Say good-bye to the node.
For 39 years, the node endured as the elusive and yet universally accepted metric that semiconductor specialists used to indicate how small their transistors were. Like depth readings on a wild descent into the infinitesimal, node figures were plotted out for the near future in a ”road map” released annually by the semiconductor industry associations of Europe, Japan, Korea, Taiwan, and the United States. That map was, and is, a collection of the global semiconductor industry’s best ideas about how it was going to fulfill the Moore’s Law prophecy of a 30 percent shrink in transistor size—and consequent doubling in density—on chips every two years.
But now, in the first tremor of what promises to be a tectonic shift in the semiconductor industry, the node is no more. For decades, makers of logic chips used the concept of the ”node” not only to measure their transistors but also to indicate how advanced their chip-fabrication lines were. Memory chipmakers, meanwhile, used a different measure, halfâ¿¿pitch, for the same purpose. Now lithography, the printing process at the heart of chipmaking, is being pushed to extremes to get to the end of that road map. These extremes will affect different devices on different timescales, but the end of the road looks the same for every device.
We’re pulling out all the stops for the current generation of chips. And if that sounds like a platitude you’ve heard before, consider this fact: Nothing significant that we’re using now will work to create the chips we plan to produce commercially just five or six years from now—least of all the current method of lithography. The next generation of chips won’t be possible without the next generation of lithography. And that, in turn, means that the next generation of lithography will depend critically on what happens to the different chipmakers. For example, memory technology, an industry that sees prices falling at the staggering rate of about 40 to 50 percent per year, faces significant pressure to scale up faster than logic devices do.
Industry observers will not be surprised by the death of the node, as the node and half-pitch, once synonymous, have been diverging for some time. This minor change foreshadows a big change in the way the lithography business will deal with memory versus logic. For the first time, lithography will apparently have to adjust to follow both microprocessors and the different memories—including NAND flash, DRAM, and SRAM—down their respective paths, which have been diverging for decades.
Optical lithography, the most important and technologically demanding aspect of chipmaking, is a pillar that won’t be easily toppled. But the technology is at a critical point. The technique, which uses radiation with about half the wavelength of purple light, is fast approaching steep, if not insurmountable, obstacles. Unfortunately, none of the various technologies proposed over the years to replace it has inspired confidence that it will be ready when the time comes.
Nevertheless, one thing is clear. From now on, the relationship between chips and lithography will be two-way. Not only will the fate of chips depend on the future of lithography, but also the reverse will be true.
Let’s start by defining our terms. Today’s most advanced microprocessors use a 32-nanometer process, and thus are said to be at the 32-nm node. To get a sense of how infinitesimal 32 nm is, consider that to span the width of the lowercase letter l on this page, you would need to bunch together more than 9500 32-nm objects. Node in this context has historically been used to refer to the size of the smallest parts of the transistors on the chips. Until the late 1990s, that was typically a feature called a gate. But there is a very fuzzy relationship between the technology node’s number and the actual dimensions of the gate it purports to signify. In fact, the International Technology Roadmap for Semiconductors, the industry’s guide star, abandoned the term in 2005, but its usage has persisted.
In both logic and memory chips, each of the vast profusion of transistors acts like a switch that allows electrons to flow through the device. A metal-oxide semiconductor field-effect transistor (MOSFET), the kind found on virtually all chips, has three main parts: a source, a drain, and a gate. A voltage applied to that gate lets the electrons flow from source to drain. Physically, the gate sits between the source and the drain.
On a chip, that translates to the distance between the parallel metal lines, called interconnects, that carry the electrons through the chip. These interconnects are stacked today in multiple levels, and as many as 10 can populate a chip (a cutting-edge chip could have 10 kilometers of interconnects). The distance between these metal lines at the first level is called the pitch, and logically, the half-pitch is half that distance.