Integration of CO2 Capture and Mineral Carbonation by Using Recyclable Ammonium Salts

A new approach to capture and store CO2 by mineral carbonation using recyclable ammonium salts was studied. This process integrates CO2 capture with mineral carbonation by employing NH3, NH4HSO4, and NH4HCO3 in the capture, mineral dissolution, and carbonation steps, respectively. NH4HSO4 and NH3 can then be regenerated by thermal decomposition of (NH4)2SO4. The use of NH4HCO3 as the source of CO2 can avoid desorption and compression of CO2. The mass ratio of Mg/NH4HCO3/NH3 is the key factor controlling carbonation and the optimum ratio of 1:4:2 gives a conversion of Mg ions to hydromagnesite of 95.5 %. Thermogravimetric analysis studies indicated that the regeneration efficiency of NH4HSO4 and NH3 in this process is 95 %. The mass balance of the process shows that about 2.63 tonnes of serpentine, 0.12 tonnes of NH4HSO4, 7.48 tonnes of NH4HCO3, and 0.04 tonnes of NH3 are required to sequester 1 tonne of CO2 as hydromagnesite.


Introduction
Carbon dioxide capture and storage (CCS) is considered to be one of the main solutions for reducing anthropogenic CO 2 . CCS includes CO 2 capture from a point source, such as power plants, and transportation to a suitable site, where it is stored permanently and safely. Many projects for CO 2 storage are based on direct injection of CO 2 into underground formations (geologic sequestration) where it is stored by hydrodynamic, solubility, or mineral trapping. [1] However, the development of CO 2 geological storage has been slow with respect to potential environmental impact and regulation for CO 2 injection and monitoring. [2] Moreover, some countries, such as Finland and India, do not have sufficient storage capacity or lack suitable storage formations. [3] Therefore, there has been increasing interest in mineral carbonation.
The concept of CO 2 sequestration by mineral carbonation is based on accelerating the weathering of rocks. CO 2 reacts with alkaline earth oxide containing minerals to form insoluble carbonates. Magnesium and calcium silicate deposits, such as serpentine and olivine, can be used for this process. Due to the availability and abundance of these minerals, the capacity for mineral carbonation to store CO 2 is estimated to be quite large. [4] Serpentine is an important source for this process due to its worldwide availability. For instance, a deposit of 30 000 km 3 of magnesium silicates found in Oman would be able to store all of the anthropogenic CO 2 generated from combustion of the world's coal reserves. [5] Clearly, one of the main advantages of this process is the permanent safe storage of CO 2 due to the thermodynamically stable nature of the solid carbonates formed. [4] Moreover, carbonation is an exothermic process, which may reduce the overall energy consumption and costs. [5] However, the slow reaction rate of mineral dissolution is the main barrier to this process. [5] Many researchers have focused on promoting the dissolution rate by using different solvents, such as H 2 SO 4 , HCl, HNO 3 , organic acids, and inorganic salts. [3,[6][7][8] For example, Maroto-Valer et al. have reported 70 % dissolution of serpentine by using 2 m H 2 SO 4 in 2 h. [6] A multistep aqueous carbonation process developed by Teir et al. used 4 m HCl or HNO 3 to dissolve Mg ions from serpentine, then NaOH was used to control the pH of the solution to precipitate high purity hydromagnesite. [9] This process achieved 79 % carbonation efficiency at 80 8C and ambient pressure. They also reported that electrolysis of the NaCl solution was used to regenerate HCl and NaOH. [9] However, this process suffered from a high energy penalty in the regeneration process, in which the energy consumption for electrolysis of NaCl is 3277 and 4361 kWh t À1 CO 2 sequestered using HCl and HNO 3 , respectively. [9] Therefore, there is a need to find low-cost, recyclable solvents that can provide high efficiency of mineral dissolution and carbonation. Recently, Krevor and Lackner tested NH 4 Cl, NaCl, sodium citrate, sodium EDTA, sodium oxalate, and sodium acetate to dissolve serpentine. [10] All experiments were carried out at 120 8C and 20 bar (1 bar = 10 5 Pa) CO 2 in a batch autoclave. On using 0.1 m citrate, EDTA, and oxalate solutions, 60 % dissolution efficiency of magnesium from serpentine was achieved within 2 h, going up to 80 % after 7 h and reaching nearly 100 % between 10 and 20 h. Therefore, mineral dissolution with organic solvents is promising in terms of dissolution efficiency, but the dissolution rate is relatively slow. Pundsack A new approach to capture and store CO 2 by mineral carbonation using recyclable ammonium salts was studied. This process integrates CO 2 capture with mineral carbonation by employing NH 3 , NH 4 HSO 4 , and NH 4 HCO 3 in the capture, mineral dissolution, and carbonation steps, respectively. NH 4 HSO 4 and NH 3 can then be regenerated by thermal decomposition of (NH 4 ) 2 SO 4 . The use of NH 4 HCO 3 as the source of CO 2 can avoid desorption and compression of CO 2 . The mass ratio of Mg/ NH 4 HCO 3 /NH 3 is the key factor controlling carbonation and the optimum ratio of 1:4:2 gives a conversion of Mg ions to hydromagnesite of 95.5 %. Thermogravimetric analysis studies indicated that the regeneration efficiency of NH 4 HSO 4 and NH 3 in this process is 95 %. The mass balance of the process shows that about 2.63 tonnes of serpentine, 0.12 tonnes of NH 4 HSO 4 , 7.48 tonnes of NH 4 HCO 3 , and 0.04 tonnes of NH 3 are required to sequester 1 tonne of CO 2 as hydromagnesite.
reported the use of NH 4 HSO 4 to dissolve serpentine and bubbled CO 2 directly into the high-concentration magnesium solution obtained with aqueous ammonia to precipitate magnesium carbonates. [11] The dissolution efficiency of magnesium for this process was 92.8 %, but the carbonation efficiency was only 35 %. Fagerlund et al. proposed a process for the production of Mg(OH) 2 from serpentine by using (NH 4 ) 2 SO 4 . [12] A solid-solid reaction of serpentine with (NH 4 ) 2 SO 4 was carried out at > 440 8C to generate MgSO 4 , which was added to aqueous ammonia to precipitate Mg(OH) 2 and regenerate (NH 4 ) 2 SO 4 . Mg(OH) 2 was then carbonated with CO 2 directly in a pressurised fluidised bed (PFB) reactor at 470-550 8C and 20 bar. However, only 20-60 % extraction efficiency of magnesium from serpentine was reported, [13] and the carbonation efficiency of Mg(OH) 2 only achieved a maximum value of 50 %. This was due to the conversion of Mg(OH) 2 into MgO at the temperature range used, at which the produced MgO cannot react with CO 2 to produce carbonates. [14] Therefore, work is needed to improve both dissolution and carbonation efficiencies.
We have developed a new pH-swing mineral carbonation process by using recyclable ammonium salts and the process route was presented in a previous paper. [15] The modified process diagram can be seen in Figure 1. In this process, aqueous NH 4 HSO 4 was used to extract Mg from serpentine. The pH of the solution was then changed by adding aqueous ammonia, resulting in iron and silicon precipitating from solution. NH 4 HCO 3 and NH 3 were then added to the solution to react with Mg and produce carbonates and (NH 4 ) 2 SO 4 , which was recycled from the solution by evaporation and then decomposed back into NH 3 and NH 4 HSO 4 . Dissolution experiments of serpentine with NH 4 HSO 4 have been previously reported. [15] It was found that 1.4 m NH 4 HSO 4 could extract 100 % Mg from serpentine, as well as 98 % Fe and 17.6 % Si in 3 h at 100 8C. In addition, the dissolution kinetics of the reaction were found to follow the model of constant size particles with a rate-limiting control step of the chemical reaction by using product layer diffusion control. [15] It must be pointed out that this process is unique in using NH 4 HCO 3 instead of direct CO 2 gas for mineral carbonation. The advantages of using NH 4 HCO 3 include avoiding CO 2 desorption in the capture step and subsequent CO 2 compression for transportation, which are energy-intensive steps in the conventional CCS process. [15] We have previously reported the dissolution of serpentine with NH 4 HSO 4 . [15] Herein, we investigated carbonation with NH 4 HCO 3 and NH 4 HSO 4 and NH 3 regeneration. The carbonation experiments were conducted at different molar ratios of Mg/ NH 4 HCO 3 /NH 3 . Finally, the mass balance of all streams in this process is presented.

Experimental Section
Preparation of magnesium salt solutions from serpentine using NH 4 HSO 4 Previous dissolution experiments conducted by us have shown that NH 4 HSO 4 is suitable for extracting magnesium from serpentine. [15] The chemical equation for dissolution of magnesium from serpentine using NH 4  For the dissolution experiments, the same procedure and serpentine sample was used as in our previous paper. [15] Different temperatures (80, 90, and 100 8C) and reaction times (1, 2, and 3 h) were used for the preparation of solutions of MgSO 4 . After dissolution, the solution was cooled to room temperature and filtered by using 0.45 mm Pall syringe filters. The filtrate is referred to as filtrate 1 ( Figure 1) and was used for the pH regulation studies described in the section below on pH regulation and removal of impurities. The solid residue was dried at 105 8C overnight and is referred to as product 1 (Figure 1). Filtrate 1 was analysed by inductively coupled plasma atomic emission spectroscopy (ICP-AES) to measure the concentration of dissolved Mg, Fe, and Si. For the purpose of this analysis, filtrate 1 was acidified with 70 wt % HNO 3 to prevent precipitation of Mg and Fe. Product 1 was sampled and sent for X-ray fluorescence (XRF) analysis to determine the Mg, Fe, and Si con- tent. The details about the instruments used for ICP, XRD and XRF, as well as the errors of these analyses, can be found elsewhere. [15] pH regulation and removal of impurities About 40 % excess NH 4 HSO 4 was used for the dissolution of serpentine to maximise magnesium extraction. After dissolution, the pH values of the solution were about 0.9-1.2. Because the carbonation reaction was favourable at high pH values, it was necessary to increase the pH of the solution to alkaline values. The chemical reaction of the pH regulation is presented in Equation (2): The reason for using aqueous ammonia is because the above reaction produces ammonium sulfate, which can be converted back into NH 3 and NH 4 HSO 4 in the regeneration step to recycle the additives (Figure 1). If a high-value product (pure magnesium carbonate) is desired, some impurities, such as Fe, Al, Cr, Zn, Cu, and Mn, need to be precipitated from the system first by increasing the pH. During pH regulation and removal of impurities, aqueous ammonia (35 wt %) was added to filtrate 1 until the pH value was neutral. During this process, the solution was stirred and an in situ pH probe was used to measure the pH value. The solution was filtered with 0.7 mm Pall syringe filters. The filtrate is referred to as filtrate 2 ( Figure 1) and was used for the carbonation experiments described in the section below on precipitation of hydromagnesite using NH 4 HCO 3 . The solid residue was dried at 105 8C overnight and is referred to as product 2 ( Figure 1). Filtrate 2 was analyzed by using ICP-AES to quantify the concentration of different elements, including Mg, Si, Fe, Mn, Zn, Cu, Al, and Cr. Product 2 was analysed by using XRF and XRD to quantify its composition and identify the mineral phases present.

Precipitation of hydromagnesite using NH 4 HCO 3
The reaction of precipitation of hydromagnesite by treating MgSO 4 with NH 4 HCO 3 and NH 3 is presented in Equation 5: The formation of magnesium carbonate species depends on temperature and pressure. [16] Nesquehonite (MgCO 3 ·3 H 2 O) can precipitate from aqueous solutions at ambient temperatures as described in Equation 5 a, whereas at higher temperatures (50 and 100 8C), nesquehonite is transformed into hydromagnesite (4 MgCO 3 ·Mg(OH) 2 ·5 H 2 O), as presented in Equation 5 b. For temperatures above 100 8C, hydromagnesite is transformed into magne-site (MgCO 3 ). In this study, hydromagnesite was produced because the experiments were conducted at 85 8C. For the carbonation experiments, filtrate 2 was added to a 500 mL three-necked glass vessel and heated to 60 8C by using a silicon oil bath. The experimental setup was the same as that previously reported. [15] The time, temperature, and pH values were recorded every 5 min. Before heating, aqueous ammonia (35 wt %) was added to filtrate 2. When the temperature reached 60 8C, NH 4 HCO 3 (as the CO 2 source) was added and the solution was heated to 90 8C. After the solution was stabilised at 90 8C, the solution was kept at that temperature for 30 min. Aliquots (2 mL) were sampled by using a needle syringe at 5, 10, 15, 30, 45, and 60 min. The liquid samples were filtered by using a mini filter unit and acidified with HNO 3 for subsequent ICP-AES analysis to measure the Mg concentration. At the end of the experiment, the solution was cooled and filtered by using 0.7 mm Pall syringe filters and the filtrate is referred to as filtrate 3 ( Figure 1). The solid residue was dried at 105 8C overnight and is referred to as product 3 ( Figure 1). The composition of product 3 was analysed by using XRF, and the mineral phases were identified by using XRD. Experiments were conducted at different mass ratios of Mg/NH 3 /NH 4 HCO 3 , in which Mg is the mass of Mg in filtrate 2, and NH 3 and NH 4 HCO 3 represent the mass of aqueous ammonia and the mass of NH 4 HCO 3 added, respectively. In addition, a preliminary experiment was conducted in which no NH 3 was added. The matrix of the experiments conducted at different mass ratios is listed in Table 1.
The carbon content of product 3 was measured by using a thermogravimetric analysis (TGA) Q500 analyzer. The temperature programme was from 30 to 950 8C at 208C min À1 under a nitrogen atmosphere. The carbonation efficiency from soluble magnesium sulfate to hydromagnesite is defined by Equation 6: in which CO 2 content (wt %) is the weight loss of product 3 during the temperature range from 300 to 500 8C, corresponding to the carbonate decomposition found in TGA studies; [17] m 3 is the mass (g) of product 3 from carbonation experiment; c 2 is the magnesium concentration in filtrate 2 from ICP-AES and V 2 is the volume of Thermal decomposition of (NH 4 ) 2 SO 4 Filtrate 3 was evaporated by using a rotary evaporator at 60 8C for 15 min. The solid collected from the rotary evaporator is referred to as product 4 ( Figure 1). The regeneration of NH 4 HSO 4 and NH 3 was conducted by thermal decomposition of product 4 in an oven at 330 8C, and the reaction is presented in Equation 7 : The thermal decomposition of product 4 was characterised by performing TGA studies by using a TGA Q500 instrument in the temperature range of 30-530 8C under nitrogen atmosphere. The temperature programme was as follows: from 30 to 230 8C at 10 8C min À1 , hold for 10 min at 230 8C, up to 330 8C at 108C min À1 , hold for 10 min at 330 8C, and finally up to 530 8C at 108C min À1 . The choice of these three heating steps was to avoid decomposition of the mixture of products. To ascertain the decomposition from the TGA of product 4, pure (NH 4 ) 2 SO 4 and NH 4 HSO 4 were also characterised by using TGA using the same heating procedure.

Results and Discussion
Preparation of magnesium salts solutions from serpentine using NH 4

HSO 4
The results from the ICP-AES analyses (  [15] for which a chemical reaction with product-layer diffusion control was found to be the rate-limiting step of serpentine dissolution in NH 4 HSO 4 .

www.chemsuschem.org
Finally, to compare our work with that of Pundsack, [11] carbonation experiments were carried out by following his procedure. CO 2 was bubbled into the prepared high-magnesium concentration solution from serpentine and excess aqueous ammonia was added to adjust the pH to a value of 9. Only 35 % carbonation efficiency was obtained. In comparison, the carbonation efficiency from this work can achieve a maximum of 95.9 % (experiment 8) due to the faster reaction rate between NH 4 HCO 3 and Mg.

pH regulation and removal of impurities
It was found that after adding aqueous ammonia (35 wt %) to filtrate 1 solution, some particles precipitated. After filtering and drying overnight at 105 8C, the resulting solid and filtrate were labelled product 2 and filtrate 2 (Figure 1), respectively. Aqueous ammonia (35 wt %) was then added to filtrate 2 until the pH value reached 8.5. Table 3 presents the XRF results of the products and the mass balance for Mg, Si, and Fe will be discussed in the section below on mass balance (Figure 9). Taking experiment 7 as an example, it can be seen that product 2 consists of 19.3 % Fe, 8.2% Si, and 2.8 % Mg. The XRD pattern of product 2 for experiment 7 (Figure 3 Tables 1 and 2 and Figure 3 are consistent with this observation, indicating that a high ironcontent precipitate was produced. Some magnesium also pre- Table 3. XRF analyses of solids produced in the experiments, and the CO 2 content from TGA. The mass balance for Mg, Si and Fe for the three products in relation to parent serpentine is also presented as the mass ratio.  www.chemsuschem.org cipitated during this procedure, causing filtrate 2 to contain 5 % less dissolved magnesium than filtrate 1. All experiments presented similar XRF, ICP-AES, and XRD results for product 2 and filtrate 2. Moreover, the iron content of product 2 was measured by using XRF to be between 16.3 and 27.5 wt % (Table 3).

Precipitation studies
Ten precipitation experiments were carried out at different mass ratios of Mg/NH 3 /NH 4 HCO 3 , as shown in Table 1. The observations and findings from these ten experiments were similar in terms of carbonation and morphology of the products. Taking product 3 of experiment 7 as an example, Figure 4 shows the presence of magnesium carbonate. This corresponds to the decrease in magnesium concentration in solution for ICP-AES results presented in Table 2. Figure 4 shows the magnesium concentration variation with time and temperature for experiment 7. The starting time was recorded as heating started. It can be seen that the pH of filtrate 2 decreased from 8.5 to 7.3 as the temperature increased during the first 20 min. As the temperature reached 60 8C, NH 4 HCO 3 was added, and the pH increased slightly to 7.6. No precipitate was formed before adding NH 4 HCO 3 . The concentration of magnesium started to drop as the temperature went up to 70 8C at 25 min. In the following 5 min, half of the Mg ions precipitated at a very high rate of 33.3 mmol min À1 . As the temperature stabilised at 85 8C after 40 min, the pH became stable, and Mg precipitated at a constant rate of 7.9 mmol min À1 . 25 min after the addition of NH 4 HCO 3 , the concentration of Mg in solution became constant and finally fell below 1000 mg L À1 .
For product 3 of experiment 7, the XRD pattern of product 3 ( Figure 5) showed that the Mg precipitated as hydromagnesite, Mg 5 (CO 3 ) 4 (OH) 2 ·4 H 2 O. Combining the results from XRF of product 3 (Table 3) and the ICP-AES results from filtrate 3 ( Table 2), it can be concluded that product 3 is 80 % pure hydromagnesite with only 0.79 wt % of Fe and 0.29 wt % Si.
The carbon content of product 3 can be calculated from the TGA profiles (Figure 6 a), as described in the Experimental Section. All samples presented one carbonate phase, according to the XRD studies. Therefore, the mass of the identified carbonate phase was estimated based on the corresponding weight loss from the TGA studies. As an example, Figure 6 a shows two peaks: the first peak below 250 8C is about 12 wt % and corresponds to the release of crystal water, [17] the second peak, located between 250 and 500 8C, accounts for 37 wt % and is due to the decomposition of hydromagnesite. [16] Finally, based on the CO 2 content (Table 3) and the Mg concentration in fil-     Ammonia captures CO 2 to regenerate NH 4 HCO 3 [Eq. 9]; this reaction is already used in CO 2 capture technology. [19,20] Ammonia can convert NH 4 HCO 3 into (NH 4 ) 2 CO 3 [Eq. 10], which can directly produce MgCO 3 [Eq. 11]. Ammonia can also react with MgSO 4 to form insoluble Mg(OH) 2 if the pH value is above 10, as shown in Equation 12. [21] Once the CO 2 is released from the decomposition of Mg(HCO 3 ) 2 [Eq. 8 b], Mg(OH) 2 can react with CO 2 to form Mg(HCO 3 ) 2 [Eq. 13]. Moreover, Mg(OH) 2 can also react with Mg(HCO 3 ) 2 directly to precipitate MgCO 3 [Eq. 14]. Therefore, the carbonation efficiency was improved by the addition of aqueous ammonia to the high-magnesium concentration solution. In experiments 1-10, in which aqueous ammonia was added, the carbonation efficiency could reach up to 95.9 % (Table 1, experiment 8).

Prevention of precipitation of magnesium ammonium carbonate
The precipitation of magnesium ammonium carbonate (MgCO 3 ·(NH 4 ) 2 CO 3 ·4 H 2 O) can reduce carbonation efficiency because MgCO 3 ·(NH 4 ) 2 CO 3 ·4 H 2 O is generated from the reaction of NH 3 and NH 4 HCO 3 with Mg ions at temperatures below 60 8C [Eq. 15]. [22] It can be seen from Figure 7  According to Equation 16, NH 3 is produced, which would decrease the carbonation efficiency due to a shortage of NH 3 . Therefore, the precipitation of MgCO 3 ·(NH 4 ) 2 CO 3 ·4H 2 O should be prevented to maintain high carbonation efficiency. Taking experiment 4 as an example, the precipitation of MgCO 3 ·-(NH 4 ) 2 CO 3 ·4 H 2 O is indicated on the top left corner of Figure 7. If the temperature increased above 60 8C, the Mg concentration increased, indicating the decomposition of MgCO 3 ·-(NH 4 ) 2 CO 3 ·4 H 2 O. The subsequent decrease of magnesium ions after 30 min indicates the precipitation of hydromagnesite. The carbonation efficiency of experiment 4 is as low as 53.4 % due to the shortage of NH 3 gas, which escaped from the reaction system during the thermal decomposition of MgCO 3 ·-(NH 4 ) 2 CO 3 ·4 H 2 O. Comparing experiments 4 and 9 by using the same mass ratio of Mg/NH 4 HCO 3 /NH 3 and same experimental conditions, the carbonation efficiency decreased from 91.5 to 53.4 % if there was precipitation of MgCO 3 ·(NH 4 ) 2 CO 3 ·4 H 2 O (Table 1). Therefore, to prevent low carbonation efficiency caused by precipitation of magnesium ammonium carbonate, Thermal decomposition of (NH 4 ) 2 SO 4 Product 4 is obtained from the carbonation step by evaporating filtrate 3 (Figure 1). Product 4 was used to generate NH 3 and NH 4 HSO 4 by thermal decomposition in an oven at 330 8C for 20 min. The released gas (NH 3 ) was collected by using water to produce aqueous ammonia. The solid residue obtained after heating was NH 4 HSO 4 . These results were verified by conducting TGA studies, as described herein. Studies of the thermal conversion of ammonium sulfate into ammonium bisulfate can be found in several patents. [23][24][25] As an example in this work, the thermal decomposition of product 4 from experiment 7, as studied by TGA, is shown in Figure 6 b . The TGA profile shows two peaks, where the first weight loss below 330 8C is about 21.7 wt %, corresponding to the release of NH 3 and the formation of NH 4 HSO 4 . [23][24][25] The second weight loss between 350 and 500 8C is 75.8 wt % and is due to further decomposition of NH 4 HSO 4 . [23][24][25] In total, the weight loss of product 4 is 97.5 wt % and the residual 2.5 wt % is due to the presence of MgSO 4 that did not react during carbonation. The TGA profile of pure (NH 4 ) 2 SO 4 (purchased from Fisher Scientific) is presented in Figure 6 c, where two peaks appear at the same temperature range as those for the TGA profile of product 4 (Figure 6 b). The TGA curve of NH 4 HSO 4 is presented in Figure 6 d and shows only one peak between 330 8C and 500 8C due to decomposition into NH 3 , H 2 O and SO 3 . The NH 4 HSO 4 and NH 3 regeneration efficiency from (NH 4 ) 2 SO 4 has been reported to be nearly 97 %. [23][24][25] Herein, the regeneration efficiency of NH 4 HSO 4 and NH 3 from product 4 is 95 %. These TGA results indicate that the reaction of thermal decomposition of (NH 4 ) 2 SO 4 should not be conducted above 330 8C to avoid further decomposition, because NH 4 HSO 4 can decompose into NH 3 , SO 3 , and H 2 O above 330 8C.
The effect of the mass ratio of Mg/NH 4

HCO 3 /NH 3 on carbonation
The mass ratio of Mg/NH 4 HCO 3 /NH 3 is the key factor controlling carbonation efficiency, as discussed herein. The stoichiometric mass ratio of Mg/NH 4 HCO 3 is 1:2, but experiment 5 shows that if a stoichiometric ratio of 1:2 is used, the carbonation efficiency is only 41.5 % ( Moreover, adding aqueous ammonia can increase the carbonation efficiency, as discussed in the section above on precipitation studies. In comparison to the preliminary experiment, experiments 1 and 2 show that carbonation efficiencies increase from 25.5 % to 53 (experiment 2) and then 71.6 % (experiment 1) if the mass ratio of Mg/NH 4 HCO 3 /NH 3 increases from 1:3:0 to 1:3:0.5 and then 1:3:1, respectively. This trend was also found in experiments 6, 8, and 9; however, if the ratio increases to 1:4:3, the carbonation efficiency does not increase any further.
Herein, the optimum mass ratio of Mg/NH 4 HCO 3 /NH 3 was determined. A 3D graph (Figure 8) is used to show the relationship of the four variables, including mass of Mg, mass of NH 4 HCO 3 , mass of NH 3 , and carbonation efficiency. Figure 8 clearly shows that a low summit of 71.6 % carbonation efficiency appears if the mass ratio of Mg/NH 4 HCO 3 /NH 3 is 1:3:1 and a high summit of 95.9 % carbonation efficiency appears if the mass ratio of Mg/NH 4 HCO 3 /NH 3 is 1:4:2. Continuously increasing both NH 4 HCO 3 and NH 3 does not result in a further significant rise of the carbonation efficiency. However, an optimum amount of NH 4 HCO 3 and NH 3 are needed to achieve the highest carbonation efficiency due to the loss of CO 2 and NH 3 in an open system.
The process studied herein presents higher carbonation efficiency than that reported previously. [8,21] For example, in a work by Gerdemann et al., [8] 64 % carbonation efficiency was achieved in direct carbonation of heat-treated serpentine at 155 8C and 115 bar CO 2 in 0.64 m NaHCO 3 and 1 m NaCl solution. In a work by Teir et al., [21] the conversion of Mg ions into hydromagnesite was 94 % using HNO 3 and 79 % using HCl at pH 9 with the addition of NaOH (1.1 g NaOH/g precipitate).
Herein, the highest carbonation efficiency is 95.9 % at 85 8C and ambient pressure within 30 min by using NH 4 HCO 3 and NH 3 . www.chemsuschem.org

Mass balance
To examine the distribution of magnesium released from serpentine in the solids formed in the process (products 1, 2, and 3) and filtrate 3, a mass balance was constructed based on the XRF and ICP-AES and the results are presented in Figure 9. It can be seen that most of the magnesium from the parent serpentine ends up in the precipitated hydromagnesite (product 3). The use of additives at the optimised ratio to improve carbonation conversion results in less magnesium remaining in the final solution after carbonation (filtrate 3, experiments 6-10). Longer dissolution times may leach more magnesium from the serpentine and therefore reduce the amount present in product 1. [7] In addition, the presence of magnesium in product 2 can be minimised by hot-water washing. [18] The mass balance for Si and Fe is presented in Table 3, as the concentration of these two elements in filtrate 3 is very small. It can be seen that most of the Si remains in the residue after dissolution (product 1). In contrast, most of the Fe ends up in both the residue after dissolution (product 1) and the precipitate after pH swing (product 2), depending on the dissolution efficiency. The concentration of Si and Fe in filtrate 3 is negligible ( Table 2). The mass balance for the three elements studied (Mg, Si, and Fe) is very good and between 99-100 % of the mass of the three elements is accounted for.
Considering that the dissolution efficiency can reach 90 % at 100 8C after 2 h and that the carbonation efficiency is 95.9 % if the molar ratio of Mg/NH 4 HCO 3 /NH 3 is 1:4:2, the net conver-sion of serpentine to hydromagnesite is calculated to be 86.3 %. Taking this into account, about 2.63 tonnes of serpentine, 8.48 tonnes of NH 4 HSO 4 , 7.48 tonnes of NH 4 HCO 3 , and 0.8 tonnes of NH 3 are required to sequester 1 tonne CO 2 , and 2.95 tonnes of hydromagnesite is produced. If 95 % regeneration efficiency of NH 4 HSO 4 and NH 3 is considered, 0.12 tonnes of NH 4 HSO 4 and 0.04 tonnes of NH 3 are consumed to sequester 1 tonne CO 2 . All of the chemicals used in this process can be obtained from (NH 4 ) 2 SO 4 . Considering that the current price for (NH 4 ) 2 SO 4 is 90 US$ tonne À1 , [26] the cost of chemicals in this process is estimated to be 18 US$ tonne À1 CO 2 . However, in a work by Teir et al., the cost of chemicals is 1300 US$ tonne À1 CO 2 using HCl and 1600 US$ tonne À1 CO 2 using HNO 3 . [21] Moreover, the cost could be brought down further by using high solid/liquid ratios and this will be the focus of future work.

Conclusions
We have studied the precipitation of hydromagnesite from prepared high-magnesium concentration solutions by using NH 3 and NH 4 HCO 3 . The regeneration of NH 3 and NH 4 HSO 4 was also investigated. Pure hydromagnesite can be produced from serpentine by using regenerated ammonium salts with a net conversion of 86.3 %. Amorphous silica can be obtained from the dissolution step. Byproducts with a maximum 27.5 wt % iron content were obtained from the pH regulation and removal of impurities step. The additives used, NH 4 HSO 4 and NH 3 , can be regenerated by thermal decomposition of (NH 4 ) 2 SO 4 at 330 8C. The addition of aqueous ammonia before carbonation could significantly improve the carbonation efficiency. It must be pointed out that NH 4 HCO 3 should be added to the solution after 60 8C to prevent the production of magnesium ammonium carbonate. The mass ratio of Mg/NH 4 HCO 3 / NH 3 was the key factor to control the carbonation efficiency, and it was found that if the mass ratio of Mg/NH 4 HCO 3 /NH 3 was 1:4:2, a carbonation efficiency of 95.9 % was achieved. From the TGA studies, the regeneration efficiency of NH 4 HSO 4 in this process was found to be 95 %. According to mass balance, about 2.63 tonnes of serpentine, 0.12 tonnes of NH 4 HSO 4 , 7.48 tonnes of NH 4 HCO 3 , and 0.04 tonnes of NH 3 are required to sequester 1 tonne CO 2 , and 2.95 tonnes of hydromagnesite is produced.