Jerry A. Stefani

Supervisory run-to-run control of polysilicon gate etch using in situ ellipsometry

Polysilicon gate etch is a critical manufacturing step in the manufacturing of MOS devices because it determines the tolerance limits on MOS circuit performance. The etch used in the current study suffers from machine aging, which causes processing results to drift with time. Performing the etch for the same time with fixed process setpoints (recipe) for all wafers would produce unsatisfactory results. Thus, an in situ ellipsometer was employed with a new run-to-run supervisory controller, termed predictor corrector control (PCC), to eliminate the impact of machine and process drift. A novel modeling technique was used to predict uniformity from the ellipsometry data collected at a single site on the wafer. Predictive models are employed by the PCC supervisory controller to generate optimal settings (recipe) for every wafer which will achieve a target mean etch rate, while maintaining a spatially uniform etch. A 200 wafer experiment was conducted to demonstrate the benefits of process control. Implementation of PCC resulted in a 36% decrease in standard deviation from target for the mean etch rate. In addition, the data indicates that controlling etch rate may improve the control and uniformity of the line width change. >

Advanced process control of a CVD tungsten reactor

An advanced multivariable in-line process control system, which combines traditional statistical process control (SPC) with feedback control, has been applied to the CVD tungsten process on an Applied Materials reactor. The goal of the model-based controller is to compensate for shifts in the process and maintain the wafer-state responses on target. The controller employs measurements made on test wafers to track the process behavior. This is accomplished by using model-based SPC, which compares the measurements with predictions obtained from process models. The process models relate the equipment settings to the wafer-state responses of interest. For CVD tungsten, a physically-based modeling approach was employed based on the reaction rate for the H/sub 2/ reduction of WF/sub 6/. The Arrhenius relationship for the kinetic model was linearized so that empirical modeling techniques could be applied. Statistically valid models were derived for deposition rate, film stress, and bulk resistivity using stepwise least-squares regression. On detecting a statistically significant shift in the process, the controller calculates adjustments to the settings to bring the process responses back on target. To achieve this, two additional test wafers are processed at slightly different settings than the current recipe. This local experiment allows the models to be updated to reflect the current process state. The model updates are expressed as multiplicative or additive changes in the process inputs and a change in the model constant. This approach for adaptive control also provides a diagnostic capability regarding the cause of the process shift. The adapted models are used by an optimizer to compute new settings to bring the responses back to target. The optimizer is capable of incrementally entering controllables into the strategy, reflecting the degree to which the engineer desires to manipulate each setting. The capability of the controller to compensate for induced shifts in the CVD tungsten process is demonstrated. Targets for film bulk resistivity and deposition rate were maintained while satisfying constraints on film stress and WF/sub 6/ conversion efficiency. The ability of the controller to update process models during routine operation is also investigated. The tuned process models better predict the process behavior over time compared to the untuned models and lead to improved process capability.

First‐wafer effect in remote plasma processing: The stripping of photoresist, silicon nitride, and polysilicon

We have identified a first‐wafer effect for photoresist ashing and silicon nitride‐polysilicon stripping in remote plasma reactors. The first‐wafer effect consists of the first wafer etching differently from the subsequent wafers in a lot. For photoresist ashing, the first wafer ashes faster than subsequent wafers. For silicon nitride and polysilicon stripping, first wafers show higher etch rates of silicon nitride and polysilicon, while silicon dioxide first wafers etch faster for the polysilicon strip process, and slower for the silicon nitride strip process. We have modeled the first‐wafer effect for photoresist ashing. We found an inverse relationship between the percentage change in the time to clear the photoresist from the wafer and the time delay between processing sequential wafers. We have included this first‐wafer effect in the on‐line statistical process control strategy for the photoresist asher in our laboratory. Examination of this first‐wafer effect suggests that it may be caused by the ge...

On‐Line Inference of Plasma Etch Uniformity Using In Situ Ellipsometry

To reduce surface damage and achieve vertical profiles during selective etching of polysilicon, uniformity of the pol silicon across the wafer after the bulk etch step must be ensured. The bulk polysilicon gate etch process on a single-wafer plasma reactor was analyzed using response surface methodology to create models to be used in model-based process control. In situ etch rate data at the wafer center was collected using a single-wavelength ellipsometer. Off-line etch rate measurements at sites across the wafer were also made. The etch rate at each site was modeled as a function of the process factors and the ellipsometer response

On-line patterned wafer thickness control of chemical-mechanical polishing

We present a gauge study of an on-line metrology system for chemical-mechanical polishing and a 600 wafer run by run (RbR) control experiment enabled by on-line wafer measurement. The variability, reliability, and accuracy of the on-line metrology system are found to be very good. We show that a simple control approach provides a root-mean-squared error of less than 100 A. In contrast, using pilot wafers and sheet film equivalents to control a process results in a 39% decrease in performance, and that using fewer sites may increase variability and lead to an incorrect controlled thickness. We outline how an 8%–80% improvement in throughput, as well as several reductions in cost of ownership, are possible using on-line metrology in conjunction with run by run control.

The interaction of ion implantation with photoresist ashing: A statistical experimental design study

We applied statistical experimental design concepts to the study of the interaction of ion implantation with remote‐plasma photoresist ashing. A 28–4IV fractional‐factorial screening experiment revealed that implant species and resist hard bake conditions were not significant factors for resist ash rate and uniformity. We obtained quadratic polynomial response surface models using a G‐optimal experimental design as a function of the remaining four ash variables (temperature, pressure, oxygen, and hydrogen mass flow rates) and two implant factors (implant dose and energy). Ion implant dose had a significant effect on ash rate while implant energy was an important factor for ash rate uniformity.

STATISTICAL METHODS FOR SEMICONDUCTOR MANUFACTURING

The sections in this article are 1 Statistical Distributions 2 Hypothesis Testing 3 Experimental Design and Anova 4 Response Surface Methods 5 Categorical Modeling 6 Summary

Open-loop predictive control of plasma etching of tungsten using an in-situ film thickness sensor

Automated control schemes would greatly improve the reproducibility of plasma-assisted etching processes. In this paper we report on the application of an in situ metal film thickness sensor to control a plasma tungsten etch process. The process consists of an anisotropic step to control line profile and remove as much tungsten as possible, followed by an isotropic step which etches through to the underlying layer. In typical operations, a pilot wafer is measured off-line to determine the initial tungsten thickness. An etch time for the first step is then calculated before processing the entire lot. Single wafer lots require the elimination of a pilot wafer. Recently, we integrated a metal film thickness sensor (based on the technology of eddy currents) into a single-wafer plasma tungsten etch module. Our control strategy uses the sensor in a feedforward manner. A measurement of the tungsten film thickness is made in situ prior to processing. Process control software adjusts the etch time for the wafer based on the measured thickness and the predicted etch rate for the equipment settings. The etch rate is calculated from an empirical model obtained using response- surface methodology. A three-fold decrease in wafer-to-wafer variability in final thickness after the etch step was realized compared to that for the deposited thickness.

Practical issues in the deployment of a run-to-run control system in a semiconductor manufacturing facility

ProcessWORKS, a factory-level run-to-run control architecture, originally developed at Texas Instruments and now a product of Adventa Control Technologies, can treat complex control problems in an automated, predictable, and repeatable fashion. ProcessWORKS is compatible with different techniques for data acquisition and analysis, model adjustment and feedback, and model optimization. ProcessWORKS is also designed to deal with practical implementation issues in the fab. In this talk we will review the benefits of ProcessWORKS run-to-run control. We will discuss some practical problems in the deployment of run-to-run control in the fab, and we will show how ProcessWORKS deals with these issues. Examples from the deployment of ProcessWORKS at Texas Instruments on state of the art semiconductor technologies will be given.

Advanced statistical process control of a chemical vapor tungsten deposition process on an Applied Materials Centura reactor

An advanced multivariable off-line process control system, which combines traditional Statistical Process Control (SPC) with feedback control, has been applied to the CVD tungsten process on an Applied Materials Centura reactor. The goal of the model-based controller is to compensate for shifts in the process and maintain the wafer state responses on target. In the present application the controller employs measurements made on test wafers by off-line metrology tools to track the process behavior. This is accomplished by using model- bases SPC, which compares the measurements with predictions obtained from empirically-derived process models. For CVD tungsten, a physically-based modeling approach was employed based on the kinetically-limited H2 reduction of WF6. On detecting a statistically significant shift in the process, the controller calculates adjustments to the settings to bring the process responses back on target. To achieve this a few additional test wafers are processed at slightly different settings than the nominal. This local experiment allows the models to be updated to reflect the current process performance. The model updates are expressed as multiplicative or additive changes in the process inputs and a change in the model constant. This approach for model updating not only tracks the present process/equipment state, but it also provides some diagnostic capability regarding the cause of the process shift. The updated models are used by an optimizer to compute new settings to bring the responses back to target. The optimizer is capable of incrementally entering controllables into the strategy, reflecting the degree to which the engineer desires to manipulates each setting. The capability of the controller to compensate for shifts in the CVD tungsten process has been demonstrated. Targets for film bulk resistivity and deposition rate were maintained while satisfying constraints on film stress and WF6 conversion efficiency.