In this thesis, we look at changes in the end-host network stack that are needed to support REACToR, a hybrid data- center network. We discuss the requirements posed by REACToR and propose a host-traffic control protocol. We then describe three different implementations of the protocol. To schedule optical circuits on a microsecond scale, a hybrid top-of-rack switch needs to control end- host traffic. The most important performance characteristic of the host-control protocol is its responsiveness, namely how quickly can a flow be paused or unpaused. The first implementation of the host-control protocol is based on priority-based Ethernet flow control. While it provides good performance by taking advantage of hardware implementations in commodity network cards, the implementation has correctness issues. The second implementation is the Software-defined Data Plane proposal which uses DPDK to provide a flexible user-land network stack. The host-control protocol can be a subset of the Software-defined Data Plane. However since the stack is implemented in software, there are performance penalties. It also requires the original network stack to be replaced, which might entail a lot of changes in end hosts. The third implementation is a reprogrammed network card built with a Netronome network flow processor. Being a hardware implementation, it has better performance and does not demand changes in the operating system. Thus it can help REACToR achieve the goal of operating "under the radar"

In the version of this Article originally published, the author Baran Eren was mistakenly affiliated with the Harbin Institute of Technology, China; it has now been corrected to Materials Science Division, Lawrence Berkeley National Laboratory, Berkeley, CA, USA.

Bimetallic and multi-component catalysts typically exhibit composition-dependent activity and selectivity, and when optimized often outperform single-component catalysts. Here we used ambient-pressure X-ray photoelectron spectroscopy (AP-XPS) and in situ and ex situ transmission electron microscopy (TEM) to elucidate the origin of composition dependence observed in the catalytic activities of monodisperse CoPd bimetallic nanocatalysts for CO oxidation. We found that the catalysis process induced a reconstruction of the catalysts, leaving CoOx on the nanoparticle surface. The synergy between Pd and CoOx coexisting on the surface promotes the catalytic activity of the bimetallic catalysts. This synergistic effect can be optimized by tuning the Co/Pd ratios in the nanoparticle synthesis, and it reaches a maximum at compositions near Co0.24Pd0.76, which achieves complete CO conversion at the lowest temperature. Our combined AP-XPS and TEM studies provide direct observation of the surface evolution of the bimetallic nanoparticles under catalytic conditions and show how this evolution correlates with catalytic properties.

The creation of metal-metal oxide interfaces is an important approach to fine-tuning catalyst properties through strong interfacial interactions. This article presents the work on developing interfaces between Pt and CeOx that improve Pt surface energetics for the hydrogen evolution reaction (HER) within an alkaline electrolyte. The Pt-CeOx interfaces are formed by depositing size-controlled Pt nanoparticles onto a carbon support already coated with ultrathin CeOx nanosheets. This interface structure facilitates substantial electron transfer from Pt to CeOx, resulting in decreased hydrogen binding energies on Pt surfaces, and water dissociation for the HER, as predicted by the density functional theory (DFT) calculations. Electrochemical testing indicates that both Pt specific activity and mass activity are improved by a factor of 2 to 3 following the formation of Pt-CeOx interfaces. This study underscores the significance and potential of harnessing robust interfacial effects to enhance electrocatalytic reactions.

Spinel transition metal oxides are important electrode materials for lithium-ion batteries, whose lithiation undergoes a two-step reaction, whereby intercalation and conversion occur in a sequential manner. These two reactions are known to have distinct reaction dynamics, but it is unclear how their kinetics affects the overall electrochemical response. Here we explore the lithiation of nanosized magnetite by employing a strain-sensitive, bright-field scanning transmission electron microscopy approach. This method allows direct, real-time, high-resolution visualization of how lithiation proceeds along specific reaction pathways. We find that the initial intercalation process follows a two-phase reaction sequence, whereas further lithiation leads to the coexistence of three distinct phases within single nanoparticles, which has not been previously reported to the best of our knowledge. We use phase-field theory to model and describe these non-equilibrium reaction pathways, and to directly correlate the observed phase evolution with the batterys discharge performance.

Colloidal chemistry holds promise to prepare uniform and size-controllable pre-catalysts; however, it remains a challenge to unveil the atomic-level transition from pre-catalysts to active catalytic surfaces under the reaction conditions to enable the mechanistic design of catalysts. Here, we report an ambient-pressure X-ray photoelectron spectroscopy study, coupled with in situ environmental transmission electron microscopy, infrared spectroscopy, and theoretical calculations, to elucidate the surface catalytic sites of colloidal Ni nanoparticles for CO₂ hydrogenation. We show that Ni nanoparticles with phosphine ligands exhibit a distinct surface evolution compared with amine-capped ones, owing to the diffusion of P under oxidative (air) or reductive (CO₂ + H₂) gaseous environments at elevated temperatures. The resulting NiP_x surface leads to a substantially improved selectivity for CO production, in contrast to the metallic Ni, which favors CH₄. The further elimination of surface metallic Ni sites by designing multi-step P incorporation achieves unit selectivity of CO in high-rate CO₂ hydrogenation.

Identification of sequence variants robustly associated with predisposition to diabetic kidney disease (DKD) has the potential to provide insights into the pathophysiological mechanisms responsible. We conducted a genome-wide association study (GWAS) of DKD in type 2 diabetes (T2D) using eight complementary dichotomous and quantitative DKD phenotypes: the principal dichotomous analysis involved 5,717 T2D subjects, 3,345 with DKD. Promising association signals were evaluated in up to 26,827 subjects with T2D (12,710 with DKD). A combined T1D+T2D GWAS was performed using complementary data available for subjects with T1D, which, with replication samples, involved up to 40,340 subjects with diabetes (18,582 with DKD). Analysis of specific DKD phenotypes identified a novel signal near GABRR1 (rs9942471, P = 4.5 × 10^-8) associated with microalbuminuria in European T2D case subjects. However, no replication of this signal was observed in Asian subjects with T2D or in the equivalent T1D analysis. There was only limited support, in this substantially enlarged analysis, for association at previously reported DKD signals, except for those at UMOD and PRKAG2, both associated with estimated glomerular filtration rate. We conclude that, despite challenges in addressing phenotypic heterogeneity, access to increased sample sizes will continue to provide more robust inference regarding risk variant discovery for DKD.