In 2005, Moore’s Law was approaching a wall. To continue doubling the transistor capacity of chips in accordance with Gordon Moore’s principle, the transistors in the next generation of microchip would have to shrink to a mere 45 nanometers or 45 billionths of a meter — about 30 percent smaller than the 65nm chips that were then state-of-the-art. But transistors that small and closely packed would exceed the inherent physical limitations of the silicon dioxide dielectrics that had been the industry-wide standard material since the late 1960s.
Intel responded by spending two years developing alternative materials that Moore, one of Intel’s founders, called “the biggest change in transistor technology since the introduction of polysilicon gate MOS transistors in the late 1960s.” Using the chemical element hafnium, Intel created a new kind of dielectric that it called “high-k,” which it then paired with a new alloy for the transistor gate. The New York Times summarized the significance of the new materials: “Intel … has overhauled the basic building block of the information age.”
The company announced the breakthrough in January 2007. (Commercial chips using the technology became available the following November.) Developed at Fab D1D in Oregon, the first 45nm prototype products using the new technology featured double the transistors of the previous generation with only a fraction of the power loss and 20 percent better transistor performance. Moreover, the new alloys were 100 percent lead-free, and by 2008 would be halogen-free as well.
The company’s commitment to having a manufacturing infrastructure in place to take advantage of the new process was as important as the technical breakthrough itself. Intel unveiled plans for a $3.5 billion high-volume 45nm fab facility — Fab 32 in Chandler, Arizona — several months before Intel had even shipped 65nm processors, and 18 months before the company announced its first successful 45nm prototype in July 2005. A complementary $3.5 billion facility in Kiryat Gat, Israel, was announced in December 2005. (Fab 11X in Rio Rancho, New Mexico, would be retooled for 45nm production following a $1.5 billion refitting announced in March 2007 — after the technology had been proven.)
These investments were indicative of the company’s confidence that it would achieve the technical breakthroughs to make use of the new facilities, and that confidence paid off. Intel wound up positioned not only to take advantage of its 45nm process but to continue doubling the transistor capacity of its microchips for generations to come.
This story is among a series running to celebrate Intel’s 50th anniversary in 2018.