IODP Proceedings    Volume contents     Search
iodp logo


Results and discussion

Site U1328: Sample 311-U1328B-2H-HydCC (Hester, 2007)

The first gas hydrate core (Sample 311-U1328B-2H-HydCC, designated hereafter as Sample H1) was cored at IODP Hole U1328B in a water depth of 1267.8 m. At this location, widespread carbonate deposits were observed on the seafloor (see the “Site U1328” chapter). The core, 1.7 m in length, was taken using a piston core, a conventional nonpressurized core, at a top depth of 4.5 meters below seafloor (mbsf) in Hole U1328B. Soupy sediment, likely because of gas hydrate dissociation during recovery, was observed upon recovery (see the “Site U1328” chapter). The core was recovered as pieces of sediment, mainly dark gray to greenish gray clay containing gas hydrate nodules (on the order of centimeters in size). Samples of the void headspace gas from the core were found to be primarily methane with 23,470 parts per million (ppm) H2S and 4,375 ppm CO2 (see the “Site U1328” chapter). With the bottom water temperature around 276.65 K and a local geothermal gradient of 53.6 ± 0.4 K/km, the in situ gas hydrate conditions for Sample H1 were around 276.9 K and 12.87 MPa. The approximate interstitial water salinity was determined to be 33.5.

For Raman analysis, six separate gas hydrate pieces (all on the order of 1 cm3) were chosen from Sample H1. Of each of these gas hydrate pieces, between two and four different spots were measured, resulting in 17 total Raman spectra collected. All spectra were collected for 50 s each. Figure F1 shows the spectral regions of interest. Only methane in the sI lattice was detected in all measurements of Sample H1. The main ν1peaks for methane are present at 2900 cm–1 (CH4in the sI 51262) and 2912 cm–1 (CH4in the sI 512) with two minor peaks at 2570 and 3045 cm–1, indicating methane in a gas hydrate phase (Sum et al., 1997; Hester et al., 2007). It should be noted that although the ν1 peaks are shifted to lower wavenumbers by 3–4 cm–1 and the 3045 peak downshifted 9 cm–1 at 77 K versus 298 K, there is no apparent temperature-induced shift of the 2570 cm–1 peak.

Using the peaks for methane at 2900 and 2912 cm–1, the relative composition in the gas hydrate can be determined. The methane occupancy ratio was found to be consistent throughout the Sample H1 core. Over 17 measurements, an average occupancy ratio was found to be 1.13 ± 0.04.

NMR spectroscopy on Sample H1 was also performed. Being a bulk technique versus the local nature of the Raman measurements, only one sample was measured (Fig. F2). As with the Raman work, only peaks for methane in the sI were measured at chemical shifts of –6.2 ppm (CH4 in the sI 51262) and –4.1 ppm (CH4 in the sI 512) (Ripmeester and Ratcliffe, 1988). The methane occupancy ratio of 1.13 ± 0.02 was in excellent agreement with the Raman work.

Site U1328: Sample 311-U1328E-2X-Hyd17

The second gas hydrate core (Sample 311-U1328E-2X-Hyd17, designated hereafter as Sample H2) was cored at IODP Hole U1328E in a water depth of 1264.7 m (see the “Site U1328” chapter). The core, 2.39 m in length, was taken using the extended core barrel at a top depth of 6.5 mbsf. Soupy sediments were also observed in this core along with a medium temperature anomaly, both indicating gas hydrate dissociation (see the “Site U1328” chapter). The core was recovered as a whole-round core with mainly dark gray to greenish gray clay sediment containing gas hydrate nodules (on the order of centimeters in size). The presence of H2S was also detected upon the core recovery in the headspace gas using gas chromatography (see the “Site U1328” chapter). With the bottom water temperature at ~276.65 K and a local geothermal gradient of 53.6 ± 0.4 K/km, the in situ gas hydrate conditions were around 277.0 K and 12.83 MPa. Salinity was not determined for this core; however, a core in close vicinity (Section 311-U1328E-2X-1) at a depth of 7.4 mbsf was determined to have an interstitial water salinity of 34.5.

Six gas hydrate pieces from Sample H2 were chosen for Raman spectroscopic analysis. As seen in Figure F3, along with peaks for sI CH4similar to Sample H1, peaks for H2S at 2593 cm–1 (H2S in the sI 51262) and 2605 cm–1 (H2S in the sI 512) were detected (Dubessy et al., 1992). Small quantities of H2S were present in every measurement of this methane-rich sI gas hydrate.

Compositional heterogeneity in Sample H2 was seen by slight variations in relative composition of CH4to H2S. Although direct quantification of Raman gas hydrate peaks is not possible, the composition of hydrogen sulfide in the gas hydrate can be obtained, as a first approximation, correcting the peaks using relative normalized differential Raman scattering (RNDRS) cross sections available in the literature (Schrotter and Klockner, 1979). The RNDRS cross section for the ν1peak of CH4 is 8.55 and 7 for the ν1 peak of H2S. The peak areas are corrected by dividing the measured peak area with the appropriate molecule and vibrational mode specific RNDRS cross section. With these corrections, the mole percent of H2S in the 12 measurements varied from 1.94% to 2.47%.

The methane occupancy ratio was also measured using both Raman and NMR spectroscopy. The average occupancy ratio from the Raman measurements was 1.16 ± 0.02. Excellent agreement was found with the NMR-determined occupancy ratio of 1.18 ± 0.02. Previous work has shown methane occupancy ratios were quantitative using Raman for pure methane gas hydrate but only qualitative when a second guest was present (Subramanian, 2002; Wilson et al., 2002). However, in the previous studies, the second guest was in much greater concentration than the few percent here. In the present work, the results show that small amounts of other guests still allow for quantification of CH4by Raman spectroscopy. The average hydrogen sulfide relative cage occupancy was found from Raman spectroscopy to be 0.49 ± 0.03.

Cage occupancy and hydration number

Estimates of the absolute cage occupancies and hydration number from the experimentally determined occupancy ratios can be made (Ripmeester and Ratcliffe, 1988; Sum et al., 1997; Uchida et al., 1999) using the gas hydrate statistical mechanics model originally developed by van der Waals and Platteeuw (1959). The statistical mechanics model was derived assuming gas hydrates were an ideal solid solution, in which guest occupancy of the cages lowers the gas hydrate chemical potential. If the water lattice is not distorted and guest-guest interactions are negligible, the chemical potential of the sI gas hydrate (ΔμwH) can be given as



  • θL,i= occupancy of the large cage by component (i),
  • θS,i= occupancy of the small cage by component (i), and
  • Δμwβ = chemical potential of the hypothetical empty lattice. A generally accepted value for Δμwβ is 1297 J/mol (Handa and Tse, 1986).

By combining the experimentally determined occupancy ratios and Equation 1, it is possible to solve for absolute cage occupancies. These can be used to solve for the hydration number, n, such that


which gives the ratio of water molecules to guest molecules in the gas hydrate. These experimentally determined cage occupancies and hydration numbers can then be compared with predictions from gas hydrate prediction programs, in this case CSMGem, which incorporates Gibbs energy minimization to eliminate the assumptions in the original van der Waals and Platteeuw model (Ballard and Sloan, 2002).

To solve for occupancies in Equation 1, quantitative experimental occupancy ratios are needed. Although NMR is inherently quantitative, Raman intensities rely on polarizability theory with parameters that are not trivial to determine (Placzek, 1934), leading to a more qualitative nature. However, cross-calibration with NMR has shown quantitative Raman methane cage occupancy ratios for pure methane gas hydrate (Subramanian, 2002; Wilson et al., 2002; Hester, 2007) and, in this study, with small amounts of H2S present.

For Sample H1, only methane was detected. The experimentally determined occupancy ratio from Raman was 1.13 ± 0.04, in agreement with NMR (1.13 ± 0.02). The 51262 was almost fully occupied (0.974) and the 512 cage had an occupancy of 0.862; this resulted in a hydration number of 6.08 ± 0.04. Using the in situ conditions (276.9 K, 12.78 MPa, S = 33.5), the results of a flash calculation performed with CSMGem yields large and small cage occupancies of 0.961 and 0.873, respectively, and a hydration number of 6.13; these values are in good agreement with the experiment.

For Sample H2, both methane and hydrogen sulfide were present. To experimentally determine absolute cage occupancies and hydration numbers, the assumption that RNDRS cross sections can be used to estimate the gas hydrate composition was used. The mole fraction H2S determined using this method for the six samples measured and absolute cage occupancies and hydration numbers calculated using Equations 1 and 2 are listed in Table T1. CSMGem calculations using the in situ conditions (277.0 K, 12.83 MPa, S = 34.5) are also summarized in Table T1. Even with the assumptions made, both experimentally determined occupancies and the calculations agreed well. It should be noted that H2S has a greater occupancy in the 512 cage (2.9%–3.8%) versus the 51262 (1.5%–1.9%), both experimentally and calculated, possibly because of its greater stabilizing effect of the 512 cage over methane.

For Samples H1 and H2, the average hydration number was 6.08 ± 0.04 and 6.08 ± 0.02, respectively. Recent work on recovered northern Cascadia margin gas hydrates (Lu et al., 2005) yielded a hydration number of 6.1 ± 0.1, very similar to the results of this work. Laboratory work using the combination of spectroscopic occupancy ratios and statistical mechanics have obtained values including 6.05 ± 0.06 (Ripmeester and Ratcliffe, 1988), 6.04 ± 0.03 (Sum et al., 1997), and 6.2 ± 0.3 (Uchida et al., 1999) for synthetic and natural methane gas hydrates. Careful direct laboratory hydration number measurements of synthetic methane gas hydrate have shown values of 5.99 ± 0.07 (Circone et al., 2005) and 6.00 ± 0.01 (Handa, 1986). The results from the synthetic samples are in good agreement with the recovered natural samples.

With known formation conditions, natural methane gas hydrates are structural and compositional analogs to laboratory samples as expected from thermodynamics. In addition, current statistical mechanics methods are able to predict the composition of these gas hydrates very well. However, multicomponent natural gas hydrates, such as those from thermogenic sources, still pose challenges, such as sample heterogeneity with mixed structures and varying compositions, and warrant future study.

Presence of H2S

It was indicated by the shipboard party that H2S was detected in association with Samples H1 and H2. The spectroscopic measurements on the samples show that only Sample H2 contained H2S. It was noted that in the upper part of Hole U1328B (from which Sample H1 was recovered) dangerous levels of H2S were detected (see the “Site U1328” chapter). Because some sample dissociation occurred during recovery, this work is inconclusive as to whether H2S was present in situ in the gas hydrate from Sample H1.

Estimates of sample degradation due to recovery

Because these gas hydrate samples were from areas where temperatures would not allow for water ice, measurement of ice in quenched samples under liquid nitrogen can be attributed to the dissociation of gas hydrate or other sources, such as surrounding pore waters. Indication of gas hydrate dissociation was present with observation of a “soupy” texture in the recovered core. Raman measurements of the water phase could possibly be used to distinguish between gas hydrate and ice phases and assess sample degradation.

Figure F4 shows the Raman spectra for water in a synthetic sI gas hydrate and the ice phase at 77 K. The most intense O-H stretching peak for water is found at 3076 cm–1 for sI gas hydrate, whereas it is shifted 8 cm–1 to 3084 cm–1 for ice. A smaller shift between sI gas hydrate and ice was observed for the weaker water peak (3207 versus 3210 cm–1). For further analysis, the peak at 3076 cm–1 for sI and 3084 cm–1 for ice were used because they showed the greatest shift between phases. The shift of this peak can be used to determine if ice is present along with gas hydrate. When comparing the area of the C-H methane peaks to the O-H water peak, Figure F5A shows that the ratio of water to methane can vary significantly. The most probable explanations could be either (1) the water peak was a mixture of sI and ice (caused by degradation during recovery and/or frozen pore water during quenching) or (2) the hydration number (ratio of water to gas hydrate guest) of the gas hydrate was variable. Figure F5B shows that for the gas hydrate spectra with a greater water-to-methane ratio, the O-H stretching peak actually shifted to 3078 cm–1. This indicates that a mixture of sI and ice was the cause of the variation in the water-to-methane ratio, not a significant difference in the hydration number.

For these measurements at 77 K, the most intense Raman peak for pure ice was at 3084 cm–1 with a full width at half maximum (FWHM) of 26.5 cm–1 and 3076 cm–1with a FHWM of 29.8 cm–1 for pure sI methane gas hydrate. Both peaks were best fit with a Lorentzian peak shape. In order to estimate the fraction of ice present, two Lorentzian peaks with the Raman peak characteristics (peak position and FWHM) of ice and the sI methane gas hydrate, respectively, were calculated and their relative intensities were varied. Summing the intensities of these two peaks over 3000–3150 cm–1 resulted in a single peak with a Raman shift between 3076 and 3084 cm–1, based on the relative peak intensities. If the RNDRS cross sections for water as ice and sI gas hydrate were assumed to be similar, this resulted in a correlation between the Raman shift of the water peak and the mole percent sI gas hydrate measured as shown in Figure F6.

To evaluate this approach of using the Raman shift to determine the amount of sI gas hydrate measured, the water peak in spectra of the IODP gas hydrates was fit between 3000 and 3150 cm–1 with two Lorentzian peaks constraining the wavenumber and FWHM width based on pure sI methane gas hydrate and ice. The resulting peak areas were then used to estimate the molar percentage of gas hydrate measured. As shown in Figure F6, good agreement was obtained with the two-Lorentzian technique described above. This approach allows for a very straightforward estimate of the sample degradation for recovered sI methane gas hydrate. In addition, this approach could be applied to laboratory samples for a rapid estimate of gas hydrate conversion. However, in the natural samples, it needs to be recognized that the water ice present in the quenched samples could have come from other sources, such as frozen pore waters.