Applicable Fields

Materials Science, Solid-State Physics, Inorganic Chemistry, Chemical Physics, etc. Especially, Organic Chemistry and Biological Physics are out of scope.

Materials Discovery

Using the app integrated with the AI technology based on our research, you can conduct materials screening and may discover candidate compounds. Don't hesitate use the app to facilitate your frontier research.

Confidence

Every compound is given a predictive confidence according to its predictive variance. You can easily determine how reliable the system prediction is, taking a look at the confidence chart.

Customizable AI

You can increase the ensemble size of the machine learning model to make formation energy predictions more accurate, and the prototype database scale to gain more insight into the chemical system. The variance threshold is adjustable, so you can control the number of very low confident predictions.

The prediction errors are less than the theoretical database uncertainty estimated by a comparison of the Open Quantum Materials Database (OQMD) and the Materials Project (MP) database.

You can learn more about the AI technology used in the app reading the in-app documents and the reference papers.

On-Device Inference

The app uses no external resources to make a system prediction on your device. You can watch the phase diagram redrawn at the end of each stage.

Performace

Thanks to the efficient graph neural network architecture, the optimized graph generator, and the scale-invariant crystal graphs, the system prediction takes only a few seconds to minutes depending on the creation settings and your device. Moreover, optional multithreading accelerates the core of the AI tool - for example, on M1 iPad Air, the most computationally intensive tasks are sped up around 2.5x and take only about a minute.

Security and Privacy

The app is a security-first AI tool for scientists. We don't track your activity in the app nor collect your data from the app. Your privacy therefore is maintained while you use the app. More info at the app's Privacy Policy.

Requirements

The app requires iOS (iPadOS) 17.2 or later and a device with the A12 Bionic chip or later.

Get iCrucible on the App Store now and see how helpful the AI app is.

Try it free

Tips

Compound Screening

Adjust the stability threshold slider in the tool sheet to mark stable and metastable compounds with red-filled circles.

Overall Confidence

Tap the chart icon in the bottom bar to take a look at the confidence chart. You can determine how reliable the system prediction is.

Compound Table

Tap the table icon in the bottom bar to see the list of the stable and metastable compounds whose stability is lower than the stability threshold. They are sorted in increasing order by stability. You can easily pick up candidate compuounds from the table.

Preferences

Tap the settings icon to change default values for the phase diagram drawing. Especially, the initial mark size may be too small.

Support

If you have any questions or need help, please see the app's Support page.

AI Helps You Find New Compounds

❮ ❯

iCrucible

Scan to get the app

Research

Crystal Graph Gallery

Jadeite & Benitoite.

Crystal Graph Neural Networks for Data Mining in Materials Science

PDF Code

Abstract

Machine learning methods have been employed for materials prediction in various ways. It has recently been proposed that a crystalline material is represented by a multigraph called a crystal graph. Convolutional neural networks adapted to those graphs have successfully predicted bulk properties of materials with the use of equilibrium bond distances as spatial information. An investigation into graph neural networks for small molecules has recently shown that the no distance model performs almost as well as the distance model. This paper proposes crystal graph neural networks (CGNNs) that use no bond distances, and introduces a scale-invariant graph coordinator that makes up crystal graphs for the CGNN models to be trained on the dataset based on a theoretical materials database. The CGNN models predict the bulk properties such as formation energy, unit cell volume, band gap, and total magnetization for every testing material, and the average errors are less than the corresponding ones of the database. The predicted band gaps and total magnetizations are used for the metal-insulator and nonmagnet-magnet binary classifications, which result in success. This paper presents discussions about high- throughput screening of candidate materials with the use of the predicted formation energies, and also about the future progress of materials data mining on the basis of the CGNN architectures.

(Published 27 May 2019)

1/6

CGNN Architecture

2/6

Diamond Cubic Si

3/6

α-SiO₂

4/6

Pyrope

5/6

Li₃Nd₃(WO₆)₂ garnet

6/6

OQMD-MP Volume Errors

1/6

CGNN Architecture

2/6

Diamond Cubic Si

3/6

α-SiO₂

4/6

Pyrope

5/6

Li₃Nd₃(WO₆)₂ garnet

6/6

OQMD-MP Volume Errors

❮ ❯

OQM9HK: A Large-Scale Graph Dataset for Machine Learning in Materials Science

PDF Code

Abstract

We introduce a large-scale dataset of quantum-mechanically calculated properties of crystalline materials for graph representation learning that contains approximately 900k entries (OQM9HK). This dataset is constructed on the basis of the Open Quantum Materials Database (OQMD) v1.5 containing more than one million entries, and is the successor to the OQMD v1.2 dataset containing approximately 600k entries (OQM6HK). We develop the graph creation algorithm to produce a binary edge-labeled (BEL) graph representing a crystalline material. The BEL graph has higher representability of crystal structure than the edge-unlabeled ones. In materials property prediction tasks, crystal graph neural networks trained on the BEL graph dataset perform better than ones on the other graph datasets. The OQM9HK graph dataset is available at the Zenodo repository, here.

(Published 30 September 2022, Updated 9 October 2022)
First version

1/5

Histograms of Properties

2/5

Ice I_h

3/5

Li-O in Li₃Nd₃(WO₆)₂ garnet

4/5

Nd-W in Li₃Nd₃(WO₆)₂ garnet

5/5

BEL CGNN Architecture

1/5

Histograms of Properties

2/5

Ice I_h

3/5

Li-O in Li₃Nd₃(WO₆)₂ garnet

4/5

Nd-W in Li₃Nd₃(WO₆)₂ garnet

5/5

BEL CGNN Architecture

❮ ❯

About Us