Magnetic digital microfluidics (MDM) offers an attractive approach to miniaturised in vitro diagnostics, but most conventional platforms remain limited to predefined, planar droplet handling without the ability to rectify operation failures. Here, we present an intelligent MDM platform in which magnetic soft millirobots (MSMRs) function as reconfigurable robotic end-effectors for 3D droplet manipulation under closed-loop control based on 2D visual detection and height-dependent perspective correction. A magnetic pick-and-place mechanism exchanges a 3D-printed MSMR gripper and pipettor on demand. These MSMRs transport, merge, dispense, and split droplets, as well as collect and resuspend bacterial colonies in an oil-immersed workspace. The MSMR gripper captured 10–30 μL droplets with a success rate above 95%, and the MSMR pipettor dispensed 5.19 μL droplets with a coefficient of variation of 2.1%. Both tools retained their droplet-handling function after 10 000 reversible actuation cycles. The computer vision module enables real-time recognition and tracking, spatial calibration, operation verification, failure rectification, and result interpretation. With samples, reagents, and protocols preloaded, the platform autonomously completed the Carba NP assay for detecting carbapenemase-producing Enterobacteriaceae. These results establish MSMRs as a promising strategy for adaptive, failure-aware MDM platforms for diagnostic automation.
Polyetheretherketone (PEEK) is attractive for patient-specific load-bearing implants, but its low absorption at near-infrared radiation (NIR) fibre-laser wavelengths and narrow processing window make powder bed fusion challenging. Here, Ti3C2Tx MXene nanosheets were conformally assembled on PEEK particles by electrostatic adsorption to construct Ti3C2Tx-sheathed powders with improved NIR absorption, heat transfer, and flowability. The modified powders enable stable multilayer fibre-laser powder bed fusion (f-LPBF) at ultralow volume energy densities. For PEEK/M1, the required energy density was reduced from 26.667 to 0.333 J⊕mm−3, representing a 98.75% reduction relative to pristine PEEK, while maintaining a slightly expander processable window and suppressing carbonisation-induced instability. The printed S-PEEK/M1 composites showed enhanced interlayer fusion, smoother surfaces, increased crystallinity, and a transcrystalline microstructure, resulting in a tensile strength of 71.71 ± 5.06 MPa, a flexural strength of 87.35 ± 8.02 MPa, and a Vickers hardness of 39.78 ± 3.22 HV. This work demonstrates a powder-level MXene engineering route that couples optical-thermal processability with mechanical reinforcement, providing a general strategy for high-throughput f-LPBF of high-performance polymer composites for biomedical and engineering applications.
Extrusion-based additive manufacturing is an emerging approach for fabricating architecturally complex biodegradable metallic bone implants, enabling zinc (Zn) to be processed into intricate scaffold architectures. Here, triply periodic minimal surface (TPMS) Zn scaffolds with gyroid, diamond, and I-WP architectures were fabricated by extrusion-based additive manufacturing followed by debinding and sintering. Scaffolds with solid fractions of 20, 30, and 40 vol% were produced, and the 30 vol% scaffolds were selected for characterisation. All architectures retained TPMS geometries and exhibited interconnected pore networks with absolute porosities exceeding 45%, while the gyroid showed the highest shape fidelity. Degradation rates decreased from 0.070–0.087 mm/year at day 7 to 0.039–0.052 mm/year at day 28, with the diamond degrading slowest. Before immersion, the diamond exhibited the highest compressive strength (5.5 ± 0.4 MPa) and elastic modulus (0.67 ± 0.19 GPa). During immersion, compressive strength increased due to corrosion-product deposition. Finite element and computational fluid dynamics analyses revealed architecture-dependent mechanical and transport behaviour, with the gyroid showing the highest permeability and lowest pressure drop. All scaffolds supported preosteoblast attachment, with the gyroid showing the highest metabolic activity. These findings demonstrate that TPMS architecture strongly governs the structural, degradation, mechanical, transport, and biological performance of extrusion-printed Zn scaffolds.
The high-temperature creep resistance of rapidly solidified aluminum (Al) alloys depends on the evolution of their strengthening phases under sustained loading. Here, we map the crack-free laser powder bed fusion (L-PBF) processing window of an Al85Cu5Li4Mg3Zn3 lightweight Al-based entropy alloy (LAEA) and probe how thermally established phases evolve during high-temperature compressive creep. After thermal exposure at 300°C, continuous nano-sized cellular eutectic networks are retained together with S-Al2CuMg and TB-Al7Cu4Li phases, whereas exposure at 400°C decomposes the networks into coarse and dispersed σ-Al5Cu6Mg2 and R-Al5CuLi3 phases. The alloy exposed at 300°C exhibits the lowest minimum creep strain rate over the common stress range at 300°C, with effective threshold stresses of 80.2 and 69.6 MPa at 200 and 300°C, respectively. Retained cellular interfaces confine dislocation motion and promote the rearrangement of accumulated dislocations into regularly polygonal interfacial dislocation networks while allowing dynamic recovery. Local and incomplete TB-to-R and S-to-θ transformations are further observed after creep at both temperatures, supporting a role for creep-generated defects in solute redistribution and interface migration. These results link the retained nano-sized cellular eutectic networks to dislocation confinement and rearrangement during creep, while showing that creep-generated defects can further assist local phase transformations.
The accuracy and efficiency of wire arc additive manufacturing (WAAM) depend on the geometrically consistent translation of a target geometry into multi-layer multi-pass bead layouts and robot-executable toolpaths under prescribed process parameters. To bridge this design-to-deposition gap, this study introduces a target-geometry-driven framework that integrates four capabilities: weld bead geometry prediction, primitive-based deposition geometry modelling, layout parameter optimisation, and toolpath generation. The weighted mean of vectors (INFO) algorithm was integrated with a cascade-forward backpropagation (CFBP) neural network to predict bead height, width, and cross-sectional area, yielding [Formula: see text] values of 0.08914 mm, 0.1160 mm, and 0.3874 mm², respectively. Based on these predictions, a multi-layer multi-pass deposition geometry model incorporating constraint-state decomposition and an overlap-ratio criterion was developed to rapidly determine the number of layers, beads per layer, inter-bead overlap relationships, and deposition toolpaths. Validation experiments on four component types yielded height and width forming errors relative to the target geometry of -4.3527% to 4.7484% and -6.1409% to 7.9370%, respectively. The manufacturing execution errors relative to model-predicted dimensions were -3.9984% to 5.1799% and -4.0145% to 4.6739%. The proposed framework can rapidly generate deposition layouts and toolpaths from prescribed process parameters and specified target geometries within the investigated process, material, and geometric ranges.
