This paper presents a new approach for predicting thermodynamic properties of perovskites that harnesses deep learning and crystal structure fingerprinting based on Hirshfeld surface analysis. It is demonstrated that convolutional neural network methods capture critical features embedded in two-dimensional Hirshfeld surface fingerprints that enable a quantitative assessment of the formation energy of perovskites. Building on our recent work on lattice parameter prediction from Hirshfeld surface calculations, we show how transfer learning can be used to speed up the training of the neural network, allowing multiple properties to be trained using the same feature extraction layers. We also predict formation energies for various perovskite polymorphs, and our predictions are found to give generally improved performance over a well-established graph network method, but with the methods better suited to different types of datasets. Analysis of the structure types within the dataset reveals the Hirshfeld surface-based method to excel for the less symmetric and similar structures, while the graph network performs better for very symmetric and similar structures.
more »
« less
Machine-learning structural and electronic properties of metal halide perovskites using a hierarchical convolutional neural network
Abstract The development of statistical tools based on machine learning (ML) and deep networks is actively sought for materials design problems. While structure-property relationships can be accurately determined using quantum mechanical methods, these first-principles calculations are computationally demanding, limiting their use in screening a large set of candidate structures. Herein, we use convolutional neural networks to develop a predictive model for the electronic properties of metal halide perovskites (MHPs) that have a billions-range materials design space. We show that a well-designed hierarchical ML approach has a higher fidelity in predicting properties of the MHPs compared to straight-forward methods. In this architecture, each neural network element has a designated role in the estimation process from predicting complex features of the perovskites such as lattice constant and octahedral till angle to narrowing down possible ranges for the values of interest. Using the hierarchical ML scheme, the obtained root-mean-square errors for the lattice constants, octahedral angle and bandgap for the MHPs are 0.01 Å, 5°, and 0.02 eV, respectively. Our study underscores the importance of a careful network design and a hierarchical approach to alleviate issues associated with imbalanced dataset distributions, which is invariably common in materials datasets.
more »
« less
- Award ID(s):
- 1809085
- NSF-PAR ID:
- 10418331
- Date Published:
- Journal Name:
- npj Computational Materials
- Volume:
- 6
- Issue:
- 1
- ISSN:
- 2057-3960
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Abstract Modern machine learning (ML) and deep learning (DL) techniques using high-dimensional data representations have helped accelerate the materials discovery process by efficiently detecting hidden patterns in existing datasets and linking input representations to output properties for a better understanding of the scientific phenomenon. While a deep neural network comprised of fully connected layers has been widely used for materials property prediction, simply creating a deeper model with a large number of layers often faces with vanishing gradient problem, causing a degradation in the performance, thereby limiting usage. In this paper, we study and propose architectural principles to address the question of improving the performance of model training and inference under fixed parametric constraints. Here, we present a general deep-learning framework based on branched residual learning (BRNet) with fully connected layers that can work with any numerical vector-based representation as input to build accurate models to predict materials properties. We perform model training for materials properties using numerical vectors representing different composition-based attributes of the respective materials and compare the performance of the proposed models against traditional ML and existing DL architectures. We find that the proposed models are significantly more accurate than the ML/DL models for all data sizes by using different composition-based attributes as input. Further, branched learning requires fewer parameters and results in faster model training due to better convergence during the training phase than existing neural networks, thereby efficiently building accurate models for predicting materials properties.more » « less
-
The unique physical properties of two-dimensional (2D) metal halide perovskites (MHPs) such as nonlinear optics, anisotropic charge transport, and ferroelectricity have made these materials promising candidates for multifunctional applications. Recently, fluorine derivatives such as 4,4-difluoropiperidinium lead iodide perovskite or (4,4-DFPD, C 5 H 10 F 2 N) 2 PbI 4 have shown strong ferroelectricity as compared to other 2D MHPs. Although it was previously addressed that the ferroelectricity in MHPs can be affected by illumination, the underlying physical mechanisms of light–ferroelectricity interaction in 2D MHPs are still lacking. Here, we explore the electromechanical responses in 4,4-(DFPD) 2 PbI 4 thin films using advanced scanning probe microscopy techniques revealing ferroelectric domain structures. Hysteretic ferroelectric loops measured by contact-Kelvin probe force microscopy are dependent on domain structures under dark conditions, while ferroelectricity weakens under illumination. The X-ray diffraction patterns exhibit significant changes in preferential orientation of individual lattice planes under illumination. Particularly, the reduced intensity of the (1 1 1) lattice plane under illumination leads to transitioning from a ferroelectric to a paraelectric phase. The instability of positive ions, especially molecular organic cations, is observed under illumination by time-of-flight secondary ion mass spectrometry. The combination of crystallographic orientation and chemical changes under illumination clearly contributes to the origin of light–ferroelectricity interaction in 2D (4,4-DFPD, C 5 H 10 F 2 N) 2 PbI 4 .more » « less
-
Structural hierarchy, in which materials possess distinct features on multiple length scales, is ubiquitous in nature. Diverse biological materials, such as bone, cellulose, and muscle, have as many as 10 hierarchical levels. Structural hierarchy confers many mechanical advantages, including improved toughness and economy of material. However, it also presents a problem: Each hierarchical level adds a new source of assembly errors and substantially increases the information required for proper assembly. This seems to conflict with the prevalence of naturally occurring hierarchical structures, suggesting that a common mechanical source of hierarchical robustness may exist. However, our ability to identify such a unifying phenomenon is limited by the lack of a general mechanical framework for structures exhibiting organization on disparate length scales. Here, we use simulations to substantiate a generalized model for the tensile stiffness of hierarchical filamentous networks with a nested, dilute triangular lattice structure. Following seminal work by Maxwell and others on criteria for stiff frames, we extend the concept of connectivity in network mechanics and find a similar dependence of material stiffness upon each hierarchical level. Using this model, we find that stiffness becomes less sensitive to errors in assembly with additional levels of hierarchy; although surprising, we show that this result is analytically predictable from first principles and thus potentially model independent. More broadly, this work helps account for the success of hierarchical, filamentous materials in biology and materials design and offers a heuristic for ensuring that desired material properties are achieved within the required tolerance.more » « less
-
more » « less
Largely due to superior properties compared to traditional materials, the use of polymer matrix composites (PMC) has been expanding in several industries such as aerospace, transportation, defense, and marine. However, the anisotropy and nonhomogeneity of these structures contribute to the difficulty in evaluating structural integrity; damage sites can occur at multiple locations and length scales and are hard to track over time. This can lead to unpredictable and expensive failure of a safety-critical structure, thus creating a need for non-destructive evaluation (NDE) techniques which can detect and quantify small-scale damage sites and track their progression. Our research group has improved upon classical microwave techniques to address these needs; utilizing a custom device to move a sample within a resonant cavity and create a spatial map of relative permittivity. We capitalize on the inevitable presence of moisture within the polymer network to detect damage. The differing migration inclinations of absorbed water molecules in a pristine versus a damaged composite alters the respective concentrations of the two chemical states of moisture. The greater concentration of free water molecules residing in the damage sites exhibit highly different relative permittivity when compared to the higher ratio of polymer-bound water molecules in the undamaged areas. Currently, the technique has shown the ability to detect impact damage across a range of damage levels and gravimetric moisture contents but is not able to specifically quantify damage extent with regards to impact energy level. The applicability of machine learning (ML) to composite materials is substantial, with uses in areas like manufacturing and design, prediction of structural properties, and damage detection. Using traditional NDE techniques in conjunction with supervised or unsupervised ML has been shown to improve the accuracy, reliability, or efficiency of the existing methods. In this work, we explore the use of a combined unsupervised/supervised ML approach to determine a damage boundary and quantification of single-impact specimens. Dry composite specimens were damaged via drop tower to induce one central impact site of 0, 2, or 3 Joules. After moisture exposure, Entrepreneur Dr, Raleigh, North Carolina 27695, U.S.A. 553 each specimen underwent dielectric mapping, and spatial permittivity maps were created at a variety of gravimetric moisture contents. An unsupervised K-means clustering algorithm was applied to the dielectric data to segment the levels of damage and define a damage boundary. Subsequently, supervised learning was used to quantify damage using features including but not limited to thickness, moisture content, permittivity values of each cluster, and average distance between points in each cluster. A regression model was trained on several samples with impact energy as the predicted variable. Evaluation was then performed based on prediction accuracy for samples in which the impact energies are not known to the model.
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Journal Article:
- https://doi.org/10.1038/s41524-020-0307-8
