Neural Networks Predict Crystal Stability

SAN DIEGO, Sept. 21, 2018 — Researchers at the University of California, San Diego (UCSD) are using neural networks to predict the stability of materials in two classes of crystals: garnets and perovskites.

They trained artificial neural networks to predict a crystal’s formation energy using just two inputs: electronegativity and ionic radius of the constituent atoms. Based on this work, they developed models that can identify stable materials in two classes of crystals. According to the team, its models are up to 10× more accurate than previous machine learning models and are fast enough to efficiently screen thousands of materials in a matter of hours on a laptop.

“Garnets and perovskites are used in LED lights, rechargeable lithium-ion batteries, and solar cells. These neural networks have the potential to greatly accelerate the discovery of new materials for these and other important applications,” said researcher Weike Ye.

The team has made their models publicly accessible via a web application at http://crystals.ai so that others can use the neural networks to compute the formation energy of any garnet or perovskite composition on the fly.

“Predicting the stability of materials is a central problem in materials science, physics and chemistry,” said professor Shyue Ping Ong. “On one hand, you have traditional chemical intuition such as Linus Pauling’s five rules . . . On the other, you have expensive quantum mechanical computations to calculate the energy gained from forming a crystal . . . What we have done is to use artificial neural networks to bridge these two worlds.”

For more information: doi:10.1038/s41467-018-06322-x




Flat optics: from high-performance metalenses to structured light

In this keynote presentation, Federico Capasso, professor of applied physics at Harvard University, presents advances in dielectric metalenses in the visible, which correct spherical, coma, and chromatic aberrations.

Capasso begins his talk by reminding the audience that conventional lenses still require a very complex type of technology as it takes several lenses to correct aberrations. “Basically, it’s 19th century technology perfected for the 21st century,” says Capasso. “So it’s really polishing, grinding, and so forth with some really expensive machines.”
Capasso describes metaoptics as a different way of looking at diffractive optics. “Metalenses have advantages over traditional lenses,” says Capasso, noting that metalenses are thin, easy to fabricate, and cost effective, and these advantages extend across the whole visible range of light.

Principle of metalenses: Controlling wavefront using nanostructures.

The metalenses developed by Capasso and his team use arrays of titanium dioxide nanofins to equally focus wavelengths of light and eliminate chromatic aberration. The metalenses are designed to provide spatially dependent group delays such that wavepackets from different locations arrive simultaneously at the focus and with the same width.
Metalens research seeks to achieve wavefront shaping of light using optical elements with thicknesses on the order of the wavelength. This miniaturization could to lead to compact, nanoscale optical devices with applications in cameras, lighting, displays, and wearable optics.

For more information: http://spie.org/newsroom/pw18_plenaries/pw18_capasso-




Our new paper in Sylwan Nano Journal

Congratulations for the publication of paper”Thermoplasmonic response of Au@SiO2 core–shell nanoparticles in deionized water and poly-vinylpyrrolidone matrix”

Maher Abdulfadhil Gatea, Hussein A. Jawad, M. Mosleh, S. M. Hamidi

Metal-dielectric core–shell nanoparticles strongly absorb light and convert into an efficient localized heat source in the presence of electromagnatic radiation at their plasmonic resonance. This process can be enhanced depending on the size, shape, structure, and surrounding media. This study theoretically and experimentally investigated the thermoplasmonic effects of Au@SiO2 core–shell nanoparticles immersed in water and poly-vinylpyrrolidone prepared through laser ablation in liquid. Two lasers (532 nm cw Nd:YAG and 520 nm fs pulsed ytterbium fiber) were used to illuminate the prepared samples. The theoretical thermoplasmonic response of the samples was estimated based on the finite element method of COMSOL multiphysics V5.2a. The generated heat difference of Au@SiO2 in both media with fs pulsed laser irradiation was higher than that of cw laser regarding the power used due to the heat confinement during the time of the pulse that cannot be disspated. This study can serve as a basis for using plasmonic core–shell nanoparticles as a nanoheat source in medical applications.




Our new paper in Journal of magnetism and magnetic materials

Congratulations for the publication of paper “Relaxation time dependencies of optically detected magnetic resonance harmonics in highly sensitive Mx magnetometers”

Ranjbaran, M.M. Tehranchi, S.M. Hamidi, S.M.H. Khalkhali

Measurement of extremely weak magnetic fields in double-resonance atomic magnetometers based on resonant optical excitation has been an active area of research in recent years. Magnetometer sensitivity can be improved via detection of higher harmonics of the magnetic resonance, a resonance which has a maximum sensitivity under conditions where the ratio of the amplitude to the line-width of the resonance signal is maximized. Based on the Bloch theorem, we analyze the time evolution of the spin polarization corresponding to each harmonic component of the resonance signal and measure this progression experimentally. Our results revealed that there is an optimal harmonic number for achieving the highest sensitivity. We have shown that the longitudinal and transverse relaxation times of spin polarization can manipulate the harmonics with the best sensitivity while the excitation frequency is detuned from the Larmor frequency.




3-D Printing Graphene Aerogels

A U.S. research team from Virginia Tech and the Lawrence Livermore National Lab (LLNL) has demonstrated a light-based approach for 3-D printing strong, lightweight, porous graphene aerogels—at a resolution an order of magnitude finer than other techniques. 3-D printing is well advanced for polymer foams, with some notable success, but is still an active area of research for graphene foams. Researchers have published schemes for printing 3-D graphene using a number of approaches, such as extrusion. As a result, according to the researchers, these techniques have generally printed out 3-D graphene structures with relatively weak, bending-dominated configurations, such as stacked “woodpile” structures, and relatively large achievable feature sizes (greater than 100 microns). That’s a far cry, they say, from the high-resolution, complex structures that could open up applications in areas such as energy storage and conversion.

In particular, they opted for a form of 3-D printing called projection micro-stereolithography (PμSL)—a light-based technique that allows the resin feeding the 3-D printing process to be shaped into fine-scale, intricate forms via patterned light. Using this technique, an entire layer of 3-D-printing resin can be cured, at very fine scales, via a single UV flash. That advantage, the team reasoned, could potentially overcome some of the toolpath and sequence issues experienced by other approaches to 3-D printing of graphene. The big trick was to figure out a way to make a photocurable graphene resin—one that would quickly firm up under a light beam, but that also was sufficiently runny to be slathered layer by layer on the workpiece. To get there, Hensleigh spent some time in the chemistry lab, developing a porous graphene-oxide hydrogel with cross-linked sheets, and then using ultrasound to disperse the cross-linked graphene oxide into a dilute, 1-weight-percent suspension.

The result is exquisite, intricate, airy structures, such a lattices of octet trusses, with feature sizes on the order of 10 microns—an order of magnitude finer, according to the team, than other 3-D-printed graphene structures. They’re lightweight enough to balance on a single filament of a strawberry blossom (see image at top of story). And they’re also strong; as measured by their Young’s modulus, the strength of the PμSL-printed structures seems to hold up better than that of most other 3-D graphenes and other carbon aerogels as the density of the structure decreases.

For more information:  doi: 10.1039/C8MH00668G




Our new paper in Journal of magnetism and magnetic materials

Congratulations for the publication of paper “Faraday rotation in a coupled resonator magneto-plasmonic structure Tamm plasmon boosting

M. Hamidi, R. Moradlou

The present study aimed to evaluate the magneto-optic Faraday rotation of one-dimensional coupled resonator magnetoplasmonic structure by metallic cover layer in each resonator. To this purpose, transfer matrix method was used where SiO2 and Bismuth substitute garnet thin films playing main building block and the gold or silver layer use to reach ((SiO2/ Bi:YIG)n /(Au or Ag)/ SiO2)m structure; where n and m are considered as the repetition and resonator numbers, respectively. Tamm plasmons related to the phase cancelation in photonic band gap are identified by optical and magneto-optical spectra and accordingly the phase change in the structure. Based on these modes, a wide range of wavelengths is detected by which the figure of merit increases due to the interaction of light with Tamm plasmons and surprisingly the flat optical window in this region in addition to the main resonance. These structures can open a new gate for enhancing performance of the magneto-optic devices.

For more information: https://doi.org/10.1016/j.jmmm.2018.08.083




New Plasmonic Metamaterial Could Revolutionize Solar Cells

A recent discovery at the University of California San Diego could change the field of photonics. A team of engineers has fabricated a plasmonic metamaterial that could change the way we look at optical transmission.

The study was led by electrical engineering professor Shaya Fainman at the UC San Diego Jacobs School of Engineering, and was published in Nature Communications.

“We’re offsetting the loss introduced by the metal with gain from the semiconductor. This combination theoretically could result in zero net absorption of the signal — a lossless metamaterial,” Commented Joseph Smalley the first author of the study.

Their metamaterial works because a light emitting semiconductor replaces the lost light, though creating it is much more complex than describing it.

This plasmonic metamaterial is created by growing the crystal of a indium gallium arsenide phosphide semiconductor on a substrate. They then use plasma to etch 40-nanometer-wide trenches, that are spaced 40-nanometers apart. Then the trenches are filled, and create tiny stripes of silver and semiconductor.

“This is the first material that behaves simultaneously as a metal and a semiconductor. If light is polarized one way, the metamaterial reflects light like a metal, and when light is polarized the other way, the metamaterial absorbs and emits light of a different ‘color’ like a semiconductor,” Smalley added.

Clearly this is an amazing new process which offers us higher efficiency optical transmission. While it is still in the laboratory phase, the manufacturing method could potentially be expanded to commercial production.




Magnetic nanoparticles deliver chemotherapy to difficult-to-reach spinal tumors

Researchers at the University of Illinois at Chicago have demonstrated that magnetic nanoparticles can be used to ferry chemotherapy drugs into the spinal cord to treat hard-to-reach spinal tumors in an animal model. The unique delivery system represents a novel way to target chemotherapy drugs to spinal cancer cells, which are hard to reach because the drugs must cross the blood-brain barrier.

Spinal cord tumors are a challenge to treat because they are difficult to surgically remove due to their proximity to healthy spinal tissue and because chemotherapy drugs must cross the blood-brain barrier in order to reach them. Intramedullary spinal cord tumors account for 8 percent to 10 percent of all spinal cord tumors and are common among children and adolescents. Average survival for patients with these tumors is 15.5 months.

The researchers, whose results are published in the journal Scientific Reports, used a unique rat model with implanted human intramedullary spinal cord tumors to show that magnetic nanoparticles could successfully be used to kill tumor cells.

First, they created nanoparticles made up of tiny, metallic magnets bound to particles of doxorubicin. Next, they implanted a magnet just under the skin covering the spinal vertebrae in the rat models. Then they injected the magnetic nanoparticles into the space around the spinal cord where the tumor was located.

The magnet implanted in close proximity to the tumor guided the nanoparticles to the tumor sites. The researchers were able to show that tumor cells took up the nanoparticles and underwent apoptosis — in other words, they were effectively destroyed. The impact of the nanoparticles on nearby healthy cells was very minimal, Mehta said.

“This proof-of-concept study shows that magnetic nanoparticles are an effective way to deliver chemotherapy to an area of the body that has been difficult to reach with available treatments,” he said. “We will continue to investigate the potential of this therapy and hope to enter human trials if it continues to show promise.”

For more information: DOI: 10.1038/s41598-018-29736-5




Toward All-Optical Artificial Neural Networks

Training an artificial neural network for a specific task can be a computationally intensive and energy-consuming feat. Researchers at a U.S. university have demonstrated that such training can be accomplished on a silicon photonic chip.

A new way to train

In previous experiments on optical neural networks, other researchers performed the network training on a traditional computer and then transferred the results onto a photonic chip. Here, the Stanford group performed the algorithm physically by propagating an error signal through the circuits of the chip. According to Hughes, this method “should make training of optical neural networks far more efficient and robust.”

For hardware, the Stanford team used a silicon photonic architecture similar to a programmable processor describe last year at the Massachusetts Institute of Technology, USA. Basically, it’s a mesh of tiny, tunable Mach-Zender interferometers. For software, the researchers derived the algorithm from the mathematics of the optical circuit, going all the way back to Maxwell’s equations.

Fine-tuning

The “teaching” of the network involves sending a laser pulse one way through the optical circuit, measuring how the signal was changed from the predicted signal, then adjusting the circuit and sending the optical signal back. Based on the received signal, the artificial neural network adjusts itself by tweaking its circuitry via optical phase shifters. This tuning happens by “applying an electrical voltage to a heating element on the chip’s surface,” says Hughes, “which changes the optical properties of the waveguide slightly.” Tiny photodetectors near the phase shifters measure the intensity of the signal passing through the chip, giving the algorithm the gradient information needed for training and optimization.

For more information: Optica, doi:10.1364/OPTICA.5.000864




Could Excitons Aid Optical-Electronic Interconnects?

A nagging efficiency bottleneck in today’s communications networks is the need to convert between the optical signals that transmit data over long distances, and the electrical signals used in data processing. One potential solution lies in devices that manipulate not electrons or photons, but “excitons”—the bound electron-hole pairs formed when photons excite electrons in a semiconductor. But thus far, the “excitonic” devices demonstrated using bulk semiconductor materials have had to operate at frigid temperatures, a disadvantage that has held back practical applications. Now, a Swiss-Japanese research team has used an ingenious stack of 2-D materials to develop a key component for practical excitonics: an excitonic transistor that can operate at room temperature.

The heterostructure difference

At the heart of the system are layers of two atomically thin TMDs, molybdenum disulfide (MoS2) and tungsten diselenide (WSe2). Because of the differing band structure of the two materials, when an exciton is created in the heterostructure (for example, by absorption of a photon), the electron tends to reside in the MoS2 layer, while the hole stays in the WSe2 layer. The result is a system in which the exciton “lives” not in a single 2-D material layer, but between the two layers.

Such an interlayer exciton, it turns out, has a spatial separation between the electron and the hole that’s large enough to allow the exciton to survive 100 times longer than it would in a single 2-D material layer. Yet the exciton can still exist and thrive at room temperature. Further, the two-layer structure means that the exciton has a built-in out-of-plane dipole moment. That means it can be manipulated and controlled by an electric field and voltage bias in ways that would be impossible with excitons in a single 2-D layer.

Graphene gates

The team found that the interlayer excitons were sufficiently long-lived to diffuse across a distance as long as five microns within the structure before recombining and emitting light. Further, the flux of excitons could be controlled and manipulated electrically by applying different voltage biases using the graphene electrodes, in transistor-like fashion.

For more information: Nature, doi: 10.1038/s41586-018-0357-y