Journal of Materials Science

, Volume 52, Issue 3, pp 1789–1796 | Cite as

Crystallizing glass seals in the system BaO/ZnO/SiO2 with high coefficients of thermal expansion

  • Marita Kerstan
  • Christian Thieme
  • Amr Kobeisy
  • Christian Rüssel
Original Paper


Glasses in the system BaO/ZnO/SiO2 with different BaO/ZnO ratios were melted and powdered and subsequently sieved to different grain size fractions. The powders were sintered to compact samples, then crystallized at temperatures in the range from 850 to 1000 °C and afterwards studied with respect to their phase composition and their thermal expansion behavior. In the glass with large BaO/ZnO ratio, the predominant phase was BaSi2O5. Besides, BaZn2Si2O7 occurred, which became the predominant phase in samples with small BaO/ZnO ratio. Dilatometric measurements showed a steep increase in length up to temperatures in the range from 325 to 375 °C. Then a kink was observed and at temperatures above, the coefficient of thermal expansion was somewhat smaller. The mean thermal expansion coefficients of many crystallized glasses were in the range from 12 to 14 × 10−6 K−1.


Thermal Expansion Size Fraction Cristobalite Glass Composition Glass Ceramic 


Glasses are frequently used for the joining of materials, such as metals or ceramics, especially if beneath the gas tightness electrical insulation is required [1]. If the materials to be joined have comparably small coefficients of thermal expansion (CTEs) (≤8 × 10−6 K−1) or the application temperatures are ≤600 °C, a large variety of glass seals are available [2]. If, however, materials with CTEs >11 × 10−6 K−1 have to be joined and at the same time application temperatures >800 °C are required, then suitable stable glass seals are no longer available, because there seems to be a certain correlation between the CTE and the viscosity, i.e., also the softening temperature of the glass [2]. In many cases, glasses with high softening temperatures (>800 °C) possess fairly low CTEs [3, 4, 5], while glasses with high CTEs (>12 × 10−6 K−1) exhibit low viscosity and hence fairly small softening temperatures [2].

The case that high CTEs and high softening temperatures are required is of special importance for high-temperature reactors or solid oxide fuel cells (SOFCs), which typically are used at temperatures in the range from 700 to 1000 °C [6]. In SOFCs, metals and ceramic materials are to be joined, which commonly have CTEs >11 × 10−6 K−1. For example, perovskites such as La1–xSrxMIIO3±δ (MII = Mn, Co, Fe) are used as materials for the cathodes and possess CTEs of typically 11.7–20.5 × 10−6 K−1 [7]. As anode material, usually a cermet of metallic nickel and stabilized tetragonal zirconia is used, which has a nickel concentration of 30–45 wt% and a CTE (25–1000 °C) of 12.7 to 13.3 × 10−6 K−1 [8]. The interconnect in SOFCs is usually a heat- and oxidation-resistant metal alloy, which has a CTE > 10.0 × 10−6 K−1 [9].

Hence, also the sealing materials have to match these high CTEs. Furthermore, the seals have to withstand the high application temperatures mentioned above. Seals which are able to fulfill these two requirements contain a glassy phase and one or more additional crystalline phases. According to the state of the art, the problem can be overcome by two different strategies: in the first, a crystalline compound with high CTE is added to the glass and then the arrangement is thermally treated, which causes sintering by viscous flow [10]. This procedure can be applied if the volume percentage of the crystalline phase is not too high (<30 %) because otherwise fully densified materials can hardly be obtained [11]. The second strategy is to use a homogenous glass. During thermal treatment, the glass sinters by viscous flow and subsequently a phase with a high CTE is crystallized predominantly at a time where the densification by sintering is nearly completed [12]. These two steps might also be carried out at different temperatures. Seals prepared using the two as mentioned strategies are usually denoted as “composite seals” or “crystallizing seals”, respectively [1, 13].

In order to control densification and crystallization of a crystallizing seal, the glass composition has to be tailored and the temperature/time schedule has to be optimized. Furthermore, the grain size and grain size distribution have to be adjusted to the respective glass composition, procedure, and application.

The final glass ceramic seal should contain crystal phases with high CTEs and phases with low CTEs should be avoided [14, 15]. Also the long-term stability with respect to reactions with materials from the metallic and ceramic reactor materials is of major importance [16, 17].

This paper reports on sealing glasses within the system BaO/ZnO/SiO2. It predominantly describes the phase formation and the resulting CTE.

Experimental procedure

Three glasses with different BaO/ZnO ratios from the system BaO/ZnO/SiO2 were prepared using reagent grade raw materials: SiO2 (Quartz powder C, SCHOTT AG), BaCO3 (chemical pure, REACHIM), and ZnO (pro analysis, LACHEM/CHEMAPOL). Three different glass compositions were chosen (see Table 1), denoted as A, B, and C with decreasing BaO/ZnO ratios. The glass batch (150 g) was mixed using an agate mortar and subsequently melted in a platinum crucible at a temperature of 1400 °C for 3 h. Then the crucible was transferred to another furnace with MoSi2 heating elements and melted at a temperature of 1500 °C kept for another 30 min. Subsequently, the melt was cast on a copper block and then transferred to a preheated muffle furnace allowing the glass to cool down slowly (cooling rate ~3 K/min).
Table 1

Studied glass compositions

















The values are given in [mol%]

The glasses were characterized with respect to their densities using a helium pycnometer (MICROMERITICS Accupyc 1330). The thermal expansion behavior was studied with a dilatometer (NETZSCH Dil 402PC) applying a heating rate of 5 K/min.

The melted glasses were crushed using a steal mortar and sieved to three different grain size fractions <25 μm (small), 25–100 μm (medium) and <100 μm (large), indexed as s, m and l in the last digit of the glass ceramics designations. The glass powders were given to a platinum/gold mold and slightly densified using a spatula. The glass ceramics are also indexed with A, B, and C according to the chemical composition of the glass (see Table 1). Sintering was performed at temperatures in the range from 850 to 1000 °C (indexed in the four digits after the letter for the glass composition) for 0.5, 1, 2, or 4 h (indexed in the following digit 0, 1, 2 and 4, respectively). This means that a sample denoted as A-0950-1-s has the composition of glass A, which was sieved to a grain size < 25 µm (denoted as s) and which was afterwards heat treated at 950 °C kept for 1 h.

The phase analysis was performed with an X-ray diffractometer (RIGAKU MiniFlex 300) and Ni-filtered Cu Kα radiation. The software DIFFRAC.EVA from BRUKER was used for the attribution of the appearing peaks to certain crystalline phases.


Figure 1 shows dilatometric curves for the three studied glasses. The slope of the curves increases with increasing BaO concentration. The glass transition temperature, T g, is in the range from 671 to 696 °C. The results from thermal analysis and pycnometry are summarized in Table 2. The dilatometric curves of the glasses are linear and the CTE decreases slightly below T g. The CTEs are in the range from 7.2 to 9.1 × 10−6 K−1 and increase with increasing BaO concentration. The densities are in the range from 3.794 to 3.931 g·cm−1 and increase within the series C-B-A, i.e., with increasing BaO concentration.
Figure 1

Dilatometric curves of the glasses A, B, and C

Table 2

Densities, ρ, glass transition temperatures, T g, dilatometric softening temperatures, E g and linear coefficients of thermal expansion α 100–500 °C





ρ in g/cm3

3.931 ± 0.002

3.871 ± 0.002

3.794 ± 0.002

T g in  °C

696 ± 5

689 ± 5

671 ± 5

E g in  °C

747 ± 5

724 ± 5

717 ± 5

α 100–500 °C in 10−6·K−1

9.1 ± 0.1

8.1 ± 0.1

7.2 ± 5

The prepared glasses are X-ray amorphous. From the three different compositions, different phases precipitate as illustrated in Fig. 2 for samples heat treated at 1000 °C kept for 1 h. The phase composition is not strongly affected by the particle size and hence, only the patterns of samples with a particle size <25 µm are displayed. On the left side, the diffractograms are shown between 10° and 50°. The whole measuring range was 10°–60°, but the theoretical lines of BaZnSi3O8 are only available up to 50° (see Ref. [18]). This phase is solely described in Ref. [18], and the crystal structure and thermal expansion are unknown. This phase appears in high concentrations in sample C together with the low-temperature phase of BaZn2Si2O7 denoted as LT-BaZn2Si2O7. The latter appears in all three studied glass compositions, but the 2θ-positions of some of the peaks are slightly shifted. Furthermore, the samples A and B also contain LT-BaSi2O5 (Sanbornite). Cristobalite was found in the samples B and C, which was identified by the characteristic peak at 2θ = 21.8°.
Figure 2

X-ray diffraction patterns of crystallized samples obtained from the glasses A, B, and C. The samples were crystallized at 1000 °C. In the lower part, the theoretical XRD-lines of LT-BaSi2O5 (Sanbornite, ICSD 15486, Ref. [24]), LT-BaZn2Si2O7 (Ref. [18]), BaZnSi3O8 (Ref. [18]), and SiO2 (Cristobalite, ICSD 34933, Ref. [25]) are illustrated. The left part shows a 2θ-range from 10° to 50°. The right part shows a narrower range from 11° to 37° and an attribution of all peaks to the respective crystalline phases

A quantitative phase analysis of the patterns illustrated in Fig. 2 was not performed because of the lack in crystal structure information of the BaZnSi3O8 phase.

The results of the dilatometric measurements of glassy and crystallized samples are shown in Figs. 3, 4, 5, and 6. The CTEs of the crystallized samples in any case are larger than those of the corresponding glasses. In principle, crystallized samples show the same tendency as the glasses: the highest CTEs are observed in the samples of series A and the smallest in the samples of series C with the exception of the sample C-950-1-l. The most dilatometric curves show a kink at temperatures in the range from 325 to 375 °C.
Figure 3

Thermal expansion of samples with the composition A prepared from different grain size fractions and crystallized at different temperatures for 1 h

Figure 4

Thermal expansion of samples with the composition A prepared from the small grain size fraction and crystallized at different temperatures for different periods of time

Figure 5

Thermal expansion of samples with the composition B prepared from different grain size fractions and crystallized at different temperatures for different periods of time

Figure 6

Thermal expansion of samples with the composition C crystallized at different temperatures for different periods of time

In the case of crystallized glasses from the sample series A, the shape of the curves does not strongly depend on the particular crystallization time and temperature as shown in Figs. 3 and 4. In Fig. 3, the thermal expansions of samples with the composition A are shown, which were all crystallized for 1 h, but different grain size fractions were used and different crystallization temperatures (950 or 1000 °C) were supplied. If the fine powder fraction was used, the CTE is larger. The samples obtained from the grain size fractions <100 μm and 25–100 μm both crystallized at 950 °C for 1 h show very similar dilatation.

In Fig. 4, samples with the composition A are shown, which were crystallized at three different temperatures (900, 950, and 1000 °C) for 1 h. Furthermore, samples with the composition A crystallized at 950 °C for 30 min, 1 and 2 h are shown. The curves of the samples crystallized for 1 h at 950 and 1000 °C and that crystallized for 2 h at 1000 °C are very similar. The thermal expansion of the sample crystallized for 30 min at 950 °C denoted as A-0950-0-s is clearly not as large and that of the sample crystallized at 900 °C for 1 h is even smaller.

Samples with the composition B (see Fig. 5) are much more different than those of the series A. A crystallization temperature of 850 °C supplied for 4 h results in a less pronounced kink and some more kinks occur at temperatures in the range of 690–880 °C. The thermal expansion in the whole studied temperature range is much smaller than those of samples crystallized at higher temperatures. The samples crystallized at 950 or 1000 °C show a kink at 390 and 370 °C, respectively. Higher crystallization temperatures lead to higher thermal expansion, while the effect of the grain size fraction is negligible if a crystallization temperature of 1000 °C is supplied.

The thermal dilatation behavior of crystallized samples C is shown in Fig. 6. In analogy to sample series B, here also a strong dependency on the crystallization temperature is seen. Surprisingly, the sample crystallized at 950 °C shows the largest thermal expansion of all studied samples. Here, not only the kink at 335 °C occurs, but also a stepwise increase between 175 and 335 °C. A similar effect, but much less pronounced is also observed in the sample crystallized at 1000 °C. A different behavior is observed in the sample crystallized at 850 °C. This sample has a similar expansion as that crystallized at 1000 °C, however, it does not show a kink, but nevertheless at a temperature of 960 °C reaches the dilatation of the sample crystallized at 950 °C.

Table 3 summarizes the CTEs of the crystallized samples in different temperature ranges. According to the shape of the curves, the CTEs depend on the temperature range. The largest CTE is observed in the range from 25 to 300 °C of the sample C crystallized at 950 °C (20.8 × 10−6 K−1). The CTEs of the two samples B crystallized at 1000 °C are also very large (19.1 × 10−6 K−1). The samples A prepared from the smallest grain size fraction and crystallized at 950 °C have somewhat smaller values of 16.3 and 16.1 × 10−6 K−1. In the temperature range from 25 to 600 °C, the samples B crystallized at 1000 °C for 1 h show CTEs as large as 15.3 × 10−6 K−1. Samples with the composition B crystallized from the large grain size fraction at 950 or 1000 °C show CTEs in the range from 14.4 to 14.9 × 10−6 K−1.
Table 3

Technical coefficients of thermal expansion of crystallized glass powders


α 25–300 °C [10−6 K−1]

α 25–600 °C [10−6 K−1]

α 25–900 °C [10−6 K−1]





























































In the temperature range from 25 to 900 °C, the highest CTEs in the range from 13.5 to 13.9 × 10−6 K−1 are found in samples A crystallized at 950 and 1000 °C from the small grain size fraction. The sample C crystallized at 950 °C for 1 h from the large grain size fraction has a CTE of 12.0 × 10−6 K−1.


The dilatometric curves of the glasses are linear (see Fig. 1), but the CTEs decrease slightly below T g, which is supposedly due to a not ideally relaxed glass, i.e., too fast quenching on a copper block. An increase of the CTE as well as of the densities of the glasses with increasing BaO concentration is straightforward and due to the higher atomic weight and the appearance of BaO as a network modifier, respectively.

However, the prediction of the thermal expansion of the glass–ceramics is not possible without the knowledge about the appearing crystalline phases. Interestingly, LT-BaZn2Si2O7 appears in all glass compositions as one of the major phases, independent from the BaO/ZnO ratio of the base glass. At high BaO concentrations, primarily BaSiO5 precipitates, which may also appear in two different phases—a low- and a high-temperature phase. It is described in the literature that higher crystallization temperatures (above around 950–1000 °C) lead to the formation of the thermodynamically stable LT-phase. Lower crystallization temperatures lead to the stabilization of the HT-phase [19]. However, both phases exhibit similar thermal expansion properties and hence, it should have only a minor influence on the thermal expansion which of the two phases, LT- or HT-BaSi2O5 occurs [17, 20].

The thermal expansion behavior of the two mainly occurring phases, LT-BaSi2O5 and LT-BaZn2Si2O7, has already been reported [14, 21]. Both phases occur as the major crystalline phases in the crystallized glasses A and B. Furthermore, both phases undergo phase transitions. According to the phase diagram, the phase transition of BaSi2O5 occurs at around 1350 °C [22]. Hence, it is not surprising that this transition is not observed in the dilatometric curves, i.e., a pronounced volume effect which runs parallel to the phase transition is not detected. The CTE of BaSi2O5 was reported from high-temperature XRD to be 14.4 × 10−6 K−1 and is approximately constant in the entire range up to 1000 °C [14]. In the case of BaZn2Si2O7, the phase transition from the LT- to the HT-phase is clearly detected by dilatometry of the pure phase as well as by high-temperature X-ray diffraction [21]. The phase transition occurs at a temperature of approximately 280 °C and is accompanied by a drastic volume increase of approximately 3.6 %. Below the phase transition temperature, the thermal expansion measured by HT-XRD was reported to be as large as 17.6 × 10−6 K−1 while it is small at temperatures above the phase transition [21].

In the crystallized glasses, a steep increase in volume is not detected at a temperature of 280 °C, but a kink is observed at about 370–400 °C. This kink is due to the as mentioned phase transition of BaZn2Si2O7, which was proven via in situ measurements in Ref. [19]. According to the fact that BaZn2Si2O7 was found in all compositions, also the kink appears in all the three studied compositions (see Figs. 3, 4, 5 and 6).

In the case of the sample A, the preparation parameters have only a minor effect on the thermal expansion of the glass—ceramics. Higher crystallization temperatures and longer holding times generally lead to a higher CTE due to the lower amount of residual glassy phase. In the case of the samples B, the grain size fraction also has only a minor influence on the thermal expansion. The samples from the series B and C also show a kink at around 200 °C or slightly below, which can be attributed to the phase transition of cristobalite. The phase transition of the tetragonal low- to the cubic high-temperature cristobalite according to the phase diagram occurs at 275 °C [23] and is in the case of the glass–ceramics slightly shifted to lower temperatures (see for example C-1000-1-l in Fig. 6). The higher amount of cristobalite leads to significantly higher thermal expansion measured between 25 and 300 °C (see. Table 3) of some of the samples with the compositions B and C.

The measured coefficients of thermal expansion of all studied glass–ceramics are in the range from 7.4 to 20.8 × 10−6 K−1, which makes these glasses suitable as sealing glasses, where these materials are joined to other high thermal expansion materials such as metal alloys. The heat treatment of the base glasses should be well adjusted, because otherwise, dilatation curves of the compositions B and C exhibit two phase transitions (cristobalite and BaZn2Si2O7), which should strongly decrease the resistivity to thermal cycling.


Samples in the BaO/ZnO/SiO2 system with different BaO/ZnO ratios show good glass formation. The powders can be sintered to compact samples and crystallized at temperatures in the range from 850 to 1000 °C. Their coefficients of thermal expansion are in the range from 12 to 14 × 10−6 K−1. In BaO-rich glasses, LT-BaSi2O5 as well as BaZn2Si2O7 predominantly crystallize. The glasses with lower BaO concentrations tend to the formation of disadvantageous cristobalite, which undergoes a phase transition at around 200 °C. The thermal expansion of all samples is generally high up to around 400 °C, then the samples show a kink and a somewhat smaller coefficient of thermal expansion. The high thermal expansion coefficients are advantageous especially if metals with high coefficient of thermal expansion are to be sealed.


Compliance with ethical standards

Conflict of Interest

The authors declare that they have no conflict of interest.


  1. 1.
    Mahapatra MK, Lu K (2010) Glass-based seals for solid oxide fuel and electrolyzer cells—a review. Mater Sci Eng 67:65–85CrossRefGoogle Scholar
  2. 2.
    Donald IW (1993) Preparation, properties and chemistry of glass- and glass-ceramic-to-metal seals and coatings. J Mater Sci 28:2841–2886CrossRefGoogle Scholar
  3. 3.
    Hunger A, Carl G, Gebhardt A, Rüssel C (2008) Ultra-high thermal expansion glass–ceramics in the system MgO/Al2O3/TiO2/ZrO2/SiO2 by volume crystallization of cristobalite. J Non-Cryst Solids 354:5402–5407CrossRefGoogle Scholar
  4. 4.
    Hunger A, Carl G, Gebhardt A, Rüssel C (2010) Young’s moduli and microhardness of glass-ceramics in the system MgO/Al2O3/TiO2/ZrO2/SiO2 containing quartz nanocrystals. Mater Chem Phys 122:502–506CrossRefGoogle Scholar
  5. 5.
    Ghosh S, Das Sharma A, Kundu P, Basu RN (2008) Glass-ceramic sealants for planar IT-SOFC: a bilayered approach for joining electrolyte and metallic interconnect. J Electrochem Soc 155:B473–B478CrossRefGoogle Scholar
  6. 6.
    Apfel H, Rzepka M, Tu H, Stimming U (2006) Thermal start-up behaviour and thermal management of SOFC’s. J Power Sources 154:370–378CrossRefGoogle Scholar
  7. 7.
    Sun C, Hui R, Roller J (2010) Cathode materials for solid oxide fuel cells: a review. J Solid State Electrochem 14:1125–1144CrossRefGoogle Scholar
  8. 8.
    Skarmoutsos D, Tsoga A, Naoumidis A, Nikolopoulos P (2000) 5 mol% TiO2-doped Ni–YSZ anode cermets for solid oxide fuel cells. Solid State Ionics 135:439–444CrossRefGoogle Scholar
  9. 9.
    Yang Z, Weil KS, Paxton DM, Stevenson JW (2003) Selection and Evaluation of Heat-Resistant Alloys for SOFC Interconnect Applications. J Electrochem Soc 150:A1188–A1201CrossRefGoogle Scholar
  10. 10.
    Wang S-F, Lu C-M, Wu Y-C et al (2011) La2O3–Al2O3–B2O3–SiO2 glasses for solid oxide fuel cell applications. Int J Hydrogen Energy 36:3666–3672CrossRefGoogle Scholar
  11. 11.
    Sakuragi S, Funahashi Y, Suzuki T et al (2008) Non-alkaline glass–MgO composites for SOFC sealant. J Power Sources 185:1311–1314CrossRefGoogle Scholar
  12. 12.
    Lara C, Pascual MJ, Durán A (2004) Glass-forming ability, sinterability and thermal properties in the systems RO–BaO–SiO2 (R = Mg, Zn). J Non-Cryst Solids 348:149–155CrossRefGoogle Scholar
  13. 13.
    Brochu M, Gauntt BD, Shah R et al (2006) Comparison between barium and strontium-glass composites for sealing SOFCs. J Eur Ceram Soc 26:3307–3313CrossRefGoogle Scholar
  14. 14.
    Kerstan M, Rüssel C (2011) Barium silicates as high thermal expansion seals for solid oxide fuel cells studied by high-temperature X-ray diffraction (HT-XRD). J Power Sources 196:7578–7584CrossRefGoogle Scholar
  15. 15.
    Kerstan M, Müller M, Rüssel C (2014) High temperature thermal expansion of BaAl2Si2O8, CaAl2Si2O8, and Ca2Al2SiO7 studied by high-temperature X-ray diffraction (HT-XRD). Solid State Sci 38:119–123CrossRefGoogle Scholar
  16. 16.
    Thieme C, Rüssel C (2016) Interfacial reactions between a crystallizing sealing glass from the system BaO–ZnO–NiO–SiO2 and Crofer 22 APU. J Mater Sci 51:756–765CrossRefGoogle Scholar
  17. 17.
    Thieme C, Rüssel C (2014) Cobalt containing crystallizing glass seals for solid oxide fuel cells – A new strategy for strong adherence to metals and high thermal expansion. J Power Sources 258:182–188CrossRefGoogle Scholar
  18. 18.
    Segnit ER, Holland AE (1970) The ternary system BaO-ZnO-SiO2. Aust J Chem 23:1077–1085CrossRefGoogle Scholar
  19. 19.
    Thieme C, Rüssel C (2015) High thermal expansion of crystallized glasses in the system BaO–ZnO–NiO–SiO2. Ceram Int 41:13310–13319CrossRefGoogle Scholar
  20. 20.
    Oehlschlegel G, Kockel A, Biedl A (1974) Anisotrope Wärmedehnung und Mischkristallbildung einiger Verbindungen des ternären Systems BaO–Al2O3–SiO2, Teil I. Messungen an Strukturen mit zweidimensionaler Verknüpfung von (Si, Al)O4-Tetraedern und Angaben über experimentelle Grenzen. Glastech Ber 47:24–30Google Scholar
  21. 21.
    Kerstan M, Müller M, Rüssel C (2012) Thermal expansion of Ba2ZnSi2O7, BaZnSiO4 and the solid solution series BaZn2−xMgxSi2O7 (0 ≤ x≤2) studied by high-temperature X-ray diffraction and dilatometry. J Solid State Chem 188:84–91CrossRefGoogle Scholar
  22. 22.
    Roth RS, Levin EM (1959) Phase Equilibria in the Subsystem Barium Disilicate-Dibarium Trisilicate. J Res Natl Stand 62:193–200CrossRefGoogle Scholar
  23. 23.
    Hatch DM, Ghose S (1991) The α-β phase transition in cristobalite, SiO2: symmetry analysis, domain structure, and the dynamical nature of the β-phase. Phys Chem Miner 17:554–562CrossRefGoogle Scholar
  24. 24.
    Douglass RM (1958) The crystal structure of sanbornite, BaSi2O5. Am Mineral 43:517–536Google Scholar
  25. 25.
    Peacor DB (1973) High-temperature single-crystal study of the cristobalite inversion. Z Kristallogr 138:274–298CrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media New York 2016

Authors and Affiliations

  • Marita Kerstan
    • 1
  • Christian Thieme
    • 1
  • Amr Kobeisy
    • 1
  • Christian Rüssel
    • 1
  1. 1.Otto-Schott-InstitutUniversität JenaJenaGermany

Personalised recommendations