Papers

AVS2001 Session MI+EL-MoM: Magnetic Devices

Monday, October 29, 2001 9:40 AM in Room 110

Monday Morning

Start	Invited?	Item
9:40 AM		MI+EL-MoM-1 High Speed 256k Magnetoresistive RAM S. Tehrani (Motorola Corporate) Magnetoresistive Random Access Memory (MRAM) has the potential to be a high-speed, low-voltage, high-density, nonvolatile solid state memory. MRAM is based on magnetic memory elements integrated in a backend process with standard CMOS semiconductor circuitry. Key attributes of MRAM technology are nonvolatility and unlimited read/write endurance. Our bit architecture is based on a minimum-sized active transistor as the isolation device in conjunction with a magnetic tunnel junction element (MTJ) defining the MRAM bit. Our MTJ material stack is composed of two magnetic layers separated by a thin dielectric barrier with the polarization of one of the magnetic layers pinned in a fixed direction. The resistance of the memory bit is either low or high dependent on the relative polarization, parallel or anti-parallel, of the free layer with respect to the pinned layer. In this talk we will summarize our progress on MRAM based on MTJ integrated with CMOS circuitry. We have demonstrated MTJ material in the 10 kOhms-um^2 range with MR values up to 50 %. The MRAM module is inserted in the back-end-of-line (BEOL) interconnect using four additional lithography steps. The source and isolation are shared between neighboring cells to minimize cell area. In this particular architecture, the cell size is 7.2um^2, corresponding to 9f^2, where f is one-half the metal pitch. We have developed a 256kb (16k x 16) MRAM memory based on 0.6 um CMOS with a 1T1MTJ (one transistor and one MTJ) cell. Nonvolatile data storage and read cycle times of 35 ns have been demonstrated. Read power consumption at 3.0V and 20MHz is about 24mW. These results show that MRAM based on MTJ has the potential to be a competitive memory with the attributes of high-speed read and write, as well as nonvolatility. The progress, potential and challenges of MRAM technology will be discussed. This work in funded in part by DARPA.
10:20 AM		MI+EL-MoM-3 Dry Etching of MRAM Device Structures R.A. Ditizio, G. Beique (Tegal Corporation) Magnetic Random Access memory (MRAM) has experienced a rise in interest in recent years as an alternative to other non-volatile memory devices. As efforts continue to improve the electrical performance of these devices, parallel efforts are underway to meet the stringent requirements for the fabrication of MRAM device structures at high densities. In this report, recent efforts that have been undertaken to apply conventional etch practices to the unique requirements for the patterning of MRAM device structures in an inductively coupled plasma source are discussed. In particular, improvements in optical emission detection and the subsequent correlation of endpoint traces to films in the device structure will be presented. Correlation of the film structure to the optical emission trace is necessary, for example, as a means to identify the specific time in the etch process at which to stop on the thin insulating layer across which a magnetic tunneling junction might typically be formed. Post-etch corrosion control of the completed device structure using an integrated rinse module is also discussed.
10:40 AM		MI+EL-MoM-4 Spin Dependent Tunneling Devices for Nonvolatile Latch Memory M. Tondra, D. Wang, D.J. Brownell, Z. Qian, C. Nordman, J. Daughton (NVE Corp.) Operation of integrated magnetoresistive nonvolatile latch cell memory using spin dependent tunneling (SDT) junctions has been demonstrated. These SDT devices were fabricated on top of commercially processed CMOS silicon circuit wafers. Fabrication of these devices presents many challenges to thin film deposition process developers. Process temperature compatibility and surface roughness are prime examples. In spite of these and other technical challenges, there is significant motivation to continue developing SDT fabrication processes. In particular, SDT devices provide resistance changes on the order of 50% (large signal), a wide range of resistance values (for low power applications), and magnetic switching speeds beyond 1 GHz. Furthermore, the SDT cell density is potentially competitive with commercial SRAM and DRAM. Recent success in SDT integrated device fabrication has been a result of using new approaches to surface preparation. Specifically, chemical mechanical polishing (CMP) has been employed to create a sufficiently smooth surface for SDT deposition. In-process atomic force microscopy (AFM) measurements suggest that a pre-deposition RMS substrate roughness of 0.2 nm is sufficient to allow successful SDT fabrication. This paper will discuss device-specific process details of SDT latch cells and their impact on the potential for near-term commercialization.
11:20 AM		MI+EL-MoM-6 Effects of Interfacial Electronic States and Roughness on Tunnel Magnetoresistance J. Inoue, H. Itoh (Nagoya University, Japan) Recently, active researches on tunnel magnetoresistance (TMR) are under progress with the objects of its technical applications such as magnetic sensors and magnetic random access memories. In spite of these studies, the understanding of the electronic states at the interfaces of the ferromagnetic tunnel junctions and of the effects of roughness on the tunnel conductance and TMR is not still complete. Quite recently, numerical results on the TMR in the first principles band calculations have been reported for junctions with clean interfaces. The calculated results show that the contribution of the states with certain wave vectors parallel to the interface becomes dominant, which are known as hot spots. We calculate the dependence of the tunnel conductance and TMR on the barrier thickness including the interface roughness, and show that the contribution of the hot spots to the tunnel conductance is reduced by the roughness. We further argue the possibility of appearance of interfacial state due to amorphous-like barrier structure and its effects on the TMR.
11:40 AM		MI+EL-MoM-7 Performance of the BARC Magnetoresistive Sensor* J.C. Rife, R.J. Colton, M. Miller (Naval Research Laboratory); M.A. Piani (Nova Research, Inc.); C.R. Tamanaha (Geo-Centers, Inc.); P.E. Sheehan, L.J. Whitman (Naval Research Laboratory) The Bead ARray Counter (BARC) is a microfabricated chip for quantitatively detecting and identifying biological molecules using giant magnetoresistive (GMR) sensors.¹ The assay is based on highly selective biomolecular binding to the surfaces of numerous individually addressable GMR sensors, followed by labeling of captured molecules with magnetic beads. An externally applied AC magnetic field magnetizes the beads and a lock-in amplifier detects changes of 10^-⁷ in resistance of the GMR sensors, limited by Johnson and 1/f noise. Overall sensitivity is a convolution of chemical and magnetic/electronic sensitivities. Our current sensors can determine target concentrations from 10 fM to 1 nM with loading of one hundred to more than a thousand beads. In principle, each sensor could detect one bead/one captured molecule. Present electronic sensitivity is restricted, in part, by the properties of the commercial 2.8 µm diameter composite polymer/ferrimagnetic beads that result in signal levels a factor of ten below the electronic noise floor. We find 10 to 100x improved signal with solid, soft ferromagnetic beads of the same size that yield the theoretical susceptibility of solid magnetic spheres, but chemical functionalization of the surfaces is not yet resolved. We have measured bead signals versus magnetizing field to have an approximately square-law dependence determined by the magnetoresistance response curve. We have also measured the bead signal versus position across the 2 µm wide GMR sensor and generated a simple model of the local resistivity change. Finally, We have developed an overall model for the GMR sensor response that agrees in large part with the measurements. The model should enable sensor and magnetic physical design to be optimized for maximal chemical and electronic sensitivity. *Supported by the Defense Advanced Research Projects Agency. ¹ Edelstein et al., Biosensors & Bioelectronics 14, 805 (2000).