Search

Electrochemical cells employing Sodium (Na) and Zinc (Zn) electrodes and a chloride salt electrolyte have been imaged by neutron radiography during cycling. The use of such abundant raw materials confers a very low energy-normalised cost to the Na-Zn system, but its operation requires them to be entirely molten, and therefore to be operated at 600 {\deg}C. To suppress the self-discharge that results from this all-molten configuration, porous ceramic diaphragms are used to partition the electrolyte and thereby impede the movement of the Zn2+ ions responsible towards the Na electrode. Neutron images reveal large gas bubbles trapped beneath these diaphragms, formed during the cell fabrication process due to the large volume change that accompanies melting/solidifying of the electrolyte. Cycling data confirm that these bubbles interfere with cell operation by substantially increasing ohmic resistance. They indicate the need for either a new diaphragm design, or a cell fabrication process that prevents their formation in the first instance., Comment: 8 pages, 3 figures

Analog compute-in-memory (CIM) in static random-access memory (SRAM) is promising for accelerating deep learning inference by circumventing the memory wall and exploiting ultra-efficient analog low-precision arithmetic. Latest analog CIM designs attempt bit-parallel schemes for multi-bit analog Matrix-Vector Multiplication (MVM), aiming at higher energy efficiency, throughput, and training simplicity and robustness over conventional bit-serial methods that digitally shift-and-add multiple partial analog computing results. However, bit-parallel operations require more complex analog computations and become more sensitive to well-known analog CIM challenges, including large cell areas, inefficient and inaccurate multi-bit analog operations, and vulnerability to PVT variations. This paper presents PICO-RAM, a PVT-insensitive and compact CIM SRAM macro with charge-domain bit-parallel computation. It adopts a multi-bit thin-cell Multiply-Accumulate (MAC) unit that shares the same transistor layout as the most compact 6T SRAM cell. All analog computing modules, including digital-to-analog converters (DACs), MAC units, analog shift-and-add, and analog-to-digital converters (ADCs) reuse one set of local capacitors inside the array, performing in-situ computation to save area and enhance accuracy. A compact 8.5-bit dual-threshold time-domain ADC power gates the main path most of the time, leading to a significant energy reduction. Our 65-nm prototype achieves the highest weight storage density of 559 Kb/mm${^2}$ and exceptional robustness to temperature and voltage variations (-40 to 105 $^{\circ}$C and 0.65 to 1.2 V) among SRAM-based analog CIM designs., Comment: This manuscript has been accepted to IEEE Journal of Solid-State Circuits (JSSC)

Neural representations have shown spectacular ability to compress complex signals in a fraction of the raw data size. In 3D computer graphics, the bulk of a scene's memory usage is due to polygons and textures, making them ideal candidates for neural compression. Here, the main challenge lies in finding good trade-offs between efficient compression and cheap inference while minimizing training time. In the context of rendering, we adopt a ray-centric approach to this problem and devise N-BVH, a neural compression architecture designed to answer arbitrary ray queries in 3D. Our compact model is learned from the input geometry and substituted for it whenever a ray intersection is queried by a path-tracing engine. While prior neural compression methods have focused on point queries, ours proposes neural ray queries that integrate seamlessly into standard ray-tracing pipelines. At the core of our method, we employ an adaptive BVH-driven probing scheme to optimize the parameters of a multi-resolution hash grid, focusing its neural capacity on the sparse 3D occupancy swept by the original surfaces. As a result, our N-BVH can serve accurate ray queries from a representation that is more than an order of magnitude more compact, providing faithful approximations of visibility, depth, and appearance attributes. The flexibility of our method allows us to combine and overlap neural and non-neural entities within the same 3D scene and extends to appearance level of detail., Comment: 10 pages

Yamdb (Yet another materials data base) addresses the need to provide thermophysical properties of liquid metals and molten salts in an easily accessible manner. Mathematical relations describing material properties - usually determined by experiment - are taken from the literature. Equations and their coefficients are stored separately. The former can be implemented in any programming language (Python and Go in this case) and the latter are kept in YAML files together with additional information (source, temperature range, composition, accuracy if available, etc)., Comment: 22 pages, 5 figures

We propose a self-consistent explanation of Rieger-type periodicities, the Schwabe cycle, and the Suess-de Vries cycle of the solar dynamo in terms of resonances of various wave phenomena with gravitational forces exerted by the orbiting planets. Starting on the high-frequency side, we show that the two-planet spring tides of Venus, Earth and Jupiter are able to excite magneto-Rossby waves which can be linked with typical Rieger-type periods. We argue then that the 11.07-year beat period of those magneto-Rossby waves synchronizes an underlying conventional $\alpha-\Omega$-dynamo, by periodically changing either the field storage capacity in the tachocline or some portion of the $\alpha$-effect therein. We also strengthen the argument that the Suess-de Vries cycle appears as an 193-year beat period between the 22.14-year Hale cycle and a spin-orbit coupling effect related with the 19.86-year rosette-like motion of the Sun around the barycenter., Comment: 32 pages, 13 figures

Motivation: Microsatellite instability (MSI) is a cancer biomarker associated with cancer prognosis and chemotherapy sensitivity. Since the discovery of MSI, polymerase chain reaction (PCR)-based testing has been considered the gold standard for MSI detection. However, with the decrease in sequencing costs, software that calculates MSI based on next-generation sequencing (NGS) data has been widely applied. Results: In this study, we evaluated the performance of the MSIsensor detection software, focusing on the limitations of the chi-square test algorithm in determining microsatellite stability under high-depth sequencing data. We demonstrated that the chi-square test algorithm is insufficient for accurately as-sessing microsatellite stability in this context. Furthermore, we explored the application of downsampling techniques to enhance the accuracy of MSIsensor detection. Our findings provide insight into the limita-tions of current methods and offer potential improvements for more reliable MSI detection based on NGS data.

Context: The old question of whether the solar dynamo is synchronized by the tidal forces of the orbiting planets has recently received renewed interest, both from the viewpoint of historical data analysis and in terms of theoretical and numerical modelling. Aims: We aim to contribute to the solution of this longstanding puzzle by analyzing cosmogenic radionuclide data from the last millennium. Methods: We reconsider a recent time-series of $^{14}$C-inferred sunspot data and compare the resulting cycle minima and maxima with the corresponding conventional series down to 1610 A.D., enhanced by Schove's data before that time. Results: We find that, despite recent claims to the contrary, the $^{14}$C-inferred sunspot data are well compatible with a synchronized solar dynamo, exhibiting a relatively phase-stable period of 11.07 years, which points to a synchronizing role of the spring tides of the Venus-Earth-Jupiter system., Comment: 8 pages, 7 figures

In this work a 1D finite volume based model using coupled meshes is introduced to capture potential and species distribution throughout the discharge process in a lithium bismuth liquid metal battery while neglecting hydrodynamic effects, focusing on the electrochemical properties of the cell and the mass transport in electrolyte and cathode. Interface reactions in the electrical double layer are considered through the introduction of a discrete jump of the potential modelled as periodic boundary condition to resolve interfacial discontinuities in the cell potential. A balanced-force like approach is implemented to ensure consistent calculation at the interface level. It is found that mass transport and concentration gradients have a significant effect on the cell overpotentials and thus on cell performance and cell voltage. By quantifying overvoltages in the Li Bi cell with a mixed cation electrolyte, it is possible to show that diffusion and migration current density could have counteractive effects on the cell voltage. Furthermore, the simulated limiting current density is observed to be much lower than experimentally measured, which can be attributed to convective effects in the electrolyte that need to be addressed in future simulations. The solver is based on the open source library OpenFOAM and thoroughly verified against the equivalent system COMSOL multiphysics and further validated with experimental results. It is openly available at https://doi.org/10.14278/rodare.2313.

For longer than one decade, liquid metal batteries (LMBs) are developed with the primary aim to provide economic stationary energy storage. Featuring two liquid metal electrodes separated by a molten salt electrolyte, LMBs operate at elevated temperature as simple concentration cells. Therefore, efficient mass transfer is a basic prerequisite for their economic operation. Understanding these mechanisms cannot be limited at the single layer level. With this motivation, the effects of solutal- and thermally-driven flow are studied, as well as the flow coupling between the three liquid layers of the cell. It is shown that solutal convection appears first and thermal convection much later. While the presence of solutal flow depends on the mode of operation (charge or discharge), the occurrence of thermal convection is dictated by the geometry (thickness of layers). The coupling of the flow phenomena between the layers is intriguing: while thermal convection is confined to its area of origin, i.e. the electrolyte, solutal convection is able to drive flow in the positive electrode and in the electrolyte., Comment: preprint

This study is a continuation of the combined experimental and numerical investigation [1] of the flow of the eutectic GaInSn alloy inside a cylindrical vessel exposed to a constant electrical current. The emerging electrovortex flow driven by the interaction of the current, which is applied through a tapered electrode, with its own magnetic field might have both detrimental and advantageous effects in liquid metal batteries. While the former work [1] was mainly concerned with time-averaged results, this paper focuses on the transient behaviour of the electrovortex flow which becomes most relevant under the influence of an external axial field. The additional Lorentz force generated by the interaction of the imposed current with the vertical component of the geomagnetic field bz drives the ordinary electrovortex jet flow into a swirling motion. Velocity distributions and motion characteristics, such as spiral streamlines and shortened and irregularly swinging jet regions, are investigated. The mechanism is analysed in detail for bz = 25.5 {\mu}T. The maximum angular velocity of the rotating jet is basically linearly dependent on bz, at least for the values studied here. A good agreement between the transient simulation and the experimental result is shown.

Lithium-bismuth bimetallic cells are amongst the best explored liquid metal batteries. A simple and fast quasi-one-dimensional cell voltage model for such devices is presented. The equilibrium cell potential is obtained from a complex two-dimensional fit of data drawn from multiple studies of equilibrium cell potential and rendered congruent with the phase diagram. Likewise, several analytical and fit functions for the ohmic potential drop across the electrolyte are provided for different battery geometries. Mass transport overpotentials originating from the alloying of Li into Bi are modelled by solving a diffusion equation, either analytically or numerically, and accounting for the volume change of the positive electrode. The applicability and limitations of the model are finally illustrated in three distinct experimental settings.

We study numerically localised short circuits in Li||Bi liquid metal batteries. In the prototype of a classical, three liquid-layer system, we assume a perceptible local deformation of the Li-salt interface. We find that there always exists a critical current at which a Li-droplet is cut off from this hump, and transferred to the Bi-phase. In a second case, we assume that the molten Li is contained in a metal foam, and that a small Li-droplet emerges below this foam due to insufficient wetting. This droplet is deformed by Lorentz forces, until eventually being pinched off. Here, the critical current is slightly lower than in the three-layer system, and both, a droplet transfer and complete short circuits are observed. Finally, we discuss the relevance of our simulations for experimentally observed short circuits and non-faradaic Li-transfer.

Realizing today's cloud-level artificial intelligence functionalities directly on devices distributed at the edge of the internet calls for edge hardware capable of processing multiple modalities of sensory data (e.g. video, audio) at unprecedented energy-efficiency. AI hardware architectures today cannot meet the demand due to a fundamental "memory wall": data movement between separate compute and memory units consumes large energy and incurs long latency. Resistive random-access memory (RRAM) based compute-in-memory (CIM) architectures promise to bring orders of magnitude energy-efficiency improvement by performing computation directly within memory. However, conventional approaches to CIM hardware design limit its functional flexibility necessary for processing diverse AI workloads, and must overcome hardware imperfections that degrade inference accuracy. Such trade-offs between efficiency, versatility and accuracy cannot be addressed by isolated improvements on any single level of the design. By co-optimizing across all hierarchies of the design from algorithms and architecture to circuits and devices, we present NeuRRAM - the first multimodal edge AI chip using RRAM CIM to simultaneously deliver a high degree of versatility for diverse model architectures, record energy-efficiency $5\times$ - $8\times$ better than prior art across various computational bit-precisions, and inference accuracy comparable to software models with 4-bit weights on all measured standard AI benchmarks including accuracy of 99.0% on MNIST and 85.7% on CIFAR-10 image classification, 84.7% accuracy on Google speech command recognition, and a 70% reduction in image reconstruction error on a Bayesian image recovery task. This work paves a way towards building highly efficient and reconfigurable edge AI hardware platforms for the more demanding and heterogeneous AI applications of the future., Comment: 34 pages, 14 figures, 1 table

Fluid flows in liquid metal batteries can be generated by a number of effects. We start with a short overview of different driving mechanisms and then address questions specific to the metal pad role instabilities in three-layer systems. We focus on the role of the conductivity distribution in the cell, noting at the same time that interfacial tension should be considered as well for smaller cells. Following this discussion, numerical results on the excitation of interfacial waves in two-layer liquid metal systems with miscibility gaps bearing an interface normal electric current are presented. Confirming recent results from the literature, we find that magnetic damping plays a decisive role for strong vertical magnetic fields. In addition, boundary conditions for the electric field strongly influence critical currents and growth rates.

Liquid metal electrodes are one of the key components of different electrical energy storage technologies. The understanding of transport phenomena in liquid electrodes is mandatory in order to ensure efficient operation. In the present study we focus our attention on the positive electrode of the Li||Bi liquid metal battery. Starting from a real experimental setup, we numerically investigate the charge transfer in a molten salt electrolyte and the mass transport in the positive electrode. The two phenomena are tightly coupled, because the current distribution influences the concentration field in the positive electrode. The cell is studied during charging when compositional convection becomes apparent. First results of compositional convection from a non-uniform current distribution are presented, highlighting its capability to affect the flow in the positive electrode and the cell performance.