An Overview on the Development of Electrochemical Capacitors and Batteries – part II

In the second part of the review on electrochemical energy storage, the devolvement of batteries is explored. First, fundamental aspects of battery operation will be given, then, different materials and chemistry of rechargeable batteries will be explored, including each component of the cell. In negative electrodes, metallic, intercalation and transformation materials will be addressed. Examples are Li or Na metal batteries, graphite and other carbonaceous materials (such as graphene) for intercalation of metal-ions and transition metal oxides and silicon for transformation. In the positive electrode section, materials for intercalation and transformation will be reviewed. The state-of-the-art on intercalation as lithium cobalt oxide and nickel containing oxides will be approached for intercalation materials, whereas sulfur and metal-air will also be explored for transformation. Alongside, the role of electrolyte will be discussed concerning performance and safety, with examples for the next generation devices. Finally, a general future perspective will address both electrochemical capacitors


INTRODUCTION
In part I of this series of two reviews, we introduced electrochemical energy storage devices, their importance in recent strategies of decentralization of the energy grid and on mobile applications. On top of that, two kinds of devices were highlighted in part I introduction: electrochemical capacitors and batteries. Their main differences were discussed, then part I continued with details of electrochemical capacitors and materials designs for both electrochemical capacitors and batteries. In this part II, more details in batteries operation and different technologies will be presented, and a general future perspective will be given.

BATTERIES
Batteries are another class of electrochemical energy storage device. They contain three key components: a positive electrode, a negative electrode and an electrolyte, similarly to the electrochemical capacitor physical design.
The chemical reaction between the negative electrode and the positive electrode has two components: (i) electronic and (ii) ionic. The ionic component is driven by the electrolyte that forces the electronic component to circulate through an external circuit. Current collectors at the negative and positive electrodes deliver the electronic current to the external circuit. When the potential is suffi ciently positive it can extract electrons from electrode which is compensate by ions from solution. The opposite happens when the potential suffi ciently negative, it will inject electrons into the electrode and the movement of ions will also compensate the change in local charge (Goodenough 2013).
The open circuit potential (V oc ) can be determined by the difference between the negative electrode chemical potential (μ -) and the positive electrode potential (μ + ) divided by the magnitude of the electronic charge (e), as shown in equation 1: The battery's positive and negative voltage limits will depend on the energy gap (E g ) between the HOMO and the LUMO of the electrolyte. μmust be lower than the LUMO and μ + greater than HOMO of electrolyte, otherwise the electrolyte will be reduced on the negative electrode or oxidized on the positive electrode, forming a passivating solid electrolyte interphase at the negative electrode (SEI) or the cathode electrolyte interphase (CEI) fi lm (Liu et al. 2016a) at the positive electrode, as showed in Figure 1.
During battery operation, some mechanisms behind the electrochemical process occur -the intercalation and the transformation processes. The intercalation process occurs by the insertion of metal ion during charging/ discharging processes. During discharge, the ion moves within the electrolyte toward the positive electrode and interleaves into the material. During battery charging, the reverse process occurs, the ion now moves toward the negative electrode, which is the reason why this is known as a reversible process (Figure 2). In contrast to an intercalation material, transformation reactions can also take place in batteries, with the break and formation of chemical bonds during battery charge and discharge.

Rechargeable Batteries
A rechargeable battery makes use of a reversible electrochemical reaction in order to restore the original chemical constitution of electrode components, by means of an electrical current that fl ows in the opposite direction of the one during discharge. Since this is a non-spontaneous process, an external energy source is required to perform the charge process (Winter & Brodd 2004). Lead-acid, nickel-cadmium and metalion are among the most important rechargeable electrochemical systems. This review will focus on metal-ion batteries, especially those based on Li and Na (Silva et al. 2016).

Metal-ion batteries
Metal-ion batteries are based on the ion insertion/extraction into/from electrodes, and include Li + , Na + , Mg 2+ , and Al 3+ , being increasingly researched in both academy and industry. Among these metal-based systems, lithium has emerged as the battery of choice for both higher energy density and lighter weight. Since production capability became available, Li-ion batteries rapidly replaced the Ni-MH batteries (Liu et al. 2016a). Details concerning electrochemical mechanism, components constitution and materials properties will be further discussed, divided in negative electrodes, positive electrodes and electrolytes.

Negative electrodes
The negative electrode material is the one that carries the electrons to the external circuit and oxidizes during the electrochemical reaction. In a battery case, at the negative electrode, an oxidation reaction occurs during discharge. The process is reversed during charge, when a reduction reaction occurs allowing lithium or sodium ions to enter by different ways (intercalation into a crystalline structure or transformation forming Li 2 O or Na 2 O). In most cases, the potential is so low that electrolyte reduction can take place, forming a film. However, the formation of this surface film, which is electronically insulating but ionically conductive, stops further decomposition of the solvent but allows the electrochemical process of metal ion insertion to continue. The research and understanding of the characteristics of negative electrode interfaces were particularly promoted by the pioneering work of Peled et al. (1995), who called this passivation layer SEI, and later by Jeong et al. (2001), who demonstrated its formation, and have been of fundamental importance for the understanding of battery electrode processes. We are looking for the lower working potential with a greater cell voltage and consequently a higher battery energy density.

Metallic
The ideal negative electrode for lithium batteries is undoubtedly, metallic lithium, since it presents enormous theoretical specifi c capacity (3860 mA h g -1 ), low density (0.59 g cm -3 ) and low working potential (-3.04 V vs SHE). On the other hand, rechargeable batteries based on metallic lithium negative electrodes are not commercialized due to a practical problem: dendrites are formed during numerous charge/discharges cycles, which can result in loss of metallic lithium (low coulombic effi ciency) and generation of internal On the other hand, sodium metal has a much lower specifi c capacity, only 1165 mA h g -1 , due to its higher mass in comparison to lithium. Its working potential is also smaller, being -2.71 V vs SHE. However, the element's availability has made it be studied as a battery negative electrode since the 1960s (Zheng et al. 2019). The issues to be overcome for the commercializing of sodium metal negative electrodes are similar to those of lithium but aggravated by the high reactivity of sodium. The SEI layer is unstable and re-grown in every cycle below 1 V vs Na/Na + , causing low coulombic effi ciency, gas evolution, and electrolyte exhaustion that eventually leads to cell failure (Lee et al. 2019). Moreover, sodium plating is also accompanied by dendritic growth (Zheng et al. 2019). New insights on the stabilization of the SEI layer and the use of solid electrolytes (Gao et al. 2018) have been investigated in order to make metal sodium negative electrodes safe and effi cient.
In this review we approach two possible strategies for replacing lithium or sodium metal negative electrodes in order to overcome the aforementioned safety problems. In the last decades, a series of alternatives have been investigated, and they can be divided into two main types: lithium/sodium-ion intercalation or insertion materials and transformation materials. However, new knowledge is necessary to build better lithium negative electrodes, since there are two kinds of next-generation batteries with high-energy-density which are those Li-S and Li-Air. For sodium-ion batteries (SIBs), the technology is still at an early stage and multiple materials are still being introduced as negative electrodes.

Intercalation
In order to identify materials that could resist long-lasting cycling but avoiding the deposition of metallic lithium, lithium-ion insertion materials (or intercalating electrodes) were developed in 1978 by M. S Whittingham, to be used as positive electrodes. They are compounds that can reversibly incorporate and release lithium ions from their open structure, and at the same time assume different oxidation states (Whittingham 1978). These conditions are satisfi ed by carbonbased materials (e.g. graphite) and transition metal compounds among others.

Carbon-based materials
The use of graphite as a negative electrode can be considered the key advance that opened the way to commercial lithium-ion batteries (LIBs). Indeed, the choice of graphite as a negative electrode is quite surprising, since the electrochemical intercalation of lithium ions takes place outside the stability window of most organic solvents. The solvent, as we described above, decomposes on the graphite surface during the reduction (discharge) process, forming the SEI, when electronic charge is stored in the graphite network while lithium ions are inserted between the carbon sheets. Therefore, graphite is thermodynamically unstable but kinetically protected. The correct use of graphite as negative electrode in LIBs applications was demonstrated by Jeong et al. (2001). However, 10 years before the concept had reached a practical application with a battery introduced by Sony in 1991 (Blomgren 2017). The key feature of Sony's battery, called "lithium-ion battery" was the choice of suitable electrode materials, using graphite as a negative electrode and a cobalt and lithium oxide (LiCoO 2 ) as a positive electrode, as proposed by Yoshino (2000) and Mizushima et al. (1980) in Goodenough's group, respectively. In the present, many studies have been made to improve the life-cycling of graphite. For instance, recently Chen et al. (2018) have investigated a modification of a typical synthesis of graphite by microwave irradiation. The authors inform a long-cycling performance of 410 cycles obtaining 370 mA h g -1 , which corresponds to a 99.46% of the theoretical capacity. They associate this important improvement to the state-of-the-art in graphite negative electrodes to the formation of nano-graphite starting from flake graphite and expanded graphite edge as well as a stable SEI formation.
Although graphite is the state-of-theart negative electrode in LIBs, it presents a low specific capacity towards sodium-ion intercalation (30 mA h g -1 ) . This is because the formation energy of NaC x is energetically unfavorable (Lenchuk et al. 2019, Liu et al. 2016c, Moriwake et al. 2017. A few groups have been working on tuning the electrolyte solvent as an approach to increase graphite capacity toward sodiumion intercalation (Goktas et al. 2018, Jache & Adelhelm 2014. Hard carbons, on their turn, have specific capacities varying from 100 to 300 mA h g -1 , depending on their structure. They are considered disorganized forms of carbon. If the structure can be transformed into graphite (organized) under high temperature, the material is said to be non-graphitic. If the disorganized form is the most thermodynamically stable structure at all temperatures, the material is considered nongraphitizable (Dou et al. 2019). Their structure can be described as composed of fragments of bundled non-planar graphenic sheets, where graphene layers are locally stacked by van der Waals forces (Dou et al. 2019). Hard carbons are commonly obtained by controlled pyrolysis of organic compounds, which include polymers and biomass (Baldinelli et al. 2018). The process can lead to highly porous carbons with high surface area. The possibility to use biomass waste as carbon source makes of hard carbons a cheap and sustainable option for SIB negative electrodes. However, it requires care, since composition and homogeneity of precursor materials are a key factor affecting structure and properties of the obtained hard carbon (Dou et al. 2017, Gomez-Martin et al. 2019. Table I shows different hard carbons used as SIB negative electrodes from works published in the past year. For more examples, specific reviews on hard carbons can be found in references (Dou et al. 2019, Hou et al. 2017, Xiao et al. 2019. The mechanism of sodium ions intercalation in hard carbons is still under debate (Anji Reddy et al. 2018, Dou et al. 2019, Gomez-Martin et al. 2019. During discharge, there is a sloping line between 1.2 V and 0.1 V, and a plateau at very low potentials (<0.1 V vs Na/Na + ). The mechanism has been described as a "falling cards" model (Dahn et al. 1997), where first there would be sodium insertion in disordered carbon layers (sloping line region) and then the adsorption of sodium into the material's nanopores (plateau region). The oxidation state of adsorbed sodium in the plateau region has been observed as zero (metallic sodium) (Stevens & Dahn 2000) or nearly zero (Stratford et al. 2016). Although the very low insertion potential is an advantageous property of hard carbons, it also brings safety concerns. At high charge/discharge rates, it could cause irreversible sodium plating.
Not long ago, with the discovery and development of graphene plus the numerous reports of improved active materials for positive and negative electrodes modified with graphene, many researchers started believing that the capacity of LIBs can only be enhanced significantly by replacing the typical graphite negative electrode by pure graphene. This material shows superior conductivity (compared to metal), high surface area (2630 m 2 g -1 ), and would present the possibility of incorporating lithium on both faces of the sheets (forming the Li 2 C 6 stoichiometry with a theoretical specific capacity of 744 mA h g -1 ). Nevertheless, despite the considerable investment of money and time, this goal has not been achieved for pure graphene materials. The mechanism of lithium incorporation on single-layer graphene is still under debate both experimentally and theoretically (Ji et al. 2019, Kühne et al. 2017, Lee & Persson 2012. For instance, Pollak et al. (2010) studied the interaction between lithium ions and a single-layer graphene (SLG) and few-layers graphene (FLG). The authors found that the carbon-Li interaction in FLG resembles that of typical bulk graphite, while SLG performs radically differently, without the formation of the common stage 1 of LiC 6 phase. They associate this behavior to lower binding energies of lithium to carbon and strong Coulombic repulsion of the lithium atoms on the opposite sides of the graphene, leading to low surface coverage of SLG. Furthermore, Ji et al. (2019) found that there is no significant difference between bilayer graphene, FLG and graphite electrodes in the Li-storage mechanisms and kinetics behavior. Therefore, the use of graphene as a negative electrode for LIBs is controversial.
The adsorption of Na + ions onto the surface of pure SLG, as onto graphite, is energetically unfavorable according to theoretical studies (Yoon et al. 2017). Experiments also have shown the poor performance of SLG in SIBs: the capacity of SLG electrodes was very close to that of the bare copper foils on which SLG was grown by CVD (Ramos et al. 2015). However, the high theoretical capacity (300 -500 mA h g -1 ) arising from the possibility of incorporating Na + to both the sides of graphene sheets (Zhang et al. 2020) encourages the search for modifications of graphene electrodes that could change the energetics of the graphene-Na + interaction. For instance, Yang et al. (2017b) have studied two different types of defects, protrusions and holes, created by doping multi-layer graphene with P and N, respectively. P-doped graphene (with protrusions) achieved a capacity of 350 mA h g -1 at 50 mA g -1 , while nitrogen-doped graphene (with holes) showed a capacity of 211 mA h g -1 at the same rate. Despite the efforts aforementioned, the main use of graphene in battery electrodes is in the form of composites with an electroactive material. Associating a graphene framework to materials such as nanostructured transition metal oxides can lead to electrodes which are chemically more stable, with better mechanical properties, and higher electrical conductivity (Wang et al. 2020).

Alkali titanates
Lithium titanate, Li 4 Ti 5 O 12 (LTO), is a material that uses titanium oxide as raw material and results of great interest to be employed as negative electrode for LIBs. LTO has the property of (de) intercalating reversibly up to 3 lithium ions per formula at 1.55 V vs. Li/Li + , with a theoretical capacity of 175 mA h g -1 according to the equation 2: Li 4 Ti 5 O 12 + 3 Li + + 3 e -⇄ Li 7 Ti 5 O 12 (2) This reaction occurs at higher potentials than that of the decomposition of organic solvents and consequently there is no formation of a SEI layer. LTO presents interesting properties in terms of long-life cycling of charge/discharge in comparison with other negative materials. In addition, LTO is considered a zero strain material, due to the negligible structural change during the charge/discharge process (a reduction of the unit cell of only 0.2%). For instance, Chauque et al. (2017) showed the impact of a ball-milling treatment on the structure and crystallinity of LTO and its energy storage. The authors tried to decrease the particle size of LTO (synthesized by a typical ceramic method at 950 °C) in order to improve the diffusion path of lithium ions. The use of high-energy milling for different times showed that the specific capacity increased as the particle size decreased. But when the ballmilling applied time affected the crystallinity of the LTO, the capacity decayed drastically. However, the addition of graphite to LTO during grinding showed an improvement on specific capacity and rate capability response as well as a higher diffusion coefficient. Another example of the use of LTO as negative electrode is provided by Kim et al. (2017) who focused on improving the low electronic conductivity of LTO by adding graphene. The authors prepared a single-layered graphene-wrapped LTO (~200 nm) with a specific capacity of 130 mA h g -1 at high current densities of (de)lithiation of 30 C (1 C rate is the necessary current to charge/discharge the battery full capacity in one hour). The authors associate the remarkable performance to the improvement of the electronic conductivity of the final composite being ca. 1.6 x 10 -3 S cm -1 . The use of LTO as a negative electrode for LIBs application is closely linked to its great stability during numerous cycles of charge and discharge even at high current densities. However, it is still limited due to its high potential plateau (1.55 V vs. Li/Li + ) as well as low specific capacity compared to graphite or transformation materials.
For SIBs, the semiconductor layered sodium trititanate Na 2 Ti 3 O 7 (NTO) draws attention for its low sodium-ion insertion potential of 0.3 V vs Na/Na + (Senguttuvan et al. 2011). Its theoretical specific capacity is 178 mA h g -1 with the uptake of two sodium ions per formula unit and reduction of 2/3 of Ti 4+ to Ti 3+ , according to equation 3 (Senguttuvan et al. 2011). The compound consists of a layered zig-zag structure of [TiO 6 ] octahedra with sodium ions occupying sites in the interlayer region. Similar to other sodium titanates (Andersson & Wadsley 2002), NTO is interesting due to its low toxicity, high availability of precursor materials (TiO 2 and Na compounds), and easy fabrication. However, the semiconductor has low electronic and ionic conductivity. The addition of carbon as particle coating (Hwang et al. 2019) or composites (Yan et al. 2015, Zhou et al. 2016) is a common method to improve the electronic conductivity. Nanostructuring NTO into nanosheets or nanotubes by an alkaline hydrothermal method (Anwer et al. 2017, Kasuga et al. 1998, Ko et al. 2017, Wang et al. 2015b showed an improved performance at high rates due to the higher contribution of surface process in the overall electrode response (Leite et al. 2020). Na 2 Ti 3 O 7 + 2Na + + 2e -⇄ Na 4 Ti 2

Transformation
To replace the current intercalation-type negative electrode materials, conversion-type electrode materials are very promising because of their high theoretical capacity and low working potential. Among the candidates that have been reported in bibliography, we highlight two ways to incorporate lithium or sodium: the iron oxides (such as Fe 3 O 4 ) and the formation of lithium (or sodium)-alloys (i.e. silicon).

Iron Oxides
Transition metal oxides (TMOs) can be reduced with concomitant lithium participation to maintain electroneutrality, as shown in the following conversion-type reaction in equation 4: where M is a transition metal that can deliver high specific capacities, much higher than graphite. Fe 3 O 4 (magnetite) has long been considered a promising negative electrode material due to its high theoretical capacity (934 mA h g -1 ). In Table II, we summarize some previously published works, remarking the specific capacities obtained as well as the number of cycles performed.
The insertion of sodium in conversiontype materials usually causes higher volume expansion than lithium (Nayak et al. 2018) going up to 250% (Klein et al. 2013), but the high theoretical capacity associated to the process makes of these materials promising alternatives to sodium metal. Only a few oxides have been

Alloys compounds
Silicon, having a theoretical specific capacity around 10 times larger than that of the stateof-the-art graphite, has been regarded as one of the most promising materials for the next generation of LIBs. The Li-Si alloy with the highest lithium concentration, the Li 22 Si 5 phase with a theoretical specific capacity around 4200 mA h g -1 , is more Li-rich than fully lithiated graphite LiC 6 (372 mA h g -1 ) (Zhang et al. 2016b). Nevertheless, the process of commercializing silicon as a negative electrode is not straightforward due to two important drawbacks: 1) Pulverization of silicon particles (mainly in those with particle size above 150 nm) as a consequence of silicon volume expansion upon lithiation process and the resultant loss of electrical contact (Liu et al. 2012); 2) the continue formation a fresh SEI layer upon cycling (Andersen et al. 2019).
To date, tremendous effort has been made to overcome these problems. For example, strategies such as nanostructured silicon, i.e. nanowires, hollow nanostructures, and clamped hollow structures, Si-C yolk-shell structures  . 2012) have been studied and significant improvement has been seen. The silicon-oxide family (in which oxides such as SiO, SiO 2 , nonstoichiometric SiO x and Si-O-C are included) is an interesting alternative due to its low cost, easy synthesis and small volume change during cycling compared to silicon. In the case of SiO x , its main outstanding property is its theoretical specific capacity in the fully lithiated phase, which is 3172 mA h g -1 . However, these kinds of materials suffer from poor electrical conductivity, in addition to the fact that their initial Coulombic efficiency (ICE) is around 50-80% in the bare SiO x . This low ICE is a consequence of the formation of lithium oxide (Li 2 O) and Li silicate (Li 4 SiO 4 ) in the first cycle, which is considered an irreversible transformation .
Among the strategies to overcome these drawbacks are the use of carbon as coating or incorporating different carbon sources during the synthesis of silicon oxide materials. For instance, in SiO x /C hybrids materials, sol-gel of siloxanes (e.g. tetraethoxysilane ( T E O S ) , 1 , 3 , 5 , 7 -t e t r a m e t h y l -1 , 3 , 5 , 7tetravinylcyclotetrasiloxane (TTCS), hydrogen silsesquioxane) have been widely used. For example, David et al. (2016) synthesized SiO x /C composites by mixing SiOC with graphene at different ratios. The electrode with a 60:40 ratio (SiOC:graphene) delivered a charge capacity of 588 mA h g -1 at 0.1 A g -1 and a capacity of 200 mA h g -1 at high current (1.6 A g -1 ). Moreover, the 3D porous reduced graphene structure around the SiOC acts as an effective current collector and electron conductor. Similarly, in order to improve the electrochemical performance of SiO x Li et al. (2015b) reported a route to anchor the SiO x -C on the surface of graphene nanoplatelets (GNPs). The derived SiOx-C/graphene nanoplatelets composite showed that the capacity remained above 600 mA h g -1 at 0.1 A g -1 after 250 cycles, higher than that of 380 mA h g -1 for the bare SiO x -C.
S o m e r e l e v a n t e l e c t r o c h e m i c a l performances of the above-mentioned materials and other silicon-based/C composites are shown in the Table III. There are others strategies to improve ICE, as pre-lithiation of the silicon-based anode and the replacement of polymeric binder (in electrode formulation) and electrolyte ). In the case of pre-lithiation, which refers to the addition of active lithium to the cell before operation, Yoo et al. (2018) were able to improve the low ICE of 2D nanostructured Si/SiO x from 33% to 82% after a pre-lithiation of 6 hours. By this means, the inserted lithium ions can compensate the active lithium loss, e.g. caused by SEI formation, which leads to a reduction of the first cycle capacity loss . With continuous and intensive worldwide efforts, we may expect significant advances in the application of silicon and SiO x materials in the near future. Silicon could form a NaSi alloy with a theoretical capacity of 954 mA h g -1 at very low potentials vs Na/Na + (Nayak et al. 2018). However, crystalline silicon has limited sodium storage ability (Song et al. 2019b). Amorphous silicon, on the other hand, was shown to produce a specific capacity of 240 mA h g -1 after 100 cycles when used as a 50 nm thick film (Jangid et al. 2017). In comparison, tin (Sn) is a more promising material for sodium energy storage, having a theoretical capacity of 847 mA h g -1 with the uptake of 3.75 Na per Sn (Na 15 Sn 4 ) (Song et al. 2019b). A study has demonstrated that the electrode suffers from a higher volume expansion upon sodiation than upon lithiation. However, in the opposite process, (de)lithiation causes pulverization of the material while (de)sodiation promotes a microstructural stabilization that could make the technology viable (Wang et al. 2015a), but still with the need to overcome the problems of a large volume change during the (de)sodiation process. Nanostructuring and/or encapsulating Sn particles are useful approaches to prevent electrode pulverization. For example, Sn nanodots (1-2 nm particles) encapsulated in carbon nanofibers delivered a capacity of 633 mA h g -1 at 0.2 A g -1 . Another approach to improve Sn electrode performance is tuning the electrolyte. Ether-based electrolytes such as glyme (Zhang et al. 2016a) seem to have an effect on stabilizing Sn electrodes for thousands of cycles (Kim et al. 2018). Other materials, such as antimuonium (Darwiche et al. 2012) and phosphorus (Zhang et al. 2019), also have been contemplated as alloying-type negative electrodes for SIBs. Although these materials show very high capacities in comparison to intercalation materials, to make them viable for commercialization is still challenging. More research on SIB electrodes is needed in order to have better understanding of the involved mechanisms and how to improve the overall electrode performance.  Figure 4 shows the capacity and operating potential for these classes of materials.
Considering layered compounds, LCO is the most used intercalation material. Cobalt and lithium are at octahedral sites in alternating layers and form a hexagonal symmetry. The LCO has a high theoretical specific capacity of 274 mA h g -1 and a volumetric capacity of 1363 mA h cm -3 , a high discharge voltage of 3.8 V, a robust cycle performance and reduced selfdischarge (Cho et al. 2003). To reduce costs, LNO was proposed but it did not show improvement regarding specific capacity. Furthermore, Li + tends to replace Ni 2+ sites during the process of synthesis and deletion, blocking Li + diffusion pathways (Rougier et al. 1996). Replacing Co with Mn could be an interesting idea due to lower toxicity and low cost. But there are some problems: (i) dissolution of Mn occurs when transformed into Mn 3+ , and the compound is disproportionated into Mn 2+ and Mn 4+ in all Mnpositive electrodes. (ii) during cycling, Mn at LMO can leach out, and (iii) the layered structure tends to convert into a spinel structure during the delithiation. Furthermore, Mn solubilization can lead to a SEI destabilization at the negative electrode (Armstrong & Bruce 1996, Gu et al. 2012, Tu et al. 2006). The NMC formed by Li/Ni/ Mn/Co has a theoretical specific capacity of 280 mA h g -1 , similar to LCO, but the cost is reduced due to the lower amount of cobalt. One of the most used NMC is the LiNi 0.33 Co 0.33 Mn 0.33 O 2 (Bak et al. 2014   There are many compounds in the spinel family, presenting many stable and robust materials. Spinel LiMn 2 O 4 was developed and commercialized since 1980s, and it stands out due to the high theoretical specific capacity of 148 mA h g -1 . Li is located in tetrahedral 8a sites and Mn in octahedral 16d sites in a ccp matrix of oxygen anions. The lithium ion can diffuse through the vacancies of these interstitial sites (Nitta et al. 2015, Thackeray et al. 1984. Recently, Huang et al. (2018) reported the synthesis and electrochemical application of hybrid spinel/ layered oxide (Li 1.15 Ni 0.20 Mn 0.87 O 2 ) as a superior positive electrode. The synergistic effect of nanostructure and spinel/layer heterostructure enhances charge transfer, lithium ion diffusion, and structural stability, resulting in improved ICE (near 100%), reversible capacity (150 mA h g −1 at 5 C), and good cyclability (258 mA h g −1 after 70 cycles at 0.2 C). The olivine crystal used on positive electrodes is recognized for their high thermal stability and high capacity. Most olivines contain Fe, Mn and Co: LiFePO 4 ; LiMnPO 4 and LiCoPO 4 with theoretical specific capacities of 170, 171 and 167 mA h g -1 , respectively. In LiFePO 4 , Li + and Fe 2+ ions are located in octahedral sites whereas P is in tetrahedral sites. This compound has relatively low intercalation voltage and poor ionic conductivity. LiMnPO 4 offers 0.4 V higher intercalation voltage compared to olivine with Fe, leading to higher specific energy, but with lower ionic conductivity. Recently, El Khalfaouy et al. (2019) have developed a new compound using yttrium-substituted phospho-olivine in different proportions LiMn 1-x Y x PO 4 /C. When x = 0.01, the material shows an excellent capacity and stability during charge/discharge processes. The initial specific discharge capacity can reach up to 156.84 mA h g -1 at C/20, with a coulombic efficiency of about 96.11%, which is 14% higher than that of the non-doped material. The authors suggest that yttrium has been incorporated into the host material structure, enhancing the electronic conductivity and improving lithium ion mobility inside the structure.
Polyanionic compounds were proposed in order to improve the ionic conductivity, compared to olivine compounds. These materials have tavorite structures of the type LiMPO 4 (OH) x F 1−x (M = Al, Ga, V, Fe, Mn, and Ti) where some M 3+ compounds are abundant in the earth's crust (Fe 3+ and Al 3+ ) (Ramesh et al. 2010). The fluorophosphates tavorites LiVPO 4 F, Li 2 CoPO 4 F and Li 2 NiPO 4 F are good positive electrodes in Li-ion batteries with specific discharge capacity around 150 mA h g -1 . Other compounds of the polyanionic class are the oxyphosphates LiVOPO 4 (110 mA h g -1 ) and fluorosulfates LiFeSO 4 F, LiCoSO 4 F and LiNiSO 4 F with specific discharge capacities around 130 mA h g -1 (Li et al. 2017b).
Both the electrochemical mechanism and the component constitution of SIB and LIB are very similar, except for their charge carries, which involves an interstitial intercalation of a guest species into a host material. As for the positive electrodes specific challenges must be considered: sodium has a less negative reduction potential when compared to lithium (-2.71 V vs SHE and -3.02 V vs SHE, respectively) which affects its maximum energy density; another point is that sodium ions are bigger than lithium (ionic radius 1.02 Å and 0.76 Å, respectively) which creates incompatibility with some traditional intercalation materials (Hwang et al. 2017).
In SIBs, the positive electrode consists of a material that can reversibly perform Na + intercalation/de-intercalation reactions at a voltage considered reasonably greater than 2 V (vs. Na/Na + ). While these electrodes work as a host matrix for Na + , the change in volume should be as minimal as possible during the Na + intercalation/(de)intercalation cycles, and this ability is critical for long-term battery cycling performance (Slater et al. 2013). During the last years, many different materials have been studied for SIBs positive electrode, including layered transition metal oxides, tunnel type oxide structures, sulfate and phosphate based polyanionic compounds (Sun et al. 2019).
Layered transition metal oxides are structures composed by alternating layers of alkali ion and transition metal (TM)-ion, and for the sodium case they are usually referred as Na x TMO 2 . Due to the large size of the Na ion, when compared to Li ion, some differences in the structures are noticed. For instance, the anion framework could either be a close-packed or non-close-packed, while for Li it is usually close-packed. When the sodium ion occupies octahedral interstitial sites in the layered Na x TMO 2 it is said to have an O-type stacking sequence; when it occupies trigonal-prismatic interstitial sites it has a P-type stacking sequence (Sun et al. 2019). The first studies on the electrochemical behavior of Na-ion intercalation in layered transitionmetal oxides were published by Mendiboure et al. (1985), early in the 1980s. Among the most interesting aspects of these compounds one can find high theoretical capacities, suitable working voltages and a diversity of synthetic routes (Chen et al. 2019).
Layered manganese oxides, Na x MnO 2 , are materials that have the flexibility of forming structures with either O-or P-type stacking sequences depending on the concentration of sodium, besides presenting the advantages of low cost and earth abundance. A layered NaMn 3 O 5 material with a Birnessite structure was synthesized by a simple route consisting of a redox reaction followed by a hydrothermal treatment was reported by Guo et al. (2014). The obtained NaMn 3 O 5 positive electrode showed a high capacity of 219 mA h g -1 (72.5% of its theoretical capacity), and exhibited a good rate capability of 115 mA h g -1 at a 5 C rate.
For the cases where Na x TMO 2 contains several stacking sequences, it is usual for the P-type phases to have an insufficient amount of sodium, i.e. x < 1, while the oxides with O-type phase usually shows an amount of sodium close to the sufficient, i.e. x ~ 1. This means that the O-type phase has a superior specific capacity, since it can have more sodium extracted from its structure than the P-type. However, the structural configuration of two adjacent face sharing sites in the trigonal prismatic structure found in the P-type lattice facilitates the diffusion of sodium ions, which means a better performance of rate capability for P-type based electrodes over O-type (Sun et al. 2019). A P2/O3 biphasic material, Na 0.67 Ni 0.33 Mn 0.57 Sn 0.1 O 2 was synthesized and used as a positive material for SIBs, providing an initial capacity of 155 mA h g -1 at 15 mA g -1 , with good rate and cycling performance. This electrochemical response was attributed to the mixture of the two phases which promoted a synergistic effect (Li et al. 2019a).
Some transition metal oxides studied as positive electrodes for SIBs have a especial arrangement with connected octahedral and square-pyramidal sites, which forms a tunnel shaped lattice that can promote rapid sodium ion diffusion and consequently superior storage performance (Yabuuchi et al. 2014). This is the case of single crystalline Na 0.44 MnO 2 nanowires synthesized by Cao et al. (2011) using a polymerpyrolysis method. When applied as SIBs electrodes, it produced a capacity of 128 mA h g -1 at 0.1 C, with 77% capacity retention after 1000 cycles at 0.5 C.
Doping tunnel structured oxides with fluorine is an interesting option to enlarge the tunnel lattices and to facilitate the mobility of the sodium ions, since ions containing fluorine have higher electronegativity and smaller ionic radius in comparison to ions containing oxygen. A series of F-doped Na 0.66 Mn 0.66 Ti 0.34 O 2- x F x were synthesized and studied by Wang et al. (2018); diffusion coefficient of the sodium ions confirmed the enlargement of the tunnels, which resulted in a good rate performance of the electrodes. The highest obtained capacity was 97 mA h g -1 at 0.2 C, and the electrode provided 85 mA h g -1 at 2 C for 1000 cycles.
The presence of polyanion groups, e.g. phosphates and fluorophosphates, in transition metal oxides increases the electrode operating voltage (vs Na/Na + ) due to inductive effect promoted by these groups. This interesting characteristic can be explored in order to improve the SIBs energy density (Yabuuchi et al. 2014). Na 3 V 2 (PO 4 ) 3 is a very promising electrode material due to its Na super ionic conductor (NASICON) structure, theoretical capacity (117.6 mA h g -1 ) and flat potential plateaus around 3.4 V (vs Na/Na + ). However, it has poor conductivity that leads to a limited rate capability. In order to avoid the conductivity hindrance, Huang et al.
(2019) synthesized and studied a carbon coated Na 3 V 2 (PO 4 ) 3 composite doped with nitrogen through a sol-gel process followed by sintering. The presence of a nitrogen-doped carbon layer helped obtaining high sodium diffusion coefficients by shortened diffusion lengths, and improving the electronic conductivity. The prepared electrode exhibited capacities of 109.2 mA h g -1 at 0.2 C and 87.2 mA h g -1 at 20 C, with a capacity retention of 91.2% after 500 at 2 C.
Sodium vanadium fluorophosphate family materials, Na 3 V 2 (PO 4 ) 2 O 2x F 3-2x (0 ≤ x ≤ 1), combine the electronegativity of the PO 4 3groups with Fions. The result is electrodes with operating voltages close to 4.0 V (vs Na/Na + ), nevertheless its theoretical capacity is close to 130 mA h g -1 , depending on V oxide state. Xu et al. (2019a) developed a hydrothermal method to synthesize carbon coated Na 3 V 2 (PO 4 ) 2 O 2 F nanoparticles and studied their electrochemical properties. The small particle size coupled with electronic conductivity enhancement from the carbon layers helped the electrode to provide a capacity of 121.5 mA h g -1 , at 1 C (3 -4.5 V vs Na/Na + ).

Transformation
Positive conversion electrodes are being widely studied as potentially high-energy-density alternatives to intercalation-based materials to increase specific capacity and enable effective application in batteries (Grimaud et al. 2016). Moreover, these compounds are capable of storing between 2 and 3 Li per atom (Grimaud 2017, Wu & Yushin 2017). LIBs have two types of reactions, shown in equations 5 and 6: Type 1: M + X z + yLi ⇌ M + zLi (y/z) X (5) Type 2: yLi + X -⇌ Li y X Where, M + can be Fe 3+ , Fe 2+ , Ni 2+ , Cu 2+ , Mn 3+ , M is the reduced material, and Xcan be F -, Cl -, Br -, I -, S 2-, Se 2-. In general, the most commonly used substances for this type of electrode are some of the most abundant and environmentally friendly materials such as O, S, Fe, and Cu. As mentioned, two types of reactions can occur: type 1, with true conversion, and type 2, chemical transformation.
For a type 1 reaction, in general the reduction of metal halides to the elemental form of the metal requires more than one lithium, which leads to theoretical capacities ranging from 500 to 800 mA h g -1 to CoF 3 , CuF 2 , NiF 2 , FeF 3 , FeF 2 , VF 3 . This conversion reaction transforms a single phase (M + X z ) in two phases (LiX and M).
For type 2, the reaction is a chemical transformation of a single phase in another single phase. During the charge/discharge process there is an alternation between the phases with the elemental material (X -) and the lithium aggregate state (Li y X). Recently, chalcogen and chalcogenide materials including Se e Li 2 Se , Song et al. 2019a, Te e Li 2 Te, and S and Li 2 S (Meini et al. 2014, Wild et al. 2015 have received considerable attention as promising positive electrode candidates for next-generation rechargeable lithium-ion and lithium batteries. Both reaction types are exemplified in Figure 5 and specific capacity and operating voltage are shown in Figure 6. Lithium sulfur batteries will be highlighted in the review due to their higher specific theoretical capacity compared to LIBs. The electrolytes for Li−S battery system are usually ether-based solvents such as 1,3-dioxolane (DOL), 1,2-dimethoxyethane (DME), and tetra-(ethylene glycol) dimethyl ether (TEGDME), since carbonate-based electrolytes react with polysulfides irreversibly.
In addition to being low cost, recovered from industrial waste, nontoxic and abundant, sulfur has a theoretical capacity approximately 5 times higher than transition metal oxides and phosphates, and it has an energy density of 2600 W h kg -1 . These are the main facts that favor the large scale application of sulfur (Ma et al. 2017b, Tao et al. 2016, Worthington et al. 2017. Electrochemical reduction of sulfur at the positive electrode produces low oxidation state polysulfides according to the following equations (7 to 11). Two plateaus are clearly defined in a typical discharge profile when used in a cell with liquid electrolytes (Medenbach & Adelhelm 2017, Talaie et al. 2017).
Numerous attempts have been made to overcome these challenges, such as optimization of the positive electrode structure, nanostructuring of active materials, including additives to electrolyte and/or electrode, and modifying the polymer separator. Some of the widely explored materials to increase the electrode conductivity and provide physical confi nement/entrapment of lithium polysulfi des are micro-nanostructured sulfur encapsulate materials, nano conductive and porous carbon materials such as: carbon nanotubes, carbon nanowires, porous carbon, graphene, carbon fi bers, as well as compounds such as conductive polymers and porous silica (Chen et al. 2013, Lyu et al. 2015, Shao et al. 2013, Wang et al. 2013. Recently, Metal-Organic Frameworks (MOFs) have been studied to encapsulate sulfur (Hong et al. 2018, Zhao et al. 2016, Zhou et al. 2014. MOFs are crystalline materials composed of metal ion junctions and organic ligands in infinite matrices, diverse geometries, and possess great synthetic versatility . ZIF8 (Zeolitic Imidazolate Framework-8) is a well-studied MOF that plays a very promising role, because it is electrochemically inert and stable. Some authors have reported specific capacities around 1000 mA h g -1 at 0.1 C (Zhou et al. 2014).
Na-S batteries are usually constituted of molten sodium and sulfur as negative and positive electrodes, respectively, and β-alumina as a sodium conducting ceramic electrolyte. They were fi rst proposed by the Ford Company in the 1970's and since then have been studied aiming stationary applications. Among the attractive characteristics of these cells one can fi nd an interesting theoretical specifi c energy of 760 W h kg -1 , low self-discharge rate, temperature stability, low-cost of the components, and safety due to solid electrolyte. The big hindrance of this type of cell is that part of the produced energy is used to maintain their high operating temperatures (300-350 °C), which reduces the The elemental sulfur used as positive electrode in Na-S batteries has a high theoretical specific capacity of 1675 mA h g -1 , considering a two-electron reaction as shown in the equation 13. However, due to the formation of electrical insulator by-products during the discharge process, i.e. polysulphides, the value cannot be reached and the active species performance is limited by the gravimetric capacity of 836 mA h g -1 (Hueso et al. 2013, Lu et al. 2010.
Molten sulfur and polysulphide compounds are highly corrosive, which implies in the use of protective steel layers and constitute a challenge in the search for a current collector. An extremely vigorous reaction between sodium and sulfur might occur if there is any rupture in the β-alumina membrane, which can result in fire and explosion (Hueso et al. 2013, Lu et al. 2010.
The use of a transition metal chloride as the positive electrode, along with a secondary molten electrolyte of NaAlCl 4 (melting point, 170 °C), can assure some operational advantages over traditional Na-S battery such as higher theoretical energy density (788 W h kg -1 ) and operating voltage, as well as better tolerance against overcharging. In these cells, the molten electrolyte works as a Na transport medium between the β-alumina and the positive electrode. It also makes the system safer due to the less vigorous reaction with molten sodium, in case of broke β-alumina (Hueso et al. 2013, Lu et al. 2010.
The positive electrode, which must be insoluble in the molten electrolyte, is usually a porous structure of nickel chloride impregnated with NaAlCl 4 salt. The addition of iron to the cell increases the power response. Equation  14 shows the charge/discharge reactions of the nickel configuration, which provides an electromotive force of 2.58 V, while the charge/ discharge reactions of the iron configuration can be seen in the equation 15, which generates an electromotive force of 2.35 V (Hueso et al. 2013, Lu et al. 2010 Both safety and electrochemical issues concerning high temperature Na-S batteries have led to the search for room temperature systems. Nevertheless, these batteries have some other challenges such as low sulfur utilization and dissolution of polysulfide intermediates which can cause a "shuttle" of the negative sulfur electrode. Microporous carbon has been found suitable for both to improve electric conductivity and to provide good confi nement/immobilization for sulfur and its reduced products . Kumar et al. (2019) obtained a high energy density room temperature Na-S battery by incorporating MnO 2 nanoneedle arrays on a fl exible carbon cloth substrate containing Na 2 S 6 , which worked as a multifunctional positive electrode and provided an initial energy density of 946 W h kg -1 , dropping to 728 W h kg -1 after 500 cycles.
Metal-Air batteries can also be highlighted due to their high specific capacity, and approaching the energy density of gasoline when Li is used as negative electrode. This kind of battery can be based on aqueous or nonaqueous electrolytes, and both of them rely on the electrochemistry of O 2 at the positive electrode. The electrolyte will play a major role on the products formed during battery discharge. While the reduction of O 2 in aqueous electrolyte involves O-O bond cleavage, forming OH -, the reaction in non-aqueous electrolytes forms peroxide, and there is no bond cleavage (Bruce et al. 2012).
As mentioned in a previous section, the negative electrode of a metal-air battery is either lithium or sodium, for Li-Air or Na-Air batteries, respectively. On the other hand, the positive electrode is composed of a material that allows the oxygen reaction to occur reversibly. Carbon is widely used for this matter. It is the oxygen electrochemical reaction at the positive electrode where lies one of the major problems on metal-air batteries. The pathway for the redox reactions is not the same, which generates a hysteresis between charge and discharge plateaus (Figure 7), causing a low energy effi ciency. On top of that, carbon can also go through corrosion when cell is charged to voltages higher than 3.5 V ( , Lim et al. 2016.

Electrolytes
In general, the electrolyte of a metal ion battery must be a solvent stable enough through the faradaic reactions (as discussed previously), and have a metal ion salt in high concentration. The transport number of the metal ion must be as high as possible. However, the chemistry at each of the different electrodes may require certain specificity on the electrolyte. Nowadays, the state-of-the-art electrolyte for Li-ion battery is composed of a mixture of carbonate solvents, as ethylene carbonate and dimethyl carbonate (50:50 v/v), with 1.0 mol L -1 of LiPF 6 , and proportions can vary (Guyomard & Tarascon 1995). The electrolyte also plays an important role on the formation of the SEI on early cycles, and PF 6 anions are involved on the mechanism (typically forming LiF). Additives can also be added to the electrolyte to assist on the formation of such layer, including the CEI. Most common additives are vinylene carbonate and vinyl ethylene carbonate, but many others are also used (Zhang 2006). Different anions can be used for the electrolyte preparation, and most of them present fluorine in its formulation, for the SEI formation. LiTf 2 N (bis(trifluoromethylsulfonyl) imide anion), LiFSI, LiBF 4 , and LiCF 3 SO 3 are a few examples in this category. The same anions can also be used for the preparation of Na-ion battery electrolytes.
In order to increase the battery operating voltage, electrolytes with higher electrochemical stability towards oxidations is necessary. For instance, Zhang et al. (2013) showed that through fluorination of organic carbonates, stability up to 5 V can be achieved. Another alternative to increase electrochemical stability is the development of ionic liquid-based electrolyte. Elia et al. (2016) used Tf 2 N-and FSI-based ionic liquids in Li-ion batteries and showed that extended cycle life could be achieved. Moreover, they observed that the rate capability was improved when electrolyte contained FSI ionic liquid. Ionic liquid electrolytes also show high thermal stability, and a large variety of chemical functionalization is possible, which can play important role on battery operation (Galiote et al. 2017), however, one must take in account any incompatibility between chemical groups in the ionic liquids with the cell components. For instance, imidazolium cations can show an acid hydrogen in the five-member cycle, which can react with lithium. The substitution of these hydrogens with a methyl group remove the reactivity with metal, making it a more compatible electrolyte , Wang et al. 2007.
Solid electrolytes were proposed to increase safety and remove the possibility of electrolyte leakage if a cell is compromised. They have been studied since the 1970s, especially using poly(ethylene) oxide (PEO) (Armand 1994, Xue et al. 2015, but no formulation has reached the market yet. The major problem that hampers their application is the low ionic conductivity when compared with their liquid counterparts. Many progresses have been made nowadays in gel electrolytes, for instance using polymers and ionic liquids, or even poly(ionic liquids) (MacFarlane et al. 2016, Torresi et al. 2018. More recently, high voltage aqueous LIBs were proposed using highly concentrated aqueous electrolyte. Suo et al. (2015) showed that a solution of 21 molal of LiTf 2 N in water completely changes the interactions between water molecules and Li + . The stronger interaction due to low availability of water expands the electrolyte stability (Martins & Torresi 2020). Since then, cells containing highly concentrated aqueous electrolytes (also called water-in-salt electrolyte, WiSE) were shown to be stable for batteries using different positive electrodes (Suo et al. 2016, Yang et al. 2019a) and even Li-O 2 (Dong et al. 2018) and Sulfur batteries (Yang et al. 2017a). Dubouis et al. (2018) showed that water reduction at the negative electrode and so the production of hydroxyl groups was responsible for the production of LiF and the formation of SEI, and not the direct reduction of the anions Tf 2 Nas previously proposed.

FUTURE PERSPECTIVES
The Nobel prize in Chemistry was awarded in 2019 to Goodenough, Whittingham and Yoshino for the development of the Li-ion battery. Their findings paved the way for the electric revolution in which we live today. From devices that enabled information access nearly everywhere to electric vehicles, they are all powered by an electrochemical energy storage device like a lithium-ion battery. The significant increase in the demand for electrochemical energy storage devices escalated the search for new materials. Many strategies have been used to find cheaper electroactive materials in morphologies that can favor a better performance and also more efficient redox reactions; that is, new research and development projects are underway where enormous efforts are being made to increase the quantity (energy stored by mass or volume of the device) and the operation speed (power to store and extract energy). The major driving force to achieve these objectives are related to the synthesis of new materials for the electrodes (with strong indication that nanostructure can enhance performance) where the kinetics of the redox reaction is faster (current increase) and the quantity of electrons that can be stored is greater (energy). In this regard, it is also important to increase the operation of the devices (greater potential where the redox reaction occurs (for batteries) or greater electrolyte/electrode stability (for electrochemical capacitors)). Therefore, it is important to develop new electrolytes that are electrochemically stable in the potentials where the electrodes undergo to the redox reaction; an example, it is the development of ionic liquids. In this case, in addition to its electrochemical stability, its thermal stability makes them suitable for use in safer devices. The safety of electrochemical energy storage devices is a key point as they are all high energy density devices. Avoiding chemical decomposition that can lead to flammable products is a non-negotiable requirement for stationary (for example, smart electricity distribution networks) and nonstationary (electric vehicles) applications. Finally, considering the availability of chemicals to synthesize different materials, it is possible to assume that in the future there will be a coexistence of different technologies for electrochemical energy storage, where the type of application will contribute to the definition of what type of storage will be more efficient for each application.