Phase-Change Technology Enters The Memory Market

Oct. 1, 2009
Phase-change memory (PCM) is a new class of nonvolatile memory technology. This Engineering Essentials article will give you the basics of PCM so that you can consider it for upcoming nonvolatile memory applications.

Phase-change memory (PCM) is a new class of nonvolatile memory technology. Like most new technologies, it offers benefits to those who know where and when to apply it. To understand where PCM fits today and to appreciate its potential value, we need to evaluate its relative cost, reliability, and performance compared to incumbent technologies such as single-level cell (SLC) and multilevel cell (MLC) NAND flash, as well as system solutions including hard-disk drives and solid-state drives (SSDs).

As we examine where PCM fits in the memory landscape, it is important to view it not as a replacement technology for other types of memory, but as a supplemental technology capable of providing key benefits when the system requirements are right. Whether it is supplementing RAM or supplementing NAND flash, PCM can be used in just the right amounts to deliver improvements in reliability and performance in high-end applications such as enterprise computing and e-commerce.

PCM BASICS PCM is a class of nonvolatile memory devices that employ a reversible phase change in materials to store information. Matter can exist in multiple phase states such as solid, liquid, gas, condensate, and plasma. PCM relies on differences in the electrical resistivity exhibited by different phases of a material.

Numonyx is using an alloy of germanium, antimony, and tellurium (Ge2Sb2Te5) called “GST.” In its amorphous phase, the molecular structure of GST is highly disordered, which results in relatively high resistivity. In contrast, the polycrystalline phase of GST has an ordered structure and exhibits low resistivity. PCM is based on the resistivity difference between the two phases of the material.

Engineers can induce the phase change by injecting an electrical current, which causes intense localized Joule heating of the material. The end phase of the material can be modulated by adjusting the amount, voltage, and duration of the application of the applied current.

PCM has some interesting characteristics:

• Like NOR and NAND flash, PCM is a nonvolatile memory technology, so it requires no refresh power for data retention.
• PCM is bit-alterable. Information stored in memory can be switched from a 1 to 0 or a 0 to 1 without a separate erase step.
• PCM features fast random access times. This enables code to execute directly from memory, without an intermediate copy to RAM. The read latency of PCM is comparable to single-bit-per-cell NOR flash, while the read performance can match DRAM.
• PCM can achieve write speeds comparable to NAND flash, but with lower latency since there is no separate erase step needed. By comparison, NOR flash features moderate write speeds but long erase times.

PCM IN THE TECHNOLOGY LIFECYCLE If solid-state drives based on NAND flash technology are currently at an early stage of adoption in many applications, PCM can be found at an even earlier stage of the adoption curve, where the most innovative developers work. Part of my job is to spend a lot of time talking with innovators to understand how they think and how we can help meet their needs.

When they evaluate a technology, these engineers view cost and price as very important. But when they consider how a memory technology can enhance the capabilities of a new system under development, they talk about reliability and performance before they ask questions about cost.

COMPARING PCM TO OTHER MEMORY TECHNOLOGI ES When we compare PCM with DRAM and NAND flash, bear in mind that write and read latency define memory performance, and endurance is a measure of reliability (see the figure):

• Write latency and write endurance: PCM is not quite as good as DRAM, but clearly faster than NAND.
• Read latency and read endurance: PCM approaches the speed of DRAM and is clearly better than NAND.
• Cost: A theoretical die cost comparison (300- mm wafer) of SLC PCM, DRAM, and SLC NAND flash shows that NAND is the cheapest. PLC is about 1.2 times the cost of NAND, and DRAM comes in at about 1.4 times the cost of NAND. So the cost of PCM is clearly better than DRAM but not nearly as cheap as NAND, but the total process mask count of many of these technologies is beginning to converge.

Compared to SLC NAND, DRAM costs about 40% more and PCM costs about 20% more. When considering a new memory technology, the cost of the replacement technology should be eight to 10 times lower than the incumbent technology it replaces. So it is clear that from a pure cost standpoint, PCM cannot be justified as a replacement for NAND flash.

RELIABILITY: WHERE PCM CHANGES THE GAME Most of us own multiple devices such as MP3 players and USB memory devices based on NAND flash technology. In normal use, these applications require writing to memory 10, 50, or even 100 times, but never thousands of times. The NAND flash technology that has been shrinkwrapped around these applications has consequently driven the cost of the technology to extremely low levels through the use of leading-edge lithography.

Higher-end applications, such as wireless communications, computing devices, and solid-state drives, may also use NAND flash, but they impose quite different usage conditions and requirements. A technology that may be perfect for an MP3 player is going to impose limitations in an enterprise-class server.

Continue on Page 2

With any nonvolatile floating-gate memory device, the more we cycle the device, the more failures we tend to observe, and the less data retention we get. What is interesting about PCM is that retention is decoupled from endurance. This means that if we cycle a PCM memory device a million times or just one time, the data retention will be identical.

That characteristic requires a fundamental change in how we think about using and managing the technology. Data retention is clearly a critical requirement in many demanding applications. At Numonyx, we have demonstrated 10-year data retention in PCM devices.

We have observed another phenomenon that makes it easier to use PCM in system-level designs. When we see a failure, it always occurs during a write. So if we are writing data to the device and a write-verify shows that the data is not there, the data can be immediately written again to another location. Because of the exceptional reliability attributes of PCM, we will see this technology adopted first in applications with the most critical requirements.

LOW-POWER MEMORY FOR WIRELESS We are in the process of bringing up a new generation of smart-phone users who demand three essential qualities: instanton, ease of use with long battery life, and great performance with multimedia, games, and Internet computing applications.

PCM’s excellent read latency means it can execute code with the performance that the wireless generation demands. It also enables the storage of large amounts of code and data in memory, with the ability to sustain millions of read-write cycles. And we should not forget that PCM is nonvolatile memory technology, so it can simplify power management in wireless devices while helping to eliminate nagging performance/battery-life tradeoffs. Microprocessor cores with LPDDR2 memory controllers will enable system designers to benefit from the low power of PCM, using a single nonvolatile memory device to simplify cell-phone architecture.

PCM AND PERFORMANCE We can use memory technology to improve computing performance in two basic ways: by adding bandwidth to move more data, and by reducing latency to move the data faster. In high-end applications where performance is most critical, adding to bandwidth involves adding more RAM, hard drives, and servers.

To achieve latency improvements we can replace hard drives, which work in milliseconds, with SSDs, which respond in microseconds. This approach tends to improve total cost of ownership, because it is more costeffective to add a single low-latency server than to add three or four servers.

When we compare write latency, PCM, at about 1 µs, is typically about 100 times faster than SLC NAND flash and approximately 100,000 times faster than a hard disk. The same advantages hold true when we look at read latency. At 50 to 100 ns, PCM is about 100 times faster than SLC NAND and about 100,000 times faster than a hard disk.

The fundamental value of a technology will determine the price users are willing to pay for it. In the world of e-commerce, even modest improvements in latency can be extremely valuable. According to Amazon, every 100 ms of added latency can cost 1% in sales. According to Google, an extra 500 ms in search page generation time can drop traffic by 20%. And, the analyst firm Tabb Group reports that a broker could lose $4 million per millisecond if its electronic trading platform is just 5 ms behind the competition.

Is there value in putting PCM into a system between RAM and SSD? I think so, because placing low-latency PCM technology in high-end systems can deliver incredible performance gains. We are now working with customers to implement PCM solutions that deliver value by supplementing other memory technologies within server systems.


To join the conversation, and become an exclusive member of Electronic Design, create an account today!