EESA Scientists were able to use the new synchrotron Infrared Nano Spectroscopy (SINS) capability at Advanced Light Source.
Above: Diagrams of the setup of (a) the Germanium-hemisphere enhanced attenuated total reflection (Ge micro-ATR) and (b) the resonance enhanced SINS.
Berkeley Lab scientists have identified a way to use machine learning to connect fine- and large-scale measurements of shale – the most abundant sedimentary rock in the Earth’s subsurface and one known for its ability to trap fluids such as carbon dioxide and methane. Results of the research team’s recent study, “Cross-Scale Molecular Analysis of Chemical Heterogeneity in Shale Rocks,” were published in the Nature publication Scientific Reports.
Zhao Hao, a physicist, and Peter Nico, a soil and environmental biogeochemist, work within the national lab’s Earth and Environmental Sciences Area. Hao led the researchers in using molecular imaging methods based on infrared spectroscopy to observe chemical and physical diversity in shale samples. They collected chemical data at the Advanced Light Source on the scale of nanometers using synchrotron infrared Nano spectroscopy and at the millimeter scale using Germanium hemisphere infrared spectroscopy then built a machine-learning model to connect the nanoscale measurements to the millimeter scale ones.
According to Nico, the ability to characterize the chemical and physical properties of shale rock at the nanoscale level and then translate the understanding across scales is key to improving technologies that depend upon predicting the behavior of subsurface rock under various conditions.
“For technologies that require an understanding of the structure of rock – particularly how rock fractures – the more we know about this structure on a small scale, the better,” says Nico. “Because shale is associated with a number of energy production technologies, knowledge about how shale behaves under pressure and other conditions is particularly important.”
Subsurface rock is stimulated to cause fractures that enable the release of hydrocarbons, in the case of hydraulic fracturing, and heat, in the case of enhanced geothermal systems (EGS), for conversion to energy. Carbon capture and sequestration, on the other hand, relies upon subsurface rock staying sufficiently intact to prevent carbon leakage. The association between subsurface rock and energy production has made modeling subsurface rock fractures a key research focus within the Energy Geosciences Division.
Shales are sedimentary rocks made up largely of fine-grained mineral particles intermixed with organic matter. Together these form a nanoporous network giving shale its limited permeability. Hao’s team wanted to better understand the flow of fluids through this shale during energy production or weathering reactions, which requires information about shale’s nanoscale composition and mesoscale diversity. Until now it has been difficult to obtain sufficient information because of the difficulty involved with capturing ultra high-resolution images of sedimentary rock that include chemical information and are also relevant to the large scale because of the time necessary to image even small areas with nanometer resolution.
“We were able to utilize the new synchrotron Infrared Nano Spectroscopy (SINS) capability at the ALS and translate that information to the macroscale images taken using the Germanium hemisphere approach through the machine learning approach,” Nico says. “This allowed us to connect two methods that had previously been used separately but not together.
Changes that occur at the nanoscale in subsurface rock do so in response to variables like temperature, pressure, or chemistry. Being able to connect the nanoscale to the macroscopic scale allows us to make better predictions about how rocks will respond in the real world as these variables change.”