April 21, 2009, 9:59 AM — A Taiwanese research group has turned to RRAM (Resistive-RAM) as the latest possible Holy Grail of memory chips, one that can replace both DRAM and NAND flash memory.
DRAM has been the main memory type used in computers for decades and is valued for its ability to handle data at high speeds. NAND flash memory is newer but its market has grown fast because of the large amounts of songs, pictures and other data it stores in iPods, iPhones, digital cameras and other products.
Researchers at Taiwan's publicly funded Industrial Technology Research Institute (ITRI) believe RRAM shows enough promise that it could be ready for the embedded chip market within the next few years.
"We're still in the early stage of development," said Tsai Ming-jinn, research director of the Nanoelectronic Technology Division at ITRI.
"Right now we cannot compete with DRAM on reliability," he added.
Most memory chip research initiatives focus on dethroning either DRAM or NAND flash memory because they hold sway over such huge markets. The DRAM market alone was worth nearly US$24 billion last year, according to market researcher iSuppli.
Most attempts to beat these two memory chips often fail.
Phase-change memory (PRAM), for example, is one kind of memory chip that ITRI plans to reduce its research efforts on this year.
Despite early promise, ITRI has found PRAM difficult to manufacture. The research group plans to finish a few PRAM-related research projects that run through the end of this year, but then phase them out of its focus.
Researchers often run into hurdles on the way to developing a new technology, but the bar for a DRAM or NAND replacement is even higher. DRAM was invented by IBM decades ago and perfected over the years by other companies and research groups. The chips dominate PCs because of the high speeds they're able to handle data, but they do have a drawback. Once the power is off, DRAM chips forget all their data.
NAND flash memory is different in that it can hold a large amount of data whether a device is on or off, but it runs too slow to replace DRAM.
Beating these chips on their technological merit is one issue. The second hurdle will be making a new chip so much better than DRAM or NAND that it actually makes sense to replace the chips.
That won't be easy considering the billions of dollars companies have spent to build DRAM and NAND factories and the investment PC component makers made to ensure their components, from microprocessors to motherboards, work with DRAM.
"The entry barrier is very high," Tsai said.
That's why he hopes RRAM can find an entry into electronics devices in the embedded chip market.
It's easier for memory chips to compete in the embedded market, where chip makers combine the functions of several chips onto one, called an SoC (System on chip), because there is no reason for companies to depend only on DRAM.
RRAM is also good for smart cards and SIM cards used in mobile phones.
The chips are speedy like DRAM but unlike DRAM, they retain data when a device's power is shut off.
But ITRI has a long way to go with RRAM.
The group has already produced 1Kbit prototype chips and successfully manufactured the chips on 8-inch wafers, two key steps along the development path. It will still take years to create chips with enough storage capacity to entice even the embedded chip market.
ITRI is currently in talks with a few startups about working on the chips, but they are not Taiwanese companies, which is a problem. ITRI is sponsored by the Taiwan government, and is required by its mandate to first give local companies a shot at a partnership on any new technologies it develops.
Tsai said the group will continue to work on the chips as it talks with various groups. The chips won't be ready for market for a few years.