Search

In this work, p-type Bi0.5Sb1.5Te3 alloys were fabricated by high-energy ball milling (MA) and spark plasma sintering. Different revolutions per minute (RPM)s were used in the MA process, and their effect on microstructure, and thermoelectric properties of p-type Bi0.5Sb1.5Te3 were systematically investigated. The crystal structure of milled powders and sintered samples were characterized using X-ray diffraction. All the powders exhibited the same morphology albeit with slight differences find at 1100 RPM conditions. A slight grain size refinement was observed on the fracture surfaces from 500 to 1100 RPM specimens. The temperature dependence of Seebeck coefficient, electrical conductivity, and power factors were measured as a function of temperature with different RPM conditions. The power factor shows almost same (~3.5 W/mK2 at RT) for all samples due to unchanged Seebeck and electrical conductivity values. The peak ZT of 1.07 at 375K is obtained for 1100 RPM specimen due to low thermal conductivity.

NiFe/FeMn/NiFe trilayer structure forms an integral part of many conventional and tunneling magnetoresistance spin valve structures with FeMn antiferromagnetic layer. A systematic investigation of the exchange bias variations of the seed and top pinned NiFe layers in the NiFe/FeMn/NiFe trilayer structure is reported as a function of thickness of all the three constituting layers, in multilayers prepared by rf magnetron sputtering. X-ray diffraction patterns show the (111) texture for the NiFe and FeMn layers, necessary for the development of antiferromagnetic γ-fcc phase. In thickness variation studies of all the three magnetic layers, seed NiFe layer shows greater bias (150 Oe) than the top pinned NiFe layer (80 Oe only). The exchange bias shows the expected 1/t behavior for increasing NiFe layer thickness after initial maxima at low thickness. In the FeMn antiferromagnet layer thickness variation on the other hand, the large bias values attained around 5 nm thickness is nearly retained up to a thickness of 25 nm and the bias for the top NiFe layer is again substantially lower. The greater bias observed for the seed NiFe layer in all the three thickness variation studies is attributed to its growth over a saturated (111) oriented NiFe seed layer, which induces formation of interfacial FeMn layers with a net parallel spin ordering, in presence of the constant applied field. On the other hand, at the top FeMn/NiFe surface, the rigid FeMn surface with compensated bulk spin ordering formed already, is not easily biased and reoriented along the top NiFe layer, to develop as much parallel net spins in the antiferromagnetic material, and hence lower bias. [ABSTRACT FROM AUTHOR]

The usefulness of applying singular value decomposition (SVD) on triaxial accelerometer data for one-dimensional (1D) convolutional neural network (CNN) based fall and activity recognition is investigated. Three-dimensional reduction methods, namely, SVD, sparse principal component analysis, and kernel principal component analysis, are compared for their effectiveness in extracting useful features for fall and activity recognition. Experiments conducted on three public falls and activities of daily living datasets show that SVD applied to acceleration data coupled with raw acceleration data or acceleration signal magnitude vector exhibited better 1D CNN fall and activity recognition accuracy than those using other principal component analysis based acceleration features. [ABSTRACT FROM AUTHOR]