Source Materials: The clay standards used were sourced from The Clay Minerals Society Source Clays Repository, and they have well-known chemical compositions. As the materials are a natural clay, these substances are assumed to be slightly impure, e.g. <7% Ti and/or Fe-oxides (Guggenheim, 2001). Montmorillonite and kaolinite were selected due to their environmental relevance, e.g. river plume systems in northern Gulf of Mexico (Grim and Johns, 1954; Pinsak, 1958; Scafe and Kunze, 1971; Sionneau et al., 2008; Ghaisas et al., 2021). SWy-3 is a Na-rich Montmorillonite sourced from Crook County, Wyoming, USA. KGa-2 is a Kaolinite sourced from Warren County, Georgia, USA. Before use, clays were sterilized for 30 minutes inside an Ultraviolet (UV) light box to degrade bacteria and avoid confounding results due to biotic activity. Samples were not ground, chemically- or heat-treated prior to experiments, as these processes are known to increase reactivity by decreasing surface crystallinity. Clays were not size fractionated as little variation in reactivity occurs when separating the < 2 micrometer (μm) size fraction from bulk clays (Siever and Woodford, 1973).

Seawater Collections: Natural seawater, which contains a complicated blend of dissolved organic and inorganic chemical compounds and species (e.g. macronutrients, trace metals) was used in these experiments as in previous studies (Siever, 1968a; Siever and Woodford, 1973). Bottom water (33 meters) was collected just above the sediment-water interface during Spring 2017 aboard the R/V Pelican (cruise ID: PE17-20) in the northern Gulf of Mexico (Krause et al., 2023); the sample location (28°30’ N, 90°50’ W) is ~150 kilometers west of the main Mississippi River outflow channel (Southwest Pass). The samples in this dataset were collected on May 6, 2017. Water was collected using 10-liter (L) Niskin bottles attached to an Ocean Instruments MC-900 Multi-corer; Niskin bottles were triggered concurrently with the multi-corer release and thus did not capture significant resuspended sediment from the Multi-corer landing. Shortly after recovery, the water was gravity-filtered directly from the Niskin through a 0.2 μm Whatman polycap filter into acid pre-cleaned 10 L carboys for storage (in the dark, room temperature) and future use. In the lab, the salinity (ppt) of the filtered seawater was checked and adjusted with 18.2 MΩ *cm deionized water to either 4 ppt or 32 ppt. The water was then sterilized with a UV Steripen (Adventure Opti Model) for 10 minutes to degrade bacteria and a subsample was analyzed for dSi using a commonly used spectrophotometric molybdate-blue method (e.g. Pickering et al., 2020).

Laboratory Experiments: Laboratory experiments were conducted from October 1 to October 14, 2018. In the lab, homogenized clay standard samples were weighed and mixed with 250 milliliters (mL) of adjusted seawater to achieve sediment dry-weight solid-to-solution ratios (SSR, grams per liter (g L-1)) of 0.1, 3 and 10; this was done for both clays (SWy-3 and KGa-2) and both salinities (4 ppt and 32 ppt). Procedural blanks were also prepared without the addition of sediment. A Teflon-coated stir bar and stir plate were used to keep the sediment suspended during subsampling. 7 mL of suspended sediment slurry was pipetted into 15 mL polypropylene plug seal centrifuge tubes and 10,000 DPM (167 Bq) of 32Si(OH)4 (104 Bq (µg Si)-1) was then added to each tube. Following the addition of 32Si to each tube, samples were sealed and immediately inverted to ensure the complete mixing of the radioisotope tracer into the slurry. Tubes were placed sideways on a shaker table and incubated in the dark at 15 degrees Celsius (°C) for 144 hours. Samples were vented every 12 hours to ensure oxygenation (i.e. exchange 8-mL ambient air headspace). Triplicate sacrificial sampling of each SSR occurred at the following time points; 0, 12, 24, 48, 72, 96, 120, and 144 hours as described below (i.e., for each clay type and SSR, 24 individual tubes sampled over time).

For each time point, individual samples were partitioned to quantify particulate and dissolved constituents. After mixing thoroughly using a vortex mixer, 3 mL of suspended sediment solution was filtered onto a polycarbonate membrane filter (25-millimeters (mm) diameter - 0.6 μm pore size). This filter was placed on a nylon disk planchette, dried overnight, covered with a mylar film, and both the filter and mylar were secured to the planchette using a nylon ring. Next, the tube was centrifuged for 10 minutes at 1500 x g. Following centrifugation, 2 mL of supernatant was subsampled in 1-mL aliquots. The first 1 mL was placed into a 20 mL plastic liquid scintillation vial. The second 1 mL was analyzed for dSi using a spectrophotometric molybdate-blue method as described above. The remaining 2 mL of solution was frozen for future analysis and contained ~57% of the original sediment, as liquid was removed without sediment from the sample tube for the two 1-mL aliquots. Blanks were treated in the same manner and run in tandem with each sediment (SWy-3, KGa-2) type totaling 48 - 4 ppt and 48 - 32 ppt blanks.

32Si Quantification: 32Si activity was quantified after the samples were aged into secular equilibrium with 32P, the short-lived (t½ = 14.28 days) and high energy (Emax = 1709 keV) daughter isotope of 32Si (t½ = 144 years, Emax = 227 keV). While the International System derived unit of radioactivity is the becquerel (Bq), we report activity in disintegrations per minute (DPM) (1 Bq = 60 DPM) as it is a functional unit that relates directly to instrument-derived counts per minute (CPM) and counting efficiency. For particulate 32Si, activity was measured using a GM-24 Multicounter (Risø DTU National Laboratory, Denmark) following Krause et al. (2011). For solution 32Si, sample activity was quantified using liquid scintillation counting (LSC). Given the quantity of samples, we chose a more commonly used and inexpensive liquid scintillation cocktail than traditionally used for LSC, e.g. Ultima GoldTM XR (Brzezinski and Phillips, 1997). To the 1 mL solution subsamples, 9 mL of EcoLumeTM (MP Biomedicals LLC, USA) was added, mixed, and total counts in solution were quantified after samples sat 2 hours (to reduce chemoluminescence) using a TriCarb 3110 TR liquid scintillation counter (Perkin Elmer).