Thermal magnesium: is it economically viable?
00/07/06  Contents

The potential benefits of an improved thermal magnesium process
The thermal magnesium process description
Mintek pilot plant development project
Economic analyses

Introduction (back to top)

Editor's comment

This paper, by Nic Barcza, Mike Freeman and Albert Schoukens, which examines the feasibility and commercial viability of a route to magnesium metal, was delivered at the recent 2nd Annual Australian Magnesium Conference 2000, held on 30-31 March 2000 in Sydney, Australia.

Dr Nic Barcza kindly allowed permission to reproduce this work.

There is a lot of interest in strong light metals like magnesium and aluminium and their alloys. The aluminium industry is very active in Southern Africa, and is currently also examining that industry.

This paper on the Mintek-developed magnesium process looks at a thermal method, which will still take advantage of the cheap electricity price in South Africa, as the heat is generated using a DC open-arc furnace.

It will be interesting to see whether magnesium and aluminium are in competition in technical applications - there may, in point of fact, be more synergy than we realise.

Dick Copperthwaite

Magnesium metal is produced using two very different processes. The thermal process operates at elevated temperatures (over 1200C) and uses a metallothermic reduction reaction in which silicon and/or aluminium extract magnesium as a vapor from the oxide. The alternative process is based on fused salt electrolysis of anhydrous magnesium chloride. There are several variants of both the thermal and electrolytic processes. The three main thermal processes are as follows:

The Pidgeon process which uses an externally heated retort, the Magnetherm process that uses electrical resistance heating via an electrode and the Brasmag (or Bolzano) process in which a bed of briquettes comprising of ferrosilicon and dolime fines are reacted in a furnace using internal electrical heating.

The use of a vacuum is common to the above thermal processes. However, the Magnetherm process is the only one in which the spent products are tapped from the furnace as a slag. Magnesium is condensed from the vapour phase in these processes.

Some of the advantages of the thermal process include:

  • The main source of magnesium is dolomite, which merely requires calcining, unlike the complex purification that the electrolytic route requires to produce anhydrous magnesium chloride feed.
  • The reducing agent ferrosilicon (or ferrosilicon-aluminium) can be cost effectively produced, using a standard submerged arc furnace process and utilising much of the plant infrastructure as the magnesium plant.

The main disadvantages of the thermal process include:

  • The relatively high cost of the reducing agents ferrosilicon and (particularly) aluminium.
  • The vacuum necessitates batch operation that causes lower productivity and results in air ingress leading to a loss of magnesium with the corresponding formation of waste that must be treated prior to disposal.
  • The slag resistance heating in the Magnetherm type of process imposes constraints on the selection of feed materials which detracts from the optimum magnesium production.

Alternative reducing agents, such as carbon and calcium carbide, have been investigated mainly because the former would have very significant cost benefits. However, avoiding the back reaction between the CO product gas and the magnesium vapor and liquid is a major challenge. The rapid quenching of the magnesium vapour to the solid phase is necessary to avoid the back reaction between Mg and CO as proposed by using hydrogen in the Hansgirg process(1) .

"The principle that is fundamental to the MTMP is its ability to achieve efficient magnesium production at atmospheric pressure."
Mintek decided to conduct R&D into improving the thermal magnesium process during the mid 1980s, because South Africa has a surplus of relatively inexpensive electrical energy and has a developing aluminium industry. The work progressed to the stage of demonstrating that high quality magnesium metal could be produced in a pilot plant DC arc furnace using atmospheric operation(2,3). Limited larger scale work was carried out in which high extraction levels of magnesium were achieved. However, adequate condenser performance was not attained. Recent developments in the market prospects and Mintek's growing experience in DC arc furnace and condenser technology resulted in a new magnesium initiative being launched in the mid 1990s.

A new magnesium project has been initiated at Mintek, funded by a consortium of interest parties as discussed below. The objective of this work is to develop a process and plant design, comprising a DC arc furnace and condenser that could be used for a totally new magnesium project on a greenfield site.

The potential benefits of an improved thermal magnesium process (back to top)

There is considerable scope to improve the conventional thermal process for magnesium which is far less capital intensive than the electrolytic type of process. These improvements have the potential to lower the operating costs of the thermal route and make it the most cost effective production method for magnesium. The improved process can be implemented on existing Magnetherm-type plants, by modifying the conventional AC furnaces to operate under DC open arc conditions instead of the electrode being immersed in the slag. We have called this the Advanced Thermal Magnesium Process (ATMP). A greenfield plant, however, offers greater scope to change the design of the DC arc furnace and condenser and to optimize the plant layout and infrastructure. We have called this the Mintek Thermal Magnesium Process (MTMP).

The benefits of the MTMP (and to a significant degree, the ATMP) magnesium process include the following:

  • Only a few large DC arc furnaces are required to produce economically viable quantities of magnesium compared to a conventional MgCl2 preparation plant and the large number of cells needed in the electrolytic route.
  • The use of atmospheric pressure permits continuous operation of the furnace and tapping of liquid magnesium from the condenser compared to the batch operation of the conventional thermal process.
  • The conventional single-phase AC slag resistance heated furnace is limited in scale up to probably not much greater than 10 MW. A DC open arc furnace could in principle be scaled up to 20 MW, in the near future (three to five years) and even 30 to 50 MW or higher in the longer term (next ten years).
  • Conventional three phase AC supply systems require groupings of three single phase AC furnaces to balance the power load whereas this is not required for a DC arc furnace.

The most critical criteria for demonstrating the advantages of the MTMP are the achievement of a good extraction of magnesium from the feed (over 85 percent), a high condenser efficiency of at least 90 percent, crude magnesium recovery into ingot of over 97 percent, and a crude product containing over 99 percent magnesium. The extraction of magnesium from the magnesia feed material under atmospheric conditions requires selection of a slag composition that results in a high partial pressure of magnesium vapor and/or the use of a high operating temperature. These parameters require optimisation by conducting testwork on a large-scale pilot plant at power levels of between 0.5 and 1.5MW.

The thermal magnesium process description (back to top)

The principal process for the thermal production of magnesium was developed in the 1960s by Pechiney Electrometallurgie and is known as the Magnetherm process. The process is based on the silicothermic reduction of dolime (calcined dolomite) in an AC furnace at 1550C and 0.05 atm. Resistance heating of a molten slag generates the energy for the endothermic process. Dolime is reduced by ferrosilicon in the presence of alumina. The magnesium produced evaporates and is recovered in a separate condenser vessel, connected via an "elbow" to the furnace.

In the Mintek thermal magnesium processes (MTMP), under development at Mintek, South Africa, a DC open-arc furnace is used instead of the AC submerged-electrode furnace employed in the Magnetherm process, and the reaction is carried out at atmospheric pressure instead of in a vacuum of 0.05 atm. The supply of energy via a DC open-arc system offers the advantage of a wide range of recipe choices and slag compositions, as the electrical resistivity of the slag becomes a variable of less importance than in the Magnetherm process.


A schematic arrangement of the Mintek Thermal Magnesium Process (MTMP)

Figure 1 illustrates the process flow sheet in schematic form. The reaction zone at the slag surface is heated directly, and appropriately high temperatures can be achieved to obviate the need for vacuum operation. Proper selection of the feed mix keeps the temperature at which efficient extraction of magnesium is possible within the 1600 to 1800C range. Silicon can be replaced by aluminium to a certain extent because aluminium oxide, produced during aluminothermic reduction of magnesium oxide is required for the process, to lower the liquidus of the slag and to increase its MgO activity.

Part of the dolime can be substituted by magnesia as the source of magnesium, and lime can be employed to adjust the slag basicity. Some argon may be used to purge the feed system and to lower the partial pressure of magnesium. Higher magnesium extractions are achieved when operating with lower pressures of magnesium, under the same conditions of temperature and feed mix.

Nicholas Adrian Barcza

Current Position:
General Manager: Business Development and Technology (since 1 May, 1998)
Commercialisation, Mintek, Randburg, South Africa.

Academic and Professional Qualifications
BSc Engineering (Metallurgy), University of the Witwatersrand (1969).
MSc Engineering (Metallurgy), University of the Witwatersrand (1972).
PhD Engineering, University of the Witwatersrand (1977).
Professional Engineer registered with ECSA since 6 November 1978 (Registration No. 780321)

Current Responsibilities:

  • Planning of Mintek's activities and submission of Mintek's Annual Budget.
  • Reporting on Mintek's annual performance to Stakeholders via the Director's Report.
  • Commercialisation of Mintek's technologies and the development and management of Mintek's Business ventures , and
  • Mintek's Marketing and Sales and Planning and Techno-economic Divisions.

Nic Barcza is directly involved in Marketing of Mintek products and services to clients in Industry.

Background Information
During the late 1970s Nic worked for Mintek as a Senior Research Engineer, but spent most of his time in industry at several ferro-alloy and iron and steel plants on projects to develop and assist to implement computer-based metallurgical modeling and control systems to improve furnace smelting performance. This work resulted in an improvement in production by over 150 percent in ferro-chromium alloy output. He was promoted to Group Leader and subsequently Assistant Director in the Process Development Division.

In the early to mid 1980s he managed several major programmes of R&D at Mintek. These covered the fields of ferro-alloys, iron steel and stainless steel, platinum group metals and base metals. He also leads a major project that culminated in the successful development of the DC plasma-arc technology for the direct smelting of ferro-chromium from chromite fines and introduced the application of fluidized bed reactors to pre-treat fine feed materials prior to the smelting of these in DC plasma arc furnaces.

This technology has now been applied to ilmenite smelting on a commercial scale and resulted in several other major developments such as the improved recovery of cobalt from slags and improved processing of PGM concentrates. Nic currently has over 100 publications and is the inventor of 10 patents and provisional patents.

He was promoted to Director of the Pyrometallurgy Division at Mintek in 1985 and to the position of Vice-President: Technology Commercialization in 1989. He has consulted locally and overseas to assist clients to evaluate how to improve their metallurgical operations.

Operating at atmospheric pressure, besides requiring lower capital expenditure, has the advantage that vacuum-leakage problems and consequent reoxidation of the magnesium product are minimised. The MTMP can be carried out in a continuous mode whereas the Magnetherm process is cyclic with about 15 percent downtime due to breaking and restoring of vacuum during slag tapping, and because of condenser change-overs.

Mintek pilot plant development project (back to top)

The principle that is fundamental to the MTMP is its ability to achieve efficient magnesium production at atmospheric pressure. This has been demonstrated at a 70kW Mintek DC arc furnace(2,3). Magnesium extractions from the feed materials of about 85 per cent were reached. Analysis of the crude magnesium indicated that a product of high purity (more than 99.8 percent Mg), is attainable by thermal reduction in a DC open-arc furnace and direct condensation of the magnesium vapor. Improved efficiencies of magnesium condensation have still to be demonstrated at a larger scale of operation (500 kW to 2MW) before a commercial plant can be built.

The MTMP project is funded by a consortium comprising Anglo American, the South African electricity supply utility Eskom, Mintek and the South African Department of Arts, Culture, Science and Technology (DACST). A furnace designed to operate at between 0.5 and 1.5MW and a condenser rated at 50 to 100kg/h will be installed. This facility will allow a commercial plant comprising one or more furnaces rated at 1 to 2t/h of Mg to be designed. Only three such 2t/h or four such 1.5t/h DC arc furnaces would be required to produce about 50kt/a of Mg, which is less than half the number of AC furnaces for a conventional Magnetherm plant of this capacity.

The objectives of this development program are to demonstrate the MTMP at the 100kg/h magnesium scale and to obtain design data for a commercial magnesium plant. In this case many of the constraints of an existing plant and current operation can be avoided. Process and equipment parameters which distinguish this greenfield from a brownfield development program pertain to items such as: furnace dimensions, slag depth, electrode seal design, power flux (MW/m2 hearth area), furnace lining, anode configuration, condenser design, and feed materials and recipe.

The project is currently in the design phase and pilot plant testwork is planned for the period November 2000 to April 2001. The proposed equipment includes a 1.2MW power supply, a DC arc furnace and a liquid magnesium condenser. The furnace will be operated at 0.5 to 1.2MW with a production capacity of up to 100kg/h magnesium.

Economic analyses (back to top)

Thermal processes

This section concentrates on the potential benefits obtainable by changing from a process where both furnace and condenser are operated at a low pressure of around 0.05 atm to one where the pressure is essentially atmospheric. This permits several technological improvements and we illustrate the possibilities by sketching a progression of five designs as described below. The basis in each case is a greenfield plant producing 50 kt/a of magnesium ingot in Australia. Despite the location chosen, the currency used for convenience is that of US dollars.

Brief description of cases considered for economic analysis

  1. Magnetherm Process
    We have taken the classic Pechiney process which is well described in the literature as the benchmark. The provision for this analysis was therefore 12 x 5 MW AC furnaces.

  2. ATMP - The Advanced Thermal Magnesium Process
    This process represents an improvement of the Magnetherm-Vacuum process and was described above. The benefits anticipated from the ATMP process include enhanced furnace utilisation and reduced losses of magnesium by virtue of minimal air leakage into the system as a result of avoiding running under vacuum. We have used four DC arc furnaces rated at 15MW.

  3. MTMP 1
    The next stage in the progression is the Mintek Thermal Magnesium Process (MTMP) also described above. In the MTMP 1 scenario the condenser is operated in cycles instead of continuously as in MTMP 2 and MTMP 3. A new condenser is put in place after each cycle of about 20 hours.

    For the 50 kt/a Mg plant we opted for 4 x 15 MW DC furnaces each producing about 1.5t/h Mg. This power rating represents a fairly conservative scale up from the proposed 100kg/h Mg vapor demonstration plant.

  4. MTMP 2
    The next logical development is to achieve continuous operation of the condenser, a breakthrough that will reduce downtime, all but eliminate contact between Mg vapor and air and avoid freezing and subsequent re-melting of the metal prior to refining. To make the comparison as meaningful as possible, this case is also based on the use of 4 x 15 MW DC furnaces. However it is conceivable that the magnesium output could increase to about 55kt/a. We have not credited the analysis with this higher figure.

  5. MTMP 3
    One potentially enormous advantage of a DC system is that very much larger furnaces are conceivable than is the case with AC operation. For the comparative evaluation we chose a single large 56 MW furnace linked to a continuously tapped condenser.

Comparative economics

To preserve confidentiality, mainly of current operations, no absolute costs are reported. Instead we resorted to a series of indexed capital and operating costs referenced to Magnetherm as the benchmark. The table below shows the data along with the internal rates of return (IRR) for 50 kt/a plants, after tax and ungeared.

The principal assumptions used for the economic analysis are listed below:

  • Internal rate of return (IRR) calculated over 20 productive years.
  • Capital spent in Years 1 and 0 with production at 60 percent of nameplate capacity in Year 1, 80 percent in Year 2 and full output in Year 3.
  • Tax rate 35 percent.
  • Plant and machinery depreciated at 20 percent per annum and buildings at five percent per annum.
  • Full equity funding assumed unless otherwise stated.

Table 1

Financial comparisons of current and improved thermal magnesium processes

Table 1 shows the results for the profit and IRR for the five scenarios considered. The variations on the atmospheric process are all potentially more attractive than the currently used vacuum processes.

Figures 2, 3 and 4 graphically illustrate the change in IRR as capital costs, operating costs and magnesium selling prices vary.

Figure 2

Capital cost sensitivities

Figure 3

Operating cost sensitivities

Figure 4

Selling price sensitivities

In all cases, the sensitivity to changes in capital is not great. Scenarios ATMP through to MTMP3 still achieve an IRR of double figures at a 20 percent increase in capital expense.

Operating cost and selling price variations paint a different picture. Case MTMP3 can weather a 20 percent increase in cash costs but all the processes studied go into single IRR figures at a 20 percent lower selling price ($1/lb Mg).

In thermal processes, the cost of reductants is significant. Several of the scenarios evaluated utilise both ferrosilicon and aluminium metal. We have taken the input costs at LME values, viz. US$0.45/lb for 75 percent ferrosilicon and US$1500/t for aluminium. It may be feasible to captively produce a Fe-Si-Al alloy using low cost feed materials, e.g. clays/alumino silicates and quartz. The provisional calculations show that the savings more than offset the additional capital, the net effect being to increase the IRR of the ATMP and all MTMP cases above (in table 1) by nearly 3 percent.

Thermal versus fused salt electrolysis technologies

An open arc DC furnace permits the use of much higher power ratings than is the case with a single-phase AC slag resistance furnace. We therefore believe that the new generation thermal plant will compete very favourably with electrolysis plants.

The survey published by Salomon Smith Barney(4)
on 1 October 1999, predicts IRR values ranging from about 8 to 13 percent for various potential electrolysis plants in Australia. These results are based on operating costs between US$0.63 and $0.75/lb. At the Inaugural Australian Magnesium Conference (Sydney, 30 June to 1 July 1999) even lower operating costs were presented in several papers. Notably those by Elliot(5) and Laughton(6). Mintek has carefully analysed a project where the cash cost is predicted to be $0.60/lb. We consider this figure to be unrealistic and estimate that $0.73/lb is more reasonable and possibly even optimistic. For comparative purposes, we have calculated two IRRs, viz. the one at $0.60/lb and the other at $0.73/lb, called FSE1 and FSE2 respectively.

Table 2 below compares the two electrolysis plants with selected thermal plant scenarios. They are ranked in ascending IRR order.

Table 2

Comparative economics of electrolysis and thermal magnesium processes

Even if the FSE1 case costs are realistic, the MTMP1 and MTMP3 scenarios compare very favourably.

All IRR numbers reported earlier were on the basis of full equity funding. We thought it worthwhile to introduce gearing in the evaluation to ascertain whether the comparative pattern changes for electrolysis projects, that have significantly higher capital costs, and the lower cost atmospheric pressure thermal routes. The scenario adopted was for 75 percent loan capital, repayable in equal instalments over 10 years at interest rates from 7.5 percent to 15 percent per annum. To provide a datum we included IRR figures for the respective full equity funding cases, called "Base" cases. We chose to compare the FSE1, FSE2 and MTMP 1 scenarios, the latter deliberately selected as by no means the lowest cost thermal process option.

Figure 5

Comparison of thermal and electrolysis projects at 75 percent gearing

Figure 5 shows an interesting pattern. Even at a 15 percent interest rate, MTMP 1 yields an IRR 4 percent higher than the base case. At this level, both the electrolysis processes are equal to or less than their respective base cases. Thus, gearing materially improves the relative economics of the thermal plant.

Conclusions (back to top)

Our analysis has shown that, provided a DC arc open bath furnace can be operated continuously under atmospheric conditions with a liquid phase condenser that can be tapped on line, it will offer a very cost effective method to produce magnesium. The MTMP process is to be demonstrated at the 50 to 100 kg Mg/h at Mintek towards the end of 2000. It will be very important to achieve the following performance targets:

  • Mg extraction from feed - over 85 percent
  • Si and Al utilisation - over 85 percent and 95 percent respectively
  • Mg condenser efficiency - over 90 percent
  • Mg recovery to ingot from crude Mg - over 97 percent
  • Energy efficiency - over 80 percent
  • Purity of crude Mg product - over 99 percent
  • Furnace availability - over 90 percent

It is generally quite difficult to operate a pilot plant facility at the 1 MW scale continuously for long periods. However, Mintek has achieved this level of operability when extracting and condensing zinc metal and we believe that we are well placed to face the above challenge. Hopefully we will be able to report on our progress at next years MAGCON conference.

Acknowledgments (back to top)

Thanks are due to the following organisations and persons:

DACST, Anglo American, and Eskom for their support of the MTMP development project.

Titaco for engineering design and project support, and Jim Sever of Alpha Omega Engineering for valuable comments on this paper.

References (back to top)

  1. PANNELL ERNEST V. Magnesium, Its production and use, 2nd edition. Sir Isaac Pitman and Sons, Ltd, London, 1948, pages 40-49
  2. SCHOUKENS A.F.S. A plasma-arc process for the production of magnesium. Extraction Metallurgy '89, London, 10-13 July, 1989. pp 209-223
  3. BARCZA N.A. AND SCHOUKENS A.F.S. Thermal Production of magnesium. Assignee: Mintek. US Patent 4,699,653. October 1987. 5pp
  4. SALOMON SMITH BARNEY. Australian magnesium, report on the Internet graeme.newing , October 1, 1999.
  5. ELLIOT P. The South Australian magnesium metal project, Inaugural Australian Magnesium Conference, Sydney, 30 June 1 July, 1999.
  6. LAUGHTON C. Magnesite and serpentinite, Golden Triangles feedstocks for magnesium metal production, Inaugural Australian Magnesium Conference, Sydney, 30 June 1 July, 1999.