In Situ Transmission Electron Microscopy Studies of Electrochemical Reaction Mechanisms in Rechargeable Batteries
- 76 Downloads
Rechargeable batteries dominate the energy storage market of portable electronics, electric vehicles and stationary grids, and corresponding performance advancements are closely related to the fundamental understanding of electrochemical reaction mechanisms and their correlation with structural and chemical evolutions of battery components. Through advancements in aberration-corrected transmission electron microscopy (TEM) techniques for significantly enhanced spatial resolution, in situ TEM techniques in which a nanobattery assembly is integrated into the system can allow for the direct real-time probing of structural and chemical evolutions of battery components under dynamic operating conditions. Here, open-cell in situ TEM configurations can provide the atomic resolution imaging of the intrinsic response of materials to ion insertion or extraction, whereas the development of sealed liquid cells can provide new avenues for the observation of electrochemical processes and electrode-electrolyte interface reactions that are relevant to real battery systems. And because of these recent developments in in situ TEM techniques, this review will present recent key progress in the utilization of in situ TEM to reveal new sciences in rechargeable batteries, including complex reaction mechanisms, structural and chemical evolutions of battery materials and their correlation with battery performances. In addition, scientific insights revealed by in situ TEM studies will be discussed to provide guiding principles for the design of better electrode materials for rechargeable batteries. And challenges and new opportunities will also be discussed.
KeywordsIn situ TEM Liquid cell Reaction mechanism Rechargeable battery
Electrochemical rechargeable batteries, including lithium-ion batteries (LIBs) [1, 2, 3, 4], lithium-sulfur batteries (LSBs) [5, 6, 7], sodium-ion batteries (NIBs) [8, 9, 10], potassium-ion batteries (KIBs) [11, 12, 13] and metal-air/O2 batteries [14, 15, 16, 17], have been intensively investigated over the recent decades for growing application in portable electronics and electrical vehicles as well as in stationary power backup devices to mediate fluctuating energy sources in smart grid systems. Overall, the performance of rechargeable batteries primarily depends on the structural and chemical dynamics of electrode materials and electrolytes [1, 3, 18, 19, 20, 21], and although traditional electrochemical tests can be widely used to characterize battery performance, the understanding of internal reaction mechanisms remains difficult due to complex intrinsic systems with multi-components and interfaces. Therefore, the probing of microscopic mechanisms, reaction kinetics and structural evolutions of electrode materials and electrolytes during cycling is crucial to shed light on fundamental sciences and the rational design of high-performance rechargeable batteries.
Nowadays, conducting measurements under operating conditions is considered to be one of the most promising and efficient approaches. Of these measurements, ex situ imaging and spectroscopy techniques have been widely applied for the characterization of LIB compositional and structural evolution but suffer from unexpected reactions due to the removal of materials from native environments and thus cannot reveal the dynamic processes of cyclic charging and discharging [22, 23, 24, 25, 26, 27, 28, 29]. Alternatively, in situ techniques can allow for the direct monitoring of these dynamic processes during operation, providing the ability to directly link these processes to the electrochemical responses of corresponding batteries . Furthermore, transmission electron microscopy (TEM) is a powerful analytical tool to address many unique scientific problems in various disciplines such as material sciences and has been widely used in the analysis of rechargeable battery systems. In addition, the analysis of elemental quantities and distributions as well as phase identifications can be successfully realized by using equipped accessories such as energy-dispersive X-ray spectroscopy (EDX) and electron-energy-loss spectroscopy (EELS) . And by benefiting from the recent and tremendous progress in the development of aberration-corrected TEM [32, 33, 34, 35, 36, 37, 38, 39], the atomic-scale imaging of materials can now become routine practice, further expanding application in the characterization of rechargeable batteries.
The development of in situ TEM as summarized by Harks et al.  is a major breakthrough in the field of in situ techniques for battery research in which the fabrication of nanobatteries inside TEMs can allow for the atomic-scale observation of dynamic evolution in operating batteries and can completely mimic true environments in commercial battery systems [25, 40]. And in conjunction with advanced in situ TEM methods including open-cell and sealed liquid-cell configurations, dynamic processes during battery operation can be visualized in real time at unprecedented levels of spatial resolution. As a result, deeper understandings at the atomic- and nanoscale can be obtained for various mechanisms inside rechargeable batteries  including: (1) Li (Na, K, etc.)-ion insertion/extraction processes during continuous electrochemical cycles; (2) phase transition of active materials along with ion insertion and extraction; (3) structural evolution of electrode materials during cyclic charging/discharging; (4) formation processes of the solid electrolyte interface (SEI) layer between the electrode and the electrolyte; and (5) degradation mechanisms and mechanical behaviors of electrolytes during cycling. Based on all of this, this review will focus on recent significant achievements and advances of in situ TEM methods in the investigation of internal reaction mechanisms of electrodes, electrolytes and their interfaces in electrochemical rechargeable batteries. In addition, this review will also briefly discuss the development of in situ TEM techniques along with existing limitations to propose possible development directions. Most importantly, this review will highlight the sciences uncovered by in situ TEM and their role in providing guiding principles to tackle either the barriers of existing electrode materials or the design of new materials for better batteries.
2 Development of In Situ TEM for Batteries
In situ open-cell TEM techniques have enabled the investigation of structural and chemical evolutions during electrochemical cycling in a range of anode materials including Si [50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74], Ge [75, 76], Sn , Ga , SnO2 [79, 80, 81, 82, 83], Fe2O3 , Fe3O4 , MoO3 , Co3O4 , MnFe2O4 , RuO2 , CeO2 , TiO2 [91, 92], CoS2 , Co9S8 , MoS2 , CuO , Al2O3 [97, 98], ZnO [74, 99, 100, 101, 102], graphene [103, 104, 105, 106] and carbon nanotubes [94, 107], as well as cathode materials such as LiMn2O4 [108, 109, 110]. However, shortcomings of in situ open-cell TEM techniques include the lack of multiple contact points between the electrode and the electrolyte, the different charge-transfer resistances and the bypassing of some fundamental processes that occur in real systems such as SEI formation. To address these shortcomings, more attention has been paid in the design and development of a new sealed configuration (a liquid cell) for in situ TEM studies of rechargeable batteries by using real liquid electrolytes.
3 Application in Rechargeable Lithium-Ion Batteries
Due to the development of open-cell and liquid-cell configurations, many exciting and significant results have been achieved in the in situ TEM investigation of rechargeable batteries, especially LIBs, allowing for deeper understandings. These investigations include the electrochemical reaction mechanism of specific electrode materials, the structural evolution of electrodes during cycling, the interface between the electrode and the electrolyte, the formation of the SEI, the control of the nucleation and growth of dendrites, the capacity fading caused by repetitive cycling as well as other issues.
3.1 Charging/Discharging Reaction Mechanisms
Intercalation, alloying and conversion reactions are three typical mechanisms generally recognized to dominate Li+ storage in battery electrodes. These reactions direct the charge storage of electrode materials differently, leading to distinct electrode capacities, morphology and structures [3, 144]. To reveal the details of these charge/discharge mechanisms, extensive in situ TEM studies have been conducted.
3.1.1 Intercalation Mechanisms
Commercially available carbonaceous anodes are also typical electrode materials based on intercalation reactions. For example, Liu et al.  investigated individual graphene nanoribbons and found that lithiation/delithiation occurred mainly on the graphene surface. Wang et al.  in their study showed that the electrochemical dynamic process of layered MoS2 during lithium intercalation underwent a trigonal prismatic (2H)-octahedral (1T) phase transition with a Li ion occupying the interlayer S-S tetrahedron site in 1T-LiMoS2. Furthermore, Gao et al.  observed that the lithiation of amorphous TiO2 nanomaterials started with the valence reduction of Ti4+ to Ti3+ to form a LixTiO2 intercalation compound and that the continued intercalation of Li+ ions in TiO2 nanotubes triggered an amorphous to crystalline phase transformation, suggesting that the detailed lithiation process depended on the atomic structure and crystal feature of specific intercalating materials.
3.1.2 Alloying Process
Electrode materials based on alloying mechanisms are believed to be lithiated by forming direct bonds between inserted Li+ and the host element A (A: Si, Ge, Sn, etc.) with the formation of Li-A alloys. For example, Si is one of the most promising anode materials and possesses an ultrahigh theoretical capacity of 4200 mAh g−1 through the alloying reaction of 4.4 Li+ + 4.4 e− + S = Li4.4Si. Despite this ultrahigh theoretical capacity however, the detailed reaction process and phase transformation of this alloying reaction are highly complex and remain unclear. To resolve this, McDowell et al.  studied the lithiation kinetics of crystalline Si nanoparticles using in situ TEM and found that the mechanical stress generated during Li insertion can affect the nature of the Li-Si alloying reaction in addition to causing fractures. In another study, Liu et al.  studied the atomic behavior of electrochemical interfaces during the lithiation of crystalline Si and found that lithiation kinetics were controlled by the orientation-dependent migration of the interface and that the amorphous LixSi alloy was produced through the layer-by-layer peeling of the (111) atomic facet. Furthermore, by measuring the rate of growth of a surface layer of amorphous LixSi during the first lithiation using in situ TEM, these researchers also found that the self-limiting lithiation of crystalline Si attributed to the retardation effect of lithiation-induced stress . As for the formation of amorphous LixSi alloys in response to Li-ion insertion, Wang et al.  revealed that local electron-rich conditions governed the mechanisms of the electrochemically driven solid-state crystalline-amorphous transition in Li-Si systems in which the researchers observed that if x in amorphous LixSi reached a critical value of 3.75, amorphous Li3.75Si transforms to crystalline Li3.75Si. Furthermore, Wang et al.  found that the crystallization of Li15Si4 (i.e., x = 3.75) from amorphous LixSi is a spontaneous and congruent phase transition process without phase separation or large-scale atomic motion, which is drastically different from what is expected from a classic nucleation and growth process. Liu et al.  further showed that in contrast to the slow and incomplete lithiation of intrinsic Si nanowires with the formation of an amorphous LixSi alloy, C-coated and/or P-doped Si nanowires can exhibit charging rate 1–2 orders of magnitude faster, ultimately forming a crystalline Li15Si4 phase (Fig. 4c). Moreover, Lu et al.  observed in their in situ TEM study that the incorporation of Sn with Si nanowires can effectively improve delithiation kinetics in which surface diffusion can be enhanced by two orders of magnitude and bulk lithiation by one order of magnitude, resulting in a sequential surface-then-core lithiation mechanism.
Similar to Si, Liu et al.  and Liang et al.  used in situ TEM to reveal that initial crystalline Ge undergoes a two-step phase transformation process involving the formation of an intermediate amorphous LixGe and a final crystalline Li15Ge4 phase. Furthermore, the lithiation behaviors of Sn , Ga  and Se  with the formation of a crystalline alloy phase were also observed by using in situ TEM. In particular, Leenheer et al.  observed phase boundary propagation in Li-alloying battery electrodes using in situ liquid-cell TEM in which thin films of amorphous Si, crystalline Al and crystalline Au immersed in a commercial liquid electrolyte were lithiated and delithiated at controlled rates. Here, the results revealed that amorphous Si demonstrated laterally uniform behaviors, whereas for the crystalline films, a lithiation front spread laterally from initial nucleation points with continued grain nucleation along the growing interface, indicating that electrochemically induced solid-solid phase transformations during alloying reactions can lead to highly concentrated stress at the laterally propagating phase boundary. And overall, these observations demonstrate that electrochemical processes can trigger mechanical effects, which in turn will closely couple with electrochemical processes to affect the behavior of materials in batteries.
3.1.3 Conversion Reaction
Nanostructured transition metal oxides, sulfides and fluorides (MX, M = Fe, Co, Cu, etc., and X = O, S, F) have emerged as promising alternative anode candidates for LIBs with high reversible capacities through conversion reactions [1, 153] and typically demonstrate additional reversible capacities beyond theoretical capacities. However, these metal oxides also suffer from severe capacity fading and low Coulombic efficiencies during initial cycles. Here, the application of in situ TEM has provided new avenues to further understand external reaction mechanisms. For example, Luo et al.  used in situ TEM to observe a lithium-inserted Co3O4 phase consisting of nanosized Co-Li-O clusters as the intermediate product prior to the subsequent formation of Li2O crystals during the lithiation of Co3O4. Furthermore, the researchers also observed that during delithiation, the reduced metal nanoparticles formed a network and broke down into even smaller clusters that acted as catalysts to prompt the reduction of Li2O with CoO nanoparticles being the products of the deconversion reaction. And overall, the use of in situ TEM in this study allowed the researchers to conclude that the electrochemical decomposition of Li2O catalyzed by nanosized Co clusters is an important process that determines reversible capacity. In another example, Gregorczyk et al.  used in situ TEM to investigate an intermediate crystal phase (LixRuO2) during the first lithiation of RuO2 and reported that after the first cycle, the reaction becomes partially reversible between amorphous RuO2 if completely delithiated and a nanoscale network of Ru and Li2O if fully lithiated. And based on the in situ TEM investigations of Wang et al.  on CuO nanowire anodes, the severe capacity loss and low Coulombic efficiency during the first cycle were found to be possibly caused by irreversible volume inflation, phase change and the incomplete conversion of Cu to CuO. Furthermore, in situ TEM results from Su et al.  showed that upon lithiation, single-crystalline Fe2O3 nanoparticles can transform to multi-crystalline nanoparticles consisting of many Fe nanograins embedded into the Li2O matrix. In addition, the results in this study revealed that the delithiated product was FeO rather than Fe2O3, accounting for the irreversible electrochemical process and the large capacity fading in the first cycle. As for the subsequent charge-discharge process, this was observed to be a fully reversible electrochemical phase conversion between Fe and FeO nanograins accompanying the formation and disappearance of the Li2O layer. Similar conversion reactions of Fe3O4  and single-crystalline MnFe2O4  have also been observed using similar in situ TEM techniques. And by using a novel and simple in situ electrochemical TEM cell based on traditional copper grids, Wang et al.  tracked lithium transport and conversion in FeF2 nanoparticles based on nanoscale imaging, diffraction and spectroscopy. Here, the researchers found that lithium conversion initiates at the surface and subsequently propagates into the bulk with a morphological evolution resembling spinodal decomposition through a “layer-by-layer” process.
Different from single reaction mechanisms, lithiation processes of some binary metal oxides can proceed through two or three mechanisms of intercalation, alloying and conversion reactions. For example, Li et al.  used in situ TEM to characterize the two-step lithiation behavior of α-MoO3 in which Li intercalation in layered α-MoO3 leads to the formation of crystalline Li2MoO3 in the early stages of lithiation and further Li insertion converts LixMoO3 into metallic Mo and amorphous Li2O. Here, the researchers found that the intercalation process was thermodynamically more favorable and induced only minor volumetric changes whereas the conversion reaction was kinetically slow and induced large deformations. This two-step intercalation-conversion lithiation process has also been observed in CeO2 , CoS2 , Fe3O4  and FeF2 . In another example, He et al.  captured the atomistic conversion reaction process during Li insertion into a WO3 single-crystalline model electrode (Fig. 4d) using in situ TEM. Here, the researchers reported that according to their results, an intercalation step occurs prior to conversion. Following this, the inserted ion–oxygen bond formation destabilizes the transition metal framework, which gradually shrinks, distorts and finally collapses into an amorphous W and MxO composite structure. Furthermore, various studies have verified the formation of LiZn alloys during the lithiation of ZnO electrodes using in situ TEM, all of which confirm that the lithiation of ZnO involves a combination of conversion and alloying mechanisms [99, 100, 101].
In general, SnO2 is an ideal anode material for lithiation studies because of multiple lithiation mechanisms involving a combination of intercalation, conversion and alloying. Here, Li+ ions are initially intercalated into the SnO2 lattice to form a LixSnO2 intermediate phase following the reaction: SnO2 + xLi+ + xe− → LixSnO2. Subsequently, an irreversible conversion of SnO2 to metallic Sn occurs: 4Li+ + 4e− + SnO2 → 2Li2O + Sn and further lithiation is a reversible reaction between Sn and LixSn alloy: xLi+ + xe− + Sn → LixSn (0 \(\leqslant\)x\(\leqslant\) 4.4). The detailed reaction mechanisms of this overall process have been investigated in many studies by using in situ TEM [79, 80, 81, 83], and all of the results collectively indicate that irreversible conversion during the initial cycle is a common phenomenon in conversion-type electrode materials.
3.2 Volumetric Changes and Mechanical Behaviors of Electrodes During Cycling
The majority of electrode materials, especially Si anodes based on alloying reactions, suffer from severe volumetric change and structural degradation during charge/discharge, leading to detrimental pulverization, rapid capacity decay and limited application potential. Therefore, the elucidation of detailed structural evolution and mechanical degradation mechanisms is critical to the design of optimal electrode materials for next-generation rechargeable batteries.
In situ TEM with nanomechanical testing can also be used in the study of the mechanical behaviors of electrode materials, which is vital for practical application in LIBs. For example, Wang et al.  observed that partially lithiated Si nanowires demonstrated a striking contrast of brittle fractures in the unlithiated Si core as compared with the ductile tensile deformation in the lithiated Si shell, demonstrating the high damage tolerance of electrochemically lithiated silicon. In another study, Kushima et al.  conducted in situ tensile strength measurements of fully lithiated Si nanowires inside a TEM and found that the lithiated nanowire possessed a much smaller Young’s modulus and tensile strength as compared with pristine nanowires at room temperature, revealing large permanent plastic strains after fracturing, which was not seen in pristine Si nanowires.
Numerous efforts have been made to mitigate volume expansion, improve mechanical stability and enhance cycling performance of electrodes, all of which focus on two main strategies involving composition with highly conductive materials such as carbon and fabrication of well-designed nanostructures. Here, in situ TEM can be useful in the elucidation of the internal mechanisms and practical functions of these two strategies. For example, to determine the function of carbonaceous materials, Su et al.  compared two types of lithiation behaviors in CoS2 anodes using in situ TEM and reported that different from the side-to-side conversion process of pure CoS2 particles with large- and anisotropically sized expansion (47.1%), the CoS2 particles anchored onto reduced graphene oxide (rGO) sheets exhibited a core-shell conversion process involving smaller- and homogeneously sized expansion (28.6%) as well as fewer fractures due to the excellent Li+ conductivity of rGO sheets. In another study, Gu et al.  found that the lithiation behavior and response of Si/C composite systems to Si nanoparticle volume change depended on the spatial correlation of the silicon nanoparticle and the carbon materials (Fig. 5c). The researchers in a subsequent study also clarified using in situ TEM that the enhanced cycling stability of the conductive polymer Si composite was associated with the mesoscale concordant function of Si nanoparticles and the conductive polymer . Furthermore, Wang et al.  studied the microstructural evolution of an amorphous Si/C nanofiber composite during the charge/discharge process and found that the Si layer was strongly bonded to the carbon nanofiber and no spallation or cracking was observed during the early stages of cycling.
The significant role of graphene to the integrity of crystalline Si nanoparticles during lithiation/delithiation can also be observed by using in situ TEM [64, 70]. For example, Shan et al.  observed that graphene possesses three roles in a strongly coupled NiO/graphene hybrid during lithiation involving increasing the Li+ diffusion rate, improving Li+ reaction kinetics with NiO and severely restricting the expansion of NiO near the interface. Similarly, this was also observed by Xie et al.  on single PbSe/rGO nanosheet systems. Moreover, the interrelated effects of Cu coatings and Si crystallinity on Si nanowire volumetric changes during lithiation , the stabilizing mechanisms of ZnO coatings on Si anodes during cycling  and the functions of naturally oxidized Al2O3 surface layers on Al nanowire electrodes  have all been visualized by using in situ TEM as well.
The design of novel nanostructured materials is considered to be another feasible method to relieve issues of pulverization. For example, by using in situ TEM, Li et al.  observed that the volume change in Si walls during charge/discharge is primarily accommodated by the inner pores in well-designed mesoporous Si sponges, leading to only ~ 30% expansion in the full particle size (Fig. 5d). In situ TEM results from Liu et al.  showed that hierarchically structured Si pomegranates with 30–40 nm gap sizes in which single silicon nanoparticles are encapsulated by a conductive carbon layer can exhibit little change to both the carbon shell and secondary particle size during cycling. Furthermore, McDowell et al.  found that amorphous Si spheres with a novel two-phase lithiation mechanism can demonstrate superior energy storage properties than crystalline Si spheres due to larger critical fracture diameters.
3.3 SEI Formation Processes and Interfacial Reactions
SEI layers form at the surface of both cathode and anode electrodes during Li+ transport as a result of the interfacial reaction between the electrode and the electrolyte . In addition, the formation of a stable SEI layer has proven to be critical to the performance of LIBs because it can act as a passive layer to allow for the facile transport of ions, prevent electron transport between the electrolyte and the electrode and protect against anode corrosion and electrolyte decomposition [1, 163, 164]. Based on this, a deeper understanding of the fundamental formation mechanisms of SEIs and the effects on battery performance is necessary.
For all-solid-state LIBs without liquid electrolytes, the solid electrode-solid electrolyte interfacial impedance is currently regarded as the main limiting factor in performance; therefore, understanding the role of interfaces is critical to improvements in battery performance. Based on this, Santhanagopalan et al.  fabricated electrochemically active all-solid-state nanobatteries using focused ion beams and highlighted that the interfaces may significantly limit lithium transport in solid-state batteries in which their obtained in situ TEM results provided the first evidence of lithium accumulation at the anode/current collector and the cathode/electrolyte interface, which can account for irreversible capacity loss. In another study, Wang et al.  also reported the successful in situ STEM-EELS characterization of a cathode/electrolyte interface by galvanostatically biasing a solid-state nanobattery. Here, the researchers discovered a disordered interface layer derived from layered LiCoO2 inherent to the LiCoO2/LiPON interface, suggesting that chemical instability leading to the formation of an ionic resistive layer is the main mechanism of interfacial impedance.
3.4 Electrolyte Degradation and Dendrite Growth
Safety is a crucial consideration for the practical application of rechargeable batteries. However, unavoidable dendrite growth on electrodes occurs after repeated charge and discharge, which can result in short circuiting between two electrodes, leading to battery failure [139, 173]. In addition, electrolyte degradation as a side reaction can limit the battery lifespan and increase safety issues [174, 175]. To resolve these issues, much effort have been placed in the reduction in metal dendrite formation and in the unraveling of dendritic growth mechanisms through the use of in situ TEM, especially with the development of the liquid-cell platform.
Aside from Li dendrites, Sun et al.  investigated in real time the growth mechanisms of Pb dendrites deposited on electrodes under applied potential and found that the Pb concentration in the electrolyte can drastically influence the morphology of dendritic formation. In their TEM images, it was revealed that Pb dendrites developed through the fast protrusion of Pb branches in the electrolyte and tip splitting and that the fast growing tip of the dendritic branch was composed of polycrystalline nanograins which eventually developed into a single-crystalline branch. In another study, White et al.  found that Pb can be induced to deposit in a compact coating or as dendrites from an aqueous solution of lead(II) nitrate with the latter being more likely to form with abrupt potential changes (Fig. 7b). As for Mg-ion battery systems, Wu et al.  studied Mg cathodic electrochemical deposition on Ti and Au electrodes and observed a uniform Mg film deposited on the electrodes during charging, which was consistent with the intrinsic non-dendritic nature of Mg deposition in Mg-ion batteries. And based on the fact that electron beams can cause localized electrochemical reactions to allow for the observation of electrolyte breakdown in real-time, Abellan et al.  were able to reveal the stability of five different electrolytes using in situ liquid-cell STEM. Here, the results suggest that electrolyte degradation, dendrite growth, SEI formation and interfacial reactions are closely correlated, even playing a synergetic role in the performance of batteries.
4 Application in Other Electrochemical Rechargeable Batteries
Aside from LIBs, in situ TEM techniques have also been widely employed in the probing of other rechargeable battery systems to reveal reaction mechanisms, interfacial reaction processes, structural evolutions and mechanical behaviors.
4.1 Lithium-Sulfur Batteries
To overcome the charge-storage limitations of traditional oxide cathodes in LIBs, lithium-sulfur batteries (LSBs) have been widely explored. This is because aside from being one of the most abundant elements on earth, the capacity of sulfur, originating from the conversion of sulfur to form lithium polysulfides, is an order of magnitude higher than those of oxide cathodes, with a theoretical capacity of 1672 mAh g−1 . However, volume change during the electrochemical cycling of LSBs and the dissolution of polysulfides in electrolytes remain acute challenges for practical LSBs.
Although the TEM is a powerful tool for the observation of microstructures, the imaging of S with TEM is challenging because S possesses a relatively low vapor pressure and is sensitive to high-energy electron beams . Therefore, to conduct in situ TEM investigations of S, researchers have developed several effective methods and obtained remarkable results. For example, by confining sulfur within 200-nm cylindrical inner pores of carbon nanotubes, Kim et al.  realized the in situ TEM observation of the dynamical process of lithium insertion into sulfur cathodes and demonstrated the presence of electron pathways at the Li2S/S interface. Alternatively, S particles can be hermetically encapsulated in 2D materials with unique properties, which not only effectively improves cycling performances but also enhances TEM observations. For example, Tang et al.  demonstrated that the high flexibility and strong van der Waals force in MoS2 nanoflakes can allow for the effective encapsulation of S particles and prevent their sublimation. In another example, Xu et al.  found that the lithiation product Li2S can be contained within the electrode with only ~ 35% volume expansion and that the carbon host can remain intact without fracture in porous carbon nanofiber/S cathodes. Similarly, Zhou et al.  provided visible evidence that porous carbons can effectively confine S without the use of any other coatings. More specifically, Yang et al.  observed that the S/Li2S phase separation phenomenon in their designed solid-state Li-S nanobattery reduced diffusion distances and provided S/Li2S interface networks. The researchers also observed the direct transformation of S into Li2S without detectable intermediate products of lithium polysulfide, which is different from liquid LSBs with a two-step electrochemical lithiation reaction at the S cathode.
4.2 Sodium-Ion Batteries
The application of LIBs in renewable energy storage systems is restricted by the limited distribution of lithium resources. Recently, however, sodium-ion batteries (NIBs) have appeared as one of the more promising state-of-the-art battery systems because of the abundant nature of sodium [23000 ppm (1 ppm = 1 μmol/mol) as compared with 20 ppm for lithium] [9, 10]. The inherent characteristics of sodium are different from lithium and cause different reaction mechanisms and battery performances . In particular, the larger ionic radius of Na+ results in more substantial electrode volumetric changes during reactions as compared with LIBs and these changes must be understood and controlled. Therefore, the clarification of these dynamic reaction processes and structural evolution mechanisms is essential for the practical engineering of NIBs.
Based on this, in situ TEM investigations have been conducted to reveal the reaction mechanisms of electrodes during sodiation/desodiation. For example, results from Yuan et al.  revealed that the first sodiation process of α-MnO2 starts with tunnel-based Na+ intercalation followed by the formation of Na0.5MnO2 as a result of tunnel degradation and ends with the Mn2O3 phase. Here, the inserted Na+ can be partially extracted out of the sodiated products and the subsequent cycles were found to be dominated by a reversible conversion reaction between Na0.5MnO2 and Mn2O3. In another example, Gao et al.  observed that Na intercalation into MoS2 nanostructures followed a two-phase reaction mechanism from trigonal prismatic 2H-MoS2 to octahedral 1T-NaMoS2 with a phase boundary of ~ 2 nm thick in which the velocity of the phase boundary was one order smaller than that of Li+ diffusion, suggesting sluggish kinetics for Na+ intercalation. This was also probed by Zhang et al.  and Li et al. . Moreover, Zhang et al.  observed that the final sodiation product of a CuO nanowire was nanocrystalline Cu mixed with Na2O and Cu2O in which nanocrystalline Cu first oxidized into Cu2O and transformed back to nanocrystalline CuO upon extraction of Na+. And by comparing the electrochemical reactions of Cu2S with Na and Li, Boebinger et al.  were able to observe different transformation pathways and concluded that the larger atomic radius of Na as compared with Li can cause substantial volume change during reactions, which do not necessarily cause accelerated capacity decay but can cause the reaction to proceed through a different pathway. Similarly, the sodiation mechanisms of WO3 , SnS2 , SnO2 , FeF2 , van der Waals stacked Sb2S3 , C- and Au-coated CuO , etc. also have been revealed by using in situ TEM. In particular, Cui et al.  investigated the pseudocapacitive Na storage mechanism of SnS2 using in situ TEM and unveiled that the adsorption of Na+ ions onto the Sn edge and the surface of SnS2 nanoplatelets together with the ultrafast intercalation of Na+ ions into the SnS2 layers were the origins of the pseudocapacitance.
Apart from layered transition metal dichalcogenides and van der Waals stacked crystals, Lu et al.  also found that the sodiation of amorphous germanium (a-Ge) nanowires formed Na1.6Ge instead of NaGe and that the nanowires retained their structural integrity over several cycles, suggesting that the potential of a-Ge for NIB applications may have been previously underestimated. Furthermore, the intercalation process of solvated Na+ into graphite was studied by Goktas et al.  who found that the electrolyte decomposition products were soluble and/or volatile and no solid SEI was formed. Alternatively, reversible metal cluster formations on rGO anodes , electrochemical Na plating/stripping dynamics through the use of amorphous carbon nanofibers as a current collector  and Na+ storage mechanisms of soft carbon anodes  were all presented. And similar to the function of carbonaceous materials in composite electrodes of LIBs, Pan et al.  confirmed using in situ TEM that the high reversible capacity of MoS2/C microtubes was mainly the result of the formation of a MoS2/C heterointerface in the MoS2/C nanosheets, which can effectively stabilize MoS2 and the reaction product Mo and prevent the coarsening of Mo nanoparticles. In another example, Li et al.  demonstrated the synergetic effects of two nanoscale components with Co3O4 nanoparticles decorated on CNTs, providing insights into a new sodiation mechanism facilitated by Na diffusion along a CNT backbone and CNT-Co3O4 interfaces. Moreover, two types of sodiation behaviors involving notable axial elongation for open CNTs and major radial expansion for closed CNTs were observed in Co9S8-filled CNTs by Su et al.  using in situ TEM.
4.3 Potassium-Ion Batteries
Similar to NIBs, potassium-ion batteries (KIBs) are another promising alternative to LIBs in large-scale renewable energy storage systems due to large natural abundance and low costs . Therefore, it is also necessary to gain insight into the structural evolution of KIB electrodes to design more robust electrodes. For example, Liu et al.  conducted an in situ TEM study of the electrochemical potassiation of carbon nanofibers and revealed that mechanical degradation during potassiation occurred through the formation of longitudinal cracks near the c-C/d-C interface. In another study, Wu et al.  found that during the entire potassiation process, the electrode of red P on nitrogen-doped porous hollow carbon nanofibers demonstrated stable structural change, revealing the robust structure integrity of the electrode (Fig. 8c). In addition, Zhang et al.  reported the in situ TEM observation of the charge/discharge reaction process of a K-CO2 battery by using CO2 as the cathode in which during discharge, K2CO3 was generated in the cathode, which converted back to metal K during charge with good cyclability. Here, it is evident that CO gas was released during discharge and the carbon cathode was consumed continuously during charge.
4.4 Metal-Air/O2 Batteries
Metal-air/O2 batteries are the combination of a high-energy density metal anode (Li, Na, K, etc.) and an open-structured air electrode that can draw cathode active materials (i.e., O2) from air or directly through the use of pure O2. These batteries represent a class of promising power sources for application in next-generation electronics, electrified transportation devices and energy storage systems of smart grids . Here, in situ TEM observations can provide in-depth understandings of metal-air/O2 energy storage systems.
For example, Zhong et al.  reported the first in situ TEM observations of electrochemical oxidation for Li2O2 supported on multiwall carbon nanotubes (MWCNTs) and revealed that the reaction preferentially took place at the Li2O2/MWCNT interface, indicating that electronic conductivity is the limiting factor during Li-O2 battery charging. In another study, Luo et al.  imaged the product morphology evolution on the CNT cathode of a solid-state Li-O2 nanobattery and found that the oxygen reduction reaction (ORR) on CNTs initially produced LiO2, which subsequently disproportionated into Li2O2 and O2. In addition, they found that the release of O2 can create a hollow nanostructure with a Li2O outer-shell and a Li2O2 inner-shell surface. Furthermore, the linear nature of the growth kinetics of metal nanoparticle catalysts for ORR in metal-air batteries was studied by in situ catalyzed oxidation experiments in TEM  and Li2O2 was confirmed to be the main product of the discharge reaction in Li-air batteries based on hierarchical mesoporous perovskite La0.5Sr0.5CoO2.91 nanowire electrocatalysts . And based on the crucial role of liquid electrolytes for controlling battery performance and cyclability in real batteries, in situ liquid-cell TEM observations of the charge/discharge reactions of non-aqueous Li-O2 battery cathodes were performed by Kushima et al.  in which it was found that Li+-ion diffusivity/electronic conductivity was the limiting factor in discharging/charging (Fig. 8d), allowing the researchers to conclude that enhancements in the binding force between reaction products and current collectors to maintain robust electronic conduction are key for improved battery performances. Subsequently, Yang et al.  reported observations of cathodic reactions in the presence of redox mediator tetrathiafulvalene (TTF) in an electrolyte, providing direct evidence that TTF plays a role in promoting the decomposition of Li2O2 as a soluble charge-transfer agent between the electrode and Li2O2. Furthermore, to explain the origins of sluggish ORR and OER kinetics in Li-O2 battery systems, He et al.  observed the nucleation of Li2O2 at the carbon electrode/electrolyte interface during discharge and found that the growth process exhibited Li+ diffusion-limited kinetics whereas the growth of Li2O2 isolated in the electrolyte was O2− diffusion limited. Overall, with the help of in situ TEM techniques, the reaction mechanisms inside metal-air/O2 batteries are becoming increasingly clear.
5.1 Beam Effect
The electron beam used in TEM can cause temporary or permanent change in the surface or bulk structure of specimens and can trigger side reactions and limit the spatial resolution of electron beam imaging. This property is the main limitation to the application of TEM, and especially in aberration-corrected TEM with a field-emission source because of the obtainable high current densities [218, 219]. Therefore, the effects of the imaging electron beam on electrochemical processes need to be taken into consideration during in situ TEM imaging procedures.
For in situ TEM investigations of rechargeable batteries by using the open-cell platform, electron beams have been reported to accelerate or retard the lithiation process by decomposing Li2O to release Li and lithiate nearby electrodes depending on the material and electron dose . Furthermore, Liu et al.  observed that the electron irradiation of SnO2 nanowires and Li2O junctions on a TEM grid can lead to the lithiation of SnO2 and that the electron beam irradiation of Li2O can lead to the decomposition of Li2O into elemental Li and volatile gases, which can be alleviated by reducing the electron dosage below ~ 1 A cm−2, thus suppressing chemical lithiation. In another study, by taking into account the sensitivity of discharge products to electron beams in metal-air batteries, Basak et al.  reported experimental optimizations in the study of Li-O2 batteries in TEM through the use of in situ EELS to evaluate experimental procedures and the use of low-dose-rate STEM and a “graphene cell,” which can encapsulate samples within the graphene sheets and protect the sample against electron beam damage. Overall, material behaviors observed by in situ TEM using the open-cell platform reflect real reaction kinetics, structural changes and chemical evolution and have been verified by blocking electron beams if not recorded and compared with ex situ results .
As for the liquid-cell platform, radiolysis and electrolyte degradation induced by electron beams are the biggest experimental limitations for in situ TEM imaging because of alterations to the chemistry of the suspending liquid environment and the interfaces [114, 134, 222, 223, 224, 225, 226, 227, 228, 229, 230]. In particular for rechargeable batteries, conventional battery electrolytes and formed SEIs are sensitive to high-energy electron beams . For example, Holtz et al.  demonstrated the negligible effect of irradiation on electrochemical measurements by comparing the cyclic voltammetry curves between beam on and beam off. And in studies with water-based electrolytes, high-energy electron interaction with water molecules can lead to water radiolysis and the formation of bubbles and many different species, all of which can alter measurements by changing the pH and oxidation/reduction environments and lowering the electrochemical potential through absorption in the specimens [223, 229]. As for commercial organic battery electrolytes, aside from radiolysis, electron beams can also generate heat, which can result in thermal decomposition [231, 232]. According to results from Abellan et al. , in the presence of a direct electron beam, commonly used organic electrolytes (LiPF6 in EC/DMC) can generate gaseous products and LiPF6 can break down to LiF. And the subsequently breakdown of LiF under irradiated conditions can result in various Li microstructures due to high-energy electron-generated crystal defects and the ionization of atoms [233, 234]. This microstructural growth may further cause errors in the TEM analysis of electrochemical systems using battery electrolyte salts, and beam-induced surface film formations can alter Li microstructures . In addition, bubble formation along with electrolyte degradation can lead to leakage and other safety issues inside TEMs; therefore, flow-liquid-cell designs appear to be beneficial . For example, Schneider et al.  proposed a model to compute the concentration of radiolysis products as a function of electron beam irradiation parameters such as time, space and solution composition, which can be useful for the design of experiments that minimize unwanted beam effects. In practice, the reduction in electron beam doses can directly circumvent beam-effect limitations but require countermeasures to compensate for the loss in image quality. For example, the introduction of scavenger species that reacts with solvated electrons can be an efficient approach in some situations . Alternatively, new low-dose acquisition methods for (S)TEM in combination with computer-based image analyses, faster readouts and the use of direct electron detection cameras can allow for drastic decreases in electron beam doses needed for image formation and provide better control over radiolysis .
5.2 Distances from Real Battery
In the open-cell platform for in situ TEM, the electrode possesses only one contact point with the electrolyte instead of full immersion in real batteries. In addition, the large overpotential used to drive Li ions into the electrode may alter electrode kinetics and phase behaviors . Furthermore, the application of ionic liquids or solid electrolyte alternatives ignores critical processes that occur in real systems and can lead to different charge-transfer resistances. And although recent liquid-cell designs can largely mimic real conditions with the use of real electrolytes, there is still distance from real batteries, especially in electrochemistry system designs, which may become limitations in the in situ TEM investigation of rechargeable batteries. In addition, existing electrochemical liquid-cell systems usually fabricate only two or three electrodes inside using linear or concentric configurations without considering the electrochemical relationship between electrodes [111, 126, 128]. However, the configuration of electrodes has great influence on the interface features between the working electrode and the electrolyte. In addition, the electrochemical potential condition of working electrodes is controlled by the ability of the reference electrode to hold a constant potential and the counter electrode to supply current . The uniformity of the electric field distribution along the working electrode is influenced by the position of the counter electrode. Here, non-uniformity of the electric field gradient can cause localized hot spots and affect the deposition rate, with uncontrolled electrochemical potentials of the working electrode leading to unreliable electrochemical test results . Furthermore, electrode materials in liquid-cell platforms are still limited to Au, Pt, Ti and, currently, graphite, which are different from those used in real batteries, and Au electrodes have been reported to possess electrochemical reactions with commercial electrolytes (LiPF6/EC/DEC) for LIBs . And although some real electrode materials have been fabricated on electrodes by using assistive techniques such as FIB and electrodeposition [156, 165], the direct observation and the characterization of real electrode materials under electrochemical biasing are still limited due to the complicated processes.
5.3 Resolution of the Liquid-Cell Platform
Limited by the thickness of the SiNx viewing window and the liquid layer in which multiple scattering from the liquid can dominate signals as well as the controlled dose rate for the beam effect, regular liquid cells still suffer from poor spatial resolution and contrast. Overall, it is difficult to acquire atomic resolution images and detailed structural features. And although the development of graphene liquid cells has resolved the resolution limitations to some extent , equipping with electrochemical biasing inside graphene liquid cells has emerged as another problem, which appears to be necessary for investigations of rechargeable batteries.
6 Conclusion and Perspectives
Benefiting from the developments of in situ TEM nanobatteries based on well-designed open-cell and liquid-cell platforms, significant findings into the internal scientific mechanisms of electrochemical rechargeable batteries have been realized. In addition, valuable knowledge including reaction details for intercalation, alloying and conversion reactions during repeated cycling, phase transformation and structural evolution principles of electrode materials, dynamic formation of SEI layers and their characteristics, growth processes of dendrites and real electrode-electrolyte interfaces has been visualized. Because of all of this, a deeper understanding into the internal reaction mechanisms of electrochemical rechargeable batteries has been reached.
Over the years, fundamental sciences with respect to rechargeable batteries have been revealed by using in situ TEM, which have not only pinpointed the critical challenges and barriers of electrode materials but also led to valuable inspirations to fabricate novel, optimal electrode materials with enhanced properties. Typical cathodes of rechargeable batteries are mostly intercalating materials with layer structures and according to in situ TEM results, the atomic interlayer structures and specific phase transition behaviors play crucial roles in energy storage performances. Therefore, instead of pursuing new materials, the design of charge/ion transport avenues for current materials in the atomic scale by combining materials of different characteristics and appropriate atomic doping to adjust electronic structures, as well as other more direct approaches may be viable methods to break the charge-storage limitations of traditional cathodes. As for LIBs, Si anodes are one of the most promising choices due to a high theoretical capacity from alloying reaction mechanisms. And in terms of anisotropic swelling, strong size dependence of fractures as well as C-coating and/or P-doping enhancements on high-rate complete lithiation, and size-controlled Si electrodes with C-coating and P-doping are potentially ideal anode materials. Apart from traditional strategies involving fabricating composites based on synergistic effects, nanoscaled structural designs and rational morphology, more attention should be paid to crystal characteristics and detailed size effects in specific materials in the design of electrodes. In addition, due to similar atomic structures to Si, other elements such as Ge and Sn in the Si group of periodic table elements deserve more attention in rechargeable battery systems. And based on complex conversion reaction mechanisms, transitional metal oxides, sulfides and fluorides (MX, M = Fe, Co, Cu, etc., and X = O, S, F) with high reversible capacities and low costs have been widely investigated as alternative anode candidates for rechargeable batteries, including LIBs, NIBs, KIBs, etc. Overall, the biggest challenges for the potential application of these electrode materials are multifaceted, including severe capacity loss and low Coulombic efficiency during initial cycles, which have been shown to be a result of the irreversible phase change and the incomplete conversion process. Therefore, the development of efficient reaction inducers or catalysts to facilitate these intrinsic conversions can be a unique and valid approach to further improve application performances.
In broad terms, electrochemical processes can trigger both thermal and mechanical effects, which in turn closely couple with electrochemical processes to critically control the performance of electrochemical systems. Here, electrode volume change, mechanical deformation, dendrite growth, electrolyte degradation and SEI formation are all typical examples during the charge-discharge process in almost all current rechargeable batteries. And although each example appears to be an independent process, essentially, they are closely correlated, either beneficially or detrimentally, in a nonlinear fashion. For their study, in situ TEM results can not only allow for the direct visualization of the dynamics of these reactions under operando processes, but also simultaneously allow for the establishment of correlations among these phenomena. Therefore, instead of adjusting the individual phenomenon through rational design and taking into account all reactions and phenomena, the control of these processes in coordination with each other is a viable approach to build better rechargeable batteries. This means more attention should be paid to achieving good matching between electrode materials and electrolytes, designing rational materials with more consideration on the actual service environment and finding a balance between high performance and high reliability.
In summary, open-cell platforms possess great advantages in the probing of intrinsic responses of material to metal-ion insertion or extraction of high-spatial-resolution imaging, but show limitations in the direct correlation of electrochemical characteristics. In addition, the development of closed liquid cells opens new options for in situ TEM studies on structural and chemical evolutions of electrochemical processes that are closer to real working systems. However, their application in batteries remains limited due to electron beam effects, low image resolution and rough electrode configurations as well as difficulties in compositional and structural characterizations. Despite this, continuing improvements in equipment and experimental capabilities will enable more in-depth observations of fundamental sciences, allowing for great progress on the rational design of high-performance rechargeable batteries. Based on this, several possible future research directions are proposed as follows: (1) further enhancing image resolution without sacrificing the electrochemical biasing setup; (2) fabricating more rational electrode configurations inside liquid cells for reliable electrochemistry; (3) combining with more standard electrochemical techniques to further reveal the electrochemical mechanisms in rechargeable batteries; (4) developing facile fabrication methods or improved liquid-cell platforms to investigate various electrode materials; (5) performing studies on more complex battery systems and exploring new battery systems; (6) further expanding the application of all-solid-state cells; and (7) developing more stable and flexible in situ TEM platforms with controlled liquid and gas conditions, accompanying with heating/cooling choices.
This work was supported by the Assistant Secretary for Energy Efficiency and Renewable Energy, Office of Vehicle Technologies of the U.S. Department of Energy under Contract No. DE-AC02-05CH11231, Subcontract No. 18769 and No. 6951379 under the Advanced Battery Materials Research (BMR) program. This work was supported by U.S. DOE (Department of Energy) EERE (Energy Efficiency and Renewable Energy) BETO (Bioenergy Technology Office) (grant No. DE-EE0008250) to BY with the Bioproducts, Science and Engineering Laboratory, Department of Biological Systems Engineering at Washington State University. This work was conducted at the William R. Wiley Environmental Molecular Sciences Laboratory (EMSL), a national scientific user facility sponsored by DOE Office of Biological and Environmental Research and located at PNNL. PNNL is operated by Battelle for the Department of Energy under Contract DE-AC05-76RLO1830. Ms. X. Wu was supported by China Scholarship Council for Overseas Studies.
- 27.Gabrisch, H., Yazami, R., Fultz, B.: A transmission electron microscopy study of cycled LiCoO2. J. Power Sources 119, 675–679 (2003)Google Scholar
- 215.Shi, C., Feng, J., Huang, L., et al.: Reaction mechanism characterization of La0.5Sr0.5CoO2.91 electrocatalyst for rechargeable Li-air battery. Int. J. Electrochem. Sci. 8, 8924–8930 (2013)Google Scholar