Development of magnesium/calcium oxalate cements

2.1. Materials

Low-grade magnesia (MgO) was purchased in powder form. As this powder dissolves too quickly in an acidic solution, it was calcined for 1 h at 1500 °C to obtain dead-burned magnesia (MgO1500) (4444. Bilginer, B.A. (2018) Development of magnesium potassium phosphate cement pastes and mortars incorporating fly ash. M.S. Thesis, Middle East Technical University, Ankara, Turkey.
). After calcination, the hard mass obtained was ground to a powder. Ground granulated blast furnace slag (GGBFS) was received, in powder form, from Kardemir Iron and Steel Plant in Karabük, Turkey. The magnesia and slag were used as the alkaline component of the acid-base mixtures. The fly ash (FA) used was received from Afşin-Elbistan Thermal Power Plant in Turkey. Technical grade oxalic acid dihydrate (OxAc) was received from Balmumcu Chemical Industries in Ankara, Turkey. Reagent grade borax, Na₂B₄O₇·10H₂O (Merck), was also used, as a set retarder. The oxide compositions of the main ingredients (MgO, MgO1500, FA, and GGBFS) determined using X-ray fluorescence spectrometry (XRF, Rigaku ZSX Primus), are provided in Table 1.

As expected, calcination does not greatly affect the oxide composition of the magnesia but mainly decreases the measured CO₂ content. Loss on ignition measurements parallel this loss upon calcination. Nevertheless, MgO1500 still contains some CO₂. This could be due to some organic impurities in the as-received magnesia forming carbon upon calcination or due to problems with detecting/measuring the lightweight (and low x-ray fluorescence yield) element carbon with semi-quantitative XRF.

The fly ash is “high-lime” as per ASTM C 618 (45) with ~21 % CaO and > 50 % SiO₂+Fe₂O₃+Al₂O₃. More than 3 million tons/year of this ash are produced, and its high SO₃ content makes it unsuitable for use as a pozzolan in PC systems. The slag is “hydraulically active” with (CaO + MgO) / SiO₂ > 1.0, as per EN 197-1 (4646. CEN 197-1. (2012) Cement - Part 1: Compositions and conformity criteria for common cements. Brussels, Belgium.
). The fly ash was used to produce a low-cost intermediate oxalic acid salt (FAOx), analogous to KH₂PO₄ used in magnesium phosphate cements. It was used to provide oxalate ions to the system in a controlled fashion. Figure 1 summarizes the production method for FAOx.

FA:OxAc:water are mixed at 1.0:1.5:1.0 by mass as in (1515. Erdoğan, S.T.; Bilginer, B.A.; Canbek, O. (2022) Preparation and characterization of magnesium oxalate cement. Engrxiv.https://doi.org/10.31224/2298.
). The resulting paste heats up and within minutes sets into a strong but water-soluble solid. The paste is oven-dried at 105 °C for 24 h, crushed, and then ground for 45 min in a laboratory ball mill to obtain a powder. The specific gravity of FAOx was measured as 1.96. Figure 2 shows the mineralogies of the raw materials and FAOx, determined using x-ray powder diffraction (XRD). Periclase is the main phase in MgO1500 with major peaks at ~42.5 and ~37 º2θ. Forsterite (Mg₂SiO₄) is formed by the solid-state reaction of MgO and SiO₂ at high temperatures (4747. Brindley, G.W.; Hayami, R. (1965) Kinetics and mechanism of formation of forsterite (Mg₂SiO₄) by solid state reaction of MgO and SiO₂. Phil. Mag. 12 [117], 505-514. https://doi.org/10.1080/14786436508218896.
). FAOx shows peaks for whewellite (~24 º2θ) and quartz (~27 º2θ). Whewellite (CaC₂O₄·H₂O) is a calcium oxalate formed by calcium in the FA used and oxalic acid. Quartz carries from the fly ash itself. The slag is amorphous, with a characteristic hump around ~30 º2θ.

The particle size distributions for MgO1500, slag, and FAOx were measured using laser diffraction (Malvern Mastersizer 2000) on dry powders (Figure 3). All three powders have median particle sizes 10-20 µm.

2.2. Methods

The mixture proportions given in Table 2 were used to prepare mortar and paste (same proportions but without sand) samples for various characterization tests.

The mixture proportions in Table 2 were chosen to achieve a low estimated carbon footprint and adequate mechanical performance. The FAOx-to-alkaline binder ratio influences the strength and water resistance of oxalate binders and a ratio close to 2 is close to optimum when MgO1500 is used (1515. Erdoğan, S.T.; Bilginer, B.A.; Canbek, O. (2022) Preparation and characterization of magnesium oxalate cement. Engrxiv.https://doi.org/10.31224/2298.
). The calcination of MgCO₃ to produce 1 g MgO emits ~1.1 g chemical CO₂. Assuming 0.4 g fuel-related CO₂ is released to obtain dead-burned MgO (at a temperature similar to PC clinker production) (4848. Benhelal, E.; Zahedi, G.; Shamsaei, E.; Bahadori, A. (2013) Global strategies and potentials to curb CO₂ emissions in cement industry. J. Clean. Prod. 51, 142-161. https://doi.org/10.1016/j.jclepro.2012.10.049.
), the total emitted CO₂ becomes ~1.5 g per 1 g of MgO1500. FAOx (1 part FA and 1.5 parts OxAc) has FA:C₂O₄ ^2- ~1.0, which means half of the FAOx mass is equivalent CO₂. Hence, carbon neutrality/negativity can be achieved only when FAOx:MgO1500 ≥ 3.0. Slag has a chemical carbon footprint of zero but operations like grinding make it slightly positive. Hence the two SlOx mixtures and the MgSlOx mixture in Table 2 may be carbon-negative but the two MgOx mixtures are not. Acid-base mixtures exhibit different behavior than portland cement pastes in the fresh state. Their stronger shear-thinning character and their short setting times do not provide a long time period over which flow is constant (especially for SlOx mixtures). The water-to-binder ratio (W/B) of 0.30 chosen for all mixtures is roughly the lowest value that allows adequate mixing and compaction of fresh mortars. The sand-to-binder ratio was chosen as 2.0 to give a paste/aggregate ratio similar to that in standard PC mortars. 5 % borax was used in MgOx to sufficiently retard setting. SlOx mixtures react more rapidly than MgOx, so a higher amount of borax was needed.

3.1 Setting times of oxalate mortars

The setting times of the mixtures are given in Table 3.

SlOx mixtures have shorter setting times than MgOx mixtures. The alkaline powder reacted with the acid salt affects the initial pH of oxalate cement pastes (Figure 5), which changes the relative amounts of the different oxalate species present in the solution. SlOx pastes have higher initial pH and form calcium oxalates while MgOx have lower initial pH and form magnesium oxalates, resulting in the observed differences in setting time. Addition of borax prolongs setting times of all mixtures. The mechanism is likely similar for MgOx and SlOx to the retardation it causes in MPCs. At low pH, tetraborate ions from borax dissolution are adsorbed on the surface of MgO particles. This produces a layer which slows the dissolution of MgO (5252. Ma, C.; Wang, F.; Zhou, H.; Jiang, Z.; Ren, W.; Du, Y. (2021) Effect of early-hydration behavior on rheological properties of borax-admixed magnesium phosphate cement. Constr. Build. Mater. 283, 122701. https://doi.org/10.1016/j.conbuildmat.2021.122701.
).

3.2. Influence of mixture proportions on strength development

The strength development of the mortars are compared in Figure 4.

The very early-age (4 h) strengths of both types of mortars are < 8 MPa. Although this appears low in comparison with the analogous MPC, most such high early strengths reported for MPC are measured on pastes, which allows very low W/B to be employed (2020. Haque, M.A.; Chen, B.; Maierdan, Y. (2022) Influence of supplementary materials on the early age hydration reactions and microstructural progress of magnesium phosphate cement matrices. J. Clean. Prod. 333, 130086. https://doi.org/10.1016/j.jclepro.2021.130086.
, 2424. Mestres, G.; Ginebra, M.P. (2011) Novel magnesium phosphate cements with high early strength and antibacterial properties. Acta Biomater. 7 [4], 1853-1861. https://doi.org/10.1016/j.actbio.2010.12.008.
). Initial comparisons indicate that oxalate cements have slightly greater water need than phosphate cements. 1-d strength is higher for MgOx than SlOx, reaching ~17.5 MPa for MgOx-7/3. The higher amount of borax used in the SlOx mixtures contributes to the lower early strength. Borax not only retards the reactions but also contributes some water (W/B becomes ~0.31 and ~0.33 for mixtures with 5 % and 10 % borax). MgOx mortars reach their ultimate strengths at ~7 d, and MgOx-7/3 even shows a slight drop in strength beyond 7 d. The fact that MgOx-6/4, which reaches > 30 MPa ultimate strength, does not show a similar drop in strength after 7 d, suggests a critical magnesium-to-oxalate ratio may have been exceeded in MgOx-7/3, leading to volume instabilities. The strength of SlOx mortars reach 24-27 MPa, and MgSlOx, with an equal mass combination of the two alkaline powders has the highest 28-d strength among all mortars. More importantly, the strengths of all slag-containing mortars have an upward trend from 7 d to 28 d, indicating their ultimate strengths may be even higher. Much as early-age strength gain is related with the formation of crystalline reaction products, later-age strength is related with the formation of pore-filling amorphous phases and this may explain the delayed but continuing strength gain of slag-bearing mortars.

3.3. Change in pH

The change in the pH of the pastes is compared in Figure 5. The increase of the low initial pH of acid-base paste mixtures is related with the amount of reaction hence corresponding neutralization.

The initial pH of FAOx in water is low (~1), due to the dissolution of oxalates in water. MgOx pastes have lower initial pH (1-2) than SlOx pastes (~4). Initial pH is related to early mortar strength, higher for SlOx than MgOx. For both mixture types, pH quickly rises as Mg²⁺ and Ca²⁺ from the dissolution of the alkaline powders react with the available oxalate species (the oxalate and hydrogen oxalate anions). The pH of MgOx rises quickly to 8-9 within 24 h, after which it more or less plateaus, similar to the pH development reported for ferrous oxalate cement pastes (3939. Luo, Z.; Ma, Y.; He, H.; Mu, W.; Zhou, X.; Liao, W.; Ma, H. (2021) Preparation and characterization of ferrous oxalate cement - A novel acid-base cement. J. Am. Ceram. Soc. 104 [2], 1120-1131. https://doi.org/10.1111/jace.17511.
). The rise in pH is much less for SlOx, only to ~5, suggesting less complete neutralization reactions, and explaining the lower 7 or 28 d strengths. Despite little change in pH after 1 h for SlOx pastes, strength increases up to 28 d (Figure 4). The moderately acidic environment is hence suitable for continued dissolution of slag particles and formation of more reaction products. The ultimate pH of MgSlOx paste is ~7, in between the values for MgO1500-only and slag-only pastes. This neutral value suggests a better balance between acidic and alkaline starting powders and a lower amount of materials left unreacted after the rapid reactions.

3.4. Influence of mixture proportions on mineralogy

Figure 6 compares the mineralogical development of the pastes. MgOx pastes, as expected, contain a hydrated magnesium oxalate, glushinskite (MgC₂O₄·2H₂O) as their main reaction products. Whewellite (CaC₂O₄·H₂O) and quartz are also found in the final paste. Quartz carries from the unreacted part of FAOx. Whewellite is also present in FAOx but it is not clear whether it partly dissolves in the acidic condition achieved when FAOx is added to water and then reforms as pH increases. Also present are unreacted magnesia (periclase) and forsterite, coming from MgO1500. Small amounts of magnesiowüstite (Fe_xMg_1-xO), and magnesioferrite (Fe₂MgO₄) are also detected but their peaks partly overlap with those of whewellite and forsterite. The diffractograms of the SlOx pastes suggest whewellite and another calcium oxalate, weddellite (CaC₂O₄·2H₂O), as well as a small amount of portlandite.

MAUD (5353. Lutterotti, L. (2000) Maud: a rietveld analysis program designed for the internet and experiment integration. Acta Cryst. A. 56, s54. https://doi.org/10.1107/S0108767300021954.
) was used to analyze the 28-d XRD spectra using the Rietveld method. Glushinskite, whewellite, periclase, quartz, and forsterite were considered as the only phases existing in MgOx pastes. Whewellite, weddellite, quartz, and portlandite are considered as the crystalline phases in SlOx, with additional amorphous content. Diffraction data of the selected phases taken from the “crystallography open database” were loaded into MAUD. Background substraction, scaling, and noise cancellation operations were performed. Analysis of MgOx-6/4 suggested ~35 % glushinskite, ~20 % whewellite, and ~25 % periclase, as well as ~4 % quartz and ~16 % forsterite. The fit parameter, R_w, was 7.1 %. The amorphous content of MgOx pastes was determined, using an internal standard, to be negligible. For MgOx-7/3, slightly greater amounts of the oxalates (~40 % glushinskite and ~25 % whewellite), and less periclase (~18 %) are calculated, consistent with its higher FAOx/MgO1500. The amounts of the remnant phases also change as expected, quartz increases to 5 % and forsterite decreases to 11 %. R_w was calculated as 8.9 %. Further inspection of these results suggested that the calculated amounts of phases may be slightly under/overestimated. The amount of non-volatile oxides in the starting mixture should equal their amounts in the final paste. For example, in the starting mixture for MgOx-6/4, MgO is mostly in MgO1500, SiO₂ is in MgO1500 (in Forsterite) and in FAOx (from FA), while CaO and C₂O₄ are mostly in FAOx. Since the amounts of each material in the starting mixture is known (Table 2), using the oxide composition for each (Table 1), and assuming FAOx contains 49 % FA and 51 % C₂O₄ (verified approximately by XRF tests), the amounts of glushinskite, whewellite, periclase, quartz, and forsterite (total 100 %) that best satisfy the “mass of oxide in initial mixture equals mass of oxide in final paste” objective for MgO, SiO₂, CaO, and C₂O₄ are calculated using the Solver add-in program in MS Excel as 35 %, 22 %, 18 %, 11 %, 14 %. Hence, there is probably a smaller amount of (unreacted) periclase and a greater amount of quartz in the reacted paste than quantified from the diffractogram in Figure 6a. The same analysis for MgOx-7/3 calculates 38 %, 25 %, 10 %, 11 %, 14 % for of glushinskite, whewellite, periclase, quartz, and forsterite, again suggesting unreacted periclase was initially overestimated and quartz was underestimated. This may be partly related with certain minor phases not being considered for the quantification or measurement parameters leading to insufficient intensity for some peaks i.e. low signal-to-noise ratio. Reviewing these calculated amounts of each phase and Tables 1 and 2, it can be deduced that the amount of forsterite increases while quartz decreases from the starting mixture to the final paste. Hence, some of the quartz in FAOx (from FA) reacts with dissolved MgO1500 to form new forsterite. It is unclear whether the forsterite initially available in MgO1500 ever partly dissolves in the acidic solution. This makes it difficult to calculate a degree of reaction for MgO1500. However, the initial fractions of MgO1500 in the starting mixture and the final unreacted periclase contents indicate that the degree of reaction of MgO1500 is higher in MgOx-7/3 than in MgOx-6/4, consistent with its higher 7-d strength (Figure 4, ignoring the subsequent drop in strength due to durability problems). Tracking MgO only (starting in MgO1500 and ending up in Glushinskite, Forsterite, and unreacted Mg1500), degrees of hydration for MgO1500 in MgOx-6/4 and MgOx-7/3 are 45 % and 55 %, respectively. Similar low degrees of reaction for MgO have been reported for magnesium phosphate cements (5454. Yu, J.; Qian, J.; Wang, F.; Qin, J.H.; Dai, X.B.; You, C.; Jia, X.W. (2020) Study of using dolomite ores as raw materials to produce magnesium phosphate cement. Constr. Build. Mater. 253, 119147. https://doi.org/10.1016/j.conbuildmat.2020.119147.
). The amorphous natures of the slag and some reaction products complicate the quantitative interpretation of XRD for the SlOx and MgSlOx pastes. The amorphous content in SlOx-7/3 is determined as ~45 %. Even if the slag used is fully amorphous, this would mean part of the reaction products, ~20 % of the total mass, is amorphous as well. The weddellite/whewellite ratio is higher in SlOx-6/4 than in SlOx-7/3 as expected due to its lower FAOx content (Figures 6c and d). Whewellite is the more stable one of the two calcium oxalates. The initial crystallization phase from aqueous solution is a calcium oxalate trihydrate, which loses water of crystallization to either the monohydrate or dihydrate, depending on conditions, such as the calcium to oxalate ion ratio or the presence of substances which form complexes with either calcium or oxalate ions, such as citric acid and magnesium (5555. Gadd, G.M. (1999) Fungal production of citric and oxalic acid: importance in metal speciation, physiology and biogeochemical processes. Adv. Microb. Physiol. 42, 47-92. https://doi.org/10.1016/S0065-2911(08)60165-4.
). Weddellite precipitates under excess of calcium ions in the medium. This is consistent with no weddellite being detected in MgSlOx, which contained much less slag, hence much less calcium, than either SlOx. The total whewellite and weddellite content is ~46 % in SlOx-6/4 and ~49 % in SlOx-7/3. The total amount of whewellite and weddellite that can be formed can be estimated considering the total calcium in the slag and in the fly ash used to produce FAOx. 1 g of the slag in Table 1 can produce 1.06 g whewellite or 1.17 g weddellite. Similarly, 1 g of FA in Table 1 can produce 0.60 g whewellite or 0.66 g weddellite. Assuming equal amounts of each are formed, the maximum total amount of whewellite and weddellite that can be produced in SlOx-6/4 is calculated as ~49 %. Although the calculated total is below this value, it is probably high, since it is unlikely that all the calcium forms one of the two calcium oxalates.

3.5. Thermogravimetric analyses

Figure 7 presents the mass loss and heat flow measured for paste samples heated to 900 °C.

Mass loss takes place in three main steps for all pastes, which contrasts with the one step decomposition of K-struvite in MPCs (5656. Chauhan, C.K.; Vyas, P.M.; Joshi, M.J. (2011) Growth and characterization of struvite-K crystal, Cryst. Res. Technol. 46 [2], 187-194. https://doi.org/10.1002/crat.201000587.
). Although glushinskite decomposes in two steps, the presence of whewellite coming from FAOx in the MgOx pastes causes a third mass loss step. Figure 8 summarizes the decomposition steps for the calcium and magnesium oxalates and the theoretical mass losses associated with each step (57-5957. Frost, R.L.; Weier, M.L. (2003) Thermal treatment of weddellite - a Raman and infrared emission spectroscopic study. Thermochim. Acta. 406 [1-2], 221-232. https://doi.org/10.1016/S0040-6031(03)00259-4.
58. Stephens, W.E. (2012) Whewellite and its key role in living systems. Geol. Today. 28 [5], 180-185. https://doi.org/10.1111/j.1365-2451.2012.00849.x.
59. Frost, R.L.; Adebajo, M.; Weier, M.L. (2004) A Raman spectroscopic study of thermally treated glushinskite--the natural magnesium oxalate dihydrate. Spectrochim. Acta A Mol. Biomol. Spectrosc. 60 [3], 643-651. https://doi.org/10.1016/S1386-1425(03)00274-9.
).

Up to ~100 °C, mass loss is related with free water in all pastes. The rest of the mass loss in step 1 is due to loss of crystal water in the magnesium and calcium oxalates. Figures 7b and 7f show two mass loss peaks for MgOx, one at 150 °C due to the decomposition of whewellite and another at 210-220 °C (higher than suggested in Figure 8). For SlOx, the dominant loss peak is at ~160 °C, as expected for whewellite (Figures 7d and 7h). Although XRD suggests presence of weddellite, a related lower temperature peak is not observed. The second step (400-500 °C) is related with the decomposition of anhydrous magnesium and calcium oxalates. The break-down of MgC₂O₄ releases CO and CO₂ leaving behind MgO. Decomposition of CaC₂O₄ releases CO and leaves behind CaCO₃. The main peaks for SlOx (whewellite decomposition) in this step have shoulder peaks (weddellite decomposition) on their lower temperature sides, most noticeable for SlOx-7/3, in Figure 7h. The third step (600-800 °C) breaks down the remaining carbonate, releasing CO₂.

3.6. Influence of mixture proportioning on the microstructure

Figure 9 presents SEM images of reacted pastes.

Many loosely connected prismatic crystals of 4-5 µm size are observed in MgOx-6/4 among unreacted magnesia particles (Figure 9a). Crystals in MgOx-7/3 are smaller and cube-like (Figure 9b). SlOx shows smaller, mostly sub-micrometer crystals dispersed in a glassy background (Figures 9c and 9d). Notably, cracks are observed in MgOx-7/3 (Figure 9b), which could explain the strength loss recorded for this mixture from 7 to 28 d. In MgSlOx (Figure 9e), the amorphous material produced by reactions of the slag envelops the crystals, leading to a denser microstructure explaining the higher ultimate strength measured on mortar samples.

3.7. Pore size distribution and porosity

Figure 10 presents the pore size distributions of the prepared pastes.

MgOx-6/4 has the lowest porosity (Figure 10a) among all mixtures. MgOx-7/3 has the highest of all mixtures which could be related with the cracks observed in Figure 9b. This paste also contains an unexpectedly high amount of pores of nearly 1 µm size which could be microcracks. All samples appear to contain two dominant sizes of pores, larger than ~0.1 µm and smaller than ~0.02 µm. Both MgOx pastes have greater average pore size than SlOx samples, as well as a wider range of pore sizes. The amorphous products in the slag-containing pastes may be refining their larger pores. Differences in the amount of small and large pores in Figure 10 may explain the differences in strength better than total porosity. MgSlOx, the paste with the highest strength, has a smaller amount of such large pores and a greater amount of smaller pores.

3.8. Temperature changes

There are two distinct heat-evolving steps in the reactions of MgOx and SlOx: i) the wetting and dissolution of FAOx in water, and ii) the contact of the alkaline powder (MgO1500 or slag) with the FAOx and water mixture. The contact of the alkaline powder alone with water evolves heat but is insignificant in comparison with the other two steps. Figure 11a shows the change in temperature of the various oxalate pastes prepared. Temperature rises sharply by 7-8 °C when FAOx and water are mixed (time zero). This is considerable for the small paste volume (~7 cm³) used. After ~1 min, the slurry begins to cool slowly.

The addition of the alkaline powder creates a second (maximum) temperature peak. The time delay of the alkaline powder addition affects this maximum temperature. Figure 11b shows the change in this maximum temperature with the amount of time elapsed before adding the alkaline powder. A longer wait period reduces the overlap of the two separate heat-evolving events, and a slightly lower maximum temperature is recorded. However, the FAOx + water paste begins to stiffen in time hence a very long wait period may require additional water to be added, influencing hardened properties. MgOx pastes reach slightly higher temperatures than SlOx. The temperature peak occurs 2-3 min after all materials have been added. Hence, setting which takes place after 5-10 minutes follows the temperature peak with a minor delay.

3.9. Water resistance

The resistance of MgOx, SlOx, and MgSlOx mortars to water are compared in Figure 12. MgOx mortars cured in air and then kept under water do not show significant changes in compressive strength (observed differences being within experimental uncertainty). In contrast, both SlOx mortars lose ~70 % of their air-dried strengths. This decrease could be related with the dissolution of the reaction products present. Hydraulic pressure inside pores within the sample could also negatively affect strength since these samples are tested immediately after removal from water, without any time to dry. Small increases in the strengths of MgOx-6/4 and MgSlOx may be due to continued reaction in mortars which had not reached their ultimate strengths prior to submersion in water.

3.10. Reaction mechanism of magnesium/calcium oxalate cements

There are two main steps in the reaction of MgOx cements with water. In the first step, the dissolution of FAOx in water yields oxalate ion species and H⁺, which makes the solution acidic. A diprotic acid, oxalic acid has pKa₁ = 1.27 and pKa₂ = 4.27. Hence, at the initial low pH of ~2 (Figure 5) more HC₂O₄ ^- and some H₂C₂O₄ are present in the solution (3939. Luo, Z.; Ma, Y.; He, H.; Mu, W.; Zhou, X.; Liao, W.; Ma, H. (2021) Preparation and characterization of ferrous oxalate cement - A novel acid-base cement. J. Am. Ceram. Soc. 104 [2], 1120-1131. https://doi.org/10.1111/jace.17511.
).

In the second step, the periclase in MgO1500 begins to dissociate in the acidic solution and reacts with HC₂O₄ ^-. pH rises and HC₂O₄ ^- further dissociates into C₂O₄ ^2-. At pH~4-4.5, approximately equal amounts of HC₂O₄ ^- and C₂O₄ ^2- are present in the solution (3939. Luo, Z.; Ma, Y.; He, H.; Mu, W.; Zhou, X.; Liao, W.; Ma, H. (2021) Preparation and characterization of ferrous oxalate cement - A novel acid-base cement. J. Am. Ceram. Soc. 104 [2], 1120-1131. https://doi.org/10.1111/jace.17511.
).

The continued dissolution of MgO1500 (forsterite and periclase) releases amorphous silica and Mg²⁺ into solution (6060. Qiushi, Z.; Xing, C.; Rui, M.; Shichang, S.; Lin, F.; Junhao, L.; Juan, L. (2021) Solid waste-based magnesium phosphate cements: Preparation, performance and solidification/stabilization mechanism. Constr. Build. Mater. 297, 123761. https://doi.org/10.1016/j.conbuildmat.2021.123761.
), resulting in a rapid increase in pH (the first hour in Figure 5). HC₂O₄ ^- and C₂O₄ ^2- react with the Mg²⁺ ions to form glushinskite (Figures 6a and 6b):

The role of Eqns. 4 and 5 in glushinskite formation depends on pH. As pH increases, Equation 5 predominates and glushinskite continues to precipitate at a slowing pace. Unreacted periclase is found in the final solid, as well as quartz and whewellite (CaC₂O₄.H₂O) remaining from unreacted FAOx. In the case of SlOx, the reaction of slag in FAOx + water begins at pH ~ 4-5 where C₂O₄ ^2- is abundant in the solution. The acidic conditions lead to dissolution of Ca²⁺ from the slag which react with C₂O₄ ^2- to form weddellite (Figures 6c and 6d). Calcium ions can also come from whewellite in FAOx, the solubility of which increases markedly at pH < 5 (6161. Gadd, G.M. (1999) Fungal production of citric and oxalic acid: importance in metal speciation, physiology and biogeochemical processes. Adv. Microb. Physiol. 41, 47-92. https://doi.org/10.1016/S0065-2911(08)60165-4.
). The reactions in SlOx can be simplified as:

3.11. Carbon neutrality and feasibility of MgOx and SlOx cements

The argument that oxalate cements can be carbon neutral or even negative relies on the availability of oxalic acid produced from captured CO₂. Hence, a natural supposition is that carbon-negative oxalate cements would be cost-prohibitive. Although production of oxalic acid from CO₂ has been shown at the laboratory scale, since no large-scale production exists, the overall cost of this step is not easy to predict (1515. Erdoğan, S.T.; Bilginer, B.A.; Canbek, O. (2022) Preparation and characterization of magnesium oxalate cement. Engrxiv.https://doi.org/10.31224/2298.
). However, simple calculations and various assumptions can be used to estimate a rough cost and carbon footprint for MgOx and SlOx systems. The production of the alkaline powder is the main contributor to the carbon emissions related with oxalate cements. As stated in Section 2.2, ~1.5 g chemical and fuel related CO₂ is emitted during the production of 1 g dead-burned MgO from pure MgCO₃. Hence, FAOx:MgO1500 must be greater than 3 to obtain a carbon neutral or negative mixture. The CO₂ emissions related with grinding, which are relatively small (6262. Chen, C.; Habert, G.; Bouzidi, Y.; Jullien, A. (2010) Environmental impact of cement production: detail of the different processes and cement plant variability evaluation. J. Clean. Prod. 18 [5], 478-485. https://doi.org/10.1016/j.jclepro.2009.12.014.
) or with the preparation of FAOx (mainly heating and grinding) can further increase this ratio. Conversely, this ratio can decrease if the MgO used has a low carbon footprint, like one derived from seawater (6363. Turek, M.; Gnot, W. (1995) Precipitation of magnesium hydroxide from brine. Ind. Eng. Chem. Res. 34, 244-250. https://doi.org/10.1021/ie00040a025.
) is used, or if a higher OxAc:FA is used to prepare FAOx. However, varying OxAc:FA causes changes in the rate and heat of reaction of the overall system. When slag is used as the alkaline powder, the main factor that increases the CO₂ footprint becomes the grinding of granulated slag. Since the oxalate salt portion of the system remains unchanged, SlOx cements can be carbon negative, as long as the emissions due to grinding are below 1 g CO₂ per 1 g ground slag. This reinforces the importance of identifying a low-carbon-footprint alkaline powder to replace dead-burned MgO in making oxalate cements.

The cost of MgSlOx concrete is assessed (Table 4) by making assumptions about the unit costs of various materials or operations (1515. Erdoğan, S.T.; Bilginer, B.A.; Canbek, O. (2022) Preparation and characterization of magnesium oxalate cement. Engrxiv.https://doi.org/10.31224/2298.
). 300 kg of powder (MgO1500+slag+FAOx) is assumed per 1 m³ of concrete. Using the proportions in Table 2, 210 kg of the powder is FAOx, so ~153.3 kg OxAc (dihydrate) is needed (~0.73 g OxAc is required to make 1 g of the FAOx in this study). The production of OxAc from CO₂ is not done at a large scale outside of the lab, so it is most difficult to assign a cost to this process. The current cost of sustainable OxAc production from CO₂ (including the cost of CO₂) was estimated in a European research Project (3434. Schuler, E.; Demetriou, M.; Shiju, N.R.; Gruter, G.J.M. (2021) Towards Sustainable Oxalic Acid from CO₂ and Biomass. ChemSusChem. 14 [18], 3636-3664. https://doi.org/10.1002/cssc.202101272.
) as > 1100 $/t. However the cost was projected to drop to < 450 $/t beyond 2030, with lower or negative CO₂ cost (with incentives), a reduction in cell cost, and improved current densities, and a reduced electricity price. Another study (3131. Fischer, J.; Lehmann, T.; Heitz, E. (1981) The production of oxalic-acid from CO₂ and H₂O. J. Appl. Electrochem. 11, 743-750. https://doi.org/10.1007/BF00615179.
) estimated the cost of electrochemical production of OxAc from CO₂ as being equal to the market price of oxalic acid made from other sources, which can be taken as 450-600 $/t (6464. Retrieved from: https://www.alibaba.com/product-detail/Oxalic-Acid-Dihydrate-99-6-Price_11000003073704.html?s=p (accessed 3 January 2023).
). They did not include the cost of capturing CO₂ needed to produce OxAc which is also difficult to estimate. As an example, for one of the methods, direct air capture, Keith et al. (6565. Keith, D.W.; Holmes, G.; St. Angelo, D.; Heidel, K. (2018) A process for capturing CO₂ from the atmosphere. Joule. 2 [8], 1573-1594. https://doi.org/10.1016/j.joule.2018.05.006.
) estimate levelized costs of 94-232 $/t. Based on these studies, 450 $/t, 750 $/t, and 1100 $/t are chosen as the unit price of OxAc made from captured CO₂. Hence, the cost of OxAc used to produce MgSlOx concrete becomes 69-168.6 $/m³.

A low-purity MgO1500 obtained like the one in this study, assumed to cost 150 $/t (6666. Retrieved from: https://www.alibaba.com/product-detail/Magnesium-Oxide-Magnesium-Oxide-Magnesium-Oxide_62443248013.html?spm=a2700.7735675.normal_offer.d_title.37124a0chvbcqw&s=p (accessed 3 January 2023).
), adds another 6.8 $/m³ (four tenths of the powder binder is MgO1500). Costs of 20 $/t for ground slag (6767. Retrieved from: https://www.alibaba.com/product-detail/Hot-Sell-China-Granulated-Blast-Furnace_62180223076.html?spm=a2700.7724857.normal_offer.d_title.f9817524mgz5Sq (accessed 3 January 2023).
), 10 $/t for aggregates, 5 $/t for mixing water (W/B = 0.5), and 500 $/t for borax are assumed. The cost of 102.9 kg/m³ FA (~49 % of FAOx) is assumed to be 2.1 $/m³ even though the FA in this study is a waste (not suitable for PC concrete). The production cost of FAOx is assumed as 15 $/t by comparison with similar low-temperature processes (e.g. production of gypsum) which adds another 3.2 $/m³ (for 210 kg/m³ FAOx). The total cost of MgSlOx concrete becomes ~110-210 $/m³. This does not consider mixing, delivery etc. which would have to differ from PC systems because of differences in properties like setting time, or profit. Similar calculations for MgOx-6/4 yield a cost range of 110-195 $/m³, slightly lower because of the decreased amount of oxalic acid used. In comparison, ready-mixed concrete can cost > 150 $ in many developed countries (6868. Federal reserve economic data, Federal Reserve Bank of St. Louis, Producer price index by industry: ready-mix concrete manufacturing: ready-mix concrete for west census region, St. Louis, MO. Retrieved from: https://fred.stlouisfed.org/series/PCU327320327320D (accessed 3 January 2023).
). These calculations show that the cost of oxalate cement is dominated by the cost of producing oxalates from captured CO₂ and that changing the alkaline powder does not influence cost as much as it does the carbon footprint. Nevertheless, the identification of an effective low-carbon base may allow the amount of OxAc in these systems to be slightly reduced and still achieve carbon neutrality (e.g. FAOx/basic powder < 6:4) which could further decrease the overall cost.