For over 80 years, tailored molecular assemblies (e.g., H-and J-aggregates) have been of interest for the emergence of collective phenomena in their optical spectra 1 , coherent long-range energy transport 2,3 and their conceptual similarity with natural light-harvesting complexes 4,5 . Another highly versatile platform for creating controlled, aggregated states exhibiting collective phenomena arises from the organization of colloidal semiconductor nanocrystals (NCs) into long-range ordered superlattices (SLs) 6 . Cesium lead halide perovskite NCs 7-9 have recently emerged as highly appealing building blocks, owing to their high oscillator strength 10 , slow dephasing (long coherence times of up to 80 ps) 11,12 , minimal inhomogeneous broadening of emission lines, and a bright triplet exciton character with orthogonal dipole orientation 10 , potentially enabling an efficient omnidirectional coupling. Here we present perovskite-type (ABO3) binary and ternary NC SLs by a shapedirected co-assembly of steric-stabilized, highly luminescent cuboid-shaped CsPbBr3 NCs (occupying B-and/or O-sites) with spherical Fe3O4 or NaGdF4 NCs (A-sites) and truncated-cuboid PbS NCs (B-site). Such ABO3 SLs, as well as other newly obtained SL structures (binary NaCl-and AlB2-types), exhibit a high degree of orientational ordering of CsPbBr3 nanocubes. These novel perovskite mesostructures exhibit superfluorescence (SF) -a collective emission resulting in a burst of photons. SF is characterized, at high excitation density, by emission pulses with ultrafast (22 ps) radiative decay and Burnham-Chiao ringing behaviour with a strongly accelerated build-up time.

Self-assembly of colloidal nanocrystals (NCs) holds great promise in the multiscale engineering of solid-state materials, whereby atomically engineered NC building blocks are arranged into long-range ordered structuressuperlattices (SLs)with synergistic physical and chemical properties. Thus far, the reports have by far focused on singlecomponent and binary systems of spherical NCs, yielding SLs isostructural with the known atomic lattices. Far greater structural space, beyond the realm of known lattices, is anticipated from combining NCs of various shapes. Here, we report on the co-assembly of stericstabilized CsPbBr 3 nanocubes (5.3 nm) with disk-shaped LaF 3 NCs (9.2− 28.4 nm in diameter, 1.6 nm in thickness) into binary SLs, yielding six columnar structures with AB, AB 2 , AB 4 , and AB 6 stoichiometry, not observed before and in our reference experiments with NC systems comprising spheres and disks. This striking effect of the cubic shape is rationalized herein using packing-density calculations. Furthermore, in the systems with comparable dimensions of nanocubes (8.6 nm) and nanodisks (6.5 nm, 9.0 nm, 12.5 nm), other, noncolumnar structures are observed, such as ReO 3 -type SL, featuring intimate intermixing and face-to-face alignment of disks and cubes, face-centered cubic or simple cubic sublattice of nanocubes, and two or three disks per one lattice site. Lamellar and ReO 3 -type SLs, employing large 8.6 nm CsPbBr 3 NCs, exhibit characteristic features of the collective ultrafast light emissionsuperfluorescenceoriginating from the coherent coupling of emission dipoles in the excited state.

Nanoscale electron-induced reactions being triggered by a finely focused electron beam in modern scanning electron microscopes are commonly used to pattern surfaces of thin films of irradiation sensitive material. Classical polymer and inorganic resist films allow precise masks to be defined for further deposition or etching process steps in the semiconductor industry. [1] A new, promising approach employs new film materials, among which are self-assembled monolayers of biphenyl, passivated gold nanoclusters, Langmuir-Blodgett films, or liquid precursors. The exposure of these films to electrons directly results in membranes, electrical wires, plasmonic structures, or conducting dots with nanoscale dimensions. [2] A very promising approach to electron-impact nanosynthesis is to replace the solid or liquid film and use a physisorbed monolayer that is continuously refreshed by injected volatile molecules. [3] The process can be compared to local chemical vapor deposition; however, the decomposition is due to electron-impact dissociation rather than thermal dissociation, thus keeping the reaction confined to the size of the electron beam and the active electron interaction volume. It has been proven to be a very innovative concept for direct, local, three-dimensional, and minimally invasive nanosynthesis of future photonic, [4] electronic, [5] and mechanical [6] nanodevice materials as well as for site-specific patterning of catalyst for individual carbon nanotube growth [7] and atomic layer deposition. [8] For nanoscale deposition, a focused electron beam is usually scanned over surface-adsorbed metal containing compounds that are volatile at room temperature. Electron-impact dissociation of such adsorbates by both the primary beam electrons (with keV energy) and the emitted secondary electrons (with eV energy) results in metalcontaining deposits and volatile reaction products, the latter being removed by the vacuum system. Advantageously, the same principle allows nanoscale removal of material. For example, physisorbed water on carbon surfaces dissociates under electron impact to produce highly reactive species that react to volatile carbon compounds, thus etching a nanosized hole in the substrate when a stationary focused electron beam is used. [9] Injected molecules used for electron-impact nanosynthesis so far comprise various metal-ligand compounds that contain carbon-, phosphorous-, or halogen-based ligands as well as organic compounds. [3] With the recent development of gas injection systems that allow the admission of two or more gases to the substrate surface, the nanosynthesis of binary metal alloys [10] or metal-(carbon) matrix deposits with outperforming properties [5c,e] can be envisaged. In contrast to classical vapor deposition exploiting co-evaporation, the deposit will be locally confined and the composition will depend not only on the ratios of molecule flow adsorption but also on the electron-impact dissociation efficiency of each individual molecule, giving a further degree of freedom to t...

We propose an effective deep learning model to denoise scanning transmission electron microscopy (STEM) image series, named Noise2Atom, to map images from a source domain S to a target domain C, where S is for our noisy experimental dataset, and C is for the desired clear atomic images. Noise2Atom uses two external networks to apply additional constraints from the domain knowledge. This model requires no signal prior, no noise model estimation, and no paired training images. The only assumption is that the inputs are acquired with identical experimental configurations. To evaluate the restoration performance of our model, as it is impossible to obtain ground truth for our experimental dataset, we propose consecutive structural similarity (CSS) for image quality assessment, based on the fact that the structures remain much the same as the previous frame(s) within small scan intervals. We demonstrate the superiority of our model by providing evaluation in terms of CSS and visual quality on different experimental datasets.

We propose an effective deep learning model to denoise scanning transmission electron microscopy (STEM) image series, named Noise2Atom, to map images from a source domain $\mathcal {S}$ S to a target domain $\mathcal {C}$ C , where $\mathcal {S}$ S is for our noisy experimental dataset, and $\mathcal {C}$ C is for the desired clear atomic images. Noise2Atom uses two external networks to apply additional constraints from the domain knowledge. This model requires no signal prior, no noise model estimation, and no paired training images. The only assumption is that the inputs are acquired with identical experimental configurations. To evaluate the restoration performance of our model, as it is impossible to obtain ground truth for our experimental dataset, we propose consecutive structural similarity (CSS) for image quality assessment, based on the fact that the structures remain much the same as the previous frame(s) within small scan intervals. We demonstrate the superiority of our model by providing evaluation in terms of CSS and visual quality on different experimental datasets.