Wafer asset modeling using language processing methods

Patent metadata

Field	Value
Publication number	US-2022019710-A1
Application number	US-202016933972-A
Country	US
Kind code	A1
Filing date	Jul 20, 2020
Priority date	Jul 20, 2020
Publication date	Jan 20, 2022
Grant date	—

Abstract

A computing device includes a processor and a storage device. A wafer asset modeling module is stored in the storage device and is executed by the processor to configure the computing device to perform acts identifying and clustering a plurality of assets based on static properties of a wafer asset using a first module of the wafer asset modeling module. The clustered plurality of assets is determined based on dynamic properties of the wafer asset using a second module of the wafer asset modeling module. Event prediction is performed by converting a numeric data of the clustered plurality of assets to a natural language processing (NLP) domain by a third module of the wafer asset modeling module. One or more sequence-to-sequence methods are performed to predict a malfunction of a component of the wafer asset and/or an event based on past patterns. Prediction information is stored in the storage device.

First claim

What is claimed is:

1 . A computing device comprising:
a processor; a storage device for content and programming coupled to the processor; a wafer asset modeling module stored in the storage device, wherein an execution of the wafer asset modeling module by the processor configures the computing device to perform acts comprising: identifying and clustering a plurality of assets based on static properties of a wafer asset using a first module of the wafer asset modeling module; determining a rating of one or more assets of the clustered plurality of assets based on dynamic properties of the wafer asset by a second module of the wafer asset modeling module; performing an event prediction by converting a numeric data of the clustered plurality of assets to a natural language processing (NLP) domain by a third module of the wafer asset modeling module; performing one or more sequence-to-sequence methods to predict a malfunction of a component of the wafer asset and/or an event based on past patterns, by the third module of the wafer asset modeling module; and predicting information and storing in the storage device.

2 . The computing device according to claim 1 , wherein:
the static properties of the assets comprise static data including one or more of a make, model, or manufacturing information of each of the assets; the dynamic properties of the assets comprise dynamic data including one or more of operational data, sensor data, or alarm data; and the third module is configured to convert one or more of the static data and the dynamic data to the NLP domain by finding word representations using at least one of a Bag-of-Words model, a term frequency-inverse document frequency (TF-IDF), or a word vectorization.

3 . The computing device according to claim 1 , wherein the numeric data is converted from an Internet of things (IoT) domain to the NLP domain by creating a word embedding model of the numeric data performing at least one of a cluster analysis, a difference measurement analysis, or a predictive model development.

4 . The computing device according to claim 1 , wherein the rating of the one or more assets determined by the second module comprises a description based on a conditional characteristic.

5 . The computing device according to claim 1 , wherein the numeric data is converted from a semiconductor domain to the NLP domain by creating a word embedding model of the numeric data of a semiconductor wafer fabrication, and wherein the second module is configured to determine a rating of a plurality of semiconductor wafers according to a recipe process.

6 . The computing device according to claim 5 , wherein the word embedding model of the numeric data is created using at least one of a Bag-of-Words (BOW) model, a term frequency-inverse document frequency (TF-IDF), or a word vectorization.

7 . The computing device according to claim 6 , wherein one or more of the plurality of semiconductor wafers are rated as one of good or bad according to the determined rating of the recipe process.

8 . The computing device according to claim 7 , wherein:
the third module is further configured to use the BOW model to create the word embedding model of the numeric data by determining BOW vectors based on creating rows and columns of the numeric data, and each row represents a semiconductor wafer, and each column represents a count of a particular recipe that is identified as being used in a respective wafer.

9 . The computing device according to claim 7 , wherein:
the third module is further configured to use the TF-IDF to create the word embedding model of the numeric data based on creating rows and columns of the numeric data, each row represents a semiconductor wafer, and each column represents a count of a particular recipe that is identified as being used in a respective wafer, and a numerical value is assigned to each recipe according to a uniqueness of the recipe compared to other recipes used.

10 . A computer-implemented method for wafer asset modeling, the method comprising:
clustering a plurality of wafer assets based on static properties of a wafer comprising static data including at least one of a make, a model, or a manufacturing information of each of the wafer assets; determining a rating of one or more assets of the plurality of wafer assets based on dynamic properties, wherein the dynamic properties comprise dynamic data including one or more of operational data, sensor data, or alarm data; predicting a target event or a failure event by converting a numeric data of the plurality of the wafer assets to a natural language processing (NLP) domain; and performing one or more sequence-to-sequence methods to predict alarms and target events based on past patterns.

11 . The computer-implemented method according to claim 10 , further comprising converting one or more of the static data and the dynamic data to the NLP domain by finding word representations using at least one of a Bag-of-Words model, a term frequency-inverse document frequency (TF-IDF), or a word vectorization.

12 . The computer-implemented method according to claim 10 , wherein converting the numeric data is performed from an Internet-of-things (IoT) domain to the NLP domain by creating a word embedding model of the numeric data performing at least one of a cluster analysis, a difference measurement analysis, or a predictive model development.

13 . The computer-implemented method according to claim 10 , further comprising: determining the rating of the one or more of the plurality of wafer assets based on a conditional characteristic.

14 . The computer-implemented method according to claim 10 , wherein:
the wafer asset comprises a plurality of semiconductor wafers, and the method further comprises:
creating a word embedding model by converting the numeric data from a semiconductor domain to the NLP domain; and
determining a rating of a plurality of semiconductor wafers according to a recipe process.

15 . The computer-implemented method according to claim 14 , further comprising:
creating the word embedding model of the numeric data by using at least one of a Bag-of-Words (BOW) model, a term frequency-inverse document frequency (TF-IDF), or a word vectorization.

16 . The computer-implemented method according to claim 15 , further comprising:
creating the word embedding model of the numeric data by the BOW model includes determining BOW vectors by creating rows and columns, wherein each row represents a wafer, each column represents a count of a particular recipe that is identified as being used in a respective wafer.

17 . The computer-implemented method according to claim 15 , wherein:
creating the word embedding model of the numeric data by the TF-IDF includes creating rows and columns, and each row represents a wafer, each column represents a count of a particular recipe that is identified as being used in a respective wafer, and wherein a numerical value is assigned to each recipe according to a uniqueness of the recipe compared to other recipes used.

18 . The computer-implemented method according to claim 14 , further comprising:
rating one or more of the plurality of semiconductor wafers as one of good or bad according to the determined rating of the recipe process.

19 . The computer-implemented method according to claim 18 , further comprising:
determining for each good wafer of the plurality of semiconductor wafers one or more closest wafers based on similar numerical values for each recip

Assignees

• IBM

Inventors

• Khabiri Elham
• Bhamidipaty Anuradha
• Baseman Robert Jeffrey
• Reddy Chandrasekhara K
• Jayaraman Srideepika

Classifications

• G01N21/9501

  Semiconductor wafers (manufacturing processes per se of semiconductor devices implementing a measuring step H10P74/20) · CPC title

• G06F2119/02

  Reliability analysis or reliability optimisation; Failure analysis, e.g. worst case scenario performance, failure mode and effects analysis [FMEA] · CPC title

• G06F30/30 Primary

  Circuit design · CPC title

• G06F30/27

  using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model · CPC title

• G06F40/40

  Processing or translation of natural language (natural language analysis G06F40/20; semantic analysis G06F40/30) · CPC title