Researchers from the University of Illinois Urbana-Champaign (UIUC) and Northwestern University have directly imaged nanoparticle crystallization at the single-particle level, in real time. As reported in Nature Materials, their work yields new details on the pathways to crystallization. More generally, the method offers a framework for imaging dynamic nanoscale phenomena in liquid environments at high resolution.

The process by which a crystal nucleates and grows impacts many of its properties. Understanding the parameters that govern this behavior could have significant implications for materials design. However, imaging this process at the nanoscale in real time is technologically challenging.

Cryogenic electron microscopy, currently one of the most promising nanoscale imaging methods, requires samples to be immobilized in amorphous ice, but crystallization and other nanoscale phase behaviors often occur in liquid suspensions. Another prominent method, super-resolution optical microscopy, can image liquid samples but cannot resolve the dynamics sufficiently in time. Transmission electron microscopy (TEM) is generally incompatible with liquid samples because it requires the sample to be stable in high vacuum.

In this new research, a collaboration led by Qian Chen at UIUC and Erik Luijten at Northwestern utilized a technique known as liquid-phase TEM. The technique “makes liquids compatible with the nanometer resolving power of TEM,” Chen says. This is usually done by containing the liquid in a protective cell that has a “window” transparent to the electron beam. Capturing nanoscale crystallization with liquid-phase TEM required pushing the limits of the technology, in part by minimizing beam artefacts in the data—an effort Chen has been working on for years.

As a model nanoparticle system, the research team chose gold nanoprisms 100 nm per side and 7.5 nm thick, suspended in a buffer solution. The prisms were coated with negatively charged ligands to ensure that the nanoparticles were evenly distributed and to provide a mechanism for inducing crystallization.

In order to image the system with TEM, a thin layer of solution was loaded into a closed cell formed by two 50-nm-thick microchips and featuring a silicon nitride membrane window.

The researchers placed the cell inside a conventional TEM and imaged the sample in real time with a low-dose-rate electron beam (3.7–8.9 e^– Å^–2 s^–1). The results showed that when the beam turned on, nanoprisms in the illuminated region began to self-arrange into stacked vertical columns. Over time, this behavior spread to regions farther and farther from the beam, eventually resulting in a hexagonal lattice of evenly spaced dark, circular discs as seen by TEM.

Crystallization was initiated by the imaging beam, says Chen. The electron dose rate was not high enough to directly influence nanoparticle interactions, but it was high enough to dissociate the water molecules in its path. This led to a rapid increase in the ionic strength of the illuminated solution, and therefore a change in the electrostatic forces acting on local nanoprisms. This change triggered self-assembly.

Using liquid-phase TEM, the researchers were able to capture real-time movies of the crystallization process. They analyzed the structural development of more than 110,000 columns by applying single-particle tracking techniques to these movies, which revealed three distinct stages in the crystallization pathway. Initially, columns were sparse and characterized by high orientational disorder—the prisms in each stack were angularly misaligned. In the second stage, the density of the columns increased and the columns arranged into an amorphous and positionally disordered structure. In the third stage, the columns rearranged into a stable crystal. The prisms in each column remained angularly misaligned, causing the formation of a hexagonal rather than a triangular crystal.

The intermediate stage was a surprising observation. This stage is not predicted by classical nucleation theory; instead it is characteristic of a two-step crystallization process. A statistical-mechanical analysis by Luijten revealed that the orientational disorder stemmed from competition between the forces governing the behavior of the nanoprisms, namely electrostatic repulsion and van der Waals attraction. Monte Carlo simulations by Luijten further supported this finding, suggesting that the orientational disorder facilitated positional order and crystallization.

“This plastic mesophase, with simultaneous orientational randomness and positional ordering, is akin to structures common in molecular solids, yet arises here without the conventional conditions of high axial symmetry and long-range repulsion,” write the authors. Refining the theories of pathways to nucleation and growth is an important step in harnessing the crystallization processes to produce materials with desired properties.

According to Chen, the framework for imaging, tracking, and analyzing nanoscale crystallization demonstrated in this research can be a roadmap for exploring dynamic nanoscale phenomena like melting, protein transformations, and glass transitions. She would like to see the community engage in “a full charting of the ‘elusive’ nanoscale dynamics, from equilibrium to non-equilibrium, from synthetic systems to biological cellular machineries,” now that the high-resolution imaging of such samples has become accessible.

“This is an exciting study at the forefront of nanoparticle assemblies,” says Nicholas Kotov, an expert in nanoparticle self-assembly at the University of Michigan who was not involved in the study. “Its strength is visualization and modeling nanoparticle assembly dynamics. What was before available as snapshots of intermediate states, one can observe now as a continuous process.” This development makes it possible for researchers to identify new intermediate states, according to Kotov.

Read the abstract in Nature Materials.