1.1Post-polish point checks
Displacement probes check radial position at two fixed bevel heights during one revolution. The acquisition part records only the minimum and maximum readings at each height and compares them with dimensional limits.
Generated automatically. Not reviewed by a person. Not legal advice or an opinion.
Questions about how individual parts of claim 1 might be changed. A question may describe something different from what the claim language recites, or a version of it; no view is offered on which.
Displacement probes check radial position at two fixed bevel heights during one revolution. The acquisition part records only the minimum and maximum readings at each height and compares them with dimensional limits.
A numerical solver calculates the pressure distribution from the incoming edge geometry, pad compliance and commanded head load. It integrates a material removal equation over the dwell time using relative sliding speed and a removal coefficient measured on separate material coupons, then subtracts the calculated removal from the incoming profile.
The head retracts after each short pass, and an optical displacement gauge scans the rinsed bevel while the stage makes one revolution. The controller compares the measured profile with the target and assigns another short pass only to zones with excess material.
The controller calculates the maximum radial deviation between the predicted profile and the target over the bevel. A transfer robot loads substrates within the deviation limit into the polishing station under the fixed recipe and sends the others to an inspection buffer.
Further arrangements of the same subject matter, described here as of the publication date at the foot of this page.
A spindle holds and rotates a circular glass lens blank while a polishing head presses a compliant pad against its rim under a cerium oxide slurry. A displacement sensor traverses the rim before and after polishing to measure the radial profile across the edge thickness. A model trained on paired profiles and slurry flow settings predicts the finished rim profile for a new blank. A controller compares the prediction with the specified bevel profile and updates the slurry flow setpoint within the pump limits.
A Gaussian process regressor serves as the learned profile predictor in the controller of a rotating edge polisher. The controller resamples paired edge profiles on a common grid, represents each measured removal profile with spline coefficients, and trains a regressor for each coefficient using the incoming profile and recorded polishing settings. During operation, the controller reconstructs the predicted removal profile and subtracts it from the incoming profile to obtain the predicted finished edge. It evaluates candidate liquid flow setpoints and selects the one with the smallest predicted squared deviation from the target profile.
An encoder indexes a 300 mm substrate at 72 angular stations spaced 5 degrees apart, and a displacement sensor records 81 points across the edge at each station with 25 micrometre spacing. The acquisition system registers the before and after scans to the same angular and cross-edge coordinates. A model uses the incoming profiles and polishing settings to predict the finished profiles at all 72 stations. The controller selects a liquid flow setpoint that minimizes the root mean square deviation from the target across the full measurement grid.
A rotating edge polisher receives slurry from a recirculation loop with an agitated reservoir, a filter, a temperature sensor, and a metering pump on the tool supply branch. The acquisition system stores the measured branch flow and slurry temperature for each run alongside the paired edge profiles. The model uses the incoming edge profile and measured slurry conditions to predict the finished edge at candidate branch flow rates. The controller selects the flow rate with the smallest predicted deviation from the target and sends it to the metering pump controller.
A displacement probe sits beside the polishing head and scans the edge after the head retracts and a rinse clears the polishing liquid. Software registers the scan to the spindle encoder and passes the measured profile and candidate settings to a model trained on paired scans from individual passes. The controller selects the liquid flow setpoint for the next pass by comparing predicted profiles with the target. It repeats the scan after each pass and stops when the measured profile meets the dimensional tolerance or the pass limit is reached.
A displacement gauge measures the unloaded polishing pad surface at a fixed head position and derives pad thickness from the measured position of the backing plate. The acquisition system records pad thickness before each run together with the polishing settings and paired edge profiles. A learned model uses the current pad thickness, incoming edge profile, and candidate settings to predict the finished profile. The controller compares the predictions with the target and updates the polishing liquid flow setpoint for the measured pad thickness.
An optical displacement sensor scans the upper bevel, rim and lower bevel before and after polishing, with samples registered to spindle encoder counts and axial sensor position. The acquisition unit stores both profiles with the workpiece identifier, head force, rotation speed, liquid flow and dwell time.
The prediction unit represents each incoming profile by spline coefficients and supplies those coefficients with head force, stage speed and dwell time to a trained regression model. Training uses paired incoming and outgoing profiles from recorded runs with their polishing conditions, and the prediction unit reconstructs the outgoing profile from the predicted spline coefficients.
The correction unit compares predicted and target radii at sampled points across the bevel and calculates a dwell time adjustment for each head position using a removal sensitivity matrix. A bounded least-squares solver limits each revised dwell time to its configured range and sends the revised recipe to the polishing controller.
Passages reproduced word for word from two earlier-published U.S. patent documents, placed beside the language of claim 1. No characterization is offered. Each passage is marked A or B for the document it comes from; ¶ numbers are the paragraph numbers in the USPTO’s published full text of that document.
According to this method, the film thickness of the wafer is measured before the polishing, and the polishing conditions are so prepared that the polishing conditions such as polishing speed, polishing pressure, abrasives and the like become optimal.A · US 2008/0248723 A1 · ¶ 31
A polishing system comprising: a polishing apparatus in which an object is polished by causing relative motion between a tool and the object in a state in which abrasive grains are interposed between the tool and the object, a means for inputting the surface shape of the object prior to polishing, and a target surface shape of the object following polishing, a means for establishing a plurality of polishing conditions, a means for determining a respective polished shape of eaB · US 7,686,673 B2 · claim 9
The present embodiment is applied to a CMP apparatus that polishes a kind or two or more kinds of films to be polished, and before rough polishing or finish polishing of the wafers, the film thickness of the wafers are measured and the optimal polishing conditions are prepared, and the polishing time of the wafers is forecasted on the basis of the polishing conditions and the measured value, and further, the actual polishing time of the wafers is measured, and the polishing cA · US 2008/0248723 A1 · ¶ 47
A polishing system comprising: a polishing apparatus in which an object is polished by causing relative motion between a tool and the object in a state in which abrasive grains are interposed between the tool and the object, a shape supplier which inputs the surface shape of the object prior to polishing, and a target surface shape of the object following polishing, a correlator which determines a relationship between polishing conditions and the respective polished shape thaB · US 7,686,673 B2 · claim 23
The data of the polishing time and the polishing quantity and the like acquired by the polishing process of wafers are controlled for each polishing step, or each polishing shaft and/or each platen, the polishing conditions are corrected/changed immediately so that the polishing quantity of the wafers becomes the target value.A · US 2008/0248723 A1 · ¶ 48
If this is done, working is actually performed under the determined working conditions, and the error between the specified shape that constitutes the target and the actual worked shape is determined. Then, combinations of working conditions constituting elements that will work the amount of this error are determined, and these combinations are added to the original working conditions so that new combinations of working conditions are determined.B · US 7,686,673 B2 · ¶ 48
An empty cell means only that no passage was reproduced there.
The present invention relates to a polishing condition control apparatus and a polishing condition control method of a CMP apparatus, and particularly, it relates to a polishing condition control apparatus and a polishing condition control method of a CMP apparatus for optimizing the polishing condition of a wafer.A · US 2008/0248723 A1 · ¶ 2
In recent years, the design rules of semiconductor technologies have become toward ultrafine configurations and multilayer wiring, and the diameter of wafers has become larger, and consequently in the CMP process, requirements for polishing precision and polishing speed have become severer.A · US 2008/0248723 A1 · ¶ 4
In order to deal with such increasing shallowness of the focal depth, there is a demand for even greater planarization of the surfaces of semiconductor devices than that achieved so far.B · US 7,686,673 B2 · ¶ 3
Currently, the CMP technique is the sole method that can be used to planarize the entire surface of a wafer.B · US 7,686,673 B2 · ¶ 5
patent.fyi, US 2026/0355897 A1, claim 1, first published 2026-11-19T16:57:22-05:00, https://patent.fyi/US20260355897/, SHA-256 89346ae8294728b8e037910cf7d109a204b7a7314836adfb5e537bf113f6ae31