Oct. 27, 2023
The Scorpius Injector, pictured here, generates X-ray images of weapons-grade materials under explosive conditions. Image courtesy of Craig Fritz/Sandia National Laboratories.
Keeping weapons in check
Deep below the Nevada desert, a machine dubbed Scorpius is under construction that will use high explosives to crush plutonium to states that exist just prior to a nuclear explosion. The aim of the U.S. project is to scan this plutonium with X-rays to help check the accuracy of supercomputer models designed to predict whether the United States’ aging nuclear arsenal works.
In the first roughly 50 years of the U.S. nuclear weapons program, researchers tested whether the bombs worked by actually detonating them. However, in 1992, President George H.W. Bush signed into law a moratorium on nuclear tests.
Currently, supercomputer models help predict whether U.S. nuclear weapons might still work. However, the models’ accuracy remains uncertain. Scientists do blow up explosives to supply data for these models, but these surrogate materials possess significant differences from the plutonium typically used in nuclear bombs. This raises the question of how well these models simulate real nuclear explosions.
The aim of Scorpius is to help give supercomputer models the accurate data they need to see if they are generating realistic simulations of nuclear-weapon behavior. The machine, under construction a thousand feet beneath the Nevada National Security Site — a test area bigger than the state of Rhode Island — is expected to be up and running by late 2027.
Scorpius is a project of the Sandia, Los Alamos, and Lawrence Livermore national laboratories, as well as the Nevada National Security Site.
Students from Tracy and the East Bay use their phones to conduct experiments as part of the Lawrence Livermore National Laboratory’s Physics with Phones program. Photo courtesy of Livermore High School.
Can you hear me now?
A scientist at the Lawrence Livermore National Laboratory is using something that students love — cell phones — to teach them about physics.
Senior scientist David Rakestraw has spent the past five years developing an education program that uses different sensors on a modern cell phone to help students do physics experiments that were previously cost-prohibitive.
The program brings physics education to lower-income areas, and addresses a nationwide shortfall in science, technology, engineering and mathematics (STEM) education.
Smart phone sensors, such as accelerometers that measure acceleration and transducers that measure air pressure, allow schools to offer lab experiments at a fraction of the cost of a traditional physics classroom.
Monolithic optics like the V3 unit from LLNL shown here have used low-CTE Invar housings to ensure camera accuracy. However, today’s increased volume of small satellites require materials with lower density, cost and lead times.
Carbon nanotubes to the rescue
The space industry is growing. According to a January 2023 McKinsey report, the number of small satellites launched per year will ramp from 1,740 in 2022 to more than 3,000 in 2030, and space consulting firm NSR has projected that 24,700 satellites will be launched between 2021 and 2030.
The small satellite hardware development team within Lawrence Livermore National Laboratory’s Space Program has been advancing technology for small satellites. Since 2012, LLNL has been developing optical imaging payloads for cube satellites (cubesats, measuring 10 centimeters per side) for monitoring space debris and other missions. Compared to conventional designs with independent primary and secondary mirrors, LLNL has developed a monolithic optic that combines these mirrors into a single element. These monolithic optics fit the tight package space of small satellites yet are robust enough to handle vibration loads during launch and temperature extremes in space. However, they require a housing that has a very low coefficient of thermal expansion (CTE). Currently, that is achieved using Invar, which has a CTE of ~1.3 (micrometer/meter)/ºC. A 36% nickel alloy with iron, Invar also is heavy, expensive and requires lead times of multiple months.
In 2021, Patz Materials and LLNL teamed up to replace Invar in these monolithic optic housings with a molding compound comprising PMT-F16 epoxy resin modified with carbon nanotubes and reinforced with 6K tow high modulus carbon fiber with 60% fiber content. The project demonstrated not only the ability to meet all metallic housing performance requirements at a fraction of the weight, but also provided even further benefits when the housing was redesigned to take advantage of the composite material and molding process.
A deep-space image of the galaxy where the supernova occurred. Image courtesy of NASA.
Classifying the death of a star
A fully automated process, including a brand-new artificial intelligence (AI) tool, has successfully detected, identified and classified its first supernova.
Developed by an international collaboration led by Northwestern University and including Lawrence Livermore, the new system automates the entire search for new supernovae across the night sky — effectively removing humans from the process. Not only does this rapidly accelerate the process of analyzing and classifying new supernova candidates, it also bypasses human error.
The team recently alerted the astronomical community to the launch and success of the new tool, called the Bright Transient Survey Bot (BTSbot). In the past six years, humans have spent an estimated total of 2,200 hours visually inspecting and classifying supernova candidates. With the new tool now officially online, researchers can redirect this precious time toward other responsibilities to accelerate the pace of discovery.
DOLCE can reconstruct low-angle CT images, despite missing and limited datasets. Image courtesy of Washington University.
CT scans enlightened with AI
Because of its speed, low radiation exposure, and advantages in specific applications, low-angle CT is a commonly used technique, often applied in emergency or critical care cases, dental cone beam CT, or when patients have trouble holding their breath. But it has its drawbacks, namely incomplete data sets, which often result in artifacts and discrepancies that reduce the accuracy of scans, even when they are reconstructed with AI.
To address these limitations, engineers at Washington University in St. Louis and researchers at the Lawrence Livermore National Laboratory developed and tested DOLCE, a deep learning model-based framework that uses generative AI to produce multiple high-quality images from severely limited data. They compiled their findings in a study, “DOLCE: A Model-Based Probabilistic Diffusion Framework for Limited-Angle CT Reconstruction,” which they discussed recently at the International Conference on Computer Vision (ICCV) in Paris.
DOLCE stands for diffusion probabilistic limited-angle CT reconstruction, and unlike traditional generative AI models, such as large language model chatbots, it uses AI to reconstruct images, but also quantifies its reconstruction uncertainty, including for the type of measured data it uses and the physical properties of the underlying system. It also creates a variance map to show the different alternative reconstructions it could have produced with the available data.
