Development of a Metallurgical Process for Eramet’s Mabounié Nb-REE Project

Abstract

ERAMET, through its subsidiary COMILOG, have rights to a rich niobium and rare earth deposit in Gabon. The Mabounié deposit, is located 50 km from the town of Lambaréné, in a fairly remote location. ERAMET has been developing a process to recover value metals for several years, and approached Hatch Ltd. to design a sizeable Demonstration Plant to help them further optimize the process, and prove the flowsheet is technically viable. The process employs concentration techniques to recover an upgraded Nb/REE feed material for processing in the subsequent leach step. Rare earths and niobium are selectively leached, and subsequently precipitated. Following precipitation , rare earths undergo bulk separation and are purified to produce LRE and mixed MRE/HRE products, which will be further processed by Third Parties. This paper provides a high-level overview of the process, summarizes the motivation for a Demonstration Plant in the vicinity of the deposit, and discusses technical challenges faced during the flowsheet definition and design phase. Key challenges are focused on and summaries of the methodologies used to address each challenge, as well as the solutions to overcome the challenges, are discussed. The paper provides motivation for further process simplification and concludes with a project update.

Process Overview

The Mabounié deposit is a world-class polymetallic deposit located approximately 50 km from Lambaréné, Gabon [1–3]. Maboumine intends to not only extract the ores, but to also process the ores at the same site. In addition to the technical challenges of processing the ore , the project requires operating in a region lacking infrastructure and an industrial network. Special attention has been given to sustainable development and radioactivity management aspects of the project. Laboratory studies to develop the hydrometallurgical process began in 2008, and have progressed to continuous semi-pilot trials with several patents filed during the course of testwork. The upstream process piloting, including the ore leaching process, started in June 2011 and the downstream process piloting was also operated for several years. The Maboumine process was developed by a dedicated team from ERAMET Research, the research and development center of ERAMET Group, located in Trappes (France).

The Maboumine process is divided in two key process sections (Upstream and Downstream) and waste neutralization. The Upstream and Downstream sections each include 2 separate circuits. Therefore, there are a total of five process areas in total, as follows:

Upstream:
- Ore dressing which involves wet grinding and enrichment via magnetic concentration, which aims to produce a non-magnetic concentrate for further treatment.
- Metal recovery via hydrometallurgy and pyrometallurgical techniques, which aim to solubilize the valuable elements and produce a Nb/Ta crude concentrate and a REE /U rich liquor to be further refined in the downstream sections.
Downstream: Production of high purity concentrates, encompassing two key processes:
- Nb/Ta: aims to produce a Nb/Ta precipitate suitable to feed a FeNb refining by aluminothermy.
- Downstream REE /U: aims to produce Ammonium Diuranate and REE concentrate.
Neutralization: the objectives of this area is to neutralize the process effluents before sending them to the effluent treatment plant, in order to meet the environmental design criteria.

This paper is focused on main challenges in the Upstream Process Circuits. Figure 1 provides an overview of the main process steps within the Upstream Process Circuits.

../images/468727_1_En_198_Chapter/468727_1_En_198_Fig1_HTML.gif — Fig. 1
Overview of Eramet’s upstream circuit

Project Objective

In order to minimize project risk, particularly given the complex process being proposed, ERAMET planned to construct and operate a large pilot/demonstration plant prior to constructing a commercial plant. The pilot plant considered in the present Feasibility Study was planned to be built on the Mabounié site and was intended to test the entire process while acquiring the required data to size the commercial plant and confirm market value of the products.

The main technical goals of the Pilot Plant were to:

Validate the process flow diagram and optimize process performances.
Test the technologies at an appropriate scale that provides the design basis needed to extrapolate to commercial conditions.
Serve as a training tool for the future operation team of the commercial plant.
Serve as a tool in establishing the environmental design criteria for the commercial plant.
Maintain practices that strive for ZERO harm to personnel and equipment.
Maintain practices that strive for good community relations and minimal impact to the environment .

The Pilot Plant was designed to operate for a three-year period. The size of the pilot plant was studied in great detail, to ensure each process unit could be tested and interpreted to commercial capacity, without attracting significant scale-up risk. Equipment dimensions, scale up methodologies, and product production rates all featured in the evaluation. Due to the minimum pilot plant capacity derived, the feed tonnage required that the pilot plant be located close to the deposit.

Process Challenges

In this paper, three key engineering challenges in the Upstream Circuit are discussed. Namely, the efficient use of the reducing agent in the Leach Circuit, mixing of residue and acid ahead of Sulphation Roasting, and the correct sizing and configuration of the Sulphation Roasting/Calcination step.

Leach Circuit

After magnetic separation and upgrading the ore , the next step is acid leaching of the concentrate. Leaching is carried out at approximately 90 °C under atmospheric pressure. Sulphur dioxide (SO₂) is sparged into a series of reactors to control the redox potential and promote the reduction of ferric iron and manganese , which in turn prevents the oxidation and subsequent dissolution of niobium and tantalum . Any dissolved niobium and tantalum is precipitated due to the leach conditions. Therefore, the Leach Circuit promotes the dissolution of rare earths and uranium , while promoting the precipitation of niobium and tantalum into the residue. The reduction chemistry runs throughout the entire Leach Circuit and becomes incrementally more difficult as the circuit gets closer to its objective of 95% overall ferric reduction to avoid the oxidation and subsequent dissolution of niobium and tantalum and also reduce sulphuric acid consumption . The operating cost of the Maboumine commercial plant will be significantly impacted by the cost of elemental sulphur (to make SO₂), utilization of the SO₂ added, and treatment of the ventilation off-gas.

Sulphur dioxide utilization became an obvious critical factor in a preliminary economic evaluation, due to the quantity of iron that needed to be reduced to ensure effective separation of niobium and rare earth elements . The very high acid and ferrous iron levels present in the leach circuit interferes with the normal dissociation of SO₂ in water to form sulphurous acid. The result is that the reduction of ferric iron by SO₂ as required by the process, must be done by direct reduction with aqueous SO₂ instead of bisulphite ion and this aqueous SO₂ reaction rate is very slow.

In the initial scoping testwork completed by ERAMET, pure SO₂ was added to the reactors at 230% excess of the stoichiometric requirements for iron reduction , to keep the leach slurries saturated with SO₂ and drive the iron reduction reaction as much as possible without the reaction extent being limited by mass transfer . It is expected that with the leach slurries near saturation with SO₂ that the first pass SO₂ utilization (absorption of SO₂ between the bottom and top of the tank) would be in the order of 65%. With the high landed cost of elemental sulphur for SO₂ production and NaOH for scrubbing, a loss of 35% of the SO₂ to the reactor off-gas and sequential treatment to avoid environmental emissions could cost up to $US 136 M per year, which eroded the project economics by a material amount.

Reactor design modifications were evaluated in an attempt to improve SO₂ utilization. For instance, consideration was made to sealing the reactors and adding a top gas inducing impeller to draw the unabsorbed SO₂ from the vapor space into the solution. However, Hatch ascertained that it was necessary to operate the SO₂ reduction reactors at less than SO₂ saturation in order to get high absorption and utilization of the gas, even with gas inducing impellers. An increase of the leach residence time would distribute the iron reduction duty over a larger volume. By reducing the required SO₂ mass transfer rate per unit volume sufficiently, compensation is made for operating slightly below SO₂ saturation. It was determined that by increasing the residence time and other reactor modifications noted above, satisfactory SO₂ gas utilization can be achieved. This would represent a capital cost increase of $US 17 M.

Alternatively, higher utilization can be achieved by catalyzing the rate of ferric iron reduction so that the reaction proceeds just as fast or faster even with a lower dissolved SO₂ concentration in the slurry. Hatch suggested exploring the use of activated carbon to catalyze the reduction of ferric iron . If the SO₂ can be consumed faster by the reduction reaction process, the concentration of dissolved SO₂ can be reduced in the slurry while maintaining the iron reduction rate. With an unsaturated slurry the first pass absorption of SO₂ and re-absorption of SO₂ at the surface will increase. Hatch estimated the operating cost saving would amount to an additional $US 97 M per year.

The use of a catalyst with pure and impure SO₂ was investigated. Impure SO₂ was an attractive consideration, as it meant that conventional sulphur burner product could be directly utilized without a need for further concentration. The sulphur burner product contains about 14 vol% SO₂. The generation of pure SO₂ requires add-on technology that would cost in the order of $US 43 M. The ability to transfer SO₂ from the gas to the solution is expected to change linearly with the log mean SO₂ gas concentration (Henry’s law and gas mass transfer equation). Therefore, the ability to transfer the gas is much higher for pure versus impure SO₂. However, it was found through bench scale testwork that although in presence of activated carbon , SO₂ utilization was acceptable when using pure SO₂, using low concentration SO₂ led to significantly lower utilizations.

When considering the most important factors impacting pilot plant data collection to prove the new process and optimize the design and cost of the commercial plant leaching facility, Hatch considered the following factors as most important. These items were ultimately addressed in the design of the pilot scale reactors, such that critical parameters could be varied and data could be accurately measured:

Ability to achieve economic SO₂ utilization.
Allowing for the flexibility to assess as independently as possible the impact of varying gas retention time, aspect ratio and process retention time.
Limiting the maximum impeller tip speeds in the SO₂ sparging reactors to speeds that will allow a reasonable impeller operating life, if possible, without the need for abrasion coatings.
Limiting the maximum impeller tip speeds in the SO₂ sparging reactors to speeds that limit carbon attrition.
Optimization of the pilot scale power input per reactor that will result in a commercial plant facility with an economic operating cost with regard to agitator power draw.
Design the pilot plant to have dimensions, which will allow flow patterns reasonably similar to the commercial sized reactors.
Designing to have reasonable commercial plant agitator diameter and shaft lengths so as to have a robust and proven design providing good long term mechanical availability at commercial scale.
Providing a design with carefully considered tank to diameter ratios, which will optimize as much as possible; the reactor mass transfer capacity via dissolved SO₂ concentration, coarse solids suspension, and solids cloud height (for slurry retention). It is also preferred to provide as much of the power input as possible to the bottom impeller, where there is the most impact on SO₂ mass transfer (where the SO₂ partial pressure and reductant gas hold up are maximum). It is also necessary to try and minimize bubble coalescence and maximize SO₂ gas re-entrainment by the top impeller.
Providing an impeller design which will maximize the solids suspension capability and minimize the requirement for external transfer of coarse particles, while at the same time providing for reasonable agitator power requirements.
Maintaining as much similarity between the commercial plant and pilot plant reactor and agitator designs as possible, with respect to, and in particular, the following factors:
- Minimizing to the greatest extent possible the difference in size ratio between commercial plant and pilot plant bubble size and impeller size.
- Maintaining to the greatest extent possible the gas retention time between the commercial plant and the pilot plant .
- Providing to the greatest extent possible for the same number of reactors between both the commercial and pilot plant designs, so as to both minimize short circuiting of the process solution and to provide for a similar degree of process driving force with respect to the reagent concentration profiles (staging) between commercial and pilot plants.
- Maintaining as much similarity as possible between the commercial plant and pilot plant on scale-up with respect to slurry shear rates by the reactor impellors (impeller Reynolds number) and reactor flow patterns, by respecting as a very minimum a reactor diameter of at least 1 m in the pilot plant and preferably 1.5 m for gas mass transfer service.
- Maintaining similar aspect ratios, numbers of baffles, impeller types and, to the maximum extent economically practical, the power input per unit volume between the pilot and commercial scale plants.
- Providing for flexibility in the pilot plant design so as to be able to vary each of the similitude factors as independently as possible while correlating the various coefficients required for scale up.

Leach Residue and Acid Mixing

The Maboumine Process features several key process steps including magnetic concentration, acid leaching , and sulphation roasting. The mixing of leach residue and acid is a critical step in the process.

Following acid leaching , where most of the rare earths are extracted, the residue is filtered, washed, and dried. The dried residue is subsequently mixed with acid and fed into a rotary kiln to sulphate the relatively refractory material , at an elevated temperature . The sulphated calcine is subjected to water leaching , where niobium is solubilized. Insufficient mixing of residue with acid was found to result in poor niobium recoveries.

Mixing performance is related to the following:

The type of technology employed to mix the residue and acid;
The chemical and physical properties of the mixture; and
The maintenance of the employed mixing technology, in respects to wear and tear.

These three attributes are discussed further below.

Mixing Technology

At the onset of the project, ERAMET investigated several technologies to mix leach residue and acid. Planetary mixers were found to provide good mixing in the laboratory, but it was deemed that scale-up to commercial capacity would be impractical. Ultimately, the technology that was decided on for further evaluation was a twin shaft pug mixer. This was primarily because these units are commonly used commercially in similar applications, such as with the processing of spodumene to produce lithium carbonate by sulphation roast (Galaxy, Jiangsu, China). The device contains two counter-rotating shafts featuring paddles which mix and move the contents ultimately from the feed to the discharge end. A weir is commonly used to provide hold-back of the contents.

The pug mixer initially used in tests was selected simply based on the equipment that was available at the metallurgical test facilities employed to perform the work. This device was found to inefficiently mix residue and acid, due to the large spacing present between paddles. Based on this, the project team sought to procure a pug mixer with smaller gaps and additionally more versatility than the initial pugger. The new pug mixer allowed for paddle angles to be changed, so that different patterns could be tested in an attempt to optimize both mixing/back-mixing and throughput. A comparison of the two puggers (initial and new) is shown in Fig. 2.

../images/468727_1_En_198_Chapter/468727_1_En_198_Fig2_HTML.jpg — Fig. 2
Initial Pugger (on left) compared to New Pugger (on right)

Chemical and Physical Nature of Mixture

Following receipt of the new pugger, extensive testwork began in an attempt to optimize the niobium recovery , minimize reagent consumption, and define the process design criteria. Initial attempts to mix residue and acid yielded poor results, based on the low niobium recoveries achieved in the downstream water leach step. Initially, it was not clear what was causing the low recoveries, as several variables were at play, including the sulphation roast temperature and residence time. However, following analysis of the feed and discharge of the water leach circuit, it was discovered that the acid was not thoroughly mixed with leach residue in the pugging step.

Mixing of residue and acid resulted in an extremely exothermic reaction, which quickly elevated the temperature in excess of 140 °C. Left unmixed, the material would become solidified, similar to concrete. The fast temperature rise resulted in the formation of granules with a very hard shell. The cores of these granules were found to contain very little acid. It was therefore determined that a rapid increase in temperature was preventing thorough mixing of residue and acid in the pugger.

ERAMET hypothesized that if the temperature of the mixture could be controlled to prevent the chemistry from being activated, mixing efficiency would be improved. A cooling jacket was installed on the pugger and following several tests it was determined that maintaining the mixture temperature below 60 °C was possible, and resulted in a thoroughly mixed paste.

Given the next step in the process is sulphation roasting at an elevated temperature , there was interest in trying to reactivate the chemistry of the mixture to allow the exothermic reactions to increase the temperature of the paste as much as possible prior to feeding the rotary kiln, in order to reduce the kiln energy duty. At the same time, a granular material was desired to facilitate controlled feeding of the material into the kiln, and reduce dusting. Tests were conducted to determine if providing sufficient activation energy could initiate the exothermic reactions, within the pugger, and at the same time, via gentle mixing action produce a granular material suitable for feeding the kiln. A heating jacket was employed on the pugger, which was insulated to minimize heat loss to the environment . The paddle arrangement and rotational speed of the pugger shafts was varied to optimize the particle size distribution of the pugger product. Ultimately it was found that a hot, granular material could be consistently produced, with temperatures reaching over 200 °C in some cases.

Ultimately the flowsheet featured a “Cold Pugger” which mixed dried leach residue and sulphuric acid under a controlled temperature of 60 °C or below, followed by a “Hot Pugger” which was gravity fed, and heated indirectly to raise the temperature as high as 200 °C, producing a granular material which could be fed into the subsequent sulphating kiln. A comparison of the granules produced without temperature control and with temperature control is shown in the Fig. 3.

../images/468727_1_En_198_Chapter/468727_1_En_198_Fig3_HTML.jpg — Fig. 3
Poorly mixed granules (on left) compared to well mixed granules (on right)

Pugger Wear and Tear

Multiple tests were completed using the cold and hot pugger arrangement to optimize the mixing efficiency and sulphating kiln feed particle size. A well-mixed kiln feed material often meant a finer particle size. This is detrimental to the kiln, as finer feed inevitably means more dusting in the kiln, and increased potential for short circuiting. Optimization involved balancing the mixing efficiency and pugger product granularity. In addition to this, the puggers needed to make sufficient sample to run multiple sulphation roasting tests.

It became apparent, as the puggers were operated to generate sample, that paddle wear was a problem in a hot pugger. The Cold pugger never had a problem, as the material stayed in slurry form, and was at a much lower temperature . The duty imposes aggressive chemical conditions, at high temperatures, and in an abrasive duty. The hot pugger often faced accelerated wear of the paddle tips, which ultimately reduced final mixing and material handling efficiency to an extent that was detrimental to the process. This catalyzed extensive research on paddle materials, paddle design, and operating procedure.

Paddles of different materials and design were placed throughout the hot pugger, in a well thought out pattern that allowed each of the materials to be exposed to similar simultaneous conditions at the feed, central, and discharge zones of the pugger. Materials tested included 316SS (primarily as a reference), 17-4 PH Stainless, HVOF, and Stellite. Paddle wear was more predominant in the feed end of the pugger, likely where the maximum temperature rise would be experienced and corresponding sudden transition from a pasty material to a hardened substance. Towards the discharge end the material was very granular and friable. HVOF and Stellite were both found to wear the least, although it was recognized that periodic changing of paddle tips would likely need to be incorporated into the design of a commercial pugger.

A photo showing the paddle wear during one of the test campaigns is provided in Fig. 4.

../images/468727_1_En_198_Chapter/468727_1_En_198_Fig4_HTML.jpg — Fig. 4
Paddle wear. Stellite (S-1 in photo) demonstrated the best performance

Sulphation Roasting

The high temperature sulphation roasting is accomplished in a rotary reactor. Granular, free-flowing, and partially sulphated leach residue is fed at elevated temperatures into a rotary reactor. The rotary reactor provides adequate soak time at the targeted soak temperature to complete the sulphation of pay metals (Nb, Ta, and residual Rare Earth Elements ) that started during the mixing process in the pug mixers. The acid roast could be best represented by the following reactions,

$\begin{aligned} 2HF(SO_{4} )_{2} \cdot 4H_{2} O &= Fe_{2} (SO_{4} )_{3} + H_{2} SO_{4} + \, 8 \cdot H_{2} O_{(l)} \\ 2HF(SO_{4} )_{2} \cdot H_{2} O &= Fe_{2} (SO_{4} )_{3} + H_{2} SO_{4} + \, 2 \cdot H_{2} O_{(l)} \\ Al_{2} (SO_{4} )_{3} \cdot 17H_{2} O &= Al_{2} (SO_{4} )_{3} + 17H_{2} O_{(l)} \\ \end{aligned}$

The above reactions are highly endothermic, and the sulphation roast process has a high energy requirement. There were several important considerations made in order to design the pilot plant sulphation roast circuit, with consideration given to flexibility in order to optimize the commercial application. These are briefly reviewed in the following sections.

Sulphation Roasting/Calcination

There are two types of rotary kilns; direct and indirect fired. In a direct fired kiln, a mixture of fuel and air is injected into the kiln drum through the burner assembly, combustion occurs and process and combustion gases mix together and leave the kiln. In indirect fired kilns, feed material is processed in an externally heated sealed rotary retort. As a result, process and combustion gases never mix together. This notably reduces dust carry-over, and decreases off-gas handling duty making indirect firing more amenable for processing of fine particle leach residues. Conversely, the dominant mode of heat transfer in direct fired kilns is radiation which, compared to convection and conduction in indirect fired kilns, can deliver higher heat fluxes to the bed of material. This means that direct fired kilns can achieve the same process duty in smaller sized reactors, and are of lower cost. Considering the overall project capital cost and constructability issues, it is important to select the correct type and number of sulphation roast kilns. The project team completed a detailed evaluation of each option and decided to adopt an indirect firing configuration with direct firing capability. Figure 5 shows a typical configuration of indirect fired rotary kiln.

../images/468727_1_En_198_Chapter/468727_1_En_198_Fig5_HTML.gif — Fig. 5
Indirect fired rotary kiln (Courtesy of Feeco) A Material Inlet, B Gear/Sprocket Guard, C Riding Ring, D Furnace , E Exhaust Vent, F Air Seal, G Spring/Lead Seal, H Seal Mounting Flange, I Seal Wear Surface, J Discharge Breeching, K Gas and Air Piping, L Burners, M Advance Flights, N Inlet Breeching

Sizing

To size a rotary calciner, first the feed characteristic properties should be determined. Among others, these include feed moisture content, particle size distribution , particle and bulk densities, static and dynamic angles of repose, flowability, heat capacity as a function of temperature , and thermal conductivity. Then, to investigate high temperature conversions and establish critical temperatures, Thermal Gravimetric Analysis (TGA) and Differential Thermal Analysis (DTA) should be performed. In addition, either high temperature X-ray Diffraction (HT-XRD) should be performed to track transformations, or Time Temperature Transformation (TTT) tests should be completed to establish soak time and temperature . The project team completed a comprehensive material characterization program to facilitate kiln sizing and establish kiln operating conditions. Figure 6 shows a typical high temperature XRD spectra.

../images/468727_1_En_198_Chapter/468727_1_En_198_Fig6_HTML.gif — Fig. 6
High temperature X-ray diffraction

Material of Construction (MOC)

Selection of the material of construction is a critical issue for the design of a rotary kiln. Using proprietary Computational Fluid Dynamic (CFD ) and Discrete Element Method (DEM) modelling tools, first freeboard and bed temperature profiles were established to identify shell maximum and nominal operating temperatures. The FactSage™ thermodynamic simulation software package was also used to identify chemical species present under the kiln operating conditions. To establish mechanical design criteria, the bed profile was then calculated to facilitate estimation of shell loading and deflection. In addition, a desktop study was completed to identify different materials of construction for testing at the pilot plant facility. The above package of work resulted in the selection of the most advantageous material of construction (Fig. 7).

../images/468727_1_En_198_Chapter/468727_1_En_198_Fig7_HTML.gif — Fig. 7
Kiln temperature profiles

Summary and Project Update

ERAMET, through it is subsidiary COMILOG, have rights to a rich niobium and rare earth deposit in Gabon. In order to minimize project risk, Maboumine planned to construct and operate a sizeable pilot plant prior to constructing a commercial plant. In this paper, key challenges and summaries of the methodologies used to address each challenge of the Upstream circuits were discussed. As part of a separate exercise, which evaluated the overall economics of the commercial application based on the process flowsheet developed, testwork completed to date, and multiple assumptions, ERAMET identified the need to optimize the flowsheet while reducing associated project risks. After careful consideration of the incremental impact each circuit has on the project economics, ERAMET has significantly simplified the process flowsheet . ERAMET is currently seeking partners to pursue the development of the Maboumine Project.