Search for: All records

Large Language Models (LLMs) with strong abilities in natural language processing tasks have emerged and have been applied in various kinds of areas such as science, finance and software engineering. However, the capability of LLMs to advance the field of chemistry remains unclear. In this paper, rather than pursuing state-of-the-art performance, we aim to evaluate capabilities of LLMs in a wide range of tasks across the chemistry domain. We identify three key chemistryrelated capabilities including understanding, reasoning and explaining to explore in LLMs and establish a benchmark containing eight chemistry tasks. Our analysis draws on widely recognized datasets facilitating a broad exploration of the capacities of LLMs within the context of practical chemistry. Five LLMs (GPT-4, GPT-3.5, Davinci-003, Llama and Galactica) are evaluated for each chemistry task in zero-shot and few-shot in-context learning settings with carefully selected demonstration examples and specially crafted prompts. Our investigation found that GPT-4 outperformed other models and LLMs exhibit different competitive levels in eight chemistry tasks. In addition to the key findings from the comprehensive benchmark analysis, our work provides insights into the limitation of current LLMs and the impact of in-context learning settings on LLMs’ performance across various chemistry tasks. The code and datasets used in this study are available at https://github.com/ChemFoundationModels/ChemLLMBench. 

Dissolved organic nitrogen (DON) is the dominant form of bioavailable nitrogen in the euphotic zone of subtropical gyres, where nitrate (NO3-) concentrations are low. However, the spatial distribution of DON production and consumption in the surface ocean remains poorly resolved due to the relatively narrow range in euphotic zone DON concentrations. Recently, the stable isotopic composition (d15N) of DON has been used to identify DON production and consumption in the surface ocean, making isotopic measurements a more sensitive indicator of DON cycling than concentration measurements alone. Here we report DON concentration and d15N measurements in the upper ~300 m from a zonal transect along ~30˚S in the South Pacific (GO-SHIP P06-2017), including samples in the Western South Pacific (154˚E-170˚W), in the oligotrophic South Pacific Subtropical Gyre (110˚W -170˚W), and overlying the Oxygen Deficient Zone (ODZ) in the east (78˚W-110˚W). We observed small variations in surface DON concentrations. Surface DON in Western South Pacific, oligotrophic South Pacific Subtropical Gyre and above the ODZ are 4.6±1.0 µM, 4.3±0.7 µM, and 4.8±0.5 µM, respectively. d15N of DON in the euphotic zone is lower in the west and higher in the east, consistent with distributions of nitrogen fixation and denitrification, respectively, in the South Pacific. Similar decreasing trend in DON d15N in the euphotic zone and subsurface nitrate d15N was observed from the east to the west in the South Pacific, suggesting the d15N in subsurface nitrate could be imprinted in the DON d15N in the euphotic zone. Low surface ocean DON d15N in the Western South Pacific (2.4±1.8 ‰) and oligotrophic South Pacific Subtropical Gyre (2.6±1.6 ‰) compared with surface ocean DON d15N above ODZ (5.4±2.3 ‰) infer significant low-d15N nitrogen is added to the western South Pacific and oligotrophic South Pacific Subtropical Gyre, potentially from N2 fixation. Additionally, high DON d15N at ~180˚ was consistent with entrainment of subsurface NO3- into surface waters due to shallow bathymetry. Together, these observations suggest that DON production and consumption processes operate on timescales adequately fast to produce isotopic gradients across the South Pacific. Comparisons of surface ocean DON d15N with subsurface nitrate d15N constrain the locations and timescales of these processes. 

Significant rates of export production and nitrogen fixation occur in oligotrophic gyres in spite of low inorganic nutrient concentrations in surface waters. Prior work suggests that dissolved organic nitrogen (DON) and dissolved organic phosphorus (DOP) are important nutrient sources when inorganic nutrients are scarce. In particular, DOP has been shown to be an important P source for diazotrophs, which may be better suited to using low concentrations of organic vs. inorganic P. Prior modeling work has also suggested that DOP is important for supporting export production in oligotrophic gyres. However, validation of such models is limited by the number of upper ocean DOP concentration measurements, especially in the South Pacific and Indian Oceans. Here, we present measurements of DOP concentration from the 2016 GO-SHIP I08S and I09N meridional transect in Eastern Indian Ocean, and DON and DOP concentration measurements from the 2017 GO-SHIP P06 zonal transect in the subtropical South Pacific Ocean. Together with DOC and DON concentration measurements from prior occupations of the same GO-SHIP lines we evaluated changes in euphotic zone DOC:DON:DOP stoichiometry. Stoichiometry changes across these two transects are used to infer regions of preferential DON and/or DOP production and consumption. Specifically, north of 36 S in the Indian Ocean an increase in DOC:DON and DOC:DOP concentration ratios, from 11:1 to 14:1 and 118:1 to 190:1, respectively, are observed. Similarly, west of 136 W in the South Pacific Ocean significant increases in DOC:DOP and DON:DOP concentration ratios are observed, from 224:1 to 398:1 and 21:1 to 39:1, respectively. These stoichiometric shifts in upper ocean DOC:DON:DOP concentration ratios are considered in the context of ocean circulation, especially upwelling patterns in the Indian and eastern Pacific Oceans, as well as prior observations of the distribution of nitrogen fixation, especially in the western tropical South Pacific. 

The alpha-hydroxyalkyl-hydroperoxides [R-(H)C(-OH)(-OOH), alpha-HH] produced in the ozonolysis of unsaturated organic compounds may contribute to secondary organic aerosol (SOA) aging. alpha-HHs' inherent instability, however, hampers their detection and a positive assessment of their actual role. Here we report, for the first time, the rates and products of the decomposition of the alpha-HHs generated in the ozonolysis of atmospherically important monoterpenes alpha-pinene (alpha-P), d-limonene (d-L), gamma-terpinene (gamma-Tn), and alpha-terpineol (alpha-Tp) in water/acetonitrile (W/AN) mixtures. We detect alpha-HHs and multifunctional decomposition products as chloride adducts by online electrospray ionization mass spectrometry. Experiments involving D2O and (H2O)-O-18, instead of (H2O)-O-16, and an OH-radical scavenger show that alpha-HHs decompose into gem-diols + H2O2 rather than free radicals. alpha-HHs decay mono- or biexponentially depending on molecular structure and solvent composition. e-Fold times, tau(1)(/e), in water-rich solvent mixtures range from tau(1)(/e), = 15-45 min for monoterpene-derived alpha-HHs to tau(1)(/e) > 10(3) min for the alpha-Tp-derived alpha-HH. All tau(1)(/e)'s dramatically increase in <20% (v/v) water. Decay rates of the alpha-Tp-derived alpha-HH in pure water increase at lower pH (2.3 <= pH <= 3.3). The hydroperoxides detected in day-old SOA samples may reflect their increased stability in water-poor media and/or the slow decomposition of alpha-HHs from functionalized terpenes.