Carbon usage as a fuel and reductant is still dominating in current global major metal reduction processes, where a carbothermic reduction technology is considered as a mature technology. Consequently, metal production industries play a vital role in the global decarbonization. Researchers have shifted their focus to maximize the potential of carbon alternative in the last decades, and hydrogen is one of the suitable alternatives for primary reduction or metal recycling process. By the end of 2050, the European Union is targeting a significant carbon emission reduction from the industrial sector by 80–95% of emissions generated in 1990 [1]. Australia, as one of the top three hydrogen suppliers in Asia markets, supports the idea of hydrogen as a long-term alternative in the iron and steel industry. In fact, the Council of Australian Government Energy Council has established a Hydrogen Working Group with a vision of clean hydrogen usage for various industries by 2030 [2].

There are several advantages in using hydrogen as reductant and fuel in metals production. Using hydrogen can significantly reduce the carbon footprint of metal-making industries that heavily relied on carbon-based reductant and fossil fuel as energy source. Studies showed that CO2 emission in iron–steel industry can be reduced by 78 to 95% when carbon is replaced by hydrogen that is made from renewable sources [3, 4]. Hydrogen is renewable, and it does not produce harmful emissions in the metals production. There are also, however, several disadvantages of using hydrogen. The current production cost of hydrogen (especially “green hydrogen”) is still high and can lead to a much higher total metals production cost. The current main hydrogen production route through natural gas steam reforming is quite energy intensive, while other methods through electrochemical and electrolysis are quite slow and dependent on electrolyzer which were affected by electricity availability. There is also a problem of scale, where large amount of hydrogen is needed to meet the needs of the large scale of metals industries (especially for iron and steel). There are also complex issues associated with the storage and transportation of hydrogen.

Even though hydrogen is projected to replace fossil fuels for energy generation and reducing agent in metal production processes, its current use in the metal industry only reflected less than 10% of global hydrogen produced [5]. From thermodynamic perspective, there are a number of metals (e.g., W, Mo, Ni, Co, Cu, Pb, Zn, and others) that can be produced through hydrogen reduction of their oxides, however, only limited commercial metals production utilizing hydrogen as a reductant currently exist, i.e., only for the production of selected refractory metal powders (W, Mo), nickel, cobalt, and iron [6]. The limited implementations of hydrogen as reductant for metals production in industrial scale may be due to a number of reasons. For example, direct hydrogen reduction may not be the most effective method if it results in a more complex process. Hydrogen is a weaker reductant compared to carbon at high temperatures, and the endothermic nature of hydrogen reduction of oxides means that the whole energy balance in existing reactors need to be revisited. Challenges in engineering and designing high-temperature reactor and high-throughput reduction reactor still exist. The operation and maintenance are also still costly as mentioned in the previous paragraph. The interaction of hydrogen with gangue materials that can affect the whole process can also be detrimental. Nevertheless, there have been many studies on hydrogen reduction of metal oxides at laboratory level that provide the fundamental understanding and useful for the development of full-scale processes; not just for industrial processing of metal oxides from primary ores, but also from secondary resources (such as residues, slags, and by-products from the plants) which promote circular economy and resource efficiency. This paper will review the previous fundamental studies at laboratory level and current state of hydrogen utilization in the industry.

Thermodynamic and Kinetics

Thermodynamics Aspect

The Gibbs free energy change for a metal oxide reduced by molecular hydrogen into its lower oxide and its metal is shown in Eqs. 1 and 2. In this case, no significant contribution of ΔS to ΔG due to small entropy change from H2(g) to H2O(g).

$${\text{MO}}_{x} + {\text{H}}_{2} \left( {\text{g}} \right) \to {\text{MO}}_{x - 1} + {\text{H}}_{2} {\text{O}}\left( {\text{g}} \right)\quad \Delta G = \Delta G^{o} + RT\;\ln \left( {{\text{pH}}_{2} {\text{O}}/{\text{pH}}_{2} } \right),$$
$${\text{MO}}_{x} + x{\text{H}}_{2} \left( {\text{g}} \right) \to {\text{M}} + x{\text{H}}_{2} {\text{O}}\left( {\text{g}} \right)\quad \Delta G = \Delta G^{o} + xRT\ln \left( {{\text{pH}}_{2} {\text{O}}/{\text{pH}}_{2} } \right).$$

Figure 1 shows the Ellingham diagram for different oxides. It can be seen that thermodynamically, molecular hydrogen can be used to reduce a number of metal oxides, namely ZnO, CoO, NiO, PbO, Cu2O, and Fe2O3. Hydrogen in atomic and plasma states, however, is found to be a more powerful reductant, where the ΔG° of atomic and plasma hydrogen are 3 to 15 times lower than ΔG° of molecular hydrogen. At these states they can also reduce all the other unreducible oxides such as Al2O3, CaO, and MgO.

Fig. 1
figure 1

Ellingham diagram of metal oxides including the hydrogen, hydrogen plasma, and carbon lines

The thermodynamics of hydrogen plasma are highly related to the component in plasma state that consists of atomic hydrogen (H), ionic hydrogen (H+, H2+, H3+), and vibrationally exited molecular (H2*). From the perspective of Gibbs free energy, the monoatomic hydrogen (H) can readily reduce many oxides compared to molecular hydrogen due to the elimination of dissociation step in the process. From the observation of Gibbs free energy of water formation, it was reported that reduction ability of the hydrogen species in plasma follows this order: H+  > H2+  > H3+  > H > H2 [7]. No information has been reported on the ΔG° of H2*.

Kinetics and Mechanism Aspects

This section is not aimed to cover all the kinetics aspect, rather to point out the general solid-state reduction and discuss the possibility of an autocatalytic reaction when hydrogen is used as reductant for solid oxides reduction.

The general mechanism of oxide reduction using molecular hydrogen gas is through the removal of the lattice oxygen in the oxide and driven by the formation of H2O [8]. Reduction processes start with hydrogen adsorption that is activated by the dissociation energy of hydrogen molecules. Hydrogen dissociation itself can be activated by energy that is emanant from surface defect. The presence of oxygen vacancy within the lattice on the surface can reduce the energy barrier required for H2 dissociation. After molecular H2 dissociates into atomic state, it will diffuse into metal oxide structure and rupture the metal–oxygen bonds. Then, heterolytic hydrogen causes the formation of hydroxyl group and metal hydride. The volatile metal hydride will release the hydrogen to form water with hydroxyl then followed by water vapor desorption step. The next step is commonly followed by metal nucleation growth for single-step metal monoxide reduction. For metal oxides that involve intermediates formation, further suboxide bond breaking takes places. At this stage, hydrogen dissociation itself can be activated by the dispersion of reduced metal or suboxide (partially reduced of metal oxides) for some transition metals that bear catalytic behavior.

Kinetic models representing the gas–solid reaction have been presented and summarized in many previous studies (selected studies include [9,10,11,12]), while studies on the kinetics of general oxides reduction by hydrogen gas in isothermal and non-isothermal conditions have been presented in the following studies [9, 13, 14]. Many kinetics studies on reduction using hydrogen or carbon monoxide were mostly related to iron oxides system (selected studies include [15, 16]). The same trend was also observed in the case of the hydrogen plasma where many thermodynamic and kinetic studies were reported on the iron oxide reduction.

There are various factors that can affect the overall reaction kinetics, and they can change during the course of the reaction. Therefore, to fit a reduction process into only a single mechanism may not be realistic. However, a simple approach on generalizing the empirical kinetics data can be useful to provide insights on the detailed reaction mechanism, especially at early stage of the reaction. Contracting sphere or shrinking core and nucleation can be considered as some of the simplest representations and commonly observed in many solid–gas reduction reactions. General model of solid–gas reaction presented below provides a common representation of an isothermal reduction of a bulk sphere oxide with no impurities, defect, nor anisotropy. In nucleation model (Fig. 2b), the removal of oxygen ions creates anion vacancy which at some point rearranges to form a lower state of oxide and/or a metal nucleus. Reaction rate or kinetic characteristic of nucleation model shows an induction period and possibility of autocatalytic reaction (shown in Fig. 2b). Autocatalytic reaction happens when there is enhancement of hydrogen adsorption due to the presence of unsaturated reduced metal oxide (containing metal cations of lower oxidation state). When the reduced oxide grows and overlaps, the reduction follows the contracting sphere model (Fig. 2a), where the diameter of reduced layer is thin compared to the whole sphere. In the contracting sphere model, the oxide is uniformly layered by reduced product, therefore, the reduction is controlled by diffusion. The thickness of reduced layer increases with time and vice versa with the reduction rate, as it shows in the characteristic of the curve in Fig. 2a.

Fig. 2
figure 2

Metal oxide reduction by a contracting sphere (left) and b nucleation (right) model

From kinetics perspective, hydrogen plasma reduction is more favorable due to the high reactivity of hydrogen ions hence providing a stronger reducing agent compared to the molecular hydrogen. It has been postulated that the presence of vibrationally excited molecules increases the reducing potential significantly [17]. The reduction mechanism, however, varies and depends on the plasma and the process type used. In general, there are two types of plasma: thermal plasma and non-thermal plasma. In experiments, the thermal plasma can be used in two types of process set-up: in-flight process (resembles the fluidized bed) and liquid process (resembles the smelting), while the non-thermal plasma nearly adopts the direct reduction process. The overall kinetics are highly affected by the degree of atomization and ionization of molecular hydrogen. The typical atomization on the thermal plasma process is described in the hydrogen plasma composition at 1 atm as shown in Fig. 3. Murphy [18] reported that 50% of the hydrogen was dissociated at 3227 °C and 50% ionized at approximately 14,727 °C.

Fig. 3
figure 3

Thermal hydrogen plasma densities at different temperatures and 1 atm (Source: [17])

Metal Oxide Reduction

The application of hydrogen-based direct reduction process in non-ferrous metal industries is very limited. On the contrary, there has been a history of a large-scale metal production using hydrogen in the iron–steel industry. Circored (by Outokumpu) was known as technology that uses hydrogen solely as a reductant. The first commercial plant based in Trinidad started to operate in 1999 with production up to 105% of its capacity and was halted halfway in October 2001 due to poor HBI market. Then the plant re-operated in November 2004 after Mittal steel bought the asset with a maximum production of 80% of its capacity due to long shutdown. In 2005, this plant recorded to successfully produce 300,000 tons of steel [19]. It is noteworthy that the development of direct reduction of Circored technology was partly triggered by significant cost advantages of using fine ores over pellets or lumps. These low-cost fines were processed into iron with the degree of metallization > 95% and HBI densities above 5.2 g/cm3. Then, the plant was shut down in 2006 due to limited supply of hydrogen and mechanical problems. Comprehensive reviews on systematic evaluation of green steelmaking technologies and detailed fundamental studies of iron oxides reduction using H2 can be found in recent works of Cavaliere [3], and Spreitzer and Schenk [16].

The current review is not focused on hydrogen reduction in the iron–steel systems, rather in the non-ferrous metal oxide systems. Table 1 presents a summary of existing industries that utilize hydrogen for metals production in a commercial scale. The total annual productions of each of these metals using hydrogen are very small compared to the previous ferrous metals production using hydrogen, for example, it is only approximately 1.3 to 4% of the total steel produced from Circored in 2005. Typically, the production process of refractory metals (W and Mo) involves a pyrometallurgy route where metal oxides are reduced in a hydrogen atmosphere through solid-state reduction. Ni and Co are commonly co-produced through a hydrometallurgy route through pressure leaching of the ores in ammonium hydroxide, ammonia (Sherritt Gordon, Caron), or in sulphuric acid (HPAL) to form intermediate products such as carbonates and sulfates. These intermediate products can be further processed using a pyrometallurgy route such as hydrogen reduction to produce Ni, Co oxides and Ni, Co metals. There are no other commercial metals production using hydrogen other than for productions of W, Mo, Ni, and Co. However, there are significant works on hydrogen reduction in laboratory scale which are presented in the following subsections.

Table 1 List of commercial metals production using hydrogen

Tungsten Oxide

Direct reduction of tungsten bearing compound is the common method to produce tungsten metal. Tungsten is one of the metals with high melting point, which makes it impractical to produce through smelting process. Current commercial tungsten powder is dominantly produced via hydrogen reduction of the tungsten oxides. Carbothermic reduction is not preferred for producing pure tungsten, since the use of carbon is usually carried out for direct carburization process for tungsten carbide production. Tungsten oxides are also not the only starting materials for tungsten reduction, as tungsten halides have been used as well. Considering the technical importance of hydrogen reduction in tungsten industry, this section discusses the technology and studies regarding hydrogen reduction of tungsten oxides.

Early studies on hydrogen reduction of tungsten oxides encompass kinetics studies and autocatalytic behavior observation, the selected key studies are given in Table 2. The reduction process of tungsten oxide forms several intermediate products where the important types of tungsten oxides and tungsten metal involved in the process are given in Table 3. Yellow tungsten trioxide (WO3) and blue tungsten oxide (WO3−x) are the common starting materials used, where both oxides are generated from calcination of ammonium–paratungsten ((NH4)10(H2W12O42)0.4H2O) or APT in the presence and in the absence of air, respectively. The details of tungsten ore beneficiation and prior-reduction process have been thoroughly discussed by Habashi [20]. WO2.9, WO2.72, and WO2 are the common intermediate oxides observed during reduction at a controlled dry hydrogen atmosphere. At relatively high moisture content, a volatile oxide hydride (WO2(OH)2(g)) is formed as intermediate along with the other tungsten oxides. The other important phases are the allotropes of tungsten metal, α-W, and β-W. Pure tungsten (α-W) is the common final product sold in the market as opposed to its allotrope known as metastable β-W. The β-W is a tungsten phase with relatively small amount of oxygen reported to be obtained by reducing the blue tungsten at a temperature range of 547–587 °C [25, 26]. The β-W can be stabilized by the addition of elements such as K, Be, and As. β-W was also once claimed as oxides with chemical formula of W3O, but it was confirmed by Mannella and Hougen [27] and Grifis [28] that β-W is an actual tungsten phase with small amounts of WO2 impurities.

Table 2 List of selected studies on kinetics of hydrogen reduction of tungsten oxides and autocatalytic behavior of tungsten oxides reduction
Table 3 Properties of the phases that appear during W–O reduction in hydrogen atmosphere

The reduction process of tungsten oxides was reported to be affected by temperature, water vapor, hydrogen flow, and the state of the starting material. The temperature significantly affects the reduction mechanism in which variation of temperature can change the reaction step and sequence hence affecting the state of the final product. The phase transformation of tungsten oxide reduced in a dry hydrogen atmosphere is given in Fig. 4 (as summarized by Wilken et al. [37]). The state of starting material can alter the reduction step and sequence as well as the hydrogen flow and water vapor presence. Haubner et al. [35, 36] reported that WO2(OH)2(g) intermediate will be formed at a particular pH2/pH2O ratio. This high moisture condition consequently alters the reaction route. Charlton [25] observed that there was a decrease in nucleation rate of suboxides in a high water vapor condition. This phenomenon was suggested to be caused by the re-oxidizing of the suboxide’s nuclei. Therefore, adjustment on the parameters mentioned is essential in industrial practice especially to control the size of the W powder. In an industrial process, the reduction of WO3 is carried out in rotary furnaces at temperature between 600 and 1000 °C [20].

Fig. 4
figure 4

WO3 reduction evolution at dry H2 and high moisture content

Autocatalytic behaviors were observed during hydrogen reduction of tungsten oxides. Bond and Tripathi [40] reported that the addition of 1% of Pd in WO3 enhances the reduction of WO3 into W4O11. In the absence of Pd, the reduction proceeds into W° following a sigmoidal change of reduction rate which was suggested to be caused by the presence of reduced W metal. Fouad et al. [33] confirmed that the presence of W° triggers the catalysis effect. Fouad et al. reported that autocatalytic behavior is observed on the reduction of WO3 to WO2 and WO2 to W° where the chemisorption of H2 by the W° surface accelerates the reduction process of both tungsten oxides. It is also worth to be noted that atomic H was suggested to migrate to the WO3 surface upon the H2 dissociative chemisorption, then migrate to the oxygen within WO3 plane and followed by diffusion into WO3 lattice before it forms HxWO3 [29]. The formation of HxWO3 led to the reduction of WO3 that further enhances the hydrogen adsorption due to the presence of unsaturated reduced metal oxide (contains metal cations of lower oxidation state and the W° as well).

For the last two decades, studies on hydrogen reduction of tungsten oxides gradually shift towards studies on ultrafine to nanoscale of tungsten oxides production. Ultrafine tungsten oxide makes a potential candidate for material in advanced application such as material for sensing application [41,42,43], electrocatalyst [44, 45], dye sensitized solar cell and photocatalyst [46, 47], electrochromic materials [48,49,50], and electrodes for supercapacitors [51, 52]. Shubert and Lassner [53, 54] produced submicron tungsten powders via conventional hydrogen reduction technique. The study outlined the critical factors that determine the consistency of the fineness and size distribution of powder which include initial decomposition of APT to tungsten oxide and the reduction process of tungsten oxide to tungsten. Wu [55] reported empirical relations between the size of tungsten oxide precursor with the resulted tungsten when a violet oxide (W2.72) is used as precursors. The higher diameter of the precursor resulted in the higher average diameter of tungsten but with lower oxygen content. Abdullin et al. [56] reported to have experimentally obtained tungsten metal and tungsten oxide successfully with average particle size of 7 to 30 nm by controlling critical reaction temperature. In summary, every aspect of the starting material as well as the process conditions determine the size distribution of the ultrafine/nano tungsten powder produced.

Molybdenum Oxide

The major share of global molybdenum application in steel industry is ferromolybdenum alloy. International Molybdenum Association reported that in 2019, only 5% of about 260 thousand tons of Mo produce around the world is utilized as Mo metal [57]. About 69% of Mo went into steel industry and the rest into chemicals and foundry. To date, hydrogen is still being used in commercial mass production of pure Mo. Due to its high melting temperature, the production of pure molybdenum is through powder metallurgy process technique. The same reason as mentioned for the tungsten metal production, carbothermic reduction is not industrially applied on pure metal production due to the carbide formation.

The existing technologies for Mo-containing ore processing to metal production have been thoroughly reported [20, 58]. In summary, Mo is mainly extracted from molybdenite or molybdenum bisulfide (MoS2) where sulfide is converted into oxide in a multiple hearth furnace resulting in technical-grade MoO3 which typically has purity of 57%. This technical-grade MoO3 needs to be converted into chemical grade MoO3 to be readily reduced by hydrogen. There are two main conversion methods: first is by sublimating at temperature range of 1100–1200 °C; second is by leaching in an aqueous ammonia to remove the remain impurities and then evaporate to obtain ammonium molybdate crystals. The latter generates ammonium hexamolybdate ((NH4)2Mo6O19), and ammonium dimolybdate ((NH4)2Mo2O7) or ADM that are also able to be used as precursor in hydrogen reduction process. These molybdate compounds can be applied not only as Mo metal precursor but also as catalysts in petroleum refinery process.

The next stage follows a two-step reduction, in which the Mo-containing precursor is reduced into MoO2 followed by further reduction to Mo metal as shown in Eqs. 3 to 4. In an industrial process, the reduction process of MoO3 into MoO2, which is exothermic, is usually carried out at 450 to 600 °C, below melting point of MoO3 (Tm = 800 °C), to prevent agglomeration or caking. This first process (MoO3 → MoO2) is considered as a single-step reduction. Nevertheless, there has been some debates on the possibility of the existence of MoO3 suboxides (i.e., Mo9O26, Mo8O23, Mo5O14, and Mo4O11). Hawkins and Worrel [59] reported that during MoO3 reduction at 300 to 450 °C, they observed not only MoO2 but also other suboxides. In several studies [60,61,62], it was suggested that Mo4O11 exists during MoO3 to MoO2 reduction at a range of temperatures. Ressler et al. [62] speculated that Mo4O11 was formed at a reduction temperature above 500 °C not as intermediate, but as a parallel product that is generated along with MoO2. Dang et al. [60] proposed that Mo4O11 exists as an intermediate during the reduction at temperature above 440 °C, following a sequence of MoO3 → Mo4O11 → MoO2, which is in an agreement with Sloczynski [63], Schulmeyer and Ortner [64], and Latif et al. [65].

$${\text{MoO}}_{3} \left( {\text{s}} \right) + {\text{H}}_{2} \left( {\text{g}} \right) \to {\text{MoO}}_{2} \left( {\text{s}} \right) + {\text{H}}_{2} {\text{O}}\left( {\text{g}} \right),$$
$${\text{MoO}}_{3} \left( {\text{s}} \right) + 2{\text{H}}_{2} \left( {\text{g}} \right) \to {\text{Mo}}\left( {\text{s}} \right) + 2{\text{H}}_{2} {\text{O}}\left( {\text{g}} \right).$$

The second step of reduction MoO2 → Mo is known to be endothermic [65]. It is noteworthy that hydrogen waste (diluted hydrogen) from this stage can be used for the first stage of reduction in order to avoid overheating, while the second step usually uses a ‘fresh’ hydrogen. Studies reported on the kinetics MoO2 reduction into Mo metal different kinetics models at different reduction temperatures. Kennedy and Bevan [66] reported that reduction of MoO2 into Mo at 650–800 °C follows the contracting sphere model, while Schulmeyer and Ortner [64] reported the same shrinking core model to fit to reduction at 1100 °C. Majumdar et al. [67] and Kim et al. [68] reported that the reduction process at 500 to 1000 °C follows the recrystallization and nucleation models, respectively.

Selected studies related to hydrogen reduction of molybdenum oxides are given in Table 4. It has been mentioned above that the molybdenum oxides reduction follows two major reduction steps that are thermodynamically opposite. Likewise, the kinetics of each reduction step provide a different behavior. However, both reduction steps were reported to be catalyzed by the presence of lower oxide state of metallic Mo [63, 69]. The formation of the catalytic sites during the reduction is dependent on the H2O pressure, condition, and the properties of the surface. When abundance catalytic sites exist, the reduction rate will be limited by the breaking Mo–O bond. On the contrary, the H2 dissociation will be the rate-limiting step [70].

Table 4 Selected studies on the kinetics and reduction mechanism of MoO3 using hydrogen

Recent studies on hydrogen reduction of molybdenum oxides have also been focused on the preparation of ultrafine or nano-size of Mo powder. Established method of Mo metal production via powder metallurgy involves compacting and sintering step as one of the essential processes. Finer size of Mo powder is preferred in compacting process and it also increases the sinterability of Mo. Early study by Lamprey and Ripley [72] reported the production of Mo powder with range of size 10–100 nm via hydrogen reduction of molybdenum chloride. Saghafi et al. [73] introduce a high-energy ball-milling method to produced nanopowder with a size range of 50–133 nm by mechanically activating the MoO3 precursors for 20 h before hydrogen reduction. Another method by Sun et al. [74], found to successfully control the nucleation and growth of Mo powder at average size of 80 nm, was carried out via salt-assisted hydrogen reduction. There are other novel techniques reported on the production of ultrafine Mo powder such as via metallothermic reduction [75], self-propagating high-temperature synthesis [76], plasma deposition [77], and molten salt technique [74, 78].

Nickel Oxide

Nickel is also one of the few metals that is mass produced through hydrogen reduction of its metal oxides. Nickel extraction process itself consists of sequences of metallurgical processes depending on the origin ores type, i.e., lateritic/oxidic ores or sulfide ores. Hydrogen reduction is part of the overall process where it is used to reduce intermediate nickel oxide product to nickel metal. Commercial nickel products are marketed into two categories based on the chemical composition. Class I is the type of nickel with high purity that consists of > 99.0% of Ni in the form of powder, pellets, briquettes, and electrolytic nickel. And Class II is nickel compounds and alloys that are usually marketed as Ferronickel (20–50% of Ni) and metalized Nickel Oxide (76% of Ni). It is noteworthy that the nickel market is dominated by the Class I that accounts for 55% of global nickel production and about 66% of the world’s nickel goes into stainless steel industry (Nickel [79]).

Considering that Ni is a more common metal compared to the prior two refractory metals mentioned (W and Mo), there have been extensive studies on the reduction mechanism of nickel oxide. The hydrogen reduction of NiO follows and overall reaction is shown in Eq. 5, nevertheless, the detailed transformation is quite complex. There have been many studies on reduction kinetics reported on various reduction mechanism due to wide variation in experimental conditions and techniques. This can also be observed from the list of activation energy given in Table 5, where the Ea value varies up to one order magnitude. The discrepancies problem highlighted by studies leads to the development on structural transformation observation. Kinetics studies and reduction models of NiO reduction are known to be highly topological and related to morphology evolution. Paravano [80] reported that defect structure on NiO crystal lattice influences reduction rate under hydrogen atmosphere. Reduction pattern in porous NiO is highly related to the particle size, structure, and diffusivity/access for the transport of reactant or product gas. Hence, uniformity of NiO precursor and method should allow controlled structural observation. For example, Hidayat et al. [81, 82] developed a NiO preparation method that allows consistence porous-free observation from a dense oxide by oxidizing high-purity Ni foil.

$${\text{NiO}}\left( {\text{s}} \right) + {\text{H}}_{2} \left( {\text{g}} \right) \to {\text{Ni}}\left( {\text{s}} \right) + {\text{H}}_{2} {\text{O}}\left( {\text{g}} \right).$$
Table 5 Summary of activation energy of hydrogen reduction of NiO reported by various studies

The key aspects identified that control the reduction pore-free/dense NiO are the formation of water vapor and interface properties. The presence of water is suggested to significantly influence the extent of reaction, i.e., Hidayat et al. [81] reported that when steam content increased by 30%, the extent of reduction reduces about 60–85% and 20–25% at temperature 500 to 675 °C and at 800 to 1000 °C, respectively. Sample geometry and characteristic (e.g., presence of pore) will highly affect the transport of H2O product where one can expect the reduction rate on pore-free severely reduced compared to porous one. However, it has also been reported that pores formation observed during reduction might compensate the water vapor transport problem. Hidayat et al. [81] reported that porous Ni structure with average size of 10–20 nm is formed during early reduction of a dense NiO. It was also reported that different H2 partial pressures and driving force for reduction (∆G) result in different morphologies of the nickel product and different pore sizes, which can be clearly seen in the maps presented in Fig. 5a and b.

Fig. 5
figure 5

Different reaction interface structures formed in the reduction of dense synthetic NiO at 500 to 1000 °C in a H2–N2 gas mixtures (Ptotal = 1 atm); and b H2–H2O mixtures (Ptotal = 1 atm) (Source: [81])

From a kinetic study by Hidayat et al. [82], a slowdown in reduction rate was observed at 700–800 °C, and was suggested to be caused by the densification of unstable Ni porous structure, which is in agreement with the observation reported by Rhamdhani et al. [100]. Ni densification during reduction of NiO particle, as given in the model in Fig. 6, was also observed under environmental-TEM by Jeangros et al. [85]. The reduction model starts with H2 adsorption and dissociation (Fig. 6a) on NiO surface defect. Followed by Ni nucleation and water desorption (Fig. 6b), Ni seed generation on NiO surface was reported to be the initiation of slow reaction. Then Ni grows with pores formation as the compensation of volume loss of oxide (Fig. 6c). In Fig. 6d, NiO is depicted to be trapped inside Ni domain which also hinders H2 access. At the final stage (Fig. 6e), the surface energy of the system was minimized at temperature higher than 600 °C which leads to complete reduction. They also reported that the Avrami nucleation model fits the overall kinetics data and microstructure observation.

Fig. 6
figure 6

NiO powder reduction involving porous formation and densification process in mechanism a hydrogen adsorption and dissociation b Ni nucleation c Ni growth and porous formation d NiO isolation e final densification (Source [85])

Observations on pore-free NiO powder reduction have been extensively carried out in the context of understanding its catalysts behavior for solid oxide fuel cell (SOFC) application. Hydrogen reduction of partially oxidized Ni wire was used by Manukyan et al. [84] at a wide range of temperature (270–1320 °C) dedicated for that purpose. The overall structure observation is summarized in the reduction mechanism given in Fig. 7. Manukyan et al. [84] observed a complex network of macro-porous structure with unreduced NiO at low temperature (270–500 °C), while, at high temperature (> 900 °C), a rapid growth of Ni structure was observed in which this phenomenon makes the porous structure smaller and denser. Meanwhile, combination of each structure from low to high temperatures was observed at temperatures between 500 and 900 °C. They also reported that the Avrami nucleation model perfectly fits the reduction.

Fig. 7
figure 7

Porous-free NiO powder reduction mechanism proposed by Manukyan et al. [84] at different temperature ranges (Source: [84])

Zinc Oxide

Studies on the direct reduction of zinc oxide using hydrogen are commonly related to the recovery process of valuable metals from secondary sources. The reduction process is mainly to extract other valuable metals where zinc content is usually separated by sublimation. The reaction of zinc/zinc oxides in hydrogen gas was also reported to be associated with the formation of vapor phase. For example, Navarro et al. [101] reported that Zn metal was successfully reduced from secondary source EAF dusts (mainly contain Fe2ZnO4) and synthetic dust using H2 as reducing agent at 850 °C and 870 °C, respectively. Their result is an agreement with Kazemi and Sichen [102], where all the zinc metal produced was transferred to the gas stream, enabling its complete separation from iron as shown in Eqs. 69. Meanwhile, Lee et al. [103] investigated the kinetics of copper and zinc recovery from brass converter slag using hydrogen gas via pyrometallurgical process between 900 and 1050 °C. Brass secondary slag contains various elements (Pb, Sb, Fe, Ni, Si, etc.) depending on sources commonly recovered via hydrometallurgy process. Pyrometallurgical recovery is not common, yet fundamental knowledge resulting from the study can be applied in a fluidized bed process of ZnO with syngas. Lee et al. [103] managed to recover spherical zinc with an average size of 1-6 µm.

$${\text{Fe}}_{2} {\text{ZnO}}_{4} = 2/3{\text{Fe}}_{3} {\text{O}}_{4} + {\text{ZnO}} + 1/6{\text{O}}_{2} \left( {\text{g}} \right),$$
$${\text{ZnO}} + {\text{H}}_{2} \left( {\text{g}} \right) = {\text{Zn}}\left( {\text{g}} \right) + {\text{H}}_{2} {\text{O}}\left( {\text{g}} \right),$$
$${\text{Fe}}_{3} {\text{O}}_{4} + {\text{H}}_{2} \left( {\text{g}} \right) = {\text{FeO}} + {\text{H}}_{2} {\text{O}}\left( {\text{g}} \right),$$
$${\text{FeO}} + {\text{H}}_{2} \left( {\text{g}} \right) = {\text{Fe}} + {\text{H}}_{2} {\text{O}}\left( {\text{g}} \right).$$

Hydrogen reduction of ZnO can be spontaneously carried out at temperature above 1200 °C. There have been a number of extensive studies carried out on ZnO reduction in hydrogen atmosphere [104,105,106]. These studies suggested that (i) the reduction process is affected by hydrogen partial pressure; (ii) the reaction rate is proportional to ZnO surface area; (iii) the reduction mechanism involves decomposition process of ZnO into Zn and oxygen, and water formation in which the former process reported to be the rate determining step; (iv) the thermal decomposition of ZnO is influenced by the presence of a third gas, i.e., nitrogen, zinc vapor. There are also other highly relevant studies regarding kinetics and reaction mechanism of hydrogen reduction of ZnO listed in Table 6.

Table 6 Previous studies on ZnO reduction by hydrogen gas which highly related to the Zn metal extraction

These extensive studies were conducted more than half century ago, and there have been no other significant progress on hydrogen reduction of ZnO studies reported in the literature. Recent studies of hydrogen reduction of ZnO have been focusing on H2 treatment looking into ZnO photocatalysis properties [111], observing ZnO-based catalysts behavior in H2 atmosphere [112], and investigating on the effect of H2 reduction on n-type semiconductor ZnO properties [113].

Lead Oxide

Lead is one of the top five non-ferrous metal produced globally, and its main processing route involves carbon as reductant in a direct smelting method [114]. Hydrogen is not used in the current lead production technology. There are limited studies on thermodynamics and kinetics observation of solid-state reduction of lead oxide using hydrogen, summarized in Table 7. Unlike other metal oxide that commonly might form metal carbide, Pb is known as one of the few metals that has no stable carbide form.

Table 7 Lists of studies on PbO and selected studies on Pb-containing glass reduction by hydrogen

While PbO reduction by hydrogen is thermodynamically feasible there is little information regarding kinetics and mechanism. Hydrogen reduction reaction of PbO is simply shown in Eq. 10, where the reduction proceeds to Pb metal without any intermediate compound formation. Pb2O intermediate was once postulated yet later established that Pb2O was just a mixture of Pb and PbO [119]. An early study by Gallo [120] reported that PbO is initially reduced by hydrogen at 160 °C and a steady reduction takes place at 240 °C, while another study found the initial reduction takes place at 270 °C [115]. Culver et al. [119] carried out a study on the kinetics of PbO reduction by measuring H2O product at a temperature range of 475–775 °C using H2–N2 and H2–H2O mixtures. They reported that hydrogen reduction of PbO is independent of the gas flow rate but influenced by the PbO particle size [119]. The rate of PbO reduction appeared to be proportional to H2 partial pressure in which chemisorption of hydrogen is considered as the rate controlling step [119]. Water vapor formation for many metal oxides during hydrogen reduction gives a retarding effect on the reduction rate. Nevertheless, there is still no clear understanding on the effect of water vapor and hydrogen reduction of PbO. The presence of N2 in the system was reported to not alter the H2 potential but H2O presence reduced the H2 potential. Culver et al. [119] reported that energy activation of PbO during reduction in H2–N2 (167.36 kJ/mol) is higher than in H2–H2O mixture (156.9 kJ/mol). In addition, a recent study by Ivanov et al. [115] reported that tetragonal PbO and orthorhombic PbO do not show a different kinetic behavior. The study was conducted by measuring the H2O product using a gas chromatograph. They also found that the addition of Pb metal into a PbO precursor during hydrogen reduction showed no effect in kinetics as well, which indicated an absence of autocatalytic behavior.

$${\text{PbO}}\left( {\text{s}} \right) + {\text{H}}_{2} \left( {\text{g}} \right) \to {\text{Pb}}\left( {\text{s}} \right) + {\text{H}}_{2} {\text{O}}\left( {\text{g}} \right).$$

Due to high proportion of lead metal recycling, the feasibility of hydrogen reduction on lead-bearing slag has also been studied. Statistics showed that lead is one of the metals with highest recycling rate where it is estimated that about 56% of lead metal is produced by recycling process [5]. It is worth to note that some metal plants (e.g., copper smelting plant) generate a slag with substantial amount of lead in which it is economically attractive to be recycled. Study on lead recovery using hydrogen for slag cleaning will have additional benefit from an environmental point of view. However, very limited information is available on the Pb recovery from complex slag systems. Studies related to it have been performed mainly not for Pb extraction purpose but to investigate the Pb formation within glass [116, 117, 121]. Meanwhile, for the purpose of extraction study, Pal et al. [118] reported the Pb reduction behavior from simple lead silicate slag system in the mixture of H2 and N2. The main finding of their study shows that reduction rates of PbO at slag/gas and slag/refractory interfaces were higher than at slag/metal interface.

Copper Oxide

Study on CuO reduction using hydrogen has gained interest as early as 1921 with the main focus in the observation of catalytic behavior during the process [122]. The detailed kinetic aspect and reaction mechanism of CuO reduction by hydrogen, however, are still not well understood, and there have been some disagreements on the reported reduction step and sequence. Theoretically, CuO reduction follows the sequence shown in Eq. 11, where the intermediate compounds possess well-defined crystal structure. However, the condition where these suboxides appear as stable phase during hydrogen reduction remains unclear. By measuring the kinetics of oxygen loss, Li and Mayer [123] reported that the reduction of CuO proceeds by forming Cu2O suboxide under highly vacuum condition (2.6 × 10–4 Pa). Other studies reported that CuO reduces directly into metallic Cu without the formation of intermediate suboxides under a normal supply of hydrogen (gas flow rates > 15 ml/min) at temperature above 200 °C [124,125,126]. When the hydrogen supply is limited (< 1 ml/min), then suboxide Cu2O was detected during reduction. Further study is needed to comprehensively describe the CuO reduction mechanism at different ranges of conditions that include hydrogen pressure, temperature, and surface defect. A list of studies regarding the kinetics and mechanism of copper oxide under hydrogen atmosphere are provided in Table 8.

$${\text{CuO}} \to {\text{Cu}}_{4} {\text{O}}_{3} \to {\text{Cu}}_{2} {\text{O}} \to {\text{Cu}}.$$
Table 8 Selected studies on kinetics and reduction mechanism of CuO by hydrogen gas

Although current industrial process to produce copper is not involving reduction using hydrogen, it is frequently used to prepare CuO/Cu-based catalysts and Cu for microelectronic components. Novel approach on utilizing hydrogen plasma has been reported for preparing CuO/Cu for both applications mentioned [131]. They reported a method of reducing CuO to Cu using a low-energy hydrogen plasma and found that it lowers the activation energy of the process. In electronic devices, adhesion between copper and polymer is considered poor. Sawada et al. [132] introduced a novel approach to improve adhesion by plating Cu metal on polymer surface using hydrogen plasma generated via atmospheric pressure glow discharge (APG). The method utilizes low-temperature plasma hydrogen to treat copper oxide precursor on epoxy resin surface which resulting an excellent bond between metal and polymer.

Titanium Dioxide

Titanium is the seventh most abundant metal in the earth’s crust, yet its low volume production makes Ti named as ‘rare metal.’ The current major Ti metal production process, the Kroll process, is energy intensive and involves complex methods which leads to high production cost. Hence, various attempts have been reported for alternative Ti production method that is less expensive such as electrolysis or combination of electrolysis and metallothermic, which is comprehensively reviewed by Takeda et al. [133]. A hydrogen-assisted magnesiothermic reduction (HAMR) was reported to utilize a combination of hydrogen and metallothermic for TiO2 reduction [134]. In HAMR method, TiO2 is reduced by a Mg particle in the presence of Mg-bearing slag in H2 atmosphere. The main purpose of hydrogen addition during metallothermic reaction is to achieve low oxygen level, while Mg-containing salts are favorable in enhancing reaction kinetics. Lefler et al. [134] reported that the formation of MgO is a rate-limiting step that also leads to MgTiO3 formation. Meanwhile, TiO2 reduced through its suboxides and α-Ti(H)x(O)y intermediate phases with sequence given in Eq. 12. The deoxygenation of α-Ti(H)x(O)y to Ti was reported to be the slowest step from overall oxygen removal process.

$${\text{TiO}}_{2} \to {\text{Ti}}_{2} {\text{O}}_{3} \to {\text{TiO}} \to \alpha {\text{ - Ti}}\left( {\text{H}} \right)_{x} \left( {\text{O}} \right)_{y} \to {\text{Ti}}$$

Molecular hydrogen is thermodynamically unable to reduce TiO2 into Ti metal completely as shown in Ellingham diagram given in Fig. 2. This has been proven by early work of Newberry and Pring [135] where TiO2 was only reduced into suboxide (TiO) state when reduced by H2 under high pressure of 130 atm. This thermodynamic limit leads to some ideas on using mixtures of C and H2 as reductant. Reduction of TiO2 using carbon and hydrogen mixture results in different types of intermediate compounds. For example, Zhang and Ostrovski [136] carried out a study on the reduction of TiO2 by a mixture of CH4–H2–Ar gas, where TiO2 not only reduced into suboxides but also carburized into solid solution of titanium oxy-carbide (TiO–TiC)ss corresponding to the sequence shown in Eq. 13. They also reported that the presence of water provides a retarding effect on the Ti3O5 to Ti2O3 reduction. Titanium oxy-carbide then can be further reduced to Ti. It has been reported by previous studies that Ti can be obtained by electrolysis process of titanium oxy-carbide [137,138,139].

$${\text{TiO}}_{2} - {\text{Ti}}_{5} {\text{O}}_{9} {-}{\text{Ti}}_{4} {\text{O}}_{7} {-}{\text{Ti}}_{3} {\text{O}}_{5} {-}{\text{Ti}}_{2} {\text{O}}_{3} {-}\left( {{\text{TiO}} - {\text{TiC}}} \right){\text{ss}}$$

TiO2 is known as a main component of a pigment, and is also an essential material for photocatalysis, photovoltaic, self-cleaning surface, and biomaterial applications. In photocatalysis and photovoltaic application, hydrogen is widely used as reducing agent to reduce nano-size anatase to obtain hydrogenated TiO2 or known as ‘black TiO2.’ The nano-size anatase possesses higher photocatalytic effect among TiO2 polymorphs, in which hydrogen treatment is able to enhance its ability to absorb visible light spectrum due to the formation of black TiO2. The interaction of hydrogen with specific anatase surface is essential in hydrogenation process where different anatase facets provide different hydrogen adsorption behaviors [140, 141]. Hydrogenation can proceed using molecular H2 at high temperature and high pressure or using atomic hydrogen. Different approaches give different photocatalysis performance. For example, hydrogenation using plasma hydrogen was claimed to improve the electrochemical performance of TiO2 [142].

Cobalt Oxide

In industry, cobalt is co-produced in nickel and copper production processes. Cobalt is dominantly used as alloying element for aircraft superalloy, permanent magnet alloy, wear-resistant alloy, and corrosion-resistant alloy. Pure cobalt metal has few applications as chemicals for pigment and catalyst for petroleum. There are several common approaches to reduce Co3O4 into metallic cobalt, i.e., cobalt oxide reduction using reductant, pyrolysis of cobalt carboxylates, and hydrothermal treatment. Hydrogen or carbon monoxide are typically applied as reducing agent for the former.

There are numbers of studies reported on the hydrogen reduction of gray cobalt (II) oxide (CoO) and black cobalt (II) cobalt (III) oxide (Co3O4). A list of selected studies on kinetics and mechanism of cobalt oxides reduction is given in Table 9. Early study by Glaser [143] reported that Co2O3, Co3O4, and CoO start to reduce at an initiation temperature of 182 °C, 207 °C, and 228 °C, respectively; while recent experimental study by Bulavchenko et al. [144] found that the initiation temperature for Co3O4 was 190 °C. It has been reported that reduction step/sequence of cobalt oxide is influenced by temperature [145], associated elements [144], and hydrogen partial pressure [144, 146]. The sequences of various hydrogen reduction reactions are given in Eqs. 1416.

$${\text{Co}}_{3} {\text{O}}_{4} \to {\text{Co}}_{(1 - x)} {\text{O}}/{\text{CoO}} \to {\text{Co}}$$
$${\text{Co}}_{3} {\text{O}}_{4} + {\text{H}}_{2} \left( {\text{g}} \right) \to 3{\text{CoO}} + {\text{H}}_{2} {\text{O}}\left( {\text{g}} \right)$$
$${\text{CoO}} + {\text{H}}_{2} \left( {\text{g}} \right) \to {\text{CoO}} + {\text{H}}_{2} {\text{O}}\left( {\text{g}} \right)$$
Table 9 Selected studies on kinetics and reduction mechanism of cobalt oxide using hydrogen

At temperature lower than 291 °C, Co3O4 was reported to directly reduce into Co metal due to instability of CoO in this temperature range. Above 291 °C CoO was present as a stable intermediate [145]. Co3O4 is frequently used as catalyst with other phases as a support element (i.e., Al2O3, ThO2, TiO2, SiO2). The addition of a support element affects the chromium oxide reduction sequence. Bulavchenko et al. [144] reported that the reduction of a single-phase Co3O4 in the temperature range of 500 to 900 °C proceeds in a single step, however, in the presence of γ-Al2O3 support, it follows a two-step reduction process. The study also reported that the reduction occurs through CoO formation where the γ-Al2O3 support hinders further CoO reduction into cobalt metal. Hydrogen partial pressure was found to have an effect on the reduction sequence. At higher H2 partial pressure, the reduction proceeds without the formation of intermediate CoO [144]. The use of plasma hydrogen for reducing Co3O4 was reported to require less activation energy [149].

Due to its catalytic behavior, cobalt oxide is also applied in various emerging application such as sensors, materials for Li-ion batteries, electrocatalysis, and electrochromic devices. It is noteworthy that hydrogen is utilized as a reductant to prepare cobalt or cobalt alloy nano-particles for these applications [156, 157].

Chromium Oxide

Studies of chromium oxides reduction by hydrogen gas are not widely explored as the conventional route to produce metallic chrome is through electrolytic and metallothermic reductions. Thermodynamically, the Gibbs energy formation of Cr2O3 is lower than that of hydrogen oxidation at ambient pressure. Direct reduction by hydrogen was reported to fail many times due to reaction condition difficulties. However, a patent submitted by Malcolm [158] claimed to successfully reduce Cr2O3 into Cr metal in an electric tube furnace over the temperature range of 900 to 1100 °C with high-purity hydrogen gas. Cr2O3 also was reported to be reduceable above 1130 °C with oxygen free and continuous removal of water vapor [159]. By strict water vapor removal Chu and Rahmel [160] reported to successfully reduce Cr2O3 into Cr metal at 900–1200 °C at pH2O/pH2 = 10–5. Chu and Rahmel [160] also reported the effect of variables on kinetics, where the study found that the rate of chrome oxide reduction was proportional to the square root of hydrogen pressure and retarded by water vapor content. Table 10 summarized selected studies on chromium oxides reduction by hydrogen.

Table 10 Selected studies on chromium oxides reduction by hydrogen or mixture of hydrogen

There are also studies on the hydrogen reduction of chromium oxide reported in which it is mainly not for the purpose of metal extraction. Chromium oxide is extensively used as catalyst, hence, the interaction of hydrogen with chromium oxides is essential on understanding the chemisorption process. Ostrovskii et al. [161] reported the interaction of hydrogen with porous chromium oxides at a temperature of 450 °C. Hydrogen chemisorption and absorption on chromium oxide proceed in the form of water and coordination bound atoms, respectively. Another study also found that hydrogen can be utilized for scaling process in which hydrogen plasma was useful to be utilized on compact chromium oxide (Cr2O3) removal to regain electrical and thermal conductivity of metal [166]. In addition, chrome is highly used as an alloy in stainless steel production, studies on Cr2O3 reduction by methane–hydrogen were also reported as an alternative for high carbon ferrochromium production [163, 167, 168]. However, oxide was converted into carbides with the overall reaction shown in Eq. 17.

$${\text{Cr}}_{2} {\text{O}}_{3} \left( {\text{s}} \right) + 13/3{\text{CH}}_{4} \left( {\text{g}} \right) = 2/3{\text{Cr}}_{3} {\text{C}}_{2} \left( {\text{s}} \right) + 3{\text{CO}}\left( {\text{g}} \right) + 25/3{\text{H}}_{2} \left( {\text{g}} \right)$$

Manganese Oxide

Carbothermal smelting process is known to be the conventional method for manganese metal production. Manganese oxide (MnO2) cannot be reduced fully to Mn metal by using H2 gas as suggested by theoretical calculation of Ellingham diagram. However, partial reduction of MnO2 into lower oxide state is feasible at PH2O/PH2 ratio lower than 105 [169]. Total reduction to metallic Mn is thermodynamically possible to be conducted above the Mn melting point with a specific ratio of PH2O/PH2 (i.e., less than 3 × 10–4) [169].

There are few studies reported on the kinetics of hydrogen reduction of manganese oxide as summarized in Table 11. From these studies, one can see that MnO2 cannot be reduced fully into metallic Mn rather into suboxides (MnO or Mn3O4). Barner and Mantell [169] reported that the reduction of MnO2 proceeds topochemically corresponding to the reaction sequence shown in Eq. 18. At lower temperature (below 250 °C), the reaction proceeded until Mn3O4 formation occurs and the chemical reaction was reported to be the controlling mechanism. At higher temperatures, MnO was obtained and diffusional resistance and solid–gas chemical reaction were reported to be the controlling steps [169]. Bruijn et al. [170] observed that MnO2 and Mn2O3 reduced into MnO via Mn3O4. They also reported that the crackling core model (as opposed to the shrinking core model) appears to be suitable and fit with the calculation and measured data. It is noteworthy that reduction of manganese oxide shows an autocatalytic behavior [171].

$${\text{MnO}}_{2} \to {\text{Mn}}_{2} {\text{O}}_{3} \to {\text{Mn}}_{3} {\text{O}}_{4} \to {\text{MnO}}$$
Table 11 Selected studies on manganese oxide reduction by hydrogen gas

Previous studies showed that many reduction processes of manganese oxides terminate at MnO. This behavior has the potential to be utilized as an oxygen storage compound in syngas production process [177]. Manganese oxide can be used as oxygen storage in methane–steam reforming processes. Thus, there are numbers of work focusing on manganese oxides reduction using methane or mixture of CO with H2. A comprehensive review of this topic can be found in Cheraghi et al. [178]. In manganese ore processing, pre-reduction process frequently involves carbon gas mixtures inside submerged arc furnace (SAF). Many studies reported that reduction proceed until MnO and then followed by manganese carbide formation [177, 179]

Rare Metal Oxide

An early study by Bolivar et al. [180] reported that metallic rhenium (Re) can be obtained from mixing Re2O7 with Pt or Pd or Re powders and heated in hydrogen atmosphere (under 1 atm hydrogen pressure) below 200 °C. Later, rare metal Re was demonstrated to be produced in a stream of hydrogen [6, 20] through two stages of reduction from its ammonium perrhenate (NH4ReO4). The first stage was carried out at 300–350 °C to form rhenium oxide and followed by reduction at 800 °C to form Re metal powder as shown in Eqs. 19, 20. Other routes such as CVD method were also introduced for preparing ultrafine rhenium powder [181]. There are also studies reported on utilizing non-thermal hydrogen plasma to reduce rare earth metal germanium [182] and rhenium [181]. Vandroux et al. patented a process of GeO2 reduction in a germanium substrate using hydrogen-containing gas in a process chamber at 550 °C. Meanwhile, Bai et al. investigated a preparation of ultrafine rhenium powder using ammonium perrhenate as starting precursor which vaporized into ReO4 and Re2O7. The rhenium oxides then reduced into rhenium powder in a CVD process under hydrogen atmosphere.

$${\text{NH}}_{{4}} {\text{ReO}}4 + {\text{H}}_{2} \left( {\text{g}} \right) = {\text{Re}}_{x} {\text{O}}_{y} + {\text{NH}}_{3} + {\text{H}}_{2} {\text{O}}\left( {\text{g}} \right),$$
$${\text{Re}}_{x} {\text{O}}_{y} + {\text{H}}_{2} \left( {\text{g}} \right) = {\text{Re}} + {\text{H}}_{2} {\text{O}}\left( {\text{g}} \right).$$

The usual method of producing elemental germanium (Ge) is by reducing a germanium dioxide in a tubular furnace at 650 °C in a hydrogen environment. The temperature needs to be kept under 700 °C to avoid the generation of volatile germanium monoxide. After complete reduction, Ge can be melted and slowly solidified in casting [6, 20].

Miscellaneous Oxides

Other non-ferrous oxides have also been experimentally reduced using hydrogen. SnO2 reduction by hydrogen is one of them, where Kim et al. [183] reported a full SnO2 reduction by hydrogen at 800 °C and H2 partial pressure of 101.3 kPa. This study also reported that a nucleation and growth model fits to describe the kinetics of SnO2 reduction. There are very few published papers found on hydrogen reduction of SnO2, however, the use of hydrocarbon-based reductant has been studied in more detail. SnO2 was reported to be able to reduce into metallic Sn via SnO by the syngas derived from methane [184] and feasible to be efficiently recovered from secondary resources using methane [185] and alkane [186]. Meanwhile, for other metal oxides, Newbery and Pring [135] reported to experimentally reduce a wide range of metal oxides (V2O5, Nb2O5, CeO2, U3O8, ZrO2, Y2O3, ThO2) under high-pressure condition, where most of the metal oxides were reduced into lower state oxidation number except for the refractory type of oxides, zirconia, thoria, and yttria. It is noteworthy that various types of ores or compounds other than oxides were also reported able to be decomposed into desired element in a hydrogen atmosphere or reduced into metal by utilizing hydrogen as the reducing agent assisted by other elements (Cl and S). For instance, platinum and palladium metal reduction process was carried out in hydrogen atmosphere, from non-oxide compounds (chlorides) of (NH4)2PtCl6 and Pd(NH3)2Cl2, respectively [6].

Use of Hydrogen Plasma for Metal Oxides Reduction

Table 12 shows that atomic hydrogen or hydrogen plasma have been used for many metal oxides reduction. Comprehensive explanation of plasma production and process can be found on Sabat and Murphy [17]. From Ellingham diagram in Fig. 1, one can see that that atomic hydrogen and hydrogen plasma are more powerful and reductant compared to molecular hydrogen. Hydrogen plasma can resolve the thermodynamic barrier where it theoretically can reduce unreducible oxides such as SiO2, CaO, and MgO. However, studies report that some of the oxides (i.e., Al2O3, TiO2) were only reduced into suboxides.

Table 12 Various metal oxides reduced by hydrogen plasma process using hydrogen gas mixture

There are two types of artificial hydrogen plasma generated in laboratory, namely, thermal (equilibrium plasma) and cold plasma (non-equilibrium plasma). Thermal or equilibrium plasma is produced by heating up the hydrogen to create high-rate collision between its molecules, atoms, ions, and the electrons at temperatures until all species reach equilibrium. The equilibrium condition happens locally, where the temperature of translational, vibrational, and rotational of electrons, ions, atoms, and molecules are equal to the equilibrium temperature of the chemical reaction. This local thermodynamic equilibrium (LTE) represents the condition of the central of the plasma torch. The energy source for plasma generation is typically produced through electrodes connected to DC or AC voltage, but can also be generated using electromagnetic field, microwave, or radiofrequency. A comprehensive review on the experimental details of thermal hydrogen plasma can be found in the work of Nakamura et al. [196], Kitamura et al. [193], and a review by Sabat and Murphy [17].

Meanwhile, the state of cold plasma is non-equilibrium, which happened due to the temperature difference between electron and heavy substances (atoms and molecules) at low-pressure condition. When the substances are heated, the electron’s temperature is greater than the temperature of the heavy substance and the electrons cannot transfer the energy as they have low collision rate caused by the low-pressure condition. The cold plasma can be generated in an atmospheric condition by inserting dielectric between two electrodes connected to AC voltage. The charge coming from the dielectric interrupts the discharge, in which the electrons have insufficient time to transfer heat to the heavy substance. Therefore, the typical cold plasma (i.e., microwave plasma) shows a strong LTE deviation due to higher electron temperature compared to atoms or molecules. Cold plasma is typically used in thin film deposition and semiconductor etching applications.


Hydrogen has been used commercially in W and Mo production and partially in Ni and Co productions. Previous studies of hydrogen reduction for non-ferrous metals production were found to be mainly focused on understanding kinetics and the reduction mechanism. There have been some debates and inconsistencies between studies on the reduction reaction sequence for a number of metal oxide systems. For example, wide range of kinetic results and activation energies for NiO reduction were reported due to inconsistency of experimental condition and techniques. Even so, these previous studies set a foundation of knowledge for future studies and potential improvement in an industrial practice.

From the perspective of metal recycling, there is great potential for using hydrogen for recovering many metals including Zn and Pb. However, further understanding of the kinetics and detailed mechanism of these metals recovery from secondary resources (i.e., EAF dust, Pb-rich slag) is required before technology can advance.

These previous studies on hydrogen reduction of metal oxides can also be useful for wider applications other than metals extraction. Hydrogen reduction is widely applied for preparing or treating of advanced materials (i.e., preparing nano-size metal powder for sensor application, doping treatment for semiconductor metal oxide, hydrogenation process for photocatalyst materials, create strong adhesion for microelectronic component, usage in plasma application). These studies can also elucidate how materials formed so that it is useful for controlling and designing the material’s properties.