Spatial Monitoring of Wafer Map Defect Data Based on 2D Wavelet Spectrum Analysis

Round 1

Reviewer 1 Report

This paper deals with a fault detection approach using spatial SPC based on wavelet spectrum analysis. The image processing method used to detect spatial fault images in manufacturing process do not need any assumptions for pre-determined values.

The paper begins with an introduction that presents the existing work in this context. although the state of the art proposed is quite complete, it is focused on the method of vision. The introduction can be improved by references of paper review on the diagnosis in the general case, and the diagnosis applied to waffers, for example: Fault prognosis for batch production based on percentile measure and gamma process: Application to semiconductor manufacturing". in Journal of Process Control, Vol.48. pp. 72-80 (2016). 
A review of in-line/off-line defect characterization techniques applied to control and improve electronic grade silicon wafer manufacturing processes. Materials Science and Engineering B91–92 (2002) 128–135.

Then the authors present in a clear and comprehensible way the theoretical tools used, the DWT and the spectral analysis. In section 3 the method of detecting wafer faults using wavelet spectrum-based two-dimensional statistical process control (WS-2DSPC) is detailed. The fourth step of the procedure which is the estimate of the Hurst exponents of each window in three directions using wavelet spectrum analysis, used for calculating control statistics of WS-SPC can be considered as a pertinent contribution.

The method is well presented and the equations are well written and easy to understand. The application support is relevant and the results demonstrate the effectiveness of the method proposed in this paper. However the following point can be deepened:

The manufacture of a wafer is achieved by the superposition of several layers of products, so that some layers are not visible and it is difficult in this case to detect defects by image processing techniques. How the method proposed in this paper will overcome this issue.

Wafers, by their manufacturing process can have multiple defects. Does the method proposed in this paper make it possible to identify the type of defect and consequently the equipment at the origin of this defect?

As the method proposed is based on a frequency processing of the signals, the step of filtering the noise becomes important, how in this paper the levels of decomposition are identified so that the useful information does not get filtered with the noise.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

Paper is interresting Add more experimental results.
2. Quality of figure 1 is poor
3. Add more information about the manufacturing process. Add some pictures, schemes to explain the idea of this process

Author Response

Please see the attachment.

Author Response File: Author Response.docx