**4. Nanosecond recrystallization dynamics in GST and AIST**

To uncover recrystallization dynamics in GST and AIST at atomic level, we investigated the atomic structure of the amorphous phase by using a combination of advanced synchrotron radiation measurements (X-ray diffraction, EXAFS, HXPS) and reverse Monte Carlo simulation (RMC)/density function theory (DF)-molecular dynamics (MD) simulations (Matsunaga et al., 2011).

Figure 13 shows the structure factors *S*(*Q*) of AIST and GST obtained using X-ray diffraction. The crystalline forms of both materials have sharp Bragg peaks (red lines), and the amorphous forms (blue lines) have typical halo patterns. However, oscillations up to the maximum *Q* value in a-AIST indicate a structure with well defined short-range order. The total correlation functions *T*(*r*) for AIST and GST are shown in Fig. 14. The *T*(*r*) for crystalline (c-) AIST and c-GST, which are very similar beyond 4 Å. Small differences between the two crystalline forms are found at shorter distances, for example the double peak in c-AIST (2.93 Å and 3.30 Å) and a single peak in c-GST (2.97 Å). The *T*(*r*) for the amorphous materials, however, are significantly different: the first peak in amorphous (a-)

C). The crystal growth is then disturbed by the impingement of crystallites with each other (D). Our schematic model is consistent with the nucleation-driven crystallization process

Fig. 12. (a) Schematic models with TEM image for the crystallization processes in GST and

On the other hand, the nuclei of AIST can immediately transform to smaller crystallites (<< 60 nm), which form domains (F). These domains are enlarged by edge-growth crystallization (F-G) and the crystallites coalesce at the final stage (300 ns~) of crystallization in AIST (G-H). These proposed schematic models are consistent with the TEM pictures, in which GST has a grainy texture filled with 100-nm-size grains, whereas AIST has a fine

To uncover recrystallization dynamics in GST and AIST at atomic level, we investigated the atomic structure of the amorphous phase by using a combination of advanced synchrotron radiation measurements (X-ray diffraction, EXAFS, HXPS) and reverse Monte Carlo simulation (RMC)/density function theory (DF)-molecular dynamics (MD) simulations

Figure 13 shows the structure factors *S*(*Q*) of AIST and GST obtained using X-ray diffraction. The crystalline forms of both materials have sharp Bragg peaks (red lines), and the amorphous forms (blue lines) have typical halo patterns. However, oscillations up to the maximum *Q* value in a-AIST indicate a structure with well defined short-range order. The total correlation functions *T*(*r*) for AIST and GST are shown in Fig. 14. The *T*(*r*) for crystalline (c-) AIST and c-GST, which are very similar beyond 4 Å. Small differences between the two crystalline forms are found at shorter distances, for example the double peak in c-AIST (2.93 Å and 3.30 Å) and a single peak in c-GST (2.97 Å). The *T*(*r*) for the amorphous materials, however, are significantly different: the first peak in amorphous (a-)

**4. Nanosecond recrystallization dynamics in GST and AIST** 

(Zaou et al., 2000).

AIST.

texture.

(Matsunaga et al., 2011).

AIST (2.86 Å) is only slightly shorter than that found in c-AIST (2.93 Å), whereas the first peak in a-GST (2.79 Å) is shorter than that in c-GST (2.97 Å). We note also that the shoulder on the second peak in a-AIST (3.50 Å) is near that observed in the crystalline form (3.30 Å). The pronounced difference between the diffraction patterns of the two materials is strong evidence that they crystallize differently. The atomic motion and/or diffusion accompanying the phase change are larger in GST than in AIST, where the phase change is accompanied by small changes in bond lengths.

Fig. 13. Total structure factors, *S*(*Q*), for AIST and GST. Blue line, amorphous form; red line, crystalline form.

Fig. 14. Total correlation functions, *T*(*r*), for AIST and GST. Blue line, amorphous form; red line, crystalline form.

Time Resolved Investigation

mechanisms.

GST.

**5. Conclusion** 

**6. Acknowledgment** 

of Fast Phase-Change Phenomena in Rewritable Optical Recording Media 271

time resolved measurements mentioned above. A combination of advanced synchrotron measurements and computer simulation has provided insight into the atomistic differences between two phase-change materials, which is related to their different crystallization

Fig. 16. Ring statistics (a) and possible schemes for fast phase change (b) in a-AIST and a-

We described the detail of the X-ray pinpoint structural measurement system for investigation of optical recording process. Furthermore, we showed the recent progress for fully understanding the atomic structure of AIST and compare it to GST with the research combined X-ray diffraction, EXAFS, HXPS measurements and computer simulations. These demonstrated that the time resolved X-ray diffraction technique using SR is very powerful

This work was supported by Core Research for Evolutional Science and Technology (CREST) "X-ray pinpoint structural measurement project -Development of the spatial- and time-resolved structural study for nano-materials and devices-" and by the Academy of Finland and the Japan Science and Technology Agency through the Strategic Japanese– Finnish Cooperative Program on "Functional materials". The synchrotron radiation

for the structural investigation of crystal growth phenomena.

Atomic configurations for a-AIST and a-GST obtained from the RMC / DF-MD simulations are shown in Fig. 15. It is found from the atomic configurations that the local environment of the predominant element Sb (75%) is very similar to that in the crystal (A7, a distorted octahedron) in AIST whereas large fraction of ABAB squares (manifested by ring) is observed in a-GST.

Fig. 15. Atomic configurations for a-AIST and a-GST obtained from the RMC / DF-MD simulations. Ag, silver; In, magenta; Sb, blue; Te, yellow; Ge, red.

Figure 16(a) shows ring statistics in a-AIST and a-GST derived from the RMC/DF-MD models which are consistent with the results of X-ray diffraction, EXAFS and HXPS measurements. In a-GST, 40% of the rings are fourfold or sixfold, whereas the distribution in a-AIST is much broader; the most common (fivefold) rings make up only 15% of the total. On the basis of structural features mentioned above, possible phase-change schemes in both materials are shown in Fig. 16(b). In a-AIST, it is suggested that a sequence of ring reconstructions by way of bond interchanges results in sixfold rings with short Sb bonds accompanied by small changes in the bond lengths, because RMC/DF-MD model for a-AIST has locally distorted 3+3 octahedron, which resembles c-AIST. By contrast, many fourand sixfold rings in a-GST act as nuclei for crystallization and require larger atomic displacements than in a-AIST. Crystallization starts simultaneously from many such nuclei in the amorphous mark (NaCl fragments) and lead to an aggregation of small crystal grains. These features propose a "bond-interchange" model as the origin of "growth-dominated" crystallization of a-AIST, whereas the large fraction of "crystalline nuclei" in a-GST is the origin of the "nucleation-driven" crystallization in GST. The structural finding and possible phase-change mechanism at atomic level obtained in this study is in line with the results of time resolved measurements mentioned above. A combination of advanced synchrotron measurements and computer simulation has provided insight into the atomistic differences between two phase-change materials, which is related to their different crystallization mechanisms.

Fig. 16. Ring statistics (a) and possible schemes for fast phase change (b) in a-AIST and a-GST.
