By Richard K. Williams
CEO and CTO
Advanced Analogic Technologies Inc.

Innovation involves two unrelated activities of the conscious mind that occur in parallel. First, innovators gather in­formation and synthesize new ideas from it without any specific purpose in mind. Then they identify un­solved problems, without any preconception as to whether a workable solution exists. These two idea buckets may coexist in the mind for years. Then somehow, in the subconscious, an innovator realizes that a creative idea solves a lingering problem. This is the "Aha!" moment.

Here's an example. For many years, I was frustrated by the nagging problem of applying IC technology to the needs of power electronics, so as to intelligently create power-management circuits in the most versatile and reusable manner. The problem was that the highest-voltage device on the IC determined the layer thicknesses and manufacturing flow, which in turn influenced the electrical and mask design rules used to manage the layout and circuit design of the product.

Once a chip was designed for a certain process—for example, a 20-volt pro­cess integrating bipolars, CMOS and power DMOS transistors (also known as a BCD20 technology)—changing the design for use at, say, 5 V required a complete redesign, even of the 5-V circuitry. Companies were forced to maintain whole libraries of circuits useful for only a single, specific voltage process. Without voltage specificity, a library of reusable 5-V circuits would be needed for common functions, independent of the high-voltage portions of the integrated circuit.

Meanwhile, I had been studying the basis of high-volume CMOS manufacturing, crucial to achieving high electrical yields in advanced DRAMs. I gradually be­came aware of DRAM companies' almost religious adherence to the principles of low-temperature wafer fabrication and fully self-aligned device structures—pro­cesses that did not depend on any critical alignments between photo­masks and wafers. Their command of multilayer metal interconnect, wafer planarization and trench etching using specialized processing equipment in ultraclean fab facilities was impressive, but initially appeared irrelevant to the challenges in my field.

In the fall of 1998, while sitting in a coffee shop, the Aha! moment came. I realized the DRAM equipment set could be used in a way for which it was never intended—for an entirely low-temperature manufacturing method to produce power devices and power-management ICs.

It quickly became evident that high-temperature fabrication was actually the underlying origin of the voltage-specificity problem in power ICs. By eliminating all high-temperature processes—epitaxy, diffusions, oxidations—a voltage-independent, yet fully isolated, BCD technology could be realized in an advanced low-cost DRAM fab. Furthermore, by architecting the process into voltage-specific modules, mixed-voltage integrated circuits with noninteracting, reusable 3-V and 5-V circuitry could be achieved in a modular design methodology. With that observation, AnalogicTech's novel and patented Modular­BCD process technology was born.