Optimisation of the Flame Spheroidisation Process for the Rapid Manufacture of Fe3O4-Based Porous and Dense Microspheres

The rapid, single-stage, flame-spheroidisation process, as applied to varying Fe3O4:CaCO3 powder combinations, provides for the rapid production of a mixture of dense and porous ferromagnetic microspheres with homogeneous composition, high levels of interconnected porosity and microsphere size control. This study describes the production of dense (35–80 µm) and highly porous (125–180 µm) Ca2Fe2O5 ferromagnetic microspheres. Correlated backscattered electron imaging and mineral liberation analysis investigations provide insight into the microsphere formation mechanisms, as a function of Fe3O4/porogen mass ratios and gas flow settings. Optimised conditions for the processing of highly homogeneous Ca2Fe2O5 porous and dense microspheres are identified. Induction heating studies of the materials produced delivered a controlled temperature increase to 43.7 °C, indicating that these flame-spheroidised Ca2Fe2O5 ferromagnetic microspheres could be highly promising candidates for magnetic induced hyperthermia and other biomedical applications.


Introduction
From the magnetic materials available, Fe 3 O 4 -based superparamagnetic nanoparticles (SMNPs) have been most extensively investigated for localised magnetic hyperthermia applications due to their superparamagnetic expression and non-toxicity [1,2], and because iron oxide metabolism is readily achieved by the heme oxygenase-1 gene which generates haemoglobin and promotes cellular iron homeostasis [3]. Related ferrites, such as NiFe 2 O 4 [4], MnFe 2 O 4 [5], CoFe 2 O 4 [6] and Li x Fe 3-x O 4 [7], have also been investigated to improve magnetic strength and thermal stability. However, the inherent toxicity of Ni, Mn, Co and Li limits their application [3]. An alternative approach is to introduce non-magnetic Ca 2+ into the ferrite crystalline structure, to generate significant improvements in terms of biocompatibility, whilst maintaining magnetic expression and heating control [8,9]. Non-toxic calcium ferrites have been shown to metabolise safely within the body [10,11], making them appropriate for a range of biomedical applications, including magnetic hyperthermia. Biomedical investigations using calcium ferrites have been reported in relation to drug-delivery systems [8,9,12,13] and cytocompatibility [8,10,12,14]. Approaches combining metal cations and calcium ferrites for therapeutic applications have also been explored [9,11,[15][16][17][18]. The present challenge is to develop these materials into practical morphologies for enhanced biomedical investigations.

Mass Ratio
As outlined in  As outlined in Table S1 (see Supplementary Materials), a mass ratio of 1:1 produced strong signatures for Fe2O3, Ca2Fe2O5 and (unreacted) CaCO3, along with a medium signature for (unreacted) Fe3O4. A mass ratio of 3:1 (Fe3O4-rich) revealed a dominant Fe2O3 signature, with medium Fe3O4 and Ca2Fe2O5 signatures, and a weak signature for CaCO3. A mass ratio of 1:3 (CaCO3-rich) showed strong signatures for Ca2Fe2O5 and CaO, medium signatures for Fe2O3 and CaCO3, and a weak signature for Fe3O4. In particular, the progression towards higher porogen content (mass ratio from 3:1 to 1:3) was associated with a reduction in magnetite (Fe3O4) and hematite (Fe2O3) peak intensities and a consolidation of intensities attributable to srebrodolskite (Ca2Fe2O5), CaCO3 and CaO (reacted porogen).

Microsphere Magnetic Properties (Sieved)
Magnetisation measurements provided information on the magnetic properties of flame spheroidised microspheres and clarified the effect of Fe3O4:CaCO3 mass ratio on magnetic expression. For all samples, the magnetisation curves were indicative of typical ferrimagnetic behaviour. Figure 3 presents magnetisation curves for the sieved flamespheroidised products (mass ratios 3:1, 1:1 and 1:3; gas flow setting 2.5:2.5). As summarised in Table 2, progression towards increased porogen (mass ratio from 3:1 to 1:3) was accompanied by a decrease in magnetisation, consistent with a lowering of iron content.

Microsphere Magnetic Properties (Sieved)
Magnetisation measurements provided information on the magnetic properties of flame spheroidised microspheres and clarified the effect of Fe 3 O 4 :CaCO 3 mass ratio on magnetic expression. For all samples, the magnetisation curves were indicative of typical ferrimagnetic behaviour. Figure 3 presents magnetisation curves for the sieved flamespheroidised products (mass ratios 3:1, 1:1 and 1:3; gas flow setting 2.5:2.5). As summarised in Table 2, progression towards increased porogen (mass ratio from 3:1 to 1:3) was accompanied by a decrease in magnetisation, consistent with a lowering of iron content.      Figure 4a-c presents BSE images and MLA mapping analyses, revealing fine-scale morphological details and clarifying the outcomes of compositional dependencies on the precursor-porogen mass ratio of the sieved microsphere products, extracted from the sample sets presented in Figure 1.   Figure 4a-c presents BSE images and MLA mapping analyses, revealing fine-scale morphological details and clarifying the outcomes of compositional dependencies on the precursor-porogen mass ratio of the sieved microsphere products, extracted from the sample sets presented in Figure 1.  Varying levels of internal porosity were observed for all sample sets. For the case of flame-spheroidised Fe3O4:CaCO3 with mass ratio 1:1/gas flow setting 2.5:2.5 (Figure 4b), upon sectioning, direct evidence was provided for the development of low and high levels of internal porosity within dense microspheres and larger irregular-shaped particles, respectively. For the case of flame-spheroidised processed precursor-rich powder (mass ratio 3:1; gas flow setting 2.5:2.5; Figure 4a), the evidence showed the development of comparatively lower levels of internal porosity within dense microspheres and few irregularshaped particles. Conversely, for the case of flame-spheroidised processed porogen-rich powder (mass ratio 1:3; gas flow setting 2.5:2.5; Figure 4c), a variety of dense and irregularshaped developed morphologies with significantly higher levels of internal porosity was Varying levels of internal porosity were observed for all sample sets. For the case of flame-spheroidised Fe 3 O 4 :CaCO 3 with mass ratio 1:1/gas flow setting 2.5:2.5 (Figure 4b), upon sectioning, direct evidence was provided for the development of low and high levels of internal porosity within dense microspheres and larger irregular-shaped particles, respec-tively. For the case of flame-spheroidised processed precursor-rich powder (mass ratio 3:1; gas flow setting 2.5:2.5; Figure 4a), the evidence showed the development of comparatively lower levels of internal porosity within dense microspheres and few irregular-shaped particles. Conversely, for the case of flame-spheroidised processed porogen-rich powder (mass ratio 1:3; gas flow setting 2.5:2.5; Figure 4c), a variety of dense and irregular-shaped developed morphologies with significantly higher levels of internal porosity was revealed.
In particular, and as highlighted in Table 3, compositional differences were evident across the mass ratio sample set, with a strong trend towards the development of banded calcium iron oxide (CFO) compositions, the most prevalent of which being srebrodolskite (Ca 2 Fe 2 O 5 , denoted CFO-3-Ca 2 Fe 2 O 5 ; Supplementary Materials, Table S2 and Figure S1). As anticipated, a progressive decrease in precursor content or increment in porogen content (mass ratio 3:1 to 1:3) was directly accompanied by a lowering of Fe levels and elevation of Ca levels throughout the microsphere products. Table 3. Mineral proportion (wt%) of flame spheroidised products, as a function of precursor to porogen mass ratio.   Table S2). Both of these samples (Figure 4a Table S2), along with larger irregular-shaped particles comprising unreacted porogen.
Complementary, energy-dispersive X-ray spectroscopy (EDS) mappings validated elemental compositions and fine-scale details for the microsphere products, as a function of mass ratios of 3:1, 1:1 and 1:3 (Supplementary Materials, Figure S5 and Table S3).  Table S3).  As summarised in Table 4, a lower gas flow setting of 2:2 (mass ratio 1:1) resulted in a consistent yield of highly porous microspheres, along with a small dense microsphere and very few irregular-shaped particles ( Figure 5a). Comparatively, an increased gas flow setting of 3:3 (mass ratio 1:1) resulted in dense microspheres and few irregular-shaped particles ( Figure 5b). Notably, only the gas flow setting 2:2 (mass ratio 1:1) processing conditions resulted in the production of microspheres with visible evidence for porosity (pore size range 1.8-64.5 µm; mean pore size 13.1 µm, SD 12.6 µm; ImageJ software). Table 4. Size range of flame spheroidised reaction products (dense microspheres, irregular-shaped particles (ISP), and microspheres with surface porosity), as a function of gas flow setting.  As summarised in Table 4, a lower gas flow setting of 2:2 (mass ratio 1:1) resulted in a consistent yield of highly porous microspheres, along with a small dense microsphere and very few irregular-shaped particles ( Figure 5a). Comparatively, an increased gas flow setting of 3:3 (mass ratio 1:1) resulted in dense microspheres and few irregular-shaped particles ( Figure 5b). Notably, only the gas flow setting 2:2 (mass ratio 1:1) processing conditions resulted in the production of microspheres with visible evidence for porosity (pore size range 1.8-64.5 µm; mean pore size 13.1 µm, SD 12.6 µm; ImageJ software). Table 4. Size range of flame spheroidised reaction products (dense microspheres, irregular-shaped particles (ISP), and microspheres with surface porosity), as a function of gas flow setting.  Figure 6a,b show BSE images and MLA mineral mapping analyses, extracted from the sample sets presented in Figure 5, presenting fine-scale morphological details of the sieved microsphere products and clarifying the compositional dependencies on gas flow setting. Interestingly, significant levels of internal porosity were revealed for both sample sets. For the case of the higher gas flow setting 3:3, moderate levels of internal porosity were associated with the microspheres and a few irregular-shaped products (Figure 6b). Whereas, for gas flow setting 2:2, a variety of developed porosities was evident, including high levels of interconnected porosity for the case of larger microspheres (125-180 µm) (Figure 6a). sieved microsphere products and clarifying the compositional dependencies on gas flow setting. Interestingly, significant levels of internal porosity were revealed for both sample sets. For the case of the higher gas flow setting 3:3, moderate levels of internal porosity were associated with the microspheres and a few irregular-shaped products (Figure 6b). Whereas, for gas flow setting 2:2, a variety of developed porosities was evident, including high levels of interconnected porosity for the case of larger microspheres (125-180 µm) (Figure 6a).   Table S2. GFS: gas flow setting.

Mass Ratio
As highlighted in Table 5, high levels of sample homogeneity were maintained, as a function of the gas flow setting, with CFO-3 srebrodolskite (Ca 2 Fe 2 O 5 ) as the dominant phase for all mass ratio 1:1 sample sets. For the case of gas flow setting 3:3, the Ca 2 Fe 2 O 5 proportion decreased slightly compared to gas flow setting 2.5:2.5 (Tables 3 and 5), whilst a small amount of CFO-2 and CFO-4 was evident (Figure 6b; Supplementary Materials, Figure S6). Notably, for the case of gas flow setting 2:2, the highest levels of Ca 2 Fe 2 O 5 homogeneity (99.6 wt%) were returned (Figure 6a; Supplementary Materials, Figure S7). It was noted that the flame spheroidised Fe 3 O 4 :CaCO 3 (1:1 mass ratio; 2:2 gas flow setting) samples revealed the highest levels of compositional uniformity and good levels of interconnected porosity. Further magnetic characterisation and induction heating investigations of these materials were performed.   Table 6, the incorporation of porogen into the mixture led to a decrease of magnetisation saturation values. Nevertheless, it was noted that Fe 3 O 4 :CaCO 3 flame-spheroidised products still showed significant magnetic saturation values (8.9 Am 2 /kg). Table 5, high levels of sample homogeneity were maintained, as a function of the gas flow setting, with CFO-3 srebrodolskite (Ca2Fe2O5) as the dominant phase for all mass ratio 1:1 sample sets. For the case of gas flow setting 3:3, the Ca2Fe2O5 proportion decreased slightly compared to gas flow setting 2.5:2.5 (Tables 3 and 5), whilst a small amount of CFO-2 and CFO-4 was evident (Figure 6b; Supplementary Materials, Figure S6). Notably, for the case of gas flow setting 2:2, the highest levels of Ca2Fe2O5 homogeneity (99.6 wt%) were returned (Figure 6a; Supplementary Materials, Figure S7).   Table 6, the incorporation of porogen into the mixture led to a decrease of magnetisation saturation values. Nevertheless, it was noted that Fe3O4:CaCO3 flame-spheroidised products still showed significant magnetic saturation values (8.9 Am 2 /kg).

. Induction Heating Studies
The potential of Ca2Fe2O5 microspheres for magnetic-mediated hyperthermia was evaluated via induction heating (Table 7). Figure 8 shows the evolution of temperature for the most homogeneous flame-spheroidised Ca2Fe2O5 sample (mass ratio 1:1, gas flow setting 2:2), along with Fe3O4 and CaCO3 starting powders by way of control. The Fe3O4 powder showed high levels of induction heating, up to ~130 °C, but with an evident lack of heating control, whilst CaCO3 powder showed no induction heating as anticipated. Notably, homogeneous Ca2Fe2O5 microspheres exhibited highly controlled heating to a constant level of 43.7 °C which remained stable upon voltage decrease (150 to 35 V).  Table S4).

Discussion
This study reports on the optimisation of the flame-spheroidisation process parameters for the controllable production of magnetite-based porous microspheres (Fe 3 O 4 precursor powders/CaCO 3 porogen mass ratio, and gas flow setting). Modification of these parameters produced a variety of products in terms of shape (dense and porous microspheres, and irregular-shaped particles), or a mixture of these with distinct magnetic saturation levels and compositions (Fe-Ca excess/deficit). Optimised parameter conditions were identified for the manufacture of compositionally uniform and porous products, with Fe 3 O 4 :CaCO 3 (mass ratio 1:1; gas flow setting 2:2) mixtures producing Ca 2 Fe 2 O 5 microspheres with strong levels of compositional homogeneity and porosity levels (for the case of large porous microspheres). The homogeneous Ca 2 Fe 2 O 5 samples demonstrated controlled delivery of heat (43.7 • C, see Figure 8), highlighting the suitability of these candidate products for magnetic hyperthermia applications. Figure 9 provides a schematic illustration detailing the development of magnetic microspheres, as a function of mass ratio and gas flow setting parameters. Mass ratio parameters as applied to magnetite/porogen combinations revealed a direct effect on microsphere composition and magnetic properties. Importantly, Ca 2 Fe 2 O 5 (srebrodolskite) was the only calcium iron oxide phase revealed for all Fe 3 O 4 :CaCO 3 flame-spheroidised samples ( Figure 2). The suggestion is that rapid cooling and solidification mechanisms associated with the flame-spheroidisation process allowed for the formation of Ca 2 Fe 2 O 5 microsphere structures with modified compositions (excess/deficit of Fe/Ca atoms), i.e., the microsphere products retained structural integrity for all mass ratio cases but presented Fe/Ca variations according to elemental availability. This could also be attributed to the unusual capacity of Ca 2 Fe 2 O 5 to support a number of defects [44,45]. Indeed, this phenomenon was reinforced by compositional analyses of sieved samples. MLA mappings (Figure 4) highlighted a clear trend towards iron deficit/calcium excess, as the mass ratio progressed from 3:1 towards 1:3, with a mass ratio of 1:1 showing the highest levels of homogeneity (CFO-3, denoted as Ca 2 Fe 2 O 5 ). This compositional trend emphasised the importance of maximum consumption of the starting materials occurring for a mass ratio of 1:1 and gas flow setting of 2.5:2.5; in which case MLA data showed no evidence for any unreacted Fe 3 O 4 and CaCO 3 from this sample. It should be noted that sieving acted simply to improve sample homogeneity by removing excess, small, unreacted Fe 3 O 4 and CaCO 3 , and Fe 2 O 3 reacted powders. In contrast, Fe-rich samples (mass ratio 3:1) showed poor compositional uniformity, with two banded calcium iron oxide minerals observed (CFO-1 and CFO-2). Similarly, Ca-rich samples (mass ratio 1:3) were associated with low porogen consumption, along with irregular-shaped particles containing an excess of unreacted CaCO 3 and reacted CaO. In addition, the mass ratio parameter also strongly influenced the microsphere magnetic properties. An increase in porogen content (from Fe 3 O 4 :CaCO 3 3:1 to 1:3) was directly related to a decrease in magnetic saturation and remanent magnetisation. This was attributed to the incorporation of paramagnetic calcium atoms within the srebrodolskite structure (Ca 2 Fe 2 O 5 ), as a function of mass ratio. In this context, a report [46] is noted on the incorporation of Gd 3+ within nanocrystalline iron oxide particles produced by an extraction pyrolytic technique, with hysteresis loops measured via vibrating sample magnetometry (VSM; magnetic field max. 795.7 kA/m) as a function of concentration (mol%) and temperature (at much slower heating rates compared to the rapid flame spheroidisation process). An increase in Gd 3+ content (from 12.5 to 75 mol%) was related directly to a decrease in magnetic saturation values and remanent magnetisation, similar to the case of CaCO 3 . Notably, the flame spheroidisation process leads to the formation of metastable products with modified ferromagnetic properties. The parameter of the gas flow setting also influenced the development of porosity within the microspheres. Two types of porosity were identified for this sample set: i.e., internal pores (either interconnected or not) and surface pores. It is considered that surface pores formed via molten droplets trapping and releasing CO2 gas bubbles produced during porogen decomposition (CaCO3 → CaO + CO2). In contrast, internal pores were created within molten drops as a consequence of unreleased CO2 gas bubbles, during rapid solidification. Indeed, it is noted that an increment in porogen concentration combined with an elevated gas flow setting, i.e., 3:3, was associated with higher internal porosity levels. Conversely, surface pores (with interconnected porosity) were more strongly associated with increased porogen content, albeit with a 2:2 gas flow setting. This effect was attributed to the increased residence time of molten droplets within the oxy-acetylene flame, as a determining factor for the development of microsphere porosity. Considering that particle temperature is directly related to the residence time of the particle within the flame [47,48], a gas flow setting of 2:2 would facilitate CO2 trapping and release, and maximise the number of reacted precursor/porogen powders, thereby producing fewer irregular-shaped particles. Furthermore, flame length could be controlled by adjusting the gas flow ratio [49]. As illustrated in Figure 9, the flame length decreased with increasing gas flow settings (from 2:2 to 3:3), consequently influencing particle residence time within the flame and cooling rate. In addition, polyvinyl alcohol (PVA) promoted the binding of Fe3O4 precursors with CaCO3 porogen particles, by helping to hold the agglomerated masses together. Accordingly, it is suggested that porous microspheres were produced from the agglomeration of Fe3O4:CaCO3 particles, with rapid melting and coalesce leading to the production of melt pools rendered spherical by surface tension, in advance of rapid solidification and phase separation, as appropriate.
Induction heating measurements demonstrated the capability of Ca2Fe2O5 microspheres (mass ratio 1:1; gas flow setting 2:2) to deliver heat in a controllable way, addressing one of the main limitations of magnetic hyperthermia which is controlling the temperature increase to between 40-45 °C [50]. It is suggested that the mechanism of heat generation used in our study was hysteresis loss, as revealed by magnetisation curves showing remanence (Figure 7; Inset figure). This hysteresis loss mechanism is associated with multi-domain, ferro-and ferrimagnetic materials [28,34], different from Néel and Brownian relaxation, responsible for heat generation within single-domain, superparamagnetic nanoparticles (SMNPs). Importantly, the induction heating parameters used for Ca2Fe2O5 The parameter of the gas flow setting also influenced the development of porosity within the microspheres. Two types of porosity were identified for this sample set: i.e., internal pores (either interconnected or not) and surface pores. It is considered that surface pores formed via molten droplets trapping and releasing CO 2 gas bubbles produced during porogen decomposition (CaCO 3 → CaO + CO 2 ). In contrast, internal pores were created within molten drops as a consequence of unreleased CO 2 gas bubbles, during rapid solidification. Indeed, it is noted that an increment in porogen concentration combined with an elevated gas flow setting, i.e., 3:3, was associated with higher internal porosity levels. Conversely, surface pores (with interconnected porosity) were more strongly associated with increased porogen content, albeit with a 2:2 gas flow setting. This effect was attributed to the increased residence time of molten droplets within the oxy-acetylene flame, as a determining factor for the development of microsphere porosity. Considering that particle temperature is directly related to the residence time of the particle within the flame [47,48], a gas flow setting of 2:2 would facilitate CO 2 trapping and release, and maximise the number of reacted precursor/porogen powders, thereby producing fewer irregular-shaped particles. Furthermore, flame length could be controlled by adjusting the gas flow ratio [49]. As illustrated in Figure 9, the flame length decreased with increasing gas flow settings (from 2:2 to 3:3), consequently influencing particle residence time within the flame and cooling rate. In addition, polyvinyl alcohol (PVA) promoted the binding of Fe 3 O 4 precursors with CaCO 3 porogen particles, by helping to hold the agglomerated masses together. Accordingly, it is suggested that porous microspheres were produced from the agglomeration of Fe 3 O 4 :CaCO 3 particles, with rapid melting and coalesce leading to the production of melt pools rendered spherical by surface tension, in advance of rapid solidification and phase separation, as appropriate.
Induction heating measurements demonstrated the capability of Ca 2 Fe 2 O 5 microspheres (mass ratio 1:1; gas flow setting 2:2) to deliver heat in a controllable way, addressing one of the main limitations of magnetic hyperthermia which is controlling the temperature increase to between 40-45 • C [50]. It is suggested that the mechanism of heat generation used in our study was hysteresis loss, as revealed by magnetisation curves showing remanence (Figure 7; Inset figure). This hysteresis loss mechanism is associated with multi-domain, ferro-and ferrimagnetic materials [28,34], different from Néel and Brownian relaxation, responsible for heat generation within single-domain, superparamagnetic nanoparticles (SMNPs). Importantly, the induction heating parameters used for Ca 2 Fe 2 O 5 measurements were similar to that previously reported for ferromagnetic glass-ceramic microspheres (denoted P40-Fe 3 O 4 microspheres) [51] with the distinction that induced magnetic fields were higher for P40-Fe 3 O 4 products. Regarding the different magnetic saturation levels between Ca 2 Fe 2 O 5 (8.9 Am 2 /kg) and P40-Fe 3 O 4 (4 Am 2 /kg), there was a requirement to adjust the field in order to reach the target temperature (via induction coil heating). Moreover, magnetic hyperthermia effects may be achieved through the application of weak magnetic fields (<7.95 kA/m) [3]; hence, relatively low magnetic fields were used for these Ca 2 Fe 2 O 5 microsphere induction heating studies. Furthermore, the field frequency (204 kHz) used was within the clinically accepted range for magnetic hyperthermia [3,30,[50][51][52][53][54][55]. Additionally, in the present induction coil experiments the target temperature was achieved rapidly (~40 s), indicating that these microsphere products are promising candidates for reduced periods of magnetic hyperthermia exposure, thereby preventing and reducing patient discomfort [3]. These induction heating measurements showed promising results for the Ca 2 Fe 2 O 5 magnetic microspheres developed. However, for formal validation, this part of the investigation requires further study using alternating magnetic fields, similar to those found in clinical settings.
The formation mechanisms associated with Ca 2 Fe 2 O 5 porous and dense microspheres developed via the flame spheroidisation process have been established [42]. For these magnetic microspheres, Fe 3 O 4 :CaCO 3 particles were fed into a high-temperature oxy-acetylene flame (~3100 • C) where rapid melting and coalescence occurred. The molten particles acquired a spherical shape post exiting the flame due to surface tension. The development of compositionally uniform, porous and dense Ca 2 Fe 2 O 5 microspheres, upon rapid cooling and solidification, was consistent with EDS data (Supplementary Materials, Figure S5 and Table S3) and CaO:Fe 2 O 3 (2:1 molar ratio) of the Ca-Fe-O phase diagram [56]. Additionally, fine scale diffraction patterns (SAED data, Supplementary Materials, Figure S8) acquired from Ca 2 Fe 2 O 5 microsphere fragments, confirmed their polycrystalline structure.
Accordingly, to produce porous microspheres with interconnected porosity and high values of compositional homogeneity, the results suggest that optimised flame spheroidisation process conditions should be at the gas flow setting of 2:2, using a magnetite-to-porogen mass ratio of 1:1. It was noted that the formation of compositionally uniform, Ca 2 Fe 2 O 5 porous and dense microsphere products can be achieved by controlling the mass ratio and gas flow setting parameters from the flame spheroidisation process. Moreover, these Ca 2 Fe 2 O 5 microspheres show potential for magnetic hyperthermia applications due to their ability to deliver heat in a controllable way. Additionally, the elevated temperatures of magnetic hyperthermia may improve synergistically the release of certain chemotherapeutic agents, such as cisplatin, cyclophosphamide and bleomycin [57][58][59]. Hence, the compositionally uniform Ca 2 Fe 2 O 5 microspheres could be explored for drug delivery applications in combination with magnetic hyperthermia.

Materials
The starting feedstock comprised mixtures of as-supplied iron(II,III) oxide powder (Fe 3 O 4 ; ≤5 µm, 95%; Merck, Gillingham, UK) and calcium carbonate as porogen (CaCO 3 , ≤5 µm, 98%; Fisher Scientific UK Ltd, Loughborough, UK.). The magnetite (Fe 3 O 4 ) powder was mixed with CaCO 3 using a pestle and mortar and combined with droplets of 2% aqueous solution polyvinyl alcohol (PVA; Merck, UK) to act as a binder, followed by drying at 37 • C for 24 h. The processed materials were collected using glass trays, placed a short distance away from the thermal spray gun, and stored in glass vials for characterisation. Parameters to evaluate the effects of Fe3O4 precursor to CaCO3 porogen mass ratios (3:1, 1:1 or 1:3) and oxy-acetylene gas flow settings (O2/C2H2; 2:2, 2.5:2.5 or 3:3) on the resultant products are summarised in Tables 8 and 9, respectively.  Imaging of the as-acquired flame spheroidised products (unsieved) was performed by scanning electron microscopy (SEM; FEI XL30; 5 kV; spot size 2.5; 13.3 mm working distance, secondary electron (SE) imaging mode). Microsphere size distributions were established using ImageJ 1.51h software (National Institutes of Health, Bethesda, USA, NA).

Magnetic Characterisation
The microsphere products were then sieved (using a stainless-steel frame; 203 × 50 mm; ≥32 µm mesh; VWR International) to filter out surplus starting material. Complementary magnetisation measurements were performed using a superconducting The processed materials were collected using glass trays, placed a short distance away from the thermal spray gun, and stored in glass vials for characterisation. Parameters to evaluate the effects of Fe 3 O 4 precursor to CaCO 3 porogen mass ratios (3:1, 1:1 or 1:3) and oxy-acetylene gas flow settings (O 2 /C 2 H 2 ; 2:2, 2.5:2.5 or 3:3) on the resultant products are summarised in Tables 8 and 9, respectively.  Imaging of the as-acquired flame spheroidised products (unsieved) was performed by scanning electron microscopy (SEM; FEI XL30; 5 kV; spot size 2.5; 13.3 mm working distance, secondary electron (SE) imaging mode). Microsphere size distributions were established using ImageJ 1.51h software (National Institutes of Health, Bethesda, MD, USA).

Magnetic Characterisation
The microsphere products were then sieved (using a stainless-steel frame; 203 × 50 mm; ≥32 µm mesh; VWR International) to filter out surplus starting material. Complementary magnetisation measurements were performed using a superconducting quantum interference device magnetometer (SQUID; Quantum Design MPMS-3 system; VSM mode; vibration amplitude 1.5 mm; 26.9 • C).

Compositional Characterisation
The sieved microspheres were embedded in a cold epoxy resin and sectioned by sequential mechanical grinding (using 400, 800 and 1200 SiC grit papers) and polishing (6 and 1 µm diamond paste). The polished samples were then cleaned using deionised water and industrial methylated spirit (IMS) and dried before carbon coating. Backscattered electron (BSE) imaging and chemical analyses of sieved and sectioned microspheres were performed via SEM-based mineral liberation analysis (MLA), using an FEI Quanta600 MLA (20 kV; spot size 7) equipped with energy dispersive X-ray spectroscopy (EDS) for compositional analysis and associated Bruker/JKTech/FEI data acquisition software for automated mineralogy.

High Frequency Induction Heating
Induction heating studies of sieved microspheres were performed via high frequency induction (Cheltenham Induction Heating Ltd.; 35-150 V; 20-120 W; 0.6-0.8 A; 204 kHz). Glass vials containing the magnetic microspheres were placed at the centre of a water-cooled copper coil generating an alternating magnetic field, whilst the temperature was measured using a fibre optic temperature sensor (Neoptix Reflex Signal Conditioner). Control samples of starting Fe 3 O 4 and CaCO 3 powders were also investigated. All measurements were repeated three times.

Conclusions
Compositionally uniform, ferromagnetic Ca 2 Fe 2 O 5 porous and dense microspheres have been developed via the rapid, single-stage, flame spheroidisation process using feedstock powder Fe 3 O 4 :CaCO 3 combinations. Morphological, structural and compositional investigations provided evidence of the effect of Fe 3 O 4 :CaCO 3 mass ratio, and O 2 /C 2 H 2 gas flow setting parameters. Complementary SQUID magnetometry confirmed the ferromagnetic properties of flame-spheroidised products. The potential use of Ca 2 Fe 2 O 5 microspheres (1:1 mass ratio/2:2 gas flow setting) for magnetic hyperthermia applications with a simple, but significant, induction heating measurement (43.7 • C) was shown. The combination of compositional control, high levels of porosity and functional properties (i.e., magnetic and thermal) achieved opens up new opportunities, to explore the application of magnetic microspheres for a range of biomedical challenges.
Full MLA compositional analysis and modal minerology of flame spheroidisation Fe 3 O 4 :CaCO 3 (mass ratio 1:1; gas flow setting 3:3), following sieving and sectioning, demonstrating high levels of CFO-3-Ca 2 Fe 2 O 5 ; Figure S7. Full MLA compositional analysis and modal minerology of flame spheroidised Fe 3 O 4 :CaCO 3 (mass ratio 1:1; gas flow setting 2:2), following sieving and sectioning, demonstrating high levels of CFO-3-Ca 2 Fe 2 O 5 ; Table S4. Statistical analysis on induction heating experiments; Figure S8. Cumulative selected area electron diffraction (SAED) pattern for 120 stacked tilted fields of view, corresponding to compositionally uniform, flame spheroidised Fe 3 O 4 :CaCO 3 (mass ratio1:1; gas flow setting 2:2), i.e. Ca 2 Fe 2 O 5 microsphere fragments following grinding; Table  comprising  Funding: National Council of Science and Technology (Consejo Nacional de Ciencia y Tecnología) (CONACyT) and the Faculty of Engineering, University of Nottingham funded this work. The SQUID characterisation work-package, performed at the Royce Discovery Centre, University of Sheffield was funded by the Henry Royce Institute.

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Data Availability Statement:
The raw data that support the findings of this investigation is available from the corresponding author upon reasonable request.