| 17 Mar 2025
Surface area and Ω-aragonite oversaturation as controls of the runaway precipitation process in ocean alkalinity enhancement
Abstract. Ocean alkalinity enhancement (OAE) is a strategy for marine carbon dioxide removal that aims to increase the total alkalinity (TA) of seawater to sequester atmospheric CO2 in the form of dissolved inorganic carbon (DIC). An intense alkalinization of seawater resulting from OAE treatment could trigger a significant runaway carbonate precipitation process, which may lead to a loss of initially added alkalinity, thereby limiting its efficiency. Even under natural background aragonite saturation states, a continuous yet barely detectable loss of alkalinity is theoretically expected to occur in seawater. With the additional increase through OAE, time ranges to initiate an appreciable TA loss process could be reduced significantly. Therefore, predicting the alkalinity stability ranges might be a necessity for application scenarios. The main drivers of the precipitation process are i) the aragonite saturation state of seawater and ii) the available surface area for heterogeneous precipitation.
In this study, we refined the use of logistic functions to describe the temporal evolution of both drivers, with experimental datasets using natural seawater from the Raunefjorden (Bergen, Norway; Temp.: ~11 °C, Sal.: ~32.6). The observed patterns were then used to derive a process-based model for calculating TA-loss rates, focusing on the accelerated precipitation phase of the runaway process while considering saturation levels and available particle surface area. The formation of carbonate phases reduces seawater TA concentrations, inducing a delay or stopping the TA-loss process. In addition, the sinking of precipitated particles decreases the potential for further precipitation by reducing the available surface area in the system. To assess the impact of particle sinking on TA-loss, their shape and size distribution were determined. Under the environmental conditions presented here, TA-loss rates could be reduced by up to 30–40 % due to the sinking of particles, after just one day.
Integrating the proposed concepts into ocean models could enhance the accuracy of predictions regarding the fate of added alkalinity. Gaining insights into the evolution of the identified, seemingly stable TA levels can help prevent accelerated precipitation phases. Additionally, an understanding of particle sinking or dilution processes reducing the available reactive particle surface area is relevant to assess the efficacy and durability of OAE.
Competing interests: Jens Hartmann is consulting the Planeteers GmbH
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.
Suitner and colleagues tackle the important question of runaway precipitation as a potential effect of ocean alkalinity enhancement. They use experimental data to derive logistic functions to describe the progression and identify distinct phases of the runaway precipitation process. The aragonite saturation state and available nucleation surface area are identified as key drivers behind the precipitation process. The authors conclude by speculating about the impact of particle removal through sinking on the possible magnitude of runaway precipitation in natural systems.
While the authors present interesting data, I struggle to see how the study advances our understanding of runaway precipitation. My main criticism is that the logistic function presented as a conceptual model is not generalised and anchored to critical environmental factors. The coefficients for the functions are derived from each experimental dataset and are thereby not generalisable (which the authors acknowledge). No attempt is made to describe how the aragonite saturation state and available surface area for precipitation impact the coefficients of the logistic function, despite these factors being identified as key drivers behind the precipitation process in this very study. The equation also does not account for temperature, salinity, or the concentrations of known precipitation inhibitors. As such, the suggested logistic models have no predictive power outside of the specific experimental conditions from which they were extracted. I would encourage the authors to reassess their data and possibly conduct follow-up experiments to mathematically describe critical environmental and chemical factors, which would result in a true conceptual model. That runaway precipitation follows a logistic function was already concluded by Suitner et al. (2024), so I do not see this manuscript as a substantially novel contribution.
The methodology is only described very briefly and relies heavily on citations of previous work. Even if a full repetition of methods is not needed, a more extensive description is required in the current manuscript to explain the limits and results of the study.
Finally, I encourage the authors to publish the data either as a supplement to the manuscript or as a separate dataset.
Minor comments:
TA and alkalinity are used interchangeably throughout the text; be consistent.
L46-48: I am sure this is not the authors’ intention, but to me, the sentence suggests that NETs can be seen as an alternative to emission reductions. Consider rephrasing.
L91: Here, it seems like runaway precipitation is a desired phenomenon. Consider rephrasing.
Section 2.1: The description of the experimental setup is too brief; it is not enough to refer to Suitner et al. (2024). Please include information about initial TA concentrations and aragonite saturation states, number of replicates, samples collected, analysis methods and uncertainties, etc. The Gran Canaria setup should also be briefly introduced here, since seemingly new results from that experiment are presented in this manuscript.
L131-132: How was the assumed available active mineral surface area obtained? As described below?
L136: Filtered or unfiltered seawater?
L139: The BET surface area is a result, and should preferably include an uncertainty as well.
Section 3.1: The data from treatments that did not experience runaway precipitation are not presented. Please include at least in the supplementary material. Furthermore, please add a table with the coefficients of the logistic functions to the supplementary material.
L160-161: The removal of outliers needs to be described in detail here, as the entire manuscript is based on curve fitting. I also suggest including removed data points in Fig. 3 as empty symbols.
Figure 3: The filtered neq and filtered eq treatments show quite different patterns from the unfiltered neq treatment, and I think those figures should be shown in the main text.
Figure 3A-B: I suggest adding the TA concentration and aragonite saturation state of the initial seawater to the figure as well.
L192-194: Please include as a figure in the supplementary material.
L219: Confusing phrasing, there is a decrease in APP timespan with increasing initial TA.
Section 3.4: I find this section somewhat confusing. The coefficients b and c determine the shape of the logistic function, so it is only natural that they correlate well with the induction time and APP timespan (which are determined from the shape of the logistic function).
Section 3.5: The manuscript is generally well written; however, I struggled with this section. Please go over the text again.
Section 3.6: Here, it would be interesting to compare n and k between treatments, so I think it would be relevant to show data for filtered neq and filtered eq as additional panels in Figure 7 and Table 3.
L242: Please specify what is meant by “some treatments”.
L279: What is meant by “in dependence to a variable density”?
L271-285: I found it hard to follow this paragraph and to understand when particles observed by SEM, particles measured by FlowCam, and purely calculated values are referred to.
Figure 8: This figure is very busy and should be split up into multiple panels. The particle size distributions should be presented on their own with a clear x-axis. The y-axis is not easily understandable (does it represent both depth and sinking velocity, or depth divided by sinking velocity?), and I do not see what it is related to. Finally, the aragonite density is three orders of magnitude too high.
Section 4.1: As I outlined in my main comment, the current model is not predictable, except within the same environmental conditions. Since the model is not actually linked to environmental conditions, it will also not be possible to implement it in ocean models.
L316: Whitings are precipitation events, not a cause of precipitation.
L392: “section”, not “chapter”.
L399-400: Yes, and this should be reflected in the logistic equations.
L402-406: But k and n were derived from experiments in natural seawater, containing inhibitors, and based on the experimentally determined saturation state of aragonite (which is then used for the calculation of R). As such, these constants should include the potential impact of inhibitors. Is it not more likely that issues with accurately determining PSA are causing the difference?
L430-434: Again, this shows that temperature and salinity need to be considered in the logistic equation.