# US 12,112,836 B2 · Artificial intelligence directed zeolite synthesis

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

- Number: US 12,112,836 B2
- Document ID: US12112836B2
- Status: Granted
- Kind: B2
- Application number: 17/346,463
- Filed: 2021-06-14
- Published: 2022-12-15
- Granted: 2024-10-08
- Earliest claimed: 2021-06-14
- Assignees: Chevron U.S.A. Inc.
- Applicants: Chevron U.S.A. Inc.
- Inventors: Larry A. Bowden, Jr.; Dan Xie; Christopher Michael Lew; Joel Edward Schmidt
- Web page: https://patentgrep.com/patent/US12112836B2

## CPC

- G16C 20/70: Machine learning, data mining or chemometrics
- G06N 3/042: Knowledge-based neural networks; Logical representations of neural networks
- G06N 3/0455: Auto-encoder networks; Encoder-decoder networks
- G06N 3/09: Supervised learning
- G16C 20/10: Analysis or design of chemical reactions, syntheses or processes
- G16C 20/40: Searching chemical structures or physicochemical data
- G16C 20/50: Molecular design, e.g. of drugs
- G16C 60/00: Computational materials science, i.e. ICT specially adapted for investigating the physical or chemical properties of materials or phenomena associated with their design, synthesis, processing, characterisation or utilisation
- B01J 29/04: having base-exchange properties, e.g. crystalline zeolites {(B01J29/005 takes precedence)}
- G16C: COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- B01J: CHEMICAL OR PHYSICAL PROCESSES, e.g. CATALYSIS OR COLLOID CHEMISTRY; THEIR RELEVANT APPARATUS

## Abstract

A computer implemented method for designing chemical reactions for catalyst construction is described. The method includes extracting historical data including historic chemical reaction data and historic catalyst construction yield data and converting the historic chemical reaction data into graph models to represent molecular structure data. The method also includes incorporating the graph models into a chemical reaction algorithm and training a vectorized cognitive deep learning network of the chemical reaction algorithm by using the graph models and a property of the historic chemical reaction data to produce a catalyst chemical reaction model. Further, the method includes validating the catalyst chemical reaction model by inputting the historic chemical reaction data and comparing a generated property corresponding to the catalyst chemical reaction model to the property of the historic chemical reaction data. Lastly, the method includes updating the training of the catalyst chemical reaction model.

## Claims

### Claim 1

A computer implemented method for catalyst construction, the method comprising:
extracting historical data including historic chemical reaction data and historic catalyst construction yield data;
converting the historic chemical reaction data into graph models to represent molecular structure data;
incorporating the graph models representing the molecular structure data into a chemical reaction algorithm;
training a vectorized cognitive deep learning network of the chemical reaction algorithm by using the graph models and a property of the historic chemical reaction data to produce a catalyst chemical reaction model;
validating the catalyst chemical reaction model by inputting the historic chemical reaction data and comparing a generated property corresponding to the catalyst chemical reaction model to the property of the historic chemical reaction data; and
updating, based on a difference between the generated property and the property of the historic chemical reaction data being larger than a threshold value, the training of the catalyst chemical reaction model.

### Claim 2 (depends on 1)

The computer implemented method of claim 1, further comprising:
identifying raw chemical reactants and desired catalyst properties;
determining chemical reactions in accordance with the raw chemical reactants and the desired catalyst properties;
evaluating catalyst construction yields in accordance with the chemical reactions through the chemical reaction model; and
creating, from the chemical reactions, a list of catalyst formulations corresponding to a maximum catalyst construction yield.

### Claim 3 (depends on 1)

The computer implemented method of claim 1, wherein the graph models include nodes representing atoms with specific molecular properties and vertices representing intermolecular properties, wherein the molecular properties include atom surface area, atom volume, and atom surface energy, and wherein the intermolecular properties represented by the vertices include chemical bond type and chemical bond angle.

### Claim 4 (depends on 3)

The computer implemented method of claim 3, wherein the nodes included in the graph models also represent chemical reaction steps, chemical reaction parameters, and organic structure directing agents in related chemical reaction steps.

### Claim 5 (depends on 2)

The computer implemented method of claim 2, further comprising designing of catalyst chemical structures based on the chemical reactions and the catalyst formulations.

### Claim 6 (depends on 5)

The computer implemented method of claim 5, wherein the catalyst chemical structures reveal topological, geometrical, and tiling information of the catalyst.

### Claim 7 (depends on 2)

The computer implemented method of claim 2, wherein the catalyst formulations are created in accordance with a minimum catalyst construction yield.

### Claim 8 (depends on 1)

The computer implemented method of claim 1, wherein the historic chemical reaction data and the historic catalyst construction yield data are extracted from historic information including literature reports, experiment notes, and manufacturing records.

### Claim 9 (depends on 1)

The computer implemented method of claim 1, further comprising designing of chemical reactions for producing a zeolite.

### Claim 10

A computer implemented method for constructing a catalyst structure, the method comprising:
extracting historic catalyst production information from operational history;
constructing, based on the historic catalyst production information, graph data representing synthesis recipes and processes for catalyst production;
training a neural network of a catalyst production algorithm by using the graph data to generate a catalyst production model; and
validating the catalyst production model by inputting the graph data and comparing a catalyst generated from the catalyst production model with catalyst corresponding to the synthesis recipes and processes in the graph data.

### Claim 11 (depends on 10)

The computer implemented method of claim 10, further comprising:
inputting desired catalyst structure and corresponding chemical reaction information into the catalyst production model;
generating, based on the desired catalyst structure and corresponding chemical reaction information, production graph data representing catalyst production information through the catalyst production model;
converting the production graph data to the catalyst production information including production synthesis recipes and processes; and
forwarding the catalyst production information to a production line for automatic catalyst production.

### Claim 12 (depends on 11)

The computer implemented method of claim 11, wherein the corresponding chemical reaction information is designed by:
extracting historical data including historic chemical reaction data and historic catalyst construction yield data;
converting the historic chemical reaction data into reaction graph data and molecular structure data;
incorporating the reaction graph data and the molecular structure data into a chemical reaction algorithm;
training a vectorized cognitive deep leaning network of the chemical reaction algorithm by using the reaction graph data, molecular structure data, and historic catalyst construction yield data to produce a catalyst chemical reaction model;
validating the catalyst chemical reaction model by inputting the historic chemical reaction data and comparing a calculated catalyst construction yield generated by the chemical reaction model to the historic catalyst construction yield data; and
updating, based on a difference between the calculated catalyst construction yield and the historic catalyst construction yield data being larger than a threshold value, the training of the catalyst chemical reaction model.

### Claim 13 (depends on 12)

The computer implemented method of claim 12, further comprising:
identifying raw chemical reactants and desired catalyst properties;
determining chemical reactions in accordance with the raw chemical reactants and the desired catalyst properties;
evaluating catalyst construction yields in accordance with the chemical reactions through the chemical reaction model; and
creating, from the chemical reactions, a list of catalyst formulations corresponding to a maximum catalyst construction yield.

### Claim 14 (depends on 13)

The computer implemented method of claim 13, wherein the desired catalyst structure is constructed based on the chemical reactions and catalyst product formulations, and wherein the production line includes a robotic arm, ingredients, and production equipment.

### Claim 15 (depends on 11)

The computer implemented method of claim 11, wherein the constructing of the graph data representing synthesis recipes and processes is performed by an encoder of the neural network.

### Claim 16 (depends on 11)

The computer implemented method of claim 11, wherein the converting of the production graph data to the catalyst production information including production synthesis recipes and processes is performed by a decoder of the neural network.

### Claim 17 (depends on 10)

The computer implemented method of claim 10, wherein the historic catalyst production information can be further extracted from literature reports.

### Claim 18

An artificial intelligence assisted catalyst processing method implemented by a computer, the method comprising:
designing chemical reactions by training a vectorized cognitive deep learning network of a chemical reaction algorithm and inputting raw chemical reactants data and desired catalyst product properties to a chemical reaction model generated from the training of the chemical reaction algorithm;
designing a catalyst structure based on the chemical reactions that are designed in accordance with a maximum catalyst production yield and corresponding catalyst product formulations; and
producing the catalyst structure automatically by forwarding catalyst production information to a production line, wherein the catalyst production information is generated by inputting the chemical reactions that were designed and the catalyst structure that was designed into a catalyst production model.

### Claim 19 (depends on 18)

The artificial intelligence assisted catalyst processing method of claim 18, wherein the training of the vectorized cognitive deep learning network of the chemical reaction algorithm further includes:
extracting historical data including historic chemical reaction data and historic catalyst construction yield data;
converting the historic chemical reaction data into graph models to represent molecular structure data;
incorporating the graph models representing the molecular structure data into a chemical reaction algorithm;
training a vectorized cognitive deep learning network of the chemical reaction algorithm by using the graph models and a property of the historic chemical reaction data to produce a catalyst chemical reaction model;
validating the catalyst chemical reaction model by inputting the historic chemical reaction data and comparing a generated property corresponding to the catalyst chemical reaction model to the property of the historic chemical reaction data; and
updating, based on a difference between the generated property and the property of the historic chemical reaction data being larger than a threshold value, the training of the catalyst chemical reaction model.

### Claim 20 (depends on 18)

The artificial intelligence assisted catalyst processing method of claim 18, wherein the producing of the catalyst structure automatically further includes:
extracting historic catalyst production information from operational history;
constructing, based on the historic catalyst production information, graph data representing synthesis recipes and processes for catalyst production;
training a neural network of a catalyst production algorithm by using the graph data to generate the catalyst production model; and
validating the catalyst production model by inputting the graph data and comparing a catalyst generated from the catalyst production model with catalyst corresponding to the synthesis recipes and processes in the graph data.