Updates on the development of nanostructured transition metal nitrides for electrochemical energy storage and water splitting Academic research paper on "Materials engineering"

Materials Today • Volume 00, Number 00• May 2017

RESEARCH

ELSEVIER

Updates on the development of nanostructured transition metal nitrides for electrochemical energy storage and water splitting

Muhammad-Sadeeq Balogun1, Yongchao Huang1, Weitao Qiu, Hao Yang, Hongbing Ji* and Yexiang Tong*

MOE of the Key Laboratory of Bioinorganic and Synthetic Chemistry, KLGHEI of Environment and Energy Chemistry, The Key Lab of Low-carbon Chem & Energy Conservation of Guangdong Province, School of Chemistry, Sun Yat-Sen University, 135 Xingang West Road, Guangzhou 510275, China

There are wide interests in developing high-performance electrode materials for electrochemical energy storage and conversion devices. Among them, transition metal nitrides (TMNs) are suitable for a wide range of devices because they have better electrical conductivity than the oxides and excellent catalytic properties. In particular, properly designed nanostructured TMNs offer additional advantages for performance enhancement. However, reviews of the rapid utilization of metal nitrides as electrode materials are still not much. In this mini-review, we present a recent (mostly since 2015) update on nanostructured TMNs as high-performance electrode materials for energy storage devices and water splitting; we discussed how a judicious nanostructure design will lead to improving performance in lithium ion battery, supercapacitor and Li-ion capacitor, as well as in electrochemical water splitting (oxygen and hydrogen evolution reactions). Knowledge about this review on metal nitrides is aimed at sharing a wide view in recent TMNs synthetic development, applications, prospects and challenges.

Introduction

Among the key issues of life, energy storage and conversion are considered to be of high significance [1-3], because most of the primary energy sources need energy storage and conversion devices to convert them to secondary sources for daily life [4,5]. Recently, the energy storage devices basically include the super-capacitors (SCs), lithium ion batteries (LIBs) and sodium ion batteries (SIBs) [6-11] while the energy conversion systems are accompany with series of electrochemical reactions such as the hydrogen-oxygen fuel cell and the electrolytic cell [12-14]. The effective usage of these energy systems is characterized with one major parameter, namely electrode materials [15-18]. Thus, researchers' headache on the fuel cell is how to effectively catalyze the reactions on each electrode surface to achieve the lowest overpotential and highest current density possible for the energy conversion system [12,19], while the obstacle on electrochemical

*Corresponding authors: Balogun, M.-S. (balogun@mail2.sysu.edu.cn), Ji, H. (jihb@mail.sysu.edu.cn), Tong, Y. (chedhx@mail.sysu.edu.cn)

1 These authors contributed equally to this work.

energy storage (EES) systems remains how to develop active and advanced electrode materials that can store enough energy in order to achieve high energy and power densities coupled with long life span [7].

Till date, different electrode materials with different morphologies have been developed for the advancement of EES and fuel cells with high efficiency [16,20-22]. Apart from the traditional carbon materials commonly used for EES [23-25] and expensive and unreliable platinum (Pt) electrocatalysts for fuel cells [26-28], other electrode materials such as the transition metal oxides [2931], carbides [32,33], sulfides [34-36], nitrides [32,33,37] and alloy metals [38,39] have been reported and impressive progress has been made. Each electrode material exhibited worthy advantages and disadvantages and those of the EES systems have been discussed in previous works. Compared to other materials, transition metal nitrides (TMNs) have attracted enormous attention as a result of their high chemical stability, such as high resistance against corrosion, high melting points, microhardness, density and low electrical resistance [40-45]. The properties of some

1369-7021/© 2017 Published by Elsevier Ltd. http://dx.doi.org/10.1016/j.mattod.2017.03.019

RESEARCH Materials Today • Volume 00, Number 00• May 2017

TABLE 1

Comparative properties of some selected TMNs and their corresponding oxides.

Materials Melting point Electrical resistivity (105 cm) Density (g cm 3) Bandgap (eV) Young's modulus (GPa) Crystal structure Solubility Abundance of corresponding metal

TiN 3050 2.7 5.39 3.4 [252] 280.8 [253] Cubic [100] Insoluble in water 0.63%

TiO2 1830 [254] ^102 4.23 3.05 [255] 348 [256] Tetragonal Insoluble in water 0.63%

VN 2350 6.5 6.13 - 344 [253] Cubic [169] Insoluble in water 0.25%

V2O5 690 - 3.35 2.6 [257] 4.8 [258] Orthorhombic 0.8 g/L (20 °C) 0.25%

CoN - 0.5-4.5 [231] 6.18 [259] Metallic [239] 485 [253] Cubic [260] Insoluble in water 0.0029%

Co3O4 895 150 [261] 6.11 1.96 [262] 116-163 [263] Cubic Insoluble in water 0.0029%

Ni3N 680 and 700 10 [230,264] 7.66 [265] Metallic [230] 399 [253] Hexagonal [232] Insoluble in water 0.021%

NiO 1955 90 [266] 6.67 3.4-3.8 [266] 205-315 [263] Cubic Rhombohedral Negligible but soluble in KCN 0.021%

NbN 2573 6 8.47 - 360 Cubic [82] Reacts to form NH3 0.002%

Nb2O5 1512 50 [267] 4.60 3.41 [268] ~60 [269] Monoclinic Orthorhombic Tetragonal Insoluble in water but soluble in HF 0.002%

CrN 1770 64 5.9 Weak metallic [270] 0.5-0.7 [271] 503 [253] Cubic Insoluble in water 0.01%

Cr2O3 2435 167 [272] 5.22 4.0 [273] 286 [274] Rhombohedral Hexagonal Insoluble in water, ethanol, acetone and acid 0.01%

selected TMNs can be seen in Table 1. These properties were compared with their corresponding oxides since transition metal oxides (TMOs) are also promising and challenging electrode materials.

Two years ago, we published a review article summarizing research advancement on metal nitrides for energy storage devices [37]. The previous work only focuses on metal nitrides as electrode materials for lithium ion batteries (LIBs) and supercapacitors (SCs). However, research on the metal nitrides (especially transition metal nitrides) as suitable and active electrodes for energy storage and conversion systems is now becoming rampant. Xia etal. have recently discussed the transition metal nitrides (TMNs) and carbides (TMCs) in energy storage and conversion but less attention was focused on the metal nitrides unlike the carbides [32]. Meanwhile, Xie and a co-worker also addressed the recent development on TMN electrocatalysts but more attention was focused on hydrogen evolution reactions (HER) and less attention on oxygen evolution reactions (OER) [46]. As the new era of employing TMNs as electrocatalysts for HER and OER water splitting is still at the infant stage [32,43], this present review aims to be an update on our previous review [37], further describing the trending use of TMNs not only in the EES devices but also in the water splitting, basically HER and OER. Fig. 1a shows the schematic summary of TMNs' wide application discussed in this review. So far on DFT calculations and volcano plots of TMNs electrocatalysts are also highlighted. Additionally, the challenges that TMNs are facing in energy research as well as their prospects are summarized and discussed.

Nanostructured transition metal nitrides

TMNs belong to the family of interstitial compounds and their structures have been well discussed and summarized in literatures

[47-49]. Their structures and bonding enable them to exhibit high electrical conductivity [50,51]. Research works on TMNs as electrode materials for energy storage devices started in the early 90s [52] while their applications as water splitting electrocatalyst commences in late 2000 [53]. In energy storage devices, most TMNs were reported to exhibit conversion reaction with Li [37,41] coupled with fast surface redox reactions [54]. For electrochemical water splitting, the formation of metal-nitrogen bond in the density of states of the metal d-band usually results in smaller deficiency causing the metal nitrides to possess an electron donating character [55]. This could contribute to higher catalytic activity of the metal nitrides than the corresponding metals and hence, an opportunity for utilization of TMNs as electrocatalyst for water splitting [43].

Recently, in LIBs, the first-row TMNs, especially TiN, VN, Mn3N2, Fe2N, CoN and Ni3N and some other row TMNs such as nitrides of Mo, W and Nb have been widely employed as electrode materials for EES systems due to their high theoretical capacities and higher free energy of formation (G°) because higher G° suggest thermodynamically favorability of the metal nitrides towards high capacities [47,56]. Thus, the experimental capacities of these TMNs is roughly proportional to the G°? but the case remain completely different for CrN with much lower G° compared to nitrides of Fe, Co, Ni, and Mo but apparently higher capacity. This phenomenon can be related to the large polarization in the CrN system. Details will be discussed in the later section. Fig. 1b displayed the relationship between the theoretical capacities, experimental capacities and G°? of some selected TMNs. For instance, according to Fig. 1b, it was concluded that nearly the entire TMNs displayed low experimental capacity compared to the theoretical capacity. However, early TMNs such as TiN and VN could only

Materials Today • Volume 00, Number 00• May 2017

FIGURE 1

(a) Examples of several applications of nanostructured TMNs in energy storage devices and electrochemical water splitting. The summarized topics include lithium ion batteries, supercapacitors, lithium ion capacitors, oxygen and hydrogen evolution reactions. Copyright permission has been received for all the images in the figure. (b) Relationship between the theoretical capacities, experimental capacities and G° (kJ moP1) of some selected TMNs. Theoretical and Experimental capacities value Reproduced with permission [47]. Copyright 2013, Elsevier. Free Energy of Formation Reproduced with permission [56]. Copyright 1993, American Chemical Chemistry.

delivered about 10% of their theoretical capacity. Detailed analysis by Zhang et al. [57] revealed that no single transition metal were formed at the end of the discharging process, suggesting that both TiN and VN undergoes partial conversion reaction. Details about the reaction mechanism of the TMNs will also be discussed later. For TMNs with complete conversion reaction, like CrN, Fe2N, Ni3N and so on, rapid capacity decay were usually observed in the consequent electrochemical cycling, based on the results reported [37,47]. The major cause of the poor performances were related to the smaller deficiency in the d-band occupation of the metal-nitride bond formed, which could easily break during electrochemical reactions, thereby forming an oxide layer with poor conductivity [58-60]. One of the major strategies to improve the

performance of the aforementioned TMNs storage performance is reducing the size of electrode active materials into nanoscale.

Efficient electrodes should exhibit excellent tailored nanostruc-tures with high surface areas, more reaction sites, and enough contact or connecting area between reactants and electrode materials. Thus, the key target is to develop facile, effective, and environmentally friendly synthetic methods to produce electrodes with tailored nanostructures [61]. It is well known that the performance of EES systems depends on the interfacial ion transfer, motion of ion through the electrolyte and chemical diffusion accompanied by electron in the bulk of electrode material [20,62,63]. Thus, the nanomaterials become important over the microsized materials and bulk materials in achieving suitable EES systems because nanomaterials tend to shorten the diffusion length of ion in the electrode, which could attractively enhance the electrodes ionic conductivity [64-66]. Different dimensional nanostructures of the first and second row of TMNs ranging from zero dimensional (0D) to three dimensional (3D) such as nano-particles, nanowires, nanorods, nanosheets, nanocages, hollow spheres, etc. have been explored as electrode materials for EES systems with improving storage properties. Combining both the merits of excellent properties and nanostructuring advancement, nanostructured TMNs are considered suitable electrodes and catalysts for EES and water splitting systems, respectively.

Synthetic methods

The methods in the fabrication of electrode materials are also important factor considering the potential application of these electrode materials. The motive of every researcher is synthesizing electrode materials with facile approaches, scalable strategies, environmental benignity and of course considerable cost, in order to meet up with the present economic and environmental challenges. Some of the methods adopted in the synthesis of TMNs have been discussed in our previous review [37,67]. Generally, metal nitrides are often prepared by subjecting corresponding metal oxides to heating in ammonia gas [68,69], annealing the metals combined with carbon-based materials at high temperature in either nitrogen or ammonia environment [70,71], calcining along with nitrogen-based compounds [72] and some other physical approaches [33,45]. Some of these methods sometimes were counted as unfriendly to the environment in terms toxic nitrogen sources used, high temperature, likewise long period for achieving the final product in some cases and inability to produce suitable nanomaterials using some physical methods. Hence, TMNs are still facing challenges in their preparation, which demand urgent attention as a prospect for the development of metal nitrides as electrode materials for energy storage and conversion systems. Moreover, since nanotechnology become useful research area to achieve high-performance materials, controlling the morphology of some TMNs during their preparation also remains a challenge. Recently, some attempts have been made to solve the predicaments faced by maintain TMN morphologies during syntheses. For example, TiO2 was introduced on the surface of FeOOH nanorods through atomic layer deposition (ALD), and further annealing in NH3 atmosphere could result in Fe2N-Ti2N without change in the nanorod morphology [73]. Also, Fan et al. directly converted the Co3O4 to CoN by subjecting the Co3O4 to N2 RF plasma treatment at room temperature within 3 min [74]. This method is very simple

RESEARCH

and could address the issue of high temperature, lesser use of toxic nitrogen source and time consumption and can be applied to the fabrication of other metal nitrides as one of the prospects in fabricating TMNs and even other metal nitrides.

Electrochemical energy storage

As lithium ion batteries (LIBs) and supercapacitors (SCs) majorly dominate the EES systems, attention has long been paid to TMNs as electrode materials for EES devices [32,37]. The LIBs exhibit attractive energy density of about 150 Wh/kg while the SCs deliver high power density up to 10 kW/kg. Recently, hybrid of LIBs and SCs commonly referred to as lithium ion capacitors (LICs) has been developed to meet the demand of both higher energy and power densities for EES systems and power sources of hybrid-electric (HEVs) and plugin hybrid-electric vehicles (PHEVs) [75-79]. Moreover, TMN electrodes have also gained some achievements in the study of LICs [80-82]. Nitrides of transition metals such as Mo, Ti and V are very rampant in the research world, while nitrides of Fe, Ni, Mn and Zn are less attractive. Recently, due to the diverse chemical composition and different crystalline phases, Nb nitrides have been found applicable in other fields [83,84] and have also been long utilized as electrode materials in SCs, now finding their way also in LIBs and LICs. Based on these developments, we accounted for the progress of the some common TMNs especially nitrides of Ti, V, Fe, Ni, Mo, W and Nb as electrode materials for LIBs, SCs and LICs.

Lithium ion batteries

Ever since the commercialization of LIBs in 1990, its technology has become the main power sources for nearly all electronic devices and also seems to be the most reliable energy storage systems for smart grids and electric vehicles. Graphite, the LIB conventional anode with low theoretical capacity, has led to the search of new electrode materials. Generally, LIB storage mechanism can be categorized into three for anode materials, namely (i) insertion, (ii) alloying and (iii) conversion reactions [11,85]. The conventional graphite exhibits the insertion mechanism as shown in Eq. (1) [86]:

C + xLi + xe

Most of the common elements that undergo alloying and de-alloying reaction include Sn, Ge, Si, Sb, and so on. Taking Sn as an example [87], The alloying and de-alloying mechanism is also reversible and the process is shown in Eq. (2):

Sn + xLi+ + xe

LixSn(0 < x < 4.4)

The third type of reaction mechanism is conversion reaction. Most of the widely reported TMOs usually exhibit this reaction whereby the corresponding transition metal and lithium oxide (Li2O) is formed at the end of the reduction process. The typical conversion reaction mechanism is shown in Eq. (3) [86]:

MxOy + 3yLi+ + 3e~ ^ xM0 + yLi2O (3)

Upon the fact that LIB are rich in storage mechanism, each mechanism exist with few obstacles that lead to the search of new electrode materials. For example, the insertion mechanism faced the problem of low vacancies unavailable for Li ion reversible storage that account for their low capacity [7] while both the

alloying and conversion mechanisms encountered large volume expansion, low conductivity, SEI film formation and so on, which result in bad cyclic stability and poor rate capability [88]. Among the common materials that possess conversion storage mechanism, TMNs have had many applications due to their excellent conversion reaction, high capacity and good rate capability [11,37]. Research on the first-row TMNs has been more promising than other group metal nitrides owing to their considerable economical, geographical and preparation factors [32]. TMNs exhibit three types of conversion reactions, namely complete-conversion, partial-conversion and conversion-alloying reactions. The conversion-alloying reaction involves a first cathodic conversion process to the corresponding metal, followed by alloying. This was confirmed in the reaction of Li with Zn3N2 forming Li-Zn alloying after complete discharge process [89]. The other two types of reactions (complete- and partial-conversion reaction) can be related to the TMNs theoretical capacity, experimental capacity and G°? (Fig. 1b). In either type of reaction, larger capacities are observed in some TMNs due to the larger nitrogen content but limited by relatively poor stability upon cycling. The complete-conversion reaction is similar to those of TMOs, in which Li3N and the corresponding metal is formed at the end of the discharging process as shown in Eq. (4) [90,91]:

MxNy + 3yLi+ + 3e~ ^ xM0 + yLi3N (4)

For example, nitrides of Cr, Ni, Fe, etc. exhibited the complete-conversion reaction. Theoretically, CrN exhibits the highest theoretical capacity among all TMNs with capacity above 1218 mAh g_1 [47,92,93]. The complete-conversion reaction of CrN can be seen in Eq. (5):

CrN + 3Li+ + 3e ^ Cr + Li3N

Based on Eq. (5), the CrN experimental capacity (1200 mAh g ) could reach its theoretical capacity but gallantly decreases after few cycles owing to large polarization [94]. In order to improve the fade in capacity that result from large polarization, a ternary TMNs consist of CrN and FeN was synthesized by Sun's group. With the incorporation of FeN, enhanced electrochemical activity was observed due to the lower decomposition of Li3N towards FeN than CrN. The enhancement can be attributed to lower overpotential of the FeN (<2.0 V), which decreases the polarization in the redox curves to improve the polarization phenomena of CrN [94].

Meanwhile, the experimental capacity of TiN (120 mAh g_1) and VN (140 mAh g_1) are very low compared to their theoretical capacity at 1300 and 1218 mAh g_1, respectively. This could be as a result of partial-conversion process. The partial conversion of a Li-TMN cells occur in a situation where no Li3N is formed but a lithiated metal nitride at the end of the discharge process. A typical Example is VN, which shows partial-conversion reaction as shown in the electrochemical reactions below [57];

VN + xLi+ + xe

The above mentioned problems disengaged many researchers to study the effect of high current density on the performance of TMN electrodes since their experimental capacities are quite low [95,96]. One of the major strategies that was first adopted to address the large polarization in the complete- and partial-conversion process of the some TMNs is reducing the size of the materials

Materials Today • Volume 00, Number 00• May 2017

to nanoscale [37,97] because most of the earlier reported TMN electrodes are in their bulk form. Consequently, Gillot et al. work on the first nano-sized Ni3N could delivered a high initial capacity of about 1200 mAh g_1, which is 3-fold the theoretical capacity of Ni3N. This development opened opportunity for the synthesis of more TMN nanostructures to tackle the problem of low capacity [37,98]. In our previous review, we have summarized the impact and development of nanostructured TMNs for the development of LIBs and also discussed the challenges up till 2014 [37]. Thus, details about these impact and development will not emphasized in this work. Rather and herein, we will focus on the improvement made in the application of some TMN nanostructures for LIBs in the last two years.

Nitrides of titanium and vanadium

To date, series of the first-row (Ti, V, Mn, Fe, Ni,) and some of the second-row (Mo, W) TMN nanostructures have been developed. Starting from TiN, benefiting from its superior electronic resistivity and mechanical stability, TiN as anode material for LIBs have been reported, because most other developed materials apart from carbon based materials suffer from poor conductivity. However, unlike TiO2 or Li4Ti5O12, research on TiN is less attractive due to the low capacity derived from its partial-conversion reaction as well as the instability caused by pulverization after few electrochemical cycles. The partial-conversion reaction in TiN anode can be seen in Eq. (7):

TiN + xLi+ + ^ LixTiN (7)

To improve its storage mechanism, varieties of TiN nanostruc-tures such as nanoparticles [99], nanowires [41] have been reported with enhanced reversibility as a result of the morphology and particle size but low capacity were still recorded. In order to boost TiN capacity, combining TiN with high theoretical capacity electrodes were developed due to the fact that such phenomenon has been employed for other TMOs such as low capacity TiO2. For example, Yousefi etal. fabricated TiN/graphene composite (TiN/G) via the mixture of TiCl4 and NaN3 in benzene and then annealed NH3 gas at 1000 °C [100]. An initial capacity of 350 mAh g^1 was delivered owing to capacity contribution from the graphene. The TiN/G composite delivered a capacity of 48 mAh g_1 at a current density of 7.7 C (10 A g_1) with attractive cyclic stability up to 250 cycles at different current densities. Interestingly, both the cyclic and rate performance of the TiN/G composite can be compared to some of the well-established TiO2 anodes at the same testing condition [101].

Similar to TiN, VN also exhibited partial-conversion reaction as displayed in Eq. (6). Various VN nanostructures have been synthesized by different methods. Compare to previous VN thin film, it has been established that electrochemical performances of the VN depends on the morphology and particle size. A typical example is the recent work of Cao's group, in which VN hollow spheres were assembled from porous VN nanosheets by a facile and scalable carbon spheres-template assisted strategy as shown in Fig. 2a [102]. The hollow structure with high porosity and surface area can be seen in Fig. 2b. Impressively, a capacity retention of 100% after 40 cycles was obtained with a capacity of 720 mAh g_1 at a current density of 0.05 Ag_1, while the bulk one could only achieved a capacity of 300 mAh g_1 (Fig. 2c). The VN hollow spheres could

deliver a reversible rate capacity of 400 mAh g_1 after 80 cycles at a current density 2 A g_1, while its bulk counterpart retains about 100 mAh g_1 at the same testing conditions (Fig. 2d). Even at high current density of 4 A g~1, the VN hollow spheres still displayed his 100% capacity retention after 100 electrochemical cycles (Supplementary Information - Fig. S1). It is indisputable that the porous hollow structure is particularly beneficial for the mass transfer and offers more active sites for Li+ accommodation. The lithiation kinetics and charge transfer in VN nanostructure were attributed to the stable solid-electrolyte interface (SEI) film formed during the initial discharging-charging cycles, which in-turn served as the buffer for the volume change during discharging-charging

[103]. Other VN nanostructures such as VN nanowires [139] shows superior performance when compare to the corresponding oxide at the same testing condition, especially vanadium oxide (V2O3)

[104] (Table 2), because most other vanadium oxides such as V2O5, VO2, V6O13 are LIB cathode materials. Additionally, the synthesis of different morphologies is also effective in improving the lithium storage properties of Fe and Ni.

Nitrides of iron and nickel

Compared with TiN and VN, Fe and Ni displayed lower theoretical capacity, but exhibit complete-conversion reaction. The mechanism of lithium insertion in nitrides of Fe and Ni has been previously studied [90,105]. Irrespective of the valence state of Fe in Fe3N, Fe2N or FeN, a complete conversion reaction usually occurs with reasonably reversibility [105]. Fu et al. utilized X-ray diffractometry (XRD) and X-ray photoelectron spectroscopy (XPS) to study the lithium behaviour of Fe3N [105]. The XRD spectra confirmed the presence of metallic Fe (110) but the Li3N could not be detected due to small particle size and the atomic scattering factors of Li and N. However, ex situ XPS study could affirm the presence of the Li-N bond after the end of the cathodic reaction. Indeed, similar reaction was also found in Ni3N owing to the

TABLE 2

Comparison and summary of the recently reported TMNs and their corresponding oxides at nearly the same applied current and potential range of 0.01-3.0 V: composites, morphologies and lithium storage performances.

Materials Morphology Current (mA) Rate performance (mAh g_1)

VN [139] Nanowires 1.9 803

V2O3 [104] Microspheres 1.5 310

Fe2N [109] Hollow nanofibers 2.0 395

Fe2O3 [109] Hollow nanofibers 2.0 350

Fe3O4 [109] Hollow nanofibers 2.0 <100

MoN [119] Nanoparticles 1.35 ~600

MoO2 [119] Nanoparticles 1.35 ~300

MoN [123] Nanochexes 3.9 ~600

MoO2 [122] Hollow nanostructures 4.0 ~109

WN [124] Nanoplates 1.0 124

WO3 [275] Nanoplates 0.96 78

Ni3N [69] Nanosheets 1.8 550

NiO [276] Nanosheets 1.7 600

RESEARCH

FIGURE 2

(a) Schematic illustration of the preparation process of VN HSs and applications in lithium storage and the oxygen reduction reaction. (b) TEM image ofVN HSs. (c) Cycling performance of VN HSs and bulk VN at a current density of 50 mA g-1 and (d) rate capability of VN HSs and bulk VN at different rates at a voltage range of 0.01-3.5 V. Reproduced with permission [102]. Copyright 2016, Royal Society of Chemistry. (e) TEM image of Fe2N NPs. Inset is the SAED pattern. (f) Rate capability profiles of the Fe2N electrodes and (g) Cycling profiles of the Fe2N electrodes at 6000 mAg-1 up to 300 cycles at a voltage range of 0.01-3.0 V. Reproduced with permission [68]. Copyright 2015, Elsevier.

similar valence states of Fe3N and Ni3N. The redox reaction that occurs in the Fe3N and Ni3N are displayed in Eqs. (8) and (9), respectively;

ForFe3N, Fe3N + 3Li++ xe" ^ 3Fe + Li3N (8) ForNi3N, Ni3N + 3Li+ + xe" ^ 3Ni + Li3N (9)

Meanwhile, with respect to their lithium storage properties, less attention was given to Fe3N due to their low theoretical capacity (421 mAh g"1) after the first work by Fu et al. in 2004 [105]. Recently, Fe3N nanoparticles coated will carbon (Fe3N@C) was reported by Huang and Co-workers [58]. The initial capacity of the Fe3N@C anode attained 676 mAh g"1, with impressive cyclic stability up to 500 cycles at 0.1 A g"1 as well as attractive rate

Materials Today • Volume 00, Number 00• May 2017

performance of 244mAhg~1 (1Ag_1). The performance of the Fe3N@C outperformed those of Fe@C and base Fe3N. The main reason for the enhanced performance was due to charge transportation and dual-capacitive interface of the carbon shell and the enhanced surface area of the nanoparticles. Nevertheless, the storage capacity of the Fe3N still call for improvement when compare to other Fe-based nanostructured anodes [106-108]. Moreover, the synthetic methods of Fe3N seem to be more complex during nitridation process [58].

Unlike Fe3N, Fe2N possesses higher theoretical capacity (about 900 mAh g-1), which can be easily obtained by direct annealing of Fe-based oxide in NH3 [68,109,110] and have wide application in other field of studies such as sodium ion batteries [110] and oxygen reduction reaction [111,112]. Moreover, different Fe2N morphologies have been fabricated with enhanced storage properties [109,110]. For instance, Dong's group prepared Fe2N hollow nano-fibers by thermal treatment of Fe2O3 hollow nanofibers in NH3 gas [109]. It showed an initial capacity of 750 mAh g_1 at 0.1 A g_1, considerable cyclic performance (438 mAh g_1 after 300 cycles at 0.1 A g_1) and enhanced rate capability (395 mAh g_1 at 1 A g_1). The rate performance of the Fe2N hollow nanofibers are better than those of Fe2O3 and Fe3O4 hollow nanofibers (studied under the same testing condition) (Table 2) but the stability test still require further improvement. In this regard, some researchers synthesized composites of Fe2N with carbon based materials in order to stabilize or reduce the capacity fade during electrochemical reactions. A typical example is the synthesis of Fe2N nanopar-ticles on three-dimensional carbon cloth (CC) by simple hydrothermal reaction and post-NH3 calcination process at different temperature reported by our group [68]. Fig. S2a displayed the synthetic route for the 3D Fe2N nanoparticles-CC composites. The nanoparticle morphology was well characterized in Fig. S2b and S2c with pure Fe2N single crystalline structure, which was confirmed by the XRD spectra as well (Fig. S2d). Low magnified transmission electron microscopy (TEM) image of the Fe2N confirmed its nanoparticle nature (Fig. 2e), likewise selected-area electron diffraction (SAED) pattern in the inset of Fig. 2e also affirmed the single crystalline nature of the Fe2N nanoparticles. Among the different Fe2N nanoparticle composites obtained from different annealing temperature, the Fe2N annealed at 600 °C (Fe2N-600) electrode exhibited the highest initial discharge capacity of 900 mAh g^1, best rate capability of 243 mAh g_1 at 6 A g_1 (Fig. 2f) and excellent cyclic stability up to 300 cycles at a current density of 6 A g_1 (Fig. 2g). The cyclic stability of was further improved by engineering Fe2N with larger surface area morphology and higher conductive materials. As a result, Yu etal. fabricated a novel composite that composed of Fe2N coated carbon microspheres grown of reduced graphite oxide (Fe2N@C-RGO) [113]. The Fe2N@C-RGO microsphere exhibited a good rate capability up to 1 A g_1 (Fig. S3a) and retained cyclic stability capacity of 760 after 100 cycles at 0.1 Ag_1 corresponding to 90% of its initial capacity (Fig. S3b). The storage properties of Fe2N@C-RGO are better than those of Fe2N@C and bare Fe2N and they are also comparable to or better than some Fe oxides and their carbon composites [114-116]. The excellent performance was concluded to be the increase the surface are of the electrode and the carbon layer support on the surface of Fe2N. Such modification could avert the structure decay caused by the volume change during the

lithiation/delithiation process. Furthermore, the existence of RGO layers on the Fe2N also facilitate conductivity enhancement and therefore shorten the lithium-ion diffusion path.

Like other TMNs, nitrides of Ni also exhibit poor cyclic stability and rate capability. For instance, nitrogen-doped graphene was used to improve the cyclic stability of NiN up to 80 cycles with 100% capacity retention at low current density of 50mAhg~1 [117]. In our recent work, we utilized the 3D CC support for the growth of Ni3N nanosheets in order to improve its storage properties [69]. Interestingly, a high initial capacity of 593 mAh g_1 was obtained at a current density of 0.4 A g_1, retaining 304 mAh g_1 when its current density was increased to 4.2 A g_1. Owing to the porous and thin nature of the nanosheets, the Ni3N-CC thin nanosheets remain the best performance ever for Ni3N-based LIB anode. However, compared to their oxides of Ni, the Ni3N nanostructure still needs further improvement (Table 2). There are less reports on nickel nitride as anode material for LIBs. Nevertheless, other reports have shown that the poor stability and lower rate capability of the TMNs can be enhanced by coupling with carbon-based materials. To justify the enhanced stability of the TMNs, when other TMNs such as FeN and CoN combined with graphene [117], Mo2N-N-doped graphene composites [95], VN combined with carbon [118], MoN combined with CNTs [119], they all accomplished excellent lithium storage properties with improved cyclic and rate performance over their corresponding pristine counterparts [120]. The enhanced performance could be attributed to rapid charge transfer resistance [119], favoured kinetics during Li insertion [118] and improved in ionic mixed conducting network of the electrodes [95,117]. Such strategy in improving the storage performance of the above mentioned TMNs could be extended to nickel nitrides. As a matter of fact, we concluded that the effects of surface area on the performance of electrode materials are crucial and dependent on the structural dimensions, morphologies, electrical conductivity and composite formation.

Nitrides of molybdenum and tungsten

Besides the first-row TMNs, Mo and W as part of the environmental friendly elements in the second and third rows transition metals have recently and widely explored as anode materials for LIBs. Precisely, the syntheses of different Mo and W nitrides nanostruc-ture have also been reported due to the fact that nanostructured materials are expected to exhibit higher surface area, which could improve the performance of electrode materials. Various reports on Mo and W nitrides with different dimensional nanostructures were depicted in Fig. 3. For example, one dimensional (1D) Mo2N nanobelts (Fig. 3b) was prepared by nitrification of MoO3 nano-belts. The Mo2N nanobelts displayed high specific capacity and high rate cycling performance up to 4 A g_1 over the commercial Mo2N [121] and even the corresponding metal oxides, MoO2 nanoparticles (Table 2) [122]. In another example, Xu etal. fabricated MoN nanochexes (Fig. 3c) from single-crystal 1D nanowires [123]. The attractive performance of the resulting rate capability over the MoN nanoparticles was due to low transport resistance and high lithium ion diffusion coefficient of the nanochexes. Similar to Mo2N and MoN, better lithium storage performance was also reported for 2D WN nanoplates with good reversible capacity and rate capability over the commercial WN. Likewise,

MATTOD-906; No of Pages 27 ARTICLE IN PRESS

RESEARCH Materials Today • Volume 00, Number 00• May 2017

(a) SEM image of MoN nanoparticles. Reproduced with permission [119]. Copyright 2016, Elsevier. (b) SEM image of the mesoporous Mo2N nanobelts. Reproduced with permission [121]. Copyright 2014, Elsevier. (c) SEM image of the MoN nanochex. Reproduced with permission [123]. Copyright 2015, American Chemical Chemistry. (d) TEM image of the WN nanoplate. Reproduced with permission [124]. Copyright 2016, Elsevier.

WN nanowires that were grown on a CC delivered an initial capacity of 418mAhg_1 at a current density of 0.2Ag_1 [96]. Compared to other WN electrodes, the capacities delivered by the 3D WN nanowires seem low. This result was related to the high mass loading (about 8 mg cm~2) of the WN nanowires. Upon high mass loading, the capacity was still comparable to that of conventional graphite (372 Ah g_1). We arrived at a conclusion that there existing rapid transport of ions and electrons, low electrical resistance and high diffusion coefficient for lithium ions in the TMN nanostructures compared to the bulk ones [124].

TMNs as supportive materials

The good electrical conductivity of the TMNs has actually expanded their application beyond using them as anode materials for LIBs. They have been used as backbone materials in lithium-sulfur batteries to reduce the dissolution of sulfur into electrolytes [125,126], and also to improve the conductivity of LIB cathode

materials [127,128]. As discussed in the prospect and challenges of our previous review, coating of metal oxides or other electrodes with metal nitrides are now becoming rampant in the research industries [37]. The method adapted in synthesizing these composites is basically low temperature NH3 treatment [129,130] or high temperature mixed gas treatment [131,132] of the corresponding oxides. Among the commonly available TMNs, TiN has been largely employed as support for other LIB anode materials. Recent development in the fabrication of other various anode material nanostructures created more opportunities for the utilization of TiN as support or backbone. For instance, benefiting from the excellent conductivity of the TiN, our previous work proved that TiN coated on the surface of TiO2 (TiO2@TIN) nanowires improved the electrical conductivity of the TiO2. At a high current density of 10 A g"1, the TiO2@TIN anode delivered excellent rate capability of 136 mAh g"1, and also higher than that the pristine TiO2 nanowires counterpart (24 mAh g"1) [131]. Similar work and

Materials Today • Volume 00, Number 00• May 2017

related electrochemical performance was also confirmed by Kure-Chu et al. for porous anodic TiO2-TiN composites [129].

On the other hand, the case of TMNs forming composites with other materials, which are not the corresponding oxides, also provide more stable support and better conductive pathways to the other materials [133,134]. The first report was from Kim etal. in which TiN was used to suppress the drastic volume change of Si [135]. Such strategy is now promising for reducing the high volume expansion of high capacity alloy materials such as Si and Ge [136,137]. Figs S4 and S5 showed figures of some of the recently reported TiN-Alloy material nanocomposites with TiN serving as substrate for the deposition of Si nanorods. Uniform coating of the Si with TiN was observed in the energy dispersive X-ray spectroscopy (EDS) data of the TiN support Si nanorods (TiN@Si NRs) (Fig. S4a) [137]. As expected, the TiN@Si NRs exhibited good cyclic stability and rate capability, which outperformed that of TiO2 support (TiO2@Si NRs) (Fig. S4b and S4c) [137]. In another example, with the deposition of TiN/Ti and Ge on Si

substrate, a 3D multi-layered Si-TiN/Ti-Ge nanocomposites were successfully fabricated (Inset in Fig. S5) [138]. This unique multilayer 3D Si-TiN/Ti-Ge (STTG) nanocomposite anode displayed superior cycling stability over the previous work on double layer nanocomposite (Fig. S5) [137]. Furthermore, an enhanced conductivity of the TiN/Ti/Si nanorod arrays over the bare Si nanorods was also reported [134]. It can be concluded that the TiN holds a great contribution in suppressing the large volume change of the allow materials.

Besides TiN, significant improvement was also observed for other TMNs, such as VN, used as a back-up owing to its good conductivity. Our group recently demonstrated the coating of SnS2 nanosheets on VN nanowires [139]. VN are less reported as supporting TMN for other electrode materials unlike TiN. The VN nanowires could serve as a support for easy flexibility of the SnS2 nanosheets as seen in Fig. 4a and b. Compared to the pristine VN nanowires and SnS2 nanosheets, the VN@SnS2 nanocomposites displayed a good cyclic stability with 102% capacity retention after

FIGURE 4

(a) TEM image of the VN@SnS2 and (inset) corresponding SAED pattern. (b) HRTEM image of the VN@SnS2 and (insets) well-resolved lattice spacings of the VN and SnS2. (c) Cycling profiles up to 100 cycles at a current density of 0.65 Ag_1 and (d) rate capability profiles of the electrodes at different discharge and charge rates at a voltage range of 0.01-3.0 V. Reproduced with permission [139]. Copyright 2015, American Chemical Chemistry.

RESEARCH

100 cycles (Fig. 4c) and a rate capability of 349 mAh g"1 at 13 A g"1 (Fig. 4d). The excellent performance can be attributed to the conductive VN substrate support that provided short Li-ion and electron pathways, accommodated large volume variation, and also contributed to the capacity and structural stability. These reports offer opportunities for the development of other metal sulfide anodes, other anode materials for LIBs and likewise in lithium sulfur batteries [125,126].

In short, TMN nanostructures in their nano-size form were superior to the bulk ones, which can be attributed to high specific surface areas that is beneficial for insertion of Li ion and stress reduction. Upon the development of different nanostructures to achieve high performance LIBs, the problem of poor cyclic stability cannot be overlooked. To rectify this problem, TMNs formed composites with other conductive materials especially carbon-based materials, which is sufficient to reduce pulverization and polarization problems of the TMNs. Finally, TMNs are good

supportive materials for protecting the active layer of other materials and enhancing the Li+ diffusion pathway, thereby achieved good rate performance. The lithium storage properties of some TMNs nanostructures have shown superior performance as a replacement for the conventional graphite anode (Fig. S6). Details on the morphologies, electrochemical performances of TMN composites and comparison with some corresponding oxides and single metal nitrides under the same testing conditions as LIB anodes are displayed in Table 3.

Transition metal oxynitrides (TMONs) and other metal nitride anodes

Resulting from the positive effect of oxide layer on the surface of metal nitrides, oxygen vacancies and surface defects in most TMOs and other electrode materials, researches on the development of TMONs has been facing some promising improvement as LIB anodes [140]. Most TMONs can be prepared by annealing the

TABLE 3

Comparison and summary of the recently reported TMNs composites and their corresponding oxides and single nitrides at nearly the same applied current and potential range of 0.01-3.0 V: composites, morphologies and lithium storage performances.

Materials Morphology Cyclic performance (mAh g"1) Rate performance (mAh g 1)

TiN@C [99] Nanoparticles 76 after 200 cycles at 0.05 A g"1 56 at 2.5 A g"1

TiN [99] Nanoparticles 54 after 200 cycles at 0.05 A g"1 90 at 2.5 A g"1

Fe2N-CC [68] Nanoparticles 225 after 300 cycles at 6 A g"1 243 at 6 A g"1

Fe2N powder [68] Nanoparticles 59 after 300 cycles at 6 A g"1 243 at 6 A g"1

Fe2N@C-RGO [113] Microspheres 760 after 100 cycles at 0.1 Ag"1 510 at 1 Ag"1

Fe2N@C [113] Microspheres ~575 after 100 cycles at 0.1 Ag"1 ~450 at 1 A g"1

Fe2N [113] Microspheres <100 after 100 cycles at 0.1 Ag"1 <175 at 0.3 A g"1

FeN/N-doped GO [117] Nanoparticles 698 after 50 cycles at 0.05 A g"1 510 at 0.1 Ag"1

CoN/N-doped GO [117] Nanoparticles 667 after 80 cycles at 0.05 A g"1 -

NiN/N-doped GO [117] Nanoparticles 730 after 80 cycles at 0.05 A g"1 -

FeCoN/N-doped GO [117] Nanoparticles 520 after 60 cycles at 0.05 A g"1 -

Crystalline VN-C [118] Nanoparticles 320 after 100 cycles at 0.4 A g"1 200 at 0.4 A g"1

Nanocrystalline VN [118] Nanoparticles 108 after 100 cycles at 0.4 A g"1 -

MoN/CNT [119] Nanoparticles 1232 after 100 cycles at 0.1 Ag"1 989.5 at 1.5 A g"1

MoN [119] Nanoparticles 708 after 100 cycles at 0.1 Ag"1 ~600 at 1.5 A g"1

Porous TiN/N-doped Carbon [120] Nanospheres 310 after 400 cycles at 2 A g"1 281 at 2 A g"1

Porous TiO2/N-doped Carbon [120] Nanospheres - ~120 at 2 Ag"1

TiO2@TiN [131] Nanowires 203 after 650 cycles (10 C) 126 at 30 C

TiO2 [131] Nanowires 116 after 650 cycles (10 C) ~30 at 30 C

MnO@Mn3N2/C [130] Nanocomposite 626 after 60 cycles (0.1 Ag"1) ~300 at 2 Ag"1

Mn3N2/C [130] Nanocomposite 329 after 60 cycles (0.1 Ag"1) ~100 at 2 Ag"1

MnO/C [130] Nanocomposite 261 after 60 cycles (0.1 Ag"1) ~150 at 2 Ag"1

TiNb2O7-Ti1 _xNbxN [132] Nanofibers 174 after 500 cycles (5 C) 184 at 100 C

TiNb2O7 [132] Nanofibers 170 after 500 cycles (5 C) 172 at 100 C

TiN@Si [137] Nanorods 3259 after 200 cycles (1 Ag"1) 2257 at 10 A g"1

TiO2@Si [137] Nanorods 2460 after 200 cycles (1 Ag"1) ~1250 at 10 A g"1

VN@SnS2 [139] Nanosheets 791 after 100 cycles (0.65 A g"1) 349 at 13 A g"1

SnS2 [139] Nanosheets 270 after 100 cycles (0.65 A g"1) 117 at 13 A g"1

CC@TiOxNy@SnS2 [142] Nanowire-nanosheet composites 612 after 100 cycles (0.65 A g"1) 219 at 12.9 A g"1

CC@SnS2 [142] Nanosheets 480 after 100 cycles (0.65 A g"1) 100 at 12.9 A g"1

Materials Today • Volume 00, Number 00• May 2017

TMOs in NH3 which are temperature and time dependent [131]. Recently, Ti and Mo oxynitrides were used as anode material for LIBs with impressive performance. For example, the size of the titanium oxynitride (TiOxNy) film (denoted as N35) derived from TiO2 nanospindles (denoted as N0) (Fig. 5a) reduces with increasing oxynitride content formation (Fig. 5b). After 300 cycles, the TiOxNy thin-film could delivered 150 mAh cm-2 at current density of 125 mA cm-2, which is higher than that of TiO2 thin film at ~70 mAh cm-2 at the same current density (Fig. 5c) [141]. The high performance can be attributed to decrease in the nanostructure sizes and enhanced conductivity from the oxygen vacant sites in the TiOxNy electrode. Additionally, benefiting from the oxygen vacant sites and enhanced conductivity of titanium oxynitride, the continuous deteriorating performance of SnS2 nanosheets was significantly improved by designing a TiOxNy nanowires coated with SnS2 nanosheets (CC@TiOxNy@SnS2) [142]. XPS survey spectra confirmed the presence of Sn, S, Ti, O and N in the nanocom-posite (Fig. 5d), likewise the characteristic redox peaks of both

TiOxNy and SnS2 can be identified in the cyclic voltammetry (CV) curve of both the CC@TiOxNy@SnS2 nanocomposite (Fig. 5e). The nanocomposite displayed stable cyclic stability (612 mAh g-1 after 100 cycles at 0.65 A g-1) than the pristine CC-SnS2 nanosheets (480 mAh g-1 after 100 cycles 0.65 A g-1) (Fig. 5f). Furthermore, molybdenum oxynitride (MoO2.31N0.24) nanoparticles that was derived by temperature programmed reaction between bulk MoO3 and NH3 also delivered a good rate performance than the bulk MoO3 (Fig. 5g) [143]. Nuclear Magnetic Resonance (NMR) spectroscopy analysis that was used in probing the lithium insertion/extraction mechanisms confirmed the presence of Li2O and Li3N at discharge voltage of 0.005 V indicating that the MoO2. 31N0.24 anode exhibited the conversion reaction of both metal oxides and nitrides, respectively (Fig. 5h). Higher rate properties was achieved as Mo oxynitride formed composites with higher conductive graphene [144]. Till date, only few metal TMONs and their composites have been reported as anode material for LIBs. Hence, the recent progresses on TMONs could create

FIGURE 5

Properties of TMONs. (a) SEM micrographs of TiO2 nanospindles and (b) TiOxNy thin films. (c) Cycle life tests of TiO2 and TiOxNy thin film electrodes at a current density of 125 mA cm-2 at voltage range of 0.05-3.0 V Reproduced with permission [141]. Copyright 2015, Elsevier. (d) XPS survey of TiN and CC@TiOxNy@SnS2. (e) CV curves of CC@TiOxNy@SnS2 electrode between 0.01 and 3.0 Vat a scanning rate of 0.1 mVs-1. (f) Cyclic performance of CC@TiOxNy@SnS2, CC@TiN, CC@SnS2 and SnS2 electrode at a current density of 0.65 A g-1 between voltage range of 0.01-3.0 V. Reproduced with permission [142]. Copyright 2016, Royal Society of Chemistry. (g) Cycling behaviour in the range of 0.005-3 V vs. Li+/Li at different current densities. (h) 7Li MAS NMR spectra (black solid lines) of the MoO2 31N024 which was lithiated to 0.005 V with a simulation (red dashed line) decomposed into each individual component (blue solid lines), as well as commercial Li2O and Li3N. Reproduced with permission [143]. Copyright 2014, Royal Society of Chemistry.

RESEARCH

more opportunity for the development of TMONs-based LIB anodes.

Besides the TMNs and TMONs, researches have been carried on other metal nitrides such as silicon nitrides [145-147], tin nitride [148], ruthenium nitride [149], boron nitrides [150-152], lithium nitrides [153], carbon nitrides [154-157], lithiated metal nitrides [158,159], oxynitrides [141,160], and so on. Moreover, promising development has been made. However, some of these nitrides face the challenges of cost in terms of production and fabrication [145,149,160], complexity in their preparation [146,148], requirement of much higher temperature for annealing [145] and so on.

Supercapacitors

Supercapacitors are energy storage devices that could bridge the low energy density of the traditional capacitors and the low power density of LIBs. The storage mechanism in the SCs is either (i) electrochemical double layer capacitors (EDLCs) or (ii) pseudoca-pacitors. The main difference between these two storage mechanisms is that the energy stored in the EDLCs comes from the electrode and electrolyte interface, while the pseudocapacitors exhibit a reversible faradaic redox process coupled with charge transfer [23]. Due to the non-faradaic reaction of the EDLCs, the capacitance of the pseudocapacitor electrodes is usually higher than that of EDLCs [15]. Thus, TMOs, TMNs and conducting polymers that undergoes pseudocapacitor storage mechanism usually display higher capacitance than the EDLC electrodes, basically carbon-based materials [4]. More importantly, these electrodes in their nanoscale form have been reported to show enhanced electrochemical performance than their bulk counterparts [161,162].

Limited to the scope of this review, the utilization of TMNs as electrode materials for SCs has been in existence since 1998 when examining the electrochemical charging and discharging behaviour of several MoxN films which serve as a possible substitute for RuO2 electrodes [163]. From then till now, a lot of advancement has been made and some of these advancements have been summarized in our previous review [37], by Cui's group [47] and also by Fan etal. [32]. The majority of the reports about TMNs in most reviews include nitrides of Ti, V, Mo, and W. Less attention were focus on other TMNs such as Ni, Fe and Nb nitrides. In the last two years, there have been some reports on Ni3N, Fe2N and NbN nanostructures with significant advancements. We will deeply discuss the nitrides of Ni, Fe and Nb in the next section.

Nickel nitride (Ni3N)

Impressive consideration has been given to Ni3N in recent times due to their cost-effectiveness, abundance, high theoretical capacitance as well as attractive faradaic pseudocapacitance. The synthesis of these TMNs nearly remains the same with other reported TMNs. Interestingly, Ni3N show a unique redox activity as electrode material for SCs, which allow it to deliver high specific capacitance [164]. The redox behaviour of the Ni3N nanoparticles was characterized in 6m KOH at a current rate of 20 mV s"1 and voltage window of "0.05 to 0.5 V as shown in Fig. 6a and b. The redox peaks of the electrode are wide and consist of two pairs (Fig. 6a). During the oxidation process, the anodic peak at 0.30 V corresponds to the oxidation of Ni(I) to Ni(II) and the one at 0.37 V denotes the oxidation of Ni(II) to Ni(III). The reduction peaks of

the Ni3N electrode (0.05 and 0.12 V) from Ni(III) to Ni(I) can also be observed, which confirmed the excellent faradaic reversibility of Ni3N. The formation of the different valence states of Ni in Ni3N after anodic reaction was further affirmed by the Ni 2p XPS spectra of the Ni3N electrode (Fig. 6b). The redox reaction of the Ni3N electrode could be summarized as;

Ni(I) ! Ni(II) ! Ni(III) cathodicreaction Ni(III) ! Ni(II) ! Ni(II) anodicreaction

Ni3N electrodes for SCs available presently are reported in their nanoscale form. For example our group prepared Ni3N nanosheets on a CC (denoted as 3D Ni3N/CC) (Fig. 6c) [69]. The storage mechanism of the 3D Ni3N/CC can be well characterized with two broad pair of redox peaks at lower scan rates (Fig. 6d), which also corresponds to the work of Yu et al. [164]. The capacitance of the 3D nanosheets was 990 F g"1 at a scan rate of 10 mA cm"2. A capacity retention of 81% was recovered as the scan rate was increased to 40 mA cm"2 (Fig. 6e). The high rate performance of the 3D Ni3N/CC composite electrode could be attributed to the electrode architectures that demonstrated fast electron transport via direct connection and facile ion diffusion path ensuring the participation of the active materials in the ultrafast electrochemical reaction.

Iron nitride (Fe2N)

Besides Ni3N, Fe2N is also another electrode materials for SCs. However, during the synthesis of Fe2N in their nano-sized form, its morphology changes. For example, the FeOOH films coated on a graphene nanosheets changed to Fe2N nanoparticles on the gra-phene nanosheets after annealing in NH3 gas [165]. Similar situation was also observed for the preparation of Fe2N nanoparticles on a CC from FeOOH nanorods [68]. Thus, we can conclude that the synthesis of Fe2N nearly remains the same with other TMNs but the situation may be different for the Fe precursor especially when morphology requires to be preserved. Fe2N with well-designed morphology are rarely reported as electrode materials for SCs. An instance is the work on shape preserved Fe2N reported by Zhu et al. [73]. To preserve the morphology of Fe2N in Fig. 6f, Ti2N shell was coated on the Fe2N nanorod core as seen in Fig. 6g. The CV curve of the Fe2N-Ti2N electrode exhibited quasi-rectangle shape in 1 m LiCl electrolyte at a potential range of "0.8 to 0.8 V (Fig. 6h). The area capacitance of the shape-preserved Fe2N nanorod electrode delivered is higher than those of Fe2N nano-particles, Ti2N films, which automatically indicate that the performance of the Fe2N electrode greatly depends on its morphology.

Nitrides of niobium

According to the issue of Nb, niobium nitrides also have also been reported to be favourable as electrode materials for SCs when compare to their corresponding oxides. Some of the major problems of Nb nitrides are that they suffer from low potential windows and poor morphological engineering. The nanocrystal-line NbN prepared by Choi in 2011 showed narrow potential window of 0.3 V with low capacitance of 73 Fg"1 [166]. The low capacitance and narrow potential window limits its fundamental research. Recently, another nitride of niobium (Nb4N5) was reported. For example, 3D Nb4N5 nanochannels were

Materials Today • Volume 00, Number 00• May 2017

FIGURE 6

(a) CV curve of Ni3N/RGO electrode at a scan rate of 20 mVs-1, with a Pt sheet and Ag/AgCl (in saturated KCl) as counter and reference electrodes, respectively, in 6 m KOH. (b) Ni 2p3/2 XPS spectrum of Ni3N/RGO electrode materials after positive charging. Reproduced with permission [164]. Copyright 2015, Royal Society of Chemistry. (c) SEM image of the Ni3N nanosheets on the CC. (d) CV curves of the 3D Ni3N/CC. (e) Specific capacitance as a function of current density of the 3D Ni3N/CC composite electrode. Reproduced with permission [69]. Copyright 2016, Royal Society of Chemistry. (f and g) Controlled conversion of FeOOH nanorods to Fe2N as asymmetric supercapacitor electrode. (h) CV curves of ammonia-annealed carbon cloth, Fe2N particles, Ti2N thin film and FTN nanorods electrode measured in a standard three-electrode system in 1 m LiCl electrolyte. Reproduced with permission [73]. Copyright 2016, Elsevier.

fabricated directly on Nb foils by anodization and calcination in NH3. Interestingly, the voltage window reached 0.6 V owing to the high-valence state. The electrode achieved a higher capacitance of 225.8 mF cm-2 and reasonable capacity retention (70.9%) after 2000 cycles due to its high-valence state of +5, porous architecture and enhanced conductivity [167]. Such improvement in the performance of high-valence state nitride of niobium was confirmed by Gao et al. in their work on mesoporous Nb4N5 nanobelts, which could also deliver a high capacitance of 124 F g-1 [168]. In other word, to explore the potential application of Ni, Fe and Nb, their

various nanostructures should be reported as well as their application in SCs should be carried out.

Nitrides of the commonly available TMNs (V, Ti, Mo, W)

Despite the fact that lots of nanostructure have been synthesized for nitrides of V, Ti, Mo, W, the poor cyclic stability problem still remains and also limits their application as electrode materials for commercial SCs. For further improvement, researchers depends on two major strategies, which are (i) fabrication of different morphologies with various dimensional nanostructures, high surface area

RESEARCH

and porosity and (ii) composite formation with highly conductive materials. For example, 1D VN nanofibers prepared by electrostatic spinning and high temperature calcination in ammonia could deliver a superior performance (rate capability of 105.1 Fg"1 at 6 A g"1). This performance outperformed some previously reported VN nanomaterials owing to the unique nanofibrous structure and crystallinity [169]. Other commonly reported TMNs nanostructures such as VN nanoflakes [170], Mo2N nanobelts [171], nanocrystals [172], corn-Like TiNnanostructures [173], TiNnanowires [174], etc. have shown identical improved performance when compared to their bulk counterparts. Some of the recently reported TMN nanos-tructures and their electrochemical properties tested in a three-electrode configuration can be seen in Table 4. It is quite difficult to make some comparison between the TMNs and some other electrodes especially the TMOs because nearly all the mentioned TMOs and TMNs are tested in different electrolyte.

Based on composite formation, TMNs combine with some other electrodes aim at developing electrodes that will achieve the advantages of both EDLCs (high power density, long cycling life) and pseudocapacitors (high energy density) [175-177]. This could assist in improving the cyclic abilities of TMNs [178]. The synthetic methods in achieving TMNs composites vary and some of the mostly reported methods adapted include:

(i) Synthesizing the other materials first and then use the ALD or reactive DC sputtering to deposit the required TMN as shell [175,179,180]. For instance, TiN grown on vertically aligned carbon nanotubes (CNTs) by deposition of TiN on the CNT via a reactive DC sputtering exhibited a capacitance as high as 18.3 mF cm"2 at 1 Vs"1 coupled with long life-span of over 20,000 cycles [175].

(ii) Synthesizing the TMN core first and then coating or electrodepositing of the shell [177,181]. Du et al. fabricated TiN nanotubes and coated polypyrrole (PPy) on the TiN surfaces. They compared the performance of the TiN coated PPy with that of TiO2 coated PPy [181]. The electrochemical performance of the TiN-polypyrrole composites is superior to that of TiO2.

(iii) Thirdly, synthesizing both materials as mixed composites (composites usually consist of transition metal oxides, TMOs) and annealing in NH3 to achieve the TMN composites. This is commonly employed for the preparation of most graphene-TMN composites [80,164,182]. A typical example is that of mesoporous Mo2N nanobelts and reduced graphene oxide nanosheets (denoted as MMNNBs/rGO),

which shows better cyclic performance and higher rate capability than the bare Mo2N and rGO [182]. (iv) Mixing of the corresponding TMN precursor and some high nitrogen content materials followed by an annealing process in a furnace under a N2 atmosphere. This method is commonly utilized in the synthesis of TMNs-carbon composites [70,183]. More attractive and superior performances can be observed when the TMNs combined with carbon [70,183185], graphene [182,186], CNTs [175,187], conducting polymers [181,185,188], and so on [180,189]. Table 5 summarized the morphologies, synthetic methods, and electrochemical performances of most recently reported TMN composites in three-electrode configuration. The enhanced performance of these composites is associated with high surface area of the electrodes, porous structure, specific surface chemistry, improved conductivity and synergistic effect [179]. It should be noted that the composite formation strategy and employing TMNs as support for other electrode materials [179,180, 190,191] are very similar to the case of LIBs, which has been discussed above.

Besides the poor cyclic stability challenge of the TMNs as electrode materials for SCs, the electrochemical charge storage behaviour, surface chemistry effect and film morphology are other challenging factors [192,193]. Based on these factors, Kumta etal. discussed some of the parameters that affect the charge storage behaviour of VN [192]. It was concluded that VN capacitance was directly proportional to surface area and inversely proportional to particle size in which the particle size reflected as a possible difference in the surface oxidation state. The electrochemical charge storage behaviour may be different or same for various metal nitrides. For instance, Mo2N could store charge through insertion of hydrogen from aqueous acid and varying pH values as revealed by in situ X-ray absorption spectroscopy [194]. During cycling, approximately one electron per Mo from a band that was primarily Mo in character was observed. The behaviour of other TMNs may be different; thus urgent attention is required to study the electrochemical charge storage behaviour, surface chemistry effect and film morphology of other metal nitrides or develop a general method for the mechanism study.

Transition metal oxynitrides (TMON)

As discussed above based on using TMONs as anode material for LIBs, some promising improvement have also been achieved in SCs [195]. The synthetic method in both LIBs and SCs are very

TABLE 4

Summary of the recently reported single TMNs morphologies and SC storage performances in a three-electrode configurations.

Materials Morphology Potential range (V) Electrolytes Capacitance at CD

VN [169] Nanofibers "1.0 to 0.1 1 m KOH 105 F g"1 at 6 Ag"1

VN [170] Nanoflakes "1.0 to 0.1 2 m KOH 152 F g"1 at 1 Ag"1

Mo2N [171] Nanobelts "0.2 to 0.4 1 m H2SO4 160 F g"1 at 0.1 Vs"1

Mo2N [172] Nanocyrstals "0.7 to 0.1 0.5 m K2SO4 275 F g"1 at 0.002 V s"1

Ni3N [69] Nanosheets "0.05 to 0.45 1 m KOH 842 F g"1 at 0.01 Vs"1

Nb4N5 [167] Nanochannels 0.0-0.6 1 m KOH 225.8 mF cm"2 at 0.5 mA cm"2

Nb4N5 [168] Nanobelts "0.3 to 0.7 1 m H2SO4 37 mF cm"2 at 0.2 mA cm"2

CD, current density.

Materials Today • Volume 00, Number 00• May 2017

TABLE 5

Summary of the recently reported TMN composites morphologies, preparation methods and storage performances in a three-electrode configurations.

Materials Morphology Preparation methods Potential range (V) Electrolyte Capacitance at CD

TiN/CNT [175] Cauliflower C1 -0.2 to 0.5 0.5 m K2SO4 4.5 mF cm-2 at 1 Vs-1

NiCo2O4@TiN [179] Core-shell nanowires C1 -0.1 to 0.6 1 m KOH 582 mF cm-2 at 20 mA cm-2

PPy-TiN [181] Nanotubes C2 -0.2 to 0.4 1 m H2SO4 772 F g-1 at 4.5 A g-1

MnO2-TiN [177] Nanotubes C2 -0.3 to 0.3 1 m KOH 56.2 mF cm-2 at 3.75 mA cm-2

TiN@MnO2 [174] Nanowires C2 0.0-0.8 5 m LiCl 352.5 mF cm-2 at 2 mA cm-2

TiN-CNT [187] Nanoforests C1 -0.1 to 0.5 0.5 m H2SO4 81 mF cm-2 at 0.2 mA cm-2

Carbon@VN [183] Nanoparticles nanospheres C4 -1.2 to 0.0 2 m KOH 229.7 F g-1 at 1 A g-1

VN@N-doped Carbon [184] Nanowires C3 + C4 -1.2 to -0.2 6 m KOH 282 mF cm-2 at 1.44 mA cm-2

Mo2N-RGO [182] Nanobelts C3 -0.25 to 0.45 1 m H2SO4 98 mF cm-2 at 150 mA cm-2

VN/N-doped Graphene [186] Nanosheets C3 -1.2 to 0.0 6 m KOH 445 F g-1 at 1 A g-1

PANI/C/TiN [185] Nanowires C2 0.0-0.6 1 m H2SO4 1093 F g-1 at 1 A g-1

Fe2N@Ti2N [73] Nanorods C3 -1.2 to 1.2 1 m LiCl 82 F g-1 at 50 V s-1

Ni3N@RGO [164] Nanosheets C3 -0.05 to 0.45 6 m KOH 2087.5 F g-1 at 1 A g-1

VN@N-dC-1000 [70] - C4 -1.5 to 1.5 Ionic liquid 115 F g-1 at 20 mA cm-2

TiOxNi_x [197] Nanogrids C5 -0.2 to 0.6 1 m KCl 8.28 mF cm-2 at 0.1 Vs-1

TiO2 [197] Nanogrids C5 -0.2 to 0.6 1 m KCl 0.74 mF cm-2 at 0.1 Vs-1

TiOxNy [199] Nanofibers C5 0.0-0.8 2 m H2SO4 120.9 F g-1 at 1.25 A g-1

N-MoO3-x [198] Nanowires C5 -1.0 to 0.0 0.5 m H2SO4 13.5 mF cm-3 at 400 mV s-1

MoO3 [198] Nanowires C5 -1.0 to 0.0 0.5 m H2SO4 0.1 mF cm-3 at 400 mV s-1

WON [196] Nanowires C5 -1.0 to 0.0 5 m LiCl 3.35 mF cm-3 at 500 mA cm-3

CD, current density.

C1: initial synthesis of other materials followed by deposition the required TMN. C2: initial synthesis of TMN followed by deposition the required coating material. C3: synthesize of both two materials as composites followed by NH3 treatment.

C4: mixing of the corresponding TMN precursor and some high nitrogen content materials followed by an annealing process in a furnace under a N2 atmosphere. C5: annealing the TMOs in NH3 and controlling the temperature and reaction time.

similar that is TMONs can be prepared by annealing the TMOs in NH3 which are temperature and time dependent [196,197]. For example, Yu etal. reported tungsten oxynitride (WON) nanowires (prepared by annealing of WO3 nanowires at different temperature for short period of time) with impressive storage performance than the pristine WO3 and WN [196]. Similar results were reported for molybdenum oxynitride [198] and titanium oxynitrides [197,199] as well. The chemistry behind the storage of these oxynitrides has been studied by Glushenkov and co-workers [195]. Since SCs electrochemical processes mostly occur at the surface or sub-surface layers of the active electrode, Glushenkov et al. aged both WON and MoON. The results upon ageing indicated that the progressive oxidation of the samples lead to material-specific changes of the electrochemical properties. As one of the prospects, such approach in the study of oxynitrides storage mechanisms should be applied to other TMONs to build a general approach and understanding. Summary of properties of some of the recently reported TMONs can be seen in Table 5. We concluded that the excellent conductivity of the TMNs has extended their application not only to serve as support to improve the performance of other materials as many cases in LIBs discussed above but also to play some role in surface modifications, especially for metal oxides.

Lithium ion capacitors

Lithium ion capacitors (LICs) is a new type of hybrid energy storage system which have been proposed to exhibit the characteristics of both LIBs and SCs [200,201]. The aim is to obtain enhancement in both energy and power densities. This can be achieved through the fast charging rate of a SC cathode, high discharge capacity of the LIB anode, and the much wider working voltage window of the organic electrolytes [80,202]. The LIC systems use a traditional capacitor electrode as one electrode via the sorption of ions (mostly, a porous carbon cathode), and a LIB anode as the counter electrode in organic electrolytes containing Li salts [75,203]. The energy density of the LICs is usually determined by the cathode part [75,204]. The major problem LIC encounter is low energy density at high charge/discharge rate due to the mismatch between the kinetics of two electrodes and the low specific capacity of the cathode that results into huge difference in the capacity of the cathode and anode. The two common approaches to solve these predicaments also include developing new materials with improved electrochemical performance for the cathode side and utilizing pseudocapacitive materials at the anode due to the diffusion-unlimited kinetics.

As LIC cathode problem remains, TMNs play a major role as promising anode materials for LICs because of their good electrical

RESEARCH

FIGURE 7

(a) SEM image of NbN. (b) Schematic diagram of the LIC with a 4.0 V potential window using p-NbN as the anode and APDC as the cathode and (c) cycling performance of the NbN//APDC LIC at a current rate of 1.0 A g_1. Reproduced with permission [81]. Copyright 2016, Royal Society of Chemistry.

conductivity that can deliver considerably large specific capacity at high current density and their pseudocapacitive properties [205]. Among the recently reported TMNs, niobium nitride has been used as LIC negative electrode. For example, porous NbN (p-NbN) worm-hole-like pore (Fig. 7a) was obtained by direct annealing of commercial Nb2O5 in a mixture of Ar and NH3 gas [81]. When coupled with activated polyaniline derived carbon (APDC) cathode (Fig. 7b), the p-NbN//APDC LIC device exhibited a high specific capacitance of 20.7 F g_1 at high current density of 5 A g_1 and 95% capacity retention after 15,000 cycles at 1.0 A g_1 (Fig. 7c) [81]. As NbN formed composite with N-doped graphene (NbN/NG) through the traditional hydrothermal and annealing in NH3 method, the storage performance of the NbN-based LIC could be enhanced according to Cui's group [82]. The NbN/NG showed excellent cyclic stability when used as anode material for LIB, in which its LIC charge-discharge profiles showed no platform since the LIC mainly displays capacitive behaviour. Compared to Nb oxide-based LICs, the energy and power densities of NbN-based LIC is significantly larger than the corresponding oxides (Nb2O5-based LICs) [206] studied under nearly the same experimental factors such as electrolyte, mass loading and current densities as seen in Fig. S7.

The idea of forming composites with graphene was extended to VN nanowires for LICs. With the high lithium storage property of the VN-graphene anode over the pristine VN, the VN-RGO//APDC

LIC device assembled showed excellent capacity retention after 1000 cycles and high specific capacitance of 28.9 Fg-1 at high current density of 5 A g_1 as well [80]. Most of the reported V-based oxides LICs [207,208] were examined under different experimental condition. However, the result presented by the NbN and VN-based LICs shows that they are promising anode materials for LICs.

The TMNs are now rising as suitable anode for LICs due to their attractive properties mentioned above. However, researches on them are still not very much probably due to low capacity after few electrochemical cycles despite the favourable kinetics. Thus, there are still more challenges and prospects in terms of developing much more favourable and suitable both negative and positive electrodes for LICs. Such challenges can be improve by nanos-tructuring and composite formation. Benefiting from the excellent properties of the TMNs such as high resistance against corrosion, low electrical resistance and high melting points, their applications are recently not limited to energy storage devices but can be used as electrocatalysts for water splitting.

Water splitting

Splitting of water into hydrogen and oxygen is to store light or electric energy. The water splitting reactions require electron to transfer through the electrolyte/electrode interface, which requires efficient electrocatalysts for practical application. For alternative hydrogen energy implement, a facile, low-cost and

Materials Today • Volume 00, Number 00• May 2017

high-efficiency extraction of hydrogen gas is desirable [209]. Due to the enormous available quantity of sea water, one of the most straight-forward ways to achieve this is the direct electrolysis of it

[210]. Such splitting reaction is accompanied with two half reactions including HER at cathode and OER at anode. The evolution of hydrogen reaction can be chemically expressed as shown in Eq. (10) [12];

H2O ! H2 + 1/2O2 (10)

While that of oxygen evolution reaction can be seen in Eq. (11)

[211];

4OH- ! 2H2O + O2 + 4e- (11)

Unfortunately, the water splitting itself is a thermodynamically unfavourable reaction (an up-hill reaction), which requires a certain external voltage (1.23 V for ideal situation). Furthermore, high activation energy is required in practical electrode surface to successfully generate hydrogen and oxygen. Therefore, extra bias is applied to the system to overcome the energetically unfavourable nature of water splitting. Since the enthalpy change (equals to thermodynamic potential of 1.23 V) is intrinsic to the reaction [212], only activation energy, which is found to be correlated to the over-potential (h), is able to be minimized by using optimized catalysts. Therefore, intensive studies have been made to improve both the bulk and surface properties of HER or OER catalyst in hope to reduce the extra energy loss at high current density [12,14,213].

Pt and RuO2 are known as benchmark materials for HER and OER, respectively since they all have excellent onset potential for water splitting [14,44,214]. Pt, for example, has an onset potential very close to that for ideal HER (0 V vs. RHE) and a cell using such electrode is generally used as a standard to evaluate the performance of a new catalyst [215]. Superior characteristics as presented by these two benchmark electrodes, the cost of them is too high for massive production and researchers have steered their focus to the searching and development of novel non-noble metal [213,216] or even non-metal catalyst [19,211] for oxygen evolution and hydrogen evolution [12]. Fe, Co and Mn containing transition metal hydroxides/oxides/phosphates/phosphide such as FeOOH [213], NiOOH [217], CoOOH [218], MnO* [219] and phosphides [220222] have been explored or reported as high-performance HER or OER catalysts [223-225]. In this review, we intend to have a fast glance on the recent development of TMN electrocatalysts, since they possess less bulk resistance (4000-55,500 S cm-1) and also superior catalytic performance to that of traditional metal oxides [37]. It should be noted that the theoretical calculations of elec-trocatalysts performance have shown great contribution to the development of new electrocatalysts for electrochemical water splitting, as well as their nanoscale forms. The next subsection will highlight some impact of theoretical approaches to the performance of TMNs as electrocatalysts for OER and HER.

DFT calculation and volcano plot

Selecting suitable electrocatalysts to enhance the electrolytic reaction is one of the major factors for effective water splitting [226]. Interestingly, density functional theory (DFT) calculations have been demonstrated as another key development that has created prospect designing catalyst through computer-based analysis

[227]. It was reported in so many recent articles that the reaction energies, reaction barriers and related entropies that took place during catalytic reaction could be solely evaluated by DFT calculations which could also be used to determine the kinetics of the reaction processes [227,228]. Furthermore, the DFT calculations have been associated to the nanostructuring and pre- or postengineering of the electrocatalysts exposed active sites to accomplish fundamental understanding and practical catalyst design [213,226,227]]. While most of the previously reported electroca-talysts have shown comprehensive information on the relationship between the theoretically and experimentally selected electrocatalysts, DFT calculations on the theoretical and experimental results of the TMNs are still at the infant stage. Nevertheless, recent reports on TMNs electrocatalysts for HER and OER have been carried out to investigate their catalytic and excellent activities [229,230].

The DFT calculation of TMNs usually depends on the structure, Fermi level and density of states (DOS) of the electrocatalysts, which can be used to determine the metallic nature of the elec-trocatalysts [229]. The metallic nature of the TMN electrocatalysts brings excellent electrical conductivity, efficient and rapid transportation of electron between the catalyst surfaces, the current collector and the entire compound as a whole [229-231]. Recently, attention has been focused on the DFT calculation and metallic property of Ni3N. For example, Wu's work proved that the density of DOS across the Fermi level from the DFT calculation confirms the metallic nature of the Ni3N nanosheets [230]. Similar observation was reported for metallic Co4N [231] and FeNi3N [229]. Details information about the DFT calculation and the excellent catalytic activity Ni3N were also provided by Ding's group [232]. Ding's group studied the electronic structure of both bulk Ni3N and Ni3N nanosheets. Fig. 8a and b confirms that both bulk- and nano-Ni3N are metallic with different contribution of electrons near the Fermi level. It was deduced that the partial DOS of the nano-Ni3N is from the N p-orbital, which is different from that of the bulk Ni3N as seen in Fig. 8c-e. This indicate that the electrons in the p-orbitals transition metals with low scattering affinity exhibited a longer electron relaxation time for the p-electrons and in turns contributed a relatively high electrical conductivity for Ni3N nanosheets. With increasing measured temperature, the temperature-dependent resistivity of the Ni3N samples increases (Fig. 8f), which identifies the intrinsic metallic state of Ni3N samples. On the other hand, bimetallic nitride, FeNi3N was demonstrated to also show similar distinct spaces very close to the Fermi level of the Ni3N [229]. However, as adsorption energy of H2O on the catalyst is an important factor affecting the catalytic performance, the adsorption energy for H2O of the FeNi3N nanoparticles is higher and related to the DOS around the Fermi level of each element in the electrocatalysts. With the accurate determination of the absorption and binding energy from the DFT calculations (i.e. relation between the adsorption energies of HOO* vs. HO*), a reaction free energy diagrams can be analyzed, which give rise to the concept of volcano plots [233,234]. The volcano plots cover the difference between practical catalyst design and fundamental understanding

[228]. It can be obtained by plotting the catalytic activities (i.e. overpotentials) as a function of the enthalpy change for the lower-to-higher transition in the catalyst [234,235] and thus can also be termed activity-adsorption volcano plots [226]. Till date,

RESEARCH

FIGURE 8

The calculation models of Ni3N (a) bulk and (b) sheet. Calculated density of states (DOS) for Ni3N bulk and sheet. (c) Total DOS results, (d) partial DOS of single Ni and N atoms in Ni3N bulk and (e) partial DOS of single Ni and N atoms in the Ni3N sheet. (f) Temperature-dependent resistivity measured from 80 to 300 K for Ni3N nanosheets and their bulk form. Reproduced with permission [232]. Copyright 2016, Royal Society of Chemistry.

Materials Today • Volume 00, Number 00• May 2017

less reports have been found on volcano plots of TMNs. Hence, this is an important challenge and prospect for further understanding of oxygen and hydrogen evolution from TMNs.

Oxygen evolution reaction

Some traditional metal oxide OER catalysts such as NiO and FeOOH were plagued by their poor conductivity for the ongoing water cleavage reaction [213], whereas the conversion of TMOs into TMNs in ammonia flow largely attenuates the drawback [230,236]. Till date, most of the common transition elements reported for OER are nitrides of Ni and Co. In addition, these nitrides were usually prepared in nanostructured form, which could further enhance their conductivity and surface reactivity.

Nickel nitride (Ni3N)

As demonstrated by Xu et al., ultrathin metallic Ni3N nanosheets were fabricated with a relatively low-temperature reaction where nanosheet-like NiO was submitted to heat treatment in NH3 atmosphere [230]. According to the Atomic force microscopy (AFM), the nanosheets scanning heights ranged from 2.15 to 2.95 nm, indicating that the Ni3N nanosheets comprises of about 5-7 unit cells (Fig. 9a). The performance evaluation through linear sweep voltammetry (LSV) demonstrated in Fig. 9b showed that the current density given by Ni3N nanosheets obviously outperformed NiO and bulky Ni3N. Also, it is clear that the 2D Ni3N nanosheets possessed higher OER activity than NiO and bulk Ni3N, which can be confirmed by the small charge transfer resistance of the Ni3N nanosheets in Fig. 9c. This indicated that the nanoscale structure and advantage of high conductivity of metal nitrides could be very

important aspect that leads to the final performance [236,237]. Besides, the origin of better performance induced by nanosheet structure was believed to be the possible missing coordination of Ni atom and crystal disorders at the 2D catalyst surface.

Cobalt nitrides

Apart from the Ni3N, nitrides of cobalt have recently attracted improving electrocatalytic performance [71] relating to their excellent conductivity and nanostructure [238] as reported by Wu et al. [231] and Zhang et al. [74]. For 3D nanostructure, CoN nanowires grown on a CC network displayed higher catalytic performance than the corresponding oxide, Co3O4, further suggesting that the TMNs are more suitable electrocatalysts for OER than their corresponding TMOs. Additionally, cobalt nitrides with different valence states to shows different catalytic performance to OER. For example, Chen etal. synthesized different cobalt nitrides with different valence states (Co2N, Co3N and Co4N) [239]. It was revealed that the valence state and the conductivity both have effect on their catalytic performance. Due to the synergic advantages of ultrahigh electrical conductivity and modulated nitrogen content, the Co4N displayed the best OER performance and smallest Tafel slope as shown in Fig. 9d and e, respectively. Moreover, the Co4N catalyst displayed excellent stability over the Co2N and Co3N after 13000seconds electrolysis, loosing 4.5% of its initial overpotential value at 437 mV (Fig. 9f). Details on the various nanostructures and electrocatalytic performance of the recently reported TMNs as electrocatalyst for OER are shown in Table 6. Commercial IrO2, Corresponding TMOs and transition metal sulfides (TMSs) are also included for comparison.

FIGURE 9

(a) AFM image of the Ni3N nanosheets. (b) The normalized polarization curves of Ni3N nanosheets, bulk Ni3N and NiO nanosheets by the BET surface area of electrocatalysts. (c) Nyquist plots of Ni3N nanosheets, bulk Ni3N and NiO nanosheets. Z' is real impedance and Z' is imaginary impedance. Reproduced with permission [230]. Copyright 2015, American Chemical Chemistry. (d) IR-corrected polarization curves for all the cobalt nitrides catalysts. (e) The corresponding Tafel plots in 1 m KOH solution for all the cobalt nitrides catalysts. (f) Time dependence of the current density under a static overpotential of 437 mV in 0.1 m KOH solution. Reproduced with permission [239]. Copyright 2016, Royal Society of Chemistry.

RESEARCH

TABLE 6

Comparison and summary of the recently reported TMNs and their corresponding oxides and phosphides in alkaline (1 m KOH) media.

Electrocatalyst Morphology Overpotential (@ 10 mA cm-2) Tafel slope

Ni3N [230] Nanosheets - 45 mV dec-1

Bulk Ni3N [230] - - 85 mV dec-1

NiO [230] Nanosheets - 59 mV dec-1

Ni2P [224] Nanoparticles 290 mV 47 mV dec-1

Ni3N [229] Nanoparticles 430 mV 64 mV dec-1

CoN [74] Nanowires 290 mV 70 mV dec-1

Co3O4 [74] Nanowires 339 mV 82 mV dec-1

CoP [277] Nanorods 320 mV 71 mV dec-1

Co2N [239] - 390 mV 72 mV dec-1

Co3N [239] - 370 mV 80 mV dec-1

Co4N [231] Nanowires 257 mV 44 mV dec-1

Co4N [239] - 330 mV 58 mV dec-1

FeNi3N Bulk [229] Nanoparticles 320 mV 67 mV dec-1

FeNi3N NPs [229] Nanoparticles 280 mV 46 mV dec-1

CoNP@NC/NG [238] Nanocomposites 390 mV 98 mV dec-1

TiN@Ni3N [237] Nanowires 350 mV 94 mV dec-1

IrO2 [278] Nanoparticles 340 mV 47 mV dec-1

Ni3N [229] Nanoparticles - 67 mV dec-1

Ni3N [244] Nanostructures -121 mV 109 mV dec-1

NiO-CNT [279] Nanoparticles -400 mV -

Ni2P Micropellets -100 mV 118 mV dec-1

MoON [245] Nanofilms -146 mV 101 mV dec-1

MoO2 [280] Nanocompacts -120 mV 116 mV dec-1

MoP [281] Bulk form -80 mV 83 mV dec-1

NiMoN [245] Nanofilms -109 mV 45 mV dec-1

FeNi3N Bulk [229] Nanoparticles - 48 mV dec-1

FeNi3N NPs [229] Nanoparticles -158 mV 42 mV dec-1

Pt - 50 mV 29 mV dec-1

Electrolyzer

Overall water splitting

FeNi3N/NF//FeNi3N/NF [247] NiFeOx/CFP//NiFeOx/CFP [249] TiN@Ni3N//TiN@Ni3N [237] Ni3FeN//Ni3FeN [246] Co3 Fe Nx//Co3Fe Nx [248] NiCoP/rGO//NiCoP/rGO [250]

Morphology Overall voltage

Nanoparticles 1.62 V

Nanoparticles ~1.63V

Nanowires 1.62 V

Nanosheets 1.495 V

Nanowires 1.539 V

Nanocrystals 1.59 V

Hydrogen evolution reaction

Detailed evolution of hydrogen in all pH range can be seen below [13]:

Acidic solutions

Cathode 2H++ 2e" ! H2 (12)

Anode H2O ! 2H+ +1/2O2 + 2e" (13)

Alkaline and neutral solutions

Cathode 2H2O + 2e ! H2 + 2OH" (14)

Anode 2OH~ ! H2O + 1/2O2 + 2e" (15)

For the HER side, TMNs were also found to be a suitable catalyst with superb activity [240-244]. Most of the TMNs with excellent catalytic activities for HER are bimetallic nitrides of Mo, Ni and Fe because they display more excellent conductivity than the single TMNs. For instance, Zhang et al. fabricated a novel bi-metallic

NiMoN HER catalyst on self-supporting carbon cloth and compared with Ni3N, NiMo alloy and MoN in 1 m KOH [245]. They combined electrodeposition with subsequent RF plasma treatment to produce nanoporous NiMoN catalyst. The catalyst mainly consists of Ni0.2Mo0.8N phase and small amount of Ni3N. The TEM inspection of the Ni0.2Mo0.8N unravelled its nanoporous morphology and clear crystal lattice fringes with interspacing of 0.246 nm. To determine surface roughness, electrochemically active surface area (ESAS) was measured and divided by the geometric area of electrode. As a result, surface bombarded with nitrogen plasma was rugged and porous with increased roughness. This ternary composite delivered an advanced onset with low overpotential roughly 50 mV and 109 mV at 10 mA cm-2. The performance of the Ni0.2Mo0.8N also outperformed those of Ni3N, NiMo alloy and MoN.

During the last 1 year, FeNi3N has emerged as promising bimetallic TMNs due to the low charge transfer resistance, favourable kinetics in the Ni3N specie, high availability and cost-effectiveness when compared to Mo compounds [229,246,247]. According to Jia et al. the FeNi3N nanoparticles prepared by annealing of ultrathin NiFe-LDH precursor in NH3 gas environment exhibited excellent catalytic performance with considerably low overpotential for both HER and OER [229]. This is due to the nanosized and metallic character of the FeNi3N catalysts. As proposed by Wu et al. that 2D nanosheet morphology could improve the electrical conductivity of TMNs than the low dimensional or bulky morphologies [230]. Thus, FeNi3N nanoparticle-stacked porous nanosheets were prepared by Wang's group as seen in Fig. 10a [246]. The nanosheet morphology of the FeNi3N can be retained from the NiFe-LDH precursor by controlling the thermal ammonolysis condition. Benefiting from the defects and dislocations of the nanosheets that could allow the rapid evolution reaction and easy flow of electrolyte with the interfaces, the FeNi3N nanosheets showed an overpotential of 45 mV at 10 mA cm-2, which was significantly higher than those of NiFe-LDH and bulk FeNi3N (Fig. 10b). Furthermore, the FeNi3N nanosheets maintained good stability after 2000 cycles (Fig. 10c). The stability of the FeNi3N catalysts was further improved by employing an in situ preparation method whereby commercially available Ni foam as the Ni sources [247]. The in situ derived FeNi3N nanostructure displayed remarkable stability as long as 60 h at different current densities (Fig. 10d) with the FeNi3N phase unchanged after stability test (Fig. 10e). Meanwhile, the performance of the FeNi3N electrocatalysts was not superb in the evolution of hydrogen but also for oxygen evolution. Thus, they could serve as bi-functional catalysts for the design of overall water splitting which will be discussed soon.

Meanwhile, the performance of most reported TMN catalysts for HER is usually tested in the alkaline electrolytes such as KOH and NaOH. TMNs have been reported to catalyze the evolution of hydrogen in acidic electrolytes such as HClO4 and H2SO4 [240242,244]. Recently, Ding's group shows that metallic Ni3N nanosheets could deliver a high current density at low overpotentials, which is even very close to that of Pt [232]. The high performance of the Ni3N nanosheets over the previously reported TMNs in HClO4 electrolyte in acidic media [240,241] can be related to its nanostructure, metallic nature and the active surface site that is determined from the theoretical calculations. Summary on the nanostructures and electrocatalytic performance of the recently

Materials Today • Volume 00, Number 00• May 2017

FIGURE 10

(a) TEM image, (b) LSV curves and (c) Stability test of the Ni3FeN nanosheets. Reproduced with permission [246]. Copyright 2016, American Chemical Society. (d) Chronopotentiometric curves of FeNi3N/NF for OER and HER at 50 mAcm~2 and at 100 mAcm~2 (successive testing) without iR-correction. (e) XRD patterns of FeNi3N/NF after corresponding tests. Reproduced with permission [247]. Copyright 2016, American Chemical Society.

reported TMNs as catalyst for HER comparing with that Pt and some corresponding TMOs and TMSs is shown in Table 6.

Opportunities

TMNs show great opportunities as electrode materials for energy storage devices and electrochemical water splitting. In the case of energy storage devices, TMNs have also been widely employed as anode materials for energy storage devices, especially LIBs. Some of

the TMN nanostructure anodes have shown advantages over the commercial graphite and some TMOs, which has given the TMNs great opportunity as potential anode materials for LIBs [96].

Opportunities in energy storage devices

These opportunities are quite glaring as TMNs have been used as anode materials for full-cell LIBs both in the coin-cell and flexible-cell configurations, and impressive performance has been reported

RESEARCH

as well. Although, few TMNs have been demonstrated as anode materials for practical LIBs. However, some of the reported TMNs for full-cell LIB such as Fe2N nanoparticles, Ni3N nanosheets, TiN nanowires and WN nanowires have not only shown significant results for LIBs in the coin-cell configurations but also in flexible lithium ion batteries (FLIBs). For instance, Benefiting from the flexible CC support, Fe2N nanoparticles grown on a flexible CC were used as flexible anode materials for FLIB [68]. Commercial LiCoO2 (LCO) cathode was employed as cathode material. The Fe2N//LCO FLIB displayed an average discharge voltage of 2.4 V

(Fig. 11a) at the flat and bending state as well as high rate performance (Fig. 11b). The specific capacity of the cell reached 215 and 125 mAh g"1 at 200 and 1000 mAg"1, respectively (Fig. 11b). The flexible cell was also characterized with good flexibility and cyclic stability up to 40 cycles at a current density of 1 Ag"1 (Fig. 11c). In another example, flexible 3D Ni3N/CC nanosheets were employed as anode material, while LCO and LiNi05Mn15O4 (LNMO) were both employed as cathode in the full coin-cell LIB configuration [69]. The average discharge voltages of the Ni3N/CC//LCO and Ni3N/CC//LNMO reached 2.5 and

FIGURE 11

(a) Charge-discharge profiles for the device at the flat and bending position. (b) Rate performance profile of the Fe2N-600//LCO FLIB device. (c) Cycling profiles of the device at current density of 1000 mAg-1 up to 40 cycles at the flat and bending position. Reproduced with permission [68]. Copyright 2015, Elsevier. Charge-discharge profile of the (d) 3D Ni3N/CC//LCO and (e) 3D Ni3N/CC//LNMO cells. Cyclic performance profile of the (f-i) 3D Ni3N/carbon cloth// LCO and (f-ii) 3D Ni3N/carbon cloth//LNMO cells. Rate performance profile of the (g) 3D Ni3N/carbon cloth//LCO and (h) 3D Ni3N/carbon cloth//LNMO cells. (i) Lighting of the blue LED with 3D Ni3N/CC//LNMO flexible device while bent. Reproduced with permission [69]. Copyright 2016, Royal Society of Chemistry.

Materials Today • Volume 00, Number 00• May 2017

3.1V (Fig. 11d and e), respectively. Consequently, excellent capacity retention of 99% with specific capacity around 100mAhg-1 after 250 cycles at a current density of 1.4 A g-1 was achieved by both cells (Fig. 11f-i and f-ii). Due to the excellent rate capability performance of the two cells, the Ni3N/CC// LCO cell showed a high discharge capacity of 85 mAh g-1 at a current density of 2.26 A g-1 (Fig. 11g) while the LNMO counterpart delivered 80mAhg-1 at the same current density (Fig. 11h). The Ni3N/CC//LCO cell was scaled-up to light a 3.0 V blue LED to prove the flexible potential application of the TMNs (Fig. 11i). Likewise the Fe2N//LCO FLIB discussed above also lightened 3.0 V blue and green LED (inset in Fig. 11c).

For SCs, TMNs can also be employed for symmetric (SSC) and asymmetric (ASC) SC devices in order to demonstrate the potential application of the TMNs. The major difference between the SSCs and ASCs is that the ASCs utilize different electrode as anode and cathode and their operating voltage window is usually higher, while SSCs uses the same electrode as anode and cathode with lower operating voltage window. For example, all TiN-nanowire-based SSCs with voltage window up to 1.0 V could deliver energy and power densities than some carbon-based SSCs [59]. In the case of ASCs, TiN and Fe2N graphene composites used as cathode and anode, respectively could deliver much higher energy density of 15.4 Wh kg-1 and power density of 6.4kWkg-1 with broader working voltage (1.6 V) [165]. Interestingly, the operating voltage of Ti and Fe nitride composites (Fe2N@Ti2N) reaches 2.1 V in a recent work [73]. This working voltage is already approaching some LIBs voltage with high voltage anodes or low voltage cathodes, which creates more opportunity for the replacement of commercial capacitors.

Opportunities in water splitting

The use of TMNs as electrocatalysts has shown great opportunity in water splitting. TMNs could serve as both anode and cathode for the evolution of both hydrogen and oxygen, respectively in a practical Alkaline Electrolyzers. Among the commonly reported TMN electrocatalysts, bimetallic TMNs application as overall water splitting electrocatalyst have attracted much attention, especially FeNi3N [229]. Moreover, different nanostructures of FeNi3N [247] such as nanoparticles [229] and nanosheets [246] have been prepared. Owing to those advantages manifested by bimetallic TMNs,

investigation of its application in overall water splitting configurations (also known as Alkaline Electrolyzers) have been examined. Inset of Fig. 12a shows the optical image of FeNi3N nanoparticles serving as both anode (OER) and cathode (HER) for an alkaline electrolytic device [247]. A low electrolytic cell voltage of 1.62 V was delivered by the all FeNi3N device (Fig. 12a). Much lower voltage of 1.49 V can be delivered by higher dimensional nanos-tructures such as FeNi3N nanosheets as reported by Wang et al. (Fig. 12b) [246]. Other bi-metal nitrides such as Co3FeNx nanosheets could also exhibit low electrolytic voltage of 1.539 V (Fig. 12c) [248]. These shed light on a promising path towards highly efficient water electrolysis cell solely based on metal nitrides. Thus, comparing the overall water splitting performance of different FeNi3N nanostructures, FeNi3N nanosheets with overall water splitting potential of 1.49 V show superior performance over FeNi3N nanoparticles, other TMOs [249] and phosphides [250]. A table summarizing the h and the cell voltages of different TMNs-based alkaline electrolyzers compared with oxides and phosphides can be seen in Table 6. As mentioned above, most of the recently performed overall water splitting devices are usually alkaline electrolyzers. This is due to the poor catalytic performance of the electrocatalysts in the acidic or neutral media. Furthermore, there are no reports yet on acidic and neutral electrolyzers especially for TMNs. Research on acidic and neutral electrolyzers is another major challenge and prospect for electrochemical water splitting.

Summary and prospective

In conclusion, we have updated the development of TMNs with respect to their advancement through nanostructures for EES (especially in LIBs, SCs and LICs) and electrochemical water splitting (basically OER and HER). This review provides an in-depth progress of significant research discoveries achieved through various approaches for enhancing the storage and electrocatalytic performance of TMNs in the last two years. For storage devices, we focused on the nitrides of Ti, V, Fe, Ni, Mo and W and compared their performance with corresponding TMOs studied under the same experimental factors. In most cases for LIBs and LICs, the TMNs nitrides displayed higher rate capability than those of the corresponding oxides due to the higher electrical conductivity. The performance of the TMNs in SCs cannot be compared easily

FIGURE 12

(a) Polarization curve of water electrolysis for FeNi3N/NF//FeNi3N/NF with a scan rate of 5 mVs-1 in 1.0 m KOH without iR-correction. Reproduced with permission [247]. Copyright 2016, American Chemical Society. (b) LSV curves of overall water splitting for NSP-FeNi3N in a two-electrode system with a scan rate of 2 mVs-1. Reproduced with permission [246]. Copyright 2016, American Chemical Chemistry. (c) LSV curves of overall water splitting for NSP-Co3FeNX in a two-electrode system with a scan rate of 2 mVs-1. Reproduced with permission [248]. Copyright 2016, Royal Society of Chemistry.

RESEARCH

because most materials have different behaviour in various electrolytes and therefore are studied in different electrolyte. Most TMN materials reported in this review suffer from poor stability. The stability problem of the reported TMNs has been improved by nanostructuring and composite formation. Furthermore, the TMNs as highly conductive support (especially TiN and VN) for the growth of other electrodes were also addressed. They show significant support for other electrode materials both in their bulky and nanostructure forms than most TMOs (comparable to TiO2, because TiO2 exhibit excellent stability). In the case of electrochemical water splitting, Ni3N, nitrides of Co, Mo and binary metal nitrides of Ni, Fe, Co and Fe in their nanosized forms have been summarized to show excellent catalytic properties than the bulky ones and other corresponding TMOs. The catalytic performances of the TMNs are less comparable to those of transition metal phosphides (TMPs) but their performance are highly competitive during overall water splitting (especially the binary compounds). This is due to the excellent stability of most the transition metal nitrides in all pH range and during redox reactions compare to the oxides and phosphides. Recent advancements on TMONs as electrode material for LIBs and SCs were also reviewed. TMONs-based LICs are yet to be reported while their application in water splitting are still at the infant stage [251]. The TMONs have been proven to show enhanced conductive properties than the TMOs, comparative conductivity and higher chemical stability than the TMNs. We concluded that the performance of the energy systems largely depends on the morphology, size, structure and composition of the active materials as well as the excellent electrical conductivity of these TMNs. However, selecting suitable active material for the energy systems is the major research threat, which has led to scrutinizing the properties of different materials.

TMNs are not omnipotent upon their recent breakthroughs as electrode resulting into some aspects that require improvement. The outlooks on the TMNs as electrode materials for energy storage and conversion are categorized into three aspects, namely (i) new nanostructure designed as next-generation electrode, (ii) enhancing the stability and capacity in the case of energy storage devices and (iii) strategy to search for new TMNs and utilization of other cost effective transition metals as electrocatalysts for water splitting systems. Details are discussed below:

(i) New nanostructure designed as next-generation electrode. Generally, the fabrication and design of various nanostruc-tures have indicated significant improvement in the energy storage and electrocatalytic performance of TMNs, much attention should be focused on the connection between the TMNs and nanostructures for designing more suitable electrode materials for future energy application. We believe that in the near future, scalable method for nanostructure synthesis would gain industrial back due to the development of nano-engineering and technology. This suggests that potential researchers of energy storage and conversion applications should divert much research work on the development and engineering of new nanomaterials as next-generation electrodes.

(ii) Enhancing the stability and capacity of TMNs for energy storage. The excellent conductivity of TMNs has been proved to be very useful in high-rate EES system, but the capacity of

some of the family is still insufficient in LIB applications. Thus, they are either employed as a surface protective and conductive layer or a 3D scaffold for improving stability and rate performance of LIBs. For most of TMN anodes submitted to battery cycling measurement, partial conversion mechanism is more suitable for protective layer and scaffold (given the electrolyte related side reaction issue to be resolved). Because less changes in reaction means more stable. In order to develop TMNs with high electrochemical storage capability, incorporating such partial conversion reaction with alloying reaction will further enhance the specific capacity of the electrode materials. For instance, the nitrides of Zn, Sn, Ge, Sb and other metals or intermetallic elements could be alloyed with Li at certain electrochemical stage during cycling, which are promising materials to deliver ultra-high rate capacity (owing to alloying) while maintaining high rate capability (due to high conductivity). Furthermore, the chemistry behind the degradation of the TMNs still demand attention.

(iii) Strategy to search for new TMNs and utilization of other cost effective transition metals as electrocatalysts for water splitting systems. Understanding the theoretical approach determining if a catalyst will possess excellent catalytic performance is still missing for TMN electrocatalysts. TMNs recently have been proved to display more attractive catalytic activity than some of their corresponding oxides and sulfides. However, as mentioned above, the DFT calculations and volcano plots for other catalysts are mature, such progress should be extended to TMNs, which can further improve their research understanding and utilization opportunities. Additionally, another barrier is that studies on other cost-effective and multi-valence state transition metals such as Ti, V, Fe, Cu, Nb and W are less reported. The reported TMNs for water splitting are basically nitrides of Co, Ni and Mo, in which their nanostructure forms showed better performance than the bulk ones and corresponding oxides. Thus, much attention can also be given to the nitrides of Ti, V, Fe, Cu, Nb and W nanostructures since these nitrides have been widely explored in energy storage systems. With series of these cost-effective and multi-valence state TMNs been employed, the tendency of the cost of other commonly used materials will be highly reduced. Like in LIBs, the cost of Li metal increases likewise their availability also reduces, which creates more challenges for the search of new energy storage devices.

(iv) During the calculation of energy and power densities of energy sotrage devices, some parameters such as anode substrate thickness, cathode substrate thickness, separator thickness, anode thickness, cathode thickness, density of electrolyte, volume fraction of active material, density of active material, volume fraction of conductive additives and binders, density of conductive additives and binders, average discharge potential, discharge potential range or cut-off voltage, assumed first cycle losses, anode-to-cathode matching, etc.) should be considered. In most literatures, most of the parameters were not given in details, which makes it more difficult to report the actual energy and power densities of these devices. Thus, upcoming literatures should consider these parameters in determining their energy and power

Materials Today • Volume 00, Number 00• May 2017

densities and also provide the value of these parameters for readers to achieve better understanding. Finally, metal nitrides as a whole deserve much attention as they have attracted widespread application in recent years not only in the field of energy storage and conversion but also some other fields of study. Our attempt to review the recent development of TMNs as electrode materials is to explore their potential application. This present review might have some areas omitted in terms of prospective. However, the chances of improving the performance of these metal nitrides will grow along with time, initiatives and breakthroughs.

Acknowledgements

This work was supported by the National Key Research and Development Program of China (2016YFA0202604), the National Science Fund for Distinguished Young Scholars (21425627), the Natural Science Foundation of China (21461162003, 21476271 and 21425627), and the Science and Technology Plan Project of Guangdong Province (2015B010118002). Special acknowledgement to Prof. H.J. Fan of Nanyang Technological University, Singapore for his contribution. Finally, we like to acknowledge Jesse Gilkey from University of Phoenix, USA for his language editing contribution.

Appendix A. Supplementary data

Supplementary material related to this article can be found, in the online version, at doi:10.1016/j.mattod.2017.03.019.

References

[1] Z. Yang, et al. Chem. Rev. 111 (2011) 3577.

[2] S.L. Candelaria, et al. Nano Energy 1 (2012) 195. [3 J.B. Goodenough, Energy Environ. Sci. 7 (2014) 14. [4 Q. Zhang, et al. Chem. Soc. Rev. 42 (2013) 3127. [5 H. Chen, et al. Prog. Nat. Sci. 19 (2009) 291.

[6 M. Winter, R.J. Brodd, Chem. Rev. 104 (2004) 4245. [7] V. Etacheri, et al. Energy Environ. Sci. 4 (2011) 3243. [8 N. Yabuuchi, et al. Chem. Rev. 114 (2014) 11636. [9 M.D. Slater, et al. Adv. Funct. Mater. 23 (2013) 947.

[10] C. Zhu, et al. ChemNanoMat 1 (2015) 458.

[11] M.-S. Balogun, et al. Nano Res. 9 (2016) 2823.

[12] Y. Jiao, et al. Chem. Soc. Rev. 44 (2015) 2060.

[13] X. Zou, Y. Zhang, Chem. Soc. Rev. 44 (2015) 5148.

[14] Y. Xu, et al. Chem. Soc. Rev. 45 (2016) 3039.

[15] G. Wang, et al. Chem. Soc. Rev. 41 (2012) 797.

[16] C. Liu, et al. Adv. Mater. 22 (2010) E28.

[17] S.-W. Kim, et al. Adv. Energy Mater. 2 (2012) 710.

[18] S. Li, et al. Adv. Mater. 26 (2014) 3545.

[19] M.-S. Balogun, et al. Energy Environ. Sci. 9 (2016) 3411.

[20] A.S. Arico, et al. Nat. Mater. 4 (2005) 366.

[21] P. Poizot, et al. Nature 407 (2000) 496.

[22] M. Yu, et al. J. Mater. Chem. A 3 (2015) 15792.

[23] L.L. Zhang, X.S. Zhao, Chem. Soc. Rev. 38 (2009) 2520.

[24] Y. Zhai, et al. Adv. Mater. 23 (2011) 4828.

[25] Y. Zhu, et al. Science 332 (2011) 1537.

[26] M.K. Debe, Nature 486 (2012) 43.

[27] H. Yang, et al. Angew. Chem. 127 (2015) 11624.

[28] L. Li, et al. Nano Res. 8 (2015) 418.

[29] J. Jiang, et al. Adv. Mater. 24 (2012) 5166.

[30] V. Augustyn, et al. Energy Environ. Sci. 7 (2014) 1597.

[31] H. Wang, et al. Sci. Rep. 2 (2012) 701.

[32] Y. Zhong, et al. Adv. Sci. 3 (2016) 1500286.

[33] P. Afanasiev, Catal. Today 250 (2015) 134.

[34] X. Xu, et al. Nano Today 9 (2014) 604.

[35] C. Tan, H. Zhang, Chem. Soc. Rev. 44 (2015) 2713.

[36] L. Zhuo, et al. RSC Adv. 2 (2012) 5084.

[37] M.-S. Balogun, et al. J. Mater. Chem. A 3 (2015) 1364.

[38] C.-M. Park, et al. Chem. Soc. Rev. 39 (2010) 3115.

[39] M.-S. Balogun, et al. Carbon 98 (2016) 162.

[40] X. Lu, et al. Adv. Energy Mater. 4 (2014) 1300994.

[41] M.-S. Balogun, et al. J. Mater. Chem. A 2 (2014) 10825.

[42] X. Lu, et al. Nano Lett. 13 (2013) 2628.

[43] W.-F. Chen, et al. Chem. Commun. 49 (2013) 8896.

[44] A.-M. Alexander, J.S.J. Hargreaves, Chem. Soc. Rev. 39 (2010) 4388.

[45] S.H. Gage, et al. Catal. Sci. Technol. 6 (2016) 4059.

[46] J. Xie, Y. Xie, Chem. Eur. J. 22 (2015) 3588.

[47] S. Dong, et al. Coord. Chem. Rev. 257 (2013) 1946.

[48] J.M. Smith, D. Subedi, Dalton Trans. 41 (2012) 1423.

[49] M.G. Brik, C.G. Ma, Comput. Mater. Sci. 51 (2012) 380.

[50] S.T. Oyama, in: S.T. Oyama (Ed.), The Chemistry of Transition Metal Carbides and Nitrides, Springer Netherlands, Dordrecht, 1996, p. 1.

[51] A. Salamat, et al. Coord. Chem. Rev. 257 (2013) 2063.

[52] S.H. Elder, et al. Chem. Mater. 4 (1992) 928.

[53] V. Chakrapani, et al. Int. J. Hydrogen Energy 34 (2009) 9050.

[54] D. Choi, et al. Adv. Mater. 18 (2006) 1178.

[55] J.G. Chen, Chem. Rev. 96 (1996) 1477.

[56] S.H. Elder, et al. Chem. Mater. 5 (1993) 41.

[57] K. Zhang, et al. J. Mater. Chem. 21 (2011) 11916.

[58] H. Huang, et al. Nano Energy 31 (2017) 74.

[59] X. Lu, et al. Nano Lett. 12 (2012) 5376.

[60] G. Cui, et al. Chemphyschem 11 (2010) 3219.

[61] W. Li, et al. Chem. Commun. 47 (2011) 3619.

[62] H.D. Yoo, et al. Mater. Today 17 (2014) 110.

[63] C. Niu, et al. Nano Lett. 14 (2014) 2873.

[64] A. Manthiram, et al. Energy Environ. Sci. 1 (2008) 621.

[65] D.R. Rolison, et al. Chem. Soc. Rev. 38 (2009) 226.

[66] L. Mai, et al. Chem. Rev. 114 (2014) 11828.

[67] S. Desmoulins-Krawiec, et al. J. Mater. Chem. 14 (2004) 228.

[68] M.-S. Balogun, et al. Nano Energy 11 (2015) 348.

[69] M.-S. Balogun, et al. J. Mater. Chem. A 4 (2016) 9844.

[70] N. Fechler, et al. RSC Adv. 4 (2014) 26981.

[71] F. Meng, et al. J. Am. Chem. Soc. 138 (2016) 10226.

[72] M. Shalom, et al. Adv. Mater. 26 (2014) 1272.

[73] C. Zhu, et al. Nano Energy 26 (2016) 1.

[74] Y. Zhang, et al. Angew. Chem. Int. Ed. 55 (2016) 8670.

[75] B. Li, et al. Energy Environ. Sci. 9 (2016) 102.

[76] J. Ding, et al. Energy Environ. Sci. 8 (2015) 941.

[77] V. Aravindan, et al. ChemSusChem 7 (2014) 1858.

[78] H. Kim, et al. Adv. Energy Mater. 3 (2013) 1500.

[79] Y. Ma, et al. Adv. Mater. 27 (2015) 5296.

[80] R. Wang, et al. Adv. Funct. Mater. 25 (2015) 2270.

[81] P. Wang, et al. J. Mater. Chem. A 4 (2016) 9760.

[82] M. Liu, et al. Part. Part. Syst. Charact. 32 (2015) 1006.

[83] W. Chen, J.Z. Jiang, J. Alloys Compd. 499 (2010) 243.

[84] L. Yan, et al. Nanoscale 8 (2016) 8443.

[85] V. Aravindan, et al. Adv. Energy Mater. 5 (2015) 1402225.

[86] M. Armand, J.M. Tarascon, Nature 451 (2008) 652.

[87] D. Deng, et al. Energy Environ. Sci. 2 (2009) 818.

[88] W.J. Zhang, J. Power Sources 196 (2011) 13.

[89] N. Pereira, et al. J. Electrochem. Soc. 149 (2002) A262.

[90] Y. Wang, et al. J. Electrochem. Soc. 151 (2004) E162.

[91] X. Li, et al. J. Mater. Chem. A 1 (2013) 6441.

[92] Q. Sun, Z.-W. Fu, Electrochem. Solid-State Lett. 10 (2007) A189.

[93] Q. Sun, Z.-W. Fu, Electrochim. Acta 54 (2008) 403.

[94] Q. Sun, Z.-W. Fu, Electrochem. Solid-State Lett. 11 (2008) A233.

[95] B. Zhang, et al. Electrochim. Acta 150 (2014) 15.

[96] M. Zhang, et al. J. Power Sources 322 (2016) 163.

[97] F. Gillot, et al. J. Mater. Chem. 21 (2011) 9997.

[98] D.K. Nandi, et al. Phys. Chem. Chem. Phys. 17 (2015) 17445.

[99] D. Lei, et al. Sustain. Energy 2 (2014) 1.

[100] E. Yousefi, et al. Mater. Res. Bull. 84 (2016) 388.

[101] Y. Zhang, et al. ChemNanoMat 2 (2016) 764.

[102] D. Zhao, et al. J. Mater. Chem. A 4 (2016) 7914.

[103] X. Peng, et al. Electrochim. Acta 214 (2016) 201.

[104] Meng, et al. Nano Res. 9 (2016) 128.

[105] Z.W. Fu, et al. J. Phys. Chem. B 108 (2004) 2236.

[106] Y. Luo, et al. Chem. Commun. 51 (2015) 13016.

[107] M.S. Balogun, et al. J. Power Sources 308 (2016) 7.

[108] W. Qiu, et al. Electrochim. Acta 193 (2016) 32.

[109] J. Guo, et al. RSC Adv. 6 (2016) 111447.

RESEARCH

[110 [111 [112 [113 [114 [115 [116 [117 [118 [119 [120 [121 [122 [123 [124 [125 [126 [127 [128 [129 [130 [131 [132 [133 [134 [135 [136 [137 [138 [139 [140 [141 [142 [143 [144 [145 [146 [147 [148 [149 [150 [151 [152 [153 [154 [155 [156 [157 [158 [159 [160 [161 [162 [163 [164 [165 [166 [167 [168 [169 [170 [171 [172 [173 [174 [175 [176 [177 [178 [179 [180 [181 [182

S. Liu, et al. RSC Adv. 6 (2016) 86131.

L. Long, et al. Small 12 (2016) 1503305.

J. Xiao, et al. Nano Energy 24 (2016) 121.

P. Yu, et al. Chem. Eur. J. 21 (2015) 3249.

J.S. Cho, et al. ACS Nano 9 (2015) 4026.

J. Yang, et al. ACS Appl. Mater. Interfaces 9 (2017) 354.

X. Lian, et al. Ionics 22 (2016) 2015.

L. Lai, et al. Electrochim. Acta 134 (2014) 28.

D. Kundu, et al. J. Power Sources 278 (2015) 608. S.M. Abbas, et al. Electrochim. Acta 190 (2016) 988. Z. Xiu, et al. J. Mater. Chem. A 4 (2016) 4706. H.-C. Park, et al. J. Power Sources 269 (2014) 534. J. Liu, et al. Energy Environ. Sci. 6 (2013) 2691.

H. Xu, et al. ACS Nano 9 (2015) 6817. H.-C. Park, et al. Ceram. Int. 42 (2016) 1933. Z. Cui, et al. Adv. Mater. 28 (2016) 6926. N. Mosavati, et al. J. Power Sources 321 (2016) 87. M. Zhang, et al. J. Mater. Chem. A 5 (2017) 2251. J. Zhang, et al. Electrochim. Acta 132 (2014) 230. S.-Z. Kure-Chu, et al. Electrochim. Acta 212 (2016) 481. Y. Wu, et al. Nanoscale 6 (2014) 14697. M.-S. Balogun, et al. J. Power Sources 272 (2014) 946.

H. Park, et al. J. Mater. Chem. A 3 (2015) 8590.

E.M. Lotfabad, et al. J. Mater. Chem. A 2 (2014) 2504. Y. Yu, et al. ACS Appl. Mater. Interfaces 8 (2016) 3992.

I.S. Kim, Electrochem. Solid-State Lett. 3 (1999) 493. D. Tang, et al. J. Mater. Chem. A 2 (2014) 10375. H. Zheng, et al. J. Mater. Chem. A 3 (2015) 12476.

C. Yue, et al. ACS Appl. Mater. Interfaces 8 (2016) 7806.

M.-S. Balogun, et al. ACS Appl. Mater. Interfaces 7 (2015) 23205. X.-J. Wang, et al. Chem. Eur. J. 18 (2012) 5970. K.-F. Chiu, et al. Thin Solid Films 596 (2015) 29.

F. Lu, et al. RSC Adv. 6 (2016) 24366.

W. Ji, et al. J. Mater. Chem. A 2 (2013) 699.

D. Zhou, et al. Phys. Chem. Chem. Phys. 15 (2013) 16898. H. Kiyono, et al. J. Eur. Ceram. Soc. 32 (2012) 1413.

R.C. de Guzman, et al. J. Mater. Chem. A 2 (2014) 14577. S.L. Weeks, et al. Plasma Process. Polym. 13 (2016) 116. X. Li, et al. J. Mater. Chem. A 4 (2016) 5081.

B. Lalk, et al. Electrochim. Acta 164 (2015) 12. H. Li, et al. Electrochim. Acta 166 (2015) 197. T. Gao, et al. Mater. Lett. 159 (2015) 54.

F. Zhang, et al. RSC Adv. 6 (2016) 27901.

L.J. Miara, et al. J. Mater. Chem. A 3 (2015) 20338.

T.S. Miller, et al. Electroanalysis 27 (2015) 2614.

Y. Fu, et al. Nanoscale 6 (2014) 12555.

W. Lei, et al. Chem. Commun. 49 (2013) 352.

M. Hankel, D.J. Searles, Phys. Chem. Chem. Phys. 18 (2016) 14205.

E. Panabiere, et al. J. Alloys Compd. 663 (2016) 624. J.B. Ducros, et al. Electrochim. Acta 167 (2015) 20. Y. Su, et al. Solid State Ionics 282 (2015) 63.

Z. Yu, et al. Energy Environ. Sci. 8 (2015) 702. M. Yu, et al. J. Mater. Chem. A 4 (2016) 4634. T.C. Liu, et al. J. Electrochem. Soc. 145 (1998) 1882. Y. Yu, et al. J. Mater. Chem. A 3 (2015) 16633.

C. Zhu, et al. Adv. Mater. 27 (2015) 4566.

D. Choi, P.N. Kumta, J. Am. Ceram. Soc. 94 (2011) 2371. H. Cui, et al. Adv. Sci. 2 (2015) 1500126.

B. Gao, et al. Appl. Surf. Sci. 383 (2016) 57. Y. Xu, et al. Electrochim. Acta 173 (2015) 680. Z. Hou, et al. CrystEngComm 18 (2016) 3040. J. Liu, et al. Int. J. Hydrogen Energy 41 (2016) 996. S.I.U. Shah, et al. J. Power Sources 266 (2014) 456. P. Yang, et al. Adv. Sci. 3 (2016) 1500299. Y. Liu, et al. Electrochim. Acta 213 (2016) 393. A. Achour, et al. Nano Energy 7 (2014) 104. Y. Xu, et al. ChemElectroChem 2 (2015) 2020. Y. Xie, X. Fang, Electrochim. Acta 120 (2014) 273.

F. Grote, et al. J. Mater. Chem. A 3 (2015) 3465. R. Wang, et al. Electrochim. Acta 196 (2016) 611. K. Grigoras, et al. Nano Energy 26 (2016) 340.

H. Du, et al. New J. Chem. 38 (2014) 1284.

G. Ma, et al. J. Mater. Chem. A 3 (2015) 14617.

[183 [184 [185 [186 [187 [188 [189 [190 [191 [192 [193 [194 [195 [196 [197 [198 [199 [200 [201 [202 [203 [204 [205 [206 [207 [208 [209 [210 [211 [212 [213 [214 [215 [216 [217 [218 [219 [220 [221 [222 [223 [224 [225 [226 [227 [228 [229 [230 [231 [232 [233 [234 [235 [236 [237 [238 [239 [240 [241 [242 [243 [244 [245 [246 [247 [248 [249 [250 [251 [252 [253 [254 [255

Y. Liu, et al. Electrochim. Acta 211 (2016) 469.

B. Gao, et al. Adv. Mater. Interfaces 2 (2015) 1500211. Y. Xie, et al. J. Power Sources 286 (2015) 561.

J. Balamurugan, et al. J. Power Sources 308 (2016) 149. E. Kao, et al. Sens. Actuators A: Phys. 240 (2016) 160. Y. Xie, D. Wang, J. Alloys Compd. 665 (2016) 323.

C. Ji, et al. J. Mater. Chem. A 4 (2016) 2158.

C. Shang, et al. ACS Nano 7 (2013) 5430.

Z. Peng, et al. J. Mater. Chem. A 2 (2014) 10904. P.J. Hanumantha, et al. Electrochim. Acta 207 (2016) 37. A. Achour, et al. J. Power Sources 300 (2015) 525. P. Pande, et al. J. Power Sources 289 (2015) 154.

0. Kartachova, et al. J. Mater. Chem. A 2 (2014) 12940. M. Yu, et al. Adv. Mater. 27 (2015) 3085.

Z. Wang, et al. J. Power Sources 296 (2015) 53. M. Yu, et al. Angew. Chem. 128 (2016) 6874. T.-T. Chen, et al. Nanoscale 6 (2014) 5106. V. Aravindan, et al. Chem. Rev. 114 (2014) 11619. Y. Wang, Y. Xia, Adv. Mater. 25 (2013) 5336. H. Wang, et al. Small 11 (2015) 1470. L. Que, et al. J. Mater. Chem. A 4 (2016) 8716. R. Wang, et al. J. Mater. Chem. A 2 (2014) 12724. Y. Xie, et al. New J. Chem. 39 (2015) 604.

E. Lim, et al. ACS Nano 9 (2015) 7497.

Z. Chen, et al. Adv. Mater. 23 (2011) 791.

V. Aravindan, et al. ChemPlusChem 77 (2012) 570.

X. Li, et al. J. Mater. Chem. A 4 (2016) 11973.

W.-F. Chen, et al. Energy Environ. Sci. 6 (2013) 1818.

S. Chen, et al. Adv. Mater. 26 (2014) 2925.

S. Licht, et al. Int. J. Hydrogen Energy 26 (2001) 653.

J.X. Feng, et al. Angew. Chem. 128 (2016) 3758.

J.-X. Feng, et al. Adv. Mater. 27 (2015) 7051.

S. Bai, et al. Angew. Chem. Int. Ed. 53 (2014) 12120.

X.-F. Lu, et al. J. Am. Chem. Soc. 138 (2016) 8336.

D. Friebel, et al. J. Am. Chem. Soc. 137 (2015) 1305. Z. Zhao, et al. Adv. Funct. Mater. 24 (2014) 4698.

A. Bergmann, et al. Energy Environ. Sci. 6 (2013) 2745. S. Han, et al. Adv. Funct. Mater. 25 (2015) 3899.

A. Mendoza-Garcia, et al. Nanoscale 8 (2016) 3244. D. Xiong, et al. Chem. Commun. 52 (2016) 8711. X. Wang, et al. Adv. Funct. Mater. 26 (2016) 4067. L.-A. Stern, et al. Energy Environ. Sci. 8 (2015) 2347. L. Han, et al. Adv. Mater. 28 (2016) 9266.

Y. Zheng, et al. Adv. Mater. 27 (2015) 5372.

J. Greeley, N.M. Markovic, Energy Environ. Sci. 5 (2012) 9246.

J.K. Norskov, et al. Nat. Chem. 1 (2009) 37.

X. Jia, et al. Adv. Energy Mater. 6 (2016) 1502585.

K. Xu, et al. J. Am. Chem. Soc. 137 (2015) 4119.

P. Chen, et al. Angew. Chem. 127 (2015) 14923.

D. Gao, et al. J. Mater. Chem. A 4 (2016) 17363.

J. Greeley, et al. Nat. Mater. 5 (2006) 909.

1.C. Man, et al. ChemCatChem 3 (2011) 1159.

J.O.M. Bockris, T. Otagawa, J. Electrochem. Soc. 131 (1984) 290.

M. Shalom, et al. J. Mater. Chem. A 3 (2015) 8171.

Q. Zhang, et al. J. Mater. Chem. A 4 (2016) 5713.

X. Zhong, et al. J. Mater. Chem. A 4 (2016) 10575.

P. Chen, et al. Inorg. Chem. Front. 3 (2016) 236.

W.-F. Chen, et al. Angew. Chem. Int. Ed. 51 (2012) 6131.

B. Cao, et al. J. Am. Chem. Soc. 135 (2013) 19186. J. Shi, et al. Electrochim. Acta 154 (2015) 345.

Z. Xing, et al. Electrochim. Acta 210 (2016) 729. Z. Xing, et al. Electrochim. Acta 191 (2016) 841. Y. Zhang, et al. Adv. Energy Mater. 6 (2016) 1600221. Y. Wang, et al. ACS Appl. Mater. Interfaces 8 (2016) 18652.

B. Zhang, et al. Chem. Mater. 28 (2016) 6934.

S. Wang, et al. Chem. Commun. 52 (2016) 12614. H. Wang, et al. Nat. Commun. 6 (2015) 7261. J. Li, et al. Adv. Funct. Mater. 26 (2016) 6785.

C. Gebauer, et al. Nano Energy 29 (2016) 136. M.N. Solovan, et al. Inorg. Mater. 50 (2014) 40. X. Zhou, et al. J. Alloys Compd. 595 (2014) 80.

F.H. DullN, D.E. Rase, J. Am. Ceram. Soc. 43 (1960) 125. W. Weppner, Int. J. Hydrogen Energy 37 (2012) 4535.

Materials Today • Volume 00, Number 00• May 2017

[256] J.Z. Zhao, et al. Chin. Phys. Lett. 25 (2008) 4356.

[257] D.W. Bullett, J. Phys. C: Solid State Phys. 13 (2000) L595.

[258] G. Gu, et al. Nat. Mater. 2 (2003) 316.

[259] L. Maya, et al. J. Appl. Phys. 79 (1996) 7905.

[260] M.V. Reddy, et al. ACS Appl. Mater. Interfaces 6 (2014) 680.

[261] V.R. Shinde, et al. Appl. Surf. Sci. 252 (2006) 7487.

[262] V. Patil, et al. Soft Nanosci. Lett. 2 (2012) 7.

[263] J.R. Nicholls, et al. Mater. High Temp. 12 (1994) 141.

[264] N.S. Gajbhiye, et al. Phys. Status Solidi 189 (2002) 691.

[265] M. Milosavljevic, et al. Nucl. Instrum. Methods A 68 (1992) 426.

[266] P.S. Patil, L.D. Kadam, Appl. Surf. Sci. 199 (2002) 211.

[267] E.H. Greener, et al. J. Chem. Phys. 34 (1961) 1017.

[268] K. Yoshimura, et al. Thin Solid Films 281 (1996) 235.

[269] B. Hunsche, et al. Thin Solid Films 392 (2001) 184.

[270] C. Constantin, et al. Appl. Phys. Lett. 85 (2004) 6371.

[271] A. Filippetti, et al. Phys. Rev. B 59 (1998) 7043.

[272] S. Pokhrel, K.S. Nagaraja, Sens. Actuators B 92 (2003) 144.

[273] R. Mientus, et al. Surf. Coat. Technol. 200 (2005) 341.

[274] I. Saeki, et al. Mater. High Temp. 28 (2011) 264.

[275] D.-M. Kim, et al. Electrochim. Acta 163 (2015) 132.

[276] Q.-q. Xiong, et al. Nanoscale 5 (2013) 7906.

[277] J. Chang, et al. ACS Catal. 5 (2015) 6874.

[278] F. Song, X. Hu, Nat. Commun. 5 (2014) 4477.

[279] M. Gong, et al. Nat. Commun. 5 (2014) 4695.

[280] Y. Jin, et al. Adv. Mater. 28 (2016) 3785.

[281] Z. Pu, et al. Appl. Catal. B: Environ. 196 (2016) 193.