Closed loop adaptive control of spectrum-producing step using neural networks

• US 5,841,651 A
• Filed: 11/09/1992
• Issued: 11/24/1998
• Est. Priority Date: 11/09/1992
• Status: Expired due to Term

• Current Assignee: United States Of America As Represented By The Department Of Energy
• Original Assignee: United States Department of Energy (Government of the United States of America)
• Inventors: Fu, Chi Yung
• Primary Examiner(s): Elmore, Reba I.
• Assistant Examiner(s): BROWN, THOMAS
• Application Number:
  US07/973,865
• Time in Patent Office: 2,206 Days
• Field of Search: 364/148-152, 250/341, 395/21, 395/22, 395/25, 427/34, 216/59, 216/60, 216/61, 216/58, 216/62, 216/63, 216/65-68, 156/345, 156/643.1, 156/626.1, 156/625.1, 156/627.1, 204/192.13, 204/192.33, 204/298.03, 204/298.32
• US Class Current: 700/48
• CPC Class Codes: G06F 2218/12 Classification; Matching
• CPC Class Codes: G06N 3/043 based on fuzzy logic, fuzzy membership or fuzzy inference, e.g. adaptive neuro-fuzzy inference systems [ANFIS]
• CPC Class Codes: H01J 37/32935 Monitoring and controlling tubes by information coming from the object and/or discharge

Abstract

Characteristics of the plasma in a plasma-based manufacturing process step are monitored directly and in real time by observing the spectrum which it produces. An artificial neural network analyzes the plasma spectrum and generates control signals to control one or more of the process input parameters in response to any deviation of the spectrum beyond a narrow range. In an embodiment, a plasma reaction chamber forms a plasma in response to input parameters such as gas flow, pressure and power. The chamber includes a window through which the electromagnetic spectrum produced by a plasma in the chamber, just above the subject surface, may be viewed. The spectrum is conducted to an optical spectrometer which measures the intensity of the incoming optical spectrum at different wavelengths. The output of optical spectrometer is provided to an analyzer which produces a plurality of error signals, each indicating whether a respective one of the input parameters to the chamber is to be increased or decreased. The microcontroller provides signals to control respective controls, but these lines are intercepted and first added to the error signals, before being provided to the controls for the chamber. The analyzer can include a neural network and an optional spectrum preprocessor to reduce background noise, as well as a comparator which compares the parameter values predicted by the neural network with a set of desired values provided by the microcontroller.

Claims

1. A method for closed-loop adaptive control of spectrum-producing plasma processes, comprising the steps of:

producing a plasma adjacent to a subject surface of a workpiece, said plasma having characteristics dependent upon a plurality of process input parameters;

while said plasma is adjacent to said subject surface, observing the optical frequency spectrum produced by said plasma;

while said plasma is adjacent to said subject surface, controlling one of said process parameters in response to said optical frequency spectrum;

wherein said step of observing comprises passing spectral emissions from said plasma through a window in a chamber containing said plasma and said subject surface;

preventing deposition of material on the inside surface of said window; and

analyzing the frequency spectrum of said emissions as viewed through said window.

8. A method according to claim 1, wherein said step of controlling further comprises the step of adjusting a valve for said one of said process parameters in response to said control signal.

2. A method for closed-loop adaptive control of spectrum-producing plasma processes, comprising the steps of:

producing a plasma adjacent to a subject surface of a workpiece, said plasma having characteristics dependent upon a plurality of process input parameters;

while said plasma is adjacent to said subject surface, observing the optical frequency spectrum produced by said plasma;

while said plasma is adjacent to said subject surface, controlling one of said process parameters in response to said optical frequency spectrum;

wherein said step of observing comprises conducting spectral emissions from said plasma through a narrow tube from said plasma to a window in a chamber containing said plasma and said subject surface, said tube comprising no deposition surface between said plasma and said window which could filter said optical spectrum;

passing said emissions through said window; and

analyzing the frequency spectrum of said emissions as viewed through said window.

3. A method for neural-network closed-loop adaptive control of spectrum-producing plasma processes, comprising the steps of:

producing a plasma adjacent to a subject surface of a workpiece, said plasma having characteristics dependent upon a plurality of process input parameters;

while said plasma is adjacent to said subject surface, observing the optical frequency spectrum produced by said plasma;

while said plasma is adjacent to said subject surface, controlling one of said process parameters in response to said optical frequency spectrum;

wherein said step of controlling comprises using a neural network to produce a control signal for said one of said process parameters in response to said optical frequency spectrum;

wherein said step of observing comprises conducting spectral emissions from said plasma through a narrow tube from said plasma to a window in a chamber containing said plasma and said subject surface, said tube comprising no deposition surface between said plasma and said window which could filter said optical spectrum;

passing said emissions through said window; and

analyzing the frequency spectrum of said emissions as viewed through said window.

4. A method for closed-loop adaptive control of spectrum-producing plasma processes, comprising the steps of:

producing a plasma adjacent to a subject surface of a workpiece, said plasma having characteristics dependent upon a plurality of process input parameters;

while said plasma is adjacent to said subject surface, observing the optical frequency spectrum produced by said plasma;

while said plasma is adjacent to said subject surface, controlling one of said process parameters in response to said optical frequency spectrum;

wherein said step of observing generates an amplitude value corresponding to each of a plurality of wavelengths in said optical spectrum, and wherein said step of controlling comprises generating at least one second-level weighted sum of said amplitude values; and

producing a control signal for said one of said process parameters in response to said at least one second-level weighted sum.

5. The method of claim 4, wherein said at least one second-level weighted sum is a sigmoid-limited weighted sum of said amplitude values.

6. The method of claim 4, wherein said step of producing a control signal comprises the steps of:

generating a third-level weighted sum of said at least one second-level weighted sums; and

producing said control signal in response to said third-level weighted sum.

7. The method of claim 4, wherein said step of generating said at least one second-level weighted sum comprises generating at least one second-evel weighted sum in response to a learned weighting corresponding to each of said amplitude values, and prior to said step of observing:

ascribing initial values to said learned weightings;

setting said one of said process parameters to one of a plurality of known teaching values;

producing an experimental plasma, being the plasma produced in response to said one of said process parameters;

adjusting at least one of said learned weightings such that said control signal produced in response to said experimental plasma is a desired value; and

repeating the steps of setting, producing and adjusting until the control signals produced for each of said plurality of known teaching values approximate a predetermined level of precision.