In situ crystallization and magnetic measurement of hexaferrite glass-ceramics

The M-type hexaferrites, MO·6Fe₂O₃ or MFe₁₂O₁₉ (M = Ba, Sr, or Pb) are important ferromagnetic oxides that have traditionally been used as permanent magnets in applications for dielectric media because of their high values of magneto-crystalline anisotropy and saturation magnetization.^1,2 Using the compositional and manufacturing flexibility of glasses, and allowing for crystallization of functional ceramic phases, magnetic glass-ceramics with hard phases like hexaferrites can be obtained.³ Glass-ceramics (GCs) with favorable magnetic properties are formed by controlled heat-treatment of glass at elevated temperatures designed to crystallize the desired magnetic phases. Magnetic GCs are currently of interest in cancer research due to the favorable combination of biocompatibility and the potential for non-invasive hyperthermia therapy.⁴ Potentially advantageous over nanoparticles, GCs retain crystallized magnetic phases while encapsulating them to help avoid oxidation. Most current research in magnetic glass ceramics focuses on spinel ferrite (LiFe₅O₈), magnetite, or hematite as the active magnetic phases in silicate glass,^5,6 though borate glasses have been proposed as sintering aids for bulk Sr hexaferrite magnets.⁷ 

The use of magnetic ceramics, with emphasis on the hard-magnetic hexaferrites, in industrial applications such as permanent magnets, recording media, and microwave absorbers is now well-established, with ongoing research aimed at optimizing coercivity.⁸ Pure M-type hexaferrites such as SrFe₁₂O₁₉ and BaFe₁₂O₁₉ have high values of magnetization (>60 emu/g) and coercivity (H_c > 5.0 kOe).⁹ They are usually produced through synthesis techniques such as conventional solid state reaction, coprecipitation, sol-gel method, auto combustion, hydrothermal, or molten salt.¹⁰ 

Efforts to crystallize Ba-M hexaferrite from borate glass began as early as 1970,¹¹ and have continued,¹² with emphasis on compositional tailoring of the resulting crystallized hexaferrite and of the glass matrix.^13–15 Sr-M hexaferrite glass-ceramics have also been investigated,^16–18 primarily in borate glasses, but also adding Si to take advantage of glass-glass phase separation,¹⁹ adding Na to modify the glass,^15,20,21 or adding Al, La, or Co to modify the hexaferrite crystalline phase properties.^17,22,23

For example, Kazin et al.²⁴ explored the glass crystallization method by rapidly quenching a homogenous melt composed of constituents making up hexaferrite (SrO, Fe₂O₃) and glass matrix (SrO, B₂O₃). Subsequent heat treatments of the glasses produced hexaferrite particles with plate-like morphologies whose dimensions resulted in very high coercivity values.²⁴ Similar studies have validated the close link between suitable heat treatment temperatures and the crystallization of the hexaferrite phase.^16,23 Wet chemistry coprecipitation methods also aid in homogenous mixing of starting chemicals (metal hydroxides and aqueous solution of metal salts).²⁵ 

In this paper, we consider a novel method of in situ crystallization of strontium hexaferrite (SrFe₁₂O₁₉) obtained while heating and measuring the magnetic behavior in the presence of an applied magnetic field. The resulting glass-ceramic is investigated by first order reversal curve (FORC) measurements to explore the origin of the shape of hysteresis loops. Phase distribution and microstructure are evaluated by X-ray diffraction (XRD) and electron probe microanalysis (EPMA). Several candidate compositions are explored ex situ using thermal analysis and different heat treatments combined with XRD. The magnetic properties were highly sensitive to atmosphere during heat treatment (air versus argon) as well as details of the heat treatment schedule.

All starting chemicals used were reagent-grade: strontium carbonate (Sigma-Aldrich 99.9%), iron (III) oxide (Strem chemicals 99.8%), and boric acid (Alfa Aesar 99.99%). Using standard glass melting techniques, batched components were mixed, placed in a Pt10%Rh crucible, melted in a resistance furnace (DelTech) for 1 h at 1200°C, then removed from the furnace and poured onto an Inconel plate. The SrO-Fe₂O₃-B₂O₃ glass-forming region was explored to determine the most suitable compositions for bulk glass formability (compositions shown in Table I). Compositions are named as XB-YSF, with X denoting the B₂O₃ mole% and Y denoting the SrO/Fe₂O₃ molar ratio. Compositions forming only an amorphous phase (according to XRD) were further explored as glass-ceramics.

All GCs were first prepared by placing cooled crushed glass in a furnace (Sentrotech) set to 800°C, holding for 2 h, then removing to ambient. This heat treatment is denoted ‘isothermal’ (IT). A subset of three compositions (* in Table I) was also heat treated by placing fresh glass in the furnace at room temperature, heating at 10°C/min to 800°C, dwelling for 2 h, then cooling at 10°C/min. This heat treatment is denoted ‘non-isothermal’ (NIT) and was designed to better mimic the heating protocol of in situ magnetic measurements. These three as-quenched compositions were also investigated by differential thermal analysis (DTA, TA Instruments SDT-Q600) to assess glass transition temperature (T_g), peak crystallization exotherm temperature (T_x), and melting point (T_m), by heating under nitrogen at 20°C/min from ambient to 1200°C in Pt crucibles.

Glasses and GCs were investigated with an X-ray diffractometer (XRD, PANalytical X’pert Pro) using Cu K_α radiation (λ = 1.5406 Å) at 45 kV and 40 mA, and scans performed from 5-90° 2θ with step size of 0.5° and 10 s per step. XRD patterns were generated from summation of five sequential scans. Quantitative Rietveld refinement was performed on three GCs (* in Table I), using 20 wt.% of CaF₂ as an internal standard to allow quantification of the residual amorphous phase. Each sample was measured and fit twice. Some additional experiments were performed to assess reproducibility and sensitivity to temperature and time. This data and discussion is provided in the supplementary material, section S.4.

All magnetic measurements, including temperature-dependent hysteresis loops and room temperature first order reversal curves (FORCs), were obtained on a vibrating sample magnetometer (VSM, PMC3900, Lakeshore Cryotronics), using a maximum applied field of 18 kOe. FORC diagrams were processed using FORCinel (v. 3.06, running on Igor Pro 7.08, Wavemetrics).²⁶ An oven accessory heats argon gas which blows over the sample, allowing temperatures up to 800°C to be achieved during VSM measurement. A small amount of crushed glass (≈0.05 g) was mixed with zirconia cement (50/50 wt.%) and fixed to the VSM sample rod. Each sample was heated at 8.3°C min^-1 while an applied field sweeps to ±18 kOe. After measurement, these in situ crystallized samples were measured by XRD using the ZrO₂ cement as the internal standard for quantification.

An electron microprobe (JEOL JXA-8500F) using a field emission gun and wavelength dispersive spectroscopy (WDS) was used for imaging and elemental quantification, including boron. In order to obtain a sample of similar phase distribution to those investigated by XRD and VSM yet enable sample preparation for EPMA, a chunk of glass (∼5 mm on a side) was heated with the NIT protocol to 800°C, held for 2 h, then removed and air quenched. Samples were mounted in epoxy, polished, and coated for EPMA measurements. Back scattered electron (BSE) images and WDS elemental maps were obtained. Quantitative mapping and spot analyses were performed on a field emission JEOL 8500F Hyperprobe, using WDS. An accelerating voltage of 10 kV and beam current of 10 nA was used for spot analyses. For quantitative mapping, the same voltage and current were used, with dwell times of 0.3 – 1.0 seconds per pixel. Backgrounds for both spot analyses and mapping were calculated using the Mean Atomic Number correction.²⁷ 

By varying B₂O₃ concentration of the glass system and Sr/Fe ratio we found that the most suitable compositions for hexaferrite had B/Sr molar ratio ≤ 2 and SrO ≥ 43 mol%. Compositions with Fe/Sr molar ratio ≳ 1.5 resulted in crystallization of α-Fe₂O₃ phase on quenching (see Table I). Of the ten (10) compositions explored in Table I, six (6) compositions were X-ray amorphous upon quench; these were further heat treated at 800°C to investigate crystalline phase formation (Table II). Previous work showed that an X-ray amorphous scan did not guarantee absence of a magnetic phase, as VSM is considerably more sensitive to small fractions (<1 wt.%) of magnetic phases than is XRD.²⁸ This was confirmed again in this study (see below). Nonetheless, all X-ray amorphous samples were heat-treated and crystalline phase production was observed.

By XRD, strontium borate (SrB₂O₄) was observed in all investigated samples after isothermal heat treatment. Hematite (α-Fe₂O₃) was observed in most samples and magnetite (Fe₃O₄) in select samples. GCs with the preferred strontium hexaferrite phase (SrFe₁₂O₁₉) were obtained from two compositions from the 800°C-2 h air heat treatment, namely 42B-2.9SF and 30B-1.8SF, as revealed by XRD. A typical XRD result is shown in Figure 1. These two compositions and one other, 40B-2SF, were selected for further study. Variations in time at 800°C (0.5 – 5 h) were performed for 30B-1.8SF, and one heat treatment at a higher temperature, 900°C-2 h, was performed for 40B-2SF (see supplementary material, section S.4), in order to quantify any phase fraction difference for slightly different times or temperatures.

DTA (scans in supplementary material, Fig. S.1.2) suggested that T_g (≈540-570°C) and T_m (≈950-1050°C) were lowest for high Sr/B (low B/Sr) which is indicative of the depolymerization of the glass and melt at high SrO concentrations. The main crystallization peak T_x lay in the range ≈650-710°C, similar to previous observations.¹⁶ The assignments of these crystallization peaks are ambiguous, but according to the literature may include SrB₂O₄, SrFe₁₂O₁₉, other Sr/Fe oxide phases, and unidentified phases.¹⁶ However, given that the major T_x occurs in all three samples, it is mostly likely due to SrB₂O₄ which forms in all three compositions. The 40B-2SF composition is similar to compositions investigated by Kazin et al.²⁴ and Zaitsev et al.²⁹ In both these studies, SrFe₁₂O₁₉ was found in the samples, whereas in our study this phase was not seen in XRD, except at 900°C-2 h (see supplementary material, Table S.4.2), but was observed in the magnetic behavior for air treated samples (see below). In the DTA for 40B-2SF, there is a small exothermic peak ∼624°C in addition to the larger one ∼673°C. Zaitsev et al.²⁹ assign the lower T peak (602°C in their study) to crystallization of SrB₂O₄ and the higher T peak (679°C in their study) to SrFe₁₂O₁₉. All three glasses studied here show a lower exothermic T peak to a greater or lesser extent, from almost absent (42B-2.9SF) to small but present (40B-2SF).

The three down-selected glasses were subjected to in situ heat treatments in the VSM. The field and temperature histories were the same for each sample, and shown schematically in supplementary material, Fig. S.2.1, but briefly: (1) hysteresis loops were measured as glasses were heated from room temperature to 800°C, then the sample was held at 800°C for ≈30 min at zero field then cooled and measured in zero field, (2) hysteresis loops were measured again on heating, this time to 650°C, then samples were quenched, and (3) a room temperature FORC pattern was obtained.

Hysteresis measurements on initial heating of as-quenched glasses showed paramagnetism with a small nonlinearity at zero field suggesting some weak ferromagnetic component (42B-2.9SF) or a signature for more (30B-1.8SF) or less (40B-2SF) magnetite (Figure 2). The identity of magnetite was confirmed as the hysteresis disappeared in loops between 550°C and 600°C (T_C of Fe₃O₄ 585°C). For samples which did form SrFe₁₂O₁₉ on in situ heating (see below), it was not observed on initial heating. This is likely due to the crystallization primarily taking place >650°C, which is above the T_C of SrFe₁₂O₁₉. Thus cooling <450°C after heating above the crystallization was necessary to see the magnetic behavior (see supplementary material, Fig. S.2.1 for field and temperature history schematic).

Room temperature VSM was obtained and compared for starting glasses, samples heat treated isothermally and non-isothermally in air (800°C-2 h), and those heat treated in situ in argon in the VSM (as described above). The results are shown in Table III and Figure 2. There are substantial differences between samples produced under different conditions. Notably, evidence of magnetic phases is seen in the VSM even when XRD does not show it (see Table III). This is particularly true of small fractions of SrFe₁₂O₁₉ which were not detectable by XRD (Table III) but which are evident in the VSM, particularly when comparing the different heat treatments. Clearly the compositions are sensitive to the heat treatment protocol (IT versus NIT) as well as gas environment (air versus argon).

The reducing effects of heating in argon gas, as opposed to air, is evidenced in the crystallization in argon of magnetite (FeO·Fe₂O₃, iron having oxidation states Fe²⁺ and Fe³⁺), compared to the typical preference for hematite (Fe₂O₃, with only Fe³⁺) phase in air heat treatments. When more iron is reduced to Fe²⁺ in argon, the Sr-hexaferrite phase (requiring Fe³⁺) is less likely to form, yet still does form in 42B-2.9SF and 30B-1.8SF, as evident from the VSM, though not from XRD. Both these compositions heated in air show significant signature of Sr-hexaferrite in VSM and are measurable by XRD and visible in microprobe (see below). The 40B-2SF composition is unusual in that it appears to form Sr-hexaferrite only with air non-isothermal heating but not air isothermal heating, according to VSM, though XRD showed no evidence of this phase for any sample.

For the in situ measurements in argon, the shorter hold at 800°C (0.5 h) compared to furnace heat treatment (2 h) was done to protect the magnetometer. On re-heating and measuring loops from 25°C to 650°C (e.g., supplementary material, Figure S.2.2), wasp-waistedness,^30–32 suggesting multiple magnetic phases of different character (two different phases or the same phase of two different domain sizes), was noted for 42B-2.9SF and 30B-1.8SF, but not 40B-2SF. The mass-normalized magnetization (emu/g, normalized from mass of starting glass) of the GC processed in situ showed saturation magnetization (M_s) at room temperature of ≈11.0 emu/g for 30B-1.8SF, ≈4.5 emu/g for 42B-2.9SF, and ≈5.6 emu/g for 40B-2SF (Figure 2). For air heated samples, wasp-waisted loops are also observed for 30B-1.8SF (IT, NIT), 40B-2SF (IT), and 42B (IT).

Temperature-dependent data showed (Figure 3), as expected, that as temperature increased, M_s, remanence (M_r), and coercivity (H_c) decreased. Remanence and coercivity reached ≈0 at 650°C. Several kinks are observed, ≈300°C (42B-2.9SF only), ≈500°C, and ≈550°C. The transition ≈300°C is not yet understood, but ≈500°C is close to the T_C for SrFe₁₂O₁₉ (464°C)³³ and 550°C is close to the T_C for magnetite Fe₃O₄ (585°C).³⁴ The temperatures measured by the thermocouple are of the argon gas near the sample and not directly on the sample, so exact agreement is not expected. Further, it is well-known that small particles have slightly lower T_C values than bulk.^33,34

First order reversal curves of the samples heat treated in situ (Figure 4) confirmed the interpretation of the wasp-waisted magnetic behavior as being due to two magnetic phases for 42B-2.9 SF and 30B-1.8SF which formed Sr hexaferrite, but not for 40B-2SF which showed only magnetite in the in situ sample. The low coercivity phase in all FORC diagrams with H_c ≈500 Oe is characteristic of magnetite.²⁸ The high coercivity phase with H_c ≈2500 Oe can be assigned to the SrFe₁₂O₁₉.³⁵ The signature of magnetite is dominant likely due to its higher volumetric magnetization compared to Sr-hexaferrite. The effect of some hematite (α-Fe₂O₃) cannot be completely ruled out, but contributions are expected ≈1000 Oe.³² It is well-known, however, that high-magnetization components like magnetite and hexaferrite can readily mute the magnetic signature of low-magnetization components like hematite when they coexist, even at low fractions.^32,35 More proof of the identified phases could be performed in the future using FORC measurements at temperatures where phases are expected to be above their Curie points.

Finally, an investigation of magnetic glass-ceramic microstructure was made. Compositions in the quantitative maps were used to verify phase identity and compare to XRD. The observed microstructures (Figure 5) of the three GCs produced by air NIT protocols are all very different. Fully quantitative EPMA WDS maps were obtained (see supplementary material, S.3), but microstructural features were too small for reliable quantification except 30B-1.8SF, where quantitative results (supplementary material, Table S.3.1) were used to make the phase assignments in the labeled backscatter image. Phase assignments in other micrographs were made from a combination of EPMA maps and XRD results. Sample 30B-1.8SF shows large needle-shaped crystals of Sr-hexaferrite with gray SrB₂O₄ crystals and bright Sr-rich residual borate glass. 42B-2.9SF shows small star-shaped aggregates of Sr-hexaferrite, similar to microstructures reported in Zaitsev et al.,²⁹ with SrB₂O₄ crystals and residual glass (which cannot be distinguished by composition), and tiny Fe_xO_y crystals. Finally, the 40B-2SF resembles a phase-separated structure, with SrB₂O₄ crystals, Sr-rich borate glass, and Fe₂O₃. The predicted immiscibility region of SrO-B₂O₃ is at SrO concentrations <20 mol% at 900°C,^36,37 which is not directly applicable to this ≈1:1 Sr:B ratio composition. However, if some Sr partitions to the hexaferrite phase, this will reduce the concentration of Sr in the residual glass, but still not likely enough to induce separation to a low- and high-borate glass. Overall, the microstructure consists of “cells” of this spinodal type structure decorated on the edges of the cells with high average atomic number (high Sr) phases.

It is instructive to investigate these compositions in the ternary phase diagram for SrO-Fe₂O₃-B₂O₃ and the presented liquidus surface of Sato.^38,39 These diagrams are shown in supplementary material, Fig. S.1.1. Composition 30B-1.8SF, the one forming the most hexaferrite, lies in the Sr₂B₂O₅-SrB₂O₄- SrFe₁₂O₁₉ Alkemade triangle, meaning these are the expected phases on full equilibrium crystallization. In this sample, magnetite usually formed in IT and irregularly in NIT air treatments according to XRD, and was evident in VSM of starting glasses and all heat treated samples (Table III). Additionally, no Sr₂B₂O₅ crystalline phase was observed, and this is approximately the composition of the residual glass (supplementary material, Table S.3.1), one which would have comparatively low viscosity and could have allowed for large crystal growth, as was observed during heat treatment. The other two compositions lie in the Fe₂O₃-SrB₂O₄-SrFe₁₂O₁₉ Alkemade triangle. Furthermore, when examining the liquidus surface,^38,39 the 30B-1.8SF sample lies in the SrFe₁₂O₁₉ primary phase field, while the 42B-2.9SF, which formed small crystals of SrFe₁₂O₁₉, lies in the SrB₂O₄ primary phase field, and formed much larger quantities of SrB₂O₄ than 30B-1.8SF. The 40B-2SF sample has a composition which lies very close to the line separating the Fe₂O₃-SrB₂O₄-SrFe₁₂O₁₉ triangle from the Fe₂O₃-SrB₂O₄-B₂O₃ triangle, and also lies almost at the top of a saddle point between the SrB₂O₄ and SrFe₁₂O₁₉ primary phase fields, where cooling could go towards SrO-rich or B₂O₃-rich points of the phase diagram, thus potentially explaining its instability and tendency for phase separation. 40B-2SF did not form XRD measurable SrFe₁₂O₁₉ on standard air or argon heat treatment but only at higher temperatures (900°C-2 h), though VSM of the IT in air heat treatment suggests hexaferrite. In both 42B-2.9SF and 40B-2SF, iron oxides reliably formed, as expected being in the Alkemade triangle containing Fe₂O₃, but the fractions of magnetite versus hematite varied, hence the large standard deviations in phase fraction, unlike the 30B-1.8SF composition, where Fe oxide is not expected to form in equilibrium per the phase diagram. Finally, the other compositions investigated in this study are found to be in the Fe₂O₃-SrB₂O₄-B₂O₃ triangle, and those crystallizing hematite on quench were in the Fe₂O₃ primary phase field, while those remaining glassy crystallized hematite and SrB₂O₄ on heat treatment. Note that by no means were any of the heat treatments in air or argon in equilibrium, so the phase diagrams can only be used as a guide for understanding the crystallization.

Obviously, the opportunities for microstructural tailoring in the SrO-Fe₂O₃-B₂O₃ is large, given the dramatically different structures and magnetic properties obtained in only three compositions. Additionally, there are many Sr-B-O crystalline phases of different stoichiometries,⁴⁰ stable phases in the SrO-Fe₂O₃ system⁴¹ and the B₂O₃-Fe₂O₃ system,⁴² as well as the SrO-Fe₂O₃ system⁴³ (i.e., SrFe₁₂O₁₉). Particular microstructures may be desired for certain magnetic applications, and GCs such as these could be optimized with heat treatment protocols and environment for a given need.

The traditional melt-quench technique followed by high temperature treatment in air was used to determine which compositions best suited the goal of forming glass upon quench and hexaferrite phase on heat treatment. Three promising candidates were selected from ten trials for further study. Crystallization of the SrFe₁₂O₁₉ phase from borate glass was attained in an in situ argon environment in a VSM. After heating and cooling, wasp-waisted loops were observed on subsequent heating, with overall H_c > 600 Oe. FORC measurements confirm multiple magnetic phases, magnetite plus Sr-M hexaferrite. Previous studies of Sr-hexaferrite glass ceramics do not observe the wasp-waisted features due to magnetite additions, which here are present due to the argon in situ heating environment. These features were further explored with FORC diagrams for the first time. Companion hysteresis loops on air treated samples show more preference for formation of Sr-hexaferrite. Microstructures obtained for bulk air treated samples show dramatic diversity. To an extent, the phase assemblages and microstructures have been explained from the phase diagram. Additionally, in situ VSM measurements show promise for understanding crystallization of hexaferrite phases in borate glasses, but further work is needed on both composition design and heat treatment schedule to optimize desired microstructure for high coercivity in bulk glass-ceramics.

The supplementary material is available online containing the following: schematic phase diagrams showing compositions from the current study; thermal analysis data; schematic temperature and field history for in situ measurements; example temperature dependent hysteresis loops; electron probe microanalysis maps and quantitative phase composition measurements; Rietveld fit results for multiple XRD measurements including altered heat treatment schedules; graphs showing Rietveld refinements of XRD data including residuals.

Electron microprobe was performed at Washington State University’s Peter Hooper GeoAnalytical Laboratory. The authors thank the anonymous reviewer for detailed comments and insight which greatly improved this manuscript.

The data that support the findings of this study are available from the corresponding author upon reasonable request and available within the article and its supplementary material.