Metal-Organic Chemical Vapor Deposition (MOCVD) of GeSbTe-Based Chalcogenide Thin Films
- PDF / 847,577 Bytes
- 6 Pages / 612 x 792 pts (letter) Page_size
- 25 Downloads / 192 Views
0997-I10-08
Metal-Organic Chemical Vapor Deposition (MOCVD) of GeSbTe-Based Chalcogenide Thin Films Gary S. Tompa, Shangzhu Sun, Catherine E. Rice, Joe Cuchiaro, and Edwin Dons Structured Materials Ind., Inc., 201 Circle Drive N., Unit 102, Piscataway, NJ, 08854
Abstract Chalcogenide Random Access Memory (C-RAM) has shown significant promise in combining the desired attributes of an ideal memory, including: nonvolatility, fast read/write/erase speed, low read/write/erase voltage/power, high endurance, and radiation hardness. Current C-RAM production technology relies on sputtering to deposit the active chalcogenide layer. The sputtering process leads to difficulties in meeting requirements for device conformality (in particular ñ filling vias), film adherence, compositional control, wafer yield, and surface damage. Ultimately, a viable CVD manufacturing process is needed for highdensity products to realize the full potential of C-RAM. In this work, we discuss the MetalOrganic Chemical Vapor Deposition (MOCVD) tool technology used to produce the films and report the materials properties of GeSbTe-based chalcogenide thin films grown in small research scale and in large production scale MOCVD reactors. Films were grown at low pressures at temperatures ranging from 350 C to 600 C. X-Ray Fluorescence (XRF) and Auger Electron Spectroscopy (AES) were performed and determined that the film composition is controllable and uniform.
Introduction Sputtering MOCVD Nonvolatility ñ the ability MOCVD Benefits ++ Conformality to retain data in a memory cell -++ Composition Tuning for years when unpowered ñ is + + highly desirable for many Purity electronic systems. The + + Manufacturability dominant nonvolatile memory + Surface Damage technology is FLASH memory, TABLE I: Benefits of MOCVD for chalcogenide production so-called because of the ability to write data individually while erasing them in chunks. FLASH is ubiquitous in todayís cell phones, digital cameras, media cards, and PDAs, among others and is projected to be a $20 billion dollar industry by 2010[1]. But FLASH memory technology suffers from several shortcomings that hamper further device improvement, primarily a lack of scalability beyond the ~65nm node due to increased leakage current through the tunnel oxide. Secondly, FLASH memories can only be reprogrammed a limited number of times, typically on the order of a million. While this may be enough for certain applications, it makes FLASH memory ill-suited for general computing and other applications. As a consequence a number of different nonvolatile memory technologies are emerging as viable alternatives to replace FLASH; most prominent are: Ferroelectric RAM (FRAM or FeRAM), Magnetoresistive RAM (MRAM) and Chalcogenide RAM (C-RAM, but also called Ovonyx Unified Memory (OUM) or Phase-Change RAM (PRAM)). These devices share in common that they can be reprogrammed nearly unlimited number of times and be programmed
in nanoseconds rather than microseconds. While no ìidealî memory device has been developed, C-RAMs have
Data Loading...
