Acknowledgements. Our thanks go to the Programme Tellus Syster of the French Institut des Sciences de l’Univers (INSU) for funding our work. We thank Monique Gentric for the financial/administrative management, and Marion Delattre & Sandra Ventalon for the technical support of this project, as well as the LOG laboratory and the Department of Earth Sciences of the University of Lille for their support. This paper has been improved thanks to the reviews carried out by Jean-François Deconinck and François Guillot.

1. Introduction

Glauconite belongs to the family of authigenic green clay minerals often grouped under the name glaucony. It is a phyllosilicate close to illite, rich in iron (Fe) and potassium (K; Odin & Matter, 1981; Velde, 2014; Banerjee et al., 2012a, 2012b, 2016a, 2016b; Huggett et al., 2017; Huggett 2021; Bansal et al., 2020, 2022; Roy Choudhury et al. 2021a, 2021b). The work of Odin & Matter (1981) established a link between the chemical composition of glauconite and the duration of its formation during authigenesis. A K₂O content > 8% and a Fe₂O₃ content in the range of 24%-28% would indicate growth lasting between 3.10⁵ years and 1.10⁶ y. These works, often taken up by other authors, have been questioned (e.g., Chafetz & Reid, 2000; El Albani et al., 2005; Baldermann et al., 2013, 2022; Föllmi, 2016; Meunier & El Albani, 2007; Wilmsen & Bansal, 2021) and Tribovillard et al. (2023a) showed that, in some cases, glauconite minerals corresponding to the “highly evolved” stage defined by Odin & Matter (1981) were nevertheless formed during (very) early diagenesis. The present study does not address the issue of exact time requirement for formation of glauconite. See for instance the relation between K₂O concentrations and the age of glauconite in the Gulf of Guinea (Odin, 1988; Tallobre et al., 2019; Giresse, 2022 and references therein). The focus is here set on the relative chronology. For instance, early (or earliest diagenesis) means that the phenomenon occurred while live in-fauna was dwelling the sediment. It is not discussed how long such dwelling lasted (years or thousands of years?); only the successive steps of the diagenetic course are taken into consideration. Regardless of how long it is grown, glauconite is an authigenic mineral and its chemical composition should reflect (at least partly) the conditions of diagenesis. If, in addition, this mineral can form during very early diagenesis, then it may reflect the paleo-environmental conditions prevailing at the sediment-water interface, or immediately below this interface. In other words, glauconite can be considered to be a marker of depositional conditions and subsequent diagenesis episodes. In this perspective, we studied glauconites of different ages (late Jurassic to Cretaceous) present in the Boulonnais area (northernmost France) with an emphasis set on trace elements and, in particular, Rare Earths Elements (REE). One of the aims of the present work is to assess the robustness of the chemical composition of glauconitic grains, used as a tool in paleoenvironmental reconstructions, which is of interest to the community of paleoenvironmentalists and sedimentary geochemists. In particular, we examined the limestone-marl alternations of the uppermost Jurassic (Assises de Croï Formation) to determine whether the carbonate beds vs. marly interbeds were generated under contrasting environmental conditions. Relatively few studies were devoted so far to the REE distribution in glauconites (e.g., Fleet et al., 1980; Stille & Clauer, 1994; Jarrar et al., 2000; Wigley & Compton, 2007; Kechiched et al., 2018; Giresse et al., 2021) and this paper aims to decipher which information can be derived from this geochemical parameter.

2. Material, methods and a reminder of some previous results

In the Boulonnais region, glauconite is commonly observed in the geological formations of the Upper Jurassic and Upper Cretaceous, cropping out between Boulogne-sur-mer and Cap Blanc-Nez, along the Strait of Pas-de-Calais (a.k.a. English Channel; Fig. 1), and some glauconite-rich levels or geological formations have been studied recently (Tribovillard et al., 2021, 2023a, 2023b). Studies on the glauconitic levels at the base of the Cenomanian chalk showed that this mineral was not reworked and had formed in situ thanks to episodes of decreased sedimentation rate in connection with the great transgression of the Late Cretaceous (Amorosi & Centineo, 2000; Tribovillard et al., 2021). By extending their study to the formations of the middle part of the Cretaceous (glauconitic sandstones of the Aptian-Albian and transition layers with the chalk) and those of the Upper Jurassic, Tribovillard et al. (2023a) concluded that the glauconite of the deposits studied had formed in shallow marine environments. Furthermore they showed that oyster patch reefs, frequent in the Jurassic of the Boulonnais, must have been an environment conducive to the synsedimentary formation of glauconite. This work also showed that the richness of glauconite in vanadium (V) and arsenic (As) could be directly linked to the redox conditions that prevailed during authigenesis. Finally, the work by Tribovillard et al. (2023b) focused on the glauconite present in great abundance in the limestone-marl alternations of the upper Tithonian Assises de Croï Formation (Fm). The limestone beds resulted from the coalescence of limestone nodules formed during early diagenesis. This study shows that the glauconite, trapped within the nodules and beds, was even earlier than the formation of the nodules, in the course of the diagenetic stages. Based on the three studies mentioned above, we intend to know: 1) whether the glauconites that formed in different environments and at different ages show a contrasting REE signature, and 2) whether the REE distribution of such an authigenic mineral could be a useful tool for paleo-environmental reconstructions, when examined in conjunction with redox-proxying trace metals.

The samples examined in the present study are those presented in the works mentioned above (Tribovillard et al., 2021, 2023a, 2023b; plus four samples of chalk: Craie1-4); their locations are indicated in Table 1. The analytical data used here (gathered in Table 2) come from these papers (samples ONM1-4 from Tribovillard et al., 2021; samples of the Assises de Croï Fm from Tribovillard et al., 2023b; the other samples from Tribovillard et al., 2023a). The REE data of the samples of the Assises de Croï Fm examined in the present study were unpublished yet (Table 2). All the samples analyzed but four are populations of glauconite grains extracted from the rocks as explained below. Four bulk-rock samples of glauconitic chalk were added to complement our understanding of the geochemistry of the glauconite of the chalk. These samples are labeled craie1 to craie4 ; they correspond to the four facies studied by Tribovillard et al. (2021).

The chemical analyzes were carried out by the SARM-CRPG facility of the CNRS in Nancy on glauconite grains extracted from the rocks. This glauconite was isolated at Lille University after acid treatment (HCl) and magnetic separation with an isodynamic Frantz apparatus, according to the protocol described in Tribovillard et al. (2023a). The chemical analyzes were carried out by ICP-OES and ICP-MS according to the protocols described in Carignan et al. (2004). REEs have been normalized using the Post-Archean Australian Shale or PAAS (Taylor & McLennan, 1985). The anomalies in cerium (Ce) and europium (Eu) were calculated according to Rollinson and Pease (2021) and the ratios La/Lu_N, La/Sm_N and Gd/Yb_N (the subscript _N stands for normalized) were calculated to highlight possible enrichments in light, medium or heavy REE (Rollinson & Pease, 2021). As regards the determination of glauconite, phase identification, based on peak positions (d-spacing) and relative intensities, has been performed on X-ray Diffraction patterns via Rietveld refinement on disoriented powder (X’Pert high score Plus software – PDF4 database). This method was used in each of our previous studies (Tribovillard et al., 2021, 2023a, 2023b). Lastly, in an attempt to observe the core of the glauconite grains or a possible zoning, handpicked grains of glauconite were embedded in resin and a thin section was cut for SEM imaging and photonic microscope observation.

Before exposing the results for the present study, some results mentioned in our previous papers cited above must be reminded here, complemented with new illustrations. Glauconite is the only green mineral identified in the Mesozoic rocks studied in the Boulonnais. All XRD analyses yielded the same diffractograms, examplified in Fig. 2. Some previous observations performed on the grains may be summarized as follows. The glauconite grains show a homogeneous dark color, whatever formation they come from (Fig. 3). They usually show cracks, discrete or not (Fig. 4A-D; Tribovillard et al., 2023b). Last, high contents in iron, aluminum and potassium are the usual geochemical markers of highly evolved glauconite. See also the discussion herein below about the correlations between K₂O and Rb, and between Sm and Nd, also confirming the high maturity.

3. Results

3.1. Precursors of the glauconite grains and degree of maturity

Observations of glauconite grains using light or electronic microscopes did not allow deciphering whether they could originate from foraminifer infilling or fecal pellets (Fig. 4 A-D). Colpaert et al. (2021), in their stratigraphic study based on foraminifers, did not mention the presence of glauconite-infilled foraminifers but their plate 2 shows the presence of some. In addition, regarding the chemical composition of glauconite, an important aspect is the nature and composition of the minerals acting as precursor phases of the authigenic growth. This consideration motivated thin sections to be cut through glauconite grains. Whatever the samples they came from, all the grains examined here through photonic microscopes and SEM imaging yielded homogeneous inner structure and composition (Fig. 5A, B): no possible precursor of glauconite, no zoning, but only homogeneous areas are observed (Fig. 5B). In addition to their composition in potassium and iron mentioned above, the grains exhibit a number of features indicating a high degree of maturity. Indeed, XRD always yields glauconite diffractograms with well-defined peaks, to be related to a high crystallinity. Through SEM imaging, some grains showed rosette textures on the surface (Fig. 4E). Rosette-like structures are a marker of maturity according to Lopez-Quiros et al. (2019), Banerjee et al. (2012b) or Roy Choudhury et al. (2021b, 2023). A dark-green color together with the presence of cracks is commonly associated to a high degree of maturity (Giresse et al., 2021; Roy Choudhury et al., 2021b, 2023).

3.2. Distribution of the samples in two sets

In what follows, the analytical results are illustrated and interpreted in two separate sample sets. The set 1 groups glauconites of various ages and depositional environments (#1, 2, 4 and 6 of Fig. 1 and Table 1). They have been examined by Tribovillard et al. (2021, 2023a). The geological formations where these samples come from are summarized in Table 1. They span a facies range from oyster reefs to chalk via silty marls and green sands, and an age range from the Kimmeridgian/Tithonian to the Cenomanian. The deepest depositional settings correspond to the Cenomanian chalk, the other ones range between shallow shoreface and (lower?) offshore, as detailed in Tribovillard et al. (2021, 2023a). The set 2 groups the samples of limestone beds and marly beds from the Assises de Croï Fm (Tribovillard et al., 2023b; #3 of Fig. 1). The Assises de Croï Fm is dated as upper Tithonian (Albani, Glaucolithus and Okusensis ammonite zones; Townson & Wimbledon, 1979; Tribovillard et al., 2023b and references therein). With a thickness of ca. 10 m, the Assises de Croï Fm yields nodular, glauconitic limestones alternating with silty, glauconitic marls (Fig. 6). This second sample set therefore contains glauconite grain populations of the same age and deposited under the same energy and depth conditions, they differ only in their host-rock facies (limestone versus marl), expressing contrasted diagenetic processes. Separating the samples into two sets allows the interpretations to be drawn at contrasting scales: the set 1 will permit interpretations about glauconite populations spanning various depositional settings and ages, whereas the set 2 will allow the focus to be set upon small variations at closer scale.

3.2.1. Sample set 1

The K₂O contents of the glauconite samples are relatively high, distributed between 2% and 9%, most often above 6% (Fig. 7). Only the chalk samples Craie1-4 rank below 6%. This figure illustrates that most of the samples may be considered to be highly mature, according to usual nomenclature (K₂O above 8%). The K₂O contents correlate remarkably well with Rb contents (R² = 0.97; Fig. 8A). In other words, the excellent correlation between K₂O and Rb is not altered by the contrasting ages and depositional settings of the samples analyzed (see below). As regards REE, no striking point can be put forward. If we consider all the REEs, we see that the parameter ∑REE brings out from the set of samples the glauconites contained in four chalk samples (Craie 1-4) as well as the two condensed levels rich in phosphate shells (P1 & P2) and one of the four patch reefs studied (PR3; Fig. 8B shaded area). All other samples show comparatively much lower ∑REE. The patterns of REE established from glauconite grains are grouped in Figure 9A. No marked tendency stands out. Anomalies in Ce and Eu are not very marked; those in Eu are usually weakly positive, those in Ce are weakly positive or negative. The (La/Lu)_N ratio shows a fairly wide range of values, the values of the (La/Sm)_N ratio are more narrowed and a wider range of values is observed for the (Gd/Yb)_N ratio. Noteworthy, the contents (normalized or not) of Nd and Sm are closely correlated, whatever the origin of the samples (age, depositional setting). The correlation linking Sm and Nd is classical but, here, it is unaffected by the depositional or diagenetic conditions (Fig. 8C). As regards redox-sensitive trace elements, the glauconite samples of the sample set 1 show relatively high contents in vanadium (V) and chromium (Cr), these elements yielding some positive correlation (Fig. 8D). In the same way, although germanium (Ge) concentrations do not exceed 10 ppm, they generally show a fair correlation with V concentrations (Fig. 8E).

3.2.2. Sample set 2.

As for the sample set 1, the glauconites of the sample set 2 show relatively high concentrations in V and Cr, with the respective concentrations of the two elements being correlated. With respect to Ge, the glauconites of the set 2 plot in the V vs. Ge diagram of Figure 8E consistently with the ones of the set 1. The most striking correlations are observed between Nd and Sm from one part, and K₂O and Rb from the other part (Fig. 8A, C). In addition, a negative correlation is observed between Al, from one side, and K₂O or Fe₂O₃ from the other side (Fig. 10), while a positive correlation is seen between Al vs. SiO₂ and TiO₂ (Fig. 10A, B). It indicates that iron and potassium, two key elements with regard to glauconite growth, are not simply inherited from the precursor aluminosilicate nucleus the authigenic growth started on. Therefore, part of the content in both metals must have been captured from seawater during authigenesis. As for the glauconite of Set 1, no obvious relation is observed between K₂O and ∑REE (Fig. 8B). Figure 11A, B allows to compare the respective REE patterns of the carbonate and marly beds of the Assises de Croï Fm. The most visible features do not concern the carbonate vs. marly beds comparison, but instead three samples (E, F, and G). differ from their companion samples, either for carbonate vs. marly beds (Fig. 11A, B). In other words, differences are more stratigraphic than lithologic. The redox-sensitive trace metals considered here do not yield any observable trend, with neither contrast between carbonate vs. marly beds, nor stratigraphic distributions.

4. Interpretations

4.1. REE distribution and signs of maturity

Previous works (Fleet et al., 1980 and references therein; Stille & Clauer, 1994; Kechiched et al., 2018) have shown that authigenic minerals exhibit relatively complex partitioning with respect to seawater, regarding REEs. Using selective chemical extractions, Zhang and Shield (2023) confirmed the mobility of REE during early diagenesis in marine sediments. It is established that seawater has a deficit in Ce for redox reasons and the minerals formed in the ocean environment record this deficit (e.g., Piper, 1974; Murray et al., 1990; Sholkovitz & Schneider, 1991; Wilde et al., 1996; Tostevin et al., 2016). However, Fleet et al. (1980) report that this deficit does not appear in authigenic minerals formed in shallow marine environments. Glauconites from shallow seas sometimes even show weakly positive anomalies, which are attributed to the fact that glauconite preferentially incorporates Ce from seawater (sort of “cerium pump”). Similarly, seawater is relatively enriched in heavy REEs, due to the propensity of these REEs to form soluble complexes; however, this enrichment is not observed in authigenic minerals from shallow marine environments (Fleet et al., 1980; Deng et al., 2017).

The potassium content (K₂O) is a classic marker of glauconite maturity used by many authors following the pioneering work of Odin & Matter (1981). A K₂O content above 8 wt% together with a Fe₂O₃ concentration of 24%—28% is the signature of the highly evolved glauconite, the formation of which lasting over a few hundred of kyrs, or even about one million yrs. Here, the K₂O and Fe₂O₃ concentrations of the samples studied correspond predominantly to highly mature glauconite. It may be stressed that some glauconite grains of the patch reefs (samples PR) show rosette-like structures (Fig. 4E). Such structures correspond to evolved but not highly evolved stages of glauconite development, according to Banerjee et al. (2019), Oze et al. (2019), Roy Choudhury et al. (2021b) and Singh et al. (2023). The rosette-showing grains have, however, high concentrations in iron and potassium. This discrepancy could be accounted for by the fact that these samples come from oyster patch reefs that worked as a sort of glauconite factory (Tribovillard et al., 2023a). The patch reefs developed at, or close to, cold-seep vents (Hatem et al., 2016). Cold seep fluids may be sources of dissolved iron and potassium (Suess, 2014; Olu et al., 1996; Wang et al., 2017, Zhang et al., 2022). Therefore, as the glauconites of the samples PR grew close to sources of Fe and K, they might have been enriched in both metals even if the rosette structures suggest a lesser degree of maturity. Last, Banerjee et al. (2019) and Oze et al. (2019) noted that such structures may be linked to infilling (e.g., of foram chambers); however the morphologies of the grains observed in the present study did not allow such a conclusion (which does not rule out the infilling hypothesis).

Tribovillard et al. (2023b) identified precocious glauconite, that formed prior to the hosting, early-diagenetic limestone beds, although showing the geochemical characteristics of evolved glauconite. In other words, glauconite could form during earliest diagenesis in a shallow depositional setting where sedimentation rates were substantial, and yet it bears all the physical and geochemical characteristics of a glauconite supposedly grown during protracted durations. This paradox evidences that the K₂O content cannot be considered any longer to be automatically a reliable tracer of the duration of formation for glauconite (see also Chafetz & Reid, 2000). Therefore, it may be of interest to know whether REEs can be an alternative marker for this process. To examine the trace-element content as a marker of diagenetic conditions, the question about the nature of the precursor phase must be addressed: the K₂O content of a glauconite may be impacted if the precursor is a K-feldspar. In such a case, using the K₂O concentration as a maturity proxy may be pointless. From the X-ray diffractograms yielding pristine glauconite, the mineralogical nature of the precursors could not be inferred. Examining the inner structure and composition using SEM did not help either in the determination of the precursor phases. Finally, in the present case, all the glauconite grain samples studied except for the 4 bulk-rock chalk samples Craie1 to Craie4) show a high K₂O content (Fig. 7). However, it is highly unlikely that all the remaining samples all had a K-rich precursor. Therefore, we are unable to discuss the role possibly played by the precursor phases in the composition of the studied glauconite samples but our results suggest that their K richness cannot be ascribed to a single type of precursor; consequently, it will be considered that the K₂O content can be used as a marker of the degree of evolution of glauconite. Accordingly, the possible influence of the precursors upon the trace element composition cannot be meaningfully discussed.

Regarding the sample set 1, the ∑REE factor is not correlated with the K₂O concentration, but it is observed that the samples which are the richest in phosphate nodules are also those showing the highest ∑REE (7 samples: 4 chalk samples Craie1 to 4, the P1 & P2 levels of the Bancs Jumeaux Formation and one patch reef PR3; Fig. 8B, shaded area). Noteworthily, the concentrations in P₂O₅ and ∑REE are not correlated, and yet the usual presence of REE in phosphorites is well documented in the literature (Kechiched et al., 2018 and references therein). Regarding the seven samples mentioned above, the chalk samples come from the Cenomanian levels which underwent a very pronounced rise in sea level accompanied by a strong decrease in the rate of sedimentation (Amorosi & Centineo, 2000; Amédro & Robaszynski, 1999). Similarly, levels P1 & P2 correspond to prolonged gaps of sedimentation (Geyssant et al., 1993; Deconinck et al., 1996; Deconinck and Baudin, 2008) of a duration estimated to be in the order of one or two ammonite zones (Hudlestoni and Pectinatus). Thus, these six samples correspond to the most condensed episodes of sedimentation among all the cases studied in this work; it is therefore proposed that the sedimentation slowdowns are associated with an increase in the REE content of authigenic glauconite, in relation to the prolonged exchange times between the seawater and the glauconite grains being growing authigenically. Beyond the ∑REE parameters, are specific distributions of REEs observable in the patterns? Three of the four chalk samples show a slight negative anomaly in Ce (samples Craie1—3, Fig. 9), which could suggest that the glauconites had more time to incorporate Ce from the seawater. Nevertheless, the same figure illustrates that this anomaly is not observed for the fourth chalk sample and no heavy-REE enrichment is recorded in any of the four samples. The same goes for P1 & P2, showing neither Ce anomaly, nor enrichment in heavy REE. Thus, the glauconites that formed under conditions of low sedimentation rate did not play any role of “sink” for Ce and heavy REE. The other samples show hardly any interpretable trends. A slight enrichment in intermediate (medium) REE is observed for a certain number of samples (convex-up shape from La to Tb, Fig. 9). Such enrichment has already been observed by Fleet et al. (1980) and seems to be common for glauconite, without the authors offering any interpretation.

In short, on the basis of the first set of samples examined here, illustrating different depositional conditions, varying sedimentation rates and contrasted ages, it appears that the REE patterns of the glauconites do not allow for the identification of interpretable trends. At best, it is observed that the total REE content (∑REE) was higher when sedimentation rates were lower. This absence of interpretable trends in the distribution of REEs may be attributable to sampling heterogeneity. To compare samples of ages that are too far apart, or deposited in contexts differing from each other by too many parameters might not be relevant. To test this hypothesis, a second set of samples was taken into consideration. It is restricted to samples from the same formation (the Assises de Croï Fm) and therefore of approximately the same age and deposited under similar conditions (upper offshore). In addition, these similar conditions suggest that the regional-scale seawater composition may have been stable during the entire deposition of the alternating facies. However, regarding most specifically the sample set 2, the results illustrated with Figs. 8 and 11 do not allow any interpretation to be drawn, as far as REE and ∑REE are considered. No tendency could be observed, considering the samples of the carbonate or marly beds, or mixing all the samples. REE examination does not yield any significant support.

4.2. K₂O vs. Rb, Sm vs. Nd

For all the samples of glauconite studied (sample sets 1 & 2), the chemical analyses make it possible to draw remarkable correlations between the contents of K₂O and those of rubidium (Rb) on the one hand, and between the contents of samarium (Sm) and those of neodymium (Nd), on the other hand (Fig. 8). Regarding most specifically the sample set 2, the concentrations in K₂O and Rb on the one hand, and Nd and Sm on the other hand, plot at a consistent location in a K₂O vs. Rb diagram or Nd vs. Sm diagram, respectively: expectedly, the samples of Set 2 are grouped in these diagrams, in agreement with their similar origin (close in time and space), compared to the samples of Set 1. Substitution of K by Rb is common in glauconite as in many other minerals (e.g., Wiewiora & Lacka, 1985, and, more recently, Rafiei et al., 2023). The Sm-Nd couple is mainly studied for the purposes of isotopic dating but the concentrations of these two elements have hardly been examined as far as we know (Rafiei et al., 2023). The striking feature of our results is that in the two diagrams K vs. Rb and Sm vs. Nd (Fig. 8), the points align almost perfectly while the glauconites analyzed are quite heterogeneous with regard to both their age (Tithonian, Aptian-Albian, Cenomanian) and their environments of formation. Thus, despite these disparities, K and Rb on the one hand, and Sm and Nd on the other hand, were incorporated in the same proportions in the authigenically growing glauconite. This suggests that differences in the chemical composition of the seawater as well as that of the sediments did not impact the crystal chemistry of glauconite: diadochic substitutions thus seem to have obeyed very strict rules. This is interesting in as much as it is thus suggested that glauconite is not a mineral with loose specifications depending on the ambiance it grows in during early diagenesis: whatever the duration of authigenic growth and the variations in (interstitial) seawater composition that may have occurred between the Tithonian and the Cenomanian in the Boulonnais, glauconite yields constant proportions between K and Rb, and Sm and Nd. Several explanations may be put forward. The equilibrium between K₂O and Rb, and between Sm and Nd, may be interpreted as the indicator of high maturity, from a crystallochemical point of view, which demands time. To illustrate this statement, the strontium abundance of carbonates may be considered. Fast forming aragonite or calcite incorporate Sr in important and variable proportions, whereas slow-forming or recrystallized calcite has Sr low contents that fall into rather narrow concentration ranges (Morse & Mackenzie, 1990). In other respects, the outstanding correlations shows that neither ‘fractionation’ nor sort of ‘Rayleigh distillation’ occurred between the two pairs of elements. It is inferred that the glauconite growth took place in close contact with an ‘infinite reservoir’ of dissolved elements (Rb, K, Nd, Sm), that is, the open seawater reservoir. With the proximal depositional settings described here, where sedimentation rates must have been relatively high, unhindered contacts with marine water masses imply that authigenic growth must have been (relatively) rapid. However, ’earliest diagenesis’ does not necessarily implies ‘shortest duration’: the glauconite at stake here may well have formed during the earliest stages of diagenesis but the process may have lasted quite long (J.-F. Deconinck mentions evidences for some depositional gaps within the Assises de Croï Formation; personal communication 2024) . The words early or late diagenesis express here a chronological succession with no precise quantification of duration. Although no quantitative data regarding the duration of glauconite growth are available here, it may be stated that, from a qualitative point of view, the depositional settings discussed here were shallow (at most lower offshore) and, thus, the sedimentation rate was probably not null. Consequently, even if long periods of growth might be envisioned in such shelfal settings, these durations are not likely to be extreme. A way to reconcile the need for a long time of growth accompanying episodes of condensed sedimentation (van Houten & Purucker, 1985; Föllmi, 2016) and proximal settings with possible relatively high sedimentation rates would involve the role played by winnowing processes (Giresse & Wiewiora, 2001; Föllmi, 2016). Proximal settings may witness episodes of condensation prone to formation of glauconite, and episodes of deposition, with glauconite being concentrated through winnowing.

4.3. Redox proxies of the two sample sets

The glauconites of the two sample sets show rather high concentrations in both V and Cr (both being correlated) and a correlation between V and Ge. Vanadium is a long-known redox-sensitive element that has been intensively studied (Breit & Wanty, 1991; Wanty & Goldhaber, 1992; Tribovillard et al., 2006; Meunier, 1994; Huang et al., 2015; Gustafsson, 2019; Haase et al., 2024). Vanadium is also known to be incorporated into phyllosilicates during diagenesis in a reduced stage (Huang et al., 2015; Gustafsson, 2019). Therefore, as formation of glauconite has been proved to take place during earliest diagenesis in the Assises de Croï Fm, V concentrations must be a reliable marker of the redox conditions developing at the sediment-water interface (or close to it), during the initial stages of diagenesis/authigenesis.

Germanium (Ge) has been proven to be redox sensitive in sedimentary environments (Tribovillard et al., 2011, 2021), during its substitution to Si in silicates (e.g., Rouxel et al., 2006; Tribovillard, 2013). Therefore, the correlation between V and Ge in glauconites is quite logical in the present case. Chromium is also reputed to be redox sensitive (to some extent) and to be brought to the sediments together with organic matter (Tuttle et al., 2009; Semeniuk et al., 2016). Bruggmann et al. (2023) recently insisted on the importance of the role played by organic matter in the enrichment of the sediment in Cr. In the present case, organic matter is lowly abundant in the deposits of the Assises de Croï Fm (TOC < 0.5%; Tribovillard et al., 2023b). In addition, Rock Eval pyrolysis showed that the organic matter is of Type III, that is, most probably issued from terrestrial organic products or degraded marine organic matter. Thus, the richness in V, the paucity of organic matter, the absence of well-preserved marine organic matter strongly suggest that bottom conditions were reducing and marine productivity was modest. With such reducing redox conditions, high productivity would have expectedly induced significantly higher TOC values and the presence of Type-II organic matter, which is not the case. Consequently, under depositional conditions characterized by reducing conditions and low organic supply, the enrichment in Cr and the correlation between V and Cr may be ascribed to the sole redox conditions.

4.4. Grain size of the glauconites of the sample set 2 (Assises de Croï Fm)

As mentioned in previous works, the HCl-leached fraction of the rock samples contains quasi-exclusively quartz and glauconite (Tribovillard et al., 2023a, 2023b). These two mineral species were efficiently separated using a magnetic separator and could be analyzed separately. The grain size analyzes are presented and discussed in Tribovillard et al. (2023b). The glauconite grains of the alternating carbonate and marly beds of the Assises de Croï Fm have been reported to have grown during earliest diagenesis, prior to the diagenetic precipitation of the carbonate beds (Tribovillard et al., 2023b). The results of importance for the present paper are gathered in Fig. 12 showing curves of grain size distribution for quartz and glauconite, whether they come from carbonate beds or from marly ones in the Assises de Croï Fm. The populations of quartz grains show a well-sorted, unimodal size distribution, whereas the populations of glauconite grains are poorly sorted. The size distribution curves of the populations of quartz grains yield peaks between 100 µm and 200 µm for both bed and interbed facies. In contrast, the modes of the glauconite curves correspond to larger particule size in the marly beds compared to the carbonate beds. For the bed-hosted glauconites, all the curves but one are superimposed, with a main mode centered around 25 µm, whereas the curves of the interbed-hosted glauconites show modes dispersed between 30 µm and 100 µm. Thus, the glauconite particles of the marly beds are larger and more heterogeneous than those of the carbonate bed (they do not overlap so well). In contrast, the size distribution curves of the populations of quartz grains show similar patterns for the carbonate vs. marly beds, thus indicating that the level of hydrodynamic energy was similar for both environments of deposition. Thus, the discrepancy in the distribution of the glauconite grain sizes cannot be explained by contrasted hydrodynamics but it can be accounted for by contrasted duration of authigenic growth. The glauconite in the case of marly beds is larger and more homogeneous in size, as if a prolonged growth of the authigenic grains would have tended towards a maximum terminal size over time (a sort of end member). At the end of the growth process, the authigenic grains would fall into a relatively narrow range of particle size value. Contrarily, inside the carbonate beds, the glauconite is smaller and less homogeneous in size. Glauconite growth preceded that of the carbonate beds, but the precipitation of carbonates isolated the glauconite grains from the marine environment as they were growing. Being isolated from their aqueous environment, the grains could not continue their growth and did not reach the same size limit as in the case of the grains of the marly beds, i.e., the size of syngenetic glauconites would reflect the duration of their growth. This hypothesis is supported by the geochemical argument discussed above: the glauconites of the depositional milieus characterized by low sedimentation rates are the richest in REE (highest ∑REE).

4.5. Implications for the causes of the limestone-marl alternation of the Assises de Croï Fm

The two facies of this formation are strikingly contrasted upon field observation; however, they cannot be distinguished using geochemical criteria (trace metals and REE) for glauconite examination. In other words, the glauconite witnessed what was going on during earliest diagenesis but did not record any difference or variation in the redox status of the sediment-water interface. This observation is in contrast with what Bellanca et al. (1997) observed, examining the bulk-rock REE distribution in limestone-marlstone couplets from the hemipelagic Cretaceous of Venetia. These authors carried out the sole study available in the literature (to the best of our knowledge; see also a related study by Su et al., 2020), comparing REE distribution in limestone beds vs. marly beds from limestone-marl alternations. They concluded that REE were impacted by redox conditions but they did not study specifically the authigenic phyllosilicate fractions. Their results encompass what was occurring in the entire water column and not specifically what was going on at the very sediment-water interface as is done in the present work.

The grain size distributions of quartz populations of both facies do not yield any difference, indicating that the levels of energy were similar during particle settling. Consequently, the diagenetic precipitation of limestone nodules and beds, which happened after sediment accumulation and glauconite growth, was not conditioned by the grain size of the detrital supply nor by redox conditions. The carbonate fraction of the beds bears the imprint of organic carbon in that the stable isotope composition of C shows negative values, differing from seawater signature of Late Jurassic seas and oceans (Tribovillard et al., 2023b). The nodules and beds growth incorporated “light” carbon released by organic-matter remineralization, circulating per ascensum. Why the carbonate precipitation happened at specific levels within the sediments, forming continuous beds, is not evident. Nevertheless, it was not induced by the grain size distribution of the clastic particles. The role of permeability barriers opposing to fluid circulation and inducing carbonate saturation and precipitation can thus be ruled out. The carbonate supersaturation must be related to the presence of organic matter or methane, and biogenic bicarbonate ions inducing/fostering diagenetic precipitation, ending with limestone beds being aligned parallel to the sea bottom and bedding. See for instance the recent paper of Akam et al. (2023) and references therein. Such a stratal layout of authigenic carbonate beds and coalescent nodules was examined with the parallel two (occasionally three) beds of the Tithonian Bancs Jumeaux Fm of the Boulonnais (Table 1; Tribovillard et al., 2012). The explanations involved episodic spills of fluids onto the sea bottom, released by cold seepage sites. Our data do not allow this scenario to be ascertained in the present case but (at least some) alternative scenarios can be ruled out.

5. Conclusions

This work brings new elements to our understanding the complex mineral glauconite. First, the glauconites studied here show morphological and geochemical indications of maturity, according to the classical criteria, which suggests a protracted period of growth. Besides, over a range of age spanning the Late Jurassic and mid- and Late Cretaceous, and over a range of shelfal environments, the glauconite studied all show narrowly identical values for the Rb:K₂O and Sm:Nd ratios, respectively. Such constant ratios also suggest that a crystal chemical equilibrium had enough time to establish a balance, and that the authigenic growth was operating in an infinite reservoir, that is, close to the sediment-water interface. Long time, proximity to the sediment-water interface: both facts suggest slow sedimentation rates. However, in other respects, the glauconite grains of the Assises de Croï Fm terminated their growth during early diagenesis. Thus, it is concluded that glauconite can gain the signs of mineralogical maturity during early diagenesis. Early diagenesis is not incompatible with long time; nevertheless, all the glauconites at stake here formed in shelfal environments, that is, in places where sedimentation rates may hardly be close to zero during protracted periods of time. Intermittent sedimentation over time with repeated episodes of winnowing may explain a (relatively) long time for the formation of glauconite on the one hand, and an agitated sedimentary environment with terrigenic inputs, on the other hand. Anyway, as we have no means to assess the quantitative duration of the growth of glauconite, we will only refer to early diagenesis, regardless to its duration. Second, no robust information could be derived from the REE distributions but one: the total amount of REE ( ∑REE) could echo the duration of formation of glauconite. Therefore, our results reinforce the view that REE of authigenic minerals are of little help to decipher (past) environmental conditions of shallow marine milieus. Third, the limestone-marl alternation of the Assises de Croï Fm allows early diagenetic glauconite to be used as a witness of the conditions developing at or below the sediment-water interface. The facies alternation has no relation to see with alternating redox conditions affecting the sediment-water interface, nor with the hydrodynamic energy during particle settling. The reason(s) why diagenetic carbonate beds did form with a stratiform geometry remain a bit unclear but, at least, our results allow excluding some of the potential factors. Lastly, the grain size distribution of the glauconite grain populations is by far the clearest parameter allowing glauconite to be distinguished between diagenetic carbonate beds and marly interbeds. This study illustrates that the glauconite growth was hampered by the diagenetic formation of the carbonate beds that caged glauconite grains, thus preventing any further growth. Oppositely, glauconite growth was not hampered in the marly interbeds and the grains tended to an end-member size that could not be exceeded.