1. Introduction
1. 引言

The development of sustainable, efficient, and environmentally friendly catalysts, along with their corresponding catalytic production processes, is a crucial endeavor aimed at addressing energy shortages, environmental crises, and air pollution. In order to develop efficient catalysts and gain comprehensive insights into reaction mechanisms, the exploration of intricate chemical reactions,¹ synthesis and structural evolution of catalysts,^2,3 and catalytic processes has become increasingly significant.⁴ However, conventional research methods have largely involved analysis after catalysis. Researchers must interrupt the reaction, retrieve samples from the setup, and perform characterization, a process that is not only time consuming but also does not guarantee the complete observation of the same catalytic particle's morphological and structural changes. What's more important, the catalyst after the reaction might not accurately present the true state of the reaction process. To address these issues, in situ transmission electron microscopy (TEM) techniques have been employed to elucidate catalytic mechanisms at the atomic scale.^5,6
开发可持续、高效和环保的催化剂及其相应的催化生产工艺，是解决能源短缺、环境危机和空气污染的一项重要工作。为了开发高效的催化剂并全面了解反应机理，对复杂化学反应的探索，¹ 催化剂的合成和结构演变，^2,3 和催化过程变得越来越重要。⁴ 然而，传统的研究方法主要涉及催化后的分析。研究人员必须中断反应，从装置中取出样品，并进行表征，这一过程不仅耗时，而且不能保证完全观察同一催化粒子的形态和结构变化。更重要的是，反应后的催化剂可能无法准确呈现反应过程的真实状态。为了解决这些问题，已经采用了原位透射电子显微镜 （TEM） 技术来阐明原子尺度的催化机制。^5,6

Gaining a profound understanding of the relationships between the size, composition, morphology, crystal structure, surface atomic distribution, and properties of these materials is crucial for the design of new materials.^7,8 However, real-time observation of structural changes often proves to be a formidable task. Furthermore, a material's performance is directly linked to operating conditions such as temperature, gas pressure, and gas composition.^9,10 This has fueled interest in various in situ techniques to elucidate mechanisms of material structural transformation, including X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS), X-ray absorption fine structure (XAFS), and transmission electron microscopy (TEM). However, for gas solid catalysis, the in situ TEM technique can not only realize the change of the whole structure of the catalyst, but also the change of the structure of the single particle of the catalyst, which is more suitable for studying the catalytic process of the catalyst, including catalyst synthesis, catalytic behavior, thermal catalysis, and structural reconstruction. Nonetheless, the in situ TEM technique has its own limitations, such as the effect of electron beam irradiation on the catalyst and the lack of macroscopic characterization.^11,12 They are also affected by gas, pressure and temperature, which affects spatial resolution. Among them, electron beam irradiation can cause radiation damage, thermal effects, charge effects, phase transitions, and dissolution of the catalyst. Therefore, before conducting in situ TEM experiments, the electron beam dose tolerance of the catalyst can be obtained by adjusting the electron beam dose while ensuring resolution.
深入了解这些材料的尺寸、组成、形态、晶体结构、表面原子分布和性能之间的关系对于新材料的设计至关重要。^7,8 然而，实时观察结构变化通常被证明是一项艰巨的任务。此外，材料的性能与温度、气体压力和气体成分等作条件直接相关。^9,10 这激发了人们对各种原位技术的兴趣，以阐明材料结构转变的机制，包括 X 射线衍射 （XRD）、X 射线光电子能谱 （XPS）、X 射线吸收精细结构 （XAFS） 和透射电子显微镜 （TEM）。然而，对于气固催化，原位TEM技术不仅可以实现催化剂整体结构的变化，还可以实现催化剂单个颗粒结构的变化，更适合研究催化剂的催化过程，包括催化剂合成、催化行为、热催化和结构重构。尽管如此，原位TEM技术有其自身的局限性，如电子束辐照对催化剂的影响和缺乏宏观表征。^11,12 它们还受气体、压力和温度的影响，从而影响空间分辨率。其中，电子束辐照会导致催化剂的辐射损伤、热效应、电荷效应、相变和溶解。 因此，在进行原位 TEM 实验之前，可以在保证分辨率的情况下，通过调整电子束剂量来获得催化剂的电子束剂量容差。

TEM, which originated in the 1930s, is a potent technique for characterizing micro and nanostructures through imaging, electron diffraction, and spectroscopy.¹³ The electron beam's wavelength enables TEM to achieve atomic level resolution. In recent years, advancements such as aberration correctors and monochromators have led to exceptional spatial resolution (0.0405 nm) and energy resolution (4.2 meV).^14,15 By incorporating aberration corrected optics, pixelated array detectors, and full field imaging, the spatial resolution for material analysis has been further improved to 0.039 nm. This advancement allows spectroscopic techniques such as energy dispersive X-ray spectroscopy (EDX) and electron energy loss spectroscopy (EELS) to detect chemical information at the atomic scale.¹⁶ TEM is a valuable approach for exploring catalytic chemical reactions. Notably, the development of in situ TEM techniques, encompassing heating, biasing, liquid and gas environments, efficient digital recording systems, and enhanced computational capabilities, has extended the speed of image acquisition and improved time resolution, making it possible to explore rapid reaction kinetics.¹⁷In situ TEM enables real-time, direct observation of atomic scale chemical reactions under typical operating conditions, offering crucial insights into the workings principle of catalysts.¹⁸
TEM 起源于 1930 年代，是一种通过成像、电子衍射和光谱学来表征微纳米结构的有效技术。¹³ 电子束的波长使 TEM 能够达到原子级分辨率。近年来，像差校正器和光栅等进步带来了出色的空间分辨率 （0.0405 nm） 和能量分辨率 （4.2 meV）。^14,15 通过结合像差校正光学器件、像素化阵列探测器和全场成像，材料分析的空间分辨率进一步提高到 0.039 nm。这一进步使能量色散 X 射线光谱 （EDX） 和电子能量损失光谱 （EELS） 等光谱技术能够在原子尺度上检测化学信息。¹⁶ TEM 是探索催化化学反应的宝贵方法。值得注意的是，原位 TEM 技术的发展，包括加热、偏压、液体和气体环境、高效的数字记录系统和增强的计算能力，提高了图像采集速度并提高了时间分辨率，从而可以探索快速反应动力学。¹⁷原位 TEM 能够在典型作条件下实时、直接地观察原子级化学反应，为催化剂的工作原理提供重要的见解。¹⁸

In this paper, we will delve into the pivotal roles and emerging opportunities that in situ TEM offers within recent major advancements, spanning catalyst synthesis, catalytic behaviors, thermal catalysis, and structure reconstruction (Fig. 1). Firstly, we will provide a concise overview of the evolutionary progression of the in situ TEM technique and its operational mechanisms. Subsequently, we will outline the application of the in situ TEM technique in investigating gas phase catalysis. Lastly, we will explore the potential that the in situ TEM technique holds in shaping the trajectory of catalytic material evolution. By offering an atomic level perspective on catalytic processes, in situ TEM paves the way for the advancement of sophisticated catalytic materials.
在本文中，我们将深入探讨原位 TEM 在最近的重大进展中提供的关键作用和新兴机会，涵盖催化剂合成、催化行为、热催化和结构重建（图 1）。首先，我们将简要概述原位 TEM 技术的演变进程及其作机制。随后，我们将概述原位 TEM 技术在研究气相催化中的应用。最后，我们将探讨原位 TEM 技术在塑造催化材料进化轨迹方面的潜力。通过提供催化过程的原子级视角，原位 TEM 为复杂催化材料的发展铺平了道路。

2. Construction of in situ TEM
2. 原位 TEM 的构建

2.1 In situ TEM technologies
2.1 原位 TEM 技术

To enhance catalytic performance, extensive investigations have been conducted into the structures of various catalysts, including single atom catalysts²¹ and bimetallic catalysts.²² TEM has proven to be an effective technique for studying catalyst structures and chemical properties at the atomic scale, although it was traditionally limited to vacuum conditions. Currently, both environmental transmission electron microscopes (ETEM) and windowed holders enable the direct observation of catalytic reactions in high spatial resolution within gas environments.
为了提高催化性能，已经对各种催化剂的结构进行了广泛的研究，包括单原子催化剂²¹ 和双金属催化剂。²² TEM 已被证明是在原子尺度上研究催化剂结构和化学性质的有效技术，尽管它传统上仅限于真空条件。目前，环境透射电子显微镜 （ETEM） 和窗口支架都可以在气体环境中以高空间分辨率直接观察催化反应。

Different pumping systems of ETEM have undergone the following developmental stages: firstly, in 1968, Hashimoto et al. popularized and developed a high temperature gas reaction chamber employing a differential pumping system.²³ About two decades later, in the 1970s, Gai and her team achieved a TEM spatial resolution of 0.23 nm, sufficient to address lattice fringes within catalytic processes.²⁴ From then on, the use of field emission electron guns elevated the spatial resolution to the atomic level. In 1988, Suzuki et al. conducted analyses on in situ samples using TEM and EELS, demonstrating the capability to monitor the chemical and physical behavior of materials.²⁵ The current ETEM operates by employing differential pumping apertures to confine gases to the high pressure region around the sample and connecting an external pump to the microscope column (Fig. 2a and b),^19,26 This setup enables the capture of atomic scale images, with gas pressures typically below 10 Torr and time resolutions reaching up to 1 ms.^27,28
ETEM 的不同泵送系统经历了以下发展阶段：首先，在 1968 年，Hashimoto 等人推广并开发了一种采用差动泵送系统的高温气体反应室。²³ 大约二十年后，在 1970 年代，Gai 和她的团队实现了 0.23 nm 的 TEM 空间分辨率，足以解决催化过程中的晶格条纹。²⁴ 从那时起，场发射电子枪的使用将空间分辨率提高到原子水平。1988 年，Suzuki 等人使用 TEM 和 EELS 对原位样品进行了分析，证明了监测材料化学和物理行为的能力。²⁵ 当前的 ETEM 的工作原理是采用差分泵孔将气体限制在样品周围的高压区域，并将外部泵连接到显微镜柱（图 2a 和 b），^19,26这种设置能够捕获原子级图像，气体压力通常低于 10 Torr，时间分辨率高达 1 ms27,28

Compared to other in situ systems, the windowed holder (gas reaction system) enables material characterization in a gas and heating environment within a traditional TEM (as shown in Fig. 2c and Table 2). The principle involves using upper and lower windows to seal both the material and gas within the reactor of the sample holder. This prevents gas from escaping into the electron microscope chamber when inserted for characterization. The reaction zone can be heated up to 1000 °C while achieving a spatial resolution of 0.1 nm. Unlike the conventional TEM sample holder, the detachable reactor features a gas supply system to introduce a reaction atmosphere into the sample chamber. It also incorporates a temperature control system for heating the materials and gas inside the reactor.
与其他原位系统相比，开窗支架（气体反应系统）能够在传统 TEM 内的气体和加热环境中进行材料表征（如图 2c 和表 2 所示）。该原理涉及使用上下窗口来密封样品架反应器内的材料和气体。这可以防止气体在插入进行表征时逸出到电子显微镜室中。反应区可加热至 1000 °C，同时实现 0.1 nm 的空间分辨率。与传统的 TEM 样品架不同，可拆卸反应器具有供气系统，可将反应气氛引入样品室。它还包括一个温度控制系统，用于加热反应器内的材料和气体。

Table 1 provides a comparison of the two current in situ techniques. One of the significant advantages of the in situ sample holder compared to the ETEM is its compatibility with conventional standard TEM instruments, enabling more research groups to conduct in situ TEM experiments. The windowed holder also permits in situ experiments within a pressure range from several mbar up to 4.5 bar, facilitating studies of catalytic reactions under high pressure conditions by simulating such conditions within reactors.^29,30 However, the presence of SiN_x windows can lower the spatial resolution and reduce the EDX signal. Additionally, the preparation of in situ experimental samples using the in situ sample holder is more complex and potentially hazardous.
表 1 提供了两种当前原位技术的比较。与 ETEM 相比，原位样品架的显着优势之一是它与常规标准 TEM 仪器的兼容性，使更多的研究小组能够进行原位 TEM 实验。窗口支架还允许在几毫巴至 4.5 巴的压力范围内进行原位实验，通过模拟反应器内的高压条件下的催化反应来促进此类条件的研究。^29,30 然而，SiN_x 窗口的存在会降低空间分辨率并降低 EDX 信号。此外，使用原位样品架制备原位实验样品更加复杂且具有潜在危险。

Instrument  仪器	Advantages  优势	Disadvantages  弊
Windowed holder  开窗支架	(1) Low cost; (2) high pressure; (3) stable sample; (4) rapid heating （1） 成本低;（2） 高压;（3） 稳定的样品;（4） 快速加热	(1) Low resolution; (2) high danger （1） 分辨率低;（2） 高危险性
ETEM	(1) Gas directly into the lens barrel; (2) high TEM resolution （1） 气体直接进入镜筒;（2） 高 TEM 分辨率	(1) Restricted gas type; (2) high cost; (3) low pressure （1） 限制气体类型;（2） 成本高;（3） 低压

In situ system  原位系统	Unique advantages  独特优势	Application field  应用领域
Self-sealing reaction system 自密封反应系统	(1) Good stability; (2) good sealing （1） 稳定性好;（2） 密封性好	Self-assembly; metal corrosion; drug transport 自组装;金属腐蚀;药物运输
Electrochemical reaction system 电化学反应系统	(1) Superior performance; (2) easy use （1） 性能优越;（2） 使用方便	Electrochemical corrosion; small molecule 电化学腐蚀;小分子
Photocatalytic reaction system 光催化反应系统	(1) Wide usage; (2) easy to use （1） 用途广泛;（2） 易于使用	Photocatalytic reaction; solar cell; photosynthesis 光催化反应;太阳能电池;光合作用
Gas reaction system  气体反应系统	(1) Temperature control; (2) high security （1） 温度控制;（2） 安全性高	Thermocatalysis; two dimensional 热催化;二维
Liquid reaction system  液体反应系统	(1) High sensitivity; (2) good stability （1） 灵敏度高;（2） 稳定性好	Nanocatalyst synthesis; solvothermal process 纳米催化剂合成;溶剂热工艺
Solid heating system  固体加热系统	(1) High accuracy; (2) good stability （1） 精度高;（2） 稳定性好	Nanomaterials; energy materials 纳米材料;能源材料
High temperature heating system 高温加热系统	(1) Easy operation; (2) cost effective （1）作简便;（2） 成本效益	Phase transition; structure transformation 相变;结构改造
Mechanical control system 机械控制系统	(1) Good stability; (2) modular design （1） 稳定性好;（2） 模块化设计	Crystal transition; material mechanics 晶体转变;材料力学
Refrigeration system  制冷系统	(1) Temperature range; (2) good stability （1） 温度范围;（2） 稳定性好	Semiconductor devices; life sciences 半导体器件;生命科学
Freezing transfer system 冷冻输送系统	(1) Freezing time; (2) good stability （1） 冻结时间;（2） 稳定性好	Structural biology; medical treatment 结构生物学;医疗

In situ chip  原位芯片	Product parameters  产品参数			Application field  应用领域	Ref.  裁判。
Self-sealing  自密封	Top chip  顶部筹码	Size: 2.7 × 3.2 mm 尺寸：2.7 × 3.2 毫米	Window thickness: 10/25/50 mm 窗口厚度： 10/25/50 毫米	Nano materials; biological materials; magnetic materials 纳米材料;生物材料;磁性材料	35
	Bottom chip  底筹码	Size: 3 × 3.5 mm 尺寸：3 × 3.5 毫米	Window thickness: 25/50 mm 窗口厚度： 25/50 mm
Vacuum heated  真空加热		Size: 4 × 6 mm 尺寸：4 × 6 毫米	Window thickness: 10/25/50 mm 窗口厚度： 10/25/50 毫米	Solid material heating  固体材料加热	36
		Thickness: 0.2 mm  厚度： 0.2 毫米	Heating zone: Ø 180 μm 加热区：Ø 180 μm
Electrochemical  电化学	Top chip  顶部筹码	Size: 3 × 3.2 mm 尺寸：3 × 3.2 毫米	Window thickness: 10/25/50 mm 窗口厚度： 10/25/50 毫米	Electric corrosion; electrical properties; photoelectric effect 电腐蚀;电气性能;光电效应	37
	Bottom chip  底筹码	Size: 4 × 6 mm 尺寸：4 × 6 毫米	Window thickness: 25/50 mm 窗口厚度： 25/50 mm
Liquid flow  液体流动	Top chip  顶部筹码	Size: 3.5 × 7 mm 尺寸：3.5 × 7 毫米	Window thickness: 10/25/50 mm 窗口厚度： 10/25/50 毫米	Temperature observation of most liquid phase fluids 大多数液相流体的温度观察	38
	Bottom chip  底筹码	Size: 3.9 × 7.9 mm 尺寸：3.9 × 7.9 毫米	Window thickness: 25/50 mm 窗口厚度： 25/50 mm
Fluid heated  流体加热	Top chip  顶部筹码	Size: 3.5 × 6 mm 尺寸：3.5 × 6 毫米	Window thickness: 10 × 30 μm 窗口厚度：10 × 30 μm	Temperature control of gas and liquid samples 气体和液体样品的温度控制	39
	Bottom chip  底筹码	Size: 3.9 × 7.9 mm 尺寸：3.9 × 7.9 毫米	Window thickness: 25/50 mm 窗口厚度： 25/50 mm
Vacuum heated  真空加热		Size: 4 × 6 mm 尺寸：4 × 6 毫米	Heating zone: Ø 180 μm 加热区：Ø 180 μm	Nano materials; FIB flakes 纳米材料;FIB 薄片	40
		Thickness: 0.2 mm  厚度： 0.2 毫米
Electrochemical  电化学	Top chip  顶部筹码	Size: 4 × 6 mm 尺寸：4 × 6 毫米	Window thickness: 25/50 mm 窗口厚度： 25/50 mm	Electrochemistry; electrocatalysis; electrocorrosion 电化学;电催化;电腐蚀	41
	Bottom chip  底筹码	Size: 4 × 10 mm 尺寸：4 × 10 毫米	Interlayer height: 200–2000 nm 层间高度：200–2000 nm

Stimulus  刺激物	Topics  主题	Setup  设置	Ref.
Windowed holder (gas)  带窗支架（燃气）	Structural kinetics of Pt in hydrogenation of ethylene Pt 在乙烯加氢中的结构动力学	Hitachi 2700C STEM  日立 2700C STEM	79
	Observe the transformation of yolk shell into gold atoms 观察蛋黄壳向金原子的转变	Aberration-corrected STEM 像差校正 STEM	54
	Growth trajectory of interface bismuth 界面铋的生长轨迹	JEOL model JEM2100  JEOL 型号 JEM2100	80
	In situ study of atomic mechanism 原子机理的原位研究	JEOL 2100  JEOL 2100 系列	81
	In situ TEM observation of silver oxidation 原位银氧化的 TEM 观察	JEOL 2010LaB6	82
	Nanoscale kinetics of asymmetrical corrosion 不对称腐蚀的纳米级动力学	JEM-ARM200F	83
	Morphology changes of gold nanoparticle supported on CeO2 CeO2负载的纳米金颗粒的形貌变化	FEI Tecnai F20 equipped 配备 FEI Tecnai F20	84
	Dynamic structural evolution of palladium ceria core–shell catalysts 钯铈核壳催化剂的动态结构演变	JEOL JEM-2100F  杰尔 JEM-2100F	85
	In situ study of noncatalytic metal oxide nanowire growth 非催化性金属氧化物纳米线生长的原位研究	FEI Titan 80-300  飞 泰坦 80-300	86
	Gas molecules interacting with nanoparticulate catalysts 气体分子与纳米颗粒催化剂相互作用	200MC TEM  200MC 有	87
	Surface faceting of PdCu by H2 PdCu 的 H2 表面刻面	FEI Tecnai F20	88
	Sintering behaviors of Au on TiO2 Au 在 TiO2 上的烧结行为	Hitachi H-9500/Protochips 日立 H-9500/原型芯片	89
	Au single atoms and nanoporous Au for methane pyrolysis 用于甲烷热解的 Au 单原子和纳米多孔 Au	Talos F200X	90
	Reversible loss of yolk–shell structure 蛋黄-壳结构的可逆性损失	JEOL 2100F	63
	Structural changes in noble metal nanoparticles during CO oxidation 一氧化碳氧化过程中贵金属纳米颗粒的结构变化	Thermo Fisher Titan TEM	78
	Metal–support interaction of catalysts 催化剂的金属-载体相互作用	JEOL3100-R05	91
	Dynamic co-catalysis of Au single atoms and nanoporous Au Au 单原子和纳米多孔 Au 的动态共催化	Talos F200X	92
	Dynamic shape changes in supported copper nanocrystals 支撑的铜纳米晶的动态形状变化	JEOL 3100-R05  电 3100-R05	93
	Gold dissolution and tunneling across Ni2P shell 金溶解和穿过 Ni2P 壳层的隧穿	Aberration corrected STEM 像差校正的 STEM	94
	Quantification of critical particle distance for mitigating catalyst sintering 量化临界颗粒距离以减少催化剂烧结	JEM ARM200F	53
	In situ observation of the crystal structure transition of Pt–Sn Pt-Sn 晶体结构转变的原位观察	FEI Talos F200S  菲利 Talos F200S	74
	Atomic scale imaging of carbon nanofibre growth 碳纳米纤维生长的原子尺度成像	Philips CM300 FEG TEM 飞利浦 CM300 FEG TEM	4
	High entropy alloy nanocrystal 高熵合金纳米晶	JEOL 2100 FEG TEM	95
	Synthesis of high entropy alloy nanoparticles by a step alloying strategy 通过阶梯合金化策略合成高熵合金纳米颗粒	Fei Titan G2  飞泰坦 G2	96
ETEM	Formation of core–shell Pt3Co nanoparticles 核壳 Pt3Co 纳米颗粒的形成	FEI Titan 80-300  飞 泰坦 80-300	97
	Atomistic mechanisms of the interfacial reaction during alloy oxidation 合金氧化过程中界面反应的原子机制	FEI Titan ETEM  FEI 泰坦 ETEM	98
	Single walled carbon nanotubes in source gases 源气中的单壁碳纳米管	200 kV TEM  200 kV 透射电镜	99
	Probing the deactivation of NiGa nanoparticles as catalyst 探索作为催化剂的 NiGa 纳米颗粒的失活	Environmental TEM  环境透射电镜	100
	Surface reconstruction of Au nanoparticle Au 纳米颗粒的表面重构	FEI Titan ETEM equipped 配备 FEI Titan ETEM	101
	Double bilayer multistep growth in gallium nitride nanowires 氮化镓纳米线中的双双层多步生长	Environmental TEM  环境透射电镜	102
	Mesoscale oxidation mechanisms of aluminum nanoparticles 铝纳米颗粒的介尺度氧化机制	Hitachi 9500  日立 9500	103
	Atomic scale dynamic interaction of H2O molecules with Cu surface H2O 分子与 Cu 表面的原子尺度动态相互作用	FEI®Titan ETEM equipped  配备 FEI®Titan ETEM	104
	Deciphering atomistic mechanisms of the gas solid interfacial reaction 破译气固界面反应的原子机制	FEI Titan ETEM equipped 配备 FEI Titan ETEM	105
	Gas adsorbate induced Au atomic segregation and clustering from Cu 气体吸附物诱导的 Au 原子与 Cu 的偏析和聚集	Thermo Fisher Titan ETEM	106
	Dynamic atom clusters on AuCu nanoparticle surface during CO oxidation CO 氧化过程中 AuCu 纳米颗粒表面的动态原子团簇	ETEM (ETEM) chamber  ETEM （ETEM） 暗室	107
	Real time atomic scale visualization of reversible copper surface activation during the CO oxidation reaction CO 氧化反应过程中可逆铜表面活化的实时原子尺度可视化	FEI®TITAN ETEM 80-300  飞®泰 ETEM 80-300	108
	Carbon monoxide gas induced 4H-to-fcc phase transformation of gold 一氧化碳气体诱导的金 4H 到 fcc 相变	FEI Environmental TEM  FEI 环境 TEM	109
	Atomic origin of CO interaction effect of PtPb@Pt catalyst PtPb@Pt 催化剂的 CO 相互作用效应的原子来源	FEI Environmental TEM  FEI 环境 TEM	110
	Direct visualization of dynamic atomistic processes of Cu2O crystal growth 直接可视化 Cu2O 晶体生长的动态原子过程	FEI®ETEM Titan G2 80-300 FEI®ETEM 泰坦 G2 80-300	111
	Structural evolution of Cu/ZnO catalysts during water gas shift reaction Cu/ZnO 催化剂在水煤气变换反应过程中的结构演变	FEI Titan G2 80-300 飞鸿泰坦G2 80-300	112
	Facet dependent oxidative strong metal support interactions of palladium–TiO2 钯-TiO2 的刻面依赖性氧化强金属支持相互作用	FEI Titan G2 80-300 飞鸿泰坦G2 80-300	113
	In situ resolving the atomic reconstruction of SnO2 (110) surface 原位解析 SnO2 （110） 表面的原子重建	FEI Titan G2 80-300 飞鸿泰坦G2 80-300	114
	Direct in situ observation of the initial oxide nucleation and growth 直接原位观察初始氧化物成核和生长	Titan G2 80-300  泰坦 G2 80-300	115
	Growth modes of single walled carbon nanotubes 单壁碳纳米管的生长模式	FEI Titan G2 80-300 ETEM 飞扬 Titan G2 80-300 ETEM	116
	Revealing temperature-dependent oxidation dynamics of Ni nanoparticles 揭示 Ni 纳米颗粒的温度依赖性氧化动力学	Titan G2 80-300  泰坦 G2 80-300	69

2.2 Gas cell
2.2 气室

Thin electron transparent windows can provide unique gas environment conditions. In 1962, the concept of a “sealed cell” was introduced for in situ TEM observations.³³ The initial design of gas cells utilized two sample grids separated by a thin metal foil and sealed by a sample holder. However, thermal drift remained a primary challenge in this setup due to the expansion and contraction of the metal grids during heating and cooling. Subsequently, the development of microelectromechanical systems (MEMS) alleviated this challenge and extended the maximum heating temperature.³⁴ The developmental progression of in situ cells, as depicted in Fig. 3a, has evolved from open cells to liquid cells and further to gas flow cells. In comparison to open and liquid cells, gas flow cells offer several advantages: (1) control of conditions like temperature, pressure, and gas flow rate; (2) higher resolution; (3) more conducive to the environment closer to real world operating conditions. As illustrated in Fig. 3b, the concept involves introducing gas into the TEM holder. The sample will be placed in a controlled heating and atmospheric external field environment, the path of the illuminated electron beam will pass through the reactive gas and the sample, and the signals generated by interaction with the sample will be used for analysis of information such as structure and composition. As shown in Table 3, the comparison of different chips shows that the in situ chip of fluid heated is the most suitable for studying gas phase catalyst particles. Fig. 3c depicts a typical gas in situ TEM holder. Electron transparent windows are set on both upper and lower chips, facilitating precise alignment for optimal overlap and high resolution imaging. The cell design permits large angle electron diffraction and high angle dark field (HAADF) imaging. However, the thickness of the dual window membrane and the gas layer between them can reduce spatial resolution. Moreover, achieving proper alignment of windows on both top and bottom chips is intricate and challenging.
薄电子透明窗口可以提供独特的气体环境条件。1962 年，“密封单元”的概念被引入原位 TEM 观察。³³ 气体池的初始设计使用两个样品网格，由薄金属箔隔开并由样品架密封。然而，由于金属板栅在加热和冷却过程中的膨胀和收缩，热漂移仍然是该设置中的主要挑战。随后，微机电系统 （MEMS） 的发展缓解了这一挑战并延长了最高加热温度。³⁴ 如图 3a 所示，原位细胞的发育过程已经从开放细胞发展到液态细胞，并进一步发展到气体流动池。与开放式和液体流通池相比，气体流通池具有几个优点：（1） 控制温度、压力和气体流速等条件;（2） 更高的分辨率;（3） 更有利于更接近真实世界作条件的环境。如图 3b 所示，该概念涉及将气体引入 TEM 支架。样品将被放置在受控的加热和大气外场环境中，被照亮的电子束的路径将穿过反应气体和样品，与样品相互作用产生的信号将用于分析结构和成分等信息。如表 3 所示，不同芯片的比较表明，加热的流体原位芯片最适合研究气相催化剂颗粒。图 3c 描述了典型的气体原位 TEM 支架。 上下芯片上均设置了电子透明窗口，有助于精确对准，以实现最佳重叠和高分辨率成像。该单元设计允许大角度电子衍射和高角度暗场 （HAADF） 成像。然而，双窗口膜的厚度和它们之间的气体层会降低空间分辨率。此外，在顶部和底部芯片上实现窗口的正确对齐是复杂且具有挑战性的。

2.3 Spatial resolution
2.3 空间分辨率

The 5 mm thick gap in the gas reactor is equivalent to a solid carbon with a thickness of 20 nm, so the refractive index change of the gas can be ignored. Moreover, a uniformly distributed gas has no directional impact on the propagation of electron waves.⁴² Additionally, we can overlook the energy dispersion of electrons in the thin gas. Nevertheless, despite these factors, imaging in a gaseous medium still presents challenges like resolution reduction due to chromatic aberration, contrast loss, and beam spreading caused by elastic scattering.⁴³ However, we can address these difficulties by reducing the thickness of the chip window and selecting imaging modes to provide improved contrast and resolution. The resolution in both TEM and STEM modes is defined as.^44,45 (1) (2) where α represents the objective half angle, C_c stands for the chromatic aberration coefficient, E denotes the electron beam energy, T signifies the chip thickness, L is the average free path length for elastic scattering, and N₀ corresponds to the incident electron count. Clearly, it can be seen that with the decrease of chip thickness, the spatial resolution can be greatly improved.
气体反应器中 5 mm 厚的间隙相当于厚度为 20 nm 的固体碳，因此可以忽略气体的折射率变化。此外，均匀分布的气体对电子波的传播没有方向影响。⁴² 此外，我们可以忽略稀薄气体中电子的能量分散。然而，尽管存在这些因素，在气态介质中成像仍然存在挑战，例如由于色差、对比度损失和弹性散射引起的光束扩散而导致的分辨率降低。⁴³ 然而，我们可以通过减少芯片窗口的厚度和选择成像模式来提供更高的对比度和分辨率来解决这些困难。TEM 和 STEM 模式下的分辨率定义为。^44,45 元 其中 α 代表物镜半角，C_c 代表色差系数，E 表示电子束能量，T 表示芯片厚度，L 是弹性散射的平均自由程长度，N₀ 对应于入射电子数。显然，可以看出，随着芯片厚度的减小，空间分辨率可以大大提高。

2.4 Measurement of gas composition
2.4 气体成分的测量

During the course of reactions, accurate measurement of gas composition is challenging due to the lack of effective methods. Li et al.⁴⁶ employed combined approach of XAFS and STEM to study phase transitions in Pt catalysts during ethylene hydrogenation. The spatial resolution was limited to 1 nm due to the influence of SiN_x (50 nm) windows. Despite experimental constraints, this work demonstrated the feasibility of utilizing XANES, EXAFS, and STEM in combination to characterize gas components. Another study involving in situ EELS measurements of gas composition was reported.⁴⁷ The authors generated gases (H₂, CH₄, etc.) from catalytic reactions and used ETEM to directly detect catalyst structural evolution through in situ imaging and diffraction. Besides characterizing the catalysts, gas environments should also be assessed to understand the mechanisms of the catalytic process. Crozier et al. noted that inner shell EELS can measure gas composition with an accuracy of about 15%. However, due to the presence of ionization edges (around 12.5 eV), inner shell EELS cannot evaluate gas composition containing H₂. This work illustrated the potential of real time determination of gas composition and volume fractions during catalytic processes.
在反应过程中，由于缺乏有效的方法，准确测量气体成分具有挑战性。Li 等人。⁴⁶ 采用 XAFS 和 STEM 的组合方法来研究乙烯加氢过程中 Pt 催化剂的相变。由于 SiN_x （50 nm） 窗口的影响，空间分辨率被限制在 1 nm。尽管存在实验限制，但这项工作证明了结合使用 XANES、EXAFS 和 STEM 来表征气体成分的可行性。报道了另一项涉及气体成分原位 EELS 测量的研究。⁴⁷ 作者从催化反应中产生气体（H₂、CH₄ 等），并使用 ETEM 通过原位成像和衍射直接检测催化剂结构演变。除了表征催化剂外，还应评估气体环境以了解催化过程的机制。Crozier 等人指出，内壳 EELS 可以以约 15% 的精度测量气体成分。然而，由于存在电离边缘（约 12.5 eV），内壳 EELS 无法评估含有 H₂ 的气体成分。这项工作说明了在催化过程中实时测定气体成分和体积分数的潜力。

While in situ TEM offers the capability to characterize local structure and chemical information of catalysts, its research landscape is still in its infancy. An obstacle for in situ TEM studies is the limited availability of instrument resources and high costs. Additionally, the lack of expertise in both catalytic processes and electron microscopy hampers the development of in situ TEM techniques. However, some breakthroughs have been achieved in catalytic reaction research using in situ TEM, showcasing its potential and prospects.
虽然原位 TEM 能够表征催化剂的局部结构和化学信息，但其研究前景仍处于起步阶段。 原位 TEM 研究的一个障碍是仪器资源有限且成本高。此外，缺乏催化过程和电子显微镜方面的专业知识阻碍了原位 TEM 技术的发展。然而，使用原位 TEM 的催化反应研究已经取得了一些突破，展示了其潜力和前景。

2.5 The importance of in situ TEM in heterogeneous catalysis
2.5 原位 TEM 在多相催化中的重要性

Understanding the composition and surface structure of catalysts, as well as exploring the mechanisms of catalytic reactions, is of crucial importance and significance for the design and optimization of catalysts. Eren et al. observed that the (111) surface of copper (Cu) becomes unstable when exposed to carbon monoxide (CO) gas. It was further demonstrated that the inactive (111) facet of Cu, employed in the hydrolysis step crucial for the water-gas shift reaction, undergoes significant activation due to CO induced clustering.⁴⁸ Tao et al. discovered that the stepped platinum (Pt) surface undergoes extensive and reversible restructuring when exposed to CO pressures exceeding 0.1 Torr. The occurrence of large scale surface reconstruction on stepped Pt crystals highlights the close correlation between the coverage of reactant molecules under reaction conditions and the atomic structure of the catalyst surface.⁴⁹ An et al. found that the catalytic activity of different Pt/oxide systems is influenced by varying CO and O₂ ratios. In situ characterization under catalytically relevant reaction conditions using near-edge X-ray absorption fine structure (NEXAFS) and ambient pressure X-ray photoelectron spectroscopy (APXPS) revealed a strong correlation between the oxidation state of the oxide support and the catalytic activity at the oxide metal interface.⁵⁰ Crozier et al. employed the in situ electron energy loss spectroscopy (EELS) method for quantitative gas analysis under an environmental TEM. They demonstrated that the in situ gas analysis can also be utilized to reveal mass transport issues associated with differential gas diffusion through the system.⁴⁷ Crozier et al. utilized time resolved in situ aberration corrected TEM to characterize the sub-nanometer scale motion of cations, thereby pinpointing atomic level variations in the generation and annihilation rates of oxygen vacancies on the surfaces of oxide nanoparticles. The observation of the highest cation displacement at low coordination sites suggests enhanced surface oxygen vacancy activity at these sites.⁵¹ Therefore, the application of in situ TEM in the study of heterogeneous catalysis by providing real time, high resolution information offers a crucial tool for a deeper understanding of the catalyst's surface structure, active sites, and reaction mechanisms. This is indispensable for catalyst design, optimization, and the enhancement of catalytic performance. In the following sections, the application of in situ TEM in important catalytic reactions will be discussed.
了解催化剂的组成和表面结构，以及探索催化反应的机理，对于催化剂的设计和优化至关重要。Eren 等人观察到，当暴露于一氧化碳 （CO） 气体时，铜 （Cu） 的 （111） 表面变得不稳定。进一步证明，在水解步骤中用于对水气变换反应至关重要的 Cu 的非活性 （111） 面由于 CO 诱导的团簇而发生显着激活。⁴⁸ Tao 等人发现，当暴露于超过 0.1 Torr 的 CO 压力时，阶梯状铂 （Pt） 表面会发生广泛且可逆的重组。阶梯状 Pt 晶体上大规模表面重构的发生突出了反应条件下反应物分子的覆盖率与催化剂表面的原子结构之间的密切关联。⁴⁹ An 等人发现，不同 Pt/氧化物体系的催化活性受不同 CO 和 O₂ 比率的影响。使用近边缘 X 射线吸收精细结构 （NEXAFS） 和环境压力 X 射线光电子能谱 （APXPS） 在催化相关反应条件下进行原位表征， 揭示了氧化物载体的氧化态与氧化物金属界面处的催化活性之间具有很强的相关性。⁵⁰ Crozier 等人采用原位电子能量损失光谱 （EELS） 方法在环境 TEM 下进行定量气体分析。 他们证明， 原位气体分析也可用于揭示与通过系统的差分气体扩散相关的质量传输问题。⁴⁷ Crozier 等人利用时间分辨原位像差校正 TEM 来表征阳离子的亚纳米级运动，从而精确定位氧化物纳米颗粒表面氧空位的产生和湮灭速率的原子水平变化。在低配位点观察到最高的阳离子置换表明这些位点的表面氧空位活性增强。⁵¹ 因此，原位 TEM 通过提供实时、高分辨率信息应用于多相催化研究，为更深入地了解催化剂的表面结构、活性位点和反应机制提供了重要工具。这对于催化剂设计、优化和增强催化性能是必不可少的。在以下部分中，将讨论原位 TEM 在重要催化反应中的应用。

3. The use of in situ TEM in catalysts' synthesis
3. 原位 TEM 在催化剂合成中的应用

3.1 Transformation from metal particles to single atoms
3.1 从金属粒子到单个原子的转变

Single atom or active site catalysis has proven to be highly efficient compared with its traditional bulk or nanoparticle catalysis. With the development of TEM, it is possible to observe the synthesis process of single atom catalysts. In general, single atoms or small clusters easily agglomerate during thermal heating. However, Liu et al. conducted in situ TEM studies to investigate the evolution and stability of subnanometric metal species confined within a specific space.⁵² To observe the evolution of atomically dispersed Pt species, an area featuring only a few Pt clusters without Pt nanoparticles was selected for the experiments. Upon exposing the sample to H₂ (0.1 torr) at 350 °C, a few Pt clusters (below 0.5 nm) emerged (Fig. 4a). Raising the temperature to 550 °C and 700 °C under H₂ atmosphere led to the appearance of more subnano Pt (below 0.8 nm) and an increase in their average size (Fig. 4b and c). Pt cluster sizes continued to grow up to 0.8–1 nm when the temperature reached 800 °C (Fig. 4d). Upon switching from H₂ to O₂ after the reduction treatment, the behavior of Pt clusters in an oxidative atmosphere was examined. As shown in Fig. 4e, Pt clusters began to disperse after exposure to O₂ at 550 °C. Notably, both the number and size of Pt species decreased upon switching the gas from H₂ to O₂, validating the redispersion behavior of Pt species in O₂. At 700 °C in O₂, the majority of Pt clusters vanished, leaving only a few small Pt nanoparticles (1 nm) in the region (Fig. 4f). Importantly, it should be noted that the MCM-22 zeolite seemed to undergo damage from beam irradiation at high temperatures. Having studied the dynamics of Pt nanoparticles and subnano species with various gas treatments, the focus shifted to examining the stability and evolution of subnano Pt species during CO oxidation with O₂, using the 0.17%Pt@MCM-22-300 sample. As depicted in Fig. 4g–l, the highly dispersed Pt species transformed into Pt clusters as the reaction temperature for CO oxidation was raised to 100–150 °C. This agglomeration was attributed to the interaction between Pt and CO. Notable, subnano Pt clusters and small Pt nanoparticles (1–2 nm) remained stable as the reaction temperature was elevated up to 300 °C. Yin, et al. quantified the critical particle distance needed to impede catalyst sintering.⁵³ Pt/XC-72R samples were prepared at 300 °C with loadings of 1.0 wt% and 10.0 wt% for initiating in situ HAADF-STEM investigations. These samples represented instances of long and short particle distances, respectively. As anticipated, 10.0 wt% Pt/XC-72R exhibited significantly higher particle density and a shorter particle distance compared to the 1.0 wt% Pt/XC-72R counterpart (Fig. 4m and n).
与传统的本体或纳米颗粒催化相比，单原子或活性位点催化已被证明非常有效。随着 TEM 的发展，可以观察单原子催化剂的合成过程。一般来说，单个原子或小团簇在热加热过程中很容易团聚。然而，Liu 等人进行了原位 TEM 研究，以研究局限在特定空间内的亚纳米金属种类的演变和稳定性。⁵² 为了观察原子分散的 Pt 物质的进化，选择了一个只有几个 Pt 簇而没有 Pt 纳米颗粒的区域进行实验。将样品在 350 °C 下暴露于 H₂ （0.1 torr） 中后，出现了一些 Pt 簇（低于 0.5 nm）（ 图 4a）。在 H₂ 气氛下将温度提高到 550 °C 和 700 °C 导致出现更多的亚纳 Pt（低于 0.8 nm）并增加它们的平均尺寸（ 图 4b 和 c）。当温度达到 800 °C 时，Pt 簇大小继续增长至 0.8-1 nm（ 图 4d）。还原处理后从 H₂ 切换到 O₂ 后，检查 Pt 簇在氧化气氛中的行为。如图 4e 所示，Pt 簇在 550 °C 暴露于 O₂ 后开始分散。 值得注意的是，当气体从 H₂ 切换到 O₂ 时，Pt 物质的数量和大小都减少了，验证了 Pt 物质在 O₂ 中的再分散行为。在 O₂ 中 700 °C 时，大多数 Pt 簇消失，在该区域只留下少量小的 Pt 纳米颗粒 （1 nm）（ 图 4f）。 重要的是，应该注意的是，MCM-22 沸石似乎在高温下受到光束照射的损坏。在研究了各种气体处理下 Pt 纳米颗粒和亚纳米物种的动力学后，重点转移到使用 0.17%Pt@MCM-22-300 样品检查 O₂ 氧化过程中亚纳米 Pt 物种的稳定性和演变。如图 4g-l 所示，当 CO 氧化的反应温度升高到 100–150 °C 时，高度分散的 Pt 物质转化为 Pt 簇。 这种团聚归因于 Pt 和 CO 之间的相互作用。值得注意的是，当反应温度升高到 300 °C 时，亚纳米 Pt 簇和小 Pt 纳米颗粒 （1-2 nm） 保持稳定。 Yin 等人量化了阻碍催化剂烧结所需的临界粒子距离。在 300 °C 下制备 ⁵³ 个 Pt/XC-72R 样品，负载量分别为 1.0 wt% 和 10.0 wt%，用于启动原位 HAADF-STEM 研究。这些样本分别表示长和短粒子距离的实例。正如预期的那样，与 1.0 wt% 的 Pt/XC-72R 相比，10.0 wt% 的 Pt/XC-72R 表现出明显更高的颗粒密度和更短的颗粒距离（ 图 4m 和 n）。

3.2 Yolk shell transforming to single atoms and clusters
3.2 蛋黄壳转化为单个原子和团簇

Precious metal particles and atoms, due to their inert nature, do not react with other media even at elevated temperatures. Herein, Cai et al. unveiled the diffusion behavior of gold within the Ni₂P matrix, leading to the formation of individual gold atoms and clusters. The researchers employed in situ STEM to directly observe the dynamic process of atom diffusion.⁵⁴ As depicted in Fig. 5a, an Au yolk is present within the Ni₂P shell, with the initial size of the hole in Ni₂P measuring approximately 16 nm. In Fig. 5b, Au atoms initiate diffusion into Ni₂P at 350 °C, resulting in the emergence of numerous bright clusters and individual atoms within the Ni₂P matrix. With prolonged temperature exposure, these Au atoms continue to diffuse into Ni₂P, giving rise to an increasing number of Au single atoms and bright clusters, both within the Ni₂P interior and on its surface. By further elevating the temperature to 500 °C for 5 min (Fig. 5c), the Au yolk undergoes complete dissolution, causing the entire Ni₂P particles to adopt a uniform and luminous appearance. Detailed high resolution images in Fig. 5d reveal a lattice spacing of 2.89 Å, corresponding to the (110) plane of Ni₂P. However, the fast Fourier transform (FFT) diffraction pattern in Fig. 5e exhibits elongation and splitting, indicative of lattice distortion resulting from the diffusion of Au atoms within the Ni₂P lattice. Subsequently, Han et al. reported a distinct diffusion phenomenon within an Au–Ni₂P yolk shell structure through in situ TEM analysis.⁵⁵ In Fig. 5f, a yolk–shell structure is depicted at room temperature, where the Au yolk makes contact with the Ni₂P shell at three regions. As displayed in Fig. 5g, the bright field STEM image at lattice resolution within the dashed square region of Fig. 5f reveals the epitaxial correlation between the yolk and shell. Specifically, the (111) surface planes of Ni₂P are aligned with the (111) crystal plane of Au. Despite a significant lattice mismatch of 5.96% between the bulk Au (111) crystal plane spacing (2.35 Å) and the Ni₂P (111) crystal plane spacing (2.21 Å), no dislocations or defects are observed at the interface. Upon heating at 350 °C for 48 min and 110 min (Fig. 5h and i), a transformation in the shape of the Au yolk is noticed. Simultaneously, the upper contact regions of the Au yolk undergo enlargement. The volume of the Au yolk significantly increases (Fig. 5j). Subsequently, maintaining the temperature at 500 °C for an additional 67 min reveals the complete diffusion of the gold yolk from the Ni₂P shell, forming a new particle around the upper left yolk (Fig. 5k). In situ TEM examination of the diffusion process demonstrates that the (111) plane of the Au yolk aligns with the (001) plane of Ni₂P (Fig. 5l–n). After heating at 350 °C for 1.5 h, both the Ni₂P shell and the Au yolk retain their original microstructures, with outward Au diffusion progressing slowly (Fig. 5o–q). At 450 °C for 1 h, STEM analysis clearly showcases the dissolution of the Au yolk into the shell, with aggregated Au atoms in the Ni₂P shell (Fig. 5r). The bright field STEM image and FFT in Fig. 5s and t indicate diffraction spots corresponding to Ni₂P lattice spacing, but no gold crystal plane spacing, signifying an intermediate phase with a single gold atom state within the Ni₂P matrix due to the diffusion of Au in Ni₂P. Upon further heating at 500 °C for approximately 0.5 h (Fig. 5u–w), phase separation between Ni₂P and Au recurs, marked by the emergence of Au aggregates outside the Ni₂P shell. Notably, Fig. 5w highlights separated Ni₂P (300) and Au (220) diffraction spots, indicating a strain free and non-epitaxial interface between the Au particle and Ni₂P matrix.
贵金属颗粒和原子由于其惰性，即使在高温下也不会与其他介质发生反应。在本文中，Cai 等人揭示了金在 Ni₂P 基体中的扩散行为，导致形成单个金原子和团簇。研究人员采用原位 STEM 直接观察原子扩散的动态过程。⁵⁴ 如图 5a 所示，Ni₂P 壳中存在金蛋黄，Ni₂P 中空穴的初始尺寸约为 16 nm。在图 5b 中，Au 原子在 350 °C 时开始扩散到 Ni₂P，导致 Ni₂P 基体中出现许多明亮的团簇和单个原子。随着长时间的温度暴露，这些 Au 原子继续扩散到 Ni₂P 中，从而在 Ni₂P 内部及其表面产生越来越多的 Au 单原子和明亮的团簇。通过将温度进一步升高到 500 °C 5 分钟（ 图 5c），Au 蛋黄完全溶解，使整个 Ni₂P 颗粒呈现出均匀和发光的外观。 图 5d 中的详细高分辨率图像显示晶格间距为 2.89 Å，对应于 Ni₂P 的 （110） 平面。然而， 图 5e 中的快速傅里叶变换 （FFT） 衍射图谱表现出伸长和分裂，表明 Au 原子在 Ni₂P 晶格内扩散导致晶格畸变。随后，Han 等人通过原位 TEM 分析报道了 Au-Ni₂P 卵黄壳结构内明显的扩散现象。 ^{图 55} 中 5f，在室温下描绘了蛋黄-壳结构，其中 Au 蛋黄在三个区域与 Ni₂P 壳接触。如图 5g 所示， 图 5f 虚线方形区域内晶格分辨率的明场 STEM 图像揭示了蛋黄和蛋壳之间的外延相关性。具体来说，Ni₂P 的 （111） 表面平面与 Au 的 （111） 晶面对齐。尽管本体 Au （111） 晶面间距 （2.35 Å） 和 Ni₂P （111） 晶面间距 （2.21 Å） 之间存在 5.96% 的显著晶格失配，但在界面处未观察到位错或缺陷。在 350 °C 下加热 48 分钟和 110 分钟（ 图 5h 和 i）后，观察到金蛋黄形状的转变。同时，金蛋黄的上部接触区域变大。Au 蛋黄的体积显著增加（ 图 5j）。随后，将温度保持在 500 °C 再 67 分钟，金黄从 Ni₂P 壳层完全扩散，在左上部蛋黄周围形成新颗粒（ 图 5k）。 原位扩散过程的 TEM 检查表明，金蛋黄的 （111） 平面与 Ni₂P 的 （001） 平面对齐（ 图 5l-n）。在 350 °C 下加热 1.5 小时后，Ni₂P 壳和 Au 蛋黄都保留了其原始的微观结构，而 Au 向外扩散缓慢（ 图 5o-q）。在 450 °C 下 1 小时，STEM 分析清楚地显示 Au 蛋黄溶解到壳中，Ni₂P 壳层中有聚集的 Au 原子（ 图 5r）。图 1 中的明场 STEM 图像和 FFT 5s 和 t 表示对应于 Ni₂P 晶格间距的衍射点，但没有金晶平面间距，表示由于 Au 在 Ni₂P 中的扩散，在 Ni₂P 基体内具有单个金原子态的中间相。在 500 °C 下进一步加热约 0.5 小时后（ 图 5u-w），Ni₂P 和 Au 之间的相分离再次出现，其标志是在 Ni₂P 壳层外出现 Au 聚集体。值得注意的是， 图 5w 突出显示了分离的 Ni₂P （300） 和 Au （220） 衍射点，表明 Au 颗粒和 Ni₂P 基体之间存在无应变和非外延界面。

3.3 Growth of metal nanoparticles on the support
3.3 金属纳米颗粒在载体上的生长

Designing efficient and stable catalysts on carrier materials has captured the widespread research interest of scientists. Lang et al. demonstrated stable isolated Pt atoms through strong covalent metal support interactions with α-Fe₂O₃ during high temperature calcination.⁵⁶In situ gas holder and TEM monitored dispersion under 1 bar of flowing O₂ at 800 °C (Fig. 6a–c). Pt nanoparticles dissociated within the support, forming thermally stable single atoms. Ganzler et al. used in situ ETEM and XAFS to study Pt nanoparticles on CeO₂ under reducing and oxidizing conditions.⁵⁷ They revealed the high mobility and redispersion of Pt nanoparticles in O₂ at around 400 °C (Fig. 6e). In contrast, slight sintering of Pt nanoparticles was observed in hydrogen at 250 °C (Fig. 6d and f). Dai et al. reported the cyclic precipitation and dissolution of Rh nanoparticles on a CaTiO₃ support in alternating hydrogen and oxygen environments at 600–700 °C.⁵⁸ Most Rh nanoparticles disappeared during oxidation treatment but reappeared during hydrogen treatment (Fig. 6g–i). They also utilized an in situ gas holder and STEM to investigate the structural evolution of atomically dispersed Pt loaded on TiO₂ during oxidation and reduction processes. Under oxidation (300 °C, 760 Torr of O₂) and mild reduction (250 °C, 760 Torr of 5% H₂) conditions, Pt single atoms remained in the same positions on the TiO₂ support (Fig. 6j–l).⁵⁹ However, under severe reduction (450 °C, 760 Torr of 5% H₂), Pt atoms migrated by 1.6 nm, indicating the mobility endowed to Pt single atoms (Fig. 6l). Sun et al. achieved in situ formation of Co nanoparticles on PBMCo, offering a potential avenue for preparing supported nanocatalysts on perovskite materials.⁶⁰ They proposed a simple and feasible method for generating uniformly distributed Co nanoparticles and utilized in situ TEM to observe the precipitation of nanoparticles and the formation of ordered layered oxygen reactions. As shown in Fig. 6m, no nanoparticles were detected at 770 °C. However, the formation of Co nanoparticles was evident at 810 °C as seen in Fig. 6n–o. Upon further increasing the temperature to 850 °C, the size of Co nanoparticles increased. Fig. 6p–r show a series of TEM images of the PBMCo material. These images were captured at 850 °C, in 0.5 Pa of H₂, over a time span of 500 s. At 0 s, no significant detachment of nanoparticles was observed (Fig. 6p). After 250 s, TEM detected some nanoparticles with a diameter of approximately 5 nm (Fig. 6q). By 500 s, nanoparticles with an average diameter of 15 nm became more distinct (Fig. 6r). These results once again demonstrate that in situ TEM is a valuable tool for studying the sintering and dispersion processes of metal catalysts. It can be combined with other experimental methods for practical catalysis research.
在载体材料上设计高效稳定的催化剂已经引起了科学家的广泛研究兴趣。Lang 等人在高温煅烧过程中通过与 α-Fe₂O₃ 的强共价金属载体相互作用证明了稳定孤立的 Pt 原子。⁵⁶ 在 800 °C 下，原位储气罐和 TEM 监测了 1 bar 流动 O₂ 下的分散性（ 图 6a-c）。Pt 纳米颗粒在载体内解离，形成热稳定的单原子。Ganzler 等人使用原位 ETEM 和 XAFS 研究还原和氧化条件下 CeO₂ 上的 Pt 纳米颗粒。⁵⁷ 他们揭示了 Pt 纳米颗粒在 400 °C 左右的 O₂ 中的高迁移率和再分散性（ 图 6e）。相比之下，在 250 °C 的氢气中观察到 Pt 纳米颗粒的轻微烧结（ 图 6d 和 f）。Dai 等人报道了在 600–700 °C 的氢和氧交替环境中，Rh 纳米颗粒在 CaTiO₃ 载体上的循环沉淀和溶解 ⁵⁸。大多数 Rh 纳米颗粒在氧化处理过程中消失，但在氢处理期间重新出现（ 图 6g-i）。他们还利用原位储气罐和 STEM 来研究氧化和还原过程中负载在 TiO₂ 上的原子分散 Pt 的结构演变。在氧化（300 °C，760 Torr 的 O₂）和轻度还原（250 °C，760 Torr 的 5% H₂）条件下，Pt 单个原子保持在 TiO₂ 载体上的相同位置（ 图 D）。 6j-l）。⁵⁹ 然而，在严重还原（450 °C，5% H₂ 的 760 Torr）下，Pt 原子迁移了 1.6 nm，表明 Pt 单个原子具有迁移率（ 图 6l）。Sun 等人在 PBMCo 上实现了 Co 纳米颗粒的原位形成，为在钙钛矿材料上制备负载型纳米催化剂提供了一条潜在的途径。⁶⁰ 他们提出了一种简单可行的方法来产生均匀分布的 Co 纳米颗粒，并利用原位 TEM 来观察纳米颗粒的沉淀和有序层状氧反应的形成。如图 6m 所示，在 770 °C 时未检测到纳米颗粒。 然而，Co 纳米颗粒的形成在 810 °C 时很明显，如图 6n-o 所示。当温度进一步升高到 850 °C 时，Co 纳米颗粒的尺寸增加。 图 6p-r 显示了 PBMCo 材料的一系列 TEM 图像。这些图像是在 850 °C、0.5 Pa 的 H₂ 中，在 500 s 的时间跨度内捕获的。在 0 秒时，未观察到纳米颗粒的明显脱离（ 图 6p）。250 秒后，TEM 检测到一些直径约为 5 nm 的纳米颗粒（ 图 6q）。到 500 秒时，平均直径为 15 nm 的纳米颗粒变得更加清晰（ 图 6r）。这些结果再次表明， 原位 TEM 是研究金属催化剂烧结和分散过程的宝贵工具。它可以与其他实验方法相结合进行实际催化研究。

4. In situ TEM on structure reconstruction

4.1 Phase evolution
4.1 相变

Bimetallic nanoparticles hold great potential as catalysts. Wang et al. recently published a study on the PtPb@Pt catalyst, where they investigated the CO gas induced phase separation and the formation of ultrathin Pb nanosheets.⁶¹Fig. 7a–h illustrate the phase and morphology changes of the nanoplates during heating from room temperature to 300 °C. Initially, very thin sheets began peeling off from the catalyst, marked by red arrows in Fig. 7b. These nanosheets increased in size over time, as depicted in Fig. 7c. After 720 s at 300 °C (Fig. 7d), these ultrathin nanosheets underwent rapid growth, forming significantly larger flakes. With further heating (Fig. 7e and f), these ultrathin nanosheets merged, leading to the formation of even larger nanosheets, as highlighted in white circles. In Fig. 7f, the original particles “1” and “2” transformed into three particles labeled as “1–3”. Detailed lattice images (Fig. 7g and h) at the initial phase of separation revealed the amorphous structure of the newly formed Pb islands when they were extremely small (3.5 nm). These islands quickly crystallized as their size reached around 3.5 nm. The lattice constants of these nanosheets matched the (111) crystal planes of the FCC-Pb crystal, indicated by white lines in Fig. 7h. In situ heating of PtPb@Pt in a vacuum was conducted and compared with results obtained in a CO gas environment to illustrate the influence of CO gas on phase separation. Sharp differences were observed in Fig. 7i–n. Fig. 7i–k show the intact shape and atomic structure of PtPb@Pt after heating to 250 °C and 300 °C in a vacuum. No significant changes were observed compared to the initial core–shell state, indicating the stability of the catalyst in a vacuum at high temperatures. However, in a CO gas environment, phase separation occurred when heated to 250 °C. Due to the ultrathin thickness of the formed Pb nanosheets, their contrast was lower than that of the original PtPb@Pt particles. Fig. 7l–n demonstrated that a significant amount of ultrathin Pb nanosheets detached from the PtPb particles, while the original particles became more aggregated after heating in CO gas. Yun et al. captured the atomic shifts and rearrangements associated with diffused solid phase transformations in Pt–Sn systems to reveal the atomic mechanisms driving these transformations.⁶² The transformation of PtSn₄ to PtSn₂ was observed under TEM (Fig. 7o). When the temperature reaches 250 °C, a black transverse stripe appears in the PtSn₄ region and expands due to the transition to the PtSn₂ phase. To investigate the atomic mechanism of the phase transition, the phase transition was initiated by slowly heating PtSn₄/PtSn₂ sheets to 300 °C. First, at 150 °C, the structure of the parent PtSn₄ is modulated. These modulations are very subtle, resulting in no change in contrast in the HAADF-STEM images. However, they are clearly visible in BF-STEM images, with transverse dark contrast occurring every four to five layers of PtSn₄ (Fig. 7p). At ∼200 °C, a new intermediate phase appears in the layer between the dark bands of the BF-STEM image. Upon further heating, the mesophase develops into PtSn₂ (Fig. 7q). The atomic structure of the intermediate is not only different from that of the parent (PtSn₄) and product (PtSn₂), but also from any existing Pt_xSn_y phase.
双金属纳米颗粒作为催化剂具有巨大的潜力。Wang 等人最近发表了一项关于 PtPb@Pt 催化剂的研究，他们研究了 CO 气体诱导的相分离和超薄 Pb 纳米片的形成。⁶¹ 图 7a-h 说明了纳米板在从室温加热到 300 °C 过程中的相和形态变化。 最初，非常薄的片材开始从催化剂上剥落，如图 7b 中的红色箭头所示。这些纳米片的尺寸随着时间的推移而增加，如图 7c 所示。在 300 °C 下 720 秒后（ 图 7d），这些超薄纳米片经历了快速生长，形成了明显更大的薄片。随着进一步加热（ 图 7e 和 f），这些超薄纳米片合并，形成更大的纳米片，如白色圆圈所示。在图 7f 中，原始粒子 “1” 和 “2” 转化为三个标记为 “1–3” 的粒子。分离初始阶段的详细晶格图像（ 图 7g 和 h）揭示了新形成的 Pb 岛在极小 （3.5 nm） 时的无定形结构。这些岛屿在大小达到 3.5 nm 左右时迅速结晶。这些纳米片的晶格常数与 FCC-Pb 晶体的 （111） 晶面相匹配，如图 7h 中的白线所示。在真空中对 PtPb@Pt 进行原位加热，并与在 CO 气体环境中获得的结果进行比较，以说明 CO 气体对相分离的影响。在图 7i-n 中观察到明显的差异。 图 7i-k 显示了 PtPb@Pt 在真空中加热至 250 °C 和 300 °C 后的完整形状和原子结构。 与初始核壳状态相比，没有观察到显著变化，表明催化剂在高温真空中的稳定性。然而，在 CO 气体环境中，当加热到 250 °C 时，会发生相分离。 由于形成的 Pb 纳米片的厚度超薄，它们的对比度低于原始 PtPb@Pt 颗粒的对比度。 图 7l-n 表明，大量超薄的 Pb 纳米片从 PtPb 颗粒上分离出来，而原始颗粒在 CO 气体中加热后变得更加聚集。Yun 等人捕获了 Pt-Sn 系统中与扩散固相转变相关的原子位移和重排，以揭示驱动这些转变的原子机制。⁶² 在 TEM 下观察到 PtSn₄ 向 PtSn₂ 的转变（ 图 7o）。当温度达到 250 °C 时，PtSn₄ 区域出现黑色横向条纹，并由于向 PtSn₂ 相的转变而膨胀。为了研究相变的原子机制，通过将 PtSn₄/PtSn₂ 片缓慢加热至 300 °C 来引发相变。 首先，在 150 °C 下，母体 PtSn₄ 的结构被调制。这些调制非常微妙，导致 HAADF-STEM 图像的对比度没有变化。然而，它们在 BF-STEM 图像中清晰可见，每 4 到 5 层 PtSn₄ 就会出现横向暗对比（ 图 7p）。在 ∼200 °C 时，BF-STEM 图像的暗带之间的层中出现一个新的中间相。进一步加热后，中间相发展成 PtSn₂（ 图 7q）。 中间体的原子结构不仅不同于母体 （PtSn₄） 和产物 （PtSn₂） 的原子结构，也不同于任何现有的 Pt_xSn_y 相。

4.2 Reversible change of structure
4.2 结构的可逆变化

Bimetallic core shell catalysts have garnered extensive attention due to their exceptional catalytic performance across a wide range of electrochemical reactions. It is generally believed that their catalytic properties are determined by the synergistic interplay between the electronic and geometric features at the core and shell interface. Zhang et al. used in situ TEM and revealed that the Ni–Au catalytic system is highly selective for the formation of CO in the CO₂ hydrogenation reaction.⁶³ The Ni core remains enveloped by a complete ultrathin Au shell both before and after the reaction. The TEM image in Fig. 8a illustrates a Ni–Au nanoparticle after being heated to 450–600 °C and subsequently cooled to 400 °C, as indicated by the yellow arrow. Within the temperature range of 450–500 °C, the ultrathin Au shell becomes visibly apparent, appearing as dark edges surrounding the Ni–Au nanoparticles. At 600 °C, the disappearance of these dark edges indicates the dissolution of the outermost Au layer into the Ni core, forming a mixed Ni–Au alloy. Upon cooling to 450–400 °C, the reappearance of the darker gold edges signifies the restoration of the Ni–Au core shell structure. The identical contrast distribution between the high resolution HAADF and BF images (Fig. 8b and c) provides clear proof of the phase-contrast properties inherent in TEM imaging. In the BF image, the dark spots along the particle edges can be precisely recognized as individual Au atomic columns. Notably, the greater depth of field in the BF-TEM image suggests that the Au shell enveloping the NiAu core is more comprehensively formed than what is apparent in the HAADF image. This uniformly covered Au-rich shell additionally attributed to the CO selectivity at low reaction temperatures. Yuan et al. conducted a study focused on the in situ manipulation of the active Au–TiO₂ interface with atomic precision during CO oxidation.⁶⁴ In the course of the CO oxidation reaction, periodic cessation of CO injection was employed. Remarkably, transitioning from S⁼ to S^// was observed upon halting CO injection and reverting to an O₂ environment (1 mbar) (Fig. 8d–f). Subsequently, changing the O₂ pressure from 1 mbar (Fig. 8f) to 4 mbar did not notably alter the interface structure (Fig. 8g). Upon reintroducing CO (Fig. 8g–i), a rotation from S^// to S⁼ occurred once more (Fig. 8e–h). These findings demonstrate the reversibility of the Au–TiO₂ (001) interface's dynamic response to the external environment at elevated temperatures. Additional top view observations indicated that the rotation of the Au NP in CO and O₂ reactive environments exhibited temperature dependence. Unlike the reversible rotation behavior at 500 °C (Fig. 8j and k), the rotation of the Au nanoparticles prompted by changes in the gas environment could be “frozen” by cooling to 20 °C. Cooling the Au nanoparticles from 500 °C to 20 °C (Fig. 8k and l) in an oxygen environment preserved the S^// state. At 20 °C, CO injection did not induce Au nanoparticle rotation, and the S^// state remained constant during the 25 min observation (Fig. 8m) in both CO and O₂ reactive environments. These outcomes suggest the stability of the S^// configuration during low temperature CO oxidation. Upon elevating the temperature back to 500 °C, the dynamic shift between S^// and S⁼ was once again observed. Therefore, by combining gas and temperature control, atomic level interface tunability was achieved.
双金属核壳催化剂因其在各种电化学反应中的出色催化性能而受到广泛关注。人们普遍认为，它们的催化性能是由核心和壳层界面处的电子和几何特征之间的协同相互作用决定的。Zhang 等人使用原位 TEM 并揭示了 Ni-Au 催化体系对 CO₂ 加氢反应中 CO 的形成具有高度选择性。⁶³ Ni 核在反应前后都被完整的超薄 Au 壳层包裹。 图 8a 中的 TEM 图像显示了 Ni-Au 纳米颗粒在加热到 450–600 °C，随后冷却到 400 °C 后，如黄色箭头所示。在 450–500 °C 的温度范围内，超薄的 Au 壳层变得明显，表现为围绕 Ni-Au 纳米颗粒的深色边缘。在 600 °C 时，这些深色边缘的消失表明最外层的 Au 层溶解到 Ni 核中，形成混合 Ni-Au 合金。冷却至 450–400 °C 后，深色金色边缘的重新出现意味着 Ni-Au 核壳结构的恢复。高分辨率 HAADF 和 BF 图像之间相同的对比度分布（ 图 8b 和 c）清楚地证明了 TEM 成像中固有的相差特性。在 BF 图像中，沿着颗粒边缘的黑点可以被精确地识别为单个 Au 原子柱。值得注意的是，BF-TEM 图像中更大的景深表明，包裹 NiAu 核心的 Au 壳层比 HAADF 图像中明显的形成更全面。 这种均匀覆盖的富金壳层还归因于低反应温度下的 CO 选择性。Yuan 等人进行了一项研究，重点是在 CO 氧化过程中以原子精度对活性 Au-TiO₂ 界面进行原位纵。⁶⁴ 在 CO 氧化反应过程中，采用周期性停止 CO 注入。值得注意的是，在停止 CO 注入并恢复到 O₂ 环境 （1 mbar） 时，观察到从 S⁼ 到 S 的转变（ 图 8d-f）。随后，将 O₂ 压力从 1 mbar（ 图 8f）更改为 4 mbar 并没有显着改变界面结构（ 图 8g）。重新引入 CO 后（ 图 8g-i），再次从 S 旋转到 S⁼（ 图 8e-h）。这些发现证明了 Au-TiO₂ （001） 界面在高温下对外部环境的动态响应的可逆性。其他俯视图观察表明，Au NP 在 CO 和 O₂ 反应环境中的旋转表现出温度依赖性。与 500 °C 时的可逆旋转行为不同（ 图 8j 和 k），由气体环境变化引起的 Au 纳米颗粒旋转可以通过冷却至 20 °C 来“冻结”。 在氧气环境中将 Au 纳米颗粒从 500 °C 冷却到 20 °C（ 图 8k 和 l）保持 S 状态。在 20 °C 时，CO 注入不会诱导 Au 纳米颗粒旋转，并且在 CO 和 O₂ 反应环境中的 25 min 观察期间（ 图 8m）S 状态保持不变。 这些结果表明 S 构型在低温 CO 氧化过程中的稳定性。将温度升回 500 °C 后，再次观察到 S 和 S⁼ 之间的动态变化。因此，通过结合气体和温度控制，实现了原子级界面可调性。

4.3 Atomistic mechanisms of the interfacial reaction
4.3 界面反应的原子机制

From multiphase catalysis to stress corrosion cracking, gas solid interfacial reactions play a pivotal role in many technological applications. However, understanding how gases and solids interact to form stable interfaces has become even more crucial. Wang et al. conducted a study where they investigated the structure–activity relationship by observing the real time sulfurization processes of ZnO nanowires using in situ TEM.⁶⁵ The results from in situ TEM confirmed that there was no observable diffusion of Zn ions out of the shell layer. This observation supported the hypothesis that SO₂ molecules diffused through the formed shell layer to reach the core shell interface for the reaction to occur. As the reaction time increased, the thickness of the shell layer also increased. Simultaneously, the ZnO nanowire underwent a progressive reduction in diameter. This phenomenon is illustrated in Fig. 9b, where the shell thickness reached 5 nm after 86 s of reaction. Upon extending the reaction time to 516 s, the shell thickness expanded to 21 nm, leading to a decrease in the diameter of the ZnO nanowire to 75 nm (Fig. 9a–g). The findings presented in Fig. 9a–g clearly depict the gradual consumption of surface atoms on the nanowire, ultimately forming the shell structure. Additionally, Fig. 9h offers a high resolution lattice image. Through high resolution TEM analysis, lattice fringes corresponding to the (002) crystal plane were observed along the ZnO core. Further insight from EDX mapping is provided in Fig. 9i, indicating that S was predominantly located within the shell layer, while Zn was primarily situated within the core of the nanowire. Luo et al. directly witnessed the dynamic transformation of CeO₂-supported AuCu alloy nanoparticles at the atomic level during CO oxidation. Their observations revealed that gas molecules play a role in liberating metal atoms on the (010) surface, leading to the creation of highly mobile atomic clusters.⁶⁶Fig. 9j presents the experimental arrangement for in situ CO oxidation reactions employing CeO₂-supported AuCu catalysts within the ETEM. Fig. 9k shows a HRTEM image that captures the configuration of an AuCu nanoparticle situated on CeO₂. The nanoparticle assumes a face-centered cubic (FCC) structure, evident from both the associated FFT pattern in the upper-right and the simulated diffraction pattern in the bottom-right. These patterns collectively affirm that the AuCu alloy nanoparticle exists in a solid-solution phase. When subjected to a gas mixture of CO and O₂ at room temperature, the (010) surface undergoes activation by gas molecules, resulting in the formation of an atomic-scale “adlayer” as illustrated in Fig. 9l. The simulated HRTEM image of this “adlayer” on the alloy surface in Fig. 9m closely corresponds to the depiction in Fig. 9l. As depicted in Fig. 9n, gas molecules engage in strong interactions with the surface atoms of the AuCu nanoparticle via chemisorption, thereby releasing low-coordinated surface metal atoms through robust Au–CO bonding. The interaction dynamics of atom clusters with edge atoms are vividly portrayed in the time-resolved HRTEM images of Fig. 9o. Over a 2 s interval, the progression is evident: the dynamic atom cluster attaches (at 0.5 s) and merges (at 1 s) with edge atoms, forming an FCC atomic structure, before ultimately disintegrating/detaching (at 1.5 s) from the edge position as delineated by the dashed-line box. Concurrently, white arrows in Fig. 9o indicate that adatoms at the lower edge position maintain their positions under the reaction conditions. These observations underscore the dynamic nature of atom clusters, which could expedite gas adsorption and diffusion on the nanoparticle.
从多相催化到应力腐蚀裂化，气固界面反应在许多技术应用中起着举足轻重的作用。然而，了解气体和固体如何相互作用以形成稳定的界面变得更加重要。Wang 等人进行了一项研究，他们通过使用原位 TEM 观察 ZnO 纳米线的实时硫化过程来研究构效关系。⁶⁵ 原位 TEM 的结果证实，没有观察到的 Zn 离子从壳层扩散。这一观察结果支持了 SO₂ 分子通过形成的壳层扩散到达核壳界面以发生反应的假设。随着反应时间的增加，壳层的厚度也增加。同时，ZnO 纳米线的直径逐渐减小。这种现象如图 9b 所示，其中反应 86 秒后壳层厚度达到 5 nm。将反应时间延长至 516 秒后，壳层厚度扩大到 21 nm，导致 ZnO 纳米线的直径减小到 75 nm（ 图 9a-g）。 图 9a-g 中呈现的发现清楚地描述了纳米线上表面原子的逐渐消耗，最终形成壳结构。此外， 图 9h 提供了高分辨率的晶格图像。通过高分辨率 TEM 分析，沿 ZnO 核心观察到对应于 （002） 晶面的晶格条纹。 图 9i 提供了 EDX 映射的进一步见解，表明 S 主要位于壳层内，而 Zn 主要位于纳米线的核心内。 Luo 等人直接见证了 CeO₂ 负载的 AuCu 合金纳米颗粒在 CO 氧化过程中在原子水平上的动态转变。他们的观察表明，气体分子在释放 （010） 表面的金属原子方面发挥作用，导致产生高度移动的原子团簇。⁶⁶ 图 9j 显示了在 ETEM 中使用 CeO₂ 负载的 AuCu 催化剂进行原位 CO 氧化反应的实验安排。 图 9k 显示了捕获位于 CeO₂ 上的 AuCu 纳米颗粒构型的 HRTEM 图像。纳米颗粒采用面心立方 （FCC） 结构，从右上角的相关 FFT 图样和右下角的模拟衍射图中可以明显看出。这些模式共同证实了 AuCu 合金纳米颗粒存在于固溶相中。当在室温下受到 CO 和 O₂ 的气体混合物时，（010） 表面被气体分子激活，导致形成原子级的“外加层”，如图 9l 所示。 图 9m 中该“外加层”在合金表面上的模拟 HRTEM 图像与图 9l 中的描述非常相似。如图 9n 所示，气体分子通过化学吸附与 AuCu 纳米颗粒的表面原子发生强相互作用，从而通过强大的 Au-CO 键合释放出低配位的表面金属原子。原子簇与边缘原子的相互作用动力学在图 9o 的时间分辨 HRTEM 图像中生动地描绘了出来。 在 2 s 的间隔内，进展是显而易见的：动态原子簇附着（在 0.5 秒时）并与边缘原子合并（在 1 秒时），形成 FCC 原子结构，然后最终从虚线框划定的边缘位置解体/分离（在 1.5 秒时）。同时， 图 9o 中的白色箭头表示下边缘位置的吸附原子在反应条件下保持其位置。这些观察结果强调了原子簇的动力学性质，这可能会加速气体在纳米粒子上的吸附和扩散。

5. In situ observations of catalytic behaviors
5. 催化行为的原位观察

5.1 Size redistribution during catalytic reactions
5.1 催化反应中的尺寸重新分布

The size of nanoscale catalysts is crucial for maintaining catalytic performance, as materials often experience sintering under high temperature reaction gas conditions, resulting in the loss of active sites and ultimately leading to a decline in catalytic activity. Therefore, comprehending the sintering process of catalysts during chemical reactions holds significant importance. Simonsen et al. studied the sintering mechanism of Pt nanoparticles dispersed on an Al₂O₃ substrate over 7.5 Torr of air at 650 °C.⁶⁷ They postulated that the sintering process of Pt is governed by an Ostwald ripening mechanism (Fig. 10a–j). Initially characterized by a Gaussian distribution, the particle diameter distribution underwent a transformation from a skewed distribution to the Lifshitz–Slyozov–Wagner distribution. However, the observations for an individual nanoparticle deviated from the Ostwald model due to the influence of the local environment on the atom exchange process. Huang et al. used in situ TEM to study the particle shape plasticity and carbon graphitization enhancement of Ir/C catalysts under heating from atomic scale imaging.⁶⁸Fig. 10k–m show HRTEM images of several Ir nanoparticles monitored over a long period of time at 800 °C. Apart from morphological particle remodeling, little significant change in particle size was found at 800 °C for 30 minutes. At lower temperatures, the mobility of Ir atoms is limited by the presence of a stable carbon, which hinders the size growth of nanoparticles during Ostwald ripening. In addition, the overlayer Ir nanoparticles allow particle migration and subsequent fusion to be also largely avoided.
纳米级催化剂的尺寸对于保持催化性能至关重要，因为材料经常在高温反应气体条件下烧结，导致活性位点的损失，最终导致催化活性下降。因此，了解化学反应过程中催化剂的烧结过程具有重要意义。Simonsen 等人研究了在 650 °C 下通过 7.5 Torr 空气分散在 Al₂O₃ 衬底上的 Pt 纳米颗粒的烧结机理 ⁶⁷。他们假设 Pt 的烧结过程受 Ostwald 成熟机制控制（ 图 10a-j）。颗粒直径分布最初以高斯分布为特征，经历了从偏态分布到 Lifshitz-Slyozov-Wagner 分布的转变。然而，由于局部环境对原子交换过程的影响，对单个纳米粒子的观察结果偏离了 Ostwald 模型。Huang 等人使用原位 TEM 研究原子尺度成像加热下 Ir/C 催化剂的颗粒形状塑性和碳石墨化增强。⁶⁸ 图 10k-m 显示了在 800 °C 下长时间监测的几种 Ir 纳米颗粒的 HRTEM 图像。 除了形态颗粒重塑外，在 800 °C 下 30 分钟下，颗粒大小几乎没有显著变化。在较低温度下，Ir 原子的迁移率受到稳定碳存在的限制，这阻碍了 Ostwald 成熟过程中纳米颗粒的尺寸增长。此外，覆盖层 Ir 纳米颗粒也允许在很大程度上避免颗粒迁移和随后的融合。

5.2 Shape changing
5.2 形状变化

The dynamic response of catalyst shapes to changes in gas environments has surpassed previous expectations. In order to gain a deeper understanding of catalytic mechanisms, research into this area has become increasingly crucial. However, due to limitations imposed by permissible pressures, the aforementioned in situ ETEM experiments have not been able to adequately capture the properties of nanoparticles under the actual working conditions of varying gas pressures. With the advancement of gas containment technology, it has become possible to observe metal catalysts under high temperature environmental conditions. Wang and co-workers investigated the dynamic oxidation process of Ni nanoparticles at different temperatures by in situ TEM, revealing the temperature dependent oxidation behavior (Fig. 11a).⁶⁹ At relatively low temperatures (600 °C), the oxidation of Ni nanoparticles undergoes the classic Kirkendall process and is accompanied by the formation of an oxide shell. In contrast, at higher temperatures (800 °C), oxidation begins with a single crystal nucleus on the metal surface and proceeds along the metal/oxide interface, without forming a cavity throughout the process. Song et al. used in situ high-resolution TEM combined with molecular dynamics to find that there is tension in the asymmetric 5 folding twin of Au nanoparticles, which is due to dislocation slip on the 5 folding axis under the electron beam and the double boundary migration of dislocation reaction. The migration of one or two layers of biplanes is controlled by energy carriers. But in general, the total energy, including surface, lattice strain and boundary energy, is released after successive boundary migrations (Fig. 11b).⁷⁰
催化剂形状对气体环境变化的动态响应超出了之前的预期。为了更深入地了解催化机制，对这一领域的研究变得越来越重要。然而，由于允许压力的限制，上述原位 ETEM 实验无法充分捕捉纳米颗粒在不同气体压力的实际工作条件下的特性。随着气体封闭技术的进步，在高温环境条件下观察金属催化剂已成为可能。Wang 及其同事通过原位 TEM 研究了 Ni 纳米颗粒在不同温度下的动态氧化过程，揭示了与温度相关的氧化行为（ 图 11a）。⁶⁹ 在相对较低的温度 （600 °C） 下，Ni 纳米颗粒的氧化经历了经典的 Kirkendall 过程，并伴随着氧化壳的形成。相比之下，在较高温度 （800 °C） 下，氧化从金属表面的单个晶核开始，然后沿着金属/氧化物界面进行，在整个过程中不会形成空腔。Song 等人使用原位高分辨率 TEM 结合分子动力学发现，Au 纳米粒子的不对称 5 折叠孪晶中存在张力，这是由于电子束下 5 折叠轴上的位错滑移和位错反应的双边界迁移所致。一层或两层双翼飞机的迁移由能量载流子控制。但一般来说，包括表面、晶格应变和边界能在内的总能量是在连续的边界迁移后释放的（ 图 D）。 11b）。⁷⁰

5.3 Catalytic growth of nanocarbon
5.3 纳米碳的催化生长

Due to the extensive potential applications of carbon nanotubes in various nanotechnologies, there has been in-depth research into their growth mechanisms. Simultaneously, the growth of carbon plays a crucial role in catalytic reactions. Hofmann et al. conducted a notable study focused on the growth of carbon nanotubes.⁷¹ Their work delved into the fundamental formation mechanisms of single walled carbon nanotubes (Fig. 12a–f) and nanofibers (Fig. 12g–j) containing metallic nanoparticles, utilizing TEM. The research sheds light on how these structures develop via chemical vapor deposition. The study reveals that the nucleation of single walled carbon nanotubes is initiated by metal catalysts, and their subsequent growth is accompanied by dynamic reshaping. The formation process of carbon nanofibers involves the elongation and contraction of the nanoparticle catalyst. Wang et al. achieved a precise identification of the active phase within cobalt catalysts for carbon nanotube growth.⁷² In their work, representative TEM images (Fig. 12k–o) showcased the growth of CNTs via both the base and tangential growth modes originating from the catalyst nanoparticles. The growth rate of the CNTs averaged around 10 nm s⁻¹, which exhibit an oscillatory pattern. FFT patterns (Fig. 12p–t) of the catalyst nanoparticles revealed a single crystal nature. The study observed dynamic alterations in the nanoparticle morphology (Fig. 12k–m), transitioning from a spherical shape (Fig. 12k) to an ellipsoidal one (Fig. 12m). Additionally, the orientation of the Co₃C nanoparticles underwent shifts during CNT growth, with the zone axis transforming from [142] to [613] (Fig. 12p–q). This research meticulously investigated the phase evolution of Co catalyst nanoparticles across the incubation, nucleation, and growth stages of CNTs under near atmospheric pressure conditions, utilizing an in situ ETEM. Rigorous statistical analysis of electron diffraction patterns was conducted to accurately determine the phases of the catalyst nanoparticles.
由于碳纳米管在各种纳米技术中的广泛潜在应用，人们对其生长机制进行了深入研究。同时，碳的生长在催化反应中起着至关重要的作用。Hofmann 等人进行了一项著名的研究，重点是碳纳米管的生长。⁷¹ 他们的工作利用 TEM 深入研究了含有金属纳米颗粒的单壁碳纳米管（ 图 12a-f）和纳米纤维（ 图 12g-j）的基本形成机制。该研究阐明了这些结构是如何通过化学气相沉积形成的。研究表明，单壁碳纳米管的成核是由金属催化剂引发的，其随后的生长伴随着动态重塑。碳纳米纤维的形成过程涉及纳米颗粒催化剂的伸长和收缩。Wang 等人精确鉴定了钴催化剂中用于碳纳米管生长的活性相。⁷² 在他们的工作中，具有代表性的 TEM 图像（ 图 12k-o）展示了 CNT 通过源自催化剂纳米颗粒的基向和切向生长模式的生长。CNT 的生长速率平均约为 10 nm ^s-1，表现出振荡模式。催化剂纳米颗粒的 FFT 图谱（ 图 12p-t）揭示了单晶性质。该研究观察到纳米颗粒形态的动态变化（ 图 12k-m），从球形（ 图 12k）过渡到椭球形（ 图 12m）。 此外，在 CNT 生长过程中，Co₃C 纳米颗粒的取向发生了变化，区轴从 [142] 转变为 [613]（ 图 12p-q）。这项研究利用原位 ETEM 仔细研究了在接近大气压条件下 Co 催化剂纳米颗粒在 CNT 的孵育、成核和生长阶段的相演变。对电子衍射图谱进行了严格的统计分析，以准确确定催化剂纳米颗粒的相。

6. Application of in situ TEM in thermal catalysis

6.1 Oxidation and reduction reactions

Noble metals, as reaction catalysts, exhibit outstanding catalytic performance and good stability. However, the high cost often poses a barrier. Therefore, the search for low cost and non-toxic catalysts is currently a widely recognized concern. Supported nickel nanoparticles have garnered significant attention due to their high activity. Nevertheless, the severe deactivation caused by coking and mass loss during continuous production remains a major issue for nickel catalysts. Hence, establishing a correlation between catalytic performance and structural information is of paramount importance for understanding the deactivation mechanisms. Chenna et al. employed in situ TEM to investigate the structural and phase alterations of nickel nanoparticles under varying redox conditions.⁷³ Their findings revealed that when exposed to oxidative conditions (at 130 Pa with CH₄ : O₂ ratio of 2 : 1), the originally spherical metallic nickel nanoparticles supported on silica underwent a transformation into hollow NiO nanoparticles at around 300 °C (Fig. 13a and b). Unfortunately, these hollow NiO nanoparticles exhibited no catalytic activity. Across all methane (CH₄) environments, the NiO nanoparticles gradually transitioned into metallic Ni nanoparticles (Fig. 13c and d) at elevated temperatures. This transition led to the development of core–shell structures consisting of Ni–NiO. Nevertheless, the catalytic performance of Ni–NiO was notably poor, and significant activity recovery only occurred following reduction of the catalysts at higher temperatures. Zhang et al. conducted in situ observations regarding the crystalline structural transitions of Pt–Sn intermetallic nanoparticles during deactivation and subsequent regeneration.⁷⁴In situ TEM images are presented in Fig. 13e–g, showcasing the progressive decarbonization process. Moreover, author explored the reverse regeneration of oxidized PtSn@mSiO₂ (Fig. 13k–p). The focus was on the most prominent nanoparticles within the observed field. Over a brief interval, the constituents within the nanoparticles initiated fusion, as depicted in Fig. 13k and l. Subsequently, after approximately 11 s (Fig. 13m), the nanoparticles underwent further fusion, leading to a nearly completely merged region on the left side of the nanoparticles. The reduction process led to the expansion of this fused region (Fig. 13n–p), ultimately resulting in the formation of uniform nanoparticles (Fig. 13p). Author observations indicated that the oxidized PtSn@mSiO₂ nanoparticles could regain their original spherical shape following hydrogen reduction.

6.2 Ammonia synthesis and methane pyrolysis
6.2 氨合成和甲烷热解

Catalyst promoters are employed in industrial catalysts to enhance catalytic activity and selectivity, and they themselves remain unreactive. The morphology plays a decisive role in determining the exposed surface facets and atomic active sites, thereby enabling control over activity and stability. As a result, the catalyst's structural changes under various environmental conditions constitute a critical determinant of its catalytic performance. The alterations in material morphology during chemical reactions can be tracked through in situ TEM, thereby unveiling the underlying reaction mechanism. Hansen et al. utilized in situ TEM to investigate the effects of Ba promotion on a Ru catalyst employed in ammonia synthesis.⁷⁵ Their findings revealed that during the course of the reaction, the Ru nanoparticles' surfaces become enveloped in an amorphous coating, with the morphology of the Ru nanoparticles remaining unchanged. This observation signifies that even as the catalytic activity is enhanced, the overall count of active surface sites remains unaltered. Through analysis of TEM images and EELS spectra, the researchers observed that Ba exhibits notable mobility and predominantly exists in oxide form. Throughout the reaction, the amorphous coating on the majority of Ru surfaces incorporates Ba and oxygen (Fig. 14a–d). Xi et al. utilized in situ TEM with high spatial and temporal resolutions to monitor the dynamic process of nanoporous gold (NPG) catalyzing methane pyrolysis. Their work revealed an intricate interplay between heterogeneous catalysis processes involving both nanocatalysts and single atom catalysts, thereby demonstrating the existence of dynamic co-catalysis.⁷⁶ In the images, a carbon zone, outlined by dashed circles in Fig. 14e–g, was observed at a distance from the NPG surface. This carbon zone exhibited changing contrast within a brief 2 s interval. In contrast, when the CH₄ flow was halted, the contrast changes transformed from intense to rather gentle and eventually disappeared, as illustrated in Fig. 14e–g. This comparative analysis highlights that the pronounced contrast alterations within the distant amorphous carbon segments weren't a result of the TEM's electron beam, but rather were attributed to specific reactions occurring within the carbon regions. Further insight is provided by Fig. 14h–j, which captures an amorphous carbon region removed from the NPG surface. In this region, a nanoparticle with an Au crystal structure momentarily materialized and then swiftly vanished before 0.4125 s during CH₄ pyrolysis. Concurrently, the contrast within the carbon region experienced changes similar to those observed in the zone marked by dashed circles. As a result, the cumulative observations suggest that the catalytic reaction of CH₄ pyrolysis was jointly facilitated by Au single atom catalysts and NPG.
催化剂促进剂用于工业催化剂中以增强催化活性和选择性，并且它们本身保持无反应性。形态在确定暴露的表面刻面和原子活性位点方面起着决定性的作用，从而能够控制活性和稳定性。因此，催化剂在各种环境条件下的结构变化构成了其催化性能的关键决定因素。可以通过原位 TEM 跟踪化学反应过程中材料形态的变化，从而揭示潜在的反应机理。Hansen 等人利用原位 TEM 研究了 Ba 促进对氨合成中使用的 Ru 催化剂的影响。⁷⁵ 他们的研究结果表明，在反应过程中，Ru 纳米颗粒的表面被包裹在无定形涂层中，而 Ru 纳米颗粒的形态保持不变。这一观察结果表明，即使催化活性增强，活性表面位点的总数保持不变。通过分析 TEM 图像和 EELS 光谱，研究人员观察到 Ba 表现出显着的迁移率，并且主要以氧化物形式存在。在整个反应过程中，大多数 Ru 表面的无定形涂层含有 Ba 和氧（ 图 14a-d）。习 et al. 利用具有高空间和时间分辨率的原位 TEM 来监测纳米多孔金 （NPG） 催化甲烷热解的动态过程。 他们的工作揭示了涉及纳米催化剂和单原子催化剂的非均相催化过程之间错综复杂的相互作用，从而证明了动态共催化的存在。⁷⁶ 在图像中，在距离 NPG 表面一定距离处观察到一个碳区，图 14e-g 中由虚线圆圈勾勒出。该碳区在短短的 2 s 间隔内表现出不断变化的对比度。相比之下，当 CH₄ 流停止时，对比变化从强烈转变为相当温和，最终消失，如图 14e-g 所示。这项比较分析强调，远处无定形碳链段内明显的对比度变化不是 TEM 电子束的结果，而是归因于碳区域内发生的特定反应。 图 14h-j 提供了进一步的见解，它捕获了从 NPG 表面去除的无定形碳区域。在该区域中，在 CH₄ 热解过程中，具有 Au 晶体结构的纳米颗粒短暂出现，然后在 0.4125 s 之前迅速消失。同时，碳区域内的对比度经历了类似于在虚线圆圈标记的区域中观察到的变化。因此，累积观测表明，CH₄ 热解的催化反应是由 Au 单原子催化剂和 NPG 共同促进的。

6.3 CO oxidation
6.3 一氧化碳氧化

The dynamic structure of a catalyst dictates the accessibility of its surface active site. However, the manner in which nanoparticle catalysts reconfigure under reaction conditions and how these changes correlate with catalytic activity remain largely unexplored. Yoshida et al. presented a study on the surface reconstruction of Au nanoparticles supported by CeO₂ when exposed to CO at room temperature. This reconstruction process was visualized in real space and time, shedding light on atomic behavior.⁷⁷ The investigation involved employing TEM with lower energy electrons to explore both the adsorption and the subsequent reconstruction of the gold nanoparticle. The study revealed that under vacuum conditions, the (100) facets of the nanoparticles remained unreconstructed, as highlighted in the magnified images depicted in Fig. 15a. However, in a specific reaction environment, as shown in Fig. 15b, the atomic columns of gold on the topmost and second topmost (100) layers shifted from their normal positions on the clean surface. The reconstructed surface exhibited a distinct arrangement: the topmost Au atoms formed an undulating hexagonal lattice, while the second topmost layer adopted a slightly distorted normal square lattice. This research proposed that the modified stacking sequence of the Au (100) surface contributed to unique bonding configurations favorable for CO adsorption with high affinity. Chee et al. employed TEM to reveal intriguing findings regarding the reversible structural and activity alterations of Pd nanoparticles when subjected to heating and cooling cycles in mixed gas environments containing both O₂ and CO. Remarkably, Pt and Rh nanoparticles did not demonstrate such reversible transformations under similar conditions.⁷⁸ In Fig. 15c, Pd nanoparticles with nearly atomically flat (100) and (111) facets at 200 °C are depicted, featuring sharp corners. However, these nanoparticles adopted a more rounded shape at temperatures of 500 °C and above. Notably, when the temperature was subsequently reduced back to 300 °C, the facets of the nanoparticles reformed. In contrast, Pt nanoparticles exhibited more subtle restructuring, as depicted in Fig. 15d. Upon introducing CO into the O₂ environment at 200 °C, the Pt nanoparticles appeared to undergo slight shape changes while maintaining their rounded appearance. Elevating the temperature to 500 °C and 600 °C resulted in some faceting of the Pt nanoparticles, albeit less pronounced than the changes observed in Pd. For Rh nanoparticles (Fig. 15e), no significant structural modifications or notable changes in CO₂ production were observed between 500 °C and 600 °C. Lowering the temperature from 500 °C led to a decline in CO₂ production, yet no discernible oxidation of the nanoparticles was evident from the images.
催化剂的动态结构决定了其表面活性位点的可及性。然而，纳米颗粒催化剂在反应条件下重新配置的方式以及这些变化与催化活性的相关性在很大程度上仍未得到探索。Yoshida 等人提出了一项关于在室温下暴露于 CO 时由 CeO₂ 支持的 Au 纳米颗粒表面重建的研究。这个重建过程在真实空间和时间中可视化，揭示了原子行为。⁷⁷ 该研究涉及使用具有较低能量电子的 TEM 来探索金纳米粒子的吸附和随后的重建。研究表明，在真空条件下，纳米颗粒的 （100） 个面仍未重建，如图 15a 所示的放大图像所示。然而，在特定的反应环境中，如图 15b 所示，最顶部和最第二层 （100） 上的金原子柱从它们在清洁表面上的正常位置移动。重建的表面呈现出独特的排列：最顶层的 Au 原子形成起伏的六边形晶格，而第二层最顶层采用略微扭曲的正常方晶格。本研究提出，Au （100） 表面的修饰堆叠序列有助于形成有利于高亲和力 CO 吸附的独特键合构型。Chee 等人使用 TEM 揭示了关于 Pd 纳米颗粒在含有 O₂ 和 CO 的混合气体环境中受到加热和冷却循环时可逆结构和活性变化的有趣发现。 值得注意的是，Pt 和 Rh 纳米颗粒在类似条件下没有表现出这种可逆转变。⁷⁸ 在图 15c 中，描绘了在 200 °C 时具有近原子平坦 （100） 和 （111） 面的 Pd 纳米颗粒，具有尖锐的角。然而，这些纳米颗粒在 500 °C 及以上的温度下采用了更圆润的形状。值得注意的是，当温度随后降低到 300 °C 时，纳米颗粒的面发生了变化。相比之下，Pt 纳米颗粒表现出更微妙的重组，如图 15d 所示。在 200 °C 下将 CO 引入 O₂ 环境中后，Pt 纳米颗粒似乎发生了轻微的形状变化，同时保持了其圆形外观。将温度提高到 500 °C 和 600 °C 导致 Pt 纳米颗粒出现一些刻面，尽管没有在 Pd 中观察到的变化那么明显。对于 Rh 纳米颗粒（ 图 15e），在 500 °C 和 600 °C 之间没有观察到 CO₂ 产生的显着结构改变或显着变化。 将温度从 500 °C 降低导致 CO₂ 产生量下降，但从图像中没有明显的纳米颗粒氧化。

7. Summary and challenge
7. 总结与挑战

The advancement of in situ transmission electron microscopy (TEM) presents promising prospects for delving into the catalytic mechanisms of materials. This technique enables the direct observation of material metamorphosis and facilitates spectroscopic scrutiny of reactive and conversion procedures. This paper succinctly outlines the extensive array of applications for in situ TEM methodologies in the investigation of catalytic materials, particularly in recent times. Furthermore, it offers our perspectives on the forthcoming enhancements of this potent methodology. The compilation of notable contributions can be found in Table 4. By employing the in situ TEM technique, scientists have succeeded in creating a controlled environment around the specimen. This environment allows for the acquisition of atomic scale insights into materials' structure, composition, and chemical characteristics. Undoubtedly, this approach has ushered in a revolutionary shift in research within these domains, enabling a profound comprehension of processes that were previously elusive to other technologies. Despite the remarkable strides made over the last decade, several challenges persist. For instance, the resolution of this method still requires enhancement due to the scattering effects of materials like SiN_x and the presence of gaseous media. Furthermore, the utilization of these media introduces instability and human associated factors into microscopic observations, placing physical constraints on the spatial and temporal resolutions achievable through TEM. From our viewpoint, sustained improvements are necessary in the following areas to unlock greater research possibilities in the future.
原位透射电子显微镜 （TEM） 的进步为深入研究材料的催化机制提供了广阔的前景。该技术可以直接观察材料，并有助于对反应和转化程序进行光谱检查。本文简洁地概述了原位 TEM 方法在催化材料研究中的广泛应用，尤其是在近代。此外，它还为我们提供了对这种有效方法即将到来的增强的看法。值得注意的贡献汇编可以在表 4 中找到。通过采用原位 TEM 技术，科学家们成功地在样品周围创造了一个受控环境。这种环境允许获得对材料结构、组成和化学特性的原子级见解。毫无疑问，这种方法为这些领域的研究带来了革命性的转变，使人们能够深刻理解以前其他技术难以理解的过程。尽管在过去十年中取得了显著的进步，但仍然存在一些挑战。例如，由于 SiN_x 等材料的散射效应和气体介质的存在，这种方法的分辨率仍然需要提高。此外，这些介质的使用将不稳定性和人为相关因素引入显微镜观察，对通过 TEM 可实现的空间和时间分辨率施加了物理限制。在我们看来，以下领域必须不断改进，以在未来释放更大的研究可能性。

Firstly, in situ TEM experiments have showcased their capacity to manipulate atmospheric conditions. However, pinpointing the precise atomic structure at the surface remains a challenge within gas nano-reactors. This challenge arises due to electron scattering stemming from both the gas layer and the upper and lower window films. Typically, gas nano-reactors employ SiN_x, with a thickness ranging from 10 to 50 nm, while the gas channel's height is maintained at less than 1 mm. In the foreseeable future, there arises a need for ultrathin and diminutive electron scattering films for the reactor's windows. Concurrently, the prospect of reducing the gas channel's height holds promise for elevating the spatial resolution of the resultant images.
首先， 原位 TEM 实验展示了它们纵大气条件的能力。然而，在气体纳米反应器中精确定位表面的精确原子结构仍然是一个挑战。这一挑战是由于气体层和上下窗膜产生的电子散射而出现的。通常，气体纳米反应器采用 SiN_x，厚度范围为 10 至 50 nm，而气体通道的高度保持在 1 mm 以下。在可预见的将来，反应器的窗口需要超薄和小巧的电子散射膜。同时，降低气体通道高度的前景有望提高结果图像的空间分辨率。

Secondly, while in situ TEM proficiently captures dynamic insights into catalyst microstructures and chemical compositions within chemical environments, a comprehensive comprehension of catalysis requires more than just microscopic investigations. Given the large number of catalytic reactions and their complex mechanisms, a holistic approach requires extensive and statistically meaningful measurements, including both catalysts and reactants. To achieve effective and thorough analyses, the integration of in situ TEM with other characterization methods emerges as a valuable strategy. This amalgamation presents an opportunity to glean broader perspectives on catalytic reactions. Of paramount importance is the real-time qualitative and quantitative assessment of evolving reactants, a key aspect in establishing the structure–activity relationships. Nevertheless, due to the scant quantities of catalysts used in in situ TEM studies, conventional techniques like gas chromatography and quadrupole mass spectrometry often fall short in detecting minute product quantities. In response, an array of other in situ characterization methodologies have rapidly emerged, including in situ Raman spectroscopy, in situ infrared spectroscopy, in situ X-ray diffraction (XRD), in situ nuclear magnetic resonance (NMR), in situ X-ray photoelectron spectroscopy (XPS), and in situ X-ray absorption fine structure (XAFS). These techniques have complemented the field, addressing the limitations of conventional methods. Furthermore, the synergy between advanced in situ methodologies and traditional characterization instruments paves the way for the development of novel characterization approaches, enabling a comprehensive understanding of catalytic reactions.
其次，虽然原位 TEM 能够熟练地捕捉化学环境中催化剂微观结构和化学成分的动态见解，但对催化的全面理解需要的不仅仅是微观研究。鉴于催化反应的数量及其复杂的机理，整体方法需要广泛且具有统计学意义的测量，包括催化剂和反应物。为了实现有效和彻底的分析，将原位 TEM 与其他表征方法相结合是一种有价值的策略。这种合并为收集催化反应的更广泛视角提供了机会。最重要的是对演化的反应物进行实时定性和定量评估，这是建立结构-活性关系的关键方面。然而，由于原位 TEM 研究中使用的催化剂数量很少，气相色谱法和四极杆质谱法等传统技术往往无法检测微小的产物量。作为回应，一系列其他原位表征方法迅速出现，包括原位拉曼光谱、 原位红外光谱、 原位 X 射线衍射 （XRD）、 原位核磁共振 （NMR）、 原位 X 射线光电子能谱 （XPS） 和原位 X 射线吸收精细结构 （XAFS）。这些技术补充了该领域，解决了传统方法的局限性。 此外，先进的原位方法和传统表征仪器之间的协同作用为开发新的表征方法铺平了道路，从而能够全面了解催化反应。

Thirdly, in situ observations based on a cryo bias holder in cryo-TEM can dynamically capture the nucleation and growth of sensitive materials. Advances in TEM in terms of resolution significantly benefit in situ characterization. By combining with high resolution three dimensional reconstruction techniques, the spatial distribution of materials can be visualized. Additionally, developing cryo-TEM coupled with a spectrometer can yield the real time structure evolution of materials and synchronous regional component analysis during operando experiments. However, the cryogenic temperature is a significant challenge for the operation of typical instruments.
第三，在冷冻透射电镜中基于冷冻偏压支架的原位观察可以动态捕捉敏感材料的成核和生长。TEM 在分辨率方面的进步显着有益于原位表征。通过与高分辨率三维重建技术相结合，可以可视化材料的空间分布。此外，开发冷冻透射电镜与光谱仪相结合可以产生材料的实时结构演变和原位实验期间的同步区域成分分析。然而，低温是典型仪器运行的一个重大挑战。

Finally, the temporal resolution stands as a crucial factor in capturing transient reaction dynamics effectively. In recent times, researchers have made notable progress by utilizing an independently developed electronic detection camera. This advancement has yielded an impressive resolution of 0.04 nm under low voltage (80 kV) imaging conditions. However, it's important to ensure a sufficient number of electrons to generate clear and informative images. This requisite hinges upon the brightness of the electron gun and the interaction between the electron beam and the specimen. Comparatively, electron diffraction offers a faster acquisition process without compromising quality when contrasted with imaging and spectroscopy. A particularly noteworthy domain in this regard is ultrafast electron microscopy, particularly ultrafast electron diffraction. This field is experiencing rapid growth as it delves into material structure changes with significantly elevated temporal resolutions.
最后，时间分辨率是有效捕获瞬态反应动力学的关键因素。近年来，研究人员通过使用自主开发的电子检测相机取得了显着进展。这一进步在低电压 （80 kV） 成像条件下产生了令人印象深刻的 0.04 nm 分辨率。然而，重要的是要确保足够数量的电子来生成清晰和信息丰富的图像。这要求取决于电子枪的亮度以及电子束与样品之间的相互作用。相比之下，与成像和光谱相比，电子衍射提供了更快的采集过程，而不会影响质量。在这方面，一个特别值得注意的领域是超快电子显微镜，特别是超快电子衍射。该领域正在经历快速增长，因为它深入研究了时间分辨率显著提高的材料结构变化。

In the foreseeable future, we anticipate that the progression of in situ TEM techniques will continue to advance on several fronts: (1) incorporation with more in situ capabilities: the expansion of in situ TEM methods is likely to encompass a broader range of capabilities, including magnetic field control and measurement, laser pump-probe techniques, and the ability to operate within varying environmental atmospheres. (2) Enhanced temporal resolution: as electron detection technologies evolve, the temporal resolution of in situ TEM is poised for improvement. The development of swifter electron detectors and innovative dynamic TEM techniques holds the promise of achieving finer temporal resolutions. (3) Robust data analysis tools: as the complexity and volume of data associated with high spatial and temporal resolution or information rich characterizations like 4D-STEM increase, the need for robust and standardized data analysis tools becomes paramount. The field is expected to develop efficient tools to handle high throughput raw data, facilitating meaningful insights from advanced characterizations. Overall, these anticipated developments are set to usher in an exciting phase of growth and innovation in the realm of in situ TEM.
在可预见的未来，我们预计原位 TEM 技术的发展将在几个方面继续向前发展：（1） 结合更多的原位功能： 原位 TEM 方法的扩展可能包括更广泛的能力，包括磁场控制和测量、激光泵浦探测技术以及在不同环境环境中运行的能力。（2） 提高时间分辨率：随着电子检测技术的发展， 原位 TEM 的时间分辨率有望得到提高。更快的电子探测器和创新的动态 TEM 技术的发展有望实现更精细的时间分辨率。（3） 强大的数据分析工具：随着与高空间和时间分辨率或信息丰富的特征（如 4D-STEM）相关的数据的复杂性和数量增加，对强大和标准化的数据分析工具的需求变得至关重要。该领域有望开发高效的工具来处理高通量原始数据，从而促进从高级表征中获得有意义的见解。总体而言，这些预期的发展将在原位 TEM 领域迎来一个令人兴奋的增长和创新阶段。

Conflicts of interest

The authors declare no conflict of interest.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (52064035) and the Key Research and Development Program of Gansu Province (21YF5GA078). The authors thank the Electron Microscopy Center of Lanzhou University.

References

Footnote