CMP users look at test technology, reducing COO

By Debra Vogler

WaferNews Technical Editor

The annual CMP Users’ Group symposium (Northern California chapter of the American Vacuum Society) held recently in San Jose was rich with new technologies and industry progress. Although the CMP market projections presented by Tom Tucker of Laredo Technologies were grim – second half revenues are expected to be down 50% from 1H01 – he did highlight several technologies that will come to the fore as the industry scrambles to innovate its way out of the downturn.

On Tucker’s list of technology advances are KLA-Tencor’s Precice System, which uses eddy currents for endpoint control and in situ film thickness measurements; Eco Physics M17 monitoring system that detects trace amounts of ammonia; and the Center for Tribology’s (CETR) coefficient of friction and acoustic emission sensors. Tucker noted that fixed abrasives are now being used for some demanding leading edge processes in first production processes but they have not yet reached levels originally projected for usage. Also, nanotopography will be important for advanced STI processes. Regarding the status of alternatives to CMP, Tucker made special mention of NuTool’s electrochemical mechanical deposition (ECMD) process and ACM Research’s stress free electropolishing process.

CETR showcased its bench top CMP tester that uses real-time high-frequency simultaneous measurement and analysis of friction force, torque, and contact high-frequency acoustic emission and defining a rate of material removal and polishing non-uniformity to provide in situ information about the polishing process. The micro-tribological parameters are processed in real-time at the rate of thousands of data points per second. The tool accommodates and provides functional testing of polishing pads, slurry, conditioning discs and wafers. Norm Gitis, president and CEO of CETR, notes that the system can measure delamination of low-k material and reports that the company’s first generation bench tester had been ordered and used by two major manufacturers.

NT International, a subsidiary of Entegris, displayed information concerning its closed-loop flow control solution for slurry. The company cites problems with peristaltic pump control in general use: frequent calibration to correlate flow rate and pump speed, changes in facility operating pressures can cause variations by as much as 10 to 30%, and pump failures can cause low-flow or no-flow conditions resulting in damaged or scrapped wafers.

A study had been conducted by Texas Instruments (MICRO, July/August 2001, Michael S. Wang) using NT International’s pressure transducer and flowmeter in combination with a Mykrolis depth filter on an Applied Materials’ Mirra polisher. Wang concluded that the flow instrumentation scheme was able to detect filter life and slurry flow and eliminate process interruptions along with decreased downtime. The author projected a savings of up to six hrs/mo in downtime associated with filter maintenance for high-volume production fabs.

Micron’s Scott Meikle discussed the company’s experiences with COO (cost of ownership) reduction for CMP. Saying the company had overcome the majority of yield impact problems it experienced in the early to mid-1990s, it now wants to focus on productivity issues such as improved consumable stability, improved hardware reliability, and higher throughput. In particular, the polish rate for PMD, while improving, still requires too many monitor wafers to verify the throughput. Meikle specifically noted that longer pad life and either no pad conditioning or better-controlled pad conditioning were needed.

Interestingly enough, Micron appears to be looking at NuTool’s ECMD as a way to reduce the overburden associated with conventional electrochemical deposition (ECD) and thereby reduce COO incurred by the subsequent CMP process. Meikle presented data that showed reduced overburden for Cu CMP in which NuTool’s process had been used on the evaluation sample prior to CMP.

Seeking to integrate CMP elements into one model, David Dornfeld of UC Berkeley presented the university’s current version. The model is an attempt to simulate wafer level, feature level, die level, non-uniformity, slurry/water usage, abrasive concentration, other chemical effluent, and energy. While work is continuing on perfecting the model, it will be interesting to watch its progress since the industry would benefit from such a tool.



Easily post a comment below using your Linkedin, Twitter, Google or Facebook account. Comments won't automatically be posted to your social media accounts unless you select to share.