4 January 2010—Though flexible devices such as roll-up displays have been promised for several years, their commercialization has been stalled by a missing ingredient: a flexible form of flash memory. But researchers at the University of Tokyo have recently developed an organic, floating-gate nonvolatile memory that behaves like flash memory, which may solve that problem.
While silicon-based flash memory is fine for the mass data storage found in cellphones, digital music players, and thumb drives, fabricating it requires high processing temperatures, thus ruling out its production on flexible substrates like plastic. Organic semiconductors, however, can be processed at temperatures well below the melting point of most plastics.
What's more, "the cost of flash memory is too high to use in applications that require large arrays of memory," says Tsuyoshi Sekitani, an assistant professor in the University of Tokyo's department of electrical and electronic engineering and one of the researchers who developed the new memory. "But we can print our organic memory on flexible substrates and over large areas using inkjet printers. So costs will be low."
He says that most memory employing organic semiconductors uses ferroelectric polymer film for its memory cells. This type of memory stores bits in an electric field in the polymer film. It typically employs rigid substrates and dielectrics between 100 and 500 nanometers thick, requiring high voltages—from 10 to 50 volts—for programming and erasing.
Given such drawbacks, the researchers chose to work with new dielectric materials that had the right electrical properties even in very thin layers. They produced a hybrid dielectric of metal oxide grown from a plasma and a self-assembled monolayer (SAM) of aliphatic phosphonic acid. SAMs arrange themselves into 1-molecule-thick layers on surfaces. The combined dielectric film was just 6 nm thick.
Like ordinary flash memory, the flexible flash transistor has a "floating gate" embedded inside the dielectric and a control gate atop the dielectric. When data is being written to the cell, charge accumulates in the floating gate. Flushing out the charge erases the bit.
The thin dielectric layers enabled the researchers to produce floating-gate transistors with a program and erase voltage of just –6 V and +3 V, respectively. They used a layer of pentacene laid down on a 125-micrometer-thick plastic substrate as the semiconductor material.
The researchers tested the durability of the dielectric with up to 6 V on the gate—sufficient for unambiguous readouts—10 000 times without any sign of irreversible damage.
"This shows that flexible devices could be operated directly by a battery of just 6 V," says Sekitani. "Imagine a portable sheet-type scanner, for instance, that could retain scanned data without the need for a computer."
But the researchers still face the major challenge of retaining data in the memory device for some practical length of time, as the charge from the floating gate is prone to leak through the dielectric. In a one-transistor memory cell they fabricated, data retention lasted only a few hours, with most of the charge loss occurring through readout operations.
So the researchers designed a two-transistor memory cell in which an "access" transistor and a readout transistor share a common floating gate measuring 400 by 1000 μm2 in area. The access transistor is active only when the memory cell is being written to or erased. Unlike the access transistor, the readout transistor operates without a control gate, cutting off a potential path for charge to leak out of and boosting retention time to between 12 and 24 hours.
"Retention time is determined by the length of the molecule of the gate dielectric," says Sekitani. "If we can synthesize a different material of longer length, we can improve data retention." Further miniaturization of the device "will also increase retention," he adds, "so we have two strategies to [tackle] this problem."
To demonstrate the device, the researchers fabricated a large-area flexible sensor with an effective area of 50 by 50 millimeters to show where mechanical pressure had been applied after the pressure was removed and the current switched off. The sensor was fabricated on three flexible sheets of polyethylene naphthalate (PEN), containing 676 two-transistor memory cells arranged in a 26-by-26 array. On top of the memory array was a pressure-sensitive rubber sheet, and on top was a PEN sheet with copper electrodes.
Applying pressure on the structure decreases the electrical resistance through the rubber between the electrode layer and the memory cells. That resistance difference at each point on the sheet is stored as charge in the floating gates of the memory cells beneath that point. Though memory retention decreased over time, the researchers could still tell where pressure had been applied, even after 12 hours.
Sekitani says this principle could be used to produce a Braille display for the blind that would enable them to read books and other information. He also offers a more futuristic application: "security flooring." This would read the footsteps of people not entered into a household database and so would signal the presence of strangers to a security company, eliminating the need for intrusive cameras.
More significantly is the role that flexible flash memory could play in helping bring a variety of flexible products to market. "Flexibility is important for a number of reasons," says Sekitani. "Flexible devices are more resistant to shock, you can carry them in your pocket, and large devices can be folded up, making them more convenient." He estimates that the research team will be able to perfect its flexible flash memory within five years.
About the Author
John Boyd writes about science and technology from Japan. In November 2009, he reported on the technology behind the winning solar car in the 2009 Global Green Challenge.