Generalisable 3D printing error detection and correction via multi-head neural networks

Material extrusion is the most widespread additive manufacturing method but its application in end-use products is limited by vulnerability to errors. Humans can detect errors but cannot provide continuous monitoring or real-time correction. Existing automated approaches are not generalisable across different parts, materials, and printing systems. We train a multi-head neural network using images automatically labelled by deviation from optimal printing parameters. The automation of data acquisition and labelling allows the generation of a large and varied extrusion 3D printing dataset, containing 1.2 million images from 192 different parts labelled with printing parameters. The thus trained neural network, alongside a control loop, enables real-time detection and rapid correction of diverse errors that is effective across many different 2D and 3D geometries, materials, printers, toolpaths, and even extrusion methods. We additionally create visualisations of the network's predictions to shed light on how it makes decisions.

Real-Time Imaging of Self-Organization and Mechanical Competition in Carbon Nanotube Forest Growth

The properties of carbon nanotube (CNT) networks and analogous materials comprising filamentary nanostructures are governed by the intrinsic filament properties and their hierarchical organization and interconnection. As a result, direct knowledge of the collective dynamics of CNT synthesis and self-organization is essential to engineering improved CNT materials for applications such as membranes and thermal interfaces. Here, we use real-time environmental transmission electron microscopy (E-TEM) to observe nucleation and self-organization of CNTs into vertically aligned forests. Upon introduction of the carbon source, we observe a large scatter in the onset of nucleation of individual CNTs and the ensuing growth rates. Experiments performed at different temperatures and catalyst particle densities show the critical role of CNT density on the dynamics of self-organization; low-density CNT nucleation results in the CNTs becoming pinned to the substrate and forming random networks, whereas higher density CNT nucleation results in self-organization of the CNTs into bundles that are oriented perpendicular to the substrate. We also find that mechanical coupling between growing CNTs alters their growth trajectory and shape, causing significant deformations, buckling, and defects in the CNT walls. Therefore, it appears that CNT-CNT coupling not only is critical for self-organization but also directly influences CNT quality and likely the resulting properties of the forest. Our findings show that control of the time-distributed kinetics of CNT nucleation and bundle formation are critical to manufacturing well-organized CNT assemblies and that E-TEM can be a powerful tool to investigate the mesoscale dynamics of CNT networks.

Crystal engineering of zeolites with graphene

Achieving control over the morphology of zeolite crystals at the nanoscale is crucial for enhancing their performance in diverse applications including catalysis, sensors and separation. The complexity and sensitivity of zeolite synthesis processes, however, often make such control both highly empirical and difficult to implement. We demonstrate that graphene can significantly alter the morphology of titanium silicalite (TS-1) particles, in particular being able to reduce their dimensions from several hundreds to less than 10 nm. Through electron microscopy and molecular mechanics simulations we propose a mechanism for this change based on the preferential interaction of specific TS-1 surfaces with benzyl-alcohol-mediated graphene. These findings suggest a facile new means of controlling the zeolite morphology and thereby also further demonstrate the potential of graphene in hybrid materials. Moreover, the generality of the mechanism points the way to a new avenue of research in using two-dimensional materials to engineer functional inorganic crystals.

En route to controlled catalytic CVD synthesis of densely packed and vertically aligned nitrogen-doped carbon nanotube arrays

The catalytic chemical vapour deposition (c-CVD) technique was applied in the synthesis of vertically aligned arrays of nitrogen-doped carbon nanotubes (N-CNTs). A mixture of toluene (main carbon source), pyrazine (1,4-diazine, nitrogen source) and ferrocene (catalyst precursor) was used as the injection feedstock. To optimize conditions for growing the most dense and aligned N-CNT arrays, we investigated the influence of key parameters, i.e., growth temperature (660, 760 and 860 °C), composition of the feedstock and time of growth, on morphology and properties of N-CNTs. The presence of nitrogen species in the hot zone of the quartz reactor decreased the growth rate of N-CNTs down to about one twentieth compared to the growth rate of multi-wall CNTs (MWCNTs). As revealed by electron microscopy studies (SEM, TEM), the individual N-CNTs (half as thick as MWCNTs) grown under the optimal conditions were characterized by a superior straightness of the outer walls, which translated into a high alignment of dense nanotube arrays, i.e., 5 × 10(8) nanotubes per mm(2) (100 times more than for MWCNTs grown in the absence of nitrogen precursor). In turn, the internal crystallographic order of the N-CNTs was found to be of a 'bamboo'-like or 'membrane'-like (multi-compartmental structure) morphology. The nitrogen content in the nanotube products, which ranged from 0.0 to 3.0 wt %, was controlled through the concentration of pyrazine in the feedstock. Moreover, as revealed by Raman/FT-IR spectroscopy, the incorporation of nitrogen atoms into the nanotube walls was found to be proportional to the number of deviations from the sp(2)-hybridisation of graphene C-atoms. As studied by XRD, the temperature and the [pyrazine]/[ferrocene] ratio in the feedstock affected the composition of the catalyst particles, and hence changed the growth mechanism of individual N-CNTs into a 'mixed base-and-tip' (primarily of the base-type) type as compared to the purely 'base'-type for undoped MWCNTs.

Nitrogen-induced catalyst restructuring for epitaxial growth of multiwalled carbon nanotubes

The ability to simply and economically produce carbon nanotubes (CNTs) with a defined chiral angle is crucial for the exploitation of nanotubes for their electrical properties. We investigate a diverse range of nitrogen sources for their ability to control CNT chiral angle via epitaxial growth from highly ordered catalyst particles. Through the use of in situ mass and infrared spectrometry, we elucidate the mechanism by which these ordered catalyst particles are formed, showing that ammonia is a key intermediate in the process. Subsequently, the direct addition of a small amount of ammonia to an otherwise standard CNT synthesis is shown to be able to form catalyst particles that grow single chiral angle multiwalled carbon nanotubes. Variation in the ammonia concentration clarifies the catalyst restructuring necessary for the epitaxial growth of carbon nanotubes and subsequent chiral angle control. The simple addition of a nitrogen source is an attractive route for chiral angle control; however, the model also suggests further ways to optimize CNT chiral angle distributions as well as to improve CNT and graphene yield and crystallinity. This understanding also explains the action of ammonia in its widely used role in activating catalyst prior to CNT growth. Finally, this work highlights the uses of novel surface geometries that are achievable through multiphase catalysts.