Steel Slag Characterisation—Benefit of Coupling Chemical, Mineralogical and Magnetic Techniques

Round 1

Reviewer 1 Report

This study will be a valuable contribution and reference point in the field of steelmaking slag research and practice (storage, reuse, environmental issues): it showcases a solid combination of characterisation techniques applied to EAF slag and provides some very interesting, detailed observations on the mineralogy and chemistry of these materials of clear relevance in terms of environmental considerations (including toxicity). It is refreshing to see some competent mineralogical characterisation of steelmaking slags! Naturally, there are some areas that could be improved, but these largely fall under the category of future work that could be done to extend the current study. Fundamentally, the article is sound and with some fairly minor revisions is eminently publishable.

 

Some detailed comments / questions and suggestions for improvements follow below. I would suggest that these would not really require re-reviewing.

 

Minor linguistic point: e.g. line 42 ‘However, the important volumes of former slag…’. In this and a few other instances, the word ‘important’ should be replaced with ‘significant’ or ‘most significant’ depending on the context.

 

Materials and methods: line 93: could you provide a little more detail on how the samples were crushed and screened to create the set of grains for microscopic and magnetic analysis (0.5-1.0 mm fraction). It is clear that the different sieve fractions of the original (non-crushed) slag have strongly varying mineralogy and chemistry. During coarse crushing of the material, inevitably some material will be lost into the < 0.5 mm fraction, while it is not clear from the description whether all of the coarsest material is crushed so that it is < 1 mm. There is therefore the potential for some fractionation of material when removing the final 0.5 – 1 mm fraction. Do the authors have any comment on how they addressed this? It is a non-trivial challenge in this kind of material, which others also face. What qualifies as ‘representative’ – a set of grains with proportional contributions from the original different size fractions (coarser than the minimum of 0.5 mm)? This is difficult to achieve in practice, but theoretically this is what would have to be aimed for to make a truly representative modal analysis of the bulk slag (coarser than 0.5mm).

 

2.2.1: bulk chemical analyses

Comment - It’s a pity that there were no titration analyses for the oxidation state of iron. Fe3+ and Fe2+ (and any metallic) proportions would be a very useful addition to the overall dataset, to complement the mineralogical analyses. Something to consider for future work – the oxidation state of Fe appears to change over the different fractions, based on the XRD results (relative abundance of wustite and spinel solid solution).

Comment – the median grainsize of 20 micron is rather coarse for analyses, especially XRD.

 

2.2.2. Mineralogical characterization

Question – why no attempt at quantitative phase analysis, e.g. Rietveld refinement, with the raw XRD patterns? The ‘semi-quantitative’ results shown are still useful but a fully quantitative analysis is certainly possible and highly advisable.

 

EMPA analyses: could you please list which primary calibration standards were used?

 

Comment: line 140 – the step size in the EMPA mapping was 0.4 micron, which at accelerating voltage used is smaller than the horizontal diameter of the excitation volume in most oxide / silicate materials. Just to alert the authors to the fact that the true spatial resolution of the maps is limited by the latter, rather than the step size in this case. The imagery is sharp in the figures but please bear in mind that the actual phase analyses obtained from points (or pixels in the mapping) where phases are intergrown on the scale < 1 micron will inevitably be contaminated by adjacent phases. They can only be indicative in such cases and should be regarded more critically than analyses placed within large crystals / domains of a single phase.

 

2.2.4. Principle Component Analysis – Line 165: could you please clarify in the wording here which analyses were performed on the two aliquots of each sample? On each aliquot, the chemical analyses and magnetic susceptibility measurements were performed, or there were two magnetic measurements performed per sample with only one chemical analysis? I find it a little ambiguous as currently worded.

Question: which software was used for the PCA analysis? Or otherwise, please cite the source for the calculation procedure used.

 

3. Results

Requested revision / addition: I would strongly suggest the authors include the full tabulated data of all chemical analyses and EMPA analyses in an appendix / supplementary data repository.

Specifically, for Table 1, showing the mean values for all 30 slag samples is fine as it shows a single meaningful composition, but the min and max values for all the individual elements are somewhat uninformative when taken out of the context of their co-variation with one another. What would be more informative would be show a selection of ‘end member’ compositions, because in this way the mutual covariation of the elements can be seen. The full data would be better. Precisely this same comment applies to the EMPA data shown in Table 3.

3.2. Mineralogical composition

Please display the actual y-axis scale (intensity) to give an indication of total counts.

The indexing of the patterns is more convincing in combination with the independent microscopic identification of corresponding phases: Ca-chromite is identified based on only one reflection peak. The phases ‘ss-Spl’ and ‘Chr’ are difficult to index but again there is some independent verification of their presence in the materials.

As per the comments on the materials and methods section – it is a pity that no quantitative phase analysis has been made (Rietveld refinement). If attempting this, it would be strongly advisable to grind the material much finer before analysing it (below 10 micron).

The semi-quantitative results in Table 1 are described in the text but thereafter not really discussed in terms of what the explanation is for these very clear trends. Is it only weathering, or partly related to the cooling practice of the liquid slag prior to final storage?

 

Line 199: comment - the XRD analyses imply that calcite is predominantly present in the very finest fractions of the slag, which were not examined by microscopy. If the authors think it likely that calcite might have dissolve during sample preparation, I would suggest taking a sample and preparing it water free (e.g. with isopropanol or ethanol and water-free polishing agent).

 

Lines 202-204: tri-calcium silicate (C3S) may decompose to an intergrowth of C2S + lime during cooling. It is no coincidence that C2S with micro-inclusions of lime gives an overall composition the same as C3S, since the latter is the parent phase of the currently observed intergrowth. Please amend the wording of line 203-204 to reflect this well-known and standard phenomenon.

 

Figure 2 and line 201: where do the authors consider the wustite to be present in the microstructures seen under the microscope? It is clearly identified by XRD so it must be there but it is not labelled. It is likely hiding in full view intergrown with spinel solid solution, though robust identification is indeed difficult due to the fine lengthscales of intergrowth. The rounded shape of the spinel solid solution grains, referred to in line 210, is an important clue about where the wustite is / was…

 

Figure 2. Colour balance seems to be very different between (a) and (b). It’s not critical but it would be better to make some kind of colour balance correction bring the two images somewhat more in line with one another, given they are showing similar phases.

Line 207: non-sequential figure callout. Referring to Figure 4 after Figure 2. If this can be avoided, better. If not, it’s not critical.

 

Table 3: see comments above about the use of the min, max values. The average and standard deviation values are useful enough, but it would be far better to provide the full tabulated data in an appendix.

 

Figures 3 and 4 and description / discussion thereof: these are very nice microstructural figures and well described in relation to the correlated Raman analyses.

One question / comment: the identification of maghemite seems to be based largely on Raman analysis. Is this really a good match between the spectrum from the sample versus the maghemite reference? At high Raman shift it seems to deviate massively between the sample and the reference – can this be explained? Is it unimportant for the identification of the maghemite?

 

Line 265: do the authors suggest explicitly that the vanadium in the brownmillerite and chromite is in the form of V3+, or is this just the chosen oxide form to report total vanadium concentration? This is not a trivial question to answer, at least not in the brownmillerite phase.

 

3.4 onwards: Figure numbers in the text are not up to date. They refer to figure numbers one less than what they should refer to (probably an extra figure was inserted and the hyperlinks were not updated). Figure 4 in the text refers to Figure 5, and so on for the rest of the article.

 

Question: what is the magnetitic susceptibility of wustite? As the dominant / abundant Fe oxide phase in the coarser fractions of the slag, it would be interesting to see how it compares to the susceptibility of the bulk.

 

Line 294: ‘That means that the spinel solid solution…’ - this could / should be explained a little more fully.

Line 301: pure magnetite has not been observed ‘neither by XRD nor by electronic microprobe’ – in the case of XRD it is not possible to distinguish pure magnetite clearly from some spinel solid solution phases. Pure magnetite may well have been identified in the optical microscopy in the form of the bright rims around the lamellar packages, rather than this being pure wustite. In fact you refer to this in Line 247. That it would not be clearly identifiable with EMPA as being pure Fe-oxide (either wustite or magnetite) is not surprising given the small lengthscale of intergrowth and the impossibility of a clean phase analysis without contamination from the adjacent phases.

 

Line 303: ‘go and back’ – presumably there is a standard expression for this in the technical English jargon of magnetic susceptibility measurements. Please find it and use it. ‘Go and back’ is clear enough  but it doesn’t seem like correct terminology in English.

 

Line 313: ‘… of the total information’ – would be better to say ‘total variance’?

Regarding the interpretation of the results: if Mo is independent of the other variables, what is the basis for regarding Si and Al as correlated and Mg and Ca as correlated with one another? They are as far apart as Mo from the magnetic susceptibility and remaining elemental concentrations in the correlation circle plot. It is obvious that Si-Al and Mg-Ca are in opposite quadrants, but Mo is in the same quadrant as four other variables? How is this ‘independent’ of the other variables, based purely on this plot? Please explain.

 

4. Discussion.

There seems to be no overall discussion of the significance of the variations in mineralogy and chemistry over the sieve fractions in the slag. What are the reasons for the trends seen and what are the implications for environment and toxicity?

 

Conclusions.

Fully agreed with Line 384: it is indeed very important know not only elemental concentrations but also their speciation and solubility of their host minerals.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

Please see attached document.

   

Comments for author File: Comments.pdf

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 3 Report

The manuscript "Steel slag characterization: benefit of coupling chemical, mineralogical and magnetic techniques" presents a multi-approach case study of steel slag.  The main focus of the study is on magnetic measurements and their correlation with data obtained from phase studies. As the main conclusion is about the applicability of magnetic methods it is a shame that a map and more thorough characteristic of sampling sites are not presented. As sampling include 30 specimens it should be shown on a map. Another main conclusion is that for a given site magnetic methods are a good tool to delineate slag presence within soils, but the reader is only referred to a general Figure 5 with compositions of soils, rocks and slags somehow overlapping. How could we be sure that the magnetic methods are applicable for this particular study site? Therefore, my major comment is that the study should be presented more in the regional context. Also, the issue that is not discussed is vertical variability within the slag heap and how it affects magnetic measurements. The samples are taken from the top of the heap, which has up to 9 meters high, is similar composition and mineralogy to be expected also at lower levels?

I have found also minor mistakes in English grammar and structure of the manuscript:

line 33 - is 400 mln for both iron and slag, the two values should be presented separately

line 39 - "are often concerned" - does not sound right to me

line 72 - "remain unusable for" - not clear

line 77 - "to better know" - should be to better understand

line 79 - "realised" - not the right word

line 114 - "were to analyze" - does not sound right

line 195 - should be different not difference

line 200 - "logically" - sounds not right

line 223 - "electronic microprobe" - should be electron

lines 275, 282, 292, 310 - Figure 4, 5, 6, 7 are wrongly referred to in the text

 

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Round 2

Reviewer 3 Report

The authors considerably improved the manuscript and I recommend to accept it in its present form. However, I think that vertical diversity with depth should be also considered.