Sintering is one of the oldest and most energy-intensive routes to densify ceramics and metals, and lowering its temperature has long been a central goal of materials processing [
1]. The recent study by Cui et al. [
2], which combines
in situ transmission electron microscopy with precise mechanical manipulation of a single bismuth (Bi) nanocluster, offers a strikingly simple thermodynamic principle that speaks directly to this goal. Trapping the cluster inside a tunable nanoscale gap, the authors show that under strong geometric confinement the stability of a phase is governed not by its volume but by its aspect ratio (
l/d), where
l is the thickness perpendicular to the contact interfaces and
d is the lateral dimension parallel to them. When the cluster is compressed to
l/d < 0.4, it forms a quasi-amorphous nanodisc; in the window 0.4 <
l/d < 1.5 it crystallizes into a single-crystal nanowire; and once
l/d exceeds ~1.5 it melts back into a liquid nanodroplet (Fig. 1A). The quasi-amorphous state should be distinguished from a liquid: it is a disordered, solid-like phase that retains short-range order while exhibiting liquid-like atomic mobility, and it is not a fluid. These boundary values are specific to the confined Bi nanocluster, and in other material systems they may shift with surface-energy anisotropy, interfacial chemistry, temperature, pressure, defects, and crystallographic anisotropy. The qualitative principle that confinement geometry controls the balance among amorphous-like, crystalline, and liquid states should therefore be distinguished from the exact Bi-derived thresholds. Because such confined, low-aspect-ratio geometries are precisely what particles experience at their contacts during densification, this aspect-ratio phase map, rooted in the interplay between surface-energy anisotropy and interfacial energetics, provides a unifying thermodynamic framework for understanding and engineering low-temperature sintering. This perspective applies that phase map as a unifying framework for low-temperature sintering across four settings: nanopowder consolidation, flash and ultrafast sintering, liquid-phase sintering, and biomineralization. The discussion is a conceptual transfer of the thermodynamic picture to sintering-related geometries rather than a claim of direct experimental evidence in each system, and the framework’s material-dependent limits are considered below.
1 Nano-Powder Sintering: Quasi-Amorphous Contact Layers and Ultrafast Crystallization
In the sintering of nanopowders, the geometry of the particle contact region largely determines its phase behavior and hence the kinetics of densification. Here,
l is the thickness of the compressed contact region perpendicular to the particle interface and
d is its lateral extent along the interface. During hot pressing or spark plasma sintering (SPS) [
3], where an external pressure is applied, nanoparticles are forced into intimate contact, and the contact zone is compressed along the direction perpendicular to the interface. This produces a low-aspect-ratio configuration (
l/d < 0.4), which, according to the phase map of Cui et al. (the numerical boundaries are material-dependent), corresponds to a quasi-amorphous state: laterally disordered yet retaining short-range order, and behaving in many respects like a liquid. Atoms in such a layer enjoy exceptionally high mobility and diffusivity, so they can promote particle rearrangement and neck formation at temperatures far below the bulk melting point (Fig. 1B). The size-induced depression of the melting point further lowers the temperature at which such a layer can form, providing an additional route to mild-condition densification. Classical sintering models, including surface and grain-boundary diffusion, viscous flow, and evaporation-condensation, describe the mass-transport pathways and kinetics that produce neck growth and densification, but they do not by themselves predict which phase state is thermodynamically favored in a strongly confined contact. The aspect-ratio framework supplies this missing thermodynamic criterion: it identifies when the contact region should prefer a quasi-amorphous or liquid-like state with high atomic mobility, allowing those transport mechanisms to operate at unusually low temperatures. It is therefore complementary to, rather than a replacement for, established sintering mechanisms.
Flash sintering and ultrafast high-temperature sintering (UHS) operate at the opposite extreme, relying on extremely rapid heating and localized overheating [
4]. In flash sintering, the electric field concentrates Joule heating at the particle contacts; in UHS, radiative heating momentarily raises the particle surface to a very high temperature. Both mechanisms create strongly confined molten zones at the nanoscale contacts, that is, regions with
l/d > 1.5 (the numerical boundary is material-dependent). Here,
l is the thickness of the molten neck normal to the particle contact and
d is its lateral dimension along the interface. These molten pockets should be viewed as transient precursors rather than end states. When field pulses or the applied pressure readjust the local geometry so that the aspect ratio re-enters the 0.4–1.5 crystalline window, the melt rapidly crystallizes under constraint (Fig. 1B). The surface-energy inversion reported by Cui et al. [
2], in which the lateral surface energy of the Bi nanowire (0.26 J/m
2) is lower than that of liquid Bi (0.378 J/m
2), shows that once this crystalline window is entered, low-energy orientations can be stabilized at low temperature, substantially lowering the recrystallization barrier. Molten regions in ultrafast sintering are therefore not merely fast diffusion channels; they are precursors whose crystallization can be locked in by geometry, suggesting a route to rapid, low-temperature densification that is difficult to achieve by conventional sintering.
2 Liquid-Phase Sintering: Stability of Grain-Boundary Films and Solidification at Triple Junctions
In liquid-phase sintering, the distribution and ultimate fate of the residual liquid dictate the final microstructure. The l/d phase map of Cui et al. provides a quantitative framework for understanding both the stability and the solidification of this liquid. Consider a thin liquid film trapped at a grain boundary. For this film, l is the film thickness between the two grains and d is its in-plane extent. Because it is strongly compressed between neighboring grains, its aspect ratio is extremely small (l/d << 0.4) (the Bi-derived boundaries are indicative and may shift with material and interfacial chemistry), placing it deep in the quasi-amorphous region of the phase map. As long as the grain spacing and the interfacial chemistry (contact angles) remain unchanged, the film rests near a free-energy minimum and can persist for extended periods, largely insensitive to variations in overall volume or temperature (Fig. 1C). This offers a simple explanation for the residual liquid frequently observed in liquid-phase-sintered materials after long-term high-temperature service: the thermodynamic stability of the film is guaranteed by the confinement geometry rather than by a delicate temperature-composition equilibrium.
Triple junctions present a different and instructive situation. Liquid channels there are usually slender, with l/d >> 1, and therefore reside in the liquid regime of the phase map. For such a channel, l is the length along the channel and d is its width across the junction, so the large l/d ratio reflects its elongated shape. During late-stage sintering, however, two processes can drive these channels toward solidification: diffusional transport may shorten the channel (decreasing l), or grain rearrangement may widen it (increasing d). In either case, the local aspect ratio can fall into the 0.4–1.5 window and trigger confinement-induced solidification (Fig. 1C). More importantly, the crystallographic facets of the three surrounding grains create a multifaceted constraint. Solidification within such a pocket may therefore force the emerging phase to expose specific low-energy facets, analogous to the orientation locking ([]) observed in the confined nanowires of Cui et al., or even to form metastable phases and special grain-boundary structures. These considerations provide a thermodynamic basis for the controlled solidification of residual liquid and for the orientation design of grain-boundary phases in liquid-phase-sintered ceramics.
3 Biomineralization: A Natural Blueprint for Low-Temperature Densification
Biomineralization is arguably Nature’s most accomplished example of low-temperature densification [
5]. In the nanoscale gaps between collagen fibrils, mineral precursors first deposit in a quasi-amorphous form (
l/d < 0.4), where
l is the gap thickness between fibrils and
d is the lateral extent of the gap. As the gap geometry is fine-tuned, the local aspect ratio enters the 0.4–1.5 window and the amorphous phase crystallizes into hydroxyapatite, with the [0001] axis locked along the fibril direction [
6] (Fig. 1D). No high temperature is required: densification and texture development are achieved at body temperature through the synergy of geometric confinement and surface-energy anisotropy, offering a compelling biomimetic blueprint for artificial low-temperature sintering.
4 Applicability and Limits
The framework should be read as a thermodynamic guide rather than a quantitative law. The reported thresholds were measured for a single Bi nanocluster, and they are expected to shift with material-dependent surface energies, interfacial chemistry, temperature and pressure, defect populations, and crystallographic anisotropy. It also addresses thermodynamic preference rather than kinetic accessibility: rapid heating or cooling, limited atomic mobility, or competing reactions can trap a system away from its equilibrium state, so the observed phase depends on both geometry and processing path. Several predictions follow: a low
l/d should favor a quasi-amorphous, high-mobility contact layer and reduce the neck-growth onset temperature in oxide nanoparticle contacts; adjusting pressure or field pulses to move the local
l/d into the intermediate window should promote constrained crystallization during flash or ultrafast sintering; and facet-mediated solidification at triple junctions should produce crystallographically oriented grain-boundary phases. Broader experimental and simulation studies across materials and interfaces, together with process-level evidence from SPS, flash sintering, and cold sintering [
3,
4,
7,
8], will be needed to establish how widely these numerical boundaries and mechanisms apply.
5 Concluding Remarks
Cui et al. establish a principle of considerable generality: in a geometrically confined nanospace, the aspect ratio
l/d, rather than volume, is the central order parameter of phase stability, and its influence may extend across many areas of sintering. In nano-powder sintering, low-
l/d quasi-amorphous contact layers supply the liquid-like atomic mobility that hot pressing exploits, whereas flash and ultrafast sintering create localized melts whose constrained crystallization can be locked into the crystalline window. In liquid-phase sintering, both the long-term stability of grain-boundary films and the solidification at triple junctions can be read directly from the
l/d phase map. In biomineralization, the same principle achieves densification and texture at body temperature. In solution-mediated cold sintering, densification proceeds through transient liquid films confined at particle contacts, which correspond to the low-
l/d region of the phase map. The same geometric criterion may therefore help explain the stability and re-crystallization of these films, reinforcing the outlook toward low-energy, high-precision manufacturing. Looking forward, combining DFT-based screening of low-energy facets with interfacial and coating design may allow us to sculpt microstructures under mild conditions in much the same way organisms do, a vision that resonates with emerging cold-sintering strategies [
7,
8] and points toward an era of low-energy, high-precision materials manufacturing.
The Author(s). This article is published by Higher Education Press.
PDF
(1700KB)
