中科微点-全球专利检索

  1. Login
  2. Sign Up

Search Results

8 records (took 0.015 seconds)

    END POINT DETECTING METHOD FOR CHEMICAL MECHANICAL POLISHING
    1.
    发明专利
    END POINT DETECTING METHOD FOR CHEMICAL MECHANICAL POLISHING 审中-公开

    公开(公告)号：JP2002083793A

    公开(公告)日：2002-03-22

    申请号：JP2001193210

    申请日：2001-06-26

    Applicant: IBM

    Inventor： LI LEPING ,  BARBEE STEVEN GEORGE ,  LEE ERIC JAMES ,  MARTIN FRANCISCO A ,  CONN WAY

    IPC: H01L21/302 ,  B24B37/04 ,  B24B49/12 ,  G11B5/31 ,  H01L21/304

    Abstract: PROBLEM TO BE SOLVED: To provide a method for detecting an end point in chemical mechanical polishing(CMP) of a cloisonne structure. SOLUTION: A nitride layer 31 is stuck on the upper surface of a metal structure 21, at least and a metal oxide layer 22 is stuck on a metal structure 21 and the nitride layer 31. Next, the metal oxide layer 22 is polished using a CMP process while using a slurry, and the nitride layer 31 on the upper surface of the metal structure 21 is exposed. By polishing the nitride layer 31, ammonia is generated inside the slurry. The ammonia is extracted from the slurry as a gas, and according to the concentration of ammonia, a signal is generated. Then, according to the change of the signal, the CMP process is ended. As a suitable execution example, the metal oxide layer 22 is oxidized aluminium, the nitride layer 31 is aluminium nitride, and the nitride layer 31 is stuck on a substrate and the metal structure 21 as a conformal layer.

    Endpoint detection in chemical mechanical polishing
    2.
    发明专利
    Endpoint detection in chemical mechanical polishing 未知

    公开(公告)号：GB2365808A

    公开(公告)日：2002-02-27

    申请号：GB0115775

    申请日：2001-06-28

    Applicant: IBM

    Inventor： LI LEPING ,  BARBEE STEVEN GEORGE ,  LEE ERIC JAMES ,  MARTIN FRANCISCO A ,  WEI CONG

    IPC: H01L21/302 ,  B24B37/04 ,  B24B49/12 ,  G11B5/31 ,  H01L21/304 ,  B24B49/00

    Abstract: An endpoint indicating film 31 is provided between a metal oxide layer 22 overlying a substrate. During chemical mechanical polishing, the endpoint of the process is determined by extracting gaseous reaction product from the slurry (14, fig. 5) arising from a chemical reaction between the slurry (14, fig. 5) and the indicating film 31 and generating a signal in accordance with the concentration of the extracted reaction product, wherein a change in the signal indicates the endpoint. The extracted reaction product may be converted into a different product which emits chemiluminescence which is detected to provide the signal. The metal oxide layer 22 is preferably aluminium oxide, the endpoint indicating film 31 is preferably aluminium nitride and the reaction product ammonia. The method can be used to form a cloisonn’ structure.

    3.
    发明专利
    未知

    公开(公告)号：DE3668188D1

    公开(公告)日：1990-02-15

    申请号：DE3668188

    申请日：1986-10-07

    Applicant: IBM

    Inventor： BARBEE STEVEN GEORGE ,  DEVINE GREGORY PAUL ,  PATRICK WILLIAM JOHN ,  SEELEY GERARD

    IPC: C23C16/40 ,  C23C16/448 ,  C23C16/52 ,  C23C14/54 ,  C23C14/24 ,  C23C14/34 ,  C23C16/44 ,  H05K3/16

    Abstract: The system for forming a uniform layer of a material from a vapor phase onto the surface of an object at a high rate of deposition includes a reservoir (10) for heating a material from which gas is propagated; a low pressure reactor (20) wherein said layer is formed; connecting means (18) between said reservoir and said reactor; means connected to the outlet of said reactor for creating a vacuum therein and causing said gas to flow from said reservoir through said connecting means to said reactor; first pressure control means at the outlet of said reservoir for maintaining the pressure of said gas at the reservoir end of said connecting means at a first value; and second pressure control means at the inlet of said reactor for maintaining the pressure of said gas thereat at a second value by adjusting said means for creating a vacuum, thereby providing a constant mass flow rate of said gas into said reactor. The method of forming a layer of material from a gas upon the surface of an object comprises: heating a reservoir to a temperature sufficient to vaporize a material from which said gas is formed; creating a vacuum in a reactor containing said object to cause said gas to flow from said reservoir through a connecting means to said reactor; and maintaining the reservoir outlet pressure and the reactor inlet pressure at constant values, thereby providing a constant mass flow rate of said gas into said reactor.

    IDENTIFIZIEREN VON OPTIMALEN GEWICHTUNGEN ZUM VERBESSERN EINERVORHERSAGEGENAUIGKEIT BEI METHODEN FÜR MASCHINELLES LERNEN
    6.
    发明专利
    IDENTIFIZIEREN VON OPTIMALEN GEWICHTUNGEN ZUM VERBESSERN EINERVORHERSAGEGENAUIGKEIT BEI METHODEN FÜR MASCHINELLES LERNEN 未知

    公开(公告)号：DE112020005610T5

    公开(公告)日：2022-09-01

    申请号：DE112020005610

    申请日：2020-11-10

    Applicant: IBM

    Inventor： XU JING ,  HAN SI ER ,  BARBEE STEVEN GEORGE ,  ZHANG XUE YING ,  YANG JI HUI

    IPC: G06N20/00

    Abstract: Bereitgestellt werden ein computerrealisiertes Verfahren, System und Computerprogrammprodukt zum Verbessern einer Vorhersagegenauigkeit bei ML-Methoden. Ein Lehrer-Modell wird erzeugt, wobei das Lehrer-Modell eine Gewichtung für jeden Datenfall erzeugt. Das aktuelle Schüler-Modell wird daraufhin unter Verwendung von Trainingsdaten und der Gewichtungen trainiert, die durch das Lehrer-Modell erzeugt werden. Nach dem Trainieren des aktuellen Schüler-Modells erzeugt das aktuelle Schüler-Modell Zustandsmerkmale, die durch das Lehrer-Modell verwendet werden, um neue Gewichtungen zu erzeugen. Daraufhin wird ein Kandidaten-Schüler-Modell unter Verwendung von Trainingsdaten und dieser neuen Gewichtungen trainiert. Eine Belohnung wird durch ein Vergleichen des aktuellen Schüler-Modells mit dem Kandidaten-Schüler-Modell unter Verwendung von Trainings- und Testdaten erzeugt, die zum Aktualisieren des Lehrer-Modells verwendet werden, wenn eine Abbruchregel nicht erfüllt wurde. Nachdem eine Abbruchregel erfüllt wird, werden die durch das Lehrer-Modell erzeugten Gewichtungen als die „optimalen“ Gewichtungen betrachtet, die an den Benutzer zurückgegeben werden.

    Identifying optimal weights to improve prediction accuracy in machine learning techniques
    7.
    发明专利
    Identifying optimal weights to improve prediction accuracy in machine learning techniques 未知

    公开(公告)号：AU2020385049A1

    公开(公告)日：2022-04-28

    申请号：AU2020385049

    申请日：2020-11-10

    Applicant: IBM

    Inventor： XU JING ,  HAN SI ER ,  BARBEE STEVEN GEORGE ,  ZHANG XUE YING ,  YANG JI HUI

    IPC: G06N99/00 ,  G06N3/04 ,  G06N3/08

    Abstract: A computer-implemented method,system and computer program product for improving prediction accuracy in machine learning techniques. A teacher model is constructed: wherein the teacher model generates a weight for each data case.The current student model is then trained using training data and the weights generated by the teacher model. After training the current student model, the current student model generates state features: which are used by the teacher model to generate new weights. A candidate student model is then trained using training data and these new weights. A reward is generated by comparing the current student model with the candidate student model using training and testing data,which is used to update the teacher model if a stopping rule has not been satisfied. Upon a stopping rule being satisfied, the weights generated by the teacher model are deemed to be the "optimal"weights which are returned to the user.

Patent Agency Ranking