Physics-aware deep learning models for predicting the heterogeneous mechanical properties of polymeric nanostructured materials.

A methodology for determining the local mechanical properties of model atomistic glassy polymeric nanostructured materials

We study the influence of spherical, triangular, and rod-like nanoparticles on the mechanical properties of a polymer nanocomposite (PNC), via coarse-grained molecular dynamics simulations. We focus on how the nanoparticle size, loading, mass, and shape influence the PNC's elastic modulus, stress at failure and resistance against cavity formation and growth, under external stress. We find that in the regime of strong polymer-nanoparticle interactions, the formation of a polymer network via temporary polymer-nanoparticle crosslinks has a predominant role on the PNC reinforcement. Spherical nanoparticles, whose size is comparable to that of the polymer monomers, are more effective at toughening the PNC than larger spherical particles. When comparing particles of spherical, triangular, and rod-like geometries, the rod-like nanoparticles emerge as the best PNC toughening agents.

Polymer nanocomposites have improved mechanical, optical and thermal properties compared to traditional thermoplastics. To study the impact of grafted brush–matrix interactions on the mechanical properties of nanoparticle filled polymers, uniaxial tensile testing with digital image correlation (DIC) was done on polystyrene (PS) grafted SiO2 in low molecular weight (MW) (N ∼ P), N = degree of polymerization of grafted brush, P = degree of polymerization of matrix, and high MW (2N = P) matrices at different loadings. The low matrix MW composites had higher strength than high MW composites at high loading but lower strength than the pure matrix. The high matrix MW composites had higher toughness at low loadings. SEM images of fracture surfaces revealed particle debonding and plastic void growth in the high toughness samples. The results show that the P/N ratio of polymer grafted nanoparticle composites is important for controlling the mechanical properties of polymer nanocomposites.

Uncertainty quantification for multiscale modeling of polymer nanocomposites with correlated parameters

Nanocomposites, where nanoscale inclusions are embedded in a matrix material, have attracted increasing research attention in recent years. Here, we limit our focus to polymer nanocomposites. Nano-reinforcements (for instance, nanoparticles, nanowires, and nanotubes) have largely disappointed us in manufacturing lightweight, high-strength, and high-toughness polymer composites because of their agglomeration in the matrix—one of the major drawbacks limiting their reinforcing effect. Dispersion of nano-reinforcements is not trivial and often requires expensive facilities and/or processing procedures. It has proven difficult to disperse nano-reinforcements in the matrix via conventional chemical, mechanical, and physical methods. To date, a perfect dispersion of nanoreinforcements in a matrix material has not been achieved. Despite considerable efforts, the mechanical properties of polymer nanocomposites are still far below their theoretically predicted potential. On the other hand, nano-reinforcements and their composites produced through the currently available methods are usually too expensive for use in large-scale industrial applications. To solve these two critical problems, we turned our attention to investigate new concepts for nano-reinforcement dispersion and new synthesis techniques that enable low-cost, high-throughput manufacturing of nanocomposites. Boron carbide (B4C), a lightweight refractory semiconductor, is the third hardest material known to man at room temperature and becomes the hardest above 1100 8C. It has many unique properties such as low density (2.5 g cm ), a small thermal extension coefficient (5.73 10 6 K ), a high melting point (>2400 8C), high resistance to chemical attacks, high thermal stability, a high Seebeck coefficient, and a large neutron absorption cross-section. The combination of these superior properties gives rise to numerous applications, especially under extreme conditions, for example, lightweight body armor, aircraft armor, abrasive wear-resistant materials, solid-state neutron detectors, and, potentially, power generation in deep-space flight applications. Recently, 1D B4C nanostructures, including nanowires and nanorods, have attracted significant attention. To synthesize B4C nanowires, the carbon source is essential. Among most of the available synthesis methods for 1D B4C nanostructures, artificial materials such as carbon black, carbon nanotubes, and activate carbon are selected as carbon source. A stepwise synthetic procedure was usually adopted: metal-catalyst nanoparticles were first synthesized and subsequently added to the precursors as the catalyst. Cotton has been used by mankind for at least 7000 years. It is the oldest and most commonly used nanoporous material, which is constructed from polysaccharide chains arranged into amorphous and crystalline regions. Today, the world production of cotton is approximately 20 million tons per annum, mainly for clothing, paper, and medical uses. In this work, a commercial cotton T-shirt was used as both the template and the carbon source to synthesize large quantities of radially aligned B4C nanowires on carbon microfibers. The as-synthesized B4C-nanowire/carbon-microfiber hybrid structures were then simply dipped in an epoxy resin bath to achieve a high dispersion of B4C nanowires in the matrix and highthroughput manufacturing of laminated polymer nanocomposites. A digital camera image of the 100% cotton T-shirt is shown in Figure 1a. A striking and extremely useful feature of cotton is its ability of absorbing large quantities of liquids, particularly water. Figure 1b is a representative digital camera image of a piece of the cotton T-shirt after the absorption of a solution with Ni(NO3)2 6H2O and amorphous boron powders. Figure 1c shows 6 pieces of final textiles. A typical scanning electron microscopy (SEM) image of carbon microfibers covered with B4C nanowires is shown in Figure 1d. The B4C nanowires, with a diameter ranging from 80 to 200 nm and a length greater than 4mm, grew radially on the entire length of the carbon microfiber (Fig. 1e). The SEM image (Fig. 1f) and transmission electron microscopy (TEM) image (Fig. 1g) together reveal a catalyst particle on the tip of each nanowire. Most of the catalyst particles were spherical and had a diameter distribution of 90 to 250 nm. TEM image (Fig. 1g) also shows that the diameter of the catalyst is larger than that of the corresponding B4C nanowire. Energy dispersive spectroscopy (EDS) reveals that the catalyst particle on the nanowire tip is nickel boride. This suggests that a catalyst is essential for growing such B4C nanowires. Our experimental results proved that most of the catalyst particles can be easily removed by acid treatment (HNO3þHF or HNO3). The B4C

The effect of fullerene and carbon fillers on the mechanical properties of polymer nanocomposites based on thermoreactive (epoxy resin) and thermoplastic (polyamide-12) matrices was investigated. It was found that the introduction of these fillers did not affect the properties of the thermoreactive blends, but Young’s modulus and the tensile strength of the thermoplastic ones increased by about 30-40% upon addition of 0.02-0.08 wt.% fullerene materials. The best results were obtained for a mixture of C 60/C 70.

Multi-factor analysis of the effects of graphene oxide nanoplatelets on self-healing polymer composites based on micromechanical FE simulation

This research work studies the effects of processing conditions on mechanical properties of polymer nanocomposites. Polypropylene (PP) nanocomposites reinforced with 0.5 and 2.5 %wt multiwall carbon nanotubes (MWCNTs) were prepared via melt compounding and formed by injection molding. 2k Full Factorial design was used to plan the experiments and determine the influences of the processing conditions on mechanical properties and carbon nanotube dispersion in the nanocomposites. These conditions consist of five factors: (a) %wt content of MWCNTs (0.5 and 2.5 %wt), (b) barrel temperature (190 and 220°C), (c) injection velocity (25 and 45 mm/sec), (d) screw rotational speed (75 and 227 rpm) and (e) holding pressure (45 and 65 bar) while injection pressure and cooling time were set at 75 bar and 50 sec, respectively, for all conditions. The samples were examined by Young’s modulus and tensile strength using a Universal Testing Machine (UTM). In addition, Scanning Electron Microscopy (SEM) was applied to study the dispersion of carbon nanotubes in the nanocomposites. The results showed that PP/MWCNT nanocomposites had Young’s modulus of 1,740 MPa and tensile strength of 34.5 MPa while original PP had 1,450 MPa and 28 MPa, respectively. Therefore, the mechanical properties were improved significantly with the content of MWCNTs. Full Factorial experiments investigate that significant factors are %wt, barrel temperature, injection velocity, and screw rotational speed. Moreover, SEM showed that a high content of MWCNTs led to a highly oriented skin layer with well-dispersed MWCNTs.

Temperature dependence of dynamic and mechanical properties in poly(acrylic acid)/graphene oxide nanocomposites

COVID-19 is highly infectious and causes acute respiratory disease. Machine learning (ML) and deep learning (DL) models are vital in detecting disease from computerized chest tomography (CT) scans. The DL models outperformed the ML models. For COVID-19 detection from CT scan images, DL models are used as end-to-end models. Thus, the performance of the model is evaluated for the quality of the extracted feature and classification accuracy. There are four contributions included in this work. First, this research is motivated by studying the quality of the extracted feature from the DL by feeding these extracted to an ML model. In other words, we proposed comparing the end-to-end DL model performance against the approach of using DL for feature extraction and ML for the classification of COVID-19 CT scan images. Second, we proposed studying the effect of fusing extracted features from image descriptors, e.g., Scale-Invariant Feature Transform (SIFT), with extracted features from DL models. Third, we proposed a new Convolutional Neural Network (CNN) to be trained from scratch and then compared to the deep transfer learning on the same classification problem. Finally, we studied the performance gap between classic ML models against ensemble learning models. The proposed framework is evaluated using a CT dataset, where the obtained results are evaluated using five different metrics The obtained results revealed that using the proposed CNN model is better than using the well-known DL model for the purpose of feature extraction. Moreover, using a DL model for feature extraction and an ML model for the classification task achieved better results in comparison to using an end-to-end DL model for detecting COVID-19 CT scan images. Of note, the accuracy rate of the former method improved by using ensemble learning models instead of the classic ML models. The proposed method achieved the best accuracy rate of 99.39%.

PurposeTo conduct a head-to-head comparison between deep learning (DL) and radiomics models across institutions for predicting microvascular invasion (MVI) in hepatocellular carcinoma (HCC) and to investigate the model robustness and generalizability through rigorous internal and external validation.MethodsThis retrospective study included 2304 preoperative images of 576 HCC lesions from two centers, with MVI status determined by postoperative histopathology. We developed DL and radiomics models for predicting the presence of MVI using B-mode ultrasound, contrast-enhanced ultrasound (CEUS) at the arterial, portal, and delayed phases, and a combined modality (B + CEUS). For radiomics, we constructed models with enlarged vs. original regions of interest (ROIs). A cross-validation approach was performed by training models on one center’s dataset and validating the other, and vice versa. This allowed assessment of the validity of different ultrasound modalities and the cross-center robustness of the models. The optimal model combined with alpha-fetoprotein (AFP) was also validated. The head-to-head comparison was based on the area under the receiver operating characteristic curve (AUC).ResultsThirteen DL models and 25 radiomics models using different ultrasound modalities were constructed and compared. B + CEUS was the optimal modality for both DL and radiomics models. The DL model achieved AUCs of 0.802–0.818 internally and 0.667–0.688 externally across the two centers, whereas radiomics achieved AUCs of 0.749–0.869 internally and 0.646–0.697 externally. The radiomics models showed overall improvement with enlarged ROIs (P < 0.05 for both CEUS and B + CEUS modalities). The DL models showed good cross-institutional robustness (P > 0.05 for all modalities, 1.6–2.1% differences in AUC for the optimal modality), whereas the radiomics models had relatively limited robustness across the two centers (12% drop-off in AUC for the optimal modality). Adding AFP improved the DL models (P < 0.05 externally) and well maintained the robustness, but did not benefit the radiomics model (P > 0.05).ConclusionCross-institutional validation indicated that DL demonstrated better robustness than radiomics for preoperative MVI prediction in patients with HCC, representing a promising solution to non-standardized ultrasound examination procedures.

Electricity load forecasting is a crucial undertaking within all the deregulated markets globally. Among the research challenges on a global scale, the investigation of deep transfer learning (DTL) in the field of electricity load forecasting represents a fundamental effort that can inform artificial intelligence applications in general. In this paper, a comprehensive study is reported regarding day-ahead electricity load forecasting. For this purpose, three sequence-to-sequence (Seq2seq) deep learning (DL) models are used, namely the multilayer perceptron (MLP), the convolutional neural network (CNN) and the ensemble learning model (ELM), which consists of the weighted combination of the outputs of MLP and CNN models. Also, the study focuses on the development of different forecasting strategies based on DTL, emphasizing the way the datasets are trained and fine-tuned for higher forecasting accuracy. In order to implement the forecasting strategies using deep learning models, load datasets from three Greek islands, Rhodes, Lesvos, and Chios, are used. The main purpose is to apply DTL for day-ahead predictions (1–24 h) for each month of the year for the Chios dataset after training and fine-tuning the models using the datasets of the three islands in various combinations. Four DTL strategies are illustrated. In the first strategy (DTL Case 1), each of the three DL models is trained using only the Lesvos dataset, while fine-tuning is performed on the dataset of Chios island, in order to create day-ahead predictions for the Chios load. In the second strategy (DTL Case 2), data from both Lesvos and Rhodes concurrently are used for the DL model training period, and fine-tuning is performed on the data from Chios. The third DTL strategy (DTL Case 3) involves the training of the DL models using the Lesvos dataset, and the testing period is performed directly on the Chios dataset without fine-tuning. The fourth strategy is a multi-task deep learning (MTDL) approach, which has been extensively studied in recent years. In MTDL, the three DL models are trained simultaneously on all three datasets and the final predictions are made on the unknown part of the dataset of Chios. The results obtained demonstrate that DTL can be applied with high efficiency for day-ahead load forecasting. Specifically, DTL Case 1 and 2 outperformed MTDL in terms of load prediction accuracy. Regarding the DL models, all three exhibit very high prediction accuracy, especially in the two cases with fine-tuning. The ELM excels compared to the single models. More specifically, for conducting day-ahead predictions, it is concluded that the MLP model presents the best monthly forecasts with MAPE values of 6.24% and 6.01% for the first two cases, the CNN model presents the best monthly forecasts with MAPE values of 5.57% and 5.60%, respectively, and the ELM model achieves the best monthly forecasts with MAPE values of 5.29% and 5.31%, respectively, indicating the very high accuracy it can achieve.

Recently, highly conductive polymer nanocomposites, particularly soft polymer nanocomposites, have received extensive attention as promising material candidates for wearable devices. Compared with the cases of the wearable devices based on conventional rigid electronic materials, the wearable devices based on polymer nanocomposites exhibit excellent conformal contacts with the skin due to the soft mechanical properties of these nanocomposites; therefore, soft polymeric nanocomposites can be applied to stretchable wirings, electrodes, and sensor units in various on-skin electronics. The types of polymers and nanofillers used for the synthesis of these nanocomposites are critical factors determining the properties of polymer nanocomposites. The overall physical properties of nanocomposites depend on the type of polymer used, whereas the electrical properties of nanocomposites are governed by the type of nanofiller employed. Herein, we review the latest studies on the polymer nanocomposites constructed using different polymers and nanofillers that are applied to wearable devices. We have classified the polymers into non-elastic polymers, hydrogels, chemically crosslinked elastomers, and physically crosslinked elastomers and the nanofillers into C, liquid metal, Ag, Au, and other emerging nanomaterials. Detailed characteristics, fabrication methods, applications, and limitations of these nanocomposites are reviewed. Finally, a brief outlook for future research is provided.

Carbon-nitride nanosheets have attracted remarkable attention in recent years due to their outstanding physical properties. C7N6 is one of the hotspot nanosheets which possesses excellent mechanical, electrical, and optical properties. In this study, the coupled thermo-mechanical properties of the single nanosheet C7N6 are systematically investigated. Although temperature effects have a strong influence on the mechanical properties of C7N6 monolayer, thermal effects were not fully analyzed for carbon-nitride nanosheet and still an open topic. To this end, the presented contribution aims to highlight this important aspect and investigate the temperature influence on the mechanical stress-strain response. By using molecular dynamics (MD) simulation, we have found out that the C7N6 monolayer’s maximum strength decreases as the temperature increase from 300 K to 1100 K. In the current contribution, 5% to 15% volume fractions of C7N6/P3HT composite were employed to investigate the C7N6 reinforcing ability. Significantly, the uniaxial tensile of C7N6/P3HT composite reveals that 10%C7N6 can enhance the maximum strength of the composite to 121.80 MPa which is 23.51% higher than the pure P3HT matrix. Moreover, to better understand the enhanced mechanism, we proposed a cohesive model to investigate the interface strength between the C7N6 nanosheet and P3HT matrix. This systematic study provides not only a sufficient method to understand the C7N6 thermo-mechanical properties, but also the reinforce mechanism of the C7N6 reinforced nanocomposite. Thus, this work provides a valuable method for the later investigation of the C7N6 nanosheet.

The effect of different types of fillers on morphology and mechanical properties of polymer nanocomposites has been investigated using ethylene-1–octene copolymer (EOC), a polyolefin based elastomer, as matrix and various nanofillers {such as multi-walled carbon nanotubes (MWCNT), layered silicate (LS) and boehmite (OS2)}. The morphological structures were studied by scanning electron microscopy (SEM) while the mechanical properties were characterized by tensile testing and microindentation hardness measurements. It has been shown that the nature of the nanofiller may have significant influence on the mechanical properties of the samples. Among the nanocomposites studied so far, the MWCNT filled samples showed the highest reinforcing effect followed by layered silicate. The least reinforcing effect was obtained for the samples filled with boehmite nanoparticles. Nepal Journal of Science and Technology Vol. 13, No. 2 (2012) 103-108 DOI: http://dx.doi.org/10.3126/njst.v13i2.7721