Insertion-type materials involve most cathode materials and some anode materials (such as graphite, Li₄Ti₅O₁₂, and TiO₂). The first generation of LIB uses LiCoO₂ and graphite as the positive and negative electrodes; the redox operation of both versus lithium is based on intercalation reactions. Alternative materials such as LiMn₂O₄, LiFePO₄, and Li₄Ti₅O₁₂ have also reached the market at different levels, bringing about incremental improvements in performance 32. Nonetheless, all these materials have intrinsic capacity limitations, which are derived from their redox mechanisms and structural aspects, i.e., the intrinsic redox activity of the transition metals and the changes the crystal structure can withstand. Such limitation handicaps the device in terms of energy density. Power density of these cathode materials with bulk sizes is also generally low due to the high level of polarization at high charge/discharge rates 32. Therefore, the selective crystallization approach was introduced to overcome these shortcomings by decreasing diffusion paths for mass transport and increasing the surface area for charge transfer.

The characteristic structure of insertion-type materials is an ionic diffusion path. It is clear that morphological control of nanocrystalline materials is significantly important. The selective exposure of surfaces normal to the most facile pathway for lithium-ion conduction is preferred, and the length of particles along the ionic diffusion direction should be decreased since it can enhance electrochemical performance by reducing transport path lengths as well as enhance the electrode/electrolyte contact 33. For example, in an olivine structure where ionic diffusion paths are along the b-axis (Figure 3), a smaller dimension along the ionic diffusion direction can be obtained via the realization of three types of morphologies: (1) zero-dimensional spherical nanoparticles; (2) one-dimensional (1D) nanorods oriented with long axes along the a or c direction; or (3) two-dimensional (2D) nanoplates with a (101) basal plane.

Based on this idea, nanoparticles of LiFePO₄ olivine materials have been synthesized by conventional solid-state synthetic methods and polyol synthesis 3435; 1D nanorods of LiMn_1-xFe_xPO₄ with a radial [010] direction can be selectively crystallized by a controlled hydrolysis method (Figure 4a, b, c) 36. Also, by employing solvothermal synthesis, 2D nanoplates of LiFePO₄ with large exposure of (010) face can be obtained (Figure 4d, e, f, g, h, i) 37. These nanostructured olivine electrodes exhibited better rate performances than bulkier materials due to smaller diffusion length, indicating that selective crystallization is an effective way to obtain high-electrochemical-performance materials.

The spinel-type LiMn₂O₄ has also attracted continued interest. LiMn₂O₄ is favorable for its safety and intrinsic rate capability and has been established as cathode materials for electric vehicle applications 38. However, the gradual capacity loss due to Jahn-Teller distortion of Mn³⁺ and Mn dissolution in the electrolyte has hindered its application 39. A solution to this problem is selective crystallization of nanostructured spinels such as LiMn₂O₄ nanorods and nanowires. In a nanostructured spinel, the phase-boundary energy dramatically changes with the particle size, which can promote the solubility and solid solution properties and thus changes the kinetics and thermodynamics of Li-insertion reactions. Freestanding, single-crystalline LiMn₂O₄ nanorods have been fabricated using single-crystalline MnO₂ nanorods as precursor via a simple solid-state reaction 40, as shown in Figure 5. LIB testing showed that LiMn₂O₄ nanorods can deliver a capacity of 100 mAh g^-1 at a high current density of 148 mA g^-1 with high reversibility and good capacity retention; after 100 cycles, more than 85% of the initial capacity was maintained. An extended method to prepare porous LiMn₂O₄ nanorods has also been reported, using porous Mn₂O₃ nanorods resulting from the thermal decomposition of MnC₂O₄ as precursor 41. The as-synthesized porous nanorods exhibited high rate capability and cyclability. An initial discharge capacity of 105 mAh g^-1 was obtained at a 10 C rate, and capacity retention of about 90% was obtained after 500 cycles. The authors attributed this durable, high rate capability to the unique, porous 1D nanostructure that gave rise to fast Li-intercalation kinetics and good structural stability for the spinel electrodes.

Vanadium oxide (V₂O₅) is a typical intercalation compound as a result of its layered structure; orthorhombic, crystalline V₂O₅ consists of layers of VO₅ square pyramids that share edges and corners (Figure 6). For LIB applications, V₂O₅ offers the essential advantages of low cost, abundant source, easy synthesis, and high energy densities 42. The crystallization of V₂O₅ into specific nanostructures also plays an important role in improving the electrochemical performance of rechargeable LIBs. For controlling the nanostructure of V₂O₅, Liu and Xue reported a solution route to fabricate single-crystalline V₂O₅·xH₂O nanorings and microloops (Figure 7a, b, c, d, e, f, g, h) 43, the formation of rolling structures are caused by the cation-induced asymmetric strain on layered-structure V₂O₅·xH₂O nanobelts. This work demonstrates that the novel nanoring structure, which has been observed previously only for polar surface-dominated structure materials, can also form in compounds without anion- and cation-terminated surfaces. This proposed cation-induced strategy extends the existing formation mechanism of nanorings and can be applied to other materials. Recently, Liu and Xue also reported a scalable, highly reproducible, and template-free process to crystallize a yolk-shell V₂O₅ microsphere cathode material (Figure 7i, j, k, l) 44. This featured material has a high specific capacity (280 mAh g^-1) in the initial discharge process and excellent retention of the initial capacity after 30 cycles. During the whole cyclic process, the Coulombic efficiency steadily kept the values higher than 95%. The enhanced electrochemical performance is closely related to the selectively crystallized yolk-shell microstructure. The hollow-structured feature facilitates the electrolyte transport, and the small granularity and cavity among individual nanoparticles (porous shell) can effectively prevent the amorphization of the electrode, which is the main cause of the capacity fading of V₂O₅ during cycling.

Besides increasing capacity, a high cell voltage resulting from a high (cathode) and low (anode) standard redox potential of the respective electrode redox reaction can also greatly improve the energy density of LIB. Due to the redox potential of an intercalation compound which mainly tracks the iono-covalency of the metal-X bonding, selective crystallization of cathode materials with fluorine substitution appears to be quite an attractive route to increase the material redox voltage as F is very electronegative 16. Recently, Barpanda et al. reported a polyanionic material that crystallizes in the triplite structure Li(Fe_1-δMn_δ)SO₄F 45. An open-circuit voltage of 3.9 V has been achieved, exceeding that of LiFePO₄ by 450 mV. Also, this new triplite phase is capable of reversibly releasing and reinserting 0.7 to 0.8 Li ions with a volume change of only 0.6% (compared with 7% and 10% for LiFePO₄ and LiFeSO₄F, respectively), to give a capacity of 125 mAh g^-1. Such a material could become a promising cathode to replace LiFePO₄. These new types of polyanionic compound, having the triplite structure, provide valuable information in the search for even better cathode materials.

In the case of anode materials, the formation of a solid-electrolyte-interface (SEI) layer on which metallic lithium is deposited during a fast charge of the battery should also be a concern. The dendrites can grow to short-circuit the battery and ignite the electrolyte. Therefore, safety concerns have led to a search for anode materials having a redox couple in the range of 1.0 to 1.5 eV below the Fermi energy of lithium. The spinel Li₄Ti₅O₁₂ is reported to be a stable anode operating on the Ti(IV)/Ti(III) redox couple located at 1.5 V versus Li⁺/Li. It is capable of a fast charge and a long cycle life because no SEI layer is formed 464748. However, it has a low specific capacity (approximately 140 mAh g^-1), and the high redox potential (1.5 V) reduces the energy density of a cell using this anode. On the basis of these considerations, different niobium-based oxides such as KNb₅O₁₃ and K₆Nb_10.8O₃₀ have been investigated, which exhibit a reversible Li insertion toward the targeted voltage range of 1.0 to 1.5 V versus Li⁺/Li 4950. More recently, the mixed titanium-niobium oxides such as TiNb₂O₇ have been selectively crystallized as the anode for lithium batteries with some similar electrochemical properties 5152. Notably, carbon-coated TiNb₂O₇ gives a reversible specific capacity of approximately 285 mAh g^-1 cycled between 1.0 and 2.5 V versus Li+/Li with a Coulombic efficiency over 98% 52. These new-type anodes provided promising candidates for batteries with high rate, high cycle life, and better safety.

From the above discussion, it is evident that selective crystallization of insertion-type materials into nanosized particles and shapes with specific facets can effectively enhance the lithium-ion diffusion rate and improve their cycling performance; these improvements are all closely related to the crystal structure and reaction mechanisms of this kind of materials. As an alternative route, tuning the crystal structure by selective atom substitution can also improve the energy density of LIB by increasing the cell voltage.

Some main-group elements (e.g., Si, Ge, Sn, Al, Bi, Zn, and Sb) can alloy with lithium at a low potential. Representatively, Si and Sn can form lithium alloys with Li compositions up to Li_4.4M, giving theoretical specific capacities as high as 4,200 and 993 mAh g^-1, respectively 535455. Unfortunately, huge volume changes occur during the electrochemical lithiation/delithiation process 56. For example, a volume expansion on the order of 400% occurred during the formation of Li_4.4Si alloy, which causes cracking and eventual pulverization of the electrode and results in rapid capacity decline (Figure 8a).

As a solution against the volume change problem, selective crystallization of alloy materials into different nanoforms has been suggested because the toughness and adhesion effect within grain boundaries may increase at nanoscale and nanostructured alloys are more flexible to accommodate the strain induced by volume variations 57. Yang et al. first reported that nanosize Sn electrodes can exhibit much better cycling performance than bulk Sn electrodes due to a smaller absolute volume change 58. Recently, 1D nanowires of alloy materials have received more attention. Chan et al. reported that Si nanowire arrays grown directly on a current collector exhibited stable cycle performances, because the nanowires were not pulverized or broken due to facile strain relaxation of the nanowire geometry (Figure 8b) 59. Also, the efficient 1D electron transport and good contact with current collector also contribute to their improved LIB performance. Cao et al. presented a concept of using Cu-Si and Cu-Si-Al₂O₃ nanocable arrays directly grown from the current collector as LIB anodes (Figure 9) 60. The conductive Cu cores that are anchored to the copper foil act as both current collectors and structural reinforcements for the Si shell of the nanocables. The outer surface of the nanocables is readily modified by an additional sheath of Al₂O₃, which provides a stable Si/electrolyte interface. Both nanocables show excellent electrochemical performance including high specific capacity and cycling stability (1,890 and 1,820 mAh g^-1 under a discharge and charge current density of 0.3 A g^-1). After coating with Al₂O₃ sheaths, the Cu-Si-Al₂O₃ nanocables showed a remarkable high rate capability and delivered a capability of 1,140 mAh g^-1 at 7 A g^-1.

Carbon protection has been proven to be an effective route to enhance the cycle stability of alloying-type electrode materials 61. Because of their high capacity and wide availability, Si-based nanocomposites are among the most attractive anode materials, and much progress has been made in this regard. Some recent examples include graphene-grafted Si nanoparticles 62 and Si nanowires with carbon nanotube coatings 63. Notably, selective crystallization of Si on the hierarchical carbon spheres with irregular channels has been reported. The hierarchical composite possesses an interconnected, aperiodic, porous network with internal channels, enabling high accessibility of the active Si for fast lithiation. Large Si volume changes can be accommodated by the particle's internal porosity. A reversible capacity of 1,950 mAh g^-1 and a stable performance are attained 64. In the case of tin-carbon composite materials, porous structures including Sn nanoparticles confined in hollow carbon capsule and coaxial SnO₂@C hollow spheres proved to be promising anode candidates for highly reversible lithium storage 6566. Recently, Sn nanopillar arrays embedded between graphene sheets were assembled using a conventional film deposition and annealing process (Figure 10) 67. The as-formed three-dimensional (3D) multilayered nanostructure can be directly used as an anode material without adding any polymer binder and carbon black. This composite showed high reversible capacity (714 mAh g^-1) and excellent cycling performance at a high current density of 5 A g^-1 and demonstrated that a highly functional nanocomposite can also be fabricated by employing conventional top-down manufacturing methods and self-assembly principles.

To tackle the volume change problem, selective crystallization of alloying-type materials into specific structures has also been employed. Some peculiarly designed nanoscale electrodes, such as hollow nanostructures, have more free space and can endure larger stresses 68. For example, Liu et al. successfully designed a template-free route to fabricate double-shelled SnO₂-V₂O₅ nanocapsules 69. The formation mechanism of these double-shelled, hollow nanocapsules is a combination of both inward and outward Ostwald ripening processes, which is shown in Figure 11a. Ostwald ripening firstly took place at the surface of these solid nanospheres, which results in the void formation between two layers. Following this inward ripening process, the solid core of nanospheres ripened outward, leading to a double-shelled V₂O₃-SnO₂ hollow structure. Finally, double-shelled V₂O₅-SnO₂ nanocapsules were obtained by the calcination of these V₂O₃-SnO₂ nanocapsules in open air. Figure 11b, c, d, e, f, g, h, i shows the structural characterizations of these composite nanocapsules. The double-shelled structure is obvious from these pictures, and SnO₂ crystallines are homogenously distributed in the V₂O₅ matrix. These V₂O₅-SnO₂ hollow nanocapsules showed a large reversible capacity of 947 mAh g^-1 as an anode material and good cycling stability (can deliver a reversible capacity of 673 mAh g^-1 after 50 cycles). The reversible capacity as an anode can be further improved by increasing the content of SnO₂ in nanocomposites. With 15 wt.% SnO₂ content, the first discharge capacity can reach 1,776 mAh g^-1; after 20 cycles, the reversible discharge capacity maintains 1,046 mAh g^-1 without obvious capacity fading except for the first cycle. The excellent capacity retention was largely attributed to the large free space between the shells which can effectively accommodate the volume variation. It is worth noting that tin-based anodes have already been applied in the commercialized rechargeable LIBs. Therefore, the development of alloy anodes has opened a new avenue in the fabrication of advanced LIBs.

Interstitial-free, 3D transition-metal oxides (M_xO_y, M = Fe, Co, Ni, Mn, Cu, etc.) are capable of incorporating more than one Li per 3D metal, hence giving high Li storage capacities 70. The Li storage mechanism of the M_xO_y differs from the Li-intercalation and Li-alloying mechanisms. Transition-metal oxides are reduced to metal in the lithiation process (Equation 4). During the first reduction of the metal oxide, highly reactive metallic nanodomains embedded in a Li₂O matrix can be generated in situ, which contributed to the reversibility of this reaction 70. Based on this mechanism, reversible lithium storage proceeds more easily with the nanostructured oxides. Therefore, selective crystallization of transition-metal oxides has a significant role for their LIB performances. Similar to the Li-alloying process, the conversion reaction leads to volume variation upon the electrochemical cycling. Conceptually, approaches such as constructing hollow structures or nano-compositing are applicable to conversion-type anode materials as well.

Recently, Liu et al. reported a novel self-templated method to fabricate anisotropic Co₃O₄ porous and hollow nanocapsules from CoCO₃ precursors 71. The selective crystallization is based on the inside-out Ostwald ripening process. Figure 12a illustrates the transformation process from CoCO₃ precursors to anisotropic Co₃O₄ porous and hollow nanocapsules. During the solvothermal process, two spherical CoCO₃ colloids aggregated and fused together under the driving force of the magnetic dipole interaction between these spherical precursors, forming anisotropic, dumbbell-like structures. Subsequently, these dumbbell-like colloids underwent a ripening process to form nanorods, and the subsequent heat treatment of these CoCO₃ nanorods led to the formation of Co₃O₄ shells; the generation of nanoporous can be attributed to the release of CO₂ from the CoCO₃ nanocrystals along different directions. Figure 12b, c, d shows the surface morphology and microstructure of CoCO₃ precursors and corresponding Co₃O₄ porous and hollow nanocapsules; these nanocapsules have nanoporous shells with dense nanopores. The LIB testing of these Co₃O₄ porous nanocapsules showed that this kind of structure exhibited superior performances with good cycle life and high capacity, at a discharge/charge current density of 110 mA g^-1; these anisotropic Co₃O₄ porous and hollow nanocapsules showed a high reversible capacity of 1,018 mAh g^-1, and a capacity of 1,000 mAh g^-1 can be obtained after 20 cycles. These high performances can be attributed to their porous and hollow nanostructures and small size of building blocks. Self-supported Co₃O₄ mesoporous nanowires directly grown on a Ti substrate as current collector were prepared using a template-free ammonia evaporation-induced method 72. Due to the novel hierarchical structure that favors electrolyte diffusion and electric contact, the mesoporous nanowire array delivered a first discharge capacity of 1,124 mAh g^-1 and maintained a stable capacity of 700 mAh g^-1 after 20 cycles, and a considerable capacity of 450 mAh g^-1 can be obtained at a high rate of 20 C.

Compared to Co, Fe is of lower cost, lower toxicity, and higher abundance, rendering iron oxides as more attractive anode materials. In particular, the spinel Fe₃O₄ demonstrates high electronic conductivity and is suitable for potential high-power application. A 3D composite has been constructed by selectively crystallizing Fe₃O₄ nanoparticles encapsulated within carbon shells onto reduced graphene oxide (RGO) sheets (Figure 13a, b) 73, which exhibited enhanced anode performances in LIBs with a specific capacity of 842 mAh g^-1 and superior recycle stability after 100 cycles; these can be attributed to the unique 3D structure of the composite; the 2D layered structure of RGO combined with the close structure of carbon shells provided a rigid and highly conductive matrix for Fe₃O₄ nanoparticles. Besides Fe₃O₄, Fe₂O₃ has also attracted much interest. For example, selective crystallization of α-Fe₂O₃ hollow spheres with sheet-like subunits can be achieved by a quasiemulsion-templated method (Figure 13c, d) 74. Quasiemulsion microdroplets of glycerol were dispersed in water to serve as soft templates for the deposition of the α-Fe₂O₃ shell. When tested as anode materials for LIBs, these α-Fe₂O₃ hollow spheres showed a high reversible capacity of 710 mAh g^-1, even after 100 cycles. A two-step electrode design consisting of the electrochemically assisted template growth of Cu nanorods onto a current collector followed by electrochemical plating of Fe₃O₄ was also proposed (Figure 13e) 75. Using such electrodes, a factor of six improvement in power density over planar electrodes while maintaining the same total discharge time can be achieved. The capacity at the 8 C rate was 80% of the total capacity and was sustained over 100 cycles. Such findings will help pave the way for the application of conversion reaction electrodes in LIB.

The use of transition-metal oxides as conversion-type electrodes holds the promise of higher energy density and wealth of compounds, but capacity fading needs to be overcome before practical use in LIB; bulk electrodes fail within a few charge/discharge cycles due to the large volumetric change that occurs during lithiation and delithiation. Selective crystallization into specific structures and composites can have a major impact on the performance and cyclability of the conversion-type anode. Nanoscale morphologies have the potential to achieve long cycling lifetimes and good reversibility as stress management and formation of a stable passivation layer during cycling can be achieved.