AU2008339656A1 - Carbothermic aluminium production process - Google Patents

Description

WO 2009/081091 PCT/GB2008/003971 CARBOTHERMIC ALUMINIUM PRODUCTION PROCESS Ths invention relates to production of aluminium by carbothermic reduction of alumina in a single vessel continuously at steady state without cyclic variations in temperature or pressure. It is distinguished immediately from all other previous related carbothermic aluminium process patents in that a slurry of fine carbon particles in carbon-saturated aluminium is force circulated through a reactor to effect reduction of alumina directly to aluminium. Furthermore, at no stage is a "slag" melt containing dissolved aluminium carbide (A14C 3 ) in association with molten aluminium oxide (A) 2 0 )involved as a process intermediate. As pointed out in a recent patent (US 6,849,101 BI) there is a long history of patents relating to the carbothermic production of aluminium, and almost invariably A1 4

C

3 formation at high temperatures, generally between 1900 0 C and 2000"C is featured, followed by a second step taking place at higher temperatures of 2050 0 C or above, which employs a "slag" melt to effect the reaction: A1 4

C

3 + A 2 A3 =6 Al+ 3 COW. (1) The production of aluminium at these high temperatures is accompanied by evolution of carbon monoxide as well as significant quantities of aluminium vapour (Alo) and gaseous aluminium sub-oxide (A1 2 0)), which react exothermically at lower temperatures, resulting in very large losses of energy from the high temperature reacting system and blockages of gas off-takes with sticky deposits of oxycarbide, unless steps are taken to avoid this. The established method comprises reacting the off-gas with solid carbon to form non-sticky A14C. The teachings of US 6,849,101 B2 indicate that the use of wood charcoal having a porosity of from about 50 to 85% is preferred, whers US patent 6,530, 970 B2 advocates the use of reactive carbon generated in situ by the cracking of hydrocarbon compounds. However, a different approach is preferred in the present invention. The aluminium metal producing reactor (MPR) system in the present invention avoids altogether the interaction of special forms of solid carbon with A120/A1,. Using an iterative computational procedure, conditions have been identified for producing aluminium in which only a single metallic melt phase comprised of carbon-saturated liquid aluminium is himodynamically stable and thus melt containment in permanent graphite structures is entirely feasible. Such newly identified operating conditions permit liquid aluminium to be produced directly by feeding solid A1 2 0 3 into a single graphite or baked carbon reactor containing carbon-saturated liquid aluminium. This mode of metal production is analogous to smelting reduction in iron and steelmaking and is believed to be the key to lower energy primary aluminium in comparison with the traditional Hall-Heroult electrolytic route. By operating above atmospheric pressure, the adverse energy implications associated with concurrent evolution of alumious vapours (A 2 O(, Alw) can be reduced to manageable proportions. For example, employing reactor pressures typical of those used commercially for the gasification of coal, say 30 atmospheres, the total aluminium content of the evolved A1 2 0 and Alw is estimated to be less than 13% of the total aluminium content of the solid feed material. At a more modest operating pressure of say 5 bar, this is increased to around 26%a in comparison WO 2009/081091 PCT/GB2008/003971 with close to 60% evolution of aluminium as A12OW and AlW at atmospheric pressure. This suggests that operating a homogeneous single liquid phase system of carbon-saturated aluminium fed with preheated alumina is a technically feasible option, provided the operating pressure is in the region of 2 to 5 bar, whereas at atmospheric pressure the losses would be totally unacceptable. In chemical reaction engineering terms, the new method for producing aluminium comprises principally a single continuous stirred tank reactor (CSTR) system and those skilled in the art will recognize that CSTR systems and other fuly back-mixed reactors embrace a whole range of physical possibilities. These include a simple stirfed tank with mechanical agitation using an impeller nude from a carbon fibre reinforced graphite or carbon composite material with enhanced mechanical properties at very high temperatures in the region of say 2160*C to 2200*C and a gas bubble agitated melt using bottom sparking of a non-reactive gas to effect complete backmixing within the liquid phase and establish conditions for intensive diffusional mass transfer and chemical reaction kinetics. Alternatively, the simple addition of preheated A1 2 0 3 to turbulently flowing carbon-saturated aluminium can itself promote vigorous gas evolution and essentially complete back-mixing within the whole extent of a metal producing reaction zone. Also as proposed by the applicant in relation to continuous steelmaking (WO 2004/007778), the energy input to sustain the endotherinic reaction of metal production could be supplied advantageously by AC electrical conductive heating of the liquid metal phase. In this case the CSTR reactor forms the central component of a melt circulation loop with sufficient electrical resistance to ameliorate the adverse effects of inherently low resistivity of metallic melts in general and in this particular case carbon-saturated aluminium. In addition, as discussed in the continuous steelmaking patent application (WO 2004/007778), it may be advantageous to operate three "lines" of reactors or melt circulation loops, if appropriate, interconnected together in order to balance the load on a three phase AC power supply. Balancing the load between individual phases of a 3-phase AC power supply is a criterion, which under certain circumstances can be extremely important, whilst in other situations the availability of associated electrical loads may render such a consideration relatively unimpotant. For this reason the description, which follows will cover both situations by considering two preferred embodiments of the same basic newly proposed technology. In other words, a single MPR with three external melt circulation loops each serviced by its own pumping arrangement is described as well as a single MPR with effectively two melt circulation loops in parallel, one each side of the centrally located MPR, which services the pumping requirements of both the parallel external melt circulation loops. This latter arraMgement is described in detail in terms of 1 million tonnes per annum (1 Mtpa) of aluminium production for the single WR. In this case, if load balancing is important, then on technical grounds an overall production of 3 Mtpa aluminium is thus implied on a single site. In fully back-mixed reactor systems, such as those discussed by the applicant in relation, for example, to co production of steel and titanium in "swimming pool" dimensioned melt circulation loops (as outlined in the 2 WO 2009/081091 PCT/GB2008/003971 applicant's published International Application WO 2007/122366 Al, which is entitled "Co-production of steel, titanium and high grade oxide", it is feasible to conduct very high temperature processes close to thermodynamic equilibrium such dat the melt phases involved are contained within unmelted shells of material of the solidus composition to provide permanent linings, which can be replenished at wil, with minimal disruption to operation. With metallic aluminium melts this is clearly not an option. The melting point of aluminium is 660"C so establishment of a frozen ledge or maintenance of an un-melted shell would result in enormous heat losses. Accordingly, in the present invention, operating conditions are stipulated to ensure the stability of graphite or baked carbon throughout for melt containment, which demands that melts are virtually in equilibrium with carbon at unit activity, throughout the entire MPR overall system. Aluminous solid feed to the MPR, comprising refined alumina, such as Bayer alumina, reverted A1 2 0 and reverted A1 4

C

3 recovered on cooling the carbon saturated aluminium down eventually to about 670*C and thus containing only a few ppm carbon are preferably preheated to at least 900"C by radiant heat exchange between an open channel containing recirculated molten aluminium at about 1000*C in a closed loop with heat resistant alloy steel belts transporting A1 2 0 along side one arm of the melt circulation loop and A1 4

C

3 on the other, all enclosed in a oxidation protective atmosphere at the MPR operating pressure so that continuous transfer of preheated charge into or onto the molten carbon saturated aluminium in the MPR is simplified. Alternatively, the new pellet decomposer approach outlined later in this description recycles A1 4 C3, which has been deposited directly from the liquid metal on to spherical pellets of carbon or graphite, back into the MPR melt at 2000-2050"C and thereby significantly reduces the thermal energy requirements There would appear to be strong parallels between the projected functioning of the MPR and a process for gasification of coal in a liquid iron bath (I. Barin, M. Modigell and F. Sauert, Met. Trans B, Vol. 18B, 1987, p. 347-354). 'he kinetics of the gasification process were investigated by analysing and assessing the basic reactions for bottom-blowing reactors. The bath turbulence induced by the injected gas and by the product gas results in intense mixing and dispersion of the reactants and their intermidiate products. The authors point out that these phenomena create extremely large mass transfer areas and extend the retention time of the reactants in the liquid iron bath, resulting in high conversion rates relative to the volume of the reactor. Because of relatively small velocity differences between dispersed particles and the melt, they conclude that dissolution of carbon particles is controlled by mass transfer and this sub-process above all others determines the gasification rate. The applicant reaches the same conclusion for the complex physical interactions pertaining to the MPR. Assuming that the diffusivity of carbon in aluminium is of the same order as for the diffusion of carbon'in an iron melt, a carbon particle with an initial diameter of less than 0.1 mm will dissolve completely in less than two seconds, for example. In the meantime, the prevailing turbulence will cause liquid A1 2 0 3 to be transported into the bulk liquid, facilitating the key metal producing reaction between A1 2 0 and carbon dissolved in the liquid aluminium by reaction (2): A1 2 %+ 3C= 2 Al+ 3 COW (2) 3 WO 2009/081091 PCT/GB2008/003971 It is the evolution of 3 moles of product gas (COw) that ensures a high degree of turbulence as the chemistry involved is likely to be extremely rapid at bath temperatures between say 2160 0 C and 2200C. In the coal gasification process, liquid ferrous oxide (FeO) is formed as an internediate in the injection zone, which is subsequently widely dispersed in the liquid iron bath and rapidly decomposed by dissolution and subsequent reactions with dissolved carbon. Exactly the same arguments apply to aluminium metal production in the MPR. Intense bath movement can be expected resulting in dispersion of carbon particles and liquid A1 2

%

3 , thus producing emulsion and foam states, all of which result in the significant enhancement of mass transfer rates in the liquid metal. According to Barin et al. this explains the fast completion of reactions they observed in their experimental programme, leading to the high conversion efficiencies observed. Gases leaving the MPR react on cooling to generate elemental carbon and initially a fine mist or aerosol of liquid A1 2

%

3 according to equations (3) and (4): Al 2 OW+ 2 COW = Al 2 0 3 + 2C (3) 2AlW + 3 COW= AM20 3 + 3C (4) A serious accretion formation problem is inevitable, if the gases are allowed to contact solid walls before the aerosol solidifies and cools enough to render the solid material non-sticky. This type of problem is well recognised in other high temperature processes and one established method for dealing with it is to employ what are termed "fluid wall" configurations. For example, US Patent 7,033,570 B2 advocates that to combat deposition on reactor walls and plugging of the reactor leading to eventual shutdown of the process, the inside wall is at least partially fabricated with a porous refractory material through which a compatible "fluid wai" gas is forced to flow inwards, thus providing a blanket of gas which prevents deposition of particles on the inside wall. In the present invention, the reactor off-gas needs to be quenched from temperatures approaching 2200*C or possibly somewhat higher down to 1860-1960C, requiring a substantial amount of quench gas to be forced through a porous wall and thus adverse energy consumption implications associated with the required pressure drop to overcome friction. This is exacerbated by the exothermicity of reactions (3) and (4) and typically to quench the off-gas requires the ratio of molar flow of quench gas to off-gas to be at least around 2/1. Accordingly, "fluid wall" configurations are not tenable and a more direct methodology is required. It is thus proposed that the off-gas is induced to leave the reactor upwards as a turbulent axisymmetric jet issuing into a chamber or plenum containing the lower temperature quench gas with refactory walls wel removed from the jet until the temperature is reduced by entrainment and sufficient residence time is given for the aerosol liquid particles to solidify and become non-sticky before they have the opportunity to contact containing walls. The fuindamental fluid dynamic aspects for this approach are covered in the now classic work of iticou and Spalding in a paper entitled "Measurement of entrainment by axisymmetric turbulent jets" (F., D. Ricou and D. B. Spalding, J. FluidMech., 11, 1961, 21-32). 4 WO 2009/081091 PCT/GB2008/003971 In the above connection, a single slot jet can be an attractive option but in the preferred embodiment a multiplicity of circular turbulent jets issuing forth into individual quench gas plenums spaced along the flow path of the circulating melt ensures that sandy alumina primary feed along with the reverted A 2 0A formed by chemical reaction within the gas quench plenum by chemical reactions between evolved A1 2 OW and Aly with the carbon monoxide atmosphere are returned almost immediately (at most a few seconds) into the MPR and assimilated into the vigorously reacting melt. This is achieved by means of entrained flow of solid preheated alumina particles in a venturi-type direct gas solid heat exchanger in association with a cyclone separator which discharges the high temperature alumina particles directly into the MPR. The MPR off-gas then proceeds to further entrained low direct gas/solid venturi-type contactors together with their associated cyclone separators to effect at least one or probably two stages of calcined alumina preheating in advance of the unit or units attached directly to the MPR gas quench plenum. This embodiment contributes significantly to the delivery of low energy aluminium metal production without incurring losses associated with circulating fluidised bed preheaters and the like. Clearly, there is little point in expending energy merely to heat up incoming feedstocks, when pressure losses amounting to only a few millibar are associated with suspension contacting with virtually zero or very little slip between the gas and the solid particles. The central theme of this invention concerns the production of aluminium by carbothermic reduction of alumina using melt circulation and electrical conductive heating of the metal phase, employing a molten slurry of fine carbon particles dispersed in carbon-saturated aluminium. This yields a technically viable, economic and environmentally advantageous process in comparison with the current Hall-Heroult electrolytic mute to primary aluminium. Normally, the final stage of carbothermic reduction processes is a purification zone or metal separation unit in which aluminium containing dissolved carbon/carbon compound material and possibly entrained solid A1 4

C

3 can be treated to produce pure aluminium. In this connection, it has been reported in the technical literature (D.. Gerogiorgis, excerpt from the Proceedings of the COMSOL Multiphysics User's Conference 2005, Boston, USA) that this important step can be satisfied by "propriety technology". To the applicant's knowledge this new "proprietary technology" is not available in the public domain. For this reason, the present invention includes two quite distinct options, which when incorporated into the metal producing central theme, yield the desired pure aluminium as the product and thus satisfies the applicant's objective of introducing a new carbothermic aluminium process in its totality. U.S. Patent 6,475,260 B2 has drawn attention to problems associated with liquid aluminium contaminated with aluminium carbide (A1 4

C

3 ). It is stated "a primary difficulty in the carbothermic production of aluminium is caused by the substantial solubility of the carbon in the metal, about 20 atom % C, when the metallic melt is in equilibrium with solid carbon." It is further stated "a severe practical difficulty arises in attempting to purify aluminium contaminated with A1 4

C

3 in significant amounts, because the mixture becomes non-pourable unless extremely high temperatures are maintained." According to U.S. 6,475,260 B2, the problems are alleviated if aluminium metal or aluminium scrap is added as solid coolant to reduce the temperature from above 2000 0 C down 5 WO 2009/081091 PCT/GB2008/003971 to about 900-1000*C. By this treatment, A14C3 is precipitated and removed by filtration, decanted or fluxed with a salt, to yield and aluminium metal product containing 5% by wt or less A1 4

C

3 . The aforementioned level of residual A1 4

C

3 is problematic, if a refined grade of aluminiurm metal product is desired, as indeed are the steps such as filtration etc. advocated. The present invention overcomes the problem at source, i.e. while the molten aluminium is still at an extremely high temperature, say about 2160"C to 2200 0 C in the preferred embodiments. In the present invention not only is an excess of carbon above the stoichiometric requirement for carbon saturation purposely added to the MPR system but also the particle size of this excess carbon is carefully regulated so that an appropriate number of somewhat larger solid carbon particles remain in the aluminium melt removed to satisfy product requirements, before the temperature of this product off-take is lowered. These carbon particles serve as seed material for the growth of A1 4

C

3 particles of such dimension that they can be readily separated out using, for example, a vibratory screen at temperatures approaching the melting point (670"C) of aluminium and thus product filtration is avoided. Accordingly, in this first preferred embodiment the overflown or otherwise removed high temperature carbon saturated aluminium produced in the MPR is quenched virtually instantaneously into a vast excess of molten aluminium in a forced-circulated melt circulation loop at about 1000"C saturated with about 45 ppm of A1 4

C

3 and in this highly diluted and relatively low temperature condition A1 4

C

3 is not nucleated homogeneously but rather A1 4

C

3 is formed heterogeneously on existing substrate surfaces of carbon particles dynamically entrained in the circulating aluminium melt controlled by liquid phase diffusion, such that dense A1 4

C

3 is deposited. This permits straightforward liquid metal/solid carbide separation without appreciable aluminium cross-contamination of the solid ALC 3 , which is reverted to the MPR system. It also avoids entirely the formation of a cellular structure of A1 4

C

3 entrapping liquid aluminium rendering the melt difficult to pour, referred to in U.S. Patent 6,465,260 B2. However, it should be pointed out that the procedure just outlined although avoiding filtration, suffers from a disadvantage in energy consumption terms in comparison with the option now to be introduced. The other preferred option for treating the carbon-saturated aluminium for product removal, which does not rely on a rapid quench from 2160"C to around 1000*C as just outlined, but admittedly is burdened by a high temperature filtration, is the adaptation of technology utilised by the nickel industry as one component of what is known as the Mond Process. An analogous approach to a nickel carbonyl pellet decomposer is proposed for decarbonisation of the melt. Carbon- saturated aluminium, if contacted with slightly cooler graphite pellets countercurrently in a packed bed of pellets moving slowly downwards with the melt as the continuous phase moving upwards, will deposit carbon initially on the pellets. Then A1 4

C

3 will be deposited as the aluminium melt is cooled down to the temperature at which A1 4

C

3 becomes thermodynamically stable, provided homogeneous nucleation is avoided and heterogeneous diffusion growth is promoted. By the time the melt discharges from the top of the moving packed bed reactor, a very low carbon aluminium product is assured. Decarbonisation of the melt by this approach is inherently a low energy option, because the A1 4

C

3 is available for immediate return to the melt circuit at say 2050"C as material coated on graphite spherical pellets. By these means, high-grade thermal energy is not dissipated by quenching down to much lower 6 WO 2009/081091 PCT/GB2008/003971 temperatures. Despite this, relative cost-effectiveness is not readily assessed without very detailed analysis, which is not presently available. Irrespective of cost considerations, further details of the pellet decomposer approach are given later in this description, because concerns about global climate change may dictate that energy efficiency is of paramount importance. Fig. I represents the operating cycle UVX of a traditional conceptual approach to carbothermic reduction for aluminium production, superimposed on a phase diagram of the system A1 2 O-A1 4

C

3 Fig. 2 is a schematic flow diagram illustrating the essential overall features of the new carbothermic aluminium process. Fig. 3 is a computer-generated equilibrium diagram using the HSC4 program, which demonstrates that for the input conditions, shown in Table 1, no A1 2 0 3 / A1 4

C

3 remains stable at 2170 0 C/2.8 bar for an aluminium melt containing free carbon into which the product aluminium accumulates and is continuously withdrawn or overflown as a carbon-saturated aluminium ready for subsequent decarbonisation. Fig. 4 is a schematic plan view of a single pressure vessel containing the metal producing reactor (MPR) and three independent external melt circulation loops connected thereto, through which molten carbon-saturated aluminium is force circulated and subjected to electrical conductive heating over an extended length in order to supply the major portion of the endothermic enthalpy of reaction for smelting reduction of alumina to aluminium metal with all three loops coming together at one end of the pressure vessel, flowing through the principal hearth of the MPR and discharging back into their respective external closed loops. Fig. 5 is a plan view of one alternate arrangement for circulating carbon-saturated aluminium through the external melt circulation loops, which recognises that a single pumping device, such as an electromagnetic or submerged graphite-based centrifugal pump, for each loop may not be practical for the severe service conditions envisaged, but rather a number of relatively simple gas lift pumping arrangements ar needed, placed at intervals throughout the extended flow paths. Fig. 6 is an elevation view of a single MPR reactor through which the circulating carbon-saturated melt containing dispersed fine carbon particles traverses with in-line melt decarbonisation employing a pellet decomposer, which discharges A1 4

C

3 coated graphite pellets directly into the MPR vessel and a combined gas quench/venturi-type gas-solid entrained flow contactor and its associated cyclone separator. Fig. 7 is a schematic half sectional elevation of portion of the MPR in Fig. 6. which shows the pellet decomposer, candle filters and heat recovery system involving inert gas contacting of a moving packed bed of pellets, all of which are associated with melt decarbonisation to produce an ultra-low carbon aluminium product. 7 WO 2009/081091 PCT/GB2008/003971 Fig. 8 is a sectional end elevation based on Fig. 7 showing details of the pellet decomposer and the associated moving packed bed gas solid heat exchange system for continuously contacting graphite pellets to efect melt decarbonisation. Fig. 9 is a schematic diagram that illustrates the essential features of a preferred design, based on use of the ATP' family of graphite materials for a candle filter element. Fig. 10 shows the basic design in cross section of one of the modular units used in the pipe heaters. Fig. II compares in sectional elevation two alternative approaches for quenching hot gas from the MPR reaction zone using turbulent axisymmetric jets for entraining cooled quenched gas well removed from walls to avoid accretion problems. Fig. 12 is a schematic flow diagram which tracks the progress of both calcined alumina and MPR off-gas through three stages of solid/gas contacting. Fig. 13 is a sectional elevation of the four-stage gas-lift pumping arrangement depicted in Figs. 18 and 19. Fig. 14 is a schematic sectional elevation of the means for accommodating differential thermal expansion between an external pipe loop and the central MPR in-line processing units, which employs a pool of melt into which the "snorkel" feed to the external pipe loop is free to expand in all directions to take up thermal expansion movement of unconstrained graphite pipework. Fig. 15 is a plan view of the arrangement depicted in Fig. 14. Fig. 16 is a schematic plan view associated with the alternative approach to melt decarbonisation, employing modification of the melt pool shown in Figs. 10 and 11 to facilitate phase separation of coarse carbon particles from the melt. Fig. 17 is a flow sheet of the preferred embodiment of the invention, which involves quenching essentially particle free carbon-saturated aluminium in an ancillary melt circulation loop containing a vast excess of aluminium at about 1000*C to which coarser carbon particles settled out from the principal MPR melt circulation have been added as seed material for A14Cs growth. Fig. 18 shows two views of one strand of a three strand carbothermic aluminium smelting complex for a nominal three million tonne per annum aluminium production, employing a single slot jet and a gas quench plenum chamber for each strand as opposed to the multiple gas-off-take approach adopted for the production strand shown in Fig. 19. 8 WO 2009/081091 PCT/GB2008/003971 Fig. 19 is an elevation of a development of the MPR system depicted in Fig 6, further extended to multiple gas off-takes along the length of the MPR and incorporating in-line multistage impeller assisted draw-down of fine carbon particles and, in this case for example, four stages of gas-lift pumping providing the input to two straight line pipe heaters, one on each side of the MPR. Referring to Fig. 1, it has long been recognised (US Pat. 2,829,961) that the overall reaction A1 2 0 + 3C= 2 Al +-3 COw (i) takes place or can be made to take place in two stages A1 2 0 + 9- = 2 A1 4

C

3 +6 COW (ii) and AI4C 3 + A1 2 O% =6 Al+ 3 CO (iii) Fig. I is copied from US Pat. 4,099,959 by Dewing et al., who describe the ideal operating cycle as follows: "Molten slag after separation from product Al and CO (at approximately 1 atm total pressure) has a temperature and composition corresponding to point U. On coming into contact with carbon feed in the low temperature reaction zone, reaction (ii) takes place, enriching the slag in A1 4

C

3 and lowering its temperature (since the reaction is endothermic) until point V is reached. The enriched slag, from the low temperature reaction (ii) is then heated. Reaction (iii) comnmences in the high temperature zone, releasing CO and Al when the reaction pressure of the liquid equals the local static pressure at point X; thereafter continuing heat input and/or decrease in local static pressure (due to the liquid/gas mixture rising) causes reaction (iii) to proceed, the A1 4

C

3 content of the slag dropping. In steady-state operations conditions return to point U." Referring now to Fig. 2, the plant comprises a refractory-lined vessel 1 enclosing the MPR graphite fabricated system containing a bath of carbon-saturated molten aluminium 2. Carbon-saturated molten aluminium containing fine carbon particles in suspension is recycled in suspension around one or more melt circulation loops and heated by electrical conductive heating in external carbon/graphite pipes 3 over extended flow paths such that the combined sensible heats of the slurry adequately supply the bulk of the endothennic reaction enthalpy required in the MPR. Within the extended MPR reaction zone 4 aluminium metal is formed by the reaction of dissolved carbon with liquid Ai 2 0A under dispersed contact conditions induced in the MPR reaction zone 4 by both the turbulence of the circulating molten metal and the copious evolution of carbon monoxide. Calcined alumina 5 and reverted Al 2 0 5 are added together independent of reverted A1 4

C

1 . Alternatively, both AI 2 0 and A1 4

C

3 could be added together provided the preheating temperature is less than that required for chemical reaction between the two solid materials. Make-up carbon 7 is preferably added remote from the MPR reaction zone to avoid any possibility of losses due to entrainment in product gases. Alternatively, inert gas injection could be used for pneumatic transport of preheated carbon into the MPR vessel I as well as providing the means for ensuring adequate dispersal into the melt prior to its progress into the external conductively heated pipes 3. Multi-stage impeller assisted carbon particles draw-down is conducted in 8 to assimilate the carbon make-up requirement 7 into the slurry, which for direct carbothermic reduction of A1 2 0 3 is 3 mol C per I mol Al 2 0 3 reacted or 2 mol of Al product. Various options for closed loop melt circulation are available, which are collectively shown as 9. These include multi-stage gs-lift pumping, mechanical and electromagnetic pumping systems. The thermal 9 WO 2009/081091 PCT/GB2008/003971 energy recovery system 10 is centred around preheating calcined alumina feed using round turbulent axisymmetric jets of hot MPR off-gas 11, which entrained cooled off-gas in the first instance followed then by an arrangement of entrained flow venturi-type heat exchangers coupled with cyclone separators. Alumina particles at around 1950*C flow directly from the base of the last cyclone into fluidised loop seals to overcome the negative pressure gradient to permit continuous discharge of solids into the MPR via the down-comers 12. Recirculation of solids is not necessary in simple heat exchange to calcined sandy alumina, which is predominantly comprised of .particles 75-125 pm with possibly a small fraction (less than 5%) greater than 150 pm and less than 10% smaller than 45 pm. Finally, carbon-saturated aluminium is decarbonised to produce a continuous stream of ultra-low carbon aluminium product 13. Referring now to Fig. 3, this a graphical representation of the multiphase chemical equilibria pertaining to a MPR reaction zone at 2160*C and total pressure of 2.8 bar. It is computer generated using the Outokumpu HSC Chemistry* for Windows Program. The input raw materials are: A1 2 0 3 1.514 mol; C 245 mol dissolved in aluminium; free C 248 mol dispersed in carbon-saturated aluminium. These input data are those for a melt circulation system containing carbon-saturated aluminium with about 10 vol% dispersed carbon particles and are typical of a circulating melt entering the MPR at just under 2200*C and due to extensive back-mixing dropping almost immediately to the uniform reaction temperature of 2160*C in this example. This temperature drop takes into account not only the sensible heat transferred but also the chemical regular solution thermal consequences of the melt saturation carbon content difference dropping from saturation at 2200*C to saturation at 2160*C The solid feed in this example is preheated to 1950*C and although the basis for the evaluation is the steady state formation of 2 mol Al from I mol new calcined A1 2 0 in the feed, the actual input to the reaction zone evaluated is 1.514 mol A1 2 O, which reflects the sum of the new feed A1 2 0 plus that reverted via the chemical reactions occurring in the quench zone involving AlzOW and Alw with COW as the temperature decreases. To evaluate the total endothermicity to be supplied by conductive heating in the external melt circulation loops in which heat has to be added not only to increase the sensible heat but also that associated with the carbon actually dissolved in the melt to maintain saturation as the temperature increases up to just below 2200*C. Again thermodynamic regular solution behaviour can be assumed, but of course this heat solution added to the melt in the pipe heaters is returned to the system, when the circulating slurry enters the MPR. Thus the actual maximum temperature of the melt required in the external pipe heaters is decreased by the heated solution effect. In other words, without the beneficial effect of heated solution and total reliance on sensible heat to supply the endothermicity would require a greater temperature differential AT. Very approximately, this means in the present example that rather than a AT of 50*C being required on sensible heat transfer alone, a AT of 40*C is adequate. The HSC evaluation shown in Table I gives the endothermicity of chemical reactions taking place in the reaction zone as being 2047 kJ. Downstream of the chemical reaction zone but still within the MPR further thermal demands have to be added to this figure when evaluating total energy requirements. In the present case these are (i) within the decomposer sump A1 4

C

3 are returned to the melt 2050*C and the graphite pellets subsequently removed at 2160C, which consumes about 66.1 kJ/2 mol Al and (ii) addition of fine carbon particles preheated to 2000'C which are heated to 2100*C by the melt with a thermal demand of 12.7 kJI2 mol Al. This additional thermal load must be added to the chemical reaction endothermicity of 2047 kJ shown in Table 1 to give a total demand of 2126 kJ2 mol Al. 10 WO 2009/081091 PCT/GB2008/003971 TABLE I HSC4 Equilibrium Evaluation Basis: I mole Al20 consumed or production of 2 moles Al (54 g) File Name: C:\HSC4\O11AL.OGI S Temperature 2433.15 K Pressure 2.8 bar Volume 3.371-01 m3 ( NPT Reaction enthalpy 2.047E+03 W Reaction entropy 8.787E+02 J/K Iterations 17 Liit 100 INPUT AMOUNT EQVIL AMOUNT MOLE FACT ACTIVITY ACTIVITY PHASE 1: mwl mol COEFFICI 02(g) 0.0000E+00 1.5492-013 3.277-14 1.00400 3.2774 0(g) 0.00003+00 5.2024E-009 i-l00E-09 1.003400 1.100E-09 Og)0.0000E+00 4.1208E+000 0.716E-01 1.003+00 8.716E-01 CO2(g) 0.00003+00 4.8114E-005 1.018E-05 1.00E+00 1.028E-05 A12019) 0.000034-00 4.2109E-001 8.906Z-02 1.003+00 8.906E-02 Al(g) 0.00009+00 1.86039-001 3.935E-02 1.00E+00 3.935E-02 A10(g) 0.0000E+00 4.94363-005 2.0463-05 1.003+00 1.046-05 Total 0.0000+00 4.7200+000 PHASE 2 MOLE FRACT A14c3 0.00003+00 0.000034000 0.0003+00 2.50E+00 0.0003+00 A1203 1.5140E+00 0.00003+000 0.000K+00 1.00E+00 0.000E+00 Total 1.5140+00 0.00003+000 PHASE 3 MOLE FRACT C 2.0718E+02 2.0107E+002 1.736E-01 5.76E+00 1.0003+00 Al 9.5507E+02 9.5707E+002. 8.264E-01 8.99E-01 7.429E-01 Total 1.1623+03 1.1581E+003 PHASE 4 MOLE FRACT C 2.8591E+02 2.87909+002 1.0003+00 1.003+00 1.0003+00 Total -2.8591E+02 2.8790E+002 Referring now to Fig. 4, the solid feed 5 of calcined alumina and reverted Al)% is fed independently of reverted A1 4

C

3 6 into the MPR I. T1he off-gas issues from Slot 16 running lengthwise along the MPR reaction zone 4 in which copious gas evolution is talking place to form a turbulent axial-symmetric slot jet. Conditions are established downstream from the principal reaction zone 4 in the MIPR 1, once the teodnmclyunstable Al1)% and A1 4

C

3 are Moy consumed, to enable the larger particles of carbon afte many circulation cycles to settle downwards and form a Sediment with the larger particles in the bottom region from which the bottom layer material 9 is removed prior to the remaining carbon particles in the sediment being redispersed by the impeller system 8. Carbon-saturated almiWn 1I is withdrawn continuously and queched immediately in a vast excess of Al melt in a melt circulation loop at about 1000TC, to which an appropriate amount of the relatively large carbon particle sediment 9 is added resulting in individual carbonn particles stably dispersed in the bulk of the liquid alumninium flowing at an enhanced velocity to ensure that the larger seed particles of carbon or subsequent A1 4

C

3 coated particles remain in suspension. Given sufficient retention time at around 1000TC enables AI 4 C3 8 .787+02 J/ WO 2009/081091 PCT/GB2008/003971 growth to facilitate eventually straightforward solid/liquid separation using a vibratory scren or other appropriate means in advance of reversion of A1 4

C

3 to MPR at 6. In this preferred embodiment illustrated in Fig. 4, the MPR has to be designed so that once the chemical reactions subside, the cross section area of the melt channel is increased to reduce the velocity down to the saltation velocity for 50pm diameter carbon particles, which is estimated to be about 0.16 n/s, for example, and then sufficient residence time provided at the lower velocity to permit the required amount of the largest particles to settle out to form a bottom layer or sludge, which preferably would be continuously thickened somewhat possibly with mechanical assistance. An appropriate amount of solids in the sediment bed of coarser particles is withdrawn in association with a predetermined amount of aluminium melt corresponding to the product aluminium make and quenched from,. say 2260 0 C to about 1000*C in the auxiliary A1 4 Ci growth aluminium melt circulation loop. The remaining population of larger Al particles continue along with the melt to the zone where the channel is returned to its former dimensions to increase the melt circulation velocity, which together with the assistance of the impellers re-disperse the solids into the bulk melt before passage to the external pipe melt circulation loops. Some or all of the required carbon make-up is also added at this stage. The added material will be preferably ultra-fine in comparison with the larger particles, already discussed. If the thermal decomposition of methane is the source of the make-up carbon in association with decarbonisation of natural gas for a future hydrogen economy, then particle sizes down to around 0.05 pm could be involved, but larger agglomerates produced by what is termed spray pelletisation would probably predominate. Carbon fibre reinforced carbon or graphite would appear to be a composite material ideal for such high temperature service as well as the implied mechanical device, possibly needed to indne a degree of continuous thickening associated with the establishment of a sediment bed of coarser carbon particles from which the seed material for A1 4 C, growth could be extracted. In an auxiliary molten aluminium melt circulation loop (not shown), the 50 pm particles of carbon referred to earlier now dispersed in the aluminium circulating around the closed loop at about 1000C are coated by diffusion and interfacial chemical reaction processes to form A1 4 C3 coated spheres, which eventually sink to the bottom of the melt circulation channels in a lower velocity zone in which the saltation velocity of about 0.4 m/s is reached and A1 4

C

3 coated particles grown to about 3 mm diameter sink to the bottom forming a layer of settled particles. These are mechanically induced into a cul-de-sac in which the temperature is allowed to decrease to say 670"C before the thickened sludge is removed to a vibratory screen for solid/liquid phase separation to yield product aluminium melt and A14Cs for reverting back to the MPR after preheating by direct radiation whilst being transported back to say 900C on a travelling belt running alongside the auxiliary melt circulation loop at about 10OOC. The preheated solids then discharge via a chute into the MPR onto the surface of the circulating carbon saturated aluminium in the high intensity reaction zone 4. Both graphite submerged centrifugal and electromagnetic pumps are already widely used in the secondary aluminium industry. Given the significant developmental strides being made in carbon fibre reinforced carbon or 12 WO 2009/081091 PCT/GB2008/003971 graphite materials, it seems reasonable to expect the existing commercial pumping technology could perhaps be adapted to melt circulation for primary aluminium production under the very high temperature conditions under discussion. On this basis, the arrangement shown in Fig. 4 with a single pump 4 for each of the three loops would be the preferred choice For these loops transporting carbon-saturated aluminium with a relatively smal in theological terms solids volume fraction of say two percent, the melt circulation rate required in each loop is calculated to be 1.74 m 3 /s for a temperature rise of 20"C. For graphite linedpipes of I m inside diameter by 1100 m in length, this equates to a pumping head requirement of about 4 m with a theoretical total combined energy consumption of about 0.7 MJ/kg Al. The particular plant under consideration in Fig 4 is envisaged to be one of three identical strands, producing between them an anticipated 0.5 Mtpa Al. Referring now to Fig. 5, the substitution of a number of gas lift pumps placed at intervals along-the melt circulation paths rather than the preferred submerged centrifugal or electromagnetic units has to take into account that gas lift pumps usually require a submergence at least twice that of the head to be pumped. Accordingly, for a friction head loss of 6.75 m, for example, and a realistic submergence of say up to 2.7 in, a minimum number of five pumping stations are required as depicted in Fig. 5. Of course, if the head loss is decreased by increasing the inside diameter to reduce friction losses, then to secure the required electrical resistance requires a greater length for each of the two arms of the parallel electric circuits involved in each of the three melt circulation loops. Also the increased diameters and lengths demand increased consumption of graphite and a greater "footprint" for the instalation. These are matters that can only be resolved by detailed cost optimisation. Referring now to Fig. 6, this reflects a shift in emphasis towards carbothermic aluminium production with an absolute minimum of energy consumption. Melt decarbonisation is effected using a pellet decomposer 17 analogous to the nickel industry's units for producing a refined nickel product. This replaces the formation of a bottom layer 9 of sedimented larger carbon particles referred to in these two figures and continuous withdrawal of carbon-saturated aluminium II for quenching in a melt circulation loop at about OOO 0 C for ultimate product recovery. The pellet decomposer 17 is associated with a moving packed bed beat exchanger 22, which uses inert gas to recover heat from the pellets before being recycled back to the top of the pellet decomposer 17 using a bucket elevator 21 or similar device. Fig. 6 also shows a single pair of venturi-type entrained suspension contactor 19 and cyclone separator 20 into which the hot MPR off-gases flow. In principle, it is immaterial whether the high temperature gaseous off-take is in the form of a slot jet 16 or a circular jet, provided both are axisymmetric and possess a high degree of turbulence (Na,> 2.5 x 104). Referring now to Fig. 7, graphite or carbon pellets, typically 1-5 cm in diameter (not shown) are transported by bucket elevator 21 or similar device from the base of the moving packed bed inert gas(solid heat recovery unit 22 to the top of the pellet decomposer 17 in which molten aluminium is the continuous liquid phase, which moves upwards counter cunrent to the downward flow of pellets. Simultaneous heat and mass transfer takes place within the moving packed bed 23 as the molten aluminium contacts the cooler pellets, which become coated with Al4C 3 whilst the melt is progressively decarbonised. At its base 24 the pellet decomposer 17 is fed with carbon saturated aluminium melt at the MPR reaction zone 4 temperature, which has been filtered to remove the fine carbon particles associated with the melt slurry 2 using pressurised candle filter elements 25 in-line within an. 13 WO 2009/081091 PCT/GB2008/003971 enlarged section 18 of the MPR vessel 1, down-stream of the gaseous off-take. The MPR I is refiactory brick lined 26 with controlled heat loss so that the temperature gradients across the dense graphite structural components, such as the hearth channel 27 and channel ceiling 28, are sufficient to reduce the temperature safely below the 2200"C upper limit recommended for structural stability of rigid graphite insulation (not shown) and similar material. To facilitate this controlled heat loss, steam boiler tubes are preferably located internally within the length of the MPR I and the low-pressure steam so generated used for ancillary thennal demands. As throughout the whole process plant, graphite components and structures, such as 27 and 28, must be unconstrained so that stresses due to differential thermal expansion are reduced to an absolute minimum. Accordingly, to permit movement, the hearth channel 27, which nms along the length of the MPR 1, is skid-mounted or preferably provided with an arrangement of rollers or rail-track wheels (not shown) all of graphite or carbon-fibre carbon composite construction. The MPR reaction zone 4 is separated from the sump of the pellet decomposer 17 by an underflow weir 29 so access is not permitted to this region by gaseous reaction products, i.e. COs Al 2 Oy and Ales. The principal melt circulation of slurry 2 flows under the weir 29 and continues its progress from left to right in Fig. 7 in the graphite channel. An inert gas such as argon is admitted downstream of the underflow weir 29 to balance the total pressure in the pellet decomposer sump region with that upstream in the MPR reaction zone 4. Movement of the graphite pellets in the moving packed bed 23 is controlled by the rotary valve 30. Pellets discharged from the rotary valve 30 are collected in the open mesh hopper 31 and transported out of the MPR 1 with the screw conveyor 32. To effect heat recovery within the moving packed bed 33 inert gas is admitted through duct 34. Referring now to Fig. 8, this cross sectional end elevation of the pellet decomposer 17 and its associated plant complements the sectional elevation in Fig. 7. It clarifies the relative locations of the two moving packed beds 23 and 33. Graphite or carbon pellets are contacted firstly with molten carbon saturated aluminimn over the temperature range of the MPR reaction zone 4, say 2160*C down to nominally 700*C in order to effect decarbonisation to yield a stream 13 of ultra-low carbon aluminium product on a continuous basis. The pellets are projected into the MPR I via the rotary valve 30 to thermally decompose A1 4

C

3 to its constituent elements. Fig. 8 also gives a clearer picture of the functioning of the screw conveyor 32, which transports the initially A1 4

C

3 coated pellets upwards out of the shury melt 2 into the gas free board 35 immediately below the rotary valve 30 and then discharges pellets depleted of their aluminium content onto the top of the moving packed bed 33 of pellets. By the time the substantially AlsCrfree pellets reach the bucket elevator charge system 36 at the base of the packed bed inert gas/solid heat recovery unit 22, their temperature is reduced to nominally 300"C or less depending on considerations relating to pellet densfication requirements, as already discussed. The inert gas enters through duct 34 at the base of 22 and by design discharges at the top through duct 35 preferably with a pressure loss of less than I bar within 50* or so of the temperature of the inlet hot pellets, which are highly insulated throughout passage in the screw conveyor 32 to minimise heat losses. Referring now to Fig. 9, the assembled candle filter unit shown in cross-sectional elevation makes use of a commercially available family of synthetic isomoulded graphites. Isostatically moulded UCAR*O Grade ATJ4 has been an industry standard for years. It is a fine-gramin, high strength advanced graphite material with a permeability of 0.002 Darcy and thus appears suitable for the outer cylindrical container 37, which for example is 14 WO 2009/081091 PCT/GB2008/003971 about 60 cm outside diameter Preliminary evaluation indicates that this should have no difficulty in withstanding an intemal pressure of 30 bar, if the internal diameter is 40 cm or in other words the wall is 10 cm, in thickness. The conceptual design is based on four candles 38 on square pitch inside the ATJm cylindrical container 37. Each candle element 38 is fabricated from Grade ATAT isomoulded graphite, which is the lowest density version of isostatically moulded, fine grain UCAR grade ATPM family of graphite materials with a permeability of 0.026 Darcy. If the effective length of each candle filter element 38 is 1 m, then an estimated minimum of eight individual candle units 38 need to be on-line in operation at any instant, so the preferred arrangement shown in Fig. 7 should be viable, provided a rapid cyclic changeover between units on-line is achieved. It is conceivable, however, that the four units 25 depicted in Fig. 7 may need to be duplicated by a second row of identical units. There is a considerable technical challenge involved in this area and skilful adaptation of cutting-edge technology will be required using ball valves, solenoid-operated ramrods, pressure surges, etc., and the like will need to be employed. There is also need for development of a viable means for filter cake removal from the filter elements 38 and its flushing out through the downcomer 39, making use of the flushing action of the incoming slurry under pressure as it enters the assembled filter unit through the input duct 40. A preferred approach for filter cake removal from the candle filter elements 38 is to employ a loosely fitting hollow disc-shaped device (not shown), rather similar to the tube sheet of a shell and tube heat exchanger with four tubes, which is poorly fitting around the four elements 38 and which floats on top of the incoming slurry and at the end of each cycle is thrst downwards by mechanical means or a pressure surge and in so doing scrapes the filter cake into the conical funnel 41 in preparation for its flushing back into the pellet decomposer sump within the MPR vessel I via the downcomer 39. The filtrate of particle-free carbon-saturated aluminium leaves the candle filter unit 25 through the discharge pipe 42 to join discharge from other units on-line at the same time and enter into the pellet decomposer 17 through the inlet 24, both of which are shown in Fig 7. Referring now to Fig. 10, this is a sectional elevation of a preferred means for assembling the graphite external pipe heaters for conductively heating the circulating melt slurry, which employs a series connection of individual graphite modules 43, which are joined together one after the other to form a conduit for circulation of the melt external to the MPR. Technology developed for the construction of columns of large graphite electrodes (up to 800 mm diameter) for electric arc furnaces, which are assembled into columns, usually three to a column to give an overall length of 2.8 m, can be adapted to joining the individual modules to each other, analogous to the tapered sockets and double tapered machine threaded nipple arrangement used in electrode column assembly. Although a relatively small fraction of the total electric current flows through the graphite wall of the module in comparison with the major current flow in the melt slurry, which is mainly within the so-called "penetration depth" associated with alternating current (AC), it is still important for stability of operation that the electrical circuit integrity is preserved by appropriate graphite jointing techniques. Notwithstanding electric current aspects, the over-riding concern is to ensure that the joints are leak proof. For this, adaptation ofjoint sealing methods and materials currently used in aluminium electrolytic cells is clearly the basis of the preferred approach. Hollow cylinders of rigid graphite felt 4 thermally insulate the conductively heated modules 43, but careful design is essential so that the rigid felt composite nde from graphite fibres and a carbon-binder do not have electrical contact with the heater modules 43. For pipe heaters, which well may be in excess of 150 m in length, differential thermal expansion must be addressed from the outset. The preferred embodiment uses an arrangement of rail 15 WO 2009/081091 PCT/GB2008/003971 track wheels to permit lateral movement along the straight length of heater, which is transmitted to the metal pool/snorkel configuration to be introduced in Fig. 14 and 15. The rigid graphite felt hollow cylinders 44 are preferably also provided with rollers or wheels 46, or alternatively, skid mounted to simplify construction. The rail system 47 clearly has to be supported on a secure refractory foundation (not shown), which is preferably contiguous with the insulating firebrick or perhaps monolithic refractory material, which lines the containment vessel (not shown) running the length of the assembled pipe heater. This maintains the protective inert gas atmosphere at a predetermined pressure, normally close to atmospheric pressure. All components shown in Fig. 10 are of graphite or carbon fibre reinforced graphite construction, except perhaps the rails 47, which subject to experimental evaluation may be fabricated from ceramic material. The components shown in Fig. 10 are all installed asymmetrically within a refractory lined shell running the length of the pipe preheater to permit access by personnel for inspection and maintenance. Now referring to Fig. 11, this compares two preferred approaches for gaseous off-take from the MPR I; (a) multiple circularjets 48 paced along the length of the MPR or (b) a single slot jet 49 located downstream of the vigorous MPR reaction zone and thus probably requiring a somewhat longer MPR vessel than (a). Both approaches are compared for a specific example, say 1 Mtpa Al priuction, both employing graphite hearth channels 27 of 2 in width transporting the slurry melt some 150 m for (a) and about 190 n for (b). In both cases the refractory hearths 27 are enclosed within refractory lined pressure vessels maintaining an inert gas pressure slightly greater than 2.8 bar in the present example with the total pressure within the MPR nominally 2.8 bar to prevent egress of aluninous vapour out into the refractory lining 26. In both configurations the off-gas is first quenched by entrainment of cooled quenched gas via perforated refractory enclosures 50 to which is admitted the quench gas through inlet ports 51. le cooled quenched gas is recycled from the off-gas stream after sensible heat recovery and charge solids preheating duties with the remainder proceeding to a shift reactor in advance of gas turbine-based combined cycle power generation. In (a), for this example, the off-gas leaves the MPR reaction zone extending over most of the vessel initially as six round jets 48, each 1.46 m in diameter with a velocity of 20 m/s, whereas in (b) a single slotjet 49 0.75 m x 40 in length discharges the entire off-gas at a velocity of 6

.

7 m/s into a gas quench plenum. The Reynolds Numbers NR, for each are > 2.5 x I0 and the off-gas is quenched from 2160"C to 1955 0 C. The entrainment length for each is about the same because the mean hydraulic radius x 4 of the slot is almost the same as the diameter of the circular jet. For the single slot jet discharge takes place into a gas quench plenum, which is a refractory-lined vessel 52 of similar diameter to the MPR itself and extending in this example for a length of greater than 40 in so that the expanded jet is accommodated until it is quenched to 1955*C after which it flows into a header 53 leading to an alumina preheating system. Clearly, if an entrained venturi-ye contactor coupled to a cyclone separator is used for both (a and (b) the volumetric flow in (b) is six times larger and thus for the same gas velocity say 5 m/s in the entrainment contactor, the diameter has to be 2.45 times that of the multiple units employed with the round jet alternative. The multiple units need to be about 5 m diameter internally, which means the corresponding single slot jet equivalent requires 12.25 m diameter. In (a) the partially heated calcined alumina is admitted into the quench jet at the level 54 by a downcomer from the preceding cyclone separator and heat recovery begins immediately with the entrained solids and gas proceeding to the cyclone separator (not shown) through 55. There are many interactions involving complex issues to be 16 WO 2009/081091 PCT/GB2008/003971 resolved in making a finn choice between (a) and (b), so for the present purpose both are viewed as preferred embodiments. TABLE 2 HSC4 Equilibrium Evaluation for 2.8 bar operating pressure Basis: I mole A1 2 0 3 consumed or production of 2 moles Al (54 g) Position 56 57 58 59 60 61 62 63 as per Fig. 12 MPR Solids to Gas to Calcined Off-gas MPR HRSG A1 2 0 Temperature 2160*C 1950"C 1320*C 1000*C 1950 0 C 1320*C 1000*C 100*C COW 4.121 12.924 13.058 12.583 CO2 4.81x10- 5 3.57x10 4 1.13x10- 2 2.49x10' A120h 4.211x10' 6.05x10 3 - 2.13x10-4 0 AlW 1.860x10' 7.26x10 3 l.05x104" 0 A1 2 0 1.514 1.010 1.000 1.000

AI.C

3 5.24x10- 5 0 0 C 1.189 0.271 0.237 1 1_1___1 Now referring to Fig. 12, this identifies the locations of the various gas and entrained solid streams listed in Table 2. The hot gas emerging from the MPR is 56 and the preheated solids comprised of calcined alumina, and reverted A1 2 0 3 along with elemental carbon arising principally from reasons in the quench zone is 57. The streams 57 to 59 emerging from the cyclones with a relatively high solid content in comparison with the suspended solid/gas within the entrained contactors are considered to be in thermal and chemical equilibrium with the cleaned gas leaving the cyclones 60 to 62. The initial feed of calcined alumina is introduced at 63 via a screw feeder or similar device from a pressure lock hopper system. Streams 57 to 59 flow opposed to the static pressure gradient and use loop or siphon seals or similar devices. The exit gas 62 finally leaving the system proceeds to a heat recovery steam generator (HRSG) before being filtered with some of the gas being recycled for entrainment into the turbulent axisymnetric jet leaving the MPR. The balance of the gas proceeds to a shift reactor in advance of the combustor of a gas turbine for combined cycle power generation, involving a second HRSG, probably based on supercritical steam generation for addition to that from the other HRSG and elsewhere in the heat recovery circuit Referring now to Fig. 13, carbon or graphite blocks are used to fabricate a series of underflow weirs 64 and overflow weirs 65 with connecting passages arranged in series to create a multistage gas-lift pumping system. Vertical graphite lances 66 admit inert gas to the melt at the base of the two-phase gas/liquid region 67 though a sparger of high permeability or similar arrangement involving numerous small diameter holes connected to a central duct running the length of the lance. A preferred arrangement employs techniques developed for the assembly of column graphite electrodes for electric arc furnaces to provide the length needed to satisfy submergence requirements associated with gas-lift pumping in general. Normally submergence twice the liquid 17 WO 2009/081091 PCT/GB2008/003971 head to be developed per stage is specified. This cross-section shows a single lance in each of the four two-phase "homogenous" compartments or regions 67 comprised of bubbles of inert gas flowing with the melt slurry to effect density reduction. The head developed over the four stages as illustrated, for example, is the difference in height of the melt slurry at exit level 68 less that at inlet 69. Contingent upon laboratory testing further rows of lances 66 may need to be deployed besides those shown in this cross-section to satisfy the endothermicity of alumina reduction to metal with a pre-determined drop in slurry melt temperature. 'Te whole of the assembly shown in Fig 13 must be free to move to accommodate thermal expansion, so the individual stages are effectively graphite structures free standing on refractory foundations insulated by rigid graphite felt boards 71 to drop the temperature from around 2160*C to at least IS C and preferably to around 1650 to 1700*C. To prevent oxidation of the graphite or carbon structures, provision of a protective inert gas atmosphere is provided, involving insuating refiactory-lined shels (not shown) to limit the heat losses, but since the maximum graphite temperature is safely below the recommended 2200C upper limit recommended for graphite rigid felt insulation, there is no necessity to implement a controlled heat loss strategy. The gas pressure is preferably slightly above atmospheric pressure to preclude air infiltration into the inert protective gas contained within the shell enclosure or enclosures. Referring to Fig. 14, thermal expansion must be taken into account for pipe loops installed initially at ambient temperature and then experiencing temperatures up to 2200*C in steady state operation. It is suggested that overall and differential changes in length may be accommodated using sumps containing the melt into which there are immersed pipe inlets or "snorkels" with freedom to move in all directions with adequate clearance of melt around them in conjunction with connections serving the pipe heaters or the barometric legs referred to previously. The system illustrated in Fig 14 permits differential thermal expansion in excess 2m to be accommodated by movement in a melt pool just after the gas-lift pumping arrangement with the MPR end of the pipe heaters anchored in fixed positions. The floating end for expansion provision is rather like a conventional siphon, requiring a reduced pressure to be applied to initiate fow to or from the pool of liquid metal. The graphite pipe modules 43 which make up the external pipe heaters are joined to a graphite bridge 72 with the cross sectional area for flow being gradually increased by the divergent channel 73 (included angle <7*) in advance of three right-angled bends which permit a graphite snorkel 74 to be immersed in a pool 75 of the melt slurry. The arrangement shown is necessary to avoid unacceptably large head losses incurred for fluid fow through 90* short elbows. A futer preferred embodiment incorporates the arrangement shown in Fig 14 at both ends of the straight line pipe heaters where the flow through the assembled modules 43 jans the mainstream common flow circuit of the MPR and ancillaries. Thermal expansion in the long straight pipe heaters is taken up by the movement of the snorkel 74 in the melt pool 75. The arrangement shown in Fig. 14 is connected to the melt discharged from the gas-lift pumping system shown in Fig. 13 with the last two stages 76 and 77 being fully visible. The final stage 76 discharges melt directly into the liquid pool 74. At the other end of the pipe heater a barometric leg (not shown) is connected to an analogous snorkel/melt pool system, again to take up theral movement of the straight-line pipe heaters, one on each side of the MPR. With this arrangement the centre of the straight4ine pipe heaters is in a fixed position for electrical currnt introduction via a hub arrangement (not shown). 18 WO 2009/081091 PCT/GB2008/003971 Now referring to Fig. 15, this is a plan view of the arrangement shown in Fig. 14 comprising graphite pipe modules 43, divergent channels 73, associated bridge 72, snorkel 74 and melt pool 75. Referring now to Fig. 16, the melt from the last stage of the gas-lift pumping on its way to the snorkel 74 enters via channel 78 the modified melt pool 75 with the major flow bypassing most of the pool 75 via the channel 79. Flow into the pool-75 is restricted by the baffle system 80 so that the melt velocity in the pool 75 is reduced below the salvation velocity of the coarser carbon particles, permitting phase separation and ultimate recovery of these particles for use as seed material subsequently for melt decarbonisation. The dotted outline of 74 schematically illustrates the new position of the snorkel once operation at the steady state temperature is reached. Referring now to Fig. 17, this is an overall process flowsheet of a preferred scheme for decarbonisation of a carbon-saturated melt, which involves quenching the melt after initial gravity phase separation of carbon particles from carbon-saturated aluminium at MPR reaction temperature, perhaps in a cul-de-sac off the main stream MPR circulation followed then by quenching the equivalent amount of aluninium produced by carbothermic production into a vast excess of recycled molten aluminium in a force-circulated melt circulation loop at about 1000'C, to which coarser carbon particles settled out from the main stream melt circulation are added as seed material for A1 4

C

3 growth. Provided nucleation followed by precipitation does not occur at the very dilute dissolved carbon levels involved (around 45 ppm) in this ancillary loop, A1 4

C

3 diffusive growth on seed carbon particles eventually increases their size so that phase separation by a simple screening operation as opposed to filtration becomes feasible to yield a low carbon aluminium product As pointed out previously, energy is degraded in this quenching route, so it is less energy efficient than the pellet decomposer method, but on the other hand in theory, it is very much simpler. Referring now to Fig. 18, both an elevation and a plan view are shown of one production line or strand, which by itself has a nominal output of one million tonne per year alumnium (1 Mtpa Al). Normally to balance the electrical load, three strands are preferred in an aluminium smelting complex producing 3 Mpta Al by the carbothermic smelting process. The preferred embodiment shown in Fig. 18 employs a single slot jet and quench gas plenum 52, which services the entire MPR 1. The two straight-line pipe heaters 81 are placed one on each side of the MPR I with both ends of the pipe heater 81 being connected via a snorkel to a melt pool 75, containing the slurry of fine carbon particles dispersed in carbon-saturaled aluminium in order to respond to thermal expansion of the pipe heaters and permit free movement of the snorkel within the melt pool 75. This arrangement also looks after any differential thermal expansion between the pipe heaters and the principal flow circuit comprised of the MPR 1 and its ancillaries. The electrical current is introduned into the pipe heater via the hub 82. Melt circulation is provided by four stages of gas-lift pumping 9 in this example, and make-up fne carbon particles are drawn down into the circulating melt shiny with the sWiane of impellers 8. Referring now to Fig. 19, the elevation shown depicts a single strand nominally capable of producing 1 Mpta Al, which normally for electrical load balancing reasons would be one of three such strands in an aluminium smelting complex producing 3 Mpta Al by carbothermic reduction of calcined alumina. The metal producing reactor MPR 1 is provided with two pipe heaters 81 one on each side of MPR 1. In this preferred embodiment, multiple units 19 WO 2009/081091 PCT/GB2008/003971 comprised of a venturi-type entrained suspension contactor 19 and an adjacent cyclone separator 20 are spaced along MPR I to quench the off-gas from the extended length of the MPR reaction zone, whilst effecting preheating of solids, which discharge directly into the MPR reaction zone via loop seals or similar devices located at the base of the cyclone separator 20. As in Fig. 18 the electrical current is introduced into the pipe heaters via a hub 82 (not shown). Melt circulation is provided by four stages of gas-lift pumping 9, make-up fine carbon is impeller assisted drawn down in 8 and differential expansion is accommodated in the melt pool 75 at each end of the pipe heaters S1. In this preferred embodiment, decarbonisation of the carbon-saturated melt is effected by a pellet decomposer 17, heat recovery from the carbon or graphite pellets is conducted in the moving packed bed/inert gas contactor 22 and the pellets are circulated from 22 to the top of 17 by a bucket elevator 21. SUPPLEMENTARY DETAILS A. Electrcal Conductive Heating By analogy with published data on iron-carbon melts, the low resistivity of aluminium is anticipated to be only marginally increased by saturation with carbon. From the outset, recognition of the implications of this for electrical conductive heating must be acknowledged in process design and it has to be accepted that the "footprint" for a commercial plant will reflect this situation. However, according to a magazine article (Gulf Industry Magazine, vol. 14, No. 10, October 2005, Al Hilal Publishing & Marketing Group) one of the world's longest potlines is 1.2 km with 336 cells, so on this basis, a large modern Hall-Heroult plant is certainly not insignificant in terms of "footprint". Provided proximity to equilibrium can be maintained, there will be no alumina slag present to attack graphite linings in the MPR and its associated graphite pipe melt circulation loops. As the sole energy input is conductive heating from hydro, nuclear or geothermal power, all attention must be focused on dissolution of graphite walls and hearths by molten carbon containing aluminium. B. Protection of Graphite Pipe and Wall facess The use of ultra fine carbon, such as carbon black or soot produced as a result of the thermal decomposition of natural gas in addition to recirculated carbon particles up to say 0.05 mm diameter, is believed to offer a reliable mechanism to secure virtually permanent graphite reactor walls. If ultra fine material is dispersed into the melt, any departure from saturation is rapidly made-up by particle dissolution as opposed to dissolution of graphite wall surfaces. Since the dissolution of carbon in molten aluminium at very high temperatures is controlled exclusively by liquid phase diffusion, the mass transfer coefficient applicable to individual spherical particles can be quantified. It is determined by the Ranz-Marhall equation given in Eq. (5) in terms of the dimensionless groups Sh, Re and Sc assembled from the variables: mass transfer coefficient k., chemical diffusion coefficient or interdiffusivity Dj, 20 WO 2009/081091 PCT/GB2008/003971 particle diameter d,, together with liquid velocity v, density p and viscosity p. The Sherwood number Sh ke D. M/d,, the Reynolds number Re = d, vp/s and the Schmidt number Sc = s/pD..m. Sh = 2+0.6 Re" 2 Sc"3 (5) For the relatively small velocity differences anticipated between the carbon particles and meft, the mass transfer rate for fine carbon particles is characterized by Sh=2 k d,/Deu where k, is the mass transfer coefficient, d, is particle diameter and D.m is the diffusivity of carbon in the melt. The corresponding mass transfer equation for dissolution of carbon at the walls under turbulent flow with Re > 10,000 is given by Eq. (6): Sh =0.023 Re"-" Sc"' (6) The strategy of circulating a dilute dispersion of fine carbon particles can be tested by evaluating the relative dissolution rate, which is the ratio of carbon particle dissolution to wall dissolution. For example, consider an ideal case in which a 10-volume % C particle population is mainnainer at steady state such that a uniform particle size is 50 pm. Accordingly, the relative dissolution rate is given simply by Eq. (7): Relative Dissolution Rate = [(k)A2 particle/[(k),Al] wall (7) For the 50pm diameter carbon particles, this ratio works out at 4388/1. Similarly, the relative dissolution rates for 1Opm and 20jm particles are 1.1 x 10W/I and 2.74 x 10'II, respectively. On the other hand, if a population of 100 nm (i.e. 0.1 pm) diameter soot particles derived from methane thermal decomposition could be maintained at steady state at the 10-volume % level, the relative dissolution rate would be 250,000 times as large as the figure for 50 pm particles, so these ultra fine particles would demonstrate a propensity for dissolution some 1.1 x 10' times that of the graphite lining constituting the wall. C Drawdown and Dispersion of FIne Carbon Particles Dispersion of very fine powder into molten metals is known to be difficult when the metal does not wet the powder. Kobashi et al. (Nippon Kinsoku GakkaishilJownal of the Japan Institute ofMetas, vol. 54, n. 8, Aug. 1990, pp. 933-940), for example, state that it is impossible to incorporate SiC of ultra-fine ZrC was not a problem, because of good wettability. However, wetting or non-wetting is only an issue for ultra-fine particles. Clearly, steps must be taken so that the nano-size carbon is in an agglomerated condition on initial admission to the melt circuit, with release back to discrete particles once the agglomerates have been drawn down into the turbulent flowing carbon-saturated aluminiun. Various studies reviewed by Joosten et aL (&ansLChem.E., vol. 55, 1977, pp. 220-223) and Bakker and Frijlink (Chem.Eng.Res.Des.. vol. 67, March 1989, pp. 208-210) suggest the creation of a vortex is energetically the most efficient way of suspending floating solids. Using dimensional analysis, Joosten et al. correlated the minimum Froude Number for their preferred baffle arrangement (one wide baffle) to create an eccentric vortex for drawing 21 WO 2009/081091 PCT/GB2008/003971 down floating solids. In summary, the design requirement is to provide and arrangement downstream of the MPR, which will consume a minimum of about 320 kW for 1 Mtpa Al in establishing at least twice N. so that carbon solids added continuously are drawn down into the carbon-saturated alumin. D. Pressurisation Aspects Pressurisation, even to a relatively modest degree, dramatically reduces the plant footprint. The MPR pressure vessel and the pellet decomposer sump preferably operate at around 2.8 bar absolute. Barometric legs connect to other ancillary plant within the overall melt circulation loop and are all at close to atmospheric pressure including the two vitally important conductive heating external pipe loops. For 3 phase mains electricity supply, clearly the ideal arrangement for phase balancing would be 3 production lines in parallel to produce annually 3 million tonnes of aluminium, nominally occupying say 50 in x 260 in, not including the gas booster for gas quench recycle and advanced power plant Also excluded are the solids storage and lock- hopper feeding systems. These latter components all comprise adaptations of current commercial practice. E. Gas-Lift Pumping Pumping molten aluminium using graphite submerged centrifugal pumps or by electromagnetic means is well established in both the primary and secondary aluminium industries, so this technology is immediately available for start-up purposes, switching over to gas-lift pumping at higher temperatures once the whole system is completely charged with molten aluminium and functioning according to design. The actual inert gas injection from the overhead lance system shown in Fig. 13 can be readily evaluatedL When inert gas (probably argon) is injected into the homogeneous two-phase flow compartment, bubbles continue downward initially and heat is transferred very rapidly. Assume that dispersed gas is virtually at bath temperature by the time bubbles begin to rise and flow homogeneously with the melt as "hydrostatic" pressure decreases. The dispersed gas effectively expands isothermally throughout most of the two-phase gas/liquid region. The theoretical energy input per unit mass of inert gas is thus given by: J) RT P -W = jPdy= -"In-2 (B) In the present case, lift gas injection from overhead lances occurs where the pressure is 1.3 bar plus the "hydrostatic" pressure equivalent to say 2.7 in submergence in the undisturbed melt prior to gas injection, which is 0.528 bar, making the total pressure P2 1.828 and P, = 1.30 bar. The injected argon thus expands from a pressure of 1.828 bar, where in this example the avenge temperature in the pumping zone is 2160'C, to 1.30 bar and in so doing inputs energy into the system at the rate of: W - (8.314X2458)(in(l.82 8 /1.

3 0)/( 39

.

95 ) - 174.36 kJ/kg I, for example, W input required for melt circulation = 999.1 kJ/s 22 WO 2009/081091 PCT/GB2008/003971 Total argon flowrate needed -999.1/174.36 kg/s = 5.73 kg/s = 0.143 kmoVs The thermal quench effect of this injected argon ifno pre-heat, i.e. 15'C to 2185*C is 6.45 MW, which is excessive, so a heat exchanger is preferably installed, not only to preheat the incoming argon but also to condense out and recover aluminium from the saturated lift gas. F. Melt Decarbonisation Carbon-saturated melt containing 10 vol% fine carbon particles in the particular example under review passes from the reaction zone in the MPR via an underliow weir to the sump of the pelet decomposer. This sump can be arranged, for example, so that it is maintained at a pressure of 2.8 bar with an argon gas atmosphere. By having the pellet decomposer sump at 2.8 bar, carbon-saturated aluminium cannot be directly separated from the 10 vol% of dispersed carbon particles using a candle filter. Filtration using candle filters requires infiltration of the melt into the open pore structure of the graphite. The relationship between the applied pressure and the pore size into which a non-wetting liquid will intrude is given by Eq. (9), the Washburn equation, concerning the dynamics of capillary flow: P =2yCosO/r (9) where P is the applied pressure, r is the pore radius, 7 is the surface tension and 9 is the contact angle between the liquid and the pore wall. In the absence of A 4

C

3 formation, it is well known that liquid aluminium does not wet graphite. As there can be no A1 4

C

3 stability at 2160*C according to the phase diagram, it is reasonable to expect contact angles in the range 120-140 degrees as observed at low temperatures in the absence of any thin oxide sdn, which hinders wetting, but before reaction between Al and C takes place because of kinetic constraints. Molina et al.(Scripta Materialia, vol. 56, 2007, pp. 991-994) have recently studied the infiltration of Al-Si eutectic at 680"C into graphite preforms, which had around 10% open porosity, 2% dosed porosity, a bulk density of 1830kg/m 3 and a 78% degree of graphitisation. The pressures for first penetration were reported as 0.95 A 0.1 MPa for mercury and 2.3 * 0.2 MPa for the AI-12 wt% Si melt. Candle filters are conventionally long hollow cylinders with one end closed and with the flow inwards during a filtration cycle, during which time a filter cake built is up on the external surfaces. After a prescribed time the gas pressure is released and the filter cake removed from the candle surfaces. In common with other surface filters they are operated cyclically. A second unit is brought on stream just as soon as filtration in the first unit is terminated. Ideally, a virtually constant stream of carbon-saturated melt is forwarded to the base of the pellet decomposer. It is important to recognise that carry-over of fine particles is undesirable, because these will act as nuclei for subsequent solids deposition. Accordingly, it may be nsary to conductively heat the melt above the saturation 23 WO 2009/081091 PCT/GB2008/003971 temperature as it proceeds to the pellet decomposer, downstream from the candle filter so that a prescribed portion of any fines carry-over is re-dissolved in the melt before entry into the decomposer. At the base of the decomposer substantial temperature gradients can be expected and depending on the temperature of the external surface of the pellets, either carbon or AC will plate out on the pellets. Initially, some carbon may be deposited followed by major growth A1 4

C

3 on the pellets as the temperature fals. A1 4

C

3 coated pelets at 2000-2050 C and associated melt from the rotary feeder discharge in free fall into the gas space in the decomposer sump. Throughout the preceding unit operations the melt remains the continuous phase until discharge into the sump by the rotary feeder. As the pelets are somewhat denser than the melt, they continue their freefall through the freeboard gas into the melt and continue falling downwards in the melt constrained by a trommel-type screen, constructed of graphite, to enter a screw feeder, also of graphite construction which elevates the pellets gradualy upwards back into the gas freeboard and then out through the wall of the sump, finally discharging into the top of a moving packed bed of pellets. When first entering the melt in the sump, the A1 4

C

3 coating on the graphite pellets being non-resistant to thermal shock can be expected to shatter, whilst the graphite pellets themselves should be virtually immune to thermal shock of the magnitude involved In any event, AL.C 3 rapidly decomposes back to elemental carbon and aluminium at 2160 0 C and any residual A1 4

C

3 continues to be exposed to high temperature melt as the pellets proceed upwards through the mainstream flowing melt within the screw conveyor, so designed that melt has ready access to the pellets throughout their passage back to the gas freeboard. Any associated melt can drain back into the main stream circulating melt once the gas freeboard is reached. Based on the assumption that at the very high temperatures at which most of the A14C3 deposition on pellets occurs, rate control is exclusively liquid phase diffusion, mass transfer taking place within the pellet composer can be assessed using the j-factor empirical correlation given by Sherwood, Pigford and Wilke (Mass Transfer, McGraw-Hill, 1975, pp. 241-247). The density of molten aluminium close to its melting point (660"C) is 2385 kg/in, whereas the theoretical density of graphite is 2150 kg/in. In other words, graphite spherical pellets will tend to float upwards in the upper regions of the pellet decomposer. On the other hand, as the temperature increases, the density of molten aluminium falls below that of graphite and dense graphite pellets will naturally fall downwards in the lower regions of the pellet decomposer. Ibis tendency will be reinforced by the deposition of a layer of A1 4

C

3 on the outside of the pellets, which will gradually increase in thickness as the pellets continue their downward path. Thus there is no problem with coated pellets sinking in the decomposer sump, where the slurry melt is expected to have a density of about 1985 kg/In 3 . Attention has to be focussed on the upper regions of the peHet decomposer and in this regard there would appear to be two options. Either rely on the downward thrust of a column of pellets maintained above the melt discharge level to ensure downwards progression of the moving packed bed, or artificially make sure that the graphite 24 WO 2009/081091 PCT/GB2008/003971 pellets entering the top of the pellet decomposer have a density greater than the density of aluminium as it discharges sat at 700"C as product with very low carbon content, say 10 ppm C). Whilst the first option is simple and should be tried experimentally, relying on downward thrust of a packed bed is problematic, as is well known to designers of bunkers and hoppers for granular solid material. If necessary the graphite pellets can be densified artificially. As graphite is not wetted by most liquid metals, infiltration of a dense low-volatility liquid metal is a viable means for overcoming the potential relative density problem. The prefered liquid metal is molten tin, which has a density of 7000 kg/m 3 at its melting point 232'C. The internal connected pores of a fine synthetic graphite can be filled up with a molten liquid metal under applied pressure and once the pressure is released the infiltrated metal remains in place. This has been demonstrated, for example by the recently published work of Molina et al. (Scipia Materialia, vol. 56, 2007, pp. 991-994). If the surface layers of the graphite pellets after pressure infiltration with liquid tin are subsequently reacted topochemically to move tin from the outer region of the pellets, artificially dense graphite pellets are formed with their outer shells rendered somewhat porous but still fined-grained with minimum porosity. As an extra precaution it may be advisable to re-impregnate the outer layers of the pellets with carbon using well established techniques. Molina and colleagues have demonstrated conclusively that no liquid aluminium will penetrate into fine-grained graphite under operating pressures as low as the 2.8 bar, which pertain to conditions within the pellet decomposer sump. Thus there is no possibility of contamination of the aluminium melt with tin. It needs to be established if the mechanical properties deteriorate if molten tin is allowed to solidify within the pellets and then raised in-situ above the melting point. If stresses induced by solidification and subsequent melting of the tin are unacceptably high, then the preferred modus operandi is to keep the pellets above the tin melting point 232*C at all times. 6. Off-Gas Quench/Energy Recovery In one preferred embodiment, the MPR off-gas flows into a gas quench plenum, a cylindrical pressure vessel mourned downstream of the reaction zone and immediately above the flowing melt surface in order to avoid direct contact of the hot off-gas with solid walls The Ricou and Spalding (F. D. Ricou and D.B. Spalding, J. Fluid Mec., vol. II, 1961, pp. 21-32) analysis for turbulent axisymmetric jets is based on: (m /m,) = 0.32(x/d,)(p, (p)" (10) where m m,+ m, and subscripts I and o refer to the quench gas and the off-gas mass flowrates, respectively and x is the vertical distance required above the axisymmetric jet source (a slot in the present case) with an equivalent diameter d. andp is the gas density. 25 WO 2009/081091 PCT/GB2008/003971 H. Pipe Conductive Heaters A key issue accordingly to Bruno (M.J. Bnmo, Light Metals 2003, Proceedings of the Technical Sessions, presented by the TMS Aluminum Committee, 132 "D TMS Annual Meeting, 2-6 March 2003, pp. 395-400) to be solved for carbothermic production of aluminium is how to deliver energy efficiently to attain temperatures of 2000-2200"C. This upper temperature can of course be reduced at the expense of increased melt circulation rate as already it has been shown that the melt circulation energy requirement is of minor significance in comparison with the endothermicity of the chemical reactions involved. The preferred design presented reflects sizes of graphite commercially available from UCAR and material currently posted on the Internet by SGL Carbon Group concerning rigid graphite felts for thermal insulation of high-temperature furnaces. In particular, it is noted that Sigratherm* RFA rigid felt composite is stable under inert atmospheres up to 3000 0 C but above 2200*C physical properties will change, i.e. thermal conductivity will increase and the material will shrink I. Heavy Current Distribution Current distribution is based solely on an aluminium bus system with molten pure aluminium at one end of current distribution launders at say 700*C and by controlled heat removal solid aluminium at the other cooled down to around 300*C. Conventional aluminium busbars are welded directly to the solidified Al in the launders and thus virtually eliminate contact resistance altogether. The concept depends on establishing a hub into which five, for example, short (say 4 m or less) launders containing molten Al at 700*C are connected with a single launder, which carries all the current to the close-by high temperature melt. To prevent passage of carbon-saturated melt by convection and to counter molecular diffusion of carbon back into the hub and subsequent accumulation of precipitated A1 4

C

3 in the cooler regions leading to blockage and serious restriction of current flow, a very small flow of aluminium must be established in the reverse direction. This can be achieved by melting scrap in the hub or alternatively using existing electromagnetic or submerged centrifugal pump technology to return a small "reflux" of the molten aluminium product to the hub. In both cases, to protect the graphite lining of the very short launder connecting the hub to the principal melt circulation, it will be necessary to incorporate some fine carbon particles into the melt using a single impeller drawdown. This is not to be seen as extra carbon consumption, but rather diversion of a small fraction of the make-up carbon away from the principal addition region. J. Saltation Velocity of Carbon Particles For horizontal transport of dispersed solids the minimum fluid velocity for keeping all solid particles in suspension is referred to as the saltation velocity. If the melt circulation velocity is reduced to the saltation velocity the largest particles begin either to settle or float out of the dispersed state. To a first approximation the carbon particles can be expected to have a density of about 2150 kg/m?, whereas the density of carbon-saturated 26 WO 2009/081091 PCT/GB2008/003971 aluminium at 2185*C is estimated to be 1945 kg/m 3 . in which case carbon particles will begin to sink to the bottom of a melt flowing in a open channel, such as the MPR, once the velocity is reduced below the salvation velocity. In the event of concern about large particles dominating the particle size distribution of the dispersed carbon particles, somewhere in the overall circuit would need to be designed so that the cross sectional area of the melt channel is increased to reduce the velocity down to the saltation velocity for say 50pn diameter carbon particles, for example, and then sufficient residence time provided at the lower velocity to permit the required amount of the largest particles to settle out to form a bottom layer or sludge. Preferably the sludge would be continuously thickened possibly with mechanical assistance. An appropriate amount of solids in the sediment bed of coarser particles is withdrawn in association with a predetermined amount of aluminium melt, probably corresponding to the product aluminium make. External milling of the larger carbon particles would be necessary in this scenario. Alternatively, in the longer term, synthetic diamond or very hard diamond-like carbon grinding media may become commercially available to realistically permit in-situ attrition milling of the settled sludge in an in-line vertical stirred mill. Also there is an element of uncertainty about the particle/melt actual relative density. Rather than carbon particles settling out to form a sediment, they may float upwards to form a surface dross. None of the actions in the preceding paragraph need to be implemented, if in fact the actual or mathematically modelled particle size distribution remains adequate for the protection of graphite surfaces. A decision on this therefore has to be deferred at this stage. K. In-plant Power Generation Consider now recovery of energy from the by-product CO in the cooled off-take gas from the MPR. In theory, the cooled gas contains about 98% CO and 2.0% CO, and at first sight may appear a premium fuel for combined cycle gas turbine (CCGT) power generation. The LHV of CO is 283 kJ/mol compared with the LHV of hydrogen at 242 kJ/mol. However, combustion of CO is far from straightforward and limited by serious kinetic constraints, which preclude comparison with H2 and even dismiss altogether pure CO as a gas turbine fuel unless it is mixed with H 2 . For this reason it is essential to incorporate a shift reactor into the flowsheet using superheated steam generated in-plant A conventional approach to advanced power generation using well-established technology has been undertaken by the applicant and is to be submitted for publication in a technical paper on lower energy primary aluminium production. Overall energy consumption figures 30.9% lower than today's most modem Hall-Heroult electrolytic plains are predicted with 52.5% less purchased electricity, supplemented with 1.06 times stoichiometric elemental carbon. Whereas particular embodiments of this invention have been described in the above for purposes of illustration, it will be obvious to those skilled in the art that variations of the present invention may be made without departing from the invention as defined in the appended claims. 27

Claims (11)

1. A process for continuously producing, under steady-state conditions of temperature and pressure without cyclic variation, primary aluminium by carbothermic reduction of refined alumina within a metal producing reactor, which operates above a temperature of 2150*C so that a separate phase containing aluminium carbide is not thermodynamically stable, thereby facilitating direct interaction of the alumina feed material with a continuous melt phase of carbon-saturated aluminium containing dispersed carbon particles typical in the region of 0.01 mm to 0.05 mm (a slurry of carbon particles in liquid aluminium) normally with 5-15 volume % solids, which is force-circulated through an external closed loop or loops, in which the carbon-saturated aluminium and its dispersed carbon particles are directly heated electrically by passage of an alternating current through the melt in order to satisfy the thermal demands of aluminium metal production within the metal producing reactor on returning the melt slurry from the external melt circulation loop to the lower back-mixed temperature of the intensive reaction zone.

2. A carbothermic aluminium process according to Claim 1, in which the metal producing reactor operates typically at a pressure of 2-5 bar to reduce to manageable proportions the adverse energy implications associated with concurrent evolution of aluminous vapours (Al 2 0) and Alw), which correspond to about 60% of the aluminium content of the alumina feed, if the process were to be conducted at atmospheric pressure, but reduced to about 26% at 5 bar pressure.

3. A carbothermic aluminium process according to Claim 1, which benefits from the enhanced mechanical properties of graphite and carbon fibre reinforced graphite or carbon at elevated temperatures by having the metal producing reactor temperature typically in the region of 2150 2170C, whilst permitting somewhat higher maximum temperatures, normally in the region of 2200*C in the external melt circulation system and hence availability in sensible heat terms of temperature rises of 30*C to say 50*C for transporting the thermal energy requirements to the metal producing reactor and accordingly dictating the mass flow rate of melt necessary to satisfy the thermal demands of the chemical reaction endothermicity for metal production from A1 2 0 3 , A1 4 C 3 thermal decomposition and evolution of aluminous vapours.

4. A carbothermic aluminium process according to Claim 1, in which graphite, baked carbon or carbon fibre reinforced carbon/graphite can be used throughout the process for melt containment and for ancillary items, such as filters, impellers, screw conveyors, rotary valves and forced circulation melt pumping systems such as submerged centrifugal, electromagnetic or gas-lift pumps, because of the protection against solution provided by the excess of fine carbon particles always stably present, which ensure availability of a kinetically more favourable option to bulk surfaces for maintaining local saturation of the aluminium melt with carbon at all times throughout the entire melt circulation system. 28 WO 2009/081091 PCT/GB2008/003971

5. A carbothermic aluminium process according to Claim 1, in which the off-gases from the metal producing reactor are partially quenched down to 1860-1960*C with recirculated carbon monoxide by entrainment into turbulent axially symmetric circular or slot jets well removed from solid surfaces to preclude potential accretion problems associated with initial deposition of sticky solid particles formed as the off-gases containing Al 2 Og) thermally decompose in the CO atmosphere as the temperature is lowered.

6. A carbothermic aluminium process according to Claim 1, in which the off-gases from the primary partial quench referred to in Claim 5, are cooled further for efficient energy recovery using multi-stage countercurrent gas/solid contacting in the entrained state employing venturi4ype contactors so that the alumina feed and associated reverted A1 2 0 3 , carbon and A1 4 C 3 stemming from thermal decomposition of aluminous vapours, arrive at the metal producing reactor for distribution above the melt surface near the melt slurry return region of the metal producing reactor at a nominal temperature of 1850C or preferably up to about 1950"C, whilst the off-gas at the cooler end of the multi-stage is reduced down to around 1000C before passing to a heat recovery steam generator.

7. A carbothermic aluminium process according to Claim 1, in which preheated Al2O 3 feed material, and A1 2 0 3 carbon and A1 4 C 3 formed during cooling of the off-gas from the metal producing reactor, discharge from the cyclone separator at the hot end of a multi-stage verturi contacting system at a nominal temperature of at least 1850C and preferably up to about 1950*C into the gas free board immediately above the highly intensive reaction zone within the metal producing reactor, which results from the evolution of three moles of carbon monoxide gas per mole Al2 3 combined with the high turbulence of the circulating melt slurry.

8. A carbothermic aluminium process according to Claim 1, in which the metal producing reactor is above atmospheric pressure in accordance with Claim 2 but elsewhere throughout the whole circuit the pressure is returned to atmospheric pressure by incorporation of so-called "barometric legs" at the melt slurry inlet and outlet ends of the metal producing reactor, thus simplifying impeller drawdown of fine carbon particles into the melt, gas-lift or equivalent pumping, provision for differential thermal expansion of the lengthy external pipe loop system for melt circulation and electrical current input via heavy duty hubs from a three-phase AC mains electricity supply.

9. A carbothermic aluminium process according to Claim 1, in which physical solid/liquid phase separation is conducted above 2150"C on the requisite amount of melt slurry to yield the equivalent amount of aluminium product in the form of carbon-saturated aluminium corresponding to the aluminium in the alumina feed, whilst leaving the associated carbon particles at operating temperature and thus available for recirculation in the carbon-saturated aluminium without incurring a thermal penalty. 29 WO 2009/081091 PCT/GB2008/003971

10. A carbothermic aluminium process according to Claim 9, in which a proportion of the carbon-saturated aluminium at the metal producing reactor temperature is contacted with spherical pellets of A1 4 C 3 coated carbon or graphite within a recirculated moving packed bed pellet decomposer to provide effective heat and mass transfer to cool the carbon-saturated melt, whilst depositing an A1 4 C 3 coating on the pellets as the melt cools and the pellets continue their downward progress to eventually deliver A1 4 C 3 coated pellets at the hot end of moving packed bed directly into the circulating melt in the metal producing reactor for thermal decomposition of A14C 3 and provision of cleaned pellets stripped of their A1 4 C 3 coating ready for removal by a screw conveyor or similar device for thermal energy recovery using countercurrent inert gas/contacting in a moving packed bed and then recycling of pellets back to the top of the pellet decomposer, whilst at the cooler end of the countercurrent contact pellet decomposer delivering a low-carbon liquid aluminium product.

11. A carbothermic aluminium process according to Claim 9, in which as an alternative to Claim 10, conditions necessary for sedimentation of larger carbon particles 50pm or greater are provided in an enlarged cross-section area region within the melt after intensive gas evolution reactions have subsided and well removed from the impeller system for carbon particle drawdown, and then the "thickened sludge" of larger carbon particles or "seed particles" in carbon saturated aluminium at a temperature above 2150C is continuously withdrawn at a rate equivalent to that necessary for ultimate aluminium product retrieval and quenched immediately in a vast excess of Al melt in a melt circulation loop at about 1000"C and given sufficient retention time within the melt to secure A1 4 C 3 growth on the carbon "seed particles" to assist solid-liquid/liquid separation using a vibratory screen, filter or other appropriate means in advance of reversion of preheated A1 4 C 3 via countercurrent radiative heat exchange on a belt conveyor to the melt circulation reactor and recovery of low-carbon aluminium product from the "filtrate" resulting from the physical solid/liquid separation device. 30