Mechanochemical Synthesis of Nanocrystalline Olivine-Type Mg₂SiO₄ and MgCoSiO₄

Abstract

Nanocrystalline olivine-structured Mg₂SiO₄ and MgCoSiO₄, with an average particle size of 27 nm and 31 nm, respectively, were successfully synthesized from oxide precursors via mechanochemical methods. The two nanocrystalline products were obtained after milling for 360 min and displayed high concentrations of Mg₂SiO₄ (>94%) and MgCoSiO₄ (>95%), together with minor amounts of WC (~3%) contaminant originating as debris abraded off milling balls and chambers. The macroscopic temperature monitoring of the grinding jars during milling trials recorded a peak temperature of 75 °C. A combination of analytical techniques that included XRD, TEM, SAED, and EDS were employed for the characterization of the synthesized products.

1. Introduction

Olivine-type orthosilicates (M,N)₂SiO₄, where M and N are mainly Fe and Mg, with small amounts of Co, Mn, Ca, and Ni present, represent a fundamental class of minerals of mafic and ultramafic igneous rocks in the crust and upper mantle of the Earth [1,2,3]. While forsterite (Mg₂SiO₄) and fayalite (Fe₂SiO₄) are the predominant end members, other binary silicates—i.e., larnite (Ca₂SiO₄); liebenbergite (Ni₂SiO₄); tephroite (Mn₂SiO₄); and Ca-containing ternary silicates, i.e., kirschsteinite (CaFeSiO₄), monticellite (CaMgSiO₄), and glaucochroite (CaMnSiO₄)—are also naturally occurring minerals. The olivine structure is among the most robust crystallographic arrangements and able to accommodate a variety of metal cations, together with different types of anionic groups, i.e., germanate GeO₄²⁻, SiS₄²⁻, phosphate PO₄³⁻, and GeS₄²⁻ [4,5,6,7].

The naturally occurring solid solution series of olivine (Fe_1−xMg_x)₂SiO₄ has been the subject of intensive research for refractories, battery materials, mineralogy, and carbon sequestration [8,9,10,11,12]. Mg₂SiO₄ exists in three well-established equilibrium polymorphs: orthorhombic Pnma forsterite α-Mg₂SiO₄, orthorhombic Imma wadsleyite spinelloid β-Mg₂SiO₄, and cubic Fd-3m spinel ringwoodite γ-Mg₂SiO₄ [13]. The transformation of α-Mg₂SiO₄ to β-Mg₂SiO₄ and γ-Mg₂SiO₄ occurs at elevated pressure and temperature, yet both stabilize at ambient conditions upon quenching.

Natural olivines incorporate small amounts of Co; the parameters controlling this process, e.g., diffusion rates, and the effects of Co incorporation on the physical properties of olivine have also been studied [14,15]. The end member of the (Mg,Co)₂SiO₄ solid solution, Co₂SiO₄, is not a recognized mineral, but can be produced synthetically. It also exists in three polymorphic forms, analogous to Mg₂SiO₄ phases. Ringwood et al. found that the transformation of Co₂SiO₄ from the olivine structure to the ringwoodite structure took place at 7 GPa and 700 °C, and Morimoto et al. reported on the syntheses and crystal-chemical relationships of the three Co₂SiO₄ polymorphs [16,17].

Recently, olivine solid solutions, MgM’SiO₄ (where M’ = Fe, Co, Mn), have received renewed interest as potential cathode materials for rechargeable magnesium batteries due to their high theoretical energy densities. The synthesis of these olivine solid solution series utilizes various conventional methods including a combination of modified sol–gel, molten salt, and water-based solution chemistry, all of which involve multi-step synthesis, while some required heating to above 450 °C [18,19,20,21,22,23,24]. For example, Li et al. prepared MgFeSiO₄ through a flux method utilizing stoichiometric amounts of MgO, FeC₂O₄·2H₂O, and SiO₂ and heat treatment at >800 °C in a controlled Ar atmosphere [19]. Qafoku et al. reported the synthesis of nanosized Fe(II)-rich olivine, fayalite, (Mg_0.33Fe_0.66)₂SiO₄, and (Mg_0.66Fe_0.33)₂SiO₄, using a metal ion-sucrose solution approach, with calcination at >850 °C under a flow of CO₂/CO/Ar balanced gas mixture and the removal of excess iron metal from isopropanol suspensions [23]. Feng et al. prepared Mg_1.03Mn_0.97SiO₄ by milling a stoichiometric mixture of MgO, MnCO₃, and SiO₂, followed by sintering at 900 °C for 24 h [18]. Mori et al. synthesized MgMnSiO₄ using a flux method, starting from a 1:1:1 molar ratio mixture of MgO, MnCO₃, and SiO₂ ball-milled in ethanol at 300 RPM with subsequent calcination at >500 °C under a flux of 3% H₂/Ar balance [24]. Nuli et al. and Zheng et al. synthesized MgCoSiO₄ utilizing water-based solution chemistry and a molten salt/mixed solvothermal approach in combination with heat treatment at 1000 °C and 1300 °C, respectively [20,21].

Mechanochemical synthesis by high-energy ball milling has become an important solid-state synthesis method for the production of nanomaterials, with advantages including large-scale capabilities (grams) and low reaction temperatures. The one-pot synthesis of olivine-type Fe₂SiO₄ highlights the early work of Šepelák et al., starting from Fe₂O₃, Fe, and SiO₂ and milled under Ar inert atmosphere [25]. Mechanochemically synthesized nanocrystalline Co₂SiO₄ has also been reported recently by our lab using a 2:1 molar ratio mixture of CoO and SiO₂ [26]. In an effort to standardize the Mg-Co silicate solution series preparation, we report here the results of our investigation into the mechanochemical synthesis of olivine-type Mg₂SiO₄ and ternary olivine MgCoSiO_4, starting with stoichiometric amounts of MgO, SiO₂ and CoO, MgO, SiO₂, respectively.

2. Experimental Methods

General Experimental Details. All sample loading was conducted on a benchtop under ambient conditions. Powder XRD analysis of the substrates revealed that the CoO material contained 12.88 wt.% of spinel-structured mixed-valence cobalt oxide, Co₃O₄ (known as mineral guite) (see Supplementary Materials Figure S1). SiO₂ starting material was analyzed in the same manner as the cobalt oxide and was found to consist of pure phase α-quartz with no measurable impurities (see Supplementary Materials Figure S2). Prior to the mechanochemical processing, the MgO starting material was heat-treated at 400 °C for 4 h to dehydrate any Mg(OH)₂ that was formed upon reacting with water in the ambient atmosphere. Powder XRD analysis of the sample after heat treatment indicated a high concentration of periclase MgO phase with trace amount of calcite CaCO₃ (see Supplementary Materials Figure S3).

Powder X-ray Diffraction (PXRD). The sample was analyzed by high-resolution X-ray powder diffraction on a Bruker D8 Advance diffractometer with a 3 kW CuKα source and a LynxEye XE detector in Bragg–Brentano parafocusing geometry, mounted on a 9-position automated sample changer using either standard puck-type mounts or zero-background Si-wafer mounts when only a small quantity of the processed sample was available. A 0.020 mm Ni filter was used to reduce CuKβ radiation. Typical powder scans were conducted over a range of 5–85 degrees with a step size of 0.02 degrees and an exposure time of 1 s per step.

Rietveld refinement analysis. Rietveld analysis was conducted using Bruker TOPAS 5 [27]. Starting models of MgCoSiO₄, Mg₂SiO₄, SiO₂, WC, CoWO₄, Co₃O₄, MgSiO₃, MgO, and CaCO₃ structures were taken from PDF 04-017-1377, PDF 00-034-0189, PDF 00-046-1045, PDF 04-001-2755, PDF 00-015-0867, PDF 00-042-1467, COD 9007016, PDF 00-045-0946, and PDF 00-066-0867, respectively [28]. Refinement included the optimization of background, phase fractions, unit cell parameters, peak profiles (controlled by grain size and strain models), site occupancies for non-oxygen atoms, and atomic displacement parameters. Fractional atomic coordinates were kept fixed at literature values.

Transmission Electron Microscopy (TEM). In order to determine the elemental composition and to assess the chemical homogeneity of the final milled product, we used an 80–300 keV high-base Titan (FEI Thermo-Fisher) scanning transmission electron microscope equipped with a solid-state Si(Li) energy-dispersive X-ray detector (Genesis 4000, EDAX). Brightfield and darkfield images were acquired in scanning transmission (“STEM”) mode and the crystal structures of grains were assessed using selected area electron diffraction (SAED). Compositions were measured by energy-dispersive X-ray spectroscopy (“EDS”) and quantified using a Cliff–Lorimer thin-film approximation and correction factors (“K factors”) derived from thin-film standards. This procedure assumes that the specimen is sufficiently thin that any X-ray absorption by the specimen is minimal.

Preparation of Mg₂SiO₄/MgCoSiO₄. The synthetic procedure was based on the methodology developed previously by our group for the synthesis of the nanocrystalline Co₂SiO₄ [26]. A mixture of a 2:1 molar ratio of magnesium oxide (0.62 g) and silicon oxide (0.47 g) and a second mixture of a 1:1:1 molar ratio of magnesium oxide (0.25 g), cobalt oxide (0.47 g), and silicon oxide (0.37 g) were loaded into 2 separate 25 mL tungsten carbide (WC) jars and mechanically milled under dry conditions with two 15 mm diameter WC balls (40:1 ball to powder ratio) oscillating at a frequency of 30 Hz for various durations of time, ranging from 0 to 360 min. Milling experiments performed in air or under a controlled atmosphere (Ar) produced no difference in the processing outcomes.

3. Results and Discussion

To set up a baseline methodological approach, forsterite Mg₂SiO₄ was prepared by the high-energy milling of a 2:1 molar ratio mixture of MgO and SiO₂ for 180 min. Our synthesis follows the general procedure of Nguyen et al. for the synthesis of olivine Co₂SiO₄ [26]. Figure 1 illustrates the results of the Rietveld refinement of 180 min milled Mg₂SiO₄, which converge to a final figure of merit R_wp = 6.524. Small quantities of unreacted quartz SiO₂ (3.03%) and qusongite WC (2.68%) that were shaved off the milling media were detected alongside the Mg₂SiO₄ (94.29%). A good agreement between calculated and measured diffraction patterns was obtained based on the Rietveld refinements. The average crystallite size for the synthesized Mg₂SiO₄ forsterite, based on the Rietveld refinement, was 26 nm.

Figure 2 shows a comparison between the powder XRD patterns of the product from the 1:1:1 molar ratio mixture of MgO, CoO, and SiO₂ milled for different lengths of time. Short milling times (up to 45 min) only produced changes in the average grain size of the sample and lattice strain, which are indicated by the increasing diffraction peak width, but no new peaks were observed. It was also observed that MgO turns amorphous after 15 min into milling. A weak pattern of orthorhombic olivine MgCoSiO₄, indicated by the characteristic peak at 35.7°, was observed after 90 min of milling time. The formation of MgCoSiO₄ from this starting mixture was complete after 360 min of milling, as indicated by the complete disappearance of the strong SiO₂ quartz diffraction peak at 26.8°. The diffraction peak width of the product phase is similar to that of the peaks of the precursors after 45 min of milling, indicating sub-micrometer grain sizes. WC debris, produced by the wear of the grinding elements, was also detected in the final product.

The milling process in an oscillating mill is believed to proceed at only slightly elevated average temperatures (<100 °C), although significant localized heating may occur from individual ball and sample impacts [29,30]. Attempts to monitor the temperature of the milling jar with an attached thermocouple registered that a maximum temperature of 75 °C was reached during the milling trials. We further investigated the effect of temperature on the synthesized MgCoSiO₄ olivine by annealing the milled powder at 1000 °C for 12 h.

Figure 3 illustrates the results of the Rietveld refinements on the 360 min milled samples with and without subsequent annealing, which converge to final figures of merit R_wp = 2.219 and 2.906, respectively. The Rietveld refinements of powder both before and after the annealing display marked changes in the peak width of the olivine phase, signifying grain growth. Annealing at 1000 °C further induced a side reaction of the WC debris with the olivine MgCoSiO₄ sample, resulting in minor impurities of CoWO₄ (1.07%) in the final product. Unreacted guite Co₃O₄ (0.05%) and clinoenstatite MgSiO₃ (3.60%) were also detected as minor phases. Unit cell parameters of olivine Mg_2−xCo_xSiO₄ (x = 0, 0.5, 1) show the expected linear trend of composition, confirming the successful diffusion of Mg and Co into the olivine lattice (

Table 1. Unit cell parameters obtained from the Rietveld refinement of the synthesized MgCoSiO_4, before and after heating in comparison with literature values.

	This Study			Literature
	Mg2SiO4	MgCoSiO4		Mg2SiO4	Mg2SiO4	MgCoSiO4	MgCoSiO4	MgCoSiO4	Co2SiO4
	360 min	360 min	Milled-1000 °C	Matsui et al. 1400 °C [31]	Akimoto et al. 1700 °C [32]	Rinaldi et al. 1200 °C [33]	Miyake et al. 1500 °C [34]	Sommer et al. 1200 °C [35]	Morimoto et al. 1500 °C [17]
a (Å)	4.756 (6)	4.77 (3)	4.775 (2)	4.7553 (6)	4.756 (1)	4.77572 (8)	4.771 (1)	4.7713 (1)	4.782 (4)
b (Å)	10.224 (12)	10.26 (7)	10.267 (5)	10.1977 (14)	10.197 (1)	10.27159 (17)	10.245 (1)	10.2533 (2)	10.302 (4)
c (Å)	5.993 (7)	5.99 (4)	5.999 (3)	5.9820 (7)	5.982 (1)	6.00235 (10)	5.988 (1)	5.9911 (2)	6.003 (4)
V (Å3)	291.4 (6)	294 (3)	294.1 (2)	290.09 (11)	290.1	294.441 (6)		293.09 (3)
[Co_1]	0	0.606 (17)	0.637 (13)			0.698 (9)	0.675 (2)	0.666 (2)
[Co_2]	0	0.409 (11)	0.325 (11)			0.302 (9)	0.278 (21)	0.334
Si-O (Å)	1.639 (2)	1.642 (5)	1.6428 (1)				1.638		1.627
* M-O (Å)	2.115 (5)	2.122 (1)	2.1233 (1)				2.113		2.134
Rwp	6.525	2.219	2.906				2.131

Standard deviations in parentheses are in unit of the last digit *; M = Mg/Co, [Co_1] = Co atomic occupancy in M1 site.

). Further analysis of the occupancies on the M1 (edge-sharing tetragonally elongated octahedra) and M2 (corner-sharing trigonally elongated octahedra) metal sites suggests a strong preference for Co²⁺ over Mg on the M1 site in both annealed and unannealed samples. Annealing after milling further promoted the diffusion of Mg²⁺ into the M2 site, resulting in an increase in the Mg site occupancy factor from 0.591 to 0.675 on M2. Unit cell parameters of samples with and without annealing are comparable to literature values [17,31,32,33,34,35].

TEM images of the synthesized Mg₂SiO₄ product, obtained at different magnifications, are compared to synthetic MgCoSiO₄ in Figure 4. The particle size for MgCoSiO₄ ranged from 23 to 50 nm (average 27 nm), while a greater size variation between 8 to 67 nm (average 31 nm) was observed in Mg₂SiO₄. Many of the particles appeared to be present as clumps, signifying aggregation during milling. Attempts to break apart these clumps ultrasonically produced little change in the end product.

The milled samples (Figure 5) showed high degrees of spottiness in the diffraction rings, consistent with the polycrystalline nature of the as-synthesized Mg₂SiO₄ and MgCoSiO₄. The indexing of the SAED patterns revealed signature orthorhombic olivine crystal structures. The EDS analysis of selected particles of the 180 min milled Mg₂SiO₄ indicated the presence of (as expected) Mg, Si, and O, as well as trace amounts of Ca (from CaCO₃ in the MgO starting material), and Co as a carryover contaminant from previous milling trials (where a small amount of Co embedded itself in the WC grinding components) (Figure 6). Quantitative analysis of elemental mass percentages based on the EDS spectrum suggested an empirical formula of Mg₂SiO₄. EDS analysis of the MgCoSiO₄ sample, although within the margin of error of the measurement, indicated some degree of variation in the Mg:Co:Si proportions. For example, the Mg:Si ratio of forsterite was 2.03:1; the Mg:Co:Si ratios of MgCoSiO₄ covered the range of (0.61–0.93):(0.67–0.81):1. The amount of Si detected by XRD was significantly lower than that determined from the EDS analysis of the synthesized MgCoSiO₄. This could be due to selective (X-ray) absorption or to a residual amorphous Si-containing component that was not readily detected by XRD analysis [36].

4. Conclusions

This report presents a straightforward method for synthesizing nanocrystalline Mg₂SiO₄ and MgCoSiO₄. The ability to synthesize and finetune the chemical series of Mg²⁺ and Co²⁺ represents a unique contribution to olivine reactivity research. The synthesis technique was shown to be efficient at ambient temperatures, without contingent inert atmospheric control. Given the nanocrystalline nature of the products and the simple reaction activation by milling, research on a potential application, i.e., the battery performance testing of the mechanochemically synthesized Mg₂SiO₄ and MgCoSiO₄, is currently underway. The contamination of the synthesized olivine with WC, which in turn resulted in the undesirable formation of tungstate when annealing the milled product at an elevated temperature, remains an unresolved concern for further material development. Thus, a robust method for removing WC from mechanically synthesized olivine is necessary and will be addressed in a follow-up publication.