Assessment of Chalk as an Archive for the Lithium Isotope Composition of Seawater

The understanding of silicate weathering and its role as a sink for atmospheric CO2 is important to get a better insight into how the Earth shifts from warm to cool climates. The lithium isotope composition (δ7Li) of marine carbonates can be used as a proxy to track the past chemical weathering of silicates. A high‐resolution δ7Li record would be helpful to evaluate the role of silicate weathering during the late Cretaceous climate cooling. Here, we assess chalk as a potential archive for reconstructing Late Cretaceous seawater Li isotope composition by comparing Maastrichtian chalk from Northern Germany (Hemmoor, Kronsmoor) to a Quaternary coccolith ooze from the Manihiki Plateau (Pacific Ocean) as a lithological analog to modern conditions. We observe a negative offset of 3.9 ± 0.6‰ for the coccolith ooze relative to the modern seawater Li isotope composition (+31.1 ± 0.3‰; 2SE; n = 54), a value that falls in the range of published offsets for modern core‐top samples and for brachiopod calcite. Further, the negative offset between the Li isotope compositions of Manihiki coccolith ooze and modern planktonic foraminifera is 2.3 ± 0.6‰. Although chalk represents a diagenetically altered modification of pelagic nannofossil ooze, manifested by changes in the composition of trace elements, we observe a consistent offset of Li isotope data between Maastrichtian chalk and Maastrichtian planktonic foraminiferal data (−1.4 ± 0. 5‰) that lies within the uncertainty of modern values. We therefore suggest that chalk can be used as a reliable archive for δ7Li reconstructions.


Introduction
Past warm periods are suitable for studying the mechanisms and processes of the Earth's thermostat.One important mechanism to withdraw CO 2 from the atmosphere on geological timescales is the chemical weathering of silicates (Berner et al., 1983).The climate of the Late Cretaceous is characterized by an 8-10°C decline in global mean temperatures that terminated the global hothouse condition of mid-Cretaceous times (Friedrich et al., 2012;B. T. Huber et al., 2018;Linnert et al., 2014).The causal mechanisms of this cooling are not well understood (Friedrich et al., 2012;B. T. Huber et al., 2018;Linnert et al., 2014), although continental silicate weathering or low-latitude ophiolite weathering may have played a critical role (Chenot et al., 2018;Corentin et al., 2023;Jagoutz et al., 2016).The role of silicate weathering as a thermostat and sink for atmospheric CO 2 during the Late Cretaceous has not been sufficiently investigated so far.
Leaching experiments with silicates show no significant Li isotope fractionation during rock dissolution (Penniston-Dorland et al., 2017;Pistiner & Henderson, 2003;Wimpenny et al., 2010).The wide range of δ 7 Li values in river waters is caused by the clay mineral formation during continental weathering, which is the most prominent process for Li isotope fractionation (e.g., Dellinger et al., 2015;Pogge von Strandmann & Henderson, 2015).Clay minerals are a sink for Li whereby the light Li isotope ( 6 Li) is preferentially incorporated, driving river water to isotopically heavier compositions, which is a typical signal of incongruent weathering (e.g., Bouchez et al., 2013;Dellinger et al., 2015;Pogge von Strandmann & Henderson, 2015).If no or only little secondary clay mineral formation occurs, the Li isotope composition of the river water reflects the composition of the source rock (lower δ 7 Li values), which is taken as a typical signal of congruent weathering (e.g., Dellinger et al., 2015;Krause et al., 2023).
So far, a variety of modern calcifiers and calcareous sediments have been studied for their Li isotope composition to assess their potential as archives, such as planktonic foraminifera (Hall et al., 2005;Hathorne & James, 2006;Misra & Froelich, 2009), benthic foraminifera (Marriott et al., 2004;Roberts et al., 2018) and brachiopods (Dellinger et al., 2018).Positive and negative Li isotope fractionations (noted as Δ 7 Li) relative to modern seawater δ 7 Li are observed in most biogenic carbonates and may be related to vital effects (Dellinger et al., 2018;Hathorne & James, 2006;Vigier et al., 2015).The extent of vital effects in modern planktonic foraminifera, brachiopods, and low-Mg bulk carbonates seems to be small compared with other skeletal carbonates.Thus, the aforementioned carbonates are suitable archives for δ 7 Li sw reconstructions (Dellinger et al., 2018).Other influencing factors such as temperature and seawater pH have been investigated on biogenic carbonates with still controversial results (Dellinger et al., 2018;Füger et al., 2019Füger et al., , 2022;;Roberts et al., 2018;Rollion-Bard et al., 2009;Seyedali et al., 2021;Vigier et al., 2015).The mineralogy of the carbonates has an influence on its Li isotope composition as well.Aragonite and calcite have different negative offsets toward the modern seawater Li isotope composition, whereby aragonite has lower δ 7 Li values than calcite (Pogge von Strandmann et al., 2019).
Modern planktonic foraminifera show the lowest offset toward the seawater Li isotope composition, but it is not clear if they are affected by species-specific vital effects (Hathorne & James, 2006;Misra & Froelich, 2012).In contrast, bulk carbonates have a higher offset toward the seawater Li isotope composition.Pogge von Strandmann et al. (2019) could not find any correlation between δ 7 Li values and the latitude, the water depth, or carbonate content of their bulk carbonate samples and therefore suggested bulk carbonates to be a reliable δ 7 Li archive.
Like for every other geochemical proxy, diagenesis may alter the Li isotope composition of fossil carbonates.Dellinger et al. (2020) and Wei et al. (2023) investigated the influence of meteoric diagenesis, burial diagenesis and dolomitization on shallow marine platform carbonates.Thereby, they observed that the recrystallization of aragonite into low-Mg calcite during meteoric diagenesis causes a decrease in Li/(Ca + Mg) as well as a shift to lower δ 7 Li carbonate compositions.The recrystallization of aragonite into calcite during burial diagenesis and dolomitization under fluid-buffered conditions appears to cause a reset in the δ 7 Li carbonate values (Dellinger et al., 2020).Attributed to this reset, the δ 7 Li carbonate values get close to the seawater Li isotope composition.Andrews et al. (2020) investigated the influence of authigenic clay mineral formation in carbonate-rich sediments on the Li isotope composition of bulk carbonates, and concluded that under closed-system condition, the formation of authigenic clay minerals can diagenetically alter the Li isotope composition of the bulk carbonates leading to an increase in δ 7 Li.Accordingly, a thorough diagenetic screening is necessary in order to determine the Li isotope composition of ancient carbonates.However, there is no uniform methodology because every archive is different and needs a specific treatment.Nevertheless, the analysis of stable Li isotopes in marine carbonates have been established as a proxy to determine silicate weathering in Earth's past (Cao et al., 2022;Crockford et al., 2021;Hathorne & James, 2006;Kalderon-Asael et al., 2021;Lechler et al., 2015;Misra & Froelich, 2012;Pogge von Strandmann et al., 2013, 2017;Pogge von Strandmann, Jones, et al., 2021;Sun et al., 2018;Ullmann et al., 2013;Wang et al., 2021;Washington et al., 2020).
So far, Li isotope data from the Late Cretaceous are rare, providing a rather fragmentary record (Kalderon-Asael et al., 2021;Misra & Froelich, 2012;Pogge von Strandmann et al., 2013).To generate a coherent long-term record, chalk could be a suitable archive.Chalk is a diagenetically modified pelagic low-Mg calcite carbonate mainly composed of small (0.5-5 μm) calcite coccolith plates as well as micro-crystalline calcite particles (<5 μm), called micarbs (Hancock, 1975).Since the 1970s, the term "micarb" is used in the literature, but with different definitions and different explanations for the formation of these particles (see Beltran et al., 2009 and references therein).Tagliavento et al. (2021) strongly support a rather early diagenetic origin of micarbs when the pore fluid was still in contact with seawater.Other subsidiary components of chalk are mainly foraminifera but also fragments of bivalves, echinoderms, bryozoans, and rarely ostracods and octocorals (Hancock, 1975).In contrast to pelagic limestones, chalk is soft, has a high porosity and is less lithified (Hancock, 1975).However, in comparison to carbonate ooze, chalk is more compacted and lithified.In previous studies, chalk has been used as an archive for a variety of different isotope systems such as δ 13 C, δ 18 O and 87 Sr/ 86 Sr (e.g., Jarvis et al., 2006;Jenkyns et al., 1994;McArthur et al., 1992;McLaughlin et al., 1995;Schönfeld et al., 1991;Voigt et al., 2010).
The advantage of chalk over the commonly used foraminifera as an archive in deep time applications is the easily obtainable large sample size of several milligrams.Pogge von Strandmann et al. ( 2013) measured δ 7 Li in chalk across the interval of Oceanic Anoxic Event 2 (OAE 2) to assess changes in the silicate weathering flux and intensity during this event.However, they did not address the issue of the general suitability of chalk as an archive for stable Li isotopes and did not consider clay mineral dissolution as an alternative interpretation for their very low δ 7 Li values across the Plenus Marls (Seyedali et al., 2021;Taylor et al., 2019).
Here, we evaluate white chalk from the Maastrichtian as a potential archive for the Li isotope composition of Late Cretaceous seawater from a relative comparison to other archives.We use Quaternary coccolith ooze from the Manihiki Plateau (Pacific Ocean) as a close analog to chalk and compare their trace elemental and Li isotope compositions.We further compare the Li isotope compositions of modern planktonic foraminifera relative to modern seawater and investigate the magnitude of offsets between the different archives for the Maastrichtian boreal white chalk from the outcrops Kronsmoor and Hemmoor (Northern Germany) and Maastrichtian planktonic foraminifera (Misra & Froelich, 2012).

Sample Material
Three coccolith ooze samples were taken from a piston core (SO225-8-3 KOL) at the Manihiki Plateau (northern Western Plateaus; 7°11.8990′S,165°3.1810′W) in 3,589.00m water depth obtained during RV Sonne expedition SO225 (Werner et al., 2013).The samples were taken at 1.0 cm intervals between 1,051 and 1,054 cm.The samples have an age of ∼1.85 Ma determined by a δ 18 O-based age model (Raddatz et al., 2017).The samples from the Manihiki Plateau were chosen because of the low terrigenous input, which is similar to the depositional setting of the Maastrichtian chalk, and because of the similar carbonate factory of both locations.
For the Maastrichtian chalk, a total of 31 samples were selected, which originated from the two boreal white chalk outcrops of Northern Germany, Kronsmoor (53°54′08.0″N,9°35′08.4″E)and Hemmoor (53°41′54.3″N,9°0 8′06.6″E).The samples are part of a sample series used for the analysis of stable carbon and oxygen isotopes in chalk and planktonic and benthic foraminifera (Stenvall, 1997;Voigt et al., 2010) and strontium isotopes (McArthur et al., 1992;McLaughlin et al., 1995).The chalk from these outcrops is classified as a very pure, extremely fine-grained and poorly cemented, foraminifera-bearing nannofossil chalk (Ehrmann, 1986;Schönfeld et al., 1991;Schulz et al., 1984) with a carbonate content between 96% and 99% (Stenvall, 1997;Voigt et al., 2010).During the Late Cretaceous, the sections were part of the boreal shelf sea and were located near the center of the North Sea Basin (Tyson & Funnell, 1987).The water depth was between 50 and 200 m and the terrigenous input was low (Ernst, 1978;Wilmsen & Niebuhr, 2017).Due to tectonic uplift caused by the elongated salt structure of Krempe, the sections of Lägerdorf, Kronsmoor, and Hemmoor are exposed near earth's surface today (Schulz et al., 1984) and have been used for industrial chalk quarrying.The Hemmoor samples belong to the collection of Friedrich Schmid stored in the Federal Institute for Geosciences and Natural Resources (BGR) today (Schmid, 1982).The Kronsmoor samples derive from the Schönfeld collection stored at the core repository at GEOMAR Kiel (Voigt & Schönfeld, 2010).The samples range from the belemnite sumensis biozone (lower Maastrichtian) into the baltica/danica biozone (upper Maastrichtian) and cover a time between 71.2 and 66.1 Ma.The samples ultimately used for this study were selected with a spacing of approximately 200-300 kyr.

Sample Preparation, Dissolution, and Chromatography
The three wet Quaternary coccolith ooze samples from Manihiki were dried in a compartment drier for 24 hr and pulverized with an agate mortar, which was cleaned with 100% methanol (distilled analytical grade) before each use.Subsequently, all three samples were mixed together in equal parts within the mortar to obtain a homogeneous powder, which will henceforth be referred to as the coccolith ooze sample.The Maastrichtian chalk samples were drilled with a conventional stainless steel microdrill with a round (Ø 1.8 mm) diamond-grinding bit to obtain several milligrams of rock powder from a hand sample several cm in size.The surface of each hand sample was examined for macrofossil remnants.Drilling was done under a binocular in an area that appeared as pure as possible.The sample powder as well as the drilled borehole were examined to see if any remnants of the drilled macrofossils were avoided.
The following sample preparation and column chemistry were performed in a clean laboratory housed at FIERCE (Frankfurt Isotope and Element Research Center, Goethe-University Frankfurt, Germany).A pre-leaching procedure based on the Tessier sequential extraction (Tessier et al., 1979) was applied to all samples to remove adsorbed Li and Li bound to clay, which are attached to the sample surface.Therefore, 50 ± 0.1 mg of the powdered sample were pre-leached in 2 mL of 1 M ammonium acetate for 30 min.Ammonium acetate was used instead of sodium acetate in order not to add more Na to the sample prior to the column chemistry.Samples were then centrifuged for 6 min at 11 × 10 4 revolutions per minute (rpm) and the supernatant was discarded.Subsequently, the samples were washed three times with 2 mL of deionized water with a resistivity of 18.2 MΩ cm (Milli-Q ® ), which was buffered with ammonium acetate to a pH of 7. The Milli-Q ® water was buffered to prevent clay mineral leaching through Milli-Q ® with a lowered pH.To break up any clumps formed during the washing step and to re-suspend our powders, the samples were sonicated for 30 s. Afterward, the samples were dried on a hot plate at 60°C for 24 hr.For the main leaching process, the dried carbonate samples were pulverized again and 10 ± 0.1 mg of each sample were leached in 2 mL of 0.05 M HNO 3 for 2 min while the suspension was agitated.Subsequently, the samples were centrifuged for 6 min with 11 × 10 4 rpm and 1 mL of the supernatant was separated from the residual and used for column chemistry.
Li purification follows the procedure reported by Seitz et al. (2004).Prior to the column chemistry, all dissolved samples were evaporated on a hot plate at 130°C and redissolved in 180 μl 5 M HNO 3 and 720 μl of 100% methanol (distilled analytical grade) within Teflon ® beakers.Following this, all samples were passed through 1.4 mL exchange columns filled with resin (BioRad AG50W-X8, mesh size 200-400) to separate Li from the other elements, mainly Na.In agreement with a Na separation test (Text S1 and Figure S1 in Supporting Information S1), the main cut covered a total volume of 11 mL starting after a pre-cut of 4 mL.Pre-and post-cuts were collected and saved for further element analyses.An internal seawater standard and a procedural blank were processed with each set of samples to monitor the purification routine.The purified samples as well as the pre-and post-cuts were carefully dried on a hot plate at 130°C within Teflon ® beakers.Following that the purified samples were redissolved in 1 mL 5% HNO 3 for Multi-Collector Inductively Coupled Plasma Mass Spectrometer (MC ICP-MS) and the pre-and post-cuts in 1 mL 2% HNO 3 for Inductively Coupled Plasma Optical Emission Spectrometry (ICP-OES) measurements in order to matrix match the samples with the standards.

Sample Analyses
The Li isotope and concentration measurements were performed on a MC ICP-MS Neptune Plus from Ther-moFisher Scientific at FIERCE (Frankfurt Isotope and Element Research Center, Goethe-University Frankfurt, Germany).Measurements were conducted at dry plasma conditions using a Cetac Aridus ® desolvator fitted with a PFA pray chamber and an ESI micro-concentric nebulizer with an uptake rate of 50 μl/min.The sample gas was dried at 160°C before introduction into the plasma.Beam intensity of 10-20 pA (1-2 V using 10 11 Ω resistor) for 7 Li at 5 ng/g concentration level was achieved with standard cones (H-Cones).
The analytical blank (distilled 5% HNO 3 ) was usually 4-6 mV on 7 Li.Sample analysis was carried out sequentially by "bracketing" the sample with the L-SVEC standard (Flesch et al., 1973) to correct the instrumental mass bias.The total integration time for each measurement was ∼2 min.In addition, a chemistry (or full procedural) blank was also frequently measured during the data acquisition period and its value was subtracted offline from sample and standard analyses.The blank typically had values around 15 mV and thus corresponded to less than 1% of the sample signal.
Elemental concentrations of Al, Mg, Mn, Sr, and Ca were measured on the pre-and post-cuts of column chemistry on an ICP-OES using a Thermo Scientific iCap 6300 (dual viewing) at FIERCE.Briefly, samples were diluted to have 1 mg/l of Y and 50 mg/l Ca in order to account for matrix effects during measurements.The measured intensity data were background subtracted, standardized internally to Y and normalized to Ca. Calibration standards were mixed from single-element standard solutions to match the expected element concentrations of the samples (cf., Rosenthal et al., 1999).The limestone standard ECRM 752-1 (non-centrifuged; Greaves et al., 2008) was measured after every tenth sample to allow for drift correction and monitor measurement quality.Relative precision of the measurements were calculated from the repeated measurements of ECRM 752-1 and amounts of 0.97 ± 0.1 mmol/mol for Al/Ca, 0.22 ± 0.03 mmol/mol for Fe/Ca, 3.67 ± 0.1 mmol/mol for Mg/Ca, 0.14 ± 0.002 mmol/mol for Mn/Ca and 0.17 ± 0.004 mmol/mol for Sr/Ca.All of them are in agreement with the reported values of Greaves et al. (2008).The Li/Ca ratios were calculated with the Li concentration from the MC ICP-MS measurement and the Ca concentration from the ICP-OES measurement.All results of the investigated samples are present in Table S1 for the ooze samples and Table S2 for the chalk samples.
The measured δ 7 Li values of Maastrichtian chalk vary between +18.1 and +28.3‰.The mean δ 7 Li for the Maastrichtian chalk is +23.6 ± 0.7‰ (2SE; n = 33) (Figure 1).The Li concentration within the chalk samples varies between 170 and 272 ng/g with a mean value of 211 ± 6.2 ng/g (2SE; n = 32).The Li concentrations and the element/Ca ratios of the Maastrichtian chalk are plotted against their corresponding δ 7 Li values in Figure 2. The Li/Ca ratios vary between 22.2 and 37.43 μmol/mol with a mean of 27.7 μmol/mol.The Al/Ca ratios are below 0.4 mmol/mol and vary between 0.14 and 0.37 mmol/mol.In case of the Al/(Ca + Mg) ratio, values vary between 0.14 and 0.36 mmol/mol.The mean value for both is 0.22 mmol/mol.The Mg/Ca ratios vary between 4.93 and 10.1 mmol/mol with a mean of 7.06 mmol/mol.The Mn/Ca ratios vary between 0.18 and 1.35 mmol/ mol with a mean of 0.31 mmol/mol, and the Sr/Ca ratios vary between 0.56 and 1.11 mmol/mol with a mean of 0.85 mmol/mol.The Mn/Sr ratio ranges between 0.16 and 1.31 with a mean of 0.37 mol/mol.

Pre-Leaching Procedure to Eliminate Silicate Contamination
The main problem using chalk as a potential archive for the seawater Li isotope composition is its purity.In addition to calcareous components, chalk contains silicates in the form of clay minerals-mostly smectites (e.g., Chenot et al., 2016).However, clay minerals are a sink for Li (μg/g-level) and preferentially incorporate the light Li isotope ( 6 Li), while Li concentrations in carbonates are generally significantly lower (ng/g-level) (e.g., Chan et al., 1992).Therefore, remaining clay fractions within the chalk will bias the Li isotope signature, driving it to lighter δ 7 Li compositions.In order to address this problem, Pogge von Strandmann et al. ( 2013) tested the preleaching procedure of Tessier et al. (1979) using samples from the English chalk, but could not observe significant differences between pre-leached and non-treated samples.However, we decided to test a pre-leaching using ammonium acetate (CH 3 COONH 4 ) and different leaching techniques using various acids and acidities (0.05, 0.1 and 0.2 M HNO 3 , 0.125 and 0.25 M HAc), and variable leaching times (seconds up to 1 hr).First, we analyzed our chalk samples without a pre-leaching procedure (data in Table S3).We then analyzed the same sample material after applying the ammonium acetate pre-leaching method and ascertained that the δ 7 Li values shifted to heavier isotope compositions (with an increase of up to 5.4‰, average 2.8‰) (Figure 3a).In addition, we observed significantly higher Li concentrations in the non-pre-leached samples compared to the pre-leached samples (Figure 3b).This pattern suggests that the higher Li concentrations result from leached clay minerals in untreated samples.We therefore conclude that a proper pre-leaching procedure is essential.Additional to the increase in the δ 7 Li chalk values and the decrease in the Li concentration, we observe a decrease in the Al/Ca ratios of the pre-leached chalk samples (Figure 3c).On average, the Al/Ca values decrease by 0.15 mmol/mol.So far, the Al/Ca and Al/(Ca + Mg) ratios have been used as a tracer for clay mineral contamination within carbonate samples (Dellinger et al., 2018;Pogge von Strandmann et al., 2013).2020), all of our non-pre-leached samples would have been classified as falsely reliable (Figure 3c).Bastian et al. (2018) already pointed out that the Al/Ca ratio may not be the best tracer for silicate contamination and concluded that no common threshold can be defined.Our data support their point.However, the Al/Ca ratio can be used as a rough indicator of silicate contamination but is not suited to decide on the basis of a certain threshold whether a sample is considered contaminated or not.Therefore, more reliable methods for detecting silicate contamination need to be determined.

Chalk Diagenesis
Chalk is a fine-grained pelagic carbonate that formed by compaction and lowgrade lithification of coccolith-rich low-Mg nannofossil ooze.During compaction, the porosity within the ooze is decreasing while secondary micrite fills the former open space (e.g., Schlanger & Douglas, 1974;Scholle, 1977).Hence, chalk is not a pristine sediment anymore.Nevertheless, it can still preserve geochemical signatures.Different studies from the Northern German sections Lägerdorf, Kronsmoor and Hemmoor concluded that chalk still preserves relative paleoenvironmental changes (e.g., Engelke et al., 2018;McLaughlin et al., 1995;Voigt et al., 2010).These authors determined this by comparing the stable isotope composition of chalk with the stable isotope composition of co-occurring macrofossils.Additionally, they observed no significant correlation between δ 13 C and δ 18 O isotopic data measured on the chalk samples and thus concluded that chalk still shows paleoenvironmental rather than diagenetic signatures.In fact, the δ 13 C and δ 18 O values for the Maastrichtian chalk samples (Voigt et al., 2010) shows no strong positive correlation (R 2 = 0.4), commonly interpreted as an indicator of strong pore fluid-rock interaction, indicating only low diagenetic overprinting (Figure 4).
The presence of micrites and micarbs contributes to the diagenetic overprinting of bulk carbonates.They are a considerable part of chalk that can  dominate the <5 μm grain-size fraction (Tagliavento et al., 2021).During all diagenetic stages, recrystallization redistributes calcite by dissolving ions and re-precipitating them at stable surfaces to form micrites (Hjuler & Fabricius, 2009).Most of the micrite formation is assigned to burial diagenesis, probably occurring after the dissolution of the coccoliths (McLaughlin et al., 1995;Schönfeld et al., 1991;Voigt et al., 2010).Micarbs are calcite particles <5 μm with an unknown origin and without a distinctive morphology (Beltran et al., 2009;Cook & Egbert, 1979;Hashim & Kaczmarek, 2019;Hasiuk et al., 2016;Minoletti et al., 2005;Westphal et al., 2004;Wise & Kelts, 1972).Tagliavento et al. (2021) studied the chemical composition of micarbs in the Campanian and Maastrichtian chalk from the Danish Basin and analyzed the Mn/Ca and Sr/Ca ratios within the 5-10 μm (coccolith-rich) and 1-5 μm (micarb-rich) fractions.They observed only insignificant differences and interpreted the micarbs as an early diagenetic neoformation of calcite within the upper 100 m of sediment burial, when pore water is still able to exchange with seawater.This observation provides evidence for a relatively low degree of chemical alteration of micarbs during diagenesis.
Element geochemical data can yield further information about diagenetic influences.Our dataset does not show any clear correlation between the Li isotope composition and the different element/Ca ratios (Figure 2).Instead, we observe a homogenous pattern in most cases.The absence of a strong correlation also indicates no major diagenetic alteration trends.Nevertheless, the Maastrichtian chalk and the Quaternary coccolith ooze exhibit a significant difference in the E/Ca ratio with the expectation of the Al/Ca ratio (Figure 2c).In general, high Al concentrations in carbonates indicate contamination by silicates that were not removed during sample processing (Dellinger et al., 2018;Pogge von Strandmann et al., 2013).The low and homogenous Al/Ca ratio in the chalk and the coccolith ooze indicates none or just a minor influence of leached silicate minerals.
In comparison to the coccolith ooze, the chalk samples are enriched in Li, Mg, and Mn (Figures 2b, 2d, and 2e) but depleted in Sr (Figure 2f).This pattern is most likely an expression of the light diagenetic overprinting of the chalk samples.The Li/Ca ratio in the coccolith ooze fits with the Li/Ca values reported for Emiliania huxleyi (Langer et al., 2020).Therefore, the chalk samples are enriched with Li.This enrichment could be an indicator of clay mineral leaching since clays yield higher Li concentrations than carbonates.However, we lack the geochemical methodology to prove the clay contamination unequivocally.The higher content of Li in chalk compared to the coccolith ooze may originate from its higher Mg content, that is, the increased incorporation of Mg leads to more defects in the calcite lattice, which thus may favor the co-precipitation of other, less compatible, elements such as Li (Knight et al., 2023).The enrichment in Mg may indicate a diagenetic effect.Mg can be incorporated from pore fluids into calcites during recrystallization.Thereby, an elevated Mg concentration can be achieved within the pore fluid through the dissolution of high-Mg calcites, such as echinoderms.Thus, the abiogenic calcite formed via early diagenesis-induced recrystallization would incorporate more Mg compared to the primary biogenic calcite.
The enrichment could also be related to an increased incorporation of Mn into the abiogenic calcites.The presence of Mn in carbonates is an indicator of recrystallization under anoxic conditions.Under oxic conditions, Mn forms a solid Mn oxy-hydroxide (III, IV) phase, but it is reduced to soluble Mn (II) under anoxic conditions (Burdige, 1993) and can migrate to the pore fluid from where it can be incorporated into the calcite crystal lattice.The enrichment of Mn within the chalk samples compared to the coccolith ooze most likely results from this process.Sr is a part of all biogenic calcites since it substitutes Ca 2+ in aragonites and calcites (McIntire, 1963).The Sr concentration differs within the different types of biogenic calcites but is typically higher than that in nonbiogenic calcites (Baker et al., 1982;Elderfield et al., 1982).However, it can be depleted during diagenesis since Sr is removed from the crystal lattice (Baker et al., 1982).Hence, the Sr concentration in biogenic calcites varies (e.g., through vital effects); therefore, it is not suitable to detect light alteration within calcites.Within previous studies, the comparison between Mn and Sr has been widely used to determine diagenetic alteration, whereby a clear correlation between Mn and Sr indicates late diagenetic overprinting (e.g., Brand & Veizer, 1980, 1981;Ullmann & Korte, 2015;Ullmann & Pogge von Strandmann, 2017).The Maastrichtian chalk samples do not show a clear correlation between Mn/Ca and Sr/Ca nor between Mn/Sr and the age of the Maastrichtian samples.Therefore, we can assume that the samples did not undergo late-diagenetic overprinting.Nevertheless, the early diagenetic influence is evident.
The degree of diagenetic alteration of chalks can also be assessed by comparing its geochemical properties with those of pristine macrofossils (e.g., Harlou et al., 2016;Voigt et al., 2004).The direct comparison between the Maastrichtian chalk samples from Northern Germany and pristine, Maastrichtian brachiopod shells from the work of Harlou et al. (2016) shows that, on average, the chalk samples have higher Mn and lower Sr values (Figure 5).This underlines our observation, that the chalk samples show signs of beginning burial diagenesis.Compared to the coccolith and micarb fractions of Tagliavento et al. (2021), the average Mn/Ca and Sr/Ca values within the Maastrichtian N-German chalk samples are lower.Compared to the Cenomanian-Turonian chalk from Eastbourne by Pogge von Strandmann et al. ( 2013), the N-German chalk shows lower Mn/Ca ratios.The Sr/Ca ratios are comparable, but the chalks from Eastbourne show a larger uncertainty.Pogge von Strandmann et al. ( 2013) also provided Cenomanian-Turonian chalk samples from South Ferriby.These show significantly higher mean Mn/Ca values compared to the Maastrichtian chalk samples as well as the Eastbourne samples.The Sr/Ca values show an opposite picture, whereas South Ferriby exhibits the lowest Sr/Ca values.It is possible that the chalk samples from South Ferriby have experienced a more extensive burial diagenesis than the Eastbourne samples.Nevertheless, it should be emphasized once again that the samples from Eastbourne and South Ferriby are of Cenomanian-Turonian age and originate from different locations than the Maastrichtian N-German chalk.Differences in element geochemistry between these three chalks could therefore be related to the age difference and thus, be due to changes within the ocean chemistry.
The geochemical data clearly show an early diagenetic influence on the chalk samples.However, this alone provides no information whether the Li isotope composition is influenced during diagenesis as well.The formation of micrite typically occurs during burial diagenesis.In shallow marine carbonates, marine burial diagenesis and dolomitization have no significant impact on the Li isotope composition of the carbonates as long as they are under fluidbuffered conditions (Dellinger et al., 2020;Wei et al., 2023).Otherwise, the Li isotope composition will decrease toward lower values.However, pelagic sediments have not been analyzed regarding their Li isotope composition during burial diagenesis so far.For a better understanding about the behavior of Li isotopes during recrystallization, Seyedali et al. (2021) performed inorganic experiments, where vaterite has been recrystallized into calcite within an aqueous solution.Thereby, a correlation between the Li isotope composition and pH was observed, and it was concluded that the results of the study are important for early diagenetic reactions.However, the results of this study are not applicable for our study because the experimental setup cannot be transferred to natural conditions (aqueous solutions with major differences to the real chemical composition of modern seawater plus recrystallization of vaterite, a metastable phase that rarely occurs in nature).Therefore, we cannot give an answer to the question if and/or how the formation of micrite has an influence on the Li isotope composition of the bulk chalk samples.An important influence on the δ 7 Li carbonate signal has the diagenetic formation of authigenic clays within carbonates.More in detail, authigenic clay formation influences the Li concentration within the pore fluid and therefore on δ 7 Li carbonate during recrystallization processes.Thereby, the δ 7 Li value of carbonates increases, when the Li concentration decreases because 6 Li is incorporated into the authigenic clays while the pore fluid becomes enriched with 7 Li (Andrews et al., 2020).In our dataset, we do not observe a negative correlation between δ 7 Li carbonate and the corresponding Li concentration (Figure 2a).Based on this, we consider that both the coccolith ooze and the Maastrichtian chalk are not influenced by authigenic clay formation in a closed system.
Overall, the behavior of Li during diagenesis is not well constrained.Only a few studies have addressed this issue but focused on modern settings only.However, a good diagenesis screening is needed for the analysis of Li isotopes on ancient sediments with respect to the past seawater isotope composition.Consequently, more studies are needed to analyze whether and to what extent Li is affected during diagenesis.It is crucial to perform these studies under different geochemical conditions that match the different geochemical conditions during the Earth's past.

Assessment of Uncertainties
Although we assume that the chalk samples record a reliable paleoenvironmental signal, the Li isotope composition of Maastrichtian chalk shows a large scatter.These scatter results in a second (more negative value;  (Harlou et al., 2016).Coccolith-rich and micarb-rich are grainsized-dependent analyses of Campanian-Maastrichtian chalk from the Danish Basin (Tagliavento et al., 2021).N-German chalk values are results of this study, and chalk compositions of Eastbourne and South Ferriby refer to Cenomanian-Turonian samples in England (Pogge von Strandmann et al., 2013).n = 7) and third population (more positive values; n = 2) in our dataset.These nine samples (depicted as light gray symbols in Figure 2) stand out since their δ 7 Li deviates by >2‰ from the Li isotope composition of stratigraphically neighboring samples.It is unlikely that these nine samples represent changes in seawater δ 7 Li.The temporal spacing between the samples of approximately 200 ka is distinctly lower than the modern residence time of Li in the ocean (∼1.2 Ma; Misra & Froelich, 2012).The difference in δ 7 Li of our outliers compared to stratigraphically neighboring samples is (a) larger than our lab precision, that is, the error bars do not overlap, and (b) a decrease in δ 7 Li to the observed extent over such a short amount of time with an equally large relapse is simply unrealistic in a geological context.Such large short-term changes in seawater δ 7 Li are known only for example, major carbon cycle perturbations so far, the occurrence of which is unknown for the time interval of interest.However, there are several other possibilities that could cause this scatter: 1. Analytical procedure and acquisition.Sample preparation was performed in a clean laboratory.Contamination with other Li-containing material is unlikely, since no samples other than the coccolith ooze and chalk samples were processed at the same time.Chemicals contaminated with Li should not affect only individual samples, since they were used equally for all samples.Another major concern is the separation of Li and Na during column chemistry since Na affects the ionization of Li and thus may lead to anomalous fractionations during analysis.While calibrating our columns for carbonates, we could not detect any measurable Na leakage in our selected main cut samples (Text 1 and Figure S1 in Supporting Information S1).Therefore, we consider that our one-stage column chemistry effectively separates Li from other cations.However, we cannot fully rule out the possibility of a Na contamination of lab wares or the analytical instruments.Furthermore, the precision on our carbonate matrix (coccolith ooze) has a 2σ uncertainty of 1.4‰, which is higher than the 2σ uncertainty of our seawater measurements (0.5‰).Nevertheless, the combination of our external precision and the precision on our carbonate samples could cause uncertainty of >2‰ and therefore be responsible for the second and third data population.2. Sample heterogeneity.Chalk is not a homogenous carbonate material, that is, it consists mainly of coccoliths but may contain fragments of foraminifera and mollusk shells.Vital effects may cause different δ 7 Li signatures in each of these groups of calcifying organisms.Most calcitic mollusks, for example, show positive fractionations relative to seawater (Dellinger et al., 2018).A partial contribution of mollusk-derived carbonate could be a plausible explanation for the enrichment of δ 7 Li values.3. Clay contamination.Clay minerals are enriched with 6 Li.When clays are leached or dissolved during sample processing, they can contribute to the lithium isotope composition of the carbonate samples and lower its isotopic signature.We tried to overcome this issue by applying a pre-leaching procedure based on the Tessier sequential extraction.Samples with depleted δ 7 Li values do not show any indications for clay contamination in their Al/Ca and Li/Ca ratios in comparison to the non-depleted samples (Figures 2b and 2c).Since we cannot pin down markers for clay contamination on elemental geochemistry, this explanation is difficult to verify or falsify.4. High porosity/fluid-rock interaction.The chalk from Northern Germany has a porosity of 40%-50% (Scholz, 1973).Therefore, pore fluids can circulate through the sediment and interact with it.Samples with unusually low Li isotopes could derive from horizons, which are more marly and/or bioturbated or from the proximity to flint beds.Flint horizons represent a diagenetic environment of more intense macrofossil fragment and coccolith replacement reactions initiated by organic matter decay during periods of lower sedimentation rates, which fixed the redox boundary and the microbial metabolic zones at a specific depth below the sea floor (Madsen & Stemmerik, 2010).Further, marly bioturbated horizons can promote an environment of differential diagenesis.Thereby, the pore fluid can be enriched with 6 Li when clays are dissolved and can result in increased incorporation of 6 Li, which lowers the overall Li isotope composition of the bulk chalk.The latter appears to be the best explanation for the unusually low lithium isotope ratios even if there is no clear indication from other geochemical data.

Coccolith-Rich Sediments as an Archive for the Li Isotope Composition of the Seawater
The Quaternary coccolith ooze from the Manihiki Plateau is the closest analog to the Maastrichtian chalk since there is no modern chalk nor Cretaceous ooze.The coccolith ooze has been sedimented in a similar carbonate factory as the chalk and in a similar depositional setting with a low terrigenous input.Even with an age of 1.85 Ma, the coccolith ooze can be compared to modern setting, because the Li isotope composition of the seawater did not vary significantly within the last 4 Ma (Hathorne & James, 2006;Misra & Froelich, 2012).
Therefore, the mean δ 7 Li value of the coccolith ooze (+27.2 ± 0.5‰; 2SE; n = 9) represents a 3.9 ± 0.6‰ offset from the modern seawater Li isotope composition of +31.1 ± 0.3‰ (2SE; n = 54).This negative offset between the Li isotope composition of seawater and coccolith ooze is similar to the offset between seawater and modern calcareous core-top samples from the Atlantic Ocean, which ranges between +23.4 and + 26.8‰ with a mean value of +25.3‰ (Figure 6, and references noted therein The Li isotope composition of the Manihiki coccolith ooze can also be compared with the δ 7 Li composition of modern planktonic foraminifera, which are about 1.6 ± 0. 5‰ lighter than seawater (Hall et al., 2005;Hathorne & James, 2006;Misra & Froelich, 2009).Thus, they can be used as an archive for the Cenozoic seawater δ 7 Li record (Hathorne & James, 2006;Misra & Froelich, 2012).
Such inter-archive comparison allows the assessment of the Maastrichtian chalk as a potential recorder of the Li isotope composition of Maastrichtian seawater.The Δ 7 Li between modern planktonic foraminifera and Quaternary coccolith ooze is +2.3 ± 0.6‰.Misra and Froelich (2012) published δ 7 Li data for planktonic foraminifera from the late Maastrichtian (68.2-65.3Ma) with a mean δ 7 Li value of +26.1 ± 0.3‰ (2SE; n = 12).Using the mean of the Maastrichtian chalk δ 7 Li (+24.7 ± 0.4‰; 2SE; n = 10) from an equivalent time span, the offset between both archives (Δ 7 Li pf-chalk ) is +1.4 ± 0.5‰.Based on the close similarity of Δ 7 Li pf-ooze and Δ 7 Li pf-chalk in the Quaternary and Maastrichtian, we assume this offset to represent a time-independent feature related to the taxonspecific Li isotope fractionation during the biomineralization of planktonic foraminifera and coccolithophores, which is not substantially influenced by the diagenetic processes affecting the chalk samples.In analogy, we further assume that the isotopic effects causing the Δ 7 Li between coccolith-rich sediments and seawater remained similar constant over time, and that the isotopic effects causing the Δ 7 Li between the coccolith-rich sediments and seawater remained constant over time.
Although this study is based only on a single site, we assume that the Maastrichtian oceans were well mixed for Li because of its residence time and that shelf waters represent a true open ocean signature.Therefore, we can reconstruct the mean seawater Li isotope composition for the Maastrichtian.By adding the Quaternary Δ 7 Li sw- ooze value of +3.9 ± 0.6‰ to the mean Maastrichtian δ 7 Li chalk (+23.6 ± 0.7‰; 2SE; n = 33), the coeval seawater Li isotope composition was +27.5 ± 1.0‰.By adding the modern Δ 7 Li sw-pf value of +1.6 ± 0.5‰ to the mean value of the late Maastrichtian δ 7 Li pf (+26.1 ± 0.3‰; 2SE; n = 12), we obtain a value of +27.7 ± 0.6‰ for the late Maastrichtian seawater Li isotope composition.The two values are in good agreement with each other.To verify our results, further studies from other locations (e.g., Pacific Ocean) need to be done.

Conclusions
In this study, we investigated the Li isotope composition of Quaternary coccolith ooze from the Manihiki Plateau (Pacific Ocean) and Maastrichtian white chalk from two quarry outcrops in Northern Germany (Kronsmoor, Hemmoor).Thereby, we wanted to assess inasmuch chalk can be used as a reliable archive to determine the Li isotope composition of Late Cretaceous seawater.We observed that chalk provides a large potential for uncertainties caused by clay contamination.Carbonate samples should always be treated with a pre-leaching Modern calcites: planktonic foraminifera (Hall et al., 2005;Hathorne & James, 2006;Misra & Froelich, 2009), benthic foraminifera (Marriott et al., 2004;Roberts et al., 2018), mollusks and brachiopods (Dellinger et al., 2018), core-top sediments (Pogge von Strandmann et al., 2019), and cave carbonates (Day et al., 2021).Quaternary calcite: coccolith ooze (this study).Maastrichtian calcites: Planktonic foraminifera (Misra & Froelich, 2012), and chalk (this study).The boxes represent the interquartile range with the median expressed as the horizontal lines and the whiskers reflecting the extreme data points.The black dots represent the outliers.
procedure to reduce the amount of Li contamination through adsorbed Li and Li bound to clays.However, it proves to be extremely difficult to detect clay contamination via the Al/Ca ratio.Unfortunately, other methods have not yet been developed but are in need for a good sample screening.Also, the high porosity of chalk can be a problem and horizons with a higher diagenetic environment as flint or marl layers should be avoided and bulk carbonates should be as pure as possible.Besides this, chalk yields the potential to be a good archive for Li isotopes.
In this study, Quaternary coccolith ooze served as an analog to the Maastrichtian chalk.We measured a δ 7 Li of +27.2 ± 0.5‰ (2SE; n = 9) for the ooze, which results in a 3.9 ± 0.6‰ offset toward the modern seawater Li isotope composition and a 2.2 ± 0.6‰ offset toward the modern planktonic foraminifera.Moreover, we could compare our late Maastrichtian chalk data to published late Maastrichtian planktonic foraminifera data.Hereby, an offset of 1.4 ± 0.5‰ could be determined.Since these offsets are indistinguishable within uncertainty, we consider this as a robust time independent feature and consider chalk a suitable archive for the Li isotope composition of ancient seawater.Therefore, we are able to reconstruct a mean value of +27. ).We also thank Branwen Williams for the editorial handling and two anonymous reviewers for their constructive comments, which greatly improved our manuscript.Open Access funding enabled and organized by Projekt DEAL.

Figure 1 .
Figure 1.Box-Whisker plot showing the results from the Quaternary (Q) coccolith ooze and the Maastrichtian (Ma) chalk samples.The left y-axis shows the isotopic difference from the modern seawater Li isotope composition.The right y-axis shows the measured δ 7 Li values.The numbers on the x-axis show the number of replicate measurements and the total amount of analyzed samples.The boxes represent the interquartile range with the median expressed as the horizontal lines and the whiskers reflecting the lower and upper quantiles of the data.The black dots represent outliers.

Figure 2 .
Figure 2. Li isotope (δ 7 Li) compositions from the coccolith ooze (yellow diamonds) and the Maastrichtian chalk (green and light gray circles) plotted against Li concentration and element/Ca ratios.The light gray color represents the outliers discussed in chapter 4.3.Error bars represent 2SE.The coefficient of determination (R 2 ) was determined based on all chalk values.The red solid line in (c) represents the proposed Al/Ca threshold of 0.8 mmol/mol by Pogge von Strandmann et al. (2013), and the red dashed line represents the proposed Al/(Ca + Mg) threshold of 0.45 mmol/mol byDellinger et al. (2018).The Al/Ca and Al/(Ca + Mg) results do not vary significantly from each other so that both lines can be plotted in the same graphic.
However, we have to call this method into question.Pogge von Strandmann et al. (2013) used an Al/Ca threshold 0.8 mmol/ mol as a tracer for silicate contamination in their chalk samples.Dellinger et al. (2020) used another threshold with an Al/(Ca + Mg) value of 0.45 mmol/mol for their biogenic carbonates.Even if we apply the lower threshold of Dellinger et al. (

Figure 3 .
Figure 3.Comparison between pre-leached (blue) and non-pre-leached (orange) samples showing the effects of the pre-leaching procedure.(a) Li isotope composition (δ 7 Li).K1, K3, K5, K8, and K10 are the sample names.(b) Li isotope composition (δ 7 Li) plotted against their Li concentration.(c) Li isotope composition (δ 7 Li) of sample K3, K8 and K10 plotted against their Al/Ca ratios.The red solid line in (c) represents the proposed Al/Ca threshold of 0.8 mmol/mol by Pogge von Strandmann et al. (2013), and the red dashed line represents the proposed Al/(Ca + Mg) threshold of 0.45 mmol/mol by Dellinger et al. (2020).The Al/Ca and Al/(Ca + Mg) results do not vary significantly from each other-both lines are depicted in panel (c).

Figure 4 .
Figure 4. Cross plot of δ 13 C versus δ 18 O of the chalk samples.Data measured by Voigt et al. (2010).

Figure 5 .
Figure 5.Comparison of published mean Mn/Ca and Sr/Ca ratios from Late Cretaceous chalk-derived carbonates.Error bars represent 2SE.Gisilina jasmundi, Argyrotheca bronnii and Magas chitoniformis are all pristine Maastrichtian brachiopod shells (Harlou et al., 2016).Coccolith-rich and micarb-rich are grainsized-dependent analyses of Campanian-Maastrichtian chalk from the Danish Basin (Tagliavento et al., 2021).N-German chalk values are results of this study, and chalk compositions of Eastbourne and South Ferriby refer to Cenomanian-Turonian samples in England (Pogge von Strandmann et al., 2013).
5 ± 1.0‰ for the Maastrichtian seawater Li isotope composition.This research was funded through the VeWA consortium (Past Warm Periods as Natural analogs of our high-CO 2 Climate Future) by the LOEWE program of the Hessen Ministry of Higher Education, Research and the Arts, Germany.FIERCE is financially supported by the Wilhelm and the Else Heraeus Foundation and by the Deutsche Forschungsgemeinschaft (DFG: INST 161/921-1 FUGG, INST 161/ 923-1 FUGG, and INST 161/1073-1 FUGG), which is gratefully acknowledged.This is FIERCE contribution No. 151.Dirk Nürnberg is acknowledged for providing the Manihiki samples.Sonne Cruise SO225 was funded by the German Ministry for Education and Research (BMBF-Bundesministerium für Bildung und Forschung) in the framework of the joint project Manihiki II (BMBF Project Number 03G0225A (Day et al., 2021;nn et al. (2019)of the Manihiki coccolith ooze is 2.0‰ heavier compared to the mean value of the modern bulk core-tops byPogge von Strandmann et al. (2019).However, both archives are within the range of uncertainty to each other.Both samples differ in their mean Mg/Ca ratio (coccolith ooze: 2.55 ± 0.1 mmol/mol (2SE; n = 12); bulk core-tops: 27.6 ± 4.3 mmol/mol), with the core-top samples being enriched in Mg/Ca compared to the coccolith ooze.Accordingly, this difference could be related to the carbonate mineralogy since the core-top samples probably yield a higher content of high-Mg calcite than the low-Mg coccolith ooze.High-Mg calcites are typically lower in their Li isotope composition than low-Mg calcites(Day et al., 2021; Dellinger et al., 2018 and references therein).