Sustainable and Durable Calcium Sulfoaluminate Binders Enabled by Multi-Physics Characterization and Theory-Guided Machine Learning

在全球范围内，波特兰水泥（PC）混凝土是使用最广泛的材料之一。虽然基础设施的不断发展保证了对混凝土的需求不断增加，但PC的生产消耗了大量的能源，并产生了大量对环境有害的二氧化碳。基于硫铝酸钙组合物（CSAC）的水泥已被认为是PC的可持续替代品。CSAC的生产产生的二氧化碳比PC少得多，并且可以使用低成本和广泛可用的原材料生产。然而，关于CSAC粘合剂（CSAB）的使用的研究有限。本研究旨在进行基础材料研究，以提高CSAC的利用率，为混凝土配制可持续和耐用的粘合剂。机器学习（ML）技术和由热力学指导的实验将被用来揭示驱动CSAB物理化学行为的机制和潜在的组成-性能联系。此外，CSAC和丰富的废物材料，如煅烧粘土和粉煤灰之间的协同作用，以产生CSAB，表现出上级的强度和耐久性将进行研究。该研究对推进CSAC在基础设施中的实际应用具有潜在意义。知识传播活动，包括对代表性不足的学生和女学生的培训，以及向专业人员和公众宣传可持续混凝土的社会经济影响的外联活动，将扩大这项研究的影响。研究战略围绕两个主题展开。第一个研究主题是CSAB微观结构的优化。针对这一主题，一个全面的热力学模型，对实验进行了验证，并结合理论指导的机器学习（ML）将被用来提供指导，调节化学CSAC，使CSABs与最佳相组合获得。术语“最佳”涉及以钙矾石和AFm相的最佳分布为特征的CSAB。此外，为了优化CSAB微观结构，将开发使用相关纳米晶种来增强针状钙矾石晶体的互锁和成核以及低密度AFm相的生长的新型活性微观结构设计方法。第二个研究主题将开发CSAB性能指标的实验数据集，包括大量不同的前体化学品，使用ASTM方法以及嵌入式光纤物理（例如应变传感器）和化学（例如原位拉曼光谱）传感器。CSAB性能的数据集，与前体化学，混合物设计和固化条件有关，将使用理论指导的ML进行处理。该平台将能够高保真预测CSAB的性能，使用容易测量的理化信息有关的粘合剂作为输入。ML平台最终将用于确定CSAB的最佳化学和加工条件，与PC对应物相比，CSAB具有上级强度和耐久性，同时将CSAB的CSAC含量限制在50%。通过预测和优化性能，这一努力将促进CSAB在建筑中的实际应用。该奖项反映了NSF的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

On a global scale, Portland Cement (PC) concrete is one of the most widely used material. While the continual development of infrastructure warrants that demand for concrete is increasing, the production of PC consumes considerable energy and produces large amount of carbon dioxide that is detrimental to the environment. Cements based on calcium sulfoaluminate compositions (CSACs) have been considered as sustainable alternatives to PC. Production of CSAC yields significantly less carbon dioxide than PC and can be produced using low-cost and widely-available raw materials. Studies on use of CSAC binders (CSAB), however, are limited. This research aims to pursue fundamental material research to enhance the utilization of CSAC to formulate sustainable and durable binders for concrete. Techniques of machine learning (ML) and experiments that are guided by thermodynamics will be employed to reveal mechanisms, and underlying composition-performance links, that drive the physicochemical behavior of CSABs. In addition, synergistic interactions between CSAC and abundant waste materials such as calcined clay and fly ash to produce CSABs that exhibit superior strength and durability will be investigated. This study has potential of advancing practical utilization of CSAC for infrastructure. Knowledge dissemination activities, including training of underrepresented and female students and outreach activities to inform professionals and public of socioeconomic impacts of sustainable concrete, will extend the impact of this research. The research strategy is premised on two themes. The first research theme focuses on optimization of CSAB microstructure. Towards this theme, a comprehensive thermodynamic model that is validated against experiments, and in conjunction with theory-guided machine learning (ML) will be employed to provide guidance on regulating chemistry of CSAC, such that CSABs with optimum phase assemblages are obtained. The term 'optimum' pertains to CSABs featuring optimal distribution of both ettringite and AFm phases. Additionally, for optimization of CSAB microstructure, novel active microstructure design methods that use relevant nano-seeds to enhance inter-locking of acicular ettringite crystals and nucleation, and growth of low-density AFm phases will be developed. The second research theme will develop experimental datasets of performance metrics of CSABs encompassing large numbers of different precursor chemistries using ASTM methods as well as embedded fiber-optic physical (e.g. strain sensors) and chemical (e.g. in-situ Raman spectroscopy) sensors. The datasets of CSABs performance, in relation to precursor chemistry, mixture design and curing conditions, will be processed using theory guided ML. The platform will enable high-fidelity prediction of CSAB performance, using readily measured physicochemical information pertaining to the binder as inputs. The ML platform will ultimately be leveraged to determine optimum chemistry and processing conditions of CSABs that exhibit superior strength and durability compared to their PC counterparts, while restricting the CSAC content of CSABs to 50%. By enabling prediction and optimization of performance, this effort will stimulate practical utilization of CSABs in construction.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.