Analog in-memory computing (AIMC) is an emerging paradigm that can enable energy efficient computing orders of magnitude beyond what is currently possible. Memory candidates for AIMC include SONOS (semiconductor oxide nitride oxide nitride), emerging resistive memory (ReRAM) and electrochemical memory (ECRAM). Electrical requirements for these memories are different than traditional digital memories in that the exact conductivity state of every device is used in every calculation. Effects including programming error and state drift are incorporated in the algorithm output. This new set of requirements has forced the development of a novel, holistic methodology for the electrical characterization and benchmarking of these devices. This talk will discuss these characterization and benchmarking methodology, and its application to SONOS, ReRAM, and ECRAM. SNL is managed and operated by NTESS under DOE NNSA contract DE-NA0003525.
Variational inference (VI) is an approximation of statistically valid Bayesian inference which is well-suited for analog accelerators, and stochastic nanomagnetic devices in particular are a strong candidate to implement this feature by exploiting tunable randomness in magnetic thin-films that can be run quickly and with a low power-draw. In this work, we a) discuss how VI can be reliably implemented with a combination of low-noise nanomagnetic synapses and tunable noise generating sources (magnetic tunnel junctions (MTJs) in a single analog design and b) summarize efforts to characterize the state-dependent noise profiles of various MTJ designs for various applications.
Magnetic domain-wall devices, modulated by the spin-transfer torque or the spin-orbit torque effect, can implement logical operations in a manner that is inherently compact and cascadable. Using circuit simulations with micromagnetics-validated compact models, we evaluate the device requirements for domain-wall logic that has low latency, outperforms scaled CMOS logic in energy efficiency, and remains robust to process variations. We further show how the inherent non-volatility of these devices can be leveraged to construct stateful logic circuits that save energy and area relative to their CMOS counterparts and propose novel logic architectures that exploit the unique advantages of domain-wall devices.
Refrigeration is an intrinsic feature of light-emitting diodes, a fact that was recognized decades ago but has so far eluded direct experimental observation at practical power densities. The problem is insufficient external luminescence efficiency; for net cooling to occur, the losses in the device must be close to zero, and a sufficiently efficient LED has yet to materialize. We propose a possible structure for such an LED, and predict that with existing optoelectronic material quality and device processing, electroluminescent refrigeration is not only possible but is potentially more efficient than its solid-state alternatives, particularly at low temperature.
To approach the Shockley-Queisser limit, a solar cell must embody the principles of efficient light-emitting diode (LED) design. Here we describe how ultra-high luminescence efficiency, both internal and external, is the basis for the present efficiency records in solar energy conversion. These developments have provided an impetus for new energy technologies, which rely on the same design strategies to reach their theoretical limits. Thermophotovoltaics, the conversion of terrestrially produced thermal radiation to electricity, can now approach >50% efficiency. Ultra-efficient photovoltaics and LEDs also enable optoelectronic refrigerators with the potential to surpass other methods of solid-state cooling in energy efficiency.
The new breakthrough in photovoltaics, exemplified by the slogan “A great solar cell has to be a great light-emitting diode (LED)”, has led to all the major new solar cell records, while also leading to extraordinary LED efficiency. As an LED becomes very efficient in converting its electrical input into light, the device cools as it operates because the photons carry away entropy as well as energy. If these photons are absorbed in a photovoltaic (PV) cell, the generated electricity can be used to provide part of the electrical input that drives the LED. Indeed, the LED/PV cell combination forms a new type of heat engine with light as the working fluid. The electroluminescent refrigerator requires only a small amount of external electricity to provide cooling, leading to a high coefficient of performance.
We present the theoretical performance of such a refrigerator, in which the cool side (LED) is radiatively coupled to the hot side (PV) across a vacuum gap. The coefficient of performance is maximized by using a highly luminescent material, such as GaAs, together with device structures that optimize extraction of the luminescence. We consider both a macroscopic vacuum gap and a sub-wavelength gap; the latter allows for evanescent coupling of photons between the devices, potentially providing a further enhancement to the efficiency of light extraction. Using device assumptions based on the current record-efficiency solar cells, we show that electroluminescent cooling can, in certain regimes of cooling power, achieve a higher coefficient of performance than thermoelectric cooling.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.