It has been twenty-eight years since the first metal–organic framework (MOF) was crystallized using transition metal-ions (Co²⁺) and negatively charged organic linkers [trimesate = 1,3,5-C₆H₃(COO)₃^3–]. This was followed by proof of permanent porosity in MOF-2 and ultrahigh porosity in MOF-5. These contributions opened the door to what has become one of the largest fields of chemistry, MOF chemistry, where it is now common practice for a researcher to imagine a structure and identify the molecular building blocks which could be linked into that structure. Overcoming the crystallization challenge by designing the appropriate conditions for the crystallization of MOFs was a dream come true for chemists, especially since it allowed for definitive characterization of MOFs using X-ray and more recently electron diffraction techniques. The strong metal-charged linker bonds holding MOFs together ensured architectural and chemical robustness, thereby giving rise to postsynthetic modifications of MOFs without loss of their permanent porosity and crystallinity. In essence a MOF is an extremely large molecule, which can be treated with chemical reactions as one would with a molecule in solution, except the MOF “molecule” encompasses space within which matter can be further controlled and transformed. The vast number of possible MOF structures and the flexibility of modifying them coupled with their many applications in climate, environment, energy, water, and health support the fact that this chemistry is vast. What is the future of MOFs and how do we harness the power and full potential of this new chemistry? We contend that at least part of the future of MOFs will involve developments in three directions. First, multivariable MOFs: MOFs are intrinsically chemical in nature and accordingly the concepts and techniques of molecular chemistry will continue to be of interest in the study of MOFs. The best way to appreciate this aspect in a MOF is through the “eye of the molecule”: imagine a molecule passing through the pores and observing events taking place inside the MOFs. Such events include binding of substrates (e.g., carbon dioxide and water) to framework atoms, reacting substrates on catalytic metal or organic sites which are integral to the frameworks, growing polymers within or from the pores to improve the processability of MOFs, and so on. In addition to these “bread and butter” type pursuits and the resulting useful chemistry, an important opportunity emerges from the recognition that the interior of the MOF is a highly heterogeneous environment especially when chemistry is taking place within the pores. For example, a reaction carried out within the MOF (postsynthesis), let us say involving an organic functionality bound to the backbone, creates a heterogeneous spatial arrangement (multivariate) of the reacted and unreacted functionalities. In a multistep postsynthetic modification of a MOF, this heterogeneity is exponentially increased with the number of functionalities and therein lies the question and opportunity: what spatial arrangement is underlying this heterogeneity and could it be used to create new chemistry. Preliminary evidence shows that such multivariable MOFs outperform their “pure, homogeneous” analogues in affecting programmed drug release, highly selective separations, and catalysis akin to those carried out by enzymes. Almost every aspect of MOF chemistry yields multivariability and therefore opportunities to describe, study, and decipher multivariate MOFs, and the spatial sequences underlying such systems will continue to become important for applications and for growing the basic science of the field. Could such sequences be designed, without losing their multivariate nature, and could their chemistry be controlled? Second, the emerging MOF innovation cycle: MOFs have allowed us to develop a cycle of innovation in which molecules are linked into frameworks, engineered in a form factor, and integrated into devices, which demonstrate impact on society. Such an innovation cycle requires knowledge of chemistry and engineering to ensure the full functioning of the MOF and the device, for example, in capturing carbon dioxide or water harvesting from air. Once the MOF is integrated into the device to be maximally exposed to incoming guests, consideration of air flow, heat, and mass transfer become paramount for achieving the highest efficiency under real world operating conditions. In such an innovation cycle a close link is created between the MOF and the device performance, which can be optimized well because of the designability of the MOF and the correlation between molecular based behavior and device performance. This was recently illustrated in water harvesting using a MOF in which two carbon atoms were added to a linker to produce a MOF with larger pore volume delivering 50% more water than the state-of-the-art MOF. Such close connection between the molecular design and performance can also allow for lower desorption temperature and harvesting water at low humidity. The elements of the MOF innovation cycle, being entirely implemented in a chemistry laboratory, also provide for better, more current, and more relevant education of our researchers. Third, digital reticular chemistry: Computational chemistry has emerged as an integral component of modern chemistry with the advances of digital computing. Its significance in reticular chemistry lies in its ability to determine structures at the molecular level and forecast novel structures that exhibit enhanced capabilities across various applications. In this field, a wide array of computational methodologies is employed, encompassing electronic structure calculations predominantly employing density functional theory (DFT), as well as classical molecular dynamics (MD) and Monte Carlo (MC) simulations. These methodologies enable researchers to investigate and determine energy landscapes, charge transport, adsorption and gas separation, guest–host interactions, thermodynamic properties, etc. It is indeed crucial to integrate computational calculations with experimental information and data science techniques to accelerate the discovery process in MOF chemistry. However, it is equally important to recognize the limitations of these computational approaches. The reliability and accuracy of computer simulations heavily rely on the chosen method and model. Therefore, it is essential to establish physically motivated models that appropriately represent the system under study. For example, when performing electronic structure calculations on a catalyst supported on a MOF node, a decision needs to be made regarding its representation either as a single atom catalyst or a nanoparticle. This choice may not always be informed by experimental data. Additionally, the catalyst structure might undergo changes along the reaction pathway. In the case of metal-based catalysis, it is not uncommon for the nuclearity and spin state of the metal to vary during the reaction. These factors should be considered when designing computational models to accurately capture the catalytic processes. Similarly, in MD and MC simulations, the presence of a guest molecule can lead to structural changes in the MOF framework. It is crucial to account for these changes in the simulations or, at the very least, be aware that the model system used may not perfectly represent the real system. In such cases, focusing on trends resulting from minor structural modifications may be more insightful than attempting to predict absolute values of specific properties. By acknowledging these limitations and utilizing computational methods in conjunction with experimental data, researchers can gain valuable insights into MOF chemistry, identify trends, and make informed predictions. This interdisciplinary approach, combining theory, experiment, and data science, can significantly enhance the discovery and development of MOF materials with tailored properties and functionalities. The integration of AI into computational chemistry has the potential to revolutionize the field, not only in reticular chemistry but also in the wider chemistry. However, it is essential to acknowledge that AI operates within the chemical space in which it is trained and does not reliably extrapolate beyond that space. Humans must still make critical decisions on when to introduce drastic changes in the exploration process, guided by their domain knowledge and expertise. An important aspect that will shape the future of AI in computational chemistry is the reproducibility of computations and AI-generated results. Researchers must strive to make their data and algorithms fully available to the scientific community, ensuring transparency and facilitating the verification and validation of computational findings. By leveraging the power of AI and embracing a culture of reproducibility, the integration of computation and AI in reticular chemistry and chemistry at large can lead to transformative advancements, accelerating discovery, optimizing performance, and fostering collaborations within the scientific community. Imagine being able to use Learning Language Models (LLMs) such as GPT-4 to mine information reliably and tabulate for example the reaction conditions under which MOFs could be made and crystallized or using machine learning algorithms to predict new MOF and correlate their structures with specific properties and applications. Imagine being able to tabulate the information on all the calculations ever performed on MOFs and compute, with the aid of ChatGPT, at different scales, the microscopic as well as the macroscopic properties of the MOF innovation cycle. We believe that the fast-emerging field of AI for science will have a revolutionary impact on chemistry and that reticular chemistry is ideally suited, because of its modularity and definition, to exemplify the power of AI in automating the MOF innovation cycle in its entirety. This is being facilitated by LLMs to provide easy entry for researchers into this vast and yet untapped reticular potential. Let us face it, chemistry needs to catch up with the times and it needs to become more sustainable and impactful on society. For the first time, we have an opportunity to use AI to be more relevant and more widely sought after as scientists. As chemistry is about the study of change so should we have the ability to change as chemists. It is a matter of survival rather than convenience. Reticular chemistry is an arena ripe for this change which we’re delighted to say is just beginning to show its impact. Indeed, the advancements in computations and AI have the potential to democratize the field of reticular chemistry and make it more accessible to researchers worldwide. With the aid of computational tools, apps, AI, and LLMs, researchers can model and simulate reticular systems more easily and efficiently. These technological advancements enable researchers to explore and analyze MOFs, covalent organic frameworkss, and other reticular structures regardless of their geographical location. The availability of computational tools, AI algorithms, and LLMs allows for the creation of a networked community where information can be shared, collaborations can be fostered, and knowledge can be collectively built. The accessibility of reticular chemistry through computations and AI empowers researchers to contribute to the field irrespective of their resources or physical proximity to specialized facilities. This inclusivity can lead to diverse perspectives, faster knowledge dissemination, and accelerated progress in reticular chemistry research. This article is cited by 1 publications.

中文翻译：

---

金属有机框架的三个未来方向

自从使用过渡金属离子 (Co ²⁺ ) 和带负电荷的有机连接体 [均苯三酸 = 1,3,5-C ₆ H ₃ (COO)结晶第一个金属有机框架 (MOF) 以来，已有二十八年了₃ ^3–]。随后证明了 MOF-2 的永久孔隙率和 MOF-5 的超高孔隙率。这些贡献为 MOF 化学这一最大的化学领域打开了大门，研究人员现在通常的做法是想象一个结构并识别可以连接到该结构的分子构建块。通过设计适当的 MOF 结晶条件来克服结晶挑战对于化学家来说是梦想成真，特别是因为它允许使用 X 射线和最近的电子衍射技术对 MOF 进行明确的表征。将 MOF 固定在一起的强金属电荷连接键确保了结构和化学稳定性，从而在不损失其永久孔隙度和结晶度的情况下对 MOF 进行合成后修饰。本质上，MOF 是一种非常大的分子，可以像处理溶液中的分子一样通过化学反应进行处理，不同之处在于 MOF“分子”包含可以进一步控制和转化物质的空间。大量可能的 MOF 结构及其修改的灵活性，再加上它们在气候、环境、能源、水和健康方面的许多应用，证明了这种化学作用是巨大的。MOF 的未来是什么？我们如何利用这种新化学物质的力量和全部潜力？我们认为，MOF 的未来至少有一部分将涉及三个方向的发展。它可以像处理溶液中的分子一样通过化学反应进行处理，不同之处在于 MOF“分子”包含可以进一步控制和转化物质的空间。大量可能的 MOF 结构及其修改的灵活性，再加上它们在气候、环境、能源、水和健康方面的许多应用，证明了这种化学作用是巨大的。MOF 的未来是什么？我们如何利用这种新化学物质的力量和全部潜力？我们认为，MOF 的未来至少有一部分将涉及三个方向的发展。它可以像处理溶液中的分子一样通过化学反应进行处理，不同之处在于 MOF“分子”包含可以进一步控制和转化物质的空间。大量可能的 MOF 结构及其修改的灵活性，再加上它们在气候、环境、能源、水和健康方面的许多应用，证明了这种化学作用是巨大的。MOF 的未来是什么？我们如何利用这种新化学物质的力量和全部潜力？我们认为，MOF 的未来至少有一部分将涉及三个方向的发展。大量可能的 MOF 结构及其修改的灵活性，再加上它们在气候、环境、能源、水和健康方面的许多应用，证明了这种化学作用是巨大的。MOF 的未来是什么？我们如何利用这种新化学物质的力量和全部潜力？我们认为，MOF 的未来至少有一部分将涉及三个方向的发展。大量可能的 MOF 结构以及对其进行修改的灵活性，再加上它们在气候、环境、能源、水和健康方面的许多应用，都证明了这种化学作用是巨大的。MOF 的未来是什么？我们如何利用这种新化学物质的力量和全部潜力？我们认为，MOF 的未来至少有一部分将涉及三个方向的发展。首先，多变量 MOF：MOF 本质上是化学的，因此分子化学的概念和技术将继续成为 MOF 研究的焦点。理解 MOF 中这一方面的最佳方法是通过“分子之眼”：想象一个分子穿过孔隙并观察 MOF 内部发生的事件。此类事件包括底物（例如二氧化碳和水）与框架原子的结合、底物在与框架不可分割的催化金属或有机位点上反应、在孔内或从孔中生长聚合物以提高MOF的可加工性等。除了这些“面包和黄油”类型的追求以及由此产生的有用的化学反应，认识到 MOF 内部是一个高度异质的环境，特别是当化学反应在孔内发生时，出现了一个重要的机会。例如，在 MOF（后合成）内进行的反应，假设涉及与主链结合的有机官能团，会产生反应和未反应官能团的异质空间排列（多元）。在 MOF 的多步合成后修饰中，这种异质性随着功能数量的增加呈指数级增加，其中存在问题和机遇：这种异质性背后的空间排列是什么以及它是否可以用于创造新的化学。初步证据表明，这种多变量 MOF 在影响程序化药物释放方面优于其“纯、均质”类似物，高度选择性的分离和类似于酶进行的催化。MOF 化学的几乎每个方面都会产生多变量，因此有机会描述、研究和破译多变量 MOF，并且此类系统背后的空间序列将继续对该领域的应用和基础科学的发展变得重要。是否可以在不失去其多元性质的情况下设计这样的序列，并且可以控制它们的化学性质吗？二、新兴的MOF创新周期：MOF 使我们能够开发出一个创新周期，其中分子被连接到框架中，以某种形状因数进行设计，然后集成到设备中，这对社会产生了影响。这样的创新周期需要化学和工程知识，以确保 MOF 和设备的充分发挥作用，例如捕获二氧化碳或从空气中收集水。一旦 MOF 集成到设备中以最大程度地暴露于即将到来的客人，为了在现实操作条件下实现最高效率，对气流、热量和传质的考虑就变得至关重要。在这样的创新周期中，MOF 和器件性能之间建立了紧密的联系，由于 MOF 的可设计性以及分子行为与器件性能之间的相关性，可以很好地进行优化。最近在使用 MOF 进行水收集时说明了这一点，其中将两个碳原子添加到连接体中，产生具有更大孔体积的 MOF，比最先进的 MOF 多提供 50% 的水。分子设计和性能之间的这种紧密联系还可以实现较低的解吸温度并在低湿度下收集水。MOF 创新周期的要素完全在化学实验室中实施，也为我们的研究人员提供了更好、更最新、更相关的教育。最近在使用 MOF 进行水收集时说明了这一点，其中将两个碳原子添加到连接体中，产生具有更大孔体积的 MOF，比最先进的 MOF 多提供 50% 的水。分子设计和性能之间的这种紧密联系还可以实现较低的解吸温度并在低湿度下收集水。MOF 创新周期的要素完全在化学实验室中实施，也为我们的研究人员提供了更好、更最新、更相关的教育。最近在使用 MOF 进行水收集时说明了这一点，其中将两个碳原子添加到连接体中，产生具有更大孔体积的 MOF，比最先进的 MOF 多提供 50% 的水。分子设计和性能之间的这种紧密联系还可以实现较低的解吸温度并在低湿度下收集水。MOF 创新周期的要素完全在化学实验室中实施，也为我们的研究人员提供了更好、更最新、更相关的教育。三、数字网状化学：随着数字计算的进步，计算化学已成为现代化学不可或缺的组成部分。它在网状化学中的重要性在于它能够确定分子水平的结构并预测在各种应用中表现出增强功能的新颖结构。在该领域，采用了多种计算方法，包括主要采用密度泛函理论（DFT）的电子结构计算，以及经典分子动力学（MD）和蒙特卡罗（MC）模拟。这些方法使研究人员能够研究和确定能量景观、电荷传输、吸附和气体分离、客体相互作用、热力学性质等。将计算计算与实验信息和数据科学技术相结合对于加速 MOF 化学的发现过程确实至关重要。然而，认识到这些计算方法的局限性同样重要。计算机模拟的可靠性和准确性在很大程度上取决于所选择的方法和模型。因此，有必要建立适当代表所研究系统的物理驱动模型。例如，当对 MOF 节点上支持的催化剂进行电子结构计算时，需要决定其表示形式是单原子催化剂还是纳米颗粒。这种选择可能并不总是由实验数据决定。此外，催化剂结构可能沿着反应路径发生变化。在金属基催化的情况下，金属的核和自旋态在反应过程中发生变化并不罕见。在设计计算模型以准确捕获催化过程时应考虑这些因素。同样，在 MD 和 MC 模拟中，客体分子的存在会导致 MOF 框架的结构变化。在模拟中考虑这些变化至关重要，或者至少要意识到所使用的模型系统可能无法完美地代表真实系统。在这种情况下，关注微小结构修改所产生的趋势可能比尝试预测特定属性的绝对值更具洞察力。通过承认这些局限性并结合实验数据利用计算方法，研究人员可以获得有关 MOF 化学的宝贵见解、确定趋势并做出明智的预测。这种跨学科方法结合了理论、实验和数据科学，可以显着增强具有定制特性和功能的 MOF 材料的发现和开发。人工智能与计算化学的整合有可能彻底改变该领域，不仅在网状化学领域，而且在更广泛的化学领域。然而，必须承认人工智能是在其接受训练的化学空间内运行的，并且不能可靠地推断出该空间之外的情况。人类仍然必须在其领域知识和专业知识的指导下，就何时在探索过程中引入重大变化做出关键决策。塑造人工智能在计算化学领域的未来的一个重要方面是计算和人工智能生成结果的可重复性。研究人员必须努力将他们的数据和算法完全提供给科学界，确保透明度并促进计算结果的验证和确认。通过利用人工智能的力量并拥抱可重复性文化，计算和人工智能在网状化学和整个化学中的集成可以带来变革性的进步，加速发现，优化性能，并促进科学界内的合作。想象一下，能够使用 GPT-4 等学习语言模型 (LLM) 可靠地挖掘信息，并列出 MOF 制造和结晶的反应条件等，或者使用机器学习算法来预测新的 MOF 并将其结构与特定的相关联。特性和应用。想象一下，能够将在 MOF 上执行的所有计算的信息制成表格，并借助 ChatGPT 在不同尺度上计算 MOF 创新周期的微观和宏观特性。我们相信，快速新兴的科学人工智能领域将对化学产生革命性影响，而网状化学由于其模块化和定义，非常适合体现人工智能在整个 MOF 创新周期自动化方面的力量。法学硕士正在推动这一点，为研究人员提供轻松进入这一巨大但尚未开发的网状潜力的机会。让我们面对现实吧，化学需要跟上时代，它需要变得更加可持续和对社会产生影响。作为科学家，我们第一次有机会利用人工智能变得更加相关并受到更广泛的追捧。由于化学是关于变化的研究，所以我们作为化学家应该有能力做出改变。这是生存问题而不是便利问题。网状化学是这一变化的成熟舞台，我们很高兴地说，它的影响才刚刚开始显现。事实上，计算和人工智能的进步有可能使网状化学领域民主化，并使全世界的研究人员更容易接触到它。借助计算工具、应用程序、人工智能和法学硕士，研究人员可以更轻松、更高效地对网状系统进行建模和仿真。这些技术进步使研究人员能够探索和分析 MOF、共价有机框架和其他网状结构，无论其地理位置如何。计算工具、人工智能算法和法学硕士的可用性允许创建一个网络社区，在这个社区中可以共享信息，可以促进合作，可以共同构建知识。通过计算和人工智能实现网状化学的可访问性使研究人员能够为该领域做出贡献，无论他们的资源如何或与专业设施的物理距离如何。这种包容性可以带来多样化的观点、更快的知识传播并加速网状化学研究的进展。