Mechanochemical processing of serpentine with ammonium salts under ambient conditions for CO₂ mineralization

Abstract

This paper assesses the suitability of mechanochemistry as a convenient low-energy processing option in CO₂ mineralization. Whereas some success has been reported in milling alkaline earth-containing minerals under gaseous CO₂, this work focuses instead on a purely solid-state approach towards two key objectives: (a) Mg extraction from serpentine using ammonium bisulfate; and (b) direct or indirect CO₂ sequestration using ammonium bicarbonate in a natural extension of its role as “CO₂ carrier” in the chilled ammonia scrubbing process. In Mg extraction work, dry milling of serpentine with ammonium bisulfate gave respectable yields (>60% Mg) as boussingaultite [(NH₄)₂Mg(SO₄)₂·6H₂O] in 2 to 4 h. In CO₂ sequestration, dry milling anhydrous magnesium sulfate with ammonium bicarbonate yielded only mixed sulfate products. Carbonation of the heptahydrate, epsomite, was found to proceed via ammonium magnesium carbonate hydrate [(NH₄)₂Mg(CO₃)₂·4H₂O], which dissolves incongruently to yield nesquehonite [MgCO₃·3H₂O]. The modest conversion (∼30%) is probably due to equipartition of Mg into the double sulfate co-product. A similar route is followed in magnesia and brucite, in which the existence of an amorphous native carbonate precursor to nesquehonite in the same molar ratio (Mg : CO₂ = 1) was inferred from inconsistency in the XRD intensities. This was largely responsible for the high carbonation yields in the unwashed products, ∼70% and ∼85% in MgO and Mg(OH)₂, respectively, as confirmed by TG-FTIR. The same intermediate is probably formed in serpentine, but it is apparently soluble in the aqueous mineral environment. When the unwashed product is subjected to mild thermal consolidation, stable hydromagnesite [Mg₅(CO₃)₄(OH)₂·4H₂O] is formed in ∼20% yield after milling for 16 h. Possible identities for the amorphous precursor are briefly considered.

---

1. Introduction

Ball milling is an established method in nanotechnology, such as in the production of nanopowdered metals and oxides by extreme mechanical attrition. Further, it can be used to activate minerals for easier down-stream processing due to the introduction of energetic structural defects into the bulk and creation of more reactive surfaces. Mechanochemistry uses the same shearing and welding forces to induce desirable chemical change in mixtures of solid reactants. It first emerged in the 1990s in metallurgical processing, notably in the “mechanical alloying” of immiscible metals and related endothermic processes aiming for metastable products with novel material properties.^1–3 It is more recently being explored as a “green” or solvent-free alternative for syntheses in the fine chemicals field.⁴ An obvious attraction is that it is a simple form of solid-state processing that works close to ambient temperature. On the other hand, its amenability to scale-up and realistic evaluation of process efficiency remain serious challenges.

In the field of CO₂ mineralization, the mechanochemical approach has been relatively little explored. Most activities to date have focussed on mechanical activation of serpentines or olivines,^5,6 generally as a preliminary to one-pot carbonation by aqueous dissolution under high-pressure CO₂.^7,8 As regards the techno-economics of solid-state processing, a recent LCA analysis of a future mineralization process, based on serpentinite rocks shipped to Singapore for carbonation and disposal (land reclamation), the energy cost associated with mining and crushing is substantially less than for shipping and only 1–2% of the total.⁹ Nevertheless, extended milling is more energy-intensive than crushing, and a cautionary note has been sounded that milling should be restricted to the time required to reach a particle size threshold of <75 μm, requiring roughly 10 kWh ton⁻¹, or less than 15% of total plant energy demand.⁷ In other words, the scalability of “process cost vs. milling time” is complex and soon reaches a minimum. However, this analysis strictly applies only to “wet” carbonation. In contrast, mechanochemistry can drive uphill chemical reactions, thereby offering prospects of considerable cost saving, e.g., by substituting for energy-intensive thermal activation.^7,10

Interesting results have been reported in mechanochemically-activated (gas/solid) carbonation of various Ca/Mg silicate minerals. A “pestle & mortar” type grinding device was used under 1 bar CO₂, but the carbonate yield was just 10% after 36 h.^11–13

In our companion paper in this journal,¹⁴ we elucidate the chemistry underlying a staged process involving ammonium sulphate as a cheap and potentially recyclable flux. Due to the high temperatures necessary for good kinetics of Mg extraction (T > 400 °C), the atom efficiency is currently restricted by irreversible loss of the flux (probably as sulphamic acid) by sublimation and/or decomposition. Thus, low temperature methods also need to be explored, such as aqueous solutions of ammonium bisulphate as originally claimed by Pundsack.¹⁵ Recent progress has been reported in this “wet” extraction process but operation at 100 °C during 3 h was necessary for full Mg recovery.¹⁶ As a novel and potentially cheap alternative, one aspect of the work reported here was to explore mechanochemical activation of Mg extraction by ammonium bisulfate in the solid-state at ambient temperature.

Our primary objective was to evaluate the role of solid ammonium bicarbonate as a “carbonator”; directly by milling with a Finnish serpentinite, or indirectly by milling with intermediates in the staged process,¹⁴viz., the sulfate, hydroxide, and oxide of magnesium. The background to this study is the recent emergence of the “chilled ammonia” process as a more efficient alternative in CO₂ trapping (from flue gas) to conventional MEA scrubbing as currently practised. In the “chilled ammonia” process,^17–21 CO₂ is sequestered in two stages involving ammonium carbonate and ammonium bicarbonate:

2NH₃ + CO₂ + H₂O → (NH₄)₂CO₃ (1)

(NH₄)₂CO₃ + CO₂ + H₂O ↔ 2NH₄HCO₃ (2)

The equilibrium of eqn 2 is easily shifted to the left under process control such that CO₂ release (for storage or permanent disposal) is from the bicarbonate, while the carbonate is recycled. Lackner's “artificial tree” is based on a similar carbonate–bicarbonate half-cycle.²² One key advantage of this technology over amine scrubbing is that the release step requires less energy. It is increasingly recognised that, where logistics are favourable, any process linking efficient CO₂ capture to safe and permanent disposal, e.g., via mineralization, may have tangible operating advantages.²³ In this context, ammonium bicarbonate acts effectively as a “CO₂ carrier.” While precipitation of the carbonate directly from Mg²⁺ ion is the next logical stage in a fully liquid-phase process,^24,25 extraction and carbonation are chemically incompatible, adding to process complexity (pH-swing, etc.).

2. Experimental

2.1 Materials

The serpentinite was received in lump form from a nickel mine in Hitura, Finland, and milled briefly to <45 μm (^#325 mesh). In XRD, it showed a diffraction pattern fitting best to the hexagonal Lizardite polymorph [(Mg,Al)₃(Si,Fe)₂O₅(OH)₄− ^#00-050-1625] with weak additional lines of magnetite [Fe₃O₄− ^#03-065-3107]. XRF analysis showed the serpentinite to be of high purity, having a Mg content (as MgO) of 43 wt.% [theor. = 43.6%]. The BET surface area [N₂ physisorption, 77 K] was ∼30 m² g⁻¹, indicating the presence of substantial porosity. All other chemicals were reagent grade (>99%, Aldrich) and used as supplied. The magnesia (MgO, ^#325 mesh) and brucite [Mg(OH)₂] had BET surface areas of 204 and 22 m² g⁻¹, respectively. Mg extraction studies were made with ammonium bisulfate. Carbonation studies were made with ammonium bicarbonate, NH₄HCO₃, also known as teschemacherite (see ESI Table S1†). Ammonium carbonate, (NH₄)₂CO₃, was not investigated as it is known to be unstable, spontaneously transforming to the bicarbonate upon exposure to ambient air.

2.2 Methodology

2.2.1 Milling

After preliminary mixing (hand-grinding) of appropriate weights of components, samples (5–10 g) were divided into two batches and loaded equally into opposing agate-lined steel pots (125 ml), together with 10–12 agate balls (1 cm diameter) for centrifugal balance. These were then clamped into the carousel of a Retsch PM 400/2 planetary mill and spun at 250 rpm for various times with periodic sampling after gas-venting. Unless otherwise stated, milling was done in ambient air at 1 bar.

2.3 Analytical techniques

2.3.1 TG-FTIR. Thermogravimetry (TG) coupled to Fourier-transform infrared (FTIR) spectroscopy is a powerful tool in the characterization of solid materials. A controlled heating ramp is applied to a sample suspended on a sensitive microbalance under carrier gas flow and the weight change recorded. Any infrared-active evolved gas products are continuously swept into the IR spectrometer, identified and monitored by transmission in a suitable optical cell. In this laboratory, a Setaram Setsys 12 TG unit is linked to a Digilab Excalibur series FTIR spectrometer fitted with a KBr beamsplitter and a mid-range (liquid N₂-cooled) MCT detector. The interface unit comprises a transfer line and 10 cm pathlength gas cell, both thermostatted at 220 °C. The experimental methodology has been described elsewhere.¹⁴

2.3.2 XRD. X-ray patterns were obtained from lightly pressed powder samples on a Bruker D8 powder diffractometer with a Cu–Kα source (λ = 1.5418 Å) over the range 2θ = 10–80° at a step size of 0.008°.

2.3.3 ICP-OES. Aqueous extracts from milling experiments, obtained from 50 ml suspensions (∼200 mg sample) suitably filtered and diluted (×100) in deionized water, were analyzed for elemental Mg and S (1–10 ppm calibrated range) in a Varian Vista-MPX inductively-coupled plasma optical smission spectrometer.

3. Results and discussion

3.1 Mechanochemical promotion of Mg ion extraction

Initial milling experiments were made on lizardite–ammonium bisulfate mixtures conforming to the stoichiometry:

Mg₃Si₂O₅(OH)₄ + 3NH₄HSO₄ → 3MgSO₄ + 2SiO₂ + 3NH₃ + 5H₂O (3)

Dry and wet studies were made simultaneously by adding an equal weight of water to half the amount of dry mixture to give counterbalance (10 g in each pot), and these were sampled during milling after 30, 120 and 225 min. However, instead of obtaining the simple target sulfate, dry milling quickly developed the XRD pattern of boussingaultite [(NH₄)₂Mg(SO₄)₂·6H₂O]. Otherwise, only a weak residual pattern of the serpentinite (lizardite) was visible (see ESI Fig. S2†). The ICP-OES analyses of Mg (ion) recoveries are shown in Fig. 1. These were quite good, ranging from over 50% after 30 min, increasing steadily to 70% after 3 h. It was surprising that wet milling barely improved Mg yield since the empirical formula unit of boussingaultite includes six molecules of water. Furthermore, a balanced equation cannot be written for this product at the 1 : 3 proportions used, implying a case of limiting reactant. Only a doubling of ammonium bisulfate would assure full yield of Mg ion according to:

Mg₃Si₂O₅(OH)₄ + 6NH₄HSO₄ + 13H₂O → 3(NH₄)₂Mg(SO₄)₂·6H₂O + 2 SiO₂ (4)

Fig. 1 ICP-OES analyses of Mg recovery from milled serpentine–ammonium bisulfate mixtures. [■—1 : 3 dry mill; ■—1 : 3 wet mill; ■—1 : 6 dry mill; •—1 : 3 dry manual, •—1 : 6 dry manual].

Unfortunately, milling of serpentine in the enriched bisulfate did not improve Mg ion recovery. The yield remained close to that obtained from the standard (1 : 3) mixture (eqn 3), reaching 65% in 2 h. This is attributed tentatively to a moderation of the interfacial frictional forces between component particles in the mixture. Ammonium bisulfate is not only a softer material than serpentine but it is also hygroscopic. This textural issue could be explored in future work by adding a hard but inert and insoluble promoter such as SiC. In contrast, control experiments where only short (5–10 min) manual grinding in a pestle & mortar was applied to dry mixtures showed that superficial extraction of Mg is remarkably easy. As shown in Fig. 1, recoveries of 30% and 40% were obtained from standard (1 : 3) and bisulfate-enriched (1 : 6) mixtures, respectively, by such a gentle treatment. This result encourages the view that a scaled-up process involving simple blending or kneading in a moving bed may be viable. Further options for processing the double sulfate towards stable magnesium carbonate products are under investigation and summarized in our companion paper.¹⁴

3.2 Solid-state mechanochemical carbonation

3.2.1 Anhydrous & hydrated magnesium sulphate. Previous modelling work²⁵ has shown that the reaction:

MgSO₄ + 2NH₄HCO₃ → MgCO₃ + (NH₄)₂SO₄ +CO₂ + H₂O (5)

is thermodynamically favourable. Although endothermic, conversion to products is driven by the entropy factor. As shown in Fig. 2, milling an anhydrous MgSO₄ : NH₄HCO₃ (1 : 2 mol : mol) mixture produced boussingaultite and ammonium sulfate. No crystalline carbonates were detected by XRD even after 16 h. This was despite the virtual elimination of the bicarbonate structure [reference diffractograms for both ammonium sulfate and ammonium bicarbonate are available in ESI† (Fig. S2)]. Transfer of NH₄⁺ ions to form double sulfates implies the creation of free bicarbonate (HCO₃⁻) ions. Evidently these may have decomposed, releasing CO₂ in the process. The milling pots did release an odourless gas upon cooling and venting. In contrast, milling the hydrated salt epsomite (MgSO₄·7H₂O) and NH₄HCO₃ in the same proportions produced a new pattern alongside that of boussingaultite. After 2 h milling, this was already identifiable as ammonium magnesium carbonate hydrate, (NH₄)₂Mg(CO₃)₂·4H₂O. Extended milling resulted in further development of this phase, as shown in Fig. 3. Reflections assigned to the double carbonate are denoted ‘A’ for ease of visualisation. Epsomite, a hydrate reactant, evidently favours the formation of a hydrate product under the dry milling conditions applied here. The influence of water, in free or crystal form, in directing mechanochemical change has been recognized elsewhere.^4,26 After aqueous washing, a white insoluble residue was obtained and identified as nesquehonite (MgCO₃·3H₂O) by XRD, as shown in Fig. 4. Ammonium magnesium carbonate hydrate evidently dissolves incongruently according to:

(NH₄)₂Mg(CO₃)₂·4H₂O → MgCO₃·3H₂O + 2NH₄⁺ + CO₃²⁻ + H₂O (6)

Fig. 2 XRD of milled product from MgSO₄ : NH₄HCO₃ (1 : 2) [B = boussingaultite (blue), AS = ammonium sulfate (red) ABC = ammonium bicarbonate (green)].

Fig. 3 XRD of milled product from MgSO₄·7H₂O : NH₄HCO₃ (1 : 2) [B = boussingaultite (blue), A = ammonium magnesium carbonate hydrate (red)].

Fig. 4 XRD of residue after aqueous washing of milled product from MgSO₄·7H₂O : NH₄HCO₃ (1 : 2) [N = nesquehonite (MgCO₃·3H₂O (red)].

Thus, the overall chemistry follows the half-cycle of the “chilled ammonia” process. Half the carbon is recycled as ammonium carbonate but the CO₂ released from bicarbonate (eqn 2) is now sequestered as magnesium carbonate. Estimates of conversion by gravimetry and ICP-OES (insoluble Mg by difference) were consistent at ∼35% after 12 h milling, rising from 30% after 8 h, and 22% after 2 h. The upper limit for the carbonation yield is probably 50%. Elemental balance requires co-formation of the double carbonate and double sulfate, between which Mg²⁺ is equi-partitioned:

MgSO₄·7H₂O + 2NH₄HCO₃ → ½[(NH₄)₂Mg(CO₃)₂·4H₂O] + CO₂ + 3H₂O + ½[(NH₄)₂Mg(SO₄)₂·6H₂O] (7)

Although nesquehonite is slightly soluble in water,²⁷ it is known to transform into the more stable hydromagnesite under mild heating,^28,29 with slight loss of CO₂:

5MgCO₃·3H₂O → Mg₅(CO₃)₄(OH)₂·4H₂O + CO₂ + 10H₂O (8)

A TG-FTIR experiment on the nesquehonite product confirmed a small CO₂ evolution feature at ∼150 °C, but the 1st stage weight loss (∼37%) still remained close to that expected (39%) from only water evolution. As seen in Fig. 5, most of the carbonate was thermally stable, the main CO₂ evolution not starting until ∼350 °C. This 2nd stage weight loss, complete by 600 °C, was ∼31%, i.e., close to the value expected for nesquehonite decarbonation (32%). Hydromagnesite (not shown) displayed a similar CO₂ evolution profile above 350 °C. This contrasts with a recent independent in situ XRD study of nesquehonite,³⁰ which reported preliminary dehydration to MgCO₃·2H₂O, followed by partial decarbonation to form magnesium oxycarbonate, MgO·2MgCO₃.³¹ Regardless of the intermediate product(s), the small loss of CO₂ observed by TG-FTIR, although non-ideal, does not detract seriously from an effective sequestration process. On a spectral area (CO₂ marker) basis, the loss was less than one tenth of the total sequestered, and less than half that expected according to eqn 8. Both of these bulk analytical methods are clearly important and complementary. Some XRD evidence for magnesium oxycarbonate has also been seen in gas–solid carbonation studies in our Turku laboratory.³²

Fig. 5 TG-FTIR analysis of MgCO₃·3H₂O residue after aqueous washing of epsomite–ammonium bicarbonate (1 : 2) milled for 12 h [Inset—FTIR evolved gas traces—arbitrary response].

3.2.2 Brucite and magnesia. Brucite and magnesia mixtures with ammonium bicarbonate were made up deliberately at 1 : 1 stoichiometry:

Mg(OH)₂ + NH₄HCO₃ → MgCO₃ + NH₃ + 2H₂O (9)

MgO + NH₄HCO₃ → MgCO₃ + NH₃ + H₂O (10)

where eqn 9 and 10 correspond to theoretical weight losses of 38.6% and 29.4%, respectively. This was done primarily to evaluate the feasibility of obtaining magnesite or nesquehonite directly by milling. These proportions also disfavour formation of a double carbonate intermediate and/or dependence on the half-cycle observed for epsomite (see eqn 7). The atom efficiency of the carbonator should ideally be as high as possible for a viable process. As no Mg ions could be ‘lost’ (as sulfates) from brucite or MgO, higher yields of magnesium carbonates were anticipated.

Substantial weight losses were registered after a 16 h milling, viz., 32% and 22% for the Mg(OH)₂ and MgO mixtures, respectively. This suggested loss of gas products (on venting) in accordance with eqn 9 and 10. XRD showed weak patterns of each magnesian phase, but no evidence for residual ammonium bicarbonate or magnesite. Surprisingly, ammonium magnesium carbonate hydrate was also present, albeit at lower levels as compared to that derived from epsomite. The intensity of the principal peak of the double carbonate at 2θ = 29.0° [d₂₂₀ = 3.08 Å, see Fig. 3] decreased in the order: ex MgSO₄·7H₂O [900] > ex Mg(OH)₂ [540] > ex MgO [100]. By analogy with the sulfates, the greater yield from brucite may be linked to the ‘elements of water’ which are absent in MgO. This was despite wide-ranging surface areas, tending to favour magnesia reactivity (see section 2.1). In contrast, strong patterns of monoclinic nesquehonite were observed after washing, but in reverse order for brucite and MgO. The main peak intensity for nesquehonite at 2θ = 13.7° [d₁₀₁ = 6.48 Å; see Fig. 4] was now: ex MgSO₄·7H₂O [6400] > ex MgO [5200] > ex Mg(OH)₂ [2800]. This suggests that ammonium magnesium carbonate hydrate is not the only source of nesquehonite. There is evidently a precursor state; a native carbonate formed in the dry milling process that is barely soluble and amorphous to X-rays. In the previously cited milling work on various Ca/Mg-containing minerals under gaseous CO₂, amorphous products were also obtained. These authors used FTIR to verify the reaction by monitoring the build-up of bands diagnostic of carbonate(s).^11–13

To better characterize the amorphous precursor, TG-FTIR analyses were made and the results are presented in Fig. 6. This compares weight loss curves for the unwashed products from MgO–ABC (denoting ammonium bicarbonate), Mg(OH)₂–ABC, and nesquehonite as reference. The FTIR evolved gas trace for the MgO mix is representative in showing mainly CO₂ evolution. Its TG scale is also offset to show the similarity in the 2nd stage weight loss (∼30%) around 400 °C in all samples. To gauge the level of CO₂ derived from any double carbonate, the relative IR marker areas for NH₃ and CO₂ were established using pure NH₄HCO₃. This was found to be A_NH3 : A_CO2 = 0.27. The NH₃ : CO₂ area ratio in the FTIR trace is an order of magnitude lower. As the double carbonate [(NH₄)₂Mg(CO₃)₂·4H₂O] and NH₄HCO₃ are equivalent in terms of incipient gas content (NH₃ : CO₂ = 1 : 1), it follows that a large amount of CO₂ was sequestered in the amorphous phase. The low FTIR response for NH₃ was consistent with the weak XRD pattern for the double carbonate. It also confirmed the gravimetric result, indicating that most of the NH₃ had evolved during milling, and escaped upon venting. This augurs well from the viewpoint of NH₃ recycling for subsequent CO₂ trapping in the “chilled ammonia” process.

Fig. 6 TG-FTIR analyses of unwashed products ex MgO : NH₄HCO₃ (1 : 1) and Mg(OH)₂ : NH₄HCO₃ (1 : 1) milled for 16 h. [Inset—FTIR evolved gas marker traces for MgO : NH₄HCO₃].

Reliable estimates of carbonation efficency were obtained by similar TG-FTIR studies on a fixed sample weight (42 ± 1 mg). This ensured that the CO₂ marker areas were directly comparable, needing only slight adjustment for the weight fraction of CO₂ (ex. NH₄HCO₃) in the original mixtures. Taking nesquehonite as representing “100% carbonation” (Mg : CO₂ = 1 : 1), the degree of carbonation of brucite and magnesia was found to be ∼85% and ∼70%, respectively. In contrast, if conversions are inferred based on relative intensities of the nesquehonite pattern after washing, these would be seriously underestimated, apparently lying below that of epsomite (<35%).

3.2.3 Serpentine. Direct gas–solid carbonation of serpentine and related minerals is viable thermodynamically (see eqn 4 in ref. 14), so it was of interest to explore a mechanochemical route to the same end-product, magnesite, according to:

Mg₃Si₂O₅(OH)₄ + 3NH₄HCO₃ → 3MgCO₃ + 2SiO₂ + 3NH₃ + 5H₂O (11)

The thermodynamics of eqn 11 may be favoured by the entropy increase as compared to the gas–solid process, which is mildly exothermic. It is also the analogous reaction to sulfation by NH₄HSO₄ (eqn 3), which has been found to proceed under milling (see section 3.1).

After 16 h milling, XRD analysis on the unwashed sample showed a very weak pattern of ammonium magnesium carbonate hydrate, along with strong reflections of serpentine (lizardite see ESI Fig. S2†). The intensity of the d₂₂₀ peak in the former (∼200 counts) lay between those obtained from Mg(OH)₂ and MgO. However, after washing no pattern of nesquehonite could be discerned and a large weight loss was recorded (∼50%). TG-FTIR analysis on the unwashed sample is shown in Fig. 7. In contrast to brucite and magnesia, the FTIR trace for the mineral mixture is dominated by a structured evolution of NH₃, H₂O, and CO₂ below 150 °C, due to unimolecular decomposition of the bicarbonate:

NH₄HCO₃ → NH₃ + H₂O + CO₂ (12)

Fig. 7 TG-FTIR analysis of unwashed product ex serpentine : ammonium bicarbonate (1 : 3) milled for 16 h. [Inset—FTIR evolved gas marker traces].

which is known to be thermally unstable.³³ If the 40% weight loss (see Fig. 7) is associated solely with volatilization of unreacted ammonium bicarbonate, this implies that about 80% of the mineral remained intact. A weak CO₂ evolution peak was also recorded at ∼400 °C, with an associated weight loss of 5.6%. The CO₂ marker areas for low- and high-temperature evolution were close to 4 : 1, which was also the ratio of the weight losses when corrected for NH₃ and H₂O evolution from NH₄HCO₃. The total of these (CO₂-linked) weight losses was 27%, close to the weight fraction of CO₂ in the original mixture (25.7%). Thus, based on this good material balance for carbon, the fraction of CO₂ sequestered was estimated as ∼20%. This is also consistent with the level of unreacted serpentine (vide ultra). When the milled product was subjected to prior washing, TG-FTIR showed no CO₂ evolution whatsoever. Evidently, any native carbonate was thermally consolidated adventitiously during the TG-FTIR analysis. In addition, ICP-OES of the filtrate registered that nearly 16% of the Mg ion was extracted in soluble form, whereas a control extraction of serpentine milled alone yielded just 1–2%. Hence, the mineral is indeed activated by milling in the presence of ammonium bicarbonate. The consistency in levels of soluble Mg and CO₂ in the absence of prior heat treatment (Mg : CO₂ ≈1) can hardly be coincidental and suggests that the same native carbonate is formed as in the magnesium (hydr)oxides. However, it is apparently soluble in the aqueous mineral environment.

As possible candidates for the amorphous precursor, magnesium bicarbonate [Mg(HCO₃)₂] is certainly implicated in nesqehonite formation. However, it is claimed to exist only in solution,^34,35 and, furthermore, its composition (Mg : CO₂ = 2) does not fit. The Mg(H₂O)_nHCO₃⁺ ion has been modelled,^36–38 while surface bicarbonates on MgO are well-known from solid-state infrared spectroscopy.³⁹ It should be noted that the univalent Mg(OH₂)⁺ species is implicated in Mg dissolution from its oxide, hydroxide, and hydroxycarbonates.^40,41 Thus, a reasonable conjecture is that the neutral species Mg(H₂O)_n(OH)HCO₃, probably of sparing solubility, may also exist in a dry and basic environment with a source of hydroxyl ion, such as the brucite–NH₄HCO₃ system. Nesquehonite may then be derived by washing or thermal activation, both facilitating intramolecular proton transfer from the carbonate oxygen to the hydroxyl group. In a weak Brønsted acid environment, as engendered by the silica component in serpentine, such an intermediate would likely be prone to dissociation and dissolution. Coincidentally, the empirical formula for nesquehonite can sometimes be found written as Mg(OH)HCO₃·2H₂O, although there seems to be no crystallographic basis for its existing as anything other than the trihydrate.⁴² Nevertheless, Frost and Palmer⁴³ have recently used FTIR and Raman spectroscopy to show that there are indeed infrared features characteristic of both HCO₃⁻ and OH⁻ groups, and that these are retained while the water of crystallization is preferentially driven off under mild heating. This also implies loss of the monoclinic structure and prompts a follow-up in situ XRD study to help resolve the issue. In other words, the amorphous precursor may be a lower hydrate of this putative hydroxybicarbonate, though some polymerization must also be invoked to satisfy the normal 6-coordinate environment of the magnesium ion. The conditions prevailing during dry milling may engender its formation due to the frictional heat created.

Despite the foregoing complexities, mild heat treatment per se should not introduce a significant energy penalty into any sequestration process. Further work is necessary to establish the minimum temperature required to consolidate the native carbonate against aqueous leaching in the mineral environment. Although this initial survey of prospects for solid–solid mechanochemistry in CO₂ mineralization has yielded encouraging results, the issues of scale-up in ball milling and the development of mechanical alternatives more amenable to high throughput or flow operation⁴⁴ remain a major technical challenge.

Conclusions

By virtue of their rich chemistry with magnesium, ammonium salts will likely become ubiquitous in the field of CO₂ mineralization. The sulfate and bicarbonate are particularly suited to silicate mineral activation (Mg extraction) and solid-state carbonation, respectively. Mechanochemical extraction of Mg²⁺ ion from serpentine with ammonium bisulfate is efficient, exceeding 30% even after brief manual grinding. Apart from anhydrous magnesium sulfate, dry milling various Mg-based solids with ammonium bicarbonate achieved CO₂ sequestration at ambient temperature. The ultimate formation of stable carbonate(s) progressed via ammonium magnesium carbonate hydrate, which dissolved incongruently to yield nesquehonite. The restricted yield from epsomite (30%) was probably due to equipartioning of Mg ions between the double carbonate and double sulfate co-products. Carbonate levels from unwashed magnesia and brucite mixtures were much higher, viz., 70% and 85%, respectively, and linked to an amorphous native carbonate precursor, as inferred from anomalous XRD intensities. The same intermediate is believed responsible for ∼20% sequestration in serpentine after 16 h milling. However, when the milled product was washed prior to TG-FTIR analysis, no CO₂ evolution was detected, showing that the native carbonate is soluble in an aqueous mineral environment.

Acknowledgements

JH and RZ are grateful for a joint A*Star/Tekes operating grant [FI-10-014 - NEACAP project 2010–2013] under which this work was conducted. JF is grateful to the Harry Elvings legacy to enable a 4 month research secondment at ICES Singapore in 2011.

References

1. See, e.g., C. Suryanarayana, Prog. Mater. Sci., 2001, 46, 1.
2. (a) C. Suryanarayana, Mechanical aloying and milling, Marcel Dekker, Norwich, NY, USA, 2004; (b) Nanostructured materials: processing, properties & applications,ed. C. C. Koch, William Andrew Publishing, Norwich, NY, USA, 2nd edn, 2007.
3. (a) See, e.g., P. Baláž, Extractive metallurgy of activated minerals, Elsevier, Amsterdam, The Netherlands, 2000; (b) K. Tkáčová, Mechanical activation of minerals, Elsevier, Amsterdam, The Netherlands, 1989; (c) L. Takacs, Hyperfine Interact., 1998, 11, 245.
4. S. L. James, C. J. Adams, C. Bolm, D. Braga, P. Collier, T. Friščić, F. Grepioni, K. D. M. Harris, G. Hyett and W. Jones, et al. , Chem. Soc. Rev., 2012, 41, 413.
5. C. A. Summers, D. C. Dahlin, G. E. Rush, W. K. O'Connor and S. J. Gerdemann, Minerals & Metallurgical Processing, 2005, 22, 140.
6. P. Baláž, E. Turianacová, M. Fabián, R. A. Kleiv, J. Briančin and A. Obut, Int. J. Miner. Process., 2008, 88, 1.
7. S. J. Gerdemann, W. K. O'Connor, D. C. Dahlin, L. R. Penner and H. Rush, Environ. Sci. Technol., 2007, 41, 2587.
8. K. L. Sandvik, R. A. Kleiv and T. A. Haug, Adv. Powder Technol., 2011, 22, 416.
9. H. H. Khoo, P. N. Sharratt, J. Bu, Y. Yeo, A. Borgna, J. G. Highfield, T. Björklöf and R. Zevenhoven, Ind. Eng. Chem. Res., 2011, 50, 11350.
10. See, e.g., M. J. McKelvy, A. V. G. Chizmeshya, J. Diefenbacher, H. Bearat and G. Wolf, Environ. Sci. Technol., 2004, 38, 6897.
11. E. V. Kalinkina, A. M. Kalinkin, W. Forsling and V. N. Makarov, Int. J. Miner. Process., 2001, 61, 289.
12. A. M. Kalinkin, E. V. Kalinkina, A. A. Politov, V. N. Makarov and V. V. Boldyrev, J. Mater. Sci., 2004, 39, 5393.
13. A. M. Kalinkin, E. V. Kalinkina, O. A. Zalkind and T. I. Makarova, Colloid J., 2008, 70, 33.
14. J. Highfield, H. Q. Lim, J. Fagerlund and R. Zevenhoven, RSC Adv., 2012 10.1039/C2RA01347A.
15. F. L. Pundsack, US Patent, US3338667, 1967, Recovery of silica, iron oxide and magnesium carbonate from treatment of serpentine with ammonium bisulphate..
16. X. Wang and M. M. Maroto-Valer, Fuel, 2011, 90, 1229.
17. E. Gal, World Intellectual Property Patent WO 2006022885, 2006, Ultra cleaning combustion gas including removal of CO₂.
18. J. T. Yeh, K. P. Resnik, K. Rygle and H. W. Pennline, Fuel Process. Technol., 2005, 86, 1533.
19. G. Valenti, D. Bonalumi and E. Macchi, Energy Proc., 2009, 1, 1059.
20. F. Kozak, A. Petig, E. Morris, R. Rhudy and D. Thimsen, Energy Proc., 2009, 1, 1419.
21. V. Darde, K. Thomsen, W. J. M. v. Well and E. H. Stenby, Int. J. Greenhouse Gas Control, 2010, 4, 131.
22. K. S. Lackner, Washing carbon out of the air, Scientific American, June 2010, 66.
23. R. J. Hunwick, Australian patent, AU2008/000232, 2008; AU2010/000214, 2010, System, apparatus and method for carbon dioxide carbonation.
24. X. Wang and M. M. Maroto-Valer, ChemSusChem, 2011, 4, 1291.
25. T. Björklöf. MSc (Eng) Thesis. Åbo Akademi University. Turku, Finland. 2010.
26. Ü. Kersen and M. R. Sundberg, J. Electrochem. Soc., 2003, 150, H129.
27. M. Hänchen, V. Prigiobbe, R. Baciocchi and M. Mazzotti, Chem. Eng. Sci., 2008, 63, 1012.
28. P. J. Davies and B. Bubela, Chem. Geol., 1973, 12, 289.
29. J. Lanas and J. I. Alvarez, Thermochim. Acta, 2004, 421, 123.
30. P. Ballirano, C. De Vito, V. Ferrini and S. Mignardi, J. Hazard. Mater., 2010, 178, 522.
31. G. Hladky, Neues Jb. Miner. Monatsh., 1975, 3, 115.
32. J. Fagerlund, J. Highfield and R. Zevenhoven, ChemSusChem. , submitted.
33. J. E. House Jr., Thermochim. Acta, 1980, 40, 225.
34. Y. Zhao and G. Zhu, Hydrometallurgy, 2007, 89, 217.
35. G. Zhu, H. Zhang and Y. Zhao, Hydrometallurgy, 2008, 92, 141.
36. D. D. Tommaso and N. H. De Leeuw, Phys. Chem. Chem. Phys., 2010, 12, 894.
37. D. D. Tommaso and N. H. De Leeuw, Cryst. Growth Des., 2010, 10, 4292.
38. S. Petrie, Int. J. Mass Spectrom., 2006, 254, 136.
39. G. Busca and V. Lorenzelli, Mater. Chem., 1982, 7, 89.
40. O. S. Pokrovsky and J. Schott, Geochim. Cosmochim. Acta, 2004, 68, 31.
41. O. S. Pokrovsky and J. Schott, Geochim. Cosmochim. Acta, 1999, 63, 881.
42. G. W. Stephan and C. H. MacGillavry, Acta Crystallogr., Sect. B: Struct. Crystallogr. Cryst. Chem., 1972, 28, 1031.
43. R. L. Frost and S. J. Palmer, Spectrochim. Acta, Part A, 2011, 78, 1255.
44. C. Medina, D. Daurio, K. Nagapudi and F. Alvarez-Nunez, J. Pharm. Sci., 2010, 99, 1693.

---

Footnote

† Electronic Supplementary Information (ESI) available: a glossary of empirical formulae, mineral names, XRD reference codes, etc. See DOI: 10.1039/c2ra20575k/

---