Content uploaded by Robert H. Lander
Author content
All content in this area was uploaded by Robert H. Lander on Jun 08, 2015
Content may be subject to copyright.
A model for fibrous illite
nucleation and growth
in sandstones
Robert H. Lander and Linda M. Bonnell
ABSTRACT
We have developed a model for the formation of fibrous illite
in sandstones where kaolinite is a primary reactant and po-
tassium is derived from in-situ K-feldspar grain dissolution or
imported into the model reference frame. Illite fiber nuclea-
tion and growth are modeled using Arrhenius expressions that
consider saturation state in addition to temperature and time.
Nucleation occurs on pore walls, and muscovite and detrital
illite may be defined as energetically favorable substrates. The
model is integrated with other Touchstone™ models to ac-
count for the influence of other diagenetic processes on sur-
face area and reactant volumes and to provide input for per-
meability simulations.
We evaluated the illite model performance on two data
sets: (1) Jurassic quartzose samples from offshore mid-Norway
with maximum temperatures ranging from 108 to 173°C (226
to 343°F) and (2) Miocene lithic samples from offshore South-
east Asia that have maximum temperatures ranging from 157
to 182°C (315 to 360°F). The model matches measured abun-
dances of illite, kaolinite, and K-feldspar in both data sets using
identical kinetic paramet ers. Pre dicted K-Ar ages are consis-
tent with available data given uncertainties associated with
detrital contaminants. Although no illite particle-size data are
available from the analyzed samples, modeled crystallite thick-
nessesfromtheNorwaydatasetarecomparabletopublished
measurements of 0.004 to 0.012 mm from North Sea samples
with similar temperature histories.
AUTHORS
Robert H. Lander Geocosm LLC, 3311
San Mateo Drive, Austin, Texas 78738;
roblander@geocosm.net
Rob Lander develops diagenetic models for
Geocosm LLC. He obtained his Ph.D. in geology
from the University of Illinois in 1991, was a
research geologist at Exxon Production Research
from 1991 to 1993, and worked for Rogaland
Research and Geologica AS from 1993 to 2000.
He is also a research fellow at the Bureau of
Economic Geology.
Linda M. Bonnell Geocosm LLC, 3311
San Mateo Drive, Austin, Texas 78738;
lmbonnell@geocosm.net
Linda Bonnell develops diagenetic models and
conducts reservoir quality prediction studies
for Geocosm LLC. She received her Ph.D. in
geology from the University of Illinois in 1990
and subsequently was a research scientist at
Washington University, Rice University, Roga-
land Research, and Geologica AS. She is also a
research fellow at the Bureau of Economic Ge-
ology. Linda was an AAPG Distinguished Lecturer
in 2003–2004.
ACKNOWLEDGEMENTS
This work was funded by Geocosm ’ s Consortium
for Quantitative Prediction of Sandstone Reser-
voir Quality (RQC), which is currently supported
by the State of Alaska Department of Natural
Resources, Anadarko, BHP-Billiton, BP, Chevron,
Cobalt International, ConocoPhillips, Devon,
Eni, ExxonMobil, Hess, Instituto Mexicano del
Petróleo, Maersk, Petrobras, Saudi Aramco,
Shell, Statoil, Total, and Woodside. We thank
Thomas Moore, Arthur Trevena, Reed Glasmann,
and Marek Kacewicz for providing the natural
data sets. Richard Larese, Reed Glasmann, and
Arthur Trevena, in particular, provided us with
invaluable insights regarding these data sets and
controls on illitization in general . Reviews by
David Awwiller, Stephen Franks, and Olav
Walderhaug h elped us improve the article.
The AAPG Editor thanks the fol lowing reviewers
for their work on this article: David N. Awwiller,
Stephen G. Franks, and Olav Walderhaug.
Copyright ©2010. The American Association of Petroleum Geologists. All rights reserved.
Manuscript received July 11, 2009; provisional acceptance October 26, 2009; revised manuscript received
March 8, 2010; final acceptance April 21, 2010.
DOI:10.1306/04211009121
AAPG Bulletin, v. 94, no. 8 (August 2010), pp. 1161– 1187 1161
INTRODUCTION
Accurate predictive models for the occurrence and
properties of fibrous illite would be useful for hy-
drocarbon exploration and production given that
illite can have a severe detrimental effect on res-
ervoir properties. In particular, the high surface
area, pore-bridging texture, and significant associ-
ated microporosity of illite fibers (Figure 1)act
t o reduce permeability while increasing irreduc-
ible water saturation and capillary entry pressure
(McHardy et al., 1982; Kantorowicz, 1984, 1990;
Bjørlykke et al., 1986, 1992). Illite fibers gener-
ally have thicknesses, widths, and lengths that are
on the order of 0.05, 0.5, and 50 mm, respectively
(Figure 1) (Güven et al., 1980; Nagy, 1 994; Lanson
et al., 1996, 2002). The resulting specific surface
areas (areas per solid volume) are at least two orders
of magnitude greater than cements such as calcite
or quartz that are composed of larger, blockier
crystals (Panda and Lake, 1995). Illite fibers tend
to grow as independent strands that extend well
into the pore space of host sandstones (Figure 1)
where they may significantly increase flow-path
tortuosity (Stalder, 1973; Pallatt et al., 1984; Cocker,
1986; Panda and Lake, 1995). Because illite fibers
frequently extend farther into pores t han other
authigenic clays, they tend to cause larger perme-
ability reductions for a given bulk volume. More-
over, the higher microporosity of fibrous illite
compared to most other authigenic clays (Nadeau
and Hurst, 1991) means that a comparable solid
volume will invade a larger proportion of macro-
pore volume.
Figure 1. Scanning electron microscope photomicrographs illustrating the texture and morphology of fibrous illite in sandstone.
(A) Illite fibers growing into an intergranular pore. The white dotted outlines indicate the zoomed-in views shown in panels B and C.
(B) Illite fibers typically show a range in sizes, suggesting that the crystallites did not all nucleate simultaneously. (C) Illite fibers appear
to have nucleated on authigenic quartz cement. (D) Illite fibers growing into a moldic pore that likely formed in resp onse to dissolution
of a feldspar grain.
1162 Fibrous Illite Model
Fibrous illite appears to form mainly by reac-
tion of kaolin minerals and K
+
. Potassium may be
derived from local K-feldspar dissolution (e.g.,
Bjørlykke et al., 1986, 1992; Bjørkum and Gjelsvik,
1988; Chuhan et al., 2001; Franks and Zwingmann,
this issue) or external sources such as migrating
fluids (e.g., Robinson et al., 1993; Lanson et al.,
1996; De Ros, 1998; Zwingmann et al., 1999) or
dissolved solutes from adjacent shales (Gaupp et al.,
1993; Berger et al., 1997; Thyne et al., 2001; Clauer
et al., 2008). The reac tio n pathway leadi ng to fi-
brous illite seems to require higher thermal ex-
posures compared to illite that forms by smectite
illitization (Lander et al., 1990; Stroker and Harris,
2009). The rate of the reaction leading to fibrous
illite growth, once it begins, appears to be rapid. One
line of evidence for this interpretation is the narrow
present-day temperature range (120 to 140°C [248
to 284°F]) from incipient to pervasive illitization in
Jurassic sandstones of the North Sea and Halten-
banken (e.g., Bjørlykke et al., 1986, 1992, 1995;
Ehrenberg and Nadeau, 1989; Glasmann, 1992;
Ramm and Rys eth, 199 6; Midtbø et al., 2000).
MODEL FORMULATION
The model simulates the following through geo-
logic time: (1) the kinetics of illite crystal nuclea-
tion and growth by considering saturation state,
temperature, and the properties and areas of nu-
cleation substrates; (2) reactant (kaolin and, op-
tionally, K-feldspar) and product (illite cement, illite
replacement, and K-feldspar dissolution porosity)
volumes; (3) the dimensions, volumes, and areas for
the illite crystallite population; and (4) the K-Ar age
for the authigenic illite with explicit consideration
made for the effect of detrital contaminants.
Illite Saturation State
The kinetics of crystal nucleation and growth both
depend strongly upon saturation state (Lasaga,
1998). Variations in the saturation state of illite
within reservoir sandstones are mainly determined
by the activities of SiO
2(aq)
,Al
+3
,K
+
,andH
+
(Bjørkum and Gjelsvik, 1988; Aagaard et al., 1992;
Bjørlykke et al., 1995). In geologic settings where
fibrous illite is actively forming, the activities of
these species are likely to be controlled by mineral
buffers present within the host sandstone (Hutcheon
et al., 1993; Bazin et al., 1997a, b; Berger et al., 1997;
Palandri and Reed, 2001). I n formation waters
above 100°C (212°F), SiO
2(aq)
concentrations tend
to be at or near expected equilibrium values for
quartz (Kharaka et al., 1985; Bjørlykke et al., 1995;
Bazin et al., 1997a, b; Palandri and Reed, 2001).
Kaolinite acts as an Al buffer and, like quartz, also
buffers pH (Bazin et al., 1997a). Bazin et al. (1997b)
argued that it is reasonable to assume that K-
feldspar, when present, buffers K
+
activities in the
absence of externally derived fluids.
We determined the likely saturation state of
illite as a function of temperature in the presence
of the quartz, kaolinite, and K-feldspar mineral buf-
fers. The first step in this analysis is to calculat e
equi librium activities for SiO
2(aq)
,Al
+3
, and K
+
as
a function of temperature over the range of inter-
est for illite formation in reservoir sandstones (we
used Geochemist’s Workbench™ version 6.0 for
these calculations; Bethke, 1996). Although the sil-
ica activity for the buffered system increases with
temperature, the opposite is true for dissolved alu-
minum (Figure 2A). Muscovite can be considered
a thermodynamically stable proxy for illite. When
the system is in equilibrium with quartz, the K-
feldspar and kaolinite stability fields will be sepa-
rated by the muscovite field over the full range of
temperatures where illitization is likely to occur
(Figure 2B) (Aagaard et al., 1992). Thus, there will
be a thermodynamic drive for illite to form in the
presence of K-feldspar and kaolinite in sandstone
reservoirs. The next step in the analysis involves
the determination of the equilibrium constant for
muscovite with temperature (Figure 2C). The final
step is to estimate the muscovite supersaturation
state given its equilibrium constant and the activ-
ities of SiO
2(aq)
,Al
+3
,andK
+
. Our calculations in-
dicate that muscovite supersaturation in the pres-
ence of the mineral buffers increases systematically
with temperature (Figure 2C). The magnitude of
this increase is about a factor of 1.5 as temperature
increases from 100 to 200°C (212 to 392°F).
Lander and Bonnell 1163
Several workers suggest that illite formation in
Rotliegende sandstones may have been influenced
by externally derived evaporitic brin es (Rossel, 1982;
Macchi et al., 1990; Lanson et al., 1996). Brines
with elevated K
+
activities could induce faster rates
of illite formation by increasing illite supersatura-
tion (Lanson et al., 1996). Consequently, we have
formulated the model so that K
+
activities may
be defined explicitly as an alternative to assuming
control by a K-feldspar buffer. In such cases, the
Figure 2. Calculated fluid compositions
and extent of supersaturation for musco-
vite (a proxy for illite) as a function of
temperature based on buffering by kaoli-
nite, K-feldspar, and quartz. (A) Activities
of Al
+3
,K
+
,andSiO
2
(aq). (B) Stability
fields for muscovite, kaolinite, and K-
feldspar assuming equilibrium with quartz.
(C) Muscovite equilib rium constant (K)
and supersaturation state (Q/K-1).
1164 Fibrous Illite Model
model still considers that the quartz and kaolinite
buffers control the SiO
2(aq)
and Al
+3
activities when
determining the muscovite saturation state. The
model also assumes in this case that K-feldspar, if
present, is not a reactant. Thus, all Al in the fibrous
illite is derived from kaolin and the system is fully
open with respect to K.
Crystal Nucleation
Wilkinson and Haszeldine (2002a) argued that nu-
cleation is the primary control on the occurrence
of fibrous illite cement in sandstones. Controls
on crystallite nucleation, however, are the most
poorly constrained aspect of kinetic models that
could potentially be used to predict fibrous illite
characteristics in sandstones because nucleation
theory remains in the developmental stage and is
difficult to test given that the nucleation process
occurs at the atomic scale (Lasaga, 1998).
Illite fibers, like most diagenetic phases in sand-
stones, appear to form by heterogeneous nuclea-
tion, where nuclei adhere to the surfaces of pre-
existing solids because of the free energy benefits
associated with the resulting lower surface area
(Lasaga, 1998; Wilkinson and Haszeldine, 2002a).
Experimentally grown illite fibers preferentially
nucleate on muscovite substrates when they are
present b ut eventually will form in the absence
of such substrates (Chermak and Rimstidt, 1990).
Little work has been done to date to rigorously
document the nature of nucleation substrates for
illite fibers in natural sandstones. However, docu-
mented examples of illite fibers nucleating on
authigenic kaolin exist (Środoń and Eberl, 1984;
Lanson et al., 1996, 2002) as well as quartz ce-
ment (Figure 1C) (Bonnell et al., 1999), and in
some sandstones, illite pervasively covers all avail-
able pore-wall surfaces irrespective of composition
(Lanson et al., 1996; Bjørlykke, 1998). Franks and
Zwingmann (2010, this issue), in their study of il-
lite in Permocarboniferous sandstones from Saudi
Arabia, find that although illite fibers nucleate
on detrital quartz grain and noneuhedral quartz
cement surfaces, they do not occur on euhedral
quartz faces. The scanning electron microscope
(SEM) photomicrograph in Figure 1C, however,
illustrates one instance where illite fibers appear to
have nucleated on euhedral quartz faces. Thus, it
appears that although illitic or micaceous mate-
rial is the favored nucleation substrate for fibrous
illite, such substrates are not an absolute require-
ment for nucleation. Furthermore, whereas there
might be a tendency for illite fibers to avoid nu-
cleating on euhedral quartz faces, it is not impos-
sible for them to do so.
The critical nucleus size is an essential con-
cept for illite nucleation models (Wilkinson and
Haszeldine, 2002a). The critical nucleus represents
a maxim um in the Gibbs free-energy chan ge over
a spectrum in atomic cluster size that ranges from a
few atoms to many millions of atoms in a crystal.
The cluster size of the critical nucleus denotes the
point where the increase in thermodynamic stabi-
lity that comes with additions of atoms promotes
continued crystal growth whereas removal of atoms
increases the potential for nucleus dissolution.
The abundance of critical nuclei is given by Lasaga
(1998) as
N
n
¼ N
0
e
G
n
RT
ð1Þ
where N
0
is the total number of moles in the me-
dium (predominantly water), G
n
is the energy
needed to form a critical nucleus consist ing of n
molecular units, T is temperature, and R is the
gas constant. By analogy with equation 1, we simu-
late the number of potential illite nuclei that form
during a time interval Dt (s) as
N ¼ A
n
e
Ea
n
RT
Dt
Q
K
1
S
A
ð2Þ
where A
n
is a pre-exponential probability factor
(nucleation sites/cm
2
/s), Ea
n
is the activation en-
ergy for ill ite fiber nuclea tion (kJ/mol), Q is the
saturation state for muscovite, K is the equilibrium
concentration for muscovite, and S
A
is the surface
area available for nucleation. Increases in illite
nuclei are thus expe cted with increasing s uper-
saturation, temperature, and elapsed time. Increased
nucleation rates could also result from lower activa-
tion energies for nucleation on s ubstrates made
Lander and Bonnell 1165
up of micaceous grains or detrital illitic clay com-
par ed to other surfa ce types. Consequently, Ea
n
and S
A
may be defined independently for mica-
ceous and nonmicaceous substrates in the model
implementation.
We allow crystallites to nucleate on intergran-
ular pore walls (Figure 1B, C)aswellasonsec-
ondary pore walls associa ted with dissolution of
K-feldspar grains ( Figure 1D). The intergranular
surface area is simulated through time by (Merino
et al., 1983; Lichtner, 1988; Canals and Meunier,
1995; Lander et al., 20 08)
S
ig
¼ V
g
6
D
f
f
0
2=3
ð3Þ
where V
g
is the grain volume in the model refer-
ence frame (cm
3
), D is the mean grain dia meter
(cm), f
o
is the intergranular porosity at the time
of deposition (volume fraction), and f is the cur-
rent intergranular porosity (volume fraction). The
fibrous illite model is integrated into Touchstone
7.0, which simulates the intergranular porosity
through time based on the sandstone compaction
state as well as the volume of illite and other ce-
ments. Touchstone also uses a proprietary algo-
rithm to estimate the f
o
value based on mean grain
size, sortin g, and detrital matrix abundance. The
nucleation area for moldic pores that form by K-
feldspar grain dissolution is given by
S
ksp
¼ f
ksp
6
D
ð4Þ
where f
ksp
is the volume of K-feldspar grains that
have dissolved in the model reference frame (cm
3
).
As discussed above, the formation of critical
nuclei that could serve as the basis for new crystals
requires a larger change in Gibbs free energy than
is needed for an existing crystal to continue to
grow. Therefore, in the near vicinity of a growing
crystal, the degree of illite supersaturation will likely
be somewhat lower compared to otherwise barren
pore walls given that the crystal represents a sink
for the dissolved reactants (Figure 3). This local-
ized reduction in supersaturation would change
the local free energy, making new nuclei less
likely to form adjacent to growing crystals com-
pared to crystal-free areas, and is analogous to what
has been proposed for the control on nucleat ion
of carbonate concretions at larger length sc ales
(Walderhaug and Bjørkum, 1998). As ne w nuclei
form and grow , the proportion of pore walls that
are energetically favorable for creation of addi-
tional nucl ei declines in the absence of changes in
othe r factors (Figure 3). We consider this effect
on nuclea tion by defining a characteristic radius
of influence arou nd each crystallite within which
no new nuclei are allowed to form. The size of this
area is likely to be a function of the diffusion rate
(Kittel and Kroemer, 1980).
D
c
¼
kT
6pam
ð5Þ
where D
c
is the diffusion coefficient, k is the Boltz-
mann constant, a is the size of particles in solution,
and m is the viscos ity of the aqueous solution. In
relatively dilute solutions, diffusion rates from
equation 5 double fro m 100 to 160°C (212 to
320°F) thereby halving the expected radius of in-
fluence. Thus, we define the exclusion radius for
nucleation of new crystallites in terms of the value
at 100°C (212°F) and adjust the size using equa-
tion 5. Simulations can effectively neglect this po-
tential effect on nucleation, if desired, by using a
vanishingly small value for the radius of influence.
In a simulation, we define representative areas
for nucleation on three types of surfaces: (1) inter-
granular pore walls that are made up of micaceous
or illitic materials, (2) intergranular pore walls char-
acterized by other materials, and (3) moldic pores
after K-feldspar dissolution. We adjust the areas of
these surfaces with each time step to account for
changes in intergranular and secondary pore vol-
umes. If illite reactants are present, we then de-
termine the number of potential new nuclei for
each area type and distribute them randomly over
the currently available nucleation surface for that
area. If a potential nucleus is located within the
radius of influence of an existing crystal, however,
it is not allowed to form.
An intriguing idea proposed by Wilkinson and
Haszeldine (2002a) for fibrous illite, and discussed
1166 Fibrous Illite Model
at greater length by Meunier (2006) for clay min-
erals in general, is that as crystallites become pro-
gressively larger, they eventually stop growing be-
cause of high strain energies that arise due to the
cumulative influence of crystallographic defects.
In s uch a scenario, it become s energetically more
favo rable for a ne w crys tal to nucleate and grow
compared to continued growth on a la rge crys tal.
Figure 3. Schematic illustration of the
potential influence of growing illite fibers
on the nucleation of new crystalli tes.
(A) Prior to crystallite nucleation, a given
pore wall is likely to be exposed to fluid
with a uniform degree in the extent of
supersaturation for illite. (B) Once a crys-
tallite nucleates, it acts as a sink for illite
solutes thereby potentially reducing the
extent of supersaturation in its immediate
vicinity (gray region). (C) New crystals
may be less likely to nucleate on the pore
wall in the near vicinity of growing crys-
tallites due to the lower local extent of
illite supersaturation. (D) As additional
crystallites form, th e surface area that is
amenable to the formation of new nuclei
declines.
Lander and Bonnell 1167
Crystallite size distribution data for illite that
formed in hydrothermal zones (Bove et al., 2002)
and by smectite illitization (Środoń et al., 2000),
however, show log-normal distributions that have
been interpreted to result from faster surface-area
normalized growth rates on larger crystallites com-
pared to smaller counterparts (Eberl et al., 1998).
This pattern of faster growth on larger crystals is
the opposite of that expected from a scenario
where increasing strain energy with size results in
diminished potential for continued growth of larger
crystallites. Regrettably no quantitative size dis-
tribution data are available to determine whether
fibrous illite crystals in sandstone reservoirs also
show log-normal size distributions. In light of these
conflicting views and the lack of illite fiber data, we
have elected not to incorporate any dependencies
between crystallite size and the rates of nucleation
or growth in our current model implementation.
Crystal Growth, Reactant Dissolution, and
Mass Balance
We assume in our model formulation that the rate
of growth of fibrous illite crystals is slower than the
potential rate of reactant dissolution and the po-
tential rate of diffusion of dissolved reactants to
illite growth sites. Consequently, the extent of re-
actant dissolution for a modeled time interval is
determined by the volume of illite that grows in
this precipitation rate limited system. These as-
sumptions are consistent with the conclusions of
Altaner (1986), who showed that K-feldspar dis-
solution rates are significantly faster than rates of
smectite illitization.
We enforce the conservation of Al within the
modeled frame of reference and permit illite to
grow only when a source for both K and residual
kaolin in the system exists. A simulation may be
set to be either open or closed with respect to K. If
the system is closed with respect to K, then K is
exclusively derived from K-feldspar, which also
provides an additional source of Al. For simulations
that are defined as open with respect to K, we as-
sume that in-situ K-feldspar dissolution does not
contribute to illite formation and that kaolin there-
fore is the exclusive source of illite Al. Given that
quartz cementation is ubiquitous in illite-bearing
sandstones (at least when quartz nucleation area is
available), we assume that there is never a shortfall
in the supply of the Si needed for illite formation.
Güven (2001) showed that illite fibers are elon-
gated along the crystallograph ic a axis. Although
illi te fibers tend to taper somewhat along the a
axis (Lee, 1984; Nagy, 1994), we approximate the
crystallite shape as a rectangular volume defined
by length, thickness, and width. We assume that
the crystallite proportions remain constant as they
grow given that the growth surfaces for illite fibers
are defined by euhedral crystal faces from early
growth stages, and Nagy (1994) found a width to
thickness ratio of around 17 for illite fibers over a
range in particle sizes. A ssessing the ratio of fiber
length to width is more difficult. Dispersed par-
ticles as analyzed by transmission electron micros-
copy (TEM) and atomic force microscopy are likely
broken during the sample preparation procedure
and therefore only provide a m inimum constraint.
Data from these methods and SEM images of in-
situ illite, however, suggest that a value on the or-
der of 100 is reasonable.
We determine the extent of growth along the
length of the crystal (a axis) over a time interval Dt
(s) as follows.
I
a
¼
m
r
A
g
e
Ea
g
RT
Dt
Q
K
1
ð6Þ
where I
a
is the increase in fiber length (cm), m is
the molecular mass of illite (g/mol), r is the den-
sity of illite (g/cm
3
), A
g
is a pre-exponential con-
stant (mol/cm
2
/s), Ea
g
is the activation energy for
illite precipitati on (kJ/mol), Q is the saturation
state for muscovite, and K is the equilibrium con-
stant for muscovite. After determining the extent
to which a crystal lengthens, we modify its thick-
ness and width proportionally. The thicknesses of
the crystallites, however, are restricted to integral
values of 0.001 mm given that this represents one
unit cell for illite along the crystallographic c*
direction. The crystallite volume (V
c
) is deter-
mined as
V
c
¼ L
c
W
c
T
c
ð7Þ
1168 Fibrous Illite Model
where L
c
is the crystal length along the a axis, W
c
is the crystal width perpendicular to the a axis,
and T
c
is the crystallite thickness along the c* axis.
As the crystal grows, the model keeps track of ex-
cess thickness and adds a new illite unit cell to T
c
once the excess reaches 0.00 1 mm. By excluding
this excess thickness in the volume calculation,
we obtain a more accurate illite K mass determi-
nation fo r simulation of K-Ar ages. We calculate
the crystallite wetted surface area (A
c
) by assum-
ing that the entire surface of the crystallite is in
contact with the pore fluid except for where it is
attached to the pore wall.
A
c
¼ 2 L
c
W
c
ðÞþ2 L
c
T
c
ðÞþT
c
W
c
ð8Þ
The model tracks the dimensions, volume, and
surface area for each simulated crystallite in addi-
tion to the nucleation time and the time at which it
ceased to have the potential to grow due to local
porosity occlusion. A crystal is not permitted to
continue to grow, even if reactants are available,
when it is located in the part of the simulated nu-
cleation area that has been removed due to po-
rosity loss that arises from cementation or con-
tinued compaction. In accordance with Nagy’s
(1994) observations, we assume that no Ostwald
ripening occurs for the crystallite population. Thus,
once formed, a simulated illite fiber never dis-
solves. This assumption appears to be reasonable
over the temperature range experienced by many
hydrocarbon reservoirs. At temperatures in excess
of approximately 150°C (302°F), however, this
assumption may not be valid because fibrous illite
may begin to dissolve while more equant crystal-
lites form (Lanson et al., 2002) such that eventually
the 2M
1
polytype predominates over the 1M poly-
type characteristic of fibrous illite (Hunziker, 1986;
Jahren and Aagaard, 1989).
Illite K-Ar Ages
Much literature is concerned with the use of illite
K-Ar ages for inferring the timing and likely driving
forces for illite formation. Thus, K-Ar ages provide
an important potential means to test and calibrate
the fibrous illite model. Consequently, the model
determines K-Ar ages by considering the volume
fraction of illite that formed through the burial
history of the sample. The K-Ar calculations are
made using the decay constants of Steiger and
Jäger (1977).
Although the illite K-Ar systematics are well
understood (Lee, 1984; Lee et al., 1985, 1989;
Hamilton et al., 1989, 1992; Clauer and Chaudhuri,
1995), the inte rpretation of the meaning of illite
K-Ar ages is not so straightforward (e.g., Ehrenberg
and Nadeau, 1989; Clauer et al., 1992; Matthews
et al., 1994; Bjørlykke, 1998; Zwingmann et al.,
1998; Pevear, 1999; Ylagan et al., 2000; Środoń
et al., 2002; Wilkinson and Haszeldine, 2002b;
Meunier et al., 2004; Szczerba and Środoń, 2009).
Two commo n assumption s in the analysis of il-
lite fiber K-Ar ages are that (1) illite in the finest
size fraction represent s the last formed illite and
(2) the K-Ar age of this illite represents the time of
cessation of illite growth (Lee, 1984; Hamilton
et al., 1989). Two potential problems with these
assumptions exist. First, even if this fraction of the
illite was the last to form, this does not necessarily
mean that the growth duration was geologically
instantaneous. Thus, in the absence of contami-
nants, the date reflects the volume-weighted average
of when the analyzed illite fibers grew (Hamilton
et al., 1989). Second, the extent to which the smallest
illite fraction actually represents the last formed illite
is unclear. Little is known about how the sample
preparation procedure impacts the in-situ particle
size distribution or if early formed small crystallites
that stopped growing due to interference with
other solids (including other illite fibers) may also
be found within the smallest size fraction.
Even the smallest size separates used for K-Ar
analysis are likely to contain some extent of de-
trital contamination. This contamination in some
cases may lead to K-Ar ages that are significantly
older than what would be derived from pure au-
thigenic illite, particularly for sandstones that con-
tain detrital clay matrix or grains rich in K-feldspar,
mica, or detrital illite (e.g., Hamilton et al., 1989,
1992; Gl asmann, 1992; Zwingm ann et al., 1998,
1999; Girard et al., 2002). Consequently, we also
determine an additional K-Ar age that incorporates
the ages and volume abundances of the following
Lander and Bonnell 1169
cont aminant phases: K-feldspar, muscovite, and
detrital illite .
In light of this discussion, we determine sev-
eral different age values in the simulations for com-
parison with K-Ar measurements: (1) the K-Ar age
for all authigenic illite fibers integrated over the
complete illitization history, (2) the K-Ar age with
detrital contaminants, (3) the times when the 10th,
50th, and 90th volume percentiles of authigenic
illite form, (4) the time of the peak rate for illite
growth, and (5) the time of the cessation of illite
formation in association with the exhaustion of re-
actants or the occlusion of pore space needed for
continued growth.
Reconstructing Reactants
We have designed the model so that it may be used
to simulate the geologic progress ion of the reac-
tion for already illitized s amples. The model re-
constructs K-feldspar for the case where it is con-
sidered to be the illite K source as well as the
amount of kaolin that reacted to form illite. The
parent materials for kaolinite also are reconstructed
by Al mass balance. These mass-balance calcula-
tions account for the stoichiometries and densities
of the associated minerals in addition to micro-
porosity values for the authigenic kaolinite and il-
lite (see Table 1 for the values used for simulations
in this article).
For simulations that assume that the illite K is
derived from K-feldspar, the starting point is to
reconstruct the depositional composition by deter-
mining the amount of K-feldspar dissolution needed
to account for the K bound in illite. This value is
then compared to the amount of K-feldspar dis-
solution indicated by secondary porosity from
K-feldspar dissolution together with K-feldspar
replacement by illitic phases. Even if K actually
is conserved within the sample frame of reference,
it is unlikely that these two measures of the volume
of dissolved K-feldspar will be in precise agreement
given the uncertainties in the petrographic data,
microporosity values, and illite stoichiometry used
for the analysis. Thus, the amount of K-feldspar to
add to the depositional composition may be based
either on the illite derived value or on the value
determined from K-feldspar secondary porosity
and illite replacement.
The next step in the reconstruction is to con-
duct an Al mass analysis. Here, we determine the
Al associated with measured kaolinite and illite
abundances in light of their microporosities, stoi-
chiometries, and densities. If K-feldspar is con-
sidered to be the illite K source, then we subtract
the illite Al that is derived from K-feldspar dis-
solution from the total Al bound in illite. We then
assume that the remaining illite Al is derived from
kaolinite dissolution. We add to this value the Al
associated with any measured kaolin in the sample
to determine the total Al produced by dissolution
of plagioclase feldspar, muscovite, biotite, or addi-
tional K-feldspar. We compare this Al mass to the
petrographically determined extent of dissolution
and replacement of these additional potential Al
sources. As with the K mass balance, we do not
expect a precise agreement between the Al mass
based on the petrographic evidence for grain dis-
solution and replacement and the Al mass based on
the authigenic clays. Consequently, we have de-
signed the model so either of these two calculation
methods may be used for the Al reconstruction.
Table 1. Stoichiometry, Density, and Microporosity Values Used for Mass-Balance and K-Ar Calculations*
Mineral Stoichiometry Density (g/cm
3
) Microporosity (vol.%)
Illite K
0.91
Al
2.21
Fe
0.04
Mg
0.15
Si
3.51
O
10
(OH)
2
2.75 65
Kaolinite Al
2
Si
2
O
5
(OH)
4
2.60 50
K-feldspar KAlSi
3
O
8
2.56 0
Plagioclase feldspar K
0.05
Na
0.60
Ca
0.35
Al
1.35
Si
2.65
O
8
2.67 0
Muscovite KAl
3
Si
3
O
10
(OH)
2
2.82 0
*Illite stoichiometry is from Lander et al. (1990) and mineral densities are from Deer et al. (1966).
1170 Fibrous Illite Model
The reconstruction process adds feldspar and
mica to sample-measured values to derive the de-
positional abundances of these minerals. We then
use paragenetic rules (defined in terms of burial
depth, temperature, or absolute geologic time con-
straints) to determine when these minerals react
to form kaolinite. Illite may then form according
to the model characteristics and in put data de-
scri bed above.
MODEL BEHAVIOR
Although the veracity of any geological model is
best determined by testing its predictions using
natural data sets, synthetic simulations of simpli-
fied geologic scenarios are useful for understand-
ing the model behavior. Thus, as a prelude to our
evaluation of the model performance, we use syn-
thetic simulations as a means to explore model
sensitivity to nucleation and growth kinetics, tem-
perature history, reactant volumes, K
+
activities,
and the nature of nucleation substrates.
Approach for Constraining Kinetics
Perhaps the greatest controversy in the application
of chemical kinetic models for simulation of dia-
genetic processes has to do with the appropriate
approach for determining rate constants. Although
laboratory studies provide a critical understand-
ing of the controls on kinetic rates, growing evi-
dence shows that the associated rates may be much
faster than those found in geologic environments,
at least for silicate phases. The dichotomy between
natural and laboratory-derived rate constants has
been well documented, for example, for feldspar
dissolution (e.g., Blum and Stillings, 1995; White
and Brantley, 2003; Zhu, 2005; Maher et al., 2006)
and quartz precipitation (e.g., Walderhaug, 1994,
1996, 2000; Oelkers et al., 2000; Worden and
Morad, 2000; Lander et al., 2008) where labora-
tory rates may be as much as five orders of mag-
nitude greater than rates derived from geologic
observations. Several studies have attempted to
apply laboratory-derived rate constants for fibrous
illite growth (e.g. , Chermak and Rimstidt, 1990)
to geologic settings. These rate constants lead to
rapid, pervasive illi tization at much lower tem-
peratures than are observed in nature. For exam-
ple, Brosse et al. (2000) found that laboratory-
derived rates suggest that pervasive illitization
would occur in about 20,000 yr at 105°C (221°F)
in North Sea reservoirs but point out that parts of
the reservoir that are at present-day temperatures
of 110°C (230°F) contain only trace amounts of
illite while having abundant K-feldspar and kaoli-
nite reactants. Likewise, Berger et al. (1997) found
that laboratory kinetics predict pervasive illitiza-
tion to occur in typical formation waters within
4.5 m.y. at 50°C (122°F) and in a mere 15,000 yr
at 80°C (176°F), despite ample field evidence
showingthecoexistenceof kaolinite and K-feldspar
reactants over millions of years at these and higher
temperatures (Berger et al., 1997). However, Nagy
(1994) and D. N. Awwiller (unpublished presen-
tation to the SEPM Clastic Diagenesis Research
Group in 1998) suggested that changes in particle
dimensions with depth imply illite fiber growth
rates that are similar to the much slower rates of
quartz cement growth that are implied by em-
pirical calibration to geologic data.
Our view is that a model that aims to predict
the occurrence and properties of illite in reservoir
sandstones should incorporate kinetic parameters
that are based on empirical calibration to geologic
observations instead of laboratory rates that are in
apparent conflict with these observations. Thus,
in the following analysis, we use kinetic parameters
that lead to pervasive illitization over time and
temperature ranges that are consistent with geo-
logical constraints on illite occurrence.
Baseline Case
We begin our sensitivity analysis by considering
suites of kinetic parameters that lead to illitization
patterns that are broadly consistent with data from
Norwegian Shelf reservoirs of Jurassic age. In these
sandstones, incipient illite formation is observed at
present-day temperatures near 120°C (248°F), and
pervasive illitization is characteristic of reservoirs
with present-day temperatures of 140°C (284°F)
(e.g., Bjørlykke et al., 1986; Ehrenberg and Nadeau,
Lander and Bonnell 1171
1989; Glasmann et al., 1989; Ramm and Ryseth,
1996; Chuhan et al., 2001). The simulations in-
corporate a series of temperature histories with a
common depositional time of 180 m.y. but with
maximum temperatures that range from 110 to
150°C (230 to 302°F) (Figure 4A). For model
Figure 4. Input and results for the baseline set of synthetic illitization models that are designed to roughly approximate illite formation
in Jurassic reservoirs in the North Sea and Norwegian Shelf. (A) Temperature history input. (B) Simulated illite cement abundance with
geologic time for each of the temperature histories. (C) Modeled change in intergranular volume (IGV, a proxy for compaction state),
intergranular porosity, quartz cement, and illite cement with geologic time for the maximum 150°C simulation. (D) Comparison of modeled
rates of the pore volume loss due to compaction, quartz cementation, and illite cementation with geologic time for the maximum 150°C
simulation. (E) Evolution in the volume abundance of illite reactants and products with time for the maximum 150°C burial history.
1172 Fibrous Illite Model
input, we use a well-sorted, upper medium-grained
sandstone with a subarkosic composition that is
based on the characteristics of petrographically
analyzed samples from the Garn Formation. The
present-day sandstone composition has 3.2 and
1.2 vol.% of K-feldspar and p lagioclase feldspar,
respectively. In addition, we assume that 3 vol.%
of kaolinite formed early in the burial history from
the dissolution of equal volumes of plagioclase
feldspar and K-feldspar. The simulations account
for the effect of compaction and quartz cemen-
tation on the porosity and nucleation surfa ce area
by the coupling of the illite model with Touch-
stone 7.0 models for these processes (Lander et al.,
2008).
The kinetic parameters we use not only lead
to pervasive illitization (all available reactants are
consumed) by 140°C (284°F), but are also in agree-
ment with the data for these reservoirs that in-
dicate that samples wi th maximum temperatures
of 120°C (248°F) h ave experienced only incipi-
ent illite formation (Figure 4B). The value f or t he
activation energy for illite precipitation (Ea
g
)of
73 kJ/mol is substantially higher than the 60.7 kJ/
mol value we used for quartz cementation, which
is based on empirical calibration to a Jurassic data
set from the Norwegian Shelf (Lander et al., 2008)
(we used an A
g
value of −9×10
−12
mol/cm
2
/s for
both the illite and quartz models). For the nucle-
ation parameters, we used values of 8000 sites/
cm
2
/m.y. for A
n
and73kJ/molforEa
n
. We also
assumed that no new illite nuclei would form
within 0.002 mm of an existing crystallite at a ref-
erence temperature of 100°C (212°F) (we adjusted
this range with temperature using equation 5).
In the simulation using a burial history with a
maximum temperature of 150°C (302°F), illite
formation begins much later than compaction or
even quartz cementation (Figure 4C). Once illi-
tization begins in earnest, however, it reduces in-
tergranular porosity at rates that greatly exceed
loss rates associated with compaction or quartz
cementation (Figure 4D). Although illitization be-
comes significant much later than quartz cemen-
tation, the simulation shows that quartz cementa-
tion is likely to continue after the illitization process
is complete provided that adequate nucleation sur-
face area remains (Figure 4C). This result is con-
sistent with observations that indicate that al-
though illite fibers nucleated and grew on quartz
overgrowths in Norwegian Shelf reservoirs, they
were engulfed by subsequent quartz cement growth
(Bonnell et al., 1999).
Thefateofthereactantsandproductsinvolved
in the illitization reaction is shown with time for the
maximum 150°C simulation in Figure 4E.Weset
up the simulation so that kaolinite formed in asso-
ciation with K-feldspar and plagioclase (not shown)
dissolution over a temperature range of 50 to 80°C
(122 to 176°F). The coexisting kaolinite and re-
sidual K-feldspar remained relatively unaffected
until about 28 Ma when the sandstone reached
120°C (248°F). From this time onward, these phases
dissolved and illite began to form. Although the
dissolution of K-feldspar produced additional sec-
ondary porosity, this porosity was partly occluded
by illite that grew within these pores, creating a re-
placement texture. Likewise, the intergranular po-
rosity increase associated with kaolinite cement
dissolution was more than compensated for by in-
tergranular porosity loss from both illite and quartz
cements (Figure 4C). The reaction ceased when all
kaolinite was consumed leading, in this scenario, to
residual K-feldspar.
In addition to matching the illitization pattern
with present-day temperature, the nucleation ki-
netic parameters also lead to predicted fiber thick-
nesses that are consistent with data from North
Sea reservoirs. Average particle thicknesses for
illite-size separates obtained by Nagy (1994) tend
to thicken with the present-day reservoir temper-
ature (Figure 5). The data shown in Figure 5 are
average values derived from measurements made
on illite fibers in two size separates: less than
0.02-mm equivalent spherical diameter and 0.02-
to 0.2-mm equivalent spherical diameter. We have
overlain on this figure the simulated present-day
10th, 50th, and 90 th percentile (P10, P50, and
P90) crystallite thicknesses (by crystallite count)
for the simulations. These simulated ranges span
the measurements and also show an increase with
temperature. Although the P50 and P90 sizes show
a larger increase with temperature than the mea-
surements, such a discrepancy seems logical given
Lander and Bonnell 1173
that the size separation procedure, by design, is
biased toward the finer part of the crystallite size
distribution.
Sensitivity to Kinetic Parameters
We evaluated the sensitivity of model results to
the activation energies for illite nucleation and
growth using the 150°C maximum burial history
(Figure 4A) and the baseline sandstone composi-
tion and texture. For these simulations, we ad-
justed the kinetic parameters such that the peak
rate of illitization occurs at 140°C (284°F) for all
scenarios. Results show that slower crystal growth
rates (larger Ea
g
values) must be counterbalanced
by faster nucleation rates (smaller Ea
n
values)
to maintain this peak illitization temperature
(Figure 6A). Stated in another way, if the crystal
growth rate slows, then growth must occur on
more crystals to maintain a fast net precipitation
rate. Consequently, when the crystal growth rate
decreases in these runs, the simulated crystal sizes
(as exemplified by the P50 thicknesses) decline
systematically (Figure 6B). The simulations also
show a systematic increase in the duration of illi-
tization (as indicated by the elapsed time between
the precipitation of the 10th and 90th volume
percentiles) as the crystal growth rate decreases
(Figure 6B).
Sensitivity to Temperature History
We evaluated t he model s ensitivity to thermal
exposure by maintaining the same temperature
Figure 5. Comparison of the average values for measured illite
fiber thicknesses for North Sea reservoirs (Nagy, 1994) with sim-
ulated values for the baseline case set of synthetic models. The
simulated values are shown as the 10th, 50th, and 90th percentile
values (P10, P50, and P90, respectively) for the particle size dis-
tribution where the percentiles are based on the number of sim-
ulated illite fibers.
Figure 6. Sensitivity of simulation results to the activation en-
ergies for illite nucleation (Ea
n
) and growth (Ea
g
) where all model
scenarios are constrained to lead to a peak rate of illitization
at 140°C (284°F) for the maximum 150°C temperature history
(Figure 4A ) and baseline sands tone compos ition and texture.
(A) Slower cryst al growth from larger growth activat ion ener-
gies must be counterbalanced by faster crystal nucleation (lower
nucleation activation energy) to maintain the same peak tem-
perature for illite formation. (B) The P50 crystal thicknesses de-
crease systematically as growth rate decreases (Ea
g
increases)
and nucleation rate increases. Faster crystal growth rates also
result in faster overall rates of reaction as exemplified by the
time duration between when the 10th and 90th volume per-
centiles (P10 and P90, respectively) of the present-day illite vol-
ume formed.
1174 Fibrous Illite Model
history values used for the 150°C maximum sce-
narioforthebaselinecase(Figure4A) while scaling
the time duration such that the time of deposition
ranged from 720 to 90 Ma. We used the same
kinetic parameters and sandstone composition and
texture as used for the baseline case. The simu-
lated temperature of pervasive illitization drops by
roughly 10°C (18°F) for each doubling of the burial
history time duration (Figure 7A). The simulations
also sho w progressively greater peak illitization
rates as the time duration decreases (Figure 7B).
This pattern occurs because the shorter residence
time within a given temperature range for the
younger burial histories means that less reactant is
consumed before progressing to the higher tem-
peratures associated with the next burial history
step. Consequently, greater volumes of reactants
persist to higher temperatures where illite nucle-
ation and growth rates are faster.
Sensitivity to Reactant Volumes
In this set of sensitivity runs, we varied the amounts
of kaolinite and K-feldspar reactants while keeping
all other factors the same as the baseline case
with the maximum 150°C burial history scenario
(Figure 4A). Results show that although the rate of
illitization is equivalent for each simulation while
reactants are available, the reaction is able to con-
tinue to higher temperatures for simulations with
greater reactant volumes (Figure 8). Thus, in the
absence of other differences, results indicate that
sandstones with greater reactant volume should
have younger integrated ages of illite formation
while having higher peak illitization rates at hotter
temperatures (Figure 8).
Sensitivity to K
+
Activity
Increasing K
+
activity leads to faster rates of both
crystal nucleation and growth because it causes a
higher degree of supersaturation for the muscovite
proxy to illite. We used the maximum 150°C burial
history scenario from the baseline case (Figure 4A)
together with log K
+
activities of −3, −2, and −1to
assess the impact of this factor on the simulated
timing of illitization with respect to a case where K
+
Figure 7. Sensitivity to thermal exposure for scenarios where
the same temperature values are used as the maximum 150°C
baseline scenario (Figure 4A) but the time durations ch ange.
(A) The temperature of pervasive illitization increases roughly
by 10°C (18°F) when the tim e d uration i s halved. (B) The rates
of illitization increase progressively with decreasing time dura-
tion for those scenarios wi th pervasive illitization.
Figure 8. Sensitivity to reactant volumes for scenarios where
other compositional and textural characteristics are constant
using the baseline maximum 150°C burial history (Figure 4A).
The illite cementation rates are ind ependent of reactant vol-
umes so long as reactants are available. For samples with greater
reactant volumes (greater final illite abundances), the reaction
takes longer to complete, occurs at higher temperatures, and
reaches faster peak rates.
Lander and Bonnell 1175
activity is controlled by equilibrium with K-feldspar.
The −2 log K
+
activity scenario is at the upper end
of the range shown by formation waters reported
for several Mesozoic and Tertiary reservoirs, whereas
the −1 value corresponds to the upper end of val-
ues determined from fluid inclusions in halite crys-
tals (Figure 9). The simulations also assume that
the system is open with respect to K
+
and that K-
feldspar is not a reactant.
Simulation results, as expected, are profoundly
influenced by increasing K
+
activity. Illitization is
complete by 95° C ( 194°F) for the log a K
+
−2
scen ario compared to 138°C ( 275°F) for the K-
feldspar equilib rium baseline case (Figure 10A).
At tem perat ures above 60°C (140°F) in the log a
K
+
−1 scenario, kaolinite is consumed more rapidly
than it forms (Figure 10B). Thus, the time of the
reaction completion for this geologically unrealistic
scen ario is determined by the constraints used to
define the formation of the kaolinite reactant.
The overall abundance of illite that forms is lower
for these high K
+
activity scenarios compared to
the baseline case because K-feldspar does not con-
tribute Al for illite growth (Figure 10A).
Figure 9. Calculated K
+
activities from
published fluid compositional analyses
collected from sandstone reservoirs and
from fluid inclusions in halite. We calcu-
lated the activities with Geochemist’s
Workbench™ 6.0 (Bethke, 1996) using
the Debye-Hückel method except for the
fluid derived from halite inclusions where
we used the Pitzer method. The shaded
region indicates the range in K
+
activities
expected for equilibrium with K feldspar.
Figure 10. Sensitivity to K
+
activity using the baseline composition and texture and the maximum 150°C temperature history (Figure 4A).
(A) Simulated illite cement with paleotemperature. The temperature of illitization drops greatly with increasing K
+
activity. The final illite
abundance is greater for the K-feldspar equilibrium run because the feldspar provides an additional source for illite Al. (B) Simulated
kaolinite cement with paleotemperature. Although the same volume of kaolinite forms in all runs, in the scenario with a log K
+
activity of
−1, the rate of the reaction is sufficiently fast enough that it is consumed faster than it forms at temperatures greater than 70°C (158°F).
1176 Fibrous Illite Model
Sensitivity to Nucleation Substrate Properties
The model is designed to consider preferential
nucleation of fibr ous il lite on substra tes such as
micaceous or illitic surface s by providing a means
to specify the activation energy benefit for nucle-
ation on such substrates compared to other surface
types and by accounting for the fraction of the in-
tergranular pore area that is made up of such sub-
strates. To illustrate the model sensitivity to these
two factors, we defined muscovite grains to be a
preferential nucleation substrate. We considered
scenarios where we decreased the activation energy
for nucleation by 4 and 8 kJ/mol for muscovite
surfaces compared to other substrate types for
simulated samples with 5 and 25 vol.% of mus-
covite grains. Other factors in the simulations are
as defined for the baseline case with the maximum
150°C burial history (Figure 4A).
Results indicate that preferential nucleation on
micaceous substrates leads to pervasive illitization
at lower paleotemperatures (earlier in the burial
histories) compared to the baseline case (Figure 11),
although the crystal growth kinetics are held con-
stant. The faster overall illitization occurs in the
preferential nucleation scenarios because growth
occurs on a larger number of crystals. Additionally,
preferential nucleation substrates promote illite for-
mation as cement compared to replacement (where
the crystals grow within secondary pores after
K-feldspar dissolution). This preference for cement
occurs because the preferenti al nuclea tion sub-
strates border intergranular pore s. The signifi-
cance of preferential nucleation on illite texture
and timing, however, is comparatively minor ex-
cept when the surface area made up of preferen-
tial substrates is a significant fraction of the overall
area, and the activation energy benefit for nucle-
ating on such surfaces is very large.
Summary of Sensitivity Test Results
The sensitivity analysis reveals several noteworthy
characteristics of the fibrous illite model. (1) Other-
wise comparable sandstones with similar maxi-
mum temperatures but longer thermal residence
time s will tend to become pervasively illitized at
lower temperatures and at slower overall rates.
(2) Incursion of potassic brines into kaolinite-
bearing sandstones leads to dramatic decreases in
the temperatures of pervasive illitization compared
to scenarios where K is derived from in-situ dis-
solution of K-feldspar. (3) Illitization takes longer
to complete in sandstones wit h more reactants.
(4) The presence of illitic or micaceous substrates
that could act as preferential nuclei for illite crys-
tals is unlikely to significantly affect the timing
of illite formation except for sandstones that are
highly micaceous, that have abundant illite-rich or
muscovite-rich rock fragments, or that have grains
Figure 11. Sensitivity of illite cement
abundance to preferential nucleation on
muscovite grains using the baseline texture
and the maximum 150°C temperature
history (Figure 4A). Preferential nucleation
on muscovite leads to a lower tempera-
ture for the completion of the illitization
reaction and a greater fraction of the illite
occurrence as cement (in intergranular
pores) compared to “replacement” (in grain
dissolution pores). The magnitude of the
effect, however, is minor for the two sce-
narios with 5-vol.% muscovite. In the
maximum effect case (25-vol.% muscovite
and an activation energy for nucleation
on muscovite that is 8 kJ/mol lower than
on other substrates), the temperature for
completion of the reaction is about 5°C
(9°F) cooler than the baseline.
Lander and Bonnell 1177
with well-developed detrital or infiltrated illitic
coatings.
APPLICATION TO GEOLOGIC DATA SETS
Although the synthetic simulations are useful for
demonstrating model behavior, the model utility
can only be determined by assessing the degree
to which it reproduces geologic observations. We
evaluated the performance of the model on two
illite-be aring data sets: Jurassic sandstones from
the Norwegian Continental Shelf (NCS) and Mio-
cene sandstones in offshore Southeast Asia (SEA).
In these data sets, the reacti on of kaolin and K-
feldspar appears to be the main source for authi-
genic illite given t hat bot h reactants are present,
some illite appears to have inherited the morphol-
ogy of kaolin that it replaced, and most authigenic
illite has a fibrous habit (as opposed to a morphol-
ogy more consistent with replacement of a smectitic
precursor phase). Samples in both data sets have
mean grain sizes ranging from very fine to very coarse
and have sorting ranging from well to poor. The
SEA samples are dominantly fluvial in origin, whereas
the NCS samples are mainly shallow-marine sand-
stones with some tidal influences. The NCS data set
is composed of 189 samples from 12 wells, and the
SEA data set is made up of 62 samples from three
wells. Sample framework-grain compositions are
showninFigure12, and the range in sample tem-
perature histories is shown in Figure 13.
Our objective for this work is to determine
whether model results are consistent with the ob-
served patterns of occurrence and abundances of
illite, kaolinite (or dickite for higher temperature
samples), and K-feldspar in the two data sets. Addi-
tionally, we compare predicted illite K-Ar results
with available data and simulated fiber thicknesses
with measurements from North Sea sandstones.
The simulations we describe below use the mass-
balance option that ensures that each simulated
sample has sufficient reactants to account for the
measured volumes of authigenic kaolin and illite.
Thus, additional feldspar reacta nts are added at
the time of deposition to maintain mass balance if
the Al or K in the illite and kaolinite exceeds the
amount associated with feldspar secondary poros-
ity and replacement. Although this model config-
uration ensures sufficient reactants to account
Figure 12. Quartz, feldspar, and rock fragment (QFR) ternary diagrams showing sample data for the two analyzed data sets using the
classification scheme of Folk (1980). (A) Plot based on petrographic data. (B) Plot of reconstructed compositions for the time of de-
position. The additional reconstructed rock fragment and feldspar values are based on (1) the interpretation of parent grains for
petrographically determined grain dissolution and replacement and (2) the additional feldspar dissolution, if required, to account for Al
and K in kaolin and illite assuming mass balance on the thin-section scale.
1178 Fibrous Illite Model
for the measured amount of illite, the simulated
amount of illite may be less than the measured
amount when the model does not predict a suffi-
cient extent of the reaction. The modeled amount
of illite may exceed the measured amount when
(1) the modeled extent of illitization is extensive
and the measured extent of feldspar dissolution
and replacement by illite exceeds that required to
account for measured illite volume or (2) the sam-
ple contains both kaolinite and K-feldspar reac-
tants and the model pre dicts more extensive illite
formation than has been measured for the sample.
Unless stated otherwise, we use the same activa-
tion energ y values for nucleation and growth and
assume no activation energy benefit for nuclea-
tion on micaceous substrates. All simulations use
an A
g
value of −9×10
−12
mol/cm
2
/s, an A
n
value
of 8000 sites/cm
2
/m.y., and an exclusion radius of
0.002 mm at 100°C (212°F) for fiber nucleation
near existing crystallites. We adjusted quartz ce-
mentation kinetic and compaction model param-
eters for each sample to match the measured quartz
cement and intergranular volumes, respectively.
The purpose of this procedure is to provide the
most accurate basis possible for reconstructing in-
tergranular nucleation surface area.
Norwegian Continental Shelf
The Haltenbanken region of offshore Norway is
well suited for diagene tic s tudies because post-
depositional rifting in this area has caused Jurassic
sandstones with similar initial compositions and
textures to have experienced a broad range in ther-
mal exposures. Kaolinite and K-feldspar reactants
show a marked decline in abundance at present-
day temperatures greater than 120 to 130°C and
cease to coexist at present-day temperatures in
excess of 140°C (Bjørlykke et al., 1986; Ehrenberg
and Nadeau, 1989; Chuhan et al. 2000, 2001).
Authigenic illite, however, shows an inverse pat-
tern of occurrence. Chuhan e t al. (2000, 2001 )
concluded that the illitization reaction is m ostly
isochemical with respect to both K and Al based
on their bulk chemical analysis of samples with a
range in the extent of illitization. Formation-water
analyses from this area have K
+
concentrations that
do not exceed what would be expected for equi-
librium with K-feldspar (Egeberg and Aagaard,
1989; Bjørlykke et al., 1995).
Simulation results indicate that apart from
sample-specific input (i.e., composition, texture,
and thermal history), a single set of input param-
eters accurately predicts the rapid transition from
incipient to pervasive illitization with present-day
temperature (Figure 14A) as well as the associated
pattern of K-feldspar and kaolin-reactant abun-
dances. These results were obtained with an activa-
tion energy of 73.2 kJ/mol for both crystal nucle-
ation and growth. Simulations that use an activation
energy value of 76 kJ/mol for these two parameters,
however, systematically underpredict illite abun-
dances for samples at temperatures less than 160°C
(Figure 14A). By contrast, increased reaction rates
Figure 13. Temperature histories used in the simulations for the two natural data sets. (A) Norwegian Continental Shelf data set.
(B) Southeast Asia data set.
Lander and Bonnell 1179
Figure 14. Comparison of the pattern in authigenic illite abundance with maximum temperature from petrographic data and with calculated values for the 68-, 73.2-, and 76-kJ/mol
simulations described in the text. Sample measurements with reactants include both kaolin and K-feldspar, whereas samples lacking reactants may contain one of these two phases but not both.
1180 Fibrous Illite Model
associated with activation energy values of 68 kJ/
mol lead to simulated illite abundances that sig-
nificantly overpredict measured values for cooler
samples that have coexisting K-feldspar and kao-
linite reactants (Figure 14A).
Although the simulation results are consistent
with the present-day distribution of illite, fiber
K-Ar ages provide an independent means to assess
the veracity of the simulated timing of illite for-
mation. Ehrenberg and Nadeau (1989) made K-Ar
determinations on illite from 16 Garn Formation
samples in 9 wells in the study area and obtained
a range in values of 31 to 55 Ma. Of these wells,
one overlaps with those in our data set and two
samplesanalyzedfromithaveK-Aragesof33
and 47 Ma. The simulated K-Ar ages range from
11 to 20 Ma in the 73.2 kJ/mol run for pure, bulk
authigenic illite for the five samples in our data
set (Figure 15). Although the simulated ages are
significantly younger than the measurements,
Ehrenberg and Nadeau (1989) stated that their
analyzed samples may have a problem with de-
trital contaminants. Recent U/Pb dating of detrital
zircons in the Garn Formation shows ages ranging
from approximately 400 to 3300 Ma (Morton et al.,
2009). The potential effects of muscovite con-
tami nants on the modeled K- Ar dates are shown
as a function of the volume fraction of three rep-
resentative ages of contaminants in Figure 15.The
results indicate that as little as 1 to 3 vol.% of
Proterozoic or Archean muscovite would result
in K-Ar measurements that are consistent with
model results. Even smaller contaminant volumes
would be required to account for the K-Ar results
if detrita l K-feldspar of these ages were present
due to the mi neral’s higher K content (Ham ilton
et al., 1989). The presence of such minor contam-
inants seems reasonable given th at sandstones
from this well in our data set contain K-feldspar
(0 to 1 vol.%) and muscovite grains (0 to 0.7 vol.%)
as well as clay matrix (0.7 to 3.7 vol.%), shale rock
fragments (0.7 to 1.3 vol.%), and various other
sedimentary, metamorphic, and plutonic rock frag-
ments (1.7 to 3.6 vol.%) that contain these po-
tential contaminants.
Although the 73.2-kJ/mol simulation predicts
K-Ar ages that appear to be consistent with mea-
surements, results from the other two simulations
are not. The 76-kJ/mol simulation predicts only
incipient illitization with K-Ar ages of around 6 Ma
for the pure bulk sample. However, the simulated
K-Ar ages for the 68-kJ/mol run range from 60 to
64 Ma for pure authigenic illite. In this case, even
without contaminants, the simulated K-A r ages
are much too old to be consistent with the mea-
surements. The 73.2-kJ/mol simulation thus pro-
vides the best fit to the illite abundance and pattern
of occurrence and K-Ar ages.
This simulated timing for illitization is con-
siderably older than some other workers have
suggested for this region. Ehrenberg and Nadeau
(1989) and Bjørlykke et al. (1995) proposed that
the bulk of the illite formed during the past few
million years based on the abrupt transition from
incipient to pervasive illitization over present-day
temperatures of 120 to 140°C (248 to 284°F)
and a pulse of rapid burial during the Pliocene–
Pleistocene. One line of evidence that s upports
older illitization consistent with our model results
Figure 15. Impact of detrital contaminants on the simulated
bulk K-Ar ages for five Garn Formation samples from a well in
the Norwegian Continental Shelf data set using an activation
energy of 73.2 kJ/mol. The horizontal lines indicate K-Ar mea-
surements reported by Ehrenberg and Nadeau (1989) for two
different Garn Formation samples in this well. The shaded re-
gions indicate the range in simulated bulk K-Ar values for the five
modeled samples as a function of the age and volume of a
potential muscovite contaminant. The detrital contaminant ages
used for the calculations are representative of values for Garn
Formation zircons reported by Morton et al. (2009). Results
indicate that only small amounts of Precambrian contaminants
are required for the simula ted results to be consistent with the
measurements.
Lander and Bonnell 1181
is the considerable volume of quartz cement that
postdates the initiation of fibrous illite nucleation
and growth in sandstones from the study area.
Bonnell et al. (1999) found illite fibers that were
enveloped by up to 50 mm of quartz cement over-
growths in Viking Graben samples with similar
burial histories. Using quartz cementation mod-
els with empirically calibrated kinetic parameters,
Bonnell et al. (1999) concluded that it would
have taken tens of millions of years to account for
the volume of the quartz cement that encases the
illite fibers.
Southeast Asia
Like the NCS data set, the analyzed sandstones
in the SEA data set were d eposited in a prerift
depositional system, and nearby formation-water
K
+
concentrations do not exceed what would
be expected from equilibrium with K-feldspar
(Lundegard and Trevena, 1990). These sandstones,
however, tend to be much richer in micaceous
metamorphic rock fragments (Figure 12) and have
experienced much greater heating rates com-
pared to NCS samples (Figure 13). Despite these
differences, the pattern of correspondence between
calculated and measured illite abundances for
the 68, 73.2, and 76 kJ/mol activatio n energy
scenarios produces similar patterns (Figure 14B).
Compared to the NCS simulations, however, the
68-kJ/mol simulation results in a much lower max-
imum extent of overprediction of illite abundance.
The clay minerals in this data set were ana-
lyzed rigorously by J. Reed Glasmann in an un-
published study (2001, personal communication)
that incorporates K-Ar results in addition to op-
tical thin-section petrography, whole rock and ori-
ented clay x-ray diffraction (XRD), oxygen iso-
topic analysis, and SEM and TEM characterization.
James Aronson conducted K-Ar analyses on the less
than 0.2-mm equivalent settling diameter fraction-
size separates made by Glasmann from four sam-
ples. Glasmann’ s XRD and TEM investigations
indicate that K-feldspar and 2M
1
polytype illite
contaminants occur in coarser size fractions as
well as one of the less than 0.2-mm fraction sam-
ples. Problems with detrital contamination are to
be expected with these samples given the sub-
stantial abundances of micaceous and K-feldspar-
bearing grains (mica grains: 0.3 to 5.2 vol.%, K-
feldspar grains: 0 to 32.5 vol.%, metamorphic rock
fragments: 2.0 to 42.5 vol.%, and plutonic rock
fragments: 0 to 29.7 vol. %) as well as detrital clay
(clasts: 0 to 6.1 vol.%, matrix: 0 to 5.0 vol.%).
Thus, the measured K-Ar values likely represent a
maximum integrated age for the authigenic illite,
even for the less than 0.2-mm-size fraction samples
that lack clear indications of detrital contamination.
The simulated K-Ar dates for pure, bulk illite
for the two wells with less than 0.2-mm fraction data
without detected contaminants show a good corre-
spondence with three measurements lacking ob-
vious contamination, particularly for the 68-kJ/mol
activation energy run (Figure 16). Although the
Figure 16. Correspondence between three measured K-Ar
dates for the Southeast Asia data set (J. R. Glasmann, 2001, per-
sonal communication) and simulated values. The measurements
were made on less than 0.2-mm equivalent settling diameter-
size fractions. The two measurements with values of approxi-
mately 13 Ma are from different samples in the same well, and
the 7-Ma value comes from a sample in a different well. Cal-
culated results are shown for bulk, pure illite (no contaminants)
for the 73.2- and 68-kJ/mol runs (the 76-kJ/mol run did not create
sufficient illite to si mulate K-Ar values). The calculated values
are based on simulations for 10 samples from each well that are
within a few meters of the sample with the K-Ar measurement.
The symbol y-axis position shows the average calculated value,
and the vertical bar shows the range in calculated values.
1182 Fibrous Illite Model
73.2-kJ/mol run results underpredict the measured
values by 5 to 6 m.y. (Figure 16), this result seems
reasonable given the likely detrital contamination
problems discussed above.
Natural Data Set Discussion
The illite model reproduces the present-day pat-
tern of occurrence and abundances of illite, kaolin,
and K-feldspar in both data sets. Moreover, simu-
lation results made using the same kinetic param-
eters (73.2 kJ/mol) provide a good match to both
data sets despite their large differences in geologic
histories and grain compositions. Optimizing the
activation energy independently for each data
set resulted in only minor differences (73.4 and
73.0 kJ/mol, respectively, for the NCS and SEA data
sets). By comparison, optimized activation energies
for quartz precipitation of 60.7 and 55.5 kJ/mol,
respectively, for the NCS and SEA data sets, show
much larger differences (Lander et al., 2008).
The simulated K-Ar dates from the 73.2 kJ/mol
runs also appear to be consistent with the mea-
surements given their uncertainties. Additionally,
the particle dimensions from these simulations are
broadly consistent with Nagy’s (1994) data from
sandstones from the United Kingdom North Sea
(Figure 17). The reconstructed depositional feld-
spar abundances for both data sets significantly ex-
ceed present-day abundances if we assume that Al
and K are conserved on the thin-section scale and
if we use illite and kaolin abundances for reference
(Figure 12).
In these simulations, we assumed no activation
energy benefit for nucleation on micaceous grains.
Both data sets, however, contain micaceous grains,
and they are especially abundant in metamorphic
rock fragments in the SEA data set. We therefore
ran simulations for a scenario where we assumed a
20-kJ/mol activation energy benefit for nucleating
on micaceous surfaces compared to other surface
types. After parameter optimization, this scenario
has a slightly higher activation energy for crystal
growth of 73.6 kJ/mol while having a comparable
quality of fit to the data. Integrated times of illiti-
zation are slightly older (0.2 to 2.6 m.y. older on
average, respectively, for SEA and NCS), and P50
widths are somewhat smaller (by 12 to 19%, re-
spectively, for NCS and SEA), but the results are
otherwisesimilar.Thus,themodelisrelativelyin-
sensitive to an activation energy benefit for nucle-
ation on micaceous grains as long a s the kinetic
parameters are optimized to match the measured
pattern of illite occurrence.
CONCLUSIONS
Our kinetic model for fibrous illite nucleation and
growth reproduces the pattern in occurrence of
fibrous illite as well as kaolin and K-feldspar reac-
tants in two geologically distinct natural data sets
using a single set of model parameters. Simulation
results also appear to be at least broadly consistent
with available K-Ar ages and data on illite fiber
dimensions. We have shown that if this reaction is
kinetically controlled, then the temperature range
where illitization occurs will be a strong function
of the thermal history. Therefore, the illitization
process should not be expected to be a universal
function of a simple temperature threshold of 120
Figure 17. Comparison of average fibrous illite particle thick-
nesses reported by N agy (1994) for two size separate s (dark
symbols) with simulated values for the two natural data sets
using the 73.2-kJ/mol scenario (light symbols). The symbols for
the simulation results indicate the predicted 50th percentile (P50)
thickness value by crystallite count for each sample. The burial
histories from the Norwegian Continental Shelf (NCS) d ata set
are similar to those experienced by Nagy’s (1994) samples, but
the Southeast Asia (SEA) samples experienced much higher
heating rates.
Lander and Bonnell 1183
to 140°C (248 to 284°F) as has been suggested
by other workers (Ehrenberg and Nadeau, 1989;
Bjørlykke et al., 1995; Bjørlykke, 1998). This kinetic
behavior is well illustrated by the tendency for the
simulations of NCS samples to have cooler inte-
grated temperatures of formation and longer dura-
tions of illit ization than SEA sa mples that have
been exposed to far greater heating rates (Figure 18).
Model sensitivity tests indicate greatly acceler-
ated rates of illitization in the presence of potassic
brines. Results for a synthetic Jurassic-aged sample
show a dramatic drop from 135 to 80°C (275 to
176°F) in the paleotemperature of pervasive illiti-
zation when K
+
activity increases from levels ex-
pected for equilibrium with K-feldspar to that as-
sociated with fluid inclusions in naturally occurring
halite. Sensitivity tests also predict that otherwise
similar sandstones with comparable maximum
temperatures but longer thermal residence times
are illitized at lower temperatures and at slower
overall rates and that completion of the reaction
takes longer in sandstones with more reactants but
similar thermal exposures.
REFERENCES CITED
Aagaard, P., J. S. Jahren, and P. K. Egeberg, 1992, North Sea
clastic diagenesis and formation water constraints, in Y.
K. Kharaka and A. S. Maest, eds., Water-rock interac-
tion: Proceedings of the 7th International Symposium,
Park City, Utah, July 13–18.
Altaner, S. P., 1986, Comparison of rates of smectite illitiza-
tion with rates of K-feldspar dissolution: Clays and Clay
Minerals, v. 34, p. 608–611, doi:10.1346/CCMN.1986
.0340517.
Bazin, B., É. Brosse, and F. Sommer, 1997a, Chemistry of oil-
field brines in relation to diagenesis of reservoirs: 1. Use
of mineral stability fields to reconstruct in situ water
composition. Example of the Mahakam Basin: Marine
and Petroleum Geology, v. 14, p. 481–495, doi:10.1016
/S0264-8172(97)00004-4.
Bazin, B., É. Brosse, and F. Sommer, 1997b, Chemistry of oil-
field brines in relation to diagenesis of reservoirs: 2. Re-
construction of paleo-water composition for modeling
illite diagenesis in the Greater Alwyn area (North Sea):
Marine and Petroleum Geology, v. 14, p. 497–511,
doi:10.1016/S0264-8172(97)00005-6.
Berger, G., J. C. Lacharpagne, B. Velde, D. Beaufort, and B.
Lanson, 1997, Kinetic constraints on illitization reac-
tions and the effects of organic diagenesis in sandstone/
shale sequences: Applied Geochemistry, v. 12, p. 23–35,
doi:10.1016/S0883-2927(96)00051-0.
Bethke, C. M., 1996, Geochemical reaction modeling—Con-
cepts and applications: New York, O xford University
Press, 397 p.
Bjørkum, P. A., and N. Gjelsvik, 1988, An isochemical model
for formation of authigenic kaolinite, K-feldspar and illite
in sediments: Journal of Sedimentary Petrology, v. 58,
p. 506–511.
Bjørlykke, K., 1998, Clay mineral diagenesis in sedimentary
basins—A key to the prediction of rock properties. Ex-
amples from the North Sea Basin: Clay Minerals, v. 33,
p. 15–34, doi:10.1180/000985598545390.
Bjørlykke,K.,P.Aagaard,H.Dypvik,D.S.Hastings,and
A. S. Harper, 1986, Diagenesis and reservoir properties
of Jurassic sandstones from the Haltenbanken area, off-
shore mid Norway, in A. M. Spencer, ed., Petroleum geol-
ogy of the Northern European margin: Norwegian Petro-
leum Society, London, Graham and Trotman, p. 285–292.
Bjørlykke,K.,T.Nedkvitne,M.Ramm,andG.C.Saigal,
1992, Diagenetic processes in the Brent group, in A. C.
Morton,R.S.Haszeldine,M.R.Giles,andS.Brown,
eds., Geology of the Brent Group: Geological Society
(London) Special Publication 61, p. 263–288.
Bjørlykke, K., P. Aagaard, P. K. Egeberg, and S. P. Simmons,
1995, Geochemical constraints from formation water
analyses from the North Sea and the Gulf Coast basins
on quartz, feldspar and illite precipitation in reservoir
rocks, in J. M. Cubitt and W. A. England, eds., The geo-
chemistry of reservoirs: Geological Society (London)
Special Publication 86, p. 33–50.
Blum, A. E., and L. L. Stillings, 1995, Feldspar dissolution
kinetics, in A. F. White and S. L. Brantley, eds., Chemical
weathering rates of silicate minerals, v. 31, p. 291–346.
Figure 18. Comparison of the integrated temperature of illiti-
zation with the duration of illite formation for samples from the
Norwegian Continental Shelf and Southeast Asia data sets using
the 73.2-kJ/mol run scenario. The duration of illitization is based
on the time span between the formation of the 10th and 90th
volume percentiles, whereas the integrated temperature is the
volume-weighted average temperature for illite formation. The
much higher thermal exposures of the Southeast Asia data set
have resulted in simulated illitization occurring at significantly
faster rates and higher temperatures compared to Norwegian
Continental Shelf samples.
1184 Fibrous Illite Model
Bonnell, L. M., C. J. Lowrey, and A. A. Bray, 1999, The timing
of illitization, Haltenbanken, mid-Norway (abs.): AAPG
Annual Meeting Program with Abstracts, v. 8, p. A14.
Bove, D. J., D. D. Eberl, D. K. McCarty, and G. P. Meeker,
2002, Characterization and modeling of illite crystal par-
ticles and growth mechanisms in a zoned hydrothermal
deposit, Lake City, Colorado: American Miner alogist,
v. 87, p. 1546–1556.
Brosse,E.,J.Matthews,B.Baxin,Y.LeGallo,andF.Sommer,
2000, Related quartz and illite cementation in the Brent
sandstones: A modeling approach, in R. H. Worden and
S. Morad, eds., Quartz cementation in sandstones: In-
ternational Association of Sedimentologists Special Pub-
lication 29, p. 51–66.
Canals, M., and J. D. Meunier, 1995, A model for porosity
reduction in quartzite reservoirs by quartz cementation:
Geochimica et Cosmochimica Acta, v. 59, p. 699–709,
doi:10.1016/0016-7037(94)00355-P.
Cendón, D. I., C. Ayora, J. J. Pueyo, and C. Taberner, 2003,
The geochemical evolution of the Catalan potash sub-
basin, South Pyrenean foreland basin (Spain): Chemical
Geology, v. 200, p. 339–357, doi:10.1016/S0009-2541
(03)00195-5.
Chermak, J. A., and J. D. Rimstidt, 1990, The hydrothermal
transformation of kaolinite to muscovite/illite: Geochimica
et Cosmochimica Acta, v. 54, p. 2979–2990, doi:10
.1016/0016-7037(90)90115-2.
Chuhan, F. A., K. Bjørlykke, and C. Lowrey, 2000, The role
of provenance in illitization of deeply buried reservoir
sandstones from Haltenbanken and north Viking Graben,
offshore Norway: Marine and Petroleum Geology, v. 17,
p. 673–689.
Chuhan, F. A., K. Bjørlykke, and C. J. Lowrey, 2001, Closed-
system burial diagenesis in reservoir sandstones: Exam-
ples from the Garn Formation at Haltenbanken area, off-
shore mid-Norway: Journal of Sedimentary Research,
v. 71 , p. 15–26, doi:10.1306/041100710015.
Clauer, N., and S. Chaudhuri, 1995, Clays in crustal en-
vironment: Isotop e dating and tracing: Heidelberg,
Springer, 359 p.
Clauer, N., J. D. Cocker, and S. Chaudhuri, 1992, Isotopic
dating of diagenetic illites in reservoir sandstones: Influ-
ence of the investigator effect: SEPM (Society for Sed-
imentary Geology), v. 47, p. 5–12.
Clauer, N., N. Liewig, B. Ledesert, and H. Zwingmann,
2008, Thermal history of Triassic sandstones from the
Vosges Mountains-Rhine Graben rifting area, NE France,
based on K-Ar illite dating: Clay Minerals, v. 43, p. 363–
379, doi:10.1180/claymin.2008.043.3.03.
Cocker, J. D., 1986, Authigenic illite morphology: Appear-
ances can be deceiving: AAPG Bulletin, v. 70, p. 575.
Deer, W. A., R. A. Howie, and J. Zussman, 1966, Introduction
to the rock-forming minerals: London, Longman, 528 p.
De Ros, L. F., 1998, Heterogeneous generation and evolution
of diagenetic q uartarenites in the Silurian–Devonian
Furnas Formation of the Paraná Basin, southern Brazil:
Sedimentary Geology, v. 116, p. 99–129, doi:10.1016
/S0037-0738(97)00081-X.
Eberl,D.D.,V.A.Drits,andJ.Ś rodo ń , 1998, Deducing
growth mechanisms for mineral s from the shapes of
crystal size distributions: American Journal of Science,
v. 298, p. 499–533.
Egeberg, P. K., and P. Aagaard, 1989, Origin and evolution of
formation waters from oil fields on the Norwegian Shelf:
Applied Geochemistry, v. 4, p. 131–142, doi:10.1016
/0883-2927(89)90044-9.
Ehrenberg, S. N., and P. H. Nadeau, 1989, Formation of dia-
genetic illite in sandstones of the Garn Formation, Hal-
tenbanken area, mid-Norwegian Continental Shelf: Clay
Minerals, v. 24, p. 233–253, doi:10.1180/claymin.1989
.024.2.09.
Folk, R. L., 1980, Petrology of sedimentary rocks: Austin,
Texas, Hemphill Publishing Company, 184 p.
Fran ks, S. G., and H. Zwingmann, 2010, Origin and timing
of late diagenetic illite in the Permian–Carboniferous
Unayzah sandstone reservoirs of Saudi Arabia: AAPG
Bulletin, v. 94, p. 1133–1159, doi:10.1306/0421100914 2.
Gaupp, R., A. Matter, J. Platt, K. Ramseyer, and J. Walzebuck,
1993, D iagenesis and fluid evolution of deeply buried
Permian (Rotliegende) gas reservoirs, northwest Ger-
many: AAPG Bulletin, v. 77, p. 1111–1128.
Girard, J. P., I. A. Munz, H. Johansen, J. C. Lacharpagne, and
F. Sommer, 2002, Diagenesis of the Hild Brent sand-
stones, northern North Sea: Isotopic evidence for the
prevailing influ ence of deep basinal water: Journal of
Sedimentary Research, v. 72, p. 746–759, doi:10.1306
/040102720746.
Glasmann, J. R., 1992, The fate of feldspar in Brent Group
reservoirs, North Sea: A regional synthesis of diagen-
esis in shallow, intermediate, and deep burial environ-
ments, in A. C. Morton, R. S. Has zeldine , M. R. Giles,
and S. Brown, eds., Geology of the Brent Gr oup: Geo-
logical Society (London) Special Publication 61, p. 329–
350.
Glasmann, J. R., P. D. Lundegard, R. A. Clark, B. K. Penny,
and I. D. Collins, 1989, Geochemical evidence for the
history of diagenesis and fluid migration: Brent sand-
stone, Heather field, North Sea: Clay Minerals, v. 24,
p. 255–284, doi:10.1180/claymin.1989.024.2.10.
Güven, N., 2001, Mica structure and fibrous growth of illite:
Clays and Clay Minerals, v. 49, p. 189–196, doi:10
.1346/CCMN.2001.0490301.
Güven, N., W. F. Hower, and D. K. Davies, 1980, Nature of
authigenic illites in sandstone reservoirs: Journal of Sedi-
mentary Petrology, v. 50, p. 761–766.
Hamilton, P. J., S. Kelley, and A. E. Fallick, 1989, K-Ar dat-
ing of illite in h ydrocarbon reservoirs: Clay Minerals,
v. 24, p. 215–232, doi:10.1180/claymin.1989.024.2.08.
Hamilton, P. J., M. R. Giles, and P. Ainsworth, 1992, K-Ar
dating of illites in Bren t Grou p reserv oirs: A regional
perspective, in A. Morton, R. Haszeldine, M. Giles, and
S. Brown, eds., Geology of the Brent Group: Geologi-
cal Society (London) Special Publication 61, p. 377–
400.
Hunziker, J. C., 1986, The evolution of illite to muscovite:
An example of t he behavior of isot opes in low-grade
metamorphic terrains: Chemical Geology, v. 57, p. 31–
40, doi:10.1016/0009-2541(86)90092-6.
Hutcheon, I., M. Shevalier, and H. J. Abercrombie, 1993,
pH buffering by m etastable mineral-fluid equilibria
Lander and Bonnell 1185
and evolution of carbon dioxide fugacity during burial
diagenesis: Geochimica et Cosmochimica Acta, v. 54,
p. 1017–1027, doi:10.1016/0016-7037(93)90037-W.
Jahren, J. S., and P. Aagaard, 1989, Compositional variations
in diagenetic chlorites and illites, and relationships with
formation-water chemistry: Clay Minerals, v. 24, p. 157–
170, doi:10.1180/claymin.1989.024.2.04.
Kantorowicz, J., 1984, The nature, origin and distribution of
authigenic cla y minerals, from Middle J urassi c Raven-
scar and Brent Group sandstones: Clay Minerals, v. 19,
p. 359–375, doi:10.1180/claymin.1984.019.3.08.
Kantorowicz, J. D., 1990, The influence of variations in illite
morphology on the permeability of Middle Jurassic Brent
Group sandstones, Cormorant field, U.K. North Sea:
Marine Petroleum Geology, v. 7, p. 66–74, doi:10.1016
/0264-8172(90)90057-N.
Kharaka,Y.K.,R.W.Hull,andW.W.Carothers,1985,
Water-rock interactions in sediment ary basins, in D. L
Gautier,Y.K.Kharaka,andR.C.Surdam,eds.,Role
of organic matter in sediment diagenesis: SEPM Short
Course 17, p. 79–176.
Kittel, C., a nd H. Kroemer, 1980, Thermal physics: New
York, Freeman, 473 p.
Lander, R. H., S. Bloch, S. Mehta, and C. D. Atkinson, 1990,
Burial diagenesis of paleosols in the Giant Yacheng Gas
field, People’s Republic of China: Bearing on illite reac-
tion pathways: Journal of Sedimentary Research, v. 61,
p. 256–268.
Lander, R. H., R. E. Larese, and L. M. Bonnell, 2008, Toward
more accurate quartz cement models: The importance of
euhedral versus noneuhedral growth rates: AAPG Bulle-
tin, v. 92, p. 1537–1563, doi:10.1306/07160808037.
Lanson,B.,D.Beaufort,G.Berger,J.Baradat,andJ.C.
Lacharpagne, 1996, Illitization of dia genetic kaolinite-
to-dickite conversion series: Late-stage diagenesis of the
Lower Permian Rotliegende sandstone reservoir, off-
shore of the Netherlands: Journal of Sedimentary Re-
search, v. 66, p. 501–518.
Lanson, B., D. Beaufort, G. Berger, A. Bauer, A. Cassagnabère,
and A. Meunier, 2002, Authigenic kaolin and illitic
minerals duri ng burial diagenesis of sandstones: A re-
view: Clay Minerals, v. 37, p. 1– 22, doi:10.1180
/00098550237 10014.
Lasaga, A. C., 1998, Kinetic theory in the earth sciences:
Princeton series in geochemistry: Princeton, New Jersey,
Princeton University Press, 811 p.
Lee, M., 1984, Diagenesis of the Permian Rotliegendes sand-
stone, North Sea: K-Ar,
18
0/
16
O and p etrographic evi-
dence: Ph.D. thesis, Case Western Reserve Uni versity,
Cleveland, Ohio, 362 p.
Lee, M., J. L. Aronson, and S. M. Savin, 1985, K/Ar dating of
Rotliegendes sandst one, Netherlands: AAPG Bulletin,
v. 68, p. 1381–1385.
Lee, M., J. L. Aronson, and S. M. Savin, 1989, Timing and
conditions of Permian Rotliegendes sandstone diagen-
esis, southern North Sea: K/Ar and oxygen isotope data:
AAPG Bulletin, v. 73, p. 195–215.
Lichtner, P. C., 1988, The quasi-stationary state approxima-
tion to coupled mass transport and fluid-rock interaction
in a porous medium: Geochimica et Cosmochimica
Acta,v.52,p.143–165, doi:10.1016/0016-7037(88)
90063-4.
Lundegard, P. D., and A. S. Trevena, 1990, Sandstone diag-
enesis in the Pattani Basin (Gulf of Thailand): History of
water-rock interaction and comparison with the Gulf of
Mexico: Applied Geochemistry, v. 5, p. 669–685,
doi:10.1016/0883-2927(90)90064-C.
Macchi, L., C. D. Curtis, A. Levison, K. Woodward, and C. R.
Hughes, 1990, Chemistry, morphology, and distribution
of illi tes from Morecambe gas field, Irish Se a, offshore
United Kingdom: AAPG Bulletin, v. 74, p. 296–308.
Maher, K., C. I. Steefel, D. J. DePaolo, and B. E. Viani, 2006,
The mineral dissolution rate conundrum: Insights from
reactive transport modeling of U isotopes and pore fluid
chemistry in marine sediments: Geochimica et Cosmo-
chimica Acta, v. 70, p. 337–363, doi:10.1016/j.gca
.2005.09.001.
Matthews, J. C., B. Velde, and H. Johansen, 1994, Significance
of K-Ar ages of authigenic illitic clay-mine rals in sand-
stones and s hales from the North Sea: C lay Minerals,
v. 29, p. 379–389, doi:10.1180/claymin.1994.029.3.09.
McHardy, W. J., M. J. Wilson, and J. M. Tait, 1982, Electron
microscope and x-ray diffraction studies of filamentous il-
litic clay from sandstones of the Magnus field: Clay Miner-
als, v. 17, p. 23–29, doi:10.1180/claymin.1982.017.1.04.
Merino, E., P. Ortoleva, and P. Strickholm, 1983, Generation
of evenly spaced pressure-solution seams during (late)
diagenesis : A kinetic theory: Cont ributions to Mineral-
ogy and Petrology, v. 82, p. 360–370
Meunier, A., 2006, Why are clay minerals small?: Clay Miner-
als, v. 41, p. 551–566, doi:10.1180/0009855064120205.
Meunier, A., B. Velde, and P. Zalba, 2004, Illite K-Ar dating
and crystal growth processes in diagenetic environments:
A critical review: Terra Nova, v. 16, p. 296–304, doi:10
.1111/j.1365-3121.2004.00563.x.
Midtbø, R. E. A., J. M. Rykkje, and M. Ramm, 2000, Deep
burial diagenesis and reservoir quality along the eastern
flank of the Viking Graben. Evidence for illitization and
quartz cementation after hydrocarbon emplacement:
Clay Minerals, v. 35, p. 227– 237, doi:10.1180
/000985500546602.
Morton, A., C. Hallsworth, D. Strogen, A. Whitham, and M.
Fanning, 2009, Evolution of provenance in the NE
Atlantic rift: The Early–Middle Jurass ic succes sion in
the Heidrun field,