Journal of Geophysical Research: Solid Earth

RESEARCH ARTICLE The Elastic Properties of Clay in Shales

10.1029/2018JB015600
Colin M. Sayers1 and Lennert D. den Boer2
Key Points: 1
Schlumberger, Houston, TX, USA, 2Schlumberger, Calgary, Canada
• To accurately characterize shales,
rock physics models must account
for anisotropic clay minerals
• Due to compliant regions between Abstract The mechanical properties of clay minerals are important in many diverse scientific disciplines,
clay platelets, the elastic stiffness of including soil mechanics, civil engineering, materials science, and petroleum exploration. Rock physics
clay in shales is much less than that provides a link between the elastic properties of rocks and their constitutive properties such as mineralogic
of clay minerals
• The clay in shales can be modeled as composition, porosity, and pore-fluid content. To accurately characterize shales, rock physics models must
anisotropic clay platelets embedded account for the anisotropic properties of clay minerals. Due to more compliant regions between clay
in a softer interparticle region particles, the elastic stiffness of clay in shales is significantly less than that of its constituent clay minerals. In
containing clay-bound water
this paper, the clay in shales is modeled as anisotropic clay platelets surrounded by a softer interparticle
region consisting of clay-bound water and interparticle contacts. Inverting for the elastic properties of this
interparticle region indicates that its effective bulk modulus is like that of water. However, it has a nonzero
Correspondence to:
effective shear modulus that is smaller by an order of magnitude, consistent with the expected shear
C. M. Sayers,
[email protected] modulus of clay-bound water. Owing to its simplicity and robustness, it is anticipated that this model of
shales, based on the properties of clay minerals and the interparticle medium, will find use in many rock
physics applications, including seismic imaging, seismic inversion, and geomechanics.
Citation:
Sayers, C. M., & den Boer, L. D. (2018).
The elastic properties of clay in shales.
Journal of Geophysical Research: Solid
Earth, 123, 5965–5974. https://doi.org/ 1. Introduction
10.1029/2018JB015600
Understanding the mechanical properties of clay minerals is important in many disciplines, including soil
Received 1 FEB 2018
mechanics, civil engineering, petroleum exploration, carbon sequestration, and materials science, where
Accepted 3 JUL 2018 clays are used in polymer matrix-based nanocomposites (Paul & Robeson, 2008), with applications in micro-
Accepted article online 8 JUL 2018 electronics, biomaterials, catalysts, polymer films, and fuel cells, for example. Unconventional hydrocarbon
Published online 28 JUL 2018
reservoirs such as the Bakken, Barnett, Eagle Ford, Fayetteville, Haynesville, Marcellus, Wolfcamp, and
Woodford shales contain an appreciable volume fraction of clay minerals. These minerals have a strongly
layered structure, leading to large elastic anisotropy due to strong covalent bonds within layers versus
weaker electrostatic bonds between (Bailey, 1966; Weaver & Pollard, 1973). The presence of anisotropic clay
minerals generally causes shales to also be elastically anisotropic.
Rock physics provides an essential link between elastic rock properties, measured using ultrasonic, sonic, or
seismic methods, and constitutive rock properties such as mineralogic composition, porosity, and pore-fluid
content (Mavko et al., 2009). Since the amplitude variation with angle (AVA) of a seismic P wave reflected
from an interface between two geologic layers depends on the contrast in P impedance (IP = ρVP), S
impedance (IS = ρVS), and density (ρ) between the layers, AVA can be used to extract information about rock
properties. Ursin and Tjäland (1996) show that P and S impedance are the two best-resolved parameters
derivable from inversion of seismic P wave AVA data. Consequently, to obtain rock properties from seismic
inversion, quantitative seismic interpretation commonly employs rock physics templates, which integrate
seismic inversion results with rock physics predictions in the form of graphs, for example, IS versus IP, VP/VS
(or IP/IS) versus IP, μρ (or IS 2) versus λρ (or IP 2-2IS2), and so on (Avseth & Odegaard, 2004; Goodway et al.,
1997; Vernik & Kachanov, 2010).
To construct a rock physics template, the elastic properties of individual rock components are required.
Whereas the elastic properties of pure minerals like quartz, feldspar, calcite, dolomite, pyrite, anhydrite,
etc. have been published (Mavko et al., 2009), the elastic properties of an assemblage of anisotropic clay
minerals, as found in shales, are also required. In this paper, such a mineral assemblage is referred to as
the clay matrix. The elastic anisotropy of the clay matrix is thus important in seismic imaging, reservoir
characterization, drilling, and hydraulic fracturing. It is required not only for rock physics modeling in
support of seismic imaging and AVA interpretation but also for geomechanics applications such as bore-
©2018. American Geophysical Union. hole stability analysis, hydraulic fracture design, and estimation of stress changes due to changes in
All Rights Reserved. pore pressure.

SAYERS AND DEN BOER 5965

 21699356, 2018, 7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2018JB015600 by CAPES, Wiley Online Library on [18/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Solid Earth 10.1029/2018JB015600

Table 1 In this paper, we present a method for estimating the elastic stiffness
Comparison of a Best-Fitting TI Medium (Sayers & den Boer, 2016) Determined tensor of the clay matrix in shales as a function of just four parameters:
From First-Principle Calculations of Militzer et al. (2011), With Elastic Stiffnesses
of a Best-Fitting TI Medium For Muscovite, Determined From The Measurements
the elastic moduli of a soft interparticle medium containing bound
Of Alexandrov and Ryzhova (1961), McNeil and Grimsditch (1993), and Vaughan water, and the volume fraction and aspect ratio of stiff clay particles.
and Guggenheim (1986) These parameters are derived via inversion of available data. The elastic
Elastic Militzer Alexandrov & McNeil & Vaughan & stiffnesses of the clay matrix are then predicted for arbitrary values of
stiffness (GPa) et al. Ryzhova Grimsditch Guggenheim these parameters, enabling the resulting rock physics model to be
applied in general.
C11 = C22 181.3 178.0 180.7 184.3
C33 60.1 54.9 60.9 59.1
C44 = C55 20.3 12.2 14.1 16.8 2. The Anisotropic Elastic Properties of Clay
C66 66.3 67.8 68.0 69.5
C12 48.8 42.4 44.8 45.3 Unfortunately, experimental measurements of single-crystal elastic
C13 = C23 24.8 14.5 21.6 22.7 moduli are not available for clay minerals, due to technical difficulties
Note. TI = transversely isotropic. associated with their very small grain size (Katahara, 1996). However,
the complete elastic tensors of several dry clay minerals and micas,
including kaolinite, illite-smectite, and muscovite, have been derived by Militzer et al. (2011), using
first-principle quantum mechanical calculations. The composition and crystallographic structure of illite are
similar to muscovite (Katahara, 1996; Rothbauer, 1971; Tosaya, 1982), for which the elastic moduli have been
measured (Alexandrov & Ryzhova, 1961; McNeil & Grimsditch, 1993; Vaughan & Guggenheim, 1986). Owing
to their strongly layered structure, the elastic stiffness tensor of clay minerals can be approximated to good
accuracy as a transversely isotropic (TI) medium, with rotational symmetry axis perpendicular to the layers
(Sayers & den Boer, 2016). A TI medium has five independent elastic stiffnesses. Using conventional two-
index notation (Nye, 1985), and assuming coordinate axis x3 is the rotational symmetry axis, the nonzero elas-
tic stiffness coefficients are C11 = C22, C33, C12 = C21, C13 = C31 = C23 = C32, C44 = C55, and C66 = (C11-C12)/2.
Table 1 compares the elastic stiffnesses of muscovite for the best-fitting TI medium, derived by Sayers and
den Boer (2016) from measurements of Alexandrov and Ryzhova (1961), McNeil and Grimsditch (1993),
and Vaughan and Guggenheim (1986), with the best-fitting TI medium derived by Sayers and den Boer
(2016) from Militzer et al., indicating good agreement.
Ulm and Abousleiman (2006) and Ortega et al. (2007) combined ultrasonic velocity measurements with nano-
indentation and microindentation measurements of the clay matrix, to deduce its elastic properties at the
scale of a few thousand to few hundred nanometers. Nano-indentation and microindentation are techniques
for determining the elastic properties of a natural composite material from its response to reversal of contact
loading. An indenter such as diamond, with known mechanical properties and geometry, is used to measure
the continuous change in indentation depth as a function of changing load. A continuum scale mechanical
model is then applied to derive the elastic stiffnesses from indentation modulus and hardness. By calibrating
these nano-indentation and microindentation measurements with ultrasonic velocity measurements, which
sample elastic stiffness at a larger scale, Ulm and coworkers were able to experimentally derive the effective
elastic stiffnesses of the clay matrix. The objective of their analysis was to determine how the observed ani-
sotropy of the clay matrix relates to that of its constituent clay minerals.
These authors found that above the percolation threshold, the clay matrix can be described as a TI medium,
characterized by a unique and evidently universal set of elastic stiffnesses: c11 = c22 = 44.9 GPa, c33 = 24.2 GPa,
c13 = c23 = 18.1 GPa, c44 = c55 = 3.7 GPa, and c66 = 11.6 GPa. Moreover, these stiffnesses appear to be little
affected by the specific clay minerals present. Comparing them with the elastic stiffnesses of muscovite
(Table 1) reveals that they are significantly smaller. This discrepancy is also observed for other clay minerals.
Table 2 shows the best-fitting TI medium obtained from first-principle calculations of Militzer et al. (2011) for
dry illite-smectite and kaolinite, together with the elastic stiffnesses found by Ulm and coworkers (Ortega
et al., 2007; Ulm & Abousleiman, 2006). It is thus evident that the elastic stiffnesses of these minerals are simi-
larly much greater than those of the clay matrix.
Rather than the elastic stiffness of the constituent clay minerals, Ulm and coworkers (Ortega et al., 2007; Ulm
& Abousleiman, 2006) concluded that the contact regions between clay platelets and clay packing density
determine the mechanical behavior of shales. Accordingly, they suggest that shale anisotropy results from
depositional anisotropy of clay platelets. However, in addition to contacts between clay particles, interparticle
regions in shales are also expected to contain clay-bound water.

SAYERS AND DEN BOER 5966

 21699356, 2018, 7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2018JB015600 by CAPES, Wiley Online Library on [18/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Solid Earth 10.1029/2018JB015600

Table 2 In this paper, a simple method is proposed for describing the impact
Comparison of Elastic Stiffnesses for a Best-Fitting TI Medium (Sayers & den Boer, of these interparticle regions between clay platelets. Owing to its sim-
2016) Derived From First-Principle Calculations of Militzer et al. (2011) for Dry
Illite-Smectite, Kaolinite And Muscovite, With Elastic Stiffnesses of the Clay Matrix
plicity and robustness, it is anticipated that this model will provide
Deduced by Ulm and coworkers (Ortega et al., 2007; Ulm & Abousleiman, 2006) insight into the elastic properties of clay in shales. The model may
thus find application in the rock physics calculations required to sup-
Elastic stiffness Illite- Ideal Kaolinite Clay
(GPa) Smectite kaolinite 2M
a
Muscovite matrix port seismic imaging, AVO analysis and geomechanics applications,
for example.
C11 = C22 180.5 185.9 186.1 181.3 44.9
C33 67.6 83.7 82.5 60.1 24.2
C44 = C55 21.3 14.8 14.7 20.3 3.7 3. Elastic Stiffness Model of Clay in Shales
C66 69.4 58.8 59.1 66.3 11.6
C12 41.7 68.2 67.8 48.8 21.7The anisotropic clay fabric of shales is the source of their anisotropy.
C13 = C23 17.7 5.3 5.9 24.8 18.1Electron microscopy studies typically show that individual clay plate-
a lets in shales vary in orientation, but are locally aligned (Hornby et al.,
Standard mineralogical notation for a specific crystal structure of kaolinite.
1994; Schoenberg et al., 1996; Swan et al., 1989). In their studies of an
illite clay, Aylmore and Quirk (1960) found that the porosity was
mostly contained in pores with radius less than 42 Å. Since the platelets of this mineral are 700 Å across, much
larger than the platelet separation, these authors view the pore radius as a measure of platelet separation.
They conclude that illite clay may be regarded as a series of parallel crystals, whose separation varies between
10 to 30 Å, although smaller separations may occur at some points. Aylmore and Quirk (1959) refer to a local
parallel arrangement of clay platelets as a domain. Their findings indicate that the pores in illite clay reside
mostly in domains, with little interdomain porosity.
Figure 1 shows a schematic representation of clay platelets in shale. Actual scanning electron microscope
(SEM) images showing local alignment of clay particles can be found in Hornby et al. (1994) and Vernik
and Milovac (2011), for example. While the occurrence of planar or curved platelets is dictated by the stress
and depositional history of the host rock, curved platelets are common, since their very thin nature renders
them susceptible to deformation as stress increases during burial. Numerical simulation confirms that clay
platelets are flexible rather than rigid, and that alignment of platelets occurs during the process of compac-
tion (Sherwood & Van Damme, 1994). Clay platelets are aligned locally, but vary in orientation from domain to
domain, in a way determined by the depositional and stress history of the rock.
To investigate the elastic properties of the clay matrix, we assume that clay platelets in a domain can be
represented as oblate spheroids embedded within a softer interparticle region. The aspect ratio of these
oblate spheroidal inclusions is c/a, where c is the polar dimension and a is the equatorial dimension.
Given a material comprised of two phases, whose small-scale geometry can be described by statistical distribu-
tion functions, a method proposed by Ponte Castañeda and Willis (1995) allows the effective elastic moduli
of the composite medium to be estimated. Their method offers the advantages of not only being closed form,
but also that the shape of inclusions may be independent of their spatial distribution. Accordingly, to calculate
the effective elastic stiffnesses C eff
ijkl of clay, including the interparticle regions, clay platelets are assumed to be
identical oblate spheroidal inclusions aligned with minor axes parallel to axis x3, and having a fixed elastic stiff-
ð1 Þ
ness tensor C ijkl . For the case when the distribution of inclusions is described by a two-point correlation

Figure 1. Schematic representation of domains of locally aligned clay platelets in a vertical cross section of shale. Irregular
shaped ovals represent silt grains on the order of 5–10 μm across. Both planar (undeformed) and curved (deformed) pla-
telets are visible. Spaces between platelets correspond to a soft interclay medium (clay matrix) containing clay-bound
water.

SAYERS AND DEN BOER 5967

 21699356, 2018, 7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2018JB015600 by CAPES, Wiley Online Library on [18/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Solid Earth 10.1029/2018JB015600

function that is itself ellipsoidal, with the same ratio of principal axes as the inclusion, the expression for the
effective elastic stiffness tensor C eff
ijkl given by Ponte Castañeda and Willis (1995) reduces to that of Willis
(1977, 1978):
  1
ð0 Þ ð1 Þ ð0Þ 1
ijkl ¼ C ijkl þ υ
C eff C ijkl  C ijkl þ ð1  υÞPijkl : (1)

ð0 Þ
Here υ denotes the volume fraction of the inclusions, and C ijkl is the effective elastic stiffness tensor of the
interparticle region. The components of the fourth-rank tensor Pijkl are given by
 0
 0
Pijkl ¼ ∫V Gik;lj x  x dx (2)
ðij ÞðklÞ

Subscripts l and j denote spatial derivatives with respect to coordinates xl and xj, while symbol (ij) denotes
symmetrification over the included indices, ij. The elastic Green’s function G(x) corresponds to that of an infi-
nite anisotropic medium, and integration is over the inclusion volume V. If underlying statistical assumptions
are honored, equation (1) constitutes a lower bound when the inclusions are the stiffest phase, and an upper
bound when they are the most compliant phase. If statistical assumptions are only approximately satisfied,
equation (1) provides instead an estimate of the elastic properties. This estimate is identical to that obtained
using the effective field theory of Sevostianov et al. (2005) for the same assumed shape and spatial distribu-
tion of inclusions, which has been shown to give a good description of the elastic properties of organic-rich
shales (Sayers, 2013).
The interparticle medium is assumed to be isotropic, with bulk modulus K and shear modulus μ. Assuming
the elastic stiffnesses of clay minerals as given in Tables 1 and 2, estimates of K, μ, volume fraction υ, and
aspect ratio c/a of clay platelets can be jointly determined from equation (1), by assuming C eff
ijkl is equal to
the elastic stiffness of Ulm and coworkers (Ortega et al., 2007; Ulm & Abousleiman, 2006), and assuming
ð1Þ
C ijkl is equal to the elastic stiffness tensor of the clay mineral.

4. Results
Employing the assumptions and approach of the previous section, optimal values of K, μ, υ, and c/a were
obtained numerically, by minimizing an objective function F (equation (3)). The objective function is the
sum of the squared difference between C eff
ijkl , predicted by equation (1), and the elastic stiffnesses of the clay
matrix Cijkl(Table 2), given by Ulm and coworkers (Ortega et al., 2007; Ulm & Abousleiman, 2006). In two-index
notation (Nye, 1985), F is given by

2
2
2
2
2
11  C 11
F ¼ C eff þ C eff
33  C 33 þ C eff
13  C 13 þ C eff
55  C 55 þ C eff
66  C 66 (3)

The parameter intervals containing a global minimum value of F were initially determined via an exhaustive
brute-force grid search, by uniformly sampling each parameter within a specified nominal interval and eval-
uating F for every possible distinct combination of parameter values. Bulk and shear modulus were searched
over the intervals 1.0 to 5.0 GPa, and 0.0 to 1.0 GPa respectively, while volume fraction and aspect ratio were
searched over the intervals 0.5 to 1.0 and 0.0 to 0.5, respectively. This initial grid search demonstrated that a
global minimum of F lies within a narrow range of aspect ratios, between 0.2 and 0.25, irrespective of the
other parameter values. Too small an aspect ratio results in overprediction of C11 and C66 compared to the
values of Ulm and coworkers (Ortega et al., 2007; Ulm & Abousleiman, 2006), whereas too large an aspect
ratio results in underprediction of the elastic anisotropy of the clay matrix. The fact that the optimal aspect
ratio is confined to such a narrow range, independent of the other parameters, effectively reduces the
number of parameters to be minimized.
It was further found that optimal values of the volume fraction and shear modulus are similarly confined
to relatively narrow intervals, between 0.9 to unity, and 0.1 to 0.25 GPa respectively. For bulk modulus K,
multiple minima are observed between 1.0 to 3.0 GPa, all yielding similar objective function values.

SAYERS AND DEN BOER 5968

 21699356, 2018, 7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2018JB015600 by CAPES, Wiley Online Library on [18/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Solid Earth 10.1029/2018JB015600

However, given that the interclay medium is expected to have prop-

erties similar to water, it is reasonable to expect K greater than
2 GPa, corresponding to a fully saturated state, whereas K less than
2 GPa represents partial saturation. Accordingly, an interval between
2.0 and 3.0 GPa was chosen for bulk modulus. Figure 2 shows the
variation in F as a function of aspect ratio c/a for optimal values of
K, μ, and volume fraction υ, assuming the best-fitting TI elastic stiff-
nesses of muscovite derived by Sayers and den Boer (2016) from
the first-principle calculations of Militzer et al. (2011) for the
clay particle.
Having identified an interval bracketing the global minimum value of F
for each parameter, the exact location of the minimum was then deter-
Figure 2. Plot of objective function F (equation (3)) versus aspect ratio c/a. A glo- mined using the quasi-Newton Broyden–Fletcher–Goldfarb–Shanno
bal minimum occurs near c/a = 0.23, irrespective of other parameter values. (BFGS) minimization algorithm (Battiti & Masulli, 1990). To test the
robustness of the underlying assumptions, four different sets of mus-
covite elastic stiffnesses (Table 1) were employed, corresponding to the best-fitting TI medium derived
by Sayers and den Boer (2016) from the experimental measurements of Alexandrov and Ryzhova
(1961), McNeil and Grimsditch (1993) and Vaughan and Guggenheim (1986) and the first-principle calcu-
lations of Militzer et al. (2011) based on density functional theory. Table 3 lists the best-fit parameters
thus derived.
Similarly, Table 4 lists the best-fit values for K, μ, υ, and c/a obtained from equation (1) assuming the
elastic stiffnesses of the clay mineral as given by the best-fitting TI elastic stiffnesses (Sayers & den Boer,
2016) for dry illite-smectite, and kaolinite, derived from the first-principle calculations of Militzer et al.
(2011; Table 2).
Having determined the optimal values of K, μ, υ, and c/a shown above, the variation in effective properties
as they diverge from these optimal values can be analyzed. For example, the curves in Figure 3 show the
variation in elastic stiffnesses as a function of clay mineral volume fraction, assuming the best-fitting TI
elastic stiffnesses for muscovite derived from Militzer et al. (2011), a clay platelet aspect ratio of 0.227,
and a bulk and shear modulus of the interparticle medium of 2.610 and 0.227 GPa respectively. Similarly,
the curves in Figure 4 show the variation of elastic stiffnesses as a function of clay platelet aspect ratio,
assuming the best-fitting TI elastic stiffnesses for muscovite derived from Militzer et al. (2011), a clay mineral
volume fraction of 0.925, and a bulk and shear modulus of the interparticle medium of 2.610 and
0.227 GPa, respectively.
Finally, the curves in Figure 5 show how the predicted Thomsen (1986) anisotropy parameters ε, γ, and δ,
and the anellipticity parameter η of Alkhalifah and Tsvankin (1995) vary as the aspect ratio of the clay pla-
telets changes, assuming the best-fitting TI elastic stiffnesses for muscovite derived from Militzer et al.
(2011), with a clay mineral volume fraction of 0.925, and a bulk and shear modulus of the interparticle
medium of 2.610 and 0.227 GPa, respectively. These anisotropy parameters are defined by the
following equations:

Table 3
Best-Fit Bulk Modulus K and Shear Modulus μ of the Interparticle Medium, Plus Volume Fraction υ and Aspect Ratio c/a of Clay
Platelets, Obtained From Equation (1) by Assuming the Elastic Stiffness of Ulm and Coworkers (Ortega et al., 2007; Ulm &
Abousleiman, 2006) as the Clay Matrix (Table 2), and the Best-Fitting TI Elastic Stiffness of Dry Muscovite as the Clay
Mineral (Table 1)
Best-fit Alexandrov & McNeil & Vaughan & Standard
parameter Militzer et al. Ryzhova Grimsditch Guggenheim Average deviation

K (GPa) 2.610 2.240 2.670 2.160 2.420 0.258

μ (GPa) 0.227 0.169 0.230 0.183 0.202 0.031
υ 0.925 0.945 0.925 0.940 0.934 0.010
c/a 0.227 0.242 0.229 0.237 0.234 0.007
Note. TI = transversely isotropic.

SAYERS AND DEN BOER 5969

 21699356, 2018, 7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2018JB015600 by CAPES, Wiley Online Library on [18/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Solid Earth 10.1029/2018JB015600

Table 4 C 11  C 33 C 66  C 55
Best-Fit Bulk Modulus K and Shear Modulus μ of the Interparticle Medium, Plus ε¼ ;γ ¼ ; (4)
2C 33 2C 55
Volume Fraction υ and Aspect Ratio c/a of Clay Platelets, Obtained From
ðC 13 þ C 33 ÞðC 13 þ 2C 55  C 33 Þ εδ
Equation (1) by Assuming the Elastic Stiffness of Ulm and Coworkers (Ortega δ¼ ;η ¼ :
et al., 2007; Ulm & Abousleiman, 2006) as the Clay Matrix, and the Best-Fitting TI 2C 33 ðC 33  C 55 Þ 1 þ 2δ
Elastic Stiffness Derived From Militzer et al. (2011) for Dry Illite-Smectite and
Kaolinite as the Clay Mineral (Table 2) The dashed lines in Figure 5 indicate the anisotropy parameters calcu-
Best-fit Illite- Ideal Kaolinite Standard lated from the elastic stiffnesses of the clay matrix obtained by Ulm and
a
parameter Smectite kaolinite 2M Average deviation coworkers (Ortega et al., 2007; Ulm & Abousleiman, 2006).
K (GPa) 2.300 2.208 2.390 2.299 0.091 Whereas anisotropy parameters ε and γ are related in a simple way
μ (GPa) 0.181 0.174 0.190 0.182 0.008 to the velocity of wave propagation parallel and perpendicular to the
υ 0.935 0.935 0.930 0.933 0.003 symmetry axis of a TI medium, anisotropy parameter δ is more com-
c/a 0.203 0.188 0.189 0.193 0.008
plex. However, δ explains the difference between small offset normal-
a
Standard mineralogical notation for a specific crystal structure of kaolinite. moveout velocity and vertical velocity of P waves, as well as small-offset
PP-AVO response (Thomsen, 1986). The anellipticity parameter η
describes the divergence of quasi-SV phase slowness from a circle, and quasi-P phase slowness from an
ellipse. Here “quasi” refers to the fact that in an anisotropic medium, the polarization of the P wave may
not be parallel to the direction of propagation, and the polarization of the SV-wave (the shear wave with
polarization in the vertical plane) may not be perpendicular to the direction of propagation.

5. Discussion
As evidenced by Table 3, the results obtained using the four different sets of elastic stiffnesses for muscovite
are all remarkably consistent. In each case, the volume fraction of clay is approximately 0.934, the aspect ratio
of clay particles is about 0.234, while the bulk and shear modulus of the interparticle medium are of order
2.42 and 0.20 GPa, respectively. The volume fraction occupied by the interparticle medium is thus of order
0.066, with a bulk modulus similar to that of water (2.25 GPa), while the shear modulus of the interparticle
medium is seen to be nonzero. Katahara (1996) points out that illite particles are typically 7 nm thick. An
adsorbed monolayer of water of thickness 0.3 nm per particle thus implies 4% adsorbed water by volume,
which is of the same order as the estimates in Table 3.
The aspect ratio of clay particles of 0.234 shown in Table 3 is significantly higher than typical aspect
ratios for clay platelets of between 0.1 and 0.05. One explanation for these higher aspect ratios is that
several clay platelets may act together as a stack of clay particles. Another possibility is that since clay

Figure 3. Predicted elastic stiffnesses (curves) as a function of clay mineral volume fraction, assuming the best-fitting
transversely isotropic elastic stiffnesses for muscovite derived from Militzer et al. (2011), with a clay platelet aspect ratio
of 0.227, and a bulk and shear modulus of the interparticle medium of 2.610 and 0.227 GPa, respectively. Dashed lines
indicate elastic stiffnesses of the clay matrix deduced by Ulm and coworkers (Ortega et al., 2007; Ulm & Abousleiman, 2006).

SAYERS AND DEN BOER 5970

 21699356, 2018, 7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2018JB015600 by CAPES, Wiley Online Library on [18/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Solid Earth 10.1029/2018JB015600

Figure 4. Predicted elastic stiffnesses (curves) as a function of clay platelet aspect ratio, assuming the best-fitting transver-
sely isotropic elastic stiffnesses for muscovite derived from Militzer et al. (2011), with a clay mineral volume fraction of
0.925, and a bulk and shear modulus of the interparticle medium of 2.610 and 0.227 GPa, respectively. Dashed lines denote
elastic stiffnesses of the clay matrix deduced by Ulm and coworkers (Ortega et al., 2007; Ulm & Abousleiman, 2006).

platelets tend to be curved rather than planar, an appropriate value of the horizontal dimension a is
the length over which the local tangent to curvature is a reasonable approximation to the clay platelet
(see Figure 1).
The similarity of the results obtained for dry illite-smectite and kaolinite (Table 4) to those obtained for
muscovite (Table 3) is striking. This agreement supports the conclusion of Ulm and Abousleiman (2006), that
the elastic stiffnesses of the interparticle medium are independent of the particular clay mineral forming the
platelets, and hence universal. Accordingly, such consistent reproduction of K, μ, υ, and c/a for different
clay minerals suggests that the underlying assumptions of the proposed model are both physically sound
and robust.

Figure 5. Predicted Thomsen anisotropy parameters ε, γ, and δ, and anellipticity η of Alkhalifah and Tsvankin (curves) as a
function of clay platelet aspect ratio, assuming the best-fitting transversely isotropic elastic stiffnesses for muscovite
derived from Militzer et al. (2011), with a clay mineral volume fraction of 0.925, and a bulk and shear modulus of the
interparticle medium of 2.610 and 0.227 GPa, respectively. Dashed lines denote anisotropy parameters calculated from
elastic stiffnesses of the clay matrix deduced by Ulm and coworkers (Ortega et al., 2007; Ulm & Abousleiman, 2006).

SAYERS AND DEN BOER 5971

 21699356, 2018, 7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2018JB015600 by CAPES, Wiley Online Library on [18/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Solid Earth 10.1029/2018JB015600

Figure 3 shows how the elastic stiffnesses of the clay matrix vary as a function of clay mineral volume fraction.
Good agreement is seen between the predicted elastic stiffnesses and those of Ulm and coworkers (Ortega
et al., 2007; Ulm & Abousleiman, 2006), for a clay mineral volume fraction of 0.925. In contrast to elastic stiff-
nesses C11, C33, C12, C55, and C66, which all decrease monotonically as the volume fraction of the interclay
medium increases, C13 initially increases, then declines.
Figure 4 shows how the elastic stiffnesses of the clay matrix vary as the clay platelet aspect ratio changes.
Again, good agreement is observed between the predicted elastic stiffnesses and those of Ulm and
coworkers (Ortega et al., 2007; Ulm & Abousleiman, 2006), for a clay mineral aspect ratio of 0.227. The in-plane
elastic stiffnesses C11, C12, and C66 decrease monotonically as the aspect ratio increases, while C13 and C55
increase. Stiffness C33 shows a more complicated behavior, initially decreasing then increasing. However,
in contrast to the other elastic stiffnesses, C33 is relatively insensitive to clay platelet aspect ratio.
Figure 5 shows how the predicted Thomsen (1986) anisotropy parameters ε, γ, and δ, and the anellipticity
parameter η of Alkhalifah and Tsvankin (1995), vary as a function of clay platelet aspect ratio. Whereas
anisotropy parameters ε, γ, and η are all positive, and decrease as the aspect ratio increases, δ is initially
negative for the clay mineral, gradually becoming less negative as the aspect ratio increases, before finally
turning positive. This behavior helps explain why δ tends to be positive for shales, but negative for
clay minerals.
Due to their anisotropic rock fabric, shales cause seismic wave velocities to vary with direction, leading to seis-
mic anisotropy. Reliable seismic estimation of reservoir fluid, lithology, and pore pressure thus depends on
accurate determination of Thomsen’s parameters for correct quantification of shale anisotropy. Thomsen’s
parameters are also required for many basic seismic processing operations, such as normal-moveout
(NMO) and dip-moveout (DMO) corrections, migration, time-to-depth conversion and amplitude variation
with offset (AVO) analysis. Failing to account correctly for seismic anisotropy in these processes leads to
significant data processing errors.
Shale anisotropy also impacts the variation of rock stresses, which in turn control the propagation and con-
tainment of hydraulically induced rock fractures. Application of an isotropic stress model to anisotropic rocks
like shales, leads to incorrect prediction of stress barriers, and hence failure to predict fracture containment.
Thus, Thomsen’s parameters are also needed for correct determination of minimum and maximum horizon-
tal stress, without which hydraulic fractures cannot be reliably designed.
Molecular dynamics calculations of Skipper et al. (1991) and Karaborni et al. (1996) show structurally ordered,
bound, or adsorbed water molecules near clay surfaces. To study the effect of interlayer water on the elastic
properties of smectite, Ebrahimi et al. (2012) used molecular dynamics to calculate the elastic stiffness coeffi-
cients of Na-montmorillonite (a member of the smectite group) as a function of basal layer spacing and
hydration state. During formation of the first water layer, Ebrahimi et al. (2012) observe an increase in shear
stiffness, suggesting that water bonded to the clay surface may have an ordered structure with nonzero shear
modulus. Holt and Kolstø (2017) argue that elastic wave propagation will be affected by this bound or
adsorbed water, and that the bound water will have an enhanced viscosity, resulting in an effective shear
modulus at finite frequency. Holt and Kolstø (2017) use a generalized Maxwell model to relate the real part
of the shear modulus of bound water to the high-frequency limit of bound-water shear stiffness μBW(∞),
the viscosity ηBW and angular frequency ω:

μBW ð∞Þ
μBW ðωÞ ¼ (5)
1 þ ðμBW ð∞Þ=ωηBW Þ2

Assuming μBW(∞) is finite, equation (5) thus implies that this enhanced viscosity will be a source of dispersion
and attenuation.
Finally, it is worth noting that the method presented in this paper explicitly accounts for the particulate
nature of shales and the discontinuities between clay particles. Many current models presume individual clay
particles are connected, whereas electron microscopy images clearly show this is not the case. If the inherent
anisotropy of clay particles is correctly taken into account, such models will overpredict the in-plane elastic
stiffness coefficients C11, C12, and C66.

SAYERS AND DEN BOER 5972

 21699356, 2018, 7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2018JB015600 by CAPES, Wiley Online Library on [18/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Solid Earth 10.1029/2018JB015600

6. Conclusion
Since many unconventional reservoirs contain not only enormous quantities of hydrocarbons but also a
significant volume of clay minerals, there is much interest in trying to derive reservoir properties from seismic
inversion. However, reliable estimation of rock properties requires a determination of the elastic moduli and
density of rock constituents, which in turn requires a rock physics model that can correctly predict the elastic
properties of an assemblage of anisotropic clay minerals. Clay minerals exhibit strong elastic anisotropy due
to their layered structure, which causes shale properties to also be elastically anisotropic. Yet, due to the pre-
sence of more compliant regions between clay platelets, the elastic stiffness of clay in shales is significantly
lower than that of the constituent clay minerals.
We present a model for the elastic properties of clay in shales, in which clay platelets are assumed to be
embedded within a softer interparticle matrix, consisting of clay-bound water and interparticle contacts.
Inversion for the elastic properties of this interparticle region indicates that its effective bulk modulus is simi-
lar to water, while its shear modulus is an order of magnitude lower than the bulk modulus, consistent with
the expected properties of clay-bound water. Once these properties are determined, the rock physics model
can be extended by predicting the elastic stiffnesses of the clay matrix for other values of the clay particle
volume fraction and aspect ratio and/or elastic moduli of the interparticle medium. Moreover, this flexibility
means the model can also account not only for variations in fluid saturation, but even for different fluid satur-
ants. The remarkable consistence with which the inverted properties are reproduced, regardless of the parti-
cular clay mineral assumed as platelets, suggests the model is physically sound and robust. This supports the
conclusion of earlier studies that the elastic properties of the clay matrix are universal, and hence indepen-
dent of clay mineralogy. The proposed method explicitly takes into account the particulate nature of shales
and the discontinuities between clay particles. Models which assume an interconnected clay phase will over-
predict the in-plane elastic stiffness coefficients C11, C12, and C66, if the anisotropy of clay particles is correctly
taken into account.
Accordingly, owing to its simplicity and wide range of potential applications, it is anticipated that this com-
pact model of shale, based on the properties of both clay minerals and the interparticle medium, may be use-
ful in rock physics for seismic imaging, AVO analysis and geomechanical applications such as borehole
stability analysis, hydraulic fracture design, and estimation of stress changes due to petroleum production.

Acknowledgments References
The experimental data for muscovite
used in this paper are listed in column 3 Alexandrov, K. S., & Ryzhova, T. V. (1961). Elastic properties of rock-forming minerals, II: Layered silicates. Izvestiya Academy of Sciences USSR,
of Table 2 in Alexandrov and Ryzhova Geophysics Series, 9, 1165–1168.
(1961), column 2 of Table 1 in McNeil Alkhalifah, T., & Tsvankin, I. (1995). Velocity analysis for transversely isotropic media. Geophysics, 60(5), 1550–1566. https://doi.org/10.1190/
and Grimsditch (1993), and column 2 of 1.1443888
Table 3 in Vaughan and Guggenheim Avseth, P. A., & Odegaard, E. (2004). Well log and seismic data analysis using rock physics templates. First Break, 22, 37–43.
(1986). The elastic stiffnesses of the clay Aylmore, L. A. G., & Quirk, J. P. (1959). Swelling of clay-water systems. Nature, 183(4677), 1752–1753. https://doi.org/10.1038/1831752a0
matrix are obtained from column 2 of Aylmore, L. A. G., & Quirk, J. P. (1960). Domain or turbostratic structure of clays. Nature, 187(4742), 1046–1048. https://doi.org/10.1038/
Table 4 of Ortega et al. (2007). 1871046a0
Calculated elastic stiffnesses for the dry Bailey, S. W. (1966). The status of clay mineral structures. In Proceedings of the 14th National Conference on Clays and Clay Minerals (pp. 1–23).
clay minerals studied by Militzer et al. Battiti, R., & Masulli, F. (1990). BFGS optimization for faster automated supervised learning. In Proceedings of the International Neural Network
(2011) were obtained from their paper Conference (INNC 90) (pp. 757–760). Paris, France.
as follows: column 5 of Table 2 for Ebrahimi, D., Pellenq, R. J. M., & Whittle, A. J. (2012). Nanoscale elastic properties of montmorillonite upon water adsorption. Langmuir, 28(49),
muscovite, column 2 of Table 3 for illite- 16,855–16,863. https://doi.org/10.1021/la302997g
smectite, column 5 of Table 4 for ideal Goodway, B., Chen, T., & Downton, J. (1997). Improved AVO fluid detection and lithology discrimination using Lamé petrophysical
kaolinite, and column 7 of Table 5 for parameters “λρ”, “μρ”, “λμ fluid stack” from P and S Inversions. 1997 SEG Annual Meeting, Society of Exploration Geophysicists.
kaolinite 2 M*. Holt, R. M., & Kolstø, M. I. (2017). How does water near clay mineral surfaces influence the rock physics of shales? Geophysical Prospecting,
65(6), 1615–1629. https://doi.org/10.1111/1365-2478.12503
Hornby, B. E., Schwartz, L. M., & Hudson, J. A. (1994). Anisotropic effective-medium modeling of the elastic properties of shales. Geophysics,
59(10), 1570–1583. https://doi.org/10.1190/1.1443546
Karaborni, S., Smit, B., Heidug, W., Urai, J., & van Oort, E. (1996). The swelling of clays: Molecular simulations of the hydration of montmor-
illonite. Science, 271(5252), 1102–1104. https://doi.org/10.1126/science.271.5252.1102
Katahara, K. W. (1996). Clay mineral elastic properties. 1996 SEG Annual Meeting. Society of Exploration Geophysicists.
Mavko, G., Mukerji, T., & Dvorkin, J. (2009). The rock physics handbook: Tools for seismic analysis of porous media. Cambridge: Cambridge
University Press.
McNeil, L. E., & Grimsditch, M. (1993). Elastic moduli of muscovite mica. Journal of Physics: Condensed Matter, 5, 1681–1690.
Militzer, B., Wenk, H.-R., Stackhouse, S., & Stixrude, L. (2011). First-principles calculation of the elastic moduli of sheet silicates and their
application to shale anisotropy. American Mineralogist, 96(1), 125–137. https://doi.org/10.2138/am.2011.3558
Nye, J. F. (1985). Physical properties of crystals. Oxford: Oxford University Press.
Ortega, J. A., Ulm, F.-J., & Abousleiman, Y. (2007). The effect of the nanogranular nature of shale on their poroelastic behavior. Acta
Geotechnica, 2(3), 155–182. https://doi.org/10.1007/s11440-007-0038-8

SAYERS AND DEN BOER 5973

 21699356, 2018, 7, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2018JB015600 by CAPES, Wiley Online Library on [18/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of Geophysical Research: Solid Earth 10.1029/2018JB015600

Paul, D. R., & Robeson, L. M. (2008). Polymer nanotechnology: Nanocomposites. Polymer, 49(15), 3187–3204. https://doi.org/10.1016/j.
polymer.2008.04.017
Ponte Castañeda, P., & Willis, J. R. (1995). The effect of spatial distribution on the effective behavior of composite materials and cracked
media. Journal of the Mechanics and Physics of Solids, 43(12), 1919–1951. https://doi.org/10.1016/0022-5096(95)00058-Q
Rothbauer, R. (1971). Untersuchung eines 2M1-muskovits mit neutronenstrahlen. Neues Jahrbuch für Mineralogie Monatshefte, 4, 143–154.
Sayers, C. M. (2013). The effect of kerogen on the elastic anisotropy of organic-rich shales. Geophysics, 78(2), D65–D74. https://doi.org/
10.1190/geo2012-0309.1
Sayers, C. M., & den Boer, L. D. (2016). The elastic anisotropy of clay minerals. Geophysics, 81(5), C193–C203. https://doi.org/10.1190/geo2016-
0005.1
Schoenberg, M., Muir, F., & Sayers, C. M. (1996). Introducing ANNIE: A simple three-parameter anisotropic velocity model for shales. Journal of
Seismic Exploration, 5, 35–49.
Sevostianov, I., Yilmaz, N., Kushch, V., & Levin, V. (2005). Effective elastic properties of matrix composites with transversely-isotropic phases.
International Journal of Solids and Structures, 42(2), 455–476. https://doi.org/10.1016/j.ijsolstr.2004.06.047
Sherwood, J. D., & Van Damme, H. (1994). Nonlinear compaction of an assembly of highly deformable platelike particles. Physical Review E,
The American Physical Society, 50, 3834–3840.
Skipper, N. T., Refson, K., & McConnell, J. D. C. (1991). Computer simulation of interlayer water in 2:1 clays. The Journal of Chemical Physics,
94(11), 7434–7445. https://doi.org/10.1063/1.460175
Swan, G., Cook, J., Bruce, S., & Meehan, R. (1989). Strain rate effects in Kimmeridge Bay shale. International Journal of Rock Mechanics and
Mining Sciences and Geomechanics, 26(2), 135–149. https://doi.org/10.1016/0148-9062(89)90002-8
Thomsen, L. (1986). Weak elastic anisotropy. Geophysics, 51(10), 1954–1966. https://doi.org/10.1190/1.1442051
Tosaya, C. A. (1982). Acoustical properties of clay-bearing rocks, (PhD thesis). Stanford University.
Ulm, F.-J., & Abousleiman, Y. (2006). The nanogranular nature of shale. Acta Geotechnica, 1(2), 77–88. https://doi.org/10.1007/s11440-006-
0009-5
Ursin, B., & Tjäland, E. (1996). The information content of the elastic reflection matrix. Geophysical Journal International, 125(1), 214–228.
https://doi.org/10.1111/j.1365-246X.1996.tb06547.x
Vaughan, M. T., & Guggenheim, S. (1986). Elasticity of muscovite and its relationship to crystal structure. Journal of Geophysical Research,
91(B5), 4657–4664. https://doi.org/10.1029/JB091iB05p04657
Vernik, L., & Kachanov, M. (2010). Modeling elastic properties of siliciclastic rocks. Geophysics, 75(6), E171–E182. https://doi.org/10.1190/
1.3494031
Vernik, L., & Milovac, J. (2011). Rock physics of organic shales. The Leading Edge, 30(3), 318–323. https://doi.org/10.1190/1.3567263
Weaver, C. E., & Pollard, L. D. (1973). The chemistry of clay minerals. Amsterdam, London, New York: Elsevier.
Willis, J. R. (1977). Bounds and self-consistent estimates for the overall moduli of anisotropic composites. Journal of the Mechanics and Physics
of Solids, 25(3), 185–202. https://doi.org/10.1016/0022-5096(77)90022-9
Willis, J. R. (1978). Variational principles and bounds for the overall properties of composites. In J. W. Provan (Ed.), Continuum models for
discrete systems (pp. 185–215). Waterloo, Canada: University of Waterloo Press.

SAYERS AND DEN BOER 5974