Search

Magnetic random-access memory (MRAM) is a promising memory technology due to its high density, non-volatility, and high endurance. However, achieving high memory fidelity incurs high write-energy costs, which should be reduced for large-scale deployment of MRAMs. In this paper, we formulate a biconvex optimization problem to optimize write fidelity given energy and latency constraints. The basic idea is to allocate non-uniform write pulses depending on the importance of each bit position. The fidelity measure we consider is mean squared error (MSE), for which we optimize write pulses via alternating convex search (ACS). We derive analytic solutions and propose an alternating water-filling algorithm by casting the MRAM’s write operation as communication over parallel channels. Hence, the proposed alternating water-filling algorithm is computationally more efficient than the original ACS while their solutions are identical. Since the formulated biconvex problem is non-convex, both the original ACS and the proposed algorithm do not guarantee global optimality. However, the MSEs obtained by the proposed algorithm are comparable to the MSEs by complicated global nonlinear programming solvers. Furthermore, we prove that our algorithm can reduce the MSE exponentially with the number of bits per word. For an 8-bit accessed word, the proposed algorithm reduces the MSE by a factor of 21. We also evaluate MNIST dataset classification supposing that the model parameters of deep neural networks are stored in MRAMs. The numerical results show that the optimized write pulses can achieve 40% write-energy reduction for the same classification accuracy. [ABSTRACT FROM AUTHOR]

The transient electric current of surfactants dissolved in a nonpolar solvent is investigated both experimentally and theoretically in the parallel-plate geometry. Due to a low concentration of free charges the cell can be completely polarized by an external voltage of several volts. In this state, all the charged micelles are compacted against the electrodes. After the voltage is set to zero the reverse current features a sharp discharge spike and a broad peak. This shape and its variation with the compacting voltage are reproduced in a one-dimensional drift-diffusion model. The model reveals that the broad peak is formed by a competition between an increasing number of charges drifting back to the middle of the cell and a decreasing electric field that drives the motion. After complete polarization is achieved, the shape of the peak stops evolving with the further increase of the compacting voltage. The spike-peak separation time grows logarithmically with the charge content in the bulk. The time peak is a useful measure of the micelle mobility. Time integration of the peak yields the total charge in the system. By measuring its variation with temperature, the activation energy of bulk charge generation has been found to be 0.126 eV. [ABSTRACT FROM AUTHOR]

Non-volatile memory is a promising alternative to present memory technologies. Oxygen vacancy diffusion has been widely accepted as one of the reasons for the resistive switching mechanism of transition-metal-oxide based resistive random access memory. In this study, molecular dynamics simulation is applied to investigate the diffusion coefficient and activation energy of oxygen in amorphous hafnia. Two sets of empirical potential, Charge-Optimized Many-Body (COMB) and Morse-BKS (MBKS), were considered to investigate the structural and diffusion properties at different temperatures. COMB predicts the activation energy of 0.53 eV for the temperature range of 1000–2000 K, while MBKS predicts 2.2 eV at high temperature (1600–2000 K) and 0.36 eV at low temperature (1000–1600 K). Structural changes and appearance of nano-crystalline phases with increasing temperature might affect the activation energy of oxygen diffusion predicted by MBKS, which is evident from the change in coordination number distribution and radial distribution function. None of the potentials make predictions that are fully consistent with density functional theory simulations of both the structure and diffusion properties of HfO2. This suggests the necessity of developing a better multi-body potential that considers charge exchange. [ABSTRACT FROM AUTHOR]

Reflective displays are advantageous in applications requiring low power or daylight readability. However, there are no low-cost reflective technologies capable of displaying bright colors. By employing photoluminescence to more efficiently use ambient light, we created a prototype display that provides bright, full color in a simple, low-cost architecture. This prototype includes a novel electrokinetic shutter, a layer that incorporates patterned luminescent red, green, and blue sub-pixel elements, and a novel optical out-coupling scheme. The luminescent elements convert otherwise-wasted portions of the incident spectrum to light in the desired color band, resulting in improved color saturation and lightness. This prototype provides a color gamut that is superior to competing reflective display technologies that utilize color filters in single-layer side- by-side sub-pixel architectures. The current prototype is capable of switching in <0.5s; future displays based on an alternative electro-optic shutter technology should achieve video rate operation. A transflective version of this technology has also been prototyped. The transflective version utilizes its backlight with a power efficiency that is at least three times that of a conventional liquid crystal display. These photoluminescence-based technologies enable a host of applications ranging from low-power mobile products and retail pricing signage to daylight readable signage for outdoor advertising segments. [ABSTRACT FROM AUTHOR]