Gas hydrates have many unusual properties and explain many curious happenings
If chlorine gas is passed into a dilute solution of CaCl2 at about 0 °C, greenish, feathery crystals appear that can be removed from the solution, dried, and kept indefinitely at room temperature. If these crystals are dissolved in water, chlorine gas is liberated. If they are heated at one end of a closed tube, the other end of which is immersed in ice water, liquid chlorine appears at the cold end. These crystals are chlorine hydrate, discovered by Sir Humphrey Davy in 1811 shortly after proving that chlorine is an element, and studied in more detail by Michael Faraday in 1823. The formula of this substance is Cl2·H2O.
Wroblewsky observed ice-like crystals in carbon dioxide-water systems with the apparent formula CO2·8H2O in 1892, which was confirmed by several other observers soon afterwards. At 0°C, this substance was stable for a CO2 pressure above 12.2 atm. At 10°C, the pressure was 44.3 atm. As we shall see, clathrate hydrates, of which these are examples, actually have the formula 6X·46H2O. The cage formed by 46 water molecules has 6 "cells" that can hold a molecule. If all the cells are filled, the "8" in the formula is really 7.66.
Natural gas pipelines in Kazakhstan were found to be clogged with an ice-like substance in 1929. It had been observed that pipelines were often clogged this way in winter, when the water vapor in the natural gas was assumed to condense and freeze, so it was general practice to dehydrate gas to prepare it for the pipeline. However, in this case the 'ice' formed at temperatures well above the freezing point of water when the gas pressure was high, say above 10 atmospheres. When the 'ice' was heated, it was found to release natural gas. Therefore, it was realized that the 'ice' was a crystalline substance composed of water and gas that formed under high pressure, and that was unusually stable. It was called a gas hydrate, and its similarity to chlorine hydrate was realized. Gas hydrates are much more troublesome than mere ice to the production and transportation of natural gas, which usually takes place under rather high pressures for economic reasons.
A recent (1998) pipeline blockage near Telluride, Colorado was ascribed to substandard gas delivered to the pipeline, probably wet gas that formed hydrates under the high pressure and near-freezing temperatures. Hydrates were not mentioned in the news reports, but this is not surprising.
Another place where high pressures accompany water near the freezing point is in ocean depths. At a depth of 500m, the pressure is about 50 atmospheres, and the temperature 4 - 5 °C, ideal for gas hydrate formation. Wherever suitable gases, usually hydrocarbons, are released from sediments, these hydrates have been found to exist, and in relatively huge amounts. Hydrates are known not only in Arctic regions, such as the Alaska north slope, but also off places like South Carolina. Hydrates also exist in permafrost regions near the surface, as well as in sedimentary formations where hydrocarbons, water, and low temperatures are found. Enthusiasts have proposed that there are great reserves of methane in such hydrates, for example on the Atlantic continental shelf. Production would be extremely difficult, however. More recent estimates of the possible amounts available are much smaller than the original enthusiastic figures.
The wreck of a steam trawler has recently been discovered at the bottom of the North Sea, 150 km northeast of Aberdeen, Scotland, apparently in a crater. It is suspected that a release of methane produced a water-methane bubble mixture of low density, in which the trawler sank when it lost buoyancy. This region, called Witch's Hole, is known to be pockmarked with such craters. It is mentioned that certain sulphur-oxidizing bacteria produce methane, so this may be another source than petroleum for the methane. Hydrates seem a better source of catastrophic releases than simple bubbles of methane in the sediments (which have not been observed otherwise).
In the Caspian sea, mud buried more than 25 km deep is squeezed upward in mud diapirs that approach the surface, like Gulf of Mexico salt domes. Gas hydrates can form on the crests of these diapirs, then decompose as they are raised into conditions of lower pressure and higher temperature. The released gas produces 'mud volcanoes,' and it is believed that the energy released can raise the temperature locally to the ignition point of methane-air mixtures (537 °C), creating dangerous explosions and hazards to the development of the local oil fields. It has even been suggested that the phenomena of the Bermuda Triangle are also due to the decomposition of gas hydrates.
Gas and rock blowouts are a hazard in coal mining, for example in the Don basin in Russia, where they begin to occur at depths of about 180 m. Yuri Makogon has suggested that their cause can be traced to the formation of gas hydrates (methane is frequently associated with coal) in the Pleistocene, when the climate was much colder and wetter, and favorable to hydrate formation. The current temperature, far above freezing, has decomposed the hydrates under confinement, which produces very high pressures, even 2000 atmosperes and more. The rocks affected must, of course, be well-sealed lenses to retain the high pressure.
Explosive release of methane from hydrates is a suggested cause of the Paleocene-Eocene Thermal Maximum. At this period, atmospheric CO2 concentrations rose to 2000-3000 ppm (380 ppm currently), but methane is a very efficient greenhouse gas. The idea that global warming might release large quantities of methane frightens some enthusiasts, but the possibility is extremely remote.
Methane hydrates in Lake Vostok, a large Antarctic subglacial lake, have been suggested. The water is under large pressure, but at only -3° C.
It is clear that gas hydrates are a very different thing than the salt hydrates formed when water makes up part of the crystal lattice formed by ions. Salt hydrates have definite, stoichiometric, compositions depending on the demands of the crystal structure, and at any temperature possess an equilibrium water vapor pressure. If the ambient water vapor pressure is less than this value, the crystal gives up water and decomposes to a powder. If it is higher, the crystal deliquesces and may dissolve in the condensed water. The crystals may be metastable, changing their water content only slowly when out of equilibrium, however.
Gas hydrates have a variable, nonstoichiometric composition depending on the conditions of formation. The gas content is usually not far from a maximum value, which is never exceeded. They are a type of substance called a clathrate by Powell in 1948. The word comes ultimately from kleiw, meaning to shut up or confine, and from Latin clatratus, latticed. In a clathrate, the host substance has voids, in which the guest substance is retained by steric (as opposed to chemical) forces. [Steric forces are the repulsions between the electron clouds of atoms and molecules.] Crystalline water is the host in a gas hydrate, while small molecules like Ar, Kr, Xe, CH4 CO2 H2S, N2 N2O, Cl2 Br2 C2H6 C2H4, C2H2, SO2, PH3, CH3Cl, C3H8, and others are the guests. Molecules that are too small, such as He, can escape through holes in the host molecule. Molecules that are too large to fit cannot form hydrates, of course.
What is the structure of the water host? Ordinary ice (hexagonal ice I and cubic ice Ic) are relatively open structures with voids, but the voids are not nearly large enough to hold the large variety of guest molecules that is observed. Two possible structures were proposed by W. F. Claussen in 1951. The first, now known as Structure II, was a cubic unit cell of about 17.4 Å on a side, containing 136 water molecules, and two kinds of voids, small ones formed by dodecahedra of water molecules, and large ones formed by a figure consisting of 4 hexagonal and 12 pentagonal faces. This is actually the more difficult structure to visualize, but Claussen found it first, regarding it as a slightly deformed dodecahedron of water molecules with two opposite vertices placed at the locations of the carbon atoms in the diamond lattice.
Claussen found the second structure, Structure I, a few months later. It contains 46 water molecules, arranged to form 2 small dodecahedral voids, and 6 medium voids formed by a figure consisting of two hexagonal and 12 pentagonal faces, with a cubic unit cell of 12.0 Å lattice constant. This structure consists of two interpenetrating simple cubic lattices of water dodecahedrons, the dodecahedrons on the two lattices differing in orientation about a 2-fold axis, together with 6 extra water molecules per unit cell that hold the dodecahedra together by tetrahedral hydrogen bonds. A methane molecule in a medium Structure I void is shown in the illustration.
H. R. Müller soon verified these structures by X-ray analysis of several gas hydrates in his 1951 thesis. If these structures existed without guest molecules, they would be new forms of ice, less dense than ordinary ice, and crystallizing in the cubic system. However, they are not stable relative to ordinary ice, and are not observed. The guest molecules offer energetically favorable van der Waals interactions with the water molecules surrounding them, and make possible the crystallization of these new forms. Considerable pressure may be required, however, to persuade the molecules into the voids.
All the bonds in gas hydrates are hydrogen bonds. Water's readiness to make these bonds is responsible for the stability of gas hydrates (as it is for the existence of liquid water, and ice). Linus Pauling suggested in 1959 that liquid water could be conceived as, in effect, water hydrate. Of course, this would not be a crystal, but a dynamically changing environment of partial dodecahedra and free water molecules. This model does, in fact, give a pretty good account of many water properties, but cannot be the whole story, since it does not explain the heat capacity or the behavior of non-polar solutes without some additional assumptions. The evanescent existence of dodecahedra in liquid water would seem to facilitate the formation of gas hydrates.
Hydrates are formed quite readily by Cl2 and H2S, even at pressures of 1 atm near 0 °C. Methane hydrate requires about 26 atm to form at 0 °C, and nitrogen hydrate an even higher pressure. Heavier hydrocarbon gases form hydrates at lower pressures than methane does. The small voids hold molecules up to about 5.2 Å diameter, medium 5.9 Å, and large 6.9 Å. Voids may be filled with different molecular species, and more or less completely.
A few gas hydrates have different structures than the common 12 and 17 Å ones. Some, like cyclopropane, can crystallize in either Structure I or II, depending on the temperature. Hydrates can form over ice as well as over water, at very low temperatures.
With the structure known, some of the characteristics of gas hydrates are easily explainable. For example, Cl2 is too big to fit in the small voids, but the medium voids in Structure I can easily accommodate it. Since there are 6 such voids in a unit cell, the stoichiometric formula of chlorine hydrate should be 6Cl2·46H2O, or Cl2·7.67H2O. The formula Cl2·8H2O was proposed by Rosenbaum in 1884, and has long been accepted. Cl2·7.3H2O is a recently-determined, more precise value, indicating that the medium voids are 95% filled. Unlike salt hydrates, gas hydrates do not have a definite composition.
The density of gas hydrates is easily found if it is known which structure they possess, I or II, since the lattice constants are always close to the same values, 12.0 or 17.4 Å. The average distance between gas molecules in a hydrate of Structure I is only 6.0 Å, which can be compared with the distance 33.4 Å between molecules in a gas at STP, and 3.1 Å between molecules in water. Therefore, when the hydrate is formed, the gas is condensed to a density nearly that of a liquid. Gases can be stored as hydrates without the necessity of cryogenic refrigeration or high pressure.
Let us investigate what happens when a hydrate decomposes in a limited volume. One mol of 8CH4·46H2O, methane hydrate, weighs 988 grams. The unit cell of Structure I has a volume of 12.03 cubic Å, so one mole occupies 6.022 x 1023 times as much, or 1040 cm3, roughly a litre. The density of the hydrate is then easily found to be about 0.95 g/cm3, between ice and water. When the hydrate decomposes, 46 moles of water, or 828 g, result, which occupy about 828 cm3, leaving only 212 cm3 for the methane. At STP, 8 mols of methane occupy 179 200 cm3. If methane were an ideal gas at all pressures, about 845 atmospheres, or 12 000 psi, would be necessary to confine it to the small volume available. These considerations sufficiently explain the explosive hazard of gas hydrates when confined at higher temperatures. This effect also offers an easy way of attaining high pressures without requiring mechanical compression. A few hundred psi suffices to form the hydrate at low temperature. When the hydrate is heated, its decomposition supplies the high pressure without further effort. It is considered foolhardy to hold a metastable hydrate in a tight container.
Hydrates formed in salt water take only the water, leaving the salt behind. This provides a convenient means for desalinization of sea water, in which the hydrate-forming gas is recycled. A patent for the process was granted as early as 1959. The energy and space requirements for this procedure are low, making it especially attractive for shipboard installations.
The anaesthetic gases chloroform, CHCl3, and xenon, Xe, are believed to work by forming gas clathrates in the brain, which gently interrupt the signals maintaining consciousness. It is remarkable that xenon, which can have no ordinary chemical effect whatsoever, is effective. These gases probably facilitate the creation of clathrates that include other, biologically active, molecules at the same time.
Gas hydrates form at temperatures above those of pure ice formation, which is an interesting point for discussion. Let us see if qualitative considerations can lead to an understanding of this. When ice forms from liquid water, the decrease in entropy is relatively small (0.292 Btu/lb-R) because water itself is fairly ordered, so the energy made available by the more efficient hydrogen bonding in ice relative to water is enough, when dissipated into the surroundings, to make the net entropy change positive, as it must be for a spontaneous process. At 4°C, methane forms hydrates above 551 psia. At 0°C, the required pressure falls to 370 psia.
When a gas hydrate is formed, there is a considerable reduction in entropy because the gas is, in effect, condensed into a small volume. This is offset by the availability of the kinetic energy of the gas molecules, and a contribution from the van der Waals attraction between host and guest, which can be turned into entropy in the surroundings (exothermic reaction). The net result, as observed, is that the solid can be formed at temperatures above those of the formation of pure ice. It is reasonable that it is more difficult for small molecules, such as N2 and CH4, to form hydrates because their van der Waals interaction is weaker than for larger or more polarizable molecules such as H2S or C2H6.
The gas molecules incident on the liquid water must also find partially-formed dodecahedral or larger sites, dissipate their kinetic energy by collision, and discover themselves trapped before they can again accumulate sufficient kinetic energy to escape. It might be assumed that this liberation of kinetic energy, aided to some degree by the van der Waals attraction, is what makes the formation of the hydrates possible.
The trapped gas molecules find themselves in a very small volume where they can bounce around. If they acted like a one-molecule ideal gas, their kinetic energies, by equipartition, would be similar to those in the free gas at the same temperature, and their pressures correspondingly high, which cannot be true, or there would be no way to offset the large decrease in entropy (the van der Waals contribution being assumed important, but not large enough). In the condensation of water vapor, there is the contribution of hydrogen bonding (explaining the anomalous existence of liquid water at fairly high temperatures), but it is still probably true that the kinetic energy of water molecules in water is less that that in the vapor at the same temperature.
Methane hydrates defeated an attempt to mitigate the giant Deepwater Horizon oil spill of 20 April 2010 in the Gulf of Mexico, when a production platform exploded due to an unexpected release of methane, and the blowout preventer failed to operate. The pressure at the ocean floor at 5000 ft depth is about 2230 psi and the temperature not far above 0°C, ideal for the formation of hydrates. When a large dome was placed over the well, it rapidly filled with slushy methane hydrate that rendered it buoyant and blocked the outlets, defeating the attempt to collect the leaking oil.
Another example of a clathrate from a totally different field is presented by "filled" skutterudites. Skutterudite is a mineral named after Skutterud, Norway, where good crystals are found. It is a hard, heavy cubic crystal composed of cobalt, nickel and arsenic, typically (Co,Ni)4As12. The tin-white to grey metallic-looking crystals are found in cubic, octahedral, dodecahedral and pyritohedral habits. In the skutterudite we shall discuss, the arsenic is replaced by antimony. In this material, dodecahedral structures of 8 Co and 12 Sb atoms are formed that can clathrate additional atoms. In particular, thallium can be clathrated to give a material of composition Tl0.8Co4Sb11Sn. Some tin replaces antimony for electrical neutrality as Tl+ are added. This is called a "filled" skutterudite, and the Tl ions rattle around in the clathrate cage.
The interest in these materials comes from a search for thermoelectric materials that have a high electrical conductivity but a low thermal conductivity. Since high thermal conductivity usually accompanies high electrical conductivity, the search is not a trivial one. A low thermal conductivity would contribute to a better thermal efficiency of thermocouples, which typically have low efficiency. In the filled skutterudite, the rattling thallium ions reduce the phonon mean free path to roughly the distance between thallium ions, from its larger value in unfilled skutterudites. This reduction in mean free path reduces the thermal conductivity more than it is increased by the addition of the thallium. Electrons are not affected by the replacement, so the electrical conductivity is not changed. For details, see the Reference.
Composed by J. B. Calvert
Put in HTML 19 November 2000
Last revised 27 May 2010