GB2048237A - Glass for sealing beta-alumina in electrochemical cells - Google Patents

Abstract

A glass for sealing a beta- alumina tube in a sodium-sulphur cell has a composition suitable to resist attack by sodium at elevated temperatures and a coefficient of thermal expansion suitable for use with beta-alumina. The glass consists essentially of 28-48 mol % B2O3, 0-20 mol % SiO2, 16-28 mol % Al2O3, together with 18-33 mol % of at least one alkaline earth metal oxide selected from BaO, SrO, CaO and MgO, the proportions of the constituents being such that the combined total of B2O3 and SiO2 is 40 to 60 mol % and furthermore being such that 0.517 A1 + 0.0354 A2 - 0.0063 A3 + 0.168 A4 + 0.1336 A5 + 0.098 A6 + 0.1597 A7, lies between 5.7 and 6.4 where A1, A2, A3, A4, A5, A6 and A7 are the respective molar percentages of B2O3, SiO2 Al2O3, BaO, CaO, MgO and SrO. A cell sealed with such a glass is described.

Description

SPECIFICATION Glass seals for sealing beta-alumina in electro-chemical cells or other energy conversion devices, glasses for use in such seals and cells or other energy conversion devices with such seals This invention relates to glass seals for beta-alumina in electro-chemical cells and other energy conversion devices containing sodium and to glasses for use in such seals.

In cells, such as for example sodium-sulphur cells, and other devices (e.g. sodium-sodium thermoelectric generators) where beta-alumina ceramic is used as a solid electrolyte through which sodium ions can pass, it is necessary to seal the beta-alumina in the cell structure. Sodium sulphur cells utilise a molten alkali metal and have to operate at elevated temperatures where the electrode materials are liquid. Seals necessary to seal these electrode materials within the cells have therefore not only to withstand highly reactive materials at these elevated temperatures but are also subjected to temperature cycling.As a typical example, a sodium sulphur cell might contain a beta-alumina electrolyte tube closed at one end and having sodium on one face of the tube, preferably the outer face, and the sulphur/polysulphides on the other face. the cell has to be sealed to prevent escape or mixing of these materials and a number of proposals have been made for the sealing of such cells. Compared with most metals, ceramic materials are generally weak, particularly in tensile strength and it is necessary therefore in any seal for such a cell to ensure that the ceramic material is not overstressed.

It is well-known in a sodium-sulphur cell to provide a beta-alumina electrolyte tube with an alphaalumina tubular extension at its open end so that the end portion is ionically insulative. The alphaalumina and beta-alumina have similar coefficients of thermal expansion and the alpha-alumina extension can be secured to the beta-alumina with a glass seal.

The present invention is concerned more particularly with such glass seals. The use of glass seals is described, for example in United States Specifications Nos. 3928071. 3826685 and 3868273.

These references however do not discuss the glass composition. It is convenient, in sodium-sulphur cells and similar electro-chemical cells to use a glass as a bonding agent between ceramic materials or between a ceramic material and a metal member. In a sodium-sulphur cell, the closure may be effected by sealing the ceramic electrolyte material to a closure member or to the housing. This may readily be done with a glass seal. The alpha-alumina extension then has to be sealed to the housing or to the closure member. The closure member may be a part of a current collector. Thus glass may be employed, in sealing a cell, as a bond between solid electrolyte material, e.g. beta-alumina ceramic, or an insulating ceramic, e.g. alpha-alumina, and a metal component or components such as a current collector, an intermediate component, or an external housing.The glass-to-metal bond is formed by a reaction between the glass and an oxide layer on the metal.

Both for sealing alumina to metal and for sealing beta-alumina to alpha-alumina, the glass employed must be a sodium-resistant glass. The metal material has to be chosen in accordance with both mechanical and chemical requirements. In particular, it must resist attack by sodium at elevated temperatures. It is the practice in sodium-sulphur cells to use mild steel or stainless steel for the housing, in contact with the sodium.

In particular, use has heretofore been made of a glass, commonly known as BAB glass. As is described in United States Specification No. 3275358, this glass, which has a composition by weight of about 40% B2O3, 25% Awl203 and 35% BaO can be used for providing a fused hermetic seal around lead wires of tantalum and other similar metals. Expressed in molar percentages, the composition of this glass is about 54.8% B203,23.4% Al2O3 and 21.8% BaO.

This BAB glass however does not fully meet the difficult service conditions of seals for sodiumsulphur cells, particularly alpha to beta-alumina seals, which place a severe demand on the performance of the sealing glass used. The required characteristics of such a glass in terms of its physical and chemical propeties are many and sometimes conflicting and so the development of a glass composition approaching the desired behaviour is of necessity a complex process. it must produce a reliable hermetic seal and hence must wet both the alpha and the beta-alumina. The glass should be chemically compatible with cell materials at cell operating temperatures. The thermal expansion characteristics of the glass and the other seal components must be compatible to give low seal stresses over the whole working temperature range and lifetime of the cell.

In general, most stable glasses will satisfactorily wet both alpha and beta-alumina and so could potentially satisfy the first criterion. The requirement as to chemical compatability considerably constrains the choice of glasses to the use of the alumino-borate glasses because of their known good sodium resistance. The requirements about thermal expansion characteristics are not satisfactorily met by BAB Glass.

It is necessary to consider in further detail the problems of seal stresses. The theory and practice of glassed seals, notably glass-to-metal seals, is very well established:- See, for example, "Glass-to Metal Seals" by J.H. Partridge, Soc. for Glass Technology, Sheffield 1 949 and "Glass-to-metal Seals" by A. W. Hull and E. E. Burger, Physics 5. 384-4051934. One of the most important factors contributing to successful seal performance is the close thermal matching of the seal components. Differential thermal strain causes seal stresses which, if sufficiently high, can result in seal failure.

At high temperatures when a seal is made, the glass is fluid and wets the ceramic components. On cooling, the viscosity of the glass rapidly increases and eventually it becomes so high that the glass behaves as a rigid material. It is convenient for analysis to idealise the transition between the fluid and rigid regions as occurring suddenly at a single temperature called the set point, even though in reality the change takes place over a small range of temperatures. Above the set point any thermal stresses developed in the glass are relieved by viscous flow. However, at the set point the glass becomes rigid and mechanically constrains the seal components, so that on further cooling, stresses are developed due to the subsequent differences in thermal strain characteristics.

A convenient graphical construction to asses the magnitude of the differential strain is shown in Figure 1 of the accompanying drawings where the expansion/contraction characteristics (dimension change plotted as ordinate against temperature as abscissa) of the seal component are plotted and the characteristics are vertically displaced so as to intersect the set point. Assuming that the seal components are elastic, the stresses generated at any temperature are proportional to the vertical separations of the two curves in Figure 1, and so the stress-temperature characteristic has the shape shown in Figure 2, in which the stress is plotted as ordinate against temperature as abscissa.

For sodium-sulphur cells, the properties of the ceramics to be joined are effectively fixed. The main seal stresses in the glass are material dependent only and are not greatly influenced by seal geometry except at points close to the seal edges. The composition of the sealing glass does however have an important effect on the seal stress characteristics.

Because of its brittle nature, glass has a much higher strength in compression than in tension.

Thus, for good seal strength, a high crossover temperature (see Figures 1 and 2) is desirable so that the seal stresses are compressive both at room temperature and operating temperature (typically 300--400"C). This is not the only criterion for satisfactory seal stresses because seal cracking and failure can also occur if the room temperature stress is too highly compressive.

All glasses have non-linear thermal strain characteristics and these are such that a sealing glass will always be in tension at high temperatures. By using a glass with a high set point, it is possible to have both a high crossover temperature and a moderate room temperature compressive stress as is shown in Figure 3, which is a diagram similar to Figure 2 but for two further different glasses.

Seals made with BAB glass have a generally tensile stress characteristic and tend to crack easily in thermal cycling of sodium sulphur cells. An additional disadvantage of this glass is a high densification rate. Densification, sometimes called stabilisation, is a process of molecular reordering of a glass at temperatures below the glass transition point, resulting in a gradual shrinkage of the material.

When constrained within the seal components, the sealing glass is thereby subjected to increase tensile stresses, which can cause seal failure if they grow sufficiently large. It might appear that improvement could be obtained by altering the relative proportions of the three constituents in this glass. However, as will be apparent from consideration of the data given by C. Hiramaya, J. Amer. Ceram. Soc.6 1 1 944(12), 602-6 on this ternary system that the particular composition of BAB glass is on the edge of the glass forming region of the phase diagram of this three-phase system and it has not been found possible to satisfactorily lower the expansion coefficient of the glass of this ternary system by altering the proportions of the components.

According to the present invention, a glass for sealing to beta-alumina in an electro-chemical cell or other energy conversion device containing sodium, consists essentially of:- (a) 40-60 mol% glass former consisting of 28-48 molar B203 and 0--20 mol % SiO2 (b) 1628 mol % Al2O3 (c) 1833 mol % of at least one alkaline earth oxide selected from the group consisting of BaO, CaO, MgO and SrO, where the proportions of the constituents furthermore are such that:: 0.0517 A, + 0.0354 A2 - 0.0063 A3 + 0.168 A4 + 0.1336 A5 + 0.O98A5 + 0.1597 A7 lies between 5.7. and 6.4, where A1, A2, A3, A4, A5, A6 and A7 are the respective molar percentages of B203, SiO2, Al203, BaO, CaO, MgO and SrO.

The coefficients of Ar, A2 etc. quoted above are, as will be explained later, truncated values of regression coefficients relating to effect of the constituent proportions on the main linear coefficient of thermal expansion.

The proportion of Al2O3 would usually be between 1 7 and 28 mol %.

The glass may contain one or more of the alkaline earth oxides; preferably the proportions of these oxides lie in the ranges O25 mol % BaO, 0--1 5 mol % SrO, 0--30 mol % CaO and 0--20 mol % MgO with a minimum total of 18 and a maximum total of 33 mol %.

It will be appreciated by those skilled in the art that some of the compositions defined above lie close to the boundary of the glass-forming region. These boundaries, such as those discussed with reference to alkali alumina borate glass systems by Hiramaya in the above-mentioned reference, are diffuse. Care has to be taken in the use of compositions in the boundary region because of the possibility of devitrification.

The invention furthermore includes within its scope a hermetic seal between an alpha-alumina component and a beta-alumina component in a sodium sulphur cell comprising a glass body fused between surfaces of these components, the glass consisting essentially of 28-48 mol % B203, 0--20 mol % SiO2, 1628 mol % Al2O3, together with 1833 mol % of at least one alkaline earth oxide selected from the group consisting of BaO, SrO, CaO and MgO, the proportions of the constituents being such that the combined total of B203 and SiO2 is 40 to 60 mol % and furthermore being such that 0.0517 Ar + 0.0354 A2-0.0063 A3 + 0.168 A4 + 0.1336 As + 0.098 A6 + 0.1597 A7, lies between 5.7 and 6.4 where A1, A2, A3, A4, AsS A6 and A7 are the respective molar percentages of B203, SiO2, Awl203, BaO, CaO, MgO and SrO.

The glass composition described above may employ silica to reduce the thermal expansion coefficient compared with that of a pure BAB glass and to improve its glass forming properties.

Additions of other alkaline earth oxides CaO, MgO and SrO within the prescribed range result in a glass having suitable seal stresses for operations in sodium/sulphur glassed seals and superior densification properties compared with a three component BAB glass.

For the range of compositions described above, various glass properties can be expressed as a linear algebraic summation of the kind.

Where a is the property in question.

Aj is the linear coefficient for compound i.

ai is the molar percentage of the compound i.

Implied in this relationship is the assumption that the contribution of each component to the overall glass behaviour is linear and independent of the amount and kind of other components present.

This is unlikely to be generally true but, from experimental results this assumption appears to be acceptable over the composition range of interest.

The expression 0.057A1 + 0.0354 A2-0.0063 Aj + 0.168 A4 + 0.1336 A5 + 0.098 A6 + 0.1597 A7 when multiplied by 1 o-6 give the value of the mean linear coefficient of expansion between 25 and 50000 (a 35500) and thus the above-defined glass would have a coefficient between 5.7 and 6.4 x 1 O-e/OC. Using this expression, the possible glass compositions for any required coefficient of expansion can be determined and it is thus possible to determine the composition in accordance with a specific physical requirement, e.g. to match a measured coefficient of expansion of the beta-alumina.

A number of physical properties of glass compositions as defined above may be expressed in a similar way as a series of coefficients representing the molar effects of the various constituents which can be summed in a similar way to obtain the magnitude of a required parameter.

In the accompanying drawings: Figures 1, 2 and 3 are graphical diagrams referred to above, for explaining real stresses due to differing temperature characteristics of materials; Figure 4 is a trilinear coordinate composition diagram for glasses showing the molar proportions of glass-forming oxides comprising B2O3 and SiO2 (marked GO). Al2O3 and alkaline earth oxides (marked RO); Figure 5 shows a number of examples plotted onn an enlarged part of the diagram of Figure 4; Figures 6 and 7 illustrate respectively the forming of two different seals between an alpha-alumina and a beta-alumina component; and Figure 8 illustrates diagrammatically a sodium sulphur cell embodying the invention.

A number of glass compositions within the present invention are set out in Table 1. This tables also includes, for comparison puposes, a BAB glass constituting example 597. For convenience, the thirty-one glasses in the table have been given, in the left-hand column, the Case Nos. 1-31. Example 597 is Case No. 1.

Figures 4 and 5 are trilinear co-ordinate composition diagrams. The shaded area in Figure 4 indicates the general range of sealing glass compositions which are within the scope of this invention as set out in claim 1. The further part of the definition in terms of the coefficients At to A7 in claim 1 may further restrict the range of compositions according to the mixture of glass forming oxides GO and alkali oxides RO which is used.

The glasses of the present invention include, as explained above, B203 and SiO2 which are referred to as the glass-forming oxides (GO in Figures 4 and 5). In the glass compositons of the present invention, the higher the total of boron oxide and silicon oxide, the better are the glass-forming properties. However, the resistance to sodium attack is decreased. Increasing the boron to silicon ratio gives better sodium resistance, better wetting properties and lower firing temperature but decreases the working range and gives a higher densification rate.

Increase of Awl203 results in a lower expansion coefficient but reduces the glass-forming properties.

It also raises the crossover temperature and glass transition temperature.

The preferred sodium oxide content is as low as possible, within the confines of the impurities of the raw materials used. Enhanced densification results from the inclusion of Na2O. The effect of deliberate addition of sodium oxide is shown in Table 3. Example 32 in that table is a glass having a composition, in molar percentages, of 46.63 B203, 6.75 SiO2, 20.51 Al203, 10.26 BaO, 6.52 CaO, 6.53 MgO and 2.8 Na2O. Example 33 has the composition 46.73 B203, 6.54 SiO2, 20.56 Awl203, 7.48 BaO, 6.55 CaO, 6.54 SrO and 5.61 Na2O.

It will be seen that the problems of high expansion, high densification rate and reactivity to molten sodium which occur with BAB glass can be overcome by modifications of the glass composition in the various ways described.

Table 1 below give experimental and calculated data for thirty-one different glasses, identified by case numbers in the left-hand column, of which properties have been measured. All of these glasses, except Case 1 (Example 597), which is included for comparison purposes, lie within the scope defined above.

As has been stated above, a number of physical properties of glass compositions may be expressed, like the coefficient of thermal expansion, as a series of coefficients representing the molar effects of the various constituents, which can be summed to obtain the magnitude of the required parameter. Tables 2(a), 2(b), 2(c), 2(d), 2(e) and 2(f) set forth coefficients and compare observed and predicted values for the expansion coefficient, transformation temperature, deformation temperature, room temperature, seal stress, crossover temperature and change in seal stress due to densification respectively for the 31 glasses listed in Table 1. In each of Tables 2(a), 2(b), 2(c), 2(d), 2(e) and 2(f) Case No. 1 is the BAB glass which is outside the scope of the present invention and which is included for comparison purposes only.

It will be understood that, although the shaded area of Figure 3 shows the composition ranges within which the glasses of the present invention lie, it does not constitute a definition of suitable glasses. The actual composition has to be chosen, as previously described, to obtain the required properties, e.g. expansion coefficient. As previously explained some of the composition lie close to the boundary of the glass-forming region and care has to be taken, in the use of compositions in the boundary region, because of the possibility of devitrification.

It will be noted from Table 1 that the densification rates of all the glasses listed (except the BAB glass No. 597 which is included for comparison purposes) are below, and most are substantially below that of BAB glass.

For sealing alpha-alumina to beta-alumina, the above-described glass can be used in known ways, e.g. by the use of a glass powder or a glass preform. A very convenient way of sealing a cylindrical alpha-alumina element to an end of a beta-alumina tube is to introduce a ring of glass between these surfaces, the glass is then being fused to seal the alumina elements together. Such an arrangement is illustrated in Figure 6 where there is shown the upper end of a beta-alumina electrolyte tube, which forms a separator between molten sodium and molten sulphur/polysulphides in a sodium sulphur cell.

The end surface of the electrolyte tube 10 is sealed by glass 1 1 to a shoulder 12 formed in an alphaalumina ring 13. Figure 7 is another example of a seal; in this case the lower end of a beta-alumina electrolyte tube 20 is sealed by glass 21 to a shoulder 22 formed in an alpha-alumina closure plate 23.

In each of these examples, the glass is fused and forms a hermetic seal between the beta-alumina and alpha-alumina components.

From Table 1 it will be seen that all the glasses tested have a transformation temperature Tg which is such that the sealing operation can be carried out at a relatively low temperature. The magnitude of the difference between the values of Tg and Td is indicative of the rate of change of viscosity with temperature and hence of the working range of the glass, which lies at a higher temperature between Td and the temperature at which the viscosity is 103 pascal seconds. All the glasses listed in Table 1 have a long working range, making them easy to form. The working range, for example, in every case is adequate for rod drawing.

In Table 2(a), the regression coefficients showing the effect on the expansion coefficient of variations in the amounts of the various components are set forth at the bottom of the table. These enable the expansion coefficient of a glass (within the range of possible compositions) to be calculated from the molar percentage composition. The above-mentioned expression 0.0517 A1 +0.03 54A2-O.0063A3+O.168A4 + O.1336A5+0.098A6+0.1597A7 is equal to the expansion coefficient multiplied by 106. It will be seen that the various coefficients in this expression are the regression coefficients of Table 2(a).

Using Table 2(b) in a similar way, the transformation temperatures can be calculated from the regression coefficients; the transformation temperature is 5.26 A1 + 6.30 A2 + 7.27 A3 + 4.93 A4 - 6.97 As + 7.27 A6 + 5.87 A7.

Similarly from Table 2(c) the calculated deformation temperature is 5.45 A1 + 6.92 A2 + 7.62 A3 + 5.58 A4 + 7.18 As + 7.71 A6 + 5.87 A7.

Table 1 shows the transformation and deformation temperatures calculated in this way and also the observed values.

Table 2(d) shows the room temperature seal stress. Using the regression coefficients of this table, a room temperature seal stress of less than a predetermined value, e.g. 50 N/mm2 can be ensured by using a composition in which 0.38 A1 + 1.47 A2 + 4.25 A3 - 5.41 A4 - 2.98 A5 - 0.78 A6 - 4.30 A7 is less than the predetermined value, e.g. 50 N/mm2. It will be noted that, in all the examples in Table 1., the room temperature seal stress is compressive.

Table 2(e) shows the crossover temperature and the regression coefficients whereby this may be calculated. If the crossover temperature is to exceed a predetermined value, e.g. 2400 C, then the composition is chosen so that 1.27A1+9.24A2+31.26A3-28.39A4-15.10A5-2.54A6-23.29A7 is in excess of said predetermined value.

Table 2(f) shows the change in seal stress due to densification at 4000C for the first 100 hours.

This stress change subsequently slows down, but may continue to increase even after a long period. For all the examples in Table 1, the change in seal stress due to densification is either to increase a tensile seal stress or decrease a compressive seal stress at 4000C. Using Table 2(f), the stress change due to densification at 4000C for the first 100 hours, can be calculated from the regression coefficients; the stress change due to densification is 0.322 At + 0.0016 A20.218 A3 + O.139A4-O.190A5-O.238A6-0.036A7.

In applications of this invention, the densification rate should be kept as low as practicable in keeping with the other requirement described in this specification.

Figure 8 is a diagrammatic representation of a sodium-sulphur cell. In this particular cell, a cylindrical housing 30, typically a metal housing, surrounds a beta-alumina tube 31 which separates a first annular region 32, between the housing 30 and outer surface of the tube 31, from a second annular region 33 between the inner surface of the tube 31 and a co-axial current collector 34. One of these regions is the anodic region and the other is the cathodic region. In this particular example, the cell is of the central sulphur type, and has sulphur/polysulphides forming the cathodic reactant in the region 33 and has sodium in the region 32. The beta-alumina tube 31 is open at the top end and, around this end of the tube, is an alpha-alumina collar 35.A first closure 36 extends between the collar 35 and housing 30 to seal the annular region 32 whilst a second collar 37 extends between the collar 35 and current collector 34 to seal the region 33.

The collar 35 is sealed to the beta-alumina tube 31 by a glass seal using a glass as has been described above. The techniques described above enable a suitable glass to be chosen which will not unduly stress the beta-alumina tube. The same glass may also be used for other seals in the cell, if glass seals are employed. The closure members 36, 37 would usually be metal and may be welded to the housing 30, if the latter is of metal and also welded to the current collector 34 if that is metal. The seal between the closure member and the alpha-alumina collar may use glass or may employ diffusion bonding.

TABLE 1

Densi Trans- Expansion fication mol% formation Deformation Coefficient Room Temp Crossover Rates Temp( C) Temp( c) (25500)x Seal Stress Temperature N/mm/100 Tg Td 10-5 N/mm C Hrs@400 C Case Glass No. No. B2O3 SiO2 Al2O3 BaO CaO MgO SrO Obs Calc Obs Calc Obs Calc Obs Calc Obs Calc Obs Calc 1 597 54.83 0 23.38 21.78 0 0 0 571 566 606 598 6.5 6.36 4.2 2.3 132 182 15.8 15.6 2 644 47.46 7.85 23.13 21.55 0 0 0 576 572 611 608 6.1 6.21 12.9 11.1 263 243 - 13.2 3 667 40.25 15.54 22.89 21.32 0 0 0 578 581 613 620 6.01 6.08 21.9 20.0 345 304 10.0 10.9 4 668 42.85 15.26 22.48 19.44 0 0 0 577 581 613 619 5.91 5.89 27.1 29.0 343 345 12.6 11.6 5 671 47.71 7.37 23.27 15.48 6.17 0 0 588 586 618 619 5.83 6.02 22.8 25.7 323 323 - 11.3 6 676 48.52 6.96 20.46 13.03 6.51 6.51 0 591 594 626 627 6.26 6.23 19.4 19.9 298 278 9.6 9.6 7 681 37.12 7.20 25.97 14.85 7.42 7.42 0 612 608 649 643 6.23 6.24 25.0 26.8 384 373 3.8 5.2 8 688 42.08 6.48 21.50 7.46 14.96 7.49 0 610 614 642 645 6.28 6.27 23.0 26.0 320 329 5.7 5.3 9 689 42.13 6.39 21.52 7.48 7.48 14.98 0 622 616 653 649 6.04 6.01 39.8 42.4 387 422 4.4 4.9 10 690 41.94 6.80 21.43 12.11 5.59 12.13 0 604 605 639 640 6.25 6.26 21.5 25.4 307 327 6.9 6.6 11 692 42.10 6.48 21.51 0 7.48 7.48 14.97 607 613 632 637 6.40 6.40 20.0 24.4 305 305 5.1 5.2 12 698 45.0 14.25 21.23 13.52 6.01 0 0 588 589 624 624 5.89 5.78 26.8 37.2 360 378 - 10.6 13 706 44.0 13.74 20.70 10.79 10.78 0 0 591 597 628 630 5.93 5.89 32.6 34.4 365 361 8.8 9.1 14 720 34.4 14.09 24.06 13.76 6.82 6.87 0 613 610 647 647 6.04 6.04 36.5 35.9 412 415 5.2 4.8 15 722 37.08 7.32 25.93 16.69 3.71 9.26 0 593 605 630 641 6.27 6.24 27.9 26.5 386 372 6.6 5.7 16 723 37.34 6.98 26.12 11.20 14.63 3.72 0 615 614 650 647 6.25 6.22 28.7 28.3 388 380 4.4 4.2 17 725 37.22 6.95 26.04 9.30 13.03 3.73 3.72 607 615 637 645 6.22 6.28 27.6 26.9 385 369 4.3 4.1 18 726 37.23 6.95 26.04 7.44 11.17 3.73 7.44 623 614 645 643 6.29 6.31 26.5 26.5 380 363 4.1 4.1 19 730 41.10 10.0 25.0 10.76 2.39 8.37 8.37 601 597 636 628 6.11 6.03 35.4 33.3 390 383 - 8.0 20 733 42.9 10.0 21.3 7.73 3.87 7.73 6.45 600 662 830 834 6.16 6.05 34.4 34.3 360 364 - 7.5 21 734 28.1 20.0 21.9 9.0 4.5 9.0 7.5 624 618 664 654 6.11 6.23 31.2 31.7 - 384 - 2.3 22 735 44.9 10.0 17.0 4.22 5.61 14.05 4.22 609 609 642 640 6.03 6.09 35.0 35.3 370 392 9.0 6.9 23 736 42.8 10.0 20.7 10.6 1.33 10.6 3.89 598 600 633 634 6.01 6.08 43.8 32.2 360 353 - 7.8 24 737 46.2 10.0 19.4 0 24.4 0 0 621 617 650 644 5.79 5.88 49.9 42.0 394 391 - 6.0 25 738 30.4 20.0 16.9 4.91 6.54 16.35 4.91 623 626 667 662 6.24 6.27 31.0 32.9 374 358 - 1.5 26 740 28.0 20.0 21.2 12.32 1.54 12.32 4.62 619 616 662 659 6.29 6.26 32.4 29.4 - 371 - 2.7 27 741 32.0 20.0 19.7 0 28.3 0 0 634 635 664 666 6.00 6.02 42.5 41.0 430 416 - 0.7 28 743 37.65 15 17.00 4.57 6.06 15.20 4.57 618 619 646 651 6.17 6.18 34.39 34.3 - 351 - 4.1 29 745 35.4 15 20.95 11.46 1.44 11.46 4.30 609 608 647 644 6.31 6.17 30.83 30.8 - 362 - 5.3 30 747 32.0 20 18.7 0 29.03 0 0 637 634 668 665 6.29 6.16 32.33 33.8 - 369 - 0.7 31 750 44.90 10 17.00 10.96 0 17.36 0 608 603 642 638 6.13 6.13 31.89 31.2 - 325 - 8.2 Note:- Room Temperature Seal Stress is compressive and densification stress change is tensile.

TABLE 2(a) Expn. Coeff. Expn. Coeff.

(x10-5) (x10-5) Case No. Observed Value Predicted Value Residual % Deviation 1 6.5### 6.3636 .13638 2.14 26.1###6.2#756.2#75-#1#754-1.73 36.#1##6.#1##6.#851-#75##77E-#1-1.23 45.91##5.8938#.16214E-#1#.28 55.83##6.#195-#.18946-3.15 66.26##6.2346#.25382E-#0#.41 76.23##6.24#4-#.1#369-#1-#17 86.28##6.2714#.85928E-#2#.14 96.#4##6.#115#.28462E-#1#.47 1#6.25##6.2612-#.112#3E-#1-#18 116.4###6.4##4-#.384#3E-#3-#.#1 125.89##5.7838#.1#6231.84 135.93##5.8945#.35542E-#1#.6# 146.#4##6.#368#.32455E-#2#.#5 156.27##6.2375#.32455E-#1#.62 16 6.2 588 6.2224 #.27599E-#1 .44 176.22##6.2813-#.61272E-#1-#.98 186.29##6.3119-#.21869E-#1-#.35 196.11##6.#299#.8#135E-#11.33 206.16##6.#588#.1#4241.72 216.11##6.2286-#.11859-1.9# 226#3##6.#927-#.62698E-#1-1.#3 236.#1##6.#846-#.746#5E-#1-1.23 245.79##5.8849-#.94874E-#1-1.61 256.24##6.2732-#.33152E-#1-#.53 266.29##6.2594#.3#637E-#1#.49 276.####6.#234-#.2344#E-#1-#.39 286.17##6.1842-#.14247E-#1-#.23 29 5.31## 6.1727 .13735 2.23 3 6.29 6.1635 #.1265# 2.5 316.13##6.13#4-#.43482E-#3 STANDARD ERROR OF THE ESTIMATE = 8.65618 E2

Independent Regression Variable Coefficient AB2O3A1#.517888E-#1 BSiO2A2#.354561E-#1 CAl2O3A3-#.6389#5E-#2 DBaOA4#.16866# ECaOA5#.133673 FMgOA6#.9854##E-#1 GSrOA7#.1597#6 TABLE 2(b) Transformation Transformation Temperature (0C) Temperature (0C) Case No. Observed Value Predicted Value Residual % Deviation 1 571 565.6 5.3966 #.95 2 576 571.82 4.1788 #.73 3 578 581.1# -3.#983 -#.63 4 577 58#.6# -3.5953 -.62 5 588 585.88 2.1192 #.36 6 591 594.14 -3.1412 4.53 7 612 68.31 3.6946 .61 B 61# 614.1# -4.#987 -#.67 9 622 616.25 5.7536 .93 1# 6#4 6#5.99 -1.9914 -#.33 11 6#7 612.9# -5.8974 -#.96 12 588 589.34 -1.3359 -#.23 13 591 596.82 -5.8225 -#.98 14 612 6#9.93 2.#7## #.34 15 593 6#5.05 -12.#51 -1.99 16 615 613.94 1.#624 #.17 17 6#7 614.52 -7.5178 -1.22 18 623 614.2# 8.8#27 1.43 19 6#1 597.#5 3.9463 #.66 2# 6## 6#2.66 -26614 -#.44 21 624 618.18 5.8161 .94 22 6#9 6#9.53 -#.53344 -#.#9 23 598 6##.45 -2.449# -#.41 24 621 617.17 3.8344 .62 26 623 626.2# -3.2#15 -#.51 26 619 615.52 3.4798 .57 27 634 634.94 -#.94428 -#.15 28 618 618.29 -#.285#1 -#.#5 29 6#9 6#8.#2 #.98#53 #.16 637 634.65 2.3521 .37 31 6#8 6#2.86 5.1412 #.85 STANDARD ERROR OF THE ESTIMATE = 5.1633

Independent Regression Variable Coefficient A 8203 Al 5.257#3 B SiO2 A2 6.3#455 C Al203 A3 7.27117 D BaO A4 4.92931 E CaO AS 6.97477 F MgO A6 7.26598 G SrO A7 5.865## TABLE 2(c) Deformation Deformation Temperature ( C) Temperature ( C) Case No.Observed Value Predicted Value Residual % Deviation 1 6#6 598.54 7.4556 1.25 2 611 6#7.83 3.1727 .52 3 613 62#.22 -7.2244 1.16 4 613 618.7 -5.6986 -#.92 5 618 619.11 -1.1145 -#.18 6 626 627.29 -1.2857 7 649 643.43 5.5745 .87 8 642 645.#4 -3.#447 -#.47 9 653 648.92 4.#775 .63 839 64#.18 -1.1778 -#.18 11 632 637.45 -5.4513 -#.86 12 624 624.24 -#.24##4 -#.#4 13 628 63#.26 -2.2612 -#.36 14 647 647.## #.17929E-#2 #.## 15 63# 641.45 -11.452 -1.79 16 65# 646.46 3.5426 .55 17 637 645.47 8.4734 -1.31 18 645 643.56 1.44#3 #.22 19 635 628.41 6.5866 1.#5 241 6391 633.82 -3.8239 21 664 654.27 9.7274 1.49 22 642 64#.49 1.514)6 .24 23 633 633.97 -#.973#8 -#.15 24 65# 644.15 5.8468 .91 25 657 662.4)5 -5.#536 -.76 26 662 654.35 7.65(i)2 1.17 27 664 666.18 -2.1789 -#.33 28 646 651.71 -5.7#95 -#.88 29 647 644.2# 2.8#26 #.44 3# 668 665.74 2.2552 #.34 31 642 638.47 3.5295 #.55 STANDARD ERROR OF THE ESTIMATE = 5.7767

Independent Regression Variable Coefficient A B203 Al 5.45372 B SiO2 A2 6.91713 C Al2O3 A3 7.61673 D BaO A4 5.57566 E CaO AS 7.1826# F MgO A6 7#93# G SrO S7 5.865#2 TABLE 2(d) Room Temp. Room Temp.

Seal Stress Seal Stress (N/mm2) (N/mm2) Case No. Observed Value Predicted Value Residual % Deviation 1 4.2000 2.3(684 1.8916 81.94 2 12.9## 11.17 1.7931 16.14 3 21.9## 2#.##2 1.8976 9.49 4 27.1## 29.##6 -1.9#59 -6.57 5 22.8## 25.714 -2.9142 -11.33 6 19.4## 19.91# -#.5#976 -2.56 7 25.### 26.81# -1.8#97 -6.75 8 23.### 26.#77 -3.#767 -11.8# 9 39.8## 42.4#6 -2.6#55 -8.14 1# 21.5## 25.4#9 -3.9#95 -15.39 11 2#.### 24.4#2 -4.4#17 -18.#4 12 28.8## 37.2#1 -8.4##8 -22.58 13 32.6## 34.399 -1.7993 -5.23 14 36.5## 35.98 .59161 1.65 15 27.9## 26.464 1.4359 5.43 16 28.7## 28.25# #.44986 1.59 17 27.60 26.917 .68296 2.54 18 26.5## 26.558 -#.58458E-#1 -#.22 19 35.4## 33.294 2.1#58 6.32 2# 34.4## 34.328 #.72313E-#1#.21 21 31.2## 31.739 -#.539#2 -1.7# 22 35.#1# 35.382 -#.37236 -1.#5 23 43.8## 32.239 11.561 35.86 24 49.97 42.88 7.8817 18.73 25 31.#5# 32.9#1 -1.8511 -5.63 26 32.4## 29.395 3.##5# 1#.22 27 42.5## 41.41 1.459 3.55 28 34.39# 34.3#6 #.84491E-#1 #.25 29 3#.83# 3#.8#2 #.27737E-#1 - #.#9 3# 32.38# 33.817 -1.4337# -4.25 31 31.89# 31.323 #.65735 2.1# STANDARD ERROR OF THE ESTIMATE = 3.9348

Independent Regression Variable Coefficient A B2O3 A1 #.381181 8 SiO2 A2 1.46994 C Al2O3 A3 4.24762 D BaO A4 -5.41328 E CaO AS -2.97645 F MgO A6 -#.775451 G SrO A7 -4.30683 TABLE 2(e) Crossover Crossover Temperature ( C) Temperature ( C) Case No. Observed Value Predicted Value Residual % Deviation 1 132 181.94 -49.942 -27.45 2 263 243.43 19.566 8.4 3 345 3#4.73 441.273 13.22 4 343 345.97 -2.9699 -.86 5 323 323.89 -#.88649 -.27 6 298 278.62 19.379 6.96 7 384 373.45 1#.553 2.83 8 32# 329.39 -9.3863 -2.85 9 387 422.#3 -35.#62 -8.3# 1# 3#7 327.33 -2#.33# -6.21 11 3#5 3#5.65 -0.64716 -#.21 12 35# 378.3# -28.2#1 -7.46 13 365 361.56 3.4433 .95 14 412 415.32 -3.32#4 -#.8# 15 386 372.417 13.931 3.74 16 388 38#.35 7.6496 2.4)1 17 385 369.26 15.738 4.26 18 38# 363.65 16.354 4.5# 19 39# 383.51 6.4855 1.69 2(6 36# 364.84 -4.8424 -1.33 21 35# 384.29 -34.29# -8.92 22 37# 342.6# 27.399 8.## 23 3641 353.17 6.8289 1.93 24 394 391.4)6 2.94#8 .75 25 374 358.13 15.87# 4.43 26 3541 371.12 -21.122 -5.69 27 43# 416.23 13.772 3.31 28 35# 351.67 -1.67## -#.47 29 35# 362.#7 -12.#71 -3.33 3# 3591 369.96 -19.959 -5.39 31 35# 325.5(6 24.496 7.53 STANDARD ERROR OF THE ESTIMATE = 22.723

Independent Regression Variable Coefficient A B2O3 Al 1.26972 B SiO2 A2 9.23752 C Al2O3 A3 31.2554 D BaO A4 -28.3943 E CaO A5 -15.#136 F MgO A6 -2.53581 G SrO A7 -23.2933 TABLE 2(f) Densification Rates (N/mm2 at 4000C in first 100 hours) Case No.Observed Value Predicted Value 1 15.8 15.59 2 - 13.5 3 10 10.96 4 12.6 11.61 5 - 11.28 6 9.6 9.56 7 3.8 5.19 8 5.7 5.29 9 4.4 4.94 10 6.9 6.58 11 5.1 5.14 12 - 10.62 13 8.8 9.13 14 5.2 4.84 15 6.6 5.71 16 4.4 4.23 17 4.3 4.11 18 4.1 4.08 19 - 7.97 20 - 7.45 21 - 2.29 22 9.0 6.79 23 - 7.84 24 - 6.03 25 - 1.52 26 - 2.75 27 - 0.66 28 - 4.14 29 - 5.29 30 - 0.69 31 - 8.16

Independent Regression Variable Coefficient A 8203 Al - 0.322 B SiO2 A2 0.0016 C Al203 A3 -0.218 D BaO A4 0.139 E CaO AS -0.190 F MgO A6 ' -0.238 G SrO A7 -0.036 TABLE3

Transformation Deformation Expansion Room Temp Crossover Densification Rate Temp( C) Temp( C) Coefficient Seal Stress Temperature N/mm/100hrs @ Tg Td (25500)x10-4 N/mm C 400 C Case Glasse No No Obs Calc Obs Calc Obs Calc Obs Calc Obs Calc Obs Calc 32 693 466 564 603 599 6.32 6.60 5.6 2.0 146 166 14.5 14.5 33 694 547 537 580 574 6.67 5.93 -16.9 -14.9 None None 20.9 20 The negative values of the room temperature seal stress indicate tension instead of compression

Claims (16)

1. A glass for sealing to beta-alumina in an electro-chemical cell or other energy conversion device containing sodium, said glass consisting essentially of: (a) 40-60 mol % glass former consisting of 28 4 8 mol % B203 and 0--20 mol % SiO2 (b) 16-28 mol % Al203 (c) 1833 mol % of at least one alkaline earth oxide selected from the group consisting of BaO, CaO, MgO and SrO, where the proportions of the constituents furthermore are such that:: 0.0517 At + 0.0354A2-O.0O63 A3 + 0.168 A4 + 0.1336 A5 + 0.098 A6 + 0.1597 A7 lies between 5.7 and 6.4, where Ar, A2, A3, A4, A5, A6 and A7 are the respective molar percentages of 8203, SiO2, Al203, BaO, CaO, MgO and SrO.

2. A glass as claimed in claim 1 wherein the composition includes at least two alkaline earth oxides.

3. A glass as claimed in either claim 1 or claim 2 wherein the proportion of Al203 is 1728 mol %.

4. A glass as claimed in any of claims 1 to 7 wherein the proportions of the alkaline earth oxides lie in the ranges 0--25 mol % BaO, O15 mol % SrO, 0--30 mol % CaO and D-20 mol % MgO with a minimum total of 18 and a maximum total of 33 mol %.

5. A hermetic seal between a beta-alumina component and another component in an electrochemical cell or other energy conversion device containing sodium, said seal comprising a glass as claimed in claim 1 fused between surfaces of said components.

6. A hermetic seal between an alpha-alumina cQmponent and a beta-alumina component in a sodium-sulphur cell comprising a glass body fused between surfaces of these components, the glass consisting essentially of 28-48 mol % B203, 0-20 mol % SiO2, 16-28 mol % Al203, together with 1833 mol % of at least one alkaline earth oxide selected from the group consisting of BaO, SrO, CaO and MgO, the proportions of the constituents being such that the combined total of 8203 and SiO2 is 40 to 60 mol % and furthermore being such that 0.0517A1+0.0354A2-0.0063A3+0.168A4+0.1336A5+0.098A5+0.1597A7 lies between 5.7 and 6.4 where A1, A2, A3, A4, Ast A5 and A7 are the respective molar percentages of B203, SiO2, Al203, BaO, CaO, MgO and SrO.

7. A hermetic seal as claimed in claim 6 wherein the composition includes at least two alkaline earth oxides.

8. A hermetic seal as claimed in either claim 6 or claim 7 wherein the proportions of the alkaline earth oxides lie in the ranges 0-25 mol % BaO, 0--1 5 mol % SrO, 0--30 mol % CaO and 0-20 mol % MgO with a minimum total of 18 and a maximum total of 33 mol %.

9. A hermetic seal as claimed in any of claims 6 to 8 wherein the proportion of Al203 in the glass lies in the range 17-28 mol %.

10. A glass for sealing beta-alumina in an electro-chemical cell or other energy conversion device, which glass has a composition substantially as set forth in any one of the examples in Table 1 which contains SiO2.

11. An electro-chemical cell containing sodium and having a beta-alumina element sealed to another element by glass fused between the elements, said glass having a composition substantially as set forth in any one of the examples of Table 1 which contains SiO2.

12. A sodium-sulphur cell having a housing containing a beta-alumina tube forming a separator between a cathodic region containing a cathodic reactant comprising sulphur and sodium polysulphides and an anodic region comprising sodium, wherein there is provided a support for a closure for at least one of said regions, said support comprising an alpha-alumina collar around or on the end of the betaalumina tube, said collar being secured to said tube by a hermetic seal formed of glass consisting essentially of (a) 40-60 mol % glass former consisting of 28-48 mol % B203 and 0-20 mol % SiO2 (b) 16-28 mol % Al203 (c) 1833 mol % in total at least one alkaline earth oxide selected from the group consisting of BaO, CaO, MgO and SrO, where the proportions of the constituents furthermore are such that 0.0517 A1 +0.0354A2-0.0063A3+0.168A4-0.1336A5+0.098A5+0.1597A7 lies between 5.7 and 6.4 where A1, A2, A3, A4, A5, A6 and A7 are the respective molar percentages of B203, SiO2, Al203, BaO, CaO, MgO and SrO.

13. A sodium-sulphur cell as claimed in claim 12 wherein the glass composition includes at least two alkaline earth oxides.

14. A sodium-sulphur cell as claimed in either claim 12 or claim 13 wherein the proportions of the constituents furthermore are such that 0.38 A, + 1.47 A2 + 4.25 A3 - 5.41 A4 - 2.98 A5 -0.67 A6 -4.30 A7 is less than 50.

15. A sodium-sulphur cell as claimed in any of claims 12 to 14 wherein the proportions of the constituents furthermore are such that

1.27A1+9.24A2+31.26A3-28.39A4-15.01A5-2.54A5-23.29A7 is in excess of 240.

16. A sodium-sulphur cell as claimed in any of claims 12 to 15 wherein the proportion of Al203 in the glass lies in the range 1728 mol %.