Kartik Ayyer

Free-standing thin liquid films offer a convenient way to probe matter in solution without a substrate, where a small sample thickness is required. The production of thin films in a vacuum environment is required for probes that cannot penetrate through a surrounding gas or in cases where the gas produces problematic background noise. Nozzles that produce continuously flowing liquid sheet jets have been employed recently for scattering and spectroscopy measurements at X-ray free-electron lasers (XFELs). However, their performance in vacuum with aqueous samples is complicated by high freezing susceptibility due to a strong evaporative cooling effect. Here, we introduce a vacuum-compatible gas-impinging liquid sheet jet nozzle capable of generating sub-100 nm aqueous films. The introduction of a second axisymmetric helium gas sheath surrounding the liquid orifice enables seamless startup and shutdown within a vacuum environment. We investigated the sheet geometry dependence under varying flow conditions and liquid viscosity and demonstrated its viability for X-ray scattering measurements at a high-repetition-rate XFEL source. Finally, we present potential strategies for reducing sample consumption. Our findings confirm that this nozzle exhibits simplified vacuum operation and may help promote the broader utilization of liquid sheet jets for several experimental applications.

Single-stranded RNA viruses co-assemble their capsid with the genome, and variations in capsid structures can have significant functional relevance. In particular, viruses need to respond to a dehydrating environment to prevent genomic degradation and remain active upon rehydration. Theoretical work has predicted low-energy buckling transitions in icosahedral capsids, which could protect the virus from further dehydration. However, there has been no direct experimental evidence, nor a molecular mechanism, for such behavior. Here, we observe this transition using X-ray single particle imaging of MS2 bacteriophages after aerosolization. Using a combination of machine learning tools, we classify hundreds of thousands of single-particle diffraction patterns to learn the structural landscape of the capsid morphology as a function of time spent in the aerosol phase. We found a previously unreported compact conformation as well as intermediate structures that suggest an incoherent buckling transition that does not preserve icosahedral symmetry. Finally, we propose a mechanism for this buckling, where a single 19-residue loop is destabilized, leading to the large observed morphological change. Our results provide experimental evidence for a mechanism by which viral capsids may protect themselves from dehydration upon aerosolization. In the process, these findings also demonstrate the power of single-particle X-ray imaging and machine learning methods in studying biomolecular structural dynamics.

Single-particle small-angle X-ray scattering (SP-SAXS) at X-ray free electron lasers (XFELs) enables quantitative analysis of morphological heterogeneity that is fundamentally inaccessible to ensemble-averaged in situ techniques. By recording diffraction snapshots from isolated particles, SP-SAXS resolves low-contrast, less abundant, or transient species within heterogeneous particle populations that would otherwise remain hidden to conventional X-ray techniques. We demonstrate this unique capability by investigating the solvothermal formation of CoO nanocrystal assemblies from a Co(acac)₃ precursor in benzyl alcohol. The single-particle data revealed amorphous, uniform-density Co(acac)₂ spheres as transient intermediates that directly crystallize into cavernous CoO nanocrystal assemblies, explaining why CoO forms as hierarchical aggregates rather than as isolated nanocrystals. These results establish SP-SAXS as a uniquely powerful framework for uncovering nonclassical nanoparticle formation pathways hidden in ensemble measurements.

Single-particle small-angle X-ray scattering (SP-SAXS) enables quantitative morphological analysis by recording diffraction snapshots from isolated particles using X-ray free-electron laser (XFEL) pulses. Unlike conventional X-ray techniques, which average over the entire illuminated sample volume, SP-SAXS resolves low-contrast, less abundant, or transient species within heterogeneous particle populations that would otherwise remain hidden. Here, we apply SP-SAXS to investigate the solvothermal formation of CoO nanocrystal assemblies from a Co(acac)$_3$ precursor in benzyl alcohol. The single-particle data reveal amorphous, uniform-density Co(acac)$_2$ spheres as transient intermediates that directly crystallize into cavernous CoO nanocrystal assemblies, which explains why CoO forms as hierarchical aggregates rather than as isolated nanocrystals. These results demonstrate that SP-SAXS provides a powerful framework for disentangling morphological heterogeneity in nanoparticle formation processes.

The Outstanding optoelectronic properties of lead halide perovskites have been related to the formation of polarons. Nevertheless, the observation of the atomistic deformation brought about by one electron-hole pair in these materials has remained elusive. Here, we measure the transient structure of single CsPbBr₃ quantum dots (QDs) after resonant excitation in the single exciton limit using serial femtosecond crystallography (SFX). By reconstructing the three-dimensional (3D) differential diffraction pattern and building on density functional theory (DFT) calculations, we assign the lattice distortion after photoexcitation to the combined presence of a delocalized electron and a localized hole, forming a mixed large/small exciton-polaron. This result creates a clear picture of the polaronic deformation in CsPbBr₃ QDs and demonstrates the exceptional sensitivity of SFX to lattice distortions in few-nanometer crystallites. We plan to use this experimental platform for future studies of electron-lattice interactions.

X-ray free-electron lasers offer unique capabilities for measuring the structure and dynamics of biomolecules, helping us understand the basic building blocks of life. Notably, high-repetition-rate free-electron lasers enable single particle imaging, where individual, weakly scattering biomolecules are imaged under near-physiological conditions with the opportunity to access fleeting states that cannot be captured in cryogenic or crystallized conditions. Existing X-ray single particle reconstruction algorithms, which estimate the particle orientation for each image independently, are slow and memory-intensive when handling the massive datasets generated by emerging free-electron lasers. Here, we introduce X-RAI (X-Ray single particle imaging with Amortized Inference), an online reconstruction framework that estimates the structure of 3D macromolecules from large X-ray single particle datasets. X-RAI consists of a convolutional encoder, which amortizes pose estimation over large datasets, as well as a physics-based decoder, which employs an implicit neural representation to enable high-quality 3D reconstruction in an end-to-end, self-supervised manner. We demonstrate that X-RAI achieves state-of-the-art performance for small-scale datasets in simulation and challenging experimental settings and demonstrate its unprecedented ability to process large datasets containing millions of diffraction images in an online fashion. These abilities signify a paradigm shift in X-ray single particle imaging towards real-time reconstruction.

Achieving the highest possible intensities of pulses at X-ray free-electron laser (XFEL) facilities entails focusing the X-ray beam to the smallest possible dimensions. Multilayer Laue lenses are volume diffractive optics that hold promise to achieve high intensities due to their high numerical aperture at hard X-ray wavelengths. These lenses are made by layer deposition and hence have a small aperture that is usually less than 100 μm. Consequently, they must withstand high intensities and heat loads if used to focus pulses of high energies. Here, we demonstrate the focusing of XFEL pulses using lenses made from layers of Mo₂C and SiC, which were chosen for their high efficiency and low beam heating. We show that lens alignment and characterisation can be carried out using attenuated XFEL pulses using the method of ptychographic X-ray speckle tracking, which also provides an approach for high-resolution projection imaging with XFEL pulses. The approach also gives quantitative information about the distribution of intensity in the focal plane, which is analysed here using plots of the encircled energy describing the proportion of the beam energy concentrated into a certain diameter, showing the possibility to reach 1.8 × 10²⁰ W cm⁻² at a photon energy of 16.9 keV. No measurable change in the wavefront aberrations of the lenses could be detected after their use in an experiment for many days with unattenuated XFEL pulses, showing that the materials and mounting scheme presented here make these lenses suitable for sustained use at XFEL facilities.

The outstanding opto-electronic properties of lead halide perovskites have been related to the formation of polarons. Nevertheless, the observation of the atomistic deformation brought about by one electron-hole pair in these materials has remained elusive. Here, we measure the diffraction patterns of single CsPbBr$_3$ quantum dots (QDs) with and without resonant excitation in the single exciton limit using serial femtosecond crystallography (SFX). By reconstructing the 3D differential diffraction pattern, we observe small shifts of the Bragg peaks indicative of a crystal-wide deformation field. Building on DFT calculations, we show that these shifts are consistent with the lattice distortion induced by a delocalized electron and a localized hole, forming a mixed large/small exciton polaron. This result creates a clear picture of the polaronic deformation in CsPbBr$_3$ QDs, highlights the exceptional sensitivity of SFX to lattice distortions in few-nanometer crystallites, and establishes an experimental platform for future studies of electron-lattice interactions.

X-ray free electron lasers capture single-particle diffraction patterns at scale. We showcase a classification algorithm that enables online monitoring and heterogeneity analysis, accelerating insights from massive, structurally diverse datasets.

Data throughput from x-ray free-electron laser (XFEL) facilities has increased dramatically over the last few years as their pulse repetition rates rise from tens to thousands of patterns per second. Such throughput creates unprecedented opportunities for serially imaging ensembles of particles that exhibit functionally relevant heterogeneity ¹ . Unlike conventional Single-Particle-Imaging ^{2, 3} where the goal is to reconstruct a handful of average structures of the ensemble ⁴ , here we explicitly seek the statistical distribution of variations amongst particles. Formally we wish to estimate the high-dimensional structural posterior distributions of these million-particle ensembles. Such posterior estimation, however, is fraught with computationally expensive inference of missing latent variables from shot-noise-limited diffraction patterns ^{2, 4– 8} . This inference is made even more challenging with the much larger datasets from Megahertz XFEL facilities ^{9, 10} on particle ensembles that are structurally heterogenous. Moving forward, the important question is “ can we scale up and scale out posterior estimation to million-particle XFEL datasets and beyond? ” This presentation describes several relevant strategies. In terms of modeling, we show how structural variations can be efficiently parametrized to reduce the number of dimensions of the search space by several orders of magnitude – this conveniently allowed us to match real-space structures to diffraction patterns without performing Fast Fourier Transforms. We accelerated the posterior estimation with Markov Chain Monte Carlo, coupled with a combination of “hard” and “soft” classification on latent parameters. We represented intermediate distributions in the high-dimensional latent spaces sparsely, which enabled us to efficiently distribute the inference over many commodity Graphics Processing Units (GPUs).