CSEDI: Searching For Hadean Impacts: Clues From the Sudbury Impact Basin and Machine Learning Approaches

该奖项全部或部分根据2021年美国救援计划法案（公法117-2）资助。地壳形成机制提供了有关行星环境的重要信息。例如，如果地壳在冥古宙（45 - 40亿年前）以与现代地球相同的方式形成，这可能表明早期地球有板块构造。这对早期地球的环境有着重要的影响，因为板块构造促进了挥发性循环和化学风化，这反过来又为生命提供了关键成分。然而，也有可能早期地壳的很大一部分是由其他机制形成的，例如陨石撞击形成的岩浆结晶。这个假说并不要求早期地球有板块构造。关于这两种机制的信息都很难获得，因为我们对40亿年前保存下来的陆地材料的访问极其有限。我们关于冥古宙地壳的大多数直接信息都是从一种叫做锆石的极耐腐蚀的矿物中获得的，其年龄接近44亿年。锆石含有微量杂质，如钛和稀土元素。它们的浓度可能受到锆石结晶的岩浆形成机制的影响，例如，与板块构造和陨石撞击有关。在这里，我们将通过多管齐下的方法研究早期地壳的形成机制，包括对冥古宙锆石中微量元素化学的详细分析。然后，这些将与预测由于陨石撞击和早期地球熔化而导致的锆石中微量元素化学的模拟进行比较。有了这些结果，我们将能够估计可能的冥古宙地壳的组成，从而其成因，由于影响有关的过程。该项目还将支持学生在现代计算方法，机器学习和现代矿物分析方法方面的培训。其中两名学生将就此项目撰写毕业论文，为学生提供参与这一前沿研究项目的机会。研究成果将在罗切斯特博物馆和科学中心（RMSC）展示，以加强当地社区的参与。这将是首次探索机器学习方法，根据微量元素丰度来表征锆石的起源。我们将用起源众所周知的最近形成的锆石来训练我们的机器学习模型，然后将该模型应用于Hadean锆石以识别它们的起源。我们建议开展以下工作：（1）我们将通过分析史密森尼学会提供的样品来表征由撞击引发的岩浆形成的锆石中的微量杂质，（2）我们将从萨德伯里撞击盆地收集岩石样品，该盆地形成于18.5亿年前，拥有地球上最大的由撞击引发的岩浆结晶而成的保存地壳，（3）我们将进行撞击模拟和熔体演化计算，并将其与萨德伯里盆地样品的全岩化学和锆石微量元素进行比较。一旦开发出化学、物理和机器学习模型，我们将把它应用于Hadean锆石数据，以确定一些锆石是否是由撞击形成的。最后，根据我们的模型和微量元素丰度在冥古宙锆石，我们将探讨冥古宙地壳的组成和其潜在的关系，影响相关的过程。我们独特而全面的方法将提供新的见解，以了解早期地球的不良约束环境。通过了解冥古宙期间的地壳形成过程，我们可以更好地限制早期地球的环境，包括地壳生产率和挥发性循环。此外，由于机器学习是该项目的关键组成部分，数据保存和归档将成为重中之重。我们计划以与现有GeoRoc数据库类似的格式存档我们分析的样本数据，该数据库将公开访问。我们将专门为这个项目和学生参与一个类将是这个拟议的工作的一个关键方面。通过课堂活动，学生将学习撞击过程，萨德伯里地质学，以及岩石（X射线荧光）和矿物（激光消融电感耦合等离子体质谱）分析中涉及的地球化学技术。该奖项反映了NSF的法定使命，并通过使用基金会的智力价值和更广泛的影响审查标准进行评估，被认为值得支持。

This award is funded in whole or in part under the American Rescue Plan Act of 2021 (Public Law 117-2). Mechanisms of crustal formation provide crucial information on the planetary environment. For example, if Earth’s crust formed during the Hadean eon (4.5-4.0 billion years ago) in the same way as the modern Earth, this might indicate that early Earth had plate tectonics. This has significant implications for the early Earth’s environment, given that plate tectonics promote volatile cycling and chemical weathering, which in turn provide key ingredients for life. However, it is also possible that a significant fraction of early Earth’s crust formed by other mechanisms, such as crystallization from magmas that formed by meteorite impacts. This hypothesis does not require early Earth to have had plate tectonics. Information about either mechanism is challenging to obtain because of our extremely limited access to preserved terrestrial materials older than 4 billion years. Most of our direct information about the Hadean crust has been obtained from the extremely resistant mineral called zircons, with ages that approach 4.4 billion years. Zircons contain trace impurities, such as titanium and rare earth elements. Their concentrations can be affected by the formation mechanisms of the magmas from which zircons crystallize, e.g., related to plate tectonics v. meteorite impacts. Here, we will investigate early Earth’s crust formation mechanisms though a multi-pronged approach that includes detailed analysis of trace element chemistry in Hadean zircon. These will then be compared with simulations that predict the trace element chemistry in zircon due to meteorite impact and melting of the early Earth. With these results, we will be able to estimate possible Hadean crust composition and hence its genesis due to impact-related processes. This project will also support students training in modern computational methods, machine learning, and modern methods of mineral analysis. Two of the students will write their senior theses on this project, providing opportunities for the students to participate in this cutting-edge research project. The outcome will be presented at the Rochester Museum and Science Center (RMSC) to enhance local community engagement.This will be the first time machine learning methodologies have been explored to characterize the origin of zircons based on their trace element abundances. We will train our machine learning model with more recently formed zircons whose origins are well known, and subsequently apply the model to the Hadean zircons to identify their origins. We propose to conduct the following tasks; (1) we will characterize trace impurities in zircons formed from impact-induced magma by analyzing samples provided by the Smithsonian Institute, (2) We will collect rock samples from the Sudbury impact basin, which formed 1.85 billion years ago and has the largest preserved crust crystallized from impact-induced magma on Earth, and (3) we will conduct impact simulations and melt evolution calculations, which will be compared with whole rock chemistry and zircon trace elements from the Sudbury basin samples. Once chemical, physical and machine learning models are developed, we will apply it to Hadean zircon data to identify whether some of the zircons formed by impact. Finally, based on our model and trace element abundances in Hadean zircons, we will explore the Hadean crust composition and its potential relation to impact-related processes. Our unique and comprehensive approach will provide new insights to understand the poorly constrained environment of the early Earth. By understanding the crust formation process during the Hadean, we can better constrain the early Earth's environments, including crust production rate and volatile cycling. Moreover, with machine learning being a key component of this project, data preservation and archiving will be a top priority. We plan to archive our analyzed sample data in a similar format as the extant GeoRoc database, which will be publicly accessible. We will dedicate one class for this project and student participation will be a key aspect of this proposed work. Through the class activity, the students will learn impact processes, Sudbury geology, and the geochemical techniques involved in rock (x-ray fluorescence) and mineral (laser ablation inductively coupled plasma mass spectrometry) analysis.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.