Memristors are a seriously hot topic at the moment--we've seen several announcements about these tiny slivers of semiconductor which are the future of electronics, and now HP's got news too. Their memristors will beat flash memory, apparently.
For those of you who've no idea what a memristor is, here's a quick recap ('cause they're going to be so important in the future). The name is a bit of a giveaway--they're kind of "memory transitor/resistors." A transistor is a tiny electrical valve with a variable control over the electrons flowing through it, whereas a resistor is just a constriction in the system, if you like, that limits the flow of electrons past it. Each is a valuable piece of pretty much every modern electronic system. But neither can "remember" what it just did--once a transistor has been thrown open, all previous action is "forgotten." A memristor is different--it too is an electronic valve, but due to the tricks of its design it can, for example, remember how long it's been switched to "open." This tiny difference opens up possibilities for all sorts of electronic complexity (even approaching the way neurons work in our brains) that would otherwise be impossible.
HP's particular memristor magic may be the most accessible so far. According to HP's research team, they've improved the speed at which the memristors they produce can switch from on to off--up to a level that matches today's transistors. That makes the possibility of high-speed memristor circuitry seem even sooner than may have been thought. Better yet, they think that inside three years they'll have memristor-based circuitry that's "better than the competitors."
The competition in this particular space is flash-based memory circuitry. Memristor chips would be better because they can actually be produced on a smaller scale--current chip fabs are based on a 30 to 40nm scale, whereas HP can make reliable fast-switching memristors on a 3nm scale. HP's science team has also discovered how to stack memristors in multiple layers. The practical result is that memristor memory chips could be smaller than flash-based ones. This means two things: Potentially much increased storage capacity in portable devices or memory cards, perhaps by a factor of ten, and significantly lower memory prices (as more memristor chips could be cut from one production wafer--one big deciding factor in chip prices.)
HP's tech is based on slightly dislocating atoms in a vanishingly-thin layer of titanium dioxide, and has the added bonus of requiring extremely low power. For all these reasons and more, expect to hear much more about these tiny devices as time goes by.
To keep up with this news and more like it, follow me, Kit Eaton, on Twitter. That QR code on the left will take you to my Twitter feed too.