Þ nm for Ic (space group Fd-3m), a ¼ 1.1847ð8Þ nmfor sI hydrate (space group Pm-3n), and a ¼ 1.7161ð9Þ nm for sII hydrate(space group: Fd-3m).

Table 1. The average percent (H3N⋯H-OH) hydrogen bonding configurations calculated from RDFs and the enthalpy ofhydrogen bond formation for pure sI and sII NH3 hydrates and binary sII CH4 (large cages) + NH3 (small cages) hydrateat different temperatures

Guests 100 K 120 K 140 K 160 K 170 K ΔHHB∕kJ·mol−1

NH3 (pure sI, large cages) 4.6 12.9 34.5 54.4 68.9 7.66NH3 (pure sI, small cages) 1.4 2.1 15.0 32.0 49.8 8.90NH3 (pure sII, large cages) 7.8 17.1 28.3 51.5 61.8 5.85NH3 (pure sII, small cages) 1.4 4.3 6.1 24.4 33.6 6.95NH3 (binary sII, small cages) ∼0 0.6 1.3 1.8 2.9 5.09

Shin et al. PNAS ∣ September 11, 2012 ∣ vol. 109 ∣ no. 37 ∣ 14787

CHEM

ISTR

Y

Dow

nloa

ded

by g

uest

on

Dec

embe

r 14

, 202

0

decompose under the pressure-temperature conditions of thesimulation. As temperature and vibrational amplitudes of thewater molecules in the hydrate lattice increase, these lattice de-fects lead to the breakdown of the water framework structure andinstability of the hydrate phase.

From Table 1, the ammonia molecules in the large sI and sIIcages form hydrogen bonds more readily than in the corre-sponding small cages. In addition, the larger hydrogen bondingformation enthalpy in the sI hydrate compared to the sII hydrate,reveals that the water lattice in the sI phase is less susceptibleto hydrogen bonding with ammonia and is preferred when theammonia forms a clathrate phase with amorphous ice uponannealing.

In contrast to the pure NH3 hydrates, binary CH4 þNH3 sIIhydrate shows a H3N⋯H-OH hydrogen bond probability of onlyapproximately 3% even at 170 K. Incorporating methane in thecages of the binary clathrate hydrate apparently stabilizes thecages which encapsulate ammonia. The molecular dynamicssimulation results are consistent within the framework of experi-mental knowledge of the ammonia hydrate system describedabove. Strong hydrogen bonding of ammonia with water in thesolid phase is expected. The stability of the ammonia hydrate isdependent on the low temperature and/or stabilizing effect ofmethane help gas.

DiscussionThe presence of both ammonia and methane give a clathratehydrate that is much more stable than that of the pure ammoniahydrate (stability up to 180 K for the mixed clathrate as comparedto 150 K for the pure NH3 clathrate). On the other hand, thepresence of ammonia is critical in forming a clathrate in a tem-perature region where methane by itself does not form a clathratehydrate. The activation of ice surfaces by ammonia at low tem-peratures is a well known phenomenon (35), so it may well be thatsynergistic behavior of ammonia and methane on the amorphousice surface produce a reasonably stable clathrate: ammonia toinitiate the reaction by introducing defects and methane to sta-bilize the clathrate lattice by limiting guest-host hydrogen bond-ing. Experiments with ethane and ammonia gave similar results(Fig. S8). From the work presented it is clear that ammonia playsan important role, so far unrealized, in forming solid clathratephases at low temperatures in planetary environments. It maybe difficult to quantify the effects of ammonia on a clathratehydrate-forming environment as it is clear that some ammoniaphases have time and temperature-limited regions of existence.

Recent spectrophotometric studies by Nelson and coworkershave shown variations in the surface reflectance on Titan which

have been attributed to episodic cryovolcanic effusion eventswhich bring NH3 from the interior of Titan to its surface in theform of a fog or frost (36). Most work on ammonia present inTitan focuses on ammonia in aqueous solution in the “magma”and its effect on methane clathrate hydrate inhibition and decom-position in that region (5). Our work shows that under tempera-ture and pressure conditions similar to those found on Titan’ssurface (5) (T ≈ 90 K and Patm ≈ 1.6 bar), considering the pre-sence of methane in the atmosphere of Titan and availability ofice on its surface, it is possible that the extruded ammonia vaporforms binary clathrate hydrate phases with methane. Clathratehydrate formation after ammonia vapor deposition in the pre-sence of methane gas and annealing to about 100 K can providea mechanism for the removal of ammonia and methane fromTitan’s atmosphere. Ethane clathrate hydrate formation may alsobe catalyzed by the presence of ammonia.

Similar cryovolcanic activity is also seen in Saturn’s moonEnceladus where plumes of ammonia, methane, and water vaporare extruded from the South Pole (37). The sources of thesemethane extrusions in both moons have been related to clathratehydrate dissociation (5, 38, 39).

The broader significance of the catalyzing effect of ammoniaon methane hydrate formation under vapor deposition conditionsmay be to the conditions of primordial ices in planets, comets,and planetesimals in solar system formation models. The ammo-nia can catalyze the formation of nonstoichiometric gas hydrateswhich are then trapped in icy grains and accreted into planetaryobjects.

In addition to subsurface processes, ammonia-bearing liquidsand vapors may interact directly with surface clathrate hydratephases on Titan. In situ clathrate hydrate formations (not invol-ving ammonia) have been considered important in determiningthe surface composition and structures of this moon (40–42) andthe presence of ammonia may play a role in these processes.

Polar, water miscible molecules such as ammonia are tradition-ally considered as clathrate hydrate formation inhibitors. Ourexperiments and computations show that the inhibition effect isnot due to inherent instability of the solid clathrate hydrate phaseof these guest molecules, but is rather the stabilizing effect ofthese guests on the aqueous phase from which the clathratehydrate is formed.

Materials and MethodsSample Preparation. Vapor codepositions of water and guest materials(ammonia, methane, and 50 mole % ammonia/methane mixture) wereperformed in an evacuated chamber at approximately 0.01 mbar. The waterwas vaporized from a pipette into the evacuation chamber by low pressure

Fig. 4. Sample configurations of the NH3 guests hydrogen bonding with the water molecules of the large (L) and small (S) cages of the sI and sII clathratehydrate (Top: sI, Bottom: sII cages). The snapshots of the cages were extracted from the periodic simulation cell. In cases where NH3 is incorporated into thewater lattice, the displaced water molecule is also shown.

14788 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1205820109 Shin et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 14

, 202

0

and deposited on a copper plate which was cooled down to approximately16 K by DISPLEX (Air Products). In each experiment a total of 600 mmHg ofthe guest gas (ammonia, methane or 1∶1 ammonia∶methane mixture) wasinjected into a 1 L bulb and slowly injected into the deposition chamber. Thewater and guest gas vapor flows were controlled by leak valve and after 24 happroximately 2 g of water and 0.5 g of guest vapors were depositedon the copper plate. The copper plate covered with the product depositwas removed from the evacuation chamber and quickly immersed in liquidnitrogen with minimal exposure to air. Significant moisture condensation onthe sample during the short transfer period is not expected. For the variouscharacterization experiments the materials were transferred to the powderX-ray diffraction and Raman instruments.

In the single crystal experiments, a THFþ NH3 binary sII hydrate was pre-pared from a 1∶17 solution of THF in 5% aqueous ammonia. The synthesisconditions give an approximate mole ratio of 1∶0.946∶17 for THF∶NH3∶H2Oin this solution. Upon cooling the solution to 263 K masses of crystalsappeared. Crystals suitable for diffraction were identified by examinationunder a microscope mounted in a cold box. The structure obtained fromthe single crystal diffraction experiment was used to determine that 39%of the small cages were occupied by ammonia which gives an approximateclathrate hydrate composition of 1∶0.78∶17 for THF∶NH3∶H2O. Different am-monia small cage occupancies are likely to be found starting from differentinitial compositions. Whether small cages could be filled with a more ammo-nia-rich liquid is uncertain and should be the subject of a separate study.

X-Ray Data Collection and Structure Solution. Single crystal X-ray diffractiondata weremeasured on a Bruker Apex 2 Kappa diffractometer at 100 K, usinggraphite monochromatized Mo Kα radiation (λ ¼ 0.71073 Å). The unit cellwas determined from randomly selected reflections obtained using theBruker Apex2 automatic search, center, index, and least squares routines.Integration was carried out using the program SAINT, and an absorption cor-rection was performed using SADABS (43). The crystal structures were solvedby direct methods and the structure was refined by full-matrix least-squaresroutines using the SHELXTL program suite (44). All atoms were refined ani-sotropically. Hydrogen atoms on guest molecules were placed in calculatedpositions and allowed to ride on the parent atoms.

Powder X-Ray Diffraction (PXRD). The PXRD pattern was recorded on aBRUKER AXS model D8 Advance diffractometer in the θ∕θ scan mode usingCu Kα radiation (λ1 ¼ 1.5406 Å, λ2 ¼ 1.5444 Å, and I2∕I1 ¼ 0.5). The samplesstored in liquid nitrogen were quickly transferred to the X-ray stage cooleddown to approximately 100 K in air and diffraction patterns were measuredwith increasing temperature. The experiment was carried out in the stepmode with a fixed time of 2 s and a step size of 0.03014° for 2θ ¼ 8 to42° at each temperature. The powder pattern was refined by the Le Bailmethod using the profile matching method within FULLPROF (45).

Raman Spectroscopy. The Raman spectra were measured using a dispersiveRaman microscope instrument with the focused 514.5 nm line of an Ar-ionlaser for excitation. The scattered light was dispersed using a single-gratingspectrometer (1,200 grating) and detected with a CCD detector (−90 °C,cooled by liquid nitrogen). The sample kept in liquid nitrogen was loadedonto a brass sample holder, which was cooled down to approximately 100 Kin air and measured with increasing temperature. The sample holder was

covered by quartz glass and dry nitrogen gas was blown over the surface dur-ing measurements in order to keep ice from condensing.

Computational Methods. Molecular Dynamics simulations were performedwith DL_POLY version 2.20 (46) using the leapfrog algorithm with a time stepof 1 fs. The positions of the water oxygen atoms of the structure I and struc-ture II clathrate hydrate phases were taken from X-ray crystallography andthe water hydrogen atoms assigned to oxygen atoms in the unit cell in such away as to simultaneously satisfy the ice rules and minimize the net unit celldipole moment. The centers of mass of the guests are initially put in the cen-ter of the hydrate cages. For simulations of the sI clathrate hydrate, 3 × 3 × 3

replicas of the unit cell are usedwith 162 large cages and 54 small cages in thesimulation cell. For simulations of the sII clathrate hydrate, a 2 × 2 × 2 replicaof the unit cell is used with 64 large cages and 128 small cages.

In the simulations, the van der Waals and electrostatic intermolecular in-teractions of water are modeled with the TIP4P potential (47), THF with theAMBER force field (48) NH3 with the OPLS force field of Rizzo and Jorgensen(49), and the TRAPP potential for CH4 (50). The potential parameters in theforce fields are given in Table S1. The partial electrostatic charges on theatoms of THF are calculated using the CHELPG method (51) and the otherguest molecules include partial atomic charges as part of the force field.The water and guest molecules are given freedom of motion in the simula-tions but their internal structure is considered rigid. A cutoff of 13 Å is usedfor the long-range forces in the simulation.

For the THF (in large cages) + NH3 (in small cages) binary sII clathratehydrate three sets of simulations at 100 K, (temperature of experimentalsingle crystal structure determination), 183 K, and 200 K were performed.For each simulation, one set of simulations with the all small cages occupiedby NH3 guests and a second set with 39% of the small cages occupied by NH3

guests (in accordance to the experimental values) were performed. In addi-tion, seven simulation sets were performed at 100, 120, 140, 160, 170, 200,and 240 K for the pure NH3 sI and sII clathrate hydrates and the CH4 (largecages) + NH3 (small cages) binary sII clathrate hydrate to study the stability ofthese phases. Simulations for each clathrate hydrate are performed for 6 nswith 100 ps of temperature scaled equilibration to bring the clathrate hy-drate with each guest combination to the desired volume and temperaturein the constant pressure—constant temperature (NPT) ensemble.

The hydrogen bonding probabilities between NH3 guest and host water,p, were calculated from the radial distribution functions (RDF), gðrÞ betweenthe corresponding atoms on the guest ammonia and cage water molecules,

p ¼Z

rmin

0

ρgðrÞ4πr2dr: [1]

In Eq. 1, rmin is the first minimum in the RDF which appears after the hydro-gen bond peak of ∼2 Åwhich represent hydrogen bonded NH3 with the cagewater molecules.

ACKNOWLEDGMENTS. We thank G. L. McLaurin and S. Lang for technical sup-port. We thank an anonymous reviewer for providing constructive commentson this manuscript. We also thank the National Research Council of Canadafor financial support.

1. Lunine JI, Stevenson DJ (1987) Clathrate and ammonia hydrates at high pressure:Application to the presence of methane on Titan. Icarus 70:61–77.

2. Fortes AD, Choukroun M (2010) Phase behavior of ice and hydrates. Space Sci Rev153:185–218.

3. Loveday JS, et al. (2001) Stable methane hydrate above 2 GPa and the source of Titan’satmospheric methane. Nature 410:661–663.

4. Tobie G, Lunine JI, Sotin C (2006) Episodic outgassing as the origin of atmosphericmethane on Titan. Nature 440:61–64.

5. Choukroun M, Grasset O, Tobie G, Sotin C (2010) Stability of methane clathratehydrates under pressure: Influence on outgassing processes of methane on Titan.Icarus 205:581–593.

6. Cable ML, et al. (2012) Titan tholins: Simulating titan organic chemistry in the Cassini-Huygens era. Chem Rev 112:1882–1909.

7. Fortes AD (2012) Titan’s internal structure and the evolutionary consequences. PlanetSpace Sci 60:10–17.

8. Castillo-Rogez J, Lunine JI (2010) Evolution of Titan’s rocky core constrained by Cassiniobservations. Geophys Res Lett 37:L20205.

9. Choukroun M, Tobie G, Grasset O (2006) Ammonia, a methane hydrate inhibitor—Implications for Titan’s atmospheric methane. Lunar and Planetary Science, XXXVII;Papers Presented to the Thirty-Seventh Lunar and Planetary Conference p 1640.

10. Sloan ED, Koh CA (2008) Clathrate Hydrates of Natural Gases (CRC Press, Boca Raton,FL), 3rd Ed,, pp 230–232.

11. Uras N, Devlin JP (2000) Rate study of ice particle conversion to ammonia hemihydrate:Hydrate crust nucleation and NH3 diffusion. J Phys Chem A 104:5770–5777.

12. Uras N, Buch V, Devlin JP (2000) Hydrogen bond surface chemistry: Interaction of NH3

with an ice particle. J Phys Chem B 104:9203–9209.13. Kurnosov A, Dubrovinsky L, Kuznetsov A, Dmitriev V (2006) High Pressure/high

temperature behavior of the methane—ammonia water system up to 3 GPA.Z Naturforschg 61b:1773–1576.

14. Nelmes RJ, Loveday JS, Guthrie M, Francis DJ (2000) Molecular systems and mixturesunder pressure. ISIS Experimental Report RB 10795 (Rutherford Appleton Laboratory,Oxfordshire, UK).

15. Dong T, et al. (July 6–10, 2008) Experimental determination of methane hydrateformation in presence of ammonia. Proceedings of the 6th International Conferenceon Gas Hydrates (ICGH 2008).

16. Dong T, et al. (2009) Experimental study of separation of ammonia synthesis vent gasby hydrate formation. Petrol Sci 6:188–193.

17. Wang L, Liu A, Guo X, Chen G, Li G (2008) Experimental determination and calculationof methane hydrate formation in the presence of ammonia. J Chem Ind Eng (China)59:276–280.

18. Fortes AD, Wood IG, Brodholt JP, Vočadlo L (2003) The structure, ordering and equa-tion of state of ammonia dihydrate ðNH3 · 2H2OÞ. Icarus 162:59–73.

19. Fortes AD, Suard E, Lemée-Cailleau MH, Pickard CJ, Needs RJ (2009) Crystal structureof ammonia monohydrate phase II. J Am Chem Soc 131:13508–13515.

Shin et al. PNAS ∣ September 11, 2012 ∣ vol. 109 ∣ no. 37 ∣ 14789

CHEM

ISTR

Y

Dow

nloa

ded

by g

uest

on

Dec

embe

r 14

, 202

0

20. Mak TCW (1965) Hexamethylenetetramine hexahydrate: A new type of clathratehydrate. J Chem Phys 43:2799–2805.

21. Koga K, Tanaka H (1996) Rearrangement dynamics of the hydrogen-bonded networkof clathrate hydrates encaging polar guest. J Chem Phys 104:263–272.

22. Alavi S, Susilo R, Ripmeester JA (2009) Linking microscopic guest properties to macro-scopic observables in clathrate hydrates: Guest-host hydrogen bonding. J Chem Phys130:174501.

23. Buch V, et al. (2009) Clathrate hydrates with hydrogen bonds. Phys Chem Chem Phys11:10245–10265.

24. Shinohara H, Nagashima U, Tanaka H, Nishi N (1985) Magic numbers for water-ammo-nia binary clusters: Enhanced stability of ion clathrate structures. J Chem Phys 83:4183.

25. Diken EG, Hammer NI, Johnson MA, Christie RA, Jordan KD (2005) Mid-infrared char-acterization of the NH4

þ · ðH2OÞn clusters in the neighborhood of the n ¼ 20 “magic”number. J Chem Phys 123:164309.

26. Khan A (2001) Theoretical studies of NH4þ · ðH2OÞ20 and NH3ðH2OÞ20Hþ clusters.

Chem Phys Lett 338:201–207.27. Willow SY, Singh NJ, Kim KS (2011) NH4

þ Resides inside the water 20-mer cage asopposed to H3Oþ , which resides on the surface: A first principles molecular dynamicssimulation study. J Chem Theory Comput 7:3461–3465.

28. Jeffrey GA (1984) Hydrate Inclusion Compounds in Inclusion Compounds, eds JLAtwood, JED Davies, and DD MacNicol (Academic Press, London), Vol 1.

29. Hallbrucker A, Mayer E (1990) Unexpectedly stable nitrogen, oxygen, carbon monox-ide and argon clathrate hydrates from vapor-deposited amorphous solid water: AnX-ray and two-step differential scanning calorimetry study. J Chem Soc Faraday Trans86:3785–3792.

30. Ripmeester JA, Ding L, Klug DD (1996) A clathrate hydrate of formaldehyde. J PhysChem 100:13330–13332.

31. Monreal IA, Devlin JP, Maşlakci Z, Çiçek MB, Uras-Aytemiz N (2011) Controlling non-classical content of clathrate hydrates through the choice of molecular guests andtemperature. J Phys Chem A 115:5822–5832.

32. Schicks JM, Ripmeester JA (2004) The coexistence of two different methane hydratephases under moderate pressure and temperature conditions: Kinetic versus thermo-dynamic products. Angew Chem Inter Ed 43:3310–3313.

33. Pironon J, Grimmer J, Teinturier S, Guillaume D, Dubessy J (2003) Dissolved methanein water: Temperature effect on Raman quantification in fluid inclusions. J GeochemExplor 78–79:111–115.

34. Hernandez J, Uras N, Devlin JP (1998) Coated ice nanocrystals from water-adsorbatevapor mixtures: Formation of ether—CO2 clathrate hydrate nanocrystals at 120 K.J Phys Chem B 102:4526–4535.

35. Delzeit L, Powell K, Uras N, Devlin JP (1997) Ice surface reactions with acids and bases.J Phys Chem B 101:2327–2332.

36. Nelson RM, et al. (2009) Saturn’s Titan: Surface change, ammonia and implications foratmospheric and tectonic activity. Icarus 199:429–441.

37. Porco CC, et al. (2006) Cassini observes the active south pole of Enceladus. Science311:1393–1401.

38. Fortes AD, Grindrod PM, Trickett SK, Vočadlo L (2007) Ammonium sulfate on Titan:Possible origin and role in cryovolcanism. Icarus 188:139–153.

39. Fortes AD (2007) Metasomatic clathrate xenoliths as a possible source for the southpolar plumes of Enceladus. Icarus 191:743–748.

40. Fortes AD, Stofan ER (March 14–18th, 2005) Clathrate formation in the near-surfaceenvironment of Titan. 36th Lunar and Planetary Science Conference (abstr 1123).

41. Osegovic JP, MaxMD (2005) Compound clathrate hydrate on Titan’s surface. J GeophysRes 110:E08004.

42. Choukroun M, Sotin C (2012) Is Titan’s shape caused by its meteorology and carboncycle? Geophys Res Lett 39:L04201.

43. Sheldrick GM (2002) SADABS Version 2.03 (University of Gottingen, Germany).44. Sheldrick GM (2000) SHELXTL, Version 6.10 (Bruker AXS Inc., Madison, WI).45. Rodriguez-Carvajal J (1990) FULLPROF: A program for rietveld refinement and pattern

matching analysis. Abstracts of the Satellite Meeting on Powder Diffraction of the XVCongress of the IUCr (International Union of Crystallography, Toulouse, France) p 127.

46. Smith W, Forester TR, Todorov IT DL_POLY 2.20 (CCLRC, Daresbury Laboratory,Daresbury).

47. Abascal JLF, Sanz E, Garcia Fernandez R, Vega C (2005) A potential model for the studyof ices and amorphous water: TIP4P/Ice. J Chem Phys 122:234511.

48. Cornell WD, et al. (1995) A second generation force field for simulation of proteins,nucleic acids, and organic molecules. J Am Chem Soc 117:5179–5197.

49. Rizzo RC, Jorgensen WL (1999) OLPS all-atom model for amines: Resolution of theamine hydration problem. J Am Chem Soc 121:4827–4836.

50. Martin MG, Siepmann JI (1998) Transferable potentials for phase equilibria. 1. Unitedatom description of n-alkanes. J Phys Chem B 102:2569–2577.

51. Breneman CM, Wiberg KB (1990) Determining atom-centered monopoles from mole-cular electrostatic potentials. The need for high sampling density in formamide con-formational analysis. J Comput Chem 11:361–373.

14790 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1205820109 Shin et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 14

, 202

0