News & Views | Published:

Materials science

Changing face of the chameleon

Chalcogenide materials form the basis of CD and DVD technologies. But an identity crisis looms in the wider field: what role do atomic reconfiguration, electronic processes and ionic movement play in these materials?

The chalcogens — the elements in group VI of the periodic table, particularly sulphur (S), selenium (Se) and tellurium (Te) — react with more electropositive elements, such as silver, to form chalcogenides. These are chameleon compounds: they can be crystalline or amorphous, metallic or semiconducting, and conductors of ions or electrons. Already important in optical storage discs and fibres, they are now being proposed as the basis for solid-state memory technologies. Two recent conferences — EFootnote 1PCOS 05 in Cambridge * , UK, and Euromat 2005 in Prague , Czech Republic — have demonstrated that devices using chalcogenides hinge on thermal and dynamic phenomena involving electronic, atomic and ionic processes. The links between these phenomena are not fully established, so unsuspected technological opportunities may well lie in store.

Electrical switching in chalcogenide semiconductors came to prominence in the 1960s, when the amorphous chalcogenide Te48As30Si12Ge10 was found1 to display sharp, reversible transitions in electrical resistance above a threshold voltage (Fig. 1a). The switching mechanism remains unclear, but seems2 to be initiated by fast3,4, purely electronic processes. If current is allowed to persist in the material, it heats up, changing its atomic structure between the amorphous and crystalline states — equivalent to information being written on it. A crystalline region may be driven to become amorphous by exposure to a brief, intense heat pulse, leading to melting. The subsequent rapid withdrawal of heat sends the temperature plummeting so quickly that the melted region solidifies with the atoms still disordered. Conversely, a lower-intensity heat pulse of longer duration will crystallize an amorphous region — the crystalline state is more stable and the heat allows the atoms to mobilize just enough to assume crystalline order.

Figure 1: Two types of chalcogenide device.

a, Electronic switching associated with phase changes. At moderate frequencies, the alternating current–voltage (I–V) characteristic of a suitable amorphous chalcogenide material shows symmetric switching above a threshold voltage Vt from a highly insulating regime (red) to a more conducting regime (blue, gradient reduced for clarity). (Modified from ref. 1.) Optically addressed phase-change materials form the basis for CD and DVD technologies. b, Ionic (electrolytic) switching. A solid electrolyte of two chalcogenide phases can also encode information. In the cycle shown, the insulating state (red) switches at VON to a conducting state (blue) when nanoscale bridges of silver form between electronically conducting islands in an ionically conducting matrix. These bridges persist until they are dissolved at a sufficiently negative voltage, VOFF. The asymmetric I–V characteristic shown here is reminiscent of symmetric I–V characteristics in phase-change random-access memory, where read operations in a working device would also be performed at low voltages. (Modified from ref. 10.)

This thermally driven, amorphous–crystalline phase change, which can encode binary information, is already of great commercial significance5, but uses thin films of chalcogenides that are switched locally by optical rather than electrical means. In write-once and rewritable CDs and DVDs, a laser beam supplies the heat pulse for write operations, whereas the read process exploits the relatively low optical reflectivity of amorphous data spots, or marks, in the chalcogenide film compared with that of the crystalline background. Contributions to E * PCOS 05 reporting optical-disc capacities as high as 112 gigabytes (A. Nakaoki, Sony Corp.) and data marks as fine as 20 nm (D.-P. Tsai, National Taiwan Univ.) show that current industrial trends could well continue.

Attempts1 to induce the amorphous–crystalline transformation of chalcogenides by electrical means form the basis of phase-change random-access memory (PC-RAM). This incipient technology is on the brink of commercial exploitation by, among others, the pioneering company ECD Ovonics. For write operations, an electric current supplies the heat pulse. The read process is performed at sub-threshold voltages, and exploits the relatively large difference in electrical resistance between the amorphous and crystalline states. Recent advances in this area include a low-power device set up as a narrow line of material with the active region thermally isolated from the electrical connections, so that these are not degraded by heating6. There is also the prospect of much smaller data marks written with a beam of electrons7 and, as announced at E * PCOS 05, continuing improvements in programming times (below 30 ns: K. Attenborough, Philips Res. Leuven) and writing currents (below 0.75 mA: B. J. Kuh, Samsung Electronics).

Although the electronic transitions and atomic rearrangements relevant to both optical discs and PC-RAM featured strongly at E * PCOS, contributions from ions were not considered — even though amorphous chalcogenides can have significant ionic conductivities. At Euromat 2005, however, it was shown that ionic transport can be useful for data storage in a solid chalcogenide electrolyte. At the nanoscale, this electrolyte consists of crystalline metallic islands of silver selenide (Ag2Se) dispersed in an amorphous semiconducting matrix of germanium selenide (Ge2Se3) (M. N. Kozicki, Arizona State Univ.). The performance of a prototype electrolytic chalcogenide cell is described in Figure 1b.

Technologies exploiting phase-change and electrolytic chalcogenide devices are evolving convergently. Although the microscopic mechanisms seem rather different, the two technologies display similar measurable characteristics. The sudden onset of conduction — at Vt in Fig. 1a and VON in Fig. 1b — is, for example, associated with the formation of filamentary conducting pathways in both cases. In PC-RAM this onset is thought to arise electronically via the injection of charge carriers2, but stems in the electrolytic device from the deposition of metal atoms (electrodeposition) to form conducting bridges between the islands. Another example of the similarity in characteristics but variation in underlying processes of the two types of chalcogenide device is that both show incremental changes in their structure, cumulative over time, when they are operated below their threshold voltages. These changes give rise to controllable intermediate conductivities and are in effect precursors to the binary memory effects that make chalcogenides useful as storage materials. In PC-RAM, this cumulative behaviour is readily explained by crystal growth; in the electrolytic variant, it is explained by electrodeposition.

Both chalcogenide technologies present exciting opportunities that are not restricted to memory, but include cognitive computing5,8 (E * PCOS 05: S. R. Ovshinsky, ECD Ovonics) and reconfigurable logic circuits9. It is too early to tell which technology will be selected for which niche, but scientific interest alone should motivate a closer look at chalcogenide materials to investigate correlations between phase-change and electrolytic behaviour. To take one example, the migration of dissolved ions is required in the electrolytic case, but could degrade the performance of a phase-change device. Fluxes of both electrons and ions participate in electromigration — widely studied as a degradation mechanism of the electrically conducting lines for integrated circuits. Thus, a unified approach to the study of chalcogenides, assessing the roles of atoms, ions and electrons, may prove crucial for both device performance and reliability.Footnote 2


  1. 1.

    *European Symposium on Phase Change and Ovonic Science, Cambridge, 3–6 September 2005;

  2. 2.

    Symposium on Memory Storage Materials. European Congress on Advanced Materials and Processes, 5–8 September 2005;


  1. 1

    Ovshinsky, S. R. Phys. Rev. Lett. 21, 1450–1453 (1968).

  2. 2

    Mott, N. F. in Nobel Lectures, Physics 1971–1980 (ed. Lundqvist, S.) 403–413 (World Scientific, Singapore, 1992);

  3. 3

    Adler, D. et al. J. Appl. Phys. 51, 3289–3309 (1980).

  4. 4

    Vezzoli, G. C., Walsh, P. J. & Doremus, L. W. J. Non-Cryst. Solids 18, 333–375 (1975).

  5. 5

    Strand, D. J. Optoelectron. Adv. Mater. 7, 1679–1690 (2005).

  6. 6

    Lankhorst, M. H. R. et al. Nature Mater. 4, 347–352 (2005).

  7. 7

    Gibson, G. A. et al. Appl. Phys. Lett. 86, 051902 (2005).

  8. 8

    Ovshinsky, S. R. Jpn J. Appl. Phys. Pt 1 43 (7B), 4695–4699 (2004).

  9. 9

    Terabe, K. et al. Nature 433, 47–50 (2005).

  10. 10

    Kozicki, M. N., Park, M. & Mitkova, M. IEEE Trans. Nanotechnol. 4, 331–338 (2005).

Download references

Author information

Rights and permissions

Reprints and Permissions

About this article

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.