Utilizing mixed-mineralogy ferroan magnesite tailings as the source of magnesium oxide in magnesium potassium phosphate cement

Construction and Building Materials 231 (2020) 117098

Contents lists available at ScienceDirect

Construction and Building Materials journal homepage: www.elsevier.com/locate/conbuildmat

Utilizing mixed-mineralogy ferroan magnesite tailings as the source of magnesium oxide in magnesium potassium phosphate cement Arnold Ismailov a,⇑, Niina Merilaita a, Soili Solismaa b, Marjaana Karhu c, Erkki Levänen a a

Materials Science and Environmental Engineering, Faculty of Engineering and Natural Sciences, Tampere University, P.O. Box 589, FI-33101 Tampere, Finland Geological Survey of Finland (GTK), P.O. Box 1237, FI-70211 Kuopio, Finland c VTT Technical Research Centre of Finland Ltd (VTT), P.O. Box 1300, FI-33101 Tampere, Finland b

h i g h l i g h t s

g r a p h i c a l a b s t r a c t


A talc mine tailing (MT) was used as a

source for MgO in MKPC.
The pre-calcined tailing consisted

of ~80% MgCO3, the rest being mostly silicates.
Calcination of MT was carried out at 350 °C–1400 °C.
Calcination at 1200 °C and above reduced the phosphate reactivity excessively.
Recommended calcination range was found by 4-point bending and solubility testing.

a r t i c l e

i n f o

Article history: Received 20 June 2019 Received in revised form 13 September 2019 Accepted 26 September 2019

Keywords: Chemically bonded ceramics Phosphate cement Mixed-mineralogy MgO Surface area Bending strength pH

a b s t r a c t A mixed-mineralogy talc mine tailing (MT) fraction consisting of 80% ferroan magnesite (MgCO3) was studied for utilization as the source of magnesium oxide (MgO) in magnesium potassium phosphate cement (MKPC). The effects of calcination temperature of this low-grade magnesite on the composition, BET surface area and phosphate reactivity of the resulting magnesia powder were studied. The 4-point flexural strength of resulting MKPC was measured for all calcined raw material fractions that produced a solid. Based on the strength measurement results, the optimal range for calcination resided between 700 °C and 1150 °C, which is drastically lower than commonly recommended for finer magnesia sources in MKPCs. Accelerated reactivity assessment showed that phosphate reactivity behavior could not be entirely predicted by BET surface area. The presence of impurity silicates and high iron content in all the constituent minerals was posed as the reason for densification and loss of reactivity at higher calcination temperatures. Ó 2019 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Abbreviations: MT, magnesite tailing. Referring to the raw material that consists of 80% magnesium carbonate (MgCO3) before calcination; MKPC, magnesium potassium phosphate cement; KDP, potassium dihydrogen phosphate (KH2PO4); CBC, chemically bonded ceramic; CBPC, chemically bonded phosphate ceramic; MKP, magnesium potassium phosphate. A general term for the matrix phase in MKPC produced in the reaction between magnesia and potassium dihydrogen phosphate; M/P, magnesia to phosphate ratio; W/C, water to cement dry matter ratio. ⇑ Corresponding author. E-mail address: [email protected] (A. Ismailov). https://doi.org/10.1016/j.conbuildmat.2019.117098 0950-0618/Ó 2019 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

2

A. Ismailov et al. / Construction and Building Materials 231 (2020) 117098

1. Introduction Hundreds of natural and synthetic phosphate mineral structures are known [1], but only a select handful have found use in applications where mechanical properties are emphasized. Magnesium phosphate cements, or MPCs, are a group of cementitious chemically bonded ceramic (CBC) materials that have received attention as rapid repair materials in civil engineering [2,3] and cements compatible with organic fillers [4,5]. One other promising application is cold-climate concrete repair [6], since the setting of MPCs, like most other similar mixtures commonly deemed chemically bonded phosphate ceramics, or CBPCs, is noticeably exothermic [7]. This built-in heat source is beneficial for retaining the workability in cold climates, but for most room temperature applications the violent reactions behind this heat release are excessively rapid, and often the use of retardants is required to produce mechanically sound materials. Larger scale MPC bulk materials have sporadically been discussed in research, but thus far mostly regarding immobilization of hazardous waste [8,9]. The formation of a phosphate solid occurs by a through-solution reaction of reactive oxide component with phosphate ions in an aqueous solution. The formation of struvite-K (magnesium potassium phosphate hexahydrate) presented in Eq. (1), currently thought to form through an intermediary phase of newberyite (MgHPO43H2O) [10], is the main binder-forming reaction occurring in popular magnesia-potassium phosphate cements (MKPCs).

MgO þ KH2 PO4 þ 5H2 O ! MgKPO4  6H2 O

ð1Þ

The chemical composition of these common phases in MPCs and the cited reactions would lead to believe, that a molar ratio of roughly 1:1 would be optimal for producing the desired end result; a stable solid. In practice, however, a much larger surplus of magnesia is required. Earlier magnesia-phosphate cementrelated publications [11,12] have shown, that incorporating a surplus of magnesia makes the relative amount of potassium dihydrogen phosphate (KDP) and water the two rate controlling parameters and large amount of the primary magnesia in the MPC mixtures remains unreacted. The newly formed phases like newberyite and struvite-K quickly encapsulate the magnesia particles, heavily slowing down and eventually preventing further reactions with the phosphate-containing liquid phase. The exact conditions where such sealing occurs is characteristic to each mixture, being mostly a result of water-to-cement weight ratio (W/C) and magnesia-to-phosphate molar ratio (M/P), as well as the particle size distribution, reactivity and purity of the magnesia powder [13]. An exhaustive study by Ma et al. [13] demonstrated how optimal M/P of a given mixture of raw materials is tied to water content (W/C), but also showed by means of modeling how utilizing magnesia as the limiting factor is the least predictable and the least understood. In general, using an excess amount of phosphate salt would be detrimental due to it not being immobilized in a network-forming mineral structure, leaving it in leachable form and thus compromising the long-term chemical stability and mechanical integrity of the end product [14]. Lastly, an excess of water causes increased porosity and inferior mechanical properties. Additionally, phosphate reactivity of most laboratory grade divalent oxides (MgO, CaO, ZnO) is known to be excessive to form mechanically sound macroscopic structures [15], a feature that requires the use of retardants. Boron compounds are routinely used as retardants in MPCs, one of the common ones being boric acid (H3BO3) [16–18]. Magnesite mineral (magnesium carbonate, MgCO3) is a commonly utilized source of magnesium oxide (MgO), though due to magnesium being deemed a critical element by the EU, alternative

sources such as several Mg-rich silicates may become more relevant in the future. The transition of magnesite to magnesia is achieved by heat treatment at temperatures commonly above 600 °C [14]. This transition from carbonate to oxide by release of carbon dioxide (CO2) initially produces a porous microstructure [19], which can be retained if not heated further, maximizing the surface area of the resulting magnesia particles. Such high surface area variant of magnesia is known as ‘‘active magnesia” in cementrelated research [14]. Exposing this porous form to higher temperatures (>1200 °C) solidifies the structure by sintering mechanism and drastically reduces the reactive surface area as well as the defect sites that promote wetting and dissociation of magnesia in acid phosphate solutions [20]. Phosphate bonding, like other chemical bonding technologies such as alkali-activated geopolymers, is an attractive processing route for utilizing low-grade raw material sources such as mixed-mineralogy tailings, due to its high tolerance to impurities [16,21]. It has been found, that pure magnesium oxide has high solubility even in mild phosphoric acid solutions, which can result in poor strength of the final material. The reduction of surface area in magnesia by high-temperature treatment offers some level of control in phosphate reactivity. The aim of this study was to establish the effect of calcination on the phosphate reactivity of a talc mine-derived mixedmineralogy ferroan magnesite tailing, and to determine the optimal calcination temperature range for utilizing it in MKPC. Even though the decomposition behavior of the pure mineral equivalents of the main constituents is relatively well known, the appropriate temperature ranges for this specific tailing fraction needed to be determined experimentally.

2. Materials and methods 2.1. Magnesite tailing The magnesite-rich mine tailing was acquired from a talc mine located in Sotkamo, Eastern Finland. The main minerals of these talc deposit rocks are talc (40–65%) and ferroan magnesite (35– 55%) [22]. The ore is located in the middle of mica schists and sulphide-bearing blackschists. In addition to talc, the mine produces Ni concentrate as a side product. The mine tailings are formed when the talc ore is crushed, milled and chemically processed in order to recover the talc and Ni concentrate from the gangue material. The key constituent minerals of this tailing sample are presented in the Table 1. Mineralogical characterization was performed by SEM/EDS according to the methodology described in previous study [23]. The tailing consists mainly of ferroan magnesite and it is devoid of quartz and alkali feldspar minerals which is typical for tailings rich in carbonate minerals. EDS analysis of individual magnesite grains in Table 2 shows that the main constituent of the tailing is not stoichiometric MgCO3, but it has a varying amount of iron (Fe) substituting Mg in the lattice. It is known, that there is a solid solution series between magnesite (MgCO3) and siderite (FeCO3) [24]. Ferroan magnesite with iron carbonate content between 5% and 30% is called breunnerite [25], which is typical form of magnesite in Finland [22]. For the sake of clarity, ‘‘ferroan magnesite” and plain ‘‘magnesite” will be used when referring to this Fe-containing magnesium carbonate mineral in this paper. The minority minerals listed in Table 1 are mostly Mg-silicates also with a significant amount of Fe substitution, making iron content an overarching characteristic of this tailing.

3

A. Ismailov et al. / Construction and Building Materials 231 (2020) 117098 Table 1 Magnesite tailing (MT) mineral composition (% of total area by EDS). Ore deposit type

Mg-Fe silicates

Talc 1) 2)

6.90

Magnesite

1)

Dolomite

80.41

Mg silicates

1.83

10.39

2)

Other/undetermined 0.47

chlorite (Mg,Fe)3(Si,Al)4O10 and serpentine (Mg, Fe)3Si2O5(OH)4 groups talc Mg3Si4O10(OH)2, olivine (forsterite) Mg2SiO4

Table 2 Elemental composition (EDS) of magnesite grains. Element

Mg

Fe

Si

O

Ca

Mn

Al

Cr

Mass %

37.86 ± 2.56

8.82 ± 3.64

4.05 ± 1.77

52.54 ± 2.19

0.51 ± 0.25

0.77 ± 0.14

1.60

0.30

2.2. MKPC sample preparation The magnesite tailing fractions were heat treated in a muffle furnace (Carbolite RHF 15/15) in the temperature range of 350 °C to 1400 °C in 50-degree increments. The temperature range was selected so that it would cover wide range of morphology for the magnesia, from barely carbonate-free porous ‘‘reactive” magnesium oxide to highly densified, less reactive product. Heating and cooling rates of 5 °C/min were targeted for the heat treatments. Dwell time in the peak temperature was three hours for all samples and the batch size of tailing for heat treatment was 40 g. The mixing of the cement pastes was performed manually in plastic containers, a batch of 29.2 g at a time. Based on the DSC/ TG measurement (Fig. 3), where mass reduction corresponds with magnesite-to-magnesia transformation, an estimated magnesia content of 66% in the heat-treated tailings was assumed. The magnesia-to-phosphate molar ratio of M/P = 6.0 and water-tocement weight ratio of W/C = 0.26 were used for all mixtures. Potable water was used as the base for all of the mixtures. A freeflowing reagent grade potassium phosphate monobasic powder (potassium dihydrogen phosphate, KDP, KH2PO4, Sigma Aldrich) was used as the source for phosphate ions in the mixture. A small amount, 0.86 wt% of the dry components, reagent grade boric acid (H3BO3, Merck, Germany) powder was used as the retarder to increase the workability of the mixtures. The dimensions of the stainless-steel mold were such that individual samples sized 10 mm  10 mm  50 mm were produced. The mixtures were left to set in the mold for the initial 24 h, demolded after that and continued to dry for 144 h in ambient conditions before being finally subjected to load in the 4-point bend test after 7 days of setting in total. 2.3. Characterization and testing BET surface area of magnesite tailing and calcined samples were measured with Micromeritics 3Flex Physisorption surface characterization system using nitrogen probe gas. Particle size of the as-received magnesite raw material determined by a laser diffractometer (Malvern Mastersizer 3000) with water as the dispersant. The surface area estimation provided by the diffractometer (Table 3) assumes spherical particles, and due to the tailing particles being mostly angular in shape, the value correlates well with the BET surface area of the dry untreated tailing (see Table 4).

An X-ray diffractometer (XRD, Empyrean, PANalytical) with a Cu K-alpha radiation source and HighScore Plus software was used for the qualitative analysis of mineral species in heat treated magnesite tailings and the related MKPCs. TG/DSC-analysis was performed with thermogravimetry (Netzsch STA449 F1 Jupiter), using heating rate of 10 K/min to the peak temperature of 1450 °C. SEM imaging and analysis of heat-treated sand fractions was performed using a Philips XL-30 scanning electron microscope with an EDAX DX 4 EDS analyzer. Modulus of rupture tests were performed with a custommade jig (Fig. 1) similar to one described in ISO 14704/2008 [26] attached to an Instron 5967 mechanical test device. The span of top pressure points was 20 mm and the span of the articulated bottom supports was 40 mm. Loading rate of 1 mm/min was used for all testing. Strength values reported in this paper are based on at least two measurements per MKPC mixture. Additionally, an accelerated reactivity evaluation was performed for select batches of raw material in accordance to Chau et al. [27], where 7.6 g of calcined MT (providing 5.0 g of magnesite-derived magnesia) was added to 100 ml of 0.5 M orthophosphoric acid solution (Diluted from Sigma Aldrich VOLUSOL 10% solution) under magnetic stirring, and the resulting change in pH as well as temperature were logged as a function of time with a VWR MU 6100H MULTI meter. The same procedure was repeated with a 1.0 M solution of KDP. The output of accelerated reactivity assessment is a reactivity parameter R, which indicates a point in time where pH = 7 is reached. These measurements were to supplement the observations from strength data and reactivity potential predicted by BET surface area. To ensure the integrity of the test setup, reference measurements were taken with two commercial MgO powders (Sigma Aldrich, >98%, 325 mesh; Alfa Aesar, >96%, 325 mesh), producing reactivity values of R = 483 s; R = 480 s in phosphoric acid and R = 589 s; R = 478 s in KDP solution, respectively. R-values for acid solution were within 5% of values given by Chau et al., who reported R = 463 s for a reactive MgO powder in similar conditions. Our R-values for KDP solution differed more significantly from the reference of R = 690 s reported by Chau et al., possibly due to variation in particle size/grade of MgO and similar manufacturer details of KDP between our setups. The data logging frequency of 1 min was targeted where possible, but 10 min was used for measurements exceeding 24 h.

Table 3 Particle size and surface area of the as-received magnesite tailing.

*

D10 (mm)

D50 (mm)

D90 (mm)

Specific s.a.* (m2/g)

BET s.a. (m2/g)

10.7

102

213

0.220

0.242

determined by laser diffractometer

4

A. Ismailov et al. / Construction and Building Materials 231 (2020) 117098

Table 4 BET surface area of magnesite tailing after different heat treatments and the 4-point bend strength of the corresponding MKPC. Fraction

Tailing sample name

Calcination temperature

SBET [m2g1]

4-point bend strength (median) of MKPC [MPa]

Magnesite Tailing (MT) dry, untreated MT Heat-treated

MTref MT350 MT400 MT450 MT500 MT550 MT600 MT650 MT700 MT750 MT800 MT850 MT900 MT950 MT1000 MT1050 MT1100 MT1150 MT1200 MT1300 MT1400

50 °C, 24 h 350 °C, 3 h 400 °C, 3 h 450 °C, 3 h 500 °C, 3 h 550 °C, 3 h 600 °C, 3 h 650 °C, 3 h 700 °C, 3 h 750 °C, 3 h 800 °C, 3 h 850 °C, 3 h 900 °C, 3 h 950 °C, 3 h 1000 °C, 3 h 1050 °C, 3 h 1100 °C, 3 h 1150 °C, 3 h 1200 °C, 3 h 1300 °C, 3 h 1400 °C, 3 h

0.242 0.29 0.29 0.63 11.25 38.24 82.37 51.97 29.41 16.37 10.02 6.35 3.99 2.88 2.29 1.74 1.30 1.04 0.69 0.27 0.10

No solid No solid No solid No solid 7.8 9.6 5.1 13.2 13.8 8.0 7.8 9.2 11.0 12.1 8.7 10.6 12.7 7.1 4.0 7.2

Fig. 1. 4-point bend test setup.

3. Results and discussion 3.1. Characterization of MT raw material and calcined fractions The expected outcome of the heat treatments of magnesite tailing was to observe the temperature range where MgCO3 releases CO2 and transforms into periclase (MgO) with some addition of brucite (Mg(OH)2). Decomposition and transformation of minor minerals such as talc and other silicates was monitored as well. The XRD-spectra collected from the heat-treated fractions are presented in Fig. 2. The decomposition of magnesite appears to occur very sharply between 550 °C and 600 °C, overlapping with the appearance of periclase (MgO or (Mg1-xFex)O) and followed by magnesioferrite (MgFe2O4) appearing at 700 °C. This was in good agreement with previous work from Vusikhis et al., who determined these two to be the end products of talcomagnesite-derived breunnerite decarbonization [25]. The decomposition of magnesite is also displayed in the TG/DSC-data in the Fig. 3, where a mass loss of over 40% occurs between 600 °C and 695 °C. Assuming the magnesite

product product product product

content of over 80% in the tailing and factoring in the 52% mass loss associated with magnesite decomposition, the mass reduction can be completely attributed to CO2 release. The slightly endothermic feature after 850 °C with negligible mass reduction (Fig. 3) can be attributed to talc (Mg3Si4O10(OH)2) decomposing into enstatite (MgSiO3), poorly crystalline residual silica and water vapor [28,29]. Upon further heating, enstatite is said to transform into clinoenstatite (Mg2Si2O6) at 1200 °C and the silica should crystallize into cristobalite by 1300 °C, two phase transformations that do not change mass and have negligible effect on heat flow. This decomposition of residual talc at around 850 °C can be observed in the XRD data (Fig. 2) as well. However, while clinoenstatite was identified by XRD, cristobalite was not observed. The SEM/EDS compilation in Fig. 4 presents the morphology and elemental composition of the as-received Fe-containing magnesite (Fig. 4a) and at the extremes of the heat-treatment temperature range. The magnesite-derived grain after 3 h in 600 °C (Fig. 4b) retains the characteristic angular appearance similar to untreated grains, but shows severe fracturing and significant reduction in oxygen content due to CO2 escaping as part of the magnesite decomposition. Calcination in 1400 °C produced a powder that consisted of highly agglomerated particles where magnesia was covered by dendritic iron-containing phase, similar to the phase separation encountered in metallurgical slags and fly ash [30]. Due to its high atomic number compared to most other elements present in the constituent minerals, iron-containing phases can easily be distinguished by their high brightness in the BSE images. Iron content of almost 5 at.% presented in Fig. 4c is slightly exaggerated due to iron-containing phases coating the magnesia particles and dominating the grain boundaries of the agglomerates. The tailing showed increasingly noticeable color change from gray to reddish brown after heat treatments starting at 450 °C, continuing to intensify in brown shades until the maximum temperature of 1400 °C. The color change pointed towards iron-related activity, possibly the appearance of hematite. Even though most of the constituent minerals had significant iron content, none of the silicates have been reported to show major deterioration beyond dehydration in temperatures where color change was observed. However, a noteworthy mineral impurity group under other/undetermined was 0.24% of iron sulfides pyrite, pyrrhotite and pentlandite (FeS2, Fe(1-x)S and (Fe,Ni)9S8), that are known to dissociate into sulfur dioxide (SO2) gas and hematite (Fe2O3) under oxidizing atmosphere and elevated temperatures (200 °C – 650 °C)

A. Ismailov et al. / Construction and Building Materials 231 (2020) 117098

5

Fig. 2. XRD-spectra of heat-treated MT. Markers for magnesite (M), Talc (T), Dolomite (D), Chlorite (C), Magnesioferrite (F), Forsterite (O), Clinoenstatite (E), Periclase (P) and unidentified crystalline phase (U) are presented. Dwell time at peak temperature was 3 h.

Fig. 3. Thermal decomposition of as-received magnesite tailing.

[31]. Iron sulfide minerals were consistently encountered in SEM analysis of untreated tailing. Additionally, this presence of sulfides in the raw material would explain the anomalous presence of sulfur in the EDS spectrum of calcined magnesia particle in Fig. 4b.

3.2. Influence of calcination temperature on BET surface area of MT and strength of corresponding MKPC The evolution of surface area as a function of calcination temperature gives an estimate of what level of reactivity is to be expected. Adjusting the surface area of magnesia by calcination is crucial in controlling the phosphate reactivity of the tailingderived raw material, since any milling was not applied for the tailing. Beyond calcination, the amount of pre-treatments was limited to a minimum to give an accurate representation of the potential of utilizing this type of MgO-source on a larger scale as costeffectively as possible. Sample naming scheme, heat treatment

parameters, BET surface area and the median of 4-point bending are collected into Table 4 for easy reference. The increase in BET characteristic surface area around 450 °C to 600 °C is attributed to magnesite decomposition, though this range of decomposition is slightly higher than that reported for finer and higher purity magnesium carbonate [19]. The fracturing of grains due to decarbonization (see Fig. 4b) is proposed as the main source of BET surface area increase. Upon further heating, the porous magnesia structure that results from carbon dioxide removal begins to densify, thus decreasing the surface area. In this process, magnesia moves away from highly reactive ‘‘caustic” or light-burned grade towards hard-burned, a grade more commonly utilized in the phosphate cements due to its narrower range of chemical reactivity. Observing the strength of MKPC as a function of MT calcination temperature in Fig. 5 indeed shows that maximizing the surface area does not produce the highest strength. Peculiarly, calcination in 500 °C increased the surface area to 11.25 m2/g, yet no solid MKPC could be produced. This would suggest that most of this

6

A. Ismailov et al. / Construction and Building Materials 231 (2020) 117098

Fig. 4. Magnesite in as-received state (a), after 3 h@600 °C (b), and after 3 h@1400 °C (c).

effect was not due to carbonite mineral decomposition, but some other mineral in the MT undergoing a deterioration that increased the surface area. Of other minerals in the tailing, chlorites (specified as clinochlore in XRD spectra) and serpentines can contribute to the area increase during calcination. Fig. 6 shows a Mg-chlorite grain that has undergone lattice deterioration similar to that of magnesite presented earlier. Villieras et al. determined by heat-treatments of magnesian chlorite [32], that the first dimensional modifications

of chlorite crystals occur as early as 450 °C – 500 °C. Upon further heating, at 600 °C, an increase in porosity should be expected due to thermal decomposition of interlayer octahedral sheets caused by the escape of entrapped water molecules. Regarding serpentine minerals, Hrsak et al. [33] determined that lattice break up occurs at 660 °C. Heating to 800 °C  850 °C causes chlorite to dehydroxylate further and recrystallize into variants of spinel (MgAl2O4), forsterite (Mg2SiO4) and enstatite (MgSiO3) [32]. Forsterite is the main product of heating serpentine to 800 °C and above as well.

A. Ismailov et al. / Construction and Building Materials 231 (2020) 117098

7

Fig. 5. 4-point bend strength of solid MKPC pieces and the corresponding BET surface area of heat-treated MT.

Fig. 6. An iron-containing Mg-chlorite grain after 3 h in 600 °C (MT600).

The XRD-spectrum of MT1400 in Fig. 2 shows, that forsterite is one of the few remaining constituents of the almost dead-burnt MT, showing only periclase, magnesioferrite and oxides of iron as the other crystalline phases. While surface area is a key factor in determining the solubility behavior of magnesia in phosphate solution, increase in surface area by thermal decomposition does not affect Mg-silicates enough to show an increase in phosphate reactivity. This would explain why the increase in BET surface area did not correlate with the calcination range required to produce solid MKPC test pieces. Dolomite (CaMg(CO3)2) comprises less than 2% of the tailing, but its decomposition could be observed at 700 °C, likely minutely adding to the overall periclase yield, though this is not reliably verifiable at such low concentration. As with the other constituent minerals in MT, the exact decomposition behavior of dolomite depends on the purity and particle size. According to OlszakHumienik et al. [34] natural dolomite decomposes between 550 °C and 861 °C, first by dissociation of Mg-portion, followed by deterioration of Ca-portion, both accompanied by varying levels of CO2 release and finally producing a mixture of MgO and CaO when completely decarbonized. Calcination at low temperatures can leave residual carbonate mineral matter in the sand, enabling

CO2 release as a side product of the carbonate dissolution into acidic phosphate medium to form phosphate salts [35]. This is a potential source for additional porosity of a final MKPC material. Another potentially detrimental effect the presence of carbonate minerals could have on phosphate binder formation is based on their fairly high solubility in acids. This means, that when residual carbonate is present in the wet MKPC mixture, phosphate ions from KDP are more readily spent on carbonate-related phosphate salt formation instead of interacting with the less reactive primary oxide source [23] to produce a load-bearing microstructure. Since KDP and water generally are the limiting reagents when producing MKPCs, this rapid spending of KDP will have a profound effect on the structural integrity of the final material. This interference from residual carbonates could explain why the magnesite calcined in relatively low temperatures produced low-strength pieces despite there being plenty of surface area available. In addition to the measurable outcomes, the calcination temperature of the MT also affected the workability of the MKPC paste, subjective amount of exothermic heat observed during mixing and the color of the paste. Contrary to the assumed implications of BET surface area data, the highest heat under manual mixing as well as the shortest workability window was observed with MT fractions

8

A. Ismailov et al. / Construction and Building Materials 231 (2020) 117098

Fig. 7. XRD spectra of MKPC prepared from MT650, MT1000 and MT1400. Key markers for Struvite-K (*), MgO (periclase) (P), Talc (T), Forsterite (Mg2SiO4) (F) and residual KDP (K) are presented.

calcined between 900 °C and 1050 °C. Differences in mixture workability may have contributed to some of the low strength results acquired within this region by making effective casting more difficult. Struvite-K (MgKPO46H2O) was confirmed as the main reaction product as presented by the Fig. 7. Residual magnesia was detected in all cases, as predicted by using a large surplus of MgO source, thus making the amount of KDP and water the limiting factors for binder formation. In agreement with the XRD-data of the heat-treated MT, residual talc can be observed in the sample made of caustic grade raw material MT650. In the case of MT1400, the resulting MKPC shows residual KDP in the XRD spectrum, clearly indicating that even with the 6:1 excess of MgO to KDP ratio is not enough to spend the amount of KDP in the mixture due to magnesia being highly agglomerated and not very effectively available for dissolution. 3.3. Accelerated reactivity assessment The accelerated reactivity assessment provides an insight on how the differently calcined MT raw material batches behave in aqueous acid phosphate environment. Fig. 8 presents the accelerated reactivity of select MT-samples in 0.5 M orthophosphoric acid solution. As predicted by other characterization results, MT1400 neutralized the acidic solution very slowly compared to the two less severely calcined samples. The reactivity parameter of MT1000, R = 17 h and 32 min, was closer to that of MT1400 (R = 29 h 45 min) than MT600 (R = 54 min). Judging by R alone, the behavior of MT1000 and MT1400 should be similar. However, this was clearly not the case considering the XRD-data and the strength measurement results presented earlier. Cropping the graph to initial two hours supplemented with temperature data, as is presented in the insert of Fig. 8 reveals, that despite having the R-parameter closer to that of MT1400, the dissolution behavior of MT1000 shows more similarity to MT600 during the first minutes of the test. The initial temperature increase arising from the exothermic reaction of phosphate formation occurs rapidly and is close to 10 °C for MT600 and MT1000, whereas MT1400 produces an increase of 4 °C that is far more delayed. The pH increase for MT1400 was sluggish as well, taking one hour to reach pH = 4. The dip after initial rise in pH shown by all MT samples in the acid

solution currently has no clear explanation. It is not equipmentrelated, since this was not observed with the commercial MgO powders during reference measurements. Changes in temperature did not correlate with this feature for all samples: MT600 and MT1000 showed a drop in pH roughly after peak temperature, but for MT1400 this did not apply. The accelerated reactivity assessment results for 1.0 M KDP are provided in Fig. 9. The starting pH was much higher and the expected severity of reactions thus lower than those observed with acid solution. For MT1400 the remarks from acid testing carry over to KDP test: the heat output and the increase of pH were the slowest by far of the three samples. The pH evolution for MT600 and MT1000, on the other hand, showed very similar trends and their reactivity parameters were practically identical, R = 1 h 56 min for MT600 and R = 1 h 53 min for 1000MT, respectively. However, when cropped to the initial two hours and presented with temperature data in the insert of Fig. 9, some significant differences can be observed, especially within the first 30 min. The sharp increase both in temperature and pH was similar to those observed in the acid solution, but there was also a sudden near-stagnation of both for MT600 after just a few minutes, resuming an upward trend after about 20 min. The MT1000 did not show this behavior, and even though the initial rise in pH and temperature were slower, the absence of this stagnation step below pH = 7 allowed it to reach the measurement limit as quickly as the more reactive MT600. It was noted by Chau et al. [27] that the formation of phosphate salt species around the magnesia particles during dissolution greatly increases the time it takes for MgO to neutralize an acidic phosphate solution, resulting in a near-horizontal feature in pH-trend lines. The step-like feature in the Fig. 9 for MT600 and the nearhorizontal phases observed for MT1000 and MT1400 in Fig. 8 would be explainable by this phenomenon. The peak temperature change for MT600 and MT1000 for MKP-solution was effectively the same, though the latter appeared to cool down more quickly, even more than MT1400 at the 2-hour mark. The better retention of elevated solution temperature by MT600 in our opinion supports the interpretation that the reaction continues after a short period of stagnation. Other than showing the relative reactivity of select MT fractions, the accelerated reactivity measurements also showed the effect of large particle size of MT raw material. Despite having a

A. Ismailov et al. / Construction and Building Materials 231 (2020) 117098

9

Fig. 8. Accelerated reactivity measurements in H3PO4-solution for MT600, MT1000 and MT1400. Dotted line denotes temperature measurement for a solid pH trend line of corresponding colour.

Fig. 9. Accelerated reactivity measurements in KDP-solution for MT600, MT1000 and MT1400. Dotted line denotes temperature measurement for a solid pH trend line of corresponding colour.

very high BET surface area of over 80 m2/g, MT600 showed a relatively high reactivity parameter R = 3300 s, which is significantly higher than values commonly reported in literature [12,27]. The BET surface area range that produced relatively high-strength MKPCs was roughly 1.0–16.3 m2/g, while common values reported for specific surface area of finer magnesia elsewhere are below that [18,36], though Soudée and Péra have occasionally used higher values in their studies [20,37].

The accelerated reactivity results only allow subjective comparison of sample fractions in different environments, as demonstrated by the relative differences in behavior of MT1000 in the two solutions. It could be argued, that in the MKPC mixtures the low amount of water and different M/P-ratio would amplify the differences in reactivity and heat output. The MT600 could react even more vigorously and encapsulate the magnesia from further reactions with the solvent, which ultimately results in

10

A. Ismailov et al. / Construction and Building Materials 231 (2020) 117098

unpredictable workability characteristics that are reflected on the large spread in 4-point bend strength values presented in the Fig. 5. The lower reactivity of MT1000 could be just enough to avoid some of that effect while providing a more sustained source of magnesia and consequently a more consistent, higher integrity MKP matrix. The results surrounding MT1400 show a clear loss of reactivity caused by excessive calcination, to a point where residual KDP could be observed in the final MKPC product, pointing towards incomplete reaction that ultimately manifests in reduced mechanical integrity displayed by the low bending strength. It should be noted, that even though boric acid was used in the MKPCs, the retarding effect has more to do with limiting the formation of Mg-K-phosphate species rather than limiting the solubility of the periclase phase [17]. This gives some confidence to maintain that, even though boric acid was not used in the accelerated reactivity tests, the main features of the results are cautiously applicable for explaining the behavior of calcined MT-fractions in the MKPC-pastes. 4. Conclusions A ferroan magnesite mine tailing (MT) originating from a Finnish talc ore mine was investigated as a potential raw material for magnesium potassium phosphate cement (MKPC). The activation of magnesia (MgO) was conducted by calcination in temperatures between 350 °C and 1400 °C for 3 h. The following conclusions were drawn from characterization results of the heat-treated tailing raw material and the corresponding MKPC mixtures:
An increase in measured BET surface area occurred well before the magnesite decomposition temperature derived from TG data. This also did not directly translate into tangible phosphate reactivity at low temperatures (450 °C – 500 °C). Solid MKPC pieces could only be cast from MT calcined at 550 °C and above. This was explained by crystal fracturing of chlorite and serpentine by dehydroxylation at low temperatures, what contributed to increase in surface area without increasing the phosphate reactivity of the heat treated MT.
The high amount (20%) of Mg/Fe-silicates, the large particle size (D50 = 102 mm) and significant (5%–12%) Fe-substitution in magnesite resulted in the optimal calcination temperature to be significantly lower than suggested in many other cases for MKPC use. According to 4-point bend test results, the preferred calcining temperature for the magnesite tailing under investigation would be around 700 °C  1150 °C, resulting in what would be classified as hard-burned magnesia in case of a pure raw material.
Struvite-K was the dominant binder phase in all mixtures that produced a solid MKPC.
Calcining the magnesite at temperatures above 1150 °C reduced the available surface area of the MgO to a level where residual KDP was observed in final MKPC. This was again explained by the presence of (Mg,Fe)-silicate minerals that caused high level of agglomeration and subsequent sintering-like densification of the Fe-containing magnesia particle clusters, which resulted in very poor solubility of Mg and ultimately poor strength of final MKPC.
Reactivity parameters observed in accelerated reactivity assessment supported the strength measurement results and XRD analysis of solid MKPC products. The maximum surface area acquired in 600 °C produced excessively rapid reaction that was not sustained after initial exothermic surge. Calcination above 600 °C allows the newly formed periclase to densify further, reducing the reactivity to a level where solubility is still

sufficiently high to facilitate MKP formation without the full effect of encapsulating magnesia particles and isolating them from the acidic medium. Utilizing a mixed-mineralogy MgO-source such as the mine tailing discussed in this paper does not follow the recommended heat treatment range commonly accepted to control the level of MgO reactivity for MKPC use. These findings are specific to the test parameters and mix ratios presented herein, and different results could be acquired by varying the W/C- and/or M/P-ratios, as well as the amount of retarder used, providing many potential routes for future research. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgements This work was done within the CeraTAIL project as part of MISU-program financed by the Academy of Finland. We thank the Academy of Finland (grant number 292564) for financial support. We sincerely thank Dr. Marja-Liisa Räisänen at Geological Survey of Finland (GTK) Kuopio for contribution to mineralogical characterization. We would also like to extend our gratitude to Miss Nelli Palmu at Tampere University (TAU) for accommodating prolonged pH measurements and MSc Leo Hyvärinen at Tampere University (TAU) for help with XRD analysis. Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi.org/10.1016/j.conbuildmat.2019.117098. References [1] J.P. Attfield, Phosphates, in: Encycl. Mater. Sci. Technol., 2001: pp. 6896–6901. [2] Y. Li, W. Bai, T. Shi, A study of the bonding performance of magnesium phosphate cement on mortar and concrete, Constr. Build. Mater. 142 (2017) 459–468, https://doi.org/10.1016/j.conbuildmat.2017.03.090. [3] J.W. Park, K.H. Kim, K.Y. Ann, Fundamental Properties of Magnesium Phosphate Cement Mortar for Rapid Repair of Concrete, Adv. Mater. Sci. Eng. 2016 (2016), https://doi.org/10.1155/2016/7179403. [4] M.A. Haque, B. Chen, Research progresses on magnesium phosphate cement: a review, Constr. Build. Mater. 211 (2019) 885–898, https://doi.org/10.1016/ j.conbuildmat.2019.03.304. [5] M.R. Ahmad, B. Chen, S. Yousefi Oderji, M. Mohsan, Development of a new biocomposite for building insulation and structural purpose using corn stalk and magnesium phosphate cement, Energy Build. 173 (2018) 719–733, https://doi. org/10.1016/j.enbuild.2018.06.007. [6] Q. Yang, B. Zhu, X. Wu, Characteristics and durability test of magnesium phosphate cement-based material for rapid repair of concrete, Mater. Struct. 33 (2000) 229–234, https://doi.org/10.1007/BF02479332. [7] A.S. Wagh, S.Y. Jeong, Chemically Bonded Phosphate Ceramics: I, A Dissolution Model of Formation, J. Am. Ceram. Soc. 86 (2003) 1838–1844, https://doi.org/ 10.1111/j.1151-2916.2003.tb03569.x. [8] W.A. Ibrahim, H.A. Sibak, M.F. Abadir, Preparation and Characterization of Chemically Bonded Phosphate Ceramics (CBPC) for Encapsulation of Harmful Waste, J Am. Sci. 7 (2011) 543–548. [9] A. Viani, A.F. Gualtieri, Preparation of magnesium phosphate cement by recycling the product of thermal transformation of asbestos containing wastes, Cem. Concr. Res. 58 (2014) 56–66, https://doi.org/10.1016/j. cemconres.2013.11.016. [10] M. Le Rouzic, T. Chaussadent, G. Platret, L. Stefan, Mechanisms of k-struvite formation in magnesium phosphate cements, Cem. Concr. Res. 91 (2017) 117– 122, https://doi.org/10.1016/j.cemconres.2016.11.008. [11] D.A. Hall, R. Stevens, B. El Jazairi, Effect of water content on the structure and mechanical properties of magnesia-phosphate cement mortar, J. Am. Ceram. Soc. 81 (1998) 1550–1556. [12] B. Xu, H. Ma, Z. Li, Influence of magnesia-to-phosphate molar ratio on microstructures, mechanical properties and thermal conductivity of magnesium potassium phosphate cement paste with large water-to-solid

A. Ismailov et al. / Construction and Building Materials 231 (2020) 117098

[13]

[14]

[15] [16]

[17]

[18]

[19]

[20] [21]

[22] [23]

[24]

ratio, Cem. Concr. Res. 68 (2015) 1–9, https://doi.org/10.1016/j. cemconres.2014.10.019. H. Ma, B. Xu, Potential to design magnesium potassium phosphate cement paste based on an optimal magnesia-to-phosphate ratio, Mater. Des. 118 (2017) 81–88, https://doi.org/10.1016/j.matdes.2017.01.012. S.A. Walling, J.L. Provis, Magnesia-Based Cements: A Journey of 150 Years, and Cements for the Future?, Chem. Rev. 116 (2016) 4170–4204, https://doi.org/ 10.1021/acs.chemrev.5b00463. A. Wagh, Chemically Bonded Phosphate Ceramics, 1st ed., Elsevier, Argonne, 2004. J. Formosa, J.M. Chimenos, A.M. Lacasta, M. Niubó, Interaction between lowgrade magnesium oxide and boric acid in chemically bonded phosphate ceramics formulation, Ceram. Int. 38 (2012) 2483–2493, https://doi.org/ 10.1016/j.ceramint.2011.11.017. H. Lahalle, C.C.D. Coumes, A. Mesbah, D. Lambertin, C. Cannes, S. Delpech, S. Gauffinet, Investigation of magnesium phosphate cement hydration in diluted suspension and its retardation by boric acid, Cem. Concr. Res. 87 (2016) 77–86, https://doi.org/10.1016/j.cemconres.2016.04.010. Q. Yang, X. Wu, Factors influencing properties of phosphate cement-based binder for rapid repair of concrete, Cem. Concr. Res. 29 (1999) 389–396, https://doi.org/10.1016/S0008-8846(98)00230-0. J.K. Bartley, C. Xu, R. Lloyd, D.I. Enache, D.W. Knight, G.J. Hutchings, Simple method to synthesize high surface area magnesium oxide and its use as a heterogeneous base catalyst, Appl. Catal. B Environ. 128 (2012) 31–38, https:// doi.org/10.1016/j.apcatb.2012.03.036. E. Soudée, J. Péra, Influence of Magnesia Surface on the Setting Time of Magnesia - Phosphate Cement, Cem. Concr. Res. 32 (2002) 153–157. P. Kinnunen, A. Ismailov, S. Solismaa, H. Sreenivasan, M.-L. Räisänen, E. Levänen, M. Illikainen, Recycling mine tailings in chemically bonded ceramics – a review, J. Clean. Prod. 174 (2018), https://doi.org/10.1016/j. jclepro.2017.10.280. M. Lehtinen, Industrial Minerals and Rocks, Miner. Depos. Finl. (2015) 685– 705. S. Solismaa, A. Ismailov, M. Karhu, H. Sreenivasan, M. Lehtonen, P. Kinnunen, M. Illikainen, M.L. Räisänen, Valorization of finnish mining tailings for use in the ceramics industry, Bull. Geol. Soc. Finl. 90 (2018), https://doi.org/ 10.17741/bgsf/90.1.002. L.L.Y. Chang, R.A. Howie, J. Zussman, Non-silicates: Sulphates Carbonates, Phosphates, Halides, Rock-Forming Miner. 5B (1998) 383.

11

[25] A.S. Vusikhis, R.I. Gulyaeva, L.I. Leont’ev, L.A. Ovchinnikova, E.N. Selivanov, Kinetic features of breunnerite decarbonization, Russ. Metall. 2016 (2016) 793–797. doi:10.1134/S0036029516090172 [26] ISO: International Organization for Standardization, ISO 14704:2008 Fine ceramics (advanced ceramics, advanced technical ceramics) – Test method for flexural strength of monolithic ceramics at room temperature, 2nd ed., 2008. [27] C.K. Chau, Z. Li, Accelerated reactivity assessment of light burnt magnesium oxide, J. Am. Ceram. Soc. 91 (2008) 1640–1645, https://doi.org/10.1111/ j.1551-2916.2008.02330.x. [28] J. Res. Natl. Bur. Stan. 15 (5) (1935) 551, https://doi.org/10.6028/jres.015.039. [29] K. Belgacem, P. Llewellyn, K. Nahdi, Thermal behaviour study of the talc, Optoelectron. Adv. Mater. – Rapid Commun. 2 (2008) 332–336. [30] R.R. Lloyd, J.L. Provis, J.S.J. Van Deventer, Microscopy and microanalysis of inorganic polymer cements. 1: remnant fly ash particles, J. Mater. Sci. 44 (2009) 608–619, https://doi.org/10.1007/s10853-008-3077-0. [31] S.K. Bhargava, A. Garg, N.D. Subasinghe, In situ high-temperature phase transformation studies on pyrite, Fuel. 88 (2009) 988–993, https://doi.org/ 10.1016/j.fuel.2008.12.005. [32] F. Villieras, J. Yvon, J.M. Cases, P. De Donato, F. Lhote, R. Baeza, Development of microporosity in clinochlore upon heating, Clays Clay Miner. 42 (1994) 679– 688, https://doi.org/10.1346/CCMN.1994.0420604. [33] D. Hrsak, J. Malina, A. Hadzipasic, The decomposition of serpentine by thermal treatment, Mater. Tehnol. 39 (2005) 225–227. http://193.2.15.2/Revija/izvodi/ mit056/hrsak.pdf. [34] M. Olszak-Humienik, M. Jablonski, Thermal behavior of natural dolomite, J. Therm. Anal. Calorim. 119 (2015) 2239–2248, https://doi.org/10.1007/s10973014-4301-6. [35] I.L. Shashkova, N.V. Kitikova, A.I. Rat’ko, A.G. D’yachenko, Preparation of calcium and magnesium hydrogen phosphates from natural dolomite and their sorptive properties, Inorg. Mater. 36, 2000, pp. 826–829. [36] B. Xu, B. Lothenbach, A. Leemann, F. Winnefeld, Reaction mechanism of magnesium potassium phosphate cement with high magnesium-to-phosphate ratio, Cem. Concr. Res. 108 (2018) 140–151, https://doi.org/10.1016/j. cemconres.2018.03.013. [37] E. Soudée, J. Péra, Mechanism of setting reaction in magnesia-phosphate cements, Cem. Concr. Res. 30 (2000) 315–321, https://doi.org/10.1016/S00088846(99)00254-9.