Well-characterized samples from the Podhale Basin, southern Poland, formed the basis for exploring and illuminating subtle diagenetic changes to a mudstone toward the upper end of the diagenetic window, prior to metamorphism. Transmission electron microscopy (TEM) performed on dispersed grains and ion-beam thinned preparations, selected area diffraction patterns, and chemistry by TEM-EDS (energy dispersive spectra) augmented mineralogy and fabric data. The deepest samples show no change in their percent illite in illite-smectite (I-S), yet I-S–phase octahedral Fe3+ and Al3+ are statistically different between samples. A decrease in the Fe3+ concentration in the octahedral sheet correlates with an increase in I-S fabric intensity and apparent crystallinity. The D-statistic from the Kolmogorov-Smirnov test on TEM-EDS data describes statistical differences in the I-S chemistry. Previous work on these samples showed a significant increase in the preferred orientation of the I-S phase across the smectite to illite transition and a significant slowdown in the rate of development of preferred orientation beyond the termination of smectite illitization. Lattice fringe images describe an I-S morphology that coalesces into larger and tighter packets with increasing burial temperature and a decrease in I-S packet contact angle, yet some evidence for smectite collapse structures is retained. The deepest sample shows the thickest, most coherent I-S packets. We propose that the deepest samples in the Podhale Basin describe the precursor stage in phyllosilicate fabric preferred orientation increase from diagenesis into metamorphism, where continued evolution of crystallite packets and associated crystallinity create higher I-S fabric intensities as the structural formulae of I-S approach an end-member composition.

The physical, chemical, and mineralogical changes that transform muds into mudstones and ultimately metamorphic pelites have been studied for many years (Sorby, 1853; Rieke and Chilingarian, 1974; Weaver, 1989; Bjørlykke and Høeg, 1997). With respect to mineralogical change, particular attention has been paid to the major reactions involving clay minerals, most obviously the transformation of smectite to illite and additional mineral changes, for example quartz and chlorite precipitation, associated with that reaction (e.g., Perry and Hower, 1970; Hower et al., 1976; Boles and Franks, 1979; Nadeau et al., 2002). While X-ray diffraction has charted the mineralogical changes, transmission electron microscopy (TEM) has been used to examine the microstructural and chemical changes involved not only in the smectite to illite transformation (Ahn and Peacor, 1986; Bell, 1986; Klimentidis and Mackinnon, 1986; Inoue et al., 1987a, 1987b; Jiang et al., 1994; Hover et al., 1999; Masuda et al., 2001; Nadeau et al., 2002; Kim et al., 2004), but also in the reactions involved in low-grade metamorphism (Merriman and Peacor, 1998; Merriman, 2002).

Other work has focused on the processes by which the initially random arrangement of phyllosilicate minerals in mud becomes organized into the highly aligned fabric observed in metapelites (Oertel and Curtis, 1972; Curtis et al., 1980; Ho et al., 1999; Jacob et al., 2000; Aplin et al., 2006; Day-Stirrat et al., 2008a, 2008b). Discussion has centered on the relative roles of mechanically driven rearrangement of particles and mineralogical changes in which neoformed phyllosilicate minerals grow normal to the principal effective stress. While laboratory compaction experiments show that mechanical rearrangement of phyllosilicates is feasible (Djéran-Maigre et al., 1998; Haines et al., 2009; Voltolini et al., 2009; Day-Stirrat et al., 2011), both Ho et al. (1999) and Day-Stirrat et al. (2008a) observed a major enhancement to the preferred orientation of illite-smectite (I-S) across the smectite to illite transition in the Gulf of Mexico and the Podhale Basin of southern Poland, respectively. Day-Stirrat et al. (2008a) suggested that the enhanced fabric intensity through the smectite to illite transition window was indicative of dissolution of smectite and growth of new illite perpendicular to principal effective stress.

The changes in phyllosilicate fabric observed at the end of the main smectite to illite transition are nevertheless substantially lower than those observed in low-grade metapelites (Jacob et al., 2000). This implies continuing rearrangement of fabric at temperatures and stresses higher than those associated with the smectite illitization. In contact metamorphism, enhancement of phyllosilicates fabrics close to the heating body have been recorded (Ho et al., 1995). In the Podhale Basin (Fig. 1), Day-Stirrat et al.’s (2008a) data tentatively suggested a continued increase in I-S (and chlorite) fabric intensity beyond the apparent termination of the smectite to illite transition or certainly the mineral reaction slowdown (Fig. 2). This slowdown occurs over an additional burial of 2 km and a temperature increase of 40 °C (∼115 °C to 150 °C). Because the porosity of these deeply buried samples is low and pore sizes are smaller than grain sizes, increases in the alignment of phyllosilicate grains are unlikely to result from mechanical processes, but rather from dissolution and reprecipitation processes that may be revealed by changes in I-S chemistry or microfabric. Changes in mineral chemistry have implications for density, and microfabric impacts anisotropy and velocity. The practical importance of these changes is that both density and velocity are key parameters in estimating both porosity and pore pressure in mud-rich sequences, particularly in circumstances where sediments have been unloaded (Bowers, 1995; Lahann and Swarbrick, 2011; Goulty and Sargent, 2016; Goulty et al., 2016). Because relationships between vertical effective stress (VES) and porosity and density are difficult to constrain in diagenetically mature samples (Yang and Aplin, 2004), an examination of the detailed chemistry of mineral change in a low-porosity system is timely.

In this study, therefore, we take some well-characterized samples from the deepest part of the Podhale Basin and perform TEM on both dispersed grains and ion-beam thinned preparations. The TEM-EDS (energy dispersive spectra) data allow us to look for compositional changes at and beyond the smectite to illite transition, while lattice fringe images allow a visual description of the change in crystallite morphology, thickness, and fabric. The samples in this paper thus represent a part of the history a mud takes on its journey to a metamorphic pelite.

The Paleogene Podhale Basin of southern Poland is situated between the Pieniny klippe belt to the north and the Tatra Mountains to the south (Fig. 1). The basin is filled with what is termed the Podhale Flysch (Olszewska and Wieczorek, 1998), deposited by submarine fans (Westwalewicz-Magilska, 1986) and covering a Mesozoic basement that is exhumed in the Tatra Mountains.

X-ray diffraction results (Table 1) on the Paleogene mudstones (Środoń et al., 2006a) describe an extremely homogeneous detrital mineral composition, with regular and clear diagenetic trends with depth. Based on grain-density trends, Środoń et al. (2006b) argued that two wells, Chochołów PIG-1 and Bukowina Tatrzańska PIG-1, can be seen as a continuous burial profile, in which a ∼500 m overlap produces a continuous trend in percentage of smectite in the mixed-layer phase I-S, as well as predictable increases in quartz and chlorite and decreases in kaolinite and potassium feldspar. Present-day and calculated paleo-geothermal gradients are similar in both wells (∼20–25 °C km–1). The overlap proposed by Środoń et al. (2006b) is consistent with thermal maturity data from Poprawa and Marynowski (2005). The maximum burial of the Podhale Basin was achieved at ca. 17 Ma, based on K-Ar dates from clay mineral separates from bentonites (Środoń et al., 2006b), and maximum burial was deeper than present-day burial. Marynowski et al. (2006) presented a burial history profile through the center of the basin that shows rapid Oligocene burial followed by Miocene uplift. Porosity data recorded by Day-Stirrat et al. (2008a) show a consistent decrease through the established synthetic profile. The area was the subject of an apatite fission-track analysis by Anczkiewicz (2006), who also concluded that the top of Bukowina Tatrzańska PIG-1 had previously been much deeper (totally reset tracks) than the top of Chochołów PIG-1 (partially reset tracks) and the overlying material subsequently was eroded.

The smoothness of the diagenetic trends (Środoń et al., 2006b), the continuity of physical trends in grain density (Środoń et al., 2006b), phyllosilicate preferred orientation and porosity (Day-Stirrat et al., 2008a), and rapid burial and uplift (Anczkiewicz, 2006) suggest that the submarine fan depositional system described by Westwalewicz-Magilska (1986) was fed by a consistent source area over the period of deposition. The rapid burial of the fore-arc basin system (Tari et al., 1993) of the Podhale Flysch probably mitigated significant progradation of the submarine fans, leading to the consistent trends noted above due to a consistent provenance.

Samples

The sample set consists of four fragments of cores selected from two boreholes in the Podhale Basin (Fig. 1): Chochołów PIG-1 in the west (samples Chochołów-06 and Chochołów-60) and Bukowina Tatrzańska PIG-1 in the east (samples Bukowina Tatrzańska–06 and Bukowina Tatrzańska–41). These samples cover a maximum temperature and original burial depth range of ∼50–150 °C and ∼2500–7000 m (Table 1). Additional detailed sample information can be found in Marynowski et al. (2006), Środoń et al. (2006b), and Day-Stirrat et al. (2008a). Sample Chochołów-06 has 50% illite in I-S, whereas Chochołów-60, Bukowina Tatrzańska–06, and Bukowina Tatrzańska–41 all have 76% illite in I-S.

Transmission Electron Microscopy

Chemistry of individual I-S phyllosilicate particles was determined using TEM-EDS; crystallite images and selected area diffraction (SAED) patterns were also obtained. Samples were examined at the University of Leeds (UK) using a Philips/FEI CM200 electron microscope equipped with a field emission gun and a Gatan imaging filter. The extinction voltage was set at 3.21 kV, giving a typical energy resolution of 0.8 eV.

Mudstone samples were disaggregated using a gentle freeze-thaw method, which does not crush individual particles (Yang and Aplin, 1997). The <2 μm fraction of the sample was then separated by centrifugation. Selected <2 μm fractions were prepared for TEM-EDS by dispersing 0.2 g of sample in excess ethanol. Approximately 10 μL of the dilute suspension was placed on a carbon-coated 200-mesh copper grid and allowed to evaporate to dryness. This technique assumes that phyllosilicate particles are aligned with (00l) planes approximately perpendicular to the electron beam (c*-axis parallel to beam). Care was taken to obtain SAED patterns from thin grains, free from the overlap of other grains. Magnification was at 50,000×, and EDS data were acquired at between 1000 and 3000 counts s–1 with a live time of 50 s using a 75Å beam diameter on the same spot as the SAED. Biotite and paragonite standards were used to obtain K-factors for the transformation of intensity ratios to concentrations (Cliff and Lorimer, 1975). K-factors relate the measured characteristic X-ray intensities to the composition of the specimen. Oxygen was not measured as it is strongly affected by differences in sample thickness. Atomic concentration ratios were converted into normalized mineral formulae using an anionic charge of 22 (O10[OH]2) and assuming that all iron occurs as Fe3+ (Weaver, 1989; Moore and Reynolds, 1997). Oxide weight percents were calculated by normalizing the atomic ratios to 95 wt% (Merriman et al., 1995). Because alkali loss, particularly that of potassium, is a significant problem in TEM-EDS analysis (van der Pluijm et al., 1988), a consistent 50 s count time was used for all samples. A loss of potassium was assumed, and the sodium content was not included in the normalization calculations. Omitting Na+ from the mineral formula does not seriously affect interlayer charge as it comprises <0.1 cations per unit formula. All Mg2+ and Fe3+ were assigned to the octahedral sheet. Estimated uncertainties in the atomic proportions are: Si4+ and Al3+, ∼±0.1 cations per unit formula; Fe3+, Mg2+, Ti4+, ∼±0.05 cations per unit formula; K+, ∼±0.2 cations per unit formula (Peacor, 1992; Warren and Ransom, 1992).

A thin section previously prepared for backscattered electron imaging (Day-Stirrat et al., 2008a) produced a sample stub that was cut 200–400 μm thick for high-resolution X-ray texture goniometry (Day-Stirrat et al., 2008a). The same sample stub was prepared for lattice fringe imaging with a LR White resin treatment (Kim et al., 1995) and was used prior to sample preparation in order to prevent the collapse of smectite layers in the high-vacuum environment of the TEM. The preparation aimed to look at I-S in its a or b planes (c* perpendicular to beam). TEM in this mode has the ability to image the size of crystallite packets and document layer terminations between crystallites. Aluminum washers of 3 mm diameter were attached to randomly selected areas on the prepared TEM thin section, and the sample was ion-beam thinned and carbon coated for TEM observation. Lattice fringe observations were obtained using a Phillips CM12 scanning transmission electron microscope (STEM) at the University of Michigan (USA). The STEM was operated at an accelerating voltage of 120 kV and a beam current of ∼10 nA.

Statistical Analysis (Kolmogorov-Smirnov Test)

The Kolmogorov-Smirnov (K-S) test is a method that expresses the similarity or difference between two data sets (Stuart et al., 1999). The test was used on TEM-EDS data from the Podhale Basin samples. The K-S test is non-parametric and does not require a particular distribution of data (e.g., data normally distributed). It can be used on small data sets (8–13 results), where simply presenting the arithmetic average of a result implies a normal distribution, and a visual appraisal of the similarity of data sets. For a data set of 20 points, the test is very simple but powerful: the data are ordered and the lowest value plotted at 0.05 (1/20), the second lowest value plotted at 0.1 (2/20), and so on up to 1 to complete the cumulative distribution. Distributions can be compared visually, and a D-statistic is calculated as the maximum difference or separation between two cumulative distributions and expressed as a percentage.

Selected Area Diffraction (SAED) Patterns

Morphologies of grains for which SAED patterns were obtained range from euhedral crystallites (Fig. 3) to subhedral crystallites (Fig. 4). The associated SAED patterns, taken at thin edges of crystallites, have a strong hexagonal arrangement of single-crystal diffraction spots in all samples. These spots correspond to (h,k,l) reflections. The presence of sharp hexant reflections implies coherence between individual layers, and the absence of diffuse diffraction rings is consistent with a lack of turbostratic defects, the latter of which is characteristic of smectitic interlayers.

Typical SAED patterns for each sample are presented in Figure 5. Chochołów-06 is the only sample that deviates from the hexagonal single-crystal diffraction patterns observed for Chochołów-60, Bukowina Tatrzańska–06, and Bukowina Tatrzańska–41. Chochołów-06 has a well-defined coherence of layers in its mixed-layer crystal particle with varying orientations of these particles around c* (or Z) producing a slight ring effect. However, one mixed-layer crystal is thick enough to define the dominant single-crystal pattern.

High-Resolution TEM Imaging

Lattice fringe images of illite and I-S in samples from Bukowina Tatrzańska–06 and Bukowina Tatrzańska–41 are presented in Figure 6. Bukowina Tatrzańska–06 shows thin illite packets which, based on a 10Å lattice spacing, are typically around five layers and are situated adjacent to I-S packets of similar thickness. Some I-S mixed layers show some lattice defects such as layer terminations (Fig. 6A), while others are straight crystals (Fig. 6B). In comparison, the samples from Bukowina Tatrzańska–41 typically reveal thicker I-S particles of 10–15 layers (Figs. 6C and 6D). Here, some diagenetic crystallite packets are terminated against thick illite minerals of, presumably, detrital origin. Illite-smectite crystallites show some edge dislocations defined by terminations of layers of illite, probably inherited from highly imperfect smectite precursor structures. In general, the samples in Figure 6 show substantial I-S growing adjacent to authigenic and detrital illite minerals. These I-S mixed layers display variable lattice morphologies with some collapse structures, detailing the probable prior existence of an expandable smectite component.

Chemistry

Bulk mineralogical data from Chochołów PIG-1 and Bukowina Tatrzańska PIG-1 are synthesized from Środoń et al. (2006b) and Day-Stirrat et al. (2008a) (Table 1). Standard structural formulae for an I-S half-cell and associated elemental concentrations expressed as weight percent oxides are presented in Tables 25 for Chochołów-06, Chochołów-60, Bukowina Tatrzańska–06, and Bukowina Tatrzańska–41, respectively. Both the octahedral cation totals (range = 1.92–2.11) and the chemical compositions are within the previously published range for I-S; some K+ values are outside the range of illite and even muscovite (>1) and are, therefore, unrealistic and probably related to the noted mobility of potassium under an electron beam (Ahn and Peacor, 1986; Brusewitz, 1986; Ramseyer and Boles, 1986; Środoń et al., 1986; van der Pluijm et al., 1988; Weaver, 1989; Jiang et al., 1990; Li et al., 1997; Hover et al., 1999; Masuda et al., 2001). The Si:Al ratio in the tetrahedral sheet is consistent with illitic material rather than pure mica (Fig. 7).

In order to compare the chemical composition of I-S from Chochołów-06 (50% illite in I-S) with that of the other samples (all 76% illite in I-S), the Kolmogorov-Smirnov test is employed as there are not enough data to adequately define averages by arithmetic means. Chochołów-06 has a broader range of tetrahedral Si4+ values but includes values which are similar to or lower than those in the more illitic I-S from Chochołów-60 (Fig. 8A). Gulf Coast, onshore southeastern United States, data previously published by Ahn and Peacor (1986) are presented as a reference frame for progressive illitization and demonstrate the utility of the K-S test; however, it should be noted that their samples are not analogs for those from the Podhale Basin.

The octahedral cation chemistry (Figs. 8B and 8C) of Chochołów-06 displays a similar range to that in the more illitic samples from Chochołów-60, albeit with more samples relatively enriched in Fe3+ and Mg2+. Total Al3+ is reflective of tetrahedral Si4+ occupancy and octahedral substitution. The relative difference between sample data is described as a D-statistic in Table 6.

Illite-smectite from Bukowina Tatrzańska–41 has a more homogeneous tetrahedral composition and a much more aluminous octahedral composition than that from Bukowina Tatrzańska–06, which is richer in Fe3+ (calculated Kolmogorov-Smirnov D-statistic of 43%; Fig. 8F). Furthermore, the octahedral composition of I-S in Bukowina Tatrzańska–41 is much more homogeneous than that in Bukowina Tatrzańska–06. The D-statistic shows that all samples analyzed are statistically different in terms of their octahedral cation compositions. The distributions (Fig. 8) and the D-statistics (Table 6) show that Bukowina Tatrzańska–41 has less Fe3+ and Mg2+ and more Al3+ in its octahedral sites than Bukowina Tatrzańska–06.

The composition of I-S reflects both that of the initial detrital supply and also changes resulting from diagenesis, which ultimately transform I-S to illite, and perhaps to chlorite and quartz (Hower et al., 1976; Boles and Franks, 1979; Weaver, 1989; van de Kamp, 2008). From a general chemical perspective, the illitization of smectite results in the export of Fe3+ (or Fe2+) and Mg2+ from I-S to chlorite or perhaps late-diagenetic ankerite, an increase in the concentration of Al3+ within illite, and the formation of quartz as the more siliceous smectite is converted to illite. It is commonly understood that Al3+ is conserved (Land et al., 1997; Land and Milliken, 2000) within a diagenetic system. While these general trends are well documented, they may be masked in a single case by natural variations in the composition of detrital I-S related to provenance changes. A common provenance and well-documented, progressive, and predictable diagenetic trends (Środoń et al., 2006b) make this data set from the Podhale Basin ideal for high-resolution study. Furthermore, illitization of smectite has been documented to play an important role in the development of an oriented alignment of neoformed clay minerals (Day-Stirrat et al., 2008a) in the Podhale Basin.

The chemical and mineralogical data presented here show that in the deeper parts of the basin (5000 m to 7000 m of maximum burial) the rate of smectite illitization has slowed or terminated, such that the percent illite in I-S does not change further (Table 1). Nevertheless, detailed analysis of I-S chemistry (Fig. 8) and morphology (Fig. 6) suggests continued recrystallization with increasing depth, observed as an increase in the size and coherency of I-S crystallite packets. In terms of the use of sonic velocity as a method to estimate porosity and pore pressure, the implication of increased preferred orientation and thicker, more coherent I-S packets is increased velocities at a constant porosity and pore fluid pressure.

Further, with increasing burial depth, K2O in the whole rock (Środoń et al., 2006b) is approximately conserved (Table 1). By assuming that K-feldspar contains 15% K+ in its structural formula (i.e., 0.1 Na per unit formula; Środoń, 2009) and assigning an appropriate percentage of the K2O to K-feldspar, the rest of the K2O can be assigned to K-bearing 2:1 clays (mica, illite, and I-S; Table 1). These calculations show that the K2O content of the 2:1 clay fraction does not evolve down the profile, staying between 5.6% and 7.4%. This consistency, despite a clear smectite illitization trend, can be explained only by a redistribution of K2O within the 2:1 fraction (dissolution of detrital illite and/or mica providing K2O for neoformed illite). Dissolution of illite with increasing burial depth is probably unreasonable as illite would be in equilibrium with smectite illitization.

It is well known that smectite produces concentric ring patterns in SAED associated with turbostratic disorder (Moore and Reynolds, 1997), resulting from the weak mutual attraction between hydrated cations in the interlayer space and adjacent 2:1 layers and the resultant lack of “keying” effects which allows more random layer positioning. X-ray diffraction (XRD) indicates that the most diagenetically immature sample in this study contains randomly interstratified (R0) I-S with 50% illite layers (Środoń et al., 2006b), but only limited turbostratic disorder (Fig. 5). The nature of the smectite and illite interfaces in interstratified mixed-layered crystallite packets can affect SAED patterns (Bell, 1986), as the boundary between smectite and illite layers may be layer terminating, changing the crystal lattice planes on the scale of the electron beam and producing what appears to be small amounts of turbostratic disorder (see Fig. 5). The three more diagenetically mature samples contain R1-ordered I-S (Środoń et al., 2006b) with essentially identical (76%) proportions of illite in I-S; these samples have correspondingly similar SAED patterns which are also similar to those observed in previous studies of similar material (e.g., Ahn and Peacor, 1986; Jiang et al., 1990). However, the ordering suggested by XRD is not entirely matched by the high-resolution TEM observations, which show progressive ordering from Bukowina Tatrzańska–06 to Bukowina Tatrzańska–41 and an increase in crystallite size. This suggests that I-S continues to recrystallize beyond the level implied by XRD data, revealed by TEM because this technique can discern a packet of crystallites within a size fraction, whereas XRD gives the average crystallographic response of all of the crystallites in that fraction. Small differences in SAED patterns of the three more mature samples most probably relate to (1) the coherency of “crystallite packets”, with larger packets producing more clearly identified single-crystal patterns (Ahn and Peacor, 1986; Li et al., 1997), and (2) the octahedral substitution of Fe3+ (or Fe2+) and Mg2+ for Al3+ in the octahedral layer, with Fe3+ being the most significant substitution due to the size of the atom relative to Al3+. The most coherent SAED patterns for I-S are thus seen in Bukowina Tatrzańska–41, which contain, according to lattice fringe images (Fig. 6), the thickest I-S crystallites which also have the least Fe3+ and Mg2+ in the octahedral sheet (Fig. 8).

It has been previously shown that on a 1 mm2 scale, the preferred orientation of I-S crystallites (Fig. 2) in these wells increases substantially during the main phase of illitization (Day-Stirrat et al., 2008a); a similar phenomenon was observed in the Gulf of Mexico by Ho et al. (1999). The change in preferred orientation implies that illitization occurs as a dissolution-reprecipitation reaction and that the neoformed mineral grows perpendicular to maximum effective stress (Day-Stirrat et al., 2008a). In the closed system implied by the whole rock chemistry (constant K2O) and mineralogy of these mudstones (Środoń et al., 2006b), potassium for illite is supplied from K-feldspar, with additional Al3+ and Si4+ from kaolinite. Changes in the proportion of illite in mixed-layer I-S halt at ∼5 km paleo burial (Table 1; Fig. 2). More deeply buried samples, such as Bukowina Tatrzańska–41, which has a maximum burial depth of 7.1 km and a maximum paleotemperature of close to 150 °C (Środoń et al., 2006a), have essentially identical percent illite in I-S. However, in the apparent absence of change in percent illite in I-S, the preferred orientation of I-S in the more deeply buried samples is somewhat higher. These data could imply (1) mechanical rearrangement of phyllosilicates as a result of higher effective stresses; (2) continued recrystallization but with no change in percent illite within I-S; or (3) formation of a variably aligned phyllosilicate fabric during the main phase of illitization, with no further recrystallization during continued burial.

It is highly unlikely that the somewhat enhanced preferred orientation is due to mechanical rearrangement. Firstly, the samples have low porosities, and secondly, mercury injection porosimetry data suggest that most pores are <∼20 nm (Day-Stirrat et al., 2008a). The lack of physical space precludes substantial mechanical reorientation of particles which are larger than the pores. Given that the diagenetic system here appears to be closed, any change in the microfabric is likely to result in a shift of load from matrix grains to pore fluid, a decrease in effective stress, an argument recently discussed by Goulty et al. (2016). Given that the pore volume to matrix volume is weighed heavily in favor of the matrix, any change could have proportionally large effects on pore fluid. Therefore, in high-pressure–high-temperature wells, the loading-unloading behavior (Bowers, 1995), decrease in effective stress, may be extremely complex and potentially differ from illite compaction trends defined for lower effective stresses (Lahann, 2002; Lahann and Swarbrick, 2011).

Unfortunately, our data cannot unequivocally differentiate hypotheses (2) and (3) above. However, lattice fringe images suggest that I-S crystallite packages in Bukowina Tatrzańska–41 are larger than those in Bukowina Tatrzańska–06, implying continuing crystal growth without the destruction of smectite layers in mixed-layer I-S (Fig. 6). We infer that the I-S does not become more illitic due to a lack of supply of K+, or that residual smectite layers are physically occluded from interacting with cations in solution. A clear compositional difference (Fig. 8) between I-S in Bukowina Tatrzańska–06 and that in Bukowina Tatrzańska–41 supports the idea of continued recrystallization. The octahedral occupancy of I-S in Bukowina Tatrzańska–41 is much more aluminous than that of Bukowina Tatrzańska–06, essentially more mica like, and the overall composition of I-S in Bukowina Tatrzańska–41 is much more homogeneous than that in Bukowina Tatrzańska–06. Stated very simply, Bukowina Tatrzańska–41 has a distribution of I-S chemical formulae that are simplified relative to Bukowina Tatrzańska–06, consistent with the progressive conversion of illite toward a mica composition at significantly greater temperatures (van de Kamp, 2008).

The data in this study, plus those from Środoń et al. (2006b) and Day-Stirrat et al. (2008a), indicate continued recrystallization and export of Fe3+ from I-S as, during late diagenesis, it transforms toward a more muscovite-like composition. We propose that this is part of a series of diagenetic steps (Fig. 9) that converts a broadly isotropic fabric inherited as a result of the deposition of clay floccules to the highly aligned fabric observed in low-grade metamorphic pelites (Haines et al., 2009). Reorientation of the clay fabric is restricted during the main stage of mechanical compaction, during which water is expelled, but is enhanced during the main stage of smectite illitization. In this study, at higher levels of diagenesis, we see that while recrystallization of illite continues, there is a limited change in the orientation of the illite fabric. As diagenesis gives way to low-grade metamorphism, there is once again a more striking development of an aligned phyllosilicate fabric, reflecting continuing clay mineral recrystallization and growth (Fig. 9). Abrupt diagenetic steps are not applicable to all major diagenetic reactions. For example, once quartz cementation reaches a kinetically favorable activation energy (with the presence of a clean quartz surface) and temperature, the reaction simply runs until there is no more space for quartz cementation (Taylor et al., 2010).

TEM-EDS data and statistical tests presented here describe a systematic change in I-S chemistry with increased burial temperature beyond the termination of the smectite to illite transformation. TEM-EDS data are accompanied by SAED patterns that show well-defined coherence of layers in I-S packets, single-crystal patterns, and an absence of turbostratic disorder. Concomitant with this change in mineral formulae is an increase in crystallinity observable in lattice fringe images and a change from high-angle contacts between discrete I-S packets to more coalesced crystallites. We propose that the data presented here describe the continued increase in the preferred orientation of I-S beyond the smectite to illite transformation. This involves a change in crystallite chemistry reflected by (1) the progressive removal of Fe3+ and Mg2+ from the octahedral sheet of I-S, (2) a decrease in crystallite layer rotation, and (3) the coalescence of I-S crystallites in a high-effective-stress regime. We suggest that these relatively subtle changes are one of a series of diagenetic steps that convert chemically diverse I-S with a broadly isotropic phyllosilicate fabric into chemically homogeneous illite with a highly anisotropic fabric in low-grade metamorphic rocks. These results have implications for loading-unloading trends in diagenetically mature siliciclastic systems and for predicating porosity and pore pressure from wireline logs.

We thank the UK Natural Environment Research Council (NERC) and BP for supporting RJD-S’s Ph.D. which generated the TEM-EDS data, and the electron microbeam analysis laboratory at the University of Michigan for use of their TEM and sample preparation equipment for lattice fringe imaging. Two anonymous reviewers are thanked for their detailed reviews and Raymond M. Russo, Ph.D., is thanked for his editorial support.