Toward the Ultimate Limit of Phase Change in Ge 2 Sb 2 Te 5 R. E. Simpson,* ,† M. Krbal, † P. Fons,* ,†,‡ A. V. Kolobov, †,‡ J. Tominaga, † T. Uruga, ‡ and H. Tanida ‡ † Center for Applied Near-Field Optics Research, National Institute of Applied Industrial Science and Technology, Tsukuba Central4, 1-1-1 Higashi, Tsukuba 305-8562, Japan, and ‡ SPring-8, Japan Synchrotron Radiation Research Institute (JASRI), Mikazuki Hyogo 679-5198, Japan ABSTRACT The limit to which the phase change memory material Ge 2 Sb 2 Te 5 can be scaled toward the smallest possible memory cell is investigated using structural and optical methodologies. The encapsulation material surrounding the Ge 2 Sb 2 Te 5 has an increasingly dominant effect on the material’s ability to change phase, and a profound increase in the crystallization temperature is observed when the Ge 2 Sb 2 Te 5 layer is less than 6 nm thick. We have found that the increased crystallization temperature originates from compressive stress exerted from the encapsulation material. By minimizing the stress, we have maintained the bulk crystallization temperature in Ge 2 Sb 2 Te 5 films just 2 nm thick. KEYWORDS Phase change memory, Ge 2 Sb 2 Te 5 , scaling, PCRAM, stress T he ever-increasing demand for greater memory den- sities is driving the development of new memory concepts and materials. Phase change RAM (PCRAM) is an emerging technology which, unlike silicon-based tech- nologies, does not suffer from problems associated with the storage of charge. 1 Ge 2 Sb 2 Te 5 is the leading candidate ma- terial for such technology, 2 and in contrast to Si, “bits” of data are stored in the form of structural differences in a thin film of the material. The ability of Ge 2 Sb 2 Te 5 to crystallize with speeds of less than 50 ns 3,4 and yet retain the amorphous state for dura- tions of years may seem contradictory, but it is this impor- tant attribute which also allows the material to stably store data in cell volumes far less than those of electron trapping in silicon oxide. Changes in the rate of crystallization have been observed for films thinner than 30 nm; however, the true limit to which Ge 2 Sb 2 Te 5 can be scaled yet still retain the ability to change phase needs to be proven. Scaling Ge 2 Sb 2 Te 5 to smaller volumes has the added virtue that the switching power, accomplished by Joule heating, linearly improves with reducing cell size; 5 clearly this is beneficial with respect to the recent increase in portable devices which require large solid-state memories. Ge 2 Sb 2 Te 5 can exist in two crystalline phases, the meta- stable cubic phase and the equilibrium hexagonal phase; in addition it can also exist in an amorphous phase. The cubic and hexagonal phases are formed by increasing the tem- perature of the as-deposited amorphous material to ap- proximately 150 and 300 °C, respectively. The amorphous phase can be formed by either sputtering or rapid heating and quenching; the need for quench rates on the order of 10 10 Ks -1 have been reported. 6 Large optical and electrical differences manifest as a result of the atomic scale structural differences between the amorphous and cubic crystalline phases; generally, the crystalline phase exhibits a higher refractive index, optical absorption, and electrical conductiv- ity in comparison to the amorphous phase. Thus changing phase between the amorphous and the metastable cubic crystalline state causes a large optical and electrical contrast which can be utilized in an optical disk or electrical memory. 7 The hexagonal crystalline phase is not utilized in Ge 2 Sb 2 Te 5 phase change memory, and therefore in this work we concentrate on the amorphous to cubic crystalline phase transition. From a materials perspective the phase change properties are known to be fundamentally different as the thickness of the phase change material is reduced and the surface to volume ratio increased. Recently, efforts to grow single crystal Ge 2 Sb 2 Te 5 revealed that the Ge 2 Sb 2 Te 5 at the sub- strate interface is initially amorphous. 8 Further, the material which interfaces the Ge 2 Sb 2 Te 5 can change the crystal growth rate, activation energy, 9 threshold quench rate for amorphization, and the domain sizes. 10 Raoux et al. studied in situ X-ray diffraction from ultrathin films of Ge 2 Sb 2 Te 5 , GeSb, and SbTe as a function of temperature. 11,12 It was found that for Ge 2 Sb 2 Te 5 films capped with Al 2 O 3 , the crystallization temperature sharply increased with decreas- ing film thickness while the crystal grain size decreased. From this analysis, the limiting film thickness for phase change was suggested to be 4 nm. However X-ray diffraction (XRD) measurement of very thin films at high temperature is challenging due to both the Debye-Waller effect and a reduction in the coherence length thus making definitive * To whom correspondence should be addressed, robert.simpson@aist.go.jp and paul-fons@aist.go.jp. Received for review: 08/25/2009 Published on Web: 12/30/2009 pubs.acs.org/NanoLett © 2010 American Chemical Society 414 DOI: 10.1021/nl902777z | Nano Lett. 2010, 10, 414-419