Search for: All records

Perennial grasslands, including prairie and pasture, have declined with tremendous environmental and social costs. This decline reflects unequal policy support for grasslands and managed grazing compared to row crops. To create a resource for community partners and decision-makers, we reviewed and analyzed the policy tools and implementation capacity that supports and constrains grasslands and managed grazing in the U.S. Upper Midwest. Risk reduction subsidies for corn and soybeans far outpace the support for pasture. Some states lost their statewide grazing specialist when the federal Grazing Lands Conservation Initiative lapsed. The United States Department of Agriculture, Natural Resources Conservation Service support for lands with prescribed grazing practices declined after 2005 but remained relatively steady 2010–2020. These results reveal the policy disadvantage for grasslands and managed grazing in comparison with row crop agriculture for milk and meat production. Grassland and grazing policies have an important nexus with water quality, biodiversity, carbon and outdoor recreation policy. Socially just transitions to well-managed, grazed grasslands require equity-oriented interventions that support community needs. We synthesized recommendations for national and state policy that farmers and other grazing professionals assert would support perennial grasslands and grazing, including changes in insurance, conservation programs, supply chains, land access, and fair labor. These policies would provide critical support for grass-based agriculture and prairies that we hope will help build soil, retain nutrients, reduce flooding and enhance biodiversity while providing healthy food, jobs, and communities. 

To enable greater control over thermal atomic layer deposition (ALD) of molybdenum disulfide (MoS 2 ), here we report studies of the reactions of molybdenum hexafluoride (MoF 6 ) and hydrogen sulfide (H 2 S) with metal oxide substrates from nucleation to few-layer films. In situ quartz crystal microbalance experiments performed at 150, 200, and 250 °C revealed temperature-dependent nucleation behavior of the MoF 6 precursor, which is attributed to variations in surface hydroxyl concentration with temperature. In situ Fourier transform infrared spectroscopy coupled with ex situ x-ray photoelectron spectroscopy (XPS) indicated the presence of molybdenum oxide and molybdenum oxyfluoride species during nucleation. Density functional theory calculations additionally support the formation of these species as well as predicted metal oxide to fluoride conversion. Residual gas analysis revealed reaction by-products, and the combined experimental and computational results provided insights into proposed nucleation surface reactions. With additional ALD cycles, Fourier transform infrared spectroscopy indicated steady film growth after ∼13 cycles at 200 °C. XPS revealed that higher deposition temperatures resulted in a higher fraction of MoS 2 within the films. Deposition temperature was found to play an important role in film morphology with amorphous films obtained at 200 °C and below, while layered films with vertical platelets were observed at 250 °C. These results provide an improved understanding of MoS 2 nucleation, which can guide surface preparation for the deposition of few-layer films and advance MoS 2 toward integration into device manufacturing. 

Short-range properties of Na–NaLi enable precise control of chemical reactivity using magnetic fields and quantum interference. 

Structured non-convex learning problems, for which critical points have favorable statistical properties, arise frequently in statistical machine learning. Algorithmic convergence and statistical estimation rates are well-understood for such problems. However, quantifying the uncertainty associated with the underlying training algorithm is not well-studied in the non-convex setting. In order to address this short-coming, in this work, we establish an asymptotic normality result for the constant step size stochastic gradient descent (SGD) algorithm—a widely used algorithm in practice. Specifically, based on the relationship between SGD and Markov Chains [1], we show that the average of SGD iterates is asymptotically normally distributed around the expected value of their unique invariant distribution, as long as the non-convex and non-smooth objective function satisfies a dissipativity property. We also characterize the bias between this expected value and the critical points of the objective function under various local regularity conditions. Together, the above two results could be leveraged to construct confidence intervals for non-convex problems that are trained using the SGD algorithm.