Patent US5467883A successfully refreshed from Google Patents.
US5467883A
Expired / Lapsed
US
⏱️ 0.0 years remaining
Active neural network control of wafer attributes in a plasma etch process
Assignee: Bell Semiconductor LLC
•
Filing: 1993-11-17
•
Granted: 1995-11-21
•
Expires: 2013-11-17
Abstract
Abstract The present invention is predicated upon the fact that an emission trace from a plasma glow used in fabricating integrated circuits contains information about phenoma which cause variations in the fabrication process such as age of the plasma reactor, densities of the wafers exposed to the plasma, chemistry of the plasma, and concentration of the remaining material. In accordance with the present invention, a method for using neural networks to determine plasma etch end-point times in an integrated circuit fabrication process is disclosed. The end-point time is based on in-situ monitoring of the optical emission trace. The back-propagation method is used to train the network. More generally, a neural network can be used to regulate control variables and materials in a manufacturing process to yield an output product with desired quality attributes. An identified process signature which reflects the relation between the quality attribute and the process may be used to train the neural network.
Inventors
- 👤 Robert C. Frye
- 👤 Thomas R. Harry
- 👤 Earl R. Lory
- 👤 Edward A. Rietman
Bibliographic Details
Application Number
—
Jurisdiction
US
Data Source
google_patents
Last Synchronized
2026-10-29 13:06
CPC Classifications
G; G06; G06F; G06F17/00; G05; G05B; G05B13/00; G05B13/02; G05B13/0265; G05B13/027; H; H01; H01J; H01J37/00; H01J37/32; H01J37/32917; H01J37/32935; H01J37/32963
Claims Analysis
Total Claims: 25
Claim 1
Independent Claim
127 words
A method for using a neural network to control a set of one or more variables in a process to yield a product resulting from said process, said product characterized by a set of one or more quality attributes, said method comprising the steps of: a. measuring a set of one or more process signatures during said process to form a record of said process signatures reflected by a relation between said set of one or more quality attributes and said set of one or more variables; b. providing data from said record to said neural network; and c. using said neural network to control said set of one or more variables in said process to yield said product resulting from said process based on said data.
Claim 2
Independent Claim
127 words
A method for using a neural network to control a set of one or more variables in a process to yield a product resulting from said process, said product characterized by a set of one or more quality attributes, said method comprising the steps of: a. measuring a set of one or more process signatures during said process to form a record of said process signatures reflected by a relation between said set of one or more quality attributes and said set of one or more variables; b. providing data from said record to said neural network; and c. using said neural network to control said set of one or more variables in said process to yield said product resulting from said process based on said data.
Claim 3
Dependent Claim
35 words
The method of claim 1 wherein said neural network is trained with measurements of said set of one or more quality attributes and with prior data of said set of one or more process signatures.
Claim 4
Dependent Claim
15 words
The method of claim 1 wherein said neural network is trained with the back-propagation technique.
Claim 5
Dependent Claim
29 words
The method of claim 1 further comprising the step of using said neural network to control said set of one or more variables based on information from a database.
Claim 6
Independent Claim
145 words
A method for controlling a set of one or more process variables in a plasma etching process to yield a product characterized by a set of one or more quality attributes, said method comprising the steps of: a. identifying a set of one or more process signatures reflected by a relation between said set of one or more process variables and said set of one or more quality attributes; b. measuring said set of one or more process signatures to form a record; c. providing data from said record to a neural network; d. training said neural network with 1) prior data from records of sets of one or more process signatures and 2) prior data of sets of one or more quality attributes; e. controlling said set of one or more process variables based on said training and on said data from said record.
Claim 7
Independent Claim
145 words
A method for controlling a set of one or more process variables in a plasma etching process to yield a product characterized by a set of one or more quality attributes, said method comprising the steps of: a. identifying a set of one or more process signatures reflected by a relation between said set of one or more process variables and said set of one or more quality attributes; b. measuring said set of one or more process signatures to form a record; c. providing data from said record to a neural network; d. training said neural network with 1) prior data from records of sets of one or more process signatures and 2) prior data of sets of one or more quality attributes; e. controlling said set of one or more process variables based on said training and on said data from said record.
Claim 8
Dependent Claim
17 words
The method of claim 5 wherein said set of one or more process variables is etch time.
Claim 9
Dependent Claim
55 words
The method of claim 5 wherein said etching process selectively removes a first layer of material to expose an underlying second layer of material, said underlying second layer of material having a thickness associated with it, and wherein said quality attribute is said thickness associated with said second layer upon completion of said etching process.
Claim 10
Dependent Claim
18 words
The method of claim 5 wherein said set of one or more process signatures is a plasma emission.
Claim 11
Dependent Claim
32 words
The method of claim 5 wherein data is gathered during said measuring of said set of one or more process signatures after said set of one or more process signatures is normalized.
Claim 12
Dependent Claim
15 words
The method of claim 5 wherein said neural network is trained with the back-propagation technique.
Claim 13
Dependent Claim
25 words
The method of claim 5 further comprising the step of controlling said set of one or more process variables based on information from a database.
Claim 14
Independent Claim
196 words
A method for controlling etch time in a plasma etching process, wherein said etching process selectively removes a first layer of material to expose an underlying second layer of material, said underlying second layer of material having a thickness associated with it, said method comprising the steps of: measuring a plasma emission from said plasma etching process to form a record, b. providing data from said record to a neural network, c. training said neural network, wherein said training of said neural network comprises computing a first order linear correction to control said etch time according to the rule ##EQU7## where t ideal is the ideal etch time for a desired thickness of said underlying second layer, t obs is the actual etch time, T obs is the observed thickness of said underlying second layer of material at the end of said plasma etching process, T des is the desired thickness of said underlying second layer of material at the end of said plasma etching process, and E R is a constant representing the etch rate of said plasma etching process, d. controlling said etch time based on said training and on data from said record.
Claim 15
Independent Claim
196 words
A method for controlling etch time in a plasma etching process, wherein said etching process selectively removes a first layer of material to expose an underlying second layer of material, said underlying second layer of material having a thickness associated with it, said method comprising the steps of: measuring a plasma emission from said plasma etching process to form a record, b. providing data from said record to a neural network, c. training said neural network, wherein said training of said neural network comprises computing a first order linear correction to control said etch time according to the rule ##EQU7## where t ideal is the ideal etch time for a desired thickness of said underlying second layer, t obs is the actual etch time, T obs is the observed thickness of said underlying second layer of material at the end of said plasma etching process, T des is the desired thickness of said underlying second layer of material at the end of said plasma etching process, and E R is a constant representing the etch rate of said plasma etching process, d. controlling said etch time based on said training and on data from said record.
Claim 16
Independent Claim
151 words
An apparatus for controlling a set of one or more process variables in a plasma etching process to yield a product characterized by a set of one or more quality attributes, said apparatus comprising: means for identifying a set of one or more process signatures reflected by a relation between said set of one or more process variables and said set of one or more quality attributes; b. means for measuring said set of one or more process signatures to form a record; c. means for providing data from said record to a neural network; d. means for training said neural network with 1, prior data from records of sets of one or more process signatures and 2) prior data of sets of one or more quality attributes; e. means for controlling said set of one or more process variables based on said training and on said data from said record.
Claim 17
Independent Claim
151 words
An apparatus for controlling a set of one or more process variables in a plasma etching process to yield a product characterized by a set of one or more quality attributes, said apparatus comprising: means for identifying a set of one or more process signatures reflected by a relation between said set of one or more process variables and said set of one or more quality attributes; b. means for measuring said set of one or more process signatures to form a record; c. means for providing data from said record to a neural network; d. means for training said neural network with 1, prior data from records of sets of one or more process signatures and 2) prior data of sets of one or more quality attributes; e. means for controlling said set of one or more process variables based on said training and on said data from said record.
Claim 18
Dependent Claim
17 words
The apparatus of claim 13 wherein said set of one or more process variables is etch time.
Claim 19
Dependent Claim
55 words
The apparatus of claim 13 wherein said etching process selectively removes a first layer of material to expose an underlying second layer of material, said underlying second layer of material having a thickness associated with it, and wherein said quality attribute is said thickness associated with said second layer upon completion of said etching process.
Claim 20
Dependent Claim
18 words
The apparatus of claim 13 wherein said set of one or more process signatures is a plasma emission.
Claim 21
Dependent Claim
32 words
The apparatus of claim 13 wherein data is gathered during said measuring of said set of one or more process signatures after said set of one or more process signatures is normalized.
Claim 22
Dependent Claim
15 words
The apparatus of claim 13 wherein said neural network is trained with the back-propagation technique.
Claim 23
Dependent Claim
24 words
The apparatus of claim 13 further comprising means for controlling said set of one or more process variables based on information from a database.
Claim 24
Independent Claim
202 words
An apparatus for controlling etch time in a plasma etching process, wherein said etching process selectively removes a first layer of material to expose an underlying second layer of material, said underlying second layer of material having a thickness associated with it, said apparatus comprising: a. means for measuring a plasma emission from said plasma etching process to form a record, b. means for providing data from said record to a neural network, c. means for training said neural network, wherein said training of said neural network comprises computing a first order linear correction to control said etch time according to the rule ##EQU8## where t ideal is the ideal etch time for a desired thickness of said underlying second layer, t obs is the actual etch time, T obs is the observed thickness of said underlying second layer of material at the end of said plasma etching process, T des is the desired thickness of said underlying second layer of material at the end of said plasma etching process, and E R is a constant representing the etch rate of said plasma etching process, d. means for controlling said etch time based on said training and on data from said record.
Claim 25
Independent Claim
202 words
An apparatus for controlling etch time in a plasma etching process, wherein said etching process selectively removes a first layer of material to expose an underlying second layer of material, said underlying second layer of material having a thickness associated with it, said apparatus comprising: a. means for measuring a plasma emission from said plasma etching process to form a record, b. means for providing data from said record to a neural network, c. means for training said neural network, wherein said training of said neural network comprises computing a first order linear correction to control said etch time according to the rule ##EQU8## where t ideal is the ideal etch time for a desired thickness of said underlying second layer, t obs is the actual etch time, T obs is the observed thickness of said underlying second layer of material at the end of said plasma etching process, T des is the desired thickness of said underlying second layer of material at the end of said plasma etching process, and E R is a constant representing the etch rate of said plasma etching process, d. means for controlling said etch time based on said training and on data from said record.
References Cited (Backward Citations)
US5286947A
Cited
USPTO Annuity & Maintenance Windows
| Annuity Stage | Window Opens | Payment Deadline | Surcharge Period Until | Estimated Fee | Status |
|---|---|---|---|---|---|
| 3.5 Years (Year 4) | 1998-11-20 | 1999-05-22 | 1999-11-21 | $2000.00 | Window Closed / Lapsed |
| 7.5 Years (Year 8) | 2002-11-20 | 2003-05-22 | 2003-11-21 | $3760.00 | Window Closed / Lapsed |
| 11.5 Years (Year 12) | 2006-11-20 | 2007-05-22 | 2007-11-21 | $7700.00 | Window Closed / Lapsed |