We report the 3D microstructure analyses of natural gas hydrates sampled from Gulf of Mexico. The samples were characterized by synchrotron radiation X-ray cryo-tomographic microscopy (SRXCTM) using the ‘TOMCAT’ beam line at the Swiss Light Source (SLS). The SRXCTM demonstrates its applicability to unlock some microscopic features of the marine hydrates, in particular of crystallite size and grain boundary network. The gas hydrate domains are surrounded by a network of pores of typically a few micrometers, which are largely due to decomposition. Out of the SRXCTM data, the porosity, total volume of the voids, the void surface area and number of the total gas-filled voids have been calculated. The results reveal the capability of SRXCTM to access the 3D microstructure which is of fundamental importance to model the petrophysical properties of natural gas hydrates.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Clathrate hydrates are non-stoichiometric crystalline inclusion compounds built of a framework of hydrogen bonded water molecules that offers suitable cavities occupied by guest molecules such as methane and ethane [Sloan and Koh, 2008]. Hydrates of hydrocarbons are stable in a wide range of oceanic and permafrost environments [Max et al., 2006; Sloan and Koh, 2008]. Gas hydrate deposits in sediments have been reported from more than 100 locations worldwide [Kvenvolden and Lorenson, 2001]. The deposits are estimated to contain hydrocarbons (mainly methane), and thus pose a potential jeopardy to climate, as methane is a greenhouse gas. But also, the recovery even of a fraction of the estimated amount would provide a substantial energy resource. Besides, gas hydrates are of major concern to the oil and gas industries for their ability to block the pipelines [Sloan and Koh, 2008].
 Natural gas hydrates have been recovered from various deep-sea sites during the Ocean Drilling Program (ODP) and the Integrated Ocean Drilling Program (IODP) campaigns [Tréhu et al., 2003], as well as from continental drillings [Dallimore and Collett, 2005] and individual ship expeditions [Bohrmann et al., 2008]. However, our overall knowledge of some fundamental properties of marine gas hydrates is surprisingly limited. One reason is the precarious preservation of the retrieved samples and the difficulties of characterization in situ. If recovered without pressure maintaining devices, gas hydrate will start decomposing [Heeschen et al., 2007]. It is therefore important to sub-sample gas hydrate pieces from the inner core of a larger hydrate lump, which seems to be pristine and unaffected from decomposition. A good sample preservation is essential to study the concentration, texture, and geometry of voids (probably filled with gases) or to explore the crystallite size, the size distribution and grain boundary topology as well as the distribution and nature of embedded or surrounding minerals. An inspection of the pore structure by image analysis may provide a substantial appreciation of the role of porosity on the macroscopic properties of gas hydrates relevant to geophysical and engineering concerns. Despite the importance of these issues for a complete petrophysical investigation, means of characterization are limited in particular for the three-dimensional visualization of the microstructure. Many insights were achieved by cryo Scanning Electron Microscopy (cryo-SEM) [Staykova et al., 2003; Kuhs et al., 2004a; Stern et al., 2004; Bohrmann et al., 2007], which allowed a detailed 2D inspection down to a few micrometers. However, the possibilities of SEMs are limited in identifying the size and shape of crystallites [Klapp et al., 2007], in particular when gas hydrates are seen in their sub-μm porous form [Kuhs et al., 2000]. Moreover, the 2D imaging techniques cannot constrain higher-order geometrical attributes such as the connectivity of pore-networks [Higgins, 2000]. X-ray computerized tomography (X-ray CT) overcomes the problem providing full volumetric (3D) information. The X-ray attenuation contrast enables even to distinguish crystallites and establish their spatial distribution in the samples as shown below.
 X-ray CT is a non-destructive and non-invasive tool providing good spatial and time resolution to visualize hydrate formation and decomposition [Mikami et al., 2000; Gupta et al., 2006; Kneafsey et al., 2007]. The number of studies on gas hydrates by X-ray CT is, however, small using tube-type X-ray sources with limited resolution (e.g., 350 μm [Mikami et al., 2000] and 2000 μm [Uchida et al., 2000] pixel size). In contrast, synchrotron based X-ray tomographic microscopy at TOMCAT offers pixel sizes down to a few hundred nanometers (e.g., 370 nm–6 μm) along with cryo-facilities. In this paper we show that synchrotron radiation X-ray cryo-tomographic microscopy (SRXCTM) is a well-suited tool to analyze natural gas hydrate samples collected from shallow-buried sediments of the Gulf of Mexico with a significantly improved resolution over conventional X-ray CT. To the best of our knowledge, this is the first tomographic report of marine gas hydrates analyzed using a synchrotron microtomography beamline; the pixel size was set here to 1.4 μm. The obtained SRXCTM images demonstrate the feasibility of the method for a variety of missing information such as internal surface area, the individual volume of each inclusion and the morphology of the voids of natural gases in a specified volume. Likewise, it can give access to the crystallite shapes, crystallite size distribution and grain boundary network topology, quantities which are of prime importance for understanding any coarsening processes taking place in gas hydrates as a function of time.
2. Sampling and Methods
2.1. Hydrate Retrieval and Preservation
 The sample investigated in this study was retrieved during the R/V Meteor cruise M67 at the Chapopote Asphalt volcano [Bohrmann et al., 2008] in the Gulf of Mexico by conventional gravity coring. The gas hydrates were found in the upper first meter of Core GeoB 10618. Gas composition analysis of the hydrate revealed methane by far as the dominant gas suggesting structure I type hydrate (sI), which was confirmed by X-ray powder diffraction (unpublished data). The gravity core was recovered at latitude 21°53.95′N and longitude 93°26.21′E in 2903 m water depth, where the water temperature was 4.4°C. Preservation pressure was not maintained during the ascent of the corer; therefore the gas hydrate partially decomposed. As a consequence, the samples contain ice from the decomposition of gas hydrate as well as frozen interstitial water. To avoid further decomposition the samples were stored in liquid nitrogen (LN2) soon after the recovery and, all subsequent transfers and investigations were performed at temperatures of hydrate stability conditions. Under LN2 condition, a lump (∼1 cm3) of the samples was crushed into sub-samples (∼mm), and two pieces of the sub-samples (∼2 mm) were put into the sample holder.
2.2. Experimental Setup
 A detailed description of the polyamide-cup sample holder and the setup of SRXCTM compatible to the beamline station is available elsewhere [Miedaner, 2007]. In short, cold nitrogen gas was flowed around a sample chamber made of plexiglass. A PT-100 resistor element (2 × 2.3 × 0.9 mm, Greisinger Electronics) was placed just below the sample chamber to record the actual temperature. For cooling, a flow of cold nitrogen gas (153 K) was directed from the standard nozzle of a CryojetXL (Oxford Instruments) onto the top of the sample holder. An additional flow (2 L/min) of dry shielding nitrogen protected the cryo-jet from icing due to water condensation inside the nozzle and the sample from ice formation. The position of the cryojet was readjusted to the rotational axis of the sample holder until no temperature fluctuation was detectable during a 180° rotation.
2.3. Data Acquisition and Processing
 SRXCTM data have been acquired at the TOMCAT beam line of the SLS [Stampanoni et al., 2006]. The monochromatic X-ray beam (ΔE/E = 2.5% at 10 keV, [Ru/C]100 multilayer monochromator) was tailored by a slit system to a profile of 1.4 mm2 to confine irradiation to the region of interest (ROI). The beam energy (8–11 keV) was optimized to enhance the contrast among hydrate, ice, and the surrounding matrix. After penetration of the sample, X-rays were converted into visible light by a thin Ce-doped YAG scintillator screen (Crismatec Saint Gobain, Nemours, France). Projection images were further magnified by microscope optics and digitized by a high-resolution CCD-camera (PCO2000, PCO GmbH, Germany). The optical magnification was set to 10×, and 2× “on-chip binning” was selected to improve the signal-to-noise ratio, resulting in isotropic voxels of 1.4 μm in the reconstructed images. For our tomogram, 1001 projections were acquired at an integration time ranging from 180 to 642 ms for each projection. The reconstruction of the ROI was performed on a highly optimized FFT transformations and gridding procedures in a few minutes.
 Post processing of the reconstructed raw data was carried out with the mathematical software MATLAB®. 3D segmentation was based on K-means clustering [Spath, 1985], whereby clusters were merged according to corresponding features (e.g., gas-filled voids, gas hydrates, frozen water, salt inclusions and mineral grains). The spatial position of clusters influenced the process of feature formation (spatial neighbourhood relationships). Based on these results, parameters like porosity, surfaces, and volumes were calculated for the whole dataset and single objects.
3. Results and Discussion
 The SRXCTM 3D images of the gas hydrate sub-samples from Chapopote asphalt volcano are shown in Figures 1 and 2. The investigated specimen was sub-sampled manually under LN2 to remove sediment as much as possible. The images therefore do not display any sediment particles. In some cases, cracks are seen (Figure 1a), which could be formed either ex situ due to fragility during contraction in LN2 [Abegg et al., 2007], or in situ due to hydraulic fractures. Gravity coring preserves the sediment texture, and we assume the former cause for this type of cracks.
 Spatial distribution of salt inclusions is helpful to distinguish the type of ice originating from either quenched or interstitial seawater, or the decomposition of gas hydrates, because hydrate is composed of pure water and contains no salt. Regions of frozen brines are identified by SRXCTM, pointing to juxtaposition of the gas hydrate sample with seawater. To what extend ice derives from quenched seawater or dissociating gas hydrate cannot be answered satisfactorily. Triple junctions marked by frozen brine suggest frozen seawater, whereas macro-porous ice [Bohrmann et al., 2007] bordering the hydrate regions probably mixed with hydrate water, characterized by their boundary pores (BP in Figure 1), is likely to originate from decomposing gas hydrate. Occasionally, diffuse appearances of salt occur within the macro-porous ice, suggesting mixing of the hydrate and seawater prior to quenching in LN2. In our sample, the salt inclusions are minute, however, with a concentration high enough to make a significant absorption contrast. They are found at grain boundaries of ice and are frequently cladding the large pores (LP in Figure 1) within this ice, formed from the quenched seawater.
 The gas hydrate domains are surrounded by porous matrix. By analogy with SEM images [Kuhs et al., 2004a; Bohrmann et al., 2007] we can identify areas surrounded by a micrometer-sized porous network as gas hydrate regions. These SEM investigations have established that gas hydrates show a typical feature of sub-μm-sized porosity [Kuhs et al., 2000]. The broad bands with lower density segmenting the gas hydrate regions can be assigned to these sub-μm porous regions. The size of the segments is in the same order of magnitude as the crystallite size of marine gas hydrates [Klapp et al., 2007]; thus we ascribe these segments to individual crystallites. The sub-μm pores cannot be resolved within our spatial resolution limit at present. However, for the first time the spatial distribution of the sub-μm porous regions has been identified suggesting that they occur mainly along the grain-boundary network.
 The sub-μm porous patches, which we ascribe to gas hydrate regions, are limited by a distinctive zone between the gas hydrate and the frozen ice phase (Figure 1). Within this boundary zone, pores are larger than the sub-μm-sized gas hydrate pores but are smaller than the fewer larger pores within the ice phase; their average diameter is about 2–4 μm. These boundary pores (BP) are uniformly distributed around the zones of gas hydrate. The interface between the boundary wall and the hydrates most probably marks the active decomposition front (Figure 1) just immediately before the sample was quenched in LN2 [Bohrmann et al., 2007].
 Large pores (LP) of tens to hundreds micrometer in size are quite frequent within the frozen water (Figures 1a and 1b). These pores were known from earlier SEM assessments [Bohrmann et al., 2007]; here we see that in the third dimension these pores are of arbitrary shape and distribution. They probably arise from gas bubbles discharged from the partial decomposition of the sample while retrieving from the sea floor. The LP, being present in the ice phase only, most likely represent gas bubbles in former water derived from decomposed gas hydrate. Accordingly, regions with a concentric sequence of BP and LP provide good evidence for the previous presence of gas hydrates in this area for recovered samples. It should be noted that, in agreement with earlier work [Heeschen et al., 2007], individual pores (∼μm) can also be occasionally found within the gas hydrate (Figure 1a).
 Irrespective of size and location, most of the pores are interconnected forming nodular channel networks (Figures 2b and 2c). The channels have no defined shape and preferred orientation as visualized from the 3D imagery. Figure 2d depicts a large cavity, which is crossed by small channels (see arrow). The nature of this phenomenon is unclear, but it demonstrates that the channel networks must be mantled by slightly denser material like ice or brines.
 The porous feature of the sample is characterized by some computations within the cylindrical volume (2.3 × 109μm3) analyzed. The calculated porosity (gas-filled voids to ice/GH/substance volume) is 29%, total volume of the voids 6.7 × 108μm3, and the void surface area (= inner surface area) 1.15 × 108μm2. The number of the total calculated gas-filled voids is 51074 (minimum size >10 connected voxels) of which the biggest interconnected void (internal channel volume assuming that all the voids are completely gas-filled/free) covers 26% of the total volume, i.e., ∼89% of the total void volume is interconnected. The smaller voids (<10 connected voxels) may also be interconnected; however, a calculation was not performed due to statistical noise of the raw data and edge enhancement effects. It is important to note that the voids and the pores do not mirror the truly intrinsic volume, as the sample partially decomposed.
 The SRXCTM 3D images of the as-preserved specimen demonstrate the spatial relation among gas hydrates, frozen water, pores and their connectivity, and salt or frozen brine inclusions. The pore properties have been characterized along with some surface features. Pore studies of a fully-preserved sample, representative of the marine settings, by the SRXCTM will provide helpful information to understand the unusual seismic signals in porous rocks containing gas hydrates [Gerner et al., 2007] and may eventually lead to better estimation of the gas hydrates [e.g., Ecker et al., 2000]. Likewise, the 3D gas hydrate crystallite arrangements will give valuable insights into possible coarsening or ripening process; understanding these processes may help provide an estimation of the formation age of gas hydrates [Klapp et al., 2007]. Further work on time-resolved tomography (4D-tomography) might lead to a better understanding of decomposition scenarios under marine and permafrost conditions, also including the effect of self-preservation [Stern et al., 2003; Kuhs et al., 2004b].
 The authors gratefully acknowledge the technical support provided by G. Mikuljan at PSI-SLS. Thanks are due to S. McDonald for his helpful discussions and practical support during the experiments. The help of the crew and captain of the R/V Meteor during cruise M67 is gratefully appreciated. We sincerely acknowledge DFG via the SFB-641 (TROPEIS) for partial financial support of this research. SAK thanks the Vollkammer scholarship for support and the Glomar Graduate School for Marine Science. This is publication GEOTECH-338 of the R&D Programme GEOTECHNOLOGIEN (METRO, grant 03G0604A (BMBF)).