Research Highlights

Microscopy and crystallography have long served as two foundational pillars of experimental science. Microscopy relies on lenses to visualize local structures, whereas crystallography determines the global atomic arrangement of crystals through diffraction. Over the past two decades, these traditionally distinct methodologies have been unified through the rise of computational microscopy, particularly coherent diffractive imaging (CDI) and ptychography. These approaches replace physical lenses with coherent scattering and algorithmic phase retrieval, overcoming long-standing resolution limits in conventional imaging. In a recent comprehensive review, Miao summarized how CDI and ptychography now deliver exceptional imaging performance across nine orders of magnitude in length scale, from sub-angstrom visualization of atomic structures to quantitative phase imaging of centimeter-scale tissues, all based on the same underlying physical principles and closely related iterative algorithms.

Beyond 2D reconstruction, CDI and ptychography can be integrated with advanced tomographic methods to achieve 3D imaging with unprecedented detail. These combined techniques have enabled the 3D atomic structure determination of crystal defects and amorphous materials, the mapping of magnetic and electronic textures in quantum and energy materials, and high-resolution, non-destructive 3D characterization of nanomaterials and integrated circuits. They have also facilitated quantitative phase imaging of biological tissues, cells, viruses, and protein complexes, extending computational microscopy into the life sciences.

The rapid progress in this field is driven by the convergence of multiple technological advances, including fourth-generation synchrotron radiation sources, X-ray free-electron lasers, tabletop high-harmonic generation, state-of-the-art electron and optical microscopes, high-dynamic-range detectors, and deep-learning–accelerated phase retrieval. As these capabilities continue to evolve, CDI and ptychography are poised to make even broader and deeper contributions across physics, chemistry, materials science, engineering, biology, and medicine in the coming years.

High fidelity soft X-ray imaging using both tabletop and facility-scale light sources is very powerful, enabling static and dynamic imaging with ~10 nm spatial resolution. However, it can suffer from several challenges. First, phase-matched soft x-ray high harmonic generation (SXR HHG) produces a bright supercontinuum that is spatially and temporally coherent. However, reconstructing objects from broadband illumination poses a severe challenge to widely used phase retrieval techniques. The performance is further degraded by photon shot noise, detector noise, and parasitic scattering from optics. For example, stray scattered light can be a challenge for tomographic imaging – since at high tilt angles, this stray light can appear close to the scattered light from the sample (see image).

To address this challenge, STROBE developed the first ptychography algorithms that simultaneously model broadband illumination, shot/detector noise, and parasitic scattering. The first algorithm, PaCMAN, corrects bandwidth using numerical monochromatization, enabling ~4x faster data acquisition relative to ePIE when both are speed-optimized, and 1.5-2x lower dose when dose-optimized. Thus, PaCMAN can advance low-dose imaging of delicate materials (cells, polymers, catalysts). A variant, Ms. PaCMAN, replaces monochromatization with multiple wavelength-dependent modes, providing superior reconstructions to multi-wavelength ePIE, and should provide state-of-the-art reconstructions from broadband illumination across elemental absorption edges. Figure shows experimental data on soft x-ray ptychography of skyrmion samples from the COSMIC beamline at the ALS that was used to test the base ADMM algorithm in the monochromatic limit, demonstrating a substantial improvement relative to ePIE in the presence of shot noise and parasitic scattering.

This advance required the combination of x-ray imaging (Thrust II), new models and algorithms to optimize the image extraction (Thrust IV), and data from a COSMIC beam time.

Cryogenic electron microscopy (cryoEM), advanced through NSF’ STROBE, a National Science Foundation Science and Technology Center, enables unparalleled resolution for complex membrane proteins like the dystrophin-glycoprotein complex (DGC). DGC is critical in Duchenne Muscular Dystrophy (DMD), affecting 1 in 3,500–5,000 male newborns. This enormous membrane protein complex, with dystrophin at its core, links the extracellular matrix to the cytoskeleton, protecting muscle membranes during contraction. Its multi-subunit complexity defies traditional methods like X-ray crystallography. STROBE’s cutting-edge cryoEM techniques allowed us to image DGC from rabbit skeletal muscle directly, collecting ~27,000 high-resolution images for near-atomic resolution 3D reconstruction, bypassing recombinant artifacts and revealing structural details unattainable by prior approaches (see figure).

Zhou lab’s cryoEM study unveiled DGC’s “keychain-like” architecture. Extracellularly, a β-helix trimer (β-, γ-, δ-sarcoglycans) anchors dystroglycan to the matrix; mutations disrupting its flexible bend cause Limb-Girdle Muscular Dystrophy. Sarcospan stabilizes the transmembrane region via a β-DG-mediated interface, a promising therapeutic target. Cytoplasmically, novel interactions between β-DG, α-/δ-SG, and dystrophin’s ZZ domain, plus a conformationally dynamic WW-α-dystrobrevin interface, drive signaling. Over 110 mutations were mapped, linking structural defects to DMD, Becker, and Limb-Girdle dystrophies, clarifying disease mechanisms.

This STROBE-supported, NSF-funded cryoEM study elucidates DGC’s mechanoprotective role, guiding therapies like gene replacement and small-molecule stabilizers for muscular dystrophies. It showcases cryoEM’s transformative potential for native membrane complexes, with implications for other diseases like cardiomyopathies, advancing precision medicine.

Polymeric organic mixed ionic–electronic conductors (OMIECs) underpin several technologies where their electrochemical properties are desirable. These properties however depend on the microstructure that develops in their aqueous operational environment. In relevant experimental conditions, electrolyte-induced swelling amounts to up 20% in volume. We investigated the structure of a model OMIEC across multiple length-scales using cryogenic four-dimensional scanning transmission electron microscopy (cryo-4D-STEM) in both dry and hydrated states. 4D STEM allows us to identify the prevalent defects in the polymer crystalline regions and to analyze the liquid-crystalline nature of the polymer. The orientation maps of the dry and hydrated polymer show that swelling-induced disorder is mostly localized within discrete regions, thus largely preserving liquid crystalline order. Therefore, the liquid crystalline mesostructure makes electronic transport robust to electrolyte ingress. This study demonstrates that cryo-4D-STEM provides multiscale structural insights into complex, hierarchical structures such as polymeric OMIECs, even in their hydrated operating state.

Computational imaging reconstructions from multiple measurements that are captured sequentially often suffer from motion artifacts if the scene is dynamic. We propose a neural space–time model (NSTM) that jointly estimates the scene and its motion dynamics, without data priors or pre-training. Hence, we can both remove motion artifacts and resolve sample dynamics from the same set of raw measurements used for the conventional reconstruction. We demonstrate NSTM in three computational imaging systems: differential phase-contrast microscopy, three-dimensional structured illumination microscopy and rolling-shutter DiffuserCam. We show that NSTM can recover subcellular motion dynamics and thus reduce the misinterpretation of living systems caused by motion artifacts.

Uterine cervical remodeling is a fundamental feature of pregnancy, facilitating the delivery of the fetus through the cervical canal. Yet, we still know very little about this process due to the lack of methodologies that can quantitatively and unequivocally pinpoint the changes the cervix undergoes during pregnancy. We utilize polarization-resolved second harmonic generation to visualize the alterations the cervix extracellular matrix, specifically collagen, undergoes during pregnancy with exquisite resolution. This technique provides images of the collagen orientation at the pixel level (0.4 μm) over the entire murine cervical section. They show tight and ordered packing of collagen fibers around the os at the early stage of pregnancy and their disruption at the later stages. Furthermore, we utilize a straightforward statistical analysis to demonstrate the loss of order in the tissue, consistent with the loss of mechanical properties associated with this process. This work provides a deeper understanding of the parturition process and could support research into the cause of pathological or premature birth.

At the ultimate scaling limit, electronic memory would store bits of information by shifting atoms back and forth inside individual crystalline unit cells. Ferroelectric materials exhibit the electronic hysteresis required to realize this ideal. However, despite the perfect alignment between a material class and an economically important application, ferroelectric computer memory has almost zero commercial presence.

The materials properties that are possible in principle are not realized in practice. Strain, defects, phase competition, and inhomogenieties all confuse the experimental picture. Previously it has been difficult to look inside a ferroelectric to see what is going on.  The standard high-resolution imaging techniques, piezo-force microscopy (PFM) and transmission electron microscopy (TEM), struggle to visualize the electric and polarization fields that are the hallmarks of ferroelectricity. Understanding why ferroelectric materials have not yet lived up to their potential is widely recognized open problem.

Using scanning TEM (STEM) electron beam-induced current (EBIC) imaging, we map the electric fields in a Hf_0.5Zr_0.5O₂ capacitor, obtaining a view of the material’s ferroelectric properties that is unprecedented in its completeness. We map the whole device and  inside nanoscale domains, correlating global free currents with local polarization reversals. In individual domains we isolate and measure the remanent background E-field that does not switch, and we show that this field determines the coercive E-field required to switch the domain. These measurements connect the nanoscopic crystal structure to the mesoscopic materials properties that ultimately determine device function.

Strong coupling refers to the coherent interaction between quantum state transitions and optical modes. These hybrid states present new possibilities for applications such as single-molecule sensing, single photon emitters, and low-threshold solid state lasers. However, due to the fundamental limitation arising from the weak transition dipole moments in the infrared (IR), reaching the strong coupling regime has been limited to macroscopic ensembles. Recently, multi-quantum-wells (MQW) with quantum engineered electronic states offer a promising route towards mid-IR electronic strong coupling. However, with traditional diffraction-limited mode volumes, even for high Q-factor resonators, the strong coupling of a single quantum emitter has historically necessitated operation at cryogenic temperatures to counteract dissipation.

Here, a STROBE team from CU Boulder with collaborators from Sandia National Lab, the Walter Schottky Institute (Munich), Texas A&M, and Colgate University, achieved control of nonlinear IR light-matter interaction between a single nano-antenna and quantum well intersubband transitions. The team combined broadband synchrotron infrared nano-spectroscopy (SINS) at the ALS-LBL with intense fs/ps-pulsed IR s-SNOM imaging developed at CU Boulder for the dynamic manipulation of the antenna-quantum well hybrid states on the nanoscale. The results demonstrate the potential for localized and dynamic modification of quantum states and excitation pathways as a new regime of coherent and tunable IR electronic strong coupling in open nano-cavity systems, with the perspective of nano-scale sensing and nano-optical control of power limiters or saturable absorbers.

The development of twisted van der Waals (vdW) heterostructures—where layers of 2D materials are stacked with controlled rotation angles—has opened exciting opportunities in quantum technologies. Notably, the twist interfaces in hexagonal boron nitride (h-BN) can undergo structural transformations that support single-photon emission, making them promising for quantum sensing. However, imaging these buried interfaces using scanning transmission electron microscopy (STEM) has been challenging due to poor signal quality and geometric constraints.

In this work, we demonstrated the use of multislice ptychography (MSP), a sensitive coherent diffractive imaging technique, to visualize a twisted h-BN interface from a single-view dataset. STEM experiments were conducted on the TEAM I microscope at the National Center for Electron Microscopy, LBNL, where we acquired diffraction patterns from a 12-nm-thick twisted h-BN sample. Unlike conventional ptychographic approaches that yield a single complex image of the sample, MSP enables depth-sectioning during post-processing, producing a series of reconstructed image slices.

We successfully reconstructed 24 slices of the twisted h-BN heterostructure, resolving the top flake, interface, and bottom flake with a lateral resolution of 0.57 Å. Remarkably, a depth resolution of 2.5 nm was achieved without sample tilting—the highest reported depth resolution at the time of publication. This work highlights MSP’s potential to resolve nanoscale features in three dimensions without requiring tomographic data acquisition, paving the way for advanced quantum materials characterization.