General comments

The manuscript by McLagan et al. is well-written and describes a well-constructed study. The experiments detailed in this manuscript are well documented, frequently sampled, and analyzed to answer important research questions for Hg contamination, transport, and sorption/desorption. The resulting data is valuable and provides noteworthy contributions for future Hg research to build on. The high sorptive capacity of Hg in these sediments is an exciting find. While data on Hg sorption and desorption and presented retardation and distribution coefficients for Hg at contaminated concentrations in environments with low DOM may be the main selling points of the study, the data on Hg isotope fractionation associated with processes of sorption and desorption and evidence of Hg reduction in oxygenated soils are also noteworthy take-aways. The manuscript is recommended for publication in SOIL with minor suggested edits.



A general suggestion for improvement relates to the possible comparisons between the experiments and the real-world sediments sampled for this study. It is detailed that 74% of the material was discarded during the sieving process (Ln 130-131). In addition, there is a mention of an “optimal density“ for the column packing (Ln 147), and the Authors went through several packing methods in the preliminary experimental phase (S1) where the density appears to have been a major factor in the consideration. This raises the following questions: 1) what the target density is, and how does it relate to the real world, e.g., what system the experiments now can be said to most closely mimic, and 2) what the effect of sieving off ¾ of the total weight could have for translating the findings to real-world systems. Discussing this would be a welcome inclusion, as would a site description of where the samples were collected. 

It is also suggested that the text be edited in a few instances to improve the conciseness and structure. This includes the beginning of the introduction, which starts very broadly with little direct relevance to the study, and the section between Ln 462-475, which can be shortened to a few sentences to keep the spotlight on the major findings in the manuscript.

Specific Comments

Ln 172-174: The tested HgCl2 concentrations are presented as estimates of original concentrations during the years of kyanization activity despite being three magnitudes below this concentration level. It is clear that this is a rough estimate and that lower concentrations of stock solutions were not feasible due to experimental constraints. However, can any data or calculations be referenced to support estimating a 1000x concentration loss in the porewater over 5-6 decades?

Ln 181, Table S2.2: Based on the 24 and 48-hour analyses, a decision was made to let the columns equilibrate for seven days before starting the experiment. The analysis shows that more than two days are needed to achieve equilibration. Is there any data to back up that seven days should be enough?

Ln 242: How can the Authors know that no Hg0 loss occurs when analysis is performed on wet samples? A reference would be appreciated.

Ln 320-330, Figure 2: With the eluate concentration only reaching 91% of the stock solution concentration, the Authors state that EXP1 likely did not reach equilibrium. Despite this, the Freundlich model fit to EXP1 indicates that the maximum totHg eluate concentration was reached when ca 5.7 L solution had passed through, after which the increase in Hg in the eluate ceased. If EXP1 did not reach equilibrium, would a maximum eluate concentration be passed? Or, is there reason to doubt the appropriateness of the model for the sorption phase in EXP1?

Ln 557: How can it be known that Hg0 exceeded solubility after the 25% breakthrough if Hg0 was only qualitatively measured? If this is a speculation, that should be indicated more clearly.

Ln 571-575: While I support the conclusions, the calculations for the Hg0 production in the contaminated sediments hinge on the assumption that the experiments will mimic what is happening in the contaminated site sediments. Several factors can be envisioned that vary between these two settings and which may play a role: The grain size composition (see general comment above), oxygen abundance, and changes in, e.g., temperature and light conditions. A discussion about the implications of these differences would be a welcome inclusion. Furthermore, the calculations for the back-of-the-envelope calculation could be included in the supplementary information to clarify how this estimation was done.

S1, Background Investigations section: It is stated that the equipment was tested for DOC to investigate the origin of the discoloration. What was the result of these tests?

Technical Corrections

Ln 116: Remove the “from”.

Ln 155: “Names” should be “named.”

Ln 246: The protocol summary makes more sense if the relative rather than the absolute volume of BrCl in the modified aqua regia is specified.

Ln 338: Insert “to” in “species used to generate stock solution”.

Figure 2: The dashed lines are faint, and the color contrasts between the Freundlich and exponential decay functions are hard to discern. It is suggested that the figure be edited to make it easier to interpret.

Ln 364: “EXP” should be “EXP1”.

Ln 394: “recovers” should be “recovery”.

Ln 407: A period sign is missing.

Ln 428, Ln 476: The use of “that” or “this” in the first sentence of the paragraph, referring to the end of the previous one, impacts readability.

Ln 434, 495: “Overtime” should be “over time”.

Ln 464-464, figure 3: It is suggested to label the panels in Figure 3 a) and b) and make the corresponding references in the text.

Ln 567: “…that fraction…” should be “…that the fraction…”.

References: The Miretzky 2005 reference is not listed.

S1: There are references to Appendices A 3 and A 4, but no appendix is included apart from the supplementary information. It appears that the S1:3 and S1:4 figures, respectively, are what is meant to be referenced.

Fig S1.5: While not critical for a SI figure, the figure could be improved by shortening the X-axis (to 100 instead of 180 minutes).

S1: The “results of the preliminary test” section: The takeaway, aided by figure S1.6, is that the concentration was ramped to reach a concentration where the eluate concentration was high enough for the experiment. This is, however, not clear when reading the text and should be clarified.

S2: Table S2.2 header “Elements” should be specified to clarify that wavelengths are listed below.

S2: “Table S1.5” should be “Table S2.5".

I first congratulate the lead scientist and the entire team. The manuscript offers interesting findings providing valuable information for Hg research, especially on Hg mobility. The manuscript is well-documented and also well-written, particularly in explaining the experimental methods. I recommend this manuscript to be published in SOIL with minor suggestions.

Abstract: The conciseness of the abstract can be improved. Some details can be moved to the conclusion section, ensuring the main finding remains highlighted.

Lines 53-66: The first paragraph can be re-arranged, some of the sentences are too broad and have little link to the research 

Lines 122-124: is there any information on geologic materials and the site description where the samples were taken?

Line 140: put a space, "8 x 60 mL ..."

Line 143: DI = deionised?

Line 148: what is the desired volume or target density?

Lines 171-173: The difference between EXP1 and EXP2 can be placed at the beginning of the paragraph, so readers can directly point out the difference between the two experiments.

Lines 172-173: The amount of HgCl concentration used in the experiment was estimated by considering the current concentration (after 55 years). Is there any data or information about the loss of Hg concentration (about 1000x) over 55 years?

Figure 2: The blue and red dashed lines are not clearly seen.

Line 364: should be EXP1

Lines 369: Reference of Miretzky et al. (2005) is missing

Lines 377-379: The authors pointed out the potential role of clay minerals or Fe/Mn oxides as an important solid-phase for the sorption of soluble Hg. In this case, the authors provide the properties of the solid-phase aquifer, such as Fe, Mn, clay, silt, sand (Table 1), and metal cations (S2). The pH was neutral to slightly basic, which support the adsorption of Hg into inorganic material. However, the clay content is low (table 1). Perhaps the authors could also provide a direct link/evidence between the Hg sorption and the specific minerals or metals involved in this process. 

Conclusion: I suggest the authors include a conclusion section. The conclusion can contain a more comprehensive summary and suggestions for future work. 

The manuscript “Organic matters, but inorganic matters too: column examination of elevated mercury sorption on low organic matter aquifer material using concentrations and stable isotope ratios” uses careful mercury breakthrough column experiments to determine the transport of mercury in low organic matter aquifer material. The experiments and analytical work are well done, and I commend the authors on their careful work. The information contained within the manuscript, particularly the role of inorganic materials in mercury transport, will be of interest to the wider mercury scientific community. However, the lack details of the solute transport modelling methods raises more questions and concerns that need to be addressed, as detailed below.

Specific Comments

The methods detail the experimental methods well but do not detail the fitting of the solute transport model to the data. The lack of such section (and associated results) causes confusion as the breakthrough curves presented in figure 2 are dependent on time, while the equation presented in Table S6.1 is not dependent on time. How did the authors fit the equation to the data? Was the advection-dispersion equation fit first using the chloride tracers to estimate conservative transport parameters (i.e., hydrodynamic dispersion of the media) then the Freundlich equation fit using the Hg data? Or were the presented equations in S6 fit to the data without considering flow (i.e., a statistical fit). If the latter, then the resulting parameters have no physical meaning and are only the summation of multiple processes occurring (advective transport, surface complexation, mineral (clay) matrix diffusion, reduction to Hg0, desorption, etc.), thus not representing the Freundlich isotherm as suggested in the text (this point also applies to the flushing analysis and the exponential decay function). I also note that the Freundlich equations presented in S6 deviates from the common form (q_e=K_F*C^e^(1/n)), where qe is the solid phase concentration, Kf is the Freundlich equilibrium constant, Ce is the aqueous concentration and n is a fitting constant determined by linearizing the equation. There was no citation for the specific Freundlich equation used in the text, nor for the reasoning behind dividing the breakthrough curve into two separate analyses. The resulting units of the proposed curves are mg/L² but the Freundlich isotherm units (Kf) are mg/g (mass/mass). Fitting the advection-dispersion equation with a Freundlich or Langmuir (as discussed below) isotherm is straightforward in freely available programs such as RETC or Hydrus-1D. There needs to be a clear section on the modelling procedure, including assumptions and equations, since the results and discussion rely on these methods. This needs to be added to the main manuscript prior to publication.

Other Specific Comments

L27-28 The Freundlich model describes sorption not breakthrough curves, which is mediated by water flow. This needs to be more clearly stated here, perhaps adding “of the” in-between sorption and breakthrough.

L61 comma splice after solubility. Regardless of this small grammatical issue, perhaps changing “have” to “having” will help the readability of the sentence.

L201-202 What is the total volume taken for analysis (was it the same as the waste 10mL or was the 15mL tube mentioned in 2.2.3 filled) and how does this compare to the total pore volume? If the ratio between the sample volume and the pore volume is large, then there needs to be some consideration that a sample taken at a given time represents a range of times rather than a specific time as assumed in the analysis. The larger the ratio the less precise your results. This comment also applies to the Hg isotope samples.

L296-297 The definition of effective porosity is the proportion of total void space that is capable of transmission fluid under advective fluxes. In most cases, this value is close to the total porosity but some media it can be quite a bit lower. Given your packed columns, I suspect that your effective porosity is closer to your total porosity as you defined. However, this assumption needs to be explicitly stated or measurements of effective porosity (e.g., soil air content at -100 mb) presented to confirm your assumptions.

L301-302 Given the well-known soil texture and artificially packed nature of the columns, the effective porosity can be relatively accurately estimated using freely-available pedotransfer functions (Rosetta — ISMC (soil-modeling.org)). Such estimations would allow for KD to be estimated on all columns that achieved 50% breakthrough. These values can then be compared to your estimated values from columns C1.1-1.3 and C2.1-2.3.

L311 missing closing parenthesis.

L303-316 How much Hg could be sorbed to the walls of the syringe? Would accounting for this improve your Hg mass balance (increasing eluate concentration from ≈91% for instance)? I note you discuss this briefly, but this may be worth a small batch experiment to see if the amount of sorbed Hg onto the plastic might be significant.

L364-365 The determination of theoretical max sorption condition suggest that the sorption characteristics can be fit with a Langmuir isotherm rather than a Freundlich isotherm model. Based on the discussion and proposed multi-mechanism for Hg desorption (outer-sphere complexation vs mineral matrix diffusion), a multi-site Langmuir adsorption isotherm may best describe the actual processes rather than the simpler Freundlich isotherm. The authors seem to use a 2-site exponential decay function in the Desorption phase of EXP2 but no discussion of this is presented nor the rational. By using the Freundlich isotherm it is assumed that both processes that remove Hg from the liquid phase (complexation and mineral matrix diffusion) are occurring at the same rate and have the same potentials. The flushing phase of your column experiment and the isotopic results, with the current analysis, suggest that this is not the case. See Swenson and Stadie (2019, 10.1021/acs.langmuir.9b00154) for a good overview of Langmuir isotherms.  

L574 I appreciate the back-of-the-envelope calculations that really contextualize the magnitude of the potential Hg0 production at the contaminated site, however there needs to be a more explicit description of the mathematics and the values used in the calculation either here or in the SI.

L592 Given the reactive transport focus of the paper, I suggest replacing “three-dimensional spread” to the more appropriate terms “longitudinal and transverse dispersion”.

L598 I really like this point!

L600-615 The lower, or even different, KD for the higher concentration EXP2 is somewhat surprising, as the KD is the linear partitioning coefficient, which assumes equilibrium between the liquid and solid phases. Assuming there is no saturation of adsorption sites, KD should be close if not equal in both experiments since the materials, packing, and flow rates are the same. However, there was no explanation of these values or their differences beyond stating other literature values. Are the differences in KD due to slight variations in clay content (experimental error) or is there another explanation? I suggest that there needs to be a bit more discussion here to explain these values more mechanistically.