In situ STEM observation of the impact of surface oxidation on the crystallization of GeTe Phase Change Material thin films

Abstract

Chalcogenide phase change materials (PCMs) such as Ge‐Sb‐Te and GeTe alloys exhibit outstanding properties, which has led to their successful use as non‐volatile resistive memories in Phase Change Random Access Memories (PCRAM). PCRAM using PCMs can be switched reversibly between their crystalline and amorphous phases with different optical and electrical properties offering a unique set of features such as fast programming, good cyclability, high scalability, multi‐level storage capability and good data retention. Controlling the crystallization is a challenge and numerous studies have been conducted to probe interface and size effects on the PCM crystallization. Surface engineering has a crucial role on the crystallization temperature and mechanisms [1,2] . Temperature resolved reflectometry experiments have shown that the crystallization temperature of GeTe films (in the thickness range 30–100 nm) change drastically depending on its surface state (Fig.2). For a better understanding of this phenomenon, we performed in situ STEM experiments to observe the complete crystallization mechanisms at a nanometer scale of GeTe films with various surface states. Amorphous GeTe films were deposited by magnetron sputtering in an industrial cluster tool and were protected either by in situ deposition of a 10nm thick SiN capping layer or left uncapped before being exposed to air. For STEM analysis, a specifically adapted preparation method using focused ion beam (FIB) milling has been developed in order to perform in situ annealing and crystallization of the GeTe films directly in the microscope (Fig.1). In particular, a specific positioning of the FIB foil enables low energy cleaning despite the sample holder configuration. We will show that this new sample preparation method, combined with the precise temperature control and negligible spatial drift when using the Protochips Aduro sample holder, allows atomic resolution and quantitative analysis to be obtained during in situ annealing. Results show that the uncapped (i.e. surface oxidized) GeTe film exhibits a two‐step crystallization mechanism. First, the crystallization spreads across the sample over the top 20 nm of the initial amorphous layer. If the temperature ramp is allowed to continue, the nucleation‐growth of the remaining amorphous part of the GeTe film is triggered at 50°C above the temperature corresponding to surface crystallization (Fig.3b,d,f). We will give evidence that the GeTe film capped by a 10nm SiN layer prior to air exposure exhibits a very different crystallization temperature and mechanism. Indeed, in that sample a single‐step crystallization occurs through a one‐step nucleation‐ growth in the whole layer at a temperature corresponding to the second crystallization step of the uncapped GeTe film. By quenching before complete crystallization, crystalline nuclei were imaged at high resolution and we observed that crystallization occurred by volume nucleation within the amorphous layer (Fig.3a,c,e). We will show that if protected from oxidation, the GeTe crystallization mechanism can be a pure nucleation‐growth process happening about 50°C above previously reported values [2] . An interpretation of this crystallization mechanism will be proposed based on the elemental segregation obtained by EDS and live recording of the crystallization obtained using multiple STEM detectors. This information will be invaluable to improve reliability and data storage capability of GeTe based devices. By adapting our in situ procedure for electrical biasing, it will be possible to perform real time TEM observation of GeTe switching between ON and OFF states. Then by comparing both electrical and thermal induced crystallization, we will be able to obtain important information about GeTe switching at an atomic scale to provide better devices.