CMP Film Thickness Control

To properly understand CMP film thickness control, the CMP engineer should understand the sources of thickness variation and how they impact the total film thickness uniformity. Nonuniformity can be grouped in two categories—random variation and systematic variation. Examples of random variation include wafer-to-wafer (WTW), run-to-run (RTR), and some elements of within-wafer (WIW) variations. Elements of random variation add to the total thickness variation by their root mean square [19] 

Systematic variation is a variation that maintains its structure from sample to sample. Examples of systematic variation in CMP include within-die (WID) variation and some elements of WIW variation such as edge thickness roll-off. WID and edge roll-off are considered systematic because the thickest and thinnest points on the die/wafer are predictable. Systematic variation adds directly to the total variation. Hence, systematic effects will have a greater impact on thickness variation. In general, for CMP processes, WID and edge variation values are greater than random components. For these two reasons, systematic effects tend to be the focus of CMP thickness control efforts. 

New head designs (Chapter 4) have given the CMP engineer an additional knob to control across-wafer nonuniformities by varying the distribution of backside wafer pressure. However, the different pressure zones within these heads have a tendency to interact, and independent control of the wafer edge has been illusive. 

WTW and RTR control of thickness are improved by the use of end-point detection systems and advanced process control. End-point detection, whether mechanical or optical, monitor the state of the wafer surface (film thickness, reflectivity, etc.) or of the entire polishing system (friction, slurry by products, etc.) in an attempt to predict when the desired amount of material has been removed (i.e., the end of process). End-point detection is most successful in processes where a change in the films on the wafer surface leads to an abrupt change in the optical or mechanical properties of the wafer surface. For example, copper CMP end point is easy to detect by optical means due to the large difference in reflectivity of the copper film compared to the barrier films. In contrast, end-point detection for small amounts of ILD removal is difficult due to the lack of change in the wafer surface or the wafer-pad interface. 

To prevent material loss due to excursions such as the previous example, robust process control systems are required throughout the supply chain from the raw materials manufacturer to the pad manufacturer and the CMP module. Invariably, incident reviews of such excursions reveal that the excursion could have been prevented or limited to only a small amount of material lost if the proper statistical process control systems had been in place. Invariably, the excursion could have been detected by careful scrutiny of an in-process parameter that was either monitored or should have been monitored by the subsupplier, pad manufacturer, and/or the CMP operation. 

---