Dawei
Liu
and
Guozhong
Cao
*
Department of Materials Science and Engineering, University of Washington, Seattle, WA, USA 98195. E-mail: gzcao@u.washington.edu
First published on 16th June 2010
Lithium ion batteries have been one of the major power supplies for small electronic devices since the last century. However, with the rapid advancement of electronics and the increasing demand for clean sustainable energy, newer lithium ion batteries with higher energy density, higher power density, and better cyclic stability are needed. In addition, newer generations of lithium ion batteries must meet the requirements of low and easy fabrication cost and be free of toxic materials. There have been many novel approaches to gain high energy storage capacities and charge/discharge rates without sacrificing the battery cyclic life. Nanostructured electrodes are seemingly the most promising candidate for future lithium ion batteries. Modification of the electrode surface chemistry and the control of appropriate crystallinity are also reported to improve the electrode intercalation capabilities. The study of appropriately designed nanostructures, interfaces and crystallinity has also promoted and is accompanied with the development of thin film electrodes without the addition of binders and conductive carbon that are typically used in the fabrication of traditional lithium ion battery electrodes, simplifying the electrode fabrication process and enhancing electrode storage density. In this perspective, we summarize and discuss the efforts of fabricating nanostructures, modifying surface chemistry and manipulating crystallinity to achieve enhanced lithium ion intercalation capacities, rate capabilities and cyclic stability, as well as the direct fabrication of binderless film electrodes with desirable nano- and microstructures.
![]() Dawei Liu | Mr. Dawei Liu is a PhD candidate under the supervision of Prof. Guozhong Cao in Department of Materials Science and Engineering at University of Washington, Seattle, WA. He has published five first-authored refereed papers and one book chapter. His specific research project is focused on nanostructured electrodes for efficient lithium ion intercalation. |
![]() Guozhong Cao | Dr Guozhong Cao is Boeing–Steiner Professor of Materials Science and Engineering at the University of Washington, Seattle, WA. He has published over 250 refereed papers, written and edited five books and monographs. His research has led to the creation of two spin-off companies on energy conversion and storage. His recent research is focused mainly on nanomaterials for solar cells, lithium ion batteries, supercapacitors, and hydrogen storage. |
Broader contextIn the new century, clean and renewable energy storage devices have become the foci of both the building industry and research development. Lithium ion batteries, as one of the most promising battery technologies, have attracted much attention due to their fast boom of market share. However, the commercialized lithium ion battery has not been good enough to satiate the public need and theoretically there is much room for improvement. Research has thus been focused on developing electrode materials with high discharge capacity, large charge/discharge rate and long life cycles. To achieve these goals, a lot of effort has been devoted to fabricating structures that best facilitate the intercalation behavior of lithium ions. In this perspective, these efforts are summarized and reported. |
Advanced energy storage technologies for vehicle electrification and efficient use of renewable energy from the sun and wind are a critical part of renewable energy.13,14 Several technologies are currently under intensive study.15–17 Generating biofuels from biomass is one example of converting solar energy to chemical fuels.18,19 Storing hydrogen in the liquid or solid forms near ambient conditions is another example.20,21 However, effective utilization of variable and intermittent sources of renewable energy in meeting growing electricity demand requires much improved electrical energy storage technologies that are economically viable to enable large scale market penetration for electric vehicles and renewable electricity storage. Capacitors and supercapacitors, also known as ultracapacitors, offer high specific power and long cyclic stability and life time, and thus are excellent choices for applications requiring a burst energy source; however, they suffer from low specific energy.22,23 Batteries possess great specific energy, but suffer from low specific power and cyclic degradation.24,25 Significant breakthroughs are needed for both supercapacitors and batteries to gain broad market penetration.
Batteries are more than a century old technology and have played critical roles in many technologies and today's mobile life. Although many types of batteries have been developed with much improved energy storage performance, the fundamental structure of a battery remains the same, consisting of an anode and a cathode with electrolyte sandwiched in between.26 The advancement in battery technology has been relying on the development and use of different types of materials for electrodes and electrolytes and thus with different electrochemical reactions.27,28Fig. 1 compares different types of batteries;29 lithium-ion batteries offer a balanced combination of high power and energy density, long cyclic life, and stability. The commercialization of lithium ion batteries has witnessed the soaring market share in the energy industry, especially in powering small electronic devices such as laptops and cellular phones.30,31
![]() | ||
Fig. 1 Comparison of the different battery technologies in terms of volumetric and gravimetric energy density. It is clear that for a given energy storage, lithium ion batteries are lighter and smaller. In addition, the lithium ion batteries are more environmental benign.29 |
A lithium ion battery, just like other types of batteries, consists of three major components: an anode, a cathode, and the electrolyte between them, and works by converting chemical potential to electric energy through Faradaic reactions, which include heterogeneous charge transfer process occurring at the surface of an electrode.32 Different from disposable or primary batteries, the lithium ion battery is a rechargeable or secondary battery, and the charging process involves the energy conversion from electric energy back to chemical potential.33 In a typical secondary lithium ion battery, Faradaic reactions are accompanied with both mass and charge transfer through the electrodes as well as dimension change; therefore, the surface area and the transport distance critically determine the performance of the battery in question.34 Chemical composition, crystal structure, and microstructure will have significant impacts on the reaction rate and transfer processes as well as its cyclic stability.29,35
E = CV | (1) |
Obviously, a high energy storage density can be achieved by choosing or developing anodic and cathodic materials with high lithium-ion intercalation capacity with a large electrochemical potential difference. Although increasing the potential of cathodic materials and, thus increasing the cell voltage, with a given anode, is seemingly the most effective way to increase the energy storage density, the decomposition of electrolyte at the cathode surface becomes a problem.34 Before finding suitable electrolytes stable at high cathode potential, the research is focused mostly on the development of electrodes with high lithium ion intercalation capacity. Fig. 2 summarizes the electrochemical potential and the typical lithium ion storage capacities of both anodic and cathodic materials.29
![]() | ||
Fig. 2 Voltage vs. capacity for positive and negative electrode materials presently used for under serious consideration for the next generation of rechargeable Li-based cells.29 |
While the lithium ion intercalation capacity of anodes, e.g. Sn can stabilize well above 500 mA h g−1 even after a big capacity loss following the initial cycle,36,37 the highest capacity of cathode materials is way below 500 mA h g−1. As a result, the performance of the cathode is a bottleneck for the improvement of the whole battery system.38,39 For example, the commercialized LiCoO2 has a mere capacity of ∼140 mA h g−1,40 and its potential alternatives LiMn2O4 and LiFePO4 have capacities of merely 150,41 and 170 mA h g−1, respectively.42 The cathode compounds demand more research efforts for new lithium ion batteries with much enhanced energy storage density.
In order to accommodate more lithium ions than the theoretical limit, the reversible phase transition boundary of the bulk materials needs to be extended. The Faradaic reaction and the phase transition associated with lithium ion intercalation and de-intercalation is dependent on the nature of the electrode materials including the arrangement of ions, the type and level of impurities and defects, and the surface energy. Electrodes away from equilibrium state may favor accommodating more lithium ions through reversible intercalation and de-intercalation process than the same materials in a thermodynamic equilibrium state. A number of ways could be used to create materials away from equilibrium. For example, nanostructured electrodes with huge surface energy are in a state far from thermodynamic equilibrium.48 As a result, the reversible phase transition limit of lithium ion intercalation and de-intercalation is often extended with much greater lithium ion storage capacity. In addition to the improvement of energy storage capacity, nanostructures also improve kinetics by providing a short diffusion path for lithium ions and enhance the rate capability.49,50 Nanostructures also permit freedom for volume change during long term cycles, alleviating the negative effect which could cause capacity degradation.51,52
Besides nanostructure, surface defects would contribute to lithium ion intercalation by shifting thermodynamics and improving kinetics. The presence of surface defects increases the surface energy and could possibly serve as nucleation center and, thus, facilitate phase transition. In addition, the presence of surface defects reduces the charge transfer resistance and improves the charge transfer kinetics while at the same time preventing active electrode dissolution in the electrolyte, which could improve cyclic stability.53 After the surface charge transfer process, lithium ion diffusion from the surface into interior depends on the crystal structure as well, so crystallinity is another factor to consider and manipulate to enhance lithium ion insertion capability.54 For some intercalation compounds, an amorphous structure appears to be more open to lithium ion diffusion due to the ability to withstand insertion stress with less well packed ions. Nanostructures, highly defective surfaces, and amorphous electrodes are all away from thermodynamically equilibrium states. Hence appropriately designed nano- and microstructures and the careful manipulation of surface and bulk chemistry will change the phase transition boundary, improve the transport kinetics, and permit more freedom for long-term cyclic stability.
In addition to searching and developing electroactive materials with higher lithium ion insertion capacity, rapid charge and discharge rate, and improved cyclic stability, research has been aimed at the development and fabrication of binderless electrode films (with directly incorporated conductive components) through solution-based chemical processing.55,56 In such an approach, the commonly used polymer binders to improve the mechanical integrity of the electrodes could be eliminated and, thus, enhance the energy storage capacity of a battery, as well as simplify the fabrication process and possibly reduce the fabrication cost. In addition, the possible electrolyte permeation path blocking problem induced by excess binder, which would severely reduce the active power electroactivity, could be avoided.57 Besides excluding binder from the electrode composition, some research groups also fabricated binderless and carbon-free thin film electrodes which promise a higher energy density because of the pure electroactivity of the electrode.58–60 These thin film electrodes, circumventing routine but tedious electrode casting fabrication, have a bright application future as microbattery electrodes, and could be used in vast number of research areas since they could power micro-electronic devices.61,62
The conventional route for fabricating LiFePO4 powders was mainly through solid-state synthesis. The starting precursors consisted of stoichiometric amount of iron salt, a lithium compound and most commonly ammonium phosphate as phosphorus source. After heat treatment first at 300–400 °C to expel gases and followed by calcination at higher temperature up to 800 °C under inert or slightly reductive atmosphere, the LiFePO4 powders could be obtained.77 However, the obvious disadvantage of the conventional solid state method was the particle growth and agglomeration due to the high temperature employed and the resultant product always possessed very small specific surface area.78 To solve this problem, mechanochemical activation was introduced into the process and the resultant powders had a much larger specific surface area.79 There have also been efforts using microwave heating instead of the furnace heating which limited the calcination temperature to a lower value to avoid excessive particle growth.80 However, even with these modified methods, it was difficult to obtain LiFePO4 with particle sizes below hundreds of nanometres.
Solution based methods have been proved to be effective for producing nanostructured LiFePO4.81 Hydrothermal method and sol–gel method were the two main methods employed and the obtained nanostructured LiFePO4 exhibited noticeable intercalation capability improvement as compared with micrometre-sized LiFePO4.
Hydrothermal growth refers to crystallizing substances at elevated temperature (typically 100–200 °C) from aqueous solutions at high vapor pressures. The starting precursors for fabricating LiFePO4 were typically an iron salt such as FeSO4, phosphate acid H3PO4 and lithium base LiOH.82 Particles of sizes from tens of nanometres to hundreds of nanometres could be obtained by tuning the reaction time, temperature and pH value.83 Dumbbell-like LiFePO4 microstructures hierarchically constructed with two-dimensional nanoplates of ∼300 nm length and ∼50 nm thicknesses were fabricated via hydrothermal self-assembly and exhibited a stable discharge capacity of ca. 110 mA h g−1 over 70 cycles at a C/30 charge/discharge rate (Fig. 3).84 Recham et al. used a solvothermal–hydrothermal method to enable the growth of LiFePO4 with controlled size and morphology and the best sample with particle size of 300 and 500 nm could deliver a sustainable capacity of 150 mA h g−1 at a C/10 rate.85 LiFePO4 nanowires with diameters of a few hundred nanometres were fabricated by adding nitrilotriacetic acid and isopropanol to the precursors and could reach an initial discharge capacity of 150 mA h g−1 and retained still as high as 138 mA h g−1 after 60 cycles at a charge/discharge rate of 0.1 C.86
![]() | ||
Fig. 3 (a) Typical low-magnification SEM image of dumbbell-like LiFePO4, (b) an individual dumbbell-like LiFePO4 from the obverse side, (c) HRTEM image of the tip of an individual dumbbell shape and (d) discharge capacity vs. cycle number of dumbbell-like LiFePO4 at C/30 charge/discharge rate84 |
The sol–gel route is another commonly used solution based method to fabricate nanostructured LiFePO4. Porous nanostructured LiFePO4 powder with a particle size distribution of 100–300 nm was obtained by using an ethanol based sol–gel route with lauric acid as a surfactant. The nanoparticles could deliver a specific capacity of 157 mA h g−1 at a discharge rate of 1 C and still had 123 mA h g−1 delivered when the rate was increased to 10 C.87 Similar high rate capability for lithium ion intercalation was also reported on sol–gel prepared wired mesoporous LiFePO4.88
For the hydrothermal and sol–gel methods, there has also been quite a lot of reported work on nanostructured LiFePO4/C composites with carbon coating aiming at enhancement of electronic conductivity which had even better rate capability than single phase nanostructured LiFePO4.89–93 In addition to the hydrothermal and sol–gel method, co-precipitation,94,95 emulsion-drying96 and spray pyrolysis methods97 could also be employed to fabricate nanostructured LiFePO4 or LiFePO4/C composites.
For LiFePO4 cathode and Li4Ti5O12 anode as discussed above, both possess good capacities at low charge/discharge rate but capacities degraded severely as the rate was increased due to the poor electronic conductivity. The introduction of nanostructures improved the intercalation kinetics by providing larger electrode/electrolyte contact area and reduced path for electron transportation and lithium ion diffusion, resulting in noticeably improved capacities at high rates.
Template-based deposition involves deposition of the material of interest, or a precursor for that material, into the pores of a microporous template membrane. After deposition, the template will be eliminated by either chemical etching or thermal annealing. By using polycarbonate filtration membrane (PC) as template, Martin and co-workers have fabricated polycrystalline V2O5 nanorod arrays. In lithium ion intercalation property study, the resultant nanorods delivered three times the capacity of a thin-film V2O5 electrode at a high rate of 200 C.120 Wang et al. have successfully fabricated single-crystalline V2O5 nanorods, nanotubes and Ni–V2O5 nanocables by employing electrophoretic deposition.121–123 As-fabricated V2O5·H2O nanotube arrays demonstrated an initial high capacity of 300 mA h g−1, about twice that (140 mA h g−1) of the plain film of the same chemical composition. Ni–V2O5·nH2O core/shell nanocable arrays were prepared by a two-step electrodeposition method with Ni nanorod arrays fabricated by electrochemical deposition first and vanadium pentoxide shell deposited onto the surface of nickel nanorods through sol electrophoretic deposition. Compared with V2O5 nanorod arrays and sol–gel-derived V2O5 films, the specific power of the nanocable arrays was enhanced by 1–2 orders of magnitude as shown in Fig. 4. Similar template-based sol–gel deposition employing PC template was also used for the synthesis of MnO2 nanorods and the produced nanorods delivered an initial capacity of 183 mA h g−1 and stabilized on subsequent cycles to 134 mA h g−1.124
![]() | ||
Fig. 4 Schemes of nanorod, nanotube and nanocable array electrodes for energy storage (top). SEM images of oxide nanorod arrays (bottom right) and comparison of specific energy and specific power of vanadium pentoxide electrodes in the form of film, nanorod arrays and nanocable arrays (bottom left).123 |
The hydrothermal method is an effective method to fabricate one-dimensional nanostructures. Nanostructured α-, β- and γ-MnO2 have been synthesized through the solution and hydrothermal route.125 While α- and γ-MnO2 nanowire or nanorods exhibited good electroactivity to lithium ions by delivering capacities more than 200 mA h g−1, β-MnO2 nanostructures showed low capacity and poor cyclic stability. Other research indeed revealed that until reaching a small enough scale of particle size, the intercalation capacity of β-MnO2 was fairly low.126 TiO2 is one of the most “hydrothermally fabricated” compounds used for lithium ion intercalation application. TiO2 nanowires synthesized by hydrothermal method were reported by Armstrong et al. and the TEM images of the nanowires with the chronopotentiometric curves of lithium ion intercalation are shown in Fig. 5. The intercalation capacity was as high as 275 mA h g−1, corresponding to the lithiation of Li0.82TiO2.127,128 Nanorods and nanotubes of TiO2 were also fabricated by different groups and exhibited similar high capacities comparable to TiO2 nanowires.129–131
![]() | ||
Fig. 5 (a) Low-resolution TEM image of the TiO2-B nanowire, (b) high-resolution lattice image viewed down the [100] projection (inset: electron diffraction pattern of a TiO2-B nanowire and (c) variation of potential (vs. 1 M Li+/Li electrode) with Li content (charge passed) for TiO2-B nanowires (solid line) and bulk TiO2-B (dashed line) cycled under identical conditions. Rate: 10 mA g−1 (10 mA of charge passed per gram of TiO2-B); voltage limits: +1 and +3 V; V = potential.127 |
The fabrication of mesoporous nanostructures often involves the use of soft (surfactant) or hard templates. In the 1990s, surfactant-templated TiO2 was initially fabricated by using amphiphilic poly(alkylene oxide) triblock copolymer as the structure-directing agent in an ethanolic solution of TiCl4.135 Later application of this mesoporous structured compound as lithium ion intercalation electrodes revealed unusually fast capacitive and intercalation charging abilities.136 Besides the synthesis route just described, mesoporous TiO2 could also be synthesized by the evaporation-induced self-assembly procedure with novel poly(ethylene-co-butylene)-b-poly(ethylene oxide) polymer (KLE) used as template. The templated films after annealing at elevated temperatures transform to the anatase phase and can store a capacity of ca. 200–250 mA h g−1.137 The rate capability measurement was also carried out on mesoporous TiO2 fabricated by using amphiphilic molecule as the templating agent: the mesoporous samples delivered a capacity of 184 mA h g−1 at C/5 and 95 mA h g−1 at 30 C, possessing a much better rate capability than commercial samples.138 Lou et al. studied the storage properties of TiO2 mesoporous hollow particles and after the initial high capacity of 408 mA h g−1, the capacity stabilized at >170 mA h g−1 for 35 cycles.139
Mesoporous V2O5 was fabricated following similar surfactant-templated procedures by using VCl4 instead of TiCl4. At a very high charge/discharge rate of 50 C, the capacity delivered by mesoporous V2O5 could be as high as 125 mA h g−1, which promised its application as a low power capacitor or high power batteries owing to the good balance between specific power and specific energy.140
Besides using surfactant as soft template, mesoporous structure could also be obtained by using a hard template, e.g. mesoporous silica. Hollow LiFePO4 was fabricated by using the hard templates KIT-6 and had a large BET specific surface area of 103 m2 g−1 with pore size distribution centered at 5.6 nm.141 The mesoporous electrode demonstrated excellent rate capability and cyclic stability. At a rate of 15 C, the capacity was 153 mA h g−1, 95% of that at 0.2 C rate. In addition, no obvious capacity degradation was noticed after 80 cycles. Considering the well-known low rate capability of LiFePO4 due to its poor electronic conductivity, using mesoporous structured electrodes could be one of the possible solutions to solve this problem.
Besides the improved electrode cyclic stability and enhanced rate capability, for some compounds, the electroactivity was significantly enhanced after changing the bulk material into a mesoporous structured one. Successful fabrication of mesoporous β-MnO2 employing mesoporous silica KIT-6 was reported. While bulk β-MnO2 was for a long time assumed to be with extremely low intercalation capacity, i.e. below 60 mA h g−1,126 mesoporous β-MnO2 with a pore size centered at 3.65 nm exhibited a high capacity of 284 mA h g−1 and stabilized at 200 mA h g−1 after initial degradation at a current density of 15 mA g−1.142 The TEM and HRTEM images of the mesoporous β-MnO2 before and after cyclic reactions together with the electrochemical cyclic performance are shown in Fig. 6 and it could be clearly seen that the integrity of mesoporous structure was well maintained during the intercalation/de-intercalation cycles. This mesoporous electrode also possessed good rate capability by having 81% capacity remaining after the current density was increased to 300 mA h g−1. Mesoporous LiMn2O4 was made through a similar route by employing KIT-6 template but involved more solid-state reactions to evolve from Mn2O3 to LiMn2O4.143 Thus-obtained mesoporous LiMn2O4 nearly doubled the capacity of bulk LiMn2O4 and possessed a much better cyclic stability than nanoparticulate LiMn2O4. The surface area of these two mesoporous products were 127 and 90 m2g−1, respectively.
![]() | ||
Fig. 6 TEM and high-resolution TEM (HRTEM) images of mesoporous β-MnO2: (a, b) as-prepared; (c, d) after first discharge; (e, f) end of discharge after 30 cycles; (g, h) end of charge after 30 cycles and (i) cyclic retention for mesoporous β-MnO2 cycled at (a) 15, (b) 30 and (c) 300 mAg−1; (d) bulk β-MnO2 cycled at 15 mA g−1.142 |
In conclusion, during lithium ion intercalation/de-intercalation, mesopores readily supplied ions from the electrolyte and acted as the buffering layers to alleviate the volume change experienced during lithium ion intercalation/de-intercalation. As a result, the cyclic stability of mesoporous structured electrodes was improved and the rate capability was enhanced. Using a template-based method to fabricate mesoporous structure could produce ordered pores with sizes controllable by template tuning. In general, the BET specific surface areas of template-derived mesoporous structures are >90 m2 g−1 and the pore sizes were centered from 3–6 nm. This structure is highly favorable for lithium ion intercalation as evidenced by noticeable improvement in capacities, cyclic stability and rate capabilities. However, the introduction of template into the synthesis processes induced much higher cost and the elimination of template after reactions was also a technical challenge. Recently, successful fabrication of high surface manganese dioxide following the template-free concept has been reported by Sinha et al. by means of a wet precipitation method144 and its application in extensive air purification turned out to be a great success. However, because of the unordered mesoporosity, manganese dioxide fabricated in this way was not suitable as a large lithium-ion capacity storage device since the electrolyte penetration would be very difficult in such thick oxide films with small pores. For unordered mesoporosity, the availability of macropores is crucial to facilitate complete electrolyte infiltration into the random distributed mesopores. Various attempts have been reported in fabricating a variety of macroporous metal oxides, e.g. NiO nanoflowers.145 The combination of macroporous and mesoporous structure was also reported in V2O5146 to improve lithium-ion intercalation kinetics. In our experiments, we have successfully deposited hierarchically mesoporous MnO2 nanowall arrays onto a platinum substrate to make thin film electrodes and obtained excellent capacities and cyclic stability which will be discussed in detailed in section 5.3
Another problem of nanostructured electrodes was related with the active metal ion dissolution in the electrolyte which would cause capacity degradation. LiMn2O4 was one typical victim: Mn ions in the electrode could easily dissolve in the electrolyte and the large electrode/electrolyte interface characteristic of nanostructures badly aggravated this problem.161,162 Despite that the discharge capacity and rate capability could possibly be improved, cyclic stability was obviously a big problem for nanostructured lithium manganese oxide and related manganese oxide electrodes.125,163
To solve the excessive SEI growth and active metal ion dissolution problems, special surface chemistry design is necessary for nanostructured electrodes as we are going to discuss next.
Introducing other elements into the electrode compound was adopted as one way to enhance the surface charge transfer conductivity. The element would either substitute the transition metal ions (doping) to modify the crystal structure168 or interact with the transition metal ions on the crystal surface without entering the lattice.169 The charge transfer resistances in both situations were significantly reduced and the rate capability was obviously improved. However, despite the enhancement of surface charge transfer conductivity, the dissolution of electrode in the electrolyte was still a challenge to the performance stability of the nanostructured electrode.
To protect the electrode from dissolution, the straightforward but essential concept was based on reducing the direct contact between electrode and electrolyte, which could be realized by the method of coating a layer of porous materials on the electrode surface. The commercialized LiCoO2 was used by many research groups as the model electrode to carry out the coating experiments and many oxides have been tested as the coating layer, e.g. Al2O3,170 AlPO4,171 CeO2,172 TiO2,173 Li4Ti5O12174etc.Fig. 7 shows a TEM image of coated LiCoO2 and compares the electrochemical cyclic performance comparison of the bare electrode and coated electrodes. It could also be seen that after coating, the cyclic stability of lithium ion intercalation measurements were obviously improved.175 More detailed experiments were also carried out by using different techniques to coat TiO2 onto the LiCoO2 electrode surface and it was found that the mechano-thermal coating from pre-formed nanoparticles were preferable to sol–gel coating from an alkoxide precursor, the cyclic stabilities in both situations were noticeably improved (5-fold for sol–gel coating and 12-fold for mechano-thermal coating).176 Besides acting as the protection layer to prevent electrode dissolution, the TiO2 coating was also found to help suppress the cycle-limiting hexagonal/monoclinic/hexagonal phase transitions accompanying the charge–discharge processes. In addition to oxides coating, polymer coating was also experimented with a combined model calculation.177 Electrochemical measurements showed that coating LiCoO2 with a poly-(2EHA-Co–F) film significantly decreased the activation energy for Li+ exchange which could improve reaction kinetics and facilitate the charge transfer process. Spinel LiMn2O4 or substituted LiMn2O4 is a promising cathode material but for a long time suffered rapid capacity fading during repeated charge/discharge cycles partly due to the dissolution of active Mn ions in the electrolyte especially when the electrodes were comprised of nanoparticles.178,179 The Mn dissolution phenomenon was even more severe at elevated temperatures,180,181e.g. 55 °C. It turned out that coating oxides onto the electrode surface to reduce the direct contact between spinel and the electrolyte was an effective method to improve the cyclic stability of LiMn2O4. Kannan and Manthiram coated LiMn2O4 with LixCoO2, LiNi0.5Co0.5O2, Al2O3 and MgO using solution-based coating followed by heat-treatment.182 All the surface coated (modified) samples showed much better capacity retention at both room temperature (25 °C) and elevated temperature (60 °C) than unmodified LiMn2O4. The LiNi0.5Co0.5O2-modified sample showed superior capacity retention with only 2.8% fade in 100 cycles at 60 °C with capacity around 110 mA h g−1; the Al2O3 modified sample showed a higher capacity of 130 mA h g−1 but with a faster fading rate (16% fade in 100 cycles at 60 °C); the Li0.75CoO2 modified sample showed the best combination of capacity (124 mA h g−1) and retention (8% fade in 100 cycles at 60 °C). Because of easy synthesis and chemical stability, Al2O3 was the most used coating compound onto electrodes. There have also been several coating methods of Al2O3 such as melting impregnation,183 reactive sputtering184 and solution soaking:185 noticeably improved cyclic stability of electrode being coated over uncoated one was observed. ZnO coating on LiMn2O4 was also reported and at elevated temperature of 55 °C, the coated LiMn2O4 showed capacity retention of 97%, significantly higher than the 55% capacity retention of the bare LiMn2O4.186 The ZnO coating collected HF from the electrolyte and better preserved the interfacial morphology and ensured the stable charge transfer process. A similar ZrO2 coating also improved the high-temperature cyclic stability by screening the acidic species from the active electrode;187 in addition, ZrO2 coating improved the rate capability up to the high rate of 10 C owing to the enhanced charge transfer conductivity because coated ZrO2 can act as a highly Li-conducting solid electrolyte interface and the strong bonding to LiMn2O4 which could tolerate the lattice stress resulting from the volume expansion during lithium ion intercalation.
![]() | ||
Fig. 7 (a) TEM image of LiCoO2 surface modified with Al2O3, (b) electron diffraction of the coated LiCoO2 and (c) discharge capacity of bare and surface modified LiCoO2vs. cycle number.175 |
Compared with the coatings onto LiCoO2 and LiMn2O4 electrodes, the favorable coating compounds onto LiFePO4 electrodes were limited to conductive species. Because of the poor electronic conductivity of LiFePO4, the bare electrode could only be charged/discharged at a very low rate.188,189 Thus the adopted coating must be highly conductive in order not to aggravate the low rate performance problem of the electrode. The conductive coatings on LiFePO4 were mainly three kinds of conductive compounds: carbon coating,190,191 metal coating192,193 and conductive oxide coating.194Fig. 8 shows the TEM images and electrochemical cyclic performance at different charge/discharge rates of LiFePO4/C nanoplate coated by amorphous carbon.195 It could be seen that the coated electrodes exhibited excellent cyclic stability and the discharge capacity was still about 100 mA h g−1 at a high rate of 10 C, preserving more than 50% of the discharge capacity at 0.1 C.
![]() | ||
Fig. 8 (a) TEM image showing a uniform coverage of amorphous carbon coating around the surface of a LiFePO4/C nanoplate, (b) HRTEM image showing a nearly 5 nm thick amorphous carbon layer around the surface of LiFePO4/C and (c) capacity vs. cycle number plots of LiFePO4/C thin nanoplates at various current rates of 0.1 to 30 C.195 |
For transition-metal oxide electrodes, carbon or metal was often used as coating layers, VOx coated with carbon prepared via reaction under autogenic pressure at elevated temperatures was reported.196 Both the reversibility and rate capability was much better than V2O5 nanoparticles without carbon coating. Metal layers were also coated onto nanostructured TiO2: the silver mirror reaction was used to coat Ag particles onto hydrothermally synthesized TiO2 nanotubes and cyclic stability and rate capability was found to be improved.197 A metal film of Cu or Sn was vacuum-deposited onto the surface of mesoporous anatase TiO2 electrodes and the electrode surface modification made by thin-film deposition improved the kinetics of Li intercalation/de-intercalation and remarkably enhanced the electrochemical performances in terms of capacity, stability and rate capability.198
Derived from the coating concept, recently a new method of surface grafting was developed to modify the surface chemistry, or say, to lower the interfacial chemical reactivity of side reactions (e.g. electrolyte decomposition) which was detrimental to long-term energy storage properties. Nitro-aryl groups were electrografted onto a Li1.1V3O8 surface by in situ diazonium chemistry during lithium ion intercalation when the electrode was potentiodynamically discharged.199 A homogeneous multilayer was formed whose thickness could be modulated. The multilayer did not impede charge transfer or limit the electrochemical reactivity. However, it did decrease the chemical reactivity of the material towards the electrolyte, resulting in significant improvements of the capacity retention.
To date, coating appears to be the most effective method to improve the surface charge transfer process. However, whatever kind of coating techniques was used, the coating process is often complicated and sometimes involved delicate equipment to achieve a homogeneous coating layer with good porosity. The basic requirement of the electrolyte permeability after coating was always challenging and tricky to meet. Considering all of these practical problems, we have been attempting to create a layer of surface defects on the intercalation electrodes to play a similar role as an exterior coating. The work included TiO2 nanotube arrays and V2O5 xerogel films annealed under reducing or inert gas flow and identified with surface defects (section 3.2 and 3.3)
Titania nanotube arrays were synthesized by anodic oxidation method according to literature.201 Then they were annealed in reducing CO gas at 400 °C for 3 h.202 XPS study of the nanotube arrays annealed in CO gas confirmed carbon doping onto TiO2 surface in the form of a minor amount of Ti–C and the formation of Ti3+ point defects as shown in Fig. 9a. Compared with N2-annealed arrays which possessed an electrode resistance of 66 Ω and a charge-transfer resistance of 38 Ω, CO annealed arrays possessed an electrode resistance of 60 Ω and a reduced charge-transfer resistance around 26 Ω, indicating a higher charge-transfer rate of Li+ in the electrode. This improved charge-transfer conductivity of CO annealed TiO2 arrays could be attributed to the presence of surface Ti–C species and Ti3+ groups with oxygen vacancies which enhanced the surface conductivity of the electrode.
![]() | ||
Fig. 9 (a) Ti![]() |
Fig. 9b summarizes and compares the relationship between the discharge current density and corresponding intercalation capacity of the TiO2 nanotube arrays annealed in N2 and CO, respectively. The lithium ion intercalation capacity of the N2 annealed nanotube arrays was found to be more sensitively dependent on the current density; the intercalation capacity reduced rapidly with the increased current density. At a current density of 100 mA g−1, the capacity of the N2 annealed TiO2 nanotube array was as high as 245 mA h g−1. However, when the current density was tripled to 320 mAg−1, the capacity decreased to 164 mA h g−1, losing one third of its discharge capacity. At a current density of 1 Ag−1, the capacity was further reduced to a value of 127 mA h g−1. In comparison, the CO annealed TiO2 nanotube arrays demonstrated less sensitive intercalation capacity. For example, an intercalation capacity of 261 mA h g−1 decreased to 223 mA h g−1, less than 20% reduction, when the current density increased from 100 to 320 mA g−1. It was clear that the CO annealed TiO2 nanotube arrays possessed much higher intercalation capacities, approximately double of that of N2 annealed TiO2 nanotube arrays at high current densities, i.e. possessing a capacity of 101 mA h g−1 at 10 A g−1.
The presence of defects may contribute to the improved intercalation capability of the CO annealed TiO2 nanotube arrays, as has been reported in other intercalation oxide electrodes such as V2O5.203 In the titania system, both intercalation and de-intercalation processes involve a phase transition between tetragonal TiO2 and orthorhombic LixTiO2 through the following reaction:
xLi+ + xe− + TiO2 = LixTiO2 |
Phase transition occurs through nucleation at the interface and subsequent growth from the interface towards the interior. Besides improving the charge transfer conductivity, the presence of defects on the surface of TiO2 nanotubes could also serve as nucleation sites so as to promote the phase transition and enable more lithium ion intercalation and improve the rate capability.
Sol–gel derived V2O5 films on FTO glass were annealed in N2 and air-flow atmosphere at 300 °C for 3 h under otherwise identical conditions.204 While air annealing did not change film color much; after N2 annealing, the originally yellow film turned dark green suggesting the presence of V4+ and V3+ species on the film surface. The charge transfer resistance of the N2-annealed film was two thirds that of the air-annealed film, reduced from 125 to 86 Ω. Fig. 10 compares the long term cyclic stability and corresponding XRD patterns of the air- and nitrogen-annealed films for lithium ion intercalation measured with a current density of 600 mA g−1 for 50 continuous discharge/charge cycles. The poor cyclic stability of air-annealed V2O5 film could be caused by the relatively larger grain size.205 For N2-annealed film, before reaching the highest discharge capacity value of 158 mA h g−1 at the 24th cycle, it started with a low discharge capacity of 68 mA h g−1. After 50 cycles, the capacity was still as high as 148 mA h g−1. While the mechanism causing such a change in storage capacity, with initially low followed by a sharp increase with increased lithium ion intercalation and de-intercalation cycles in the N2-annealed sol–gel derived V2O5 film electrodes, is not clear, the same or similar results or observation have been reported in literature and it could possibly be due to the surface defects.206,207 The crystallinity of these two kinds of films were also found to change differently during cycling: while the hydrous V2O5 peak degraded noticeably in the air-annealed film, such degradation was much less severe in the N2-annealed film, suggesting the disruption of layered structure. Along with defects on the film surface, the integrity of layered structure could be a further reason for the good cyclic stability of the N2-annealed film.
![]() | ||
Fig. 10 (a) The Li-ion intercalation discharge capacity of V2O5 films annealed in air and N2 at 300 °C for 3 h as a function of cyclic number. The measurements were carried out in a potential window between 0.6 and −1.4 V vs. Ag/AgCl as the reference electrode at a current density of 600 mA g−1; X-ray diffraction patterns of V2O5 films annealed in (b) air and (c) nitrogen before and after 20 and 50 cycles of lithium ion intercalation and de-intercalation measurements.204 |
The presence of surface defects could contribute significantly to the stability improvement which can be analyzed in three aspects: (1) the interfacial charge transfer abilities of the N2-annealed V2O5 film was improved because of the presence of more conductive surface defect species. As has been found in the optical absorption and impedance analysis, both the optical and electrical conductivity of the film annealed in N2 was improved compared with film annealed in air due to the presence of V4+, V3+ ions and associated oxygen vacancies on the film surface. It was found that the intercalation capability of lithium ions into the V2O5 xerogel film annealed at high temperature was mainly determined by the interfacial reactions at the electrolyte/electrode interface rather than the lithium-ion transport in the bulk oxide electrode.208 Since the enhanced charge transfer conductivity facilitated electron transportation during lithium ion intercalation/de-intercalation at the electrolyte/electrode interface which would obviously facilitate the lithium ion intercalation process,209 the cyclic stability and rate capability improvement could be explained. Similar improvement after N2 annealing was also observed in TiO2 nanotube arrays;210 (2) in addition to the conductivity enhancement contribution, the defect layer, like coating layers, also prevented the possible dissolution of V2O5 film in the electrolyte and ensured the integrity of film surface morphology upon cycling. One of the major causes of lithium ion intercalation capacity degradation in long-term cycles was the dissolution of the active material (electrode) in electrolyte. Building up a protecting layer on the surface will effectively reduce the possibility of electrode dissolution, thus improving the cyclic stability; (3) more than just a simple protecting layer, these surface defects of lower vanadium valency and oxygen vacancies could also serve as nucleation centers in the phase transformation process that occurred during lithium ions intercalation/de-intercalation.211 For this reason, the phase transformation process in the N2-annealed film was much easier and more reversible. The crystal structure was also more stable and the stability of lithium ion intercalation capacities was better than for air-annealed V2O5.
The higher capacity of amorphous/low crystallinity substances over crystalline ones was reported and discussed as early as the study of intercalation cathode sulfides, e.g. MoS2. Whittingham et al. pointed out that this might be associated with either the more open lattice in amorphous compounds or the disordered framework which prevented the decomposition in some materials.214 In terms of nano-ionics, amorphous compounds have more interfaces within the crystal due to the small grain size and could store more lithium ions than well-crystallized compounds on the grain boundaries and interfaces. The low mobility of lithium ions inside the amorphous structure could be a limitation for high rate performance but the high energy storage capacity was indeed promising for powering small electronics. Especially given that the energy storage capacity of cathodes has not seen much impressive improvement in recent years, constructing appropriate amorphous structure might be a good direction to achieve high intercalation capacity and even high energy storage density.
Lithium transition metal oxides are often synthesized through high-temperature solid-state reactions, so the products were well crystallized structures like the layered structure of LiCoO2, spinel structure of LiMn2O4 and olivine structure of LiFePO4. However, for transition metal oxides which only consist of metal ions and oxygen, the variation of crystallinity was more controllable and the amorphous structure could be fabricated from low-temperature synthesis route.
Xu et al. synthesized amorphous manganese dioxide via a sol–gel route and the as fabricated material delivered a high capacity of 436 mA h g−1 and stored energy at a level of 1056 mW h g−1.215 However, the rate capability of this amorphous material was low due to the poor diffusion abilities of lithium ion inside the structure. To improve the rate capability, Xu et al. also made a cryogel from the sol–gel derived MnO2 aerogel via freeze drying. The nanoporous cryogel had a high specific surface area of 350 m2 g−1 and exhibited specific capacities of 289, 217 and 174 mA h g−1 at C/100, C/5 and 2 C rates, respectively, which demonstrated excellent rate capability.216 West et al. fabricated amorphous MnO2 nanowires arrays through a template-based electrodeposition and the cathode assembled from the crude material without adding any binder or conductive species exhibited a specific capacity of approximately 300 mA h g−1.217 Despite these good results obtained from amorphous MnO2, amorphous MnO2 often showed a quick fading rate during repeated cycles. Although the exact failure mechanism is not clear yet, the instability of the amorphous structure and the conglomeration of the small particles during cycling could be the possible reasons. There are also corresponding efforts to solve the problem by doping other elements such as Na,218 Cu219 or Bi220 to achieve improved cycling performance by stabilizing the local structure.
Amorphous intercalation compounds often exhibit sloping charge/discharge curves instead of possessing a well-defined plateau which was characteristic of well crystallized intercalation compound.221 As a result, the average intercalation voltage was lower. The energy density of a lithium ion battery is a product of capacity C and voltage V. The amorphous intercalation electrodes sacrifice the level of V for higher capacity C. In the case of amorphous MnO2 whose intercalation discharge curve is shown in Fig. 11,215 the sloping discharge curve possessed an average voltage of ca. 2.5 V, lower than 3 V reported from well-crystallized MnO2,124 however, the high capacity of more than 400 mA h g−1 was more than double the capacity of the well crystallized counterpart and the energy density was thus increased. In general, when calculating the energy density of amorphous intercalation electrodes, both voltage drop and capacity increase need to be considered.
![]() | ||
Fig. 11 Variation of electrode potential with lithium ion content upon insertion into the oxide at a current density of 20 μA cm−2.215 |
![]() | ||
Fig. 12 Li+-ion intercalation discharge capacity of amorphous as-grown TiO2 nanotube arrays and anatase TiO2 nanotube arrays annealed at 300, 400, and 500 °C in nitrogen for 3 h as a function of cyclic number. The measurements were carried out in a potential window between −0.6 and −2.1 V vs. Ag/AgCl as a reference electrode at a current density of 320 mAg−1.210 |
There were also some reports of composites having more amorphous structure than single oxides with good crystallinity due to the crystallization competition between oxides. TiO2 were added to V2O5 following the sol–gel route.222 While pure V2O5 xerogel exhibited typical hydrated V2O5 peaks, after introducing TiO2, no noticeable peaks were found and this suggested the amorphous state of the composites. Even after annealing at 500 °C for 1 h, the crystallinity was still poor. The intercalation capability of the amorphous composites were measured and compared with pure V2O5 film; it was found that all the composites films with different composition ratios exhibited a higher capacity and the composition of V/Ti = 80/20 exhibited a very high initial capacity of more than 400 mA h g−1. An et al. added amorphous NiO2 nanoparticles to a mixture of TiO2-B and anatase TiO2 nanotubes and both the cyclic stability and capacities at high current densities were improved.223
Silicon anode was another good system to demonstrate the advantage of amorphous electrodes. While crystalline silicon often suffers severe volume change during lithium ion intercalation/de-intercalation and possessed low cyclic stability,224 amorphous silicon showed a far less severe problem and could deliver stable high capacity over hundreds of cycles.225 Amorphous silicon thin film prepared by DC magnetron sputtering of silicon on stainless steel substrates showed a good performance with a stable capacity of about 3000 mA h g−1 and a relatively low irreversible capacity.226 Similar stability of high capacity was also reported on amorphous Si film made by pulsed laser deposition.227,228
It is somewhat arbitrary to conclude that amorphous structures deliver a higher capacity than well crystallized structure, or the opposite, because the situation varies depending on the intercalation compounds, e.g. the different story of TiO2 and V2O5 as discussed above. However, while the lithium ion storage capability of well crystallized structures is already well known and subject to little chance of ground-breaking improvement, appropriate manipulation of amorphous or low-crystallinity structures could be a good route to circumvent some of the limitations of the rigid structure and achieve significant advances.
Preparation of binderless and carbon-free thin film electrodes were reported by several groups utilizing radio frequency (rf) magnetron sputtering,238,239 pulsed laser deposition (PLD),240,241 electrostatic spray deposition (ESD),242,243 sol–gel spin coating244,245 to directly deposit electroactive materials onto conductive substrates. These thin film electrodes without any additive exhibited comparable electrochemical performance to binder and carbon-added electrodes. In our experiments, we fabricated nanostructured thin film electrodes by sol–gel derived coating and electrodeposition. To improve the rate capability of thin film electrodes, we also modified the surface chemistry of electrodes and obtained good storage capacities under large charge/discharge current density.
![]() | ||
Fig. 13 (a) TGA curve for V2O5·nH2O xerogels, (b) dependence of interlayer spacing on the n value in V2O5·nH2O, (c) dependence of grain size on the n value in V2O5·nH2O and (d) cycling performance at a current density of 100 μA cm−2 for V2O5·nH2O films obtained at 25, 110, 250 and 300 °C. The voltage ranges from 0.4 to −1.6 V vs. Ag/Ag+.205 |
V2O5 sol–gel derived film could also be synthesized through electrophoretic deposition. Electrophoretic deposition refers to the migration of colloidal particles suspended in a liquid medium under the influence of an electric field and deposition onto an electrode. It is a useful technique to deposit polycrystalline films with controlled crystalline texture and good porosity. In addition, it is simple and low cost with film thickness controllable by adjusting deposition conditions which are favorable for energy storage device electrodes. Zhitomirsky’s group has adopted electrophoretic deposition method to fabricate manganese oxide or manganese oxide–carbon nanotube composites films used as electrochemical supercapacitors with capacitances of 150 Fg−1 for pure manganese oxide and 650 F g−1 for composites being obtained.246,247 As a lithium ion battery electrode, V2O5 film could also be prepared by electrophoretic deposition from V2O5 sol using an anodic voltage 5 V followed by film calcination in air at 500 °C for 1 h.248 The resultant film exhibited an initial capacity of 250 mA h g−1 and increased to 310 mA h g−1 after 50 cycles in a potential window between 0.4 and −1.6 V.
![]() | ||
Fig. 14 (a) Scheme showing the proposed growth mechanisms of hierarchically structured manganese hydroxide nanowall arrays on cathodes due to the increased pH value resulting from water electrolysis (blue area stands for high pH): precipitation of manganese hydroxide nanoparticles from the electrolyte accompanied with the release of hydrogen gas bubbles from the cathode surface, (b) SEM image of the hierarchically structured nanowall arrays reflecting the structure proposed in the growth mechanism scheme, (c) TEM image of stacked nanoparticles in a nanowall with voids (pores) and (d) comparison of discharge capacities of anodic deposited manganese dioxide and cathodic deposited manganese dioxide in the first 10 cycles. The measurements were carried out between 0.4 V and −1.4 V vs. Ag/AgCl at a current density of 30 mA g−1.249 |
The electrochemical properties of the as-fabricated nanowall arrays were directly measured on the platinum substrate without any electrode refinement and compared with an anodic oxidized MnO2 film. In the lithium-ion intercalation test, both the cathodic and anodic MnO2 were measured for different deposition thicknesses, i.e. 0.5, 1.5 and 2.5 μm and the long time cyclic performance of different thicknesses was compared in Fig. 14d. It was obvious that cathodic deposited manganese dioxide possessed a higher discharge capacity and better stability over anodic deposited manganese dioxide for each thickness. For the mesoporous nanowall arrays, when the film thickness was 0.5 μm, the initial capacity was as high as 256 mA h g−1 and after the thickness was increased to 2.5 μm, the initial capacity was still as high as 230 mA h g−1.
The above comparison clearly revealed that the cathodically deposited manganese dioxide films possessed a favorable hierarchically mesoporous structure with higher discharge capacities and better cycle stability than anodically deposited manganese dioxide. The cyclic stability improvement could be attributed to the mesoporous structure of cathodic deposited manganese dioxide nanowall arrays. As to the high discharge capacities at large deposition thickness, the macrostructure should be the key point. Besides the large surface area and shorter diffusion path provided for lithium-ion reaction, like tube arrays, this honeycomb macroporous structure facilitated the penetration of electrolyte to the bottom of the array even when thickness was large, thus minimizing the adverse effect of large deposition thickness, i.e. difficulty of electrolyte penetration. A similar phenomenon was also found when making a comparison of the thickness effect on vanadium oxide with and without macroporous structures.252,253 There are also other reports of porous thin film electrodes made by direct electrostatic spray deposition without binder and carbon additive introduction that exhibited excellent electrochemical performance, e.g. porous Fe2O3 film,254 porous Li4Ti5O12255 and porous NiO.256
Acidic anodization has been yet another method to fabricate electrode films for lithium ion intercalation. TiO2 nanotube arrays could be fabricated by anodization of Ti foil in a two-electrode electrochemical cell with platinum foil as a cathode at a constant potential at room temperature.257,258 Polished Ti foil was anodized in different types of electrolytes for 1 h to form nanotube arrays on Ti substrate. The nanotube diameters and length could be controlled by changing the anodization voltage and time. The as-fabricated TiO2 nanotube arrays exhibited excellent intercalation properties after appropriate heat treatment in nitrogen or carbon monoxide gas as discussed in section 3.2 and 4.2.
This journal is © The Royal Society of Chemistry 2010 |