Jun 16 2016
Scientists at Hewlett Packard Enterprise (HPE) have experimentally proved the crucial aspects of how memristor, a new kind of microelectronic device, operates at an atomic scale. This was proven through experiments conducted by scientists at two Department of Energy national labs, Lawrence Berkeley National Laboratory, and SLAC National Accelerator Laboratory.
This animation shows how millions of high-voltage switching cycles can cause memristors to fail. The high-voltage switching eventually creates regions that are permanently rich (blue pits) or deficient (red peaks) in oxygen and cannot be switched back. Switching at lower voltages that would be suitable for commercial devices did not show this performance degradation. These observations allowed the researchers to develop materials processing and operating conditions that improved the memristors’ endurance by nearly a thousand times. (Suhas Kumar)
This result is a crucial step in developing the design for these solid-state devices, which will be used in future computer memories capable of functioning more rapidly with longer lasting time and less energy usage, compared to today’s flash memory. The results of the study have been featured in the February issue of Advanced Materials.
We need information like this to be able to design memristors that will succeed commercially.
Suhas Kumar, Scientist, HPE
In 1971, the memristor was theoretically proposed as the fourth basic electrical device element along with the inductor, capacitor, and resistor. The memristor has a very small piece of a transition metal oxide placed between two electrodes. The electrical resistance of the memristor dramatically decreases or increases by applying a negative or positive voltage pulse. This behavior allows the memristor to be used as a “non-volatile” computer memory that is quite similar to a flash memory as it is capable of retaining its state without the need for being refreshed with extra power.
An HPE group headed by senior fellow R. Stanley Williams has researched on varied memristor designs, behavior, and material over the past decade. Since 2009, the researchers used strong synchrotron X-rays in order to expose the movements of atoms present in the memristors during switching. Despite the enhanced understanding of the nature of the switching, crucial facts that could play a vital role in designing circuits that are commercially successful remained controversial. For instance, the forces causing the atoms to move, which indeed results in dramatic resistance changes during switching, is still a debatable topic.
Recently, the researchers analyzed memristors created with oxides of vanadium, tantalum, and titanium. The primary experiments pointed out that switching in the tantalum oxide devices can be easily monitored, so it was selected for advanced exploration at two DOE Office of Science User Facilities, Berkeley Lab’s Advanced Light Source (ALS) and SLAC’s Stanford Synchrotron Radiation Lightsource (SSRL).
The HPE researchers at ALS mapped the positions of oxygen atoms after and before switching. This process was carried out using a scanning transmission X-ray microscope and an apparatus they developed to accurately monitor the sample’s position and the intensity and timing of the 500-electronvolt ALS X-rays, which were adjusted to see oxygen.
The experiments highlighted that even the feeble voltage pulses develop a thin conductive path passing through the memristor. The path is heated during the pulse, and this results in developing a force that pushes the oxygen atoms away from the path, allowing it to be more conductive. Reversing of the voltage pulse leads to resetting the memristor by sucking a small number of oxygen atoms back into the conducting path, increasing the resistance of the device. The resistance of the memristor varies between 10-fold and 1 million-fold, based on operating parameters such as voltage-pulse amplitude. This resistance variation is dramatic enough for commercial development.
To come up with a conclusion, it was necessary for the team to understand if the movement of the tantalum atoms was along with the oxygen during switching. Tantalum imaging required higher-energy, 10,000-electronvolt X-rays, which the researchers obtained at SSRL’s Beam Line 6-2. In a single session, the researchers highlighted that the tantalum remained stable.
That sealed the deal, convincing us that our hypothesis was correct.
Catherine Graves, Scientist, HPE
Catherine Graves had worked at SSRL as a Stanford graduate student, added that discussions with SLAC experts were critical in guiding the HPE team toward the X-ray techniques that would allow them to see the tantalum accurately.
Kumar stated that the most favorable aspect of the tantalum oxide results refers to the fact that the scientists did not detect degradation in switching over more than a billion voltage pulses of a magnitude appropriate for commercial use. He further stated that this detail allowed the researchers to develop memristors that existed for almost a billion switching cycles, which indeed is a thousand-fold enhancement.
This is much longer endurance than is possible with today’s flash memory devices. In addition, we also used much higher voltage pulses to accelerate and observe memristor failures, which is also important in understanding how these devices work. Failures occurred when oxygen atoms were forced so far away that they did not return to their initial positions.
Suhas Kumar, Scientist, HPE
Kumar goes on to state that beyond memory chips, an increase in the switching speed and small size will allow them to be perfect for use in logic circuits. Extra memristor characteristics may also prove to be advantageous in the upcoming class of brain-inspired neuromorphic computing circuits.
“Transistors are big and bulky compared to memristors,” he said. “Memristors are also much better suited for creating the neuron-like voltage spikes that characterize neuromorphic circuits.”
Other partners in the study were HPE’s John Paul Strachan and Emmanuelle Merced Grafals; the group of Yoshio Nishi, professor of electrical engineering at Stanford; Johanna Nelson Weker, staff scientist at SSRL; and David Kilcoyne and Tolek Tyliszczak, beamline scientists at ALS. Funding for the work at ALS and SSRL was received from DOE Office of Science, Office of Basic Energy Sciences. The HPE group is affiliated with HPE’s research division, Hewlett Packard Labs.