Reconstructing the carbonate chemistry of the past ocean is a central goal of paleoceanography. In this regard, the boron isotope composition of marine carbonates has become a promising and widely used proxy for the pH of ancient seawater. However, a major concern in the application of this proxy is that the isotopic composition of carbonates may have also been influenced by secular changes in the isotope value of seawater (δ11Bsw). The purpose of the present thesis is to address this potential source of bias and investigate the likely changes in δ11Bsw from the Cretaceous to the present.
We do this by first experimentally examining boron adsorption on clay minerals, one of the key mechanisms by which boron isotopes are fractionated in the ocean. Cleaned samples of kaolinite, smectite and illite were suspended for 48 hours in a fixed volume of fluid (with a solid-to-fluid mass ratio of 0.04), containing boron at a known concentration and isotopic composition. The ability of each clay mineral to sorb and isotopically fractionate dissolved boron was determined by measuring the boron content of the fluid before and after interaction with the clay minerals. This basic experimental procedure was repeated multiple times under varying levels of pH, ionic strength, dissolved inorganic carbon (DIC) and major ion concentrations to probe the sensitivity of boron adsorption to fluid chemistry.
When clays are suspended in a pure aqueous boron solution without any other dissolved constituents, boron adsorption is inefficient and approximately constant over a wide range of pH. This is because clay edge surfaces, where most adsorption takes place, remain neutral regardless of the acidity of the surrounding fluid, inhibiting complexation of either B(OH)3 or B(OH)4-. In contrast, clay edges can become charged if the ionic strength of the surrounding fluid is high. Complexation reactions of boron in seawater are therefore strongly dependent on pH, as the pH sets both charge state of clay minerals (positive, neutral, negative) and the speciation of boron (B(OH)3 or B(OH)4-). We identify maximum rates of boron adsorption for all clay minerals in alkaline seawater. At these conditions, adsorbed concentrations of boron can be between six to twelve times higher than those of dissolved boron in the surrounding fluid. Isotope fractionation associated with adsorption is maximized at a pH of 8 (approximately 20 ‰ between fluid and clay) but becomes negligible at very alkaline or acidic conditions (~ 0 ‰). Complexation modeling shows that the isotopic composition of adsorbed boron is consistent with a high affinity of B(OH)4- to the clay surface, in agreement with previously published experimental work. As a result, detrital clay minerals isotopically fractionate seawater by preferentially adsorbing isotopically light B(OH)4-, not because the adsorption process itself induces a specific isotopic fractionation of adsorbed boron.
Moreover, our results demonstrate that adsorption behavior of clay minerals is not clearly altered when adsorption occurs in seawater with a major ionic composition close to that of the Eocene or Cretaceous oceans. Long-term changes in seawater chemistry consequently had little impact on the boron adsorption flux, which was instead set over time by the total mass of sediment entering the ocean as well as its mineral assemblage. To further explore the impact of these parameters on the boron adsorption sink, we model a number of plausible erosion scenarios and the associated sediment discharge for the last 100 million years. The model takes into account changes in terrestrial climate and erosion patterns to predicts the abundances of five boron-adsorbing minerals (kaolinite, chlorite, illite, smectite and iron oxyhydroxides) in the global sedimentary discharge. Our results show that changes in the terrestrial climate had a limited impact on the global-scale abundance of these minerals because a warmer and wetter climate state in the past was balanced by a reduced land mass in tropical latitudes. On the other hand, changes in sediment provenance could have significantly altered the mineral assemblage. Previous studies suggested a lack of mountainous erosion in the Cretaceous and early Paleogene which, according to our erosion model, caused discharged sediment at these times to be composed of 40 % clay minerals (compared to 15 % today). In other words, the large acceleration in sediment discharge thought to occur over the course of the last 40 million years was dominantly comprised of minerals with a small specific surface area that do not adsorb boron in significant quantities. Our results therefore show that the capacity of detrital sediments to adsorb boron from seawater has remained similar to today for most of the Cenozoic.
We finally combine these findings on the boron adsorption sink with additional advancements made for other relevant parts of the Earth system to construct a new model for the evolution of δ11Bsw. Critically, our model also includes a self-consistent representation of the lithium isotope value of seawater (δ7Lisw), as both lithium and boron are mechanistically coupled because of their shared dependence on the same geologic processes. These processes include the riverine discharge of water and sediments, precipitation of carbonate and silica in the ocean, marine productivity, early sediment diagenesis and exchange reactions between seawater and the ocean crust. The model allows us to simulate the effect of proposed changes in various lithium and boron fluxes, and test their ability to match available seawater reconstructions. We find that seawater proxy data of both boron and lithium can be explained by a hitherto unrecognized long-term rise in the pH of soil porewaters, which we attribute to aridification of the global land surface. During the Cretaceous and Paleogene, soils were acidic (pH < 7) so secondary minerals forming in soils reincorporate only small amounts of solubilized lithium and boron with a modest isotope fractionation. Rivers at this time consequently delivered large amounts of isotopically light boron and lithium to the ocean. The model produces minimum δ11Bsw values of 36 and 37 ‰ for the late Cretaceous (85 Ma) and early Eocene (50 Ma), respectively. As the planet cooled and land surfaces dried in the latter half of the Cenozoic, soils became more alkaline (pH > 7) and the propensity of soil minerals to reincorporate and isotopically fractionate solubilized boron and lithium increased. This caused the river input and, accordingly, seawater compositions of both elements to become isotopically heavier over time. The δ11Bsw increased at a steady rate of 0.06 ‰ per million years from the early Eocene to the present and did not exceed its modern value of 39.6 ‰ at any time over the last 100 million years.
