Memristors combine memory, logics set in one device

Tagged: Memristors, Technology
Source: Ars Technica - Read the full article
Posted: 4 years 23 weeks ago

Memristors, a type of circuit element based on magnetic flux, were first proposed back in 1971, but remained limited to the realm of theory until 2008. That's when some researchers from HP Labs figured out that memristors could be read and written using applied voltages, and didn't need to explicitly involve magnetic flux at all. Since then, the group has continued to develop memristor production, hoping to find a place for them alongside the existing capacitor, inductor, and resistor devices. Now, in their latest work, the researchers have shown that it's possible to simultaneously use memristors to perform a full set of logic operations at the same time as they function as nonvolatile memory.

Our past coverage goes into the actual physics of how a memristor functions, so we'll just summarize the basic operations here. The devices can adopt either high- or low-resistance states, which can be considered bits. A positive voltage above a specific threshold will set the memristor in its high-resistance state in as little as two microseconds. A negative voltage of the same magnitude toggles it to its low-resistance state.

A key feature of memristors is that these states are stable—they act as nonvolatile memory. Reading their state can be accomplished by applying voltages smaller than the critical threshold, and determining how much current flows through.

Although these devices are currently difficult to manufacture, they have some distinct advantages. For example, earlier work showed that, by amplifying the current from one memristor, it's possible to have its state set the state of others, producing devices that could potentially reprogram themselves in a manner that depends on the evaluation of other logic operations. This new work expands on that by showing that a memristor-based device can perform the complete set of logic operations.

The work is based on the existing knowledge that the full set of logic operations can be performed using a combination of NAND (not-and) gates. The new work demonstrates that it's possible to build a NAND gate using a combination of three memristors, but only if you use a frequently overlooked logical operation called "material implication." As the authors describe it, for Boolean states p and q, a material implication is "p implies q"—if p is true, then q must also be.

The work shows that it's possible to build an IMP logic gate using two memristors combined with a standard resistor; add a further memristor that acts as a false operation (it always returns false), and you get a complete set of logic operations.

Most of the work involves a collection of voltage traces that show the device operates as expected, keeping its state and providing the expected output, so the data itself isn't especially exciting. What's exciting about the work is that it's a demonstration that the unique properties of memristors can be harnessed to create a nonvolatile and programmable logic gate.

Still, the memristors aren't especially useful in isolation at the moment, as the authors had to rely on standard silicon circuitry to control them. They also don't expect that to change overnight, which has implications for how we might see the first deployments of memristors. "Applications of this technology will most likely require substantial parallel operations," the authors write, "in order to amortise any silicon-based driving circuitry."

The authors also point out that the use of implication is quite common among those who study formal logic, but is largely ignored by those designing logic schemes, who focus on what's easy to do in silicon. "The major lesson from this research," they write, "is that when confronted with a new device, one needs to determine whether it has a natural basis for computation that is different from familiar paradigms."