Review of molten carbonate-based direct carbon fuel cells

Direct carbon fuel cell (DCFC) is a promising technology with high energy efficiency and abundant fuel. To date, a variety of DCFC configurations have been investigated, with molten hydroxide, molten carbonate or oxides being used as the electrolyte. Recently, there has been particular interest in DCFC with molten carbonate involved. The molten carbonate is either an electrolyte or a catalyst in different cell structures. In this review, we consider carbonate as the clue to discuss the function of carbonate in DCFCs, and start the paper by outlining the developments in terms of molten carbonate (MC)-based DCFC and its electrochemical oxidation processes. Thereafter, the composite electrolyte merging solid carbonate and mixed ionic–electronic conductors (MIEC) are discussed. Hybrid DCFC (HDCFCs ) combining molten carbonate and solid oxide fuel cell (SOFC) are also touched on. The primary function of carbonate (i.e., facilitating ion transfer and expanding the triple-phase boundaries) in these systems, is then discussed in detail. Finally, some issues are identified and a future outlook outlined, including a corrosion attack of cell components, reactions using inorganic salt from fuel ash, and wetting with carbon fuels.


Introduction
Fuel cells (FCs) offer an environmentally friendly and highly efficient approach to energy-conversion technology for power generation. This approach has been intensively investigated for over a hundred years, since Grove et al. [1] developed the first FC in 1839. This technology is steadily evolving and improving, as the demand for electricity and new electricity generation technologies continues to grow. The FC device generates electric power from gaseous fuels (hydrogen or syngas) by reforming liquefied petroleum gas or liquid fuels. However, the complex synthesis process and the storage and transportation requirements increase the cost. At present, the utilization of carbon fuels is still a preferred option for the development of FCs, due to the abundant resources, easy access and relatively high-energy density. Direct carbon fuel cells (DCFCs), as an energy-conversion technology, generate electricity from solid carbon fuels through electrochemical reactions [2,3]. The device has a higher energy-conversion efficiency rate than traditional power generation devices of heat engines, which minimize system complexity and thus ensure a lower cost [4].
The first DCFC was reported by Becquerel [5] in the middle of the nineteenth century; the configuration was a carbon rod in a KNO 3 solution inside a platinum container. In 1896, William Jacques [6] developed a large-scale DCFC using 100 single cells, and baked coal as the anode. The electrolyte was an unspecified mixture of KOH and/or NaOH and the cathode was iron pots. Initially, the actual reaction in this system was only deemed to be the chemical reaction of carbon with nitrate [7]. Later research reported that gases from solid carbon pyrolysis are involved in the electrochemical Can Cui and Shuangbin Li contributed equally to this work.
* Cairong Jiang crjiang@suse.edu.cn * Jianjun Ma jjma@suse.edu.cn reaction [8,9]. Other researchers from Stanford Research Institute (SRI, International) investigated this technology thoroughly and invented the actual devices for electricity generation by the electrochemical oxidation of carbon [10,11]. DCFC technology is again receiving interest, especially in terms of the direct carbon oxidation mechanism. Gür [12,13] proposed that it is appropriate to name the carbon fuel cell a "direct" carbon fuel cell if the reaction is achieved in a single procedure or a single chamber. This proposal is based on the simplified reactions in the anode. However, it is much more complicated in practice. Generally, more than one reaction happens in the anode simultaneously. Among those reactions, carbon is often oxidized into carbon monoxide as an intermediate product, and then CO is oxidized into CO 2 as the final product. We use the term "direct carbon fuel cells" for most carbon fuel cells to be consistent with most publications.
One advantage of DCFC is the emission of pure CO 2 , unlike other coal-fired power plants, which reduce the technical difficulties of CO 2 capture and separation [14]. Notably, the DCFC power station has a modular structure design, which can be adjusted regarding the cost [15]. The direct utilization of various solid carbonaceous materials as fuels is the most distinctive feature of DCFC. Some solid carbon fuels, like graphite, carbon black [16][17][18][19][20][21][22][23] and different types of coal fuel [24][25][26][27][28][29] have been used in various DCFC configurations.
The carbonaceous materials obtained from renewable resources (e. g., biomass and organic waste) are sustainable fuels due to abundant, cheap and natural characteristics [30]. In recent decades, many researchers have focused on the applications of biochar in the environment and catalysis fields [31][32][33][34]. Biomass and municipal waste are also suggested as raw materials to produce clean energy in DCFCs [35][36][37]. Bio-char is suitable in DCFC for electric energy generation because the biochar obtained by pyrolysis is amorphous, which is conducive to the exposure of carbon fuel surface active sites. Another feature is the high porosity of this kind of biochar, promoting gas transportation in the reaction process [38]. Furthermore, some natural metal ions disperse uniformly in biochar, which are natural catalysts for the carbon gasification reaction [39]. It is worth mentioning that the efficiency of biomass-fueled DCFC is generally 50-60%, while it is up to 80% if heat and electricity co-generation is applied [40].
There are three types of DCFC; these are categorized according to the electrolyte material, i.e., hydroxide, carbonate and solid oxide. Of these, the molten carbonate direct carbon fuel cell (MC-DCFC) was one of the earliest cells, and it has been widely studied [41]. The molten carbonate also remains stable compared to hydroxide-based cells, even in an environment rich in CO 2 , which helps to prevent electrolyte damage [42,43]. Recently, an investigation was done on hybrid direct carbon fuel cells (HDCFCs), using mixed solid carbon and carbonate in the anode [21]. This overcomes corrosion issues associated with the molten carbonate fuel cell (MCFC) and yields slightly faster kinetics compared to the solid oxide fuel cell (SOFC) system [21]. Some reviews have been done in DCFC, with the earliest reviews of DCFC done by Howard [44] and Liebhafsky [45]. Later, Cao et al. [46,47] and Giddey et al. [48] summarized the fundamental electrochemical performance and the developments of DCFC technology. The reaction mechanism of direct carbon oxidation and the conversion was explained by Cooper et al. [49] and Gür [13,50]. Then, Rady et al. [51] reviewed the performance of various fuels used in MCFCs and SOFCs, and Zhou et al. [52] published a review paper that discussed the anode used in DCFCs. Recently, Jiang et al. [53] presented an overview of the impact of different parameters on the resistance and power output and the electrochemical behavior of DCFCs, and also summarized the challenges associated with developing DCFCs. Glenn et al. [54] reviewed the carbon electro-catalysis mechanism of alkali metal molten carbonates in DCFCs.
This paper focuses on providing a broad and timely summary of the carbonate-based DCFC: MC-DCFC, CO 3 2− + mixed ionic-electronic conductors (MIEC) and CO 3 2− + SOFC. The configurations of three cell systems and some possible reactions related to carbonate are shown in Fig. 1. This review starts with an introduction to molten carbonate species. Then, the fundamental mechanisms of different types of DCFC are provided, with the focus being on the functions of carbonate: (1) transport ions as an electrolyte; (2) catalyze carbon oxidation as a catalyst; (3) enlarge the reaction area, that is, the anode/ electrolyte reaction interface zone. The review concludes with a synopsis of some issues regarding corrosion, compatibility and wettability, and a possible future outlook regarding the utilization of carbonate in DCFC.

Species of molten carbonate in DCFC
Molten carbonate is commonly used in MCFC, and has received wide attention in DCFC [4,49]. It shows good compatibility with CO 2 when used as an electrolyte [55,56], and provides excellent ionic conductivity at a relatively low temperature [57]. Furthermore, the molten carbonate in the anode chamber can significantly enlarge triple-phase boundaries (TPBs), which favors ion diffusion to the electrochemical reaction sites [21]. Many studies have shown that doping alkali metal ions in molten carbonate can effectively accelerate the reaction rate of carbon gasification in a CO 2 gas stream [58][59][60]. In particular, potassium salt delivers high catalytic activity for the carbon gasification reaction [61].
The operating temperature of DCFCs is reduced by changing the carbonate composition while maintaining cell performance, which is another positive factor when using molten carbonate as the electrolyte [62,63]. A suitable amount of other carbonates or oxides can also further reduce the melting temperature of the carbonate [64]. For example, adding an appropriate amount of Ru 2 CO 3 and SrCO 3 to a binary carbonate reduces the surface tension of molten carbonate, which, in turn, increases the solubility of gas and reduces the melting temperature [65]. Therefore, using carbonate with different components in DCFC is discussed in detail.

Binary
Binary carbonate eutectic has been widely used as an electrolyte material in FCs. It is well known that a lower melting point is obtained with a binary carbonate eutectic (Li 2 CO 3 -Na 2 CO 3 , Li 2 CO 3 -K 2 CO 3 , K 2 CO 3 -Na 2 CO 3 ) than with a single carbonate [66,67]. Of the three binary carbonates, Li-K carbonate has the lowest melting point (below 550 °C). Therefore, it is more suitable for use as an electrolyte for the cell. However, an early study suggested that a Li-Na carbonate eutectic exhibited certain benefits in comparison with a Li-K carbonate eutectic, such as a higher ionic conductivity (the ionic conductivity of 43.5 mol% Li 2 CO 3 -31.5 mol% Na 2 CO 3 -25 mol% K 2 CO 3 , 52 mol% Li 2 CO 3 -48 mol% K 2 CO 3 and 72 mol% Li 2 CO 3 -28 mol% Na 2 CO 3 at 600 °C is 1.20 S cm −1 , 1.15 S cm −1 , and 1.79 S cm −1 , respectively) [68]. The Li-Na carbonate eutectic has a lower dissolution rate of in the cathode (NiO) than that of Ni generated by NiO reduction, and it is quickly oxidized to nickel oxide [69].
The cell life of different carbonate components is predicted based on the evaporated carbonate lost. Li-Na carbonate has a greater projected lifetime of 10.3 × 10 5 h at 600 °C, compared to the 6.4 × 10 5 h for Li-Na-K carbonate and the 6.3 × 10 5 h for Li-K carbonate [70]. Therefore, Li-Na carbonate has often been used as an electrolyte in MCFC to substitute the Li-K carbonate. Nevertheless, there may be a risk of a rapid decrease in cell voltage with Li-Na carbonate at atmospheric pressure and low temperature (≤ 600 °C) because of the relatively low solubility of oxygen in this eutectic [71]. Na-K carbonate eutectic showed the worst cell performance of the various binary carbonate-based DCFCs, probably due to its high melting temperature [72].
One investigation showed that, when using sawdust biofuel, HDCFC with a Li-K carbonate eutectic as the medium in anode showed excellent cell performance (789 mW cm −2 ) at 750 °C [46]. The low melting point of Li-K carbonate is one of the influencing factors on excellent cell performance because the viscosity of its solution can be reduced when the temperature reaches the melting point, thereby improving the diffusion of solid fuel in the medium. And also, Li-K carbonate is a good catalyst for gasification reactions [22]. Table 1 provides the melting temperatures of the common carbonates.

Ternary
A ternary carbonate eutectic of lithium, sodium and potassium carbonate as the carbon oxidation medium is a potential choice in DCFCs. Initially, the ternary carbonate eutectic was selected as an electrolyte in DCFC due to its good  397 [73] compatibility with unexpected pollutants, such as sulfur and ash contents in coal [10]. Furthermore, compared with the binary carbonate eutectic mixed DCFC, the ternary system (Li 2 CO 3 -Na 2 CO 3 -K 2 CO 3 ) has the lowest melting point [73].
(See Table 1.) Vutetakis et al. [74] employed a ternary carbonate of Li 2 CO 3 -K 2 CO 3 -Na 2 CO 3 with a weight ratio of 32.1:34.5:33.4 in DCFC with coal as the fuel, and found that it was capable of reducing the cell operating temperature to 500 °C. Jiang et al. [75] also applied a ternary carbonate of Li 2 CO 3 , Na 2 CO 3 and K 2 CO 3 in a mole ratio of 43.5:31.5:25 in coal-fueled HDCFC. The cell generated a maximum power density of 50.02 mW cm −2 with an opencircuit voltage (OCV) of 1.1 V at 700 °C.

Carbonate in DCFC
In DCFC, carbonate can contact carbon particles in the electrolyte or anode compartment directly, to catalyze the reaction of carbon oxidation. It can also be used as an electrolyte for ion transportation in MC-DCFC and CO 3 2− + MIEC. The difference is that, in the molten state, carbonate has a catalytic effect on carbon oxidation, due to carbonate mixing with carbon particles in the former system, but carbonate combines with MIEC materials is a solid electrolyte form in the latter. By introducing carbonate in the HDCFC anode, cell performance is improved, possibly due to the catalytic effect on carbon oxidation and the enlargement of the reaction interface. Therefore, it is critical to discuss the role of carbonate in DCFC. Table 2 summarizes some configurations used in carbonate-based DCFCs.

MC-DCFC
MCFC is a commercial fuel cell. Its scale is gradually expanded from kilowatt to megawatt [84]. Besides gaseous fuel, solid carbon is also used as fuel in this cell system. Some progress has been made with developing different cell structures and materials of MC-DCFC, followed by different electrode reaction mechanisms.

Configuration
Baur et al. [85] replaced all the molten hydroxide electrolytes with molten carbonate electrolytes, which opened up the study of MCFCs. In the 1970s, researchers from SRI International first dispersed a carbon particle in a molten Pb using a molten alkali metal carbonate as the electrolyte [86,87]. Following the first trial, the researchers used coal as the anode in an MC-DCFC with Li-Na-K ternary carbonate electrolyte [10]. Some reports claimed that some alkali metal salts (such as Li 2 CO 3 ) are regarded as a catalytic agent for accelerating carbon gasification [88,89], which further suggests the possibility of using carbonate in DCFC. Some researchers from Lawrence Livermore National Laboratory in the United States (LLNL, USA) then made significant progress in MC-DCFC, particularly in terms of the materials related to cathode catalysts, aerogels and xerogel carbon anodes [82][83][84][85][86][87][88][89][90][91][92][93][94].
The unconventional tilted design was invented by Cooper et al. [95] using molten carbonate (32 mol% Li 2 CO 3 -68 mol% K 2 CO 3 ) as the electrolyte (See Fig. 2a). The cell cathode is composed of a lithiated NiO, while the anode is made of the carbon-carbonate slurry mixture and the Ni current collector. The key aspect of this device is the electrolyte bed of ZrO 2 fabric filled with molten salt that conducts ions while preventing a short circuit of the electrodes. This device presents a 5-45° inclined angle, which favors a continuous fuel supply and facilitates the discharge of excess electrolyte to prevent the electrolyte from flooding the cathode. The cell surface area was expanded from the conventional 2-60 cm 2 with no significant polarization loss [48].
Later, researchers presented another carbon/air molten carbonate cell, as illustrated in Fig. 2b. The composite electrode composed of foam nickel and stainless steel made a great contribution to the stability of the cell structure [41]. This design was called 'planar MC-DCFC' due to the mixture of carbon and carbonate located above the anode where the redox reactions occur. A low cell resistance (the resistance of carbon aerogel and the anode current collector was 0.4 Ω cm 2 ) was obtained in this configuration, and therefore improved cell performance.
A schematic of the MC-DCFC prepared by Li et al. [96] is shown in Fig. 2c. The method applied was similar to that used by Vutetakis et al. [74]. This MC-DCFC consists of three electrodes: a working electrode (WE); a gold counter electrode (CE); and a 12 mm diameter alumina sheath that serves as the reference electrode (RE). An Inconel stirring bar is also introduced into the molten carbonate electrolyte to ensure the uniform distribution of carbon particles and improve mass transfer process inside the cell. However, the power density in this configuration is not good enough because the reaction area (fuel/electrode/electrolyte contact area) is severely limited. In this case, the reaction area is only a few square centimeters in the 250 g melt carbonate bath [97].
In an effort to extend the formation zone of TPBs, a fluidized bed cell with a three-dimensional (3D) electrode was adopted by Gür and Huggins [100]. Zhang et al. [98,101] showed another one with a self-designed fluidized bed electrode anode, as shown in Fig. 2d. In this design, the bubbling gas is applied in DCFC to ensure mass and heat transfer.
Recently, a tubular DCFC with a closed-end structure was conceived by Ido et al. [99], using carbonate as the electrolyte (See Fig. 2e). This design can effectively protect the short circuit between the electrodes using carbon powder. Additionally, the continuous solid fuel supply can be realized through calcinating the anode nickel particles outside the DCFC.

Mechanisms
In the MC-DCFC system, the solid carbon fuel can be oxidized directly to CO 2 (as given in Eq. 1). Thereafter, it is circulated to the cathode compartment through the molten carbonate electrolyte to achieve mass balance (as given by Eq. 2). The reactions are detailed below [102,103].
The total reaction is shown in Eq. (3). Planar MC-DCFC with the mixture of carbon and carbonate located above the anode to ensure low cell resistance; c MC-DCFC with a stirring rod in the molten carbonate to ensure uniform distribution of carbon particles and improve the mass transfer process inside the cell; d FBEDCFC using the bubbling gas to agitate a mixture of carbon and molten carbonate to accelerate mass and heat transfer; e TMC-DCFC with a closed-end structure that ensures there is no short circuit and that there is a continuous solid fuel supply. Reproduced from refs. [41,95,96,98,99] 12 Page 6 of 24 The above equations also indicate that P CO 2 could influence the open-circuit voltage (OCV) of the cell, as shown in Eq. (4) [41].
The possible oxidation reactions of carbon in MC-DCFC and the gas produced are shown in Fig. 3, which shows that: the decomposition of carbonate can also produce CO 2 ; it has an influence on the DCFC performance (Eq. 5) [104]. One investigation showed a noticeable increase in over-potential at a higher current density, owing to the mass transfer process being prevented, but it was easy for the released CO 2 gas to make contact with the carbon and ions at the anode again, with a long-term discharge recorded [105].
Generally, it requires a high operating temperature to enhance the anode reaction rate. Early experimental results show that the predominant product with a carbon anode is CO above 700 °C and this is dependent on the reversal of the Boudouard reaction (as given by Eq. 6). Expected energy from reaction (Eq. 1) is halved, because only two equivalent charges are obtained from one mole of carbon without a significant voltage change at 750 °C.
Other reactions besides the Boudouard reaction are possible, such as partial oxidation of carbon, as shown in Eqs. (7) and (8) [106].
The main product of the carbon anode is identified as CO 2 through off-gas analysis [107]. The electrochemical formation of CO is not an insurmountable problem in DCFC, as this reaction in the anode compartment occurs only in the process of cell operation [74]. The OCV value can be diminished or even eliminated if pure CO 2 is blown into the anode compartment under the cell when operating above 700 °C, or when the CO 2 residue in the open-circuit condition is minimal ( Table 2).
Cooper et al. [41,108] proposed that the mechanisms of carbon oxidation in MCFC are the same as in the Hall process, as given by Eqs. (9)(10)(11)(12)(13)(14), while MCFC also forms oxygen ions. The oxygen ions are formed by the ready decomposition of the molten carbonate at the DCFC operating temperature (as given by Eq. 5), which triggers the subsequent oxidation reactions of carbon.
The above reactions indicate that oxidation of carbon monoxide is key to the whole carbon oxidation process, because carbon is easily oxidized to CO with the one-electron transfer. The combination of carbon and O 2− is affected by the number and concentration of carbon active sites, depending on the carbon surface area. Li et al. [96] used different carbon sources as the fuel and proposed that the redox reaction rate in carbonate slurry is mainly determined by the crystallinity and surface properties, especially the surface area and the quantity of the carbon surface functional group. These results also verify the rationality of Cherepy's electrochemical mechanism to a certain extent.
Recently, more attention has been paid to tailoring the cell structure to improve the performance of MC-DCFC. Lee et al. [109] proposed that the addition of Gd 2 O 3 to a Ni anode improved cell performance due to the enlarged TPBs and the reduced charge-transfer resistance. They also concluded that the Ni: Gd 2 O 3 = 1:5 anode was an optimal value between the wettability and the electronic conductivity. Then, Lee et al. [78] also reported that the addition of Slow adsorption:Rate determining step C RS O − 2 = CO 2 (g) + e − Fast discharge and outgassing   [15,29] lanthanum strontium cobalt ferrite (LSCF) and MIEC to the Ni anode, at a molar ratio of 1:1, showed a better power density of 111 mW cm −2 at 700 °C compared to the single Ni anode. The excellent power output is attributed to the decreased ohmic and charge-transfer resistance and the expansion of TPBs. Additionally, Bie et al. [110] designed a novel syringe-type anode, which ensured extended region TPBs by pressing the carbon powder into the molten electrolyte, and preventing carbon oxidation.

CO 3 2− + MIEC
It is well known that ceria-based oxide is a typical transition metal oxide with mixed electronic and ionic conductivity (MIEC). Its doped oxides, including Gd-doped ceria (GDC) and Sm-doped ceria (SDC), have been applied to an intermediate temperature (IT) SOFC ranging from 400 to 700 °C [111]. The SDC-carbonate composite electrolyte shows excellent conductivity of 10 -2 to 1.0 S cm −1 in the range of 400-700 °C, which is better than a pure GDC or SDC (5 × 10 -3 -10 -2 S cm −1 ), and similar to the traditional solid oxide electrolyte-yttrium-stabilized zirconia (YSZ) at 1000 °C [112,113]. Additionally, this composite electrolyte is not subject to corrosion issues in the normal MCFC. This type of composite electrolyte has also now been popularized in DCFC and has shown acceptable performance. The cell performance was improved when using the composite electrolyte of SDC and carbonate because molten carbonate with mobility can expand the TPB at the anode. Therefore, reduced electrode polarization resistance was often achieved. It was also found that the redox reaction of carbon fuel was enhanced when using doped ceria materials. A diagram of the electrochemical process between the electrolyte and the electrode is presented in Fig. 4. It shows that two electrochemical mechanisms might be responsible for forming carbonate ions and oxygen ions in the cathode chamber filled with O 2 and CO 2 (Eqs. 2 and 15).
In the composite electrolyte of SDC-carbonate, it is apparent that the charged species are carbonate ions and oxygen ions and that carbonate ions are transferred in the molten electrolyte, while oxygen ions are the conducting species within the SDC [114,115]. The carbon particles in the anode cavity, part of which is immersed in molten carbonate, combine with carbonate ions, while others are in contact with the electrolyte of SDC directly and react with oxygen ions, which simultaneously release CO or CO 2 and generate electrons (Eqs.1, 7, 16 and 17). CO can also be produced from Eq. (6) through the Boudouard reaction, with further oxidation by oxygen ions [116].
Despite many studies done on the molten carbonate system to explore the oxidation reactions of carbon, the reaction mechanism is still not fully understood [117].
Varieties of carbonaceous fuels used in DCFC with the doped ceria-carbonate composite electrolyte have been reported in literature. Elleuch et al. [79] used lowcost solid carbon-containing coal coke, petroleum coke and almond shell carbonization as fuel for DCFC with an SDC-NiO anode, an SDC-(66 mol% Li 2 CO 3 -33 mol% Na 2 CO 3 ) electrolyte and Li x Ni 1-x O-SDC as the cathode. The results showed that the carbonized-almond shell with more oxygen-containing functional groups has good cell performance at 700 °C, with a power output of 127 mW cm −2 . The researchers further explored the electrochemical oxidation of carbon using graphite in the same cell [113]. They found that cell performance was improved with a maximum power density of 59 mW cm −2 at 700 °C when adjusting the anode environment (N 2 atmosphere). In CO 2 atmosphere, the value reached 37 mW cm −2 [113].
Recently, a dual 3D ceramic textile electrode was integrated into a GDC-carbonate composite electrolyte-supported DCFC, and thus, the TPB region was expanded [118]. Bian et al. [81] developed a unique electrolyte-supported DCFC consisting of a GDC-(67 mol% Li 2 CO 3 -33 mol% Na 2 CO 3 ) electrolyte, a NiO-GDC anode and a Sm 0.5 Sr 0.5 CoO 3 -GDC cathode. It exhibited unprecedented cell performance at 600 °C, with a power output of 392 mW cm −2 when using graphitic fuel, due to the enhanced charge and mass transfer on the electrode below 600 °C [81]. High fuel utilization of 87.3% was also achieved, as the carbon fuel could quickly reach the

CO 3 2− + SOFC
Recently, HDCFC was combined with SOFC technology and MCFC technology to provide a new way for carbon fuel to reach the reaction region [21]. This method ensures that the molten carbonate is incorporated into the anode cavity, which significantly expands the reaction region from a two-dimensional to a three-dimensional region, and this accelerates the mass transfer to the solid anode/electrolyte. The carbonate serves as a medium to promote the complete oxidation of carbon [23,119]. In this system, the solid oxide electrolyte separates the anode from the cathode and prevents the diffusion of carbonate to the cathode with no risk of carbonate corrosion on the cathode. Another feature is that no need for the CO 2 cycle. The cathode is exposed to air, which simplifies the cell structure.
The first HDCFC was demonstrated by SRI International [120], and was further developed by the University of St Andrews [21,121,122], Contained Energy [123] and Technical University of Denmark [124,125]. An early tubular HDCFC using a Pt cathode and YSZ electrolyte was designed by the University of St Andrews, with the anode nickel mesh placed in a mixture of carbonate and carbon [126]. However, it is difficult to observe the reaction at the anode in this configuration, as the active zones of the electrolyte/electrode are uncertain in a closed-cell above the melting point of the lithium and potassium carbonate [126,127]. The researchers at the University of St Andrews also developed a planar button cell with better sealing for improved gas purification [128]. The electrochemical reaction mechanism in the anode was also investigated in detail [21,122]. To date, the highest power density was obtained by Jiang et al. [15] using pyrolyzed medium-density fiberboard in HDCFC, which reached 878 mW cm −2 at 750 °C.

Reaction mechanism
HDCFC is based on two typical fuel cells-SOFC and MCFC-in which: the solid oxide electrolyte (which includes YSZ, GDC or SDC) separates the electrode chambers; the molten carbonate electrolyte has fluidity at a high temperature and expands the oxidation reaction area [21]. The oxygen ions reduced from the oxygen molecule are transmitted from the cathode to the anode compartment through the solid oxide electrolyte (as given by Eq. 15). In the anode compartment, carbon particles may be completely oxidized to CO 2 or partially oxidized to CO (Eqs. 16 and 17) [129,130]. The OCV value would be 1.02 V, if Eq. 16 was the only anode reaction, regardless of temperature.
An OCV of approximately 1.5 V was observed by a current study, which is higher than the theoretical voltage with working temperatures of 550-700 °C in HDCFC [119]. A higher OCV value indicates that other reactions are also underway apart from the direct electrochemical oxidation of carbon to CO 2 . Figure 5 shows that some reactions and some gas production occur in the anode compartment of HDCFC, which is full of CO 2 and CO at the same time. In the anode chamber filled with nitrogen, the reaction processes of carbon (both electrochemical and chemical) are complicated. In the anode, both oxygen ions and carbonate ions are active species of electrochemical oxidation in the slurry of carbon/carbonate [21]. When the number of oxygen ions is sufficiently high, the CO 2 could be converted to carbonate ions, which would result in a slow decline of CO 2 activity in the molten carbonate, as indicated by Eq. (18). With continued consumption of the oxygen ions, the carbonate ions oxidize carbon to CO 2 or CO, as indicated by Eqs. (7 and 8). Carbonate ions then regenerate from oxygen ions and CO 2 as indicated by Eq. (18), which maintains the electric charge balance in the molten carbonate solution. Some reports have claimed that the low activity of CO 2 could be due to it being dissolved in molten carbonate by physical or chemical methods [131,132], which may increase the Nernst potential. Nevertheless, the presence of molten carbonate, which facilitates the flow of carbon particles to the anode chamber and expands the TPBs, is expected to act as an electrochemical mediator and accelerate the oxidation reaction kinetics of the carbon particles [13,41]. In addition, the non-electrochemical reaction of the Boudouard reaction that occurs at 750 °C has a strong influence on the entire anode reaction, which consumes carbon through a chemical reaction and causes a sharp decrease in current density [28]. But in terms of long-term stability, the influence of current density is not obvious [29]. However, the CO produced by the reverse Boudouard reaction can also generate electricity through electrochemical oxidation, which contributes to higher-power output [133].
Deleebeeck et al. [124] compared the effects of different anode gases on cell performance and found that introducing pure CO 2 could reduce the mass transfer limitation by facilitating the Boudouard reaction or preventing the carbonate decomposition. In addition, introducing N 2 leads to a high OCV, due to the movement of CO 2 or CO [134].
Recently, Li et al. [135] also found that introducing CO 2 into the anode chamber could improve cell performance through the electrochemical reaction (Eqs. 19 and 20) at 700-800 °C. However, poor performance by filing CO 2 was obtained at 650 °C, as a low temperature is unfavorable for the reverse Boudouard reaction. (20) CO + CO 2− 3 = 2CO 2 + 2e − Given this prior research, Lee et al. [136] designed threefuel cells that provided different contacts with the anode and analyzed the oxidation of carbon fuel by observing the current density using different voltages. The various means of contact method between the carbon and anode are presented in Fig. 6a. Cell I, cell II and cell III indicate: direct contact, physically separated, in contact with the carbonate medium. The current density of the cell at 800 °C, when using a different voltage load, is shown in Fig. 6b. It shows that the values of i I and i II were lower than that of i III at a given voltage. Therefore, adding carbonate increases the current density significantly, proving that the carbonate ions dominate the oxidation reaction (See Fig. 6c. Recently, Jiang et al. [137] designed an experiment to investigate the possible reaction active sites at the anode chamber in HDCFC. The configuration of the HDCFC design is shown in Fig. 7a. The current collection wire could be moved to allow easy adjustment of the space between the current collector and the layer at the anode (L), to further ensure that the reactions occur in the area containing the carbon and carbonate. Figure 7b and c show the AC impedance spectra at 700 °C when using an anode current collector of Au and Pt and different Ls. This demonstrates that the ohmic resistance was determined by the location of the current collection with unchanged polarization resistance. The Fig. 6 The relationship between the different contact modes of the anode and the cell performance of DCFC: a three different contact methods for carbon fuel and the anode; b comparison of current den-sity using different voltage levels at 800 °C; and c relative contribution value of various reaction mechanisms to the total current. Reproduced from ref. [136] unchanged polarization resistance with the current collection position also indicates a possible extension from a 2D anode in the TPBs to a 3D area filled with carbon and carbonate for all instances of carbon oxidation.
Apart from the temperature, the carbon oxidation process at the anode also depends on the electrode and electrolyte. Jiang et al. [122] further explored the reaction mechanism of the anode by exchanging the carbonate content in the HDCFC system with a configuration of NiO-YSZ/YSZ/LSM, while using 62 mol% Li 2 CO 3 -38 mol% K 2 CO 3 in the anode chamber. They observed reduced polarization resistance of the cell with a carbonate content of 20 mol% or 50 mol%. A different result was found when more carbonate was applied, and higher polarization resistance was achieved when the carbonate increased to 80 mol%. This finding demonstrates that carbon particles are prevented from reaching the electrode by a high concentration of molten carbonate, which has a limited effect on various carbon oxidation reactions.
Some corrosion and sealing issues are significant concerns when using high carbonate concentrations. Cantero-Tubilla et al. [72] proposed that the liquid phase was related to the carbonate content, and therefore the rate of movement of carbon fuel slowed in the presence of high carbonate concentrations. The melt carbonate could result in a concomitant increase in polarization resistance and, consequently, a significant decline in electrochemical output.

Chemical catalysis
McKee et al. [133][134][135][136][137][138][139][140] used a definite redox cycle and postulated that alkali metal carbonates (Li 2 CO 3 , Na 2 CO 3 , K 2 CO 3 , etc.) produce catalytic activity, which is helpful in the process of gasification of carbon under an oxygen or carbon dioxide atmosphere. In a flowing oxygen atmosphere, carbon-induced decomposition of metal carbonates (M 2 CO 3 ) into metal oxides (M 2 O), as shown Eq. (21) [138]: At a higher temperature, M 2 O is further oxidized to peroxide or higher oxide (M 2 O 1+n ), and eventually reduced to M 2 O by carbon, which completes the redox cycle-see Eqs. (22,23) [139]. Nevertheless, it is unlikely that the reaction of the oxide-peroxide cycle will take place in a flowing carbon dioxide atmosphere because the conversion of M 2 CO 3 to M 2 O is difficult. Instead, M 2 CO 3 is reduced to the elemental state, while C is oxidized to CO: Then, the metal element generates metal oxide in the carbon dioxide atmosphere: After that, carbonate is formed by the metal oxide and carbon dioxide, which completes a cycle reaction [138]: Nagase et al. [141] proposed that some metal carbonates also had an influence on the reverse Boudouard reaction in an inert atmosphere, which acted as the catalytic media in the process of electrochemical oxidation or gasification of carbon.
Recently, Li et al. [22] studied the reaction mechanism of the HDCFC anode filled with N 2 . These researchers found that more CO was produced using eutectic salts (Li 2 CO 3 -K 2 CO 3 ) as the catalyst media, than the absence of carbonate, which therefore improved cell performance. Two reactions produced the carbon dioxide (Eq. 27) and the carbon monoxide (Eq. 28), with both converting M 2 CO 3 to M 2 O: Lastly, more CO and electrons were generated by series of reactions (Eqs. 29 and 30) than in an oxygen atmosphere:

Issues with the presence of carbonate
Even though DCFC is being developed continuously, some issues, like materials and technology, must be addressed. Most molten carbonates are corrosive and reactive, which raises widespread concern about the thermal corrosion and chemical stability of cell materials and the chemical composition of solid fuels. Besides the ash of solid carbon fuels, the wettability of carbonates also affects cell performance. These problems ultimately lead to a significant reduction in conversion efficiency and long-term stability.

Corrosion
The highly corrosive action of molten carbonate has prevented progress from being made with MCFC technology. From 1996 to 2006, the cell lifetime was greatly increased from just a few months to 2 years [142]. However, when various carbonates are used as the electrolyte contacted by cell components, it still results in thermal corrosion attacks (such as oxidation, carburization and flux reactions) at high operating temperatures [143]. The corrosion issue gives rise to severe challenges in improving the chemical and physicochemical stability of electrode materials for MCFC. Although some new power generation devices are emerging, stability has always been an issue when developing new technologies.
At present, nickel is one of the most widely used electrode materials because of its low price and good performance in conductivity and electro-catalytic ability. However, Ni and NiO are easily dissolved into the molten carbonate, which causes an internal short-circuit in MCFC or a short lifespan [144]. It was reported that the Ni/NiO solubility in carbonate could be reduced by adding some oxides (e.g., SrO or MgO [140][141][142][143][144][145][146][147][148]) or carbonates (e.g., CaCO 3 , BaCO 3 , or SrCO 3 [146,149]) to alkali carbonate, due to the change in the pH value of the solution.
Doping a rare-earth metal [150] or a rare-earth metal oxide (such as La 2 O 3 , Y 2 O 3 , and Yb 2 O 3 ) [146][147][148][149][150][151][152][153] is another practical approach to protect the nickel or nickel oxide electrode against dissolution in molten carbonate. For example, a Ni-Ce cathode showed more durability (136 mW cm −2 during 2100 h) than a commercial Ni cathode material (but the voltage decreased from 120 to 108 mW cm −2 in 1000 h) [150]. Further, Liu et al. [154] observed that adding the rare-earth metal Dy to a mixture of Li 2 CO 3 and K 2 CO 3 in a mole ratio of 62:38 could reduce the NiO/ Ni passive anodic current, inhibiting the outward diffusion of Ni at 650 °C, and eventually increase the resistant ability to corrosion.
In terms of HDCFC, the solid oxide electrolyte separates the two electrodes and prevents the cathode from carbonate corrosion. Nevertheless, the chemical compatibility between the electrolyte and the carbonate is essential because of the contact-attack. Suski et al. [155] reported on a double-cell with YSZ and an electrolyte of 53% Li 2 CO 3 -47% Na 2 CO 3 , with the OCV of O 2 + Ar/YSZ/Au at TPBs being similar to the OCV value of the reference electrode over 1000 h of operation. The long operation hours indicate that the kinetics of the chemical reaction between YSZ and Li 2 CO 3 -Na 2 CO 3 is very slow. However, another report used XRD analysis to determine that YSZ immersed in Li 2 CO 3 -K 2 CO 3 at 700 °C for 10 days was converted to lithium zirconate, but with no change K 2 CO 3 -Na 2 CO 3 [126].
Jiang et al. [15] did a more detailed investigation of YSZ corrosion. First, they confirmed that the YSZ of 5-10 μm did not change for 13 h in an atmosphere of 97-3% H 2 O. Then, several corrosion experiments were done on the YSZ electrolyte in molten carbonate of 62 mol% Li 2 CO 3 -38 mol% K 2 CO 3 when using different atmospheres at 700 °C for 10 h. Figure 8 shows the corrosion of the YSZ surface when using different atmospheres. With the air atmosphere, the grain boundaries of YSZ are destroyed and new particles appear, as shown in Fig. 8b, which may be due to the formation of Li 2 ZrO 3 , K 2 ZrO 3 or LiKZrO 3 [151][152][153][154][155][156][157][158]. Some wrinkles appear on the surface of YSZ and the grain boundaries become inconspicuous in a reducing atmosphere, as shown in Fig. 8c, but less significantly than in the air test. There is no significant change after testing performance using 5% H 2 -95% Ar, as shown in Fig. 8d. The insignificant change is possible because of the formation of Li 2 ZrO 3 being prevented in a reducing atmosphere [159].
The researchers used the basicity model of Lux-Flood to explain the experimental result. The formation of oxides and superoxides was more likely to occur in an oxygen environment, which was the primary reason for corrosion [43]. It is common to purge a carrier gas (e.g., N 2 or CO 2 ) in the anode chamber. The electrolyte material, YSZ, is either in the reducing environment at the interface of the electrolyte/ anode or in the oxidizing environment at the interface of the electrolyte/cathode. Therefore, it is necessary to investigate the stability of YSZ in both the reducing atmosphere and the oxidizing atmosphere to simulate both operating conditions [159,160]. According to Xu et al. [161], YSZ is a good choice of electrolyte for HDCFC, due to the reducing atmosphere of the anode chamber.
The doped cerium oxide-based material is a typical solid electrolyte in DCFC that shows good stability below 650 °C, due to the complete microstructure being maintained at 650 °C and 10% H 2 -90%N 2 for 1000 h [162]. Xu et al. [161] further explored the stability of SDC in 62 mol% Li 2 CO 3 -38 mol% K 2 CO 3 in air, and found that YSZ easily reacted with K 2 CO 3 to form Li 2 ZrO 3 at 700 °C after 10 days  [15] of testing. And, there is no new phase formation in the SDC sample when using the same conditions. This suggests that SDC is more resistant to corrosion than YSZ, when using air. Although there was no impurity formation of SDC at 700 °C after 10 days in air, cracks appeared at the grain boundaries [163]. This suggests that crystal boundaries might be the most preferred site for corrosion development [163], as shown in Fig. 9. In addition, the resistance result also showed an insignificant increase in the ohmic resistance of the SDC electrolyte at 650 °C during the 70-h testing, which may be due to the slow corrosion rate along the grain boundary. As mentioned, the corrosion of molten carbonate can be ameliorated by different methods, but cannot be avoided. Therefore, it is vital to develop nascent corrosionresistant materials, to maximize the advantages of carbonate for DCFC devices.

Ash
Solid carbon fuels come from a wide range of sources and have a unique constituency. The molten carbonate is likely to change chemically due to interaction with some ash components, including heavy metals and their oxides, sulfides and chlorides, which ultimately lead to cell failure [74]. However, some metals in ash have a positive effect on cell performance. For example, Cao, MgO and Fe 2 O 3 have a catalytic effect on the oxidation reaction of solid carbon, which increases the current density [24]. A mixture of Fe m O n , Li 2 O, K 2 O and CaO is the catalyst for the Boudouard reaction [26] and it can improve the reaction activity and electrochemical performance of the fuel [164]. Therefore, studying the effect of impurities in the ash on cell materials is key to improving the performance of DCFC.
It has been reported that coal with a low ash content shows better cell performance than high ash [29,[160][161][162][163][164][165][166][167][168]. Ju et al. [169] compared the electrochemical performance of different grades of coal, and found that ash-free coal (low ash content) exhibited the most prolonged stability (300 min at 50 mA cm −2 and 900 °C), while the raw coal was under operation for 50 min before failure due to blockage of the anode channel by ash and thus prevention of the reaction. The high ash content in raw coal is considered detrimental to the short-term stability with a Ni-YSZ anode, as the ash blocks the anode and causes contact between the reaction interface and the carbon interface during cell operation [167].
One study reported that when using Al 2 O 3 , SiO 2 and TiO 2 from coal ash, it is easy to generate a passivating membrane on the electrode interface during operation of the cell, which leads to passivation of the electrode [74]. In particular, SiO 2 is likely to react with carbonate electrolytes, to generate CO 2 and M 2 SiO 3 (where M = Li, Na, K) in MCFC, leading to additional electrochemical reactions [170]. Tulloch et al. [171] reported that coal with about 70 wt% SiO 2 ash showed a significant decrease in current density for MC-DCFC, as SiO 2 inhibits the oxidation behavior of carbon.
Recently, different pretreatment methods have been proposed to remove unwanted impurities and enhance cell performance [167][168][169][170][171][172][173][174][175]. Eom et al. [172] reported that using HCl to pre-treat coal could ensure maximum reduction of ash content and the sensitivity of cell performance to surface silicon content may be reduced when the temperature is higher than 733 °C because Li 2 SiO 3 can be completely decomposed at 733 °C [176]. Xie et al. [174] modified bituminous coal with acetic acid, and effectively removed ash of about 84 wt% Si and 64 wt% Al, to optimize cell performance. Fig. 9 Cross section of the SDC for the corrosion test in Li-K carbonate at 700 °C in an air atmosphere for 10 days: a before the corrosion test; b after the corrosion test. Reproduced from ref. [161] It has been reported that some impurities in ash may promote an electrode reaction, e.g., CaO, MgO and Fe 2 O 3 could improve the current density slightly by acting as a catalyst in MC-DCFC [24,177]. Cai et al. [39] employed orchid leaf char in SO-DCFC with yttrium-stabilized zirconia electrolyte and an Ag-GDC electrode. The researchers concluded that the natural Ca in biochar derived from orchid tree leaves exerted a catalytic effect on the reverse Boudouard reaction and enhanced the performance of DCFC. Then, they loaded 5% Ca on wheat straw char in the same cell configuration and recorded a higher-power performance (258 mW cm −2 ) compared to that of unloading char (197 mW cm −2 ) at 700 °C [178]. Hao et al. [36] further confirmed the positive effect of calcite and magnesium calcite in magazine waste paper, which indicated that more amorphous carbon exhibited a higher degree of carbon oxidation also catalyzed the gasification reaction of carbon fuel and delivered a higherpower density.
Recently, a DCFC supported by an SDC-(67% Li-33% Na) 2 CO 3 composite molten carbonate showed excellent cell performance (378 mW cm −2 ) at an operating temperature of 750 °C. The enhanced cell performance was because the KCl in raw reed ash increases the degree of structural disorder of biochar during the pyrolysis process, which leads to the high oxidation activity of reed char [179].

Wetting
Complete wetting of the carbon particles in eutectic carbonate is critical for the charge transfer at the TPBs zone [180,181], as this is a dominant factor in ensuring uninterrupted round-the-clock power generation by the DCFC. During the operating process, the eutectic carbonate is completely diffused and penetrates the carbon pores. Therefore, the potential of a stable cell can be reached [49]. When measuring the contact angle of a graphite rod in 62 mol% Li 2 CO 3 -38 mol% K 2 CO 3 at 520-560 °C, Chen et al. [182] found that wetting of carbon in carbonate is driven by capillary force, as well as being largely determined by CO bubbles produced from the reverse Boudouard reaction. The OCV increases from 0.5 to 0.81 V during the process of wetting, which contributes to the adsorption of CO on the surface. Therefore, no matter what carbonate is employed in DCFC, as long as the solid carbon is completely soaked in the molten carbonate to form the interface where the carbon oxidation reaction occurs, the electrons can be continuously generated and transferred.
The wettability of solid carbon fuel in a molten carbonate is enhanced by pre-treatment to increase the surface area and the pore volume of the carbon particles [183]. The effect of base or acid pre-treatment on the electrochemical activity of carbon in lithium and potassium carbonate has also been examined [183], with HF, HNO 3 and NaOH are the common acid and base materials used. Of all the samples, activated carbon pre-treated with HF exhibited the highest electro-oxidation activity, with an increased current of approximately 50 mA cm −2 . The enhanced current is because HF pre-treatment increases the porosity and surface area of the carbon particles, forming a double electrical layer that facilitates electron transfer of carbon during anodization.
It has also been shown that carbon particle wettability relies on carbonate composition. Watanabe et al. [184] indicated that the surface tension of solid carbon could be reduced by lowering the amount of sodium carbonate and thus improving its wettability. The test results showed that carbon particles had a better wetting effect in a ternary carbonate of 12.2 mol% Li 2 CO 3 -25 mol% Na 2 CO 3 -62.8 mol% K 2 CO 3 compared to 12.2 mol% Li 2 CO 3 -45 mol% Na 2 CO 3 -42.8 mol% K 2 CO 3 , and that a stable OCV of 0.63 V could be sustained for 45 min.
In terms of DCFC with molten carbonate, the degree of stirring is also an important variable that affects carbon particle wetting. Li et al. [96] observed that mass transfer has been significantly improved at 400 rpm, and that current density can be further increased in the whole electrode potential range when the stirring rate is increased up to 600 rpm. Improved current density when increasing the stirring rate aligns with the earlier report of Vutetakis et al. [74]. However, the performance of DCFC will be greatly reduced if the stirring rate exceeds a specific value, because the fuel splash phenomenon leads to a fuel shortage [4]. Whether the stirring improves cell performance or not depends on other factors, such as cell operating temperature and electrolyte viscosity [185].
Carbonate has good ionic conductivity but poor electronic conductivity at a high temperature. Although carbon particles with different sizes are dispersed in molten carbonate, the electrochemical oxidation of carbon occurs only in TPBs, thus the percolation limit of carbon particles must be considered, to ensure efficient and continuous power generation [186]. Using highly conductive carbon fuels such as graphite could improve electronic conductivity, but the chemical activity of graphite is very low. At present, molten metal anodes can solve this problem and ensure good conductivity [187]. However, the corrosion issue remains unresolved.

Outlook
With the increase in environmental concern and the high energy demand, DCFC may provide a clean electrochemical device for generating electricity from solid carbon. The introduction of carbonate in the DCFC system is conducive to improving cell performance of open-circuit voltage and output power. The carbonate accelerates ion transfer as a medium or is a catalyst for carbon oxidation and gasification reaction.
At present, research on basic theory is still an essential aspect of the development of DCFC. Although various DCFC systems that use carbonate are being developed rapidly, the specific reaction mechanism with carbonate involved in the whole-cell system still lacks sufficient proof of experiments due to the complexity of reactions inside the cell. Therefore, new experimental configurations should be designed. Alternatively, a combination of a theoretical model and experimental tests should be considered, which would be beneficial to cell performance if determined what substances affect the electrode reaction process.
The issue of material corrosion in molten carbonate has not been solved satisfactorily. Despite a plethora of literatures that have been published to solve the problem of corrosion of carbonate in MC-DCFC, there is still no detailed research on the corrosion of carbonate to the solid-state electrolyte for emerging HDCFCs. It is unknown whether adding some carbonates, oxides or rare-earth elements has a positive or negative effect on cell performance. It has been confirmed that the addition of carbonate to the anode favors expanding TPBs, because the fluidity of molten carbonate could promote the transfer of carbon fuel to the anode, resulting in a dramatic improvement of electrochemical reactions. However, carbonate attacks the solid electrolyte. Although SDC has better corrosion resistance than YSZ, the corrosion issue cannot be avoided. Provided that corrosion occurs during the entire cell operation, the cell will degrade quickly and cannot generate power over the long term. Hence, more attention needs to be focused on the carbonate corrosion to other cell components (including solid-state electrolytes), instead of only common electrode materials.
It is well known that the variety of solid carbonaceous fuels is one of the most significant advantages of DCFC. Nonetheless, the carbonate component in DCFC easily reacts with the inorganic salts of fuels ash-especially coal and biomass fuels with a high ash content-which shows various effects on cell performance. For example, a large amount of Fe, Mg and Ca promotes gasification of carbon, while Si and Al have a negative effect on the cell performance of DCFC with carbonate. It is difficult to identify the effect of each inorganic salt on cell performance because there is often more than one salt in the fuel.
Removing ash from non-pure carbon fuels such as biomass should also be considered seriously. Pre-treatment should be attempted as an essential treatment method applied to DCFC (including heat treatment, acid or base washing and air plasma, etc.).
Other issues worth mentioning include the effect of pre-treatment on the properties of the fuel itself and the reactivity of the pretreated fuel with carbonates to eventually explore the feasibility of using different renewable carbon-based materials in the cell. While many cell designs have been developed to provide better continuous power generation, the fluidized bed cell can provide a more convenient feeding mode. Thus, designing different cell configurations is still key to the commercialization of DCFC.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.