Ultrahigh Carbon Steel Research Articles

Automatic Microstructural Classification of Ultrahigh Carbon Steel with Vision Transformers and Convolutional Neural Networks

Microstructural classification is key to better understanding of the relationships between the microstructures of carbon steels and their properties, as well as the development of automatic models to reliably and efficiently categorize the microstructures which can largely mitigate the burden of the human experts. Methods based on transfer learning with deep neural networks, such as convolutional neural networks (CNNs) have become established in automatic microstructural classification of carbon steels. However, vision transformers (ViTs), as a new class of deep learning algorithms, have recently emerged as a competitive alternative to CNNs in many other fields, but have not been considered in this field yet. In this study, the micrographs from an open-source dataset of ultrahigh carbon steel are analyzed by use of two CNNs (GoogLeNet and MobileNetV2) and two ViTs (ViT-B16 and ViT-L32). These deep neural networks are used as end-to-end classifiers to discriminate between different microstructures. Promising results were achieved by use of ViTs, as they performed as well as, or better than CNNs, though not by a large margin. These results suggest that ViTs can be a viable alternative to CNNs in the advancement of automatic microstructural classification.

Machine Learning Approaches for Classification of Ultra High Carbon Steel Micrographs

The aim of this investigation is to analyze the performance of several supervised machine learning algorithms for solving the automatic classification problem of steel image microstructures. We conducted an experiment using a public-domain dataset of Ultra High Carbon Steel Micrographs (UHCSM). This image database consists of a collection of scanning electron micrographs (SEM) taken from samples of a commercial roll-mill casting with a nominal carbon of 2%. Heat treatments such as annealing, water quenching, air and furnace cooling were performed on steel samples so primary microconstituents could be found in micrographs. Each of these microconstituents defines each of the categories of classification to be accomplished by machine learning algorithms. The heat treatments brought about 4 usable classes (sets of images) of primary microconstituents: pearlite, spheroidite, proeutectoid cementite network, pearlite containing spheroidite. All labeled images are prepared to improve models' accuracy in a preprocessing stage so that the image dataset is ready for feature extraction. In order to develop classification models, we put to the test distinct machine learning approaches by working with Matlab's classification learner application where we perform automated training to search for the best classification model type, including Decision Trees, Support Vector Machines (SVM), Discriminant Analysis, Nearest Neighbors, Naive Bayes, Ensemble k-NN, and Neural Network classification. For obtaining the features of the images (feature extraction) we choose the method of Bag-of-features with 400 words for the first experiment, and 327 words by removing less important features for a second experiment. The experimented models reached very different accuracy values on training, with SVM as the best classifier which gets 91.6% accuracy. We can conclude that classic machine learning algorithms solve the classification, but an accuracy improvement can be reached by investigating deep learning techniques.

Data augmentation in material images using the improved HP-VAE-GAN

Since the rapid development of computer vision relies heavily on large-scale labeled data and high-performance computing equipment, therefore, image recognition in small sample datasets faces several challenges, such as difficult to implement model training. In the field of materials research, the cost of collecting image data is relatively high. In order to solve the problem of insufficient image samples in material research, an improved HP-VAE-GAN is proposed to generate material images to achieve data augmentation. HP-VAE-GAN is a single sample generation model that consists of Patch-VAE and Patch-GAN. The improved HP-VAE-GAN introduces the attention mechanism into model. By adding CBAM (Convolutional Block Attention Module) to the encoder of Patch-VAE, the feature extraction and representation capabilities of the network are further improved. Use this model to train a single image, and then generate a certain number of samples to achieve the expansion of the training set. For the classification of ultrahigh carbon steel microstructure images, experiments show that the accuracy of classification model (MobileNet, ResNet50 and VGG16) trained with real images plus generated images is improved obviously. In addition, the effectiveness of the improved HP-VAE-GAN is verified by experiments on texture images similar to material images.

Deep Learning Approaches to Image Texture Analysis in Material Processing

Texture analysis is key to better understanding of the relationships between the microstructures of the materials and their properties, as well as the use of models in process systems using raw signals or images as input. Recently, new methods based on transfer learning with deep neural networks have become established as highly competitive approaches to classical texture analysis. In this study, three traditional approaches, based on the use of grey level co-occurrence matrices, local binary patterns and textons are compared with five transfer learning approaches, based on the use of AlexNet, VGG19, ResNet50, GoogLeNet and MobileNetV2. This is done based on two simulated and one real-world case study. In the simulated case studies, material microstructures were simulated with Voronoi graphic representations and in the real-world case study, the appearance of ultrahigh carbon steel is cast as a textural pattern recognition pattern. The ability of random forest models, as well as the convolutional neural networks themselves, to discriminate between different textures with the image features as input was used as the basis for comparison. The texton algorithm performed better than the LBP and GLCM algorithms and similar to the deep learning approaches when these were used directly, without any retraining. Partial or full retraining of the convolutional neural networks yielded considerably better results, with GoogLeNet and MobileNetV2 yielding the best results.

Deformation behavior during hot torsion of an ultrahigh carbon steel containing 1.3 wt.% C

Abstract The torsion behavior of an ultrahigh carbon steel containing 1.3 wt.% C has been studied at high strain rates (2 – 26 s –1) and high temperatures (900 – 1200 °C). Adiabatic heating strongly affects the shape of the stress– strain curves. Grain size measurements at strains of ε = 2 and ε = 5 reveal that the austenite grain size does not change between these two strains and is finer than the initial austenite grain size. This is an indication that a recrystallization process took place in the interval of strains between ε peak and ε = 2 and that the deformation state associated to the strains of ε = 2 and ε = 5 can be considered as steady state. Stress–strain rate relations have been analysed for ε peak, ε = 2 and ε = 5. It has been found that data obtained at ε peak correlate well with a lattice diffusion-controlled dislocation creep process. On the other hand, analysis of the stress –strain rate relations found at the steady state indicated that the deformation occurs by the contribution of two independent mechanisms: grain boundary sliding controlled by pipe diffusion and slip creep controlled by lattice diffusion.

High deformation effects on the plasticity of ultra-high carbon steel wires

Steel wires under severe cold drawing deformation, develop high strength (5-6 GPa) and ductility. For these reasons it is relevant to increase the knowledge on the structural evolution and deformation mechanisms involved during wire drawing process, due to their critical applications such as bridges, cranes and tire cord. This paper presents a comparative study of steel wires (0.84%C) at different deformation stages. The product presents a normal behaviour under torsion test, the mentioned test is normally used to corroborate the wire aptitude. The main objective of the study is to increase the knowledge on the structural evolution after cold drawn considering the deformation mechanisms, cementite dissolution, and epsilon carbide precipitation. The microstructural study was carried out applying light and scanning electron microscopy (SEM-EBSD). The structural information was correlated with results of differential thermal analysis (DTA) and FactSage simulation. The structural study verified the presence of curling phenomenon in the wires. The interlaminar spacing (l) and the thickness of cementite lamellae in wires cold drawn from 8 mm up to 2 mm of diameter was determined. Finally, the dynamic strain aging, which is promoted by cementite destabilization and the precipitation of epsilon carbide was studied. Copyright © 2018 VBRI Press.

Heat treatment optimization of an automotive stamping die steel with low energy consumption

It has innovatively teat-treated an ultra-high carbon steel via a novel technique with low energy consumption for automotive stamping die application in this study. Formation mechanism of isothermal transformation product at low temperatures in an ultra-high carbon steel has been investigated by the means of X-ray diffraction (XRD) analysis and metallographic examination on the resulting microstructure, which is optimally designed to achieve an excellent combination of high impact toughness and high hardness. It is indicated that microstructure morphology of resulting bainite isothermally formed at temperatures below 200°C is observably different from water-quenched shear-transformed martensite; in the case of isothermal transformation above 200°C for a long time, it obtains bainite sheaf which is composed of slender bainitic ferrite plates separated by parallel parent phase films; bainite sheaf is achieved by nucleation and growth of new parallel bainite plates in the vicinity of both body sides of the already formed bainite plate.

Data-driven phase recognition of steels for use in mechanical property prediction

This paper develops a deep learning scheme of phase recognition for steel materials. A convolutional neural network classifier is established, such that the martensite phase, which has a substantial impact on the mechanical properties of steels, can be recognized from microstructure images and its volumetric fraction can also be estimated from multi-phase microconstituents. The testing results on an ultrahigh carbon steel dataset proved that the developed scheme has a rational phase recognition accuracy. The estimated martensite fraction can be used as an essential feature to predict the mechanical properties of materials in additive manufacturing.

Instance Segmentation for Direct Measurements of Satellites in Metal Powders and Automated Microstructural Characterization from Image Data

We propose instance segmentation as a useful tool for image analysis in materials science. Instance segmentation is an advanced technique in computer vision which generates individual segmentation masks for every object of interest that is recognized in an image. Using an out-of-the-box implementation of Mask R-CNN, instance segmentation is applied to images of metal powder particles produced through gas atomization. Leveraging transfer learning allows for the analysis to be conducted with a very small training set of labeled images. As well as providing another method for measuring the particle size distribution, we demonstrate the first direct measurements of the satellite content in powder samples. After analyzing the results for the labeled data dataset, the trained model was used to generate measurements for a much larger set of unlabeled images. The resulting particle size measurements showed reasonable agreement with laser scattering measurements. The satellite measurements were self-consistent and showed good agreement with the expected trends for different samples. Finally, we present a small case study showing how instance segmentation can be used to measure spheroidite content in the UltraHigh Carbon Steel DataBase, demonstrating the flexibility of the technique.

Accelerated Spheroidization of Cementite in Sintered Ultrahigh Carbon Steel by Warm Deformation

Evolution of microstructure and hardness in quenched ultrahigh carbon steel Fe-0.85Mo-0.6Si-1.4C by warm compression on a Bähr plastometer-dilatometer at 775 °C and at 0.001 to 1 s−1 strain rate range is reported. The material was prepared via powder metallurgy: cold pressing and liquid phase sintering. Independent of strain rate, the initial martenstic microstructure was transformed to ferrite and spheroidized cementite. Strain rate had an effect on size and shape of spheroidized Fe3C precipitates: the higher the strain rate, the smaller the precipitates. Morphology of the spheroidized carbides influenced hardness, with the highest hardness, 362 HV10, for strain rate 1 s−1 and the lowest, 295 HV10, for the lowest strain rate 0.001 s−1. Resultant microstructure and ambient temperature mechanical properties were comparable to those of the material that had undergone a fully spheroidizing treatment with increased time and energy consumption, indicating that it can be dispensed with in industrial processing. All our results are consistent with the Hall–Petch relation developed for spheroidized steels.

Stress-strain behavior and microstructural evolution of ultra-high carbon Fe-C-Cr-V-Mo steel subjected to hot deformation

Numerous processes have been applied to refine the microstructure and the precipitation of carbides in ultra-high carbon steel (UHCS). However, the evolution of microstructure and carbide precipitation in UHCSs during the hot deformation compression process remains poorly understood. The present study addresses this issue by applying hot deformation compression to Fe-C-Cr-V-Mo UHCS specimens at temperatures of 950–1150 °C under strain rates of 0.01–5 s−1, and evaluating the resulting flow curves and microstructures of the specimens after cooling in air. High deformation temperature and high strain rate conditions are observed to generate flow curve stress peaks, which result in dynamic recrystallization. In addition, dislocation cell formation and substructure are observed at deformation temperatures of 950 °C and 1050 °C, respectively, under a strain rate of 1 s−1. The recrystallized grain structure fraction is found to be positively correlated with the deformation temperature at strain rates less than 5 s−1, and a fully recrystallized grain structure is not obtained at a strain rate of 5 s−1 even at a temperature of 1150 °C. Moreover, the concentration of primary M7C3 carbides is observed to decrease with increasing deformation temperature, and fine precipitated M23C6 particles are observed in the matrix during hot deformation. An increased concentration of precipitated M23C6 particles is also observed in the specimen microstructure at a deformation temperature of 1150 °C and a strain rate of 5 s−1. The recrystallization process is also observed to vary according to M23C6 particle size, where large M23C6 particles are observed to stimulate the formation of recrystallization nuclei and fine M23C6 particles delay or hinder the recrystallization process.

Novel Method for Refining Coarse Eutectic Carbides in Ultrahigh-Carbon Steel

The structure and mechanical properties of an ultrahigh-carbon steel modified with amorphous ferrosilicon with a low melting point are studied. The methods of x-ray diffraction, optical and scanning electron microscopy are used to study the effect of the amorphous modifier on the structure of eutectic ledeburite in the cast steel. Tests for tensile strength and impact toughness are conducted. It is shown that the mechanical properties increase as a result of the modification without subsequent heat treatment.

Effects of Cr Concentration on Cementite Coarsening in Ultrahigh Carbon Steel

Ultrahigh carbon steels (UHCS) containing 1 wt pct Cr (1Cr UHCS) were heat treated at 1073 K, 1173 K, or 1243 K (800 °C, 900 °C, or 970 °C) for durations of 5 minutes up to 24 hours to study cementite coarsening. Results were compared to a previous study on coarsening of a UHCS with 4 wt pct Cr (4Cr UHCS) and significantly different behavior was observed. In the heat-treated 1Cr UHCS, particles clustered in zones that were tens of microns near branches of the cementite network. The opposite trend was observed in the 4Cr UHCS, in which the regions within a few microns near the cementite network were entirely denuded of particles. Causes for differences in coarsening behavior in 1Cr and 4Cr UHCS are discussed.

Data-driven high-fidelity 2D microstructure reconstruction via non-local patch-based image inpainting

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).

Computer vision approach for phase identification from steel microstructure

PurposeThe purpose of this paper is to develop an automated phase segmentation model from complex microstructure. The mechanical and physical properties of metals and alloys are influenced by their microstructure, and therefore the investigation of microstructure is essential. Coexistence of random or sometimes patterned distribution of different microstructural features such as phase, grains and defects makes microstructure highly complex, and accordingly identification or recognition of individual phase, grains and defects within a microstructure is difficult.Design/methodology/approachIn this perspective, computer vision and image processing techniques are effective to help in understanding and proper interpretation of microscopic image. Microstructure-based image processing mainly focuses on image segmentation, boundary detection and grain size approximation. In this paper, a new approach is presented for automated phase segmentation from 2D microstructure images. The benefit of the proposed work is to identify dominated phase from complex microstructure images. The proposed model is trained and tested with 373 different ultra-high carbon steel (UHCS) microscopic images.FindingsIn this paper, Sobel and Watershed transformation algorithms are used for identification of dominating phases, and deep learning model has been used for identification of phase class from microstructural images.Originality/valueFor the first time, the authors have implemented edge detection followed by watershed segmentation and deep learning (convolutional neural network) to identify phases of UHCS microstructure.

Fabrication of piston pins made of a novel aluminium-alloyed UHC steel

The vehicle industry strives for weight reduction in order to reduce fuel consumption. This leads to the development of new steel grades that are enriched with light metals, such as aluminium. In addition to the reduction in density, this means that the properties of conventional steels need to be at least replaced or ideally even surpassed. In the present work, a novel material, ultra-high-carbon (UHC) lightweight steel, was investigated for its forming behaviour. In addition to extrusion tests for this material, comprehensive material characterisation was carried out. The temperature, strain and strain rate–dependent flow behaviour were determined. Furthermore, ring compression tests were performed to identify the temperature and tool-workpiece-dependent friction factors. The findings served as a basis for a material model to perform numerical simulations in order to develop the tool for the extrusion of piston pins. The extrusion took place at 1050–1150 °C in a screw press. This work provides a basis for the forming of hard and solid steels that are difficult to process and can be applied to metals of similar properties.

High Throughput Quantitative Metallography for Complex Microstructures Using Deep Learning: A Case Study in Ultrahigh Carbon Steel.

We apply a deep convolutional neural network segmentation model to enable novel automated microstructure segmentation applications for complex microstructures typically evaluated manually and subjectively. We explore two microstructure segmentation tasks in an openly available ultrahigh carbon steel microstructure dataset: segmenting cementite particles in the spheroidized matrix, and segmenting larger fields of view featuring grain boundary carbide, spheroidized particle matrix, particle-free grain boundary denuded zone, and Widmanstätten cementite. We also demonstrate how to combine these data-driven microstructure segmentation models to obtain empirical cementite particle size and denuded zone width distributions from more complex micrographs containing multiple microconstituents. The full annotated dataset is available on materialsdata.nist.gov.

Formation of Graphite Phase in Powder Metallurgical Ultrahigh Carbon Steels with High Wear Resistance

Carbon with different forms and morphologies plays a key role in the wear resistance of ultrahigh carbon steels (UHCS). In this study, UHCS with carbon contents from 1.4 to 1.9 wt% are prepared through powder metallurgy route, and an organic compound is used as the carbon source. The microstructures, mechanical properties, and wear resistance of UHCS are investigated. Results show that the graphite phase is formed in UHCS, and the morphology of graphite changes from granular to irregular strip‐type with the increase of carbon content. The synthesized UHCS show improved wear resistance compare to other reported UHCS with no graphite. Moreover, the UHCS with 1.7 wt% C is found to have the best wear resistance with a wear rate of 4.14 × 10−6 mm3/N · m. Besides, it is found that the wear rate and the coefficient of friction (COF) of UHCS increase after heat treatments. This is mainly due to the diffusion of carbon into the matrix, which leads to the reduction of graphite. The results are much helpful for preparing UHCS with controllable graphite morphology, and better wear resistance.

Effects of Nb Modification and Cooling Rate on the Microstructure in an Ultrahigh Carbon Steel

In this study, two different melting methods were used to investigate effects of Nb modification on microstructure in ultrahigh carbon steel (UHCS). Nb-free and Nb-modified UHCS samples were produced by melting and resolidifying an industrially produced base UHCS with and without addition of Nb powder. Microstructure was characterized using scanning electron microscopy, X-ray diffraction, and electron dispersive spectroscopy. Equilibrium computations of phase fractions and compositions were utilized to help describe microstructural changes caused by the Nb additions. Nb combined with C to form NbC structures before and during austenite solidification, reducing the effective amount of carbon available for the other phases. Cementite network spacing in the Nb-free samples was controlled by the cooling rate during solidification (faster cooling led to a more refined network). Network spacing in the Nb-modified UHCS could be enlarged by NbC structures that formed cooperatively with austenite.

In situ synthesis and strengthening of ultra high-carbon martensitic stainless steels in addition of LaB6

A powder metallurgy route which is characterized by the vacuum solid state sintering is utilized to fabricate Fe-20Cr-3C and Fe-20Cr-3C-2Mo-3V-2Co ultra high-carbon martensitic stainless steels. A sintering window from 1190 °C to 1210 °C is chosen to obtain high density samples and avoid liquid phase simultaneously. The densification process is promoted by solubility effect during formation of M7C3 and grain boundary diffusion. TEM analysis demonstrates three types of reaction products: the La(BO2)3 phase, the La2O3 phase and the La(Fe0.5V0.5)O3 phase, after adding 0.1 wt% LaB6. The impurity elements such as S and O are absorbed following the LaB6 addition to form non-gaseous products, which promotes densification by decreasing the sintering resistance and thus contributes to the improvement of bending strength and hardness. Both of the electrochemical and immersion results lead to the fact that the corrosion resistance of the steels are improved by 0.1 wt% LaB6 addition, in 3.5 wt% NaCl solution. However, the corrosion resistance of Fe-20Cr-3C-0.1LaB6 still can not compete with that of 316L stainless steel. Nevertheless, Fe-20Cr-3C-2Mo-3V-2Co and Fe-20Cr-3C-2Mo-3V-2Co-0.1LaB6 after quenching at high temperature present comparable corrosion resistance with that of 316L in 3.5 wt% NaCl solution.