An official website of the United States government
Roll-to-roll printing has significantly shortened the time from design to production of sensors and IoT devices, while being cost-effective for mass production. But due to less manufacturing tolerance controls available, properties such as sensor thickness, composition, roughness, etc., cannot be precisely controlled. Since these properties likely affect the sensor behavior, roll-to-roll printed sensors require validation testing before they can be deployed in the field. In this work, we improve the testing of Nitrate sensors that need to be calibrated in a solution of known Nitrate concentration for around 1–2 days. To accelerate this process, we observe the initial behavior of the sensors for a few hours, and use a physics-informed machine learning method to predict their measurements 24 hours in the future, thus saving valuable time and testing resources. Due to the variability in roll-to-roll printing, this prediction task requires models that are robust to changes in properties of the new test sensors. We show that existing methods fail at this task and describe a physics-informed machine learning method that improves the prediction robustness to different testing conditions (≈ 1.7× lower in real-world data and ≈ 5× lower in synthetic data when compared with the current state-of-the-art physics-informed machine learning method).
more »
« less
Physics Constrained Multi-objective Bayesian Optimization to Accelerate 3D Printing of Thermoplastics
The printing outcome of vat photopolymerization (VPP) of thermoplastics largely depends on physicochemical properties of monomers and their compositions in resins, which also greatly determine the material properties, e.g., tensile strength (σT) and toughness (UT)and phase transition temperature (Tg). A methodology for optimizing the resin formulation is of paramount importance in realizing highly printable thermoplastics with balanced σT/UT and target Tg while remaining largely underexplored. Herein, we introduce a multi-objective Bayesian optimization (MOBO) algorithm under two physics informed constraints (printability and Tg) to optimize two conflicting properties: σT and UT. The two constraints are formulated as two machine learning (ML) models, which are trained with weight ratios of the six monomers and physics informed (PI) descriptors derived from their physiochemical parameters. Dimensional reduction analysis reveals that the algorithm avoids recommendation of the monomer ratios that do not pass the two constraints. The printing failure rate is reduced from 16% in the background experiments to 3% in the recommended experiments. Within only 36 iterations (72 samples), the MOBO algorithm successfully identifies five sets of ratios leading to Pareto optimal of σT and UT. Due to the constraint in Tg they show appropriate Tg for shape memory application. The partial dependence analysis indicates that σT and UT depend on both the ratios and physiochemical features of the monomers. These results underscore capability of such a smart decision-making algorithm—with constraints that are not fully understood but can be informed by prior knowledge—in planning the experiments from the vast design space, thus holding a great promise for broader applications in materials design and manufacturing.
more »
« less
- Award ID(s):
- 2154428
- PAR ID:
- 10504288
- Publisher / Repository:
- RSC
- Date Published:
- Journal Name:
- ChemRxiv
- ISSN:
- 2573-2293
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Unfunctionalized vinyl-addition polynorbornene (VAPNB) possesses many outstanding properties such as high thermal, chemical, and oxidative stability. These features make VAPNB a promising candidate for many engineering applications. However, VAPNB has a small service window between its glass transition temperature ( T g ) and decomposition temperature ( T d ), and it cannot be readily processed in a melt state. In this work, we demonstrate that the service window of VAPNBs can be tailored through the use of norbornene monomers bearing alkyl, aryl, and aryl ether substituents. The vinyl addition homopolymerization and copolymerization of these functionalized norbornyl-based monomers yielded VAPNBs with high T ′ g s (>150 °C) and large service windows ( T d – T g > 100 °C), which are comparable to other commercial engineering thermoplastics. To further establish the feasibility of melt processing, a functionalized VAPNB material with T g = 209 °C and a service window of 170 °C was successfully extruded and molded into bars. Subsequent characterization of the bars by dynamic mechanical analysis (DMA), nuclear magnetic resonance spectroscopy (NMR), and gel permeation chromatography (GPC) revealed only minor signs of polymer degradation. These studies suggest that substituted VAPNBs could be developed into a new class of engineering thermoplastics that is compatible with workhorse melt processing techniques such as extrusion and injection molding, as well as emerging techniques such as extrusion-based 3D printing.more » « less
-
Abstract Laser powder bed fusion (LPBF) is a 3D printing technology that can print metal parts with complex geometries without the design constraints of traditional manufacturing routes. However, the parts printed by LPBF normally contain many more pores than those made by conventional methods, which severely deteriorates their properties. Here, by combining in-situ high-speed high-resolution synchrotron x-ray imaging experiments and multi-physics modeling, we unveil the dynamics and mechanisms of pore motion and elimination in the LPBF process. We find that the high thermocapillary force, induced by the high temperature gradient in the laser interaction region, can rapidly eliminate pores from the melt pool during the LPBF process. The thermocapillary force driven pore elimination mechanism revealed here may guide the development of 3D printing approaches to achieve pore-free 3D printing of metals.more » « less
-
The development of new thermoplastic-based nanocomposites for, as well as using, 3D printing requires extensive experimental testing. One typically goes through many failed, or otherwise sub-optimal, iterations before finding acceptable solutions (e.g. compositions, 3D printing parameters). It is desirable to reduce the number of such iterations as well as exclude failed experiments that often require laborious disassembly and cleaning of the 3D printer. This issue could be addressed if we were able to understand, and ultimately predict ahead of experiments if a given material can be 3D printed successfully. Herein, we report on our investigations into forecasting the printing and resultant properties of polymer nanocomposites while encompassing both material properties and printing parameters, enabling the model to generalize across various thermoplastics and additives. To do so, nanocomposites of two different commercially available bio-based PLAs with varying concentrations of nanoclay (NC) and graphene nanoplatelets (GNP) were prepared using a twin-screw extruder. The thermal and rheological properties of the nanocomposites were analyzed. These materials were printed at varying temperature and flow using a pellet printer. The quality of the printing was evaluated by measuring weight fluctuation, internal diameter of cylindrical specimen, and surface uniformity. The interactions between material properties and printing parameters are complex but captured effectively by a machine learning model, specifically we demonstrate such a predictive model to forecast printability and, printing quality utilizing a Random Forest algorithm. Printability was predicted by developing a classification model with constraints based on the weight fluctuation (W ) of the printed sample w.r.t. the optimal print; defining ‘‘not printable’’ for −1.0 d W < −0.8 and ‘‘printable’’ for W e −0.8. The classification model for predicting printability, performed well with an accuracy of 92.8% and identified flow index and complex viscosity, contributing 52% to the model’s importance. Another model to predict W of the only on successful prints also showed strong performance, emphasizing the importance of viscoelastic properties, thermal stability, and printing temperature. For diameter change (Di), the Random Forest model identified flow consistency index, complex viscosity, and thermal stability as influential parameters, with crystallization enthalpy gaining increased importance, reflecting its role in crystallization and shrinkage. In contrast, the surface roughness average (RA) model had lower performance, yet revealed remarkable insights regarding the feature importance with crystallization enthalpy and complex viscosity being most significant.more » « less
-
Abstract Auxetic materials have a negative Poisson’s ratio and are of significant interest in applications that include impact mitigation, membrane separations and biomedical engineering. While there are numerous examples of structured materials that exhibit auxetic behavior, the examples of engineered auxetic structures is largely limited to periodic lattice structures that are limited to directional or anisotropic auxetic response. Structures that exhibit a three-dimensionally isotropic auxetic response have been, unfortunately, slow to evolve. Here we introduce an inverse design algorithm based on global node optimization to design three-dimensional auxetic metamaterial structures from disordered networks. After specifying the target Poisson’s ratio for a structure, an inverse design algorithm is used to adjust the positions of all nodes in a disordered network structure until the desired mechanical response is achieved. The proposed algorithm allows independent control of shear and bulk moduli, while preserving the density and connectivity of the networks. When the angle bending stiffness in the network is kept low, it is possible to realize optimized structures with a Poisson’s ratios as low as −0.6. During the optimization, the bulk modulus of these networks decreases by almost two orders of magnitude, but the shear modulus remains largely unaltered. The materials designed in this manner are fabricated by dual-material 3D-printing, and are found to exhibit the mechanical responses that were originally encoded in the computational design engine. The approach proposed here provides a materials-by-design platform that could be extended for engineering of optical, acoustic, and electrical properties, beyond the design of auxetic metamaterials.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Journal Article:
- The DOI is not currently available.
