A common experience for researchers in experimental laboratories focused on energy production, generation, or consumption is performing repetitive functions. This could be the synthesis of a library of materials with variation in composition or processing conditions, or the analysis of system performance under varied environmental conditions. In some cases, for example, charging and discharging of batteries or gas chromatography-based compositional analysis, these repetitive operations are performed automatically by standard equipment found in academic laboratories. In other cases, repetitive functions are performed manually by researchers. For example, heterogeneous catalyst and photocatalyst performance testing involves in situ pretreatment via oxidative and/or reductive environments, variations in temperature, reactant composition, photon flux or wavelength, and simultaneous product analysis by analytical equipment, all which require some level of system actuation to implement. In industrial settings, automation of lab functions such as these is common—either by commercially available equipment or through construction by an in-house automation specialist. The increase in data reliability, productivity, and safety that come with automation, and of course the current need for “social distancing”, suggest that academic researchers would benefit from spending more time on automating certain repetitive functions. The automation of experimental lab research can be applied with varying levels of control over a research effort (Figure 1). Automated, high-throughput screening for materials and chemical synthesis discovery has been an area of interest for the last few decades.(1−3) High-throughput, automated experimental analyses are now being coupled with theoretical guidance and approaches like machine learning to steer the search for new materials and synthetic pathways.(4,5) For example, recently, the use of a mobile, functional robot carrying out an unsupervised optimization of photocatalytic materials for hydrogen evolution was demonstrated.(6) The idea of an automated researcher, rather than automated equipment, is certainly appealing for performance optimization, and this demonstration is impressive. However, this is likely not a suitable approach for training Ph.D. students and postdocs to develop fundamental insights and identify potentially new technologies, as it removes the hands-on “tinkering” that is critical to develop intuition. Further, many interesting discoveries have come from making “mistakes” and identifying that the mistakes produced interesting results. Figure 1. Automation in chemical laboratory. For Ph.D. students and postdoctoral researchers, the ability to be original, creative, and adaptive in experimental design is key to new discoveries and also a part of the fun and value of educational training. Making sure measurements are reliable and repeatable is equally important.(7) Thus, automating lab functions that will be implemented regardless of regular “tweaks” to the experimental design to avoid rigidity is likely very valuable.(8−10) A simple example is the automation of spectra collection or variations in environmental conditions of a chemical reaction at controlled time intervals. Automating these tasks will certainly increase data quality and reproducibility compared to the same tasks being done manually. Another important consideration for automating academic research laboratories is maintaining minimal costs. Various approaches exist and can be implemented cheaply by integrating actuators and data acquisition/communication platforms, which are now available in open-source formats. For example, switching between various gas feeds in a process can be achieved with pneumatic valves that are easily timed and automated. The widespread capability of most undergraduate students to program in languages like python further enables the development of relatively simple automation algorithms. The opportunity to have lab functions running at all times with the ability to remotely control environmental conditions and analysis is exciting. The tools to implement this level of automation have been available for decades. With cheap, open-source hardware and software platforms and countless internet forums on automation, academic laboratories can implement advantageous automation easily and cheaply, and researchers should not be afraid to try! In addition, the introduction of courses that teach automation in chemical and materials science experimentation is likely an important addition to current curricula. For example, my colleague Prof. Mike Gordon at University of California, Santa Barbara recently introduced a class entitled Mechatronics and Instrumentation for Chemical Engineers teaching exactly these skills. While most academic laboratories will likely not be operated with robots moving around the lab in the near future, it is critical to think of what routine operations could be automated to allow researchers to read, think, and process all the extra data they are collecting. This may also allow researchers to enjoy experiments that do not produce the expected outcomes, because more time can be devoted to thinking about what the results tell them.(11) It is likely that regardless of if the next career stage for a researcher is academic, or at a start-up company, or at more mature company, the knowledge and confidence in implementing automation approaches will prove useful. Views expressed in this editorial are those of the author and not necessarily the views of the ACS. This article references 11 other publications.

中文翻译：

---

自动化学术实验室：提高可靠性，生产率和安全性

在实验实验室中，研究人员着重于能源生产，发电或消耗的一个普遍经验是执行重复功能。这可以是合成组成或处理条件不同的材料库，也可以是在变化的环境条件下分析系统性能。在某些情况下，例如，电池的充电或放电或基于气相色谱的成分分析，这些重复操作由学术实验室中的标准设备自动执行。在其他情况下，重复功能由研究人员手动执行。例如，非均相催化剂和光催化剂性能测试涉及原位通过氧化和/或还原环境进行预处理，温度，反应物组成，光子通量或波长的变化，以及通过分析设备进行的同时产品分析，所有这些都需要一定水平的系统驱动来实现。在工业环境中，实验室功能的自动化是很常见的-通过商用设备或内部自动化专家进行构建。自动化带来的数据可靠性，生产率和安全性的提高，当然还有当前对“社会距离”的需求，这表明，学术研究人员将从花更多的时间来实现某些重复功能的自动化中受益。实验实验室研究的自动化可以应用于对研究工作的不同级别的控制（图1）。自动化 在过去的几十年中，材料和化学合成发现的高通量筛选一直是人们关注的领域。（1-3）高通量的自动化实验分析现已与理论指导和诸如机器学习的方法相结合，从而引导搜索（4,5）例如，最近，展示了使用可移动的功能型机器人对光催化材料进行无监督优化以释放氢的方法。（6）自动化研究人员的想法当然，除了自动化设备之外，它对于性能优化也具有吸引力，并且这种展示令人印象深刻。但是，这可能不是培训博士的合适方法。学生和博士后，以发展基本见识并确定潜在的新技术，因为它消除了对发展直觉至关重要的动手“修补”。此外，通过做出“错误”并确定错误产生了有趣的结果而获得了许多有趣的发现。图1.化学实验室的自动化。对于博士 对于学生和博士后研究人员而言，具有独创性，创新性和适应性的实验设计能力是新发现的关键，也是教育培训的乐趣和价值的一部分。确保测量结果的可靠性和可重复性同样重要。（7）因此，自动化实验室功能（不考虑对实验设计的常规“调整”）的刚性，可能会非常有价值。（8-10）一个简单的例子是在受控的时间间隔内自动进行光谱收集或化学反应的环境条件变化。与手动完成的相同任务相比，自动化这些任务无疑会提高数据质量和可重复性。使学术研究实验室自动化的另一个重要考虑因素是保持最低成本。存在各种方法，并且可以通过集成执行器和数据采集/通信平台来廉价地实现这些方法，现在可以以开源格式使用它们。例如，可以通过易于定时和自动化的气动阀实现过程中各种气体进给之间的切换。大多数本科生使用python等语言进行编程的广泛能力进一步推动了相对简单的自动化算法的开发。能够始终运行实验室功能并能够远程控制环境条件和分析的机会令人兴奋。实现这种自动化水平的工具已有数十年的历史了。便宜的 开放源代码的硬件和软件平台以及无数的自动化论坛，学术实验室可以轻松，廉价地实现有利的自动化，并且研究人员不必害怕尝试！此外，引入课程教授化学和材料科学实验自动化的课程可能是当前课程的重要补充。例如，我的同事，美国加州大学圣塔芭芭拉分校的Mike Gordon教授最近开设了一个名为“机电一体化与化学工程师的课程”，专门教授这些技能。尽管在不久的将来大多数学术实验室可能不会使用机器人在实验室中移动来进行操作，但考虑到可以自动执行哪些常规操作以允许研究人员阅读，思考，并处理他们收集的所有额外数据。这也可能使研究人员享受无法产生预期结果的实验​​，因为可以将更多的时间用于思考结果对他们的意义。（11）不管研究人员的下一个职业阶段是否是学术阶段，都有可能，或者在初创公司，或者在更成熟的公司中，对实现自动化方法的了解和信心将非常有用。本社论中表达的观点只是作者的观点，不一定是ACS的观点。本文引用了其他11种出版物。（11）不管研究人员的下一个职业阶段是学术上的，还是在初创公司，或更成熟的公司中，实施自动化方法的知识和信心都将很有用。本社论中表达的观点只是作者的观点，不一定是ACS的观点。本文引用了其他11种出版物。（11）不管研究人员的下一个职业阶段是学术界，还是在初创公司，或更成熟的公司中，实施自动化方法的知识和信心都将很有用。本社论中表达的观点只是作者的观点，不一定是ACS的观点。本文引用了其他11种出版物。