Generation of Large-Scale 3D Microstructures from Surface Images: Application to Additive Manufacturing
Explore the source record for details and available documents.
SEARCH · Engineering Papers
Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.
Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.
Explore the source record for details and available documents.
3D information may be extracted from two 2D images by capturing a first image of a sample at a first orientation. The sample may be titled at a second or different orientation, resulting in a second image of the titled sample to be captured. Third dimension of information may be extracted from the images.
Abstract The design of materials structure for optimizing functional properties and potentially, the discovery of novel behaviors is a keystone problem in materials science. In many cases microstructural models underpinning materials functionality are available and well understood. However, optimization of average properties via microstructural engineering often leads to combinatorically intractable problems. Here, we explore the use of the reinforcement learning (RL) for microstructure optimization targeting the discovery of the physical mechanisms behind enhanced functionalities. We illustrate that RL can provide insights into the mechanisms driving properties of interest in a 2D discrete Landau ferroelectrics simulator. Intriguingly, we find that non-trivial phenomena emerge if the rewards are assigned to favor physically impossible tasks, which we illustrate through rewarding RL agents to rotate polarization vectors to energetically unfavorable positions. We further find that strategies to induce polarization curl can be non-intuitive, based on analysis of learned agent policies. This study suggests that RL is a promising machine learning method for material design optimization tasks, and for better understanding the dynamics of microstructural simulations.
To develop methods for quantifying the effects of the microstructural variations of woven ceramic matrix composites on the effective properties and response of the material, a research program has been undertaken which is described in this paper. In order to characterize and quantify the variations in the microstructure of a five harness satin weave, CVI SiC/SiC, composite material, specimens were serially sectioned and polished to capture images that detailed the fiber tows, matrix, and porosity. Open source quantitative image analysis tools were then used to isolate the constituents and collect relevant statistics such as within ply tow spacing. This information was then used to build two dimensional finite element models that approximated the observed section geometry. With the aid of geometrical models generated by the microstructural characterization process, finite element models were generated and analyses were performed to quantify the effects of the microstructure and its variation on the effective stiffness and areas of stress concentration of the material. The results indicated that the geometry and distribution of the porosity appear to have significant effects on the through-thickness modulus. Similarly, stress concentrations on the outer surface of the composite appear to correlate to regions where the transverse tows are separated by a critical amount.
Nanoparticle organic photovoltaics, a subfield of organic photovoltaics (OPV), has attracted increasing interest in recent years due to the eco-friendly fabrication of solar modules afforded by colloidal ink technology. Importantly, using this approach it is now possible to engineer the microstructure of the light absorbing/charge generating layer of organic photovoltaics; decoupling film morphology from film deposition. In this study, single-component nanoparticles of poly(3-hexylthiophene) (P3HT) and phenyl-C 61 butyric acid methyl ester (PC 61 BM) were synthesized and used to generate a two-phase microstructure with control over domain size prior to film deposition. Scanning transmission X-ray microscopy (STXM) and electron microscopy were used to characterize the thin film morphology. Uniquely, the measured microstructure was a direct input for a nanoscopic kinetic Monte Carlo (KMC) model allowing us to assess exciton transport properties that are experimentally inaccessible in these single-component particles. Photoluminescence, UV-vis spectroscopy measurements, and KMC results of the nanoparticle thin films enabled the calculation of an experimental exciton dissociation efficiency (η ED ) of 37% for the two-phase microstructure. The glass transition temperature (T g ) of the materials was characterized with dynamic mechanical thermal analysis (DMTA) and thermal annealing led to an increase in ηED to 64% due to an increase in donor-acceptor interfaces in the thin film from both sintering of neighboring opposite-type particles in addition to the generation of a third mixed phase from diffusion of PC 61 BM into amorphous P3HT domains. Finally, this study demonstrates the higher level of control over donor-acceptor film morphology enabled by customizing nanoparticulate colloidal inks, where the optimal three-phase film morphology for an OPV photoactive layer can be designed and engineered.
The Porous Microstructure Analysis (PuMA) software is a suite of tools for the analysis of porous materials and generation of material microstructures. From microstructural data, often obtained through X-ray microtomography, PuMA can determine a number of effective material properties and perform material response simulations. Version 2.2 includes capabilities for computing volume fractions, porosity, specific surface area, effective thermal and electrical conductivities, and continuum and rarefied diffusive tortuosity. PuMA can also simulate competitive diffusion/reaction processes at the micro-scale, such as surface oxidation. In this poster, recent advancements to the PuMA software are detailed, including the full refactoring of PuMA into v3.0, a new module to compute heat conduction in anisotropic materials, a particle method for simulating molecular beam experiments, a new finite-volume Laplace solver, complex fibrous material generation, woven material generation, and a coupling of PuMA with the DAKOTA software for advanced statistics.
TATB-based explosives have been investigated on length scales spanning several orders of magnitude, from just under 10 nm to larger than 1 cm. This has been accomplished using a combination of ultra-small angle x-ray scattering (USAXS), ultra-small angle neutron scattering (USANS), and x-ray computed tomography (XRCT). USAXS determines distributions the smallest structures including hot-spot voids from hundreds of nanometers to a few microns, USANS extends this range to about 10 microns, and two variants of XRCT cover sizes from microns to centimeters. Several examples are presented for LX-17, a triaminotrinitrobenzene based plastic bonded explosive using Kel-F 800. As an extension of previous USAXS results, in these proceedings, an alternate binder results in a more uniform microstructure for the PBX, useful towards design of next-generation TATB-based explosives. These data are an important step to understanding microstructural mechanisms that affect the mechanical properties of TATB-based explosives, and provide complete a comprehensive characterization of the structure of LX-17 from nanometers to centimeters that can be used as empirical input to computational models of detonation, and in determining the relationship between voids and microstructure to detonation properties.
The goal of this study is to characterize and validate the texture and grain topology of additively-manufactured anisotropic three-dimensional (3D) polycrystalline microstructures. The special focus is on developing methodologies to compare the grain shapes and orientations of two-dimensional (2D) and 3D microstructure representations using the same metric. To generate statistical data, synthetic microstructures are reconstructed from experimental data using Markov random field (MRF). The statistical similarity between the experimental and synthetic microstructures is verified by comparing their grain topologies. A universal measure to compare 2D and 3D grains is portrayed through the concept of image moments that are invariant to shape transformations. The graphical plots developed based on moment invariants to compare the 2D and 3D grains are used to verify the synthetic model
This study quantifies the interaction between volumetric defect location and size on the very high cycle fatigue (VHCF) of laser beam powder bed fused (LB-PBF) AlSi10Mg. Crystal plasticity finite element method (CPFEM) simulations were used to investigate the effects of defect location and size on the driving force for crack initiation. The CPFEM model was calibrated against uniaxial and cyclic experimental data of LB-PBF AlSi10Mg. Defect characteristics were informed by experimental data from the specimens produced in various geometries to create realistic representative volume elements (RVEs) with equivalent volume fractions of defects. By embedding defects of varying sizes and locations within the RVEs, fatigue indicator parameters (FIPs) were calculated to analyze the impact of defects’ characteristics on fatigue performance. Different combinations of defect volume and locations were generated for various microstructure instantiations, providing insight into extreme value fatigue responses. Larger defect volumes located on free surfaces consistently generated the highest FIPs, suggesting defect size and boundary proximity intensify stress concentration effects. RVEs with multiple smaller defects produced lower FIPs than those with single large critical defects. These findings underscore the critical role of defect characteristics on fatigue life, providing a foundation for future predictive modeling in fatigue-sensitive AM applications.
A multi deep learning-based framework is developed for efficient, automated microstructure reconstruction and generation of stochastic representative volume elements (SRVEs) with periodic boundary conditions (PBCs) for accurate modeling of ceramic matrix composite (CMC) response. The methodology comprises a convolutional neural network coupled with regression layers to act as a vanilla regression network for semantic segmentation of the microstructure, allowing accurate characterization of the phases and their distributions at the microscale. Scanning electron microscope and confocal microscope are used to obtain C/SiNC and SiC/SiNC CMCs micrographs for vanilla regression testing. Microstructure variability in terms of fiber volume fraction and porosity are quantified through the output regression layer, ensuring accurate representation of material variability in SRVE construction. Generative adversarial network (GAN) and its variants are designed to produce high-fidelity SRVE, spanning CMCs microstructure variability space. A circular padding algorithm is developed to generate SRVEs with PBCs during training of GANs. The accuracy of the generated SRVEs is established through micromechanics simulations, where an efficient formulation of the high-fidelity generalized methods of cells (HFGMC) approach is used to compute the effective mechanical properties. Furthermore, an iterative algorithm is implemented in the HFGMC solver to simulate time-dependent deformation of SiC/SiNC subjected to creep loading conditions.
The main goal of the proposed project is the development of validated nondestructive evaluation (NDE) techniques for in situ monitoring of ferritic-martensitic steels like Grade 91 9Cr-1Mo, which are candidate materials for Generation IV nuclear energy structural components operating at temperatures up to ≈650 °C and for steam-generator tubing for sodium-cooled fast reactors. Full assessment of thermomechanical damage requires a clear separation between thermally activated microstructural evolution and creep damage caused by simultaneous mechanical stress. Creep damage can be classified as "negligible" creep without significant plastic strain and "ordinary" creep of the primary, secondary, and tertiary kind that is accompanied by significant plastic deformation and/or cavity nucleation and growth. Under negligible creep conditions of interest in this project, minimal or no plastic strain occurs, and the accumulation of creep damage does not significantly reduce the fatigue life of a structural component so that low-temperature design rules, such as the ASME Section III, Subsection NB, can be applied with confidence. The proposed research project will utilize a multifaceted approach in which the feasibility of electrical conductivity and thermo-electric monitoring methods is researched and coupled with detailed post-thermal/creep exposure characterization of microstructural changes and damage processes using state-of-the-art electron microscopy techniques, with the aim of establishing the most effective nondestructive materials evaluation technique for particular degradation modes in high-temperature alloys that are candidates for use in the Next Generation Nuclear Plant (NGNP) as well as providing the necessary mechanism-based underpinnings for relating the two. Only techniques suitable for practical application in situ will be considered. As the project evolves and results accumulate, we will also study the use of this technique for monitoring other GEN IV materials. Through the results obtained from this integrated materials behavior and NDE study, new insight will be gained into the best nondestructive creep and microstructure monitoring methods for the particular mechanisms identified in these materials. The proposed project includes collaboration with a national laboratory partner and the results will also serve as a foundation to guide the efforts of scientists in the DOE laboratory, university, and industrial communities concerned with the technological challenges of monitoring creep and microstructural evolution in materials planned to be used in Generation IV Nuclear Energy Systems.
Powder-bed additive manufacturing (AM) processes are associated with the formation of multiple types of process-specific pores, including but not limited to lack-of-fusion (LoF) and keyhole pores. The performance of an AM component is dependent on the type of pores, their density and their proximity to the free surface, and other heterogeneities in the microstructure. In order to characterize the influence of porosity on the mechanical behavior of AM materials, it is imperative to quantitatively analyze the heterogeneous strain accumulation in the vicinity of porosity. Process-specific microstructure models are generated using SPPARKS, an open-source process simulation code. Spherical keyhole or irregular LoF pores are embedded into the microstructure models, which are meshed and input into a finite element code, ScIFEN, to solve for the heterogeneous strain localization in the vicinity of the pores. Given the non-smooth geometries of LoF pores, they readily promote strain accumulation in their vicinity thereby increasing the propensity of initiating fatigue cracks.
Conventional machining especially hard turning is commonly used in aerospace industry to produce the final shape of cylindrical parts. However, the process generates damages and microstructural alterations on the surface and at the subsurface layer, in particular the cracking of hard carbides particles. The present paper focuses on investigating the probable mechanisms responsible for carbide cracking and the formation of a softened layer at a depth between 10 and 30 µm below the machined surface. Advanced techniques, such as laser confocal microscopy and field emission gun scanning electron microscope equipped with in situ picoindenter and electron backscatter diffraction, were used to characterize and analyze the evolution of the microstructure in the affected layer and propose the possible governing mechanisms for the observations.
We investigate the generation of high aspect ratio microstructures across 0.7 mm thick glass by means of single shot Bessel beam laser direct writing. We study the effect on the photoinscription of the cone angle, as well as of the energy and duration of the ultrashort laser pulse. The aim of the study is to optimize the parameters for the writing of a regular microstructure due to index modification along the whole sample thickness. By using a spectrally resolved single pulse transmission diagnostics at the output surface of the glass, we correlate the single shot material modification with observations of the absorption in different portions of the retrieved spectra, and with the absence or presence of spectral modulation. Numerical simulations of the evolution of the Bessel pulse intensity and of the energy deposition inside the sample help us interpret the experimental results that suggest to use picosecond pulses for an efficient and more regular energy deposition. Picosecond pulses take advantage of nonlinear plasma absorption and avoid temporal dynamics effects which can compromise the stationarity of the Bessel beam propagation.
This paper presents a deep generative model-based design methodology for tailoring the structural stochasticity of microstructures. Although numerous methods have been established for designing deterministic (periodic) or stochastic microstructures, a systematic design approach that allows the unified treatment of both deterministic and stochastic microstructure design domains has yet to be created. The proposed methodology resolves this issue by learning a unified feature space that embodies diverse structural patterns with continuously varying stochasticity levels. A highly diverse microstructure database is established to incorporate various types of deterministic and stochastic microstructure patterns. A property-aware deep generative model is proposed to learn a unified feature space of the structural characteristics, as well as the relationship between structure features and properties of interest. Autoencoder (AE), Variational Autoencoder (VAE), and Adversarial Autoencoder (AAE) are compared to understand their relative merits in the property-aware learning of the unified feature space. Microstructural designs with tailorable stochasticity and properties are obtained by searching the unified feature space. Multiple design cases are presented to demonstrate the capability of designing microstructures for structural stochasticity and properties. Furthermore, the proposed method is employed to create stochastically graded structures, which manipulate the mechanical behaviors by varying the local stochasticity of the structure.
We present that the transition fatigue regime between low cycle fatigue (LCF) and high cycle fatigue (HCF) is often addressed in the design and performance evaluation of load-bearing components used in many structural applications. Transition fatigue is characterized by elevated levels of local inelastic deformation in significant regions of the microstructure as compared to HCF. Typically, crystal plasticity finite element method (CPFEM) simulations are performed to model this phenomenon and to rank-order microstructures by their resistance to crack formation and early growth in the regime of transition fatigue. Unfortunately, these approaches require significant computational resources, inhibiting their use to explore novel materials for transition fatigue resistance. Reduced-order, microstructure-sensitive models are needed to accelerate the search for next-generation, fatigue-resistant materials. In a recent study, Paulson et al. (2018) extended the materials knowledge system (MKS) framework for rank-ordering the HCF resistance of polycrystalline microstructures. The efficacy of this approach lies in the reduced-dimensional representation of microstructures through 2-point spatial correlations and principal component analysis (PCA), in addition to the characterization of the HCF response with a small set of performance measures. In this work, these same protocols are critically evaluated for their applicability to rank-order the transition fatigue resistance of the same class of polycrystalline microstructures subjected to increased strain amplitudes. Success in this endeavor requires the formation of homogenization linkages that account for the significantly higher levels of local inelastic deformation and stress redistribution in transition fatigue as compared to HCF. Finally, a set of 12 α-titanium microstructures generated using the open access DREAM.3D software (Groeber and Jackson, 2014) are employed for this evaluation.
In this paper, we examine the key factors influencing ductile failure of various grades of dual phase (DP) steels using the microstructure-based modeling approach. Various microstructure-based finite element models are generated based on the actual microstructures of DP steels with different martensite volume fractions. These models are, then, used to investigate the influence of ductility of the constituent ferrite phase and also the influence of voids introduced in the ferrite phase on the overall ductility of DP steels. It is found that with volume fraction of martensite in the microstructure less than 15%, the overall ductility of the DP steels strongly depends on the ductility of the ferrite matrix, hence pre-existing micro voids in the microstructure significantly reduce the overall ductility of the steel. When the volume fraction of martensite is above 15%, the preexisting voids in the ferrite matrix does not significantly reduce the overall ductility of the DP steels, and the overall ductility is more influenced by the mechanical property disparity between the two phases. The applicability of the phase inhomogeneity driven ductile failure of DP steels is then discussed based on the obtained computational results for various grades of DP steels, and the experimentally obtained scanning electron microscopy (SEM) pictures of the corresponding grades of DP steels near fracture surface are used as evidence for result validations.
Microstructure reconstruction problems are usually limited to the representation with finitely many number of phases, e.g. binary and ternary. However, images of microstructure obtained through experimental, for example, using microscope, are often represented as a RGB or grayscale image. Because the phase-based representation is discrete, more rigid, and provides less flexibility in modeling the microstructure, as compared to RGB or grayscale image, there is a loss of information in the conversion. In this paper, a microstructure reconstruction method, which produces images at the fidelity of experimental microscopy, i.e. RGB or grayscale image, is proposed without introducing any physics-based microstructure descriptor. Furthermore, the image texture is preserved and the microstructure image is represented with continuous variables (as in RGB or grayscale images), instead of binary or categorical variables, which results in a high-fidelity image of microstructure reconstruction. The advantage of the proposed method is its quality of reconstruction, which can be applied to any other binary or multiphase 2D microstructure. The proposed method can be thought of as a subsampling approach to expand the microstructure dataset, while preserving its image texture. Moreover, the size of the reconstructed image is more flexible, compared to other machine learning microstructure reconstruction method, where the size must be fixed beforehand. In addition, the proposed method is capable of joining the microstructure images taken at different locations to reconstruct a larger microstructure image. A significant advantage of the proposed method is to remedy the data scarcity problem in materials science, where experimental data is scare and hard to obtain. The proposed method can also be applied to generate statistically equivalent microstructures, which has a strong implication in microstructure-related uncertainty quantification applications. The proposed microstructure reconstruction method is demonstrated with the UltraHigh Carbon Steel micrograph DataBase (UHCSDB).