Memristor Device Engineering and CMOS Integration for Reconfigurable Logic Applications
This chapter reviews the application of memristive devices as reconfigurable switches. First, the fundamental performance metrics (power consumption, endurance, switch speed, IV nonlinearity, CMOS compatibility, ON/OFF ratio, etc.) of memristors as switch
- PDF / 957,469 Bytes
- 25 Pages / 439.36 x 666.15 pts Page_size
- 99 Downloads / 206 Views
Memristor Device Engineering and CMOS Integration for Reconfigurable Logic Applications Qiangfei Xia
11.1 Introduction In the past few decades the integrated circuits (IC) industry has successfully followed Moore’s Law in delivering more and more powerful computer chips with reduced cost per transistor [1]. However, CMOS (complementary metal oxide semiconductors) scaling is approaching a physical and economical limit. On the one hand, it becomes more and more challenging and expensive to build smaller transistors that are packed into a very small area. On the other hand, the leakage current associated with smaller transistors will deteriorate or even destroy the device. To sustain rapid progress in information technology, there are intensive efforts to go beyond Moore’s Law in research areas including new devices/materials, new technologies, and new architectures and algorithms. The memristor (memristive device) emerges as one of the most promising devices for the post-CMOS era. As a nonvolatile, two-terminal electronic device, the memristor has variable resistance that changes with the polarity and amplitude of the applied voltage [2–4]. High and low resistance states instead of charge storage are used to represent the logic “1” and “0” in these devices. The typical structure of a memristor consists of a layer of switching material sandwiched between two electrodes. With a cross-point structure, these devices offer great scalability since the junction area is dependent solely on the width of the two nanowires. Materials that are widely used for the switching layer include binary or ternary transition metal oxides, perovskites, and solid-state electrolytes [5]. Depending on the device switching mechanism, the electrode materials are usually inert metals such as Pt, W, or active metals such as Ag, Cu, etc.
Q. Xia () Nanodevices and Integrated Systems Laboratory, Department of Electrical and Computer Engineering, University of Massachusetts, Amherst, MA 01003, USA e-mail: [email protected] R. Tetzlaff (ed.), Memristors and Memristive Systems, DOI 10.1007/978-1-4614-9068-5__11, © Springer Science+Business Media New York 2014
327
328
Q. Xia
Although the fundamental physics for the switching mechanism is not thoroughly understood, there has been significant progress in memristive device research and development. Particularly, superior device performance has been achieved, such as high endurance higher than 1012 cycles [6], faster than 1 ns switching speed [7], sub-100 fJ energy consumption per switching event [8] and an extrapolated date retention time longer than 10 years at room temperature [9]. Because of the achievement in device performance, memristors have been proposed and demonstrated for applications in nonvolatile memory [10–12], reconfigurable switches [13], nonvolatile logic [14], and bio-inspired neuromorphic computing [15–17]. This section focuses on the memristor device engineering and integration for reconfigurable logic circuits applications. First, the requirements in particular device perform
Data Loading...
