Search for: All records

Per‐ and poly‐fluoroalkyl substances (PFAS) are interfacially‐active contaminants that adsorb at air‐water interfaces (AWIs). Water‐unsaturated soils have abundant AWIs, which generally consist of two types: one is associated with the pendular rings of water between soil grains (i.e., bulk AWI) and the other arises from the thin water films covering the soil grains. To date, the two types of AWIs have been treated the same when modeling PFAS retention in vadose zones. However, the presence of electrical double layers of soil grain surfaces and the subsequently modified chemical potential of PFAS at the AWI may significantly change the PFAS adsorption at the thin‐water‐film AWI relative to that at the bulk AWI. Given that thin water films contribute to over 90% of AWIs in the vadose zone under many field‐relevant wetting conditions, it is critical to quantify the potential anomalous adsorption of PFAS at the thin‐water‐film AWI. We develop a thermodynamic‐based mathematical model to quantify this anomalous adsorption. The model couples the chemical equilibrium of PFAS with the Poisson‐Boltzmann equation that governs the distribution of electrical potential in a thin water film. Our model analyses suggest that PFAS adsorption at thin‐water‐film AWI can deviate significantly (up to 82%) from that at bulk AWIs. The deviation increases for lower porewater ionic strength, thinner water film, and higher soil grain surface charge. These results highlight the importance of accounting for the anomalous adsorption of PFAS at the thin‐water‐film AWI when modeling PFAS fate and transport in the vadose zone.

 

With the advent of pre-trained language models (LMs), increasing research efforts have been focusing on infusing commonsense and domain-specific knowledge to prepare LMs for downstream tasks. These works attempt to leverage knowledge graphs, the de facto standard of symbolic knowledge representation, along with pre-trained LMs. While existing approaches leverage external knowledge, it remains an open question how to jointly incorporate knowledge graphs represented in varying contexts — from local (e.g., sentence), document-level, to global knowledge, to enable knowledge-rich and interpretable exchange across contexts. In addition, incorporating varying contexts can especially benefit long document understanding tasks that leverage pre-trained LMs, typically bounded by the input sequence length. In light of these challenges, we propose KALM, a language model that jointly leverages knowledge in local, document-level, and global contexts for long document understanding. KALM firstly encodes long documents and knowledge graphs into the three knowledge-aware context representations. KALM then processes each context with context-specific layers. These context-specific layers are followed by a ContextFusion layer that facilitates knowledge exchange to derive an overarching document representation. Extensive experiments demonstrate that KALM achieves state-of-the-art performance on three long document understanding tasks across 6 datasets/settings. Further analyses reveal that the three knowledge-aware contexts are complementary and they all contribute to model performance, while the importance and information exchange patterns of different contexts vary on different tasks and datasets.