Search

Hf and Zr contamination during immersion in process solutions is most likely to occur in neutral and caustic solutions. Both Hf and Zr contamination are introduced onto the wafer surface if they are present in an ammonium hydroxide peroxide mixture solution (which is caustic), but such contamination is removed using existing acid cleans. Large mounts of wafer-to-wafer cross contamination occurs in plasma etch tools. Particles can cause cross contamination in a thermal reactor during high-temperature anneals of high-k dielectric layers. Residual surface cross contamination does not diffuse into the wafers during thermal processing. If contamination remains on a wafer, gate oxide integrity degradation is only observed at high concentrations. Near surface minority carrier lifetime is also affected, but bulk lifetime is not. Index Terms--Cleaning, cross contamination, hafnium dioxide (Hf[O.sub.2]), high-k dielectric, zirconium dioxide (Zr[O.sub.2]).

This work demonstrates that copper contamination present on pre gate-oxidation silicon surfaces results in yield and reliability problems particularly at field overlap edges. Similarly, the junction leakage current associated with the junction perimeter dominates the total leakage current. These detrimental device effects are shown to be caused by copper that is present close to the silicon surface even after thermal processing. Because field overlap and junction perimeter defects become relatively more important when the critical dimensions of circuits are scaled to smaller sizes, they dominate yield loss of high-density circuits, magnifying their importance for future technology generations. Index Terms - Copper contamination, gate oxide yield, junction leakage, silicon.

Deposition of metallic impurities from HF process solutions has been investigated experimentally and explained theoretically in a qualitative manner. The depositions are shown to be electrochemical in nature in that an oxidation reduction reaction results in metal ions in solution depositing on the wafer as elements with an oxidation state of 0. The theory is only qualitative in that it can only predict which metals will deposit, not how much. Experimentally, simple transmission equations can be determined which relate metallic contamination levels on Si wafer surfaces (atoms/[cm.sup.2]) to metal concentration in the solution (ppb). Simple test structures have been fabricated with known amounts of iron and copper contamination in the pregate oxide clean of a 1.25 [[micro]meter] CMOS process. Device measurements indicate device degradation in the case of copper, confirming deposition studies that copper deposits from HF solutions. Iron contaminated wafers show no contamination related device effects, in support of theoretical predictions and deposition studies indicating iron does not deposit from HF solutions. The importance and potential usefulness of test structures as homogeneous contamination monitors is illustrated through device modeling of the contamination effects observed in the test structures that can then be used to estimate the effects of such contamination on ULSI circuit performance.

The Arizona SEMATECH Center of Excellence (SCOE) was established in May of 1988, is funded by SEMATECH and contractually monitored by the Semiconductor Research Corporation (SRC). The SCOE is engaged in research in a broad front to understand and control contamination which causes yield limiting defects in submicron ULSI circuits. Sandia National Laboratory personnel are integrated with UA personnel in the SCOE research. The focus of the research is on contaminants, both particulates and homogeneous or distributed, which originate in, are caused by, or are transported and deposited by process gases and chemicals or process equipment. Further, the work involves investigating the mapping from contaminants and contaminant levels to degradation of device properties. The resulting degradation in device properties can then be used to estimate the effects of such contaminants for submicron processes and circuits. Results achieved during the four years of the SCOE's existence are describe in this paper.