The future of ferroelectric memories

Abstract

Since 1984, ferroelectric RAMs (FeRAMs) have been demonstrated in many applications such as smart cards and low-density memories. Prior to 1984, attempts failed because of the poor quality of thin films of complex materials. Currently, two materials compete for the large-scale integration development of FeRAMs. The first is a perovskite ceramic known as PZT (PbZr/sub 1-x/Ti/sub x/O/sub 3/). The second material is known as a layered perovskite such as SBT (SrBi/sub 2/Ta/sub 2/O/sub 9/). For low-density devices which employ thin films of either material with a thickness <300 nm operated at 3-5 V, both materials yield approximately the same results. As FeRAMs enter the deep submicron realm, the ferroelectric thin-film technology is ready to support high-density integration. SBT-based devices can be integrated as capacitors in DRAM-like 1T/1C stacked cells and flash-like FeFET cells. Experience with embedded FeRAMs is positive, so that the system-on-chip as well as stand-alone high-density devices are foreseen. The possibility of 1 V operation at a few to several tens of nanoseconds write with nonvolatility brings FeRAMs to the forefront of non-volatile memories. Scaling of capacitor areas as small as 0.04 /spl mu/m/sup 2/ is possible. With capacitor and FET technologies, FeRAMs blur the line between non-volatile memories as DRAM-like destructive read-out (DRO) devices and flash-like non-destructive read-out (NDRO) devices, which compete for the highly mobile generation of Internet devices and G-3 phones.