GB2401362A - Method for eliminating top speck on float glass - Google Patents

Abstract

Top specks on float glass are produced by hydrogen in the protective gas atmosphere (6, Fig 2) reducing solid stannous sulphide deposited on surfaces above the melt level to form tiny globules of liquid tin, which then fall back onto the glass surface 7. To secure improvements, which collectively prevent stannous sulphide being deposited in the first place, attention is focused on forced circulation of protective gas externally preheated by direct contact with comprehensively de-oxidised and desulphurised molten metal extracted from the bath at the hot end 10 so that the protective gas on its admission to the float chamber 5 is hot enough to sustain temperatures above the local dewpoint temperatures for stannous sulphide deposition for all surfaces above the float bath 4. Also to preclude stannous sulphide deposition within the accessible pore structures of conventional refractory roof and wall construction, these are all replaced by impermeable linings.

Description

240 1 362 APPARATUS AND METHOD FOR E:Lllilll\ATD\G TOP SPECK ON FLOAT

GLASS In the early days of commercialization of float glass manufacture, L.A.B. Pilkington in his address to the Royal Society in London during 1969, implicated sulphur contamination of the tin bath as the source of so- called top specks on the float glass ribbon in glass manufacture and presented a schematic diagram illustrating the so-called sulphur cycle involved. A more up-to-date account has been given by C.A.

Ealleroni and C.K. Edge in a PPG publication published in 1996. According to these authors, the sulphur cycle stems from the presence of sodium sulphate, an essential ingredient in commercial float-glass.

Sulphur in the molten glass is extracted into the tin bath from the bottom surface of the glass and is also volatilised into the bath atmosphere. This leads to deposition of solid tin sulphide in the cooler regions of the chamber, particularly towards the discharge end. Apparently, the net result, after hydrogen from the bath protective atmosphere reduces the deposited solid sulphide, is that small globules of metallic tin are formed especially on the roof and these then fall onto the glass surface. Up until the present time, the consensus amongst glass manufactures has been that the occurrence of tin speck can be controlled by a number of measures being taken, which are well documented, but with more advanced glass products, there is increasing awareness that existing surface defect control technology needs to be improved.

This invention attacks the root cause of the problem by ensuring that tin sulphide solids are never allowed to form in the first place. To achieve this objective, attention must be focused on elimination of transient excursions of operating factors that lead to enhanced concentration levels of gaseous tin sulphide in the atmosphere above the bath. The principal courses of these are related to the phenomenon of so-called backmixing and longitudinal mixing both in the tin bath and in the protective gas atmosphere above the bath. Positive flow of both the gas and liquid tin phases from the cold end to the hot end ameliorates the problem but does not necessarily totally eliminate it. The only certain approach is to ensure that all surfaces within the float glass chamber are above the "dewpoint temperature" for the deposition of solid stannous sulphide. Since the float glass is typically removed at around 600 C, this temperature dictates a minimum temperature level for exposed stationary surfaces throughout the entire chamber. However, if the molten tin bath is below the dewpoint temperature in order lo effect cooling of the glass ribbon down to 600 C, some stannous suiphidc possibly could be deposited thereon, but this is arranged so that it flows away towards the hotter end so any transient surface deposition is reevaporated almost immediately or if reduced to metallic tin is assimilated into the tin bath and, in any event, globules of metallic tin on the bath surface cannot contribute directly to globules landing on the top surface of the glass.

This invention arranges for the protective atmosphere to be preheated to at least 600 C, in the present example, prior to its admission to the float glass chamber. Following on the advice given by Falleroni and Edge, some 80Yo of this preheated protective gas flows immediately out of the chamber as the glass is withdrawn into the annealing leer, while some 20% or so progresses unimpeded above the glass to be removed in the vicinity of the glass inlet by side vents as discussed for example in US patent 3, 595,635.

The headspace must be of such dimension so that this positive gas flow is turbu1cut in order to induce gas mixing and at such a velocity that longitudinal or backmixing from the hot end back to the cold end are insignificant By these means, oxygen and sulphur desorbed from the molten glass, especially at the high temperature feed end, are immediately transported away with flowing protective gas and withdrawn from the float chamber for external removal.

Because the protective gas enters at around 600 C in the proposed process modification, there is little scope for cooling the glass by contacting with the gas phase. The cooling duty is accordingly shined very much towards heat removal from below the floating ribbon of glass directly into the tin bath. This implies circulation of the bath and external cooling so that it can be returned in a cooled and purified state back into the float glass chamber principally at or near the cold end.

From a purely heat balance view point, tin melt circulation with existing float glass technology is unlikely to be large enough to yield tin velocities of sufficient magnitude to combat the very extensive longitudinal mixing that must occur in an apparently static tin bath, resulting from the surface drag of the ribbon as it flows on the tin surface along the bath. This is exacerbated by stretching to produce glass thinner than the equilibrium thickness of about 6.5mm.

Attainment of higher positive tin now velocities countercurrent to the direction of glass ribbon movement, a prerequisite to major reduction in longitudinal mixing, without increasing the energy demand for the overall process, is now identified as a the key issue to be resolved. In the present invention, the approach is focused on energy conservation and an assessment of how a float glass chamber needs to be modified away from the traditional design. Better use has to be made of the energy input to the chamber in the form of sensible heat of the molten glass fed into the chamber, typically at a temperature of about 1050 C.

Consider now a typical modern float glass line with an output of say 800 tonnes per day using a molten tin float bath say 6m wide by 60m in length. If this plant is designed to produce laminated glass by the sticking together of two or more sheets using PVB as adhesive, the thickness of the initial float glass is likely to be in the region of 3mm and assuming that some 80% of the width of the tin bath is covered by glass, the withdrawal velocity at the end of the float glass chamber can be calculated to be about 0.26 m/s. Using the standard "operating package" advocated in the Falleroni and Edge paper, the bath atmosphere is 1-5% hydrogen and the balance nitrogen flowing at SO,000-100,000, scfl,, depending on the bath size, which for the present example is relatively large, so the upper figure is taken. If the headspace is Im, towards the hot end the established total gas flow will be turbulent, assuming that some 20% of the total gas flow is extracted by the vents and the average gas velocity is in the region of 0.6 m/s. This positive gas flow should go a long way towards reducing to acceptable proportions longitudinal bachnixing in the gas phase, so there should be little opportunity for gaseous stannous sulphide, evolved at the hot end when the molten glass is first added, reporting in the colder regions of the chamber.

The total external surface area of the float glass chamber can be estimated to be about I SOOm2. Now, if the traditional chamber design is modified to incorporate low thermal mass and highly efficient insulating materials, it should be possible to achieve a total heat loss by radiation convection approaching a target figure of 1.2 MW. The heat input associated with the molten glass cooling from 1050 C to 600 C works out at about 4.5 MW so once the line is in operation, it should not be necessary to add further heat by heaters incorporated into or near the ceiling of the chamber. The control of heat removal can then be achieved by circulation of molten tin principally from the cold end towards the hot end, with optional intermediate addition of circulated tin at various positions along the line as the control means, if process conditions so dictate. Clearly, the maximum heat to be extracted from the molten glass ribbon is the difference between the heat input of 4.S MW and the heat losses of 1.2 MW, which equals 3.3 MW. Engineering calculations reveal that only a relatively small amount of this heat can be removed by the flow of the bath atmosphere, if this is preheated externally to say 600 C before admission to the float chamber. To a first approximation, if the average temperature of tin withdrawn towards the hot end is about 850 C and is cooled externally to say 550 C before re-admission to the float chamber, then a tin flow rate of about 170 tonnes per hour is required. Furthermore, if the bath is about Scm deep, the average velocity for the total of the recirculated tin towards the hot end is 0.02 m/s. The corresponding Reynolds number is loll, indicating that the tin is flowing under streamline or laminar flow conditions. In this example, the relative velocity of the glass and tin liquid phases has been increased from a nominal 0.260 m/s to 0.262 m/s with melt circulation' a change unlikely to introduce complications in terms of surface quality of the float glass product. Most important, however, is the effect of the bunk tin velocity away from the cold end towards the hot end, which again will go a some way towards ameliorating the adverse effects of longitudinal and backmixing inadvertently causing contaminated tin to report back to the cold end of the chamber. If additional heat is added to the tin, the recirculation rate can be increased at the expense of an increase in total energy consumption and the extent of backmixing can be reduced further if necessary. However, it is clear that without due attention to a reduction in heat losses, even the improvement proposed could not be implemented.

Ideally, all surfaces within the float glass chambers above the liquid tin level should be impermeable so that on start-up or during temporary shut-down oxidising gases and other gaseous contaminates are not introduced into the protective gas by gaseous diffusion fiom remote porosity over an extended period of time resulting in unacceptable levels of surface dross fon;nation. This means the elimination of refractory construction presently used above the tin bath level in all float glass lines. Provided the gas atmosphere is not contaminated unduly during its once through passage in the chamber and steps are taken as discussed above in addition to purifying the tin bath externa11y to a high degree of refinement so that it is "superclean" before returning to the float glass chamber, then surfaces above the tin bath level with temperatures below about 875 C can be sheathed with heat resistant alloy, probably nickel chromium based, and welded fabrication employed throughout. The highest temperatures involved will be associated with radiation from the feed glass at 1050 C. Therefore, towards the hot end either more exotic metal construction may be needed or alternatively, and preferably in terms of cost-effectiveness, plasma deposited thermal barrier coatings on heat resistant alloys may be employed as in advanced gas turbine combustors.

With these materials, impermeable surfaces are provided for gas containment with little or no propensity for contributing gaseous contaminates into the bath atmosphere, particularly at start-up or restarting after temporary shut-downs. Under these conditions the only contaminates introduced into the gas atmosphere in the float glass chamber are those due to Resorption from the molten glass at the feed end. This is provided, of course, that a small positive gas pressure is maintained throughout the chamber. As already discussed, these contaminants can be purged out of the chamber almost immediately provided longitudinal backmixing can be adequately controlled by forced circulation.

As well as the modifications just proposed to the internal surfaces, in the interests of economy of high purity nitrogen and hydrogen usage to maintain the required small positive pressure, it is desirable to incorporate certain additional new features into float glass production technology. These are described in co-pending patent application filed on 17 July 2002 by the applicant relating to continuous coal-based steelrnaking and a further co-pending application filed by the applicant on 22 February 2003 relating to a molten metal siphon. The "swimming pool" reactors of these prior applications are analogous in many respects to the float glass chambers currently under discussion. In particular, attention is drawn firstly, to removal of melt from a liquid metal bath from the top and its re-introduction to the bath using a molten metal siphon and second, to the "top-hat" enclosure of a molten metal bath by lightweight structures and low thermal mass insulation with trough-type sealing arrangements employing a non-volatile liquid seal. This ensures ease of access to the metal baths by lifting the top-hat enclosure up clear of the sealant troughs, while providing 100% gas tightness in the down position during operation. Finally, the electrical direct resistance or conductive heating advocated for "swimming pool" reactors containing liquid steel is clearly equally applicable to float glass technology. The bath sizes in both are comparable but there can be no suggestion that frozen shells of melt replace hearth refractories for tin baths. In the steel case, the melts are constrained to temperatures fairly close to the liquidus temperature, whereas for the tin baths the temperatures involved are vastly in excess of the tin melting point of 232 C.

This invention may be further appreciated from the drawings accompanying this description.

Pig. I is a schematic sectional plan of a float glass line employing the proposed external bath recirculation system.

Fig. 2 is a schematic sectional elevation of the float glass line shown in Fig. 1.

Referring now to Fig. I and Fig. 2, a float glass line comprises hoppers at the feed end for storing raw material mix I and cutlet 2 which are charged into the glass melting furnace 3 and delivered as molten glass at a temperature around IOSO C on to the surface of a bath of molten tin, referred to as the float bath 4, contained within the float glass chamber 5. The glass is held in a chemically controlled bath atmosphere at a high enough temperature for long enough for the surface to become completely flat and a glass ribbon 7 is formed. The ribbon is cooled down while still advancing along the molten tin bath until it is hard enough (about 600 C) to be transported into an annealing lehr 8 without marking the bottom surface. The ribbon passes through the annealing lehr 8 to be cut at 9 to customer's requirements and transported to an automatic warehouse (not shown) for stacking and despatch. A typical modern float line produces around X00 tomes per day on a continuous basis, 365 days per year for a campaign of several years. The glass directly produced has a uniform thickness with bright fire-polished surfaces.

A preferred embodiment of the apparatus and method of the present invention is also depicted in Fig. I and Fig. 2. On each side of the float chamber 5, contaminated tin is continuously withdrawn from the float bath 4, without disturbing the laminar flowing tin melt surface, into a sump 10 from which it is sucked into a barometric leg or seal vessel 11 for distribution into two vacuum-assisted desorption towers or desorbers 12.

The refined tin leaves both reduced pressure zones and is returned to atmospheric pressure via a barometric seal vessel 11, which discharges liquid tin continuousb into a single packed bed column, the heat exchange column 13, which receives refined liquid tin from all four of the desorbers and their associated barometric legs I 1. The coolant gas mixture for the heat exchange column 13 is operated on a closed circuit with in- line heat recovery using a heat recovery steam generator (HRSG) 14 or other cooling device as appropriate and gas flow is induced either by a fan or by an eductor 15 actuated by the high pressure nitrogen/hydrogen gas mixture, which is added at the same rate as preheated protective gas atmosphere is bled off the principal circuit for return to the float glass chamber' principally at inlet ports 16 but also at other locations if so desired. The cooled liquid tin at about 600 C is returned to the float bath at 17, which closes the loop for float bath circulation. Also if appropriate to control the rate of cooling, side streams of this cooled refined tin can be directed elsewhere (not shown) to the bath, possibly incorporating direct resistance or conductive electrical heating to effect the desired temperature gradient. The pumping requirements for returning the tin to the float bath at 17 and for reticulation elsewhere are not shown, but bearing in mind the corrosiveness of tin at elevated temperatures towards most metals, a gas-lift system would probably be employed for this service. The protective gas atmosphere after being added to the chamber in a preheated condition around 600 C, principally via inlet ports at 16, is vented from the float glass chamber at the hot end through an appropriate number of vents 18 (only one shown on each side).

For vacuum assisted desulphurisation and deoxidation of the molten tin extracted from the hot end of the float bath by direct countercurrent contacting with a hydrogen/nitrogen gas mixture as the strip gas, the tin needs to be at a temperature of about 850 C or even higher to limit the physical dimensions of the plant requirements This is followed by cooling of the molten tin after desulphurisation and deoxidation by a closed gas circuit direct contacting operation involving a packed bed irrigated with the molten tin to be cooled as it flows down by gravity as a dispersed phase of non-wetting droplets and rivulets countercurrent to the nitrogen/hydrogen gas flow, about twenty percent of which is added subsequently in a preheated s condition to the float glass chamber as the bath atmosphere. This make-up preheated gas, which is bled off the hot end of the closed loop tin cooling circuit at a temperature of around 600 C or slightly higher to allow for heat losses in its distribution back to the float glass chamber. During cooling of the molten tin with recirculating nitrogen/hydrogen, a worthwhile amount of process steam may be generated using a scaled down version of a heat recovery steam generator (HRSG) based on commercially proven designs for gas turbine combined cycle power generation.

The scheme outlined above is believed to be superior to a related method proposed in US Patent 3,480,420 for vacuum purification of a float glass bath, but in which the extent of refining is severely limited by the availability of a packed bed maximum height of about 1.5m. This is because the vacuum lift that can be obtained using a barometric leg approach as adopted in this US Patent is severely restricted for dense liquid metals such as molten tin. To obtained "superclean" tin, packed heights considerably greater than 1.5m are needed. For this, certain aspects of the methodology presented by the applicant for hot gas clean-up in a copending patent application, filed 14 September 2002 and entitled "Liquid Metal Transport Gasifier", are needed.

In addition, deoxidation of the molten tin in the preferred embodiment of US Patent 3,480,420 uses carbon and graphite linings, which seems to the applicant not to be compatible with the high availability required for commercial utilization. The preferred embodiment in the present proposal is based on direct desorption of dissolved atomic oxygen from the molten tin followed immediately by chemical reaction and reduction with the hydrogen in a hydrogen/nitrogen strip gas mixture at the gaslliquid metal interface. This is sustainable indefinitely and in general considered to be kinetically more favourable. Thus, although the teachings of US Patent 3,480,420 may be adequate for reducing the level of contamination of the tin bath, they are unlikely to be suitable for the high level of refining required for the present case, in which relatively large amounts of tin are recirculated in the region of 200 tonnes per hour in order to combat back and longitudinal mixing in the float bath. The tin bath circulation rates in the present proposal are typically about one order of magnitude larger than those envisaged in US Patent 3,480,420, where it is stated that molten tin can be purified at a rate of 20 tomes per hour, for example, down to about 2 to 6 parts per million of contaminants. For tin circulation rates at the level proposed in the present invention, it is necessary to achieve a degree of refinement considerably better that those attainable with the method used in US Patent 3,480,420, otherwise the propensity for returning sulphur back to the float glass bath dissolved in the return tin can be shown be shown to be excessive. Targets level of sulphur in the refined tin need to be below I part per million. Detailed mass transfer evaluations indicate that these levels are achievable with the method and apparatus now being proposed. For tin at 850 C a purge gas rate of about 0.5 molts is required at an operating pressure of about 400 microns of mercury to attain the tin purity specified.

Desorption of sulphur from molten tin for these conditions can be shown to be influenced both by gas phase and liquid phase mass transfer under what is known as mixed transport control. On this assumption, it will require in the region of 3.5 overall liquid phase transfer units for the vacuum assisted Resorption. Using mass transfer correlations developed by the applicant and recently submitted for publication, this scenario translates into a reasonably sized commercial plant installation. Using the case of a float glass line producing say BOO tonnes per day, this will require two desorbers on each side of the float glass chamber making a total of four units in parallel, each about 3m in diameter and packed to a height of about 6m with 150mm ceramic Raschig rings, for example.

In the introductory paragraphs of this description, the importance of being able to force circulate the molten tin bath at an appropriate level, while not increasing the overall energy consumption, was stressed as a key issue needing resolution. It is now appropriate to return to this aspect, having proposed that the existing conventional refractoryarched roof construction used in present day float glass chambers be replaced by lightweight removable and gas tight "top-hat" enclosures. This will dramatically reduce thermal losses by R. widespread use of low thermal mass and very efficient thermal insulating materials as backing to impermeable sheet metal fabricated internal surfaces above the liquid tin bath. It is also claimed that such modification significant" improves contamination of the gas atmosphere by prolonged diffusional outgassing of air and other contaminants contributing to unacceptable levels of dross formation. Both of these very desirable attributes are clearly dependent on being able to use relatively lightweight sheathing of internal surfaces with sheet stainless steel, for example, and all-welded fabrication. The provisions of this invention make this all a feasible reality. With "superclean" tin being returned to the feat glass chamber and control of back and longitudinal mixing in both the liquid tin bath and the gas annosphere, it is s anticipated that sulphidation of above- bath metallic linings should not be a problem. However, it may be necessary to control the hydrogen level below the 1-5% currently employed by the industry for bath protection, so that the oxygen potential in the gas atmosphere is high enough to ensure stability of the dense protective chromia surface layer on which the reputation of stainless steel for reliable high temperature service so much depends. 8 9

Claims (8)

1. I. In an apparatus and method wherein flat glass is formed on a bath of molten metal having a hotter end portion and a cooler end portion inside a chamber known as a float glass chamber, through which the glass is advanced from said hotter end portion to said cooler end portion and in which the gas space above the molten metal bath is kept filled with a protective gas atmosphere by introducing substantially all of such protective gas atmosphere into the float glass chamber at its cooler downstream end whilst removing a controlled amount of said protective gas atmosphere at the hotter upstream end to produce a gas flow from the colder to the hotter end and in which prior art includes both an essentially static molten metal bath as well as one which is force circulated from its colder end of portion towards the hotter end portion with extraction of some of the molten metal at the hotter end to effect external de-oxidation and desulphurisation of the molten metal before being re admitted to the bath at its cooler end or at various locations along the bath, the improvements needed to positively eliminate top speck on float glass comprise the following steps: (a) preheating the protective gas external to the float glass chamber by direct contact with molten metal extracted from the hot end of the molten metal bath, such hot molten metal first having been comprehensively desulphurised and de-oxidised under partial vacuum with a strip gas and then admitted via a barometric leg to the direct contact gas/liquid metal heat exchanger at substantially atmosphere pressure and in which the molten metal as the dispersed phase is in an attenuated condition and in direct contact with protective gas as the continuous phase, with the overall objective of admitting the required amount of protective gas atmosphere to the float chamber at such an elevated temperature that solid stannous sulphide, for example, is unable to form anywhere within the float glass chamber above the molten metal melt level, because all such surfaces are maintained at temperatures in excess of the so-called dewpoint temperature for such stannous sulphide deposition at the particular location relative to its propensity for adventitious contamination of the protective gas atmosphere at such location (b) removing entirely the extensive accessible pore structures within conventional refractories currently used in all roof and wall construction with current state-of-the-art float glass chambers, such void spaces having access to the protective gas atmosphere via gaseous diffusion, including; both molecular inter-diffusion and so-called Knudsen diffusion, and thus affording a haven for deposition of materials such as stannous sulphide in the internal lower temperature regions adjacent to the hot faces of such refractory roofing and wall materials and thus potentially available for adventitious contamination of the protective gas atmosphere especially at start-up and during temporary shut-down and also thus preventing possible deposition of solid stannous sulphide followed by chemical reduction in-situ within the bulk of the refractory materials by the hydrogen in the protective gas atmosphere behind the hot face so that liquid tin is eventually formed, finding its way in due course to the roof surfaces and ultimately becoming the source of hot speck contamination on the float glass surface itself.

   2. An apparatus and method as claimed in Claim I wherein all internal surfaces above the melt line within the float glass chamber are impermeable.

   3. An apparatus and method as claimed in Claim 2 wherein internal roof structures and side walls are fabricated, using all-welded construction from sheets of heat resistant metallic alloy backed by low I thermal mass insulating materials with high performance features compatible with the elevated I temperature characteristics of the float glass process.

   4. An apparatus and method as claimed in Claim 3 wherein the highest temperature regions of the roof surface utilize plasma deposited thermal barrier coatings as used in advanced gas turbine combustors to enhance the elevated temperature performance of exposed metallic surfaces.

   5. An apparatus and method for manufacturing float glass as claimed in Claims 1 - 4 in which the flow of protective gas and molten metal from the colder end to the hotter end are substantially greater than that employed in current state-of-the-art float glass technology and are also greater than that employed in prior art, which now may be considered obsolete, in order to positively control back and longitudinal mixing within the protective gas atmosphere as well as within the molten metal bath inside the float glass chamber, so that contaminants are not permitted to report adventitiously back towards the colder end portions, having originated towards the hotter end portions, where the molten glass is charged onto the molten metal bath.

   6. An apparatus and method for manufacturing float glass as claimed in all preceding Claims in which energy is conserved, protective gas sealing along the length perimeters is improved and free access to the float glass chamber internals is facilitated by the introduction of a so-called top-hat enclosure system that is readily lifted upwards and is sealed around its periphery by a trough-type arrangement employing a nonvolatile liquid seal. I 7. An apparatus and method as claimed in Claim I in which the comprehensive de-oxidation and desulphurisation of the molten metal therein referred to, is conducted external to the float glass bath I to produce "superclean" tin, for example, using gaseous desorption under reduced pressure in which a hydrogen/nitrogen gas mixture is used to countercurrently strip dissolved oxygen and sulphur from the molten metal extracted from the hotter end bath portion of the float glass chamber in a packed; bed of such predetermined height of say 6m, for example, irrigated with the molten metal so that target levels of dissolved sulphur are less then I part per million are achieved under around 3.5 overall liquid phase transfer units need to be accommodated, a performance unattainable for practical operating conditions pertaining to prior art preferred embodiments, limited to a maximum I height of around 1.5m for molten tin, for example, if a vacuum lift is employed as the means for elevating the molten metal to the top of the contacting device, in which such Resorption of dissolved oxygen and sulphur is required to take place.

   Amendments to the claims have been filed as follows 1. A method for the elimination of top speck on float glass based on the creation and maintenance of physical conditions inside the float glass chamber that totally prevent deposition of solid studious sulphide anywhere within the chamber on exposed surfaces above the molten metal bath by taking positive steps to ensure that all such exposed surfaces are maintained at a temperature in excess of the so-called dewpoint temperature for solid stannous sulphide deposition thereby precluding chemical reduction by hydrogen in the protective gas atmosphere of such deposited solid stannous sulphide to form tiny globules of liquid tin. I he distinctive steps comprise the following: a. preheating the protective gas external to the float glass chamber by direct contact with molten metal extracted from the hot end of the molten metal bath, such hot molten metal first having been desulphurised and deoxidised under partial vacuum conditions with a strip gas and then admitted via a barometric leg to an externally located direct contact gas/liquid metal heat exchanger at substantially atmosphere pressure and in which the molten metal as the dispersed phase is in direct contact with protective gas as the continuous phase at a gas flow rate commensurate with the propensity for adventitious contamination of the protective atmosphere within the float glass chamber so that the vapour pressure of gaseous stannous sulphide is maintained by dilution to be always less than that required nor the initiation of deposition of solid stannous sulphide.

   b. removing entirely the extensive accessible pore structures associated with conventional refractories currently used in all roof and wall construction with state-of-the-art float glass chambers, such void spaces having access to the protective gas atmosphere via gaseous diffusion, including both molecular inter-diffusion and so-called Knudsen diffusion, and thus affording a haven for deposition of materials such as stannous sulphide in the internal lower temperature regions adjacent to the hot faces of such refractory roofing and wall materials and thus potentially available for adventitious contamination of the protective gas atmosphere specially at start-up and temporary shut-down and also possible deposition of solid stannous sulphide followed by chemical reduction in- situ within the bulk of the refractory materials by the hydrogen in the protective gas atmosphere behind the hot face so that liquid tin is eventually formed, finding its way in due course to the roof surfaces and ultimately becoming the source of hot speck contamination on the float glass surface itself.
2. 2. An apparatus to support the method as claimed in Claim 1 wherein all internal surfaces above the melt line within the float glass chamber are impermeable.
3. 3. An apparatus as claimed in Claim 2 wherein internal roof structures and side walls are fabricated, using all-welded construction from sheets of heat resistant metallic alloy hacked by low thermal mass insulating materials. 1=
4. 4. An apparatus as claimed in Claim 3 wherein the highest surface temperature regions of the roof, namely those above 875 C, which receive radiation from the feed molten glass, utilize plasma deposited thermal barrier coatings to enhance the elevated temperature performance of such exposed metallic surfaces.
5. 5. An apparatus as claimed in Claim 3 wherein the highest surface temperature regions of the roof, namely those above 875 C, which receive radiation from the feed molten glass, utilize so-called exotic high temperature metals or alloys, as opposed to those materials used elsewhere, which are comprised of conventional standard heat resistant alloy materials, probably nickel-chromium based.
6. 6. A method for manufacturing float glass as claimed in Claim I, in which the flow of protective gas and molten metal from the colder end to the hotter end in relative terms are greater than those employed in current state-of-the-art float glass technology and are also greater than those employed in prior art, which now may be considered obsolete, in order to positively control back and longitudinal mixing within the protective gas atmosphere as well as within the molten metal bath inside the float glass chamber, so that contaminants are not permitted to report adventitiously back towards the colder end portions, having originated towards the hotter end portions, where the molten glass is charged onto the molten metal bath.
7. 7. An apparatus for manufacturing float glass as claimed in any of the preceding claims in which free access to the float glass chamber internals is facilitated by the introduction of a so-called top-hat enclosure system comprised of the aforementioned impermeable material backed with low thermal mass insulation that is readily lifted upwards and is sealed around its periphery by a trough-type arrangement employing a non-volatile liquid seal.
8. 8. A method as claimed in Claim I in which the dcoxidation and desulphurisation of the molten metal therein referred to, is conducted external to the float glass bath to produce "superclean" tin, for example, using gaseous Resorption under reduced pressure in which a hydrogen/nitrogen gas mixture is used to countercurrently strip dissolved oxygen and sulphur from the molten metal extracted from the hotter end bath portion of the float glass chamber in a packed bed of such predetermined height of say 6m, for example, irrigated with the molten metal so that target levels of dissolved sulphur are less than 1 part per million are achieved under mixed transport control conditions, in which both the gas phase and the liquid phase exert an influence on the transport phenomena such that around 3.5 overall liquid phase transfer units need to be accommodated, a performance unattainable for practical operating conditions pertaining to prior art preferred embodiments, limited to a maximum height of around 1.5m for molten tin, for example, if a vacuum lift is employed as the means for elevating the molten metal to the top of the contacting device, in which such desorption of dissolved oxygen and sulphur is required to take place.