By combining optical and electron imaging to study a plasmonic photocatalytic reaction, researchers at Stanford University have been able to achieve sub-2-nm resolution for observing a Pd-H phase change reaction in situ and in real time. This work is a first step in an effort to visualize photocatalytic chemistry while it is happening, which would be a major advance in this field.

While other research groups exploring plasmonic photocatalytic reactions have achieved resolutions as low as 15 nm, their work generally involved post-reaction transmission- and scanning electron microscope (TEM/SEM) analysis or the use of external markers such as fluorescent molecules. The Stanford group, led by Jennifer Dionne, Associate Professor of Materials Science and Engineering, performed the reaction in a custom-designed cryo-cathodoluminescence (CL) holder in an environmental TEM (ETEM) capable of both visible light and electron beam irradiation. 

“The combination of the spatial and temporal resolution—seeing the reaction as it happens and not in a post-reaction analysis step—is the most exciting result of this work,” says post-doctoral researcher Michal Vadai, the first author of the article published in Nature Communications.

Dionne, Vadai, and colleagues worked with Gatan, Inc., to design the CL holder, which has two parabolic mirrors that surround the sample and fit into the 5-mm pole piece gap of the microscope. The holder has a 500-μm wide aperture to allow the primary electron beam to pass while light is focused to/from the sample, with two optical fibers to provide and collect light. Light induces the formation of plasmons in the catalytic system, which have major effects on the reaction.

As a first test of this new setup, the researchers chose to look at a reaction they were already familiar with: the phase change from hydrogen-rich β-Pd to hydrogen-poor α-Pd by desorption of hydrogen. Years of experience with this reaction “in the dark”—that is, without photo-illumination—gave the team an excellent baseline for comparison with the photocatalytic reaction. It was also easy to follow the phase transformation visually in the ETEM’s imaging mode because the phases have different imaging contrast. Selected area electron diffraction (SAED) and electron energy loss spectroscopy (EELS) analyses also provided valuable data.

The catalyst comprised Au nanodiscs with diameters of 100-120 nm and single-crystal Pd nanocubes  with edge lengths of 30-50 nm on an Si₃N₄ membrane to form antenna-reactor pairs. The plasmonic Au antenna harvests light to power the reaction, while the Pd nanocube acts as the reactor where the catalytic reaction—in this case hydrogen desorption—takes place. When the β-phase reached equilibrium, the researchers illuminated the sample with pulses of unpolarized laser light in the visible range (550-750 nm). The Pd nanocubes began to desorb hydrogen, transitioning to the α phase, with lattice contraction monitored by SAED. The β and α phases coexisted for a while until only the α phase remained.

A major finding of this research is that there are two different reaction steps each of which responds to light differently. The “induction” step involves the desorption of hydrogen atoms from the surface of the Pd cube. The “reaction” step begins with the formation of the new α phase and involves the continuous diffusion of hydrogen atoms from inside the palladium lattice. As the illumination wavelength was increased from 550 nm to 800 nm, the reaction time remained constant while the induction time changed with wavelength, so the excitation of plasmons affects the rate of each step differently. Hydrogen desorption from the Pd surface has a higher activation energy than hydrogen diffusion to the surface and lattice rearrangement. The researchers suggest that the formation of hybrid Pd-H molecular orbitals reduces activation barriers in the induction step when plasmons are excited in the system.

“In photocatalyst optimization, it is important to know whether there are different reaction steps and if light is affecting just one step or multiple processes,” Dionne says. “Here, we have been able to deconvolve the effects of optical excitation on distinct steps of the reaction.”

This work also revealed that plasmons enable site-selectivity through localized strong electromagnetic fields known as “hot spots” that form in the space between the antenna and reactor. By investigating different orientations of Pd nanocube corners to hot spots, the researchers showed that corners closest to the hot spot were much more likely to nucleate the α phase than corners that were far away. 

Future research will be along two parallel paths: to expand the system to handle any molecular reaction in the gas phase, and to introduce liquid phase reactions.  Obviously, this will require major enhancements in the CL holder and other aspects of this ETEM process. But the Stanford researchers are optimistic.

“Now that we have this unique capability,” Dionne says, “there is a clear path forward to analyzing a multitude of industrially relevant reactions.”

“In this paper, the authors have looked into the plasmon-controlled photocatalytic activity of Pd nanoparticles to tune light absorption for site-selective and product-selective photocatalysts,” says Wei Luo of the Division of Materials Theory in the Physics and Astronomy Department of Uppsala University, Sweden, who was not involved in this research. “This work will open a new path for tailor-made photocatalysts.”

Read the article in Nature Communications.