The Use of Hydrogen as a Potential Reductant in the Chromite Smelting Industry

Abstract

The chromium (Cr) content of stainless steel originates from recycled scrap and/or ferrochrome (FeCr), which is mainly produced by the carbothermic reduction of chromite ore. Ever-increasing pressure on FeCr producers to curtail carbon emissions justifies migration from traditional FeCr production routes. The interaction between hydrogen and chromite only yields water, foregoing the generation of significant volumes of CO-rich off-gas during traditional smelting procedures. For this reason, the use of hydrogen as a chromite reductant is proposed. In addition to thermodynamic modelling, the influence of temperature, time, and particle size on the reduction of chromite by hydrogen was investigated. It was determined that, at the explored reduction parameters, the iron (Fe)-oxides presented in chromite could be metalized and subsequently removed by hot-acid leaching. The Cr-oxide constituency of chromite did not undergo appreciable metalization. However, the removal of Fe from the chromite spinel allowed the formation of eskolaite with the composition of (Cr_1.4Al_0.6)O₃ in the form of an exsolved phase, which may adversely affect the reducibility of chromite. The study includes the limitations of incorporating hydrogen as a reductant into existing FeCr production infrastructure and proposes possible approaches and considerations.

1. Introduction

Chromite ore is characterized by the unit formula (Mg, Fe²⁺)(Al, Cr, Fe³⁺)₂O₄ and serves as the only economically viable source of virgin Cr units [1,2]. The energy-intensive carbothermic smelting of chromite ore yields ferrochrome (FeCr)—a crude alloy of Cr and Fe [3,4,5]. Approximately 80 to 90% of produced FeCr is consumed by the stainless steel industry and is, therefore, considered equally important [4,6,7]. The significance of FeCr production is highlighted by the Cr constituency thereof, as the inclusion of Cr in stainless steel is the reason for the corrosion resistance thereof under standard ambient conditions [4].

Chromite smelting is typically performed in semi-closed or closed submerged arc furnaces (SAFs) and, to a lesser extent, closed direct current (DC) arc furnaces [6,8,9]. The energy consumption of a specific furnace is affected by factors such as material screening, chromite pre-heating, and pre-reduction [6]. For instance, the energy consumption of a semi-closed SAF operating on unscreened feed materials can be as high as 4500 kWh/t FeCr produced, whereas a closed SAF operating on hot-fed pre-reduced chromite can be as low as 2000 kWh/t FeCr produced [6]. In addition to this, produced FeCr contains a certain amount of Si, which originates from the unintended reduction of silica in the slag phase, which increases energy consumption [10,11].

During chromite smelting, various types of carbon sources are used as reductants, e.g., coke, coal, and charcoal. The Cr- and Fe-oxide constituencies of chromite are reduced to their metallic states or carbide form [3,12]. For all these reduction reactions, CO(g) is formed and present in significant quantities in the off-gas. Due to the highly toxic nature of CO, the CO-rich off-gas must be oxidized before its release into the atmosphere. For semi-closed SAFs, this oxidation occurs above the furnace bed concurrently with smelting operations, while closed SAFs and DC furnaces oxidize CO-rich off-gas on a furnace stack [13,14,15].

Due to the nature of semi-closed SAFs, relatively significant quantities of air enter the space above the furnace bed and off-gas duct, and these SAFs produce between 10,000 to 15,000 Nm³/t FeCr off-gas. Closed SAFs generate the lowest volumes of off-gas during smelting (i.e., 650 to 850 Nm³/t FeCr), which are cleaned in venturi scrubbers before being flared or employed as fuel elsewhere on-site. The volume and CO-content of off-gas during closed SAF operations are, however, dependent on the leakage air [13,16]. Nevertheless, significant quantities of CO₂ are produced during chromite smelting, regardless of the furnace type and smelting regime employed.

Increasing pressure on the FeCr industry to curb C-emissions calls for the development of alternative smelting processes. Such processes should achieve high levels (preferably near-complete) of chromite reduction, while reducing C-emissions (when compared with traditional smelting processes). A recent patent by Winters (2016) proposed a process to produce Cr-containing alloy without employing SAFs [17]. This process utilizes a vertical moving reactor bed with external burners that consume pre-heated reformed natural gas (CO and hydrogen) to supply heat to the reactor, while concurrently injecting reformed natural gas into the bottom of the reactor. By doing so, the chromite is reduced in a solid state. Thereafter, the Cr-containing alloy is recovered through physical separation based on differences in density and magnetic properties of the alloy. This smelting process consumes approximately a third of the electrical energy required during conventional chromite smelting. Water and CO₂ are generated during this smelting process, of which the CO₂ is captured, compressed, and supplied to various industries. Significant quantities of naturally occurring natural gas are, however, a pre-requisite for this smelting method [17].

A patent by Pheiffer and Cookson (2015) suggested that SiC can be employed as a reductant during the FeCr production process. SiC composites can be incorporated during the generation of chromite agglomerates (e.g., briquettes or pellets) destined for oxidative sintering or pre-reductive pellet curing. The proposed SiC composite comprises mainly SiC, C, and ash (various fractions of CaO, MgO, SiO₂, Al₂O₃, FeO). The reaction of SiC with chromite yielded SiO₂ and CO as by-products; SiO₂ suggests that CO formation was partially reduced [18]. However, Pheiffer and Cookson (2015) did not perform a techno-economical study, and due to the cost of SiC, the running operational cost of a furnace operating on SiC-based reductants will likely be higher than traditional smelting operations [5].

A solid-state chromite reduction process was investigated by Yu and Paktunc (2018) where segregation reduction was implemented. During the process, graphite, petroleum coke, and charcoal were used as reductants, and calcium chloride (CaCl₂) as the segregation catalyst. This flux-assisted process achieved high degrees of Cr metalization at 1300 °C after 2 h of reduction in a rotary kiln or rotating hearth furnace. The CaCl₂ initiated the chlorination system that enabled the reducible Cr and Fe species from the chromite particles to segregate before the reduction and metalization by carbon. Subsequently, individual alloy particles were formed, which made the separation during ferrochrome recovery easier [19]. The CaCl₂ also provided a liquid media for the incongruent dissolution of chromite and migration of the reducible ions (e.g., Fe²⁺, Fe³⁺, Cr³⁺). It was also possible to recover CaCl₂ in the reduced product through water-leaching; the catalyst could then be re-used and this would possibly save on material cost [20]. However, CaCl₂ could react with the siliceous gangue (e.g., clinochlore), causing the formation of Ca-bearing chromite, wadalite, and gaseous chlorides (MgCl₂, FeCl₂, and/or HCl). In the CaCl₂-assisted segregation reduction process, the ferrochrome alloy particles must be separated from the unwanted gangue present in the reduced products. Separation could likely succeed through conventional gravity or magnetic concentration techniques; however, further process development was proposed [19,21].

A patent by Sokhanvaran and Paktunc (2019) provided a method where the addition of cryolite (Na₃AlF₆) in chromite direct reduction was advantageous. The cryolite dissolved several oxides from the chromite spinel, which diminished slag growth. The forming molten cryolite layer enabled the migration of Cr and Fe ions. Impure mixtures containing cryolite (e.g., by-product of aluminum smelting processes and bath material produced as waste) were used in the chromite reduction. Using cryolite as an additive in chromite reduction involves several embodiments as well as many options of achieving these embodiments, enhancing the necessity for further research [22]. It is, however, noted that the use of F-containing reductants (as well as Cl-containing) are considered environmentally hazardous.

It has also been shown that the use of hydrocarbon gas for chromite reduction achieved higher degrees of reduction at lower temperatures when compared with traditional carbothermic smelting procedures [23].

To alleviate some C-dependence of chromite smelting, and to avoid the use of Cl- and F-containing compounds, the use of hydrogen as the reductant is proposed. Theoretically, the complete reduction of pure chromite by hydrogen is given as follows [24]:

Cr₂FeO₄ + 4 H₂(g) = 2 Cr + Fe + 4 H₂O(g)

(R1)

It is noted that Reaction (1) requires temperatures greatly exceeding the typical temperatures associated with SAF smelting. In addition, the reduction of naturally occurring chromite does, however, proceed differently as the chromite spinel contains stabilizing Mg²⁺ and Al³⁺ [25]. Nevertheless, Reaction (1) shows that the only by-product when employing hydrogen as a reductant is water. Therefore, considering the occlusion of C-based by-products, the use of hydrogen to reduce chromite was explored in this study. More importantly, the reduction of Cr- and Fe-oxides was of importance, and the metalization mechanisms thereof were elucidated. It is proposed that by employing hydrogen to metalized chromite, or at least partially, the electrical and C consumption of a chromite smelter can be reduced.

2. Materials and Methods

2.1. Materials

Kleynhans et al. (2016) stated that in 2012 approximately 28% of globally smelted chromite originated from South Africa. Thus, a significant fraction of globally produced FeCr, which is estimated to be approximately 32%, originated from South African chromite [26,27]. Considering this, the results presented in this study are of international relevance.

Metallurgical-grade chromite ore was obtained from a large South African FeCr producer and was used as the case study ore. A homogeneous and representative sample collection campaign was undertaken at the South African FeCr smelter and sample selection for experimentation was applied here as described by Du Preez et al. (2019) [2]. A thorough characterization of this case study ore was conducted by Glastonbury et al. (2015) [28], therefore characterization was not repeated here.

Table 1. Characterization of the case study chromite ore in terms of the chemical composition (ICP–OES, wt.%) [28], and mineralogical composition (XRD, wt.%).

ICP-OES		XRD
Compound	Metgrade	Compound	Metgrade
Cr2O3	44.19	Chromite	93.4
FeO	24.68	Enstatite	6.6
SiO2	2.96
Al2O3	14.71
MgO	10.31
CaO	0.16
P	0.001
Cr/Fe	1.58

summarizes the chemical and crystalline characteristics of chromite ore used here. Bulk chemical characteristics were determined by inductively coupled plasma–optical emission spectroscopy (ICP–OES) and bulk mineralogical composition by X-ray diffraction (XRD) analysis.

Of particular importance here were the Cr- and Fe-contents, and the Cr/Fe ratio. It is worth noting that South African chromite has relatively low Cr-contents. For instance, Ringdalen and Olsen (1998) indicated that Turkish, Kazak, and Brazilian chromite ore contains 0.70–0.77 mol Cr in the octahedral sites, whereas South African ore contains between 0.59–0.62 mol Cr. As a result, octahedral sites are occupied by Al³⁺ and a relatively low amount of Fe³⁺ (around 0.06 mol) [29]. As stated earlier, the presence of Mg²⁺ (in tetrahedral sites) and Al³⁺ adversely affect the reducibility of chromite.

Hydrogen, which was the investigated reductant in this study, can be produced from numerous processes, e.g., thermochemical [30,31], photocatalytic [32,33], photochemical [34,35], photo-electrochemical [36,37], metal, and metal hydride hydrolysis [38,39,40,41,42,43,44], electrochemical [45,46,47,48,49,50], ammonia and formic acid decomposition [51,52,53], and various biological processes [54,55,56]. The hydrogen used throughout this study as the reductant was generated on-site at Hydrogen South Africa Infrastructure, South Africa, by proton exchange membrane water electrolysis (PEMWE). A PEMWE stack was powered by photovoltaic energy and the generated hydrogen was stored as a pressurized gas. Therefore, the hydrogen used here is defined as green hydrogen as it is produced from renewable energy and a renewable source. Nitrogen (99% purity) was obtained from Afrox, South Africa.

2.2. Sizing of Chromite

Size partitioning of as-received chromite ore was performed by screening it into the following size fractions, i.e., <106, 106 to 250, 250 to 500, and 500 to 1000 µm. Chromite screening was performed using a Haver EML Digital Plus sieve shaker and stainless- steel Haver and Boecker sieves.

Particle size distribution of the screened chromite fractions was performed by laser diffraction particle sizing using a Malvern Mastersizer 3000. Samples were ultra-sonicated before measurements and continuously agitated by mechanical stirring at 2400 rpm during analyses. Laser obscuration was maintained between 10 to 20%. The d₉₀, d₅₀, and d₁₀ values of as-received and size fractioned are presented in

Table 2. D₉₀, d_50, and d₁₀ (µm) of as-received and size fractioned chromite.

Size Equivalent	As-Received	Sieve Sizes (µm)
		<106	106 to 250	250 to 500	500 to 1000
d90	415.0	140.2	282.1	464.2	924.1
d50	182.3	87.4	186.3	347.4	667.7
d10	71.7	51.8	120.7	260.0	487.6

.

2.3. Reduction Procedure

Reduction procedures were performed in a Carbolite vertical tube furnace; the furnace had a maximum operating and maintainable temperature of 1200 and 1100 °C, respectively. Ceramic heat shields were placed within the tube at both ends to increase the stability of the heated zone. Stainless steel caps were fitted to seal the tube ends and acted as a gas inlet and outlet. The thermocouple used to measure temperature was located at the center of the heating zone, immediately adjacent to the tube. The furnace was heated by a series of electrodes which were located adjacent to the tube.

The reduction was performed by placing 50 g of chromite in a crucible in the center of the furnace hot zone. The samples were then heated from room temperature at the maximum allowable heating rate for the specifically fitted tube, i.e., 5 °C/min, until the predetermined temperature was reached. The tube was continuously purged with N₂ at a flow rate of 75 mL/min. It was of importance to ensure no oxygen was present within the furnace seeing that hydrogen has a self-ignition temperature of approximately 572–585 °C and an explosive range of 4–75 vol.% in air at standard ambient conditions [57,58,59]. A 1 L/min hydrogen flow, in conjuration with the N₂ flow, was introduced at 600 °C to obtain an approximate 93 vol.% hydrogen atmosphere within the furnace tube. This atmosphere was maintained until a pre-determined reduction temperature was reached/maintained. Thereafter, the hydrogen flow was discontinued, and the reduced samples were allowed to cool to room temperature in an N₂ atmosphere.

2.4. Analytical Techniques

Temperature programmed reduction (TPR) was performed using an automated chemisorption analyzer AutoChem II 2920 (Micrometrics, Norcross, GA, USA). Approximately 0.25 g of sample was placed in a quartz U-tube fitted with quartz wool which was fixed to the bottom of the tube. TPR analyses were performed using 10 vol.% H₂/Ar at a flow rate of 50 mL/min. Before analysis, the sample-containing tube was purged at room temperature for 30 min using the latter mentioned gas. Thereafter, the furnace was heated at a rate of 10 °C/min to 1000 °C while continuously recording hydrogen consumption with a thermal conductivity detector. This instrument had a maximum maintainable temperature of 1000 °C.

Scanning electron microscopy (SEM) equipped with an energy-dispersive x-ray spectrometer (EDX) was used to perform surface and subsurface characterization of unreduced/reduced chromite particles in backscattered electron mode. An FEI Quanta 250 FEG SEM incorporating an Oxford X-map EDX system operating at 15 kV and a working distance of 10 mm was used. For subsurface analysis, treated chromite particles were set in resin, cross-sectioned, and polished using an SS20 Spectrum System Grinder polisher before being mounted on Al stubs using carbon adhesive tape. All samples considered for SEM analysis were coated with carbon using an Emscope TB 500 carbon coater.

Crystalline phase analysis of bulk samples was performed by X-ray diffraction (XRD) utilizing a Rigaku D/MAX 2500 rotating-anode powder diffractometer with Cu Kα radiation at 50 kV, 260 mA, a step-scan of 0.02°, and a scan rate at 1/min in 2 h from 5 to 70°. Phase identification was performed using JADE v.3.9 with the ICDD and ICSD diffraction databases.

Further analysis was undertaken by mounting cross-sectioned polished pellets on glass slides and carbon-coating them before characterization by electron probe microanalysis (EPMA). EPMA analyses were performed to determine the oxidation states of the transformation products. Analysis was conducted using a JEOL JXA-8900 electron microprobe fitted with five wavelength dispersive spectrometers at an accelerating voltage of 20 kV and a beam current of 26 nA.

The extent of Cr and Fe metalization was determined after the reduction of chromite by solubilizing Cr and Fe through hot-acid leaching. The treated chromite was leached using a 50 vol.% H₂SO₄ solution at 90 °C for 60 min. Thereafter, the solubilized Cr and Fe contents were determined using ICP-OES.

ICP-OES was performed by using a 5110 ICP-OES Instrument coupled with a VistaChip II CCD detector. The instrument was calibrated using Ultraspec aqueous certified single-element reference standards (Fe, Cr) obtained from De Bruyn Spectroscopic Solutions (South Africa).

Considering that during size fractioning, gangue mineral (i.e., non-chromite, siliceous particles) particles could collect within a certain size range, the total Cr and Fe content of the respective size fraction were determined, and the extent of Cr and Fe metalization calculated as a function of the total Cr and Fe contents present in each size fraction.

A Siebtechnik pulverizer was used to mill each size fraction. All parts of the pulverizer that made contact with the chromite during milling were made of SiC to prevent Fe contamination during the milling procedure. A sample was milled for 2 min to obtain a particle-size distribution of which 90% of the particles were below 75 µm. Total Cr and Fe were determined for each size fraction by high-temperature alkaline fusion using Na₂O₂. Alkaline fusion was performed on all size fractions by reacting 0.2 g of the milled chromite with 2 g Na₂O₂ and 0.5 g Na₂CO₃, respectively, in a zirconium crucible. The fused material was then solubilized using a 20% mixture of 1:1 water:HNO₃ and the Cr and Fe contents were determined by ICP-OES.

2.5. Expressing the Extent of Metalization

Two terms are generally used to express the extent of Cr and Fe reduction; either in terms of lowering the oxidation state (e.g., Fe₂O₃ (Fe³⁺) reduced to FeO (Fe²⁺)) or metalization (e.g., Cr₂O₃ reduced to Cr⁰) [12]. Barnes et al. (1983) defined Reduction (%R) and Metalization (%M) as the follows:

The removal of oxygen is associated with %R and is defined in terms of the oxygen mass loss expressed in the equation below:

$$\%\mathrm{R}=\frac{\mathrm{Mass} \mathrm{of} \mathrm{oxygen} \mathrm{removed}}{\mathrm{Original} \mathrm{removable} \mathrm{oxygen}}\times 100$$

(1)

in a case where solid carbon is used as a reductant, CO will form as a product in the reduction process of chromite and for this reason, the reduction can also be described as [12]:

$$\%\mathrm{R}=\frac{\mathrm{Mass} \mathrm{of} \mathrm{CO} \mathrm{evolved}}{28/16 \mathrm{x} \mathrm{original} \mathrm{removable} \mathrm{oxygen}}\times 100$$

(2)

$$\%\mathrm{M}=\frac{{\mathrm{Cr}}^0+{\mathrm{Fe}}^0}{{\mathrm{Cr}}_{tot}+{\mathrm{Fe}}_{tot}}\times 100$$

(3)

where Cr⁰ and Fe⁰ are the total amounts of metalized Cr and Fe, and Cr_tot and Fe_tot are the total Cr and Fe contents of the original chromite. Furthermore, Barnes et al. (1983) proposed that due to the possible concurrence of the reduction of Fe₂O₃ to FeO and metalization of FeO to Fe and Cr₂O₃ to Cr, there is no linear relationship between chromite’s reduction and its metalization [12].

There is, however, a close resemblance between the complete removal of oxygen and complete metalization, as 100% reduction corresponds to 100% metalization [12]. The factors explained by Barnes et al. (1983) have contributed to the statement given by Algie and Finn (1984), which explains the three stages of chromite reduction: first, the Fe³⁺ is reduced to Fe^2+, followed by its metalization; the metalization of Cr³⁺ is only initiated after approximately 50% of the Fe²⁺ is metalized. After the completion of Fe reduction, 60% of the Cr reduction could be achieved [60,61].

However, the reduction of chromite by C is relatively well understood and more descriptive expressions have been formulated. In this study, the %Fe and Cr metalization was of importance and expressed individually according to the following equation:

$$\%\mathrm{Fe} \mathrm{or} \mathrm{Cr} \mathrm{metalization}=\frac{\mathrm{Solubilized} \mathrm{Fe} \mathrm{or} \mathrm{Cr}}{\mathrm{Total} \mathrm{Fe} \mathrm{or} \mathrm{Cr} \mathrm{content}}\times 100$$

(4)

where the metalized Fe and Cr fractions of the treated chromite (defined as the fractions of leached Cr and Fe, as determined by ICP–OES analysis) are represented by the solubilized Fe or Cr value, which is divided by the total amount of Fe or Cr present in the untreated chromite ore. The metalization of Fe and Cr are presented individually.

3. Results and Discussion

3.1. Thermodynamic Considerations

When kinetic data are not available to characterize a process at elevated temperature, an alternative is to understand the thermochemical aspects that govern the behavior of the chemical species that are in reaction. Thermodynamic considerations are, however, required to understand what can be achieved, as well as the driving force for reduction to occur. The predominant area phase-stability diagrams are a tool that allows us to graphically represent the conditions of thermochemical equilibrium of compounds in reaction. The diagrams presented in this section were generated using the methodology proposed by Ramos-Hernandez et al. [62], using recently updated thermochemical data that were selected from various sources, both old and recent [63,64,65,66,67].

To graphically evaluate the thermochemical conditions of the Cr and Fe reduction processes from chromite (Cr₂FeO₄) using C or hydrogen as a reducing agent, it is necessary to identify the reduction regions of interest (where the Cr and Fe species predominate) for each of the processes in a (Cr, Fe)-C-H-O diagram, such as Figure 1, where the chemical activity of Cr and Fe are represented on the X and Y axes, respectively.

As references, the ideal point for the reduction of Cr and Fe is presented in red, considering that the chemical activity of each reduced metallic element takes the value of the unit (log aM = 1). Furthermore, the equilibrium lines in black correspond to the Cr₂FeO₄ reduction reactions in an environment containing O_2, where four regions of stability are defined where Cr₂FeO₄ predominates, and the following pairs of species: Fe-Cr, FeO-Cr, and Fe-Cr₂O₃ where the following decomposition reactions have been considered:

2 FeO (s) ⇄ 2 Fe(s) + O₂ log K (1000 °C) = 15.5563

(R2)

2 Cr₂FeO₄ (s) ⇄ 2 Cr₂O₃(s) + 2 Fe(s) + O₂ “ = 18.9956

(R3)

Cr₂FeO₄ (s) ⇄ 2 Cr(s) + Fe(s) + 2 O₂ “ = 42.6068

(R4)

Cr₂FeO₄ (s) ⇄ 2 Cr(s) + Fe(s) + 2 O₂ “ = 66.2180

(R5)

2 Cr₂FeO₄ (s) ⇄ 2 FeO(s) + 4 Cr(s) + 3 O₂ “ = 69.6572

(R6)

It is worth mentioning that there is a decomposition Reaction (7) that cannot be drawn on the diagram because it is independent of the variables being evaluated; however, its thermochemical viability can be evaluated as indicated below:

Cr₂FeO₄ (s) ⇄ 2 Cr₂O₃(s) + 2 FeO(s) log K (1000 °C) = 14.4218

(R7)

Each chemical reaction presents its corresponding value of the equilibrium constant (log K) at 1000 °C, which is related to the value of the standard Gibbs free energy as ∆G° = −RT log K. Each set of reactions was written considering this value in descending order to carry out a numerical evaluation of the thermochemical conditions of each one of them since the thermochemical viability of a reaction can be evaluated numerically considering it inversely proportional to the absolute value of log K; however, when the log K value is zero, it is established that the reaction is in thermochemical equilibrium. Also, a positive value for log K indicates the thermochemical viability of the forward reaction (as written), while a negative value implies viability for the reverse reaction.

Based on the criteria described in the previous paragraph, all reduction reactions (Reactions (2)–(6)) are not viable, considering the negative value of log K and the magnitude of its absolute values; this is why these reactions were considered an extreme case of comparison. However, relatively, it can be identified that the reduction of Cr and Fe from Cr₂FeO₄ (Reaction (4)) is more viable than the reduction of Cr from Cr₂O₃ (Reaction (5)) by 23.6 orders of magnitude. Although Reactions (4) and (5) have lower viabilities than Reactions (2) and (3), the reduction of Fe from FeO (Reaction (2)) is relatively more viable than from Cr₂FeO₄ (Reaction (3)) by 3.5 orders of magnitude. Rection 6 is the reaction with the lowest thermochemical viability and its interpretation in the case study of this work is meaningless; however, it is imperative to define the area of stability for Cr₂FeO₄ on the phase stability diagram. Reaction (7) is practically more viable, even more so than Reaction (2); this is the reason why the reduction of Fe and Cr is typically evaluated from Reactions (2) and (4).

The equilibrium lines in blue correspond to reduction reactions using C as a reducing agent and considering only CO as a by-product (normally employed in carbothermic processes):

FeO(s) + C(s) ⇄ Fe(s) + CO log K (1000 °C) = 1.4196

(R8)

Cr₂FeO₄ (s) + C(s) ⇄ Cr₂O₃(s) + Fe(s) + CO “ = 0.3000

(R9)

Cr₂O₃(s) + 3 C(s) ⇄ 3 CO + 2 Cr(s) “ = 5.5156

(R10)

Cr₂FeO₄ (s) + 4 C(s) ⇄ 2 Cr(s) + Fe(s) + 4 CO “ = 5.8157

(R11)

Cr₂FeO₄ (s) + 3 C(s) ⇄ 2 Cr(s) + FeO(s) + 3 CO “ = 7.2353

(R12)

It is well understood that during the carbothermic reduction of chromite, Fe-oxide reduction initiates before the Cr₂O₃ constituency [61,68]. Considering this, the reaction kinetics of pure Fe-oxides and Cr₂O₃ and with carbon and hydrogen was considered separately to gain insight into their expected reactivity. The carbothermic reduction of Fe- and Cr-oxides of chromite proceeds according to the following reactions (for naturally occurring chromite, both Fe³⁺ and Fe²⁺ are present) [12]:

In the first stages of reduction, Fe₂O₃ (Fe³⁺) is reduced to FeO (Fe²⁺) without any metalization,

Fe₂O₃ + C = 2 FeO + CO

(R13)

Subsequently, metalization occurs as FeO (Fe²⁺) is reduced to Fe⁰ (Reaction (8)) and lastly, Cr₂O₃ (Cr³⁺) is metalization to Cr⁰ (Reaction (10)). Considering Reactions (3) and (4), the reduction and metalization of Fe-oxides should occur at 552 and 694 °C, respectively. Considering Reaction (10), the reduction of Cr₂O₃ proceeds at 1263 °C. It is worth noting that during Reactions (8)–(13), hot CO gas is continuously generated which permeates from the smelting zone upwards through the furnace bed. The hot CO gas acts as a reductant, partially reducing chromite [69]. The effect of CO(g) partial pressure was, however, not considered here.

The reduction reactions of Fe- and Cr-oxides by hydrogen, in terms of oxidation state changes, will likely follow similar routes as Reactions (8)–(13). The equilibrium lines in green color correspond to reduction reactions and considers H₂O as a by-product:

FeO(s) + H₂ ⇄ Fe(s) + H₂O log K (1000 °C) = 0.5047

(R14)

Cr₂FeO₄ (s) + H₂⇄ Cr₂O₃(s) + Fe(s) + H₂O  “ = 2.2243

(R15)

Cr₂O₃(s) + 3 H₂ ⇄ 2 Cr(s) + 3 H₂O “ = 11.2885

(R16)

Cr₂FeO₄(s) + 3 H₂⇄FeO(s) + 2 Cr(s) + 3 H₂O “ = 13.0081

(R17)

Cr₂FeO₄(s) + 4 H₂ ⇄ Fe(s) + 2 Cr(s) + 4 H₂O “ = 13.5128

(R18)

The presence of in-situ formed H₂O affects Reactions (14)–(18), and it has to be taken into account that these reactions proceed as a function of the partial pressure of both H₂ and H₂O [70,71].

In the diagram shown in Figure 1, the reduction region of the processes that use C and hydrogen as reductants tends to be closer to the ideal reduction point, than to the reduction region of the process in an environment rich in O₂; this indicates that the use of hydrogen as a reducing agent is a thermodynamically good alternative. Furthermore, the main advantage of using hydrogen as a reducing agent is that the main by-product of the process is H₂O, which means that it is a green alternative that contrasts with the greenhouse gases generated in the process that uses C as a reducing agent.

Figure 2 shows a full predominant area phase-stability diagram for the (Cr, Fe)-H₂-H₂O system, where the Y-axis represents the partial pressure of hydrogen_, and the X-axis represents the partial pressure of H₂O. Considering the chemical activity of each of the condensed species of interest (which can be formed from the elements Cr, Fe, H, and O) as a unit, in such a way that the equilibrium lines that establish the stability region of each species correspond to the reduction Reactions (6) and (14) and another listed below:

2 Fe(OH)₃ (s) + H₂ ⇄ 2 Fe(OH)₂ (s) + 2 H₂O log K (1000 °C) = 9.6678

(R19)

2 CrH (s) ⇄ 2 Cr (s) + H₂ “ = 7.8491

(R20)

2 CrH (s) + 4 H₂O ⇄ 2 Cr(OH)₂ (s) + 3 H₂ “ = 5.9153

(R21)

2 Cr(OH)₃ (s) + H₂ ⇄ 2 Cr(OH)₂ (s) + 2 H₂O “ = 5.9027

(R22)

3 Fe₂O₃ (s) + H₂ ⇄ 2 Fe₃O₄ (s) + H₂O “ = 4.8997

(R23)

Fe(OH)₂ (s) + H₂ ⇄ Fe (s) + 2 H₂O “ = 4.8199

(R24)

Fe₃O₄ (s) + H₂ ⇄ 3 FeO(s) + H₂O “ = 1.4717

(R25)

FeO (s) + H₂O ⇄ Fe(OH)₂ (s) “ = −5.3245

(R26)

Cr₂O₃ (s) + H₂O + H₂ ⇄ 2 Cr(OH)₂ (s) “ = −13.2223

(R27)

Fe₃O₄ (s) + 2 H₂O + H₂ ⇄ 3 Fe(OH)₂ (s) “ = −14.5019

(R28)

Fe₂O₃ (s) + 3 H₂O ⇄ 2 Fe(OH)₃ (s) “ = −17.7025

(R29)

Cr₂O₃ (s) + 3 H₂O ⇄ 2 Cr(OH)₃ (s) “ = −19.1249

(R30)

Cr₂O₃ (s) + 4 H₂ ⇄ 2 CrH (s) + 3 H₂O “ = −19.1376

(R31)

2 Fe₃O₄ (s) + 10 H₂O ⇄ 6 Fe(OH)₃ (s) + H₂ “ = −58.0072

(R32)

The equilibrium lines in blue color correspond to the reactions involving Cr, while the equilibrium lines in red involve Fe. Also, the diagram shows a black dotted horizontal line that corresponds to the value of log pH₂ = 0 (that is, pH₂ = 1 atm), where two points are also located from the conditions of log pH₂O of equilibrium for the reduction reactions of Fe (reaction H, log pH₂O ≈ −0.5) and Cr (reaction K, log pH₂O ≈ −4). The above indicates that to exclusively reduce Fe and Cr in the system, there must be a partial pressure of H₂O less than or equal to 0.0001 atm; when this partial pressure is between 0.0001 and 0.3162 atm, the Fe and Cr₂O₃ species will predominate and if the partial pressure it is even greater, species such as Fe₃O₄, Fe₂O₃, Fe(OH)₃, Cr(OH)₂ will also predominate. Słowiński and Smoliński (2016) showed that the introduction of Fe to an H₂O-rich atmosphere will result in its oxidation to Fe_0.947O (shown as FeO here) [70]. It can be seen from Figure 2 that Fe is stable in a relatively wide range of H₂/H₂O partial pressure ratios at 1000 °C, with decreasing stability as the partial pressure of H₂O increases. Cr is stable in a relatively small area, which corresponds to a high H₂/H₂O ratio (dry, H₂-rich atmosphere). Chu and Rahmel stated that the reduction of Cr₂O₃ is only possible in oxygen-free and dry hydrogen environments, if the in-situ generated H₂O is continuously removed from reaction sites [71].

Most of the reactions with a log K value greater than zero are found in the region where the partial pressure of hydrogen is greater than 1 atm, while the reduction reactions with a log K value less than zero are found in the region of partial pressure less than 1 atm, which corresponds to the experimental conditions of this study.

It is important to note that the oxides considered in this study form part of the chromite spinel, i.e., [(Mg²⁺, Fe²⁺)(Al³⁺, Cr³⁺, Fe³⁺)₂O₄] [1,72]. Chromite is considered isomorphic, with Mg²⁺ and/or Fe²⁺ occupying tetrahedral sites and Al³⁺, Cr³⁺, and/or Fe³⁺ occupying octahedral sites at standard ambient conditions [73]. The presence of Mg²⁺ and Al³⁺ greatly decreases the reducibility of the chromite because these cations form highly stable oxides. It is noted that a magnesium-containing chromite spinel, i.e., (Fe²⁺, Mg²⁺)Cr₂O₄, has a greater resistance to reduction when compared with an Al-containing spinel, i.e., Fe(Cr³⁺, Al³⁺)₂O₄ [23]. Nevertheless, although the reduction reactions (Reactions (2)–(32)) are for pure Cr₂FeO₄, Fe- and Cr-oxides, the conclusions reached from the phase stability diagrams presented are consistent with the results of this work.

3.2. Reactivity of Chromite with Hydrogen

Though thermodynamic explanations of a reaction are a useful tool, they ignore kinetic effects. To determine the actual interaction between the case-study chromite and hydrogen, TPR analysis was performed on as-received chromite (Figure 3). The initiation of Fe-oxidation state reduction is included in Figure 3.

Figure 3 presents the interaction between hydrogen and chromite as a function of temperature. When considering Figure 3, the following deductions may be made: (i) the reduction of Fe₂O₃ to FeO (Fe³⁺ to Fe²⁺) initiated at approximately 500 °C; (ii) the peak observed at approximately 846 °C is likely the reduction of FeO to Fe (Fe²⁺ to Fe⁰); (iii) no magnetite (expressible as Fe(Fe³⁺)₂O₄ or FeOFe₂O₃) to FeO reduction peak is observed, likely indicating that the intermediate step proceeds rapidly during the reduction of Fe₂O₃ → FeO; (iv) the reaction rate peaked after reaching 1000 °C and proceeded at a slower rate thereafter, suggesting hydrogen diffusion was likely limited by the formation of a surficial metallic layer and restricted hydrogen–chromite reaction surface, or that the reduction reaction rate decreased as a function of decreasing FeO concentration, or that mass transfer was limited due to a build-up of H₂O in the atmosphere surrounding the crucible; and (v) the absence of addition peaks suggests that Cr₂O₃ did not undergo observable reduction, which was expected considering thermodynamic calculations.

During thermal treatment of chromite, the constituents present in the chromite spinel migrate as a function of the atmosphere. For instance, Fe₂O₃-rich sesquioxide phases form at the rim of chromite particles, as well as the cleavage planes during calcination in air [2], whereas metalized Fe forms at the rims during carbon-based solid-state reduction [74]. To evaluate constituency mobility of chromite during reduction using hydrogen, the surface of as-received, reduced, and acid-leached chromite particles were explored by SEM (Figure 4) and EDX (

Table 3. Microanalyses of chromite particles by SEM–EDX (wt.%). A minimum of eight randomly selected particles were considered and between three to five areas were analyzed with the averages and standard deviations indicated. The Cr/Fe ratio is also indicated.

Sample Surface	Detected Element (wt.%)						Cr/Fe Ratio
	O	Mg	Al	Si	Cr	Fe
As-received	36.68 ± 1.41	6.29 ± 0.30	8.45 ± 0.23	0.29 ± 0.01	30.85 ± 1.49	17.44 ± 0.52	1.77
Treated	26.32 ± 3.69	2.12 ± 0.49	13.69 ± 0.48	0.42 ± 0.04	13.94 ± 1.35	43.53 ± 5.4	0.32
Leached	28.98 ± 0.68	2.43 ± 0.23	14.93 ± 0.63	0.66 ± 0.18	49.98 ± 0.74	3.01 ± 0.37	16.60

). Reduced particles presented here were exposed to hydrogen at 1100 °C for 30 min. The acid-leached sample was imaged at a higher magnification than the as-received and treated samples to better illustrate the surface morphology.

Considering Figure 4, it is clear that morphological changes occurred during the reduction and acid-leaching procedures of chromite. Table 3 indicates that the Cr/Fe ratio of as-received (Figure 4a) and treated (Figure 4b) chromite particles are 1.77 and 0.32, respectively, suggesting that the surface was enriched with Fe after reduction. The acid-leached chromite particles (Figure 4c) have a Cr/Fe ratio of 16.60. This increase in Cr/Fe ratio suggests that the Fe-enriched surficial layer was removed during the acid-leaching procedure. Hazar-Yoruc (2007) found that cracks, dislocations, and cavities/spaces of the spinal are ideal locations for the nucleation of Fe [61]. Therefore, the cavities observed on the surface of acid-leached particles (Figure 4c) may be ascribed to the dissolution of Fe. Table 3 further suggests that only Fe was mobile and that Cr likely remained immobile. To gain further insight into the mobility of chromite constituents, treated particles were cross-sectioned and investigated by SEM (Figure 5) and SEM–EDX (

Table 4. SEM–EDX (wt.%) of the center (Area 1), transition phase (Area 2), and rim (Area 3) of a representative particle thermally treated at 1100 °C for 30 min, as well as the respective Cr/Fe, Mg/Fe, and Al/Fe ratio for each of the areas. A minimum of eight randomly selected particles were considered for EDX analysis and between three to five areas were analyzed with the averages and standard deviations indicated.

Cross-Sectioned Particle	Detected Element (wt.%)						Cr/Fe Ratio	Mg/Fe Ratio	Al/Fe Ratio
	O	Mg	Al	Si	Cr	Fe
Area 1	31.64 ± 0.39	6.19 ± 0.07	8.24 ± 0.04	0.47 ± 0.06	33.41 ± 0.04	20.07 ± 0.31	1.66	0.31	0.41
Area 2	32.42 ± 0.30	8.86 ± 0.08	8.79 ± 0.13	0.38 ± 0.02	40.29 ± 0.10	9.27 ± 0.33	4.35	0.96	0.95
Area 3	3.44 ± 0.13	0.22 ± 0.30	1.53 ± 0.04	0	4.05 ± 0.51	90.38 ± 0.35	0.04	0.002	0.017

).

Area 1 (center) represents the chromite particle core, Area 2 (transition phase) represents the Fe deprived area, and Area 3 (rim) represents the surficial Fe-enriched layer. The detected O and Cr in Area 3 may be ascribed to the nearby Area 2.

Figure 5 shows that the metalization occurred at the rim (Area 3) of chromite particles. EDX analysis indicates the rim with the composition of 90.4 wt.% Fe and 4.1 wt.% Cr, which corresponds to a Cr/Fe ratio of 0.04. The O content of 3.4 wt.% indicates that some of the Fe, Cr, Al and Mg concentrations may not be elemental, which may partially be ascribed to the presence of unreduced oxides. However, EDX is a semi-quantitative analytical technique, with O determined by difference. Nevertheless, Area 2 had a relatively low Fe content of 9.3 wt.%, which is ascribed to the migration of Fe towards the particle surface. The Cr/Fe ratio for Area 2 was 4.35, which is significantly higher than the 1.66 Cr/Fe ratio of the particle core (Area 1). A non-related study utilizing the same ore had a similar Cr/Fe ratio (i.e., 1.65) at the particle core [2], which suggests that the total Fe content at the particle center was unaffected.

The migration of Fe (as Fe²⁺ and Fe³⁺) towards the rim of chromite particles creates vacancies in the octahedral and tetrahedral sites of the chromite spinel [58]. The removal of Fe²⁺ from tetrahedral sites increased the Mg/Fe ratio from 0.31 for Area 1 to 0.96 for Area 2, whereas the removal of Fe³⁺ from octahedral sites increased the Al/Fe ratio from 0.45 for Area 1 to 0.95 for Area 2. Furthermore, the low Mg/Fe and Al/Fe ratios of Area 3 indicate that Mg and Al did not undergo any migration.

As opposed to the significant decrease in Fe observed in Areas 1 and 2 (Table 4), the Cr content remained relatively unchanged. Additionally, the observed increase in the Cr/Fe ratio of 1.66 to 4.35 for Areas 1 and 2 supports the migration of Fe towards the particle rim. This is made evident when considering the Cr/Fe ratio of 0.04 of Area 3.

Soykan et al. (1991) proposed the reduction mechanism of chromite, as well as the subsequent formation of Cr- and Fe-carbides during carbothermic smelting, where carbon is used as the reductant. The reduction mechanism proceeds according to the following four steps [68]: (i) the surface of the chromite particle serves as the location for the chemical reduction reaction where Fe²⁺ is metalized to Fe; (ii) Cr²⁺ ions migrate from the rim to the transition phase (Area 3 to Area 2) via solid-state diffusion; (iii) at the outer edge of the center (Area 1), Fe³⁺ ions are reduced to Fe²⁺ ions, as presented in the following reaction [68]:

[Fe³⁺] + (Cr²⁺) = (Fe²⁺) + [Cr³⁺]

(R33)

The round and square brackets refer to the tetrahedral and octahedral sites of the spinel lattice, respectively. The formed Fe²⁺ ions migrate from the center to the rim of the particle via solid-state diffusion. (iv) After the complete reduction of Fe, the remaining Cr²⁺ ions together with the Cr³⁺ ions are reduced to their metallic state (Cr⁰), creating Fe- and Cr- depleted spinels of MgAl₂O₄ [68].

Though the reaction mechanism proposed by Soykan et al. (1991) is for carbothermic smelting, it is likely true for the employment of hydrogen as a reductant during solid–gas reduction. The use of hydrogen is, however, limited to the reduction mechanism of Fe-oxides as Cr₂O₃ reduction is not evident when considering Table 3 and Table 4.

EPMA analysis was employed to determine the chemical compositions of the center (Area 1) and transition phase (Area 2) of treated chromite. The rim (Area 3) was not considered for EPMA as it mainly comprised metallic Fe. The results obtained from EPMA analysis are represented in

Table 5. EPMA of particle Areas 1 and 2 (as indicated in Figure 4) of ore particles reduced by hydrogen at 1100 °C for 30 min. Table 5 presents 90 spot analyses from Area 1 (center) and 21 spot analyses from Area 2 (transition phase).

Grain Area	Detected Phase (wt,%)
	TiO2	Al2O3	Cr2O3	FeO #	MnO	MgO	NiO
Untreated *	0.97 ± 0.04	11.97 ± 0.18	44.45 ± 0.34	32.14 ± 0.22	0.36 ± 0.02	4.45 ± 0.09	0.12 ± 0.02
Area 1	0.59	16.69	56.14	12.28	0.34	13.71	0.01
	±0.07	±0.25	±0.75	±0.28	±0.06	±0.17	±0.02
Area 2	3.31	21.47	71.82	0.62	0.04	0.67	0.01
	±0.15	±0.28	±0.86	±0.08	±0.03	±0.04	±0.02

* Average values of six microanalyses, data reproduced from Paktunc and Cabri (1995) [72], with permission from Elsevier; ^# total iron.

.

It is evident from Table 5 that the total Fe content decreased systematically from Untreated to Area 2, and that the total Fe in Area 2 was <1 wt.%. The decrease in total Fe indicates its migration towards the rim (Figure 5), where it occurs as a metallic Fe layer.

The Cr₂O₃ content of Areas 1 and 2 increased from 56.1 to 71.8 wt.%, respectively, which may be ascribed to the migration of Fe²⁺ from Area 2 to Area 3. This observation suggests that Cr remained immobile at the investigated reduction conditions. The migration of Fe towards the rim of the particle further leads to changes in the MgO content within the different grain areas. The presence of MgO in Area 2 is non-appreciable and according to EPMA calculations, Area 2 is dominated by eskolaite with the composition of Cr_1.4Al_0.6O₃. This further suggests that tetrahedral vacancies created in Area 1 were occupied by Mg⁺² originating from Area 2.

Changes in the composition of chromite resulting from reduction by hydrogen are shown in Figure 6 in terms of Mg fraction among the tetrahedral cations and Cr fraction among the octahedral cations in chromite. Increases in the Cr fraction from the original unreacted chromite to Area 1 shown as groups of A and B result from the reduction of Fe³⁺. With the continued reduction and migration of Fe from the chromite structure, the Mg/(Mg⁺Fe²⁺) ratio increases from A to B in Figure 6. The highest observed Mg/(Mg⁺Fe²⁺) ratio is 0.87 indicating the extent of Fe reduction under the experimental conditions.

According to Figure 6, the removal of Fe from the chromite spinel yields a partially Fe-depleted chromite phase and eskolaite with the composition Cr_1.4Al_0.6O₃ (as determined by EPMA). The extent of reduction is determined by the amount of Fe removed from the chromite spinel. To illustrate this, the following reactions were compiled based on the results presented in Figure 6.

The lowest extent of reduction is indicated as Group A (circled in Figure 6) and is expressed as follows:

Mg_0.4Fe_0.6Cr_1.2Al_0.5Fe_0.3O₄ + 0.75H₂(g) = 0.66 Mg_0.6Fe_0.4Cr_1.4Al_0.6O₄ + 0.2 Cr_1.4Al_0.6O₃ + 0.6 Fe + 0.75 H₂O (g)

(R34)

Chromite which has undergone a higher extent of reduction than the chromite shown in Reaction (9) is indicated as Group B (circled in Figure 6) and is expressed as follows:

Mg_0.4Fe_0.6Cr_1.2Al_0.5Fe_0.3O₄ + 0.92H₂(g) = 0.5Mg_0.8Fe_0.2Cr_1.4Al_0.6O₄ + 0.36Cr_1.4Al_0.6O₃ + 0.8Fe + 0.92H₂O (g)

(R35)

Chromite which has undergone near-complete removal of Fe from the spinal is indicated with the arrow as part of Group B and is expressed as follows:

Mg_0.4Fe_0.6Cr_1.2Al_0.5Fe_0.3O₄ + 0.97H₂(g) = 0.45Mg_0.9Fe_0.1Cr_1.4Al_0.6O₄ + 0.41Cr_1.4Al_0.6O₃ + 0.85Fe + 0.97H₂O (g)

(R36)

Lastly, chromite which has undergone complete removal of Fe is indicated as C in Figure 6 and is expressed as follows:

Mg_0.4Fe_0.6Cr_1.2Al_0.5Fe_0.3O₄ + H₂(g) = 0.4MgCr_1.4Al_0.6O₄ + 0.46Cr_1.4Al_0.6O₃ + 0.9Fe + H₂O (g)

(R37)

Considering Reactions (34)–(37), it is evident that the reduction of the Fe-oxides present in the chromite spinel proceeded systematically and was kinetically limited under parameters considered in this study, i.e., complete removal of Fe was not achieved. Nevertheless, it is theoretically possible to metalize chromite’s total Fe-oxide content by selective metalization using hydrogen (Reaction (37)).

3.3. XRD Characterization of Pre-Reduced Chromite Fines

To determine the mineralogical changes occurring within the chromite spinel during reduction, quantitative Rietveld refined XRD analysis was performed on as-received, treated, and leached chromite particles (Figure 7). The treated particles were reduced at 1100 °C for 30 min. The peaks of importance, i.e., eskolaite and metallic Fe, are indicated. The remainder of the unmarked peaks are ascribed to chromite and are not labelled.

Figure 7 shows the formation of metallic Fe and the eskolaite-phase after the thermal treatment of as-received chromite with hydrogen. During the reduction process, Fe is released from the chromite spinel and affords the formation of eskolaite from the spinel structure in the form of exsolution. It was further clear that the applied acid-leaching procedure effectively removed the metalized Fe from the treated chromite, and any remaining Fe may be ascribed to its encapsulation (the confinement of an individual molecule within a larger molecule in the host chromite in the form of fine and minor inclusion).

Considering the formation of the eskolaite phase, it is of importance to determine the effect thereof on the reducibility of hydrogen-treated chromite. An Ellingham diagram was compiled with the HSC thermodynamics program to indicate the change in the reduction ∆G° of pure chromite (Cr₂FeO₄) and eskolaite (shown as Cr₂O₃) by C (Figure 8).

Figure 8 indicates that synthetic chromite Cr₂FeO₄ is reduced at approximately 1200 °C, whereas Cr₂O₃ reduces at the slightly higher temperature of 1250 °C [13]. The reason is that a sesquioxide solid solution forms between ɑ-Al₂O₃ and Cr₂O₃ that stabilizes eskolaite (i.e., Cr_1.4Al_0.6O₃) against reduction. Both Al₂O₃ and Cr₂O₃ are sesquioxide with a similar corundum crystal structure that could form a range of substitutional corundum-eskolaite-type compounds, e.g., ɑ-(Al³⁺, Cr³⁺)₂O₃, at high temperatures [76]. It is, however, noted that in practice, the Cr_1.4Al_0.6O₃ phase will have a higher degree of resistance to reduction when compared with Cr₂O₃ due to the stabilizing effect of Al.

According to Kleynhans et al. (2016), the formation of Cr₂O₃ may not necessarily be beneficial, as according to thermodynamical calculations, eskolaite reduces at a slightly higher temperature when compared with pure chromite. Kleyhans et al. (2016) proposed that it is crucial to identify the maximum temperature for Fe-oxide reduction, while minimizing the extent of eskolaite-phase formation [27]. From this study, it is evident that eskolaite-phase formation is ascribed to the removal of Fe from the chromite spinel.

3.4. Effects of Reduction Time and Particle Size on Metalization

To evaluate the effect of particle size on the extent of Fe and Cr metalization, size-partitioned chromite (refer to d₉₀ values, Table 1) was reduced at 1100 °C for 30 min (Figure 9). It is evident from Figure 9 that the degree of Fe and Cr metalization increased as the particle size decreased. The smallest size partition (d₉₀ of 140 µm) had the highest %Fe and %Cr metalization, i.e., 57% Fe and 7.5% Cr metalization. The lowest percentage metalization was observed for the largest size partition (d₉₀ of 941 µm), i.e., 18% Fe and 0.25% Cr metalization.

The observed effect of particle size on chromite metalization was, however, expected, considering that the larger particle surface area is present for the smallest size partition (d₉₀ of 140 µm), affording a greater chromite/hydrogen contact boundary when compared with the largest size partition (d₉₀ of 941 µm). Moreover, during the initial phase of the reduction reaction, the reaction occurs at the surface and sub-surface of the chromite particles, which suggests a relatively short diffusion path for both Fe-movement and hydrogen penetration [61,68].

To evaluate the effect of reduction time on the extent of Cr and Fe metalization, the size-partitioned chromite was reduced isothermally at 1100 °C for between 0 and 30 min (Figure 10).

Figure 10a further shows that the %Fe reduction had a near-linear trend. The %Fe metalization increased systematically from 51 to 84% for the smallest size fraction (d₉₀ of 140.2 µm) as the isothermal reduction time increased from 0 to 30 min. A similar trend was observed for all other size fractions (d₉₀ of 282.1 to 924.1 µm). Considering the relatively linear %Fe, it is likely that the formation of the surficial Fe layer did not curtail hydrogen’s accessibility to the underlying Fe-oxides present in the bulk material.

For all four particle sizes fractions, Figure 10b indicates very little Cr metalization during the first fifteen minutes of the reduction process, followed by a relatively significant increase in metalization. This can likely be explained by the ionic diffusion mechanism proposed by Soykan et al. that suggests reduced Cr²⁺ is metalized to Cr⁰ after the complete reduction of Fe [68]. However, considering that the highest %Cr metalization did not exceed 1.52%, it is evident that the reduction of Cr-oxide present in the chromite spinel was insignificant under the investigated conditions.

It can, therefore, be concluded that the metalization of Fe-oxides present in chromite occurs readily, whereas Cr-oxide metalization was insignificant.

3.5. Practical Implementation

This study was mainly dedicated to exploring the fundamentals of incorporating hydrogen as a reductant during chromite smelting. To guide further development, it is worth mentioning the limitations of the current study and aspects that need to be considered in the future. It is further important to consider the existing infrastructure of FeCr smelters, and the chemical properties of hydrogen, when developing an approach to incorporate hydrogen as a reductant.

Firstly, existing industrial processes employing high temperatures before smelting are considered. A significant fraction of chromite is beneficiated and pelletized via two high-temperature processes [28]. The existing pellet curing processes are summarized thus: (i) oxidative sintered pellets are generated by batching chromite, 1 to 2 wt.% carbonaceous reductant fines such as (coke, gas coke, char, and anthracite), and 1 wt.% refined clay (usually bentonite). Green pellets are then generated in a pelletization drum and cured/sintered between 1400 to 1500 °C [6,77]; and (ii) pre-reduced pellets are generated by batching chromite ore, 12 to 14 wt.% carbonaceous reductant, and 3 to 4.5 wt.% clay binder [6]. The green pellets are then cured/pre-reduced between 1300 to 1400 °C [3,6]. The hot pre-reduced green pellets are fed directly without being screened to a closed SAFs [78,79].

In both pellet-curing processes, green pellets are directly exposed to thermal energy sources. The incorporation of hydrogen directly into the pellet-curing kilns as part of the kiln atmosphere is perplexed by the combustion properties hydrogen. For instance, hydrogen has a flammability range (in air) of 4–75 vol.%, an auto-ignition temperature of approximately 570 °C, and minimum ignition energy of 0.02 mJ [57,59,80,81,82]. Therefore, hydrogen ignition would be a certainty within its combustion range, if the required amount of oxygen is present within the furnace atmosphere. Nevertheless, the addition of hydrogen outside its 4–75 vol.% flammability range is possible but would be a major safety concern and operational perplexity.

Considering the latter, the incorporation of hydrogen as a reductant will require a designated process as its incorporation into existing infrastructure is unlikely. Selecting a suitable furnace is critical, and the employment of an enclosed and externally heated furnace is the evident choice. Nevertheless, internal heating using hot gas could also be considered.

The stage at which the hydrogen treatment is incorporated in the overall chromite smelting process is another critical consideration. Two scenarios are considered here: (i) run-off-mine ore (ore as recovered from mines) is treated before beneficiation and an Fe-separation step (magnetic, chemical, or density) is included as part of the beneficiation circuit; and (ii) cured pelletized chromite is subjected to hydrogen treatment. Considering the two pellet types (oxidative sintered and pre-reduced), the effects of hydrogen treatment on oxidized and pre-reduced chromite should be determined. It is, however, likely that the hydrogen treatment of pre-reduced pellets would not be beneficial, as a significant fraction of Fe would already be in a metallic state [83].

Considering this section, it is clear that further research is required on the use of hydrogen and its real-world application. It is proposed that further fundamental research be conducted to quantify kinetic parameters for the different reactions as a basis for process design. Hereafter, a comprehensive techno-economic study be performed.

4. Conclusions

In this paper, we present the reduction of naturally occurring chromite using hydrogen. It was shown that the reduction of Fe³⁺ → Fe²⁺ and Fe²⁺ → Fe⁰ for as-received chromite, initiates at 500 and 846 °C, respectively. It was found that three distinct areas were present within chromite particles after hydrogen reduction, i.e., a rim area enriched by metallic Fe, an Fe-deprived transition phase, and a center phase representative of untreated chromite. The Cr/Fe ratios of the rim, transition phase, and center were 0.04, 4.35, and 1.66, respectively.

Removal of the Fe layer by hot-acid leaching revealed an underlying eskolaite-phase (Cr_1.4Al_0.6O₃), which suggests the exsolution of at least a fraction of chromite’s Cr-oxide content. Preliminary tests showed that particle size had a relatively significant effect on the % Fe and Cr metalization. Also, the % Fe metalization increased near-linearly with increasing reduction time, while the % Cr metalization showed a relatively appreciable increase as the % Fe metalization increased. The highest % Fe metalization was achieved for chromite with a d₉₀ value of 140.2 µm, treated with hydrogen at 1100 °C for 30 min, while the % Cr metalized was approximately 1.52%. This suggests that, under the considered experimental parameters, the Cr-oxide content of chromite was essentially unaffected.

Lastly, we indicate the limitations of employing hydrogen as a chromite reductant and indicate that its incorporation into existing infrastructure is unlikely. Specialized alterations to existing infrastructure are required to prevent unwanted hydrogen ignition during operations. We further propose that significant research and development is required before hydrogen can be employed as a real-world chromite reductant.