Low intensity continuous copper smelting Noel A. Warner Emerit Professor Univers ty of Birmingham B15 2T' United K ingdom 11 December 2009 1 Low intensity continuous copper smelting This invntion relates to production of molten copper continuously from mineral sulphide concent ates, containing principally compounds of copper and iron sulphide. More particularly, the present invention is concerned with providing a practical means for treating high energy content sulphide minerals such as chalcopyrite. The objective is to provide an energy efficient means for establishing, maintaining in stable operation, taking off-line for temporary or planned shutdown and restarting a process employing a number of melt circulation loops. These loops are arranged to effect integrated smelting of concentrates to high-grade matte and converting to molten copper on a truly continuous basis. Highly intensive reactors, such as Noranda, Tienente, Mitsubishi, Isasmelt and Ausmelt, do not capture :ffectively the inherent energy in mineral sulphide concentrates to the full extent, which iE clearly vital for energy efficient processing. Vast amounts of energy are wasted by the need to install water-cooled copper elements to protect refractory linings and/or feeding in moist cl arge materials to control dust losses. Current technology, therefore, normally needs a degree c f oxygen enrichment to ensure autogenous operations even when smelting high-energy content :opper and/or nickel concentrates. Ideally, when such materials are smelted, excess energy should be made available. The alternatives include co-smelting with lower energy sulphide s or energy recovery by steam-based electric power generation to satisfy in-plant requirements and possible export to national grids, if such are available. High intensity within the process reactors, referred to in the preceding paragraph, precludes straightforward high-pressure steam generation. The full temperature difference to cooling water is normally needed to provide the driving force essential to satisfy the very large thermal fluxes necessary to protect refractory linings in the high temperature aggressive environments within s ich reactors. Accordi g to U.S. 5,607,495, a process is described for oxygen smelting of a copper/nickel sulphid ore concentrate of high intrinsic energy value, comprising the steps of forcibly 2 circulating a molten sulphide carrier composition through a closed loop extraction circuit from wt ich copper/nickel or sulphide(s) thereof can be continuously extracted at an elevated tempe re, introducing the ore concentrate into the molten carrier composition at an ore receiving station so that the ore is dissolved in or melted by the composition, contacting the molten carrier composition containing said ore with an oxidising gas containing at least 30 vol % o ygen at an oxidation station so as to oxidise at least part of the ore and/or the molten carrier composition, and utilising heat generated during the oxidation step as a result of oxidation of the ore concentrate. Most preferably in U.S. 5,607,495, the oxidising gas is technic lly pure oxygen, although an oxidising gas having a lower oxidising potential may be employed if circumstances permit. In one e bodiment of U.S. 5,607,495, the generated heat is utilised by smelting the copper/nickel sulphide ore concentrate of high intrinsic energy value with another mineral concent -ate of low or negative intrinsic energy value, e.g. high-grade zinc concentrate, high grade le d concentrate or even a bulk flotation concentrate containing both lead and zinc. Concen rates are preferably low in gangue oxides for highest thermal efficiency, but not necessaily restricted to high -grade concentrates if metallurgical efficiency dictates otherwise. With such a process, metallic copper, metallic zinc and metallic lead may be co produce I in the primary smelting circuit employing forced circulation of copper/nickel sulphide through various extraction zones, following the teachings of GB-A-2048309. The primary zinc and/or lead formation reactions consume thermal energy and so, if the ore concent ates are added in the correct proportions, the excess energy released on direct smeltin, of copper/nickel concentrate using technically pure oxygen can be balanced against the endc thermic requirements of zinc and/or lead production. This has the advantage that the energy required for zinc and lead production is provided in situ within the smelter so that no external fuel is required and all the benefits of virtually zero gas emission smelting are secured. Preferably, copper is extracted as the metal, whilst nickel is extracted as high-grade nickel s lphide. In contradistinction to the preceding paragraphs, the present invention teaches that for high copper end/or nickel sulphide ore concentrates of high intrinsic energy value, use of oxygen enrichment or technically pure oxygen should be avoided in the interests of energy efficiency, if the sole purpose is to produce copper in the most efficient way possible. However, certain steps must be taken at the process design stage. This invention identifies 3 areas warranting close attention and proposes new technology to secure the objective of maximum energy utilisation efficiency for copper smelting. Notwitk standing energy considerations, it is recognised from the outset that a move away from hih intensity reactors could significantly increase the amount of saleable material held up with n the process. For example, if there is a significant increase in the hold-up of molten copper matte within the circuit, this will have immediate financial implications. Interest will be lost (n the equivalent value of copper metal not sold to customers. For many years the inventor has been promoting swimming pool size reactors in association with generic melt circulation technology. Clearly, to minimise the hold-up of matte, the first pre-requisite, if large su-face-area reactors are dictated on other grounds, the flowing melt streams must be very shallow. In the present context, emphasis is directed towards melt depths in top blown zones not to exceed about 5 cm or so in swimming pool reactors. The corollary to the above is that top air blowing must avoid jet penetration into the melt and be cond icted in what is known as the non-splash mode. The critical conditions necessary to enforce this requirement have been the study of numerous research projects over the years. For the >resent purpose, based on available data for melt properties and likely scenarios for multipk top jetting of molten cuprous sulphide, the critical depth of the melt cavity before splashing commences is in the region of 1.5 to about 1.75 cm. Accordingly, if the melt stream is fully turbulent with a Reynolds Number exceeding say 4,000, a melt depth of 5 cm is entire y feasible. Recently Campforts et al. (Metall. & Materials Trans. B, 2007, vol. 38B, pp. 841-851) have expressed the view that the formation of a freeze layer has to be guaranteed for high intensity process s conducted in furnaces with refractory walls. Such an approach is compatible with expert o pinion, as expressed by Donaldson et al. (Proc. International symposium on Non ferrous Iyrometallurgy, Edmonton, Canada, August 23-27, 1992, Metallurgical Society of CIM, pr. 491-505), for modem high productivity pyrometallurgical furnaces, "... bottom leaks in ariably lead to catastrophic run-outs and because the hearth brickwork is not readily accessible for repairs, proper attention to the design and erection of the hearth arches is the single n ost important aspect of furnace construction." In the context of continuous copper smelting, this means that conditions are necessary by appropriate design to ensure a freeze lining of cuprous sulphide exists throughout to protect the hearth or other melt containment walls in their entirety. Also as matte oxidation generates iron oxide containing slags, the 4 sidewal s clearly need protection by a freeze lining, which is preferably contiguous with that on the arth or containment vessel. Unlike existing electric matte smelting furnaces, a matte oxidation reactor is not protected above its thin slag layer with a reasonably thick charge layer, which is customary in matte or ferronic el furnaces. Special attention is therefore necessary to accommodate differential expansi n of the roof refractory system exposed to very high temperatures, which may well be in the region of 90-130 m in overall length. An analogous problem arose in the inventor's recent p paper on co-production of steel and titanium (N. A. Warner: Trans. Inst. Min. Metall. C, 2007, vol. 116 No. 1, pp. 34-47). This challenge was met using a flat suspended arch contain g top blowing nozzles above an oxidic melt contained within a hearth comprised of solidus omposition frozen oxycarbide melt at about 2,000"C. An anal gous arrangement can be used in the matte oxidation reactors associated with continuous copper smelting. In this particular scheme, the whole roof assemblies are supported by steel pontoons attached to joist girders and floating in launders containing a dense relatively non-volatile aqueous solution or a fusible alloy, one on each side of the full length of the furnace enclosures. Conventional refractories for the hearths would no longer be used but rather melt containment would rely on the stable lining of unmelted cuprous sulphid4 encased in a steel shell in association with steam generation to satisfy the heat removal requirements, as already referred to in a preceding paragraph. For low intensity continuous copper smelting, there are two principal melt circulation loops. The firs essentially replaces the continuous high intensity matte producing reactor systems of curre: t technology. To ensure effective removal of minor impurity elements such as arsenic, antimony and bismuth, this first melt circulation loop has a carrier melt of cuprous sulphide with only a minor level of ferrous sulphide but most importantly a relatively small thermo namic activity of dissolved copper metal. This is where the incoming moderately preheated d (before chalcopyrite decomposition) copper concentrate feed is dispersed into the slag, pr duced initially on the oxidising side of the melt circulation loop and then floated into the neut -al or reducing side. To effect dispersion, mechanical agitation or alternatively inert gas spar ing is required and then phase disengagement must be achieved by gravity separation, possibly enhanced by emerging electromagnetic means. This is an extremely effective procedure for reducing the copper oxide dissolved in the slag and can be regarded as in sitl slag cleaning to ensure minimal losses of copper in discard slag. 5 The mel t is then returned to the matte oxidation zone containing the extensive array of preheated air top-blowing jets. It is crucially important that the positioning of the jets in terms of spacing and height above the shallow melt surface is such that at least 90 % oxygen utilisation is achieved, whilst operating in the non-splash mode. To meet these demanding specific itions, a large surface area is required. The second melt circulation loop replaces the multiple batch Peirce-Smith converters of the traditional converter aisle and it again is low intensity in terms of the air top-blow arrange ents, which essential mirror those already described for the first melt circulation loop. T1e matte in this loop is copper-saturated (thermodynamic activity close to unity) and a separate bottom layer of molten copper is formed, which flows by gravity down a gently sloping earth of frozen cuprous sulphide so that it does not accumulate in the shallow hearth itself but rather is collected as a pool down one end. In the event of a temporary shutdown electric l conductive heating can keep the shallow layer of molten cuprous sulphide matte, again a out 5 cm or so in thickness in the liquid state until melt circulation is resumed. Molten opper siphoned out or otherwise withdrawn continuously from the accumulated pools re erred to in the previous paragraph becomes the feed to two relatively intense dispersed gas liquid/contactors in series based on non-wetted irrigation of packed beds. Counter current flow of inert gas with controlled minor addition of air in the first packed bed is follow red then by reformed natural gas as the continuous phase in the second packed bed. The liq- id copper irrigating these packed beds goes through liquid seals at the bottom of the column, accommodating each of the packed beds to maintain the desired gas composition during c:ntactors without air infiltration before emerging into the next in-line processing step. That is, the second packed bed contactors for the first and continuous anode casting, probably Hazelett casting machine, for the second. In this arrangement all melt transfer launders operate continuously with an inert protective gas atmosphere at a pressure just slightly ibove atmospheric to prevent any possibility of air infiltration. The purpose of the first gas liquid contactor is to remove the residual sulphur on the blister copper by the reaction S + 0 = SO2g), whilst the second reduces residual oxygen down specific tion limits by the two reactions: 6 o + H 2 (g)= H2O(g) and 0 + CO(g) = C 0 2(g), where the underlining represents the elemental species dissolved in liquid copper. It is to te appreciated that another embodiment downstream of the two low intensity melt circulation loops would reduce residual sulphur in the copper by effectively degassing under reduced pressure by the reaction S + 0= SO2(g), analogous to steel vacuum degassing in the Ruhstahl Hereaus (RH) process. RH vacuum degassing is now practised worldwide for decarbu station in the steel industry in its original form or for ultra-low carbon steel product on the more recently introduced processes RH-OB (oxygen blown) or RD-KTB (recircu ating degasser with Kawasaki top blow lance). Followi :g the final matte melt circulation loop, the melt will contain an unacceptably high level of dissolved cuprous oxide if left untreated. There are several options available for conduct ng deoxidation of the melt, of which the most promising candidate is considered to be dispersed contacting in a packed bed in counter flow with a continuous phase of reformed natural Ias. An em diment of the present invention will now be described, by way of example only, with ref rence to the accompanying diagrams, in which: Figure 1 is prior art disclosure of a molten metal siphon essential for successful implementation of the present invention, illustrating: (a) a schematic sectional elevation of a molten metal siphon connecting two furnace hearths at approximately the same level, which constitute adjacent arms of two different melt circulation loops, it which the product flow is very small in proportion to the forced melt circulation impose by other means individually on both loops. (b) a sc 1ematic plan view of the arrangement shown in (a) illustrating the molten metal siphon straddlir g the corners of the two hearths containing molten metal in two side-by-side rectang lar channels (not shown in full). (c) a schematic arrangement of the gas-cooled area of one leg which is submerged just beneath the surf ce of the melt to effect the establishment of a sealant rim of frozen material, which preclud s adventitious ingress of the ambient furnace gas atmosphere into the siphon legs. 7 Figure 2 is a flowsheet plan view showing the essential features of low intensity continuous smelting of charopyrite concentrates plus silica flux to produce a liquid copper intermediate product ready for continuous anode casting. Figure 3 is a schematic plan view of the overall plant comprising three melt circulation reactors of which the initial smelting of concentrate is conducted using two melt circulation loops (loop 1 type) in parallel, each feeding via a siphon a single melt circulation converter (loop 2) to effect c nversion of high grade matte to copper metal. Figure 4 is a cross-sectional plan showing a loop I melt circulation reactor for primary high grade mtte formation from preheated concentrates. Figure 5 is a cross-sectional plan of the loop 2 melt circulation reactor or converter, which receives high-grade matte continuously from the two primary smelting reactors. Figure 6 is a cross-sectional elevation of the air top-blown smelting arm of a loop 1 melt circulati n reactor. Figure 7 is a cross-sectional elevation of the air top-blown converting arm of the loop 2 melt circulating reactor. Figure 8 is a schematic sectional elevation illustrating application of a prior art siphon to the particular r case of continuous copper smelting. In this context, matte is automatically transfer d continuously from the first of the melt circulation loops, the matte producing furnace o the second of the two melt circulation loops, the converter. Referring now to Fig. 1, vacuum lift and gas lift pumping of molten steel is now a mature technology as evidenced by the widespread use of what is known as RH or circulating degassing. There is no difficulty in adapting such technology to the massive forced circulation of melt within a melt circulation loop. Current technology covers the range from about 20 tonnes t 200 tones per minute or even higher. Such high circulation rates mean that there is a vast amount of sensible heat in the circulating melt and therefore reasonable heat loses can normally be tolerated without risk of the melt freezing. This is not, however, the case for the actual flow or throughput of product stream, where for example, a 100,000 tonnes per year of 8 metal pi oduction equates to a mere 0.19 tonnes per minute. Accordingly, technology is require to ensure that the product stream can be continuously transported between reaction or refining steps or eventually continuously tapped as product. Figure I (a) shows a molten metal siphon connecting two furnace hearths at approximately the same le el, which constitute adjacent arms of two different melt circulation loops, in which the product flow is very small in proportion to the forced melt circulation imposed by other means individt ally on both loops. The shallow pools of melt 1 and 2 are interconnected, by a siphon upleg 3 and a siphon downleg 4 by a reservoir body 5 into which melt is sucked by maintenance of a reduced pressure connection to a vacuum system (not shown) and in which the leve of melt is maintained just below an electrical heater 6 fabricated from graphite or similar material, which is controlled so that the melt is kept molten under all conditions, whether that be in actual production or in a stand-by mode. The profile of the siphon assembly in eleva ion is as low as possible, perhaps an inverted Y configuration as shown in Fig. 1 (a). The sipt on upleg 3 and siphon downleg 4 dip into the melt pools 1 and 2 only so far as is required for practical operation but, in any event, deep enough to ensure that the ambient gas atmospf eres 7 and 8 are not permitted to be drawn into the melt contained in the legs within relatively narrow concentric conduits 9, 10 encased in dense refractory material 11, probably monolithic. Admission of gas into either leg will result in a decrease in effective density and the liquid seal and siphonic action will cease, be impaired or at worst flow reversal will take place wih the siphon acting as a gas lift pumping device. The melt is contained in hearths of residual solid melt in Fig. 1 (a), which means that the melt temperature has to be close to the liquidus temperature and therefore any temperature drop has to be minimised or reduced to an absolute minimum. This is particularly the case because of the relatively very low melt throughputs involved. Therefore, positive steps have to be taken not only to add heat to the liquid reservoir body 5 via the electrical heater 6 as already referred to, but also the melt in both leg has also to be heated to counter inevitable heat losses by passage of a low-voltage heavy c rent through the melt to effect conductive heating. Figure 1 (b) shows the means for introducing the current is via at least one electrical conductor 12, which is connected to a mains frequency electrical supply such that a voltage difference can be maintained between the melt in the reservoir body 5 and the liquid pools 1 and 2 so that control ble heavy current can flow in parallel to the required extent individually through both 9 legs, w ether the melt is in motion or is merely held-up in the standby mode, if for operational reasons melt flow should inadvertently cease or otherwise be stopped as part of a planned shutdo n. In certain circumstances phase differences can be used to establish the required potentiaI difference and in others a "floating voltage" will need to be introduced via the current input c nductor 12. The design of the current input conductor 12 is crucial to the successful operatic of the molten metal siphon. There are two conflicting considerations. Firstly, current 1as to be introduced at relatively low current density with minimal electrical contact resistance. Second, the high interfacial area implied to effect low contact resistance means that a relatively large area is available for conducting heat away from the reservoir body, placing undue d mands on the electrical heater 6. Accordingly, the hot end of the input conductor 12, itself ha s to be at the liquidus temperature with a frozen crust being established in contact with the melt and according to operational requirements, either intentional or otherwise, the thickneE s of the frozen material will either grow or shrink to reach steady state. This is achieved by the provision of a computed length of copper for the input connection, such that controlled cooling of the exposed cooler end does not result in excessive heat losses, but provides a welded interface between the current connector and the melt. Forced Cooling of the annular steel tip 14 is achieved by gas impingement at high velocity through an arrangement of small bore metal tubes, two of which 15 and 16 are shown in Fig. 6 or alter atively via a slot placed adjacent in close proximity to the gas side of the tip 14. It is import it to ensure that the other side of the tip 14, which is in contact with the melt, has unimpaired access to the melt at the instant when the siphon legs 3 and 4 are first submerged. If a frozen layer is not immediately formed, either because the melt freezes away from the exposed tip or, alternatively, the exposed solid metal tip is contaminated such that a coherent bonded deposit is not formed instantaneously, then gas ingress into the liquid metal conduits of the snor cels becomes a strong possibility and siphonic action will either not be initiated or will be terminated if already commenced. It is the integrity of the solidified metallic non-porous seal est blished on the annular metal rim, which is the key to success. Provided this is attained gas ingress through the refractory block 17 from the ambient furnace atmospheres 7 and 8 is not possible, provided the metal rim itself is fully submerged in the melt 18. The refracto y block 17, which must be provided in order to protect other steel surfaces, remote from the gas-cooled rim, from excessively high temperatures in the case of melt operations, for example, is preheated to as close as possible to the temperature of the melt by holding the siphon j ist above the melt surface for as long as necessary before immersion. Also the 10 refracto y block in the immediate vicinity of the exposed tip is profiled so that melt enters the region through an entrance area, which is greater than the area of the exposed steel on which the mel is to be solidified. Finally, steps must be taken to ensure that the refractory block 17 is submerged only to the extent necessary to provide complete submergence of the solidified metal ri n formed on the steel tip 14. In certain operations, the snorkels are fabricated with double walls from concentric steel pipes with only a relatively small clearance between them to provide an annulus for coolant gas leaving he gas impingement area to return to a header or discharge manifold located remote from the high temperature region, probably via a double wall construction for the reservoir body, fcr heat extraction so that mild steel or low alloy steel can be used throughout rather than more expensive heat resistant alloys. To bring a new or replacement metal siphon on-line use is made of sacrificial heating elements to preheat both arms of the siphon as well as the reservoir body. Steel pipes or rods can be inserted into the legs and connected together either within the reservoir body or externally, with the bottom contact finally being made by dipping the rods or pipes into the melt pools once the refractory blocks and snorkel externals have been preheated above the melt surface. Electrical conductive heating can then be undertaken, but to reduce to an acceptable level the associated thermal shock, it will probably be necessary for the first part of preheating to be conduct d with the sacrificial pipes or rods connected together with a bridging conducting member, whilst preheating above the melt takes place, but then to accept destruction of this member once the snorkels are lowered into the melt. For melts close to the liquidus tempera ure, dissolution or melting of sacrificial pipes or rods will be delayed long enough to achieve a final top-up preheat, before sucking up the melt into the reservoir body to commence siphonic action with the full input then of the conductive heating circuit, as well as that provided by the heater within the reservoir body itself. Referring to Fig. 2, as an example, to produce 250,000 tpa copper product, initially a single matte oxidation reactor equivalent to around 180 m in width by 20 m in length was indicated to be re uired. However, from the viewpoint of balancing three phase AC electrical power input for conductive heating, when continuous operation is temporarily halted, it may be preferable to have three matte oxidation reactors arranged in parallel with each reactor 11 around iO m wide by 20 m long, for an overall plant to produce to 250,000 tpa copper metal product Further ore, to make the above a practical proposition, a matte circulation rate in the region of 150 nol Cu 2 S being circulated per mol CuFeS 2 feed is necessary so that the slag formed, which is carried along floating on the circulating carrier melt of cuprous sulphide, moves at a sufficiently high velocity to guarantee retention of very thin slag layer. Clearly, none of these o jectives can possibly be reached in traditional copper smelting in the absence of forced relt circulation in cross-flow across a furnace hearth. The arr ngement shown in Fig. 2 was adopted as the basis for the detailed calculations presented d in the complementary submission, "Low Intensity Copper Smelting" filed on 22 July 2008. However, further consideration now indicates that the total holdup of matte in the circuit can be appreciably reduced, if two larger matte oxidation furnaces operate in parallel with both feeding high-grade matte via siphons into a single melt circulation converter. The alternative approach to produce 250,000 tpa copper product, shown initially to require a single c oss-flow matte oxidation reactor equivalent to around 180 m in width by 20 m in length, is preferably now believed to be satisfied by having two matte reactors arranged in parallel with each reactor around 90 m wide by 20 m in length together with a somewhat smaller single converter reactor and associated plant constituting the third electrical load. Referrir g now to Fig. 3, a preheated feed 19 of chalcopyrite concentrates plus the required amount Df silica flux is split into two equal portions and then added to the surface of the slag layer fk ating within relatively narrow launders containing the carrier melt of copper matte. Provided the feed concentrate is kept below about 430 0 C prior to addition, evolution of sulphur gas takes place with the thermal decomposition of CuFes 2 to Cu 2 S and FeS. The FeS and S2(g serve to in-situ slag clean the slag produced by air top blowing jets (not shown in the cross-section illustrated in Fig. 3) to yield discard slag 24 with minimal copper content. The disc ard slag 24 is tapped intermittently or possibly continuously from the system. The rectang lar hearth regions indicated by 20 and 21 constitute the swimming pool reactor systems In Fig. 3 the matte flows longitudinally in the narrow launders to which feed 19 is distribute ed. The matte flows transversely across the full width from bottom to top of the major rectangular hearth regions 20. As matte is produced in 20, the equivalent amount is added continuously by the siphon 22 to the high-grade cuprous sulphide matte circulating in the hear hs of a cross flow across the hearth region indicated by 21 in the converter melt 12 circulat on loop to produce liquid copper 23. This liquid copper 23 is removed continuously by a sip on (not shown) dipping into the pool of molten copper, which accumulates by gravity ;eparation from the over-lying flow of matte within the narrow transverse launder running across the full width of the cross flow swimming pool reactor 20. Referrir g to Fig. 4, preheated feed 19 is distributed to slag covered matte flowing in the launder 25. The in-situ cleaned slag is then discarded at 24 either continuously by overflow with he ting provision (not shown) or intermittently slag tapping. The matte flows along the launder 25 to underflow weir 32 to enter the gas-lift pumping system 28 and then proceeds to matte fl yw straightener or hub 29 prior to entering the launder system 26 extending to 31, which i visible in Fig. 4 but on addition flowing matte is not visible, because it is submerged by the 1 yer of liquid matte overflowing from the launder system 26 to form the hearth layer 5 cm or less in depth. This flowing hearth layer of melt (not shown in Fig. 4) discharges eventua ly into the launder 25, which is virtually across the whole width of Fig. 4, by means illustrated in the cross-section shown in Fig. 6. Refractory brickwork or alternatively monolith ic castable material is shown as 30. The equivalent amount of matte produced within the swimming pool reactor is siphoned continuously into the converter reactor system shown in Fig. 5. The siphon is removed in Fig. 4 but the site of the siphon up-leg is shown as a cross 33. The hearth floor visible in Fig. 4 is of welded steel slab construction throughout. It is crit cally important that the steel slab floor is perfectly level so that a constant matte depth of 5 cm or less can be accommodated on top of the solid layer of Cu 2 S, which constitu es the freeze lining covering the entire hearth area. Seen in cross-sectional elevatio 1, the steel slab construction forming the vertical walls containing the melt within the swimmi g pool reactor is shown as 27. A prefer red means for ensuring that the whole floor area of the hearth is perfectly level is to employ i self-levelling technique, in which the welded steel slabs with their associated heat removal means in position are subjected to high current electrical conductive heating. If heat is extracted from the walls and base of a very large solid slab of iron, while a heavy alternate g current is passed from one end to the other, in due course a liquid pool will be 13 formed, which can be maintained as such by controlling the current input. This, of course, is the rati nale behind the design principle of a generic swimming pool reactor system. An alte native preferred embodiment, if a self-levelling technique is indeed necessary, assumir g that more direct mechanical grinding and related means are inadequate to ensure the degree of levelness required, is also worth further consideration. It involves lower temperature electrical conductive heating of the welded steel slabs up to surface temperatures around 250-1300 0 C in order to melt an added top layer of cast iron containing carbon in the region cf 3-4%. At carbon saturation, the melting point is 1153"C at 4.3% C. With this alternative approach, the liquid cast iron layer so formed will be self-levelling without the need to aise the steel slab surface to steelmaking temperatures around 1540*C. Now re erring to Fig. 5, this converter reactor is virtually a scaled down version of the compare ble matte producing reactor system shown in Fig. 4, from which the melt feed to the c nverter is introduced individually from both primary smelting reactors from the down-legs of siphons at locations shown by the crosses 35. An important difference in the plant illustrated in Fig. 5 is the fact that the matte circulating is copper saturated. Thus there is little scope for further removal of impurities, such as As, Sb and Li. Also the amount of slag formed is minimal so no attempt at in-situ slag cleaning is pursued, but rather the relatively small amount of slag formed 34 is reverted to the principa smelting reactor shown in Fig. 4. As a cor sequence of the matte being copper saturated, further top blowing with air produces the metallic copper, which is subsequently removed by a molten metal siphon (not shown). The up-leg position for this copper removal siphon is shown by the cross 37. The slag covered matte flowing in the launder 25 goes over a refractory base gently sloping towards the cent e so that gravity separation of copper from matte is effected and an accumulated pool 36 f copper is formed. From this pool 36, the metal siphon continuously removes liquid cc pper from the melt circulation system to undergo in-line downstream desulph risation followed by deoxidation in gas/liquid packed bed reactors (not shown). Referrin to Fig. 6, the air-top blow ducts or nozzles 39 are shown for the first time in this cross-se tional elevation, which has the central portion of the plant removed from the 14 diagr so that the linear scale used permits identification of the principal components of the matte p oducing reactor. The ducts or nozzles 39 deliver preheated air to the surface 45 of the mol en matte circulating as a thin layer no more than 5 cm deep from right to left across the slab steel base 27 of the hearth and its associated melt distribution system. The philosophy behind of the plant illustrated is to ensure reliable melt containment by permanent linings of frozen matte 44 on steel slabs 27. Once the melt enters launders, etc., readily accessi le for maintenance when required, conventional refractory linings are employed. Thus the external accessible launder or pipe system involved in the gas-lift piping system, for example, is lined with conventional refractory materials as discussed in the complementary Applica ion No. GB 0813355.5. In particular, Rodriguez GonzAlez et al. (Metall. & Materia s Trans. B, 2007, vol. 38B, pp. 167-174) report that there is zero penetration of Cu 2 S into spinel direct bonded brick (71% A1 2 0 3 ; 25% MgO). Accordingly, the preferred embodii ent makes extensive use of this refractory in the overall plant. Continu ng with reference to Fig. 6, special attention is drawn to the "drain hole" 41 from which b th matte and slag overflow into the steel slab 27 lined launder 25. To maintain the stable thickness of frozen matte, which constitutes the freeze lining 44 of the hearth and associated launder system, heat is removed by steam boiler tubes to generate high-pressure steam f r advanced steam turbine cycle generation of electrical power. These boiler tubes are loca ed in the "cavity wall" 40, which also contains (not shown) the skid mounting or possibly bogey/rail system to accommodate thermal expansion as the temperature of the matte is raised to the operating temperature level, typically 1150-1210"C. The temperature of the freeze lining surface in contact with the circulating matte is typically in the region of 1100-11 20C. Details of a comparable system to accommodate differential thermal expansion are given in Fig. 3 of US 7,279,127. Viewed from this perspective, the slab steel hearth li ing and its associated freeze lining of matte is rather analogous to a very extensive "hot pla e". Appreciable movement is inevitable because of thermal expansion, so provision must be provided for free movement to take place. The gap between the hot plate and the refractoty lining 30 at the left-hand end of Fig. 6 schematically illustrates this requirement, without showing the skid mounted support structure needed in the region of the rim 42 of the drain ho e 41 and the melt containment wall 43. The precise number of such drain holes 41 required to maintain a reasonably uniform flow across the hearth width, in this example say 90 m, neds computer evaluation, but clearly the separation between the drain holes is 15 unlikely to be greater than a couple of meters. This whole area will eventually stabilise with frozen r iatte or slag on all steel surfaces exposed directly to matte flow. Contin ing with reference to Fig. 6, the gas free board zones 47 and 48 are separated into an inert ga; region 47 and an oxidising region 48 containing un-reacted air and sulphur dioxide. Arrangements for the required preheated air input to the header region 38 extending over the entire h -arth area are not shown in Fig. 6. Neither shown in Fig. 6 is the hot gas off-take system 'equired to transport the gases to a downstream processing plant, probably a sulphuric acid plant, if local circumstances makes this economically viable. Alternatively, some other sulphur fixation means, such as exothermic reaction with dolomite or limestone, if this is available locally, must be installed to alleviate atmospheric pollution and environmental degradaion. The ine t gas region 47 shown in Fig. 6 is necessary to prevent oxidation of the steam boiler tubes ar d differential thermal expansion system. Accordingly, provision must be made to separat these two gaseous regions by a controlled flow of purge gas, reduced to a minimum compati le with maintenance of the required non-oxidising condition in 47. This implies a small difference in absolute pressure between the slightly higher pressure region 47 and the more extension oxidising region 48. To contain this slight differential pressure without inordinz te consumption of inert gas, the level of the flowing melt must be rigorously controll d so that a minimum clearance gas exists over a prescribed length of refractory cover 30 and the melt surface flowing immediately below it. The pre eated air jetting down from the ducts or nozzles 39 flows through a configuration of either t angular or square pitched openings in the refractory ceiling, some 0.5 m or so above the melt surface 45, operating in the so-called "non-splash mode", as discussed in detail in GB 0813355.5. In addition, careful attention needs to be focussed on the detailed design so that the :hemical reactions involved in matte oxidation take place on a slag-free gas/liquid interface. Referring to Fig. 7, production of metallic copper in the converter must take into account vitally important modifications over what has already been described in Fig. 6 for matte producig reactors. These modifications need emphasising to ensure reliable operation of this unit Because copper is more dense than matte, the copper produced at the slag-free 16 gas/lig id interface immediately below the top blowing jets will sink by gravity to form dropletE of liquid copper on top of the freeze lining. These droplets of metallic copper will be flusl ed along the freeze-lining surface by the circulating matte and will eventually move to the configuration of drain holes, exemplified by 41 in the prior description given in Fig. 6, if these are not modified along the lines now to be stipulated. Without modification complete coverage of the hearth by 5 cm of liquid copper would eventually occur and once established would remain in place. This would have serious consequences on the electrical conductive heating necessary to maintain the melt layer in the liquid state in the event of a shutdown or cessation of continuous operation. Accordingly, the scenario just described must not be allowed to happen. The rim 42 of the drain hole shown in Fig. 6, which causes the 5 cm of melt to form in Fig. 6, must be removed for the converter shown in Fig. 7. In its place a configu -ation of bevelled drain holes in the steel slab base itself, must be provided. This allows copper to be flushed directly down continuously with the circulating melt into a launder 25. For the converter the launder 25 is provided with a sloping base so that copper separated out by gravity can flow down to a centrally located pool or sump of liquid copper, as shown by 36 in Fig. 5. Flow through the modified drain hole arrangement must now be controll d by what is effectively a large stopper rod, so arranged that downward pressure could v rtually preclude any flow from the hearth into the melt launder 25. Release of some applied pressure could then provide the discharge pressure drop necessary to establish the "hydrostatic" head. In other words, the required depth of 5 cm of matte could be maintained through ut the swimming pool reactor in a state of dynamic equilibrium. The actual pressure drop or "hydraulic" gradient for the melt layer to flow across the hearth is negligible in compare son with the 5 cm liquid head introduced by discharge of the stoppered matte plus copper roplets flushed along by the circulating mass flow along through and out through the multiplicity of partially stoppered drain holes. Continuing with reference to Fig. 6, the preferred embodiment makes use of the commercial availability of large diameter (up to 800 mm) graphite electrodes for electric arc furnace steel melting To form a graphite column, graphite electrodes are joined together by nipples for use in e ectric furnace steelmaking. Graphite electrode manufacturers deliver these electrodes with one end of the tapered nipple already installed in place. In the present context, the protruding tapered nipple provides the stopper for controlled matte discharge with the remaining electrode up to about 2800 mm in length contributing the downwards pressure due to its own weight. On first admission of a replacement electrode, melt will 17 freeze to form a protective lining on the exposed nipple, stabilised by heat removal up the electro body, which has an advantageous high thermal conductivity eminently suitable for the purp ose. This arrangement is shown in Fig. 7 with the electrode 49 displayed in a fictitious down position with the melt and steel slab discharge port removed from view in the diagrar for reasons of clarity. To effect flow control, the electrode 49 is mechanically lifted upwards or at least partially released on its own weight and may even be oscillated to preclude any tendency towards welding of the nipple with its freeze lining of matte with the freeze lining of the discharge port itself. The sch ematic arrangement in Fig. 7 also provides the opportunity to elaborate further on the provision of the two discrete atmospheres 47 and 48. The purged gas referred to previously in the di scussion relating to purge gas flowing from 47 to 48 through a minimal clearance gap bet een the top surface of the melt 45 and the refractory immediately above is shown as 50inFi .7. Referrir g to Fig. 8, use of the prior art metal siphon 51 is highlighted. This siphon 51 facilitates the continuous transfer of matte from a matte producing furnace 20 to the converter 21. The siphon up-leg 52 sucks up matte from the launder 25 beneath the over lying slag layer floating on the matte surface as it progresses at every increasing rate towards the matte underflow weir 32 shown in Fig. 4, immediately in advance of the gas-lift pumping system:!8, which is used as the driving force for melt circulation in both the matte producing furnaces 20 and the converter 21. As the matte progresses along the launder 25, the flow increases as additional matte flows into the launder along its 90 m or so length in a matte producing reactor and 60 m or so length in the converter. This occurs as the new air-blown matte in cross flow is progressively added to the cumulative matte flow proceeding towards the matte underflow weir 32. Matte is circulating at a rate of about 150 times that of the actual r te of new matte production. Concentration differences within the matte circuit are thus insi gnificant, so the addition of feed 19 for in-situ slag cleaning makes little difference to the overall composition of the matte being transferred by siphonic action. Fig. 8 is schematic as far as the plane section is not strictly adhered to but varied for clarity and press rented as illustrated. As can be seen in the plan view shown in Fig. 3, the up-leg position 33 is remote from the underflow weir 32 leading to the gas-lift pumping system 28 in both matte-producing reactors. Accordingly, there is little prospect of feed concentrates 18 contam nating the matte siphoned off from the matte producing reactors 20. This aspect is obvious ly important in preventing As, Sb and Bi, because once the matte is added by the down-leg 22 of the siphon to the launder 25 of the converter, it is immediately in the presenc of matte saturated with copper, with little chance of further reduction in the As, Sb and Bi contamination of the matte. Whether or not it would be preferable to slope the launder 25 in the converter 21 so that the copper netal accumulates at one end rather than the middle, as depicted in Fig. 5, is still a matter f some conjecture. Gravitational movement of settled out copper droplets is assumed to dominate what happens to the copper droplets, as opposed to a possible influence of friction-d drag exerted in the direction of matte flow. On this basis, central collection of copper :netal by siphon at position 37 as shown in Fig. 5 avoids obstruction with the two matte siphons, one at each end of the launder (sites 35 in Fig. 5) in the converter. Thus, the same g neral arrangement as depicted in Fig. 8 applies also to the matte siphon at the other end oft ie converter. The prior art features of the molten metal siphon given in Fig. I are repeated in Fig. 8 for the matte si phons under discussion. The up-leg 52 sucks up matte into the reduced pressure vessel 51 to establish a matte level 54 within the body of a siphon. The reduced pressure or vacuum connection is not shown in Fig. 8, but it is implied that the siphons remain connected to the source of the low-pressure system compatible with the desired lift of melt required. This is in recognition that any adventitious entry of permanent gas could destroy the siphonic action, if not pumped away immediately. The down-leg 53 delivers matte to the converter continuously, whilst maintaining the prescribed liquid levels in the two matte producing reactors The refiactory end caps 55 cast around the external steel piping forming both up-legs and down-legs facilitate natural convection entry only of matte to the critically important forced gas-coo ed steel surfaces 56 on which the dense solidified rims of matte or copper metal are formed, following the teaching discussed in relation to Fig. 1. In addition, it should be noted that the hermal resistance of refractory end caps 55 is important in protecting the external steel sui faces from attack with either slag or molten copper, again in accord with the teaching s of the prior art discussed in relation to Fig. 1. 19 Finally, the schematic representation of Fig. 8 is a definitive way to illustrate the relative location of the freeze layer of matte 57 in relation to the steel slab 27, the flowing matte layer 54 and he upper surface of the melt 45. The existence of this arrangement of the freeze layer of matt( 57 is paramount to the whole concept of the new low intensity copper smelting plant. Concluding Remarks Low in nsity copper smelting captures very effectively the inherent energy in chalcopyrite copper concentrates. Conventional refractory hearth and wall linings are replaced by permanent freeze linings in the two matte oxidation reactors involved in the technology, which uses air top blowing in the non-splash mode of circulating shallow melts of 5 cm depth or less bowing turbulently over freeze linings of cuprous sulphide. The major energy input is the inherent energy of the mineral sulphide undergoing oxidation at the gas/liquid interface on impz ct areas cleared of slag by jet momentum, but without jet penetration into the melt or splashir g taking place. By preheating the air to an appropriate level, most of the energy required to stablilise freeze linings Of solid cuprous sulphide is provided by the exothermicity supplemented by the enthalpy of the preheated air jets. Straightforward air preheating using a recuperative heat exchange system avoids the more expensive regenerative option. The hig -pressure steam generated under benign conditions to balance the thermal flux necessa y to freeze linings for electricity generation feeds a sub-critical steam cycle employ ng three pressure levels with single reheat. For 250,000 tpa copper production the net elec ricity generated amounts to about 28 MW. One interesting possibility, which would consume the newly generated electricity at an existing copper smelting site, is a doubling of smelter output from say 250,000 tpa to 500,000 tpa, whi st retaining the infrastructure still in place. Thus, for example, it can be shown that a 250,000 tpa Isasmelt reactor with a traditional Peirce-Smith converter aisle and the associat d air separation plant, already in place for oxygen enrichment, could have all its electric 1 requirement provided by a parallel plant of 250,000 tpa capacity based on the new 20 melt ci ulation technology. With this scenario, there would still be enough electricity generat d in-plant to electro-refine about 250,000 tpa of copper product. At first sight, the large cross-flow swimming pool dimensioned-reactors may appear to have an excessive amount of molten matte holdup and have inordinate heat losses in comparison with the thermal input of the chalcopyrite feed. However, the restriction on melt depth to 5 cm or le ss means that the holdup of molten matte is no more than a conventional flash smelter with two pierce-Smith converters on-line at any time and a batch-refining furnace for 400 ton le batches of blister copper. Finally, if quality specifications for arsenic, antimony and bismuth are tightened in the years ahead, t e first melt circulation loop is ideally placed to take this in its stride. This loop operates at low copper activity, so present specification levels for As, Bi and Sb can be met without employing a R-H type vacuum degasser (analogous to that used in the steel industry) for melt circulation. At present, all that is needed is a gas-lift pumping system for each melt circulation loop with the total head no more that 0.3 m or so for slag separation. It must also be reiterated that the energy involved in melt circulation is negligible in comparison with the thermal energy terms. 21