Journal of Non-Crystalline Solids 80 (1986) 93-102 North-Holland, Amsterdam

93

REFRACTORIES FOR THE GLASS INDUSTRY V. G O T T A R D I t

lstituto di Chimica Industriale, Via Marzolo 9, Padova, Italy
After a brief review of the evolution of furnaces for the glass industry, the most important refractories used in the different parts of a furnace are described, paying particular attention to electrocast refractories. The properties of these materials are reviewed referring to the relevant chemical compositions and structures. The different parts of the glass-furnace are indicated for which these refractories are recommended, and their behaviour in the presence of different types of glass is evaluated. Recent advances performed on this topic are presented and discussed.

1. Introduction The recent advances of the glass industry, particularly in continuous glass manufacturing, are properly connected with progress both in the testing and tuning of refractory materials suitable for the different and various needs of the glass-furnace. Important factors, such as the required absence of contaminants in the glass, the use of higher temperatures, and the never ending problems of energy saving and environmental pollution are currently forcing a rapid evolution in furnace designs; on the other hand, these improvements will be surely frustrated without the parallel studies of refractory reactions.

2. Refractories in glass furnaces Ever since the discovery of glass in the remote past, it has always been associated with the dynamic concept of a particular cooling rate of the liquid which, alone, was able to permit the passage from the liquid state to the vitreous one. Even if new ways of producing glass have been proposed in recent times, it remains true that most industrial glass is still obtained today by cooling a liquid. From this consideration it is clear that the production and elaboration of this liquid is of paramount importance with regard to the properties of the end product. In fact it is essential that this liquid be as homogeneous as possible, that it be gas-free as far as possible, that its evolution be understood
+Deceased Sept. 2, 1985 in Venice.

0022-3093/86/$03.50 Elsevier Science Publishers B.V. (North-Holland Physics Publishing Division)

94

V. Gottardi / Refractories for the glass industry

and reproducible, since no other liquid is as sensitive to that series of variables which together constitute what is generally known as thermal history, as are vitrogeneous liquids. It follows that the container in which the liquid glass is treated is also extremely important: that is, the container material which comes into contact with an extremely reactive liquid at high temperatures. This material, in fact, conditions the accuracy of the desired composition, since it is the primary source of possible contamination, it affects the cost of the end product as regards its durability, and, most important, it limits the choice of possible compositions as regards the maximum temperatures to which it can usefully be subjected. Since the process of glassmaking has undergone no important changes during the last two thousand years, if we except the innovation of automation, it could be said that progress in the technology of glassmaking is synonymous with the progress made in the refractory materials used to contain glass in its liquid state, whether crucibles or continuous tanks are used. During the past fifty years alone, the temperature of furnaces for hollow ware has increased from one thousand three hundred degrees centigrade to one thousand five hundred, as can be seen in fig. 1 taken from Garstang; a similar tendency has been pointed out by Bondarev as regards furnaces for sheet glass. This increase in temperature has not only permitted changes to be made in the composition so as to improve the properties of the products, and has not

10

2
0 0

I
1930

I
1940

I
1950

I
1960

I
1970

I
1980

Fig. 1. Temperature evolutioninfurnacesforcontainerproduction(Garstang, 1971; Barton, 1982).

V. Gottardi / Refractories [or the glass industry

t.~5

only permitted improved homogeneity and refining, but has also permitted a considerable reduction in specific consumption which as regards hollow ware is reproduced in fig. 2, again according to Garstang, plus additional data for nineteen eighty-two, which shows that over a period of sixty years, the energy required today to produce a ton of glass is only a tenth of that needed in nineteen twenty. In the sheet glass furnaces it has been calculated that the increase in specific productivity is four per cent for every ten degrees centrigrade increase in temperature in the melting zone. It has also been shown experimentally that an increase of maximum temperature in the furnace from one thousand five hundred and twenty to one thousand five hundred and ninety degrees centigrade increases the specific yield of glass from the melting end by fifty per cent, and reduces the fuel consumption by fifteen percent and the operating costs per ton of glass by twenty-four per cent. A logical conclusion of this tendency would lead us to believe that an increase in temperature would automatically lead to an improvement in the quality of glass and to savings in its production, and that the only obstacle to this lies in the absence of suitable refractory materials at the present time. In actual fact, it does not appear that this conclusion is completely valid. Laboratory research carried out by Botvinkin on the time required for the complete melting for a typical sheet glass in the range of one thousand four hundred to one thousand seven hundred degrees centigrade shows that the process time is accelerated when the temperature goes from one thousand five hundred and fifty to one thousand six hundred degrees centigrade.

1600 0

0 1500

o-

1400

I 1930

I 1940

I 1950

I 1960

I 1970

I 1980

Fig. 2. Specific consumption in furnaces for container production (Garstang, 1971; Barton, 1982).

96

V. Gonardi / Refractories for the glass industry

Table 1 Increase of melting rate of sheet glass with temperature (Botvinkin) Temperature (C) Time required for complete melting

(min)
1400 1450 1500 1550 1600 1650 1700 53.7 36.4 20.9 10.0 4.0 2.4 1.6

These fifty degrees reduce the process time by two and a half times, whereas further temperature increases do not produce a corresponding decrease in the process time. If we add to this consideration the preoccupation due to increased atmospheric pollution - it is c o m m o n knowledge that increased temperatures result in a larger production of nitrogen o x i d e s - a n d with increased insulation difficulties, we come to the conclusion that with the present state of our knowledge, the optimum temperature for the production both of hollow ware and of sheet glass can be fixed between one thousand five hundred and fifty and one thousand six hundred degrees centigrade. T h e problem, then, regarding refractories for the construction of glass furnaces is not the availability of material which can stand ever increasing temperatures, but rather materials which give the best performance at the temperatures indicated above. In order to reduce such a vast subject to the limits of this paper, melting in crucibles will not be considered (these too have problems of high temperatures in the production of sophisticated glass). Instead, we shall concentrate on the results obtained and the problems which remain unsolved for refractories used in tanks. As regards the tank containing glass in its liquid state, the most delicate problem concerns the possible chemical interactions, that is, corrosion. If we assume that a certain amount of interaction is theoretically inevitable, it must be ensured that this is as slow as possible in order to increase the tank life, and that the interaction gives rise to products which can be easily digested by the molten glass, the composition of which must not be substantially altered. Since the interaction takes place on the surfaces in contact with the glass, non-sintered products are advisable, that is electro-cast materials with two per cent porosity, in place of the eighteen per cent of traditional refractories produced by heating. The first attempts made by Corning to produce a type based on alumina date from nineteen twenty-five, to be followed a few years

V. Gottardi / Refractories for the glass industry

97

later by compositions based on ZrO2 and A1203 from which a production of two hundred thousand tons per year is derived today forming an essential part of all the principal glass furnaces in the world. Does this general consensus of opinion mean that we have found the best possible composition or that we have fully understood the mechanism of corrosion of a refractory in contact with molten glasses? The answer is certainly not; and this lets us hope that more progress will be made in the future. Let us consider a classical composition containing A1203 fifty per cent, ZrO2 thirty-four per cent, SiO2 fifteen per cent and one per cent of NaeO. During the course of cooling the liquid, crystallisation takes place first of primary baddeleyite, then of corundum, eutectically associated with baddeleyte, then of free corundum or corundum growing on eutectic grains and finally a vitreous phase remains. The size and morphology of the crystals, and the entity and the composition of the vitreous phase, are functions of the speed of cooling which rarely takes place in a uniform way throughout the entire volume of a block after casting. The vitreous phase is linked to the presence of sodium. In the ternary diagram AZS, a ternary eutectic of corundum, baddeleyite and mullite is indicated in the final solidification. After prolonged heating at one thousand five hundred degrees (forty-eight hours, for example,) the vitreous phase is reduced from twenty to fourteen per cent, and the mullite formed through devitrification increases from insignificant amounts to up to twentyfive per cent. If we attempt to reproduce the same composition obtained by electrocasting with the sol-gel method, enabling situations close to equilibrium to be reached rapidly, as has been done in Padua, only ten per cent of the vitreous phase is reached, and thirty four per cent of mullite. There are products on the market with a higher percentage of zirconium than the one mentioned above, and recently products have appeared containing up to twenty-eight per cent of Cr203, which, with alumina, produces a solid solution within large ratios. If we consider that these delicate, complex structural situations are further influenced by the gradient of temperature formed in the blocks of a tank, and above all by the type of glass with which it comes into contact, we realise how complex the interpretation of the corrosion phenomena is, and how difficult it is to carry out laboratory tests able to simulate actual working conditions. Similarly, attempts to isolate the various component crystals and to study their solution in various different types of glass has provided information which is interesting but not helpful from a practical point of view. These results indicate that the corrosion of refractories by molten glass must be a diffusion-controlled process, but that the enhanced rate of solution due to upward drilling at the flux line needs to be explained. The electronic microprobe in special experimental conditions has shown that at a temperature of one thousand six hundred degrees, the diffusion of A1203 and ZrO2 in soda-lime glass is 700 mm whereas on the other hand, the

98

V. Gottardi / Refractories for the glass industry

diffusion of sodium and calcium in the vitreous phase is of six hundred and one hundred, respectively, with diffusion profiles which are not yet completely understood. From similar results obtained for other types of glass, it would seem that potassium oxide is the most likely to transfer into the vitreous phase, and the classification of the various ions- which can be deduced from experimentssupports the theory of Well on their polarisability. Many studies have been made on the surface tension, the wettability, the density and viscosity of the vitreous phase which is from ten to fifty times higher than that of the glass with which it comes into contact, and on the electro-chemical processes whose importance has tended to diminish in recent times. For a long time it was thought that in an electro-cast refractory of the AZS type, the phase most subject to corrosion was the vitreous one, but recent successes obtained in transforming corundum into a solid solution of AI203-Cr203 have demonstrated that the crystalline phases are important as well. Attempts made to alter the composition by introducing other oxides with a high melting point include the introduction of cerium oxide which would appear to involve the vitreous phase, and of magnesium oxide which forms a spinel with alumina. These attempts have been, however, rather infrequent, and have so far not produced many practical results, unlike those concerned with improving the melting and cooling technology and reducing impurities that are considered by many to be the main cause of the development of gas during contact between glass and refractories. This is one of the problems to which much attention has been paid. Notwithstanding this, there remain many points which are not entirely clear. The origin of bubbles in melted glass in contact with refractory material can be attributed to a variation of the solubility of gases dissolved in the glass, due either to a local variation of temperature or composition caused, for example, by the exudation of the vitreous phase of the refractory. On the other hand, the gas could come from the open pores or the closed pores, destroyed during the working life of the refractory, and expelled either by thermal expansion or by capillary action. There also exists the possibility that the bubbles could be the result of chemical reactions between glass and certain impurities in the refractory material. An analysis of the bubbles themselves has demonstrated the constant presence of oxygen and nitrogen, but many other gases have been found-providing proof of the multiplicity of the causes. It would appear, however, that casting in oxidising conditions causes a significant reduction in this phenomenon. The particular emphasis devoted to the problem of the interactions between refractory material and glass should not exclude at least a mention of the other properties in connection with the structure of electro-cast products.

V. Gottardi / Refractories for the glass industry

90

The presence in the AZS products of percentages above thirty per cent of baddeleyite and the well-known allotropic transformation of ZrO2 from a tetragonal form at high temperatures, into a monocline form at under one thousand degrees, a transformation which is accompanied by a marked variation of volume, warns us that particular precautions must be taken in the case of the intermittent functioning of the furnace, or, at any rate, during the periods of warming up and cooling down of the furnace. Attempts to stabilise the ZrO2 have not met with any success up to the present time, as happens with other products. Thus, the task of absorbing variations in volume is still entrusted to the vitreous phase, whose composition and behaviour on heating become increasingly critical. Another important aspect of the adoption of electro-cast refractories in the place of traditional ones, is that connected with the different thermal characteristics of a material which, due to the fact of its extremely low porosity, has a much higher thermal conductivity than that of sintered products. Bondarev, whose work has been mentioned previously, calculates that when sintered refractories are replaced by fusion-cast ones, the heat losses are increased two to three times, and because the fusion-cast refractories allow higher melt temperatures, there is a further heat loss of eight to ten per cent. Consequently, in the planning of a glass furnace, the study of insulation is of paramount importance, in order to reduce the external temperature of the walls from four hundred and forty degrees centigrade to one hundred and ten. This now creates problems for the refractory materials, to which solutions have been found which are frequently in contrast with those mentioned previously. If, on the one hand, good insulation ensures a saving in fuel consumption, increasing the furnace's performance and improving the quality of the glass produced, on the other hand it increases corrosion, exudation and the penetration of liquid in the interstices between the blocks. The technology of the production of blocks has provided some solutions with the production of blocks with shrinkage cavities placed in the least dangerous areas, or even without shrinkage cavities, but most of all with the introduction of diamond grinding of the mating surfaces. Dimensional tolerance of as little as 1 mm/m has been reached on pieces weighing three hundred to four hundred kilos, thus permitting a perfect juxtaposition of the blocks which reduces the risk of infiltration to a minimum. As far as the electric properties are concerned, it should be mentioned that electro-cast refractories have followed the growing employment of electric furnaces for the production of glass, confirming their availability for the construction of these furnaces, or of partly electric furnaces, with the presence of boosters or electrodes in the end part. If the refractories in direct contact with glass are decisive for successful

100

V. Gottardi / Refractories for the glass industry

melting, and time has been spent to discuss their properties and their problems, mention should also be made of those refractories intended for the construction of the crown, and for the regenerators. As regards the crown, the most commonly-used refractories today are those of silica, even if there are examples of uses of basic, or mullitic or of electrocast types. The reasons for this are perhaps the cost, the weight and the refusal already mentioned to contemplate higher temperatures than the ones at present being used. If this were not so, silica refractories would be the weak point as they now have to withstand near limit conditions, which is never advisable for refractories. The porosity of the materials used in the crown does not appear to be a negative quality; the ability to form a well-developed crystal structure of high purity on the working face is of considerable importance. The pores of the bricks seem to be filled with a glassy phase which migrates towards the cooler part of the brick and this reduces permeability and corrosion. Naturally the quantity of alkalies which are fixed on the surface is important; obviously, it is a function of the temperature. The greater the porosity and the greater the permeability, the easier it is for the sodium vapour to pass and the better the performance. Very often the stability of the crown does not depend so much on the bricks forming it as on the refractory mortar used in construction. Finally, the high thermal conductivity of silica refractories should be borne in mind, as a problem regarding insulation. The problem of materials for regenerators is encouraging further research, although complete solutions have not yet been found. Refractories for regenerators in glass furnaces must be highly resistant to high temperatures and resistant to thermal shock; they must also possess a high thermal capacity as well as the maximum inertia with regard to gaseous and condensed products with which they may come into contact. These requirements often find solutions in contrast with each other. Hence, different materials are sometimes used, thus creating further problems such as that of compatibility and the minimum of reciprocal interaction. The composition of the carryover that enters the regenerator is highly variable, and depends on the kind of glass being produced. Particular conditions of corrosions may result from the presence of arsenic, lead, boron, or vanadium, or else as a result of oxi-reduction phenomena. Certainly alkalies or alkaline sulphates are always present. Since 1950, the refractories most generally used for the construction of checkerworks for glass furnaces are the basic ones. They have better resistance to alkalies than the aluminosilicates, and they make it possible to increase the heat capacity of the regenerators. However, they are sensitive to the effect of water, especially during storage or during the warming up of the furnace.

V. Gottardi / Refractories for the glass industry

101

T h e attack of magnesite bricks by NazSO4 is significant in the lower courses of the checkerworks, and, to a lesser extent, in the middle courses. Resistance to thermal shock corresponds exactly to the purity of the magnesite. T r a c e s of boron, too, such as those present when the magnesite has been obtained from salt water, are able to reduce this characteristic, and they can also give rise to the formation of low melting phases. A ceramic-type link would appear to be the most suitable, but may give rise to abnormal growth of grains of peryclasium. The presence of calcium silicate between the grains is acceptable in small percentages, as long as the ratio CaO/SiO2 is not higher than 2. In these conditions too, the best resistance to creep obtains, provided that the amount of Fe203 remains as low as possible. Regenerators can be constructed entirely or in part with chromo-magnesite, forsterite, or mullite refractories. Another alternative could be to use fused-cast materials; however such a solution presents two difficulties: first the weight of the packing, and second the cost. This led to the development of the cruciform shape which allows light-weight fusion-cast blocks to be used, while ensuring the stability of the construction. T h e advantages claimed are high corrosion-resistance which should lead to a constant level of regeneration during the campaign, and increased surface area for heat exchange. It is obvious that any attempt to deal in one short paper with such a vast subject about which so much information is available, where so much progress has been made, but which also contains many unknown quantities and has many problems still awaiting a solution (like that of refractories for glass furnaces) is an impossible task, and is bound to have only modest results. Each single topic that has been only briefly dealt with deserves a far more detailed analysis with more serious discussions. At the conclusion of this extremely rapid survey, I hope that at least three points will have emerged. The first concerns the progress that has undoubtedly been made over recent decades in the field of refractories, both on the part of the makers and on the part of the consumers of the product. This progress has often consisted in small steps forward, made on the basis of experience and a statistical base which is constantly growing; these data deserve to be considered as scientifically valid where the complexity of some problems does not allow them to be studied in the laboratory. If the life-cycle of glass furnaces has been increased, and the quality of the glass improved, this is in part due to the refractory material used. T h e second conclusion is that research has a broad field available for the acquisition of new information and the submission of new solutions. It cannot be claimed that the compositions used at present are the best possible even within the temperature range mentioned. T h e influence of the presence of many elements has still to be ascertained. T h e phase diagrams at the highest

102

V. Gonardi / Refractories for the glass industry

temperatures are still not known with any degree of certainty. The problems regarding the crown and the regenerators are still awaiting definitive solutions. On the other hand, as has already been mentioned, the mechanisms of corrosion of complex structures in dynamic situations, such as those which exist in glass furnaces, still lack the theory capable of providing a full interpretation. There is no doubt that when certain phenomena at the interface have been explained, it will be possible to reduce the noxious effects and thus prolong the life of a glass oven. The third and last conclusion is that, whatever improvements we do obtain, there will always be an interaction between the liquid glass and the container, since a completely inert material does not exist, nor will one ever be invented. Therefore, in order to increase the efficiency of refractories, the period of contact at the same temperature must be reduced as much as possible. It is not impossible that alternative systems of melting glass will be discovered that function with an improved thermal yield. This is one of the most fascinating aspects of research on a material like glass which can only be obtained by rapid melting followed by rapid cooling always in critical conditions, which acquires shape and properties at the moment of its birth, and remains one of the most interesting of materials produced by man.