News & Views | Published:

Condensed-matter physics

Glasses made from pure metals

Nature volume 512, pages 142143 (14 August 2014) | Download Citation

The experimental realization of amorphous pure metals sets the stage for studies of the fundamental processes of glass formation, and suggests that amorphous structures are the most ubiquitous forms of condensed matter. See Letter P.177

On page 177 of this issue, Mao and colleagues1 report a method that allows them to achieve a long-standing goal for materials scientists — the formation of glasses from pure metals. This will enable much-needed studies of glass formation in simple systems, and allows computational modelling of the processes involved.

For thermodynamic reasons, most liquids become crystalline when they are cooled below their 'liquidus' temperature, above which substances are completely liquid. Crystallizations occur at different timescales and can be suppressed by fast cooling of a liquid, causing it to vitrify into a glass2. Vitrification occurs for various materials at widely different critical cooling rates (Rc) — the minimum rate of cooling required to form a glass.

Glass formation has been reported for metallic alloys3. An alloy's glass-forming ability increases with the number of components in the alloy, particularly if it contains elements with atomic sizes that differ by more than 12% and which have the thermodynamic impetus to mix4. Some alloys that exhibit these criteria, known as bulk metallic glasses, have remarkably good glass-forming abilities, with Rc values of less than 1,000 kelvin per second (comparable with the cooling needed to make amorphous polymers). They also have critical casting thicknesses — the largest thickness over which heat can be extracted enough to avoid crystallization — exceeding 1 millimetre. So far, hundreds of complex alloys have been reported to form bulk metallic glasses.

Pure metals do not fulfil the above criteria because they lack the complexity needed to 'confuse' crystallization5. They have therefore been considered to be poor glass formers6. Even advanced rapid-cooling techniques have been too slow to avoid crystallization of liquid pure metals, except in some specific cases7. Mao and co-workers now introduce a general ultra-rapid heating and cooling method that allows liquids of pure metals to be vitrified.

The authors used a nanometre-scale heating device that brings together two metal tips approximately 100 nm in length. Heating was accomplished using a short electrical pulse (about 4 nanoseconds in duration), which rapidly melted the tips. The heat then dissipated rapidly through the melted sample towards the device, inducing cooling rates of roughly 1014 kelvin per second at the centre of the sample. Such high cooling rates were predicted by the researchers to occur on the basis of molecular-dynamics modelling, and caused vitrification of a region of pure metals approximately 40 nm by 50 nm in size.

Metallic glasses are pursued for commercial applications because they exhibit attractive mechanical properties such as high strength, elasticity and processability8. The advent of metallic-glass formation, along with methods that allow the liquid state of metals to be studied at slow, experimentally accessible timescales, have also been exciting for fundamental science. These developments have enabled study of the properties of metallic liquids, and investigation of both the transition to the crystalline state and the glass transition. However, the fact that multicomponent alloys have been needed for glass formation has complicated the study of metallic glasses.

In multicomponent systems, glass formation depends on atomic-size differences and attraction between atoms of different elements. Glass formation is also affected by the fact that crystallization in alloys typically requires a change in atomic composition: long-range diffusion is needed to establish the difference in composition between the liquid and the growing crystalline phase. Such diffusion has a long timescale and slows down crystallization, facilitating glass formation. But it also obscures the fundamental and ubiquitous aspects of vitrification that would be observed in simple systems. Mao and colleagues' breakthrough allows glass formation to be studied in its purest form, and their findings confirm theoretical and modelling predictions that glass formation can occur in pure metals.

The researchers studied metals in which atoms adopt a 'body-centred cubic' (bcc) arrangement in the crystalline solid phase. But what would happen for metals that adopt different crystal structures, such as the common face-centred cubic (fcc) arrangement? Glass formation is limited only by crystal growth in Mao and co-workers' heating device, and crystal growth rates are slower for bcc crystals than for fcc ones. The Rc values for pure fcc metals are therefore expected to be even higher than those reported by Mao et al. for pure bcc metals.

In its most general form, crystallization involves nucleation — the initial formation of tiny crystals called nuclei — and growth. Glass formation competes with the combination of both processes. However, crystallization proceeds through growth into the undercooled liquid phase of the crystal–liquid interface in Mao and co-workers' experiments (Fig. 1). Crystal growth therefore does not depend on nucleation in their system, which means that the Rc values reported by the authors are probably an overestimate for the most general form of vitrification in pure bcc metals. To involve nucleation, direct contact of the liquid phase to a crystalline boundary has to be avoided. Such an experimental realization would allow study of the earliest stages of nucleation — one of the great mysteries of physics.

Figure 1: Ultra-rapid cooling causes a pure metal to form a glass.
Figure 1

The micrograph shows a region of molten tantalum between two crystalline regions. Mao and colleagues1 report that, on ultra-rapid cooling, the crystalline regions grow into the liquid region (blue arrows) until the growth kinetics can no longer keep up with the thermal field defined by the cooling rate. The liquid in front of the interface then 'freezes' into a glass. On heating, the crystal–liquid interface moves out into the crystalline sections of the sample (red arrows). Image: Scott X. Mao

Experimental investigations of glass formation have typically been carried out on large samples of more than 108 atoms, and at long timescales greater than 1 microsecond. By contrast, molecular-dynamics simulations have been limited to small samples of fewer than 105 atoms and short timescales (less than 1 nanosecond), because of the restrictions of available computing power. Our ability to predict experimental results from such simulations has therefore been limited because the properties of metallic glasses are affected by sample size9 and cooling rates10. Mao and colleagues' method now allows us to carry out experiments at spatial and temporal timescales similar to those in simulations. This opens the way to exploring glass formation and its competition with crystallization. Given that vitrification has previously been observed in ionic melts, aqueous solutions, alloy melts, molecular liquids and polymers, the finding that pure metals can also be glasses suggests that amorphous structures are the most ubiquitous form of condensed matter.


  1. 1.

    , , , & Nature 512, 177–180 (2014).

  2. 2.

    Science 267, 1924–1935 (1995).

  3. 3.

    , & Nature 187, 869–870 (1960).

  4. 4.

    Acta Mater. 48, 279–306 (2000).

  5. 5.

    Nature 366, 303–304 (1993).

  6. 6.

    Contemp. Phys. 10, 473–488 (1969).

  7. 7.

    et al. Nature 448, 787–790 (2007).

  8. 8.

    Phys. Today 66, 32–37 (2013).

  9. 9.

    , & J. Appl. Phys. 103, 083539 (2008).

  10. 10.

    , , & Nature Commun. 4, 1536 (2013).

Download references

Author information


  1. Jan Schroers is in the Department of Mechanical Engineering and Materials Science, Yale University, New Haven, Connecticut 06511, USA.

    • Jan Schroers


  1. Search for Jan Schroers in:

Corresponding author

Correspondence to Jan Schroers.

About this article

Publication history



Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Newsletter Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing