Automated detection of orbital angular momentum (OAM) can tremendously contribute to quantum optical experiments. We develop convolutional neural networks to identify and classify noisy images of Laguerre–Gaussian (LG) modes collected from two different experimental set ups. We investigate the classification performance measures of the predictive classification models for experimental conditions. The results demonstrate accuracy and specificity above 90% in classifying 16 LG modes for both experimental set ups. However, the F-score, sensitivity, and precision of the classification range from 57% to 92%, depending on the number of imperfections in the images obtained from the experiments. This research could enhance the application of OAM light in telecommunications, sensing, and high-resolution imaging systems.
Additive manufacturing (AM) is a crucial component of smart manufacturing systems that disrupts traditional supply chains. However, the parts built using the state-of-the-art powder-bed 3D printers have noticeable unpredictable mechanical properties. In this paper, we propose a closed-loop machine learning algorithm as a promising way of improving the underlying failure phenomena in 3D metal printing. We employ machine learning approach through a Deep Convolutional Neural Network to automatically detect the defects in printing the layers, thereby turning metal 3D printers into essentially their own inspectors. By comparing three deep learning models, we demonstrate that transfer learning approach based on Inception-v3 model in Tensorflow framework can be used to retrain our images data set consisting of only 200 image samples and achieves a classification accuracy rate of 100 % on the test set. This will generate a precise feedback signal for a smart 3D printer to recognize any issues with the build itself and make proper adjustments and corrections without operator intervention. The closed-loop ML algorithm can enhance the quality of the AM process, leading to manufacturing better parts with fewer quality hiccups, limiting waste of time and materials.
In this paper, we propose a distributed machine learning (DML) algorithm to fulfill the requirements of the smart factory (or Industry 4.0) including self-organization, a distributed control function, communication between the smart components, and real-time decision-making capability. We show the proposed DML algorithm not only enables the smart factory to adjust the components for new demands and circumstances, but also each component of the system acts smart and communicate with each other, either request or offer functions. The DML is an interactive learning mechanism among smart components and a natural way of scaling up learning algorithms. The different machines can have the best learning algorithms of their own data while the communication between different learning processes is an integration of different learning biases that compensate one another for their inefficient characteristics. As such, the size of the smart factory is scalable and the growing amount of data from additional machines has a minor effect on the communication overheat. We will elaborate on the DML model that overcomes the problems of centralized systems and increases the possibility of achieving higher accuracy, especially on a large-size domain.
Graphene is a promising material for thermoelectric application due to its large surface-to-volume ratio, high electrical conductivity, and high mechanical strength. In this paper, the thermoelectric properties of a series of narrow armchair graphene nanoribbons (GNR) in semiconducting family GNR(3p+1,0) are evaluated by using the semi-classical Boltzmann theory. It is found that the narrow GNR(7,0) exhibits small thermal conductivity and large TEP of 1170μV / K at small chemical potential μ = 0.1 eV . However, the small electrical conductivity of narrow GNR(7,0) suppresses the thermoelectric figure-of-merit ZT, such that better thermoelectric performance of ZT > 0.01 is achieved only for large chemical potentials, μ > 0.5eV . Our result shows that tuning the chemical potential with respect to ribbon chirality and orientation can enhance the thermoelectric performance of GNRs, however, further increase in thermoelectric power requires phonon engineering to reduce the thermal conductivity of graphene without significant reduction in its thermoelectric power and electrical conductivity.
Raman scattering is a well-known technique for detecting and identifying complex molecular samples. The weak Raman signals are enormously enhanced in the presence of a nano-patterned metallic surface next to the specimen. This paper reports new techniques to obtain the nanostructures required for Surface Enhanced Raman Scattering (SERS) without costly and sophisticated fabrication steps, which are nanoimprint lithography (NIL), electrochemical deposition, electron beam induced deposition, and focus ion beam (FIB). 20 nm Au thicknesses of sputtered Au were deposited on etched household aluminum foil (base substrate) for vitro application. The Raman signal were caused by the Aluminum pre-etched times. In preliminary results, enhancement factors of 106 times were observed from SERS substrate for in vitro measurements. Moreover, the ability to perform in vivo measurements was demonstrated after removing the base aluminum foil substrate. This application allows Raman signals to be obtained from the surface or interior of opaque specimens. The nano-patterned gold may also be coupled in a probe to a remote spectrometer via an articulated arm. This opens up Raman spectroscopy for use in a clinical environment.
Graphene has been extensively investigated as a promising material for various types of high performance sensors due to its large surface-to-volume ratio, remarkably high carrier mobility, high carrier density, high thermal conductivity, extremely high mechanical strength and high signal-to-noise ratio. The power density and the corresponding die temperature can be tremendously high in scaled emerging technology designs, urging the on-chip sensing and controlling of the generated heat in nanometer dimensions. In this paper, we have explored the feasibility of a thin oxide graphene nanoribbon (GNR) as nanometer-size temperature sensor for detecting local on-chip temperature at scaled bias voltages of emerging technology. We have introduced an analytical model for GNR FET for 22nm technology node, which incorporates both thermionic emission of high-energy carriers and band-to-band-tunneling (BTBT) of carriers from drain to channel regions together with different scattering mechanisms due to intrinsic acoustic phonons and optical phonons and line-edge roughness in narrow GNRs. The temperature coefficient of resistivity (TCR) of GNR FET-based temperature sensor shows approximately an order of magnitude higher TCR than large-area graphene FET temperature sensor by accurately choosing of GNR width and bias condition for a temperature set point. At gate bias VGS = 0.55 V, TCR maximizes at room temperature to 2.1×10−2 /K, which is also independent of GNR width, allowing the design of width-free GNR FET for room temperature sensing applications.
In this paper, we have explored the feasibility of a metallic single-walled carbon nanotube (SWCNT) as a radiation detector. The effect of SWCNTs’ exposure to different ion irradiations is considered with the displacement damage dose (DDD) methodology. The analytical model of the irradiated resistance of metallic SWCNT has been developed and verified by the experimental data for increasing DDD from 1012 MeV/g to 1017 MeV/g. It has been found that the resistance variation of SWCNT by increasing DDD can be significant depending on the length and diameter of SWCNT, such that the DDD as low as 1012 (MeV/g) can be detected using the SWCNT with 1cm length and 5nm diameter. Increasing the length and diameter of SWCNT can result in both the higher radiation sensitivity of resistance and the extension of detection range to lower DDD.