Data processing: generic control systems or specific application – Specific application – apparatus or process – Product assembly or manufacturing
Reexamination Certificate
2000-10-04
2004-06-01
Picard, Leo (Department: 2125)
Data processing: generic control systems or specific application
Specific application, apparatus or process
Product assembly or manufacturing
C700S030000, C700S220000, C438S710000
Reexamination Certificate
active
06745095
ABSTRACT:
BACKGROUND OF THE INVENTION
Within the semiconductor industry, an ever present need exists for improved process repeatability and control, and in particular for ways of monitoring progress of a process. For example, during the formation of a typical metal-layer-to-metal-layer interconnect, a dielectric layer is deposited over a first metal layer, a via hole is etched in the dielectric layer to expose the first metal layer, the via hole is filled with a metal plug and a second metal layer is deposited over the metal plug (e.g., forming an interconnect between the first and the second metal layers). To ensure the interconnect has low contact resistance, all dielectric material within the via hole must be etched from the top surface of the first metal layer prior to formation of the metal plug thereon; otherwise, residual high-resistivity dielectric material within the via hole significantly degrades the contact resistance of the interconnect. Similar process control is required during the etching of metal layers (e.g., Al, Cu, Pt, etc.), polysilicon layers and the like.
Conventional process monitoring techniques provide only a rough estimate of when a material layer has been completely etched (i.e., endpoint). Accordingly, to accommodate varying thicknesses of material layers (e.g., device variations) or varying etch rates of material layers (e.g., process/process chamber variations), an etch process may be continued for a time greater than a predicted time for etching the material layer (i.e., for an over-etch time). Etching for an over-etch time ensures that all material to be removed is removed despite device variations that increase the required etch time and despite process/process chamber variations which slow etch rate (and thus increase the required etch time).
While over-etch times ensure complete etching, over-etching raises a number of issues. Overetching increases the time required to process each semiconductor wafer, and thus decreases wafer throughput. Moreover, the drive for higher performance integrated circuits requires each generation of semiconductor devices to have finer dimensional tolerances, making over-etching increasingly undesirable. Overetching also prolongs exposure of the wafer to a plasma environment, affecting the heat budget of the process, generating additional particles that could contaminate the wafer, and consuming expensive process materials.
A more attractive solution is an in situ monitoring technique that more accurately identifies significant processing events such as etch endpoint, chamber clean endpoint, and chamber seasoning. However, conventional monitoring techniques do not track progress of a semiconductor fabrication process accurately enough to reduce over-etch or other over-processing times required to compensate for both process/process chamber variations and device variations (e.g., material layer thickness variations, etch property variations, etc.). Moreover, previous attempts to correlate fluctuation in output data with process events have considered changes in output data over time periods on the order of seconds.
Accordingly, a need exists for an improved method and apparatus for monitoring semiconductor processes.
SUMMARY OF THE INVENTION
The present inventors have discovered that detecting fluctuation in the output data of a semiconductor fabrication process over extremely short time periods can provide previously unavailable information concerning the progress of the fabrication process. Embodiments of the present invention allow progress of a semiconductor fabrication process to be monitored by detecting fluctuations in output from the process over an extremely short time period of 10 milliseconds or less. For example, in accordance with one embodiment of the present invention, endpoint of a plasma chamber cleaning process may be detected by measuring optical emissions from a plasma chamber at a rate of>1 kHz, and then calculating standard deviation in optical emissions based upon a local time period of one second or less. Endpoint of the chamber cleaning process is indicated when standard deviation of optical emission attains a steady-state minimum value.
Alternatively, endpoint of a plasma chamber cleaning process may be determined by performing a Fast Fourier Transformation (FFT) to resolve the emission output data into frequency and amplitude components, and then identifying the point at which total power of optical emissions taken over all relevant frequencies attains a steady state.
Yet further alternatively, where optical emissions are absent from the plasma chamber due to a lack of electrons, endpoint of a plasma chamber cleaning process may be determined by monitoring fluctuations in optical emissions of a plasma cell positioned downstream from the plasma chamber and receiving exhaust from the plasma chamber.
While the above description relates to detecting endpoint of a plasma process by monitoring fluctuation in optical data output by the process, the invention is not limited to this particular application. The progress of a process could also be monitored by detecting fluctuation of other types of output signals, including but not limited to RF power fluctuations, temperature fluctuations, pressure fluctuations, and fluctuations in readings of a mass-spectrometer receiving by-products of the process.
One embodiment of a method in accordance with the present invention comprises measuring a value of an output from the semiconductor fabrication process, characterizing a fluctuation in the value of the output over a time period of 10 milliseconds or less, and correlating the fluctuation to an event of the semiconductor fabrication process.
One embodiment of an apparatus for processing a substrate in accordance with the present invention comprises a substrate processing chamber; and a sensor operatively coupled to said substrate processing chamber to detect a value of an output from the chamber. A computer processor is operatively coupled to said sensor; and a memory is coupled to said computer processor. The memory stores a computer program in computer readable format including computer instructions to permit said processor to measure a value of an output from the process, and characterize a fluctuation in the value of the output over a time period of 10 milliseconds or less.
REFERENCES:
patent: 4312732 (1982-01-01), Degenkolb et al.
patent: 4948259 (1990-08-01), Enke et al.
patent: 5374327 (1994-12-01), Imahashi et al.
patent: 5552016 (1996-09-01), Ghanayem
patent: 5846373 (1998-12-01), Pirkle et al.
patent: 6046796 (2000-04-01), Markle et al.
patent: 6153115 (2000-11-01), Le et al.
patent: 6246972 (2001-06-01), Klimasauskas
patent: 10081965 (1998-03-01), None
patent: WO 99/21210 (1999-04-01), None
patent: WO 99/40617 (1999-08-01), None
Department of Chemical Engineering, University of Texas, Model-Based control in Microelectronics Manufacturing, IEEE 1999, p. 4185-4191.*
Sovarong Leang, A control System for Photolithographic Sequqnces, IEEE 1996, p. 191-207.*
Moshe Sarfaty et al., “Temporal fluctuations: A fingerprint of surface chemical reactions”, Applied Physics Letters, vol. 76, No. 24, Jun. 12, 2000, pp. 3641-3643.
Ben-Dov Yuval
Garachtchenko Alexander Viktorovich
Sarfaty Moshe
Applied Materials Inc.
Bahta Kidest
Picard Leo
Townsend and Townsend and Crew
LandOfFree
Detection of process endpoint through monitoring fluctuation... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Detection of process endpoint through monitoring fluctuation..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Detection of process endpoint through monitoring fluctuation... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3312938