Ferroelectric hafnium oxide for ferroelectric random-access memories and ferroelectric field-effect transistors
- PDF / 1,774,879 Bytes
- 7 Pages / 585 x 783 pts Page_size
- 9 Downloads / 224 Views
Introduction Ferroelectric materials have two nonzero spontaneous polarization states in the absence of an applied electric field. The most frequently used ferroelectric material class are perovskites with the general structure of ABO3, where A and B are cations (e.g., BaTiO3 [BTO] and PbZrxTi1–xO3 [PZT]). In PZT, the B site can be occupied by either a titanium cation or zirconium cation. Perovskites typically undergo a phase transition from a nonferroelectric cubic phase at higher temperatures to a ferroelectric tetragonal phase at lower temperatures. Figure 1a shows a PZT crystal in the tetragonal phase. It becomes clear that the smaller cation (Ti4+ or Zr4+) can have two stable positions resulting in two opposite polarization states. The blue and the green cations in Figure 1 indicate the two stable positions. Please note that the Ti4+ or the Zr4+ will occupy one of the two positions only. Therefore, the polarization of ferroelectrics can be reversed when an external electrical field greater than the coercive field Ec is applied1 (Figure 1a–d). Since the polarization reversal process is purely field driven, without a sufficient applied field, the polarization will remain in the previously set direction; therefore, ferroelectricity is ideal
for low-power binary nonvolatile memory having two stable states that represent “0” and “1” data (Figure 1b–c). All other known emerging nonvolatile memory concepts, such as spin torque transfer magnetic random-access memory or resistive random-access memory, require passing a current through the device. Consequently, there is limited efficiency in the writing process, since not every electron that passes through the structure will contribute to the switching effect.2 The field-driven polarization reversal thus gives ferroelectrics a unique selling point for nonvolatile memories. As early as 1952,3,4 the first attempts were made to realize memories based on the ferroelectric effect in barium titanate crystals. However, to mitigate problematic issues caused by the voltages applied to currently unselected cells that are connected to the same wordline or bitline of the active cells, researchers found that a selector device, which will only be turned on if the cell is operated, needed to be added. This possibility only became available after semiconductor technology reached a certain level of maturity in the 1970s and 1980s. The resulting “1 transistor—1 ferroelectric capacitor” memory (see Figure 2a) reached the market in the early 1990s.5 This success
Thomas Mikolajick, TU Dresden, Germany; and Nanoelectronic Materials Laboratory GmbH, Germany; [email protected] Stefan Slesazeck, Nanoelectronic Materials Laboratory gGmbH, Germany; [email protected] Min Hyuk Park, Nanoelectronic Materials Laboratory gGmbH, Germany; [email protected] Uwe Schroeder, Nanoelectronic Materials Laboratory gGmbH, Germany; [email protected] doi:10.1557/mrs.2018.92
340
• VOLUME 43 • MAY 2018 University • www.mrs.org/bulletin ©available 2018 Materials Downloaded MRS fr
Data Loading...
