Introduction

Electrochemical energy storage (EES) via batteries and electrochemical capacitors is increasingly ubiquitous in today’s digital world. EES devices are powering the portable electronics revolution, from cell phones to tablets and smart watches. The high efficiency of electrical-to-chemical energy conversion coupled with advances in the energy density, power density, and cost of these devices are also increasing their use in the transportation sector and sustainable grids. Continued research and development of EES will shape the future of these applications and their impact in our daily lives. For example, further advances in high-power EES systems will improve the harvesting and storage of intermittent renewable energy sources such as wind and solar. Introducing large-scale EES into the power grid will decrease the demand for fossil fuels and hopefully reduce the environmental impact from existing electricity generation methods [1, 2]. EES could also enable a revolution in electric powered transportation, reducing the amount of global petroleum combustion [3,4,5]. For EES to reach its potentially tremendous impact on sustainable technology and human well-being, there are further scientific hurdles to overcome, including achieving simultaneous high power and energy density, decreasing cost, increasing safety, and improving lifetime [6,7,8].

Most EES devices today store energy in solid-state electrodes [9,10,11]. Since the electrochemical redox reaction occurs on the solid material’s interface with the electrolyte, and because (in the case of batteries) the reacted species must also undergo solid-state diffusion into the bulk of the particle, there are several fundamental requirements for the electrodes. First, since electrochemical reactions require both electron and ion motion to and from the electrochemical interface, the electrodes must be both electronically and ionically conductive. This requirement can be difficult to satisfy simultaneously in many materials, and as a result, the electrodes contain micro- or nanoscale particles to decrease diffusion distances and increase porosity as well as conductive additives to increase electronic conductivity. Another basic requirement for solid-state EES electrodes is that they must store high amounts of energy by volume and area. To do this, it is important to utilize high gravimetric capacity active materials, but the final volumetric and areal energy densities are determined by the electrode architecture. Ideally, the electrodes are thick (hundreds of microns or more in thickness) as well as dense to provide high areal and volumetric energy densities. This is also difficult to satisfy simultaneously, and most electrodes are either thin and dense or thick and highly porous.

To meet these requirements, traditional solid-state EES electrodes are fabricated via the casting of a slurry that consists of micron to hundred nanometer-sized particles of the electrochemically active material, a high surface area conductive additive, and a polymer binder [12]. Due to the number of components, there are many factors that affect the final electrode architecture, including the active material size, conductive additive size, and binder concentration, to name a few. This approach has been optimized for micron-size, roughly spherical particles often utilized in lithium-ion batteries. However, scientists still lack a thorough understanding of how each processing parameter or condition individually affects the final electrode architecture [13], which leads to difficulties when applying this methodology to non-spherical and/or nanosized particles. Due to the many variables involved in slurry casting, there is always the possibility that the electrode architecture is not consistent from batch to batch, which can result in poor energy storage performance. Furthermore, the electrode architecture fabricated from slurries does not simultaneously promote high energy density and high power density because of sub-optimal electron and ion transport pathways. To illustrate this point, scanning electrochemical cell microscopy of single particles of LiMn2O4 showed facile de/intercalation of Li+ at rates as high as 1 V s−1—but when the particles were assembled into slurry electrodes, these rapid kinetics disappeared due to poor ion and electron transport in the electrode architecture [14]. It is hypothesized that electrode architectures that provide more control over the transport of ions and electrons—so-called deterministic architectures—would be better for providing both high power and high energy densities [15].

Recently, the United States Department of Energy Office of Basic Energy Sciences reported the need for “tunable functionality of materials and chemistries that enable holistic design for energy storage” as well as the need to achieve simultaneous high energy and high power [16]. Since design of electrodes and their energy and power performance are intimately related, these will require detailed understanding of the kinetics and mechanisms of electrochemical reactions occurring at solid-state electrode interfaces, as well as predictive control over the electrode architecture.

One promising approach in the pursuit of next-generation energy storage with simultaneous high energy and high power is through cooperative assembly of electrochemically active materials into conductive scaffolds [15, 17, 18]. Figure 1 conceptualizes the hierarchical structures that can be defined by the conductive scaffold or the ordered assembly of conductive materials. It is important to note that in such architectures, the active material is often directly bonded to the conductive material, therefore reducing the need for separate binders and current collectors. Advances in synthesis techniques, such as hydrothermal synthesis [19], electrodeposition [20], and atomic layer deposition [21, 22], have enabled conformal deposition of nanostructured active materials onto a variety of conductive scaffolds. The conductive scaffold material can also provide a robust, free-standing structure that is capable of enduring mechanical deformation, which is particularly important for high gravimetric capacity materials that can undergo significant volume changes during electrochemical cycling. If the conductive scaffold has a high surface area, this could enable high active material loading and therefore high areal capacity. Achieving high areal capacity is necessary to match the areal capacity of high energy density anodes such as Li metal and silicon [23]. This review summarizes several of the most common approaches for developing free-standing binder-free electrode architectures of transition metal oxides (TMO) that aim to achieve simultaneous high energy and high power for the next generation of electrochemical energy storage. Several other reports have summarized the importance of electrode architecture on EES [17, 24,25,26,27,28].

Figure 1
figure 1

The morphology of the commercially utilized slurry electrode for EES can result in insufficient electron and ion transport to the active material. Emerging deterministic electrode architectures, such as arrays, assemblies, and metal and carbon foams, provide well-defined electron transport paths and porosity for improved utilization of the active material, as shown schematically in the figure

Full size image

Transition metal oxides for electrochemical energy storage

TMOs have several advantages as active materials for EES. In general, they can deliver a high theoretical capacity due to the number of accessible valence states of transition metal ions, which is directly related to the number of electrons that can be stored per formula unit. Compared to transition metal sulfides, TMOs have higher ionicity which leads to higher redox potentials [29]. High redox potentials are needed for cathodes in high energy density batteries. TMOs can undergo faradaic EES via several different mechanisms: (1) pseudocapacitance, whereby the redox occurs at the surface or is not limited by solid-state diffusion [30]; (2) intercalation of ions into the bulk while the host structure retains its overall morphology [31]; and (3) decomposition and conversion, whereby the oxide undergoes reversible reduction into a suboxide or the transition metal itself [32]. The general trend is that the capacity, and energy density, is highest for TMOs undergoing conversion or decomposition and lowest for those undergoing only pseudocapacitive charge storage. The power density, energy efficiency, and lifetime tend to trend in the opposite way, with the fastest, most efficient, and durable storage obtained with pseudocapacitive materials. It should be noted that the conversion reaction mechanism is plagued by inherent voltage hysteresis even at slow rates, which results in low energy efficiency [32].

TMOs can undergo EES reactions with a variety of cations, including protons [33], Li+ [34], Na+ [35], and K+ [36], in aqueous, non-aqueous, ionic liquid, and solid-state electrolytes. Furthermore, TMOs exhibit a variety of crystalline structures, such as layered, olivine, or spinel phases, that enable fast ion diffusion pathways [30]. These structures can be tuned to improve the EES properties, by, for example, inducing oxygen vacancies [37] or introducing species such as water molecules in the interlayer [38]. The TMO surfaces can also be modified to prevent undesirable reactions with the electrolyte by depositing a more stable TMO or carbon coating [39, 40]. TMOs can be synthesized via several scalable approaches without the need for special atmospheres, making them attractive for industrial applications [19]. There are also several methods for depositing nanostructured TMOs onto a conductive scaffold, such as chemical bath deposition [41,42,43,44], hydrothermal reaction [36, 45], atomic layer deposition [46,47,48], and electrodeposition [35, 49, 50]. In many cases, such synthesis results in good adhesion between the TMO and the scaffold, which can lead to low charge transfer resistance and long cycling stability—negating the need for electrochemically inactive conductive additives and binders. Furthermore, these nanostructured TMO deposits benefit from improved tolerance to pulverization during charging/discharging [51, 52].

Metrics to evaluate performance

Evaluating and comparing the performance of advanced energy storage electrode architectures can only be possible when sufficient metrics are provided to do the analysis. However, this can be complicated as the reported metrics vary depending on the mechanism of energy storage (for example, pseudocapacitance vs. conversion reactions) and the intended application. In general, however, it is best practice to report the electrode mass loading (both active material and total mass including binder, conductive additive and/or scaffold), the footprint area, and the overall thickness. This way, gravimetric, areal, and volumetric capacities and energy densities can be readily calculated for electrodes even if they are not directly discussed in the manuscript. Moreover, future data mining of the energy storage literature (in order to comprehensively benchmark new electrode architectures) would significantly benefit from the proper reporting of these parameters.

Hierarchical arrays and assemblies for free-standing electrode architectures

Free-standing TMO-based electrode architectures have been developed through (1) the formation of nanoarrays on flat substrates and (2) assembly of materials into a hierarchical structure.

Arrays on a flat substrate

The general concept for this electrode architecture is to grow or deposit micron or nanoscale ordered arrays onto a flat substrate, thereby increasing the total surface area and decreasing diffusion distances as well as the contact resistance due to the improved adhesion of the active material to the current collector. However, the areal mass loading that can be reached without sacrificing EES performance is limited to less than 1 mg cm−2, which is directly related to the limited thickness of the array (typically less than 10 microns). This is due to the low electronic conductivity of most TMOs in their fully oxidized state [53]. Nanoarrays of TMOs can be obtained by multiple approaches. One is to directly synthesize TMO nanostructures on the surface of a conductive substrate [54,55,56,57]. Wu et al. [45] used a two-step hydrothermal synthesis to first grow Co3O4 nanowire arrays on Ti foil, and subsequently grow Fe2O3 branches, as shown in Fig. 2a. Electrochemical analysis of the conversion reactions of Co3O4 and Fe2O3 with Li+ was carried out in a non-aqueous electrolyte. The high-frequency electrochemical impedance spectroscopy revealed a lower charge transfer resistance for the 5-μm-thick Co3O4@Fe2O3 array electrode as compared to electrodes with only Co3O4 or Fe2O3 arrays on the Ti substrate. Using a similar approach, Xia et al. [42] hydrothermally grew Co3O4 nanowire cores on a conductive glass substrate followed by chemical bath deposition of NiO shells. The Co3O4/NiO arrays, with a thickness of ~ 10 μm, were electrochemically characterized in an alkaline aqueous electrolyte. It was hypothesized that the rate capability and cycling stability of the Co3O4/NiO electrode arrays were improved as compared to only the Co3O4 or the NiO arrays due to sufficient electronic conductivity of the Co3O4 nanowires and additional capacity and structural support provided by the NiO shells. Nanostructured TMOs deposited on a flat substrate can obtain good rate capability and cycling stability. However, the limited thickness of the electrochemically active material limits the achievable areal capacity.

Figure 2
figure 2

a Scanning electron micrographs of the cross-section (left) and top-down view (right) and transmission electron micrograph (inset) of Fe2O3 branches grown on Co3O4 nanowires (adapted with permission from Ref. [45], Copyright 2013 Springer Nature), b low-magnification cross-section (left) and a high-magnification (right) scanning electron micrographs of nickel nanowires with MnO2 electrodeposited on the surface (adapted with permission from Ref. [49]. Copyright 2016 John Wiley and Sons), and c scanning electron micrographs of nickel nanotubes (left) and nickel nanotubes electrodeposited with Fe3O4 and MnO2 which are uniformly distributed throughout the array electrode (adapted with permission from Ref. [50]. Copyright 2017 John Wiley and Sons)

Full size image

Metal and carbon supports can also be grown from the surface of flat substrates to produce reliable electronic conduction pathways to/from the current collector [58, 59]. For example, Xu et al. [49] grew a Ni nanowire array via a magnetic-field-driven selective deposition growth process on Ti foil, which was then used as a scaffold for the electrodeposition of MnO2, as shown in Fig. 2b. The length of the Ni nanowires was approximately 1 mm, which provided electron transport paths through thicker electrodes compared to the electrodes that were only composed of TMOs. To determine the pseudocapacitive behavior of this electrode, cyclic voltammetry was performed in a Na2SO4 aqueous electrolyte between 0 and 0.8 V versus a saturated calomel electrode (SCE). It was found that the areal capacitance increased as active material loading increased, and an electrode with an active material loading of 17 mg cm−2 had an areal capacitance of ~ 2.5 F cm−2 at 1 mV s−1. Furthermore, a Ni nanoarray electrode with 3.5 mg cm−2 MnO2 retained its initial capacitance after 20,000 cycles at 10 mA cm−2, indicating that electrodeposition onto the Ni nanowire array provided sufficiently strong adhesion to survive many charge/discharge cycles.

Another type of architecture synthesized on a flat conducting substrate is the bicontinuous inverse opal structure [60,61,62]. Liu et al. [63] synthesized an inverse opal V2O5/graphene composite electrode and investigated its EES performance as a Li+ intercalation cathode in a non-aqueous electrolyte. Polystyrene spheres deposited on tungsten foil were used as templating agents for the inverse opal architecture, followed by the deposition of Ni, graphene, and V2O5. Subsequent etching of the polystyrene and Ni left behind only a V2O5/graphene composite electrode on tungsten. Even after 2000 cycles at a rate of 5 C, a 6-μm-thick electrode retained a capacity of 203 mAh g−1 (based on the mass of the full electrode). A commercial Ni foam sacrificial scaffold was also used to synthesize a 100-μm-thick electrode. The capacity of the thick electrode after 1000 cycles at a 5 C rate was 198 mAh g−1. This study demonstrated that the bicontinuous electron transport and electrolyte diffusion pathways enabled high capacity and cycle life in the V2O5/graphene inverse opal architecture.

Another strategy to develop a hierarchical electrode architecture is to synthesize porous metal foils. Hou et al. [64] fabricated porous copper by chemically dealloying magnetron-sputtered copper manganese. Subsequently, MnO2 was plated on the nanoporous Cu using an electroless plating technique. Since the MnO2 was directly plated on the Cu surface, there was good electron transport throughout the electrode layer and the porous structure was maintained for electrolyte diffusion. The electrochemical characterization was performed by cycling in a Li+ non-aqueous electrolyte between 0.01 and 3.0 V versus Li/Li+ to determine the performance of the electrode for the full conversion reaction of MnO2 to metallic Mn. This architecture accommodated the significant volume changes during the repeated conversion between MnO2 and metallic Mn, and the capacity retention was more than 90% after 1000 cycles at 8.4 A g−1. Tubular metals have also been shown as effective electrode scaffolds for oxides undergoing conversion reactions [65]. Figure 2c shows Ni nanotubes grown by Li et al. [50] using an electrochemical alloying/dealloying process. Using two subsequent electrodeposition processes, Fe3O4 nanoparticles were grown on the Ni nanotube arrays followed by a shell of MnO2, which added to the overall capacity and was hypothesized to protect the Fe3O4 nanoparticles from pulverization brought on by the conversion reaction during electrochemical cycling in a LiPF6 electrolyte. This electrode architecture had a capacity retention of 42.6% upon increasing the rate from 0.2 to 8 C. Furthermore, between the 2nd and 1000th cycle, the capacity retention was ~ 96% when cycling at a 1 C rate. Overall, synthesizing nanostructured conductive materials on a flat substrate enables the electronic conductivity necessary for thick energy storage electrodes.

Assemblies

Electronically conductive nanostructured materials can be assembled in such a way that they form hierarchical structures with good electronic and ionic conductivity. Carbon materials such as carbon nanotubes (CNTs) [46, 66] and graphene [67] are prime candidates for this type of approach. Recently, Xu et al. [68, 69] reported a holey graphene conductive assembly with tunable porosity suitable for fast electrolyte and ion diffusion. Sun et al. [70] combined solutions of graphene oxide and Nb2O5 precursor, which formed a free-standing electrode after two freeze-drying and annealing steps (Fig. 3). This structure allowed for an oxide mass loading of up to 22 mg cm−2. The Li+ intercalation behavior of the Nb2O5/holey graphene framework electrode was compared to a Nb2O5/graphene framework electrode (graphene without nanopores) by cycling in a LiPF6 non-aqueous electrolyte. As the charge/discharge rate increased from 1 to 100 C, the Nb2O5/holey graphene framework electrode showed superior performance compared to the Nb2O5/graphene framework, as demonstrated in Fig. 3. When both were tested with a Nb2O5 loading of 6 mg cm−2 at a 100 C rate, the Nb2O5/holey graphene framework had a capacity of 75 mAh g−1, while the Nb2O5/graphene framework had a capacity of only 35 mAh g−1. The effect of mass loading on the gravimetric capacity was also investigated, and it was found that the Nb2O5/holey graphene framework had better capacity retention than the Nb2O5/graphene framework and Nb2O5/graphene (with no framework structure). These results demonstrated that holey graphene is a promising scaffold for TMOs particularly at the high mass loadings necessary for many applications.

Figure 3
figure 3

(top) Cross-sectional scanning electron micrographs of the Nb2O5/holey graphene framework before and after compression. (Bottom) The specific capacity measured during galvanostatic charge/discharge tests in a LiPF6 non-aqueous electrolyte shows that the Nb2O5/holey graphene electrodes (hollow) exhibit higher capacity than the Nb2O5/graphene electrodes (filled) at rates between 1 and 100 C and for three different Nb2O5 loadings. Adapted with permission from Ref. [70]. Copyright 2017 American Association for the Advancement of Science

Full size image

Free-standing metal oxide nano- and microstructures grown on porous substrates

Foams and sponges

Many metal foams are commercially available, while other, more advanced carbon foams can now be synthesized. These structures are in many ways ideal supports for TMOs. The hierarchical structure of a metal foam exhibits high surface area for active material support, macropores for easy electrolyte diffusion, and interconnected electronically conductive pathways. These attributes enable high active material loading per footprint area, which is a key requirement for high areal capacity. Electrically conductive foam scaffolds are compatible with several synthesis approaches, including in situ growth [71, 72], electrodeposition [42] and, most commonly, hydrothermal synthesis [36, 73,74,75,76]. These approaches enable control of morphology, size, and mass of the deposited material, allowing for good tunability of the final electrochemical behavior. The nanostructure of the TMO in contact with the foam architecture leads to good electrolyte accessibility to the electrochemically active sites and reduces the ion diffusion path lengths within the TMO.

Metal foams

Nickel foam and copper foam scaffolds are particularly interesting because they can be purchased at low cost and require very little preparation prior to TMO synthesis. The mechanically robust, macroporous architecture of the Ni foams provides open channels for electrolyte absorption and transport, as well as reliable metallic conduction throughout electrodes with initial thicknesses that can reach several millimeters. When using these foams as scaffolds for EES, it is important to keep in mind the redox activity of the foam itself, along with any native surface oxides that are present. A prominent example of this is the redox activity of Ni foam in alkaline electrolytes. Several studies demonstrated simple synthesis techniques that effectively incorporate nanostructured TMO arrays on the surfaces of Ni foam [36, 73, 74, 77,78,79,80] and copper foam [81,82,83]. For example, Wu et al. [36] used hydrothermal synthesis to grow MnO2 on the surface of Ni foam. As shown in Fig. 4a, the resulting electrode architecture consisted of MnO2 nanosheets extending from the surface of micron-sized plate-like MnO2 directly adhered to the Ni foam scaffold. The adhesion between MnO2 and the Ni foam scaffold was beneficial for efficient electron transport, while the hierarchical structure enabled good oxide mass loading, high MnO2/electrolyte interfacial area, and short ion diffusion distances.

Figure 4
figure 4

Scanning electron micrographs of TMOs grown on foam scaffolds: a MnO2 nanosheet-on-plates grown on a Ni foam scaffold (adapted with permission from Ref. [36]. Copyright 2018 Elsevier); b graphene and CNTs were initially grown on a Ni foam scaffold, followed by hydrothermal growth of δ-MnO2 on the carbon surfaces (adapted with permission from Ref. [86]. Copyright 2018 Elsevier); c core–shell Co3O4 nanowire-NiO nanosheet grown on a Ni foam scaffold (adapted with permission from Ref. [42]. Copyright 2012 American Chemical Society); d V2O3 grown on graphene foam via solvothermal synthesis (left) and V2O3 with CNTs grown on the surface via chemical vapor deposition (right) (adapted with permission from Ref. [88]. Copyright 2016 John Wiley and Sons); e V2O5 nanobelts grown on graphite foam (left) with subsequently electropolymerized PEDOT, which uniformly coated the V2O5 surface (adapted with permission from Ref. [89]. Copyright 2014 John Wiley and Sons)

Full size image

The electrochemical performance and stability of TMOs on Ni foam can be further improved by the addition of a layer or array of carbon, such as graphene nanosheets or CNTs, between the nickel and the TMO [75, 84, 85]. The Ni foam architectures with a carbon array proved to be more effective than bare Ni foam for accommodating the strain induced by large volume expansions of the active material during charging/discharging [71]. Zhai et al. [86] demonstrated the growth of both graphene flakes and CNTs from Ni foam, followed by the growth of δ-MnO2 (birnessite) via a hydrothermal reaction, as shown in Fig. 4b. The graphene and CNTs provided additional electronic transport paths and increased the scaffold surface area, which enabled the filling of more pore volume with the δ-MnO2 active material. The conversion reaction of the electrode was tested in a LiPF6 electrolyte. The high-frequency region of the Nyquist plot obtained via electrochemical impedance spectroscopy showed a reduction in the size of the semicircle. This was attributed to a reduction in the charge transfer resistance during the conversion reaction of MnO2 to metallic Mn. In a related approach, Jia et al. [87] hydrothermally synthesized metal oxide nanosheets containing manganese and cobalt onto Ni foam. Subsequent annealing of the electrode in air led to the formation of MnO2 nanosheets which had pores filled with cobalt nanospheres. These cobalt nanospheres subsequently served as catalysts and growth sites for the plasma-enhanced chemical vapor deposition of CNTs. These CNTs grew through pores in the MnO2 nanosheets and entangled between the nanosheets to form an interconnected conductive network. The capacitive energy storage performance of the CNT/MnO2 composite on Ni foam was tested in a Na2SO4 aqueous electrolyte. The CNT/MnO2 on Ni foam electrode showed better cycling stability as compared to Co-MnO2 on Ni foam, retaining ~ 95% and 63% of the initial capacity, respectively, after 10,000 cycles at 10 A g−1. Carbon nanostructures integrated with metal foams or nanostructured TMOs can lead to a higher surface area to support TMO deposits and more pathways for electronic conduction.

Another approach to improve the electrochemical performance of TMOs deposited on metal foams is by growing core–shell structures of two different TMO materials [76, 90, 91]. Xia et al. [42] demonstrated this idea through the synthesis of Co3O4 nanowire/NiO nanoflake core–shell arrays grown on Ni foam, shown in Fig. 4c. Electrochemical characterization was carried out in an aqueous KOH electrolyte. Upon increasing the rate from 2 to 40 A g−1, the capacitance retentions for the Ni foam + Co3O4/NiO, Ni foam + Co3O4, and Ni foam + NiO were 85%, 76.7%, and 66.6%, respectively. For the same electrodes, the capacitance retentions after cycling for 6000 cycles at 2 A g−1 were 95.1%, 85.5%, and 56.7%, respectively. These capacitance retentions were based on the maximum capacity that was reached after some activation period during the first ~ 1000 cycles. Two advantages of using the Co3O4/NiO core/shell architecture were (1) the Co3O4 was sufficiently electronically conductive (extending the electron transport paths to more NiO) and (2) the NiO shell improved the mechanical stability during the continuous electrochemical activation process throughout the first ~ 1000 cycles.

Carbon-based foams

In addition to providing better electrochemical stability, carbon-based foams generally provide higher surface area and lower mass density, particularly as compared to metal foams. The lower mass density is important to achieve high gravimetric energy and power densities. In comparison with metal foams, which can be 10–15 mg cm−2, carbon-based foams have achievable mass densities of < 1 mg cm−2 [89]. These free-standing, electrically conductive carbon-based materials are also suitable for the TMO synthesis protocols used for metal foams.

The carbon foam substrates that are most commonly used are graphene foam [92] and CNT foam [66, 93, 94]. Similar to what was discussed for metal foams, it is possible to grow CNTs from the surface of the graphene foam, which increase the surface area for TMO growth while still providing suitable electron transport pathways. Guan et al. [95] deposited nanocrystalline Fe2O3 on CNT forests on graphene foam. These electrode architectures demonstrated good cycling stability in an aqueous KOH electrolyte. After 50,000 cycles at 20 mA cm−2, the average capacity retention was 97%. While 50,000 cycles exceed the EES cycle life requirements for portable electronics, this result demonstrates the ability of the architecture to extend EES device lifetime, which is critical for transportation and grid storage applications. Additionally, Xia et al. [88] demonstrated that growing TMO nanostructures on graphene foam and subsequently growing CNT forests on the TMO surface is another viable approach to improve charge transport throughout thick electrodes. Figure 4d shows the graphene foam + V2O3/CNT and the graphene foam + V2O3 electrode architectures that were compared in this study. Galvanostatic charge/discharge experiments in a NaPF6 non-aqueous electrolyte were used to characterize the intercalation of Na+ into V2O3. The impact of the electrode architecture was evident at fast charge/discharge rates. Upon increasing the current density from 2 to 10 A g−1, the capacity retention of the composite with CNTs was 52%, whereas that of the composite without CNTs was only ~ 6%.

Instead of growing CNTs on the TMOs, Chao et al. [89] used solvothermal synthesis to grow V2O5 on graphene foam and then electrodeposited poly(3,4-ethylenedioxythiophene) (PEDOT) on the V2O5 surface. Control of the electropolymerization of EDOT (3,4-ethylenedioxythiophene) led to the formation of a thin layer of PEDOT on the V2O5 and graphene foam, as shown in Fig. 4e. This composite exhibited a good cycling stability in a LiPF6 non-aqueous electrolyte for the intercalation of Li+ into V2O5, with 98% capacity retention after 1000 cycles at 60 C. It was hypothesized that the PEDOT layer was able to accommodate the volume expansion of the V2O5 and potentially reduced the polarization at the electrode at high C-rates. The results highlight the utility of conductive polymer coatings on TMOs for additional electronic conductivity as well as strain accommodation. Another benefit is that most can be deposited at low temperatures, decreasing cost and avoiding thermally induced structure and morphological changes.

Chen et al. [72] modified a CNT sponge with hydrothermally grown Co3O4 nanoparticles by a densification and compaction procedure, which was a simple approach to achieve high areal capacity. To achieve higher mass density, the CNT sponge was compressed by a factor of about 10 to allow the electrode to fit into a coin cell. The increase in density is evidenced in the scanning electron micrographs in Fig. 5a, b. The Co3O4 mass was kept at approximately 60% of the total electrode mass. As the thickness of the CNT sponge was increased from 2 to 5 mm, the amount of Co3O4 mass per area increased from 7.3 to 14.3 mg cm−2. Thus, the thickness of the CNT sponge was used to tune the active material loading. The sponge electrode was subsequently compacted to a thickness between 0.2 and 0.4 mm. The compacted electrode architecture with 50% strain was tested in a LiPF6 non-aqueous electrolyte for the conversion from Co3O4 to metallic Co between 0.01 and 3.0 V versus Li/Li+. As shown in Fig. 5c, this final electrode architecture led to an increase in areal capacity with mass of Co3O4, which allowed the electrode to reach the active material loading necessary for many commercial applications (> 10 mg cm−2). Upon increasing the active material loading, impedance spectroscopy revealed a small increase in charge transfer resistance while the areal capacity increased nearly linearly. Based on these results, it was concluded that the densification led to a trade-off between increased electrical conductivity and reduced porosity (which possibly restricted electrolyte diffusion). Ultimately, this approach provides a path toward controlling the electrode architecture via a top-down method to obtain high areal capacity.

Figure 5
figure 5

Scanning electron micrographs of Co3O4/CNT sponges from 0% strain a to 90% strain b show the effect of strain on densification; c areal capacity increases with increased Co3O4 mass loading, which is controlled by the initial height of the CNT sponge. Adapted with permission from Ref. [72]. Copyright 2018 John Wiley and Sons

Full size image

Textiles

Textiles, such as carbon cloths, felts, and papers, can also serve as free-standing substrates for TMO electrodes. They are typically composed of micron-sized carbon fibers. These fibers are mechanically quite stable and as a result offer reliable electron transport pathways. Compared to foams, textiles typically exhibit lower surface area, but they offer the advantages of low-cost and scalable processing.

Carbon cloths and felts

Carbon cloths and felts are commonly used as flexible, conductive scaffolds that offer higher surface area compared to foils. They allow for continuous electron transport paths and eliminate inactive materials such as conductive additives and binders. Carbon cloths and felts are particularly beneficial for electrodes for flexible energy storage devices [35, 44, 96,97,98,99], which could be used in smart textiles. Carbon fibers can also be used to design devices in a rope or twisted configuration [100]. Furthermore, carbon cloths and felts are compatible with several TMO syntheses, such as chemical bath deposition [41, 43, 44], atomic layer deposition [47, 48], hydrothermal synthesis [47, 48, 96, 101,102,103,104] and electrodeposition [35, 97, 100].

Liu et al. [41] used chemical bath deposition to grow ZnCo2O4 nanoplate arrays on carbon cloth. The oxide nanoplate-on-carbon cloth architecture, shown in Fig. 6a, had several advantages, such as good electron conduction paths through carbon fibers, short ion diffusion lengths in the nanoplates, and adhesion between the oxide and carbon substrate. This led to good cycling stability during the conversion reaction between ZnCo2O4 and lithium in a non-aqueous Li+ electrolyte. The conversion reaction involved both Zn and Co and led to a capacity of ~ 3 mAh cm−2 after 100 cycles at 0.24 mA cm−2. Using electrodeposition, Huang et al. [35] synthesized MnO2 nanosheets on carbon cloth. At a synthesis temperature of 60 °C, ε-MnO2 nanosheets formed on the carbon cloth followed by the growth of hollandite (α-MnO2) nanorods on the surface of the nanosheets. The cycling stability and lack of morphology change of the electrode after cycling suggested that the MnO2 nanosheet/nanorod on carbon cloth architecture accommodated the strain due to Na+ de/intercalation during cycling in an aqueous 1 M Na2SO4 electrolyte. An electrode with an active material mass loading of 10 mg cm−2 showed an overall increase in capacity after 2000 cycles, which was ascribed to some type of activation process during the first 200 cycles. It was hypothesized that the interface between the α-MnO2 nanorods and ε-MnO2 nanosheets, along with the increase in active material-electrolyte interfacial area, led to better electrochemical performance as compared to a carbon cloth electrode that contained only the nanosheets.

Figure 6
figure 6

Scanning electron micrographs of nanostructured TMOs on carbon textile architectures and in carbon paper electrode architectures: a ZnCo2O4 nanoplates uniformly grown on carbon fibers in a carbon cloth electrode via chemical bath deposition (adapted with permission from Ref. [41]. Copyright 2018 Elsevier); b Co3O4 nanorod bundles hydrothermally grown on a carbon cloth architecture that was uniformly coated with polypyrrole via electropolymerization (adapted with permission from Ref. [98]. Copyright 2018 Royal Society of Chemistry); c Fe2O3 distributed throughout a binder-free graphene/CNT composite paper electrode architecture (adapted with permission from Ref. [83]. Copyright 2015 Elsevier); d electrospun composite TiO2/carbon nanofibers with Fe2O3 nanoparticles grown on the surface via a vapor–solid reaction (adapted with permission from Ref. [34]. Copyright 2019 Elsevier)

Full size image

Huang et al. [97] electropolymerized poly(vinylpyrrolidone) on carbon cloth and subsequently annealed the electrode and electrodeposited V2O5, resulting in a V2O5/carbon nanowire array on carbon cloth. This electrode architecture led to good rate capability during Li+ intercalation to/from V2O5 in an aqueous LiCl electrolyte, retaining 52.4% of the capacity when the current increased from 1 to 100 mA cm−2. This electrode architecture was used in a 2 V flexible asymmetric supercapacitor device, which retained 91% of the capacitance after 8000 cycles at a rate of 40 mA cm−2.

Conductive polymers have been incorporated into TMO/carbon cloth composite electrodes, similar to approaches taken using foam architectures. Ma et al. [98] used a hydrothermal reaction to grow Co3O4 nanorod bundles on carbon cloth followed by the electropolymerization of polypyrrole in a core–shell architecture, as shown in Fig. 6b. Electrochemical analysis was carried out in an aqueous KOH electrolyte. This architecture proved to be advantageous based on the higher capacity of the Co3O4@polypyrrole electrode compared to just Co3O4 nanorod bundles, presumably due to the polypyrrole layer facilitating electron transfer. The Co3O4@polypyrrole electrode had a capacity retention of 97.4% upon increasing the current density from 2 to 20 mA cm−2, and the capacity was stable after 2000 cycles.

Zheng et al. ball-milled LiCoO2 and incorporated it into carbon cloth along with a smaller scale conductive additive (Ketjenblack). The resulting electrode exhibited a high areal capacity of 28 mAh cm−2 for Li+ de/intercalation in a non-aqueous electrolyte [23]. It was hypothesized that in high active material loading electrodes, conductive pathways must be present at multiple length scales to effectively distribute charge [105]. Using this approach, an electrode with a LiCoO2 loading of 213 mg cm−2 reached a discharge capacity of ~ 28 mAh cm−2 at 0.8 mA cm−2. For an electrode with a LiCoO2 loading of 71 mg cm−2, the capacity retention was ~ 70% upon increasing the rate from 0.8 to 1.6 mA cm−2. The integration of conductive materials with different dimensions thus enables high active material loading and therefore areal capacity. Further research will be necessary to understand how the various conductive additives interact with the TMOs to provide sufficient electronic transport even at high rates.

Electrode papers

Free-standing electrode paper architectures can also be utilized as scaffolds for TMOs. The fabrication of the papers can be as simple as vacuum filtration [106] or via more complicated techniques such as floating catalyst chemical vapor deposition [107]. Electrode papers can also be made via a scalable electrospinning technique that has become a popular synthesis approach to obtain interconnected electron transport throughout thick electrodes [108]. Wang et al. [83] utilized the vacuum filtration procedure to synthesize a composite electrode of Fe2O3, CNTs and graphene oxide shown in Fig. 6c. The surface area of this composite was 37.8 m2 g−1, indicating that this route does not lead to very high surface area electrodes. The EES performance of these electrode papers was investigated for the conversion reaction of Fe2O3 in a non-aqueous Li+ electrolyte. After 120 cycles at 50 mA g−1, the electrode had a high reversible capacity of 716 mAh g−1, which was higher than either a graphene-CNT composite or Fe2O3 electrode. The performance improvements of the composite electrode were attributed to better electron transport and electrolyte diffusion throughout the electrode architecture. It was also hypothesized that the graphene and CNTs support the Fe2O3 nanoparticles and prevent cracking during cycling, which is particularly beneficial in conversion reactions which lead to large volume changes.

More complex techniques for fabricating electrode papers, such as electrospinning, lead to better control over the nanostructure and porosity than vacuum filtration [108,109,110]. Yang et al. electrospun TiO2 nanofibers in polyvinyl pyrrolidone and calcined them in a nitrogen atmosphere to carbonize the polymer. The resulting TiO2/carbon nanofibers were then soaked in iron chloride hexahydrate and heated in an autoclave in the presence of ammonia to obtain the nanostructured Fe2O3/TiO2/carbon nanofibers shown in Fig. 6d [34]. The fibrous electrode architecture allowed for a high electrode–electrolyte interfacial area in a Li+ non-aqueous electrolyte as well as sufficient electronic transport throughout the electrode. This supported the Li+-driven conversion reaction of Fe2O3 and intercalation reaction of Li+ into TiO2. Kong et al. [111] also used the electrospinning technique to form V2O5 nanosheets that were subsequently encapsulated in carbon using chemical vapor deposition. Electrochemical analysis in a Li+ non-aqueous electrolyte showed that the capacity was close to the theoretical value for the intercalation of 2 Li+ per V2O5, and the capacity retention was 91.7% after 200 cycles at a 0.5 C rate. Furthermore, the electrode architecture had good electronic transport and electrolyte diffusion pathways, which was demonstrated by the retention of 40% of the capacity when cycling from 0.1 to 100 C (36 s discharge or charge). Electrospun fiber papers have emerged as an interesting architecture for free-standing electrodes that exhibits flexibility, good stress accommodation, and sufficient electronic and electrolyte diffusion paths through a simple and scalable synthesis procedure.

Aerogels

Aerogels present a promising high surface area architecture that enables good electron transport and porosity for both active material deposition and electrolyte diffusion [18, 112]. Some aerogels are even reversibly compressible, making them viable candidates for flexible next-generation batteries [113]. Although there have been many examples of aerogel electrodes that use binders [114, 115], there are some successes in binder-free aerogels [116,117,118,119]. For example, Liu et al. [120] formed aerogels of vertically aligned graphene sheets separated by CNTs and octahedral SnO2 nanocrystals, as shown in Fig. 7a. This architecture led to good contact between the conductive carbon materials and the active SnO2, while ensuring good permeability of the Li+ non-aqueous electrolyte throughout the channels of the aerogel. The good contact was particularly important for the conversion reaction of SnO2. In addition to good rate capability, the aerogel electrode was capable of retaining 80% of the initial capacity after cycling 1000 times at a 0.1 C rate. Furthermore, Gao et al. [121] used a N-doped graphene aerogel framework to deposit V2O5 nanoparticles via a hydrothermal method. Figure 7b shows how the resulting aerogels contained nanoparticles that were well adhered to, and encapsulated by, the graphene sheets. These composite aerogels accommodated the volume expansion of the V2O5 during intercalation of Li+ from a non-aqueous electrolyte, with a capacity of 283 mAh g−1 and a capacity retention of ~ 78% after 500 cycles at 1 A g−1. Furthermore, the capacity retention was 71% upon increasing the current density from 100 to 1000 mA g−1. These studies indicate that aerogels could provide the electronic conductivity and ion transport needed for high-energy and high-power electrodes; however, further research is needed to develop binder-free electrodes with higher active material loading.

Figure 7
figure 7

a Scanning electron micrographs of SnO2 nanoparticles and CNTs dispersed in a vertically-aligned graphene aerogel, and an optical image of the free-standing electrode (inset) (adapted from Ref. [120]. Licensed under CC BY 4.0); b Scanning electron micrographs and optical images of a flexible, free-standing graphene framework electrode architecture with V2O5 nanoparticles uniformly dispersed on the carbon surface (adapted with permission from Ref. [121]. Copyright 2018 Elsevier); c Scanning electron micrographs and optical micrographs (insets) of a 3D-printed graphene aerogel with electrodeposited MnO2 and d corresponding electrochemical analysis that demonstrate that the areal capacitance of the electrode increased linearly with increased MnO2 areal loading and e the relatively consistent decrease in gravimetric capacitance as charge/discharge rate increased for electrodes with MnO2 loading between 2 and 45.2 mg cm−2 (adapted with permission from Ref. [122]. Copyright 2019 Elsevier)

Full size image

Yao et al. [122] reported the fabrication of a 3D printed graphene electrode architecture using a direct ink writing method that was subsequently used as the substrate/working electrode for the electrodeposition of nanostructured MnO2, as shown in Fig. 7c. The active material mass loading was controlled by the electrodeposition time. This electrode architecture demonstrated good electrochemical intercalation of Li+ in an aqueous LiCl electrolyte. Figure 7d, e shows that areal capacitance increased linearly from ~ 0.5 to 11.55 F cm−2 at 0.5 mA cm−2 when mass loading increased from 2 to 45.2 mg cm−2, while the gravimetric capacitance remained relatively constant. This proved that most of the active material was utilized for energy storage even at the higher mass loading. The scalability of this method was demonstrated with a 4-mm-thick electrode that had an MnO2 mass loading of 182.2 mg cm−2. This electrode achieved a high areal capacitance of 44.13 F cm−2 at a rate of 0.5 mA cm−2. This example highlights the potential for 3D printing to be integrated into the manufacturing of free-standing TMO electrodes. Additional research will be necessary to push the resolution limits of the 3D printing process, which currently leads to large porosity which decreases the volumetric energy density.

Electrode architectures for solid-state battery cathodes

Solid-state batteries with inorganic solid electrolytes are increasingly being considered for next-generation EES with improved energy density and safety than existing Li-ion batteries. Inorganic solid-state batteries have the potential to replace flammable organic liquid electrolytes due to recent improvements in ionic conductivity of inorganic solid electrolytes [123]. Furthermore, when such electrolytes are paired with Li metal and other Li alloys they could lead to higher energy densities than existing EES [124]. Many research efforts have focused on developing solid electrolytes with high ionic conductivity and investigating the interface stability between the electrolyte and electrodes [125,126,127,128,129,130,131]. Several promising electrolyte materials have been tested, including various sulfides, oxysalts, and oxides, perhaps most notably cubic garnet type oxides such as lithium lanthanum zirconate (LLZO) [132, 133]. Although there are several challenges that must be addressed with ISEs, especially at the electrode/electrolyte interfaces [134], it is critical to also develop new electrode architectures for cathodes with high active material loading in contact with solid electrolytes [135, 136]. Of particular importance for these new electrode architectures is to develop and maintain a high electrochemical interface area between the solid electrolyte and electrode during electrochemical cycling.

Thus far, most reports on inorganic electrolyte-based solid-state batteries use a cathode that was prepared via slurry processing. Unlike the slurry electrodes fabricated for liquid electrolyte-based EES, these consist of the cathode active material, conductive additive, and inorganic solid electrolyte powder in order to ensure high interfacial area between the cathode and electrolyte [137]. However, there are emerging attempts to develop more advanced architectures for better contact between the active material and current collector (for electronic transport) and active material and solid electrolyte (for ionic transport) [136]. The importance of increasing the interfacial contact between the electrolyte and composite electrode was recently illustrated by van den Broek et al. [138]. The electrochemical behavior of a solid-state battery with an “interface-engineered” LLZO was compared to a flat electrolyte pellet, as shown in Fig. 8a. Both cells contained cathodes fabricated from slurries of Li4Ti5O12 active material, LLZO electrolyte powder, polyvinylidene fluoride (PVDF), and conductive carbon particles. Through electrochemical impedance spectroscopy and equivalent circuit modeling, it was evident that the resistance of the interface-engineered full cells was reduced by more than a factor of two. Furthermore, the gravimetric capacity obtained from galvanostatic charge/discharge experiments showed up to a tenfold increase in capacity for the interface-engineered electrodes, which was largely attributed to a decrease in the charge transfer resistance at the active material-electrolyte interface. Wang et al. [139] developed a thermal soldering technique to engineer the interface between LiCoO2 and LLZO. LLZO particles and LiCoO2 particles were coated with Li2CO3 and Li2.3C0.7B0.3O3, mixed together, and deposited onto a Li2CO3-coated LLZO pellet. The composite electrode and electrolyte were then sintered at 700 °C, which caused the Li2.3C0.7B0.3O3 to react with the Li2CO3 and form ionically conductive Li2.3−xC0.7+xB0.3−xO3 (LCBO). The resulting electrode architecture, shown in Fig. 8b, prevented interdiffusion between the LiCoO2 and LLZO due to their separation by the LCBO matrix. Surprisingly, no additional conductive additives were necessary because apparently the LiCoO2 had sufficient electronic conductivity (even in its delithiated state). The electrochemical performance was first investigated at 100 °C with a LiCoO2 loading of ~ 1 mg cm−2 and a Li metal anode. The reversible capacity was ~ 106 mAh g−1 at a 0.05 C rate, which was close to the theoretical capacity of 115 mAh g−1 for LiCoO2 cycled between 3.0 and 4.05 V versus Li/Li+. These electrodes retained ~ 66% of the initial capacity upon increasing the rate from 0.05 to 1 C, as shown in Fig. 8b. Furthermore, the reversible capacity measured at 25 °C was ~ 83 mAh g−1 at a 0.05 C rate. There was still some mechanical degradation that occurred due to stresses/strains brought on by LiCoO2 expansion and contraction, which resulted in capacity fade. This last point highlights the need to understand the electro-chemo-mechanics of solid electrolyte–cathode interfaces and to use this knowledge to design better electrode architectures.

Figure 8
figure 8

Electrode architectures found in solid-state batteries with inorganic solid electrolytes: a Scanning electron micrographs that demonstrate the improved integration of solid-state battery slurry electrodes into a porous interface-engineered LLZO electrolyte as compared to the slurry deposited on an unmodified LLZO pellet (adapted with permission from Ref. [138]. Copyright 2016 John Wiley and Sons); b scanning electron micrographs before and after soldering of LiCoO2 particles to LLZO particles using a matrix of Li2.3C0.7B0.3O3, a schematic diagram of the interface and electrochemical analysis showing the capacity retention of an electrode cycled at 100 C at rates ranging from 0.05 to 1 C (adapted with permission from Ref. [139]. Copyright 2018 Elsevier); c scanning electron micrograph of the interface formed between a sulfide electrolyte and NCM active particles after compression of the composite pellet (adapted with permission from Ref. [140]. Copyright 2016 Elsevier)

Full size image

Sulfide-based solid electrolytes are attractive because they not only have high ionic conductivity but are mechanically more ductile than oxides, which can mitigate the capacity fading due to mechanical degradation in solid-state batteries [141]. Furthermore, these mechanical properties enable lower temperature processing, such as cold pressing, as compared to solid oxide electrolytes [142]. To prevent the formation of a resistive layer between oxide cathodes and sulfide-based solid electrolytes, several coating layers, including ZrO2 [143], LiAlO2 [144], Li4Ti5O12 [145], and LiNbO3 [146], have been investigated. Sakuda et al. [140] demonstrated this approach by spray-coating ~ 10 nm of LiNbO3 onto LiNi1/3Co1/3Mn1/3O2 (NCM). Electrodes consisting of 75 wt% NCM and 25 wt% sulfide electrolyte were pressed to the sulfide electrolyte by applying a pressure of 330 MPa at room temperature. Figure 8c shows that upon decreasing the size of the solid electrolyte particles, there was improved contact and fewer cracks between the NCM and electrolyte. At a low current density and at 30 °C, the size of the NCM and solid electrolyte particles did not affect the capacity, and all of the electrodes delivered a reversible capacity between 140 and 150 mAh g−1 at a C/20 rate. At increased current densities, electrodes with smaller initial particle sizes for NCM and sulfide electrolyte (5 μm and 4 μm, respectively) delivered the highest reversible capacity with 126 mAh g−1 at a C/2 rate.

Another strategy to increase the cathode/solid electrolyte interfacial area is to deposit thin films of solid electrolytes on top of three-dimensional cathode arrays. Sun et al. used magnetron sputtering to deposit MoO3 cathodes. Depending on the deposition temperature, the MoO3 formed either a 3D vertically oriented nanosheet morphology or a 2D planar morphology. Amorphous lithium phosphate oxynitride (LIPON) was then deposited on top of the MoO3 nanosheets followed by a ~ 2-μm-thick Li metal anode [147]. Cyclic voltammetry showed that the 3D vertically oriented electrode architecture led to a decrease in cell resistance and polarization and improved Li+ kinetics as compared to the 2D planar morphology. Furthermore, galvanostatic charge/discharge revealed that the capacity retention of the 3D thin film battery was 41% upon increasing the rate from 50 to 1000 mA g−1, whereas the 2D thin film battery only achieved a capacity retention of 23%. There was also a substantial improvement in the cycling performance of the 3D thin film battery as compared to the 2D thin film battery. Furthermore, the 3D thin film battery had a capacity retention of 92.7%, compared to 76% for the 2D thin film battery, after 1000 cycles at 500 mA g−1. The improvements in rate capability and cycling stability of the 3D thin film battery were attributed to the increased electrode/electrolyte interface area and short Li+ diffusion lengths.

Another approach is for the solid electrolyte to define the electrode architecture [136]. Vapor deposition techniques, such as ALD and CVD, enable conformal coating of the active material on surfaces with rough or high aspect ratio features such as solid electrolytes [21]. For example, Xu et al. [148] developed a porous-dense-porous trilayer LLZO electrolyte for lithium/sulfur batteries, shown in Fig. 9a, using a scalable tape-casting technique. The thickness of the porous electrode layers was about 50–70 μm, and the thickness of the dense layer separating the two electrodes was 10–30 μm. The anode porous layer was first coated with zinc oxide, which improved the wetting of the subsequently deposited lithium. The cathode porous layer was coated with CNT ink until the surface was visibly dark prior to the infiltration of molten sulfur with a mass loading of 5.4 mg cm−2. The pores within the sulfur electrode were intentionally left slightly unfilled to accommodate the volume expansion during discharge. This electrode architecture achieved a capacity of ~ 1200 mAh g−1 at 50 mA g−1 (per gram of sulfur) and maintained ~ 90% of the initial capacity after 50 cycles. Gong et al. [149] developed the LLZO textile architecture shown in Fig. 9b using a template method that involved soaking of the textile template in a ceramic precursor followed by pyrolysis to remove the organic components. This architecture was used for a lithium-sulfur battery but was also proposed for TMO cathodes. These emerging strategies will likely pave the way toward the integration of solid electrolytes into commercial applications.

Figure 9
figure 9

a Schematic illustration and scanning electron micrographs of a trilayer porous-dense-porous tape-casted LLZO electrolyte with high surface area for sufficient contact between the electrode materials (in this case Li metal and sulfur) and the electrolyte, and a photograph of a large sheet of the electrolyte demonstrating the scalability of this approach (adapted with permission from Ref. [148]. Copyright 2018 Elsevier); b schematic and scanning electron micrograph that illustrate an LLZO ceramic textile that was used to define the electrode architecture, which was demonstrated for a lithium-sulfur battery (adapted with permission from Ref. [149]. Copyright 2018 Elsevier)

Full size image

Conclusions and outlook/perspectives

Recent advancements in the synthesis of nanostructured metals, carbons, and TMOs, along with their characterization, have enabled the development of a large number of “beyond slurry” electrode architectures for EES. As summarized here, these types of electrode architectures exhibit significantly improved energy storage kinetics as compared to their traditional slurry electrode counterparts and will be necessary for future EES that combines both high power and high energy density. Figure 10 graphically summarizes several of the electrode architectures discussed in this review. First, it should be noted that it is difficult to make this type of comparison because most manuscripts do not provide sufficient detail of all the electrode parameters. Future data mining of electrode architectures would be beneficial to compare new architectures to existing ones, but this will only be possible if sufficient data are reported about the area and volumetric material loading. Figure 10a plots the areal capacity as a function of applied current, and there is great variability in the initial capacity due to different energy storage mechanisms. Figure 10b shows that when the initial capacity is normalized, most of the reported electrode architectures exhibit very similar rate capability indicating similar kinetic limitations. Figure 10c shows that when the areal capacity is plotted as a function of areal mass loading, there is a difference between conversion electrodes and intercalation electrodes. The former do not require high areal mass loading to achieve high areal capacities due to very high gravimetric capacities. Such electrode architectures need to focus more on accommodating the significant volume changes of conversion electrode materials. On the other hand, intercalation-type electrodes do need high areal mass loading and here the focus should be on how to integrate more mass per area.

Figure 10
figure 10

a Areal capacity and b normalized capacity as a function of applied current and c areal capacity as a function of areal mass loading for some of the advanced electrode architectures described here. The size of the data point in a and b indicates the relative mass loading. While there is great variability in the areal capacity, the rate capability tends to follow similar trends indicating similar kinetic limitations

Full size image

There are still several open challenges and opportunities for these types of advanced electrodes. First, slurry electrodes benefit from high reproducibility and ease of large-scale manufacturing; “beyond slurry” electrode architectures need to do the same. This has been a challenge for nanostructured materials particularly for small-batch processes. It is likely that electrode architectures that minimize the number of steps required for manufacturing will be most successful. The high cost of nanostructured materials is also an issue. Some techno-economic studies suggest that the manufacturing costs of EES electrodes with carbon felts and CNTs could be similar to current methods; however, these estimates assume decreased costs from increased production (economy of scale) and expected performance improvements such as longer battery life [3, 150, 151]. With the capacity of roll-to-roll slurry manufacturing of Li+ batteries rapidly expanding, the performance and/or cost of next-generation electrodes needs to far exceed the status quo or be a “drop-in” replacement to existing manufacturing processes. Second, the obtained electrode architectures are still far from deterministic. In this respect, computational finite element modeling should be beneficial to maximize the utilization of the electrode porosity to provide just enough volume for mass loading and electrolyte infiltration. Third, much of the advancement in these next-generation electrode architectures arises from innovation in nanostructured current collectors which need to also be highly scalable and simple to manufacture. Fourth, such architectures are leading the way for the integration of EES with flexible electronics and smart clothing and this is likely to continue. Lastly, these next-generation electrode architectures may enable multi-functionality of EES that is yet to be fully explored, by incorporating, for example, self-healing and self-diagnostic features.