Texture and anisotropy analysis of Qusaiba...

Geophysical Prospecting, 2011, 59, 536–556 doi: 10.1111/j.1365-2478.2010.00942.x

Texture and anisotropy analysis of Qusaiba shales

Waruntorn Kanitpanyacharoen1, Hans-Rudolf Wenk1∗, Frans Kets2,3,Christian Lehr3 and Richard Wirth4

1Department of Earth and Planetary Science, University of California, 307 McCone Hall, Berkeley, CA 94720, USA, 2School of Earth andEnvironment, University of Leeds, Leeds LS2 9JT, UK, 3Shell International Exploration and Production BV, Kessler Park 1, 2288 GSRijswijk, the Netherlands, and 4Geoforschungszentrum, Telegrafenberg, 14473 Potsdam, Germany

Received January 2010, revision accepted November 2010

ABSTRACTScanning and transmission electron microscopy, synchrotron X-ray diffraction, mi-crotomography and ultrasonic velocity measurements were used to characterize mi-crostructures and anisotropy of three deeply buried Qusaiba shales from the Rub’al-Khali basin, Saudi Arabia. Kaolinite, illite-smectite, illite-mica and chlorite showstrong preferred orientation with (001) pole figure maxima perpendicular to the bed-ding plane ranging from 2.4–6.8 multiples of a random distribution (m.r.d.). Quartz,feldspars and pyrite crystals have a random orientation distribution. Elastic propertiesof the polyphase aggregate are calculated by averaging the single crystal elastic prop-erties over the orientation distribution, assuming a nonporous material. The averagecalculated bulk P-wave velocities are 6.2 km/s (maximum) and 5.5 km/s (minimum),resulting in a P-wave anisotropy of 12%. The calculated velocities are compared withthose determined from ultrasonic velocity measurements on a similar sample. In theultrasonic experiment, which measures the effects of the shale matrix as well as theeffects of porosity, velocities are smaller (P-wave maximum 5.3 km/s and minimum4.1 km/s). The difference between calculated and measured velocities is attributed tothe effects of anisotropic pore structure and to microfractures present in the sample,which have not been taken into account in the matrix averaging.

Keywords: Anisotropy, Clay minerals, Preferred orientation, Shale.

INTRODUCTIO N

Shales compose a large part of sedimentary basins and areof great interest as cap rocks of hydrocarbon reservoirs (e.g.,Aplin and Larter 2005), in the context of carbon sequestra-tion (e.g., Chadwick et al. 2004) and as repositories for nu-clear waste (e.g., Mallants, Marivoet and Sillen 2001; Bossartand Thury 2007). Organic rich shales are also important assource rocks in petroleum formation and occurrence (e.g.,Tissot and Welte 1984). Constituent clay minerals are phyl-losilicates that acquire preferred orientation during sedimen-

∗E-mail: [email protected]

tation and compaction. Sedimentation and compaction alsolead to a well-developed bedding foliation and the combinedeffects are observed as anisotropy of texture dependent prop-erties such as permeability and acoustic wave propagation(as observed in seismic prospecting). Anisotropy is directlylinked to preferred orientation patterns of component min-erals, especially phyllosilicates with high single crystal elasticanisotropy (e.g., Militzer et al. 2011). Due to their small grainsize and poor crystallinity, it has been difficult to quantify pre-ferred orientation (or texture) of clays in shales with classicalmethods such as pole figure measurements, optical methodsand electron microscopy, though several studies have pro-vided important data (e.g., Aplin et al. 2006; Day-Stirrat et

al. 2008; Ho, Peacor and van der Pluijm 1999; Valcke et al.

536 C© 2011 European Association of Geoscientists & Engineers

Texture and anisotropy analysis of Qusaiba shales 537

2006). New methods that rely on synchrotron X-ray diffrac-tion have been shown to be able to successfully characterizecomposition, textures and microstructures of natural and ex-perimentally produced shales (e.g., Lonardelli, Wenk and Ren2007; Wenk et al. 2008a; Voltolini et al. 2009). From theseX-ray diffraction studies, it became apparent that there is alarge range of variability in patterns that depends on com-position, compaction and diagenesis. The range of samplesthat have been analysed by this method is still very limited. Inthis study we apply this type of orientation analysis to lowerSilurian diagenetic shales from Saudi Arabia.

Samples were first analysed with scanning and transmis-sion electron microscopy for microstructure and composition.They were then investigated with X-ray microtomography for3D particle distribution, including porosity. After conductingsynchrotron X-ray diffraction experiments, a Rietveld diffrac-tion image analysis was applied to quantify preferred orienta-tion patterns (or texture) of the constituent minerals. This isthe focus of this paper. Based on textural information, elas-tic properties of the mineral polyphase aggregate constitutingthe rock matrix were calculated by averaging single crystalelastic properties over the experimentally determined orien-tation distribution. The calculated compressional and shear-wave velocities of the mineral polyphase aggregate have thenbeen compared with velocities determined experimentally inultrasonic measurements in the laboratory and with velocitiesmeasured as part of the petrophysical measurements in theborehole.

SAMPLES

We analysed three samples from a well drilled in south-westSaudi Arabia. In this paper, the samples are referred to as Qu1,Qu2 and Qu3. The shale samples are from a depth of 3566m and are taken from the Qusaiba member of the QalibahFormation. This Lower Silurian member is the main sourcerock for hydrocarbons in the Palaeozoic section of the Rub’al-Khali basin (Al-Harbi 1998; Schenk, Pollastro and Ahlbrandt2002). It is mainly composed of claystones interbedded withsiltstones and mudstones and consists of various layers. In-formation obtained from the South Rub Al-Khali Company(SRAK) documents shows that the sample is from a leanportion of the Qusaiba, with total organic carbon (TOC) of1.5%. The maturity of the organic material is fairly low, withmaturity in the range from non-mature to early oil mature.However, the shales themselves are considered to be maturerocks with high cementation. The shales have reached temper-atures of approximately 100◦ C and diagenesis occurred over

450 million years. The shales are firmly cemented and havea low porosity. The samples studied are very fine-grained,medium to dark grey, hard and quite heterogeneous. They arevisibly anisotropic, with horizontally deposited mica plateletsand layers of kerogen rich material in a matrix of cementedmudrock.

EXPERIMENTAL METHODS

Scanning electron microscopy

A polished slice of the shale sample Qu2 was coated withcarbon and examined with a Zeiss Evo MA10 low vacuumscanning electron microscopy (SEM) equipped with an EDAXEnergy-Dispersive Spectroscopy (EDS) system (UC Berkeley).The SEM was operated with an accelerating voltage of 30kV and a probe current of 20 nA to collect images. Theshale was analysed for mineralogical composition and mi-crostructure. A backscattered (BE) SEM image shows complexmicrostructures of authigenic illite-smectite and kaolinite, de-trital illite-mica platelets and some coarse-grained quartz andpyrite with a well-developed horizontal bedding plane (Fig. 1).The brightness variation of the BE image, ranging from low(black) to high (white) is due to the contrast in atomic number,with high atomic numbers appearing bright. EDAX GenesisImaging/Mapping software was used to collect compositionalmaps for Fe, S, Mg, Si, O, C, Al, K and other elements.Only maps for Fe, S, Mg and Si are displayed in Fig. 2,

Figure 1 A backscattered SEM image of sample Qu2 showing mi-crostructures of different phases. Brightness variations indicate theatomic number: pyrite (white), quartz and clay minerals (intermedi-ate grey shades) and porosity (black).

C© 2011 European Association of Geoscientists & Engineers, Geophysical Prospecting, 59, 536–556

538 W. Kanitpanyacharoen et al.

Figure 2 EDS chemical analysis showing elemental distribution maps of Fe (yellow), S (aquablue), Mg (magenta) and Si (green) of sample Qu2(similar area as Fig. 1).

suggesting an abundance of detrital angular quartz (SiO2),Fe- and Mg-containing detrital mica and pyrite (FeS2).

Transmission electron microscope

Microstructures were also investigated with the transmis-sion electron microscope (TEM) with a much higher reso-lution than the SEM. TEM, analyses were performed usinga FEI Tecnai G2 X-Twin transmission electron microscope(TEM/AEM) equipped with a Gatan Tridiem energy filter,a Fishione high-angle annular dark field detector (HAADF)and an energy dispersive X-ray analyzer at GFZ Potsdam. Toexclude preparation induced damage, the samples for TEMstudies were prepared with a focused ion beam (FIB) device(FEI FIB200TEM) at GFZ. TEM foils with dimensions of15 μm × 10 μm × 0.15 μm were sputtered from powdersembedded in epoxy using accelerated Ga-ions (Wirth 2009).

Figure 3(a) shows a low magnification scanning transmis-sion electron microscope (STEM) image of the FIB slice ofthe sample that was investigated. The advantage of FIB sam-ple preparation is that the amount of surface defects that areintroduced is minimal, which is particularly important forthe characterization of pores. Figure 3(b) is a TEM brightfield image of a region with an equiaxed quartz grain sur-rounded by phyllosilicates. Notice the abundance of elongatedmicropores (bright), 50–150 nm in length and <10 nm wide.Figure 3(c,d) shows HAADF-STEM images of two regionsrich in phyllosilicates. The brightness in these images corre-sponds to the atomic number. In Fig. 3(c), there are againelongated nanopores parallel to adjacent phyllosilicate flakes,as well as grain boundaries (in this imaging mode the pores areblack). The bright pyrite octahedron in the lower right cornercan be compared with those identified by SEM (Fig. 1). Fig-ure 3(d) shows a kinked stack of phyllosilicates. The kinkingprobably formed due to stresses during compaction. Also here



Figure 3 TEM images of microstructures in Qusaiba shale. a) Sample slice cut perpendicular to the bedding plane, obtained by FIB sputtering.Bedding plane is vertical. b) TEM brightfield image of a quartz grain surrounded by phyllosilicates. Nanopores are bright. c) HAADF STEMimage of phyllosilicates and a pyrite grain (bright) with nanopores (black). d) HAADF STEM image of a kinked stack of phyllosilicates, indicativeof deformation during compaction.

one sees pores parallel to phyllosilicate (001) planes as well asadditional pores along the kink boundaries.

Microtomography

Small prisms were prepared and ground to cylinders 1 mmin diameter and 5 mm in height. These cylinders were in-vestigated with synchrotron X-ray radiation tomographyat the TOMCAT (TOmographic Microscopy and Coherent

rAdiology experimenTs) beamline of the Swiss Light Source(SLS) at the Paul Scherrer Institute in Villigen, Switzerland.Energy of the X-ray radiation is 16.0 keV, correspondingto a wavelength of 0.775 A. The material was investigatedwhile scanning with a voxel size of 0.38 μm × 0.38 μm× 0.38 μm. Rotating the sample in 0.3◦ increments aboutthe cylinder axis, 1500 projections were recorded. Datawere first processed with local software used at the SLS.Figure 4 shows a 2D slice through the centre of sampleQu2. The information content is comparable to that of the



Figure 4 Microtomography 2D slice of sample Qu2. High absorption minerals are bright (pyrite), intermediate shades are grey (quartz and clayminerals) and low absorption phases are very dark grey to black (pores, fractures and kerogen).

Figure 5 Tomography 3D images showing a) low absorption (porosity and microfractures) and b) high absorption (pyrites). This only illustratesa small segment (150 μm × 150 μm × 150 μm) of the total array.

backscattered SEM image (Fig. 1), indicating a resolution ofbetter than 1 μm for the tomographic reconstruction. A cubicvolume of 150 μm × 150 μm × 150 μm was selected from thearray to represent the volumetric data set and characterized bythe software Avizo (Visualization Sciences Group). Segmen-tation was done based on the threshold with correspondingx-ray absorption intensity values. Low absorption features(dark) with values ranging from 10 043–24 800 are indicativeof porosity, fractures and/or kerogen (Fig. 5a). The highly ab-sorbing particles (white) with values ranging from 44 000–65 535 are pyrite (Fig. 5b), intermediate shades representquartz, feldspars and clay minerals. Pyrite exists in the formof individual octahedral crystals (5–20 μm) comparable to theTEM image (Fig. 3d) and as fine-grained clusters of small oc-

tahedral crystallites (<1 μm). The distribution seems randomand not to be linked to the bedding plane. Porosity (includingfractures and kerogen), on the other hand, is anisotropic, or-ganized mainly parallel to bedding and with little connectivityof the flat pores in the direction perpendicular to the beddingplane. A few horizontal fractures are observed. Analysis soft-ware implemented in Avizo was also used to extract statisticaland numerical information such as volume and surface area.The volume fractions of porosity and pyrite are estimated at1.1 vol% and 0.7 vol%, respectively. Nanometer size poresare beyond the resolution of microtomography. The poros-ity derived from tomography images is lower than the valuederived from a mercury injection capillary pressure experi-ment, where 2 vol% can be attributed to pores in nm size and



Figure 6 A sketch illustrating the geometry of the synchrotron diffraction experiment. The sample is rotated around an axis perpendicular tothe incident X-ray to improve pole figure coverage. The Debye diffraction rings are recorded with a CCD detector.

1 vol% to fractures and pores in μm or tens of μm size. Thisdoes not include isolated pores on a nanometer scale that werenot detected by mercury injection capillary pressure but arevisible in TEM images (Fig. 3b–d). Ambiguities of the mer-cury injection capillary pressure method have been discussedby Diamond (2000).

X-ray texture analysis

We prepared three samples for hard synchrotron X-raydiffraction analysis by first embedding the shale in low-temperature hardening epoxy resin in a plastic cup overnightto produce an epoxy cylinder roughly 2 cm in diameter. Withaid of kerosene, the shale-containing cylinders were carefullycut with a diamond saw and polished into slices, approxi-mately 2 mm thick, suitable for X-ray diffraction experiments.The diffraction measurements were done at the BESSRC 11-ID-C beamline of the Advance Photon Source (APS) at Ar-gonne National Laboratory. Details about instrumental setupand procedures can be found in Wenk et al. (2008a).

A monochromatic X-ray beam with a wavelength of0.10779 A and a diameter of 0.5 mm was used. The highX-ray energy ensured high penetration with only minor ab-sorption. The sample slice was mounted to an aluminiumrod roughly perpendicular to the bedding plane on a go-niometer with a horizontal x-axis (Fig. 6). During exposure,the sample was translated parallel to the x-axis to averageover five different spots in 2 mm increments to obtain a

representative sample volume. Diffraction images wererecorded with a Mar345 image plate detector (3450 × 3450pixels), positioned at about 2 m from the sample. Typical im-ages, recorded on three samples in 60 s, are shown in Fig. 7,with the figures processed by Fit2D (Hammersley 1998). Im-ages record a 2θ angle range from 0–4.6◦. Intensity varia-tions along some Debye rings immediately indicate the pre-ferred orientation of corresponding lattice planes hkl. A singlediffraction image contains information about the orientationof lattice planes with poles on a circle when represented ona pole figure (#1 in Fig. 8). The pole figure coverage wasimproved by tilting the sample around the horizontal axisand recording seven diffraction images between ω angles –50◦

and +50◦ (at –50◦, –30◦, –10◦, 0◦, 10◦, 30◦, 50◦, #1–7 inFig. 8). Overall, the diffraction images of the three samplesare very similar. Some of the phases also exhibit strong pre-ferred orientation as observed in the variation of intensityalong the diffraction rings. However, the diffraction patternof Qu3 displays a noticeably stronger texture than Qu2 andQu1.

Before the diffraction images of the shale sample were anal-ysed, the instrument geometry (wavelength, sample-detectordistance, beam centre and image plate tilt) was calibrated witha CeO2 powder standard. The diffraction images obtained onthe Qusaiba shale, ranging in azimuth from 0–360◦ were di-vided into segments and integrated over 10◦ intervals to pro-duce 36 spectra, which represent distinctively oriented latticeplanes. Figure 9 (bottom: EXP) displays a stack of these spec-tra for each sample a) Qu1, b) Qu2 and c) Qu3 for the 0◦



Figure 7 Diffraction images showing variation of intensity along Debye rings, which indicate preferred orientation of kaolinite (001), illite-smectite (001), illite-mica (002) and chlorite (001). The quartz (101) diffraction ring displays uniform intensity, indicating lack of preferredorientation. The samples are inclined arbitrarily relative to the bedding plane.

tilt image. It clearly shows peak intensity variations with theazimuth. A Q range of 0.37–3.70 A−1 (d-spacing 1.80–16.98 A) was used for the refinement. Note that we ex-press spectra as a function of Q = 2π /d rather than d

(lattice spacing), where everything becomes compressed to-wards small spacings, or θ (scattering angle), which dependson wavelength. For each sample (Qu1, Qu2 and Qu3), we pre-pared 252 spectra (7 images × 36 spectra) and then processedthe data with Material Analysis Using Diffraction (MAUD)(Lutterotti et al. 1997), a very flexible Rietveld code (Rietveld1969) that allows quantitative texture analysis. In the Rietveldrefinement, one uses a least-squares approach to minimize thedifference between experimental diffraction data and a cal-culated diffraction model. The calculated model is definedby several factors such as instrumental parameters, scatteringbackground, crystal structure, microstructure, weight fractionof each phase and its preferred orientation. This technique al-lows to successfully resolve overlapping peaks of multiphasesamples. The spectrum of Qusaiba shale is indeed composedof over seven mineral phases (Fig. 10, bottom).

In the Rietveld refinement, parameters of the crystallo-graphic structures are required. We used parameters for tri-clinic kaolinite from Bish (1993), parameters for triclinicchlorite-penninite from Joswig et al. (1980), for monoclinicillite-mica from Gualtieri (2000) and for monoclinic illite-smectite (based on a muscovite-phengite composition) fromPlancon, Tsipurski and Drits (1985) by importing correspond-ing crystallographic information files from the American Min-eralogist Database and we used quartz, K-feldspar and pyritestructures from the data base contained in MAUD. Mon-

oclinic phyllosilicate phases are generally described in the‘second setting’ with b = [010] as the unique axis and (001)as the cleavage plane. However, for texture analysis the firstsetting (c = [001] as the unique axis and (100) as the cleavageplane) has to be used (Matthies and Wenk 2009) and this re-quires some transformations. Our labels on figures and valuesin Table 2 refer to the second setting. The instrumental pa-rameters (detector distance and orientation, wavelength etc.)were entered into MAUD and calibrated with the CeO2 stan-dard. Subsequently, 252 spectra of the sample were importedand refined with polynomial background functions and scaleparameters for each image in 21 consecutive cycles. Phase pa-rameters such as volume fraction and lattice parameters wererefined but atomic coordinates were kept constant. The peakshapes and widths are governed by microstructural param-eters and thus were modelled by refining an isotropic crys-tallite size and microstrain. The data quality was not suffi-cient to use an anisotropic model for crystallite size. Textureanalysis, which is the main interest of this study, was com-puted by the EWIMV algorithm (Matthies and Vinel 1982),using 10◦ resolution for the orientation distribution deter-mination, without imposing sample symmetry. The refinedmodel diffraction spectra (top: Calc.) are compared with ex-perimental spectra (bottom: Exp.) in Fig. 9, which shows aclose similarity, indicative of an excellent fit, both in inten-sities as well as position of diffraction peaks. This is furtherquantified in Fig. 10, which shows the average spectra for the0◦ tilt images with dots for experimental data and a thin solidline giving the calculated fit. Note that many diffraction peakscontribute to these spectra. Only with the Rietveld method can



Figure 8 Pole figure coverage obtained by rotating the sample aroundthe x-axis. Each line corresponds to a single diffraction profile usedin the Rietveld refinement 1) ω = 0◦, 2) –10◦, 3) –30◦, 4) –50◦, 5)10◦, 6) 30◦ and 7) 50◦.

these be adequately deconvoluted and resolved, including theillite-smectite and illite-mica peak at Q = 0.6 A−1 (see enlargedinset).

The orientation distribution, which defines the crystalliteorientation relative to sample coordinates, was then exportedfrom MAUD and imported into the BEARTEX software(Wenk et al. 1998) to further process the orientation data.It is important to emphasize that both MAUD and BEAR-TEX rely on the first-setting (c-axis is the unique axis) for themonoclinic system. But for the pole figures (Figs 11–14), wegive Miller indices for the more conventional second settingwith (001) as the cleavage plane (see Kocks, Tome and Wenk(2000) for a detailed explanation of pole figures, which rep-resent the distribution of crystal directions relative to samplecoordinates). The orientation distribution was smoothed witha 7.5◦ filter to minimize artefacts from the orientation distri-bution cell structure. Then the sample was rotated so that polefigures are represented as projections on the bedding plane.The pole densities are expressed as multiples of random dis-tribution (m.r.d.) where a value of 1 corresponds to a randomor isotropic distribution and a high value in a particular di-rection suggests a strong orientation along that direction. Poledensities are normalized in such a way that the integral overthe whole pole figure is 1.0. In clay minerals the basal plane(001) is most significant, thus we show the (001) pole fig-

Figure 9 Map 2D plots of a stack of calculated (top) and experi-mental (bottom) diffraction spectra for a) Qu1, b) Qu2 and c) Qu3samples, corresponding to images in Fig. 7. The grey shades illustratethe intensity variation in the diffraction images.

ures and in addition (100) pole figures to establish if there areconstraints on the orientation of a-axes [100] or if they haverotational freedom in the (001) plane (Figs 11–14). The samescale is used for all phases. (001) pole figures have a strongmaximum in the centre, indicating that (001) lattice planes ofphyllosilicates are more or less parallel to the bedding plane.

Velocity measurements

Elastic wave velocities and P-wave anisotropy were exper-imentally obtained for Qusaiba shale at in situ stress con-ditions at Shell’s Geomechanics Laboratory in Rijswijk, theNetherlands. Plugs were taken from the core in the cm-dmscale. The core was quite homogeneous, with a num-ber of visible horizontal fractures. Heterogeneity and lay-ering can be observed in the mm scale. A number of



Figure 10 Average diffraction spectra, obtained by integrating over the whole Debye ring, compare experimental data (crosses) and calculatedmodels (solid line) for a) Qu1, b) Qu2 and c) Qu3 in the Rietveld refinement. The inset shows the illite-smectite (001) peak being separatedfrom the illite-mica (002) peak. At the bottom are the positions of diffraction peaks for the various mineral phases.

fractures may have originated as stress relief features andfrom poor preservation (the core had been stored under am-bient conditions for a short time), or alternatively have beenopened as a result of core extraction and sample prepara-tion. As the rock was cemented and had a long history of

diagenesis in the subsurface, the probability of substantialdamage to the texture and clay structure of the sample dueto storage conditions was considered, to be minor. A cylin-drical plug with an axis normal to the bedding plane wasdrilled from this core. Note that this sample is not identical



Figure 11 (100) and (001) pole figures for kaolinite from the OD ofQu1, Qu2 and Qu3 samples. Equal area projection on the beddingplane. Contours in multiples of a random distribution (m.r.d.).

to the samples used for the X-ray diffraction study but thesamples have been taken in close proximity of each otherand can hence be expected to have similar properties onthe cm scale. The plug measured 36 mm in diameter and52.1 mm in length. Tap water was used as a coolant whiledrilling the plug and in a later stage as a saturation fluid ofthe sample during the ultrasonic measurements. Tap waterwas considered warranted because the reservoirs adjacent tothe Qusaiba samples have near fresh water conditions. Fluidcompatibility tests under ambient conditions did not revealvisible chemical reactions, swelling or disaggregation. The ma-terial was water wet, which facilitates saturation of the sampleat the beginning of the ultrasonic test.

The velocity measurements were performed in a biaxialcompaction apparatus, where the plug is mounted betweentitanium end-caps and enclosed in a Viton

R©sleeve. Within

that sleeve, the sample is wrapped in metal gauze in orderto allow radial drainage and enhanced pressure equilibrationvia a permeable shell around the circumference of the plug.This method is widely employed in shale testing and described

Figure 12 (100) and (001) pole figures for illite-mica from the OD ofQu1, Qu2 and Qu3 samples. Equal area projection on the beddingplane. Contours in multiples of a random distribution (m.r.d.).

in detail in the literature (e.g., Hornby 1998; Jacobsen et al.

2000; Dewhurst et al. 2006; Fjaer et al. 2008).The apparatus is rated at 100 MPa with an independent

control of axial stress, confining or radial stress and pore fluidpressure, in order to mimic the in situ vertical stress, horizon-tal stress and pore pressure. Pairs of piezoelectric transmittersand receivers for ultrasonic P- and S-wave are embedded inthe end caps. In addition, P-wave transducers are mountedto the outside of the Viton

R©sleeve. The pulse transmission

technique is employed to measure elastic P-wave velocity inthe radial direction (parallel to bedding) (Fig. 15a) and P-and S-wave velocities in the axial direction (normal to thebedding plane and along the symmetry axis) (Fig. 15b–d).The measurement frequency is ∼1 MHz. The signals (am-plifier output in volts) were recorded at the ultrasonic re-ceivers after pulse transmission travelling through the sam-ple and end-cap (axial), as well as through the sample andsleeve assembly (radial). The time scale has its origin at thetrigger for the pulse transmission. The original, transmittedpulse is clearly maintained in the received signals. This allows



Figure 13 (100) and (001) pole figures for illite-smectite from the ODof Qu1, Qu2 and Qu3 samples. Equal area projection on the beddingplane. Contours in multiples of a random distribution (m.r.d.).

straightforward, accurate and reliable picks of arrival times,on which the derived velocity values are based.

Within the test apparatus, the sample was evacuated, sat-urated with tap water and after two weeks loaded stepwisewith 4 MPa steps to a maximum net stress condition, fol-lowed by a low rate simultaneous increase of pore pressureand total stress while maintaining the system at a constant netstress. Note that net stress refers to the difference between theapplied total stress and pore fluid pressure. Each step was fol-lowed by a hold period of 48 hours to equilibrate pore fluidpressure and strain throughout the sample. Standard strainversus

√time plots as introduced by Hornby (1998) were not

found to be feasible to validate that the sample has reachedequilibrium. Also, the variation in velocities was relativelysmall, less than 0.5% per 48 hours during hold periods athigh stress. Sensitivity to stress was approximately 2% per50 MPa change in net stress. The total stresses, correspondingto in situ conditions, are 80 MPa in the axial direction and66 MPa in the radial direction, with a 71 MPa mean value. Thein situ condition was chosen to characterize the sample ma-

Figure 14 (100) and (001) pole figures for chlorite from the OD ofQu1, Qu2 and Qu3 samples. Equal area projection on the beddingplane. Contours in multiples of a random distribution (m.r.d.).

terial for comparison between measured ultrasonic velocities,sonic velocities as measured in the borehole and velocities cal-culated from synchrotron X-ray analysis. The maximum netstresses selected for the experiment were 27 MPa in the ax-ial direction and 13 MPa in the horizontal direction, with an18 MPa mean value. The corresponding pore pressure was53 MPa, which is higher than the actual in situ pore pres-sure of 37 MPa – the measurement was thus taken at anoverpressure situation. Measurements at lower net stresses incombination with the measurement at 53 MPa overpressurewere used to extrapolate velocities to the values at in situ con-dition. This procedure avoids inelastic irreversible effects tobe expected at or beyond in situ conditions. Velocities werecalculated from selected arrival times and the dimension ofthe sample. Calibration has been performed with aluminiumsamples. The arrival times in calibration runs and, as shownin Fig. 15 during actual measurements, could be determinedwithout problems as data quality was good for the calibra-tion material as well as for this low porosity shale and theinterference of other arrivals was minimal.



Figure 15 The amplifier output signals from the ultrasonic receivers after pulse transmission through the sample for the a) P-wave in the radialdirection, b) P-wave and c–d) S-waves in the axial direction.

Table 1 Quantitative phase proportions in weight% fractions extracted from the Rietveldrefinement of Qu1, Qu2 and Qu3

Sample Kaolinite Illite-Smectite Illite-Mica Chlorite Quartz K-feldspars Pyrite

Qu1 28.46 19.06 13.80 4.55 22.02 8.58 3.52Qu2 25.31 22.19 20.19 3.56 20.18 5.47 3.09Qu3 27.48 21.16 20.60 4.56 18.93 4.35 2.93

R E S U L T S

Composition and preferred orientation

A summary of different phase proportions in weight fractionsis given in Table 1. All samples (Qu1, Qu2 and Qu3) are richin phyllosilicates, with the illite-group as the dominant phase.There are large amounts of detrital quartz and illite-mica (frag-ments are 5–50 μm in size). Among authigenic phyllosilicateskaolinite dominates over illite-smectite. The diffraction peaks

of illite-smectite are diffuse (Fig. 10), indicating small grainsize and considerable stacking disorder. With the Rietveldmethod, it was possible to separate illite-smectite from illite-mica with diffraction peaks at similar positions (see insets inFig. 10). Other phases such as feldspars, chlorite and pyrite aresubordinate (5–10 wt%). Glycolation tests could not identifymontmorillonite and the 14 A peak is attributed to chlorite.Lattice parameters of the major phases were refined and cor-respond to those described in the literature (Table 2).



Table 2 Lattice parameters of major phases used in the refinement of Qu1, Qu2 and Qu3.Standard deviations are indicated in parentheses. Parameters of quartz, K-feldspars, pyrite andchlorite are kept constant throughout the refinement. Monoclinic phases are displayed in thesecond setting system

Phase Sample A b c α β γ

(A) (A) (A) (◦) (◦) (◦)

Kaolinite Qu1 5.181 8.956 7.440 91.863 104.999 89.911Qu2 5.184 8.980 7.444 92.134 105.058 89.309Qu3 5.184 8.980 7.444 92.134 105.058 89.309

Illite- Qu1 5.371(1) 8.936(2) 11.170(2) 90.000 100.529 90.000Smectite Qu2 5.112 9.496 11.164 90.000 95.661 90.000

Qu3 5.112 9.496 11.164 90.000 100.529 90.000Illite-Mica Qu1 5.254 9.046(1) 20.391(1) 90.000 95.366 90.000

Qu2 5.247 9.062 20.343 90.000 95.096 90.000Qu3 5.247 9.062 20.343 90.000 95.37 90.000

Chlorite 5.327 9.232 14.399 ∼90 97.160 ∼ 90Quartz 4.937 4.937 5.433 90.000 90.000 120.000K-feldspars 8.604 13.029 7.263 90.000 116.005 90.000Pyrite 5.442 5.442 5.442 90.000 90.000 90.000

The raw diffraction images clearly show strong intensitychanges with azimuth, which indicates preferred orientation(Figs 7 and 9). This was quantified in the Rietveld analysisand pole figures are displayed for kaolinite (Fig. 11), illite-mica (Fig. 12), illite-smectite (Fig. 13) and chlorite (Fig. 14),with maximum and minimum pole densities given in Table 3.In general, preferred orientations for phyllosilicate mineralsare quite strong, whereas orientation of quartz, K-feldsparsand pyrite are close to random (pole figures are not shown).All pole figures are more or less axially symmetric with the(001) maximum perpendicular to the bedding plane and rota-tional freedom of (100). Kaolinite shows a strong texture (4.4–6.5 m.r.d.). Detrital illite-mica has a stronger preferred ori-entation (3.2–6.8 m.r.d.) than authigenic illite-smectite (2.4–4.0 m.r.d.). There is some variation between the samples. Illite-mica in Qu1 has the strongest texture with a (001) maximumperpendicular to the bedding plane of 6.8 m.r.d. A similarmaximum is observed for kaolinite in Qu3 (6.5 m.r.d.). Ori-entation distributions of chlorite are quite high in all samplesbut most relevant in Qu3 with 5.8 m.r.d. (Fig. 14). The weak-est textures are observed in Qu2.

Calculations of elastic properties based on microstructure

3D orientation distributions and single crystal elastic proper-ties are required for the calculation of the elastic propertiesof the anisotropic polyphase aggregate, which constitutes theshale matrix. Elastic properties of clay minerals are not very

well-known (e.g., Mondol et al. 2008). Crystals are gener-ally too small for experimental determinations. Dependingon the approximation method, quite different values havebeen obtained for numerical estimates from experiments inwhich clay properties are calculated from an extrapolationof trends assuming clay mineral distributions that best fit thedata (Hornby, Schwartz and Hudson 1994; Wang, Wang andCates 2001). Indentation tests find relatively low values forelastic constants (Prasad et al. 2002). Katahara (1996) at-tributed differences between estimates derived from well crys-tallized phyllosilicates and other values to the presence of mi-crocracks aligned with crystals and to the presence of boundwater that would naturally align with the crystal surfaces.

We here rely on first principles calculations for single crystalelastic properties of phyllosilicates based on the local densityapproximation (Militzer et al. 2011). This recent study ofMilitzer et al. (2011) illustrates that first principles calcula-tions are very reliable for illite-mica but that there is someuncertainty for illite-smectite, particularly the C33 coefficientthat relies on the structural stacking model. Note that theillite-smectite model of Militzer et al. (2011) included struc-tural water but not additional water layers in the expandedmontmorillonite structure. For kaolinite, stacking and poly-typism does not affect elastic properties appreciably and weuse the model for ideal kaolinite. The new values are con-siderably different from the older simulations by Sato et al.(2005). Some off-diagonal coefficients for kaolinite are consis-tently large, which is expressed in an unexpected behaviour for



Table 3 Pole densities for (001) of kaolinite, illite-smectite, illite-mica and chlorite polefigures (m.r.d.)

Kaolinite Illite-Smectite Illite-Mica Chlorite

Sample Min Max Min Max Min Max Min Max

Qu1 0.29 4.81 0.19 3.97 0.30 6.75 0.17 5.61Qu2 0.33 4.41 0.41 2.42 0.44 3.23 0.17 3.56Qu3 0.19 6.49 0.29 3.31 0.36 4.56 0.12 5.77

Figure 16 The calculated a) P-wave velocities (Vp) in km/s and b) shear-wave splitting (�Vs) in m/s of the average over the three Qusaibasamples plotted as a function of the angle to the bedding plane normal. Shown are curved for kaolinite, illite-mica, illite-smectite and chlorite.The geometric mean is used for the averaging.

Table 4 Single crystal stiffness tensor coefficients (Cij) of kaolinite, illite-smectite, illite-mica, chlorite and quartz that were used in elasticproperties calculations (in GPa) for monoclinic phases second setting

Phase C11 C12 C13 C22 C23 C33 C44 C55 C66

Kaolinite(1) 169.1 66.1 15.4 179.7 10.2 81.1 17.0 26.6 57.6Illite-smec(1) 27.2 13.2 5.2 153.9 25.1 188.5 55.4 10.4 2.8Illite-mica(1) 60.3 27.2 23.5 180.9 53.4 170.0 70.5 18.4 23.8Chlorite(1) 180.9 53.4 27.2 170.0 23.5 60.3 18.4 23.8 70.5Quartz(2) 87.3 6.6 12.0 87.3 12.0 105.8 57.2 57.2 40.4

Note (1) Militzer et al. 2011; (2) Heylinger 2002.

shear waves and particularly shear-wave splitting (Fig. 16b).For quartz we used experimental data of Heyliger, Ledbetterand Johansen (2002). Table 4 lists the single crystal stiffnesscoefficients that were used.

Since orientation distributions are close to axial symme-try, we strictly imposed this symmetry (transverse isotropy)for the polyphase aggregate property calculations, reducingthe elastic tensor from 21 to 5 independent components(C11 = C22, C12 = C11 – 2C66, C13 = C23, C33, C44 =

C55, all others are zero) (Nye 1956). First, elastic proper-ties of contributing mineral phases were calculated by aver-aging single crystal elastic properties over the mineral orien-tation distributions for the individual phases in the softwareBEARTEX (Wenk et al. 1998) and then averaging over allphases according to volume fractions was performed. Val-ues of Cij for Voigt (1928), Reuss (1929), Hill (1963) aver-ages and a geometric mean average (Matthies and Humbert1993) are shown in Table 5. Note that there is a considerable



Table 5 Calculated polycrystal stiffness tensor coefficients (Cij) in GPa assuming axial symmetry by Voigt, Reuss, Hill averages and geometricmean

Qu1 Qu2 Qu3

Voigt Reuss Hill Geo. Voigt Reuss Hill Geo. Voigt Reuss Hill Geo.

C11 124.88 87.72 103.70 107.72 119.35 83.97 99.35 102.07 126.36 88.63 104.99 108.70C13 22.80 20.26 21.49 22.75 23.32 20.70 22.64 22.74 25.25 21.85 23.46 24.92C33 94.09 72.25 81.91 81.27 101.97 75.06 85.69 86.96 97.82 73.56 84.25 83.99C44 40.51 27.80 33.23 32.88 41.97 28.69 33.35 34.17 40.50 27.45 33.06 32.70C66 48.10 33.15 39.60 40.64 46.20 31.64 37.20 38.44 47.44 32.54 39.00 39.81

Table 6 Wave velocities from calculations based on single crystal elastic constants, volume fractions and orientation distributions (Qu1–Qu3,QuAVE), from ultrasonic measurement under over pressure (QuUS-OP), ultrasonic, extrapolation to in situ stress (QuUS-is) and from wire-linelogs data (QuWire). P-wave anisotropy% and Thomsen’s parameters are also displayed

Sample Clay (vol%) Averaging model Vs min (km/s) Vp max (km/s) Vp min (km/s) Ani. (%) ε γ δ

Qu1 67.15 Voigt 3.87 6.80 5.90 14.2 0.16 0.09 0.11Reuss 3.20 5.69 5.17 9.6 0.11 0.10 0.05Hill 3.50 6.19 5.50 11.8 0.13 0.10 0.07Geometric 3.47 6.28 5.44 14.3 0.16 0.12 0.10



QuAVE 72.05 Voigt 3.89 6.76 6.02 11.6 0.13 0.08 0.08Reuss 3.21 5.73 5.22 9.32 0.09 0.08 0.04Hill 3.50 6.16 5.57 10.1 0.11 0.08 0.06Geometric 3.47 6.23 5.52 12.1 0.13 0.10 0.08

QuUS-OP - - 2.10 5.31 4.03 27.4 0.37 - -QuUS-is - - 2.14 5.33 4.09 26.3 0.35 - -QuWire - - 1.96 3.76 - - -

difference between the constant strain (Voigt) and constantstress (Reuss) average for these strongly anisotropic materi-als, particularly for coefficients along the diagonal (C11, C33,C44 and C66). For example, C33 for the Voigt average is 44%larger than for the Reuss average. There is not much differencebetween arithmetic (Hill) and geometric means. From theseaggregate elastic properties, P- and S-wave velocities as well asP-wave anisotropy (Anisotropy (%) = 200(Vp max – Vp min)/(Vp max + Vp min)) were calculated (Table 6, top three rows,Qu1, Qu2 and Qu3). We then calculated an average overthe stiffness tensors of the three samples as a representativevalue for the polyphase aggregate properties of the Qusaibashale and derived corresponding velocities (Table 6, QuAVE).

Figure 16(a) shows calculated P-wave velocities and Fig. 16(b)displays shear-wave splitting for individual phyllosilicate com-ponents, averaged over the three samples, as a function of theangle to the bedding plane normal. Low P-wave velocities andno S-wave splitting is observed perpendicular to the beddingplane (0◦). In all samples, illite-mica and illite-smectite arestrongly anisotropic (11.8–22.4% and 15.4–34.0%, respec-tively), combining strong preferred orientation and strongsingle crystal anisotropy. The anisotropy of kaolinite issmaller, ranging from 12.0–17.1%. The anisotropy of chlo-rite is strongest but also has large variations among samples(9.9–22.9%). The contribution of quartz, feldspars and pyriteto anisotropy is negligible. Keep in mind that these elastic



properties and velocities are only representative of the min-eral polyphase aggregate and do not include the effects ofporosity, interlayer water, fluid or kerogen fill. This informa-tion cannot be obtained directly from the synchrotron X-raydiffraction measurements.

Experimental velocity measurements

The velocity values measured at slight overpressure condi-tions are given in Table 6 (QuUS-OP). In addition, Table 6gives velocities at in situ stress conditions that were obtainedfrom the measured values by the extrapolation procedure de-scribed above (QuUS-is). Under over-pressure conditions, themaximum P-wave velocity is 5.31 km/s while the minimum is4.03 km/s, with an anisotropy of 27.4%. The values at the in

situ stress conditions are only slightly higher. The Vp max-imum is 5.33 km/s and the minimum 4.09 km/s, yielding26.3% anisotropy. The increase with net stress is relativelysmall and has a minor impact on the P-wave anisotropy. Be-cause of the axial sample symmetry, the S-wave travellingalong the cylinder axis (axial wave) has no splitting becausethe S-wave velocity (2.10–2.14 km/s) is independent of polar-ization. The experimental uncertainty in the measured valuesis ±1%.

P- and S-wave velocities have also been recorded in thewell by wire-line measurements. Mud pressure in the well washigher than the estimated fluid pressure at that depth, whichmay have affected the P- and S-wave velocities in the bore-hole walls. One can anticipate slight overpressure conditionsnear the borehole wall but it will not affect the measurementto a significant extent. The ultrasonic measurement gives aquantitative measure of the order of magnitude of the effectto be expected. The P-wave velocity from wire-line logs in thecored well, which would be equivalent to Vp at 90◦ to the bed-ding plane in the laboratory test, was measured to be around3.76 km/s. This is about 8% lower than the velocity deter-mined from the core plug (Table 6, QuWire). Similarly, theshear velocity as measured in the wire-line is some 1.96 km/s,again about 8% lower than the velocity measured in the coreplug. This difference between sonic and ultrasonic measure-ments is substantially higher than the difference in velocitiesmeasured in ultrasonic measurements at in situ conditions andat slight overpressure conditions. The discrepancy between ul-trasonic and sonic velocities is most likely due to the differencein temperature (100◦ C in situ versus room temperature inthe laboratory, e.g., Johnston (1986) and Manafov, Holt andFjaer (2007)) and in the frequency used for the measurements(10 kHz versus 1 MHz). Diagnetic changes due to storage

of the core, having induced hardening of the rock and thushigher velocities, cannot be excluded as causing some of thediscrepancy. It should be noted that the sonic log measuresover a larger volume than the ultrasonic experiment and thusstatistical sampling may also play a role.

D I S C U S S I O N

Three diagenetic shales from the Silurian Qusaiba formationwere analysed for microstructure and texture, mainly relyingon synchrotron X-ray experiments. The three samples havesimilar mineralogical composition, with the members of theillite-group dominating (33–42 wt%). Kaolinite is also abun-dant in the samples with approximately 25–28 wt%. The mi-crostructure is somewhat heterogeneous but it should be keptin mind that only volumes of 8–10 mm3 were analysed for eachsample and an average over the three samples would seem agood representative value for the formation as a whole.

We successfully extracted orientation distributions for fourphyllosilicate phases – kaolinite, illite-mica, illite-smectite andchlorite – with 001 pole figure maxima ranging from 2.4–6.8 m.r.d (Figs 11–14). The texture strengths of illite andkaolinite observed in this study are similar to previous syn-chrotron studies for Kimmeridge North Sea shale (Wenk,Kanitpanyacharoen and Voltolini 2010, 2–6 m.r.d.), Callovo-Oxfordian shales and Opalinus clay from Central Europe(Wenk et al. 2008a,b, 2–9 m.r.d.), shales from Nigeria(Lonardelli et al. 2007, 2–5 m.r.d.) and shales from Silver Hill,Montana (Wenk et al. 2007, 10 m.r.d.). This also agrees withdata obtained by pole figure goniometry for shales/siltstonesfrom the North Sea (Valcke et al. 2006, 3–5 m.r.d.), GulfCoast mudstones (Ho et al. 1999, 2–7 m.r.d.) and Zechsteinshales (Sinbutin 1994a, 4–6 m.r.d.) but the anisotropy is con-siderably weaker than the one encountered in metamorphicslates (Sinbutin 1994b, 5–18 m.r.d. and Oertel and Phakey1972, 16 m.r.d.) and schists (Wenk et al. 2010, 7–14 m.r.d.).Our pole densities of illite and kaolinite are also consistentwith experimentally compressed clay-silt mixtures (Voltoliniet al. 2009).

Overlapping peaks of authigenic illite-smectite could besuccessfully separated from detrital illite-mica through theRietveld refinement implemented in MAUD. Fine-grained au-thigenic illite-smectite has a weaker texture than coarser de-trital illite-mica (Table 3). It is appropriate to note that a highstrength of preferred orientation is generally related to a largeclay content in the sample, as has been established in exper-iments (Voltolini et al. 2009). Qu3 has the largest amountof phyllosilicates (73.8 wt%) and displays the highest texture



strength for chlorite and kaolinite. However, the (001) polefigure maximum of illite-mica in Qu1 is the highest amongall phases (6.8 m.r.d., Table 3), even though it has a lowerclay content (65.9 wt%). Aside from the dependency on thepercentage of clay, texture strength may be related to burialand compaction history and local heterogeneities. All sam-ples were acquired within a few feet of each other and thusoverall burial is identical. Illite-smectite is more poorly ori-ented than other phyllosilicates as shown in the SEM image(Fig. 1) and observed from a broad and weak peak at ∼10A in the diffraction profile (Fig. 10). This authigenic phyl-losyllicate has a very small size and represents most of thebackground material in the SEM images, which indeed showlittle orientation, in contrast to the detrital illite-mica flakes,which can be easily distinguished and are orientated perpen-dicular to the maximum vertical compaction stress. The poledensities of illite-smectite range from 2.4–4.0 m.r.d, which isconsiderably lower than those of illite-mica. The orientationdistributions of quartz, K-feldspars and pyrite are nearly ran-dom and close to 1 m.r.d. It is noteworthy that for all samplesa considerable portion of crystallites is randomly oriented,expressed by (001) minima of 0.2–0.4 m.r.d. for kaolinite,illite-mica and illite-smectite. The minima in Qusaiba shale(65.9–73.8 wt% clays) are also comparable with the experi-mentally compressed kaolinite-illite-silt mixtures (75% clay),which are 0.4–0.6 m.r.d. (Voltolini et al. 2009).

As a number of studies suggest, preferred orientation inshales has a strong influence on elastic anisotropy and di-rectionality of acoustic velocities (e.g., Sayers 1994; Hornby1994) and this is relevant for seismic prospecting. Illite-mica,chlorite and kaolinite have higher calculated Vp velocities andstronger anisotropy in comparison to illite-smectite (Fig. 16a),consistent with higher single crystal stiffness and stronger tex-ture. The shear-wave splitting (�Vs) of chlorite, kaolinite andillite-mica is relatively high for waves propagating in the bed-ding plane (Fig. 16b, 90◦) and consistent with other obser-vations (Wenk et al. 2008b; Voltolini et al. 2009). Note therather complex behaviour for kaolinite can be explained withthe large values of non-diagonal stiffness coefficients, whichis a consequence of the triclinic kaolinite structural modelused by Militzer et al. (2011). Calculated P-wave anisotropieswith geometric mean are ranging from 9.1–14.3% (Table 6,Fig. 17). Our predicted anisotropies are comparable withother reports on calculated velocities of natural shales fromthe North Sea (Valcke 2006, 12%) and shales from Nigeria(Lonardelli et al. 2007, 10%) but relatively weaker than MontTerri shale (Wenk et al. 2008a, 20%) and measured velocitiesfor Kimmeridge shale (Hornby 1998, 38%).

Figure 17 Minimum and maximum P-wave velocities. The dashedlines indicate degrees of anisotropy. Shown are experimental velocities(Polyphase_Exp.) and calculated velocities using the geometric mean(Polyphase_Geo.), Voigt average (Polyphase_Voigt), Reuss average(Polyphase_Reuss) and Hill average (Polyphase_Hill). Also shownare velocities for the individual mineral components (kaolinite, illite-smec, illite-mica, chlorite and quartz). The values represent the aver-age over the three Qu1, Qu2 and Qu3 samples (Table 6).

The anisotropic properties can be expressed in terms ofThomsen’s parameters (Thomsen 1986), which assume trans-verse isotropy, i.e., axial symmetry around the bedding planenormal.

ε = (C11 − C33)/(2C33)

γ = (C66 − C55)/(2C55)

δ = [(C13 + C55)2 − (C33 − C55)2]/[2C33(C33 − C55)]

These parameters are used to characterize P- and S-wave prop-agation through a weakly anisotropic medium. Average valuescalculated for the Qusaiba samples are given in Table 6. P-wave anisotropy (9.1–14.3%) is associated with anisotropyparameters ε (0.10–0.16) while parameter γ (0.07–0.12) isa measure of S-wave anisotropy. The Thomsen parameterδ (0.08–0.10) is related to the near-vertical P-wave veloc-ity and is used to explain the discrepancy between verticalvelocity and small-offset normal moveout (NMO) velocityand to understand small-offset amplitude variation with anoffset (AVO) response. Our calculations show comparableanisotropy to shale samples from the North Sea, Africa andthe Gulf Coast with parameters ε (0.08–0.33), γ (0.11–0.53)and δ (–0.05–0.23) (Wang 2002) but relatively higher thanshales from the Nigerian Coast with anisotropy parameters



ε (0.03–0.07), γ (0.01–0.10) and δ (–0.03–0.03) (Lonardelliet al. 2007).

A significant difference exists between the velocities as com-puted for the polyphase aggregate and the velocities deter-mined experimentally in ultrasonic measurements and sonicmeasurements in the borehole (Table 6). Experimental valuesfor P-wave velocities in the bedding plane are 14% lower thancalculated ones (Vpmax in Fig. 17). Velocities perpendicularto the bedding plane as measured by sonic in the borehole andin the ultrasonic experiments (Vpmin) are 27% lower and cor-respondingly anisotropy is higher. This highlights the fact thatdifferent factors influence elastic properties of aggregates. Oneof them is preferred orientation of the component crystals (thepolyphase aggregate) and another one is the shape preferredorientation of the minerals. A third reason is microporosityand nanoporosity, including the porosity between phyllosil-icate crystals and a fourth one in this sample is the layeredalternation of kerogen-richer and kerogen-poorer layers.

Simple modelling of the elastic properties indicates thatthese differences are well within the limits to be expected andalso suggests causes for this discrepancy. The layered alterna-tion of kerogen-rich and less kerogen-rich layers is unlikely thecause of major differences. The elastic properties of kerogenare largely unknown and different estimates can be found inthe literature (e.g., Rundle and Schuler 1981; Vernik and Liu1997; Mavko, Mukerji and Dvorkin 1998). Modelling of theunrealistic scenario of interlayering of pure kerogen with theshale polyphase aggregate can be done by Backus averaging(Backus 1962) and the anisotropy increases in such a case. Butonly when a kerogen type with very low shear strength is se-lected, compressional and shear velocities along the symmetryaxis are obtained that are of the same order of magnitude asthe measured velocities. The fastest (radial) velocity is hardlyaffected in such geometry and the interbedding of the kerogenand polyphase aggregate can be ruled out as a single sourceof the velocity reduction.

An alternative explanation for the softening of thepolyphase aggregate is the loss of coherence between crystalsaffected by porosity, interlayer water and kerogen (Hornbyet al. 1994). Bound water and kerogen are independent mech-anisms tying crystals together in addition to the phyllosili-cate cement. It is a reasonable assumption that fluid filledporosity is anisotropic. Aylmore et al. (1960) argued thatnm scale porosity should be considered as the porosity be-tween crystal plates and hence the porosity will align with thecrystal structure. Katahara (1996) saw this effect as a reasonfor the discrepancy between elastic properties deduced fromtexture measurements and the elastic properties as deduced

from direct velocity measurements. The strong anisotropy atthe nm scale is also in-line with the observation that the μmscale porosity is anisotropic (Fig. 5a), which is related to theweakness in the axial direction relative to the radial direction.Observation of nanopores (Fig. 3b–d) strengthens this argu-ment. In a realistic averaging model, several parameters (i.e.,grain shape, grain orientation, volume, shape and alignmentof pores) need to be considered, which is not done in simple av-eraging schemes. Currently, self-consistent methods are beingdeveloped for this (e.g., Matthies 2010), but are beyond thescope of this work. These methods are based on more empiri-cal full effective medium modelling (by e.g., Hornby 1994;Hornby et al. 1994; Johansen, Ruud and Jakobsen 2004;Draege, Jakobsen and Johansen 2006) but incorporate quan-titative crystal preferred orientation. An approximation canbe obtained by using a Voigt-Reuss-Hill average to calculatethe properties of the matrix with the kerogen inclusions andadding the porosity as cracks parallel to the bedding plane,with an aspect ratio of the order of 0.03 (Cheng 1993). Giventhe uncertainties in kerogen properties, exact porosity valueand the anisotropy caused by the interbedding of kerogenlayers, such a solution is far from unique. But we estimatedthat the large difference in calculated (∼12%) and measuredanisotropy (∼27%) is largely due to anisotropic porosity andopen fracture distribution that affect mainly the lower Vpminvalues. The degree of porosity and fractures are crucial factorsthat should be considered in the system. During compactionby overburden (axial compression), porosity diminishes andclay platelets become more aligned. The more sophisti-cated models that consider porosity (Bayuk, Ammerman andChesnokov 2007), fracture distribution (Sayers 1998), satura-tion (Pham et al. 2002) and frequency dependency (Sams et al.1997), which can contribute to anisotropy, may be of use totie the calculated matrix properties to the velocities measuredin ultrasonic measurements.

The information from texture strengths can also be usedto estimate the compaction strain in the system. Based onMarch’s method (March 1932), we assume preferred ori-entation of originally randomly oriented rigid platelets em-bedded in a viscous matrix when the composite is homo-geneously strained. Oertel and Curtis (1972) later modifiedthe original March model for compaction by taking a vol-ume change into account. The compaction strain (εc) is ex-pressed in terms of the maximum pole density (ρmax) as εc =ρmax

−1/2 – 1. The maximum (001) pole densities of each phaseare used to calculate the March strain (Table 7). Our calcu-lated compaction strain values (–0.36 to –0.61) are compara-ble to that of 75–100 wt% clay in the compression experiment



Table 7 Estimated compaction strain (εc) values from maximum pole densities(ρmax) of kaolinite, illite-smectite, illite-mica and chlorite. (εc = ρmax

−1/2 – 1)

Kaolinite Illite-Smectite Illite-Mica Chlorite

Sample ρmax εc ρmax εc ρmax εc ρmax εc

Qu1 4.81 −0.54 3.97 −0.50 6.75 −0.62 5.61 −0.58Qu2 4.41 −0.52 2.42 −0.36 3.23 −0.44 3.56 −0.47Qu3 6.49 −0.61 3.31 −0.45 4.56 −0.53 5.77 −0.58

of Voltolini et al. (2009), which found that with 5 MPa com-pression on 75–100wt% clay, the compaction strains rangefrom –0.35 to –0.51. With higher pressure applied, as muchas 50 MPa, the compaction strains were –0.48 to –0.65. Thissuggests that the Qusaiba shales had undergone an extensivecompaction history. We attribute the variation between thethree samples and the phases to natural heterogeneities anddifferent crystal shapes. To estimate the compaction strain, thetexture of detrital illite-mica is probably the most reliable, be-cause they constitute relatively large platelets in a fine-grainedmatrix.

CONCLUSION

In this study, we combine microscopic observations of mi-crostructural features with an evaluation of macroscopicproperties of deeply buried shale subjected to compaction anddiagenesis. We are able to quantify preferred orientation inQusaiba shales through X-ray diffraction techniques. Withthe use of the Rietveld refinement, we obtain quantitative ori-entation distributions for kaolinite, illite-mica, illite-smectiteand chlorite. One of the main factors contributing to tex-ture strength is the volume fraction of phyllosilicates. Sam-ples rich in clay exhibit a stronger texture. Moreover, burialand compaction history can affect texture strength. Kaolinite,illite-mica and chlorite generally exhibit strong texture whilenanocrystalline illite-smectite shows weaker preferred orien-tation. From preferred orientation of mineral phases we cancalculate the elastic properties of the shale matrix. However,the anisotropy of the matrix is not the only contributor to theanisotropy of the shale. The distribution of kerogen and moreimportantly the orientation of the microfracture and poros-ity network reduce the strength of the matrix, enhancing theanisotropy and these porosity related effects are importantcontributors to the anisotropy. Through linking the matrixand porosity components, a more comprehensive model ofshale elastic properties can be developed as soon as the poros-

ity/fracture/kerogen anisotropy can be quantified. The latteris not the subject of this study.

ACKNOWLEDGEMENTS

This project is supported by NSF (EAR-0836402) and DOE(DE-FG02-05ER15637). This publication was in part sup-ported by Award No KUS-I1–004-21, made by the King Ab-dullah University of Science and Technology (KAUST), whichinitiated the contact with researchers in Saudi Arabia, espe-cially Edgardo Nebrija (ARAMCO). We would particularlylike to thank Dr Walter Voggenreiter (Exploration Manager),Geoff Pike and other staff of the South Rub’al-Khali Company(SRAK) for providing us with shale samples and useful infor-mation. We acknowledge the Ministry of Petroleum, SaudiArabia as well as Saudi Aramco, SRAK and Shell for permis-sion to publish the information obtained in this study. We areappreciative for access to beamline 11-ID-C at APS and assis-tance from Yang Ren for diffraction experiments and accessto beamline TOMCAT at SLS and assistance from RajmundMokso for tomography experiments. We also greatly appre-ciate the help from Luca Lutterotti for updating the MAUDsoftware, Marco Voltolini for suggestions with data analy-sis and Kai Chen for help with tomography data processing.Hans-Rudolf Wenk is appreciative for the hospitality and ac-cess to facilities at GFZ Potsdam. We thank David Dewhurstand two anonymous reviewers for helpful and constructivecomments on the first version of this manuscript.

REFERENCES

Al-Harbi O.A. 1998. Petrography and diagenetic controls on reservoircharacteristics in Unayzah Formation, Central Saudi Arabia. JKAU:Earth Science 10, 45–57.

Aplin A.C. and Larter S.R. 2005. Fluid flow, pore pressure, wettabilityand leakage in mudstone cap rocks. In: Evaluating Fault and CapRock Seals (eds P. Boult and J. Kaldi), pp. 1–12. AAPG.

Aplin A., Matenaar I., Matenaar D. and Van Der Pluijm B.A. 2006.Influence of mechanical compaction and clay mineral diagenesis



on the microfabric and porescale properties of deep-water Gulf ofMexico mudstones. Clays and Clay Minerals 54, 500–514.

Aylmore L. and Quirk J.P. 1960. Domain or turbostratic structure ofclays. Nature 187, 1046–1048.

Backus G.E. 1962. Long-wave elastic anisotropy produced by hori-zontal layering. Journal of Geophysical Research 67, 4427–4440.

Bayuk I.O., Ammerman M. and Chesnokov E.M. 2007. Elastic mod-uli of anisotropic clay. Geophysics 72, D107–D117.

Bish D.L. 1993. Rietveld refinement of the kaolinite structure at 1.5K. Clays and Clay Minerals 41, 738–744.

Bossart P. and Thury M. 2007. Research in the Mont Terri RockLaboratory, quo vadis. Physics and Chemistry of the Earth 32,19–31.

Chadwick R.A., Zweigel P., Gregersenc U., Kirby G.A., Holloway S.and Johannessen P.N. 2004. Geological reservoir characterizationof a CO2 storage site. The Utsira Sand, Sleipner, northern NorthSea. Energy 29, 1371–1381.

Cheng C.H. 1993. Crack models for a transversely isotropic medium.Journal of Geophysical Research 98, 675–684.

Day-Stirrat R., Aplin A., Srodon J. and Van Der Pluijm B. 2008. Di-agenetic reorientation of phyllosilicate minerals in Paleogene mud-stones of the Podhale Basin, southern Poland. Clays and Clay Min-erals 56, 100–111.

Dewhurst D.N. and Siggins A.F. 2006. Impact of fabric microcracksand stress field on shale anisotropy. Geophysical Journal Interna-tional 165, 135–148.

Diamond S. 2000. Mercury porosimetry. An inappropriate methodfor the measurement of pore size distributions in cement-basedmaterials. Cement and Concrete Research 30, 1517–1525.

Draege A., Jakobsen M. and Johansen T.A. 2006. Rock physicsmodelling of shale diagenesis. Petroleum Geoscience 12, 49–57.

Fjaer E., Holt R.M., Horsrud P., Raaen A.M. and Risnes R. 2008.Petroleum Related Rock Mechanics, 2nd edn. Elsevier.

Gualtieri A.F. 2000. Accuracy of XRPD QPA using the combinedRietveld-RIR method. Journal of Applied Crystallography 33,267–278.

Hammersley A.P. 1998. Fit2D. V12.012 reference manual. Version6.0., European Synchrotron Radiation Facility, Internal ReportESRF98HA01T. (www.esrf.ed/computing/scientific/FIT2D).

Heyliger P., Ledbetter H. and Kim S. 2002. Elastic constants of naturalquartz. Journal of the Acoustical Society of America 114, 644–650.

Hill R. 1963. Elastic properties of reinforced solids: Some theoreti-cal principles. Journal of the Mechanics and Physics of Solids 11,357–372.

Ho N.-C., Peacor D.R. and Van Der Pluijm B.A. 1999. Preferred ori-entation of phyllosilicates in Gulf Coast mudstones and relation tothe smectite-illite transition. Clays and Clay Minerals 47, 495–504.

Hornby B.E. 1994. The elastic properties of shales. PhD thesis, Uni-versity of Cambridge.

Hornby B.E. 1998. Experimental laboratory determination of thedynamic elastic properties of wet, drained shales. Journal of Geo-physical Research 103, 29945–29964.

Hornby B.E., Schwartz L.M. and Hudson J.A. 1994. Anisotropiceffective medium modeling of the elastic properties of shales. Geo-physiscs 59, 1570–1583.

Jakobsen M. and Johansen T.A. 2000. Anisotropic approxima-tions for mudrocks: A seismic laboratory study. Geophysics 65,1711–1725.

Johansen T.A., Ruud B.O. and Jakobsen M. 2004. Effect of grainscale alignment on seismic anisotropy and reflectivity of shales.Geophysical Prospecting 52, 133–149.

Johnston D.H. 1987. Physical properties of shale at temperature andpressure. Geophysics 52, 1391–1401.

Joswig W., Fuess H., Rothbauer R., Takeuchi Y. and Mason S.A.1980. A neutron diffraction study of a one-layer triclinic chlorite(penninite). American Mineralogist 65, 349–352.

Katahara K.W. 1996. Clay mineral elastic properties. 66th SEG meet-ing, Denver, Colorado, USA, Expanded Abstracts, 1691–1694.

Kocks U.F., Tome C.T. and Wenk H.-R. 2000. Texture andAnisotropy. Preferred Orientations in Polycrystals and Their Ef-fect on Material Properties. Cambridge University Press.

Lonardelli I., Wenk H.-R. and Ren Y. 2007. Preferred orientation andelastic anisotropy in shales. Geophysics 72, D33–D40.

Lutterotti L., Matthies S., Wenk H.-R., Schultz A.J. and RichardsonJ.W. 1997. Combined texture and structure analysis of deformedlimestone from time-of-flight neutron diffraction spectra. Journalof Applied Physics 81, 594–600.

Mallants D., Marivoet J. and Sillen X. 2001. Performance assessmentof the disposal of vitrified high-level waste in a clay layer. Journalof Nuclear Materials 298, 125–135.

Manafov R.T., Holt R.M. and Fjaer E. 2007. Temperature sensitivityof wave velocities in shales. 69th EAGE meeting, London, UK,Expanded Abstracts.

March A. 1932. Mathematische Theorie der Regelung nach der Korn-gestalt bei affiner Deformation. Zeitschrift fuer Kristallographie 81,285–297.

Matthies S. 2010. On the combination of self-consistent and geomet-ric mean elements for the calculation of the elastic properties oftextured multi-phase samples. Solid State Phenomena 160, 87–93.

Matthies S. and Humbert M. 1993. The realization of the concept of ageometric mean for calculating physical constants of polycrystallinematerials. Physica Status Solidi B 177, K47–K50.

Matthies S. and Vinel G.W. 1982. On the reproduction of the orien-tation distribution function of textured samples from reduced polefigures using the concept of conditional ghost correction. PhysicaStatus Solidi B 122, K111–K114.

Matthies S. and Wenk H.-R. 2009. Transformations for monocliniccrystal symmetry in texture analysis. Journal of Applied Crystallo-gaphy 42, 564–571.

Mavko G., Mukerji T. and Dvorkin J. 1998. The Rock Physics Hand-book. Cambridge University Press.

Militzer B., Wenk H.-R., Stackhouse S. and Stixrude L 2011. Elasticproperties of illite-smectite and kaolinite based on first principlesand their significance for the anisotropy of shales. American Min-eralogist 96, 125–137.

Mondol N.H., Jahren J., Bjørlykke K. and Brevik I. 2008. Elasticproperties of clay minerals. The Leading Edge 27, 758–770.

Nye J.F. 1956. The Physical Properties of Crystals. Clarendon.Oertel G. and Curtis C.D. 1972. Clay-ironstone concretion pre-

serving fabrics due to progressive compaction. GSA Bulletin 83,2597–2606.



Oertel G. and Phakey P.P. 1972. The texture of a slate from Nantille,Caernarvon, North Wales. Texture 1, 1–8.

Pham N.H., Carcione J.M., Helle H.B. and Ursin B. 2002. Wavevelocities and attenuation of shaley sandstones as a function ofpore pressure and partial saturation. Geophysical Prospecting 50,615–627.

Plancon A., Tsipurski S.I. and Drits V.A. 1985. Calculation of inten-sity distribution in the case of oblique texture electron diffusion.Journal of Applied Crystallography 18, 191–196.

Prasad M., Kopycinska M., Rabe U. and Arnold W. 2002. Measure-ment of Young’s modulus of clay minerals using atomic force acous-tic microscopy. Geophysical Research Letters 29, 1172–1175.

Reuss A. 1929. Berechnung der Fließgrenze von Mischkristallen aufGrund der Plastizitatsbedingung fur Einkristalle. Zeitschrift furAngewandte Mathematik und Mechanik 9, 49–58.

Rietveld H.M. 1969. A profile refinement method for nuclear andmagnetic structures. Journal of Applied Crystallography 2, 65–71.

Rundle J.B. and Schuler K.W. 1981. A composite model for theanisotropic elastic moduli of lean oil shale. Geophysics 46,163–171.

Sams M.S., Neep J.P., Worthington M.H. and King M.S. 1997.The measurement of velocity dispersion and frequency-dependentintrinsic attenuation in sedimentary rocks. Geophysics 62,1456–1464.

Sato H., Ono K., Johnston C.T. and Yamagishi A. 2005. First-principles studies on elastic constants of a 1.1 layered kaolinitemineral. American Mineralogist 90, 1824–1826.

Sayers C.M. 1998. Misalignment of the orientation of fractures andthe principal axes for P and S waves in rocks containing multi-ple non-orthogonal fracture sets. Geophysical Journal International133, 459–466.

Sayers C.M. 1994. The elastic anisotropy of shales. Journal of Geo-physical Research 99, 767–774.

Schenk C., Pollastro R. and Ahlbrandt T. 2002. Assessment of undis-covered oil and gas of the Lower Silurian Qusaiba-Paleozoic totalpetroleum system of the Arabian Peninsula. US Geological SurveyFacts Sheet FS-008–02.

Sintubin M. 1994a. Clay fabrics in relation to the burial history ofshales. Sedimentology 41, 1161–1169.

Sintubin M. 1994b. Phyllosilicate preferred orientation in relationto strain path determination in the lower Paleozoic Stavelot-VennMassif (Ardennes, Belgium). Tectonophysics 237, 215–231.

Thomsen L. 1986. Weak elastic anisotropy. Geophysics 51,1954–1966.

Tissot B.O. and Welte D.D. 1984. Petroleum Formation and Occur-rence, 2nd edn. Springer-Verlag.

Valcke S.L.A., Casey M., Lloyd G.E., Kendall J.-M. and Fisher Q.J.2006. Lattice preferred orientation and seismic anisotropy in sedi-mentary rocks. Geophysical Journal International 166, 652–666.

Vernik L. and Liu L. 1997. Velocity anisotropy in shales: A petro-physical study. Geophysics 62, 521–532.

Voigt W. 1928. Lehrbuch der Kristallphysik. Teubner-Verlag.Voltolini M., Wenk H.-R., Mondol N.H., Bjørlykke K. and Jahren

J. 2009. Anisotropy of experimentally compressed kaolinite-illite-quartz mixtures. Geophysics 74, 13–23.

Wang Z. 2002. Seismic anisotropy in sedimentary rocks: Part 2 –Laboratory data. Geophysics 67, 1423–1440.

Wang Z., Wang H. and Cates M.E. 2001. Effective elastic propertiesof solid clays. Geophysics 66, 428–440.

Wenk H.-R., Kanitpanyacharoen W. and Voltolini M. 2010. Preferredorientations of phyllosilicates. Comparison of fault gouge, shaleand schist. Journal of Structural Geology 32, 478–489.

Wenk H.-R., Lonardelli I., Franz H., Nihei K. and Nakagawa S. 2007.Preferred orientation and elastic anisotropy of illite-rich shale. Geo-physics 72, E69–E75.

Wenk H.-R., Matthies S., Donovan J. and Chateigner D. 1998. BEAR-TEX: A windows-based program system for quantitative textureanalysis. Journal of Applied Crystallography 31, 262–269.

Wenk H.-R., Voltolini M., Mazurek M., van Loon L.R. and VinsotA. 2008a. Preferred orientations and anisotropy in shales. Callovo-Oxfordian shale (France) and Opalinus Clay (Switzerland). Claysand Clay Minerals 56, 285–306.

Wenk H.-R., Voltolini M., Kern H., Popp T. and Mazurek M.2008b. Anisotropy in shale from Mont Terri. The Leading Edge27, 742–748.

Wirth R. 2009. Focused Ion Beam (FIB) combined with SEM andTEM: Advanced analytical tools for studies of chemical compo-sition, microstructure and crystal structure in geomaterials on ananometre scale. Chemical Geology 261, 217–229.


Date post:	16-Jul-2020
Category:	Documents
Upload:	others
View:	0 times
Download:	0 times

Texture and anisotropy analysis of Qusaiba...

Documents