The CKD and LKD were oven-dried at 40 °C for 72 h before experimental use. Grain size distribution was determined with a Malvern Mastersizer 3000 equipped with a Hydro LV dispersion system and operated at a stirring speed of 3000 RPM and no sonication prior to measurement. The dry solid phase density of the kiln dust was determined by measuring water displacement in a graduated cylinder (Dan-Asabe et al., 2013). To characterize the elemental composition, a 125 mg aliquot of dry, ball-milled (<2 µm) kiln dust was added to Teflon vessels and digested overnight at 90 °C in a heat block with a mixture of 1.5 mL HClO₄, 1 mL HNO₃, and 2.5 mL HF. After cooling, the caps were removed, and the vessels were heated to 140 °C to evaporate HF, leaving a gel-like residue. Next, 25 mL of 4.5 % HNO₃ was added to each vessel, which was then capped and heated at 90 °C for 2 h. The resulting solutions were diluted and analyzed for elemental composition using ICP-OES (Avio 500, Perkin-Elmer) at the GeoLab (Utrecht University, The Netherlands). Quality control measures included a blank, two certified river clay standards (ISE 921), and a duplicate sample. Recorded elemental concentration were between 98 % and 109 % of certified values. The mineralogical composition of the samples was determined in duplicate via quantitative X-ray diffraction (XRD) on a Bruker D8 Advance Eco diffractometer (Cu Kα, 40 kV, 25 mA) over 5–70° 2θ with 0.015° steps and 0.5 s per step. Samples were rotated at 10 rpm using a 10 mm variable divergence slit, and patterns were recorded with a LynxEye XE-T detector. Phase identification and quantification were performed using EVA and TOPAS (Bruker, V7). The BET specific surface area was determined by N₂ adsorption using a Quantachrome NOVA 2200E at the laboratory for Process Engineering for Sustainable Systems (KU Leuven, Belgium). The geometric specific surface area was calculated from the grain size distribution results (Appendix A Sect. A1). Finally, the grain morphology and presence of secondary minerals on kiln dust grains recovered from the dissolution experiments were analyzed via Scanning Electron Microscopy (SEM) using a Phenom ProX SEM equipped with an energy dispersive spectrometer (EDS), operated at an accelerating voltage of 15 kV. For SEM analysis, the dried kiln dust was mounted on aluminum (Al) pin stubs using double-sided carbon tape.

The dissolution kinetics of LKD and CKD in seawater were assessed in two separate experiments (Table 1), hereafter referred to as experiments I and II. In experiment I, the short-term dissolution behaviour was tracked via continuous pH monitoring. Filtered (<0.2 µm) seawater from the Eastern Scheldt (saline water body in The Netherlands adjacent to the North Sea; salinity 32.3 ± 0.5) was obtained from Stichting Zeeschelp (Kamperland, The Netherlands). The filtered seawater (FSW) was aerated for 24 h before use, yielding an initial seawater pCO₂ of 458–557 µatm. The initial seawater pH on a total scale (pH_T) and total alkalinity (AT) were analysed (as described in Sect. 2.3). For AT analysis, 55 mL seawater was collected in duplicate using a 60 mL plastic syringe, and filtered (0.8/0.2 µm polyethersulfone (PES) membrane). Based on preliminary tests, three concentrations of CKD (30, 130, and 309 mg kg⁻¹) and LKD (11, 48, and 113 mg kg⁻¹) were selected to target specific aragonite saturation states (ΩArg=3.6, 5.7 and 9.7) (Table 1). Experiment I was carried out in triplicate. Kiln dusts were weighed in small aluminium (Al) foil cups using a micro balance (XP26 Excellence Plus, Mettler Toledo) and then transferred to 200 mL polystyrene vials with polyethylene screw caps containing approximately 200 mL of FSW. These small-scale laboratory experiments provide a high-throughput, cost-effective first assessment of a material's suitability for OAE. Vials were weighed on an analytical balance (Sartorius TE3102S) to precisely determine the mass of added seawater. Plastic vials were rinsed with 0.5 M HCl and ultrapure water (PURELAB^® flex 3, Elga Veolia) before usage. The pH electrode was inserted through a hole in the vial cap, which fit tightly to minimize atmospheric CO₂ exchange. Vials were wrapped in tape to block light and contained minimal headspace. Subsequently, the pH_T of the suspension was measured every minute over a period 8 h (see pH_T measurement procedure below). Seawater temperature was kept constant at 20 °C during the incubation by means of a water bath (T100, Grant). Magnetic stirring was applied at a rate of 700 rotation per minute (RPM) to ensure good mixing of the suspension and to create optimal dissolution conditions. At the end of experiment I (after ∼ 8 h), vials were opened and 55 mL of seawater was collected in duplicate for AT analysis. Salinity was measured, and the samples were subsequently analysed for AT (see Sect. 2.3).

To assess the alkalinity generation potential and the possibility of secondary mineral formation, we conducted a second dissolution experiment with incubation periods of one and 15 d. The one-day (i.e. 24 h) incubation ensured complete dissolution of the reactive phases in the kiln dusts, while the 15-d incubation allowed for the verification of secondary mineral precipitation, in case this would occur. At the start of experiment II, clean 200 mL plastic vials were filled with 200 mL of FSW on an analytical balance (TE3102S, Sartorius). Different amounts of LKD (11–111 mg kg⁻¹) and CKD (30–308 mg kg⁻¹) were added to achieve a target ΩArg ranging from 3.6 to 9.7 (Table 1). A control containing only 200 mL of FSW without kiln dust was also included. Vials were closed tightly and had minimal headspace to minimize gas exchange with the atmosphere. Experiment II was conducted in triplicate at ambient room temperature (17.5–22.7 °C). Vials were subsequently incubated on bottle rollers (ThermoFisher Scientific) for 1 or 15 d at 14 RPM, a speed sufficient to keep particles suspended and ensure optimal dissolution conditions. Duplicate samples for dissolved inorganic carbon (DIC), dissolved metals, turbidity, and AT analysis were collected on both sampling days by drawing water with a syringe right after opening the vials (analytical procedures are described in Sect. 2.3). DIC samples were collected first to minimize the exposure time to the atmosphere. Nevertheless, some CO2 exchange inevitably occurred during the incubations because the vials contained a small headspace, meaning the solutions could partially re-equilibrate with the air. However, this also reflects natural deployment conditions, where atmospheric CO₂ exchange occurs alongside alkalinization, although at a slower rate. Samples for AT, DIC, and dissolved metals were filtered using 0.8/0.2 µm PES membrane filters. DIC samples (12 mL) were fixed with 10 µL of saturated HgCl₂ solution and stored at 4 °C in 12 mL exetainers until analysis. Dissolved metal samples were acidified with TraceMetal™ Grade 67 %–69 % nitric acid to a final acid concentration of 1.4 % (v/v) and stored at -20 °C prior until analysis. The remaining suspension in the incubation vials was filtered through a 0.2 µm polycarbonate membrane filter to collect solids, which were rinsed with deionized water and then oven dried at 40 °C in preparation for SEM-EDX analysis (see Sect. 2.1).

Seawater salinity and pH_T were measured with an Orion™ DuraProbe™ 4-Cell conductivity probe (Thermo Scientific) and Unitrode pH electrode (Metrohm) connected to an Orion Star A215 pH/conductivity meter. The pH electrode was calibrated using 35 salinity TRIS (2-amino-2-hydroxy-1,3-propanediol) and AMP (2-aminopyridine) buffers, and the pH_T was calculated following Dickson et al. (2007).

For AT analysis, samples were titrated with 0.1 M HCl using an automated titrator setup (888 Titrando with 814 USB Sample Processor, Metrohm). The titrant was calibrated with certified reference material (batch 209; OCADS). AT was derived from the titrant volume and electromotive force measurements recorded by the Unitrode pH electrode, using a non-linear least-squares method as described by Dickson et al. (2007). Blank FSW samples were analyzed at the start of the run, after every tenth sample, and at the end for quality control, yielding relative standard deviations (RSD) smaller than 1.1 %. The specific AT release ΔNAT(t) at a given incubation time t, and expressed in mmol AT per g of kiln dust, was derived via 4ΔNAT(t)=ATt-ATt0mFSWmKD Here, AT(t) represents the seawater alkalinity (mmol kg⁻¹) at a given incubation time, AT(t0) is the alkalinity (mmol kg⁻¹) of the blank FSW at the start of the experiment, mFSW is the mass of FSW added to the incubation vials (kg), and mKD is the mass of added kiln dust (g).

In experiment I, the AT was only measured at the begin and end of the experiment, and so only one ΔNAT(t) value can be calculated. To verify the observed alkalinity release, we calculated a theoretical value based on measured pH_T. To this end, the initial DIC of the seawater was first calculated from measured values of pH_T, AT, temperature, and salinity using the AquaEnv package in R with default settings (Hofmann et al., 2010). Under the assumption that DIC remained constant throughout the experiment, the corresponding AT was then calculated from the measured pH_T and fixed DIC. The difference between this calculated AT and the starting AT was used to compute ΔNAT(t), and the value at the end (8 h) was compared to the measured ΔNAT(t).

For experiment I, the fraction of reactive phases in the kiln dust that dissolved over time, χdiss(t) (%), was determined by normalizing ΔNAT(t) to the maximum experimentally observed specific AT increase (ΔNAT,max): 5χdiss(t)=ΔNAT(t)ΔNAT,max100 Seawater DIC concentrations were measured using a DIC analyzer (AS-C6L, Apollo SciTech) coupled to a trace gas analyzer (LI-7815, LI-COR). Measurements were repeated until the relative standard deviation (RSD) for at least three repeats was ≤0.1 %. DIC concentrations were determined using a calibration curve based on an internal standard solution (0.002 M NaHCO₃) adjusted to a salinity of 30 with NaCl and spiked with 0.05 % (v/v) saturated HgCl₂. This internal standard was calibrated against certified reference material (CRM; batch 209, OCADS). For quality control, CRM (batch 209, OCADS) was analyzed at the start and end of the sequence, and the internal standard was run at the start and after every eight samples. Quality control checks consistently yielded an RSD ≤0.25 %.

The remaining seawater carbonate chemistry parameters, including the aragonite and calcite saturation state and seawater pH_T, were calculated using the AquaEnv package in R (Hofmann et al., 2010). Measured AT, DIC, salinity, and temperature were given as input values, with all other parameters set to their default values.

Seawater samples for dissolved trace metal analysis were thawed and diluted 20-fold with 2 % (v/v) TraceSELECT™ HNO₃ (Honeywell Fluka) to a final volume of 10 mL. Before analysis, samples were spiked with 100 µL of an internal standard solution (10 ppm Y; Alfa Aesar) and analyzed using high-resolution inductively coupled plasma mass spectrometry (HR-ICP-MS; Agilent 7850) at the ELCAT group, University of Antwerp, Belgium.

Seawater turbidity was measured with a HI98713 ISO portable turbidity meter (Hanna Instruments) which was calibrated with four turbidity standards (<0.1, 15, 100, and 750 FNU, Hanna Instruments) before each use.

Saturation index (SI) values for kiln dust mineral phases were calculated using PHREEQC Interactive (version 3.7.3-15968) with the LLNL thermodynamic database (Parkhurst and Appelo, 2013). Saturation indices were converted to saturation states (Ω) according to Ω=10SI. Input parameters included measured seawater temperature, salinity, AT, and pH_T. Major constituent concentrations (Cl, Na, Mg, K, Ca, SO₄) were derived based on the average composition of natural seawater, scaled to measured salinity (Hem, 1985). Aragonite and calcite saturation states were not computed in PHREEQC but were instead calculated using the AquaEnv package in R, as described previously (Hofmann et al., 2010). AquaEnv uses carbonic acid dissociation constants from Lueker et al. (2000) and solubility product constants for CaCO₃ from Mucci (1983), which differ from the thermodynamic data used in the LLNL PHREEQC database to describe the carbonate system. We used the AquaEnv approach to remain consistent with the methodology commonly applied in most OAE studies.

Differences in seawater physico-chemistry across kiln dust concentrations and incubation times were assessed using two-way analysis of variance (ANOVA). The best fitting models were determined by the ANOVA and the lowest Akaike Information Criterion (AIC). Normality and homoscedasticity of residuals were evaluated both visually (via QQ and residual plots) and statistically (via Shapiro-Wilk and Levene's tests). Post-hoc pairwise comparisons were performed using estimated marginal means (EMMs) with Holm-adjusted p-values. Comparisons were conducted within each concentration, adjusting for incubation time, and vice versa. Data are presented as mean ± standard deviation (SD), unless otherwise specified. All statistical analyses were performed in RStudio (version 2024.12.0 + 467) using R version 4.3.3 (R Core Team, 2022).

XRD analysis revealed that the kiln dusts contained substantial amounts of calcite (CaCO₃; CKD ∼ 37 % and LKD ∼ 54 %; Table 2) and amorphous phases (∼ 34 % in CKD and ∼ 14 % in LKD). Furthermore, lime (CaO), portlandite (Ca(OH)2), quartz (SiO₂) and anhydrite (CaSO₄) were additionally present in both kiln dusts at concentrations of 0.1 %–20 %. Finally, CKD also contained 0.9 % sylvite (KCl), 9.5 % syngenite (K₂Ca(SO4)2 ⋅ H₂O), and 3.9 % Aphthitalite ((K,Na)3Na(SO4)2) (Table 2). Furthermore, minor phases including 2.1 % larnite/β-C₂S (Ca₂SiO₄), 0.68 % hematite (Fe₂O₃), and 0.56 % maghemite (γ-Fe₂O₃) were identified with low confidence in one of the duplicate CKD samples analyzed by XRD. Both materials exhibited a relatively wide grain size distribution (Table 2 and Appendix A Fig. A1), but CKD (D50 = 8.4 ± 0.1 µm) was significantly finer than LKD (D50 = 72 ± 4 µm). The observed elemental composition was generally in line with the XRD results, showing high calcium contents in both CKD (27.8 wt %) and LKD (44.9 wt %), which fall within the range previously reported for CKD (14 %–46 %) and LKD (20 wt %–49 wt %) (Collins and Emery, 1983; Pavía and Regan, 2010; Latif et al., 2015; Drapanauskaite et al., 2021; Dvorkin and Zhitkovsky, 2023). While LKD exhibited relatively low levels of trace elements, CKD showed elevated concentrations, particularly of Zn (0.65 wt %) and Pb (0.15 wt %) (Table 2). Further details on the elemental composition are provided in Appendix A Table A1.

Based on the mineralogical composition, the theoretical alkalinity release upon dissolution in seawater was calculated from the reaction stoichiometry. Saturation state analysis (Appendix B Fig. B2) show that seawater is undersaturated (Ω<1) with respect to most mineral phases present in the kiln dusts, with the exception of calcite and quartz. This indicates that dissolution is thermodynamically favourable, although it may be limited by kinetic constraints. Among the undersaturated minerals, anhydrite, sylvite, aphthitalite, and syngenite do not contribute to alkalinity upon dissolution in seawater. Furthermore, the hematite and maghemite present in the CKD are essentially insoluble under oxic conditions in natural seawater, and filtration would remove Fe-reducing bacteria capable of enhancing their dissolution (Canfield, 1989). In contrast, the dissolution of portlandite and lime each produces two moles of AT per mole (Eq. 3), while larnite (potentially present in the CKD) would yield four moles of AT per mole upon dissolution (Brand et al., 2019). Overall, complete dissolution of these phases corresponds to maximum alkalinity contributions of 8.8 mmol g⁻¹ for the LKD and 1.7 mmol g⁻¹ for the CKD, respectively.

The dissolution kinetics of CKD and LKD in seawater (salinity 32 ± 0.1, temperature 20 °C, initial pH_T of 8.05 ± 0.03) were investigated in experiment I over a period of 8 h at three different kiln dust concentrations under continuous stirring. A rapid increase in seawater pH_T was observed for both kiln dusts: about 69 ± 8 % of the total pH_T rise occurred within the first ∼ 10 min, after which the pH_T continued to slowly increase for the rest of the 8-h incubation (Fig. 1a). The final pH_T increase Δ pHT= pH_T (tend)- pH_T (t0) was greater for LKD than for CKD and increased with higher kiln dust concentrations (Fig. 1c).

Upon dissolution, the percentage of reactive phases that had effectively dissolved χdiss(t) rapidly increased with time, with 50 % of the reactive phases dissolved within 9.5 ± 3 min for LKD and within 9.9 ± 2.4 min for CKD (Fig. 1b). At the lowest kiln dust concentration, approximately 83 ± 9 % of the reactive phases in LKD and 72 ± 12 % in CKD had dissolved after one hour, which increased to 88 ± 7 % and 82 ± 17 % respectively after 8 h. At higher kiln dust concentrations, the percentage of dissolved reactive phases was generally lower (Fig. 1b). This effect is likely caused by secondary mineral precipitation, and not so much by reduced dissolution, as further discussed in Sect. 3.3.

Kiln dust dissolution in seawater resulted in a concomitant increase in alkalinity, ΔAT=AT(t)-AT(t0), which increased with higher kiln dust concentrations (Fig. 1d), and attained values ranging from 60 to 447 µmol kg⁻¹ for CKD and 110 to 416 µmol kg⁻¹ for LKD, depending on the applied kiln dust concentration. Notably, the relationship between ΔAT and kiln dust addition was non-linear and showed a saturating effect (Fig. 1d), suggesting a decreased specific alkalinity release at higher kiln dust concentrations.

In experiment II, we investigated the alkalinity release from LKD and CKD after 1 and 15 d of incubation across a range of kiln dust concentrations. Dissolution resulted in a change in seawater pH_T and AT that markedly varied with kiln dust concentration (Fig. 2a and b). After one day, the pH_T showed a non-linear (saturating) increase with the kiln dust concentrations for both LKD and CKD (Fig. 2a), consistent with the results of experiment I after 8 h (Fig. 1c). Yet after 15 d of incubation, the ΔpH_T curve showed a marked difference between LKD and CKD. For CKD, the ΔpH_T curve at 15 d showed a similar saturating shape as after 1 d, though with slightly decreased ΔpH_T values at higher kiln dust concentrations (suggesting a process at play that reduced pH_T). For LKD, the ΔpH_T curve at 15 d was entirely different compared to day 1, with a marked decrease in ΔpH_T at higher concentrations. For the highest LKD concentration examined (111 mg kg⁻¹), the pH_T after 15 d was almost the same as at the start (ΔpHT=0.17 ± 0.14) (Fig. 2a).

The observed changes in alkalinity (Fig. 2b) were congruent with those seen for ΔpH_T. After one day, ΔAT showed a monotonous increase with the CKD concentration, while for LKD, the ΔAT curve reached a maximum at 69 mg kg⁻¹ and decreased at higher application concentrations (Fig. 2b). The ΔAT curves for LKD and CKD obtained at 15 d of incubation were markedly different from those at day 1, showing reduced ΔAT values at higher kiln dust concentrations, indicative of a process that consumes alkalinity. Notable, for LKD, ΔAT became negative at higher concentrations (Fig. 2b), indicating a removal of alkalinity compared to the initial situation.

The specific alkalinity release quantifies the alkalinity release per mass of kiln dust added, and generally decreased for higher concentrations (Fig. 2c). The highest values for the specific alkalinity release were obtained at the lowest kiln dust concentrations applied (21 mg kg⁻¹ and lower for LKD, 89 mg kg¹ or lower for CKD). The maximum specific alkalinity release for LKD (8.02 ± 0.53 mmol g⁻¹) was more than three times higher than that of CKD (2.38 ± 0.16 mmol g⁻¹). Moreover, the LKD value was in good agreement with the theoretical prediction (8.8 mmol g⁻¹; see Sect. 3.1), while the CKD value deviated more substantially from the theoretical estimate (1.7 mmol g⁻¹). The specific alkalinity release after 15 d showed a clear difference with day 1, which was more pronounced for LKD compared to CKD. At the low application concentrations, the specific AT release after 15 d of incubation was not statistically significantly different from the release after 1 d (Fig. 2c), thus indicating that all AT-generating reactive phases had dissolved within the first day and no alkalinity removal took place. Yet, the specific AT release decreased substantially at higher kiln dust concentrations, with a more pronounced decline for LKD compared to CKD. In the case of LKD, the specific AT release became even negative after 15 d of incubation, thus indicating overall alkalinity consumption rather than production (Fig. 2c).

To verify whether secondary carbonate precipitation could be responsible for the observed alkalinity consumption, we calculated the seawater aragonite saturation state (ΩArg). The increase in seawater total alkalinity (AT) from kiln dust dissolution led to a corresponding rise in aragonite saturation state (ΩAg) from an initial value of 3.0 ± 0.1 at the start of the experiment II to 9.3 ± 0.3 and 10.3 ± 0.1 after 1 d of incubation at the highest concentrations of LKD and CKD, respectively (Fig. 2d). Moreover, a significant decrease in ΩArg was observed at day 15 compared to day 1 for LKD concentrations above 21 mg kg⁻¹ and for CKD concentrations above 89 mg kg⁻¹, suggesting alkalinity scavenging by secondary mineral precipitation. This decrease became more pronounced at higher application concentrations and was notably greater for LKD compared to CKD. For the highest LKD concentration investigated (111 mg kg⁻¹), ΩArg after 15 d (2.5 ± 0.6) was not significantly different from the initial ΩArg value (Fig. 2d).

The morphology of kiln dust particles was compared before and after chemical weathering (i.e., fresh material versus samples retrieved after 15 d in experiment II). Fresh lime kiln dust (LKD) consisted of a heterogeneous mixture of particles of varying size and shape, dominated by small (<10 µm), irregular, calcium-rich particles (∼ 10–20 at %) (Fig. 3a). In weathered LKD, the abundance of these fine particles decreased (Fig. 3c, e), while larger particles developed rough surface texture (representative particle indicated by red arrow in Fig. 3c), suggesting dissolution of surface phases. After 15 d of incubation at the highest LKD concentration, most particles were extensively coated with bundles of needle-like, calcium-rich precipitates (Fig. 3e).

Fresh cement kiln dust (CKD) also contained irregularly shaped particles of various sizes (Fig. 3b). In addition, spherical fly ash particles of varying diameters were prominent in both fresh and weathered CKD samples and showed no signs of weathering during experiment II (red arrows in Fig. 3b, d, f). In contrast to LKD, weathered CKD particles showed neither roughened surfaces nor secondary calcium carbonate precipitation, regardless of concentration or incubation duration (Fig. 3d, f).

The turbidity in solution after 1 and 15 d increased linearly with the applied kiln dust concentration, as expected. At equivalent application concentrations, turbidity was ∼ 2.4 times higher for CKD than for LKD, consistent with the finer grain size of the CKD (Fig. 4a). After 15 d of incubation, seawater turbidity rose from 0.33 ± 0.10 FNU in the initial filtered seawater to 281 ± 3 FNU at the highest CKD concentration and 52 ± 3 FNU at the highest LKD concentration. Turbidity was slightly but statistically significantly greater on day 15 compared to day 1 at higher kiln dust concentrations (>253 mg kg⁻¹ for CKD; >69 mg kg⁻¹ LKD), possibly due to fragmentation of unreacted kiln dust particles and/or the formation of secondary calcium carbonate providing fine particles in solution (Fig. 4a).

The accumulation of dissolved Fe, Ni, Cu, Zn, As, and Pb was low in both the LKD and CKD incubations and showed no clear dependence on kiln dust concentration (Appendix B Fig. B3). So trace metal release was limited, apart from vanadium (V) in the CKD incubation, which linearly scaled with the kiln dust concentration (Fig. 4b). After 15 d, V accumulation reached 0.51 ± 0.09 µmol kg⁻¹ at the highest CKD concentration and 0.008 ± 0.001 µmol kg⁻¹ at the highest LKD concentration, respectively. These values agreed relatively well with the expected accumulations of 0.63 and 0.0033 µmol kg⁻¹, based on the elemental composition and assuming that V belongs to the dissolvable fraction of the kiln dusts (Appendix B Fig. B4). The dissolved V accumulation after 15 d did not differ significantly from that on day 1, suggesting that V is not involved in secondary reactions. For CKD, concentration-dependent accumulation of Al, Cr, and Mn was also observed, though the accumulation represented only a small fraction of what was expected based on complete dissolution of reactive phases: 2 %–11 % for Al, 24 %–35 % for Cr, and 0.6 %–2 % for Mn (Appendix B Figs. B3–4).

The dissolution kinetics of LKD and CKD were studied in natural seawater (salinity: 32.3 ± 0.5; temperature: 17.5–22.7 °C) under continuous stirring. LKD mainly consists of calcite (CaCO₃), with smaller amounts of quartz, portlandite, lime, anhydrite, mullite, and dolomite (Strydom et al., 1996; Ban et al., 2022). CKD is more compositionally complex, typically containing calcite along with various sulfates, chlorides, silicates, and aluminates, including belite, aphthitalite, spurite, ettringite, arcanite, and ferrite (Ayman et al., 2004; Siddique and Rajor, 2012; Beltagui et al., 2017; Adekunle, 2024; Lee and Choi, 2024; Nikolov et al., 2025). The compositional complexity of kiln dusts underscores the need for detailed mineralogical and chemical characterization to properly assess the CDR potential and environmental risks in OAE applications.

Both materials generated significant alkalinity upon dissolution in seawater, with maximum values of 8.02 ± 0.53 µmol mg⁻¹ for LKD and 2.38 ± 0.16 µmol mg⁻¹ for CKD after 24 h. This alkalinity originated from a fraction of reactive phases contained within the kiln dust (25 % by mass in LKD; 29 % in CKD), which we estimated based on the mineral components that are unstable in seawater. Consequently, both LKD and CKD contained a substantial amount of unreactive phases that remained inert over the 15-d duration of our short-term experiments. In preliminary tests, replacement of the overlying seawater after 15 d did not result in further dissolution or alkalinization. The residual fraction in LKD and CKD consisted primarily of calcite, which is supersaturated in surface seawater (Appendix B Fig. B2G), preventing its dissolution (Sulpis et al., 2021).

The experimentally observed alkalinity release from LKD (8.02 ± 0.53 mmol g⁻¹) was slightly lower than the theoretical value (8.8 mmol g⁻¹), whereas CKD released substantially more alkalinity (2.38 ± 0.16 mmol g⁻¹) than predicted from its mineralogical composition (1.7 mmol g⁻¹). In LKD, alkalinity release was fully attributed to the dissolution of portlandite (Ca(OH)₂) and lime (CaO), whereas in CKD these phases explained only about half (54 ± 3 %) of the observed alkalinity release. Calcium silicates (e.g. larnite Ca₂SiO₄) are also alkalinity-generating phases that occur in minor amounts in CKD. They originate form the raw materials used in cement production (e.g. iron ore, clay, or shale) and exhibit a relatively high reactivity in water (Brand et al., 2019; Adekunle, 2024). Dissolution of the larnite present in our CKD sample (∼ 2.1 %) could account for 17 ± 1 % of the observed AT release, which hence provides a substantial additional contribution. The remaining ∼ 29 % of the alkalinity released from CKD likely originated from dissolution of amorphous phases, including (partially dehydrated) clay minerals, reactive amorphous silica, and kiln-derived materials such as fly ash or slag (Khanna, 2010; Pavía and Regan, 2010). ICP-OES analysis revealed 8.0 % Ca, 2.7 % K, 0.7 % Na, 0.9 % S, 1.6 % Fe and 3 % Al that were not accounted for by the crystalline phases detected via XRD (Tables 2 and A1). This suggests the possible presence of amorphous calcium aluminosilicates, alkali sulfates, and poorly crystalline CaO or Ca(OH)₂, which may have contributed to the remaining alkalinity upon dissolution in seawater (Hu et al., 2024; Nikolov et al., 2025). There is a possibility that some reactive phases remained sequestered in larger particles and did not react with seawater on the time scale of the experiment. However, this fraction is likely minor, since the experimentally observed alkalinity release was close to, or even exceeded, the maximum theoretically predicted values from the mineralogical composition.

The dissolution of kiln dusts was rapid, with 65 %–92 % of the reactive phases dissolving within the first hour of incubation under continuous stirring (Fig. 1b). This estimate assumes a constant DIC concentration over time, although values could have increased due to CO2 uptake from the (limited) vial headspace or decreased through secondary aragonite precipitation at higher application concentrations. Therefore, the χdiss values reported (Fig. 1b) should be considered best estimates based on available data. Nonetheless, the rapid alkalinity release highlights the potential of kiln dusts for OAE. Using our measured particle size distribution and assuming particle sinking follows Stokes' law, all CKD particles and the majority of LKD particles (85 ± 2 % v/v) will remain in the surface ocean mixed layer (assumed to be 200 m) for at least one hour, thus allowing sufficient time for most reactive phases to dissolve and generate alkalinity (Appendix A Sect. A2). However, ocean turbulence and particle aggregation induced by biological exudates can significantly accelerate particle sinking (Yang and Timmermans, 2024) and affect kiln dust dissolution kinetics. Consequently, further research is needed to quantify kiln dust dissolution rates and settling velocities under a range of scenarios that more closely mimic in situ hydrodynamic conditions. This information is essential for the careful selection of application areas, ensuring that dissolution of the reactive phases occurs within the ocean mixed layer.

Alkalinity addition to seawater should avoid triggering secondary precipitation reactions that consume alkalinity, as these reduce the overall efficiency of OAE. Specific AT release significantly decreased at concentrations above 21 mg kg⁻¹ for LKD and 89 mg kg⁻¹ for CKD, with greater reductions at higher kiln dust concentrations and longer incubation times (Fig. 2c). Furthermore, the decrease was stronger for LKD relative to CKD at equivalent aragonite saturation states (Fig. 2c and d). The observed reduction in specific AT release can be attributed to secondary mineral precipitation, as indicated by the reduction in seawater DIC concentrations (Appendix B Fig. B1) and the formation of Ca-rich needle-like structures on weathered LKD grains (Fig. 3e). These needles resemble the early developmental stage of aragonite precipitates, as described by Suitner et al. (2024). No significant AT loss occurred at ΩArg values of 5.1 ± 0.07 for CKD and 4.5 ± 0.1 for LKD, but significant loss was observed at higher kiln dust concentrations. This aligns with the previously documented ΩArg=5 threshold for secondary aragonite precipitation, when fine-grained (<63 µm) quick lime (CaO) or slaked lime (Ca(OH)₂) powder from a chemical and industrial supplier are added to natural seawater at 35 salinity (Moras et al., 2022). While the ΩArg precipitation threshold was similar for LKD and CKD, the CKD treatment showed a lower precipitation rate, possibly because CKD has a lower content of Ca-rich phases (e.g., calcite, lime and portlandite), which can serve as nucleation sites for aragonite precipitation (Pan et al., 2021; Moras et al., 2022; Suitner et al., 2024). Prolonged exceedance of critical saturation thresholds can trigger “runaway CaCO₃ precipitation”, leading to a net AT loss, as seen at the highest LKD concentration after 15 d (Fig. 2b) (Moras et al., 2022). Under natural conditions, freshly precipitated aragonite may redissolve after dilution in the ship's wake, especially when not yet fully crystallized, recovering some of the lost alkalinity due to secondary precipitation (Hartmann et al., 2023). Aragonite precipitates that settle onto the sediment at the deployment site may undergo further metabolic dissolution provided that geochemical conditions are favourable, offsetting the earlier alkalinity loss (see Sect. 4.4). However, these fine-grained precipitates could also disperse far from the deployment site, thus complicating monitoring, reporting, and verification (MRV) of CDR via kiln-dust-based OAE. So, despite that some secondary aragonite may redissolve, its formation is best minimized to maximize the alkalinization potential. Based on our temporal dissolution data (Fig. 1a, b), it is recommended to adjust the OAE dispensing and deployment procedure in such a way, that dilution to ΩArg<5 occurs within minutes as to minimize secondary mineral precipitation.

Mineral-based OAE shows promise as a CDR technique, but its effects on seawater carbonate chemistry, turbidity, and trace element concentrations could lead to adverse ecological impacts that need to be mitigated (Bach et al., 2019; Flipkens et al., 2021). For ocean liming, rapid mineral dissolution can cause acute spikes in pH_T and alkalinity right after discharge, raising potential environmental concerns (Caserini et al., 2021; Varliero et al., 2024). In our study, CKD and LKD caused fast, concentration-dependent seawater pH_T increases (Fig. 1a). Model predictions indicate that pH could rise by 1 to 1.5 units for several minutes during ship-based ocean liming (Caserini et al., 2021), which may have an impact on marine life if pH exceeds 9 (ANZECC and ARMCANZ, 2000; Pedersen and Hansen, 2003; Camatti et al., 2024). To avoid temporary exceedances of pH 9, CKD concentrations should stay below 343–502 mg kg⁻¹, and LKD below 102–149 mg kg⁻¹ under average surface seawater conditions (AT=2350 µmol kg⁻¹, DIC = 2100 µmol kg⁻¹, salinity = 35, temperature = 10–25 °C). Application concentrations must be tailored to local seawater geochemistry at the deployment site to prevent exceeding the pH 9 threshold.

Seawater turbidity rose linearly with kiln dust concentration and was greater for CKD than LKD at equivalent doses, consistent with the finer grain size of CKD (Fig. 4a). Increased turbidity could reduce primary production by obstructing light (Cloern, 1987; Köhler et al., 2013), and impair feeding efficiency in marine suspension feeders (e.g. bivalves, sponges, and tunicates) (Cheung and Shin, 2005; Bell et al., 2015) and visual foragers (e.g. most marine fish and mammals) (Lowe et al., 2015; Lunt and Smee, 2020). Additionally, the sinking rate of organic carbon to the deep sea could be enhanced by the adsorption of organic molecules onto suspended particles (Santinelli et al., 2024), which may affect ecosystem carbon cycling, but also would provide additional CDR. Turbidity guidelines are designed to protect marine life from harmful increases: for example, in Canada, seawater turbidity should not increase by more than 8 NTU over 24 h in clear water, or 5 NTU at any time in already turbid water (8–50 NTU background) (Singleton, 2021). To stay within these limits, ambient CKD concentrations must remain below 9.7 mg kg⁻¹ in clear water and 6.1 mg kg⁻¹ in turbid water, while LKD concentrations should stay below 23.7 and 14.8 mg kg⁻¹, respectively (Fig. 4b). In real applications, kiln dust will be rapidly mixed into much larger volumes of surface water, meaning that the allowable concentration in the input stream will depend on the discharge rate, the intensity of local turbulence, and kiln dust settling time (which is primarily determined by initial particle size). Accurate numerical modelling to determine suitable discharge rates therefore requires detailed knowledge of both the environmental conditions at the deployment site and the behaviour of kiln dust particles under varying hydrodynamic conditions (Fennel et al., 2023; Yang and Timmermans, 2024). This information is essential to extrapolate small-scale laboratory results to realistic field scenarios and ensure that concentrations remain below guideline levels.

At the turbidity thresholds, the maximum seawater alkalinity enhancement would range from 119 to 190 µmol AT kg⁻¹ for LKD, and only 15 to 23 µmol AT kg⁻¹ for CKD. The corresponding increases in pH_T (assuming AT=2350 µmol kg⁻¹, pCO2=420 µatm, salinity = 35, and temperature = 10–25 °C) are ΔpHT=0.18–0.30 for LKD and ΔpHT=0.002–0.004 for CKD, both remaining well below the threshold of ΔpHT=∼ 0.9 (rise up to pH_T 9). Moreover, the carrying capacity of natural coastal and shelf ecosystems appears to be large enough to execute LKD- and CKD-based OAE within the existing turbidity constraints. For example, the North Sea has a total volume 54 000 km³ and an average residence time of ∼ 1 year (Lee, 1980; Liu et al., 2019). A hypothetical one-time application of LKD-based OAE across the entire North Sea at the maximum level of 119 µmol AT kg⁻¹ for LKD, would require 823 Mton of LKD, which is ∼ 28 times larger than the global annual LKD production rate (∼ 29 Mt yr⁻¹).

The dissolution of alkaline minerals can release trace metals into the environment, which may be beneficial or toxic to marine life (Bach et al., 2019; Flipkens et al., 2021). CKD contained notable amounts of Zn and Pb, while LKD had generally low trace element levels (Appendix A Table A1). Trace metal content in kiln dusts varies with raw materials, fuels, and kiln operations, and is typically higher and more variable in CKD (Siddique and Rajor, 2012; Nyström et al., 2019). In experiment II, metal release from LKD was limited, whereas CKD showed concentration-dependent release of V, Cr, Mn, and Fe (Fig. 4b; Appendix B Fig. B3). Regulatory guidelines exist to protect aquatic life from trace metal toxicity. For example, Tulcan et al. (2021), proposed a seawater V guideline of 0.022 µmol L⁻¹, which would require CKD concentrations to remain below 14.1 mg kg⁻¹ to avoid exceedance. Higher application rates may be permissible under other guidelines for V, Cr, or Mn. Residual kiln dust mixing with surface sediments may elevate metal levels, particularly Zn and Pb from CKD. Sediment Quality Guidelines (SQGs) aim to protect benthic ecosystems (Hübner et al., 2009; Simpson and Batley, 2016). Assuming full mixing in the top 10 cm of the sediment, up to 1.4 kg CKD or 74.8 kg LKD per m² could be applied to pristine sediments without exceeding the strictest marine SQG of 30.2 mg kg⁻¹ for Pb (Appendix C). Using a kiln dust bulk density of 0.62 g cm⁻³ (Nikolov et al., 2025), this corresponds to an applied layer of approximately 0.2 cm for CKD and 12 cm for LKD. The exact layer thickness would of course be dependent on the specific kiln dust properties and local grain packing. These estimates indicate that burial risk for benthic organisms is small for CKD, but could be considerable for LKD, since the deposition of a cm-thick layer could substantially impact the resident benthic infauna and epifauna. Moreover, applying LKD at this scale is also not advisable because it may lead to changes in habitat suitability (e.g., grain size, permeability, organic carbon content) (Speybroeck et al., 2006; Flipkens et al., 2024) and alter geochemical sediment processes (see Sect. 4.4). Overall, these findings underscore the need for ecotoxicological testing and cautious application of kiln dust to avoid ecological harm.

Both LKD and CKD contained a significant amount of unreactive phases (75 and 71 wt %, respectively) that remained inert over the 15-d experimental time scale. In coastal and shelf environments, this residual material would rapidly settle to the seafloor. The residual fraction consists primarily of CaCO3 phases (52 % in CKD and 72 % in LKD). When residual CaCO₃ or freshly precipitated secondary CaCO₃ become mixed into the seabed through local hydrodynamics and bioturbation, porewater acidification resulting from microbial degradation of organic matter can trigger metabolic CaCO₃ dissolution (Rao et al., 2012; Kessler et al., 2020). This process takes place under oxic conditions and produces 2 moles of alkalinity per mole of dissolved CaCO₃. 6CaCO3+CO2+H2O→Ca2++2HCO3- In anoxic environments, organic matter mineralization generates more AT than DIC, quickly increasing Ωcal and thereby inhibiting dissolution (Morse and Mackenzie, 1990; Burdige, 2006). If kiln dusts would be applied to continental shelf waters overlying sediments with potential for enhanced carbonate dissolution, including organic-rich, carbonate-poor marine sediments (Lunstrum and Berelson, 2022; Dale et al., 2024; Biçe et al., 2025; Fuhr et al., 2025) or coastal upwelling zones (Harris et al., 2013; Fuhr et al., 2025), weathering of all calcite in the residual kiln dust could additionally produce a maximum of 10.8 mmol AT g⁻¹ LKD and 7.4 mmol AT g⁻¹ CKD. However, large-scale fining of sediment with kiln dust could reduce sediment properties, such as the permeability, solute exchange rates, and oxygen penetration depth (Speybroeck et al., 2006; Ahmerkamp et al., 2017). The latter would limit the zone in which metabolic CaCO₃ dissolution can occur. Relatively high CaCO₃ concentrations may further reduce the dissolution efficiency (i.e. dissolution rate per amount of CaCO₃ added) (Dale et al., 2024). Additionally, ecological impacts may arise through changes of the solid sediment matrix and modifications of porewater conditions (see Sect. 4.3). The potential for enhanced sedimentary alkalinity generation via residual kiln dust addition to organic-rich, carbonate-poor marine sediments therefore warrants further experimental investigation. If fully realized, the total alkalinity release potential (immediate dissolution and metabolic CaCO₃ dissolution) could reach up to 18.8 mmol g⁻¹ for LKD and 9.8 mmol g⁻¹ for CKD. By contrast, in open-ocean applications, the residual material would settle to the deep seafloor, where metabolic dissolution would occur in waters isolated from the atmosphere and thus would not contribute to CDR on relevant (year–decade) timescales.

Achieving the Paris Agreement targets will require rapid and deep CO₂ emission reductions, complemented by 12–15 Gt CO2 yr⁻¹ of carbon removal by 2100 (Rockström et al., 2017; Minx et al., 2018). Kiln dusts could potentially contribute to this CDR portfolio. Currently, most kiln dust is landfilled (El-Attar et al., 2017), while the remainder is recycled for applications such as soil stabilization, concrete mix, chemical treatment, ceramics, and brick manufacturing (Al-Bakri et al., 2022). CKD can replace 5 %–10 % of cement, or up to 20 % when combined with pozzolanic materials (fly ash, slag), reducing waste, lowering raw material and energy consumption, and cutting CO₂ emissions by a similar percentage (Huntzinger and Eatmon, 2009; Al-Bakri et al., 2022). While this represents the ideal use of CKD in terms of CO2 mitigation, high levels of alkalis, sulfate, and chloride limit the extent to which CKD can be recycled in cement manufacturing (Al-Bakri et al., 2022). Carbonation of kiln dusts, involving the reaction of metal oxides with CO₂ to form solid carbonates, has been proposed as an alternative CO₂ sequestration method (Huntzinger et al., 2009; Adekunle, 2024): 7CaO+CO2→CaCO3 This process captures 1 mol CO₂ per mol metal oxide, which is less than what can be achieved via CaO hydration and subsequent Ca(OH)2 dissociation in seawater (∼ 1.68 mol CO₂ mol⁻¹ metal oxide). In landfills, both processes naturally occur when kiln dust is exposed to rainwater (Sreekrishnavilasam et al., 2006). However, limited water availability in large kiln dust piles promotes secondary precipitation of carbonates or clay minerals, reducing the effective CO₂ sequestration. As such, the ad hoc CDR effect that occurs during landfill disposal of LKD and CKD remain uncertain. Similarly, application of kiln dust to agricultural soils for enhanced weathering purposes (as an alternative to primary mined rocks such as basalt or dunite) could contribute to CO2 removal, though restricted water availability may again increase the risk of secondary mineral formation (Buckingham and Henderson, 2024; Xu and Reinhard, 2025). The focus of this study is the usage of kiln dusts via OAE in natural marine environments. Alternatively, kiln dusts could also be used in reactor-based OAE approaches, such as accelerated weathering of limestone (Rau and Caldeira, 1999; review in Huysmans et al., 2025). These methods allow fast, controlled, and easily monitored alkalinity addition, but require higher energy inputs and dedicated infrastructure compared to ship-based distribution in natural environments (Rau and Caldeira, 1999; Rau et al., 2007; Huysmans et al., 2025).

In this study, short-term weathering of LKD and CKD in seawater produced up to 8.02 ± 0.53 and 2.38 ± 0.16 mol of alkalinity per kg of source material, respectively (Fig. 2c). On average, 1 mol of added alkalinity sequesters 0.84 mol CO2 in surface ocean waters (Schulz et al., 2023), thus resulting in 297 ± 20 g CO2 kg⁻¹ LKD and 88 ± 6 g CO2 kg⁻¹ CKD. With current global annual production of approximately 29 Mt for LKD and 287 Mt for CKD (CEMBUREAU, 2024; USGS, 2025), their maximum carbon dioxide removal potential via dissolution in seawater, assuming an average of 0.07 t of kiln dust produced per tonne of lime or cement (Al-Refeai and Al-Karni, 1999), is approximately 8.7 ± 0.6 Mt CO2 yr⁻¹ for LKD and 25 ± 2 Mt CO2 yr⁻¹ for CKD. Additional CO2 uptake through metabolic calcite dissolution in coastal and shelf sediments could potentially further increase this to 13.4 Mt yr⁻¹ for LKD and 57 Mt yr⁻¹ for CKD, although the effectiveness and time scaling of this process are uncertain. Cumulatively, this would amount to 1 and 4.3 Gt CO2 by 2100, assuming constant kiln dust production rates and complete utilization for OAE from 2025 onwards. While significant, this represents only 1.9 %–2.8 % of the 2.5–3.7 Gt CO2 emitted annually by the cement and lime industries (Simoni et al., 2022; Cheng et al., 2023). Therefore, decarbonizing these sectors remains the top priority for effective climate change mitigation (Simoni et al., 2022; Barbhuiya et al., 2024).

The CDR estimates presented here are upper-bound values, assuming that all globally produced LKD and CKD will be used for OAE, and that production rates remain constant throughout the 21st century. In practice, some kiln dust will be used for other economically viable applications (Al-Bakri et al., 2022), while on the other hand, cement demand is projected to increase by 12 %–23 % by 2050, which will increase kiln dust production (IEA, 2018). CO2 emissions from transporting kiln dust to the ocean were not considered, but their impact on net CDR is likely minor if deployment occurs near production sites with minimal road transport (Foteinis et al., 2022). These CDR estimates also assume full atmospheric CO2 equilibration of AT-enriched surface waters (γCO2≈0.84), while actual values in coastal regions are possibly lower (γCO2≈0.65–0.8), due to alkalinity transport to the deep ocean without prior atmospheric exchange (He and Tyka, 2023). Therefore, application should focus on continental shelf seas, especially those with organic-rich, carbonate-poor sediments (Lunstrum and Berelson, 2022; Dale et al., 2024), to promote metabolic CaCO₃ dissolution and maximize the CDR potential. Importantly, only one specific type of CKD and LKD were tested in this study, and mineralogical and chemical composition can vary significantly with the production process, thus affecting the CDR potential (Pavía and Regan, 2010; Siddique, 2014; Drapanauskaite et al., 2021). Tailoring application concentrations to site-specific conditions and material properties is therefore essential for safe and effective deployment.