Search

Antiferromagnetic insulators are a ubiquitous class of magnetic materials, holding the promise of low-dissipation spin-based computing devices that can display ultra-fast switching and are robust against stray fields. However, their imperviousness to magnetic fields also makes them difficult to control in a reversible and scalable manner. Here we demonstrate a novel proof-of-principle ionic approach to control the spin reorientation (Morin) transition reversibly in the common antiferromagnetic insulator α-Fe2O3 (haematite) – now an emerging spintronic material that hosts topological antiferromagnetic spin-textures and long magnon-diffusion lengths. We use a low-temperature catalytic-spillover process involving the post-growth incorporation or removal of hydrogen from α-Fe2O3 thin films. Hydrogenation drives pronounced changes in its magnetic anisotropy, Néel vector orientation and canted magnetism via electron injection and local distortions. We explain these effects with a detailed magnetic anisotropy model and first-principles calculations. Tailoring our work for future applications, we demonstrate reversible control of the room-temperature spin-state by doping/expelling hydrogen in Rh-substituted α-Fe2O3.

Federated learning enables training a global model from data located at the client nodes, without data sharing and moving client data to a centralized server. Performance of federated learning in a multi-access edge computing (MEC) network suffers from slow convergence due to heterogeneity and stochastic fluctuations in compute power and communication link qualities across clients. We propose a novel coded computing framework, CodedFedL, that injects structured coding redundancy into federated learning for mitigating stragglers and speeding up the training procedure. CodedFedL enables coded computing for non-linear federated learning by efficiently exploiting distributed kernel embedding via random Fourier features that transforms the training task into computationally favourable distributed linear regression. Furthermore, clients generate local parity datasets by coding over their local datasets, while the server combines them to obtain the global parity dataset. Gradient from the global parity dataset compensates for straggling gradients during training, and thereby speeds up convergence. For minimizing the epoch deadline time at the MEC server, we provide a tractable approach for finding the amount of coding redundancy and the number of local data points that a client processes during training, by exploiting the statistical properties of compute as well as communication delays. We also characterize the leakage in data privacy when clients share their local parity datasets with the server. We analyze the convergence rate and iteration complexity of CodedFedL under simplifying assumptions, by treating CodedFedL as a stochastic gradient descent algorithm. Furthermore, we conduct numerical experiments using practical network parameters and benchmark datasets, where CodedFedL speeds up the overall training time by up to $15\times$ in comparison to the benchmark schemes., Final version to appear in the first issue of the IEEE JSAC Series on Machine Learning for Communications and Networks

Phase change materials provide unique reconfigurable properties for photonic applications that mainly arise from their exotic characteristic to reversibly switch between the amorphous and crystalline nonvolatile phases. Optical pulse based reversible switching of nonvolatile phases is exploited in various nanophotonic devices. However, large area reversible switching is extremely challenging and has hindered its translation into a technologically significant terahertz spectral domain. Here, this limitation is circumvented by exploiting the semiconducting nature of germanium antimony telluride (GST) to achieve dynamic terahertz control at picosecond timescales. It is also shown that the ultrafast response can be actively altered by changing the crystallographic phase of GST. The ease of fabrication of phase change materials allows for the realization of a variable ultrafast terahertz modulator on a flexible platform. The rich properties of phase change materials combined with the diverse functionalities of metamaterials and all-optical ultrafast control enables an ideal platform for design of efficient terahertz communication devices, terahertz neuromorphic photonics, and smart sensor systems. Agency for Science, Technology and Research (A*STAR) Ministry of Education (MOE) National Research Foundation (NRF) The authors acknowledge research funding support from the Ministry of Education, Singapore (AcRF Tier 1, Grant RG191/17, MOE2017-T2-1-110, MOE2019-T2-1-058, and MOE2016-T3-1-006(S)), Advanced Manufacturing and Engineering (AME) Programmatic grant (A18A5b0056) from Agency for Science, Technology and Research (A*STAR) and NRF CRP on Oxide Electronics on silicon Beyond Moore (NRF-CRP15-2015-01).

In the quest for post-CMOS technologies, ferromagnetic skyrmions and their anti-particles have shown great promise as topologically protected solitonic information carriers in memory-in-logic or neuromorphic devices. However, the presence of dipolar fields in ferromagnets, restricting the formation of ultra-small topological textures, and the deleterious skyrmion Hall effect when driven by spin torques have thus far inhibited their practical implementations. Antiferromagnetic analogues, which are predicted to demonstrate relativistic dynamics, fast deflection-free motion and size scaling have recently come into intense focus, but their experimental realizations in natural antiferromagnetic systems are yet to emerge. Here, we demonstrate a family of topological antiferromagnetic spin-textures in $\alpha$-Fe$_2$O$_3$ - an earth-abundant oxide insulator - capped with a Pt over-layer. By exploiting a first-order analogue of the Kibble-Zurek mechanism, we stabilize exotic merons-antimerons (half-skyrmions), and bimerons, which can be erased by magnetic fields and re-generated by temperature cycling. These structures have characteristic sizes of the order ~100 nm that can be chemically controlled via precise tuning of the exchange and anisotropy, with pathways to further scaling. Driven by current-based spin torques from the heavy-metal over-layer, some of these AFM textures could emerge as prime candidates for low-energy antiferromagnetic spintronics at room temperature., Comment: 18 pages, 4 figures

Antiferromagnetic insulators are a ubiquitous class of magnetic materials, holding the promise of low-dissipation spin-based computing devices that can display ultra-fast switching and are robust against stray fields. However, their imperviousness to magnetic fields also makes them difficult to control in a reversible and scalable manner. Here we demonstrate a novel proof-of-principle ionic approach to control the spin reorientation (Morin) transition reversibly in the common antiferromagnetic insulator α-Fe2O3 (haematite) – now an emerging spintronic material that hosts topological antiferromagnetic spin-textures and long magnon-diffusion lengths. We use a low-temperature catalytic-spillover process involving the post-growth incorporation or removal of hydrogen from α-Fe2O3 thin films. Hydrogenation drives pronounced changes in its magnetic anisotropy, Néel vector orientation and canted magnetism via electron injection and local distortions. We explain these effects with a detailed magnetic anisotropy model and first-principles calculations. Tailoring our work for future applications, we demonstrate reversible control of the room-temperature spin-state by doping/expelling hydrogen in Rh-substituted α-Fe2O3., One major challenge for antiferromagnetic spintronics is how to control the antiferromagnetic state. Here Jani et al. demonstrate the reversible ionic control of the room-temperature magnetic anisotropy and spin reorientation transition in haematite, via the incorporation and removal of hydrogen.

Using interlayer interaction to control functional heterostructures with atomic-scale designs has become one of the most effective interface-engineering strategies nowadays. Here, we demonstrate the effect of a crystalline LaFeO3 buffer layer on amorphous and crystalline LaAlO3/SrTiO3 heterostructures. The LaFeO3 buffer layer acts as an energetically favored electron acceptor in both LaAlO3/SrTiO3 systems, resulting in modulation of interfacial carrier density and hence metal-to-insulator transition. For amorphous and crystalline LaAlO3/SrTiO3 heterostructures, the metal-to-insulator transition is found when the LaFeO3 layer thickness crosses 3 and 6 unit cells, respectively. Such different critical LaFeO3 thicknesses are explained in terms of distinct characteristic lengths of the redox-reaction-mediated and polar-catastrophe-dominated charge transfer, controlled by the interfacial atomic contact and Thomas-Fermi screening effect, respectively. Our results not only shed light on the complex interlayer charge transfer across oxide heterostructures but also provides a new route to precisely tailor the charge-transfer process at a functional interface., 40 Pages, 14 Figures, 1 Table

Federated learning is a method of training a global model from decentralized data distributed across client devices. Here, model parameters are computed locally by each client device and exchanged with a central server, which aggregates the local models for a global view, without requiring sharing of training data. The convergence performance of federated learning is severely impacted in heterogeneous computing platforms such as those at the wireless edge, where straggling computations and communication links can significantly limit timely model parameter updates. This paper develops a novel coded computing technique for federated learning to mitigate the impact of stragglers. In the proposed Coded Federated Learning (CFL) scheme, each client device privately generates parity training data and shares it with the central server only once at the start of the training phase. The central server can then preemptively perform redundant gradient computations on the composite parity data to compensate for the erased or delayed parameter updates. Our results show that CFL allows the global model to converge nearly four times faster when compared to an uncoded approach, Presented at the Wireless Edge Intelligence Workshop, IEEE GLOBECOM 2019

Spatiotemporal manipulation of electromagnetic waves has recently enabled a plethora of exotic optical functionalities, such as non-reciprocity, dynamic wavefront control, unidirectional transmission, linear frequency conversion, and electromagnetic Doppler cloak. Here, an additional dimension is introduced for advanced manipulation of terahertz waves in the space-time, and frequency domains through integration of spatially reconfigurable microelectromechanical systems and photoresponsive material into metamaterials. A large and continuous frequency agility is achieved through movable microcantilevers. The ultrafast resonance modulation occurs upon photoexcitation of ion-irradiated silicon substrate that hosts the microcantilever metamaterial. The fabricated metamaterial switches in 400 ps and provides large spectral tunability of 250 GHz with 100% resonance modulation at each frequency. The integration of perfectly complementing technologies of microelectromechanical systems, femtosecond optical control and ion-irradiated silicon provides unprecedented concurrent control over space, time, and frequency response of metamaterial for designing frequency-agile spatiotemporal modulators, active beamforming, and low-power frequency converters for the next generation terahertz wireless communications. National Research Foundation (NRF) The authors acknowledge the research funding support from National Research Foundation (NRF) Singapore and Agence Nationale de la Recherche (ANR), France- NRF2016- ANR004 (M4197003), NRF CRP on Oxide Electronics on silicon Beyond Moore (NRF-CRP15-2015-01) and Advanced Manufacturing and Engineering (AME) Programmatic grant (A18A5b0056) from Agency for Science, Technology and Research (A*STAR).

Herein, we report a systematic study of water contact angle (WCA) of rare-earth oxide thin-films. These ultra-smooth and epitaxial thin-films were grown using pulsed laser deposition and then characterized using X-Ray diffraction (XRD), Rutherford backscattering spectroscopy (RBS), and atomic force microscopy (AFM). Through both the traditional sessile drop and the novel f–d method, we found that the films were intrinsically hydrophilic (WCA < 10°) just after being removed from the growth chamber, but their WCAs evolved with an exposure to the atmosphere with time to reach their eventual saturation values near 90° (but always stay ‘technically’ hydrophilic). X-Ray photoelectron spectroscopy analysis was used to further investigate qualitatively the nature of hydrocarbon contamination on the freshly prepared as well as the environmentally exposed REO thin-film samples as a function of the exposure time after they were removed from the deposition chamber. A clear correlation between the carbon coverage of the surface and the increase in WCA was observed for all of the rare-earth films, indicating the extrinsic nature of the surface wetting properties of these films and having no relation to the electronic configuration of the rare-earth atoms as proposed by Azimi et al.

In wireless mobile edge computing platforms, such as those supported by vehicular networks, location specific machine learning models can be trained by distributing computations using resources available at the edge. The computational and communication resources at the wireless edge are heterogeneous and unreliable, which can lead to straggler effects that significantly slow down recursive learning tasks, such as gradient descent. In this paper we propose a coded computation framework, which utilizes statistical knowledge of resource heterogeneity to determine optimal encoding and load balancing of training data using Random Linear codes, while avoiding an explicit step for decoding gradients. Results show that the proposed coding framework speeds up the training time for linear regression models, by up to ten times, when compared with repetition coding schemes.

The high demand for computational and storage resources severely impedes the deployment of deep convolutional neural networks (CNNs) in limited resource devices. Recent CNN architectures have proposed reduced complexity versions (e.g,. SuffleNet and MobileNet) but at the cost of modest decreases in accuracy. This paper proposes pSConv, a pre-defined sparse 2D kernel based convolution, which promises significant improvements in the trade-off between complexity and accuracy for both CNN training and inference. To explore the potential of this approach, we have experimented with two widely accepted datasets, CIFAR-10 and Tiny ImageNet, in sparse variants of both the ResNet18 and VGG16 architectures. Our approach shows a parameter count reduction of up to 4.24× with modest degradation in classification accuracy relative to that of standard CNNs. Our approach outperforms a popular variant of ShuffleNet using a variant of ResNet18 with pSConv having 3 × 3 kernels with only four of nine elements not fixed at zero. In particular, the parameter count is reduced by 1.7× for CIFAR-10 and 2.29× for Tiny ImageNet with an increased accuracy of ∼ 4%.

Scaling up distributed machine learning systems face two major bottlenecks – delays due to stragglers and limited communication bandwidth. Recently, a number of coding theoretic strategies have been proposed for mitigating these bottlenecks. In particular, the Gradient Coding (GC) scheme was proposed to speed up distributed gradient descent algorithm in a synchronous master-worker setting by providing robustness to stragglers. A major drawback of the master-worker architecture for distributed learning is however, the bandwidth contention at the master, which can significantly deteriorate the performance as the cluster size increases. In this paper, we propose a new framework named Tree Gradient Coding (TGC) for distributed gradient aggregation, which parallelizes communication over a tree topology while providing straggler robustness. As our main contribution, we characterize the minimum computation load for TGC for a given tree topology and straggler resiliency, and design a tree gradient coding algorithm that achieves this optimal computation load. Furthermore, we provide results from experiments over Amazon EC2, where TGC speeds up the training time by up to 18.8× in comparison to GC.

The reduced electrical screening in 2D materials provides an ideal platform for realization of exotic quasiparticles, that are robust and whose functionalities can be exploited for future electronic, optoelectronic, and valleytronic applications. Recent examples include an interlayer exciton, where an electron from one layer binds with a hole from another, and a Holstein polaron, formed by an electron dressed by a sea of phonons. Here, a new quasiparticle is reported, "polaronic trion" in a heterostructure of MoS

The strikingly contrasting optical properties of various phases of chalcogenide phase change materials (PCM) has recently led to the development of novel photonic devices such as all-optical non-von Neumann memory, nanopixel displays, color rendering, and reconfigurable nanoplasmonics. However, the exploration of chalcogenide photonics is currently limited to optical and infrared frequencies. Here, a phase change material integrated terahertz metamaterial for multilevel nonvolatile resonance switching with spatial and temporal selectivity is demonstrated. By controlling the crystalline proportion of the PCM film, multilevel, non-volatile, terahertz resonance switching states with long retention time at zero hold power are realized. Spatially selective reconfiguration at sub-metamaterial scale is shown by delivering electrical stimulus locally through designer interconnect architecture. The PCM metamaterial also features ultrafast optical modulation of terahertz resonances with tunable switching speed based on the crystalline order of the PCM film. The multilevel nonvolatile, spatially selective, and temporally tunable PCM metamaterial will provide a pathway toward development of novel and disruptive terahertz technologies including spatio-temporal terahertz modulators for high speed wireless communication, neuromorphic photonics, and machine-learning metamaterials.

Recent research shows that video recordings of the user’s hand movement and his or her smartphone screen display can be used to steal sensitive information such as pins and passwords. The methods presented in the past assume the victim to be present in a well illuminated place. In this paper, we present a novel attack on the smartphone users’ pins that does not require a highly illuminated room and works even in the complete darkness. We use a DS325 Soft Kinect camera to record the users’ interaction with their smartphones while they type their pins. Using the 900 short RGBD video recording of the pin entry process from 30 different users, we show our attack was able to break 43% of the pins in the first attempt and 61% of the pins in the first 10 attempts. With the advancements in the quality and accessibility of the depth-sensing cameras day by day, we believe our work exposes a major security risk in the present and future and calls the community to take a closer look at the security measures for the usage of the smart devices.

In this manuscript, we demonstrate a method based on atomic force microscopy which enables local probing of surface wettability. The maximum pull-off force, obtained from force spectroscopy shows a remarkable correlation with the macroscopically observed water contact angle, measured over a wide variety of surfaces starting from hydrophilic, all the way through to hydrophobic ones. This relationship, consequently, facilitates the establishment of a universal behaviour. The adhesion forces scale with the polar component of surface energy. However, no such relation could be established with the dispersive component. Hence, we postulate that the force(s) which enable us to correlate the force spectroscopy data measured on the nanoscale to the macroscopic contact angle are primarily arising from electrostatic-dipole-dipole interactions at the solid-liquid interface. London forces play less of a role. This effect in is line with density functional theory (DFT) calculations suggesting a higher degree of hydroxylation of hydrophilic surfaces. This result shows that molecular simulations and measurements on an atomic scale can be extrapolated to macroscopic surface wetting problems.