4.1 A Brief History of In-Situ Heating TEM

In the past, people could only obtain static information from the sample, but advancing science and technology make it possible to apply in-situ stimuli and observe the real-time response. Microheaters allow samples to be thermally stimulated and research on dynamic properties (such as phase transitions, materials growth, sublimation, and catalysis) to rise.

The first likely dynamic STEM observation of the motion of atoms on the surface of thin substrates was reported by Crewe in 1979 [1]. The movement of clusters/atoms and the nucleation process have been observed over periods of several hours. This phenomenon was ascribed to thermal effects and inspired further development of technology to improve the contrast and resolution. Soon after, the first in-situ HRTEM imaging of dislocation formation, motion, interaction, and annihilation in gold due to electron beam irradiation was carried out using a TV system by Hashimoto et al. [2]. A more convincing example of atomic movement caused by heating from an incident electron beam was reported by Sinclair et al. [3]. They captured moving species within the crystal structure of the parent material (CdTe) rather than single atomic jumps and the motion rate became higher with longer exposure (Fig. 4.1). When the beam current density was reduced, the response of the motion rate was slow, which implied that this motion did not result from direct electron-atom collisions or ionization effects but presumably arose due to heating dissipation from the active area.

Fig. 4.1
Four greyscale photomicrographs represent the atomic motion induced by electron beam heating. They have a speckled pattern on the left top and dark granule and grid patterns below.

Atomic motion induced by electron beam heating. Four panels are images captured several minutes apart (reproduced with permission from Ref. [3], Copyright 1981, Springer Nature)

Full size image

Increasing demand for controllable elevated temperatures during in-situ observation in TEM stimulated Parker’s introduction of commercial heating holders in 1986 [4]. The holders exhibited acceptable mechanical and thermal stability to allow high-resolution imaging during the heating process. Besides, a video-rate recording was also realized to follow the dynamic behavior over long periods of time. To demonstrate the technique, Parker et al. observed silicon re-growth at temperatures between 500 and 800 °C. A silicon thin film with a thickness of 300 nm had been deposited on a sapphire substrate and treated by dual ion-implant and annealing. The corresponding TEM specimens were made using a standard cross-section technique. Heating was implemented with a commercial holder (model number PW 6592), where a platinum pad was heated by an electrical feed-through thus the 3-mm disk sample on it was thermally stimulated. There was a thermocouple attached to the pad to measure the temperature and the reading should be fairly close to the actual temperature due to the high thermal conductivities of sapphire and silicon. Reproducible imaging conditions were carefully controlled to keep any beam heating or other effects at a systematic error level. In their study, they have found that crystallization and defect reactions in silicon are analogous to their predictions that atoms would rearrange at around 600 °C and this phenomenon is similar to that observed in CdTe by electron beam heating [3]. At this point, it is reasonable to expect an atomic-level investigation of structural transformation in many materials during in-situ heating using a similar strategy.

4.2 Current In-Situ Heating TEM Technologies

4.2.1 Operation Mode

TEM can provide extraordinary imaging capability (sub-angstrom spatial resolution) and acquire diffraction and spectroscopy. It generally operates under either TEM mode or STEM mode. TEM mode corresponds to the application of a parallel electron beam while STEM mode uses a focused, small probe scanning over the sample. In TEM mode, HRTEM images are formed by interference between transmitted and diffracted electron beams (phase contrast), but owing to lens aberrations and sample thickness effects, the interpretation of HRTEM images is complicated. In comparison, the interpretation of the image contrast at STEM mode with a high-angle annular dark-field (HAADF) detector is more straightforward, because it is nearly proportional to Z2 where Z stands for the atomic number. Z-contrast of HAADF-STEM allows visualization of cation mixing at the atomic scale. Furthermore, the combination of STEM and energy-dispersive x-ray spectroscopy (EDX) or electron energy loss spectroscopy (EELS) can obtain chemical information, such as elemental distribution and oxidation state of the materials simultaneously with capture images [5,6,7]. However, since the low scanning rates may limit the acquisition of instant change of materials, the majority of in-situ observation during heating is operated under TEM mode rather than STEM mode [8].

4.2.2 Type of Heating Holders

The temperature of the specimen can be elevated by using two kinds of heating holders (Fig. 4.2): a furnace-type heating holder (left) that is designed to use standard TEM grids, or a microfabricated-heater-based holder where a localized area is heated through Joule heating [9]. The furnace-based heating holder has long been commercialized and it enables the sample to be heated up to 1300 °C [10, 11]. Unlike the in-situ mechanical experiments in TEM, achieving the sample heating inside a TEM is much easier. For example, Luo et al. investigated the thermal stability of Ni-based superalloy [12], and Liu et al. successfully attached monodisperse Cu nanoparticles onto carbon nanotubes during in-situ heating in a TEM equipped with a Gatan 652 double-tilt heating holder [13]. Nevertheless, the thermal expansion of the stage leads to a non-ignorable image shift during an increase in temperature so it is hard to realize the atomic resolution observation while heating. Especially at temperatures higher than 500 °C, the use of recirculating cooling water to protect the heating unit may also bring vibrations and thereby reduce the imaging stability and resolution. It is challenging to do tests on low-dimensional materials because their morphologies and structures are more sensitive to thermal stimulation. In contrast, the recently developed MEMS-based heating holders only go through negligible mechanical vibrations since heating is localized in a specific small area in the chip and thus allow atomic resolution imaging [14, 15]. For instance, Janish et al. studied nucleation in Ta and captured high-resolution images of nucleating crystallites [16]. In general, the advantage of using furnace-type of holders is that sample preparation is much easier, but the microchip-based holders have better performance regarding temperature control and mitigation of sample drift.

Fig. 4.2
Two schematics. The furnace heating holder lists the lock ring, T E M grid, and heating furnace. M E S device lists a heating surface, S I C membrane, biasing contacts, and an array of holey carbon windows.

Schematics show two types of heating devices: Furnace heating (left) and MEMS device (right) (reproduced with permission from Ref. [9], Copyright 2018, American Chemical Society)

Full size image

4.2.3 Microheaters

Microheaters are the core of heating holders. The heating zone is usually the region of interest and is placed at the center of the heater since the temperature there is relatively homogeneous. Generally, a microheater includes a silicon substrate or sometimes gallium arsenide and thermally insulating dielectric thin film layers on top [17]. There is an electrically conducting thin film between the substrate and the top insulating layer, where heat is produced by passing electric current provided by the voltage of the current source (This process is called Joule heating). The electrically conducting film can be metal, ceramic or doped polysilicon. Metal is the most frequently used material in microheaters. There are three main reasons. First, since metal has excellent ductility and conductivity, it enables flexible geometrical design and fast response to exerted voltages. Second, due to the near-linear relationship between temperature and resistance, precise temperature control through resistance measurements can be achieved through a four-point probe. Third, the biggest drawback of metal (i.e. high reaction activity in the environment) can be simply overcome by covering a passivation layer. Ceramic is chemically inert and has high-temperature resistance hence it can be used in tough environments. Doped polysilicon can be easily integrated into standard CMOS processes since it does not limit the following steps as is the issue with metallization [18, 19].

The performance of a microheater is evaluated based on parameters falling into two categories. First, mechanical robustness and stability include mechanical stress, thermal stress, stress distribution, and spatial sample drift. Second, temperature and heat include range and lifetime, homogeneity, accuracy, stability, power consumption, and response time [19].

4.2.4 Synergy with Heating

In addition to single thermal excitation, a combination of heating and other stimuli has shown great potential for research on the dynamic properties of materials in a complicated environment. For example, Karki et al. investigated thermal degradation of overcharged Li0.1NCA in different atmospheres using an environmental TEM (ETEM) [20]. Oxidizing, neutral, or reducing environments were created by injecting O2, He and H2 gas into the sample area inside the ETEM. They found that surface oxygen loss and structural evolution were inhibited with the oxidizing environment while greatly enhanced under reductive conditions. Moreover, the understanding thermo-electrical performance of materials is also critical to better manipulate and optimize them for practical applications. For instance, it is crucial to monitor structural changes of ReRAM materials while both heating and biasing are applied to improve their stability for the fabrication of future non-volatile memory devices. Hence, Garza et al. [21] present the design of a system for in-situ biasing and heating called Lightning System as shown in Fig. 4.3. The left panel shows the structure of the nanochips for heating and biasing at the same time. There are eight electrical contacts and half of them are used for heating while the other four are used for biasing. This design ensures 4-point probe measurements for accurate control of the temperature and voltage/current which further guarantees a reproducible response. Then, this nanochip is attached to a customized holder (Right panel in Fig. 4.3), where contact needles are used to provide stimuli from outside of the holder. The system can supply voltage up to 100 V and temperature up to 800 °C while it also enables double-tilt during observation. In-situ simultaneous biasing and heating can carve a new path for a comprehensive understanding of materials and advance the development of nano-electronics. Another typical example has been reported by Wang et al., who utilizes the mechanical tests system during increasing the temperature aiming to select the appropriate alloys for aero-engines [22]. Their work also implies that the cooperation among microscopists, physicists, chemists, and engineers is essential for developing next-generation smart materials.

Fig. 4.3
Two parts. Part a has 4 schematics of nanochip that resemble a P C B board for electric field studies, I V measurements on nanowires and lamellas. Part b is a customized holder with a mounted nanochip.

Configuration of nanochip (left) and customized holder with mounted nanochip (right) (reproduced with permission from Ref. [21], Copyright 2016, John Wiley and Sons)

Full size image

4.3 Research Based on In-Situ Heating TEM

4.3.1 Material Growth

Controllable growth of functional nanomaterials is the starting point for their application in multitudinous devices. Therefore, to explore optimal crystal structures and morphologies, it is necessary to develop methods that can help deeply understand the actual mechanisms behind their growth. The growth of one-dimensional (1D) nanostructures is usually elucidated via a vapor–liquid-solid (VLS) theory. According to the VLS process, the liquid phase dissolves atoms from a solid matrix. When it is supersaturated, growth arises at the surface of the droplet [23]. As the nanostructure is extremely small, a method with a high spatial resolution is crucial to systematically investigate the growth process. The conventional ex-situ observations using microscopy can only analyze the static nanoscale structural information, however, they are incapable of following their forming process [24,25,26]. By comparison, the in-situ methods enable direct observation of structural transformation occurring at the nanoscale or even atomic scale [27]. For instance, by equipping a heating component, modern TEM can be employed not only for improving resolution but also for making it possible to track dynamic processes during nanowire growth [28, 29], as shown in Fig. 4.4a. During crystal growth, especially in a VLS process, the motion of a semiconductor–metal molten zone is generally motivated by a temperature gradient. Yet, the temperature gradient fields have not been completely understood in earlier ex-situ characterization [30, 31]. Hence, the real-time evidence acquired via in-situ methods provides an opportunity to accurately track liquid zone motion inside the solid [32,33,34].

Fig. 4.4
Two parts of 8 greyscale photomicrographs. Part a illustrates the growth of a copper wire at 13, 18, and 23 seconds and part b with a closer view of the A u tip at 5 different time intervals.

a TEM images showing the growth of Cu nanowires during heating (reproduced with permission from Ref. [29], Copyright 2018, American Chemical Society) b TEM images exhibiting the motion of Au/Ge alloy during heating (reproduced with permission from Ref. [35], Copyright 2015, American Chemical Society)

Full size image

Figure 4.4b illustrates an Au particle initially liquifies at the Ge nanowire tip, then creating an Au/Ge alloy liquid area which subsequently moves inside the Ge nanowire [35]. It should be clear that the motion direction and position of the Au/Ge liquid can be manipulated by the applied bias; upon removal of the bias, the Ge nanowire is detached from the Au tip and the liquid Au/Ge alloy quickly hardens. These results resolve the growth kinetics of nanowires, and also inspire more explorations of the growth process of nanoparticles and nanosheets. However, sample handling of them is more complicated. It is expected that a deep investigation of physical and chemical evolutions during material growth at such high resolutions would drive the development of surface and interface engineering.

Later, Cheng et al. found that the growth processes of PbSe nanocrystals (NCs) can be controlled by changing temperature [36]. The NCs begin to grow under the oriented attachment growth mode by attaching dot-shaped NCs along a certain crystal orientation, and sequentially switch to growth with grain-boundary migration by absorption of smaller NCs by neighboring larger ones through interfacial atom reconstruction.

4.3.2 Sublimation and Surface Energy

Besides crystal growth, sublimation is another important phase transition during which the chemical bonds within the solids break-down resulting in gasification. The in-situ experiment in TEM is a direct technique to investigate the sublimation mode at the atomic scale and it should also shed light on a better understanding of the inverted growth process. Additionally, sublimation is strongly linked to material stability, and the research on sublimation of NCs would reveal the critical factors influencing the stability. For example, though the surface free energies of the major crystal planes follow (111) < (100) < (110) in face-centered cubic (FCC) structures [37, 38], the thermally stable surface of FCC metals at elevated temperatures has been under debate for a long period because high-temperature-induced relaxation of crystal planes strongly depends on crystal index [39,40,41]. Hence the observation of the stable facets of FCC metals at high temperatures has significant importance for clarifying this debate. Ding et al. [42] and He et al. [43] studied the sublimation scenarios of Ag NCs via in-situ heating TEM. The featured shapes and surfaces during sublimation provide direct information for revealing the close linkage between surface free energy and stability (Fig. 4.5). Benefiting from the low drift of the microchip during heating, the dynamic sublimation process of the Ag NC at the very initial can be effectively captured, showing an overwhelming superiority comparing with the furnace-type heating holder.

Fig. 4.5
15 grayscale photomicrographs of a cube of 50 nanometers tilted at 40 degrees illustrate the evolution of the surface in different forms. They exhibit a sphere at 3 stages. Part p presents the decrease in volume of the structure and q plots a contact angle with increasing sublimation time for 2 stages.

TEM image sequences showing the surface evolution of a cubic Ag NC with a carbon shell (reproduced with permission from Ref. [43], Copyright 2017, John Wiley and Sons)

Full size image

Aside from this temperature-related surface stability, in fact, the surface energy of a NC itself is still under fierce debate when taking into account the size effect. Theories and simulations from the thermodynamic view suggest a decreasing trend of surface energy with decreasing particle size [44,45,46], while another view which considers the size-dependent lattice parameters proposes a rather contrary conclusion [47, 48]. On this issue, the in-situ heating TEM becomes the only effective tool for performing investigations from the experimental end. Sambles et al. probed into the surface energies of Au and Ag NCs by fitting the sublimation curves with Kelvin equation [49, 50]. Later, investigations using similar in-situ heating approach achieved more results for understanding the size effect in surface energy [51]. Although the current in-situ heating TEM still cannot clarify all the inconsistencies remained, the development of more delicate microchips with high controllability and accuracy can provide opportunities for modifying/calibrating the physical parameters of nanomaterials.

In addition to the surface energy, the particle size and defects are also of great importance, which further affects the stability of the nanostructure. Cheng et al. discussed the sublimation process of PbSe nanocrystals at the atomic scale and corresponding size, surface, and interface effects [52]. Because of the interplay between the electron beam and surface organic ligands, the sizes of nanocrystals are successfully manipulated via the introduction of the electron beam at different growth periods, which allows real-time observation of the size dependence of sublimation. As shown in Fig. 4.6, the nanoparticles with a size less than 10 nm exhibit directional orientational sublimation, and those of larger size sublimate uniformly.

Fig. 4.6
Nine greyscale photomicrographs compare the sublimation of a small P b S e nanocube with a scale of 2 nanometers. The structure of the nanocube decreases with respect to time at 0.0, 3.3, 6.1, 6.5, 12.6, 23.2, 23.8, 25.0, and 26.0 seconds.

Series of TEM images showing sublimation of a small PbSe nanocube is controlled by its facets (reproduced with permission from Ref. [52], Copyright 2020, Elsevier)

Full size image

4.3.3 Failure Analysis

In in-situ heating TEM, Joule heating is not only used to create a heating region serving as the substrate, but also it is frequently exerted directly in the sample to create synergistic effects. The combination of heat and electromagnetic field frequently leads to unique behaviors in nanomaterials [53]. For example, by controlling the stimulus of atom migration with an appropriate current (i.e. to create a certain temperature and electron-wind force), the mass transport of nano-objects can be implemented at the atomic scale [54,55,56,57]. This provides a fantastic method to modulate the nanostructures to enable diverse nanodevices and applications including mass sensors and resonators [58, 59], archival memory [60], oscillators [61], and nano-welding [62]. Aside from these electromigration-based mass transportation phenomena, the combination of thermal heating with electrical stimuli can also enable ion transport [63,64,65,66] and solid-state reactions [67,68,69,70,71]. Mei et al. [72] investigated the solid state reaction/alloying process between nanometer-sized Cu and Al metallic tips. It was found that hetero-joining could be implemented by the coupling of heating and electromigration.

Group III–V nanowires prepared by metal-catalyzed chemical vapor deposition have recently attracted an explosion of interest [73], driven by the inspiring optoelectronic properties, the high crystallinity, and the potential of integrating them with silicon. Recent research on Si nanowires suggests that when the nanowire diameter is below around 150 nm, the thermal conductivity is weakened because of phonon surface scattering [74]. Thermal breakdown in Joule-heated GaN nanowires is studied by in-situ TEM experiment [75]. The thermal conductivity of the nanowires is reckoned to be below the bulk GaN value. Breakdown in a single nanowire is observed to happen at a maximum temperature of about 1000 K, and nanowire morphology adjacent to the breakdown region suggests that failure happens via thermal decomposition, which is proved by in-situ TEM images captured at the failure stage.

4.3.4 Annealing and Phase Transitions

In-situ annealing provides more dynamic information of sample’s structure evolution than the ex-situ approaches. This enables an easy combination between the treating temperatures and the intermediate structures. For example, the FeCrAl alloys are thought to be promising materials of fuel claddings in nuclear reactors. To deeply understand the transformation of helium bubbles and dislocation loops during the annealing process, He+-irradiated TEM specimens were annealed and characterized by in-situ TEM [76]. The dislocation density reduced with the rise of temperature, but the loop size presented a different behavior. When the temperature was raised to 1072 K, the bubbles grew quickly while the density reduced. Moreover, in-situ thermal annealing of PtCu3/C in TEM has also been studied, which reveals essential transformations for achieving highly-active oxygen reduction [77].

Phase transitions are a well-known temperature-dependent process in materials, in-situ techniques in TEM allow direct observation of these processes [78]. For example, polymorphs of Ga2O3 [79] are potential candidates for ultra-wide bandgap semiconductors and are widely studied because of the newly developed growth technologies. The temperature-dependent phase transformation of κ-Ga2O3 layers grown on sapphire was investigated by high-resolution TEM [80]. Annealing procedures up to 1000 °C were performed in situ in TEM. This enabled the reveal of the mechanisms of κ-to-β phase transition and corresponding atomic rearrangement.

Likewise, the complex phase diagram of Pr1−xCaxMnO3 (PCMO) leads to possibilities of tuning the physical properties for various applications. Significantly, as a result of strong correlation effects, electronic and lattice degrees of freedom are strongly coupled. Therefore, it is arguable whether the bulk phase diagrams can be directly transferred to strained thin films. Beche et al. explored phase transitions at high temperatures in PCMO (x = 0.1) deposited as a 400 nm film on a SrTiO3 (STO) substrate [81]. Combination with TEM, individual domains of the nano-twinned films can be resolved compared with macroscopic X-ray or neutron diffraction studies [82].

Recently, Tang et al. [83] investigated the phase evolution of CuAg alloy NCs by in-situ heating TEM. They found that the transition from a solid solution phase to a Janus-type separated phase leads to the formation of different interfaces related to nanoparticle size. Small nanoparticles tend to form a Cu(100)/Ag(100) interface, while the large ones tend to form a Cu(111)/Ag(111) interface. Ni et al. [84] demonstrates that in-situ annealing also helps fine-tune the phase of Au–Ag nanorods to optimize their plasmonic properties. It is interesting that the microchip can also be used as a calibrator to estimate the beam heating effect inside the TEM. In this way, the electron beam can be used as an ultra-fast heating source to explore those transient phases that are hardly revealed by conventional methods. Shi et al. [85] used this technique approach to probe into the phase segregation process of PbBi nano-alloy. They have uncovered several metastable phases which haven’t been reported before (Fig. 4.7).

Fig. 4.7
Nine greyscale photomicrographs of several dots of different sizes at a scale of 10 nanometers. A few dots are circled in parts b and j and closer views are reported in parts c, d, e, f, and h. Part g is the frequency versus facets histogram.

Different morphologies, structures, and phases of PbBi nanoparticles during evolution under in-situ heating treatments (reproduced with permission from Ref. [85] Copyright 2022, American Chemical Society)

Full size image

4.3.5 Catalysis and Battery

Ordered intermetallic nanoparticles have shown electrocatalysis with improved activity and durability for the oxygen reduction reaction (ORR) [86,87,88]. Using in-situ heating electron microscopy, morphological evolution and the creation of fully and partially ordered nanocrystals have been observed at the atomic scale [89]. This comprehensive study will have a long-lasting influence on the development of ordered intermetallic electrocatalysts for real-world applications.

Temperature is one of the key factors limiting the application of batteries, thus in-situ heating can be employed to investigate the structural and chemical transformation of battery materials at elevated temperatures. Determining the rational working temperature ranges of certain battery materials is critical for developing high-performance batteries to be used in extreme conditons [90,91,92]. Thermal stability of anode materials may lead to safety issues, so different approaches are demanded to prevent thermal runaway and burning. Nam et al. [93] employed in-situ technologies in TEM to study the thermal stability of overcharged cathode materials, as illustrated in Fig. 4.8. Their work is practically significant since the working temperature is usually changeable rather than constant.

Fig. 4.8
Four greyscale photomicrographs of overcharged nanoparticles heating at different temperatures. Each resembles a hazy pattern and is labeled R, S, and R S.

HRTEM images of overcharged nanoparticles: a before heating b heating at 100 °C c heating at 200 °C and d heating at 300 °C (reproduced with permission from Ref. [93], Copyright 2013, John Wiley and Sons)

Full size image

4.3.6 Solid-State Amorphization and Crystallization

One of the most fascinating findings in the 1980s is the formation of amorphous phases by the interplay of two metals at their interfaces [94]. As semiconductor–metal interfaces are vital structures in integrated circuits, Holloway et al. focused on Ti-Si multilayers and discovered that there is also a solid-state amorphization reaction between the interfaces [95]. Additionally, during the early periods of silicidation, an amorphous barrier is formed at the crystalline interface, leading to a high Schottky barrier height [96].

In contrast to amorphization, there is unexpected crystallization observed in some eutectic systems (Al–Si or Ag–Si). In-situ high-resolution TEM reveals that rather than a metastable phase or liquid formation, the initial amorphous phase could be crystallized during adjusting the temperature [97]. Further investigation showed that this crystallization occurred by semiconductor elements diffusing into the metal, then precipitation of the semiconductor would happen after supersaturation [98]. A more thorough research of crystallization in a promising high-k dielectric tantalum oxide enabled a typical Avrami (Kolmogorov– Johnson–Mehl–Avrami) examination of the in-situ experiment. During the observation, the reaction was all the way tracked, thus the nucleation and the growth stages were simply separated for detailed analysis [99].

4.3.7 Degradation of Perovskite Solar Cells

One of the most serious problems limiting the practical application of perovskite solar cells is the lack of thermal stability. Different possible degradation pathways exist in this material system, especially hybrid composites. Therefore, it is inevitable to examine the response of perovskite materials at elevated temperatures. Divitini et al. monitored morphology, structural and chemical changes when heating the prototypes of perovskite solar cells. They present element diffusion and structure break-down and discovered distinct degradation modes in vacuum or air [15].

4.4 Conclusions and Outlook

In summary, in-situ heating TEM plays an increasingly important role in advancing the development of materials for precise structural design and accurate performance control to adapt them to practical applications. In-situ heating offers unique direct information that is not available by other ex-situ methods [100, 101]. In-situ experiments are usually carried out in “TEM mode” to achieve both high spatial resolution and fast recording while “STEM mode” allows more detailed analysis of chemical and electronic information of materials when combined with spectroscopic methods.

However, there are still some issues about in-situ heating in TEM. First, whether the observations are representative of the corresponding bulk materials. Second, the recording rate may still limit the capture of thermally activated processes. Third, data storage and processing capability may affect the in-depth analysis. Fourth, temperature measurement at the microscale remains challenging. Five, it still lacks research on complicated or composite systems, and current studies mainly focus on simple model material systems and reactions.

In the future, in-situ heating TEM should get the following items improved.

Measurement of electron beam effects and their mitigation. High energy electron beam possibly causes destructive changes to the TEM sample and leaves artifacts affecting the objective analysis. Therefore, electron dose, the energy of electrons, vacuum conditions, and nature of materials that determine interaction with the electron beam should be paid careful attention to [102].

High spatial resolution and temporal resolution. Based on more electron-sensitive complementary metal oxide semiconductor, higher special and temporal resolution can be realized, which is crucial for revealing the ultra-fast physical and chemical changes and also for investigating the e-beam irradiation sensitive materials [103]. Scanning rates of probe aberration-corrected STEMs are becoming much faster, so HAADF-STEM in-situ observation is expected to play increasingly vital roles in tracking temperature-dependent processes at the atomic scale. Together with EELS and EDS analysis, local changes in chemical composition and valence states of the elements can be monitored during the increasing/decreasing of the temperature [104].

Rapid data recording and processing. Recently developed faster-recording devices will promote the rate/speed at which changes induced by temperature elevation can be tracked, which will result in an enormous increase in data accumulation. The in-situ experiments at nano-second rates by laser pulsing the electron source will also demand much higher data processing power and artificial intelligence should be one of the feasible solutions [105].

Precise temperature measurement down to microscale or even nanoscale. Although estimation of temperature has been achieved via resistivity measurements [106], more accurate local temperature readouts remain challenging.

Multifunctional holders for simultaneous application of different stimuli. Advancement of versatile in-situ TEM approaches that can offer coupled external stimuli. The structural transformation under multiple external stimuli such as mechanical loading, thermal activation, and electric field, or even in a gas atmosphere [107], liquid environments [108], laser illumination [109, 110], magnetic fields [111, 112] etc., have also been observed nowadays. It is expected that future TEM technologies could incorporate multiple fields that simulate the real working environments for the materials.

Expanding material systems where in-situ heating is employed. It is expected that in-situ heating TEM would continue motivating intriguing discoveries in materials science and further promote industrial development. With the development of advanced unique heating holders combined with the ongoing improvement of the transmission electron microscope, the scope of corresponding in-situ heating studies is only limited by the imagination and innovation of the researcher. It should be believed that much fundamental understanding of material properties will be obtained with an increased resolution both spatially and temporally.