Impact of green clay authigenesis on K–Mg–Fe sequestration in marine settings

Retrograde clay mineral reactions (i.e., reverse weathering), including glauconite formation, are rst-order controls on element (re)cycling vs sequestration in modern and ancient marine sediments. Here, we report substantial K–Mg–Fe sequestration by glauconite formation in shallow marine settings from the Triassic to the Holocene, averaging 4 ± 3 mmol K·cm − ²·kyr − 1 , 4 ± 2 mmol Mg·cm − ²·kyr − 1 and 10 ± 6 mmol Fe·cm − ²·kyr − 1 , which is ~ 2 orders of magnitude higher compared to deep-sea settings. Upscaling of glauconite abundances in shallow-water (< 200 m) environments predicts a global K–Mg–Fe uptake of ~ 0.05–0.06 Tmol K·yr − 1 , ~ 0.04–0.06 Tmol Mg·yr − 1 and ~ 0.11–0.14 Tmol Fe·yr − 1 . We conclude that authigenic clay elemental uptake had a large impact on the global marine K, Mg and Fe cycles throughout Earth`s history, in particular during ‘greenhouse’ periods with sea level highstand. Quantifying authigenic clay formation is key for better understanding past and present geochemical cycling in marine sediments.


Introduction
Chemical elements are supplied to the global ocean by the chemical and physical weathering of carbonate and silicate minerals on the continents, and the subsequent transport of dissolved and particulate matter by rivers, groundwater, glaciers and wind (Raiswell, 2006 The source-sink relations of the global elemental cycles are increasingly well constrained due to advances in e.g., high-precision isotope and element concentration measurements in benthic chambers, novel isotopic tracing methods and isotope-enabled earth system models combined with multivariate statistical modeling ( The long-standing view that clay mineral reactions taking place at low to ambient temperature (≤ 30°C) over much of the Earth`s surface are very slow (~ 10 5 up to a few 10 6 yr) has been increasingly challenged (Michalopoulos and Aller, 1995 settings has been shown to be signi cant, up to six-fold greater than Fe sequestration by pyrite formation in suboxic near-surface sediments (Baldermann et al., 2015), there has been no attempt to estimate the impact of glauconite formation on past and present marine geochemical cycles.
In this study, we ll this gap using a well-characterized glauconite-bearing sequence from Langenstein, in the Northern German basin (Germany), as representative example of Mesozoic and Cenozoic glauconitebearing intervals, to calculate K-Mg-Fe sequestration rates related to glauconite formation in shallowwater settings, and to compare this with published estimates for the deep-sea (Baldermann et al., 2015).
The sedimentary sequence from Langenstein ( Fig. 1) represents an authigenic glauconite deposit on a palaeo-shelf setting (Wilmsen, 2003;Wilmsen et al., 2005;Wilmsen, 2007). Here, shallow-water carbonate or sandstone lithologies contain major glauconite, with overlying shelf sediments hosting relatively smaller quantities of glauconite. K-Ar glauconite dating indicate glauconite formation was completed within < 1 Myr close to the sediment-seawater interface (Baldermann et al., 2017). This study site is representative of many modern and palaeo-shelf settings that accumulate glauconite minerals (Odin and Matter, 1981

Results
Glauconite characterization. XRD analysis identi es the green grains within the sandstone lithology as glauconite with minor admixtures of glauconite-smectite based on broad re ections at 10 Å (001), 5.0 Å (002), 4.5 Å (020), 3.3 Å (003), 2.6 Å (13 ) and 1.51 Å (060,33 ) (Fig. 2a). The glauconite within the carbonate lithologies produces identical XRD patterns, though these rocks contain calcite and quartz impurities. Well de ned re ections at 3.6 Å (11 ) and 3.1 Å (112) and the weak 'XRD hump' between 25 to 40° 2 indicate the green grains are mixtures of the 1 M and 1M d polytype structures, which correspond Chemically, the vast majority (more than 95 % of the EMPA data, cf. Table S1) of the glauconite grains has K 2 O contents exceeding 7 wt.%, frequently reaching up to > 9 wt.%, with averages of 8.3 ± 0.3 wt.%, 9.0 ± 0.2 wt.%, 8.7 ± 0.2 wt.% and 9.2 ± 0.2 wt.% for samples S4-S7, respectively (Fig. 2d), which is characteristic of 'evolved' to 'highly evolved' glauconites (e.g., Odin and Matter, 1981;Banerjee et al., 2016). The total Fe contents (de ned here as TFe; sum of Fe 2 O 3 and FeO) range from 16 to 26 wt.% (av. 21.6 ± 1.9 wt.% for S4, 24.3 ± 1.6 wt.% for S5, 23.0 ± 1.5 wt.% for S6 and 24.2 ± 1.3 wt.% for S7; Fig. 2e  The plot of the chemical composition data in the Al 2 O 3 vs TFe and K 2 O vs SiO 2 diagrams (Fig. 2d,e) indicates glauconite formation progressed through the substitution of Fe 3+ , Fe 2+ and Mg 2+ ions for Al 3+ ions in the octahedral sites and of Al 3+ ions (and eventually Fe 3+ ions) for Si 4+ ions in the tetrahedral sites, which resulted in a negative layer charge that had to be balanced by the uptake of K + ions and minor Na + and Ca 2+ ions in the interlayer sites of the glauconite (Baldermann et al., 2013). Baldermann et al. (2017) showed that the glauconites evolved in organic-rich, semi-con ned micromilieus, such as in faecal pellets and in foraminifera chambers, close to the sediment-seawater interface through the reaction of Fe(III)-smectite precursors with monosilicic acid, goethite (inherited from the sediment) and seawater-derived K + and Mg 2+ ions. As the mixed-layered glauconite-smectite clay transforms into glauconite, Na + , Ca 2+ and H + ions are released from the crystal lattice, as supported by our chemical composition data (Table S1). This mode of glauconite formation is representative of many modern and palaeo-shelf environments ( However, we note that the bulk sedimentation rate was calculated on the basis of estimates of carbonate bio-productivity and that the rate is much lower compared to the distal marine sequences (glauconitefree) of Northern Germany (~ 70 m·Myr − 1 at Wunstorf; Wilmsen, 2007). This is mainly because of low productivity of the carbonate factory and low clastic sedimentation on the palaeo-shelf at Langenstein. The carbon and oxygen isotopic composition (cf. Table S2) of calcite spar within the sandstones and the conglomerate (-7.5 to -0.8‰ of δ 13 C, VPDB, and − 7.5 to -5.4‰ of δ 18 O, VPDB), as well as the positive linear correlation between the δ 13 C and δ 18 O values (R² = 0.995), suggest a continent-derived carbon source (i.e., low δ 13 C values inherited from soil organic matter) and minor diagenetic overprinting (i.e., low δ 18 O values inherited from the interaction with meteoric or burial uids) (Fig. 3c,d). The calcite matrix within the glauconitic sandstone records the transition from continental in uences to progressing marine sedimentation (-6.0 to 0.4‰ of δ 13 C, VPDB, and − 7.3 to -5.0‰ of δ 18 O, VPDB), while the carbonate mud within the Glauconitic Pläner Limestones exhibits isotopic signatures typical for shallow marine carbonate sedimentation (1.0 to 2.1‰ of δ 13 C, VPDB, and − 3.9 to -3.4‰ of δ 18 O, VPDB) (Fig. 3c,d).
Calcite mud precipitation within the glauconite-bearing interval occurred at a temperature of around 26 ± 2°C, which is typical of a 'warm' shelf environment (Banerjee et al., 2020;Gao et al., 2021). Thus, the palaeo-depositional setting at Langenstein is representative of many modern and palaeo-shelf environments.

Discussion
Rate of K-Mg-Fe sequestration by glauconite in marine settings. Reverse weathering reactions to produce authigenic clay minerals in the ocean can signi cantly impact the ratio of element (re)cycling vs element sequestration in marine sediments. Such element sequestration reactions are commonly slower (10 5 to 10 6 yr) compared to clay mineral formation by continental weathering (10 3 to 10 5 yr) or hydrothermal-driven uid-rock interactions at mid-oceanic ridges (10 2 to 10 4 yr). Nevertheless, previous studies have identi ed 'hot spot' areas that favor clay mineral precipitation in marine sediments, such as mangrove forests, deltas and estuaries ( Herein, we focus on constraining the sequestration rates for K and Mg, as these elements are taken up directly from seawater or seawater-derived pore uids by glauconite, and on Fe, because this element is an essential micro-nutrient in the ocean and plays a key role in the sedimentary Fe redox cycle, in particular when the rate of pyrite formation is low (Raiswell, 2006;Baldermann et al., 2013; López-Quirós et al., 2019). We do not consider Na and Ca, as they are barely incorporated in glauconite (Table S1), and also exclude Al and Si from further consideration, as these elements are virtually 'absent' in modern ocean water. This is attributed to the low solubility of most Al-and Si-bearing solids and the strong concentration limitations arising from the activity of silicifying organisms, such as radiolarians, diatoms and siliceous sponges, in particular in the surface oceans ( Assuming (i) a sedimentation rate of 130 cm·Myr − 1 for the M. dixoni Zone (Wilmsen, 2007) and (ii) a density of 2.7 g·cm − ³ for the glauconitized strata (Logvinenko, 1982), and considering (iii) the K-Mg-Fe contents of the glauconites (Table S1) and (iv) the abundance of glauconite across the studied sedimentary sequence (Fig. 3a,b), element-speci c sequestration rates associated with shallow-water glauconite formation can be calculated ( Fig. 3e and Table S3). The sequestration rates range from 0.7 to 42 mmol K·cm − ²·kyr − 1 , 0.4 to 23 mmol Mg·cm − ²·kyr − 1 and 1.1 to 65 mmol Fe·cm − ²·kyr − 1 , re ecting the extremely low to very high abundances (1 vs 70 wt.%) of glauconite in the pro le.
The 'low rate' of element sequestration may re ect reverse weathering processes taking place in depositional systems that are characterized by high sedimentation rates and a short residence time at the sediment-seawater interface. Authigenic clay precipitation is ine cient under such conditions (Amorosi, 2012;Chattoraj et al., 2016) so that element sequestration is reduced. The 'high rate' of element sequestration may represent transgressive systems that are characterized by low to close to zero overall sedimentation and a prolonged residence time at the sediment-seawater interface, which favor glauconite formation ( Although the K-Mg-Fe sequestration rates reported for glauconite formation at Langenstein (shallowwater; this study) and the Ivory Coast (deep-water; Baldermann et al., 2015) may not be directly transferrable to all other marine settings that accumulated glauconite through time and space, (i) the mode of glauconite formation (Fe-smectite-to-glauconite reaction), (ii) the micro-environment (faecal pellets and foraminifera), (iii) the timing (10 5 -10 6 yr), (iv) the composition (Fe-rich, highly evolved vs nascent), (v) the abundance (5-10 wt.% vs 2-3 wt.%) and (vi) the depositional environment (warm shallow shelf vs cool deep-sea; low sedimentation rate) at the two localities are representative of the range expected for many modern and past glauconite-forming environments ( Table S4).
It is evident that glauconite formation signi cantly contributed to K-Mg-Fe sequestration in shallow marine sediments throughout Earth`s history, averaging 4 ± 3 mmol K·cm − ²·kyr − 1 , 4 ± 2 mmol Mg·cm − ²·kyr − 1 and 10 ± 6 mmol Fe·cm − ²·kyr − 1 , respectively. We note that K-Mg-Fe sequestration by glauconite formation also happened in the 'older' sediments of the Archaean, Proterozoic and early Cambrian, but this elemental uptake can be barely quanti ed given that sedimentary archives of this time are scarce and that most of the 'old' glauconites are at least partly altered to illite or chlorite minerals Using average elemental sequestration rates per geological period (Fig. 5a) and corresponding occurrences of glauconite on the shelf (Fig. 4a), as well as calculated low and high estimates of the shallow ocean areas over time ( Fig. 5b; Cao et al., 2017; Scotese and Wright, 2018), we can compute K-Mg-Fe palaeo-uxes (Tmol·yr − 1 ) associated with green clay authigenesis that progressed on the world`s shelf area (de ned here as 0-200 m water depth) over time (Fig. 5c-e; Table S5). Based on comparison with global K-Mg-Fe uxes of the modern and past ocean, we propose that the obtained 'high' K-Mg-Fe palaeo-uxes are overestimated (i.e., the shelf areas reported by Cao et al. (2017) are overestimated) and that the 'low' K-Mg-Fe palaeo-uxes (i.e., the shelf areas reported by Scotese and Wright (2018)) better portray the average elemental burial related to green clay authigenesis per geological period, which averages 0.09 ± 0.12 Tmol K·yr − 1 , 0.08 ± 0.11 Tmol Mg·yr − 1 and 0.20 ± 0.28 Tmol Fe·yr − 1 during the Triassic to Holocene. The ratio of the K-Mg-Fe palaeo-uxes associated with glauconite formation in the past vs modern ocean (Fig. 5d) indicates further that the elemental uxes were much higher from the Jurassic to the Oligocene compared to the modern ocean, averaging a factor of 2.4 ± 3.4, which we attribute to the warm and sea level highstand 'greenhouse' conditions prior to the Eocene-Oligocene transition. The lower K-Mg-Fe palaeo-uxes ever since the Oligocene are caused by the decrease of the shallow-water shelf areas and seawater temperature with the onset of the rst Southern Hemisphere glaciation (~ 34 Myr ago) and then Northern Hemisphere glaciation (~ 5 Myr ago), where glauconite formation is reduced . Although these estimates have a relatively high uncertainty and need to be better constrained in future work, it is evident that green clay authigenesis greatly affected the global marine K, Mg and in particular Fe cycle throughout Earth`s history. conservative estimates, if considering that the abundance of glauconite in shallow marine sediments was much higher in the majority of the Phanerozoic due to sea level lowstands and higher seawater temperatures under warm 'greenhouse' conditions vs modern 'icehouse' conditions that signi cantly reduce the burial uxes linked to green clay authigenesis (Banerjee et al., 2020).
K sequestration by deep-water glauconite formation is barely signi cant and accounts for ~ 0.1 % removal of the total dissolved riverine K in ux (~ 1.51 Tmol·yr − 1 ; Garrels and Mackenzie, 1971) and of the K supplied to the ocean by hydrothermal alteration of the modern oceanic crust (~ 1.51 Tmol·yr − 1 ; Berner and Berner, 2012). Contrary, shallow-water glauconite formation can play an important role in the global K cycle (Fig. 6), as it xes ~ 3-4 % of the total oceanic K inventory that is sourced from riverine and hydrothermal uxes, or ~ 3 % of the K that is removed from the ocean through low-temperature alteration of ocean-oor basalts (~ 1.99 Tmol·yr − 1 ; Sayles, 1979;Sun et al., 2016). Hence, K sequestration by glauconite formation at the shelf is at the same order of magnitude as K sequestration by Fe-illite formation taking place in the mangrove forests worldwide (~ 0.02-0.08 Tmol·yr − 1 ; Cuadros et al., 2016). An exception to this amount of K sequestration occurs when K is available locally from e.g., a K-feldspar substrate altering to glauconite, as reported in Proterozoic oceans (Banerjee et al., 2015).
The same conclusions can be drawn for the marine Mg cycle (Fig. 6): Glauconite formation at the shelf consumes ~ 1 % of the terrestrial Mg ux (~ 5.51 Tmol·yr − 1 ) that is brought to the ocean via continental weathering of Mg-bearing carbonates and silicates (Fig. 6)  In addition, glauconite acts as an important sink for Fe (Fig. 6), with shallow-water glauconite formation accounting for up to ~ 13-29 % removal of the dissolved and particulate riverine ux of highly reactive Fe to the ocean (~ 0.48-0.86 Tmol·yr − 1 and ~ 0.63 Tmol·yr − 1 ; Raiswell, 2006). Although Fe uptake by deepwater glauconite formation is less signi cant (Fig. 6), it is still equivalent to ~ 2 % removal by the total hydrothermal alteration ux (~ 0.25 Tmol·yr − 1 ) or ~ 3 % and ~ 30 % removal by the glacial ux (~ 0.20 Tmol·yr − 1 ) and the atmospheric dust ux (~ 0.02 Tmol·yr − 1 ; Raiswell, 2006). Even though oxidation and scavenging processes are the rst-order controls for the benthic Fe uxes in the ocean (Dale et al., 2015), we conclude that glauconite formation taking place in shallow and deep marine settings is an important but currently overlooked mechanism and elemental sink. These glauconite formation settings can, at least locally, affect the pore water inventory of Fe 2+ , K + and Mg 2+ ions that might otherwise be available for biogeochemical use for the ocean and the marine sediments.
We conclude that fast retrograde clay mineral reactions, which occur to wide extent on the ocean oor, are of great signi cance in the marine K, Mg and in particular Fe cycle and have to be considered in earth system models of the present and past marine element cycles. The elemental burial uxes attributed to green clay authigenesis were signi cantly higher under the sea level highstand and 'greenhouse' conditions in the majority of the Phanerozoic compared to the modern sea level lowstand and icehouse' conditions, which suppress glauconite formation. It is now up to future studies (i) to estimate how K-Mg-Fe sequestration through glauconite formation impacts the isotopic composition of the ocean, the pore water reservoir and the modern marine (deep-sea) sediments, and (ii) to assess the impact of climate change through time on the elemental burial uxes attributed with reverse silicate weathering and green clay authigenesis. Methods X-ray diffraction (XRD) patterns were recorded on powdered bulk rocks using a PANalytical X'Pert PRO diffractometer equipped with a high-speed Scienti c X'Celerator detector and operated at 40 kV and 40 mA (Co-Kα radiation source). The samples were prepared using the top-loading technique (Baldermann et al., 2021). The preparations were examined in the range 4 to 85° 2θ with a step size of 0.008° 2θ and a scan speed of 40 s. Mineral quanti cation was carried out by Rietveld analysis of the XRD patterns using the PANalytical X'Pert HighScore Plus Software and the ICSD database (Baldermann et al., 2021). The analytical error is better than ± 3 wt.% (Baldermann et al., 2013;Ra ei, et al., 2020).
The micro-texture and the chemical composition of the green grains were analyzed on polished thick sections by electron microprobe analyses (EMPA) using a JEOL JXA8530F Plus Hyper Probe at Karl-Franzens-University Graz. Analytical conditions were 15 keV accelerating voltage, 15 nA beam current and defocused beam, ~ 3 µm in size, to avoid mineral damage during the measurements. The chemical data were standardized against a range of natural and synthetic crystals, which included the following elements with their characteristic spectral lines: Al-Kα, Si-Kα and K-Kα (microcline), Mg-Kα and Ca-Kα (augite), Fe-Kα (ilmenite), Na-Kα (tugtupite) and P-Kα (LaPO 4 ). Counting times were set to 10 s on peak and 5 s on background position on each side of the element-speci c peak. Only compositions with an analytical error of less than 7 wt.% off 100 wt.% were taken into further consideration (Table S1). The chemical compositions were corrected for the average Fe(II)/Fe(III) ratio of the green grains reported by Baldermann et al. (2017) based on electron energy-loss spectroscopy (EELS) analyses. Structural formulae were calculated based on 22 negative charges, assuming (i) IV Si 4+ + IV Al 3+ is equal to 4, (ii) Fe 2+/3+ , Mg 2+ , and Al 3 + rest occupy the octahedral sheet, (iii) K + , Na + and Ca 2+ are located within the interlayer sites and (iv) the P 2 O 5 contents belong to apatite impurities and not to glauconite. Furthermore, element distribution maps (Al, Fe, K, Si, Ca, Mg, F, Na, S and P) of 1200 × 1200 pixel resolution were acquired (Fig. S1-3). The analytical conditions were as follows: focused beam, 15 keV accelerating voltage, 20 nA beam current, 3 µm pixel size and a dwell time of 13 ms·step − 1 .
The particle form and the nature of the green grains were determined by transmission electron microscopy (TEM) using a FEI Tecnai F20 instrument operated at an accelerating voltage of 200 kV and tted with a Schottky eld emitter, a Gatan imaging lter and an UltraScan CCD camera. High-resolution TEM lattice fringe images were collected parallel to the (001)-plane of the clay minerals particles. Therefore, a sub-fraction of the Glauconitic Pläner Limestones was treated with 10 % acetic acid for 1 h to dissolve carbonates. The acid-insoluble residue was washed several times with ultrapure water and subsequently the green grains were separated by hand-picking under a binocular microscope.