New Map Reveals How Cobalt Alloys Freeze Into Next-Generation Magnets

In the world of advanced materials, some of the most transformative technologies hide in the most unassuming places. Metallic glasses, the amorphous metallic alloys that lack any crystalline order, have quietly powered current transformers, magnetic sensors, and high-precision cores for decades. Among them, cobalt-rich alloys containing silicon and boron stand out for their exceptional magnetic properties and remarkable resistance to corrosion. Yet the scientists who design these materials have long worked with an incomplete map. While the iron-silicon-boron and nickel-silicon-boron systems have been thoroughly charted, the cobalt-rich corner of the cobalt-silicon-boron ternary system remained a thermodynamic blank spot, forcing alloy designers to rely on extrapolation and trial-and-error rather than fundamental data.

A new study published in the Journal of Materials Science by Ester Kelly Starick Pellegrino and colleagues at the University of São Paulo, together with collaborators at the Federal University of Itajubá and the Federal University of Rio de Janeiro, has now filled that gap. The team constructed a liquidus projection for the cobalt-rich region of the Co-Si-B system, a two-dimensional map that shows which crystalline phases are the first to solidify from the molten alloy across the entire compositional space. This map is far more than an academic exercise. When engineers want to manufacture an amorphous metallic ribbon, they need to know exactly which competing crystalline phases will try to form as the melt cools, because those crystals are the enemies of glass formation and can destroy the delicate amorphous structure that gives the alloy its superior properties.

The logic behind these alloys is elegantly chemical. Amorphous metallic alloys typically contain 70 to 85 atomic percent transition metals, with the remaining 15 to 30 percent made up of metalloids such as silicon and boron. Boron, with its small atomic radius, promotes a denser and more complex liquid structure that hinders the atomic rearrangement needed for boride crystals to nucleate. Silicon works synergistically by increasing the viscosity of the melt and stabilizing the amorphous network against premature devitrification. Together, these metalloids allow molten metal to be rapidly quenched into a homogeneous glassy ribbon rather than a polycrystalline solid. But the glass is metastable, and upon heating it will always try to crystallize into the equilibrium phases. Knowing precisely what those phases are, and at what compositions they dominate, is therefore the foundation of rational alloy design.

To build their map, the researchers first performed a thermodynamic extrapolation using Thermo-Calc software, combining the most recent optimized assessments of the binary Co-Si, Co-B, and Si-B systems and accounting for the known ternary phase. They then synthesized a series of cobalt-rich alloys by arc melting high-purity elements in a water-cooled copper crucible under argon. Each composition was remelted four times to guarantee chemical homogeneity, with total mass losses kept below one percent so that the nominal compositions could be trusted as the true alloy compositions. The as-cast samples were then interrogated with an arsenal of characterization techniques: X-ray diffraction, scanning electron microscopy with backscattered electron imaging, energy-dispersive X-ray spectroscopy, and electron backscatter diffraction, the latter used to definitively identify crystal phases in the microstructures.

The resulting liquidus projection identifies ten primary precipitation fields, meaning ten different crystalline phases that can be the first to appear as the liquid cools, depending on composition. These include the cobalt solid solution in its high-temperature and room-temperature allotropic forms, the silicides Co3Si, alpha Co2Si, beta Co2Si, and CoSi, the borides Co3B, Co2B, and CoB, and the ternary phase Co4.75Si2B. The map also traces the monovariant lines, the boundaries along which two phases crystallize simultaneously from the melt, and proposes nine ternary invariant reactions involving the liquid phase, including three ternary eutectic points where three phases solidify together. Twelve binary invariant reactions inherited from the Co-Si and Co-B edge systems anchor the projection, and the arrows along the monovariant lines point the way toward lower temperatures, tracing the solidification paths that molten alloys of any given composition will follow.

Perhaps the most striking finding concerns the ternary phase Co4.75Si2B, first identified in the literature more than sixty years ago but never before characterized in terms of its melting behavior. By preparing an alloy with exactly the stoichiometric composition of this phase, the researchers discovered that it does not melt congruently. Instead, the microstructure showed crystals of Co4.75Si2B enveloping primary grains of alpha Co2Si, a classic signature of a peritectic reaction. In other words, the ternary phase forms when the liquid reacts with pre-existing alpha Co2Si crystals rather than crystallizing directly from the melt on its own. This single observation, expressed as the reaction L plus alpha Co2Si yielding Co4.75Si2B, resolves a long-standing question about the phase and reshapes the topology of the entire liquidus surface around it.

Equally fascinating is the discovery of three thermal maxima along the monovariant lines, labeled m1, m2, and m3. These points act as thermodynamic watersheds on the liquidus surface. On one side of a maximum, the liquid composition evolves toward one ternary eutectic point during solidification; on the other side, it flows toward a completely different one. The researchers demonstrated this with pairs of alloys whose compositions straddled the maxima: alloy 41, for example, saw its liquid become richer in boron as it solidified toward the EI eutectic, while alloy 5a, on the opposite side of maximum m2, saw its liquid become richer in cobalt and terminate along a different monovariant line. Tiny shifts in the initial composition, therefore, send the solidification process down entirely different trajectories, a fact of enormous practical importance for anyone trying to control the microstructure of a cast or rapidly quenched alloy.

The monovariant line separating the alpha Co2Si and Co4.75Si2B primary fields turned out to have a dual personality. Between the thermal maximum m1 and the invariant point UIII, the line behaves as a eutectic valley, with the two phases crystallizing in a coupled fashion. Between m1 and the invariant point UVI, however, the line switches to peritectic behavior, with one phase reacting with the liquid to form the other. This kind of behavioral switch along a single monovariant line is a subtle feature of ternary phase diagrams, and detecting it required the careful microstructural detective work that characterizes the whole study, including the identification of fine-scale eutectic constituents that push the resolution limits of electron backscatter diffraction to roughly one micrometer.

The new projection is in good agreement with the partial liquidus map proposed by Omori and Hashimoto in 1977 for the region above 66.7 atomic percent cobalt, but it extends and refines that pioneering work substantially, adding fields for phases such as CoSi, CoB, Co3Si, and beta Co2Si that the earlier study could not resolve. The authors note that the actual temperatures of the invariant reactions, liquidus, and solidus were not determined in this study, leaving a clear target for future differential thermal analysis and thermodynamic modeling. Even so, the compositional framework alone provides what the field has lacked: an experimentally grounded prediction of solidification behavior across the cobalt-rich ternary space.

For the growing community working on soft magnetic materials, the implications are immediate. Cobalt-based metallic glasses are prized in high-precision applications where magnetic cores must be stable, efficient, and corrosion-resistant, and every improvement in glass-forming ability translates directly into better ribbons and devices. With the liquidus projection now in hand, alloy designers can anticipate which borides and silicides will compete with glass formation for any candidate composition, choose compositions that maximize the distance from troublesome primary phase fields, and understand why small compositional tweaks near the thermal maxima can dramatically alter the outcome of rapid solidification. It is a reminder that even in a field as mature as metallurgy, fundamental maps still hold the power to unlock the next generation of technology, one carefully melted button of alloy at a time.

Subject of Research: Phase equilibria and solidification behavior of the cobalt-rich Co-Si-B ternary alloy system

Article Title: Liquidus projection of the cobalt-rich Co–Si–B system

Article References: Pellegrino, E. K. S., Boas, S. B. V., da Silva, A. A. A. P., Ferreira, L. M., Faria, M. I. S. T., Coelho, G. C., & Nunes, C. A. (2026). Liquidus projection of the cobalt-rich Co–Si–B system. Journal of Materials Science. https://doi.org/10.1007/s10853-026-13756-y

Image Credits: AI Generated

DOI: 10.1007/s10853-026-13756-y

Keywords: metallic glasses, amorphous alloys, cobalt alloys, liquidus projection, phase diagrams, peritectic reaction, ternary system, solidification, magnetic materials, silicides, borides, thermodynamics