Mudstone pore networks are strong modifiers of sedimentary basin fluid dynamics and have a critical role in the distribution of hydrocarbons and containment of injected fluids. Using core samples from continental and marine mudstones, we investigate properties of pore types and networks from a variety of geologic environments, together with estimates of capillary breakthrough pressures by mercury intrusion porosimetry. Analysis and interpretation of quantitative and qualitative three-dimensional (3D) observations, obtained by dual focused ion beam–scanning electron microscopy, suggest seven dominant mudstone pore types distinguished by geometry and connectivity. A dominant planar pore type occurs in all investigated mudstones and generally has high coordination numbers (i.e., number of neighboring connected pores). Connected networks of pores of this type contribute to high mercury capillary pressures due to small pore throats at the junctions of connected pores and likely control most matrix transport in these mudstones. Other pore types are related to authigenic (e.g., replacement or pore-lining precipitation) clay minerals and pyrite nodules; pores in clay packets adjacent to larger, more competent clastic grains; pores in organic phases; and stylolitic and microfracture-related pores. Pores within regions of authigenic clay minerals often form small isolated networks (<3 μm). Pores in stringers of organic phases occur as tubular pores or slit- and/or sheet-like pores. These form short, connected lengths in 3D reconstructions, but appear to form networks no larger than a few microns in size. Sealing efficiency of the studied mudstones increases with greater distal depositional environments and greater maximum depth of burial.

Mudstones, rocks composed of 50% or more silt- and clay-sized material with grain sizes <62.5 μm (MacQuaker and Adams, 2003), compose the major portion of sedimentary basin fill (Blatt, 1982) and thus play an integral role in many geologic processes and systems. These rocks dominate fluid dynamics of the subsurface, affecting pore pressure distribution and local- to regional-scale patterns of groundwater and hydrocarbon transport (Aplin et al., 1999). Mudstones can act as source rock, migration pathway, and caprock for oil and gas, and even as reservoirs for shale gas (Loucks et al., 2009). Properties such as low permeability and high capillary breakthrough pressure make these rocks ideal as barriers (seals or caprocks) to the movement of single or multiphase fluids (Potter et al., 2005). Hence, they are targets, as bounding formations or caprocks, for hazardous waste storage (Marty et al., 2003; Davy et al., 2009) and subsurface containment of anthropogenic CO2 (Intergovernmental Panel on Climate Change, 2005). Despite their societal and scientific importance, only limited work has achieved a detailed characterization of pore networks in mudstones, especially in three dimensions, due to constraints in sampling, preparation, measurement resolution, and the heterogeneous nature of these rocks (Desbois et al., 2009). Greater understanding of the geometry, topology, and pore-lining material of pore networks in mudstones will yield insight into their predicted controls on transport and storage of geologic and injected fluids (U.S. Department of Energy, 2007).

Recent studies have demonstrated the utility of dual-beam focused ion beam–scanning electron microscopy (FIB-SEM) systems for imaging submicrometer-scale textures and compositions, including mudstones (Kotula et al., 2003; Kotula and Keenan, 2006; Tomutsa et al., 2007; De Winter et al., 2009; Desbois et al., 2009). These methods combine milling of smooth nanometer-scale surfaces and high-resolution imaging by field emission SEM (Yao, 2007). Successive milling and imaging yield a series of two-dimensional (2D) images, which are stacked and processed to construct a 3D pore network geometric model (Holzer et al., 2004; Tomutsa et al., 2007). Images taken in backscattered electron mode (BSE; De Winter et al., 2009) provide information on mineral and organic matter distributions by variation in mean atomic number (Z). Recent FIB-SEM studies on low-permeability geologic samples focused on defining and describing submicrometer pore types, and morphology, capillarity, fractal scaling, and fluids in pores (Tomutsa et al., 2007; Desbois et al., 2008, 2009; De Winter et al., 2009; Loucks et al., 2009).

In this paper we investigate properties of pore geometries and pore networks in mudstones from several depositional environments. Of interest is the relative importance of depositional environment, burial history, and diagenesis as primary controls of pore network properties. We incorporate high-resolution (i.e., submicrometer) 3D petrography using FIB-SEM techniques (see discussions labeled “Materials and Methods” and “FIB-SEM Imaging and 3D Image Analysis”) to identify pore and pore network types, to permit comparison of pore networks from different mudstone lithofacies, and to compile pore network statistics (see discussion in Results). Mercury intrusion porosimetry (MIP; see discussions in Materials and Methods) is applied to the same samples to determine pore throat distributions of the connected pore network and breakthrough pressures, and evaluated in light of the 3D pore network information (see discussion in Results). Our results suggest that there is a strong association between depositional environment, burial history, and capillary sealing behavior of these mudstones, all of which have been proposed as caprocks for subsurface CO2 storage (see discussion of Pore Networks and Sealing Quality).

The Southeast and Southwest Regional Carbon Sequestration Partnerships (SECARB and SWP; Litynski et al., 2008) provided mudstone samples from cores through caprock sealing sequences at recently deployed Phase II CO2 sequestration sites (Fig. 1A; see Litynski et al., 2008, for descriptions of SECARB, SWP, and the U.S. Department of Energy's Regional Carbon Sequestration Partnerships program in general). The mudstones cover a range of depositional environments from continental to marine shelf (Table 1; Figs. 1B–1D). Geologic units investigated include, from generally proximal to more distal or deeper water depositional environments, the Late Cretaceous Kirtland Formation, the Late Cretaceous Lower and Middle Tuscaloosa Group, and the Gothic shale of the Pennsylvanian Paradox Formation.

Kirtland Formation

The Late Cretaceous Kirtland Formation of the San Juan Basin, New Mexico and Colorado (Fig. 1A), is a regional aquitard and reservoir seal (Ayers, 2003). It was deposited by streams flowing toward the retreating shoreline of the Western Interior Seaway in an alluvial plain with floodplain and channel depositional environments, which were landward of the swampy environments of the underlying Fruitland Formation (Fig. 1B; Fassett and Hinds, 1971; Fassett, 2009). The Kirtland Formation is divided into upper and lower shale (i.e., mudstone rich) members and a middle sandstone-rich member, the Farmington Sandstone (Bauer, 1917; Fassett and Hinds, 1971; Stone et al., 1983; Molenaar and Baird, 1992). Throughout most of the basin the Kirtland Formation conformably overlies the coal-bearing Fruitland Formation (Fassett and Hinds, 1971), which contains the world's most prolific coalbed methane play (Ayers, 2003).

Starting in May 2008, the SWP oversaw drilling of CO2 injection well EPNG Com A Inj 1 (API No. 30–045–34305; lat 36.307735°N, long 107.251278°W; Fig. 1A) with coring of the upper and lower shale members of the Kirtland Formation from depths of 624–631 m and 820–822 m. The SWP injected CO2 into unmineable coals seams at depths between ∼889 m and 957 m for a 12 month period (i.e., from 30 July 2008 to 12 August 2009). Core samples studied for pore network characteristics herein were obtained from depths of 624.75 and 820.80 m, representing the upper and lower shale members, respectively (Table 1).

Optical petrography complemented by standard SEM, X-ray diffraction (XRD), total organic carbon (TOC), porosity, and permeability analyses of several upper Kirtland samples taken along the length of the core indicates sandy to silty argillaceous mudstones (Figs. 2A, 2B; Heath, 2010). Mottled, disorganized, and poorly laminated textures are common. Pedogenic features include illuviation envelopes that contain well-aligned clays transported into the sediments along root channels or desiccation cracks. Sand- and silt-sized grains include quartz, plagioclase feldspar, and volcanic and sedimentary rock fragments (e.g., chert-replaced volcanic clasts and clay rip-up clasts). Some lithofacies are dominated by sand-sized grains, whereas others are mostly silt- and clay-sized material. Clay minerals include smectite, illite, mixed layer smectite/illite, chlorite, and minor kaolinite. Clay expandability ranges from 20% to 23%. Authigenic minerals include minor carbonate cement (usually iron rich) and pyrite.

Macroporosity types were summarized in Heath (2010) and include abundant induced porosity from coring (i.e., pressure-release fractures), core handling, and dry-out (∼5–20 μm in size); root and organic particle casts left by the decayed organic material (2–20 μm); and matrix-hosted pores between clay flakes and within cements (<5 μm). Minor intragranular porosity within feldspar clasts also occurs. Porosity and permeability range from 6.5% to 9.9% and 7 × 10−20 to 1 × 10−19 m2, respectively, with total organic carbon ranging from 0.06 to 0.09 wt%. Vitrinite reflectance values for the Kirtland Formation for the area of interest may range from 0.8% to 1.3% Rm, with maximum depth of burial of the upper and lower Kirtland Formation being ∼1630 and 1830 m, respectively (Law, 1992).

Tuscaloosa Group

In the northeastern Gulf of Mexico Basin, the Tuscaloosa Group represents the basal strata of a thick wedge of predominately clastic Late Cretaceous sediments, which thin updip (to the north) and thicken downdip (to the south), representing at least four region-scale marine transgressive-regressive cycles on the Cretaceous shelf (Mancini and Puckett, 2005). The Tuscaloosa Group contains predominately siliciclastic sediments and represents fluvial-deltaic and marginal marine to mid-shelf open-marine depositional environments (Mancini et al., 1987; Rosen and Rosen, 2008). In the subsurface of Alabama and Mississippi, the Tuscaloosa Group contains a lower sandstone unit, a middle unit of predominately marine shale, and an upper sequence of interbedded sandstone and shale. These three Tuscaloosa divisions persist along depositional strike over a large region of the northern Gulf of Mexico Basin. The middle Tuscaloosa Group, informally called the middle Tuscaloosa Marine Shale or Marine Tuscaloosa, ranges from Cenomanian to Turonian in age and represents the maximum marine transgression during deposition of the Tuscaloosa (Liu, 2005). The deltaic and marginal marine sandstones of the lower Tuscaloosa Group (the so-called basal massive sand unit) represents a period of aggradational sandstone deposition preceding the maximum marine transgression represented by the middle Tuscaloosa Marine Shale. Thin siltstone and sandstone beds become thicker and more abundant upward in the Marine Tuscaloosa due to subsequent progradational infilling (Liu, 2005; Mancini and Puckett, 2005; Pashin et al., 2008).

In 2009, Southeast Regional Carbon Sequestration Partnership (SEACARB) conducted a CO2 injection test (Litynski et al., 2008) in the basal lower Tuscaloosa Group sandstones at the Plant Daniel site in Jackson County, Mississippi (Fig. 1C). Both the midddle Tuscaloosa Marine Shale and an interbedded mudstone unit between two thick lower Tuscaloosa Group sandstones are considered potential seals for the CO2 injected into these sandstones. Core samples for characterization of pore networks were obtained from the Mississippi Power Company No. 1 observation well (MPC #11–1; API No. 23–059–20023–00; lat 30.536902°N, long 88.558073°W). Depth intervals cored include 2409.7–2418.3 m in the middle Tuscaloosa Marine Shale, and 2600.2–2618.7 m in the interbedded mudstone unit in the lower Tuscaloosa Group. The lower Tuscaloosa Group core includes the bottom part of the upper massive sand unit from 2600.2 to 2611.8 m; interbedded and bioturbated sandstone and silt- and clay-bearing mudstones from 2611.8 to 2615.8 m; and predominately clay-rich mudstone with silt-rich mudstone laminations from 2615.8 to 2618.7 m. Pore network samples for this study are from depths 2415.7 m, 2417.6 m, 2618.2 m, and 2618.5 m. These sample depths correspond with those of the Marine Tuscaloosa and the interbedded mudstones in the lower Tuscaloosa Group.

A thin-section image and photomicrograph for the lower Tuscaloosa Group (Figs. 2E, 2F) from core depth of 2615.05 m show an interbedded sand-, silt-, and clay-rich mudstone, with individual layers that thin and pinch out. Sand- and silt-sized grains include angular quartz, micas, and possible feldspar. Opaque regions are authigenic pyrite and organics. Bioturbation (burrows) is visible at the thin-section scale (Fig. 2E). XRD from a similar depth (2613.6 m) indicates that clays are chlorite, kaolinite, and illite, and nonclay minerals are predominately quartz with lesser amounts of plagioclase feldspar and minor K-feldspar (U.S. Department of Energy, 2008).

A thin-section scan and photomicrograph of the Marine Tuscaloosa (Figs. 2G, 2H) from a core depth of 2416.27 m also show interbedding of sand-, silt-, and clay-rich layers. Clays do not show alignment under crossed polarizers as strong as that in the Lower Tuscaloosa Formation thin section. Sand- and silt-sized grains include quartz and lesser amount of micas and feldspars. XRD on a sample from 2417.6 shows that clays are chlorite, kaolinite, and illite with ∼5 wt% calcite (U.S. Department of Energy, 2008).

Petrophysical and other properties for the Tuscaloosa samples as determined by SEACARB were summarized in U.S. Department of Energy (2008). Porosity and Klinkenberg-corrected permeability at 17 MPa net confining stress for the lower Tuscaloosa Group from a depth of 2618.2 m are 9.5% and ∼7 × 10−17 m2, respectively. The TOC at depth 2617.7 m is 1.0 wt%. Marine Tuscaloosa porosity and permeability at 17 MPa net confining stress at depth 2415.7 m are 2.2% and ∼1 × 10−19 m2, respectively. The TOC for depths of 2415.4 m and 2417.2 m are 0.65 and 0.73 wt%, respectively. Calculated vitrinite reflectance values (%Ro) of the Marine Tuscaloosa core are 0.51 and 0.47 for depths of 2415.4 and 2417.2 m, respectively. Lower Tuscaloosa Group core samples, at depths of 2616.6 and 2617.7 m, have calculated vitrinite reflectance values of 0.69 and 0.65, respectively. Production index and maturity (based on the maximum temperature during Rock-Eval pyrolysis) indicate that the Marine Tuscaloosa and lower Tuscaloosa Group core samples are at low maturity (i.e., low level of conversion of kerogen). The core contains mixed type II-III kerogen. Current core depths represent the maximum depth of burial of the core samples (Mancini et al., 1999).

Gothic Shale

The Pennsylvanian Gothic shale, composing the basal unit of the Ismay zone of the Paradox Formation, occurs within the Blanding subbasin of the Paradox Basin in southeastern Utah and southwestern Colorado (Fig. 1D), overlying the Desert Creek hydrocarbon-producing zone of the Greater Aneth oil field (Peterson, 1992). Goldhammer et al. (1994) described Gothic shale lithology as a sapropelic dolomite and dolomitic shale to silty carbonate mudstone, deposited as a transgressive systems tract (TST) during fourth-order sea-level rise. At the Aneth Unit, the Gothic shale ranges from 1.5 to 8.2 m thick and is separated by an unconformity (sequence boundary) from the underlying mottled carbonates of a mound buildup complex (Chidsey et al., 2009). The Gothic shale is Goldhammer et al.’s (1994) black laminated mudstone lithofacies that represents the TST, possibly representing a deeper water subtidal zone with deposition in quiescent, reducing bottom conditions as much as ∼30 m deep.

The Gothic shale is the reservoir caprock at the Aneth Unit (Chidsey et al., 2009). As part of Phase II evaluation of CO2 injection into the Desert Creek zone for enhanced oil recovery, the SWP obtained Gothic core from the Aneth Unit H-117 well (API No. 43–037–30153; lat 37.3093°N, long 109.3035°W), which was drilled in 1974. The SWP performed a suite of tests on the core to evaluate sealing integrity (Chidsey et al., 2009). The cored interval examined by the SWP included a complete 5.09 m section of Gothic mudstone from depths of 1638.5–1643.8 m. The core sample studied herein for pore network and petrographic properties came from a depth of 1643.1 m (Chidsey et al., 2009).

Figures 2I and 2J show a thin-section scan and photomicrograph of the Gothic shale sample examined in this study, showing a fine-grained mudstone with faint lamination and well-developed fissility. Textural components include minor silt-sized quartz, calcite, dolomite, and mica in a dominant clay matrix, with common authigenic pyrite. Compacted white nodules (Fig. 2J) are composed predominately of chert. Fossils include conodonts, brachiopods, and condalarids. Besides induced porosity, macroporosity and microporosity are difficult to identify. Thus, the Gothic appears less porous and more homogeneous than the Kirtland or Tuscaloosa samples.

Porosity, pressure-decay permeability, and TOC measurements taken from four depths from the core range from 2.7% to 4.3%, 1.3 × 10−19 to 1.4 × 10−19 m2, and 2.2–4.4 wt%, respectively (Chidsey et al., 2009). Vitrinite reflectance values (calculated %Ro) range from 0.83 to 0.96. Production index and maturity (based on the maximum temperature during Rock-Eval II pyrolysis) indicate that the kerogen within the Gothic shale core is mature, containing type II and mixed type II-III kerogen. The maximum depth of burial is ∼3050 m (Nuccio and Condon, 1996).

To facilitate interpretation of geologic controls on pore network properties and capillarity, FIB-SEM techniques were used to image pores, contrasting mineralogy, organic phases, and geologic structures. Representative samples from each core were characterized first by FIB-SEM and then with MIP. The MIP was used to allow comparison of interpreted MIP pore size distributions to the direct imaging of the pore networks, as well as to obtain breakthrough pressures in an assessment of sealing quality.

Sample Preparation

We prepared 10 mudstone samples, 2 from each of the study units, from plugs (2.5 cm diameter by <1.5 cm long). The samples were taken from the same or similar rock types as those of the thin sections shown in Figure 2. Plug orientation was approximately parallel to the axis of the core and roughly perpendicular to bedding. Prior to microanalysis, the top and bottom of the plugs were lightly sanded with fine-grit paper to achieve a nominally flat surface. Due to potential fluid sensitivity of these clay- and organic-rich samples, polishing using a saline solution or oil was not performed to avoid imbibition of fluids, deflocculation, contamination, or other alteration of pore structure, organics, and other solid phases. Plug cleaning was also not performed. Significant penetration of drilling fluids into the plugs was unlikely due to low permeability of the rocks and appearance of fresh surfaces.

FIB-SEM Microscopy

Samples were gold-palladium coated and painted with silver dag to provide a current path and mitigate specimen charging. Samples were placed in mounts or vises and electrically grounded. Serial sectioning and imaging by FIB-SEM was performed with FEI Company's Helios 600 Nanolab DualBeam instrument and semiautomated Slice and View software. To promote smooth, cross-sectioned faces and avoid curtaining artifacts (i.e., uneven vertical striations) during milling (Hayles et al., 2007), and to provide a fiducial reference for image alignment, a layer of platinum was deposited over the area of sample to be milled using the FIB deposition system (Yao, 2007). During milling, the Ga+ ion beam was normal to the sample surface, and the electron beam was at an angle of 52° from the ion beam. Acceleration voltage for Ga+ was 30 kV, and the beam current was 2.8 nA. A rough-cut trough was made to provide an area for deposition of milled material and to reveal the vertical cross section. Vertical cross-sectioned surfaces perpendicular to bedding and the upper surface of the sample were milled by the ion beam at 25 nm spacing. The Slice and View software facilitated cross sectioning with automation for beam shifts and auto-focusing; however, manual adjustments were made during milling to ensure acquisition of high-quality electron micrographs. Backscattered electron (BSE) imaging mode with a through-the-lens detector was chosen for obtaining mineralogical information in terms of differences in mean atomic number, Z, which provided strong contrast in mineralogy, organics, and pores for most samples. The acceleration voltage and beam current were 1 kV and 1.4 nA, respectively. The BSE micrograph field of view was 16.00 μm × 13.81 μm.

Image Analysis

Digital reconstruction and quantification of 3D pore networks and organic distribution for selected samples required several steps. The computer software ImageJ 1.42q (Rasband, 1997–2010; Ferreira and Rasband, 2010) was employed to align, crop, and segment stacks of Slice and View images into regions of pore or nonpore or organics or nonorganics. The ImageJ plugin TransformJ was used to interpolate between segmented images (Meijering, 2008), resulting in a cubic voxel length size of 15.6 nm. The cropped regions consisted of 299 × 299 × 299 voxels, with a total cubic data volume with side length sizes of 4.66 μm. TransformJ interpolation produced gray-scale information, which necessitated further segmentation for proper delineation of pore or nonpore or organics or nonorganics. We performed 3D floodfill image rendering with calculation of pore volume, organic volume, and porosity using ScanIP by Simpleware Ltd. Segmented images of pore or nonpore regions were input into the 3DMA-Rock code (Lindquist, 1999) for further quantification of the pore networks.

The 3DMA-Rock algorithms performed the following steps in sequence (Lindquist, 1999; Lindquist and Venkatarangan, 1999; Lindquist et al., 2005; Neethirajan et al., 2008; Udawatta et al., 2008): medial axis construction of pore pathways; throat finding and pore partitioning; throat and pore characterization including pore volume, coordination number, and throat area; and geometrical construction of the pore and throat network. The medial axis construction produces a centrally located skeletonization of the pore space, which preserves topology and geometry of the pore network. Voxel paths consisting of five voxels or fewer were excluded from the final medial axis construction. Throats are minimal cross-sectional surface areas located on the medial axis, and the throat surfaces may be nonplanar. The coordination number of a pore gives the number of neighboring pores. The 3DMA analysis enabled compilation of pore statistics in the form of pore volume and pore throat radii frequency distributions and other topologic properties (see Results discussion).

The distribution of throat and pore body sizes, connectivity of the pore network, and surface area and shapes of pores have strong control on drainage and imbibition processes and fluid flow properties such as relative permeability (Neethirajan et al., 2008).

MIP was performed on a Micromeritics AutoPore IV 9500 Series porosimeter. The inner diameter of the penetrometer bulb (Sigal, 2009) was 2.54 cm, which necessitated light sanding of the 2.54 cm plugs to a smaller size for placement into the bulb. Sanding removed the gold-palladium coating and silver dag previously applied to the samples. Prior to analysis, samples were dried at 100 °C and photographed. To investigate capillary and transport properties in the direction perpendicular to bedding, some plugs were jacketed with epoxy for directional intrusion. Some plugs were too thin for proper jacketing and directional intrusion.

Closure pressure corrections (Sneider et al., 1997) followed a compressibility method that estimated the pressure and corresponding volume of mercury when the mercury first penetrated the pore network (Colombo and Carli, 1981; Almon et al., 2008). The closure volume was subtracted from the incremental volumes of mercury intruded into the penetrometer bulb. Breakthrough capillary pressure (also called bubbling pressure or sealing pressure) is the pressure at which a continuous filament of mercury extends across a MIP sample, or equivalently, the pressure when the nonwetting phase first appears on the outlet side of a sample plug (Katz and Thompson, 1987; Dullien, 1992; Dewhurst et al., 2002). For this study, it was estimated by identifying the point on the cumulative mercury saturation versus pressure curve at which it had maximum inflection upward (Dewhurst et al., 2002; Daniel and Kaldi, 2008).

FIB-SEM Imaging and 3D Image Analysis

The 3D pore network reconstructions of 101.5 μm3 regions from each mudstone are presented in Figures 3–10, arranged in order from proximal to distal from sediment source and deepening water depth of deposition. Shown for Kirtland Formation, Tuscaloosa Group, and Gothic shale samples are a single 1024 × 884 pixel 2D BSE image, usually taken from the back of an image stack, an extracted 299 × 299 segmented image to distinguish pores (in black), a 3D reconstruction of pores, and medial axis and pore throat 3D visualizations. Organic-rich samples have additional images to show 3D distributions of organic phases as layers and possible pore-filling entities. Medial axes, representing the skeletonization of pore networks, are shown using a rainbow color scale wherein red corresponds with low burn numbers, and blue and violet are higher burn numbers. Burn numbers represent the maximum norm distance in number of voxels from the medial axis to the grain boundary where a pore sharing a grain boundary has a number of one (Neethirajan et al., 2008). The cubic voxel size for all 3D images is 15.6 nm.

To allow viewing of the FIB-SEM images used to construct 3D pore network models, Animations 1–4 present sets of serial sections for the upper Kirtland Formation, the lower Tuscaloosa Group, Marine Tuscaloosa, and Gothic shale, respectively. These present a variety of distinct 3D pore types for the different geologic environments. A pore type classification scheme is introduced herein, and shown in Figure 11. Pore statistics for all analyzed digital samples are given in Figures 12 and 13.

A common artifact visible in most of the FIB-SEM images is induced pores near the upper surface (i.e., top of micrographs) of the samples where mechanical cutting and rough polishing were performed. Pore shapes at that location can be much different and nonrepresentative of those deeper in the cross section. Specimen charging, visible as bright white spots in the images, is an occasional imaging artifact and was especially prevalent on organics. It can be differentiated from high Z phases like pyrite by a dark halo around the margins of the bright areas.

Figure 3 presents the pore network of the upper Kirtland Formation, from a sample 624.75 m (below ground surface, bgs). The lack of lamination or planar clay fabric, mottled texture, and the more or less random distribution of pore and clay orientations (observable from the larger, higher Z clay packets that are likely chlorite) in the BSE image (Fig. 3A) may reflect the compacted analogues of deposited clay packets or floccules, as observed experimentally by Schieber and Southard (2009). The disorganized texture may reflect pedogenic processes of aggregation formation and weathering before deeper burial, such as by shrink-swell of clays. Pores are dominantly slit shaped, as shown by the image of the segmented pores (in black; Fig. 3B). Larger pores are subparallel to surrounding clay fabric and discernable by subtle changes in gray level, roughly perpendicular to the vertical axis. Voxel counts of the segmented 3D image (Fig. 3C), with pores shown in red, indicate a porosity of 1.04%, and 3DMA-Rock analysis indicates that 34% of pores are connected. Figures 3D and 3E show representations of pore network medial axes (rainbow colored) and pore throat shapes (in gray). Even through some of the imaged pores appear quite large (Fig. 3C), the pore throats are small. Throat size has implications for interpretation of MIP data presented herein.

An example from a carbonate-rich zone with quartz clasts (lower gray-scale value) in the lower Kirtland Formation is shown in Figure 4A. Although there are a number of induced fractures in the upper portion of the image, the planar pores observed in the BSE image in Figure 4A may actually be associated with a microstylolitic structure [this is interpreted from scanning transmission electron microscopy (STEM)–energy dispersive spectroscopy (EDS) analysis presented in Heath, 2010]. The interconnected porosity can be observed in the 3D representation shown in Figures 4C and 4D, with larger pore throat radii than in the clay matrix of Figure 3. Total porosity by voxel count is 0.72%, with 28% of that being connected.

A polygonal crack pore structure is evident in an example from the lower Tuscaloosa Group at 2618.2 m (bgs), shown in Figure 5. This may represent boundaries between depositional clay packets, but more likely represents a wetting-drying structure not unlike hexagonal fractures in mud. A 3D floodfill image (shown in blue) shows a large degree of connectivity across the sampled region. Largely due to this structure, this example has 2.64% porosity with 74% connectivity. The crack pore may be induced by coring and/or sample preparation, and may not be an in situ feature.

A sampled region in Figure 6, from the lower Tuscaloosa Group, shows a nominally horizontal microfracture along with crescent-shaped pores aligned with a possible authigenic clay fabric. It is not known if the porosity was preexistent to the clays, providing access to mineralizing fluids and concomitant authigenic precipitation, or if the porosity results from a phase change, perhaps as more dense chlorite replaced a less dense clay. Crystal habit of the clay minerals, perhaps rosettes, controls the non–slit-like pore shapes. Chlorite rosettes, as seen in traditional SEM photomicrographs by other researchers (Welton, 1984), have triangular, trapezoidal, and rectangular pore shapes between the chlorite mineral plates similar to those that are visible in Animation 2.

Figure 7 shows interlamination between organics (lower Z and thus darker gray) and clay matrix (lighter gray) from an example from the Marine Tuscaloosa at 2415.7 m (bgs). Organic layer thickness can exceed 2 μm. The boundary between the clays and organics is wavy and convolute. Some clays appear isolated within the organics in the 2D images. The organic phase has been confirmed as dominantly carbon with chloride and sulfur by EDS analysis (Heath, 2010). Because the organics are located as an intact layer, they are likely syngenetic with surrounding clays. There are tiny pores, barely discernable at the resolution of our method, seen in the segmented image in Figure 7B. The 3D reconstruction (Fig. 7D) shows that some of these are aligned to form large tubes. Although representing very little porosity (0.58%), connectivity on the 3 μm scale is high at 62%, and these pathways may represent primary migration pathways for petroleum generated within them. Connectivity on larger length scales may be poorer, but this may be an artifact of our imaging resolution.

A clastic portion of the same image data set for the Marine Tuscaloosa is shown in Figure 8. Pores from this portion appear to have a slit-like pore morphology similar to that in the upper Kirtland; although the porosity is less (0.47%), a higher percentage is connected (52%). Elongate phases (higher Z and brighter gray scale) indicate possible detrital muscovite, which is consistent with thin-section observations. XRD data indicate that clay phases may include chlorite, illite, and kaolinite (U.S. Department of Energy, 2008).

Figures 9A and 10A show BSE images of the Gothic shale at 1643.1 m (bgs) with 3D representations of the distribution of organic matter at 15.19% and 5.59% by volume, respectively (Figs. 9B, 9C, 10B, and 10C), and porosity at 0.42% and 1.33% by volume (Figs. 9E and 10E), occurring largely adjacent to clastic quartz and feldspar grains. All imaged porosity appears to be located within or adjacent to organic matter (see Animation 4). Organic content is high, although not concentrated in layers as in the Marine Tuscaloosa sample. While some pores may be relatively large (shown in blue in Fig. 9E), pore throats, shown in Figure 9G, are small, and the pore bodies are not well connected (42%). Pore morphology in these organics is slit and/or sheet like, as opposed to the circular and/or tubular pores of the Marine Tuscaloosa. The large bright circular structure to the lower right of Figure 9A is a pyrite framboid that contains pores. The orientation of clays and clasts around the framboid indicate pyrite formation prior to significant compaction.

Summary of Pore Types

Based on 2D observations of morphology and size of a FIB serial section from a Boom Clay sample, Desbois et al. (2009) identified three mudstone pore types, which they termed pore types I, II, and III: the type I pores encompass elongate pores in similarly oriented clay sheets and are <100 nm in the elongate direction. Type II includes crescent-shaped pores in folded clays and are 100–1000 nm in size. Type III pores are large, jagged pores associated with margins of larger, more competent grains and are typically >1000 nm (Desbois et al., 2009). Ideally, one would like to use our 3D observations of pores and pore networks along with those of Desbois et al. (2009) to propose a universal classification scheme for mudstone pore networks. Such a classification scheme could be based on characteristics of pores in mudstone microfacies such as morphology; sizes and size distributions of pores in networks; connectedness and/or topology of pores; size relationship of pore throats to pore bodies; roughness of pore body walls; geometrical relationships between the pores and surrounding grains; and the characteristic of being induced or present in situ. Clearly, such an ambitious scheme is beyond the scope of this paper.

For our purposes here, distinct types of pores and 3D pore networks can be distinguished and placed into at least seven descriptive groups (Fig. 11), including the Desbois et al. (2009) pore types I–III, and adding pore types IV–VII (Fig. 11). Type IV, pores in organics, includes two subtypes, IVa and IVb. Type IVa refers to circular and/or tubular pores in organics in the Marine Tuscaloosa (Figs. 7 and 11; Animation 3), whereas IVb refers to slit-like pores in organics in the Gothic shale (Figs. 9 and 11; Animation 4). Type V refers to microstylolitic or other diagenesis-related pores, as based on lower Kirtland 2692.9A in Figures 4 and 11. The pores in Figure 4 in particular have a jagged, dentate morphology that appears more natural than induced, and this is brought out more in STEM–EDS imaging analysis on this sample discussed in Heath (2010). Type VI is for microfractures, which may or may not be induced (Figs. 5 and 6). Type VII designates pores in pyrite framboids, illustrated by the Gothic shale in Figure 9.

Type I pores are abundant and dominant in the upper Kirtland Formation. Some of the larger pores exceed 1000 nm in their axial directions and have a length perpendicular to their axis of <∼130 nm. A range of smaller sizes has similar shapes. The Tuscaloosa Group mudstone samples also show an abundance of type I pores, although less commonly than the Kirtland. The Gothic shale has the most infrequent occurrence of type I pores. Desbois et al. (2009) described the orientation of type I pores as being similar to nearby clay sheets, and thus compaction-derived fabric, which is verifiable in our FIB-SEM images as clay sheets based on differences in gray scale of parallel sheets (see Figs. 3A–3E). However, some FIB-SEM images show elongate pores where the clay fabric is not clearly visible. In addition, some elongate pores contain apparent bridging material inside the pores, not noted by Desbois et al. (2009), that is approximately perpendicular or slanted relative to the margins of the pores. Some pores have sharp tips, whereas others have terminations that are curved, and thus the pores can appear as narrow ellipses. In the lower Tuscaloosa Group (Fig. 6) many of these pores are curved and follow the undulate texture of the surrounding sheet clays (see Animation 2).

Type I pores are interconnected through small pore throats at the tips of the pores, and the small throat radii undoubtedly contribute to high capillary pressures (see following discussion of MIP). In addition, due to the sheet-like or fracture-like geometries, these pores are likely enlarged in size relative to those occurring in situ, due to unloading and/or clay mineral shrinkage. The orientation of most of the observed type I pores, being subperpendicular to overburden and thus maximum principal compressive stress, could contribute to a stress-sensitive permeability of mudstones with abundant type I pore networks.

Type II pores of Desbois et al. (2009) are crescent shaped in folded clays. We extend their definition to include pores associated with clays that are in predominately nonparallel orientations. Thus, type II can include pores associated with authigenic pore-lining or pore-filling clay minerals, pores with recrystallized minerals, compacted detrital floccules (if preserved), or clays in nonhorizontal orientations that have been deformed. The lower Tuscaloosa Group contains prime examples of this pore type (Figs. 6 and 11; Animation 2). The nonparallel orientations of groups of clay crystals probably indicate that they are authigenic and formed after primary deposition. Thus, the pores may be associated with chlorite precipitation that may be a pore-filling or replacement feature. For example, replacement textures could entail conversion of less dense hydrous smectite to denser chlorite plus secondary pores.

The Desbois et al. (2009) type III represents pores associated with margins of relatively large, competent grains, which we assume are detrital, nonclay clasts (e.g., detrital quartz). Our samples do not have clear examples of this pore type; however, the Gothic shale samples appear to have compaction shadows, or rather structures associated with deformation of relatively small material around larger clasts with possible preexisting pores that are filled with organics (Figs. 9–11; Animation 4). Organic phases may have flowed into type III pores during deep burial and compaction of the Gothic shale.

Type IVa pores in Marine Tuscaloosa organics are generally circular to bulbous to ellipsoidal in two dimensions (Figs. 7 and 11), whereas type IVb pores in organics in Gothic shale are slit like (Fig. 11). Networks of type IVa pores can be arranged linearly and appear tubular. The generally circular pore cross sections indicate in situ formation that postdates most compaction, probably due to interfacial tension and minimization of energy between gases and liquids during conversion of kerogen to hydrocarbons (Loucks et al., 2009). The slit pores in organics in the Gothic shale (Fig. 11) can be located near margins of organic particles while also cutting through the interior of the particles away from the margins. The orientation of several slit pores is similar to the general subhorizontal fabric of the rock. Formation of this pore type is unclear, but probably related to maturation of the organic material within the Gothic shale.

Types V and VI represent subplanar stylolitic and microfracture-related features. These can have generally planar, slit-like morphology and high connectivity across the 3D reconstructions (Figs. 4, 6, and 11). These pore types may be induced due to pressure release during coring and desiccation of swelling clays. Their large size could contribute significantly to lower capillary pressures if well connected throughout an MIP plug sample.

Type VII pores occur within pyrite framboids (Fig. 9; Animation 4). Framboids are generally isolated in both the Marine Tuscaloosa and Gothic shale; thus, this probably does not represent a dominant, connected pore type. Pores are irregular due to the packing of the individual pyrite crystals and precipitation around the crystals.

Pore Network Statistics

Figure 12 presents relative frequency histograms of pore throat areas (ta; Fig. 12A), pore body volumes (pv; Fig. 12B), and ratios of the equivalent circular radii of the pore throat areas and pore body volumes (rt/rb; Fig. 12C), compiled from the 3DMA-Rock analysis. The histograms are organized from left to right in order of proximal to more distal, deeper water (i.e., lower energy) depositional environments. The abscissa used a log scale to facilitate examination of the large range in values. Bin sizes are the same for a particular row. The count number (e.g., n = 96) on each histogram reflects differences in pore structure as the same sample volume (i.e., 101.5 μm3) is used for all the pore network reconstructions. The histograms do not include pore throats and bodies that were unresolved (i.e., below the resolution) by the FIB-SEM imaging: BSE imaging in this study resolved features (i.e., pores) as large or larger than the 15.6 nm pixel size.

The histograms of the pore throat areas (Fig. 12A) generally exhibit positive skewness (i.e., the tail of the distribution is longer on the right side than the left) except for Marine Tuscaloosa sample 7925.5 organic, which focused on an organic-rich region (see Fig. 7), and both Gothic shale samples. The organic material in the Gothic Shale contains the majority of the pores; thus, the pore throats in the organics have different distributions, which may favor uniform or possibly log normal distributions (see sample 5390.8A in Fig. 12A). The samples with dominant pores in organics also have relatively lower count numbers, except for inorganic-focused sample in the Marine Tuscaloosa (MT 7925.5 inorganic). The histograms of pore body volume (Fig. 12B) have generally log normal distributions, except for the samples that have pores mainly in the organics (Fig. 12B, MT 7925.5 organic and GS [Gothic Shale] 5390.8B). The histograms for the ratios of the equivalent circular radii of pore throat areas and pore body volumes (Fig. 12C) have generally log normal distributions for the Kirtland samples, the inorganic sample of the Marine Tuscaloosa, and the Gothic shale sample 5390.8A (with some skewness) and uniform distributions for the lower Tuscaloosa Group samples. Two of the samples with pores mainly in the organics, the Marine Tuscaloosa 7925.5 organic and Gothic shale 5390.8B, have low count numbers and distributions that differ from those of the other samples. The circular radii histograms (Fig. 12C) indicate the degree of departure of the pore network from a cylindrical tube model (i.e., values <1 are more strongly dominated by bulges and constrictions than tubes of uniform diameter; see Dullien, 1992).

To investigate the relation between pore size and connectivity, Figure 13 presents plots of joint relative frequency and marginal relative frequency histograms of coordination number and pore body volume. Joint relative histograms give the number of occurrences of different pore body volumes associated with a particular coordination number, normalized by the total count number for a particular sample. Each row in the joint plot can be thought of as the distribution of pore body volumes for a given coordination number. The relative frequency of occurrence of a given range of pore body volumes at a given coordination number is indicated by shading from light (low frequency) to dark (high frequency). The two marginal histograms are, respectively, the summation of the relative frequencies of the rows and columns in the joint relative histogram, which give the univariate relative frequency histograms of pore body volume and coordination number. The pore body volume marginal histograms are equivalent to those of Figure 12A, although they are plotted slightly differently (i.e., bins sizes are different and the range is limited to the data) for simplicity in presentation.

For sample upper Kirtland 2047.9B (Fig. 13A), two clusters of high frequency (relatively dark areas) occur for values of log pore volume of ∼6.2 nm3, at coordination numbers of 0 and 2. This shows that the most frequently occurring pores exist both as isolated pores and as connected networks. The highest frequency in the lower Kirtland sample (Fig. 13C) is similar, occurring at a slightly larger pore volume and coordination number of 2. This suggests that conductivity in the clay matrix of the upper Kirtland and carbonate-rich portions of the lower Kirtland may not be controlled by the smallest pore bodies resolvable by the FIB-SEM imaging, but rather that the imaging method can be useful for study of transport properties in this mudstone. A similar conclusion can be reached for the lower Tuscaloosa Group samples (Figs. 13C, 13D), although the most frequent pore volumes are slightly smaller and coordination numbers differ (2–3 for area 1 in Fig. 13C and 1 for area 2, Fig. 13D). The Marine Tuscaloosa inorganic (clay matrix) sample (Fig. 13F) shows no connectivity for the most frequent pore volume, although the number of pores sampled in the interrogated mudstone volume was small compared to the lower Tuscaloosa Group and Kirtland samples.

A different picture emerges for the organic pores of the Marine Tuscaloosa (Fig. 13E) and Gothic samples (Figs. 13G, 13H). The Marine Tuscaloosa pore volumes, showing a uniform distribution, lack any connectivity at the scale of the interrogated volume. Gothic sample 5390.8A, with a slightly larger pore volume at the highest frequency, also shows little connectivity. Gothic sample 5390.8B displays frequent relatively larger pore volumes with no connectivity, but also a few smaller pore volumes with higher connectivity, with coordination numbers of 1 and 2.

For the upper Kirtland sample, connectivity may be governed by the touching of tips of the elongate type I pores that are similarly oriented (Figs. 3C, 3D). The lower Kirtland, and area 1 of the lower Tuscaloosa Group (Figs. 13B, 13C) have dominant fracture or stylolitic-like pores, types V and VI, which have peak coordination numbers of 2 and 3. The pores in organics for the Marine Tuscaloosa 7925.5 sample (Fig. 13E) are dominated by unconnected pores. The up-to-the-right diagonal trend on the joint plot of the Gothic 5390.8A sample probably indicates a scaling relationship between connectivity and pore body size, but this is not explored further here.

Graphs of cumulative mercury saturation versus pressure or Washburn pore radii for the 10 samples, corrected for closure pressure, have a range of shapes that cover approximately two orders of magnitude of pressure (Figs. 14 and 15), reflecting a broad distribution of pore throat sizes as based on the cylindrical “bundle of tubes” model (Diamond, 2000). As seen in Figure 12, however, the pore-throat-to-body ratio indicates that pores are not cylindrical tubes (for most samples), so that applying the bundle of tubes model is a simplification at best.

Upper Kirtland curves differ the most from the other samples in terms of shape and range of pressure values. There is a broad shoulder of lower pressures possibly due to lacunar pores associated with silt- to sand-sized grains (see Fig. 2; Fies, 1992). The Gothic shale curves have the highest pressure values for the corresponding mercury saturations and the highest breakthrough pressure (these are compiled for all samples in Table 2), whereas upper Kirtland samples have the lowest values (see discussion of Pore Networks and Sealing Quality).

Figure 15 compiles MIP data for the suite of mudstone samples. Linear relationships in log-log space exist for portions of the cumulative volumetric pore-density distributions (Figs. 15B, 15D, 15F), suggesting power law scaling of pore throat sizes, in possible agreement with relationships in the FIB histograms (Fig. 12). Incremental volumetric pore density distribution curves (the slope of the cumulative mercury intrusion curves; Figs. 15B, 15D, 15F) are unimodal for the Kirtland, Gothic, and one Marine Tuscaloosa sample. The other Marine Tuscaloosa and both the lower Tuscaloosa Group samples show bimodality. All modes occur at <10 nm (Figs. 15B, 15E, 15H).

Comparison of MIP and FIB-SEM Pore Properties

As MIP and FIB data sampled different numbers of pore bodies, a direct comparison of the two data sets in terms of pore volumes is not possible. MIP data do not give direct information on pore body sizes (Meyer and Klobes, 1999). Thus, we present the two data sets in a manner more conducive to investigate scaling relationships, via a different type of pore size distribution (PSD) plot. Figure 16 presents MIP data with y and x axes given by the following, respectively:
where ΔVHg-i is the sorted ith incremental mercury intrusion volume for a pressure step; the incremental volumes were sorted from largest to smallest volumes; VHg-total is the total intruded mercury volume. The new PSD plots cumulative sorted incremental volumes per total volume, a nonstandard form of saturation, by the corresponding incremental volumes. Thus, only pore volume information is presented in the plot without explicit reference to entry throats. FIB data are plotted by similarly sorting the FIB-derived pore body volumes from largest to smallest, dividing by the total volume of all the FIB-derived pore bodies, and then plotting cumulative pore body volumes versus the corresponding sorted pore body volume. The y and x axes for the new FIB PSD are given by, respectively:
where pv-i is the ith sorted pore body volume, and pv-total is the summation of all the pore body volumes.

Closure pressure corrections probably do not affect the comparison of the MIP and FIB data in terms of the slope of the PSD curves. The same closure volume of mercury was subtracted from all ΔVHg-i values of equation 1; thus, the slope of the MIP PSD curve would not change. Closure corrections affect the calculation of the largest pore throats of the pore throat size distributions (given in MIP discussion). Induced pore bodies (i.e., not found at depth under in situ conditions) include features such as cracks and/or fractures due to drying out of the samples or unloading. Both MIP and FIB estimates can include these induced pores, but we have made no attempt to remove possible induced pores from the FIB-SEM pore body data.

The slopes of the linear portions of the MIP and FIB PSD curves for sample upper Kirtland 2049.7B are similar, –0.359 and –0.394, respectively. This suggests that the MIP and FIB sampled similar pore bodies, and that they both exhibit power law scaling. The scaling probably relates to the slit- and/or sheet-like type I pores that dominate the FIB sample. The lower Kirtland 2692.9A has different slopes for the MIP and FIB data, suggesting that the FIB did not sample a representative pore type (Fig. 16A). Correspondence between FIB and MIP slopes for the Tuscaloosa Group is low, probably due to the low number of pores sampled in the FIB data sets. The Gothic shale MIP PSDs have unique shapes, probably resulting from the extremely high capillary pressures, which may be due to pore structure alteration during the MIP test. In summary, the small volumes imaged by FIB-SEM were sufficient to capture representative pore statistics as seen by MIP analysis only in the case of the upper Kirtland.

Mudstone textural properties (i.e., pore and grain size, shape, connectivity) can be subdivided into four categories of influence: type of framework (i.e., framework supported or matrix supported); degree of diagenesis (e.g., pores are filled with cement, secondary, or created by dissolution); sorting (e.g., well sorted or poorly sorted, relative abundance of silt and clay); and compaction (e.g., pore volume loss through compaction processes) (Katsube and Willamson, 1994). Textural parameters that vary as a function of these categories are typically macroscopically measured (i.e., effective porosity, permeability, tortuosity, and specific surface area) (Katsube and Willamson, 1994). From our 3D imaging of a variety of mudstone types, it is evident these parameters are macroscopic manifestations of the collection of pore body types (described in Summary of Pore Types) and the connecting pore throat radii such as imaged in Figures 3–10.

Although a full comparison of imaged pore types and macroscopic properties is beyond the scope of this paper (due to the complexity of pore types and the large degree of heterogeneity of the rocks at all scales), it is valuable to examine depositional controls (i.e., framework and sorting) and diagenetic controls (cementation, compaction, and dissolution) on the pore networks as well as MIP-estimated sealing efficiency. This information is directly applicable to subsurface CO2 storage, as an understanding of the sealing processes could enable proper design of a CO2 sequestration project with regard to performance of a proposed mudstone seal. In addition, it is applicable to other studies on multiphase flow in low permeability rock, such as efforts in recovering hydrocarbons from shale gas reservoirs. For our purposes, we equate sealing efficiency with MIP breakthrough pressures (see discussion of MIP in Results).

Framework and sorting correlate with depositional environment. Poorly sorted mudstones with a high framework-supporting silt content exist proximal to sediment sources and would be expected to have pore types and resulting macroscopic properties much different from more well-sorted clay-rich mudstones occurring more distal to sediment sources. In Figure 17, we place our mudstone samples into the shale lithofacies depositional succession devised by Schieber (1999); the lithofacies designations, as well as those discussed by Almon et al. (2005), are given in Table 3. Dominant pore types as imaged by FIB-SEM are also listed in Table 3.

The mudstone-rich portions of the Kirtland Formation, interpreted as a floodplain, overbank, and crevasse-splay lithofacies, reflect deposition in shallow water depths (∼1–2 m), proximal to sediment source. With abundant mottled texture and pedogenic features, it is equivalent to the red-gray mudstones of Schieber (1999). It represents the most proximal to source and most poorly sorted of the mudstones investigated. Note that some of the clay fabric is suggestive of compacted analogues of the clay floccule textures discussed by Schieber et al. (2007); consequently, some caution is required when comparing depositional paleoenvironments of muddy sediments. The high silt content is likely responsible for the low breakthrough pressures (Tables 2 and 3), which are likely due to possible lacunar pores occurring in silt-clay mixtures. The higher breakthrough pressures are probably a macroscopic manifestation of abundant networks of type I pores observable in Figure 3.

The lower Tuscaloosa Group, interpreted as a bioturbated shallow shelf deposit, represents deposition in ∼10 m of water (Billingsley, 1980). With abundant burrowed features and laminations, it is probably equivalent to the moderately bioturbated graded mudstone lithofacies of Schieber (1999). Breakthrough pressures for the lower Tuscaloosa Group are between the two breakthrough pressure values for the Kirtland, reflecting connectivity via a network of type I and type IV pores. The existence of type IV (microfracture) pore types probably enhances connectivity over the poorly connected network of type I pores. The formation of some type II pores (Fig. 6A) may be due to the development of authigenic chlorite.

The more shelfal Marine Tuscaloosa Formation reflects a deeper water environment, being deposited during the transgressive-regressive cycle (T-R K5) maximum regression during the Late Cretaceous (Mancini and Puckett, 2005). Being well laminated and organic rich, the Marine Tuscaloosa samples are probably equivalent to the laminated mudstone lithofacies of Schieber (1999) or microfacies 2 (organic laminated shales) of Almon et al. (2005). The relative lack of a connected network of type I pores probably is responsible for the high breakthrough MIP pressures.

The Gothic shale is an open-marine, euxinic, massive, black, organic-rich mudstone, which was probably deposited below wave base in water at least tens of meters deep (Fig. 16; Goldhammer et al., 1994). It contains the highest TOC of the mudstone types we examined. As such, it is the closest to belonging to the classic black shale lithofacies, equivalent to the carbonaceous mudstone lithofacies of Schieber (1999) or microfacies 1 (massive organic shale) of Almon et al. (2008). The dominant pore types are largely contained in organics (type IVb), which are dominantly located in the compaction shadows between larger quartz, feldspar, or other clasts or grains (Fig. 9). The small amount of porosity and very poor connectivity account for the highest breakthrough pressures of the studied mudstones (Table 3).

The maturity of organic matter may be responsible for the differences in pore types of the organic phases in the high TOC mudstones (type IVa in the Marine Tuscaloosa and type IVb in the Gothic shale). The Marine Tuscaloosa organics are at a low level of conversion, whereas the Gothic shale organics are mature. Pore type IV may factor in the generation and migration of petroleum in these rock types (Loucks et al., 2009). The Gothic shale appears to have undergone the greatest amount of compactional pore volume loss. That, and the seeming lack of type I pores, is responsible the high breakthrough pressure values for the Gothic shale (Table 3).

In summary, pore type I and a lack of well-connected pore types V and IV result in high breakthrough pressures for lower Kirtland, lower Tuscaloosa Group, and Marine Tuscaloosa samples. High breakthrough pressures for the Gothic shale are due to organics filling what otherwise might have been larger, connected pores. Pore type II, based on the evidence of its localized nature in lower Tuscaloosa Group samples, may be involved in the high breakthrough pressures as mercury intrudes from surrounding slit pores in the clay fabric. Based on matrix pore types imaged herein, we suggest that more distal mudstones make better sealing lithologies (i.e., caprocks) as a result of their higher breakthrough pressures.

The 3D visualization and quantitative analysis of small (101.5 μm3) samples of five mudstones from continental to deep-marine depositional settings revealed seven distinct pore types. The most dominant pore type is a slit- and/or sheet-like pore in clay-rich mudstones, designated type I following Desbois et al. (2009). It is elongate with sharp to curved tips and is located with parallel clay sheets. The associated coordination numbers of typically 1–4 are governed by the sheet-like geometry. Small pore throat connections between these pores limit conductivity and contribute to high (mercury) breakthrough pressures and concomitant high sealing capacity. These pore sizes scale over approximately an order of magnitude, possibly as a power law, as illustrated by MIP and FIB data (Fig. 15). Pore type I is expected in all argillaceous mudstones and may be predominant in mudstones with abundant expandable clays, as in the Kirtland Formation. The extent to which pore type I morphology is influenced by coring and concomitant depressurizing and possible dehydration is unclear; however, these pores may indicate the possibility for pore formation or enlargement if desiccation and/or dewatering in the subsurface were to occur (e.g., during injection of dry CO2).

Other pore types, such as types II (non-slit-like pores in clays) and III (jagged pores in compaction shadows), are associated with authigenic clay minerals and deformation (compaction shadows). They probably contribute to increased mercury breakthrough pressure if they result in an overall reduction in pore throat sizes (chlorite as a pore-filling mineral). Secondary (diagenetic) microfractures and stylolitic features (pore types V and VI) may not be present in situ; however, as imaged by the FIB and TEM (Heath, 2010), they may still reflect preexisting features in the subsurface that may respond to changes in stress, as a perturbation due to CO2 injection or hydraulic fracturing for shale gas production is applied to the system. Type VII, the pores in pyrite framboids, do not have strong connectivity to other pore types outside of the framboids.

Pore types IVa and IVb are examples of pores in organic phases. It is intriguing that many mudstones could have transport pathways that are both water (clay-lined pores) and oil (organics lined) wetting. These pores in our studied samples, however, are small and generally unconnected for the Marine Tuscaloosa and Gothic shale mudstones examined here. In addition, these limited pore networks do not have the scaling relationships of the pores in clay-rich portions of mudstones.

Depositional environment and burial history are both strong geologic controls on pore network properties that bear on the ability of mudstones to serve as sealing lithologies for subsurface CO2 storage. More distal, organic-rich mudstone lithofacies may make the best seals, barring any deleterious effect associated with CO2-organic interactions. Understanding how pore network topology and geometry vary as a function of depositional environment would provide an important starting point for predicting sealing behavior of mudstones. On the basis of our data set, we cannot separate depositional environment, burial history, and diagenesis independently on the basis of their effects on sealing efficiency, although a clear influence of there on pore types and connectivity has been demonstrated Unraveling geologic controls on sealing efficiency of mudstones is a relevant topic for future research, given that a priori performance of mudstone caprocks for carbon storage may be needed in portions of basins lacking in subsurface characterization.

Focused ion beam–scanning electron microscopy (FIB-SEM) imaging and subsequent three-dimensional image analysis was funded by the U.S. Department of Energy (DOE) Office of Basic Energy Sciences, Division of Chemical Sciences, Geosciences, and Biosciences. Rock samples were provided by the Southwest (SWP) and Southeast Regional Carbon Sequestration Partnerships, which are managed by the DOE National Energy Technology Laboratory. The SWP funded mercury intrusion porosimetry (MIP) analyses. We also thank the Southeast Regional Carbon Partnership and Richard Esposito of the Southern Company for making the Tuscaloosa cores from the Mississippi Power Company #1 available for viewing and core description, as well as for providing rock samples and access to core data.

We thank Joseph Michael and Michael Rye of the Materials Characterization Department at Sandia National Laboratories for performing FIB serial sectioning and image acquisition. John Neasham of Poro-Technology performed MIP measurements. TerraTek, a Schlumberger company, aided in petrographic work on Kirtland Formation and Gothic shale thin sections. Brent Lindquist provided assistance with installation and use of the 3DMA-Rock software. We thank Fabian Duque-Botero, William Dawson, and William Almon of the Seal and Trap Team at Chevron Energy Technology Company, who shared concepts of performing caprock analysis in a sequence stratigraphic framework. Dr. Duque-Botero provided assistance with interpretation of MIP data. We thank Susan Delap Heath for drafting expertise.

Sandia National Laboratories is a multi-program laboratory managed and operated by Sandia Corporation, a wholly owned subsidiary of Lockheed Martin Company, for the U.s. Department of Energy's National Nuclear Security Administration under contract DE-AC04-94AL85000.