Insights into active and passive carbon sequestration and causticity reduction in hazardous red mud slurry

Batch experiments were conducted to collect data for obtaining insights into the chemical mechanisms and kinetics of red mud neutralization by both atmospheric (passive treatment) and injected CO2 (active treatment) in the absence and presence of gypsum. Active treatments allowed effective sequestration of CO2 within 1 h. A mixing ratio of gypsum to red mud at 0.04–0.06 enabled effective control of pH rebound, completely eliminating the causticity of the red mud by reducing the pH value of red mud to < 9. The carbonation of red mud was realized through the formation of carbon-containing minerals, mainly basic aluminium carbonates (largely dawsonite), sodium bicarbonate, sodium carbonate and calcite. The importance of calcite as a carbon carrier increased when gypsum was added. Passive treatments also allowed simultaneous causticity reduction and carbon sequestration but at a much slower rate compared to the active treatments. The research findings obtained from this study have implications for developing strategies to cost-effectively manage red mud. Where flue gas is available, active treatment could be a feasible option for simultaneously reducing the harmfulness of red mud and CO2 emission. Passive treatment can be used as a natural attenuation process for low-cost management of red mud. Where off-site utilization of red mud is feasible, gypsum addition at an optimal rate could be a more appropriate option. For future study, industrial-scale experiments are required to validate the research findings obtained from this laboratory-scale study. • Injection of CO2 into red mud slurry with added gypsum allowed simultaneous causticity reduction and carbon capturing. • Passive treatment by exposing red mud slurry to atmospheric air had the same effects but at a much slower rate. • Carbon sequestration was via formation of dawsonite, sodium (bi)carbonate and calcite, which were stable upon aging.


Graphical Abstract 1 Introduction
Aluminium is one of the most important materials that gained wide applications in everyday life.The global demand for aluminium is rapidly growing, which unavoidably drives the increase in its production.Aluminium is produced from smelting of alumina, which in turn is extracted from bauxite (aluminium-containing ore).Alumina refining results in substantial amounts of solid waste, which is collectively known as red mud due to its characteristic colour.To extract alumina from bauxite, NaOH is added to solubilize bauxite-borne aluminium.The presence of residual NaOH makes fresh red mud usually having a pH between 11 and 13 and thus being highly caustic (Gräfe et al. 2011).Red mud is therefore viewed as a hazardous material that requires proper management to minimize its adverse impacts on environment and human health (Liu et al. 2014).Containment of red mud in engineered storage facilities is a general practice for red mud management (Gräfe and Klauber 2011).However, construction of these engineering structures is usually very costly, especially where there is a competitive demand for other land uses.Failure of red mud storage facilities could have disastrous consequence for the surrounding environments (Tabereaux 2010).
While acid neutralization is effective for reducing the causticity of red mud (Mishra and Rao 2020), it is costly if the acid agents (either mineral or organic acids) need to be purchased from commercial sources.In addition, safe handling of mineral acids such as H 2 SO 4 , HCl and HNO 3 for large-scale red mud treatment could be challenging.The alkaline nature of red mud can be utilized to make acid neutralizing agents for remediating acidic soils (Lin et al. 2004;Liu et al. 2009;Niu and Lin 2021) and acidic mine water (Keller et al. 2020;Qin et al. 2022).Red mud is also a good sorbent for immobilization of heavy metals (Liu et al., 2007;Ma et al. 2009;Milenković et al. 2016), radionuclides (Chen et al. 2022) and organic pollutants (Martins et al. 2020).However, such beneficial uses of red mud only allow consumption of small amounts of red mud and therefore do not play a significant role in solving the environmental problems associated with generation of red mud from the aluminium industry.Alternatively, reduction of red mud causticity could be achieved by neutralizing the red mud-borne NaOH with CO 2 , which also creates an avenue for mitigation of global warming caused by CO 2 emission (Si et al. 2013).
In theory, when CO 2 is injected into a NaOH solution, the following sequential reversible reactions take place (Yoo et al. 2013).Initially, CO 2 reacts with NaOH to produce sodium bicarbonate (NaHCO 3 ): Sodium bicarbonate can react with another NaOH molecule to further eliminate an OH − and produce sodium carbonate (Na 2 CO 3 ): When the concentration of Na 2 CO 3 in the solution exceeds its saturation point, Na 2 CO 3 could precipitate until the equilibrium concentration of NaOH is reached.After this, the incoming CO 2 will turn to Na 2 CO 3 and produce NaHCO 3 again: When the concentration of NaHCO 3 in the solution exceeds its saturation point, NaHCO 3 could precipitate until the equilibrium concentration of Na 2 CO 3 in Eq. ( 3) is reached if the supply of CO 2 is not limited.
If the supply of CO 2 is cut off, the reverse reaction of Eq. ( 1) could take place, leading to release of CO 2 back to atmosphere and rebound in solution pH due to reformation of NaOH.The reverse reaction of Eq. ( 1) could then induce the reverse reaction of Eq. ( 2), which leads to hydrolysis of Na 2 CO 3 and further reformation of NaOH.As a result, OH − activity in the solution re-increases, causing the rebound in solution pH.
Red mud is a mixture of different minerals.Apart from NaOH, other forms of alkaline materials such as Al(OH) 4 − , Ca(OH) 2 and Fe(OH) 4 − that are produced from reaction between the bauxite-borne minerals and the added NaOH are also present in red mud.These alkaline materials could also contribute to carbon capture in red mud.Therefore, the chemical mechanisms responsible for simultaneous causticity reduction and carbon sequestration are expected to be far more complex, compared to the pure NaOH solution systems.Several pieces of preliminary work on the use of red mud slurry to capture injected CO 2 have been reported.Yadav et al. (2010) investigated the effects of different particle size fractions of the red mud on sequestering injected CO 2 under high pressure conditions.Due to removal of soluble alkaline materials during separation of red mud grain fractions in their experiment, the pH values of the red mud fractions were all below 8, which deviated markedly from the alkaline nature of the original red mud.Therefore, it was likely that CO 2 sequestration by the de-alkalized red mud was through reaction of carbonic acid with red mud-borne Ca-containing minerals to form CaCO 3 .Furthermore, they did not determine the C concentration in the treated red mud.Rather, they estimated the degree of (1) carbon sequestration by measuring the weight gain of the treated red mud after CO 2 injection, which might not be reliable in terms of quantification of the captured carbon.Sahu et al. (2010) conducted batch experiments to investigate the neutralization of red mud using multiple CO 2 sequestration cycles under ambient conditions.They were able to detect the presence of CO 3 2− and HCO 3 − in the slurry and attributed the observed carbon capture to the formation of CaCO 3 through the reaction of the injected CO 2 with the Ca-containing minerals present in the red mud.However, they estimated the amount of carbon captured by the red mud solely based on SEM/EDX results, which actually only semi-quantitatively reflected the concentration of carbon on the surface of red mud particles being used for EDX analysis rather than the total C concentration in the bulk red mud.They did not detect Na 2 CO 3 and NaHCO 3 from the neutralized red mud.But this was because the Na 2 CO 3 and NaHCO 3 were soluble and thus separated from the solid phase of the treated red mud during centrifugation.Therefore, the experimental design in Sahu et al. 's study (2010) did not allow the evaluation of the CO 2 capturing via formation of Na 2 CO 3 and NaHCO 3 .Han et al. (2017) conducted an experiment to observe the pH change in red mud slurry following injection of CO 2 of varying doses and found that the magnitude of pH drop increased with increasing CO 2 doses.However, the pH of the treated red mud rebounded to above 9.6 after injection of CO 2 was stopped, suggesting that part of the captured CO 2 was released back to atmosphere through the reverse reaction described by Eq. (1).Clearly, while CO 2 sequestration by red mud can be partially realized via formation of Na 2 CO 3 and NaHCO 3 , this process is not capable of completely eliminating the causticity of red mud.
In Ca-rich red mud, NaHCO 3 could react with Ca to form practically insoluble CaCO 3 and remove the soluble NaHCO 3 and Na 2 CO 3 from the red mud slurry.This leads to enhancement of NaOH-consuming reactions shown in Eqs. ( 1) and (2).For red mud that contains insufficient amount of Ca, addition of Ca and Mg to the red mud has been suggested (Si et al. 2013;Ma et al. 2014).Dilmore et al. (2008) used Ca-containing saline water as a source of Ca to facilitate simultaneous neutralization of red mud and sequestration of flue gas-borne CO 2 .
Gypsum (CaSO 4 •2H 2 O) is a readily available source of Ca for being used in the treatment of alkaline soils and red mud (Lehoux et al. 2013;Maryol and Lin 2015).In particular, gypsum is an abundant waste from flue gas desulfurization process in coal-fired power stations (Wang et al. 2017).Therefore, the use of desulfurization gypsum and flue gas from coal-fired stations for red mud treatment allows three abundant industrial wastes being managed in a circular economy way.Passive treatment of red mud by atmospheric CO 2 in the presence of gypsum has been practised to reduce the pH of red mud for revegetation in red mud disposal areas (Wong and Ho 1993).Han et al. (2017) monitored the change in pH of the gypsum-treated red mud under ambient conditions during a period of 55 days and observed a clear trend showing a decrease in pH from initially nearly 12 to slightly higher than 8.It was shown that the total carbon concentration in the treated red mud was markedly affected by the ratio of solid to water with a lower solid/water ratio favouring the sequestration of carbon by red mud.This indicates that the length of incubation time set in their experiment might not be sufficient to allow a full carbonation of the red mud.In addition, only one mixing ratio of gypsum to red mud was set in their experiment and the ratio was at 50:50, which appeared too high, causing overdosing of the red mud by gypsum.
To develop cost-effective methods for simultaneous red mud causticity reduction and carbon sequestration, a more in-depth understanding of the interaction between red mud and CO 2 that enters into red mud either actively or passively is required.The aim of this study was to fill the knowledge gaps identified above.The specific research objectives were to (a) examine the effects of air or CO 2 injection on causticity reduction and carbon sequestration in the treated red mud; (b) examine the effects of added gypsum on halting pH rebound and stabilizing the captured carbon when red mud slurry is neutralized by injected CO 2 ; (c) observe the long-term temporal variation in pH and total C in red mud upon exposure to atmospheric CO 2 in the absence and presence of added gypsum; (d) evaluate the effects of gypsum application rate on simultaneous causticity reduction and carbon sequestration by red mud that was passively treated with atmospheric CO 2 ; (e) test the stability of the captured carbon in the treated red mud upon aging; and (f ) test the thermal stability of the captured carbon in the treated red mud.

Materials used in the experiments
The fresh red mud used in the experiments was collected from the red mud discharge point at an alumina refinery in Shandong Province, China.It had a pH of 12.05 and contained 27.5% of Fe 2 O 3 , 24.4% of Al 2 O 3 , 18.54% of SiO 2 , 5.23% of TiO 2 , 1.01% of CaO, and 0.21% of C, respectively.After collection, the red mud was ovendried at 40°C, ground to pass a 2 mm sieve and stored in a sealed plastic bag to minimize air exposure prior to use in the experiments.
The analytical grade gypsum powders (CaSO 4 •2H 2 O) used in the experiments were purchased from J&K Scientific Co., Ltd. in Beijing.The CO 2 gas used in Experiment 1 was purchased from Praxair Industrial Gas Co., Ltd. in Guangzhou.The purity of the CO 2 gas was 99.9% according to the manufacturer's specification.

Experiment 1: active treatment of red mud slurry by air or CO 2 injection
Six treatments were set for this experiment (Table S1).
Treatment 1 (TA NG ) involved injection of air while Treatments 2-5 (TI NG , TI G2 , TI G3 , TI G4 and TI G5 ) involved injection of the CO 2 gas with varying amounts of added gypsum (0, 4, 6, 8 and 10%, respectively).For each treatment, 50 g of red mud was placed in a 125 mL glass bottle with the respective amount of gypsum where applicable.100 mL of deionized water was then added into the bottle to form red mud slurry.Air or CO 2 was injected into the bottle with a flow rate of 0.5 mL/min using an Alicat standard CODA controller.
For each treatment, eight sets (each in 3 replicates) of sub-experiments were run to allow measurement of slurry pH and harvest of the carbonated red mud at the following reaction times: 1 min, 5 min, 10 min, 15 min, 1 h, 2 h, 4 h and 8 h.After harvest, each of the treated red mud samples was oven-dried at 40°C, ground to pass a 0.5 mm sieve, and stored in a sealed plastic bag to minimize air exposure prior to sample analysis.

Experiment 2: passive treatment by exposure of red mud slurry to atmospheric CO 2
The second experiment consisted of five treatments (Table S2) with varying amounts of added gypsum, corresponding to those set in Experiment 1.For each treatment, 50 g of red mud was placed in a 125 mL plastic bottle with 100 mL of deionized water added into the bottle to form red mud slurry.The bottle was not capped to allow exposure of the red mud slurry to atmosphere during the entire period of the experiment.Two sets of sub-experiments (each in 3 replicates) were performed to allow the harvest of carbonated red mud on two occasions: one on Day 11 and another at the end of the experiment (Day 168).After harvest, each of the treated red mud samples was oven-dried at 40°C, ground to pass a 0.5 mm sieve and stored in a sealed plastic bag to minimize air exposure prior to sample analysis.During the entire period of the experiment, an appropriate amount of deionized water was added weekly to compensate water loss of the slurry due to evaporation, which was done by adding the deionized water to the level (marked on the plastic bottle) equivalent to that at the beginning of the experiment when the red mud slurry was formed.
The pH and electrical conductivity (EC) in the slurry were monitored during the 168-day experiment.

Evaluation of aging effects on pH and the captured carbon in the treated red mud
To evaluate the stability of pH and the carbon captured by the treated red mud, each of the carbonated red mud samples was stored in a sealed plastic bag with sufficient air space under ambient conditions for two years before re-measurement of pH and the carbon content in the carbonated red mud.The rate of carbon loss resulted from aging (C aging ) was calculated by the following formula: where CA 1 stands for carbon content in the red mud after the carbon capturing experiment (i.e., at the beginning of the aging experiment) while CA 2 denotes carbon content in the red mud after 2-year aging.

Evaluation of thermal stability of the carbonated red mud
The thermal stability of the carbonated red mud was evaluated by thermogravimetric analysis to observe the weight change of the red mud with a temperature range from room temperature to 800 °C.In addition, the carbon loss of red mud at 105 °C was also measured to estimate the amount of carbon captured in the form of sodium bicarbonate.For each aged red mud, 10 g of the sample was moistened with 5 mL of deionized water and then heated in an oven at 105 °C for 1 day.The oven-dried red mud samples were then ground to fineness for measurements of pH and carbon content.The rate of carbon loss caused by thermal decomposition at 105 °C (C thermal ) was calculated using the following formula: where CT 1 and CT 2 represent the carbon content of the red mud before and after being heated at 105 °C for 1 day, respectively.

Analytical methods
The pH and EC in the red mud slurry were measured using a calibrated pH meter (model: SX-620) and an EC meter (model: DDSJ-308A), respectively.A TOC analyzer (Vario TOC Elementar, Germany) was used to determine the content of carbon in the treated red mud.The mineral composition of the treated red mud was determined by X-ray diffraction (XRD, Ultima IV, Rigaku, Japan).Selected carbonated red mud samples were also examined by scanning electron microscopy (SEM, EVOMA 15, Zeiss, Germany).The surface of selected red mud particles was determined by energydispersive X-ray spectroscopy (EDS, Quantax200 with X-Flash 6/100, Brüker, USA).Thermogravimetric analysis of the carbonated red mud samples was performed with a TG-DSC STA 499 F3 instrument (NETZSCH, Germany) using the heating rate set at 10 °C/min.
For measurement of the pH in red mud after aging and thermal treatment at 105 °C, 1:2.5 (red mud:water) extracts were prepared and the pH in the extracts was measured using a calibrated pH meter (model: SX-620).

Statistical methods
The experiments were performed in triplicate.All the experimental data were analyzed using IBM SPSS ® Statistics 22.0 software.One-way analysis of variance (ANOVA) with Tukey's HSD test was used to compare the effects of different treatments on pH, EC, and carbon content.The statistical significance of carbon content in the treated red mud exposed to atmospheric CO 2 between the 11 th day and the 168 th day of the experiment was determined by an independent sample t-test at the 0.05 level.The statistical significance of carbon content in red mud-borne carbon before and after 2-year aging and before and after thermal treatment was also determined by the above-mentioned independent sample t-test.

Experimental results on active treatment of red mud with air and CO 2
For the red mud slurry treated with injected atmospheric air in the absence of gypsum (TA NG ), the pH slightly decreased from initially 12.10 to 11.30 after injection of the air for 1 min.The pH then maintained at a level above 11.13 with fluctuation until the 4 th h of the experiment.After injection of the air for 8 h, the slurry pH dropped to 10.86.For the CO 2 injection treatments, all the slurry pH were significantly lower for any given length of CO 2 injection time compared to TA NG .Depending on the dosage level of the added gypsum, the magnitude of pH dropped differently.For the red mud slurry without added gypsum (TI NG ), the pH dropped to 9.87 after 1 min of CO 2 injection.The pH then decreased to below 9 after injection of CO 2 within a length of time between 5 and 15 min and below 8 within a length of time between 15 and 60 min.For the treatments with added gypsum, the pH all dropped to below 9 after 1 min of reaction.However, the pH of slurry was significantly higher in the treatment with 2 g of gypsum (TI G2 ) than in the treatments with a higher dose of gypsum (TI G3 , TI G4 and TI G5 ) and this situation maintained at least within the first 10 min of CO 2 injection.After 15 min of CO 2 injection, the pH of the treatments with added gypsum were all below or slightly above 7 regardless of the gypsum dosage level used (Table 1).

Temporal variation in total carbon in the treated red mud
For the red mud with injected atmospheric air, there was no significant change in total C content in the red mud after injection of air for 15 min.However, after 1 h of air injection, a carbon content of 0.68% was recorded in the treated red mud.It is interesting to note that the total C content increased to 0.91% at the 2 nd h of the air injection and then bounced back to 7.3% at the 4 th h of air injection and increased again to 0.84% at the 8 th h of air injection.
For the treatments with the concentrated CO 2 injection, a total C content of around 0.7% was recorded after injecting the CO 2 into the red mud slurry regardless of the dosage level of gypsum.The total C content increased from the 1 st min to the 5 th min of CO 2 injection for all the treatments, but the magnitude of increase differed among the different treatments, showing the following decreasing order: TI G3 > TI G2 > TI G4 > TI NG > TI G5 .The red mud-borne C then tended to decrease from the 5 th min to the 60 th min with the total C content being significantly lower in TI NG and TI G2 compared to the treatments with a higher gypsum dose (i.e., TI G3 , TI G4 and TI G5 ).Between the 1 st and the 8 th h of CO 2 injection, the total C content fluctuated and at the end of the experiment (the 8 th h), there was no significant difference in the total C content among the treatments with added gypsum regardless of the dosage level of gypsum, which was significantly higher compared to TA NG and TI NG (Table 2).

XRD results
The XRD results of the treated red mud without added gypsum (T NG ) showed a weak peak corresponding to calcite (CaCO 3 ) and a few peaks possibly corresponding to dawsonite (NaAlCO 3 (OH) 2 ) (Fig. 1a).For the treatment with 2 g of gypsum (T G2 ), the CaCO 3 peak became stronger compared to T NG (Fig. 1b).When the dose of gypsum was increased to 3 g (T G3 ), weak peaks corresponding to gypsum were detected (Fig. 1c).At a higher dose of gypsum (T G4 and T G5 ), the gypsum peaks became more and more evident (Fig. S1).There was no marked difference in the intensity of calcite peak among T G3 , T G4 and T G5 .The possible dawsonite peaks were detected for all the treatments with added gypsum.

Experimental results on passive treatment of red mud with atmospheric CO 2
For the control (NG, red mud without added gypsum), the pH slightly decreased from 12.05 to 11.01 after the red mud slurry was exposed to air for 1 day.In contrast, the pH in the red mud with added gypsum (G2-G5) all dropped to below 10.50, significantly lower compared to the control.For the control, the pH of the slurry showed  a gradual decreasing trend over time dropped to below 10 at least before the 56 th day of the experiment and then maintained at a level above 9.5 with fluctuation until the 168 th day of the experiment.For treatments with added gypsum (G2-G5), the pH of the red mud slurry maintained at a level of 10 at least until the 27 th day of the experiment.For the treatment with 2 g of added gypsum (G2), a pH value above 9 maintained for the whole period of the experiment while a pH below 9 was obtained on the 168 th of the experiment for all the treatments with a higher gypsum dose (G3-G5) (Fig. 2).

Variation in EC during the period of passive treatment
The EC of the red mud slurry in the control (NG) was 7.73 dS/m after 1 day of air exposure.It then showed a trend to decrease over time with fluctuation and dropped to below 3 after the 154 th day of the experiment.Addition of gypsum caused increase in the slurry EC with a value > 12 dS/m on the 1 st day of the experiment and > 10 dS/m at least before the 20 th day of the experiment.However, no evidence showed that EC increased with increasing dosage level of gypsum.Like the control, the EC in the treatments with added gypsum all showed a decreasing trend over time during the period of the experiment.At the end of the experiment (i.e., the 168 th day of the experiment), the EC value of treatments with added gypsum all dropped to below 6 dS/m (Fig. S2).

Total C in the dried red mud after exposure to atmospheric CO 2
On the 11 th day of the experiment, the total C content in the dried red mud was below 0.35% for the treatments with the added gypsum < 3 g (i.e., NG, G2 and G3).The total C content significantly increased to above 0.9% for the treatments with the added gypsum of 4 and 5 g (i.e., G4 and G5).On the 168 th day of the experiment, the total C contents in the dried red mud all increased to above 0.65% and there was a clear trend showing that the red mud-borne C increased with the increase in gypsum dose from 0 to 3 g.However, there was no significant difference in the red mud-borne C between G3 and G4 or G5.By comparison, the red mud-borne C was significantly lower on the 11 th day than on the 168 th day in the treatments with gypsum dose < 3 g, but there was no significant difference in the red mud-borne C between the 11 th day and the 168 th day in the treatments with higher gypsum doses (Fig. 3).

XRD results
For the control, a weak peak corresponding to calcite was observed for the dried red mud on both the 11 th day and the 168 th day of the experiment.The possible dawsonite peaks were also recorded (Fig. 4a).For the treatment with 2 g of added gypsum (G2), the calcite peak became stronger compared to the control (Fig. 4b).When the gypsum dose was increased to 3 g, weak gypsum peaks were observed on the 11 th day of the experiment only and no gypsum peaks were recorded for the dried red mud collected on the 168 th day of the experiment (Fig. 4c).For the treatments with higher doses of added gypsum (G4 and G5), strong gypsum peaks were observed on both the 11 th and 168 th day of the experiment (Fig. 4d and e).

Change in pH and total C content after 2-year aging
After 2-year aging, the pH of the red mud in all the injected CO 2 treatments bounced back significantly.In particular, the pH for TI NG (no added gypsum) increased from 7.27 to 10.29.For treatments with added gypsum, the pH of red mud all increased from originally below 7 to above 8.6.In contrast, the pH of red mud in the  1 and S3).Mixed results were observed among the different active treatments in terms of carbon loss due to aging.The C aging value for TA NG , T NG , T G2 , T G3 , T G4 and T G5 was 10.2, -9.6, 10.7, 1.0, 1.5 and 1.9%, respectively.However, there was no statistically significant (P > 0.05) change in red mud-borne carbon compared to that at the start of the aging experiment except for TI NG , which had a significantly higher C content (negative C aging value) after aging than before aging (Fig. 5a).
In contrast, only NG (no added gypsum) and G2 (2 g of added gypsum) showed an increase in pH after 2-year aging for the passive treatments.All passive treatments with added gypsum at a level of 3 g and above had a pH slightly lower than that at the beginning of the aging experiment (Fig. 2 and Table S3).There was a consistent trend showing that the red mud-borne C increased after 2-year aging although the difference was not statistically significant for the treatment without added gypsum (T NG ) (Fig. 5b).

Change in C content and pH of the carbonated red mud after heating at 105 °C
After heating the moistened carbonated red mud at 105 °C for 1 day, the total C in the heated red mud tended to be lower compared to its unheated counterpart for all the CO 2 treatments although there was no statistically significant difference in TA NG and TI G2 for the CO 2 injection treatment (Fig. 6a) and in NG and G4 for the atmospheric CO 2 exposure treatment (Fig. 6b).However, no significant change in pH after thermal treatment was observed for any samples from either the air/CO 2injection or air exposure experiments (Tables S3 and S4).

Thermogravimetric analysis of the carbonated red mud
The results of thermogravimetric analysis of the carbonated red mud in the control and selected treatments with added gypsum for both active carbonation treatment and passive carbonation treatment are shown in Figs.S3 and S4, respectively.There was a high level of similarity among them in terms of the thermogravimetric curve.A sharp weight loss was observed when the temperature increased from room temperature to about 150 °C, followed by the maintenance of a similar weigh loss rate until a temperature of about 350 °C was reached.The weight loss rate then decreased but with two additional peaks occurring in temperature ranges of 500-550 °C and 700-750 °C, respectively.However, the peak at 700-750 °C tended to be less evident in the control compared to those in the treatments with added gypsum.

SEM-EDS results
For the red mud treated with injected CO 2 in the absence of gypsum (TI NG ), the major elements on the surfaces of red mud particles detected by EDS included O, Al, Si, Na, C and Fe (Fig. 7).When gypsum was added, Ca was detected in addition to the above elements.A similar surface elemental composition was observed for the red mud treated passively (Fig. S5).The surface chemical composition detected by EDS deviated from the bulk chemical composition of the red mud, as detected by XRF, particularly for Fe.The C content detected by EDS was at least 5 times higher than that detected by TOC analyzer for the bulk carbonated red mud (Fig. 5).

Discussion
Under the active treatment of red mud slurry by injecting concentrated CO 2 in the absence of gypsum (TI NG ), a substantial amount of carbon was sequestered by the red mud within just 1 min and a carbon peak was attained at the 5 th min of the experiment while the pH of the slurry was 9.59.This corresponded to chemical reaction shown in Eq. ( 1), indicating that the injected CO 2 rapidly reacted with the NaOH in the slurry to form NaHCO 3 while sufficient amounts of OH − was still present in the solution to uphold an alkaline status of the slurry.Drops in pH to below 9 at the 15 th min and to below 8 at the 60 th min suggested that most of the free OH − in the solution was consumed within 1 h through the reaction shown in Eq. ( 2) without causing an increase in the captured carbon.The observed decreasing trend over time for the treated red mudborne C from the peak value at the 5 th min of the experiment suggested that the reaction as shown in Eq. ( 3) did not take place to a significant degree or the reverse reactions described by Eqs. ( 1), ( 2) and (3) took place during drying of the red mud after CO 2 injection was stopped.It is surprising that nahcolite (NaHCO 3 ) and trona (Na 2 CO 3 ) were not detected by XRD analysis, indicating that these minerals, if any, were present in the treated mud at a level under the detection limits of XRD.This suggested that precipitation of nahcolite and trona from the treated red mud slurry during drying might be impeded.The detection of possible presence of dawsonite suggested that NaHCO 3 tended to react with Al(OH) 3 to form dawsonite: This may partially explain the lack of XRD-detectable nahcolite and trona in the treated red mud.There was no detectable calcite in the original red mud.However, calcite was detected in TI NG , suggesting occurrence of reactions between the H 2 CO 3 in the solution and the red mud-borne Ca via the following possible chemical reactions: (4) The presence of quartz (SiO 2 ) and rutile (TiO 2 ) in the carbonated red mud in TI NG appeared to support the above assumption.
In the presence of gypsum, NaHCO 3 and NaCO 3 could react with gypsum to form calcite: This explains the increased presence of calcite in the gypsum-treated red mud compared to the treatments without added gypsum (Figs. 1 and S1).Increase in gypsum dose from 2 to 3 g enhanced the formation of calcite.However, a further increase in gypsum dose did not lead to further enhancement of calcite formation, suggesting that a ratio of gypsum to red mud at 0.06 was sufficient to sequester the injected CO 2 while effectively reducing the causticity of the red mud used in this study.Han et al. (2017) used a ratio of gypsum to red mud as high as 1 (50:50) to simultaneously reduce the red mud pH and capture CO 2 .It appears that such a high dosage level of gypsum is unnecessary for red mud treatment.
The reactions shown in Eqs. ( 4) and ( 7) enhanced the consumption of NaHCO 3 , which favoured the reaction shown in Eq. ( 1), leading to accelerated neutralization of OH − and thus rapid pH drop.It is interesting to note that reduction in pH to circumneutral status was achieved by the injected CO 2 no matter whether gypsum was added into the slurry or not.This appeared to suggest that formation of dawsonite as shown in Eq. ( 4) alone was sufficient to reduce the causticity of red mud to an acceptable level if injection of concentrated CO 2 (7) was practised.In contrast, the maintenance of a high pH (10.86) in the red mud treated with injected air for 8 h suggested that the amount of CO 2 entering into the red mud slurry during this period of time was far from sufficient to neutralize the OH − present in the red mud slurry.This was consistent with the observation in pure NaOH system when CO 2 gas at low concentration was used (Shim et al. 2016).To further reduce the causticity of the red mud using injected atmospheric air, extension of injection time is required.
The results obtained from Experiment 2 demonstrated that passive treatment by exposing red mud slurry to atmospheric CO 2 also resulted in simultaneous causticity reduction and carbon sequestration.However, they were much slower compared to those in active treatments.If gypsum was not added into red mud, the pH did not drop to below 9.6 even after 168 days of air exposure.This suggested that, in the absence of gypsum, the partial pressure of CO 2 in atmospheric air was too weak to drive the reactions shown in Eqs. ( 1) and ( 2) to consume a sufficient amount of OH − in the red mud and thus brought the pH down to a non-caustic level.When gypsum was added to red mud at a ratio of 0.04, the pH of the red mud was still above 9, This indicates that when the supply of Ca is limited, consumption of NaHCO 3 via formation of calcite (Eq.( 5) was impeded, failing to reduce the causticity of the red mud to an acceptable level.It is worth noting that a significant increase in the captured carbon was achieved within just 11 days of air exposure if the gypsum dose was raised to above 4 g (Fig. 3).This suggests that the supply of atmospheric CO 2 is not necessarily a limiting factor for the carbonation reaction shown in Eq. ( 1) if the supply of Ca is sufficient to accelerate the NaHCO 3 -consuming reaction shown in Eq. ( 7).While it was clearly shown that the slurry pH generally decreased over time, fluctuation did occur during the period of the 168-day experiment.The pH in the red mud slurry was controlled by CO 2 exchange between the slurry and the atmosphere, which was temperature-dependent.Therefore, the observed fluctuation in pH of slurry was likely to be largely caused by the variation in temperature during the period of the experiment.
It is worth noting that addition of gypsum caused increase in salinity, as indicated by EC (Fig. S2).Gypsum is slightly soluble, resulting in an increase in soluble Ca 2+ and SO 4 2− in the red mud slurry.However, EC value of the red mud decreased markedly over time (Fig. S2), possibly via removal of soluble Ca 2+ and Na + from the solutions by formation of practically insoluble calcite, dawsonite and sodium aluminium silicate hydrates, as confirmed by the XRD results.This aging effect on salinity reduction favours the off-site uses of the neutralized red mud.
The much higher C content detected by EDS compared to that in the bulk red mud suggested that the carbonation reactions primarily took place in the interface between the solution and red mud particles of the slurry and the newly formed C-containing compounds deposited as coatings on the surfaces of the original red mud particles after drying.Therefore, evaluation of carbon capturing capacity based on C content in the carbon-rich coating materials, as determined by SEM-EDS (Sahu et al. 2010) could dramatically overestimate the actual value.It is interesting to note that the molar ratio of C to Ca from the EDS analysis for treatments TI G3 and G3 were both greater than 20 (Figs. 7b  and S3b) and it was far greater than 1 for CaCO 3 .This suggests that calcite is not the major carbon-carrier even in the carbonated red mud with the presence of gypsum.The molar ratio of C to Al was 0.80, 0.99, 1.30 and 1.86 for TI NG , TI G3 , NG and G3, respectively (Figs. 7b and S3b).These values were much closer to 1 for NaAl(OH) 2 CO 3 .Therefore, it appears that dawsonite plays a more important role in holding the captured carbon.It is also possible that carbon is captured through formation of basic aluminium carbonate (Cao et al. 2012): The presence of NaHCO 3 in all the carbonated red mud was confirmed by the fact that the C content in the carbonated red mud was reduced after heating at 105 °C, indicating thermal decomposition of NaHCO 3 because NaHCO 3 becomes unstable at a temperature greater than 80 °C (Ball et al. 1986), which leads to the release of some CO 2 back to atmosphere: It is unlikely that this carbon loss was caused by thermal decomposition of dawsonite, which is stable at a temperature below 250 °C (Lundvall et al. 2019).The fact that the red mud-borne carbon was consistently reduced after heating at 105 o C for 1 day suggests that NaHCO 3 was present in the carbonated red mud no matter whether gypsum was added or not.The average rates of carbon loss caused by thermal decomposition at 105 o C (C thermal ) were 7.6% and 7.1% for the active carbonation treatment and passive carbonation treatment, respectively.From Eq. ( 10), it can be seen that 50% of the carbon carried by NaHCO 3 is lost after thermal decomposition at 105 o C. Therefore, the amount of carbon carried by red mudborne NaHCO 3 was estimated to be 15.2% and 14.2% of the total C in the carbonated red mud. (9) The sharp weight loss from room temperature to 105 °C, as indicated by thermogravimetric analysis could be primarily attributed to the dewatering and CO 2 emission caused by decomposition of NaHCO 3 while the weight loss in the temperature range of 105-250 °C was likely to be caused solely by CO 2 emission due to NaHCO 3 decomposition.The bigger weight loss in the latter compared to the former suggests that, under the operational conditions set for this instrument, the thermal decomposition of the red mud-borne NaHCO 3 largely takes place at the higher temperature range.The heating time changed from room temperature to 105 °C within 10 min for the thermogravimetric analysis, which was much shorter than 1 day in the experiment that determined carbon loss caused by thermal decomposition at 105 o C. Therefore, the majority of the red mudborne NaHCO 3 could only be decomposed during the period when the temperature was increased from 105 to 250 °C.In the temperature range of 250-400 °C, it was likely that the weight loss was caused by decomposition of dawsonite and gibbsite with the small peak at around 500 o C being also attributable to gibbsite dehydration (Zsolt et al. 2013).The peak occurring in temperature range of 700-750 °C is attributable to decomposition of calcite (Karunadasa et al. 2019).The much weaker calcite peaks in the control for both active (Fig. S3a) and passive (Fig. S4a) treatments were consistent with what was found in XRD spectra (Figs. 1 and 4).
The results obtained from the aging experiment indicate that, in general, the captured carbon in the carbonated red mud is stable under dried conditions.The slightly increased C content in the red mud derived from passive carbonation treatments suggested that further carbonation took place during the period of 2-year aging.However, upon mixing the red mud with water, pH rebound did occur with the red mud without added gypsum (TI NG and NG) having the pH risen to above 10.For the passively carbonated red mud with 2 g of added gypsum, the pH rose to above 9.This can be explained by hydrolysis of Na 2 CO 3 , which leads to release of OH − and thus pH rises: The rise of pH level caused by the above chemical reaction depends on the concentration of Na 2 CO 3 in the solution with a pH range of 10-10.5 corresponding to a Na 2 CO 3 concentration range of 0.1-1 mM (Nakayama 1970).Therefore, the estimated amount of carbon captured in the form of Na 2 CO 3 ranged from 0.3 to 3 g in each kilogram of red mud for the red mud without added gypsum (TI NG and NG).
The findings obtained from this study have implications for developing strategies to cost-effectively manage (11) Na 2 CO 3 + H 2 O → 2Na + + HCO 3 − + OH − red mud.Although simultaneous causticity reduction and carbon sequestration can be achieved within a short period of time by active treatment of red mud, this is limited to locations where the source of flue gas is in close proximity to alumina refineries.Without adding gypsum, injection of CO 2 into red mud slurry alone is not capable of maintaining the pH to a level that meets the requirement for off-site beneficial uses of the treated red mud (Santini et al. 2011).Unless the treated red mud is stored in red mud impoundments, active treatment of red mud requires the addition of Ca or Mg-containing materials such as gypsum to turn the red mud to a harmless material for off-site valorization (Li et al. 2020).When land is available for storage of red mud safely, the slow carbon sequestration and causticity reduction by atmospheric CO 2 without adding gypsum can be explored as a natural attenuation process for low-cost management of red mud.Air injection may be used to accelerate the causticity reduction and therefore reduce the duration needed for natural attenuation of hazardous red mud.When offsite utilization of the neutralized red mud, such as using it as construction materials, is feasible, application of gypsum to reduce the causticity of the red mud could be a more appropriate option.However, it is important to optimize the application rate of gypsum to avoid causing unnecessary increase in red mud salinity.

Conclusion
Under the set experimental conditions, active treatment of red mud slurry allowed effective CO 2 capturing within a short period of time.However, air injection was not able to bring the pH down to a non-caustic level within a period of 8 h.Although injection of concentrated CO 2 into red mud slurry was capable of reducing the pH to circumneutral status, post-treatment rebound of pH to a hazardous level took place.Addition of gypsum enabled simultaneous carbon capturing and causticity reduction in the treated red mud and effective control of the pH rebound to an acceptable level.Under the set experimental conditions, a mixing ratio of 0.04-0.06for gypsum and red mud was sufficient to achieve this goal.The carbonation of red mud was via formation of basic aluminium carbonates (mainly dawsonite), sodium bicarbonate, sodium carbonate and calcite.In the presence of gypsum, the relative importance of calcite as a carbon carrier increased.Passive treatment by exposing red mud slurry to atmospheric CO 2 also allowed simultaneous causticity reduction and carbon sequestration.However, the process was much slower compared to the active treatments.Increase in gypsum dose markedly reduced the duration needed for carbonation but not causticity reduction.The findings obtained from this study have implications for developing strategies to cost-effectively manage red mud.When flue gas is readily available, active treatment of red mud with injected flue gas is a feasible option for reducing the harmfulness of red mud while simultaneously capturing the flue gasborne CO 2 .When land is available for safe disposal of red mud, the slow carbon sequestration and causticity reduction by atmospheric CO 2 without adding gypsum can be explored as a natural attenuation process for low-cost management of red mud.Air injection may be used to accelerate the causticity reduction and therefore reduce the duration needed for natural attenuation of hazardous red mud.When off-site utilization of the neutralized red mud is feasible, application of gypsum at an optimal rate could be a more appropriate option.

Fig. 1
Fig.1XRD spectra of the treated red mud with (a) no added gypsum (TI NG ), (b) 2 g of added gypsum (TI G2 ), and (c) 3 g of added gypsum (TI G3 ) derived from treatment with injected CO 2 for different time periods

Fig. 2
Fig. 2 Temporal variation in the red mud slurry pH during the period of air exposure experiment for the control and the treatments with gypsum

Fig. 3
Fig. 3 A comparison of total carbon in the dried red mud between the 11 days and 168 days of exposure to atmospheric CO 2 .Data are presented as mean ± SD (n = 3).Different letters (a, b, c, d) above the bars of various treatments for either the 11 th day or 168 th day indicate significant difference at p < 0.05.The double asterisk sign indicates significant difference (p < 0.01) between the two different sampling times for the same treatment

Fig. 5 Fig. 6
Fig. 5 Comparison between the red mud-borne carbon before and after 2-year aging for (a) active treatment and (b) passive treatment.The asterisk sign and double asterisk sign indicate a significant difference in red mud-borne carbon between the beginning and the end of the aging experiment at p < 0.05 and p < 0.01, respectively

Fig. 7
Fig. 7 Surface chemical composition of red mud particles treated with injected CO 2 (a) in the absence of gypsum, and (b) in the presence of gypsum at a ratio of 0.06 (gypsum: red mud)

Table 1
Temporal variation in pH of red mud slurry under different active treatments with atmospheric or concentrated CO 2 injection Data are presented as mean ± SD (n = 3).Different letters (a, b, c, d, e, f ) in the same row indicate significant difference at p < 0.05

Table 2
Total carbon content (%) in the treated red mud at different treatment times under different active treatment conditions Data are presented as mean ± SD (n = 3).Different letters (a, b, c, d, e, f ) in the same row indicate significant difference at p < 0.05