Data-mining of in-situ TEM experiments: On the dynamics of dislocations in CoCrFeMnNi alloys

Zhang, Chen; Song, Hengxu; Oliveros, Daniela; Frączkiewicz, A.; Legros, Marc; Sandfeld, Stefan

doi:10.1016/j.actamat.2022.118394

Cited by 17 publications

(10 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The goal of the next steps is to segment dislocations such that each of them can be represented as a mathematical spline. This enables us to perform quantitative calculations, such as computing the velocity, position, and curvature, as demonstrated in [10]. Training a ML model for segmentation tasks can be challenging, in particular when dislocations are nearly touching each other (see, e.g.…”

Section: Learning Strategymentioning

confidence: 99%

“…These can be represented by, e.g. polynomial approximations such that even local geometrical properties can be represented as demonstrated in [9,10]. Up to date, digitization of dislocations is almost always done manually.…”

Section: Introductionmentioning

confidence: 99%

“…Dislocations can be annotated as a line by selecting points on the dislocations using these tools. In [10], such a manual method was used to extract the information of dislocations from 300 individual frames where each frame has up to 20 dislocations. This is a challenging and tedious task, difficult to reproduce, and the result may depend on the experience of the person who performs the labeling of the images.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Deep learning of crystalline defects from TEM images: a solution for the problem of ‘never enough training data’

Govind,

Oliveros,

Dlouhy

et al. 2024

Mach. Learn.: Sci. Technol.

Self Cite

View full text Add to dashboard Cite

Crystalline defects, such as line-like dislocations, play an important role for the performance and reliability of many metallic devices. Their interaction and evolution still poses a multitude of open questions to materials science and materials physics. In-situ TEM experiments can provide important insights into how dislocations behave and move. During such experiments, the dislocation microstructure is captured in form of videos. The analysis of individual video frames can provide useful insights but is limited by the capabilities of automated identification, digitization, and quantitative extraction of the dislocations as curved objects. The vast amount of data also makes manual annotation very time consuming, thereby limiting the use of Deep Learning-based, automated image analysis and segmentation of the dislocation microstructure.In this work, a parametric model for generating synthetic training data for segmentation of dislocations is developed. Even though domain scientists might dismiss synthetic training images sometimes as too artificial, our findings show that they can result in superior performance, particularly regarding the generalizing of the Deep Learning models with respect to different microstructures and imaging conditions. Additionally, we propose an enhanced deep learning method optimized for segmenting overlapping or intersecting dislocation lines. Upon testing this framework on four distinct real datasets, we find that our synthetic training data are able to yield high-quality results also on real images–even more so if fine-tune on a few real images was done. Our approach demonstrates the potential of synthetic data in overcoming the limitations of manual annotation in TEM, paving the way for more efficient and accurate analysis of dislocation microstructures. Last but not least, segmenting such thin, curvilinear structures is a task that is ubiquitous in many fields, which makes our method a potential candidate for other applications as well.

show abstract

Section: Learning Strategymentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Deep learning of crystalline defects from TEM images: a solution for the problem of ‘never enough training data’

Govind,

Oliveros,

Dlouhy

et al. 2024

Mach. Learn.: Sci. Technol.

Self Cite

View full text Add to dashboard Cite

show abstract

“…In-situ TEM straining experiments were performed on a single-phase face-centered, cubic CoCrFeMnNi alloy (also known as "Cantor alloy"). Samples were prepared as electron-transparent, stretchable thin foils following the method described in [11], [12]. Each straining experiment was performed at 96 K using a Gatan 671 straining holder arXiv:2309.03499v1 [cs.CV] 7 Sep 2023 in a JEOL 2010, which was operated at 200 kV.…”

Section: B Data Acquisitionmentioning

confidence: 99%

Instance Segmentation of Dislocations in TEM Images

Ruzaeva,

Govind,

Legros

et al. 2023

2023 IEEE 23rd International Conference on Nanotechnology (NANO)

Self Cite

View full text Add to dashboard Cite

Quantitative Transmission Electron Microscopy (TEM) during in-situ straining experiment is able to reveal the motion of dislocations -linear defects in the crystal lattice of metals. In the domain of materials science, the knowledge about the location and movement of dislocations is important for creating novel materials with superior properties. A longstanding problem, however, is to identify the position and extract the shape of dislocations, which would ultimately help to create a digital twin of such materials.In this work, we quantitatively compare state-of-the-art instance segmentation methods, including Mask R-CNN and YOLOv8. The dislocation masks as the results of the instance segmentation are converted to mathematical lines, enabling quantitative analysis of dislocation length and geometryimportant information for the domain scientist, which we then propose to include as a novel length-aware quality metric for estimating the network performance. Our segmentation pipeline shows a high accuracy suitable for all domain-specific, further post-processing. Additionally, our physics-based metric turns out to perform much more consistently than typically used pixel-wise metrics.

show abstract

“…On a side note, this also raises the question of what a statistically equivalent microstructure is and is connected to our second question. Although there are several novel techniques and approaches for observing 3D structure of dislocations via experiments [15][16][17][18][19], the precision of the state-of-art techniques is limited, therefore quantitative microstructure information can be accessed is insufficient to describe a complete dislocation network of a microstructure. Furthermore, initializing the microstructures is also a concern for CDD simulations where the initial variables other than the dislocation density are less amenable to 'guesswork'.…”

Section: Introductionmentioning

confidence: 99%

Statistical analysis of discrete dislocation dynamics simulations: initial structures, cross-slip and microstructure evolution

Demirci

Steinberger

Stricker

et al. 2023

Modelling Simul. Mater. Sci. Eng.

Self Cite

View full text Add to dashboard Cite

Over the past decades, discrete dislocation dynamics simulations have been shown to reliably predict the evolution of dislocation microstructures for micrometer-sized metallic samples. Such simulations provide insight into the governing deformation mechanisms and the interplay between different physical phenomena such as dislocation reactions or cross-slip. This work is focused on a detailed analysis of the influence of the cross-slip on the evolution of dislocation systems. A tailored data mining strategy using the "discrete-to-continuous (D2C) framework'' allows to quantify differences and to quantitatively compare dislocation structures. We analyze the quantitative effects of the cross-slip on the microstructure in the course of a tensile test and a subsequent relaxation to present the role of cross-slip in the microstructure evolution. The precision of the extracted quantitative information using D2C strongly depends on the resolution of the domain averaging. We also analyze how the resolution of the averaging influences the distribution of total dislocation density and curvature fields of the specimen. Our analyzes are important approaches for interpreting the resulting structures calculated by dislocation dynamics simulations.

show abstract

Data-mining of in-situ TEM experiments: On the dynamics of dislocations in CoCrFeMnNi alloys

Cited by 17 publications

References 28 publications

Deep learning of crystalline defects from TEM images: a solution for the problem of ‘never enough training data’

Deep learning of crystalline defects from TEM images: a solution for the problem of ‘never enough training data’

Instance Segmentation of Dislocations in TEM Images

Statistical analysis of discrete dislocation dynamics simulations: initial structures, cross-slip and microstructure evolution

Contact Info

Product

Resources

About