Printing Your Own Semiconductors

Much of the look and feel of today's PCs, tablets and smartphones was developed at PARC (Xerox's Palo Alto Research Center).  Is PARC now on the track of an innovation that could revolutionize semiconductor manufacturing?

With financing from DARPA (Defense Advanced Research Projects), PARC has developed a method of programmable electrostatic assembly, inspired by xerography.  Tiny chiplets are produced and mixed into a solution-based “ink.”  The chiplets, which have an electrical charge pattern on them, are subjected to dynamic electric fields which are used to orient and position them with micron-level accuracy.  Once assembled, the chips are put onto a final substrate with interconnects.  The technology used is similar to laser printer technology, which is essentially the assembly of large numbers of micron-sized toner particles, a Xerox development from the 1970s.    The resulting circuits can be microprocessors, memory or any other desired semiconductor chip.  Although production quantities will not be feasible for several years, could this be leading to desktop printing of almost any semiconductor?

The semiconductor industry needs this.  The depreciation costs for today's leading edge fabs are so high that the only way to cover them is to manufacture millions of identical ICs.  But, there are only a few kinds of advanced ICs needed in those kinds of quantities, primarily processors, DRAM, and NAND flash.  The demand for  other semiconductors is for smaller volumes, or for ICs that do not need to be manufactured on leading edge technology.  Many could be manufactured using four or five generations- old technology. 

The semiconductor industry has addressed this problem in several creative ways.  Multi-project wafer services combine different ICs on one wafer.  Multi-wafer lots combine different wafers in one processing lot.  But, each of these solutions imposes constraints.  A process optimized for memory chip production is not well suited for logic chip production.  And, even a logic chip process cannot be optimized for a particular IC.  Every IC must be designed to perform satisfactorily on one given set of process parameters. 

Suppose that wafers could be processed in a desk-sized volume instead of in a massive clean room.  There would be multiple advantages. The clean room cost would be much less.  The process equipment could be smaller and less expensive. The process could use smaller, less expensive wafers.  It might be possible to manufacture ICs in volumes that match the demand. 

This is the promise of the PARC technology.  The PARC technology promises IC production in the volume occupied by a laser printer.  In that small volume, the PARC device might be able to manufacture ICs economically in small quantities.  It is very easy to see why DARPA would be interested, since the demand for many military ICs is only in quantities of a few thousand.

The Semico Spin

The PARC device points to the future.  3-D printer technology is making rapid strides.  It should be possible now to print a substrate and then print metal lines on that substrate to duplicate the function of a PC board or hybrid substrate.  Soon, it may be possible to print transistors.  How long after that will it be before a 3-D printer becomes a desk-sized semiconductor fab?

By Morry Marshall and Adrienne Downey

Morry Marshall headed up Semico Research's Strategic Technologies practice, which covers emerging technologies vital to the semiconductor industry, like nanotechnology, MEMS, SiGe and RF semiconductors.  In addition to numerous custom studies, he also authored studies about personal computers, automotive electronics, SiGe markets, semiconductor packaging and numerous semiconductor end-use products.  After graduating from the United States Military Academy, West Point, NY in 1961, Morry served as an Armored Cavalry platoon leader. When a line-of-duty medical retirement ended his military career, he transitioned into the electronics industry.  During a subsequent career in electronics spanning more than thirty years, Morry accumulated progressive experience in sales, sales management and marketing management positions and was associated with breakthrough products such as semiconductor memory for the first personal computers and LED displays for the first handheld electronic calculators.  He retired in 2012.