Online End Point Detection in CMP Using SPRT of Wavelet Decomposed Sensor Data

Abstract

Efficient end point detection (EPD) in chemical mechanical planarization (CMP) is critical to quality and productivity of the wafer fabrication process. The cost of over and under polishing, and the cost of ownership of many expensive metrology-based EPD methods have motivated the researchers to seek cost effective and efficient alternatives. This paper presents a novel method for EPD, which uses a sequential probability ratio test (SPRT) on the wavelet decomposed coefficient of friction (CoF) data from the CMP process. The method is made suitable for online application by developing a moving block data processing strategy, which matches the rate of data acquisition. Tests on both oxide and copper metal CMP show that the developed methodology is uniquely capable of identifying the start and finish of the end point event.Index Terms-Acoustic emission, chemical mechanical planarization (CMP), coefficient of friction (CoF), end point, sequential probability ratio test (SPRT), wavelet-based multiscale analysis.

A Multiscale Bayesian SPRT Approach for Online Process Monitoring

“…and is the decision limit associated with level of significance (size of the critical region (6) where and are obtained by minimizing the Bayes risk , and is the prior probability. In order to consider the SPRT from a Bayes viewpoint, the overall cost or loss when is the true state of the process is given by (7) where and are binary variables, denotes the loss when is true and hypothesis is accepted, denotes the cost of sampling, and is the number of samples.…”

“…Wavelet-based multiresolution analysis followed by variance BSPRT were applied on these data sets to assess the efficacy of the EPD approach. It is to be noted that the data collected above was used in our previous work [6]. For the purpose of testing the BSPRT algorithm, the data was fed to the algorithm at the same rate at which it was generated to simulate an online data feed scenario.…”

“…This requirement is automatically satisfied when monitoring uses wavelet details, which is known to be Gaussian. The BSPRT procedure presented here is a significant improvement over our previous work that uses the regular SPRT procedure [6].…”

Simulated Effects of Measurement Noise on Contact Measurements Between Rough and Smooth Surfaces

“…Through this polishing, the surface roughness of the raw silicon wafer can be reduced from hundreds of nanometers to tens of nanometers [5]. There is evidence that the pad-wafer friction signature changes as the surface roughness is reduced [6], a possible end-point detector of the polish flatness or completion of the material removal [7].…”

Endpoint Detection in Cu-CMP

“…In CMP, in situ detection of barrier-to-dielectric layer transition is typically carried out using reflectivity measurements 1 or coefficient of friction ͑CoF͒ measurements. 2 The use of low-k materials as dielectric materials has opened up the possibility of using spectroscopic techniques such as Raman and Fourier transform infrared spectroscopy ͑FTIR͒.…”