Your next Mac will have a faster processor, snappier graphics, and better communication skills, right? And your next iPhone, iPad, and iWhatever will be similarly juiced, eh? Well, yeah--but don’t count on those improvements to continue forever, unless there’s a significant breakthrough in chip-making technology.
We’ve been lulled into complacency by steady microprocessor power bumps, but the age of easy improvements is about to run into a brick wall. The culprit is light--and, no, I don’t mean that said culprit isn’t chubby. I’m talking about the light that’s used in photolithography, the technology that carves your devices’ chips.
The wavelength of light used in today’s chip-carving photolithography--193 nanometers--is frickin’ huge compared to the size of the features it creates. Intel’s “Ivy Bridge” chips, for example--which should have debuted by the time you read this--have 22-nanometer features. Current photolithography, called 193i (the i is for immersion, since it requires a thin layer of water), uses tricks and whiz-bang pirouettes to carve at that infinitesimal size.
Photolithography is used to carve a chip’s building blocks: its transistors, the miniscule on-off switches that handle the digital language that runs your Mac and iOS devices. The smaller those transistors, the less power they require and the more goodness a chip designer can fit onto each wee slice of silicon. The billion or so transistors on an Ivy Bridge chip, for example, give that li’l guy not only multicore computing power, but also graphics, input-output, and memory-controller powers--on-chip integration impossible just a decade or so ago.
That’s all well and good, but will the shrinkage that allows that consolidation continue? Mebbe so, mebbe not.
When chip-carving tech shrinks to around 14 or 11 nanometers, 193i photolithography will be reaching the end of its rope, and no one is quite sure what will replace it. One candidate, extreme ultraviolet lithography (EUV), has been under development for decades, but it remains elusive.
EUV isn’t an evolution of 193i--it’s a whole new ballgame. To take one (greatly simplified) example, 193i’s chip-carving photons are shot from what’s called an ultraviolet excimer laser, and the process can take place in good ol’ air. EUV, on the other hand, creates its beams by firing a high-powered laser at microdroplets of tin to create high-energy plasma, and requires a deep vacuum in which to operate.
EUV’s advantage is that its wavelength is a mere 13.5 nanometers, small enough to give chipmakers some breathing room. Disadvantages include its need for a boatload of energy, its current slowness, and that experimental implementations are still one or two orders of magnitude too weak--so weak that it’s difficult to know exactly how to brew the chemistry of its target surface, what chipmakers call a resist.
While EUV’s developers struggle, a handful of companies are developing an entirely different technology called massively parallel electron beam lithography (EBL). Rather than individual EUV beams, this technology uses an array of independent electron beams directed onto multiple targets by flittering nanoscale mirrors or electromagnetic means to carve the resist. Think of a mind-numbingly precise rainstorm rather than a 13.5-nanometer Super Soaker. EBL has its problems as well, however. Like EUV, it’s currently dog slow, and electron beams have this pesky habit of scattering in the resist, degrading precision.
The bottom line: there are billions of dollars being spent on creating smaller, more-capable chips for future iOS and OS X goodies. Exactly how that will happen is not yet known, but my money is on EUV. If history is any guide, a bunch of übersmart engineers will figure out how to break through the 193i barrier. I hope.
--
Since the late 1980s, Rik Myslewski has paid his rent by keeping an eye on Apple. He was editor-in-chief of MacAddict from 2001 until its transformation into Mac|Life in early 2007, and is now a member of the snarkily sophisticated team at London’s The Register, which is “biting the hand that feeds IT” daily at www.theregister.co.uk.
stengelmedia
April 15, 2012 at 5:27pm
you wrote "Photolithography is used to carve a chip’s building blocks: its transistors, the miniscule on-off switches that handle the digital language" which is not exactly accurate. Photolithography lays the patterns of resist onto the wafer, but does not perform any of the "carving", or as the industry refers to as Etching. The etching process, either Dry Etch (which is plasma based, or Wet Etch, which is chemical based, actually carves out the features.
You are correct in the fact that the feature size limitations are due to the lithography limitations, though are dealt with by one of many different proprietary solutions.
Geoduck
April 11, 2012 at 3:21pm
This article could have been written in '05, or '00, or '95 or '90 or...
And it's been wrong every time.Sure they are reaching the end of what they can do with current photo-lithography technology but that doesn't mean they can't make chips faster. Massively Parallel Processing is the wave of the future. We've gone from single processor CPUs a decade ago to dual core to quad core. I expect over the next decade we'll see personal systems and notebooks with 8, 16, 32, even 64 core processors with a full 'system-on-a-chip'. No the individual cores won't get faster but the chips will do what we want faster, and that's all most people really care about.
