Under appropriate conditions of temperature and pressure, gas hydrates usually form one of two basic crystal structures known as Structure I and Structure II (Fig. 1; Structure II not shown). Each unit cell of Structure I gas hydrate consists of 46 water molecules that form two small dodecahedral cages and six large tetradecahedral cages, leading to the ideal formula of eight gas molecules to 46 water molecules. Structure I gas hydrates can only hold small gas molecules such as methane and ethane, with molecular diameters not exceeding 5.2 Å. The unit cell of Structure II gas hydrate consists of 16 small dodecahedral and eight large hexakaidecahedral cages formed by 136 water molecules with an ideal formula of 24 gas molecules to 136 water molecules. Structure II gas hydrates can contain gases with molecular dimensions in the range of 5.9 to 6.9 Å, such as propane and isobutane.
Structure I gas hydrates appear to be the most common hydrate structure occurring in nature (reviewed by Kvenvolden, 1993, Booth et al., 1996, Sloan, 1997, and Makogon, 1997). The ideal water/gas molar ratio (n, also called the hydrate number) of Structure I gas hydrate is 46/8, or n = 5.75, whereas the ideal water/gas molar ratio of Structure II gas hydrate is 136/24, or n = 5.67. These ideal water/gas molar ratios confirm the observation that gas hydrates contain a substantial volume of gas. For example, if all the cages of Structure I gas hydrate are occupied, each volume of gas hydrate at Blake Ridge in situ equilibrium pressures (between about 22 and 32 MPa) will contain about 173 volumes of gas when converted to STP.
Most researchers believe that a completely filled clathrate is not likely to be found in nature (reviewed by Sloan, 1997). However, gas hydrates are not stable if the clathrate cages are less then 70% occupied by gas (Holder and Hand, 1982). For the purpose of this study, we have assumed a minimum threshold hydrate gas yield of 139 m3 gas/m3 gas hydrate (70% cage occupation, 1 MPa) and a maximum hydrate gas yield of 159 m3 gas/m3 gas hydrate (94% cage occupation, 1 MPa). This range represents the likely minimum and maximum values that occur in nature.
Our direct measurements of gas hydrate are made of the volume of gas released to the volume of water from dissociated gas hydrate, a measurement that is independent of considerations of gas hydrate formation equilibrium pressure and density. The theoretical upper limit of this ratio is 216 at STP as calculated from the ideal formula for Structure I methane hydrate of C1 · n H2O where n = 5.75 for gas hydrate with 100% cage occupancy. There is reason to believe that the molar ratio in nature is greater than the ideal 5.75. Holder and Hand (1982) calculated the likely molar ratio of 6.37 (90% cage occupancy) based on considerations of enthalpy for methane in Structure I hydrate in oceanic sediments. Furthermore, a molar ratio of 5.75 is predicated on pressures exceeding the equivalent water depths of 7000 m. If one takes n = 6.37 to be the likely molar ratio for natural gas hydrate, then the upper limit of the gas to water volumetric ratio from dissociated gas hydrate likely in nature is 195. Makogan (1997) states that the likely molar ratio for natural gas hydrate should range from 5.8 to 6.0, and thus gas to water volumetric ratios of 215 and 207, respectively. If one considers the variation of n with pressure (Saito et al., 1964) at in situ pressures of the Blake Ridge study area (from about 22 to 32 MPa), then n = 6.1 with a resultant gas to water volumetric ratio of 204, or 94% cage occupancy.
At the lower limit of gas hydrate cage occupation, 70% of the gas hydrate cages must be occupied for a stable gas hydrate structure to exist. This corresponds to n = 8.2 or a gas to water volumetric ratio of 152. Thus the likely range for natural gas hydrate lie between n = 6.1 to 6.8 and gas to water ratios of 204 to 152, respectively.
Table 3 lists a range of expected values for the volumetric ratio of gas to water, gas to gas hydrate, n, cages occupied, and gas hydrate density at pressure conditions expected at the Blake Ridge (28 Mpa, ave.), then compared gas hydrate pressure conditions at sea level (about 1 Mpa).
Blake Ridge gas hydrate samples preserved in liquid nitrogen and analyzed by X-ray powder diffraction in the laboratories of the National Research Council of Canada revealed that gas hydrate from the Blake Ridge is solely Structure I (J. Ripmeester, pers. comm., 1996). A total of five gas hydrate samples were analyzed from Sites 994, 995, and 997 at depths spanning the range of all gas hydrate recoveries. The confirmation of Structure I gas hydrate eliminates the need to consider Structure II gas hydrate at the Blake Ridge.
A limited number of gas to water volumetric ratios have been made of gas hydrate from the continental margin of North and South America (Fig. 1; Table 2). The gas to water volumetric ratios, ranging from 4 to 177, are similar to the range presently observed on the Blake Ridge (29-154). There is no apparent geographic pattern to the ratios; rather we believe that all of these ratios are minimum values resulting from partial gas hydrate dissociation during recovery. The amount of dissociation during recovery is strongly dependent on the magnitude and time the sample encounters conditions outside of the gas hydrate pressure-temperature (P-T) stability field. These parameters are in turn affected by the water temperature through which the core is recovered and the core recovery transit time. Other factors include time and temperature on deck and so on—thus the amount of sample dissociation can be quite variable and location dependent (e.g., hot climate vs. cold climate).
Gas hydrate excludes all ions upon formation, thus the chlorinity of gas hydrate water reflects the amount of pore water that accompanies gas hydrate in our measurements. Therefore, pore-water contamination can be estimated by the observed chlorinity concentration in dissociated gas hydrate residual water. Chlorinity concentrations observed from dissociated gas hydrate water vary from 352 to 167 mM. When compared to nearby Cl- concentrations from sediment pore water (ranging from 520 to 997 mM; Shipboard Scientific Party, 1996b, 1996d, 1996e), the amount of pore-water contamination can be calculated by the percentage excess chlorinity in gas hydrate water. In cases of slight dilution of pore waters by dissociated gas hydrate, a value was chosen that corresponded to a chlorinity trend given by undiluted pore-water samples. The results show that the contamination ranges from 2% to 50%. If these results are applied to the gas to water volumetric ratios released by dissociated gas hydrate seen in Table 3, the chlorinity-corrected gas to water volumetric ratios vary from 29 to 204. Thus gas hydrate samples from Leg 164 may be fully saturated with gas. It should also be noted that there is a correlation of the chlorinity-corrected gas to water volumetric ratios with depth; that is, the deeper samples are more saturated with gas than the shallow gas hydrate samples. Only the deepest gas hydrate sample (Section 164-997A-42X-3, 331 mbsf) is fully saturated with methane. Whether this is a drilling and core-recovery artifact or a geological curing process is unclear.
It appears that disseminated gas hydrate cannot be recovered with conventional coring techniques, and therefore measurements have been made only on massive or vein-like gas hydrate. The geochemistry of disseminated or massive gas hydrates are likely similar in gas composition and concentration, because we observed no change in sediment gas composition in areas where disseminated gas hydrate occurs (Shipboard Scientific Party, 1996b, 1996c, 1996d, 1996e). Dissociated gas from disseminated gas hydrate would likely result in perceptible shifts in sediment gas geochemistry because C3+ gases in sediment should be diluted by methane from dissociated gas hydrate.
The distribution of hydrocarbon gases released from dissociated gas hydrates is compared to the distribution of hydrocarbon gases expelled from sediment into voids in the core liner (free gas). We make the assumption that gas incorporated into gas hydrate is derived from nearby sediments, unless there is evidence of active gas seepage from depth such as at Site 996. The concentration of gases can be normalized to 100% of C1, CO2, and total gaseous hydrocarbons (Table 1). Oxygen and nitrogen were eliminated from the calculation because they are considered contaminants from the atmosphere. In Table 1, the gas from gas hydrates is compared with sediment gas obtained close to where the gas hydrates were found. The minimum and maximum distance between sediment gas sample points and those where gas hydrate was recovered was 0.3 and 11 m, respectively. At Site 994 sediment gas released over a period of hours from whole 1.5-m core sections is used for isotopic composition of methane comparison only.
Higher C2 concentrations in gas hydrates than in core gas possibly results from the preferential inclusion of C2 in Structure I gas hydrate, where C2 acts to stabilize the forming crystal by filling the larger cages first (Sloan, 1997). In addition, higher molecular weight hydrocarbon gases (C3 to Cn, where n > 3) are excluded from Structure I hydrate. Thus C2 is expected to be more concentrated in Structure I hydrate, whereas C3+ (C3, i-C4, and n-C4) is excluded. Following this line of reasoning, higher concentrations of C2 should be found in gas hydrate relative to C2 concentrations in nearby sediment (Hand et al., 1974). In addition, C3+ should be more concentrated in surrounding sediment relative to C3+ concentrations in gas hydrate. This relationship has been observed in the Middle America (Kvenvolden et al., 1984,) and in the Gulf of Mexico (Brooks et al., 1989).
The gas analyses shown in Table 1 are less conclusive, yet favor the above supposition. In five of eight cases, gas hydrate compositions show some, although minor, enrichment of C2, whereas in two cases C2 concentrations are nearly equal, and one case the opposite is observed. In contrast, the exclusion of C3+ from gas hydrate is evident. In every case, the sum of C3+ gases from dissociated gas hydrate is less than nearby sediment gases. The presence of C3+ gases in Structure I gas hydrate is enigmatic because they are larger than the size of any cage and is attributed to contamination by minor volumes of sediment and dissolved pore-water gases that cannot be removed from gas hydrate used for analyses.
Comparison of sediment gas collected in closest proximity to gas hydrate occurrences (0.3 and 0.9 m in Sections 164-996A-8H-4 and 164-996D-6H-5, respectively) illustrate these relationships. For example, gas hydrate gas from Hole 996A is enriched in ethane by 9.8% whereas nearby sediment gas is enriched in C3+ by 63% relative to gas hydrate gas. These relationships can become less distinct or even contrary when comparing samples from greater distances apart. Nonetheless, when a group of comparative samples are analyzed with the simple statistical parameters as done in Table 4 and Figure 4, these observations are validated. For example, the average and median C2 concentration for gas hydrate gases are 694 and 750, respectively, whereas those of sediment gases are 445 and 629, or a median enrichment of C2 in gas hydrate gas of 16%. The same comparison made for C3+ gases yields a median enrichment of C3+ gases in sediment of 62%. Because of the small number of samples used for comparison, we have chosen to use the median rather than the average, however the conclusions are the same. Comparisons of C1 and CO2 median values in Table 4 and Figure 4 reveal that gas hydrate gas appears to be enriched in C1 and depleted in CO2 relative to sediment gas. This result is somewhat surprising because CO2 fits snugly into the larger Structure I hydrate cage and therefore is not expected to be selectively excluded. We suggest that CO2 gas concentration in sediment gas is artificially enriched in gaseous CO2 during core recovery. Most or all of CO2 at in situ pressure and temperature is in the form of dissolved bicarbonate ion (Paull et al., Chap. 7, this volume). Exsolution of CO2 resulting from the depressurization of the core during recovery also artificially dilutes methane concentration in sediment gas samples, and for this reason, a meaningful comparison is difficult.
Comparison of median
isotopic compositions of methane carbon and hydrogen are listed in Table
4. Median methane isotopic carbon and hydrogen of sediment gas is -65.0
and -195.8,
respectively (one sample for methane hydrogen isotopic composition), and that of
gas hydrate is -67.0
and -194.6,
respectively. The differences between the median methane carbon and hydrogen
isotopic compositions of sediment gas and gas hydrate are small and the very
limited number of samples demands prudence; thus we observe no obvious
fractionation.
In contrast, the CO2
carbon isotopic median composition is depleted in gas hydrate gas (-21.9)
relative to sediment gas CO2 (-6.5)
likely derived from nearby DIC (+6.6;
Paull et al., Chap. 7,
this volume). Thus there are significant differences between DIC, sediment gas
CO2, and CO2 gas incorporated into gas hydrate.
Combined 
13C
C1 and 13C
CO2 profiles of gas and gas hydrate from Sites 994, 995, 996, and 997
are seen in Figure 5.
Measurements from sediment gas, gas tubes, and gas from the pressure coring
system are combined in Figure 4
and show little variation (Paull et al., Chap.
7, this volume). 13C
C1 of gas hydrate gas closely resembles sediment gas at all sites and
at all depths. However 13C
CO2 of gas hydrate is depleted of 13C by about
15 from the
corresponding sediment gas 13C
CO2. This relationship is best illustrated by gas hydrate samples
from Site 996 and a single gas hydrate sample from Site 997. At Site 994, gas
hydrate 13C
CO2 is nearly that of the sediment gas and does not appear depleted.
Isotopic fractionation of CO2 between gas hydrate and sediment gas
has never been reported to our knowledge. However Finley and Krason (1986), in
their report on Site 570 from the Middle America Trench, plot the carbon
isotopic composition of C1, CO2, and total CO2
(DIC) from cores with depth. A shift of about 15-20
between the trend of CO2 and DIC with depth occurs within the hydrate
zone. Brooks et al. (1989) noted differences in 13C
CO2 of gas hydrate and 13C
of associated authogenic carbonate extracted from sediment. In two cases the 13C
CO2 of gas hydrate was heavier by 2.8-16.5
than that of nearby sedimentary carbonate. Unfortunately no analyses of sediment
13C
CO2 was made for direct comparison to our results. To explore these
observations we have attempted to constrain the fractionation processes for CO2
in these sediments.
Paull et al. (Chap.
7, this volume) discuss the offset between 13C
values of CO2 gas and DIC. They note a significant isotopic offset
(~12.5)
between the 13C
values of CO2 gas and DIC. A similar offset (~10)
was reported between these carbon pools at DSDP Site 533 (Claypool and Threlkeld,
1983). The offset is presumed to be an artifact of CO2 outgassing
during sediment recovery. Equilibrium fractionation of 8.38 ± 0.12
(at 20ºC; Emrich et al., 1970) occurs between DIC and gaseous CO2.
C1 is
supersaturated in sediments adjacent to gas hydrate, and the majority (likely
~99%) of the original C1 in the sediments below ~100 mbsf is lost
during the core-recovery process (Dickens et al., 1997; Kvenvolden and Lorenson,
Chap. 3, this
volume). Paull et al., (Chap.
7, this volume) infer that part of the DIC pool is sparged by degassing
C1, and thus much of the original in situ DIC pool vents as CO2.
Thus, the measured pore-water DIC samples reflect the residual DIC that remains
in the pore water after vigorous degassing and fractionation. 13C
values of the DIC probably lie in between the measured CO2 gas and
DIC pools, but closer to the CO2 gas values because these samples may
reflect the majority of the original DIC input.
If 13C
DIC values in sediment adjacent to gas hydrate are nearly that of sediment gas
CO2, then the same fractionation process that occurs between DIC and
sediment gas CO2 could possibly account for most of the 15
isotopic shift between sediment gas CO2 and gas hydrate CO2.
In such a scenario, a phase change of the DIC from ion to gas before inclusion
in the hydrate is required for fractionation to occur. Thus gas hydrate would
appear to form most efficiently when sediment gas is present rather than from
water saturated with a particular gas.
