Making the shift from floating-gate to phase-change in non-volatile memory

In the coming years, portable systems will demand even more NVM with high density and very high writing throughput for data storage application, or with fast random access for code execution.

While continued research on floating gate techniques should extend the current flash technology capability through the end of this decade, interest is increasing in new memory storage mechanisms and materials that have promise for scaling through at least the end of the next decade.

Among the different NVMs based on storage mechanisms alternative to the floating-gate concept, phase-change memory (PCM) is one of the most promising candidates, having the potential to improve the performance compared with flash as well as to be scalable beyond flash technology.

Some alloys based on the group VI elements (referred as chalcogenides) have the interesting characteristic of stability at room temperature both in the amorphous and in the crystalline phase. In particular, the most promising are the GeSbTe alloys, which follow a pseudobinary composition (between GeTe and Sb2Te3), referred to as GST.

In the silicon-based PCM, electric current of different magnitudes are passed from a heater element through the chalcogenide material; local Joule heating is used to change the programmable volume around the contact region (Figure 1 below).

Figure 1: Shown is a schematic cross-section of a PCM storage element (left-side) and the simulated temperature profile during the writing operation (right-side).

High current and fast quenching freeze the material into an amorphous state giving a high resistance. The time required for switching to the amorphous state is typically less than 100ns, and the thermal time constant of the cell structure is typically only a few nanoseconds. Medium current for longer pulse time is used to re-crystallize the region to a crystalline state, which has a low resistance.

The different current level to program the memory cell gives the direct write characteristics of the memory. This direct write capability simplifies the writing to and improves the performance of the memory. A much lower current with essentially no Joule heating is used for reading the memory, differentiating between the high (amorphous) and low (crystalline) resistance states.

PCMs are interesting for two primary reasons. The first is enhanced functionality: this includes faster random access time, read throughput and write throughput, as well as other features including direct write, bit granularity and high endurance.

The merger of some of the properties of today's flash and DRAM provides a new level of functionality that can result in not only replacing flash but also replacing some usages of DRAM, such as storing frequently used operating code and high-performance disk caching (Figure 2 below).

Figure 2: Technology attribute comparison between PCM, DRAM, NOR flash and NAND flash.

The second reason for interest in PCM is the small size of its memory element and its scalability. The phase-change physics shows promise to be scalable to dimensions of below 5nm, providing the opportunity to continue the rate of cost reduction and density increase established by flash memory well into the next decade.

The integration of the PCM concept, cell structure, array and product vehicle in a standard CMOS technology has been widely assessed and demonstrated. High-density, 128Mbit PCM prototypes have been demonstrated at 90nm, showing good performance and reliability.

The process integration results obtained so far and the level of comprehension of the PCM integration details allow for moving the next development steps into the Gbit-level device manufacturing at scaled technology.

Roberto Bez is a Numonyx Fellow in the R&D Technology Development at Numonyx B.V.