Catalytic Hydrogen Generation and Use for Production of Fuels

The Special Issue of the Energies journal related to the hydrogen production from formic acid decomposition was published recently by MDPI. This Editorial note contains a short analysis of the papers published in this Special Issue and some historical information connected to this reaction.

Hydrogen production from different hydrogen carriers is an important topic that has been examined in many studies, as hydrogen is considered to be a clean fuel for the future, giving only water as a product. At the same time, efforts for its storage and transportation may encounter serious difficulties related to safety. Hence, it is necessary to develop liquid and solid hydrogen carriers that will allow the safe storage/transportation of hydrogen and its liberation at mild conditions using catalysts. Therefore, many recent studies in this field have focused on the development of efficient catalysts for hydrogen production from different compounds.
MDPI published the Special Issue "Advanced Catalysis in Hydrogen Production from Formic Acid and Methanol" in the Energies journal and as a separate book, for which I served as Guest Editor. Formic acid and methanol are liquid organic hydrogen carriers (LOHCs) that can be produced from biomass [1] or by CO 2 hydrogenation [2,3]. The Special Issue included five invited research papers and two invited reviews. Unfortunately, no papers related to the production of hydrogen from methanol were presented; hence, the issue was focused on different aspects of hydrogen production from formic acid using heterogeneous and homogeneous catalysts ( Figure 1).
Studies of the catalytic decomposition of formic acid have more than 150 years of history and have contributed significantly to the science of catalysis. Formic acid decomposes through two routes, giving hydrogen and carbon dioxide (1, dehydrogenation) and carbon monoxide and water (2, dehydration).
HCOOH → CO 2 + H 2 (1) The German chemist Döbereiner was the first to report the dehydration of formic acid by heating it with anhydrous sulfuric acid in 1821 [4]. This reaction was studied in detail at the beginning of the last century. Some heterogeneous catalysts were tested. If the goal is to produce pure hydrogen, dehydration should be eliminated using catalysts. However, dehydration is important for applications where formic acid is applied as a CO source-for example, synthesis gas production [1,5].
The French chemist Marcellin Berthelot was the first to report the dehydrogenation of formic acid in 1864 [6]. He conducted the decomposition of formic acid vapor over 15 g of Pt black powder at 443 and 533 K and showed that equal amounts of CO 2 and H 2 were formed. Later, the Nobel Prize Winner in Chemistry (1912) Paul Sabatier and his co-author A. Mailhe showed that some heterogeneous catalysts such as Pt, Pd, Ni, Cu, Cd, SnO 2 , and ZnO were active in dehydrogenation reaction, while oxides of Ti, Al, Si, Zr, U, and W were active in dehydration reaction [7]. Formaldehyde was also formed sometimes. It should be mentioned that Sabatier was an assistant in the laboratory of Berthelot and received his Doctor of Science degree under his supervision in 1880 [8]. In 20th century, formic acid decomposition was studied as a model reaction used for the establishment of the bases of catalysis and for the development of spectroscopic methods. Later, Density Functional Theory (DFT) studies were performed to elucidate the mechanism of the reaction.
Formic acid contains a considerable amount of hydrogen (53 g L −1 ). Since 2008-2010, it has been considered as a substance that could be used for hydrogen storage. Homogeneous catalysts have been used intensively for liquid-phase formic acid decomposition and for formic acid production from CO 2 and H 2 [9][10][11][12][13][14]. At the same time, heterogeneous catalysts have also been used for liquid-phase [15] and gas-phase [16,17] reactions with the purpose of producing CO-free hydrogen efficiently. These catalysts possess a significant advantage with respect to homogeneous catalysts, allowing to separate them easily from the reaction mixture [18,19].
There are several ways to achieve efficient hydrogen production from formic acid using catalysts. These involve support modification, metal size optimization, the alloying of active metal with another metal, and promotion with basic compounds containing alkali metals or amines. All these methods are considered in this Special Issue. The reviews featured to discuss liquid-phase formic acid decomposition over bimetallic (PdAg) [20], molecular (Ru, Rh, Ir, Fe, Co, and others) [21], and heterogenized molecular Fe catalysts [21]. The gas-phase reaction is studied over highly dispersed Pd [22], Pt [23], Au [24], Cu [25], and Ni [26]-supported catalysts.
Stathi et al. [21], in their review, discuss the features of the production of hydrogen from formic acid over a set of homogeneous catalysts represented by metal complexes. The mechanism of the liquid phase reaction was analyzed based on DFT studies. The authors indicated the importance of the deprotonation of formic acid as a first step of the reaction. They also outlined other important factors: the nature of the solvent and ligand, temperature and pH. Moreover, they discussed the continuous operation of hydrogen production from formic acid and showed some examples where immobilized molecular catalysts demonstrated comparable properties to nonimmobilized catalysts. This latter feature is important for the creation of immobilized molecular catalysts [18].
Navlani-García et al. [20], in their review, discussed the properties of PdAg catalysts in the production of hydrogen from formic acid. The interest in these catalysts is based on their high performance in liquid-phase reactions [15,27,28]. The authors ascribed the enhancement displayed by PdAg catalysts as compared to their monometallic counterparts to several effects, such as the formation of electron-rich Pd species and increased resistance to CO poisoning. Additionally, the authors considered the photocatalytic decomposition of formic acid. They also concluded that very little is known about the performance of the catalysts in highly concentrated formic acid solutions. It would be interesting to study the decomposition of gas-phase formic acid over highly dispersed PdAg catalysts.
The nature of the catalyst's support plays an important role in the reaction when the metal is highly dispersed. Sobolev et al. [24] studied the gas-phase formic acid decomposition over Au catalysts with a mean size of 1.6-2.4 nm supported on SiO 2 , TiO 2 , and Al 2 O 3 and compared their catalytic properties with the properties of the corresponding supports. They concluded that the undesirable dehydration pathway was provided by the acid-base properties of the support. Thus, the selectivity in hydrogen production was found to be very low for the Au/TiO 2 catalyst (<20%), but it was close to 100% for the Au/Al 2 O 3 catalyst. At the same time, the conversion of formic acid over the former catalyst was entirely provided by the support, since the conversion temperature dependences were almost the same. Earlier, we also indicated the importance of the nature of the support for highly dispersed Au catalysts and this reaction [29,30]. Gold supported on an N-doped porous carbon support showed a higher hydrogen yield than gold supported on silica or alumina.
In this Special Issue, the effect of carbon doping with nitrogen species was analyzed in detail for Pt [23], Pd [22], and Ni [26] catalysts. Nitrogen was inserted into the structure of carbon supports by direct synthesis using N-containing precursors [23,26] or by the post-deposition of N-containing precursors on the surface of the carbon support with optimized properties [22]. The latter method allows the scaling of the synthesis of N-doped carbon and could be useful for industrial applications.
It is important that the N-doping of the carbon support exerts a significant promotional effect. One of the reasons for the high activity of N-doped catalysts is the improvement in metal dispersion and formation of single-atom metal sites stabilized by pyridinic N species present in the support [19]. For example, Podyacheva et al. [23] demonstrated the higher activity of single Pt atoms supported on N-doped carbon nanotubes or carbon nanofibers as compared to Pt atoms on the surface of Pt nanoparticles. A correlation with the concentration of pyridinic N atoms was shown. In a recent paper [31], this group demonstrated the same effects for Pd catalysts.
Golub et al. [22] used different N-containing precursors (bypiridine, phenanthroline, and melamine) to deposit over a graphitic carbon support (Sibunit). The best performance was discovered for Pd catalysts, for which melamine was used as the N-precursor. The deposition of melamine followed by pyrolysis led to an increase in the activity and selectivity and to a decrease in the apparent activation energy. Thus, the catalytic activity of the N-doped Pd catalyst was higher by a factor of 4 than that of the Pd catalyst supported on the N-free carbon support at 373 K.
The properties of non-noble metal catalysts (Ni and Cu) were also discussed in this Special Issue. Normally, the activity of these catalysts is lower than that of noble metals such as Pt and Pd, but their price is also lower and this is important. Nishchakova et al. [26] studied Ni catalysts supported on N-doped and N-free porous carbon prepared at different temperatures. A temperature equal to 1073 K was found to be optimal for the catalytic activity. The N-doped Ni catalysts possessed a high stability in the formic acid decomposition reaction and a slightly higher activity than the N-free catalyst, with a mean particle size of 3.9 nm. A further study in this field [32] performed by the same group demonstrated that the used N-doped catalyst was a single-atom catalyst with active Ni-N 4 sites.
Pechenkin et al. [25] used a Cu-containing catalyst supported on a CeO 2 /γ-Al 2 O 3 support. The gas-phase decomposition of formic acid was studied in detail. The authors showed a very high yield of hydrogen equal to 98% at temperatures of 473-573 K. This yield was higher than those obtained for methanol and dimethoxymethane steam reforming reactions. The catalyst was stable in the reaction conditions used.
Therefore, key problems related to catalytic activity in hydrogen production from formic acid were discussed in this Special Issue. Interestingly, the amounts of noble metals in the Pt [23] and Pd [22] catalysts discussed in this issue were by a factor of 10 5 lower than those used in the experiments carried out by Berthelot [6] at similar temperatures. This reflects the progress made in the development of catalysts during the last 150 years, indicating that modern catalysts are significantly more active.
Finally, the results reported in this Special Issue can be applied for the development of catalysts for hydrogen production not only from formic acid, but also from other organic hydrogen carriers.

Introduction
Hydrogen is mainly used for ammonia synthesis and the petrochemical industry. Its traditional production involves non-renewable sources and processes giving a significant emission of carbon dioxide leading to global warming. Among these processes are steam reforming of natural gas and gasification of coal performed at very high temperatures (>900 K). Recently, the International Energy Agency reported that the hydrogen production reached 75 mln of tons and that it was accompanied by emission of 830 mln tons of CO 2 [1]. Global demand for hydrogen increases from year to year accompanying by an increase of the carbon dioxide emissions.
Despite hydrogen is a clean energy carrier its safe transportation and storage are rather complicated. Liquid organic hydrogen carriers (LOHCs) are used for safe storage and transportation of hydrogen [2,3]. They can be produced from biomass or CO 2 thus avoiding the effect of the evolved CO 2 for global warming. Formic acid (HCOOH) is an example of such a LOHC. It contains 53.4 g L −1 hydrogen (4.4 wt %), which is by a factor of 2 higher than the content of compressed hydrogen at 350 bar at the same volume. This amount corresponds to the energy density of 2.1 kWh L −1 . In contrast to hydrogen, formic acid can be easily transported and stored and its application is much safer. An important feature of using formic acid is that it can be produced by catalytic hydrolysis/oxidation of biomass with high yields at low temperatures (<423 K) [4][5][6]. Hydrogen can be released from formic acid using catalysts at even lower temperatures ( Figure 1). Thus, transformation of biomass to hydrogen through formic acid could be considered as an efficient route, because direct gasification of biomass also giving hydrogen demands very high temperatures (>900 K) ( Figure 1). Recently, Zhang et al. [7] and Park et al. [8] demonstrated the proof of concept for such an approach. The hydrogen obtained from formic acid could be further transformed to electrical energy ( Figure 1). The development of a compact integrated 25 kW system which converts formic acid to power has been discussed [9]. Formic acid could be used also as a donor of hydrogen instead of molecular hydrogen to hydrogenate different organic substances for production of fuels and intermediates for fuels [4,10]. Thus, it could be applied for synthesis of γ-valerolactone from C6 sugars and levulinic acid [11], 2,5-dimethylfuran from 5-formyloxymethylfurfural [12], furfuryl alcohol [13] and methylfuran [14] from furfural, upgraded bio-oil from bio-oil [15], and diesel/gasoline mixtures from tar [16] (Figure 1).
Supported catalysts with nanoparticles are traditional catalysts for the hydrogen production from formic acid in gas and liquid phase. Novel single atom metal catalysts supported on N-doped carbon may provide a higher activity in the formic acid decomposition than the activity of the catalysts with nanoparticles, but the difference is not so significant [17]. The activity of homogeneous metal complexes is often higher [18][19][20][21][22][23][24]. Hence, they could be used at lower temperatures. Metal complexes also represent more uniform active sites as compared to metal nanoparticles. Hence, basing on this knowledge the design of the catalyst could be facilitated. However, there are serious problems of application of homogeneous metal complexes as catalysts for different reactions limiting their industrial applications. They include difficulties in separation of a catalyst from the reaction medium and catalyst's recovery, instability of homogeneous catalytic systems, as well as possible corrosive effects of catalyst solutions on the equipment [25].
Separation of the catalysts with noble metals could be important for production of hydrogen from reaction mixtures containing formic acid and obtained from biomass. To solve this problem, metal complexes could be supported on different supports. Serious efforts have been directed toward the immobilization of homogeneous catalysts on supports. Evidently, their catalytic properties could change due to a change of ligand environment, since after supporting the support surface sites become important ligands for metal atoms. These sites may have no direct analogs in solutions [25]. Their nature affects strongly the energy of interaction of metal complexes and resistance of the catalyst to leaching. Additionally, the support may provide surface sites for formic acid activation leading to its faster conversion.
Carbon dioxide is also produced as a by-product during the decomposition of formic acid; however, it can be further hydrogenated into formate salts at low temperatures [26,27]. Earlier, we have analyzed this reaction taking place on different catalysts, particularly on supported metal complexes [27]. In the present review, we will consider in details the catalytic properties of supported Ru, Ir and Fe complexes in the hydrogen production from formic acid. There are only a few studies performed with supported complexes of other metals (Pd, Rh) in this reaction. We have not found a specialized review related to application of supported metal complexes in the hydrogen production from formic acid. However, this subject is worth to discuss since this type of the catalysts shows excellent activity, selectivity and stability in the reaction and can be easily separated from the reaction mixture.

Supported Ruthenium Complexes
Ru complexes are among the most active complexes for the hydrogen production from liquid phase formic acid. The group of Laurenczy contributed significantly to the development of these catalysts [20]. In 2009, they reported the results of immobilization of ruthenium(II)-TPPTS (trisulfonated triphenylphosphine) complex on different supports [28]. Among them, they used an ion exchange resin containing basic trimethylammonium groups. The reaction mixture except of formic acid contained sodium formate (9:1). It is known that addition of sodium formate to formic acid should give a higher activity [19,29,30]. The results showed that this resin ionically interacted strongly with the Ru complex and no Ru leaching took place during the reaction. However, recycling of the catalyst led to a decrease in the reaction rate, but the same conversion was achieved in 3 h.
In another case, covalent interaction of the Ru species to the phosphine groups of PPh 3 -or PPh 3 -O-cross-linked polystyrene led to strong coordination of the metal. Unfortunately, the obtained catalyst was not sufficiently active in the reaction as compared to the homogeneous catalyst. The authors supposed that the reasons are related to a high hydrophobicity of the material and mass transfer limitations. Additionally, they used five different types of zeolites as supports. The activity was sufficiently good, but the adsorbed Ru-TPPTS could be removed in water from zeolites thus indicating that it was attached weakly through physical adsorption. This complicates the catalyst recycling which is necessary for a sustainable process.
The same group has noticed later that their earlier attempts to create heterogenized metal complex catalysts were only partially successful and developed another system [29]. In this case, they used a mesoporous silica support (MCM-41) with attached phosphine groups. The optimized catalytic system corresponded to MCM41-Si-(CH 2 ) n PPh 2 /Ru-mTPPTS with n = 2 and demonstrated the activity and stability comparable to those of the homogeneous catalyst ( Figure 2). Thus, the turnover frequency (TOF) of 2780 h −1 was obtained at 383 K (Table 1). TOF value corresponds to the number of hydrogen molecules obtained per one metal site per time unit. It is a major value characterizing the specific activity of the catalysts. Figure 2. Effect of the number of CH 2 groups attaching phosphine groups to the MCM-41 support on formic acid (HCOOH) decomposition catalyzed by the immobilized Ru catalysts at 363 K. Reprinted with permission from [29].
The effect of the number of CH 2 groups attaching the phosphine groups (n) is demonstrated in Figure 2. It is seen that short CH 2 chains lead to the activity higher than those for the catalysts with longer chains. The content of CO obtained as a by-product was negligible (3 ppm). It is very important to have a very low level of CO in the reaction to prevent poisoning of the catalyst in a fuel cell. In addition, the supported Ru catalysts were recyclable since they allowed performing the reaction for more than 20 cycles without loss in activity.
Later, the same group created a reactor system for continuous production of hydrogen from formic acid [31]. A Ru-mTPPTS catalyst supported on phosphinated polystyrene beads was used in this case. This catalyst provided the TOF of 270 h −1 at 378 K and the apparent activation energy of 93.6 kJ mol −1 . The low CO concentration level (<5 ppm) was reached due to a PROX reaction using a Pt/CeO 2 catalyst. Alternatively, a methanation of CO could be used to decrease the CO content [9].
Zhao [32] modified the surface of SiO 2 support (450 m 2 g −1 ) with 3-mercaptopropyltrimethoxysilane to obtain SiO 2 -SH groups, which then interacted with Ru or Pd chlorides (about 2 wt %). The obtained Ru-S-SiO 2 and Pd-S-SiO 2 catalysts showed TOFs of 344 h −1 and 719 h −1 with a 4 M HCOOH/HCOONa (9:1) mixture at 358 K, respectively. X-Ray Photoelectron Spectroscopy (XPS) studies before and after experiments showed the presence of mainly Pd 2+ ions in the catalyst indicating that they are the active sites of the reaction. The authors also showed that sulfates accelerate the reaction by up to 70%. This could be useful for practical applications.
Wang et al. [33] used a ruthenium pincer complex knitted in a porous organic polymer (810 m 2 g −1 ). Thermal gravimetric analysis revealed that the supported complex was thermally stable up to 533 K. However, the TOF of 266 h −1 obtained at 363 K was not very high. The authors proposed a mechanism for the formic acid decomposition and production based on participation of Ru hydride in the reaction (Figure 3). For the decomposition, the mechanism involves the dissociative formic acid adsorption and CO 2 release followed by H 2 release. It is interesting that N sites of the complex provide deprotonation of the formic acid through dissociation of the O-H bond. Solakidou et al. [34] showed that amino functionalized silica (H 2 N@SiO 2 ) significantly increases the TOFs of the hydrogen production from formic acid by a (Ru/P(CH 2 CH 2 PPh 2 ) 3 ) homogeneous catalyst. The maximal TOF reached 983 h −1 at 363 K. They observed a significant decrease of the apparent activation energy from 41 kJ mol −1 to 28 kJ mol −1 and supposed that the H 2 N@SiO 2 particles play a dual role: they act as a co-catalyst for deprotonating formic acid by amine groups, and they serve as a template, which stabilizes the metal complex on its surface, thus promoting formate decomposition via (Ru II -hydride) species.
Hausoul et al. [36] studied the effect of the nature of polymeric support on the properties of a Ru complex in the hydrogen production from formic acid. Polymeric analogs of PPh 3 (pTPP), 1,2-bis(diphenylphosphino)ethane (pDPPE) (Figure 4), and 1,2-(diphenylphosphino)benzene (pDPPBe) have been tested. The highest TOF of 22,900 h −1 was obtained with a RuCl 2 (p-cymene)/pDPPE catalyst at 433 K ( Table 1). The catalyst performed efficiently in solutions with up to 30 wt % formic acid. It is seen in Figure 5a that the activity of the unsupported (RuCl 2 (p-cymene)(PPh 3 )) complex is significantly lower than those of the supported complexes. The kinetics of the reaction was featured by an induction period and a pseudo-zero-order dynamics of the pressure increase. Recycling experiments revealed only low leaching and a small decrease in the activity over 7 runs. A Ru/C catalyst with nanoparticles gave a significantly lower activity, lower selectivity and leaching of Ru to the reaction solution.  The same authors studied the decomposition of formic acid in solutions with other substances (Figure 5b). It is seen that levulinic acid (LA) and sulfuric acid retard the reaction, but they do not poison the catalyst completely. This is important to know for development of future biorefineries involving the process of conversion of biomass to hydrogen through the formic acid production [7,8] (Figure 1).
Beloqui Redondo et al. [37] used a 0.4 wt % Ru phosphine complex supported on a metal organic framework (MOF) for the gas phase decomposition of formic acid at 418 K. They obtained 99% selectivity and TOF of 2300 h −1 (Table 1). This TOF is sufficiently high for the gas phase decomposition. The authors indicated that phosphine species interact with Ru species providing the formation of Ru single-sites on the MOF support. Amine linkers present in the support could activate formic acid for the reaction through deprotonation. The authors observed an induction period, which was assigned to removal of chloride ligands from the metal complex followed by coordination of formates. However, the Brunauer-Emmett-Teller (BET) surface area of their catalyst decreased significantly after the reaction. It is not clear whether this will take place further and affect negatively the catalytic reaction.

Supported Iridium Complexes
As we showed above, several Ru complexes, which are very active in the formic acid decomposition, involve phosphine ligands. Broicher et al. [38] indicated that P-based ligands are sensitive to oxidation, while N-based ligands show a great advantage allowing handling and storage of the catalyst in air. In this section, we will consider Ir complexes, which are mainly attached to N-containing ligands of the support.
Bavykina et al. [35] have used an [IrCp*(OH)](OTf) 2 complex (OTf-triflate) to deposit over a covalent triazine framework (CTF) prepared at 773 K with a high surface area (Figure 6a,b). OTf was washed out during the recycling of the catalyst, pointing that formate replaces triflate. The TOFs of the catalysts with a low concentration of the metal complex (0.2 wt %) were higher than those of the catalysts with a high concentration and corresponded to 27,000 h −1 (Table 1). The authors tested the catalyst's durability in continuous mode. Thus, a highly concentrated formic acid solution (88 wt %) was fed to a reactor at 353 K. At termination of this experiment, a turnover number (TON = the number of H 2 molecules related to the number of metal sites) of 1,060,000 was obtained, which demonstrates that the catalyst is highly durable and can be used in devices producing hydrogen. Figure 6c displays possible molecular pathways for the system [35], which consist of the three main steps: (1) formic acid deprotonation, (2) β-hydride elimination and (3) hydrogen release. The deprotonation is important and takes place on free pyridinic sites that provide basicity of the CTF support. The hydrogen release step has been proposed to be rate-determining.
Gunasekar et al. [39] studied an [IrCp*Cl 2 ] 2 complex supported on CTFs prepared at different temperatures 673 and 773 K. The activity of the supported complex was higher than that of the homogeneous complex. The TOFs were, however, lower than those obtained by Bavykina et al. [35] and corresponded to 7930 h −1 at 363 K. This could be related to a much higher concentration of metal in the samples. The TOF for the similar supported RhCp* complex was lower than that of the IrCp* complex.
Shen et al. [30] studied an IrCp*Cl 2 complex supported on porous polypyrrole particles (500 nm) with a BET surface area of 51 m 2 g −1 ( Table 1). The TOF was very high in the presence of sodium formate and equal to 46,000 h −1 at 363 K. The apparent activation energy in the formic acid decomposition in the absence of sodium formate corresponded to 63 kJ mol −1 and in the presence of sodium formate it was approximately the same (66 kJ mol −1 ). When 1.0 M formic acid solution flowed into a tubular reactor (55 × 9 mm) with a catalyst containing 250 mg of the IrCp*Cl 2 (polypyrolle) complex at 313 K, the initial hydrogen evolving rate was 5.6 mL min −1 , which could generate about 1.1 W electric power through a proton-exchange membrane fuel cell. The authors noted that this value was sufficient to drive, for example, a personal mobile phone.
Recently, Broicher et al. [38] also used an [IrCp*Cl 2 ] 2 complex as a precatalyst and a conjugated microporous polymer (CMP) with bipyridine groups as a support. The combination of those gave an Ir@CMP catalyst (Figure 7). This catalyst showed a record value of TOF of 123,894 h −1 at 433 K (Table 1), relatively high apparent activation barrier of 90 kJ mol −1 and low leaching. The CO content was stable in the range around 68 ppm. Complete conversion of formic acid could be reached. In the same conditions a commercial Ir/C catalyst with nanoparticles demonstrated a strong leaching and gave a low conversion of formic acid confirming that Ir nanoparticles are not active in the reaction. Application of the [IrCp*Cl 2 ] 2 complex without bipyridine ligands as a catalyst gave also only a low conversion. At the same time, the [IrCp*Cl 2 /2,2bipy] complex showed a high activity of 43,051 h −1 at 433 K demonstrating the need for bipyridine ligands. The activity of the supported 1 wt % Ir@CMP catalyst ( Figure 8) was close (35,246 h −1 ) indicating that the heterogenization affected only weakly. After the reaction, a significant amount of nanoparticles (up to 5 nm) was found in the sample, demonstrating that the measured activity is really caused by a low number of active sites with a very high activity. This point is also supported by a study of the effect of variation of metal loading showing that the TOFs are higher for the catalysts with a low content of metal complex ( Figure 8) in accordance with the data of Bavykina et al. [35].

Supported Iron Complexes
The use of efficient non-noble metal catalysts would be a good choice for the hydrogen production from formic acid decomposition. Boddien et al. [40] proposed different homogeneous Fe containing complexes for this reaction. Using a mixture of 0.005 mol% Fe(BF 4 ) 2 ·6H 2 O and tris[(2-diphenylphosphino)ethyl]phosphine [P(CH 2 CH 2 PPh 2 ) 3 ] to a solution of formic acid in propylene carbonate, without other additives or bases, obtained a TOF value up to 9425 h −1 and a TON value of more than 92,000 at 353 K [40]. The apparent activation energy corresponded to 77 kJ mol −1 and the CO content did not exceed 20 ppm. However, the catalyst was completely poisoned and became inactive after 16 h of the continuous reaction. This was assigned to chloride and/or water admixtures accumulation on the catalyst.
Later, Stathi et al. [41] successfully heterogenized Fe phosphine complexes on the surface of two types of silica modified with phosphines ( Figure 9a). Heterogenization of the Fe(II)/P(CH 2 CH 2 PPh 2 ) 3 system increased its TOF by 1.7 times as compared to the homogeneous catalyst and was in the range of 6000-8000 h −1 ( Table 1). The reaction was performed in propylene carbonate solvent. The apparent activation energies were significantly lower than that of the homogeneous complex and corresponded to 51 kJ mol −1 and 43 kJ mol −1 for Fe II /RPPh 2 @SiO 2 and Fe II /polyRPhphos@SiO 2 , respectively. The authors indicated that the possible rate determining steps could be hydride elimination or direct hydride transfer from formate to Fe. No leaching of iron in the reaction solution was found. The Fe II /RPPh 2 @SiO 2 catalyst showed a TON of higher than 176,000. The authors observed an inhibitory water effect (Figure 9b), but it was reversible, since the catalyst could be reactivated by a simple washing. This result is important to know, since formic acid always contains a small amount of water, which can be formed together with carbon monoxide due to self-decomposition of formic acid taking place during its storage [42]. This effect is more significant at high concentrations of formic acid. Ea-apparent activation energy, mTPPTS-meta-trisulfonated triphenylphosphine, PPh-phenylphosphine, CTF-covalent triazine framework, MOF-metal organic framework, Cp*pentamethylcyclopentadienyl, CMP-bipyridine-based conjugated microporous polymer.

Discussion
The summarized data for the key catalysts with supported Ru, Ir and Fe complexes for the hydrogen production from formic acid are shown in Table 1. Other metal complexes are almost not studied. The table can help to choose the optimal catalysts corresponding to certain conditions of the reaction. However, it is not easy to compare the activity of the catalysts (TOFs) presented in Table 1, since the conditions of the reaction and concentrations of the active component in the catalysts were different. Moreover, some experiments have been performed in the presence of sodium formate. Basic additives to the reaction mixture or basic sites of the catalysts/supports are known to promote significantly the reaction. They deprotonate formic acid to formate species. Deprotonation of formic acid can be provided also by traditional oxide supports having basic sites [43] and by introduction of alkali metals promoters to supported metal catalysts [44][45][46]. Deprotonation provided by pyridinic N sites of N-doped carbon support was also reported for the catalysts with single metal atoms [47,48]. In this case, it reminds the effect of basic amine additives.
As for catalysts with nanoparticles, the steady-state TOF values obtained for the gasphase reaction over a Pd/C catalyst doped with K ions with 3-4 nm Pd nanoparticles did not exceed 3600 h −1 at 353 K [44,45]. For the liquid-phase reaction and Pd nanoparticles (~1.4 nm) supported on N-doped carbon, the initial values were higher and reached 8414 h −1 at 333 K in the presence of sodium formate [22]. Some supported catalysts with nanoparticles (Ir/C [38] and Ru/C [36]) were used for comparison of the activity with supported metal complexes (Table 1). It was shown that their activity is negligible as compared with the activity of supported metal complexes. The disadvantage of these comparisons was that the mean sizes of nanoparticles in the catalysts with nanoparticles were not reported.
In contrast, some homogeneous complexes showed much higher TOFs than those of the supported metal complexes ( Table 1). The values in the range 250,000-322,000 h −1 for temperatures 363 and 373 K have been reported by a few groups of authors [21,23,24]. These complexes are also based on Ir [23,24] and Ru [21]. It would be useful to immobilize them on some supports in order to have an opportunity to separate easily the obtained catalyst from the reaction mixture.
As it is shown above (Figure 8), concentration of a metal complex is an important factor determining TOFs. Interesting that at a lower concentration of a metal complex higher TOFs were observed. In this case, the active sites could be stabilized by specific support sites. The nature of these active sites should be studied using advanced methods like extended X-ray absorption fine structure (EXAFS) combined with X-ray absorption near edge structure (XANES) preferably in situ. Using density functional theory (DFT) calculations may assist in understanding the structure of these active sites. There is an evident lack of such studies. The progress in understanding may lead to development of a targeted synthesis of the catalysts with these very active sites.

Conclusions
Therefore, the analysis of the literature showed that immobilization of Ru, Ir and Fe complexes on some polymers, covalent triazine frameworks, metal organic frameworks or silica modified with phosphines is promising for the hydrogen production from formic acid in terms of activity of the catalysts and possibility to separate the catalysts from the reaction mixture. Supported Ir complexes were normally more efficient than the supported Ru complexes. Thus, the maximal TOF value was reached for the IrCp* complex supported on the bipyridine-based conjugated microporous polymer and corresponded to 123,894 h −1 at 433 K. In part, the high activity of Ir complexes could be provided by the presence of basic N sites of the supports which deprotonate formic acid for further easier decomposition of the formed formate species with participation of the Ir atoms. Yet, the activities of some homogeneous metal complexes were higher and reached 322,000 h −1 .
A strong concentration effect of metal complexes was observed demonstrating that at a small concentration of supported metal complex higher TOFs are obtained than those at a high concentration. This effect is not related to nanoparticles formation. Finally, non-noble metal supported Fe complexes were efficient in the reaction provided the contents of water and chlorine ions in the solution were negligible.

Introduction
Energy demand has constantly increased in the last decades, which is closely linked to the expanding population and increasing prosperity. Currently, about 80% of the world's energy supply comes from fossil fuels (i.e., coal, oil, and natural gas). However, their utilization is inevitably associated with the emission of hazardous gases, which drives the current global warming crisis. Currently, nearly 100% of the total CO 2 emissions originate from the combustion and processing of fossil fuels [1], and its concentration in the atmosphere, which has experienced a great increase since the start of the Industrial Revolution, is nowadays higher than 400 ppm [2].
Hence, the tremendous concerns about the environmental issues related to the use of fossil fuels, together with their finite nature, is fostering the search toward the deployment of renewable energy sources. Such an ambitious goal has been the central focus of many investigations. In this sense, the use of hydrogen has tremendous hope for the use of renewable energies in different industrial applications and the transport sector. This is reflected in the constantly increasing number of publications in which its supremacy as an outstanding energy vector is highlighted [3][4][5][6][7].
The use of hydrogen, produced from renewable energy sources, has significant benefits. Its use has zero emission of greenhouse gases and it produces only water as a by-product. Additionally, it has a high energy storage capacity on a gravimetric basis (120 MJ/kg), which is much greater than those of gasoline (44.4 MJ/kg), diesel (45.4 MJ/kg), biodiesel oil (42.2 MJ/kg), and natural gas (53.6 MJ/kg) [8]. However, against all the benefits of hydrogen, its low volumetric energy density (i.e., 0.01 MJ/L at standard temperature and pressure conditions (STP)), which is much lower than those of common fuels (gasoline (34.2 MJ/L), diesel (34.6 MJ/L), biodiesel oil (33 MJ/L), and natural gas (0.0364 MJ/L)) limits its utilization as a fuel at ambient conditions. There are different options to increase the hydrogen energy density so that it can meet the target set by the U.S. Department of Energy (DOE), which fixes the ultimate onboard hydrogen storage for light-duty fuel cell vehicles at 0.065 kg H 2 /kg system and 0.050 kg/L, in gravimetric and volumetric basis, respectively [9]. The most used options encompass physical methods such as hydrogen liquefaction, compression, and adsorption in porous materials [10,11]. However, the harsh conditions (such as very high pressure or extremely low temperature), and the high cost associated with the infrastructure needed for the safe handling and storage of hydrogen are important drawbacks of such physical storage methods.
In contrast, chemical storage methods stand up as a promising option, which is particularly important for onboard automotive applications, in which the technologies used in the physical methods do not fully meet the DOE targets for safe and inexpensive hydrogen storage [12,13].
Chemical hydrogen storage refers to those processes in which molecular hydrogen is released through a chemical reaction that starts when the hydrogen-containing molecule (i.e., hydrogen carrier) is subjected to thermal or catalytic decomposition [14].
There are various hydrogen carrier molecules in both the liquid and solid phase. Notable examples of those molecules, together with their hydrogen content (in wt. %) are included in Table 1.  [31][32][33] Seeking and exploring new hydrogen storage options are in continuous progress and significant advances related to reversible hydrogen storage have been recently achieved [34][35][36][37]. In this review, we cover some of the most relevant recent strategies on hydrogen production from NH 3 BH 3 (ammonia borane, AB), which is the simplest nitrogen boron hydrogen compound [38][39][40], and one of the most fruitfully investigated solid-state hydrogen-rich molecules.
AB is a white crystalline solid at room temperature, which was first prepared by Shore and Parry in 1955 [41], and it has received tremendous attention due to several advantages compared to other hydrogen carrier molecules such as its high hydrogen content (19.6 wt. %; each equivalent yielding up to 3 equivalents of hydrogen), low molecular weight (30.87 g mol −1 ), its stability in solid-state, and high solubility in water. Furthermore, it is nonexplosive and non-flammable under standard conditions. AB has a high melting point of 112 • C and a density of 0.74 g cm −3 [38].
AB has an equal number of protic H δ+ (N-H) and hydridic H δ− (B-H) hydrogens in intra-and intermolecular interactions. It has heteropolar (N-H···H-B) and homopolar (B-H···H-B) dihydrogen interactions, which are the origin of the intra-and intermolecular dehydrogenation of AB [42]. It also has a strong B-N bond, so that the release of hydrogen is more favored than the dissociation into NH 3 and BH 3 under most conditions [43]. AB can be synthesized in three different ways (i.e., Lewis acid-Lewis base exchange, saltmetathesis followed by hydrogen release, and isomerization of the diammonate of diborane ([H 2 B(NH 3 ) 2 ] + [BH 4 ] − )) [38].
AB solvolysis can afford 3 equivalents of molecular hydrogen at moderate temperatures upon utilization of a proper catalyst, so this is the preferred option. The dehydrogenation of amine-borane adducts catalyzed by transition-metals dates back to the late 1980s [45], but its application in the production of hydrogen is still drawing great interest in the research community, as can be seen in the increasing number of publications reported per year (see Figure 1). Both homogeneous and heterogeneous catalytic systems have been explored, but the advantages of heterogeneous catalysts make these systems preferred from a practical point of view. There is vast literature reported on the hydrolytic dehydrogenation of AB, highlighting those contributions made by Yamashita et al. [46][47][48][49][50][51][52][53][54], Özkar et al. [55][56][57][58], and Xu et al. [44,[59][60][61][62].
Among the heterogeneous catalysts, systems based on ruthenium nanoparticles (NPs) have shown outstanding performances. Ru-based catalysts usually achieve complete AB dehydrogenation, producing~3 equivalents of hydrogen in short reaction times, and with thermo-controllable reaction rates. The hydrolysis rate is also frequently related to the amount of catalyst and AB, but the hydrolysis rate is frequently found as zero-order relation or quasi-zero-order relation with the concentration of AB [63].
It is worth mentioning that despite the intense efforts devoted to unveiling the mechanism of the hydrolytic dehydrogenation of ammonia borane, there are still some aspects that remain unclear (e.g., rate-determining steps, the order of bond cleavages, etc.) [64]. For instance, Xu et al. postulated that the interaction between AB molecules and the surface of the metal active phase gives rise to the formation of activated complex species, which are attacked by a molecule of H 2 O, leading to the concerted dissociation of the B-N bond and the hydrolysis of the BH 3 intermediate to form BO 2 , releasing H 2 [65]. Fu et al. proposed a mechanism that proceeds via an almost self-powered process that involves the formation of BH 3 OH − and NH 4 + , followed by the attack of adjacent H 2 O to generate H 2 [66]. Na et al. suggested that the mechanism is very similar to that of the hydrolytic dehydrogenation of sodium borohydride, and proceeds via dissociative adsorption of ammonia borane on Ru surface [67]. More recently, Liu et al. claimed that the hydrogen production from AB attained by noble metal catalysts occurs via the following steps: (1) AB molecules interact with the surface of the catalyst to form a complex; (2) A molecule of H 2 O attacks to AB-catalyst complex; and (3) AB and H 2 O each lose a hydrogen atom to form H 2 . Such a mechanism is illustrated in Figure 2 [63]. It is evident that the catalyst's nature plays a crucial role in controlling the whole reaction. Most of the studies reported on Ru-based catalysts for the hydrolytic dehydrogenation of AB are focused on elucidating the role of the features of the metal active phase (i.e., size, morphology, incorporation of a second and third metal in the nanoparticles, etc.), while less attention has been paid to the properties of the support. We divided this manuscript into several sections, which contain a review of representative catalytic systems based on monometallic Ru NPs and supports of a diverse nature, which have been used for the hydrolytic dehydrogenation of AB. As a summary, Table 2 includes representative examples of Ru-based catalysts supported on carbon materials, oxides, metal organic frameworks (MOF), and some other less explored supports, together with the turnover frequency values achieved (TOF; in mol H2 ·mol Ru −1 ·min −1 ) and the calculated activation energy (Ea; in kJ mol −1 ).

Hydrolytic Dehydrogenation of Ammonia Borane (AB) over Carbon Material-Supported Catalysts
Several catalytic supports have been explored for the synthesis of highly efficient catalysts for the hydrolytic dehydrogenation of AB. Carbon materials are one of the most intensively studied and the resulting catalysts have given very interesting results.
Akbayrak and Özkar explored the performance of catalysts supported on multiwalled carbon nanotubes (Ru(0)@MWCNT) by evaluating the activity of in-situ formed Ru nanoparticles (NPs) [72]. The resulting NPs had an average particle size range of 1.4-3.0 nm and were well-dispersed on the support. The effect of the Ru content was evaluated by checking the activity (in mL of H 2 /min) of catalysts with metal contents of 0.73, 1.47, 1.91, 2.26, and 2.83 wt. %. Among those investigated, the sample with 1.91 wt. % displayed the best activity, with a TOF = 329 min −1 (mol H2 ·mol Ru −1 ·min −1 ). That catalyst was evaluated during four consecutive reaction cycles and it preserved 41% of its initial activity. Doe et al. also checked the performance of MWCNT-supported Ru catalysts with Ru(NH 3 ) 6 Cl 3 as the metal precursor and using electrostatic adsorption (EA) and incipient wetness impregnation (IWI) methods [98]. Additionally, catalysts based on activated carbon and SiO 2 were prepared as reference materials. The Ru NPs were located on the external surface after both EA and IWI, and a smaller average size was achieved for the EA (2 and 3 nm for Ru/MWCNT-EA and Ru/MWCNT-IWI, respectively). Checking the performance of three sets of catalysts supported on MWCNTs, activated carbon, and SiO 2 with various average NPs size, it was observed that in all cases the catalysts with larger NPs attained higher reaction rates (expressed as turnover rates in mol H2 ·mol surface Ru −1 ·s −1 ), and higher TOF values were achieved for the Ru/MWCNT catalysts. Among those evaluated in that study, Ru/MWCNTs-EA produced the highest initial TOF value and the lowest activation energy, which was attributed to the hydrogen spillover taking place on metal NPs supported on CNTs.
Cheng et al. reported a simple method for the preparation of a Ru/graphene catalyst synthesized from graphene oxide and RuCl 3 using a one-step co-reducing approach with methylamine borane (MeAB) [68]. The resulting catalyst was compared to those synthesized by using different reducing agents, namely AB and NaBH 4 . It was observed that the sample prepared with MeAB provided better results than those using AB or NaBH 4 , which was attributed to a better control over nucleation and growth processes. It was determined that the average NPs size was 1.2, 1.7, and 2.0 nm for the catalysts reduced with MeAB, AB, and NaBH 4 , respectively, suggesting that the size of the NPs increased as the reducing agent became stronger. The catalyst reduced with MeAB had a TOF of 100 mol H2 ·mol Ru −1 ·min −1 , and activation energy of 11.7 kJ mol −1 , and it retained 72% of its initial activity after four reaction cycles. The same research group followed a very similar one-step co-reducing approach using ascorbic acid to synthesize Ru/graphene catalysts, which resulted in much higher TOF values of 600 mol H2 ·mol Ru −1 ·min −1 and activation energy of 12.7 kJ mol −1 [78], which pointed out the importance of the reducing agent in controlling the final catalytic performance. The high TOF value achieved by Ru/graphene was attributed to the narrow size distribution of the Ru NPs and the utilization of graphene as a suitable support.
In the study of Ma and co-workers, [75] ligand-free Ru NPs supported on carbon black were prepared in-situ from the reduction of the metal precursor (i.e., RuCl 3 ) by AB concomitantly with its hydrolysis. The resulting catalyst had an average nanoparticle size of 1.7 nm and showed a TOF of 429.5 mol H2 ·mol Ru −1 ·min −1 . In that case, the reusability of the catalyst was checked during five consecutive reaction runs, after which it preserved 43.1% of its initial activity. The results of the characterization of the spent catalyst indicated that there were neither metal leaching nor aggregation of the NPs, so that the activity loss was attributed to an increasing concentration of the reaction products (i.e., metaborate and Cl − ions) and their adsorption on the surface of the NPs. Furthermore, it was postulated that the increase in the viscosity in the solution after several reaction cycles could impede the diffusion of the reactant molecules and their collision with the Ru active sites.
Sun et al. studied the performance of poly(N-vinyl-2-pyrrolidone) (PVP)-stabilized Ru NPs loaded onto bamboo leaf-derived porous carbon (Ru/BC) [80]. In that case, the NPs were synthesized by in-situ reduction with AB using RuCl 3 ·nH 2 O as a metal precursor and with various PVP content (i.e., 0, 1, 3, 5, or 10 mg), which was used to avoid the agglomeration of the NPs. The catalysts with the best activity among those investigated (i.e., Ru/BC stabilized with 1 mg of PVP) displayed a TOF of 718 mol H2 ·mol Ru −1 ·min −1 , and retained nearly 56% of the initial catalytic activity after 10 consecutive reaction cycles. That stability was much better than that of the PVP-free catalyst, which indicated the important role of PVP in enhancing the recyclability of the catalysts by preventing the agglomeration of the NPs. The same research group also investigated the performance of nitrogen-doped (N-doped) porous carbon materials [81]. The support was prepared by hydrothermal treatment of hydrochloride semicarbazide and glucose, and it was subsequently loaded with the metal precursor. The resulting Ru/NPC catalyst completed the dehydrogenation reaction in 90 s at room temperature, reaching a TOF of 813 mol H2 ·mol Ru −1 ·min −1 . The stability of the catalyst was checked by performing five consecutive runs, after which 67.3% of the initial activity was preserved, and the activity decay was attributed to the agglomeration of the NPs.
Yamashita et al. addressed the preparation of highly efficient Ru/carbon catalysts prepared by pyrolysis of a supported Ru complex (i.e., tri(2,2-bipyridyl) ruthenium (II) chloride hexahydrate) [79]. In that study, the Ru catalysts were prepared by impregnating a commercial activated carbon with the metal precursor and subsequent decomposition of the metal complex by carrying out a heat treatment at temperatures ranging from 600 to 1000 • C (catalysts denoted as Ru/C(600)-(1000)). Transmission Electron Microscopy (TEM) analysis confirmed the formation of Ru NPs even for the lowest temperatures used for the decomposition of the metal complex. The average NP size strongly depended on the decomposition temperature and ranged from 3.8 to 13.5 nm for the Ru/C catalysts. Two additional reference samples were also synthesized: Ru/C(imp) prepared by the same protocol using Ru(NO)(NO 3 ) 3 as the metal precursor, and the Ru/SiO 2 catalyst prepared from Ru(bpy) 3 2+ and commercial fumed silica, with NP size of 3.3 and 2.0 nm, respectively. The results of the catalytic activity indicated that complete conversion of AB was attained with Ru/C(800), Ru/C(900) and Ru/C(1000), whereas it was not achieved with Ru/C(600), Ru/C(700), Ru/C(imp), and Ru/SiO 2 . That observation led the authors to conclude that under the experimental conditions used, relatively large NPs were preferred for the reaction. Among those investigated, Ru/C(800) showed the most promising activity with an average TOF number of 670 mol H2 ·mol Ru −1 ·min −1 . That catalyst achieved 100% of conversion even after four consecutive reaction runs, but the reaction rate was progressively sluggish. A summary of the results of the catalytic activity of the materials assessed in that work is plotted in Figure 3. It was also concluded that the electronic properties of Ru NPs played an important role in controlling the catalytic performance. It was claimed that Ru/C(800) had an optimum proportion of oxidized Ru species, which are important in the reaction. Gao et al. [70] studied the performance of Ru/nanodiamonds in the hydrolytic dehydrogenation of AB. That study was motivated by a wide variety of oxygen functional groups present in the nanodiamonds, which were expected to serve as anchoring points for the Ru NPs. Commercial nanodiamonds, with an average diameter of 5-10 nm were used as the support and RuCl 3 as the metal precursor, for the synthesis of catalysts with a metal content of 3.22, 4.82, 6.21, and 8.05 wt. %. The TOF values showed a volcano-type tendency, the best catalyst being that with a Ru content of 6.21 wt. % (229 mol H2 ·mol Ru −1 ·min −1 ). The recyclability tests performed with that sample indicated that total conversion of AB was achieved during four consecutive runs, but the TOF values decreased during the cycles, which was related to the increase of the NPs from 3.7 to 5.1 nm and the increasing viscosity and concentration of metaborate in the reaction solution.
Most of the catalysts used for the hydrolytic dehydrogenation of AB displayed good reusability and attained total conversion during several consecutive reaction runs. However, the reaction is frequently sluggish and longer reaction times are needed to produce 3 equivalents of H 2 per mole of AB, so the poor stability of the catalysts under reaction conditions is one of the most addressed issues.
The stability of the catalysts was shown to improve upon utilization of supports with abundant surface functional groups, which are known to increase the stability and reusability of the catalysts [8]. That was the case of Fan et al., who developed catalysts consisting of ultrafine and highly dispersed Ru NPs supported on N-doped carbon nanosheets formed by a hierarchically porous carbon material (HPCM) [76]. Ru/HPCM had an average NP size of 1.41 nm and narrow size distribution, which ranged from 0.6 nm to 2.0 nm. Such ultrafine NPs were confined into the micropores and mesopores of the support, affording numerous active sites and stabilizing the NPs from sintering under reaction conditions. X-ray Photoelectron Spectroscopy (XPS) analysis confirmed the presence of pyridinic, pyrrolic, and graphitic nitrogen. The importance of both pyridinic and graphitic nitrogen in enhancing catalytic performance was pointed out in that study. Ru/HPCM was evaluated during eight consecutive reaction cycles, after which it retained 50% of its initial activity. After that, the average NP size of the spent catalyst slightly increased to 1.47 nm and the partial activity decay was related to the catalyst loss and the passivation effect of metaborate ions formed along the recycling tests. The same group also synthesized Rubased catalysts supported on N-doped bagasse-derived carbon materials (BC-hs) [73]. That biomass residue, which has abundant negative oxygen and nitrogen functional groups, was suitable for the interaction with Ru 3+ cations of the metal precursor. Catalysts with various metal contents (i.e., 2.5, 3.5, 4.5, and 5.5 wt. %) were synthesized by in-situ reduction with AB, achieving homogeneously dispersed ultrafine Ru NPs. It was observed that the best-performing catalyst preserved 80% of its TOF value after five runs, demonstrating the suitability of the BC-hs support to stabilize the metal NPs. The partial loss of the activity was attributed to changes in the NP size and loss of the catalyst during the separation and washing steps.
Ma et al. used a support based on a N-doped porous carbon material (NC-Fe) using a facile pyrolysis of a porous organic polymer (POP) synthesized from ferrocene carboxaldehyde and melamine as the starting materials [69]. The resulting Ru catalyst (Ru/NC-Fe) achieved the total conversion of AB during five cycles, but the reaction was considerably sluggish. That catalyst contained γ-Fe 2 O 3 , so it was easily recovered with a magnet. The activity decay was attributed to both the metaborate formed in the reaction and the agglomeration of the metal NPs.
Liu et al. used a N-enriched hierarchically macroporous-mesoporous carbon support that was synthesized by a co-template evaporation-induced self-assembly approach with SiO 2 nanospheres as a macroporous hard template and F127 as a mesoporous soft template and a soft nitriding by the low-temperature thermolysis of urea [99]. The resulting support, denoted as hPCN, was loaded with Ru NPs (Ru@hPCN), and reference catalysts based on pure macroporous and pure mesoporous support were also prepared (Ru@macroPCN and Ru@mesoPCN, respectively). Ru@hPCN exhibited the best activity and total dehydrogenation was finished in 120 s, while longer reaction times were required for the reference samples and for a commercial Ru/C catalyst (160, 280, and 880 s, for Ru@macroPCN, Ru@mesoPCN, and Ru/C, respectively). Such better performance of Ru@hPCN was also evidenced by the higher TOF values achieved at 60 • C (1850, 1258, 902, and 308 min −1 , for Ru@hPCN, Ru@macroPCN, Ru@mesoPCN, and Ru/C, respectively). The reusability of Ru@hPCN was evaluated during four consecutive reaction cycles and it was observed that the NP size increased from 0.7 nm to 2.7 nm. Additionally, the amount of surface N species decreased from 13.4 at. % to 2.3 at. %, which suggested that the N-enriched species contributed to improving the catalytic performance by dissociating the electropositive H δ+ from water molecules and the breakage of the B-N bonds.
Not only N-doped carbon materials, but also other N-containing carbon-based supports were shown to be effective in stabilizing Ru NPs for the hydrolytic dehydrogenation of AB. For instance, Zheng et al. used hierarchical porous graphitic carbon nitride (g-C 3 N 4 ) nanosheets to anchor ultrafine Ru NPs [100]. In that case, the supports were prepared from melamine and various amounts of NH 4 Cl as a dynamic gas template. The NPs were encapsulated into the network of the g-C 3 N 4 by the reduction of the metal precursor with NaBH 4 to achieve a final metal content of 1.91 wt. %. The characterization of the catalyst indicated that the resulting NPs had a size that ranged from 1.9 to 5.1 nm and they exhibited uniform dispersion onto the support. Among the synthesized supports, that prepared with a melamine to NH 4 Cl mass ratio of 1:3 exhibited the highest surface area (S BET of 59 m 2 g −1 ) and its counterpart Ru catalyst was selected to assess the performance in the dehydrogenation of AB. That catalyst displayed satisfactory recyclability even after four consecutive reaction cycles. The activity loss observed from the fifth cycle was attributed to a particle aggregation of the NPs and to the accumulation of NH 4 BO 2 species in the reaction solution, which increases its viscosity and blocks the active sites of the catalyst.
Tang et al. also explored the suitability of Ru/g-C 3 N 4 catalysts for the dehydrogenation of AB [77]. In that case, the catalysts were prepared from urea and RuCl 3 . Catalysts with various metal loadings (i.e., 4.10, 3.28, 2.46, and 1.64 wt. %) were synthesized. The time needed for the reaction to be completed decreased to 3.5 min for the catalyst with 3.28 wt. %, which showed the highest TOF value (459.3 mol H2 ·mol Ru −1 ·min −1 ). Concerning the recyclability of Ru/g-C 3 N 4 , it preserved 50% of the initial catalytic activity after the fourth run, and the activity loss was related to the increase of the size of the NPs from 2.8 nm to 4.1 nm and the adsorption of B species on the surface of the NPs.
Yamashita et al. also studied the performance of g-C 3 N 4 supported Ru catalysts [101]. In that case, carbon/g-C 3 N 4 composites with various carbon contents (C(x)/g-C 3 N 4 ; "x" is the initial carbon weight per cent of 0.1, 0.5, 1.0, 2.0, and 4.0 wt. %, respectively) were synthesized from glucose and dicyandiamide by a simple experimental procedure. The obtained supports were subsequently impregnated with RuCl 3 ·3H 2 O to obtain Ru NPs after reduction with H 2 gas at 300 • C. The incorporation of carbon was reported to extend the absorption of the materials to the visible region of 480-800 nm compared to the pristine g-C 3 N 4 , so the resulting Ru/C(x)/g-C 3 N 4 were interesting photocatalysts for the dehydrogenation of AB under visible-light irradiation. Additionally, the incorporation of carbon served to achieve smaller Ru NPs than those achieved in the raw g-C 3 N 4 , which also affected the catalytic activity. Figure 4 contains information on the catalytic activity displayed by Ru/C(x)/g-C 3 N 4 . It was observed that the addition of moderated carbon contents in the catalysts enhanced the activity compared to that displayed by Ru/g-C 3 N 4 , which was related to the smaller NPs shown in those samples ( Figure 4a). It was also seen that the reaction rate of all the materials improved under visible light irradiation (Figure 4b), achieving the fastest reaction rate with Ru/C(1.0)/g-C 3 N 4 . Interesting results were also achieved by Fan et al., who synthesized phosphorusdoped carbon-supported Ru catalysts (Ru/PPC) [74]. In that study, the supports were prepared from hypercrosslinked polymer networks of triphenylphosphine and benzene, and they were subsequently used for the synthesis of Ru catalysts with various metal contents (i.e., 1.5, 2.5, 3.5, and 4.5 wt. %). Small and well-distributed Ru NPs were attained in all the materials, the smallest for Ru/PPC with 3.5 wt. % of Ru (average size of 1.13 nm). The catalytic activity was shown to be dependent on the metal content, achieving TOF values of 174, 325, 413, and 290 mol H2 ·mol Ru −1 ·min −1 for catalysts with 1.5, 2.5, 3.5, and 4.5 wt. % of Ru, respectively. The stability of the most active material was evaluated during four consecutive cycles, observing that its activity gradually decreased after the first run. Such loss of activity was attributed to the NP sintering (from 1.13 to 2.47 nm) as well as the catalyst loss in the separation and washing steps.

Hydrolytic Dehydrogenation of AB over Oxide-Supported Catalysts
As in most catalytic reactions, carbon materials are the most fruitfully studied supports. However, interesting results have also been achieved with oxide-supported catalysts, with silica (SiO 2 ) and alumina (Al 2 O 3 ) the most investigated. Representative examples of such systems are summarized in this section.
SiO 2 has been shown to serve as a suitable support for the dehydrogenation of AB. SiO 2 , with various structures and morphologies, have been nicely utilized for the development of well-performing catalysts. For instance, Zhu et al. reported on Ru NPs confined in SBA-15 (Ru@SBA-15) by using a double solvent approach (with hexane and water) [85]. Metal loadings of 0.5, 1.1, 2.1, 3.2, and 4.0 wt. % were used, which resulted in NPs with an average size of 2.0 ± 0.6 nm, 2.2 ± 0.6 nm, 3.0 ± 0.8 nm, and 3.7 ± 0.7 nm, respectively. Among investigated, Ru@SBA-15 with a metal content of 2.1 wt. % displayed the best activity, which was also superior to those catalysts with non-confined NPs supported on SBA-15 and SiO 2 . It also showed good durability after five consecutive reaction runs at room temperature.
Chen et al. designed Ru catalysts supported on cubic 3D cage-type mesoporous silica SBA-1 functionalized with carboxylic acid (Ru/S1B-C10) [84]. The carboxylic acid was incorporated by co-condensation of tetraethyl orthosilicate (TEOS) and carboxyethylsilanetriol sodium salt in the presence of poly(acrylic acid) and hexadecylpyridinium chloride. The -COOH groups were uniformly distributed within the mesopores of the support, which assisted the preparation of well-dispersed and small Ru NPs. Catalysts with metal loadings from 0.5 to 2.0 wt. % were synthesized with both -COOH free support (SIB-C0) and -COOH containing support (S1B-C10). It was observed that while the average NP size observed in Ru/S1B-C0 catalysts increased from 2.8 to 4.6 nm when the metal loading increased from 0.5 to 2.0 wt. %, it only increased from 2.0 to 3.2 nm in Ru/SIB-C10 for similar metal contents. Catalysts with 1 wt. % of Ru displayed better performance for both SIB-C0 and SIB-C10 supports than materials with other Ru loadings. Among those investigated, Ru(1)@S1B-C10 had the highest TOF value (202 mol H2 ·mol Ru −1 ·min −1 ), which was attributed to the nanosized Ru particles and their good dispersion as well as the effect of the pore confinement of the support. The reusability of the best-performing catalysts was evaluated during five reaction cycles and even though the total conversion was achieved in all cases, the reaction rates decreased with an increase in the number of cycling tests, which was related to a partial metal leaching and restricted access of reactants to the Ru active sites originated by the adsorption of metaborate on the surface of the NPs.
Yao el at. prepared core-shell Ru@SiO 2 catalysts with various metal contents (i.e., 1.0, 2.0, 3.0, 4.0, 6.0, 8.0, and 10.0 wt. %), which consisted of Ru NPs of~2 nm embedded in the center of spherical SiO 2 particles of~25 nm [83]. Among those investigated, Ru@SiO 2 with 6 wt. % of Ru loading exhibited the best performance for the production of H 2 from AB, with a TOF of 200 mol H2 ·mol Ru −1 ·min −1 . A much slower reaction rate was observed for a supported Ru/SiO 2 catalyst used as a reference sample, which was related to the easy aggregation occurring in Ru/SiO 2 . The Ru@SiO 2 core-shell catalyst displayed good recycling stability for five cycles, but the reaction was progressively sluggish during the cycles.
Özkar et al. explored the activity of catalysts formed by Ru NPs loaded on magnetic silica-coated cobalt ferrite (Ru(0)/SiO 2 -CoFe 2 O 4 ) [82]. That catalyst showed a moderated activity, with a TOF of 172 mol H2 ·mol Ru −1 ·min −1 at room temperature, but excellent recyclability during 10 reaction runs, preserving 94% of the initial activity. After each reaction cycle, the spent catalyst was isolated using a permanent magnet, and no metal leaching was detected.
[86] also developed SiO 2 -based magnetic core-shell catalysts. In that study, Ru NPs were loaded on amino functionalized silica-covered magnetic NPs (Fe 3 O 4 @SiO 2 -NH 2 -Ru), and the resulting catalysts were evaluated in the dehydrogenation of AB. It was claimed that the amine group served to increase the electron transfer from the coreshell Fe 3 O 4 @SiO 2 structure to the surface of the Ru NPs, which resulted in enhanced catalytic activity in the production of H 2 . The developed catalyst had a TOF value of 617 mol H2 ·mol cat −1 ·min −1 and showed good stability during eight reaction cycles. Al 2 O 3 has also been applied for the synthesis of Ru-based catalysts for the dehydrogenation of AB. For instance, Metin el al. reported an easy method for the synthesis of nearly monodisperse Ru NPs, which consisted in the thermal decomposition and simultaneous reduction of the metal precursor (ruthenium(III) acetylacetonate (Ru(acac) 3 ) in the presence of oleylamine (serving as a stabilizer and reducing agent) and benzylether (used as solvent) [87]. The as-synthesized NPs (with an average size of 2.5 nm) were subsequently loaded on γ-Al 2 O 3 to build a Ru@Al 2 O 3 catalyst with 1 wt. % of Ru. That material had a TOF of 39.6 mol H2 ·mol Ru −1 ·min −1 . It was observed that the TOF value increased to 83.3 mol mol H2 ·mol Ru −1 after carrying out a treatment of the catalyst with acetic acid, which removed the organics from the surface of the NPs. The treated catalysts exhibited great stability even after 10 consecutive reaction runs, and the TOF only decreased 10% of the initial value.
Chen el at. developed a microporous crystalline γ-Al 2 O 3 with a large surface area (large surface, 349 m 2 g −1 ) prepared from a microporous covalent triazine framework (CTF-1, surface area of 697 m 2 g −1 ) as a template [88]. Ru-based catalysts with 1, 2, and 5 wt. % were prepared from RuCl 3 and they were evaluated in the dehydrogenation of AB. The high surface area and the hierarchical pore structures of the micropores developed in the synthesized γ-Al 2 O 3 was claimed to be responsible for the better performance achieved by the materials studied in that work compared to those reported elsewhere for Ru/γ-Al 2 O 3 catalysts (TOF as high as 256.8 mol H2 ·mol Ru . Catalysts with metal contents from 2.52 to 4.91 wt. % were prepared in-situ by reducing the metal precursor with AB. It was observed that the time needed to achieve a total conversion of AB decreased with increasing the metal content. The stability test indicated that AB conversion was completed even during five consecutive runs, but the reaction rate decreased along the cycles, which was attributed to an increase of the average NPs size from 2.9 to 3.1 nm. Lee et al. investigated the effect of the crystal phase of Ru on the dehydrogenation of AB by using face-centered cubic (fcc) structures and hexagonal close-packed (hcp) structured Ru NPs loaded on γ-Al 2 O 3 [102]. Catalysts with 1 wt. % of Ru and different sizes of both fcc and hcp Ru NPs were prepared. The size and the crystal phase were controlled by adjusting the experimental conditions, in terms of amount and type of metal precursor, solvent, and amount of PVP. It was observed that under the experimental conditions used in that study, fcc NPs were achieved using Ru(acac) 3 while RuCl 3 originated hcp NPs. XPS analysis suggested that fcc Ru NPs were more easily oxidized than the hcp counterpart since a larger relative proportion of Ru 4+ was detected (i.e., [Ru 4+ ]/[Ru 0 ] of 28.2 and 19.2% for fcc and hcp Ru/γ-Al 2 O 3 , respectively). The catalytic activity of fcc and hcp NPs with different sizes (i.e., 2.4, 3.5, 3.9, and 5.4 nm) was evaluated by monitoring the H 2 generation profiles at 25 • C (see Figure 5). As can be seen, hcp NPs displayed better performance regardless of the size of the NPs. It was also seen that the difference between hcp and fcc NPs became smaller as the NPs increased. Additionally, the opposite tendency was seen for fcc and hcp NPs: the catalytic activity of fcc Ru/γ-Al 2 O 3 enhanced with increasing NP size, while hcp displayed worse performance with increasing sizes. Density functional theory (DFT) calculations were done to determine the adsorption energy of O 2 molecules on the (001) crystal plane of fcc and hcp Ru to get information about their easy oxidation. The results obtained suggest that the fcc Ru was easily oxidized than hcp Ru, which was consistent with the experimental results observed in that study. Thus, the authors of that study ascribed the worse performance of smaller NPs with fcc crystal phases to their higher degree of oxidation, while the tendency observed for hcp was attributed to size effects.

Other Supported Ru Catalysts
Carbon materials and oxides are the most investigated supports for the hydrolytic dehydrogenation of AB, but there are other interesting materials that have also attracted great interest in the last years. Among them, the utilization of MOF should be highlighted because of the attention drawn not only for this application, but for many other catalytic reactions [106][107][108]. They have been shown to be suitable to embed metal NPs as well as to stabilize them by incorporating additional functionalization. This was the case of Chen et al., who achieved well-dispersed Ru NPs immobilized within the pores of amine-functionalized MIL-53 by using an in-situ impregnation-reduction method [94]. It was claimed that the amino groups were located at the acid linkers in MIL-53(Al)-NH 2 , and served as Lewis bases, thus stabilizing the Ru precursor (i.e., RuCl 3 ) during the impregnation step. An amino-free catalyst was prepared for comparison purposes. The average NP size was determined to be 1.22 nm, smaller than the mean diameter of the pores of the amino functionalized MIL-53(Al) (2.07 nm) so that they were embedded in the framework of the MOF, while larger NPs were located on the external surface. Ru/MIL-53(Al)-NH 2 catalyst displayed better activity than the amino-free counterpart, and also better durability and reusability. Such better performance exhibited by Ru/MIL-53(Al)-NH 2 was attributed to the amino groups, which assisted the formation and stabilization . The MOF selected in that study was claimed to be a 3D framework with three different kind of cages as well as with thermal and chemical stability in water. Ru catalysts loaded on other supports (i.e., carbon black, SiO 2 , γ-Al 2 O 3 , and GO) were prepared for comparison purposes. Ru/MIL-96 was the most active catalysts checked in that study, with a TOF of 231 mol H2 ·mol Ru −1 ·min −1 . However, that material did not show suitable stability, since it only retained 65% of its initial activity after the fifth run, which was attributed to the increasing NP size and viscosity of the solution.
Zhu et al. explored the suitability of a N-containing microporous organic framework (POF) as a scaffold to anchor Ru NPs [95]. The resulting catalyst, which was denoted as Ru/PAF-72, has an average NP size of 1-2 nm and showed a TOF of 294 mol H2 ·mol Ru −1 ·min −1 . The most remarkable aspect of that system was its great stability even after 10 consecutive reaction cycles. For comparison purposes, the stability of a catalyst based on Ru NPs loaded onto carbon black was also evaluated, showing a significant activity decay after only four cycles.
Additionally, some other supports, which are less frequently used in catalysts, have served to develop Ru catalysts for the hydrolytic dehydrogenation of AB. For instance, Cai et al. developed a sophisticated catalyst based on Ru NPs loaded a natural polyphenolic polymer (bayberry tannin, BT) immobilized on collagen fibers (CF). The catalysts, denoted as CF-BT-Ru, were synthesized by immobilization of BT, crosslinking of glutaraldehyde, and subsequent chelation of Ru 3+ (final metal loadings of 0.58, 1.03, 1.57, and 2.12 wt. %) [71]. The procedure used is shown in Figure 6. Various reference samples were also prepared for comparison purposes (i.e., CF-BT-Ru, CF-Ru, and Ru-carbon material (Ac, GO, CNTs and g-C 3 N 4 )). It was observed that the presence of BT enhanced the dispersion of the metal active phase, due to the interaction with the phenolic groups, so that the BT containing catalyst has much smaller Ru NPs than the BT-free counterpart (i.e., 2.6 ± 0.6 and 6.5 ± 0.5 nm, respectively). XPS characterization indicated that the catalyst contained positively charged nitrogen and neutral amine groups. Such neutral amine groups were responsible for the stabilization of the Ru NPs by providing some of their electrons. The six catalysts evaluated showed total conversion of AB, but different reaction rates, with CF-BT-Ru being the most active with a TOF of 322 mol H2 ·mol Ru According to that observation, it was postulated that the Mg/Al ratio, which is related to the relative acidity of the material and the support-Ru interaction, was a key aspect in controlling the final catalytic performance. It was found that Ru/Mg 2 Al-LDH-h had more Brønsted acid sites and it showed a weaker interaction with Ru species, which enabled the existence of Ru in the metallic state, thus explaining the better activity exhibited by that catalyst. That sample preserved its stability during the first four cycles, but it slightly decayed from the fourth to the tenth consecutive run, which was attributed to the increased viscosity of the solution, which impedes the diffusion of AB molecules, thus hindering their interaction with the Ru active sites. Sun et al. also explored the performance of MgAl-LDH-supported Ru catalysts for the hydrolytic dehydrogenation of AB [97]. In this case, supports with a composition of Mg 0.5 Al 1 -LDHs, Mg 1 Al 1 -LDHs, Mg 2 Al 1 -LDHs, and Mg 3 Al 1 -LDHs were prepared and loaded with Ru NPs. Among those samples, Ru/Mg 1 Al 1 -LDHs showed better activity, completing the hydrogen release in 130 s, compared to the other catalysts that needed longer reaction times (180, 140, and 230 s, for Ru/Mg 0.5 Al 1 -LDHs, Ru/Mg 2 Al 1 -LDHs, and Ru/Mg 3 Al 1 -LDHs). In this case, such tendency was also attributed not only to the higher content of Brønsted acid sites in Ru/Mg 1 Al 1 -LDHs, but also to the higher purity of Mg 1 Al 1 -LDHs, which did not contain boehmite (AlO(OH)) and hydromagnesite (Mg 5 (-CO 3 ) 4 (OH) 2 ·4H 2 O). Ru/Mg 1 Al 1 -LDHs exhibited good stability during 10 reaction runs, and 58.1% of the initial activity remained after those cycles. DTF calculations were conducted to get information on the promotion effect of MgAl-LDHs catalysts. It was determined that Ru/MgAl-LDHs catalysts have the beneficial electronic properties to accelerate the H 2 O dissociation to form O-H bonds, activating the H 2 O molecules during the hydrolytic dehydrogenation of AB.

Conclusions and Perspectives
There is great hope for the potential of hydrogen as an energy vector, which motivates the search for alternatives that overcome the limitations that are frequently related to its storage. Chemical hydrogen storage stands up as a promising option and there are several hydrogen carrier molecules that afford satisfactory hydrogen capacity. Among them, ammonia borane has drawn much attention, and research to exploit the potential of ammonia borane as a hydrogen storage material has been intensified in the last years.
This review summarized some of the recent Ru-based heterogeneous catalysts applied in the hydrolytic dehydrogenation of ammonia borane. A perusal of the most frequently used catalysts is included, paying particular attention to those heterogeneous catalysts with carbon materials and oxides as supports. Among the vast diversity of supported Ru catalysts studied thus far, carbon material-based catalysts frequently attain the best performance.
Most of the investigations deal with the optimization of the properties of the active phase such as the nanoparticle size and morphology, while the effect of the composition of the support is less explored. Most of the catalysts experienced deactivation during few consecutive reaction cycles, which is linked to several factors: (i) aggregation of the nanoparticles; (ii) metal leaching; and (iii) accumulation of metaborate in the solution and change in the viscosity. It is expected that such phenomena involved in the deactivation of the catalysts could be partially averted by using encapsulated or nanoconfined metal catalysts upon selection of a suitable porous material serving as a host for the metal active phase. Most studies include information about the NP size and metal content of the spent catalysts, but no information has been reported about the concentration of metaborate species and viscosity of the solution before and after the reaction, which would help to verify such possible reasons for deactivation that are frequently mentioned but never confirmed. Some other interesting works evidenced the formation of different B-containing products such as B(OH) 3 , BO 2 − , B(OH) 4 − , and polyborates [64,65,[109][110][111]. Some studies have aimed at enhancing the cyclability of the catalysts by stabilizing the metallic phases using strong metal-support interaction, using, for example, a support with abundant functional groups. Nitrogen-containing supports have been widely studied, but the resulting catalysts are still lacking in stability during the cycles. Hence, the insufficient stability of the assessed catalysts is frequently the weakest point indicated in the literature for the heterogeneous catalysts used in the dehydrogenation of ammonia borane. The regeneration of the catalysts after reaction would therefore be a very interesting issue to be considered in future works.
No attention is paid to the actual cornerstone for the successful application of this hydrogen carrier molecule in the hydrogen storage scenario, which is the regeneration of ammonia borane from the products obtained in its decomposition reaction. The AB regeneration problem, which was pointed out long back [112], remains unsolved nowadays and studies dealing with the regeneration of AB are still sparse [113]. The nature of the by-products of the hydrolysis of AB, which are mainly borate species, complicates the regeneration of AB, since the stable B-O bond formed in the by-product are strong and they are not easily reconverted to the B-H bond present in AB molecules, so a strong reducing agent would be needed. Thus, multi-step reactions are required for the regeneration of AB. For instance, Liu et al. proposed a regeneration process that implied the conversion of boric acid to trimethyl borate (B(OCH 3 ) 3 by esterification with methanol. Then, B(OCH 3 ) 3 can react with NaH, generating NaBH 4 and, finally, AB was formed upon reaction of NaH with ammonia sulfate in THF [114]. Gagare et al. aimed at regenerating AB from NH 4 B(OMe) 4 using LiAlH 4 as a reducing agent and NH 4 Cl as an ammonia source [115]. Vasiliu et al. reported on the regeneration of AB from a more simplified and energy-efficient process that involved minimum reaction steps [116]. However, their starting point was not the regeneration of the by-products originating from the production of 3 equivalents of H 2 per molecule of AB, but polyborazylene was used, which is formed upon generation of two equivalents of H 2 per molecule of AB (partially spent AB). In that case, polyborazylene was converted to AB nearly quantitatively by 24 h treatment with N 2 H 4 in liquid NH 3 at 40 • C. More recently, Sharma et al. [117] studied the regeneration of AB using a digestion-based approach in the presence of methanol and subsequent regeneration with the reducing agent (i.e., LiAlH 4 ).
There is still plenty of room for improvement in both the design and development of stable and reusable catalysts as well as in the processes involved in the real application of ammonia borane as a hydrogen storage material, especially for those on-board systems. One of the main issues to be tackled is the high cost of ammonia borane compared to other hydrogen storage systems, so finding cost-effective ways for the synthesis and regeneration of ammonia borane is highly desirable.
The present review contains only information on monometallic Ru-based catalysts, which have been shown to be the most effective to attain suitable catalytic behavior, but there are other compositions that have also displayed interesting results.

Introduction
Rational use of fossil fuel resources is a necessary condition for the long-term stable development of society. In this context, the development of efficient processes for energy production from renewable sources is a topical problem that can be solved using various fundamental studies, particularly in the field of material science. Hydrogen production from formic acid (FA) is a way to involve renewable biomass sources into the energy cycle [1,2]. The FA decomposition reaction proceeds on mono-, bi-and trimetallic catalysts deposited on oxide or carbon supports [3,4]. Activity and selectivity of the catalysts towards hydrogen production are determined by the nature of their active component, its dispersion, and electronic state of metals. The application of nitrogen-doped carbon nanomaterials (N-CNMs) as the catalyst supports makes it possible to perform the tailored synthesis of metal supported catalysts with desired properties [5].
It is known that nitrogen in N-CNMs having different structures occupies standard positions: pyridinic (N Py ), pyrrolic (N Pyr ), graphitic (N Q ), and oxidized (N Ox ). According to data in the literature, supported metals can stabilize on different nitrogen centers. For example, Pt stabilization on N Q centers of nitrogen-doped carbon nanotubes (N-CNTs) leads to the formation of electron-rich Pt nanoparticles, which are active towards the hydrogenation of nitroarenes [6] and aerobic oxidation of glycerol [7]. Ombaka et al. [8] reported the formation of Pd 2+ species on the N Pyr centers of N-CNTs, which are active towards the hydrogenation of nitrobenzophenone [8]. However, a substantial body of data indicates the involvement of N Py centers in the stabilization of electron-deficient metals on the surface of various N-CNMs, which show the increased activity and selectivity in hydrogenation reactions or electrochemical reduction of oxygen [9][10][11][12]. The positive role of the electron-deficient metals deposited on ordered or disordered N-CNMs was also demonstrated for FA decomposition in a liquid or gaseous medium [4,[13][14][15][16]. Additionally, for platinum deposited on nitrogen-doped carbon nanofibers (N-CNFs) and N-CNTs, it was shown that the catalytic properties of metal in this reaction are determined by its locally one-type interaction with N Py centers [16]. Among various metals that are active towards FA decomposition, platinum and palladium are the most active catalysts for this reaction [3,17]. It should be noted that the use of N-CNMs as the supports opens new possibilities for enhancing the efficiency of catalysts via the stabilization of highly active and selective atomic metal species [18,19]. Among the wide variety of N-CNMs, bamboo-like N-CNTs are characterized by a unique combination of the graphene layer curvature and the availability of all nitrogen centers. At the same time, data on the use of these materials as a support for the catalysts of the FA decomposition reaction are quite scarce [18]. This paper is devoted to investigating the formation of the Pd catalyst on bamboo-like N-CNTs for the gas phase formic acid decomposition reaction. A detailed investigation of the properties of N-CNTs, variation of palladium concentration, and application of various research methods allowed us to describe the interaction of palladium with the surface of N-CNTs. Different active palladium species were shown to form as isolated ions and metallic nanoparticles with a size of 1.3 nm, the ratio of which were determined by the affinity of palladium for N Py centers and the palladium concentration in catalysts.

Materials and Methods
Nitrogen-doped and nitrogen-free carbon nanotubes (N-CNTs and CNTs) were synthesized by the standard CVD route [20]. An ethylene-ammonia mixture or ethylene were decomposed on the Fe-containing catalyst at 650-700 • C. The growth catalyst was removed by the boiling of N-CNTs and CNTs in 2 M hydrochloric acid. Nevertheless, X-ray photoelectron spectroscopy (XPS) revealed the presence of iron in the washed samples with the content up to approximately 0.5 at.%. Previously, we have demonstrated the low intrinsic activity of washed N-CNTs and CNTs in FA decomposition reactions compared to metal supported catalysts [16]. Palladium was deposited on carbons, which were pre-dried in Ar at 170 • C, by incipient wetness impregnation from a Pd acetate-acetone solution. The samples were dried in air at 105 • C for 8 h and then reduced in an H 2 /Ar flow at 200 • C for 1 h.
Nitrogen adsorption measurements were carried out at 77 K by the use of ASAP-2400. Prior to measurements, the samples were evacuated at 200 • C for 24 h. Acetone capacity of the N-CNTs and CNTs was measured by the standard incipient wetness impregnation method. Acetone was added dropwise under stirring to the sample dried in Ar at 170 • C until the pores of the material were completely filled. The volume of the absorbed acetone (cm 3 /g) was determined by weighing. CO chemisorption on the catalysts was measured at room temperature by a pulse technique. Before the experiments, the catalysts were reduced in hydrogen flow at 200 • C. X-ray diffraction patterns were recorded using an HZG-4 (Zeiss, Germany) diffractometer with monochromatic CuK α radiation. Electron microscopy investigation was performed using JEM-2200 FS (JEOL Ltd., Japan) and Titan 60-300 (FEI, Netherlands) electron microscopes. Raman spectroscopy with an excitation wavelength of 632.8 nm was performed using a HORIBA LabRAM HR800 (HORIBA Jobin Yvon Raman Division, France) Raman spectrometer. XPS study of N-CNTs was performed on an ES-300 (KRATOS Analytical, UK) photoelectron spectrometer with an AlKα source (hν = 1486.6 eV). The distribution of nitrogen in N-CNTs was studied by synchrotronbased photoelectron spectroscopy at 500 and 800 eV photon energy [21]. Pd catalysts were examined using a VG ESCALAB HP (ThermoScientific, UK) instrument with an AlKα source (hν = 1486.6 eV). In situ reduction of the catalysts under 1000 Pa H 2 in a preparation chamber of the XPS unit was performed at 300 • C for 1 h [22]. After the pretreatment, the sample was transferred to an analytical chamber without contact with air.
Activity of the catalysts in the decomposition of FA (5 vol. % НCООН/Не) was measured using a flow setup. A 20 mg catalyst sample was uniformly mixed with 0.5 cm 3 of quartz sand (grain size of 0.25-0.5 mm). The reaction mixture preheated to 100 • C in a special box was fed into the reactor at a rate of 20 cm 3 /min. The experiments were carried out in the temperature-programmed mode at a ramp rate of 2 • /min with chromatographic analysis of the gas mixture. Before measuring the activity, the catalyst was reduced in a 10% H 2 /He mixture at 200 • C for 1 h. The apparent turnover frequency (TOF) was calculated as a ratio of the reaction rate measured at low (<20%) conversions to the total number of palladium atoms in the catalysts.

Properties of N-CNTs
Nitrogen-doped carbon nanotubes were synthesized by the CVD method, which is widely applied to obtain N-CNTs [23][24][25][26]. As seen in Table 1, by increasing the ammonia concentration in the C 2 H 4 -NH 3 reaction mixture (25, 40, 60 and 75% NH 3 ), one can monotonically increase the degree of nitrogen doping of nanotubes from 1.8 to 6.6 at.% with a simultaneous decrease in the yield of N-CNTs from 38 to 5 g C/g of catalyst. Such an effect of N/C ratio on the nitrogen content in N-CNTs was described by other authors for different reaction mixtures [24,26,27]. In turn, a decrease in the yield of N-CNTs is related to deactivation of the catalyst due to its encapsulation (Figure 1), which has been reported by many authors for the CVD synthesis of N-CNTs and other carbon materials [25,26,[28][29][30]. Table 1. Properties of nitrogen-doped carbon nanotubes (N-CNTs): N/C and N Py /N Q ratios according to XPS data, I D /I G and I 2D /I D ratios according to Raman spectroscopy, S BET values and product yield.

Sample-NH 3 Concentration (%) N/C at.% N Py /N Q I D /I G I 2D /I D S BET , (m 2 /g) Yield, (g C/g of Catalyst)
CNTs  TEM micrographs of the 4.4 at.% N-CNTs as an example and CNTs are displayed in Figure 1. Ethylene decomposition is accompanied by the predominant formation of multiwalled carbon nanotubes. In their turn, N-CNTs have a distinct bamboo-like structure, which is known to form due to the curvature of nitrogen-containing graphene layers [31,32].
According to STEM-HAADF images with elemental mapping, nitrogen is uniformly distributed in N-CNTs.
XPS revealed (Figure 2) that nitrogen in N-CNTs is in pyridinic (N Py ,~398 eV), pyrrolic (N Pyr ,~400 eV), graphitic (N Q ,~401 eV), oxidized (N Ox ,~403 eV) and molecular (N N2 , 405 eV) states [24,26,33]. The presence of molecular nitrogen encapsulated between graphene layers or inside channels is characteristic of bamboo-like N-CNTs [26,34]. An increase in the total nitrogen content in N-CNTs is accompanied by an increase in the content of all structural species. However, this changes the predominant species and the ratio of nitrogen species. In the N-CNTs with the minimum doping degree, the main contribution is made by graphitic nitrogen, and the N Py /N Q ratio is equal to 0.5, Table 1.
As the total nitrogen content is increased, the contribution of pyridinic nitrogen grows; as a result, the N Py /N Q value achieves 1.2 in the case of 6.6 at.% N-CNTs. It is known that nitrogen incorporation in the pyridinic positions of graphene layers is accompanied by the formation of carbon vacancies, i.e., by an increase in structural defectivity. Indeed, the formation of such fragments was observed in TEM images of N-CNTs obtained at atomic resolution [23]. The formation of such structural elements was confirmed by a growth of the I D /I G ratio with a simultaneous decrease in the I 2D /I D ratio revealed by Raman spectroscopy ( Table 1). The I D /I G ratio reflects defectivity of carbon nanomaterials [35,36], while the I 2D /I D ratio is used to analyze the size of the defect-free graphene blocks [37,38] (Figure 3). In our earlier work, the XRD study of N-CNTs and their structural simulation gave grounds to suppose the formation of the fragments similar to the structural elements of the g-C 3 N 4 layer comprising the ordered carbon vacancies and pyridinic nitrogen atoms [20]; the estimated size of such a fragment of the layer is not smaller than 5 Å. However, it should be noted that the obtained XPS data do not provide information on the surface state of N-CNTs with respect to different nitrogen species, which is an important factor when these materials are employed as the catalyst supports. In this context, N-CNTs were additionally studied by synchrotron-based XPS with variation of the analysis depth using radiation energies of 500 and 800 eV. The data obtained were compared with the results of standard XPS study performed at 1486.6 eV. The radiation energies of 500, 800 and 1486.6 eV allowed us to acquire data at the analysis depths of ca. 1.6, ca. 3 and ca. 6 nm, respectively (Figure 4). According to TEM data displayed in Figure 1b, the mean wall thickness of N-CNTs is ca. 5 nm. Hence, when the depth of analysis is varied from 1.6 to 6 nm, the acquired data could provide information on the nitrogen in external/internal graphene layers, and partially in internal arches.  One can see that in both samples, nitrogen is incorporated in all structural positions of external/internal graphene layers and internal arches. The total nitrogen concentration profile along the analysis depth coincides to a greater extent with the profile of different nitrogen species, which is determined most likely by the catalytic growth mechanism of N-CNTs on the Fe catalyst with the equiprobable incorporation of nitrogen in different positions [24]. As seen in Figure 4c, when the entire tube wall (ca. 6 nm) or only its external/subsurface layers (ca. 1.5 nm) are analyzed, N Py /N Q , ratios for 4.4 at.% N-CNTs virtually do not differ. In the case of 1.8 at.% N-CNTs, as the depth of analysis is decreased, the N Py /N Q ratio increases by a factor of ca. 1.6. Nevertheless, our study of N-CNTs with the minimum depth of analysis allows a conclusion that external layers of N-CNTs, irrespective of the doping degree, are slightly enriched with graphitic nitrogen.
To choose a method of palladium deposition on N-CNTs, their textural and adsorption properties were investigated. As seen in Table 1, nitrogen doping of CNTs does not produce changes in the textural characteristics of carbon nanotubes. Indeed, their specific surface area changes in a narrow range of 148-160 m 2 /g, and pore size distribution curves for all the samples are similar and have distinct maxima in the regions of 3-4 nm and 25-40 nm ( Figure 5). Fenelonov et al. [39] proposed that pores with the size of 3-4 nm are formed due to the curvature of graphene layers, whereas larger pores emerge due to the entangled weave of N-CNTs leading to the formation of multiple voids. However, an increase in the nitrogen content in N-CNTs is accompanied by a considerable growth of their capacity towards acetone, from 3 to~6-7 cm 3 /g, which was used as a solvent in the deposition of palladium catalyst (Figure 5b). This suggests that such a high acetone capacity of N-CNTs is achieved due to the bulk filling of inner channels of the tubes in spite of multiple regular internal arches (Figure 1b). Note that the values of acetone capacity of fresh and treated in 2 M HCl carbons are the same. In turn, a good correlation between nitrogen content in N-CNTs and N Py /N Q , I D /I G and I 2D / D ratios was found (Table 1). Therefore, it can be supposed that the presence of large defects (not smaller than 5 Å) in graphene layers of the synthesized N-CNTs, which are represented by the ordered carbon vacancies surrounded with pyridinic nitrogen atoms [20], improves their wettability and makes the capillary effect more pronounced. Kumar et al. [40] calculated isotherms of water adsorption on disordered N-C materials and attributed the anomalously high adsorption to the implementation of a specific mechanism, according to which 1D, 2D and 3D nucleation, accompanied by the formation of solvent clusters which resembles the growth of crystals, was followed by bulk condensation. The filling of the inner channels of N-CNTs may indicate that during their deposition on N-CNTs by wet chemistry methods, metals can be anchored not only on the external surface but also in the inner channels of nanotubes [41]. On the other hand, Hao et al. [42] conclude that high hydrophilicity of N-doped carbon nanomaterials ensures a high dispersion of supported metallic catalysts, thus increasing their activity.

Properties of Pd/N-CNTs Catalysts
A series of Pd/N-CNTs catalysts (0.2, 0.5, 1 and 2 wt.% of Pd) were synthesized using the N-CNTs that were obtained by decomposition of the 40% C 2 H 4 -60% NH 3 mixture. The choice of support was based on the optimal combination of the nitrogen content in carbon nanotubes and the yield of N-CNTs (Table 1). To study the interaction of palladium with the N-CNTs surface, the concentration of palladium was varied from 0.2 to 2 wt.%. N-free catalysts 0.2% Pd/CNTs and 2% Pd/CNTs were synthesized for comparison purposes. The revealed high affinity of N-CNTs for acetone, which was used in our study as a solvent of palladium salt, explains the uniform distribution of palladium in the catalysts and the absence of agglomerates ( Figure 6). In addition, as seen in Figure 6b, it cannot be ruled out that individual palladium particles are located in the inner channels of nanotubes. In the X-ray diffraction patterns, a weak intensity peak related to Pd 111 was recorded only for the catalysts with a maximum metal content (2 wt.%; see Figure 7). The X-ray diffraction patterns of all other catalysts show only the maxima related to the carbon phase.
50 nm 10 nm  Table 2 and Figure 8. In the case of N-free catalysts, a ten-fold increase in palladium content was accompanied by a decrease in apparent TOF from 180 to 72 h −1 , with a simultaneous decrease in CO/Pd ratio from 33.5 to 22.1%, according to the CO chemisorption data. Indeed, the mean particle size for 0.2% Pd/CNTs was 1.2 nm, whereas for 2% Pd/CNTs it was 2.3 nm. According to the XPS, palladium was in the metallic state and the binding energy of Pd 3d was 335.6 eV (Figure 9). This value is typical of the dispersed metallic palladium particles [43,44]. The TOF values of the 0.2% Pd/CNTs and 2% Pd/CNTs catalysts calculated as a ratio of the reaction rate to the number of surface palladium atoms were 193 and 147 h −1 , respectively. The observed difference in TOF values may be due to different palladium crystallinity or its different localization in the tubes (outer surface and inner channels). Selectivity of the catalysts towards hydrogen formation did not exceed 92%.    All the nitrogen-doped catalysts, irrespective of palladium content, demonstrated the increased activity and selectivity (97-98%) in comparison with 0.2-2% Pd/CNTs catalysts. TOF values varied in a narrow range of 225-324 h −1 ; the maximum activity was observed for the 0.2% Pd/N-CNT catalyst. The dependence of CO/Pd on palladium content for this series of catalysts had a complicated shape. A decrease in palladium content from 2 to 1% was not accompanied by changes in CO/Pd, and a further decrease in Pd concentration to 0.2% produced a sharp drop in CO/Pd from 32 to 16.1%. The observed dependence of CО/Pd on palladium content did not correlate with the ТЕМdata, according to which mean particle size for the catalysts with 2% Pd and 1% Pd was equal to 1.4 and 1.3 nm, respectively (Table 2), whereas for 0.5% Pd and 0.2% Pd it could not be determined owing to their small sizes. An HAADF-STEM study of 0.2% Pd/N-CNTs showed that the catalyst consisted predominantly of isolated palladium atoms ( Figure 6). The 2% Pd/N-CNT catalyst contained not only nanoparticles but also a considerable amount of single palladium atoms. As seen in Figure 9, the binding energy of palladium in 0.2% Pd/N-CNTs was 337.7 eV, which may correspond to both the PdO [45] and the ionic palladium stabilized on pyridinic centers of N-CNTs [46]. However, oxidation of the samples was completely excluded in XPS studies; therefore, Pd 3d BE 337.7 eV can reliably be assigned to Pd 2+ -N Py species. Some studies have demonstrated that Pd 2+ -N Py species, in distinction to Pd 0 , weakly chemisorb CO [12,46]. In the case of 2% Pd/N-CNTs, palladium is in the highly dispersed Pd 0 (BE Pd 3d 335.5 eV) and Pd 2+ -N Py (BE Pd 3d 337.7 eV) states.

Discussion
Pd/N-CNTs catalysts have been synthesized and examined in the gas-phase decomposition of FA for hydrogen production. To describe the interaction of palladium with N-CNTs, palladium concentration was varied from 0.2 to 2 wt.%, and a comparison with N-free catalysts was made. XPS experiments with variation of the analysis depth demonstrated that nitrogen on the external graphene surface of N-CNTs was in all structural states.
Irrespective of the doping degree, the predominance of graphitic nitrogen was observed. However, according to XPS and HAADF-STEM studies of the 0.2% Pd/N-CNT catalyst, palladium starts interacting with pyridinic nitrogen centers of N-CNTs to produce isolated ions. Indeed, theoretical calculations show that palladium preferentially interacts with the N Py centers of N-CNTs instead of the N Q centers to form a positive +0.5 e charge [47]. The binding energy of Pd 3d in 0.2% Pd/N-CNTs is 337.7 eV, which coincides well with the BE Pd 3d values for Pd/N-CNTs catalysts reported in literature [47,48]. In these studies, DFT calculations were used to propose the models of palladium anchoring on both the single N Py center and 3N Py -V (where V is the vacancy). It seems interesting that close BE Pd 3d values of 336.9-338.5 eV were also obtained in the case of palladium deposition on g-C 3 N 4 or CN material derived from a metal-organic framework [12,19]. It was supposed that stabilization of isolated palladium ions occurs with the participation of four nitrogen centers (2N Py and 2N Pyr ) [19] or 6N Py , which forms structural elements of the g-C 3 N 4 graphene layer [12]. It should be noted that in the 0.2% Pd/N-CNTs catalyst, the N Py /Pd ratio is 25, which makes it possible to implement any of the proposed models, including Pd anchoring on fragments of the layer, which are similar to g-C 3 N 4 . The possibility of the formation of such fragments in graphene layers of N-CNTs was demonstrated in our earlier study [20].
As the content of supported palladium is increased, this most likely leads to the filling of accessible N Py centers; as a result, metallic nanoparticles start to form ( Figure 10). This assumption is supported by the obtained dependence of CO/Pd on palladium content in the catalysts. It reflects the occurrence of competitive processes: an increase in CO chemisorption owing to the growing dispersion of metallic nanoparticles, and a decrease in CO chemisorption due to formation of ionic palladium. It should be noted that the involvement of N Q centers in the stabilization of metallic nanoparticles, which are more dispersed than in N-free catalysts, cannot be ruled out; their involvement was derived, for example, by Shi et al. [6] from a 0.2 eV shift of the corresponding maximum in the N1s spectrum towards higher binding energies. We did not observe any changes in the N1s spectrum of Pd/N-CNTs catalysts in comparison with N-CNTs. However, a 0.1 eV shift of BE Pd 3d for a more dispersed 2% Pd/N-CNTs (1.4 nm) towards lower binding energies in comparison with 2% Pd/CNTs (2.3 nm) provides the basis to suppose such a possibility. As a result, the involvement of nitrogen centers in the stabilization of palladium decreases the TOF value only by a factor of 1.4 at a ten-fold increase in palladium concentration. For N-free catalysts, the TOF value decreases by a factor of 2.5. As shown by the catalytic experiments, isolated palladium ions are more active than metallic palladium nanoparticles, and the maximum TOF value is observed on the 0.2% Pd/N-CNT catalyst. A strong interaction of palladium with the pyridinic centers of N-CNTs is responsible for high stability of the catalyst in the course of the reaction as compared to the N-free catalyst (Figure 8c). Presumably, activity of the catalysts with high palladium content is determined by the ratio of two palladium species and by the size of metallic nanoparticles. In comparison with N-free catalysts, selectivity on all the N-doped catalysts increased by 5%, which corresponds to nearly a four-fold decrease in CO concentration. Additionally, the discovered stability of the catalyst 2% Pd/N-CNT compared to 2% Pd/CNT confirms the stabilization of both palladium species (isolated ions and metal nanoparticles) by nitrogen centers [49].

Conclusions
Nitrogen centers of N-CNTs were shown to be efficient in the stabilization of different palladium species in the form of isolated ions and highly dispersed metal nanoparticles that are active towards formic acid decomposition for hydrogen production. This allows a conclusion that various nitrogen-containing carbon nanomaterials are promising for use as the supports of metallic catalysts for this reaction. Taking into account an increased activity of isolated palladium ions, it is more preferable to use N-CNMs with predominant pyridinic nitrogen centers in the external graphene layers.

Introduction
Due to increasing worldwide demand for motor fuel, the need to utilize non-traditional heavy oil feedstock (HOF), including heavy oils, natural bitumen, heavy residual oil fractions (tar, black oil), bituminous sands and pyroshales is also rapidly growing. A number of methods for deep processing of heavy oil fractions, heavy oils and oil residues in the presence of catalysts have been reported. Traditional approaches to HOF processing can be divided into two main types [1][2][3][4]. The first type includes deasphalting, thermal processes: gasification and carbonization (delayed, flexicoking, etc.), viscosity breaking and catalytic cracking. The second type includes various hydrogenation processes. Therefore, there is great interest in hydrocracking of heavy residual oil fractions, particularly tar, aimed at the production of the main products of the gasoline and diesel fractions that are in great demand. The chemistry of hydrogenation processes is based on the efficient use of hydrogen. However, reactors used for hydrocracking processes operate under high pressure and require complex instrumentation. Thus, researchers are currently looking for new sources of hydrogen.
One of the actively developing areas combining the advantages of the thermal and hydrogenation refinement processes is catalytic vapor cracking of HOF [5][6][7]. In this process water acts as a hydrogen donor. Its participation makes it possible to increase the yield of the light fraction (boiling point below 360 • C) and decrease the coke yield as well as the contents of sulfur and heteroatoms in the liquid products. The use of catalysts based on Ni [8,9], Fe [10,11], Mo [8,12], Co [13] and Zr [14,15] in the catalytic vapor cracking favors stronger HOF interaction with water, including partial oxidation, low-temperature partial steam reforming and catalytic cracking processes, thus substantially improving the overall efficiency of the process. A different approach was demonstrated in other publications. It was suggested to add methane to CO 2 for cracking of resins obtained by coal pyrolysis. Due to its high H/C ratio, methane is considered to be a hydrogen substitute. Hydrogen is formed by methane reforming with carbon dioxide [16,17].
However, these approaches are complicated for practical implementation because two complex processes have to be optimized: hydrogen production and the hydrocracking itself. The use of formic acid as a hydrogen donor significantly simplifies the process. Formic acid is a high-quality organic hydrogen source, as it has relatively high hydrogen content (4.4 wt.%), low flammability and low toxicity. It is important to note that formic acid can be prepared from the biomass or from CO 2 [18][19][20].
Delayed carbonization is the most popular scheme for processing of tar and other heavy residues. In the current study, modification of the tar carbonization was attempted. The formic acid addition was suggested as a means to regulate the yield and properties of the formed carbonization products.
The goal of the current study was to investigate the effect of the formic acid addition and different catalysts on the yield and properties of liquid tar hydrocracking products.

Materials and Methods
Tar from the Omsk Oil Refinery (Omsk, Russia) was used as a feedstock. The elemental and fraction compositions of the used tar are reported in Table 1. Hydrocracking catalysts are bifunctional. Their hydrogenation-dehydrogenation function and activity in all reactions related to hydrorefining is usually associated with MoS 2 or WS 2 promoted with nickel sulfide. Amorphous or crystalline aluminosilicates, Y [21] or BEA [22][23][24] zeolites are typically responsible for the cracking function. Based on this information, a nickel-molybdenum catalyst and a BEA zeolite were used in this study for the tar hydrocracking.
The use of carbon supports for tar carbonization is desirable because they do not increase the ash content of the resulting oil coke. The artificial mesoporous carbon support "Sibunit" [25,26], which has a surface of 480 m 2 /g and a graphite-like structure, was used for the catalyst synthesis. A TEM image of the Sibunit support is shown in Figure 1. The support particles have an egg-shell shape. Sibunit is a graphite-like carbon material with an interlayer distance of d 002 = 3.52 Å and an average crystallite size in this direction of about 25 Å.
The 8% Ni-2.5% Mo/Sibunit catalyst was prepared by impregnation. Precalculated amounts of Ni(NO 3 ) 2 ·6H 2 O and (NH 4 ) 2 MoO 4 (Reakhim, 98% purity) were dissolved in 10 mL of distilled water and added to a beaker containing 10 g of Sibunit. Then, the sample was dried on a hot plate with a magnetic stirrer and calcined in a muffle furnace in air at 250 • C for 30 min. The obtained NiO-MoO/Sibunit sample was reduced in the hydrogen flow in a flow reactor at 400 • C for 1 h. After reduction the catalyst composition was 8% Ni-2.5% Mo/Sibunit. The XRD pattern of 8% Ni-2.5% Mo/Sibunit reduced in hydrogen is shown in Figure 2.   The peaks observed in the XRD pattern of 8% Ni-2.5% Mo/Sibunit can be attributed to nickel metal or a nickel-molybdenum phase ( Figure 2). The peak at 2θ~44 • is formed by overlapping of different peaks and has a shoulder. This shoulder (2θ~42-43 • ) can be attributed either to graphite or to a nickel-molybdenum phase. The position of the peak close to the shoulder (2θ~44.3 • ) is typical for the nickel metal phase. The diffraction characteristics of the active component of the Ni-Mo/Sibunit catalyst are reported in Table 2.  Tables 3 and 4.  Catalytic reactions of tar were performed in an autoclave at 350-500 • C and 1 MPa pressure. A scheme of the autoclave unit for catalytic reactions of tar is shown in Figure 3. A tar sample (~16 g) or a tar sample (~16 g) with a catalyst (0.4 g) was loaded in the sample holder basket. In another series of experiments, a tar sample (~16 g) with formic acid (0.4 g) or a tar sample (~16 g) with formic acid (9 g) and a catalyst (0.4 g) was loaded in the sample holder basket. Then, the autoclave was heated to the desired temperature and maintained at it for 2 h. The gas phase components were analyzed using a Kristall-2000M gas chromatograph (Chromatec, Yoshkar-Ola, Russia). A thermal conductivity detector was used for analysis of H 2 S, COS, SO 2 , H 2 , O 2 , CO 2 and CH 4 after their separation on a packed column (2 mm × 1.5 m) filled with SKT activated carbon "AQUACHEM" Kazan. A flame ionization detector was used for analysis of the gas phase organic components after separation on a packed column (2 mm × 3 m) filled with Hayesep Q + 0.9% PTMSP (Hayes Separations, Inc. of Bandera, TX, USA).
The concentrations of sulfur and nitrogen and the H:C ratio in the initial tar and products of its processing were determined using a VARIO EL CUBE CHNS-O-analyzer (Elementar Analysensysteme GmbH, Langenselbold, Germany). The fraction compositions of the tar and products formed from it were determined by imitation distillation according to ASTM D7169 using an Agilent 7890B gas chromatograph (Santa Clara, CA, USA).
The elemental compositions of solid phases were determined using the X-ray fluorescence spectrometer ARL Perform'X (Bruker, Germany) with a Rh anode for fluorescence excitation. Powder samples were ground in an agate mortar to a homogeneous finely dispersed state. Powder and liquid (viscous) samples were placed into a special holder covered with a Spectrolenesix polypropylene film that is transparent to the X-ray irradiation. Then, this holder was placed into the spectrometer chamber filled with helium. Automatic spectrum smoothing, background subtraction and calculation of the concentrations were performed using the UniQuant software package (Germany).
The phase composition of the samples was studied by the X-ray diffraction analysis. Diffraction patterns were recorded with a Thermo ARL X'TRA diffractometer (Thermo Fisher Scientific, Basel, Switzerland) using CuKα radiation with a wavelength of 1.54184 Å. The 2θ scan range was 5-75 • , the scanning step was 0.05 • and the accumulation time at each point was 5 s. Diffraction (Powder Diffraction Files) (PDF) and structural (Inorganic Crystal Structure Database) (ICSD) databases were used to identify the phase compositions. The average crystallite sizes were determined using the Selyakov-Scherrer formula from the integral widths of the diffraction peaks.
High-resolution transmission electron microscopy (HRTEM) images were obtained using a JEOL JEM-2010 (Tokyo, Japan) electron microscope with a lattice resolution of 0.14 nm.
The experiments on the formic acid (FA) decomposition were performed in a flow installation using a quartz reactor (inner diameter 6 mm). The catalyst loading was 20 mg. The catalyst was uniformly mixed with 0.5 cm 3 of quartz. The feed consisted of 5 vol.% FA in helium. Its flow rate was 20 cm 3 /min. The experiments were performed in a temperature-programmed mode with a temperature increase rate of 2 • C/min. The reaction progress was followed by monitoring the release of CO and CO 2 .

Thermal Transformations of Tar
The tar carbonization was studied in the temperature range of 350-500 • C. At 350 • C, the tar carbonization yielded gaseous and liquid products as well as oil coke. The amount of liquid product was about 15 wt.%. At 450 • C only gaseous and solid carbonization products were formed. The temperature increase favors conversion of liquid products into gases. The composition of gaseous products obtained at 350-500 • C is reported in Table 5. Methane, ethane and propane were the main gaseous carbonization products. The methane concentration in the products grew with carbonization time. Significant amounts of H 2 S and COS were observed in the reaction products. The contents of the sulfur-containing compounds in gaseous products formed during the tar carbonization at 450 • C are reported in Table 6. Hydrocarbon gases, as well as sulfur-containing H 2 S and COS products, were formed during the process. Gaseous products were periodically removed from the autoclave, and therefore a release of weakly bound sulfur from the tar decreased as the coking process proceeded. This explains the decreasing concentration of sulfur-containing products in the gas as the reaction time increased. Table 6. Concentrations of sulfur-containing compounds in gaseous products of tar carbonization at 450 • C as a function of reaction time.

Catalytic Decomposition of Formic Acid
Hydrogen can be produced from formic acid by a catalytic or non-catalytic reaction. The formic acid decomposition in the gas phase can follow two pathways: dehydrogenation with the formation of H 2 and CО 2 (Equation (1)), and dehydration with the formation of CОand H 2 O (Equation (2)).
Formic acid decomposition can be catalyzed by a number of metals. Palladium catalysts are the most efficient for the hydrogen production from formic acid according to Equation (1) [27][28][29][30][31]. Nickel catalysts have lower activity [32]. However, from the economic viewpoint nickel catalysts deserve more attention than noble metal catalysts. The catalyst supports play an important role in the catalytic activity, especially when a highly dispersed active metal is used. Porous carbon materials are suitable candidates, since they have all the properties required for such support.
Experiments in a flow installation with a quartz reactor were carried out to determine the pathways of the formic acid decomposition over the 8% Ni-2.5% Mo/Sibunit catalyst. The 8% Ni-2.5% Mo/Sibunit catalyst was pre-treated in a flow of 5 vol.% FA in helium at 300 • C to remove the oxide layers formed by contact of metal particles with air. The FA conversion was calculated as a ratio of the sum of the CO and CO 2 concentrations to the initial FA concentration. The selectivity to CO 2 (H 2 ) was determined as a ratio of the CO 2 concentration to the sum of the CO and CO 2 concentrations. The results of the formic acid catalytic decomposition over the 8% Ni-2.5% Mo/Sibunit catalyst are presented in Figure 4. The formic acid decomposition on the 8% Ni-2.5% Mo/Sibunit catalyst preferentially follows the pathway of Equation (1) with the formation of CO 2 and H 2 (Figure 4). Similar experiments on the formic acid decomposition were performed in an empty autoclave without a catalyst. About 3 mL of formic acid was placed into a 300 mL autoclave. The autoclave was gradually heated to 250 • C and maintained at this temperature for 30 min. Then, the temperature was increased in several steps to 350 • C. It was found that the pressure growth in the autoclave began at 120 • C. At 250 • C the pressure was equal to 4 atm. The gas phase composition was studied by gas chromatography. The dependence of the gas phase composition and pressure on temperature is reported in Table 7. Even in the empty autoclave the formic acid decomposition followed the pathway of Equation (1). This is most likely caused by the catalytic effect of the steel autoclave walls on the formic acid decomposition. A CO methanation reaction also occurred to some extent. The obtained results evidenced that formic acid can be used as a source of hydrogen during catalytic reactions of tar. The change in pressure at 250 and 300 • C occurs because the system needs time to reach a steady state. It should be noted that products of formic acid decomposition might be in equilibrium with each other. The water-gas shift (WGS) reaction described in Equation (3) is an industrial process in which water in the form of steam is mixed with carbon monoxide to obtain hydrogen and carbon dioxide.
The WGS reaction is reversible and exothermic (ΔH o = −41.2 kJ/mol). The WGS reaction is thermodynamically favorable at temperatures of 300-400 • C [33]. Iron-based catalysts are typically used industrially. Therefore, the water-gas shift (WGS) reaction also favors the formation of hydrogen and CO 2 .

Effect of Formic Acid on the Catalytic Transformations of Tar
The data on the tar, FA and catalyst loadings, and the selectivity to different fractions are summarized in Table 8. The reaction was performed at 350 • C for 2 h.
In addition to the total tar conversion, it is important to control the composition of liquid products. They were analyzed using the imitation distillation method. The fractions of gasoline (0-180 • C) and diesel (180-360 • C) fractions, vacuum gasoil (360-550 • C), vacuum residue (550-720 • C) and non-eluted residue were calculated according to ASTM D7169 on Agilent 7890B. The fraction compositions of the obtained liquid products are presented in Table 9 and Figure 5.   The presented data demonstrate that the addition of formic acid and the catalysts resulted in redistribution of the liquid products between different fractions. The lowest amount of the gasoline (10.8%) and diesel (39.1%) fractions was observed in the liquid products formed by the thermal tar cracking. The addition of formic acid alone or together with the BEA zeolite to tar increased the contributions of these fractions among the liquid products.
Substitution of the BEA zeolite for the 8% Ni-2.5% Mo/Sibunit catalyst led to a decrease of the gasoline fraction concentration from 16% to 12.7%. This means that the 8% Ni-2.5% Mo/Sibunit catalyst has lower cracking ability than the BEA zeolite. Nickel metal is known to be a good catalyst for methanation and hydrogenolysis of C-C bonds [34][35][36][37]. These properties lead to its higher selectivity to gaseous products (Table 8). Meanwhile, simultaneous addition of the BEA zeolite and the 8% Ni-2.5% Mo/Sibunit catalyst (sample 380) led to the highest yields of both the gasoline and the diesel fractions.
Changes of the sulfur and nitrogen contents in the liquid reaction products were also studied. The results are presented in Table 10. Variations of the sulfur and nitrogen concentrations were symbatic, most likely due to similarities in the mechanisms of desulfurization and denitrogenation of liquid products during the catalytic transformations of tar. The lowest sulfur concentration was observed for sample 5. This result was observed despite the fact that the original carbon support Sibunit contained 0.5 wt.% S. Ni-Mo catalysts are known to be among the best catalysts for purification of diesel fuel from sulfur. When the BEA zeolite is present together with the 8% Ni-2.5% Mo/Sibunit catalyst, it also contributes to desulfurization of the liquid products.

Conclusions
Thermal cracking of tar was studied at 350-500 • C and 1.0 MPa pressure. At 350 • C the tar carbonization resulted in the formation of gaseous and liquid products as well as oil coke. The fraction of liquid products was about 15 wt.%. Only gaseous and solid products were observed when the carbonization temperature was increased to 450 • C. Apparently, the temperature increase led to the conversion of liquid products to gases. The main gaseous tar carbonization products were methane, ethane and propane. H 2 S and COS were observed in the reaction products. Due to this fact, the sulfur concentration on the oil coke decreased in comparison with the precursor tar from 1.5 to 1.28 wt.%.
Specific features of the catalytic tar cracking in the presence of formic acid, BEA zeolite and the 8% Ni-2.5% Mo/Sibunit catalyst were studied at 350 • C 1.0 MPa pressure. The experiments showed that the formic acid addition made it possible to perform efficient hydrocracking of heavy feeds such as tar. Both the tar conversion and selectivity to light (gasoline-diesel) fractions increased in the sequence: tar < (tar -formic acid) < (tar -formic acid -BEA zeolite) < (tar -formic acid -BEA zeolite -8% Ni-2.5% Mo/Sibunit catalyst). Furthermore, significantly lower concentrations of impurities containing sulfur and nitrogen were observed for the (tar -formic acid -BEA zeolite -8% Ni-2.5% Mo/Sibunit catalyst) system. For example, the sulfur and nitrogen concentrations in the tar precursor were 1.50% and 0.86%, 36

Introduction
Currently, there is a rapid growth in the population of the Earth. Over the past 200 years, the number of people has grown by~6 billion, reaching 7.8 billion in 2020. According to forecasts, by the end of the century, the world's population will reach 11 billion people [1,2]. To meet the demand of a growing population, the world needs more and more energy every year. Fossil fuels continue to be the main source of energy resources. Humanity consumes~22 million tons of coal,~12 million tons of oil, and~10 billion m 3 of natural gas per day [3]. The constant growth in fossil fuel consumption is accompanied by an increase in the concentration of carbon dioxide in the atmosphere, which is the cause of climate change. To reduce the negative impacts on the environment and rational use of natural resources, it is urgent to develop technologies for the decarbonization of the energy system [4][5][6]. This is in line with the global strategy to reduce greenhouse gas emissions and the realization of the Paris Agreement's tasks to create a climate-neutral society by 2050 [7].
Under these conditions, hydrogen energy has already been recognized as a reasonable decision in the struggle for climate neutrality [8,9]. By 2050, it is expected that 24% of the world's energy needs will be met by hydrogen. The priority is to obtain renewable hydrogen, produced mainly with the use of wind and solar energy. In the short and medium term, low carbon footprint technologies will prevail, leading to reduced CO 2 emissions. Currently, the volume of hydrogen production in the world is estimated at 75 million tons per year and is expected to increase by 30% in the next five years [10]. Reforming natural gas is one of the main conventional ways of hydrogen production [11][12][13][14][15]. This is the least expensive and energy-efficient method, but to prevent CO 2 emissions, the use of carbon capture and storage technology is required [16,17]. It is advisable for the production of hydrogen to use biofuels (biogas, bioethanol), the source of which can be a renewable raw material-biomass [18][19][20][21][22][23][24].
Bioethanol takes the top place in the list of liquid biofuels. World bioethanol production in 2020 amounted to 26 billion gallons [25]. The world leader in the bioethanol production is the United States, which generated about 13.8 billion gallons. The secondlargest producer country is Brazil, which produced 7.9 billion gallons of ethanol. The bulk of bioethanol is derived from corn and sugar cane. Various agricultural crops with a high starch or sugar content can also become raw materials for the bioethanol production, for example, cassava, potatoes, sugar beets, sweet potatoes, sorghum, barley, etc. The raw materials can also be various agricultural and forestry waste: wheat straw, rice straw, sugarcane bagasse, sawdust. The growth of biomass is accompanied by the consumption of atmospheric CO 2 . Thus, using bioethanol as a feedstock in H 2 production reduces the consumption of fossil fuels and provides carbon neutrality of technology.
The most efficient process for producing hydrogen from ethanol is autothermal reforming (ATP of C 2 H 5 OH) [26]: The energy neutrality of this reaction makes it possible to refer it to energy-saving processes. In addition to the favorable energy balance, this process is characterized by a high yield of H 2 . The effective conversion of bioethanol requires the solution of important problems of increasing the activity of catalysts and their resistance to deactivation. The chemical formula and nanostructure of catalysts strongly affect H 2 yield and the composition of the reaction products in ATR of C 2 H 5 OH [27][28][29][30]. The mode of the first stages of ethanol transformation (dehydrogenation to acetaldehyde or dehydration to ethylene) depends mainly on the properties of the support [26]. Diffusion and transformation of C 2 -intermediates are controlled by the metal-support interface, while the decomposition of C 2 -intermediates and the conversion of C 1 -reaction products occur with the participation of metal centers of the active component. The noble metals Rh, Ag, Au, Pd, Pt, Ru, Re, as well as Ni, or Co, are used as an active metal, while for stabilization of their highly dispersed forms, various oxides (SiO 2 , MgO, La 2 O 3 , CeO 2 -Al 2 O 3 , CeO 2 -La 2 O 3 , CeMnO 2 , MgAl 2 O 4 ) are applied as support [31][32][33][34][35].
Cerium dioxide is a suitable support for ATR of C 2 H 5 OH catalysts due to its redox properties, high oxygen capacity, and the possibility of realization of the strong metalsupport interaction [36][37][38][39][40]. The intrinsic catalytic activity of CeO 2 -based supports and the degree of their interaction with the active component can be regulated by its doping. It was shown that among the tested dopants (M = Gd, La, Mg), La has a more pronounced positive effect on the state and functionality of Ni/Ce 1-x M x O y (M = Gd, La, Mg, x = 0-0.9, 1.5 ≤ y ≤ 2.0) [41][42][43][44]. The introduction of La as a modifier in the support composition enhances the metal-support interaction, which improves Ni dispersion and catalyst stability under the ATR of C 2 H 5 OH. However, it also leads to a diminution of Ni n+ reducibility that can decrease the concentration of Ni o active sites and, consequently, H 2 yield.
The introduction of a promoter is a rather wide-spread approach directed to the improvement of functional characteristics of Ni catalysts developed for various catalytic processes [45][46][47][48][49]. It is shown that bimetallic catalysts have advantages over monometallic ones in the reforming of C 2 H 5 OH [31,35,[50][51][52][53]. In particular, it is indicated [35] that for Ni/CeMnO 2 , the introduction of Cu or Fe increased the ethanol conversion from 57 to 70% and 61% consequently. Ni-Fe/CeMnO 2 catalyst provides higher hydrogen yield (60%) among the other samples because of conducting the reaction to dehydrogenation route, while the feature of Ni-Co/CeMnO 2 sample was high CO selectivity due to the impact of Co in the progress of water-gas shift reaction. It is noted that an increase of H 2 yield over Ni-Co/Ce-Zr-O and Ni-Pd/SiO 2 catalysts is correlated with an increase of reducibility of the bimetallic sample in comparison with appropriate monometallic Ni catalyst [31,50]. The Ru or Rh additives stabilized Co metallic phase under operation in oxidative conditions [54]. A study of a series of Cu-Ni/SiO 2 catalysts with different Cu/Ni molar ratios showed that the Cu-rich catalysts had a higher resistance of coking [55]. Thus, the mechanism of action of the promoter is specific and strongly depends on the composition and method of preparation of the catalyst. It is possible to increase the reducibility of the active component, improve its dispersion or resistance to coking.
This work is devoted to the development of effective bimetallic catalysts for hydrogen production through ATR of C 2 H 5 OH. To improve the functional properties of the Ni/Ce 0.8 La 0.2 O 1.9 catalysts, their nanostructure and reducibility were regulated by introducing various types and content of M promoters (M = Pt, Pd, Rh, Re; molar ratio M/Ni = 0.003-0.012).

Materials and Methods
The Ni-M/Ce 0.8 La 0.2 O 1.9 catalysts (M = Pt, Pd, Rh, and Re; molar ratio M/Ni = 0.003-0.012) were prepared by the combined incipient wetness impregnation method unless otherwise specified. For this, Ce 0.8 La 0.2 O 1.9 support was impregnated by an aqueous solution of the mixture (Ni + M) of metal precursors with a specified concentration. The description of the preparation mode and properties of the Ce 0.8 La 0.2 O 1.9 support can be found in our previous paper [56]. After the impregnation, the Ni-M/Ce 0. 8  In some specially stipulated cases, the catalysts were obtained by the sequential incipient wetness impregnation method. In this case, the Ce 0.8 La 0.2 O 1.9 support was impregnated by an aqueous solution of Ni(NO 3 ) 2 ·6H 2 O with a given concentration. After the impregnation, the Ni/Ce 0.8 La 0.2 O 1.9 catalyst was dried at 90 • C for 6 h, calcined at 500 • C for 4 h in the air, and then impregnated by an aqueous solution of a promoter (M) precursor with a specified concentration. Then Ni-M/Ce 0.8 La 0.2 O 1.9 samples were dried at 90 • C for 6 h and calcined at 500 • C for 4 h in the air.
The Ni content was equal to 10 wt.%, while the molar ratio M/Ni was 0.003 or 0.012. The samples are noted according to their composition and synthesis procedure: the number means the molar ratio M/Ni, while "C" and "S" correspond to the samples obtained by combined and sequential impregnation, respectively.
The catalysts were thoroughly studied by X-ray fluorescence spectroscopy, thermal analysis (TA) (thermogravimetric (TG), differential thermogravimetric (DTG), and differential thermal analysis (DTA)), N 2 adsorption, X-ray diffraction, transmission electron microscopy, and EDX analysis. A description of devices and conditions for studying materials by physicochemical methods can be found in our earlier publications [41,44,47].
ATR of C 2 H 5 OH was investigated in a flow setup with a quartz reactor (14 mm i.d.) at atmospheric pressure, temperature 200-700 • C, a flow rate of 230 mL/min and the molar ratio between reagents C 2 H 5 OH:H 2 O:O 2 :He = 1:3:0.5:1 according to the method described in [41]. Note that the influence of the reaction conditions (500-700 • C, C 2 H 5 OH:H 2 O = 1-4, C 2 H 5 OH:O 2 = 0.2-0.8) was preliminarily studied, and the optimal molar ratio of the reagents for the maximum hydrogen yield was selected. In contrast to our previous studies [41,57], there was no reduction of catalysts before catalytic activity tests. In this case, the active centers will be formed directly under the reaction conditions. The ability to self-activate in the reaction environment ensures that the catalyst can operate on a daily start-up and shut-down cycle without requiring activation before use [58]. It is essential for hydrogen production through reforming for fuel cell technology [59,60]. The Ni-M/Ce 0.8 La 0.2 O 1.9 samples were prepared by the incipient wetness impregnation method, in which precursors of the active component are introduced into the support matrix followed by thermal treatment. The characteristics of the decomposition of metal precursors and the formation of catalysts have been studied by thermal analysis. Figure 1 demonstrates typical TG, DTG, and DTA curves of dried unpromoted and promoted Ni catalysts. For Ni catalyst in a low-temperature region (T < 150 • C), an endothermic effect at T DTA = 125 • C is observed. It is accompanied by a weight loss (−Δm/m = 2.5%) through water desorption. In the temperature range 150-350 • C, there are two endothermic effects at T DTA = 200 and 325 • C. The weight losses of 7.8 and 7.6% were connected with the decomposition of nitrate nickel hydrate to anhydrous Ni(NO 3 ) 2 and then transformation of Ni(NO 3 ) 2 to oxide NiO, correspondingly [61]. At a temperature of 450-900 • C, the change in the weight of the sample is apparently associated with dehydroxylation of the support surface. The total weight loss is equal to 17.4%. Similar behavior was observed for all other Ni-M/Ce 0.8 La 0.2 O 1.9 catalysts. As an example, on the derivatogram of the Ni-Pt/Ce 0.8 La 0.2 O 1.9 sample, there are no effects corresponding to the decomposition of the Pt precursor due to its low content (Figure 1b).

Results and Discussion
From N 2 adsorption data (Figure 2), it follows that fresh Ni-M catalysts are mesoporous materials: the type IV adsorption isotherms with a hysteresis loop of type H3 are observed, which usually indicates the pore shape is wedged with the opening at both ends or groove pores are formed by flaky particles [62]. Hysteresis at partial pressure P/Po = 0.7-1.0 corresponds to the presence of texture mesoporosity [39]. The samples show bimodal pore size distribution with a maximum at ca. 4 and 18 nm (Figure 2c).The texture characteristics of the samples are weakly dependent on the type of promoter: S BET = 70 ± 5 m 2 /g, V pore = 0.20 ± 0.01 cm 3 /g, and D pore = 11.5 ± 0.9 nm (Table 1), which are typical values for materials of such composition [32,35,63]. When using the sequential impregnation method, there is a tendency to some decrease in specific surface area (69 → 56 m 2 /g). This is probably due to differences in the heat treatment procedure for these samples. Single calcination at 500 • C instead of two times reduces the degree of sintering of the material. Note that the specific surface area of the catalysts is 25% lower than the S BET of the Ce 0.8 La 0.2 O 1.9 support (94 m 2 /g), which is due to the partial jamming of the support pores by particles of the active component. The coherent scattering region (CSR) was calculated by the Selyakov-Scherrer method from the broadening of the diffraction peak 1.1.1 assigned to phases having a cubic structure of the fluorite type; CSR (NiO)-peak 2.0.0 NiO phases; CSR (Ni)-peak 2.0.0 phases of Ni • . The average crystallite size of CeO 2 -based solid solution is equal to 8.0 nm, and it is the same for all samples. Thus, the phase composition of the support and its structural characteristics do not change as a result of the introduction of Ni or Ni-M active component ( Table 1). The average crystallite size of the NiO phase is slightly smaller for Ni, Ni-Pd, and Ni-Rt catalysts in comparison to Ni-Pt and Ni-Re samples (16.5 vs. 19.0 nm), which indicates some effect of the second metal. The diffraction patterns of the Ni-M samples show no peaks related to the corresponding M-containing phases because of the low content of M (less than 1 wt.%) and its high-dispersed state ( Figure 3).
TEM data (Figure 4) confirms the XRD study results. On the surface of CeO 2 -based solid solution with crystallite size no greater than 10 nm, NiO particles of 10-20 nm in size were observed.   In the case of catalyst Ni-Pd-0.012 (C), we studied the local distribution of constituent elements using the EDX analysis. Figure 5 shows a HAADF-STEM image of the fresh Ni-Pd-0.012 (C) catalyst with the corresponding EDX maps. It was seen that Ce and La had almost homogeneous distribution in the studied region, which confirmed the formation of a solid Ce x La 1-x O 2 solution. The opposite situation was observed for nickel: areas with an increased nickel concentration were well seen, which indicated the formation of its separate phase. It agreed well with the XRD data ( Table 1). The palladium distribution map did not contain any areas of high concentration. This indicates the absence of individual palladium-containing nanoparticles in the studied area. However, due to the low palladium content, the PdL signal is very weak. Thus, the obtained data do not allow us to conclude whether palladium is located predominantly in the composition of Ni-rich particles or is localized on the surface of the fluorite phase. The redox properties of the active component are an important characteristic of the reforming catalyst since they determine the concentration of active Ni • centers under the reaction conditions. Typically, the samples were activated by treatment in a hydrogencontaining mixture at a high temperature before the ATR of C 2 H 5 OH reaction [31,33,35]. This allows nickel oxide to be reduced and Ni • active sites to form. The Ni • state should be preserved under reaction conditions, while re-oxidation of Ni • will lead to catalyst deactivation. A positive quality of reforming catalytic systems is their ability to selfactivate under reaction conditions [47,48]. In this case, no additional stage of activation was required, there was no consumption of H 2 , and active centers were formed directly under the reaction conditions. Thus, to study the effect of the composition of Ni-M catalyst and the method of its synthesis on the reducibility of Ni 2+ cations, the samples were studied by the thermal analysis in H 2 /He ( Figure 6). There were four temperature regions of weight loss which could be connected with water desorption (T < 200 • C), reduction of nickel oxide (300 • C < T < 600 • C), and reduction surface (200 • C < T < 300 • C) and bulk (600 • C < T < 900 • C) cerium dioxide. For the unpromoted Ni catalyst, three temperature peaks are observed during the reduction of nickel oxide species: at 405, 475, and 545 • C. The observed wide reduction region indicates the presence of various forms of Ni 2+ stabilization in the support matrix. It is known [64] that reduction of large particles of NiO, characterized by weak metal-support interaction, occurs in a low-temperature area (T < 500 • C), while highly dispersed particles of NiO, characterized by strong metal-support interaction, is in the high-temperature region (T > 500 • C). It was also shown that the reduction of Ni 2+ cations moved to a high-temperature region when the dispersion of Ni-supported nanoparticles and their sintering stability increase due to a decrease of the crystallite size of support [41]. With the introduction of a promoter, the behavior of the Ni 2+ reduction was changed ( Figure 6). In particular, in the presence of Pt or Pd, the reduction of Ni 2+ shifted to the low-temperature region with a maximum at~420 • C that could be explained by the H 2 spillover effect [46,58]. This effect was more pronounced when higher content of promoter or the co-impregnation method of synthesis in contrast to the sequential impregnation method was used (Figure 6b). In the case of Rh, the reduction region remained the same, but the part of difficult-to-reduce Ni 2+ species decreased. With the introduction of Re, on the contrary, the part of difficult-to-reduce Ni species increased, and the temperature maximum became equal to 545 • C. Note that due to the low content of promoters, their contribution to the reduction can be ignored in comparison with the reduction peak of Ni 2+ .
As a rule, the reduction of Pt, Pd, and Rh species occurs at T < 200 • C, while reduction of Re cations takes place in the same region where Ni 2+ cations are reduced [58,65]. Different reducibility of Ni-Re catalysts can be explained by the Ni-Re alloy formation under the conditions of TA in H 2 /He that changes the kinetics of the active component reduction [66].
The degree of the interaction between metals in the composition of bimetal catalysts is rather different, and it depends on the chemical composition of the catalyst, its preparation method, conditions of its activation, and exploitation [46,67]. In the case of the weak interaction between nickel and a promoter, the formation of monometal particles took place; in the case of the strong interaction, a surface or bulk alloy was formed. It was shown [58] that for the Ni-Pd catalyst, the co-impregnation promoted the formation of bimetallic particles with a more uniform Pd distribution and a lower surface Pd concentration. In the case of sequential impregnation, the formation of bimetallic particles with a high surface Pd concentration-Pd clusters on the surface of Ni particles took place.
Thus, the type of promoter practically had no effect on the textural and structural properties of fresh Ni-M catalysts. On the contrary, the introduction of even a small amount of a promoter changes the reducibility of the Ni active component. The reducibility of catalysts improves in the following sequence of promoters Re < Rh < Pd < Pt, with an increase in their content, and when using the co-impregnation method. It is expected that the ability of a catalyst to form dispersed Ni o phase and retain it under reaction conditions will control the functional properties of catalysts in ATR of C 2 H 5 OH. Figure 7a shows a typical dependence of the conversion of ethanol and the yield of the reaction products in ATP of C 2 H 5 OH on temperature. It can be seen that in the presence of the Ni catalyst, the conversion of ethanol to a hydrogen-containing gas increased with an increase in the reaction temperature, reached 100% at 400 • C, and then did not change (Figure 7a). The main reaction products are H 2 , CO, CO 2 , and CH 4 . Ethylene, acetaldehyde, and acetone were present in the reaction products in trace amounts only at T = 300 • C. With an increase in the reaction temperature from 200 to 700 • C, the hydrogen yield increased from 0 to~43%. The yield of methane (Y CH4 ) reached a maximum at a reaction temperature of 400 • C, and then at an increase in the reaction temperature decreased and became less than 1% at a T = 700 • C. The CO yield (Y CO ) increased over the temperature range 400-700 • C, and was equal to~40% at 700 • C. The behavior of the temperature dependence of Y CH4 and Y CO indicated an increase in the contribution of the reaction of steam conversion of methane with increasing temperature. The curve of the dependence of the CO 2 yield (Y CO2 ) on temperature passed through a maximum (~67%) at a reaction temperature of 400-500 • C.

Activity of Ni-M/Ce 0.8 La 0.2 O 1.9 Catalysts in ATR of C 2 H 5 OH
The temperature dependence of the activity indices of bimetallic Ni-M catalysts, in general, is similar to those observed in the Ni sample. The advantage of Ni-M systems is a lower temperature to achieve 100% conversion of ethanol and higher values of hydrogen yield (Figure 7, Table 2). Note that an increase in the promoter content has a positive effect on the process performance, while the method of introducing the metal does not matter ( Table 2).  At 600 • C, the hydrogen yield increased in the next row of promoters Pt < Rh < Pd < Re at 100% conversion of ethanol ( Figure 8). For the Ni-Re sample, the obtained parameters of ATR reaction were close to thermodynamic equilibrium values, and the maximum H 2 yield (65%) was attained (Figures 7 and 8). The study of catalyst characteristics after ATR of C 2 H 5 OH reaction shows that catalysts change their textural and structural properties. The values of S BET decreased by 2-3 times, while the average pore diameter increases ( Table 1). The behavior of the distribution of mesopores in size changed: the portion of large pores increased ( Figure 2). It is associated with the intensification of sintering processes under reaction conditions. The increase in the contribution to the porosity of large interparticle pores was observed, which was indicated by the shift of the position of the hysteresis loop to the region of higher values of relative pressure ( Figure 2). There is a tendency to increase the activity of catalysts with an increase in their resistance to sintering and retention of textural properties (Tables 1 and 2).
According to XRD data ( Figure 2, Table 1), support preserves their phase composition and changes in the phase composition of catalysts are largely associated with changes in the structure of Ni-containing phases. It can be seen that for spent samples in comparison with fresh catalysts, the cell parameter of CeO 2 -based solid solution practically does not change, but the average crystallite size increases (8 → 13-18 nm) which is associated with the intensification of sintering processes under the reaction conditions, wherein Ni sample is characterized by lower resistance to sintering, and the best is Ni-Re (13 vs. 18 nm). For all samples after the reaction a metallic nickel phase was formed, the average crystallite size was 16-20 nm (Figure 4b). There was no direct correlation between activity and reducibility ( Figures 6 and 8) in contrast to the literature data [31,50]. All studied Ni and Ni-M catalysts have the capability to self-activation under reaction conditions and can be used without preliminary reduction.
It is noted that the value of the unit cell parameter (a) of Ni o in Ni-M catalysts (M = Pd, Rh, or Re) is larger than Ni o reference data (a = 3.523 Å): 3.532 Å (M = Pd), 3.528 Å (M = Rh), and 3.538 Å (M = Re). It is for these catalysts, in contrast to the Ni-Pt sample, that a more significant promotion effect is observed. It indicates the formation of Ni-M alloy that can change the detailed structure of the nanoparticle surface as well as modify electronic metal properties, which affects the activity of the catalysts due to the synergistic effect between metals.
According to the thermal analysis of spent catalysts in air, carbonaceous deposits were formed during the reaction. Their content depended on the catalyst composition and increased in the following row of samples: Ni-Pt (0.3%) < Ni~Ni-Rh (1.3%) < Ni-Re (4.5%) < Ni-Pd (7.5%). An increase in the activity was accompanied by an increase in the yield of carbonaceous by-products. Thus, a catalysts regeneration procedure should be developed.
It was mentioned above that for the enhancement of Ni catalyst performance in H 2 production through ATR of C 2 H 5 OH, the most effective promotion was with Re (Figure 8). This sample was distinguished by resistance to sintering, reducibility in the high-temperature region, and the formation of Ni-Re alloy particles which optimized the functional properties of the catalyst due to the synergistic effect between the metals. The developed catalysts provide equilibrium values of ethanol conversion 100% and hydrogen yield 65%, which is comparable or higher than those described in the literature ( Table 3) and indicates that they are promising for use in hydrogen energy.

Conclusions
An important issue of our time is the development of technologies for the decarbonization of the energy system. This will reduce the greenhouse effect and become the key to the sustainable development of society. A promising carbon-neutral technology is a production of hydrogen from ethanol, the source of which can be a renewable raw material, biomass. Hydrogen is an important reagent and an alternative energy carrier with high ecological properties. In this work, the development of efficient bimetallic catalysts for the production of hydrogen by ATR of C 2 H 5 OH was carried out.
A series of Ni-M/Ce 0.8 La 0.2 O 1.9 catalysts were prepared at the variation of type (M = Pt, Pd, Rh, and Re) and content (molar ratio M/Ni = 0.003-0.012) of a promoter. The genesis of materials and their properties were systematically studied by thermal analysis, X-ray fluorescence analysis, N 2 adsorption, XRD, and TEM. It was found that the prepared catalysts are mesoporous materials with analogous textural properties (S BET = 70 ± 5 m 2 /g, V pore = 0.20 ± 0.01 cm 3 /g, and D pore = 11.5 ± 0.9 nm) and phase composition (NiO with an average particle size of 18 ± 1 nm, Ce-La-O solid solution with an average crystallite size of 8.0 nm). It was shown that the reducibility of Ni 2+ cations is regulated by the type and content of promoter M as well as the mode of its introduction. It enhances in the following sequence of promoters Re < Rh < Pd < Pt, with an increase in their content, and when using the co-impregnation method. The effect of promoter on the functional properties of catalysts in ATR of C 2 H 5 OH was studied, and the optimal composition of the catalyst was selected. The Ni-M/Ce 0.8 La 0.2 O 1.9 catalysts have the ability to self-activation under the