# US 2024/0316720 A1 · Substrate processing control system, substrate processing control method, and program

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2024/0316720 A1
- Document ID: US20240316720A1
- Status: Published application
- Kind: A1
- Application number: 18/734,791
- Filed: 2024-06-05
- Published: 2024-09-26
- Earliest claimed: 2016-10-18
- Assignees: Ebara Corporation
- Applicants: Ebara Corporation
- Inventors: Koichi TAKEDA; Tsuneo TORIKOSHI; Kunio OISHI; Katsuhide WATANABE; Hozumi YASUDA; Yu ISHII
- Web page: https://patentgrep.com/patent/US20240316720A1

## CPC

- B24B 37/005: Control means for lapping machines or devices
- B24B 37/12: Lapping plates for working plane surfaces
- B24B 37/14: characterised by the composition or properties of the plate materials
- B24B 49/03: according to the final size of the previously ground workpiece
- B24B 49/04: involving measurement of the workpiece at the place of grinding during grinding operation
- B24B 55/06: Dust extraction equipment on grinding or polishing machines (B24B31/12 takes precedence)
- H10P 52/00: Grinding, lapping or polishing of wafers, substrates or parts of devices
- B24B: MACHINES, DEVICES, OR PROCESSES FOR GRINDING OR POLISHING (grinding of gear teeth B23F, of screw-threads B23G1/36; by electro-erosion B23H; abrasive or related blasting B24C; tools for grinding, buffing or sharpening B24D; polishing compositions C09G1/00; abrasives C09K3/14; electrolytic etching or polishing C25F3/00; grinding arrangements for use on assembled railway tracks E01B31/17); DRESSING OR CONDITIONING OF ABRADING SURFACES; FEEDING OF GRINDING, POLISHING, OR LAPPING AGENTS
- H10P: GENERIC PROCESSES OR APPARATUS FOR THE MANUFACTURE OR TREATMENT OF DEVICES COVERED BY CLASS H10

## Abstract

A local polishing system comprises: a particle estimation unit (30) for estimating the film thickness distribution of a wafer; a local polishing region setting unit (11) for setting a local polishing region on the wafer based on the film thickness distribution; a polishing head selection unit (12) for selecting a polishing head based on the size of the local polishing region; a model storage (20) holding a recipe generating model that defines the relation between an input node and an output node, the input node being an attribute of the local polishing region, the output node comprising a recipe for a polishing process; a polishing recipe generator (13) that puts an attribute of the local polishing region set by the local polishing region setting unit (11) into the input node of the recipe generating model and determines a polishing recipe for polishing the local polishing region; and a polishing recipe transmitter (15) for transmitting data of the polishing recipe to a local polishing module (200) that performs local polishing.

## Claims

### Claim at position 1

1-47. (canceled)

### Claim 48

A method for controlling a polisher that locally polishes a substrate, the method comprising:
measuring a film thickness of the substrate and estimating the film thickness distribution of the substrate based on the measurement result;
setting a local polishing region on the substrate based on the film thickness distribution;
selecting a polishing head based on the size of the local polishing region;
storing a recipe generating model that defines the relation between an input node and an output node, input nodes including (i) an attribute of the local polishing region that includes the estimated average film thickness thereof and (ii) a head type, and the output node comprises a recipe for a polishing process;
generating a polishing recipe that puts an attribute of the local polishing region and the selected polishing head into the input node of the recipe generating model and determining a polishing recipe for polishing the local polishing region; and
transmitting data of the polishing recipe to a polisher that performs local polishing,
wherein the method controls local polishing to be performed on an entirely polished substrate, and
wherein the recipe generating model has a condition of the entire polishing as an input node.

### Claim 49 (depends on 48)

The method according to claim 48, wherein the input node comprises a plurality of input nodes that are the outline of the local polishing region.

### Claim 50 (depends on 48)

The method according to claim 48 wherein the local polishing region is set depending on whether the precision or throughput of substrate processing is given priority.

### Claim 51 (depends on 48)

The method according to claim 48, wherein the film thickness distribution is also estimated based on an etching pattern of a substrate to be polished.

### Claim 52 (depends on 48)

The method according to claim 48, further comprising:
simulating the film thickness distribution of a substrate and throughput to be achieved by local polishing according to the polishing recipe and determining whether the polishing recipe is to be adopted or not based on a simulation result.

### Claim 53 (depends on 52)

The method according to claim 52,
wherein a plurality of recipe generating models of different types are stored,
wherein a plurality of polishing recipes using different recipe generating models are determined, and
wherein a single polishing recipe is determined based on the simulation results of the plurality of polishing recipes determined by using different recipe generating models.

### Claim 54 (depends on 53)

The method according to claim 53, further comprising:
performing learning on the recipe generating model based on results of simulations performed on a plurality of local polishing regions by the simulator.

### Claim 55 (depends on 53)

The method according to claim 53, further comprising:
performing learning on the recipe generating model based on a result of local polishing actually performed according to the polishing recipe.

### Claim 56 (depends on 52)

The method according to claim 52, further comprising:
storing a previously generated polishing recipe and data on the film thickness distribution of a substrate polished by using the polishing recipe; and
reading from an actual result data storage a predetermined number of polishing recipes approximate to the film thickness distribution of a substrate estimate, and
wherein the simulating determines a single polishing recipe based on simulation results of a plurality of polishing recipes.

### Claim 57 (depends on 52)

The method according to claim 52, further comprising:
outputting the simulation result obtained by the simulator.

### Claim 58 (depends on 52)

The method according to claim 52, wherein the simulating uses a simulation model.

### Claim 59 (depends on 58)

The method according to claim 58, wherein the simulating performs learning based on a result of local polishing actually performed according to the polishing recipe.

### Claim 60 (depends on 48)

The substrate processing control system according to claim 48, further comprising:
receiving data on local polishing transmitted from a plurality of polishers connected to the data receiver via a network.

### Claim 61

A non-transitory memory storing instructions for performing steps for controlling a polisher that locally polishes a substrate, the steps comprising:
measuring a film thickness of the substrate and estimating the film thickness distribution of the substrate based on the measurement result;
setting a local polishing region on the substrate based on the film thickness distribution;
selecting a polishing head based on the size of the local polishing region;
storing a recipe generating model that defines the relation between an input node and an output node, input nodes including (i) an attribute of the local polishing region that includes the estimated average film thickness thereof and (ii) a head type, and the output node comprises a recipe for a polishing process;
generating a polishing recipe that puts an attribute of the local polishing region and the selected polishing head into the input node of the recipe generating model and determining a polishing recipe for polishing the local polishing region; and
transmitting data of the polishing recipe to a polisher that performs local polishing,
wherein the method controls local polishing to be performed on an entirely polished substrate, and
wherein the recipe generating model has a condition of the entire polishing as an input node.

### Claim 62 (depends on 61)

The non-transitory memory according to claim 61, wherein the input node comprises a plurality of input nodes that are the outline of the local polishing region.

### Claim 63 (depends on 61)

The non-transitory memory according to claim 61 wherein the local polishing region is set depending on whether the precision or throughput of substrate processing is given priority.

### Claim 64 (depends on 61)

The non-transitory memory according to claim 61, wherein the film thickness distribution is also estimated based on an etching pattern of a substrate to be polished.

### Claim 65 (depends on 61)

The non-transitory memory according to claim 61, the steps further comprising:
simulating the film thickness distribution of a substrate and throughput to be achieved by local polishing according to the polishing recipe and determining whether the polishing recipe is to be adopted or not based on a simulation result.

### Claim 66 (depends on 65)

The non-transitory memory according to claim 65,
wherein a plurality of recipe generating models of different types are stored,
wherein a plurality of polishing recipes using different recipe generating models are determined, and
wherein a single polishing recipe is determined based on the simulation results of the plurality of polishing recipes determined by using different recipe generating models.

### Claim 67 (depends on 66)

The non-transitory memory according to claim 66, the steps further comprising:
performing learning on the recipe generating model based on results of simulations performed on a plurality of local polishing regions by the simulator.

### Claim 68 (depends on 66)

The non-transitory memory according to claim 66, the steps further comprising:
performing learning on the recipe generating model based on a result of local polishing actually performed according to the polishing recipe.

### Claim 69 (depends on 65)

The non-transitory memory according to claim 65, the steps further comprising:
storing a previously generated polishing recipe and data on the film thickness distribution of a substrate polished by using the polishing recipe; and
reading from an actual result data storage a predetermined number of polishing recipes approximate to the film thickness distribution of a substrate estimate, and
wherein the simulating determines a single polishing recipe based on simulation results of a plurality of polishing recipes.

### Claim 70 (depends on 65)

The non-transitory memory according to claim 65, the steps further comprising:
outputting the simulation result obtained by the simulator.

### Claim 71 (depends on 65)

The non-transitory memory according to claim 65, wherein the simulating uses a simulation model.

### Claim 72 (depends on 71)

The non-transitory memory according to claim 71, wherein the simulating performs learning based on a result of local polishing actually performed according to the polishing recipe.

### Claim 73 (depends on 61)

The non-transitory memory processing control system according to claim 61, the steps further comprising:
receiving data on local polishing transmitted from a plurality of polishers connected to the data receiver via a network.
