Interesting Times in the Semiconductor Industry

A wise man once said, "May you live in interesting times." We are all lucky because this is certainly an interesting time for the semiconductor industry. Not only are technical challenges abounding, but the economics that drive the industry have changed quickly and dramatically.

Most experts agree that we are close to the end of the era of scaling, where it was relatively easy to gain advances in chip speed and functionality by simply shrinking dimensions. As a case in point, the new 2005 International Technology Roadmap for Semiconductors (ITRS) notes that the gate dielectric has emerged as the most difficult challenge associated with future device scaling. Continued optimization of oxynitride gate dielectric and scaling of this material toward an equivalent gate oxide thickness (EOT) of <1 nm — consistent with device reliability — is envisaged for high-performance (HP) microprocessors. Oxynitride, however, will no longer meet the strict leakage current requirement in low-power applications (low operating power and LSTP). Therefore, the introduction of higher-k materials in which tunneling current can be suppressed while maintaining the drain current will be necessary. In either case, the gate electrode material and process should be optimized so that the depletion width in the gate electrode may be minimized and the boron diffusion prevented. The former necessitates the introduction of metal gates having appropriate work function after the conventional polysilicon ceases to work. "These material changes pose a great challenge in MOSFET technology, where silicon dioxide/polysilicon has long played a central role as the most reliable gate-stack system, and also these new high-k and metal gate stacks are fundamental structures expected to be used with the future new structure CMOS, such as fully depleted SOI or multigate MOSFETs," as is noted in the ITRS.

However, even though huge challenges exist — such as those associated with the gate dielectric — the good news is that the semiconductor industry has an excellent track record in overcoming these challenges. Continued improvements will be possible using new transistor technologies, such as strained silicon, mixed-crystal orientation, high-k gate dielectrics, metal gates and structures like the finFET. Higher levels of integration will be possible through 3-D integration, where very thin die are stacked and connected with through-silicon vias, and through improved system-on-a-chip (SoC) and system-in-package (SiP) designs.

To make life more interesting, the economic drivers of semiconductor technology are also becoming more diverse, fueled by the demand of gadget-crazed consumers. This is putting new pressures on pricing, inventory control and product lifecycle. Many new and diverse applications will result in the lack of dominant industry standards, and ICs will appear in all kinds of new forms.

Semiconductor manufacturers will have to deal with continued tool cost increases. The cost of lithography tools, for example, will continue to increase at a rate of 20% per year, and people will have to pay $40M per tool because there is no other solution. Mark Liu, vice president of operations at TSMC , where more than 200 new products are processed every month, recently said that tools will also have to be customized for high-mix, low-volume production and be capable of handling small lots and frequent recipe changes.

There has been much talk of the next wafer size change to 450 mm or perhaps 675 mm. But there is still more that can be done to improve productivity and reduce cycle time with 300 mm wafers before that will happen. Liu said that the wafer throughput efficiency of 300 mm tools is lacking and sees potential for a 2× increase. Tom Sonderman at AMD has similarly said he sees a 2× increase in fab productivity possible through the use of what he calls advanced precision manufacturing, which involves the judicious use of data collection and evaluation to improve cycle time and yield. Yes, these are interesting times indeed. I would not want it any other way.