Recent Development of Advanced Electrode Materials by Atomic Layer Deposition for Electrochemical Energy Storage

Electrode materials play a decisive role in almost all electrochemical energy storage devices, determining their overall performance. Proper selection, design and fabrication of electrode materials have thus been regarded as one of the most critical steps in achieving high electrochemical energy storage performance. As an advanced nanotechnology for thin films and surfaces with conformal interfacial features and well controllable deposition thickness, atomic layer deposition (ALD) has been successfully developed for deposition and surface modification of electrode materials, where there are considerable issues of interfacial and surface chemistry at atomic and nanometer scale. In addition, ALD has shown great potential in construction of novel nanostructured active materials that otherwise can be hardly obtained by other processing techniques, such as those solution‐based processing and chemical vapor deposition (CVD) techniques. This review focuses on the recent development of ALD for the design and delivery of advanced electrode materials in electrochemical energy storage devices, where typical examples will be highlighted and analyzed, and the merits and challenges of ALD for applications in energy storage will also be discussed.


Atomic Layer Deposition
Atomic layer deposition (ALD), which is also historically named as atomic layer epitaxy (ALE), is a vapor-based self-terminating thin fi lm growth technique, which can deliver a conformal coverage of layered materials with well-controlled thickness, in particular on complex surfaces and 3D structures. [ 1 ] It is an effi cient and powerful deposition process that has been developed for deposition of various metals, metal oxides, metal nitrides, metal sulfi des and compound materials. Since 1970s, ALD has steadily been established and commercialized for various thin fi lms and surface coatings in chemical, mechanical and optical engineering as well as in microelectronics, where the best known examples are in electroluminescent displays and advanced highk metal oxides. [ 2,3 ] As ALD can effectively tailor the surface and porous structures of different materials, it has also been widely employed for the surface functionalization of materials for catalysis, fuel cells, batteries, and sensors, Electrode materials play a decisive role in almost all electrochemical energy storage devices, determining their overall performance. Proper selection, design and fabrication of electrode materials have thus been regarded as one of the most critical steps in achieving high electrochemical energy storage performance. As an advanced nanotechnology for thin fi lms and surfaces with conformal interfacial features and well controllable deposition thickness, atomic layer deposition (ALD) has been successfully developed for deposition and surface modifi cation of electrode materials, where there are considerable issues of interfacial and surface chemistry at atomic and nanometer scale. In addition, ALD has shown great potential in construction of novel nanostructured active materials that otherwise can be hardly obtained by other processing techniques, such as those solution-based processing and chemical vapor deposition (CVD) techniques. This review focuses on the recent development of ALD for the design and delivery of advanced electrode materials in electrochemical energy storage devices, where typical examples will be highlighted and analyzed, and the merits and challenges of ALD for applications in energy storage will also be discussed.

Introduction
With the rapid depletion of fossil fuels and ever-increasing demand for clean and sustainable energy sources, development of advanced electrode materials for effi cient energy storage has drawn much attention in recent years. Effi cient electrochemical energy storage devices, including those of high energy density, power density and long device stability are desperately needed for electrical and hybrid vehicles, portable and wearable electronics, as well as large scale energy storage. Atomic layer deposition (ALD) is known to be a non-solution nanotechnology for conformal deposition of nanoscale thin fi lms and surface REVIEW especially since 2000s. [ 4 ] ALD process is based on successive cycles of self-terminating gas-solid surface reactions, where a typical cycle is composed of two or more pulses of precursors. Taking ALD TiO 2 using two precursors (TiCl 4 and H 2 O) as an example, the typical ALD process is schematically illustrated in Figure 1 , where the detailed procedure of one ALD cycle is as follows: (i) a pulse of TiCl 4 precursor on substrate or predeposited fi lm reacts with the surface reactive sites (OH*), thus depositing Ti and introducing TiCl* reactive species; (ii) after the surface-saturated reaction in (i), there is purge of the unreacted TiCl 4 and by-product HCl; (iii) a pulse of H 2 O precursor then reacts with TiCl* species to deposit O and supply OH* reactive sites; (iv) this is followed by purge of the oversupplied H 2 O and by-product HCl, providing a clean surface with OH* species for the following cycles. The two half reactions in (i) and (iii) can thus be written as [ 3,5 ] : Since the uniqueness of the ALD process, as compared with other gas-phase deposition techniques, such as physical vapor deposition (PVD), chemical vapor deposition (CVD), and solution-based deposition methods such as hydrothermal and sol-gel method, it demonstrates several apparent advantages, including well-controllable thickness with high uniformity, excellent conformal deposition, and low temperature growth (normally below 300°C; some materials can be deposited at room temperature), even on complex surfaces and 3D Figure 1. A schematic ALD process for depositing multi-TiO 2 layers using TiCl 4 and water as the precursors on substrate with reactive OH* sites. A typical ALD cycles consists of: i) a pulse of TiCl 4 precursor on substrate or pre-deposited fi lm reacts with the surface reactive sites (OH*) thus deposits Ti and introduce TiCl* reactive species; ii) upon the surface-saturated reaction in (i), there is purge of the unreacted TiCl 4 and by-product HCl, leaving a new intermediate layer; iii) a pulse of H 2 O precursor reacts with TiCl* species to deposit O and supply OH* reactive sites; iv) the purge of oversupplied H 2 O and by product HCl, provides a clean surface with OH* species for the following cycles.
substrates. Figure 2 shows a representative collection of electron microscopy images of various types of selected nanostructures, where ALD was involved in at least one of the fabrication steps. Compared with typical solution-based deposition techniques, ALD relies on vapor-phased surface reaction, thus the conformal deposition can be easily realised even on complex surfaces and 3D substrates. For example, conformal TiO 2 inverse opals has been realized by ALD on templates of PS spheres (Figure 2 i), and ZnO has been conformably coated on 3D polymer templates by ALD (Figure 2 j). Since the amount of material deposition by ALD can be manipulated with the number of cycles, it delivers the desired high uniformity and accurate control in fi lm thickness. For example, TiO 2 and Pt nanotubes of considerable high uniformity have been formed with ALD Al 2 O 3 sacrificial spacer layers (Figure 2 c). As a typical example to show the structural uniformity, ALD Al 2 O 3 /ZnO fi lm was demonstrated with a small surface roughness of ≈0.15 nm with the deposition thickness of ≈62 nm. [ 6 ] Another advantage of ALD is its relatively low deposition temperature. Compared with conventional CVD techniques that are often conducted in the temperature range of 600-1050 °C, [ 7 ] ALD can be run in temperatures below 300 °C. In addition, some materials can be effectively deposited below 100 °C or even at room temperature. [ 8 ] This low temperature merit offers ALD a wide application potential, such as on temperature-sensitive substrates, especially organic nanomaterials and biomaterials. For example, ALD ZnO was successfully operated at 70 °C on poly(methyl methacrylate) (PMMA) template, [ 9 ] which helps to create a high-density periodic ZnO nanopattern. In other reports, ALD Al 2 O 3 and TiO 2 were successfully deposited on biological macromolecules at 35 °C. [ 10 ] Nanocrystalline In 2 O 3 fi lm with high conductivity could be deposited at a low temperatures of 100 °C. [ 11 ] Last but not least, since ALD can offer conformal deposition on complex surfaces and 3D textures, it is of particular interest for use on the very wide range of different substrate types. With such apparent advantage, ALD has recently been investigated for materials in a wide spectrum of chemical, energy and environmental applications, such as catalysis, fuel cells, photovoltaics, batteries, supercapacitors, fi ltration devices, sensors and membranes. There were several review articles that introduced the fundamentals of ALD and its applications in nanotechnology, [ 4,12 ] and a few others on applications of ALD in lithium ion batteries and supercapacitors. [ 13 ] Since the rapid development of ALD in energy applications, it is timely to review the recent progress, and therefore the present paper mainly focuses on the recent advances of ALD in design and fabrication of electrode materials for electrochemical energy storage devices.

Electrochemical Energy Storage
In the presently energy-concerned society, potential energy crisis, globe warming and worsening environment have aroused huge attention to search for generation and storage of clean and sustainable energy at low cost. [ 14 ] Among various energy storage techniques, electrochemical energy storage has been considered as one of the most promising, owing to the high effi ciency, versatility, high mobility, low cost, and fl exibility. There are different types of electrochemical energy storage devices that have been widely explored, for example, including rechargeable batteries and supercapacitors for electrical and hybrid vehicles, emergency power supply, portable electronics and wearable devices. [ 15 ] In principle, electrochemical energy storage devices, such as rechargeable batteries and supercapacitors, keep energy in the format of electricity, which takes place through electrochemical processes by charge and discharge of electrons and ions, such as Li + , Na + , K + , H + and OH − . A typical device of such electrochemical energy storage is composed of an anode, a cathode, and an electrolyte. During the discharge process (energy release), ions will desorb/extract from one electrode and transfer to the other, at the same time the electron thus generated fl ows pass through an external circuit; in the reverse charge (energy storage) process, the device will be under an external potential across the electrodes, thus the reverse electrochemical reactions (ions adsorption/insertion) takes place at the electrodes, and the electrochemical device becomes polarized with energy being stored after the charge process.
Rechargeable batteries can be further categorized as leadacid, metal-air, nickel-iron, lithium-sulfur, sodium-ion, and Li-ion batteries (LIBs). [ 15 ] Supercapacitors can be further classifi ed into electric double layer capacitors (EDLCs) and pseudocapacitors. The electrochemical energy storage performance of both rechargeable batteries and supercapacitors is essentially determined by the electrode materials. [ 15,16 ] Even though there have been considerable investigation effects that are devoted to the design, selection and fabrication of advanced electrode materials, many challenges still exist for development of the next-generation electrochemical energy storage devices. [ 14,17,18 ] For example, as an anode in lithium ion batteries, carbonaceous materials can only provide limited capacities. Alloyed metal oxides and Si-based materials suffer from huge volume changes, which result in poor cycling ability. Intercalation metal oxides usually show poor conductivity and low capacity. On the other hand, lithium ion battery cathode materials often suffer from limited rate performance and the side reaction with electrolyte. In Li-S batteries, the insulating sulfur and Li 2 S give rise to poor rate capability. The intermediate product of polysulfi des can dissolve in electrolyte and damage the Li metal surface. In addition, the sulfur electrode undergoing huge volume expansion (≈76%) limits the cycling life. For application as supercapacitor electrode, carbonaceous materials have relatively low capacitance, while the metal oxides/hydroxides with low electric conductivity and low surface area usually show poor rate and cycling ability.

Conformal Surface Protection for Enhanced Rate and/or Cycling Ability
Compared with other deposition techniques, such as the solution based sol-gel coating and hydrothermal reaction, the materials developed by ALD can offer much better conformal surfaces coating on electrodes and bring different surface chemical reactions, which serve as effective protection layers thus leading to better cycling ability. The ALD coating layers can be accurately controlled down to atomic level, which can benefi t better rate capability. Detailed discussions on ALD surface protection of electrodes will be presented in Section 2.1, where selected examples will be given.

Controllable Deposition for Optimised Power and Energy Density
The thickness/amount of active materials strongly infl uence the ion diffusion length and charge transfer path in electrochemical reactions, where in general a thicker layer of active material increases the energy density of the electrode but sacrifi ce the power density. Thus control in the amount of active material is essential in order to optimize the power and energy density. The thickness of active material by ALD can be easily manipulated by the number of cycles applied, which can therefore offer well-tunable deposition of active materials with optimized overall performance. In addition, electrodes with a thin ALD layer offers the unique benefi t for investigation and understanding of the charge transport and storage mechanism of the active materials. A detailed discussion on ALD of active materials for optimized electrochemical performance will be presented in Section 2.2. compared with those nanostructures obtained by other techniques, such as PVD and CVD. For example, the thin and uniform ALD layer can be ion-exchanged leading to the formation of different active materials, while the structural uniformity is preserved. Examples of utilizing ALD in the development of unique electrode nanostructures will be further discussed in Section 2.3.

Utilizing ALD for Advanced Electrode Materials
As key components in almost all electrochemical energy storage devices, electrode materials are playing a determining role in the overall device performance. In this section, discussion will be focused on the development of ALD for advanced electrode materials. On the basis of the unique functions by ALD, the following three aspects will be presented: ALD for surface modifi cation of electrodes; ALD for active materials; and ALD for construction of novel nanostructures. Typical examples with different ALD functions and the accordingly electrochemical performance are summarized in Table 1 , Table 2 , and Table 3 .

ALD for Surface Modifi cation on Electrodes
The structure and surface chemistry of electrodes greatly infl uence the electrochemical performance in energy storage devices. This section will focus on ALD surface modifi cation of different electrode materials, including lithium-ion battery electrodes, supercapacitor electrodes, and electrodes for other energy storage devices, such as Na-ion battery, Li-S and Li-O 2 batteries. The unique function of ALD surface modifi cation and the resultant contribution in enhancement of electrochemical performance will be highlighted.

ALD Surface Modifi cation for Lithium-Ion Battery Cathodes
The direct contact of cathode materials with electrolytes can bring about side reactions that may cause slow degradation of electrodes, thus proper surface modifi cation of the cathode materials have been conducted for improvement in electrochemical performance. [ 19 ] The materials studied for surface modifi cation include various metal oxides, phosphates and fl uorides. [ 20 ] These materials can be used as a physical protective layer or HF scavenger with improved ionic conductivity, thus much improved rate capability and cycling ability is achieved. [ 19 ] Compared with traditional wet chemistry processes, such as sol-gel process for surface modifi cation, ALD offers much uniform surface coverage on the electrode materials with well controlled thickness down to sub-nanometers scales.
Jung et al. have demonstrated that direct ALD Al 2 O 3 coating on micro-powdered LiCoO 2 electrode surfaces effectively protects the active material while maintaining an inter-particle electronic pathway for high rate capability. [ 21 ] It is believed that the ALD Al 2 O 3 acts as an "artifi cial" solid electrolyte interphase (SEI), thus protects the inner active material from side reaction. However, if ALD Al 2 O 3 was fi rst deposited on LiCoO 2 powder and then applied into an electrode, the electron conduction paths would be blocked thus the capacity would decrease rapidly. Scott et al. reported that the LIB performance of nano LiCoO 2 powder-based electrode could be signifi cantly improved by ALD Al 2 O 3 surface coating. [ 22 ] As shown in Figure 3 a-b, 6 cycles of ALD Al 2 O 3 with thickness of ≈1-2 nm gives rise to a uniform coverage on LiCoO 2 powder particles. The ALD surface-modifi ed electrode maintained 100% capacity after 200 charge-discharge cycles at 2.8 C, in contrast to the capacity of bared LiCoO 2 that had dropped to almost zero after the same number of cycles. The ALD surface coating also largely improves the rate capability of the LiCoO 2 electrode. For example, two ALD cycles on Al 2 O 3 coated electrode demonstrates a capacity of 133 mAh g −1 at 7.8 C, corresponding to about 250% improvement as compared to the bare LiCoO 2 electrode. The uniform assembling of the thin ALD Al 2 O 3 layer has effectively protected the electrode and maintained the high rate capability.
Other cathode materials, such as LiNi 0.5 Mn 1.5 O 4 , [23][24][25] [ 28 ] LiMn 2 O 4 , [ 29 ] have also shown much improvement with ALD Al 2 O 3 surface modifi cation. The ALD Al 2 O 3 coating layer serves as an "artificial" SEI layer that suppresses the side reaction at high voltage thus improve the overall cathode performance.
Since the insulating nature of Al 2 O 3 that retards Li + diffusion process with low rate performance, other metal oxides, such as ALD LiAlO 2 , ZnO, ZrO 2, FePO 4 and TiO 2 , have been investigated as surface modifi cation on cathode materials. [30][31][32][33][34][35] For example, as shown in Figure 3 c, [ 30 ] Park et al. have investigated ALD LiAlO 2 surface coating on LiNi 0.5 Mn 1.5 O 4 electrode. Since the ionic conductivity of LiAlO 2 is better than that of Al 2 O 3 , LiAlO 2 surface-modifi ed electrode shows much improved rate capability and cycling ability when compared to Al 2 O 3 coated electrodes. Using a commercial LiCoO 2 electrode, Sun's group conducted a systematic investigation to compare the effects of ALD coating layers of TiO 2 , ZrO 2 and Al 2 O 3 . [ 36 ] Their results suggest that 2 ALD-cycle coatings gives rise to the best improvement when compared with thicker coatings, and Al 2 O 3 leads to the best cycling ability while ZrO 2 shows the highest rate capability. Another interesting work from Liang's group reported an ALD CeO 2 modifi ed LiMn 2 O 4 electrode, in which the pin-hole free layer of CeO 2 could not only protect the electrode from side reaction but also acted as an electronic path. [ 37 ] Compared with Al 2 O 3 -and ZrO 2 -surface coated LiMn 2 O 4 , the electrode with CeO 2 surface modifi cation provides much improved rate capability at both room temperature and 55 °C. In addition, as shown in Figure 3 d-f, CeO 2 modifi ed LiMn 2 O 4 shows a stable cycling ability with a high capacity of 78 mAh g −1 being maintained even after 1000 cycles, which is also much improved compared with those of Al 2 O 3 or ZrO 2 ALD surface-coated electrodes. The Li + conductivity of ALD CeO 2 ultrathin fi lm effectively overcomes the tradeoff between Li + diffusion and the "artifi cial" SEI protection.

ALD Surface Modifi cation For Lithium-Ion Battery Anodes
As a practically valuable LIB anode material, graphite exhibits a theoretical capacity of 372 mAh g −1 . However SEI layer can be formed on the graphite surface even during the fi rst cycle,   [ 97 ] 4. SnO 2 Nano Ni foam Li-ion battery Initial discharge capacity of 546 mAh g −1 , and 505 mAh g −1 after 100 cycles at 500 mA g −1 [ 98 ] 5. V 2 O 5 AAO + ALD Ru Li-ion battery 50% of 1C capacity maintained at 150 C; 80% of initial capacity retained after 1 000 cycles at 5 to 25 C.
[ 107 ] 9. RuO x Carbon nanotube Supercapacitor 644 F/g with a high power density of 17 kW kg −1 [ 109 ] 10. NiO Nanoporous graphene Supercapacitor ≈1897.1 F/g and cycling for 1500 cycles. [ 111 ] which brings potential safety issue, especially when the cell operation temperature is high. [ 38 ] To protect graphite and reduce the SEI, surface modifi cation works have been conducted with natural graphite to improve the cycling ability and high temperature safety. For example, Jung et al. have discovered that ALD Al 2 O 3 on the surface of natural graphite could signifi cantly improve its cycling ability and safety at high temperatures. [ 21 ] It is demonstrated that the graphite electrode with 5 cycles of ALD Al 2 O 3 maintains 98% of the initial capacity after 200 cycles of charge-discharge at 50 °C, while the capacity of the bare graphite electrode drops to only 26% under the same test condition. An interesting work from Wang's group observed the infl uence of ALD TiO 2 surface coating on natural graphite, [ 39 ] where they found that the ALD TiO 2 served as artifi cial SEI to improve the stability of graphite electrode at the high temperature of 55 °C. For LIB anodes, metal oxides have been considered as promising candidates for the next-generation electrode materials, because of their generally large capacities and safety factor. [ 40 ] Alloyed anode materials, such as Si and Sn-based materials (such as SnO 2 and Sn) have been investigated for lithium storage, due to their high capacities and relatively low onset potentials. [ 41,42 ] In the electrochemical process, however, these alloyed anodes undergo severe pulverization with huge volume expansion. When continuous SEI layers are formed, the cycling ability becomes very poor. [ 42 ] Proper surface modifi cation has thus been pursued to help solve the problem. As has been reported by He et al., an ALD Al 2 O 3 coating layer on pattern Si column electrode not only prevents side reaction but also enhances the mechanic strength during the electrochemical reaction. [ 43 ] The ALD Al 2 O 3 -surface coated electrode also shows a higher Coulombic effi ciency (remains above 99.5% after few cycles) than that of the uncoated one. ALD Al 2 O 3 coated SnO 2 has also been reported, where ALD Al 2 O 3 can well serve as "artifi cial" SEI to protect side reaction from SnO 2 and electrolyte. [ 44 ] In addition, the cycling ability of the electrode is closely related to the size of SnO 2 nanoparticles and the thickness of the protective layer. Since the volume expansion of SnO 2 nanoparticles occurs during the cycling process, an protective ALD Al 2 O 3 layer of appropriate thickness buffers the stress and strain thus to maintain the capacity. Further work on ALD surface coating on Si thin fi lm [ 45 ] and Si nanowires [ 46 ] also demonstrates much improved cycling performance after the surface modifi cation.
ALD surface coating helps improve on conversional and intercalation metal oxides, such as Fe 2 O 3 , MoO 3 and Li 4 Ti 5 O 12 . [47][48][49] As has been reported by Lipson et al., a relatively thick and rough SEI is formed on the surface of the uncoated MnO anode, while ALD Al 2 O 3 coated MnO (with a thickness of 3 Å) can effectively prevent the formation of such SEI and maintain the capacity for more than 100 cycles. [ Table 3. Summary of utilizing ALD for the construction of advanced nanostructured electrodes.

ALD Materials
Functions of ALD Application Electrochemical performance Ref.
Ru As a 3D current collector for ALD V 2 O 5 Li-ion battery 50% of 1C capacity maintained at 150 C; 80% of initial capacity retained after 1 000 cycles at 5 to 25 C.
[ 100 ] 3.  [ 49 ] in which the ultrathin ZrO 2 layer is shown to prevent SEI for high stability and extend the voltage window of Li 4 Ti 5 O 12 for high energy density.
There is no doubt that ALD surface modifi cation on LIB electrodes can effectively suppress the side reactions, alleviate the stress and strain of electrodes, and prevent the decomposition of SEI especially at elevated temperatures. The surface reactions occurring in the anode and cathode are highly related, [ 51 ] thus the LIB performance can be much improved by ALD surface coating on separator and both electrodes. [ 52 ]

ALD Surface Modifi cation for Supercapacitor Electrodes
Supercapacitors is another important class of widely investigated and employed electrochemical energy storage devices, which store charges either through absorption/desorption of ions by forming electric double layers (largely based on carbonaceous electrodes) or through fast reversible surface/ near-surface Faradic reactions (based on certain transition metal oxides/hydroxides and conducting polymers) or both. [ 16 ] ALD surface coating technique on LIBs has been extended to supercapacitor electrodes, although the requirement for supercapacitors is quite different from that for LIBs. The performance of supercapacitors is largely relied on the surface/near surface reactions, thus in order to enhance the electrochemical performance, the coating materials should be thin and electrically conductive for fast surface reactions, and they preferably should be highly active and stable for electrochemical energy storage.
Carbonaceous materials have been widely employed for supercapacitor electrodes, owing to several merits such as high electric conductivity, tunable high surface area, low cost, and high electrochemical stability. [ 53 ] However, they exhibit relatively low capacitance and therefore low energy density. In order to enhance the energy density of carbon-based materials, one approach is to expand their working voltage windows. However, carbon-based materials are not stable at high-voltages. To solve this problem, ALD Al 2 O 3 layer has been investigated for coating on the active carbon surface, as has been reported by Hong et al. [ 54 ] When tested in an organic electrolyte, the ALD coated electrode showed excellent stability with a 3 V voltage window, in which the energy density was 39% higher than that for 2.5 V voltage window. The ALD surface coating protects the surface functional groups and prevents the degradation of electrolyte. Another approach to enhance the energy density of carbonaceous materials is to deposit thin layers of anode materials (such as V 2 O 5 , Fe 2 O 3 , Bi 2 O 3 ) on carbon surface, where the active material contributes additional capacitance for high energy density, but does not sacrifi ce much of the power density and cycling ability. For example, Boukhalfa et al. have studied ALD V 2 O 5 coating on CNT electrodes, one sample of which is shown in Figure 4 a,b. [ 55 ] The rather uniformly surface-coated   [ 22 ] Copyright 2010, American Chemical Society. c)Reproduced with permission. [ 30 ] Copyright 2014, American Chemical Society. d,e,f) Reproduced with permission. [ 37 ] V 2 O 5 provides more Faradic reaction for higher capacitance, and the CNTs ensure good electric conductivity for fast charge transfer. CNTs coated with 100-ALD-cycle V 2 O 5 delivers a specifi c capacitance of 1400 F g −1 at 5 mV s −1 , which is signifi cantly improved when compared with that of the bare CNT electrode (≈30 F g −1 ). The V 2 O 5 surface-coated CNTs show a capacitance of above 360 F g −1 , even when the current density is increased to 20 A g −1 , demonstrating its high rate capability. On the other hand, too thick a V 2 O 5 coating layer will block the active reaction sites and electron transport paths, and therefore leading to degraded electrochemical performance. Thus an optimized ALD thickness of V 2 O 5 coating layer is needed. Wang's group has developed an anode material of ALD Fe 2 O 3 deposited on hierarchical carbon support, one example of which is shown in Figure 4 c-e. [ 56 ] Different from the above mentioned ALD V 2 O 5 which exhibited a 2D growth mode, ALD Fe 2 O 3 was grown on CNTs under an island growth mode, where small nanoparticles of Fe 2 O 3 were assembled rather uniformly on CNTs. When tested as the anode material in supercapacitors, the C@Fe 2 O 3 delivered much improved capacitance than that of pure carbon substrate. For example, at the same current density of 20 mA cm −2 mA cm −2 , the carbon electrode coated with 400-ALD-cycle of Fe 2 O 3 showed an areal capacitance of 470.5 mF cm −2 , which is much larger than that of the bare carbon (93.8 mF cm −2 ). However, since the insulating nature of Fe 2 O 3 , too thick a coating layer would degrade the rate performance of electrodes. A proper control in ALD coating thickness is thus essential for optimized performance. V 2 O 5 and Fe 2 O 3 have been investigated as active anode materials, which are believed to help improve the performance of carbon anodes. Cathode materials (such as   [ 55 ] Copyright 2012, Royal Society of Chemistry. c,d,e) Reproduced with permission. [ 56 ] Copyright 2015, American Chemical Society. f,g,h) Reproduced with permission. [ 59 ] NiO, Co 3 O 4 ) can also be deposited on carbon support for supercapacitor applications. The carbon support contribute little capacitance in cathode performance. Further discussion will be given in Section 2.2.2, detailing ALD active materials on conductive support.
Transition metal oxides/hydroxides and conducting polymers store charges with surface Faradic reactions, when employed as supercapacitor electrode materials. [ 57,58 ] They generally provide much higher capacitances than those of the carbon-based materials, although their cycling ability is poor. To improve the stability and capacitance of conducting polymers, Xia et al. deposited ALD RuO 2 layer on polyaniline (PANI) nanowire surface, where a PANI-RuO 2 core@shell nanostructure was formed, as shown in Figure 4 f-h. [ 59 ] The capacitance, rate capability and cycling stability of PANI were all improved. For example, PANI coated with 100-ALD-cycle RuO 2 showed a specifi c capacitance of 710 F/g at 5 mV s −1 , which was much higher than that of PANI alone (564 F g −1 ). More importantly, 88% of its capacitance was maintained after 10 000 cycles at 20 A g −1 , while it was only 65% being maintained for PANI alone after the same number of cycles. The authors observed that both the capacitance and rate capability of PANI-RuO 2 could be largely reduced with too thick RuO 2 coatings. Thus an optimized thickness of ALD RuO 2 is needed to balance the overall performance including the capacitance, rate and cycling behavior.

ALD for Surface Modifi cation of other Electrochemical Energy Storage Devices
Na-ion batteries (SIBs) have redrawn considerable attention in recent years, by considering the abundance and low cost of Na and their promise for large-scale storage applications. [ 60 ] Although they share similar fundamental principles as LIBs, the Na-intercalation chemistry and surface modifi cation have not been fully explored. Nevertheless some studies have been made with sodium-based compounds. [ 61 ] ALD surface coatings on certain SIB electrodes have been shown to effectively improve the device performance. For example, ALD Al 2 O 3 coating was able to enhance the cycling ability of anode material of Na 2 C 8 H 4 O 4 . [ 62 ] As shown in Figure 5 a-b, Han et al. demonstrated that the cycling ability of Sn nanoparticles (as anode for SIB) was signifi cantly improved with an ALD Al 2 O 3 coating. [ 63 ] With the help of in situ transmission electron microscopy (TEM), the dynamic mechanical protection of ALD Al 2 O 3 coating was clearly revealed. A unique Na−Al−O layer was formed during the reaction of Al 2 O 3 and Na ions, which acted not only as a mechanical protection for the Sn inside, but also an ion transport channel for improved Na ions diffusion. The work by Jung et al., who used dynamics calculations, suggested that Na ion diffusivity in Na x Al 2 O 3 could be much higher than the Li ion diffusivity in Li x Al 2 O 3 , [ 64 ] thus the infl uence of ALD Al 2 O 3 in SIB was quite different from that in LIB. The high diffusivity of Na ions in Al 2 O 3 might therefore bring better electrochemical properties for ALD coated electrodes in SIBs.
Due to the high specifi c energy density in theory and abundance in starting materials, other battery systems, such as Li-S and Li-O 2 have received considerable attention recently. [ 18,65,66 ] For example, the theoretical capacity of S cathode is 1673 mAh g −1 , which is much larger than some of those currently known cathodes of metal oxides in LIBs (400 mAh g −1 ). [ 67 ] The theoretical energy density of Li-O 2 battery is as high as 3505 Wh Kg −1 , while commercial LIBs can only reach a fraction of this theoretical value. [ 18 ] In Li-S battery, because of the insulating nature of S and the dissolution of polysulfi de, the rate and cycling ability are usually rather poor. [ 68 ] Researchers have demonstrated that coupling S or Li 2 S with conductive carbon is an effective pathway to improve the Li-S battery performance, where proper surface coating by ALD contributes to improved electrochemical properties. [ 69 ] ALD Al 2 O 3 surface coating has been reported by Kim et al., who developed an electrode with sulfur-infi ltrated activated carbon fi bers (S-ACFs). As shown in Figure 5 c-d, [ 70 ] a signifi cant enhancement in cycling ability was demonstrated. The ALD Al 2 O 3 coated electrode retained a high specifi c capacity of 600 mAh g −1 after 300 charge-discharge cycles at 0.2 C, while the capacity of the uncoated electrode dropped to almost zero. The much improved cycling ability was attributed to the ALD layer, which effectively confi ned the polysulfi des inside within the barrier thus reduced S dissolution. The reduction in S dissolution has been further confi rmed by ex situ SEM and EDS mapping experiment of the S-ACFs electrode and the lithium anode. Other works on ALD surface coating on carbon/S electrodes [ 71,72 ] have also demonstrated much improved cycling ability, in which ALD surface coating is generally utilized as an artifi cial SEI layer and a barrier layer to prevent the S loss. [ 73 ] For example, ALD Al 2 O 3 coated on carbon cloth was successfully assembled between S electrode and the separator in the Li-S system. [ 74 ] For the cell made with ALD-Al 2 O 3 coated on carbon cloth, the initial discharge capacity was demonstrated 25% higher than the one without ALD coating. 70% of the capacity could be maintained after 40 cycles, which was also much improved compared with the one without ALD coating. SEM and EDX results further confi rmed that the coating of ALD-Al 2 O 3 on carbon cloth could effectively adsorb and reactivate the dissolved polysulfi des from electrolytes.
As another promising type of battery system for energy storage, Li-O 2 batteries are still in the infancy of their development. There are several scientifi c and technological challenges in this complex system that need to be addressed in detail, [ 75 ] including for example, the severe passivation and corrosion of anode with low Coulombic effi ciency. The electrolyte should be stabilized and have suffi cient Li + conductivity and O 2 solubility. Cathode materials with proper pore structures (e.g., pore size and distribution) and effi cient catalysis are needed in order to prevent electrical passivation from discharge products. [ 65,76 ] As an unique surface modifi cation process for electrodes of LIBs, ALD surface coating has been conducted with cathode materials of Li-O 2 battery. [ 77 ] Utilizing ALD Al 2 O 3 and Pd, Lu et al. developed a novel cathode material of palladium nanoparticles and thin alumina layer coated on carbon. [ 78 ] As shown in Figure 5 e-f, the ALD alumina layer effectively protects the carbon surface and prevents the decomposition of electrolytes, while the nanosized Pd electrocatalyst promotes the formation of nanocrystalline discharge product of Li 2 O 2 , which is benefi cial for charge transport. The Pd and Al 2 O 3 surface-coated carbon gives rise to a very small charge over-potential of ≈0.2 V. Wang's group have constructed a hierarchical structure of Li-O 2 battery cathode, in which three dimensionally ordered www.MaterialsViews.com www.advancedscience.com Adv. Sci. 2016, 3, 1500405 mesoporous carbon support was protected by an ALD FeO x layer, followed by ALD Pd nanoparticles. [ 79 ] In the structure, the carbon support localizes the Li 2 O 2 deposition, FeO x protects the inner carbon and catalyses the decomposition of Li 2 O 2 , and Pd nanoparticles serve as an oxygen reduction reaction (ORR) catalyst. With the ALD coating of FeO x and Pd, the cycling ability of the carbon support is improved from 16 to 68 cycles. Liu et al. also developed Pd-coated, ZnO passivated carbon cathode for Li-O 2 battery, in which ZnO serves as a protective layer for inner carbon and Pd acts as an effective catalyst for oxygen evolution reaction. [ 80 ] Other effi cient catalysts developed by ALD, such as Ru, Pt and Pt 3 Co, [ 81 ] have also been reported for Li-O 2 battery, with enhanced electrochemical performance.

ALD of Active Materials
Direct deposition of active materials by ALD has led to the development of effi cient electrode structures on different substrates. The studies also help understand the fundamental reaction/energy storage mechanisms of the active materials involved. This section will look at ALD active materials developed for improved energy storage, where their merits (such as the tunable thickness, uniformity and conformal surface, and interface with complex substrates) will be discussed. The relationships between the textures of resultant ALD active materials and their performance in electrochemical devices will also be reviewed.   [ 63 ] Copyright 2013, American Chemical Society. c,d) Reproduced with permission. [ 70 ] e,f) Reproduced with permission. [ 78 ] Copyright 2013, Nature Publishing Group.
There are several unique and benefi cial features of the ultrathin layer materials grown by ALD on 3D complex substrates for electrochemical energy storage. Indeed, several ALD cathode and anode materials for LIBs have been studied, for example including FePO 4 , [ 82 ] Li x Mn 2 O 4 and Li x V 2 O 5 , [ 83 ] LiCoO 2 , [ 84 ] V 2 O 5 , [ 85 ] Co 3 O 4 , [ 86 ] RuO 2 , [ 87 ] SnO 2 , [ 88 ] and TiO 2 . [ 89,90 ] Active materials of metal sulfi des, [ 91 ] such as Cu 2 S, [ 92 ] GaS x , [ 93 ] have also been developed. Similar to metal oxides, ALD metal sulfi des utilize metalorganic precursors for metal sources. The difference is however hydrogen sulfi de (H 2 S) being used as the typical sulfi de source for sulfi des, while H 2 O, O 2 plasma and O 3 are commonly employed as the oxygen sources for metal oxides. ALD has also been utilized for the deposition of active materials for supercapacitors [ 94 ] and Li-S batteries. [ 95 ] Since the thickness of thus deposited active materials can be well controlled down to atomic layers, the studies help understand the key charge transfer and energy storage mechanisms (such as ion diffusion, absorption as well as redox processes involved) in these active materials.
Østreng et al. have investigated ALD V 2 O 5 as the cathode material for LIBs by using VO(thd) 2 and ozone as the precursors. [ 96 ] As shown in Figure 6 a-f, ALD V 2 O 5 showed a highly textured surface with lots of platelets, which apparently offer large surface area for effi cient electrode-electrolyte contact. In characterization of electrochemical behavior, the electrode with 500-cycle ALD V 2 O 5 is discharged with rates up to 960 C, while maintaining 20% of the initial 1 C capacity. It depicted excellent cycling ability, which maintained more than 80% of the initial capacity at more than 1500 cycles at a discharge rate of 120 C. The work has successfully led to an electrode with high cycling stability and high power density, where the confi ned nanosized V 2 O 5 and the direct contact with the current collector contribute to the excellent high power and long-time cycling ability.
In order to develop Li 2 S in a more tunable and controllable manner, which is a promising cathode for Li-S battery, Meng et al. have successfully synthesised amorphous Li 2 S by ALD using lithium tertbutoxide (LTB, LiOC(CH 3 ) 3 ) and hydrogen sulfi de (H 2 S) as the precursors. [ 95 ] As shown in Figure 6 g-i, the resultant Li 2 S is uniformly deposited on the high aspect ratio silicon trench, and the growth rate of Li 2 S is ≈1.1 Å per cycle in the temperature range of 150-300 °C. In the electrochemical characterization, the 700-ALD-cycle Li 2 S fi lm was showed with excellent cycling ability for 500 cycles at a high current of 840 mA g −1 . It also demonstrated a high Coulombic effi ciency without the help of the normally used electrolyte additives. Since the conformal nature, ALD is a promising technique for depositing active materials on complex surfaces and 3D substrates, as has been demonstrated with 3D conductive metal substrates and hierarchical carbon supports.

ALD Active Materials on 3D Conductive Metal Nanostructures
As shown in Figure 7 a- (500, 1000, 2000, and 5000) on Si substrate. e) Simulation of a surface equivalent to the sample deposited using 5000 cycles. f) Rate performance of 500 cycles of ALD V 2 O 5 electrode between 1 and 960 C, the capacity retention is normalized to the capacity at 1 C. g-h) SEM images of ALD Li 2 S on Si substrate. i) Cycling ability and Coulombic effi ciency of the Li 2 S electrode with 500 cycles. a-f) Reproduced with permission. [ 96 ] Copyright 2014, Royal Society of Chemistry. g-i) Reproduced with permission. [ 95 ] Copyright 2014, American Chemical Society.
the pore size and thickness of the active materials can be well controlled for optimised Li + diffusion length in the electrolyte and solid state active materials. [ 97 ] A high power density of 13 KW Kg −1 with high energy density of 130 Wh Kg −1 is achieved with the sample of 225 nm in pore size and 2 nm TiO 2 surface coating. The large pore size is essential for Li + diffusion in electrolytes at high rates and the thin layer of TiO 2 benefi ts the short Li + diffusion length. Haag et al. successfully fabricated 3D nanostructures of ALD SnO 2 conformally coated on Ni nanofoam, one example of which is shown in Figure 7 c-d. [ 98 ] The 3D Ni nanofoam substrate with high surface area and high conductivity is shown to contribute more sites for fast electrochemical reaction, and the ultrathin layer of ALD SnO 2 effectively buffers the volume change for long-time cycling ability. The Ni nanofoam ALD coated with 8 nm SnO 2 expressed an initial discharge capacity of 546 mAh g −1 . It could maintain 505 mAh g −1 after 100 cycles at a current of 500 mA g −1 , showing the excellent cycling ability. Other ALD coated 3D metal substrates, such as Ni/TiO 2 , Al/TiO 2 , Ni/V 2 O 5 , have also demonstrated the combined merits from conductive substrate and the conformal coating of thin layers of active materials. [ 99 ] Using anodic aluminum oxide (AAO) for developing nanopores, Liu et al. constructed a novel all-in-one battery with parallel nanotubular arrays as electrodes, where the liquid electrolyte is confi ned within AAO nanopores. As shown in Figure 7 e-g, [ 100 ] ALD Ru could be used as nanotube current collectors and ALD V 2 O 5 served as active material for energy storage. With the elegant device design, the thin layer of V 2 O 5 was fully exposed to the electrolyte, and the confi ned connection with nano-sized Ru current collector drastically facilitated the fast ion and electron transports. In electrochemical characterization, the device was demonstrated with an excellent rate capability with ≈50% of capacity (relative to 1 C) maintained at the 150 C (24 s charge-discharge time), and good cycling ability, with more than 80% of initial capacity being retained after 1 000 cycles at 5 to 25 C.
In general, much improved electrochemical energy storage performance has been demonstrated by ALD active materials deposited on 3D conductive substrates, [ 101 ] where the substrate provides the desired electrical conductivity with high specifi c surface area, and the ultrathin layer of active materials by ALD enables short ion diffusion length, which benefi ts the fast electrochemical reaction. The ALD active materials thus deposited on conducting supports are therefore very promising candidates for electrochemical energy devices, for example, as has been demonstrated with 3D allsolid-state micro-batteries. [ 102 ]

ALD Active Materials on Carbonaceous Materials
As a typical class of electrode materials for electrochemical energy storage, carbonaceous materials exhibit high electrical conductivity and tunable high surface area, which are essential for high rate performance and high power density. On the other hand, transition metal oxides/hydroxides of high capacity/ capacitance are benefi cial for high energy density. Therefore a proper construction of hybrid structures combining carbon materials with metal oxides/hydroxides takes the advantages of each constituent components and their synergetic efforts. [ 57,103 ] Considering the conformal and controllable ALD features down to atomic scales, decorating ALD active materials on carbonaceous substrate is of particular interest for much improved device performance. [ 53 ] ALD active materials on conductive carbon support gives rise to shortened ion diffusion length thus benefi ting the high power. Together with the well tunable www.MaterialsViews.com www.advancedscience.com Adv. Sci. 2016, 3, 1500405   Figure 7. ALD active materials on 3D current collectors for electrochemical energy storage. a) Schematic of transport process of 3D nanoporous gold coated with ALD TiO 2 electrodes for lithium ion batteries. b) SEM image of the ALD TiO 2 coated 3D gold. c) SEM and d) TEM images of ALD SnO 2 coated Ni nanofoam. e) Schematic of parallel nanopore battery array by ALD V 2 O 5 and Ru on AAO template. f,g) SEM images of AAO coated with ALD Ru (f), and ALD Ru and V 2 O 5 (g). a,b) Reproduced with permission. [ 97 ] Copyright 2015, American Chemical Society. c,d)Reproduced with permission. [ 98 ] e-g) Reproduced with permission. [ 100 ] Copyright 2014, Nature Publishing Group.
thickness/mass of active materials, they lead to much improved overall electrochemical performance.
ALD with the quaternary cathode material of LiFePO 4 on conducting substrates, such as CNTs, has been successfully developed, by using 5 different precursors and carefully tailoring their surface reactions, as shown in Figure 8 a-c. [ 104 ] The LiFePO 4 /CNTs electrode thus derived demonstrated a high discharge capacity of ≈150 mAh g −1 at 0.1 C. In addition, it also expressed excellent rate property, where a high capacity of 71 mAh g −1 was maintained when the current was increased to 60 C. The cell could retrieve the capacity after the current being cycled from 0.1 to 60 C. A key parameter in evaluating the electrode performance is the cycling ability, where the LiFePO 4 / CNTs shows a high capacity retention of 120 mAh g −1 after 2000 cycles at 170 mA g −1 . The excellent rate and cycling ability of LiFePO 4 /CNTs was derived from the unique combination of carbon nanostructure and ALD layer.
Another interesting feature of carbonaceous materials is their mechanical robustness and fl exibility. They can be made into 3D supports with fl exibility. Using 3D fl exible supports made of carbon cloth decorated with carbon nanowire array, Wang et al. further decorated the carbon surface with a thin layer of ALD TiO 2 . It was studied as an anode for LIBs. As shown in Figure 8 d-e, [ 105 ] compared with planer substrates, ALD gives rise to a much higher (≈300 times) mass loading of TiO 2 on the hierarchical 3D carbon cloth/carbon nanofi ber substrate, which is essentially desirable for high volume/areal capacity. With the unique 3D design and ALD deposition, the C/TiO 2 demonstrated a high discharge capacity of 309 mAh g −1 at 0.2C with a capacity of 100 mAh g −1 being maintained at 20 C. Excellent cycling ability was also achieved for the C/TiO 2 , where a high reversible capacity of 170 mAh g −1 could be maintained even after 8000 cycles at 10 C. The excellent electrochemical behavior is believed to originate from the conformal thin layer of ALD TiO 2 , which provides a short ionic diffusion length, and the 3D conductive carbon support of high surface area.
There are several other types of 3D nanostructures of ALD active materials deposited on carbon supports that have been reported, [ 106 ] for example, including the CNT sponge/ V 2 O 5 , [ 107,108 ] CNT/Ru, [ 87,109 ] gaphene foam/ZnO, [ 110 ] and porous graphene/NiO. [ 111 ] Because of the well-controlled ALD thin layer of active materials (desirable for high energy density) and the high electrical conductivity and high surface area of the carbon support (desirable for high power density), they have been shown very promising for achieving high performance energy storage devices.

ALD Parameters and their Functions on Electrochemical Performance
ALD has been utilized in tuning the deposited materials for different crystallinities, textures and mass distribution, which in turn impact the electrochemical properties. [ 84 ] For example,   [ 104 ] d,e) Reproduced with permission. [ 105 ] Copyright 2015, Royal Society of Chemistry.
by tuning the deposition temperature, either amorphous or crystallized SnO 2 nanoparticles can be deposited on graphene nanosheets. Amorphous SnO 2 showed an apparently better cycling ability since it could better buffer the volume change than the crystallized nanoparticles. [ 112 ] The amount of ALD active materials can well be controlled with different number of deposition cycles, which infl uence the electrochemical performance. For example, the capacitance of PANI nanowires is largely increased when ALD coated with 100 cycles of RuO 2 , while a surface coating of 1500 ALD cycles coating results in a drastic drop in capacitance (Figure 4 h). [ 59 ] Tuning the type of precursors in the 2 nd half reaction strongly infl uences the nature of ALD materials and electrochemical behavior. For example, by changing the plasma precursor from O 3 to NH 3 , Mattelaer et al. observed a variation between MnO 2 and MnO, which were employed as cathode and anode for LIB, respectively. [ 113 ] Last but not least, since ALD is based on the surface-determined chemical reactions, surface properties of substrates greatly infl uence the morphologies and properties of ALD materials. [ 114,115 ] For example, unmodifi ed carbon surface (e.g., single-walled carbon nanotubes or graphene) are generally chemically inert to ALD precursor molecules, thus active materials grow at the defect sites. [ 116 ] By proper surface functionalization (such as by conjugating -NO 2 or -OH groups) on carbon surface, rather uniformly deposited ALD materials are obtained. The substrate surface properties have also shown great infl uence on electrochemical behavior and other functional behavior such as catalysis and sensing, [ 115,117 ] and therefore they impact on the device performance. To develop a proper continuous layer of ALD TiO 2 on graphene, ALD Al 2 O 3 has to be deposited fi rst on graphene as a buffer layer, otherwise only isolated nanoparticles of TiO 2 are formed with little capacity. [ 118 ]

Construction of Advanced Nanostructured Electrodes by ALD
As has been discussed above, ALD has been successfully developed for surface modifi cation of electrodes and deposition of high quality active materials for energy storage, both of which are useful in the rational design and fabrication of electrodes for electrochemical energy storage. In addition, as will be discussed in this section, ALD has been used in development of several new nanostructures, which are otherwise diffi cult to achieve by conventional processing techniques, including for example 3D types, core@shell types, and hollowed nanostructures. ALD precursors can also be successfully exchanged/ transferred into other types of materials for different functionalities, but preserve the desired uniformity and conformity.

ALD for 3D Current Collectors and Structure Support
Since the concept of all-solid-state 3D-integrated batteries was fi rst proposed for miniaturized wireless and portable electronics, ALD has demonstrated its merits in the construction of such 3D structures of high aspect-ratio for current collectors. For example, ALD TiN is deposited on patterned Si surface as a current collector for Li ion batteries, in which the TiN layer acts as a Li + diffusion barrier. [ 119 ] Other ALD materials, such as TaN and Pt, have also been reported as 3D current collectors. [ 120 ] Besides Si substrate-based 3D batteries, other micro/ nano templates have also been reported where ALD materials are used for current collectors. [ 121 ] As has been discussed previously, ALD Ru has been employed as 3D current collector for nanopore-based lithium-ion battery. [ 100 ] Using AAO template and ALD Pt, [ 122 ] Wen et al. constructed Pt@MnO 2 coreshell nanotube arrays for supercapacitor electrode, which was showed with high capacitance and rate capability.
ALD metal oxides have been employed to construct 3D current collectors using nanostructure-based templates. For example, Luo et al. developed a 3D nanostructure of TiO 2 nanotube@Fe 2 O 3 nanofl akes, which was made by ALD TiO 2 on the surface of Co 2 (OH) 2 CO 3 followed by a solution growth process. [ 123 ] As shown in Figure 9 a-b, ALD TiO 2 was developed into a uniform tubular structure for fast electron transfer, which would also benefi t for the stable deposition of Fe 2 O 3 . The hollowed nanostructure of TiO 2 @Fe 2 O 3 demonstrated an initial capacity of 840 mAh g −1 , and a capacity of 530 mAh g −1 could be maintained after 200 cycles, which was much enhanced when compared with the electrode without ALD TiO 2 support. As shown in Figure 9 c-d, the ALD TiO 2 nanotubes could be also combined with other materials, such as SnO 2 and CoS, which could lead to further enhancement in electrochemical performance. [ 124,125 ] ALD has been successfully employed for fabrication of other 3D active materials, [ 126 ] e.g., SnO 2 @TiO 2 double-shell nanotubes, derived from ALD SnO 2 and TiO 2 deposited on ZnO nanowire substrate (Figure 2 f). [ 127 ]

ALD for Construction of Hollow/Porous Nanostructures
Although metal oxides/hydroxides exhibit a theoretical energy density much higher than that of the carbonaceous counterpart, their intrinsically poor electric conductivity is a hindering parameter for application as electrode materials in supercapacitors. In addition, it is more challenging to develop a high surface area and well controlled pore structure for some of these metal oxide/hydroxide electrodes. The volume change and strain generated during the electrochemical reaction can drastically degrade their cycling ability. [ 128 ] Given that properly controlled pore and/or hollowed nanostructures have been demonstrated with enhanced rate capability and cycling stability, [ 129 ] ALD has been explored for construction of some these unique nano-/micro-hollow and porous nanostructures.
On the basis of ALD Al 2 O 3 and TiO 2 , porous and hollowed wire-in-tube nanoarrays of CoO TiO 2 have been reported by Guan et al. As shown in Figure 10 a-c, [ 130 ] in such rationally designed nanostructure, porous CoO nanowires are in direct contact with current collectors, where TiO 2 nanotubes contribute to the increased surface area of the electrode and protect the inner structure. The tunable, small gap between CoO and TiO 2 are designed as an "ion reservoir'' which facilitates the fast electrochemical reaction at high rates. Compared with CoO nanowires or solid core@shell structure of CoO@ TiO 2 , the electrode made of the new nanostructure showed much improved capacitance and rate capability, e.g., two to four times of the capacitance of the solid wires. As shown in  Figure 10 d-f, based on this new "nanogap" concept, hollowed ALD TiO 2 nanotubes have been developed for the protection of SnO 2 nanowires in LIBs. [ 131 ] When employed as an anode material in LIB, alloyed SnO 2 can generate a huge volume expansion and structure change with poor cycling ability. Although a protective coating layer enhances the cycling ability, the severe structure change can damage the protection layer. The rational design with purposely made hollow nanospace therefore tolerates and buffers the volume change of the inner material thus improving the cycling ability. As has been demonstrated, the SnO 2 TiO 2 with ≈40 nm gap showed excellent cycling ability, and a capacity of 393.3 mAh g −1 is maintained after 1000 cycles. The much improved cycling ability is attributed to the hollow nanospace that buffers the volume change, and the highly stable and uniform ALD TiO 2 layers stabilize SEI and protect the inner structures.
In addition to the above mentioned examples of new nanostructures developed by ALD, it has been demonstrated useful for creating hollowed nanospace without need for chemical etch of a sacrifi ced layer, but through solid state diffusion and Kirkendall effect. [ 132,133 ] As shown in Figure 10 g, for example, through the solid state diffusion reaction of Co(CO 3 ) 0.5 (OH) 0.11 H 2 O nanowire with ALD TiO 2 layer, Jiang et al. have successfully grown hollowed CoO-CoTiO 3 nanotube arrays. [ 134 ] Since its unique tubular core-shell structure and stable ALD CoTiO 3 , the CoO-CoTiO 3 was illustrated with excellent cycling stability with a capacity of 585 mAh g −1 being well maintained after 150 cycles, which is much better than that by CoO. A similar one-step solid state reaction between Co(CO 3 ) 0.5 (OH) 0.11 H 2 O nanowire and ALD SnO 2 layer results in the formation of CoO nanowires in CoSnO 3 nanotube structure. As further shown in Figure 10 h, [ 135 ] the unique "porous + hollowed" nanostructure of CoO᭪CoSnO 3 facilitates the electrode/electrolyte contact and provides shortened ion diffusion pathways, thus has demonstrated both better rate and cycling ability than that of CoO alone.
A recent piece of work from Fan's group has investigated several different nanostructures of metal oxide@carbon fl akes, one example of which is shown in Figure 10 i. [ 136 ] The porous carbon nanofl akes were made by the heat treatment of ALD Al 2 O 3 and glucose composite layer. The carbon coating provides conductive pathways and high surface area, which are essential for electrochemical electrodes. The carbon nanofl ake-coated CoO nanowires thus constructed showed much better capacitance and cycling ability than those of the bare one, with 98.6% of capacitance being maintained after 5000 cycles. As shown in Figure 10 j, the ALD carbon coating nanofl akes can be combined with mesoporous Ni x Co 1x O nanosheets, giving rise to much enhanced performance as LIB electrode. [ 137 ]

ALD as Sacrifi cial Layers for Ion Exchange
ALD materials have been employed as sacrifi cial layers and/ or reaction precursors by ion exchange, in deriving other active materials that are otherwise diffi cult or costly to make. For example, a rather uniform thin fi lm layer of CH 3 NH 3 PbI 3 perovskite was obtained from ALD PbS with a two-step ion exchange reaction. [ 138 ] Conformal large-scale WS 2 nanosheets are then made by a vapor-phase ion exchange reaction with ALD WO 3 . [ 139 ] In general, ALD assisted ion exchange involves an in situ reaction, thus the resultant fi lm can well maintain the uniformity of ALD seed layers.
surface of carbon substrate, which facilitated ion transport. In characterization for device performance in LIB, the electrode expressed a high capacity of 785 mAh g −1 at 1 C rate, and the capacity was well maintained after 500 cycles at 10 C rate. The same ion exchange concept has been utilized for electrodes of supercapacitors. For example, Zhu et al. developed metal nitride solid-state asymmetric supercapacitors using carbon cloth/graphene nanosheets substrate, one example of which is shown in Figure 11 f-g. [ 141 ] Through the ion exchange reaction with ALD TiO 2 and ZnO layer, thin TiN nanolayers (cathode) and Fe 2 N nanoparticles (anode) were uniformly assembled on graphene nanosheets, which provided a high surface area for fast electrochemical reaction. Using PVA/LiCl electrolyte, the full cell was demonstrated to provide a high energy density of 15.4 Wh kg −1 and a high power density of 6.4 kW kg −1 , together with excellent rate capability and cycling stability.

Conclusions and Outlook
This review has focused on the recent advances of ALD for new and improved electrode materials in electrochemical energy storage devices. High performance electrochemical energy storage has been extensively developed in recent years, with the typical key performance parameters being the high energy density, high power density and long cycling life stability. Novel electrode materials are crucial for development of the next generation high performance electrochemical energy storage devices with these superior parameters. ALD has been advancing rapidly over the past few years as a powerful nanotechnology for design and fabrication of advanced electrode materials with some of the most desirable features, which cannot be realized by other processing techniques. In this connection, this review has summarized three main aspects:  Figure 10. ALD for construction of hollow/porous nanostructures. a) SEM and b) TEM images of CoO᭪TiO 2 hollowed core-shell nanowires. c) Schematic illustration of the hollowed core-shell structure in supercapacitor, which is better than the bare core and solid core-shell structures. d) SEM image of SnO 2 ᭪TiO 2 hollowed wire-in-tube nanoarrays. e) TEM image of the hollowed nanostructure after LIB test. f) Schematics of the fabrication process of the SnO 2 ᭪TiO 2 wire-in-tube nanostructure, which has free space for volume expansion thus more stable than the solid wires. g) TEM image of a CoO-CoTiO 3 nanotube. h) TEM image of CoO᭪CoSnO 3 wire-in-tube structure. i) SEM image of CoO@carbon nanofl akes. j) TEM image of Ni x Co 1-x O@ carbon nanofl akes. a-c) Reproduced with permission. [ 130 ] Copyright 2012, Royal Society of Chemistry. d-f) Reproduced with permission. [ 131 ] Copyright 2014, American Chemical Society. g) Reproduced with permission. [ 134 ] Copyright 2013, Royal Society of Chemistry. h) Reproduced with permission. [ 135 ] Copyright 2014, Royal Society of Chemistry. i) Reproduced with permission. [ 136 ] j) Reproduced with permission. [ 137 ] Copyright 2015, IOP Publishing (i) ALD provides a unique platform for surface modifi cation of electrode materials, leading to much enhanced rate capability and cycling stability; (ii) Active materials grown by ALD on different substrates giving rise to some of the most optimized combination of electrochemical properties, where some of the reaction mechanisms and underlying principles have been visited; (iii) ALD has been successfully developed in the rational design and construction of novel nanostructures, which are otherwise impossible/diffi cult to achieve by other techniques, for electrochemical energy storage.
For the past several years, although considerable progress has been made with ALD for advanced electrode materials, it remains much room for further improvement and key understandings. One promising direction that has been undertaken and will continue is to develop novel ALD materials for surface modifi cation on electrodes, which will bring new surface and interface chemistry for better protection, ion transport and electrochemical reactions. As has been mentioned in this review, since the side reaction and instability of LIB electrodes, ALD Al 2 O 3 has been often employed for surface coating on these electrodes, while some other works suggest that LiAlO 2 and CeO 2 can lead to even further improved performance over Al 2 O 3 . There is still considerable amount of further research needed, in order to properly manipulate the ALD coating layer optimized for different electrochemical energy storage devices, such as ALD oxides, fl uorides, phosphates, and Li-containing coating materials.
Another interesting direction would be to develop new novel active materials by ALD for much improved electrochemistry, which can not only help make better fundamental understanding, but also achieve optimized device performance. It is commonly known that the performance by most of the known active materials is much below their theoretical expectations. Therefore there is still a way to go towards the most desirable structure at varying scales and dimensions. Since ALD gives much better control in materials growth, it will play an important role in the drive towards this goal. Since ALD active materials can be very controlled, they will pave the way towards further understandings on the electrochemical reaction mechanisms and phenomenon in some of these materials, for example by in situ studies.
A further interesting, and equally important future development is the advance of ALD, either by itself or by combination with other processes, in development of completely and Fe 2 N anode materials from ALD TiO 2 and ALD ZnO, respectively. g) Cycling performance of metal nitride full device with different bending situations. a-e) Reproduced with permission. [ 140 ] Copyright 2013, American Chemical Society. f,g) Reproduced with permission. [ 141 ] www.MaterialsViews.com www.advancedscience.com Adv. Sci. 2016, 3, 1500405 new nanostructures. Some of the known examples have been mentioned in this review. It will be defi nitely continue for development of novel materials with new structure for the next generation electrochemical energy storage devices.
Last but not least, with the unique and new electrode structures developed by ALD, it would be of interest to revisit even some of the "old" battery systems, such as aqueous nickel-zinc battery, which may well bring up much higher power density and improved cycling ability.
While ALD has been widely studied for design and construction of advanced electrode materials, the present and future trends are to establish various new phenomena and underlying principles, not only for energy storage, but also for energy generation and environmental devices, such as PVs and catalysts. With the structure, performance and underlying principles be established for materials developed by ALD, it would be of interest to develop new ALD systems for large mass production and at low cost for the expected wide range of applications. [ 142 ] With the steady establishment of ALD, another deposition technique using organic precursors is the molecular layer deposition (MLD), which has shown some potential for electrode surface modifi cation. [ 143 ] The technique, together with use of certain organic compounds, can bring in new electrochemical, electrical, optical, magnetic and catalytical properties. It would also be useful for some of the functional organic−inorganic hybrid materials that can be utilized for electrochemical energy storage.