- Email: [email protected]
A new formulation of barium–strontium silicate glasses and glass-ceramics for high-temperature sealant
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 1 1 3 6 0 e1 1 3 6 9
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/he
A new formulation of bariumestrontium silicate glasses and glass-ceramics for high-temperature sealant K. Sharma a,b, G.P. Kothiyal a, L. Montagne b,*, F.O. Me´ar b, B. Revel b a
Glass and Advanced Ceramics Division, Bhabha Atomic Research Centre, Mumbai 400085, India Universite´ Lille Nord de France, UCCS e Unite´ de Catalyse et Chimie du Solide e UMR CNRS 8181, Ecole Nationale Supe´rieure de Chimie de Lille, Universite´ des Sciences et Technologies de Lille, BP 108, 59562 Villeneuve d’Ascq Cedex, France b
article info
abstract
Article history:
P2O5 is a common nucleating agent in glass-ceramics. To avoid the negative effect of P2O5
Received 28 January 2012
on sealing glass properties, we combined it with Barium oxide in the glass formulation. The
Received in revised form
effect of Ba3(PO4)2 incorporation on thermo-physical, structural and sealing characteristics
11 April 2012
of a glass of composition (mol%) 30SiO2e20SrOe30BaOe10B2O3e5La2O3e5Al2O3 is reported.
Accepted 29 April 2012
The glass transition temperature decreases from 650 C to 584 C when Ba3(PO4)2 is
Available online 19 June 2012
increased from 0 to 2 mol%. Thermal expansion coefficient of the glass samples are found around 11.4 to 12.0 106 C1, which is close to that of Crofer-22 APU.
Keywords:
29
Si MAS-NMR
measurements indicate that glass network is composed of mainly Q2 units, while
31
P
0
SOFC sealant
forms Q units. XRD shows the formation of Ba2SiO4, Sr2SiO4 and BaAl2Si2O8 crystalline
Thermal properties
phases. There is no further evolution of crystalline phases when heat treated at 800 C for
XRD
500 h. Seals with Crofer-22 plates have been made and interface reveals good bonding. The
MAS-NMR
glass composition formulated by addition of 1 mol% Ba3(PO4)2 has shown good thermal and sealing characteristics, while further increase of Ba3(PO4)2 to 2 mol% leads to increased crystallization tendency. Copyright ª 2012, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.
1.
Introduction
Solid oxide fuel cells (SOFCs) convert chemical energy into electrical energy, which presents higher efficiency and lower environmental impact as compared to conventional power generation systems [1,2]. The SOFC cells operate at 600e900 C for thousands of hours, and, for efficient and effective operation, it requires a hermetic seal for channeling of fuel and oxygen (or air). In addition, sealing material has to meet several rigorous requirements, including chemical and electrical compatibility with the cell components [3]. Glass and glass-ceramics are explored as promising materials for sealing application in SOFC [4]. A major challenge for this kind of sealants is to avoid the formation of
undesirable crystalline phases during working of the cell, for instance Celsian (BaAl2Si2O8) and its polymorph hexacelsian crystalline phases having low thermal expansion coefficients (TEC), which induces stress in the seal and lead to failure of seals [5]. Further, during long-term operation, interfacial reactions of BaO with Crofer-22 APU material lead to formation of BaCrO4 phase which is detrimental for long-term stability of seal [6,7]. In order to develop suitable sealant materials for high temperature applications, several glass and glass-ceramics mainly based on silicates and borates have been studied [8e10]. BaOeCaOeAl2O3eB2O3eSiO2 (BCABS) glass system has been studied for joining dissimilar metal and ceramic SOFC materials [11]. Haanappel et al. [12] studied the behavior of
* Corresponding author. E-mail address: [email protected] (L. Montagne). 0360-3199/$ e see front matter Copyright ª 2012, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.ijhydene.2012.04.142
11361
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 1 1 3 6 0 e1 1 3 6 9
BCABS seals in presence of oxidizing and reducing atmospheres and reported excess corrosion of interconnect. Lara et al. [13,14] studied the ROeAl2O3eBaOeSiO2 (R ¼ Ca, Mg, Zn) glass system and reported their glass forming ability, sinterability, thermal and electrical properties. As mentioned above, significant amount of BaO in glasses leads to degradation of seals due to formation of BaCrO4. Thus, in order to avoid this, SrO based glasses have been studied. Tiwari et al. [15] reported bonding of SrO and ZnO based silicate glasses with Crofer-22 APU. Ojha et al. [16] studied the SrO and La2O3 based Al2O3eB2O3eSiO2 glasses and reported that SrO modified the network and resulted in change in thermo-physical properties. Thermal expansions of coefficients of glasses are repor ted around 9 106 C1. Wang et al. [17] prepared La2O3eAl2O3eB2O3eSiO2 glasses free of alkaline earth metals and reported a decrease in softening temperature of glasses with an addition of La2O3. However, there is still a need to reduce the softening temperature (Ts), so that seals can be prepared at temperatures lower than 1000 C. It is also desired to match the TEC (12 106 C1) with the metallic interconnects of SOFCs. In the present study, SrOeBaO based aluminosilicate glasses with various compositions are prepared by melt quench technique. The system is of particular interest as replacing BaO with SrO mitigates the formation of BaCrO4. We also expect that with this we may induce the formation of SrAl2Si2O8, which has higher TEC than BaAl2Si2O8. Further, originality of our approach lies in the combined addition of BaO and P2O5 to tailor the sealing and thermo physical properties of the sealants. Larsen et al. reported that phosphate based glass exhibit insufficient durability for sealing but addition of silica can increase the stability of phosphate in the glass [18]. Moreover, P2O5 can be useful as nucleating agent to avoid undesirable reaction at the metal glass interface [19]. Finally, addition of P2O5 in alumino-silicate glasses scavenges the modifier cations and increase the network polymerization, hence viscosity and sealing temperature increase, which is detrimental for the sealing application [20]. Thus a series of glasses have indeed been formulated by addition of Ba3(PO4)2 to a base glass composition. According to previous studies [21], it is expected that with this combined addition, the network polymerization will not get affected but we will get a decrease in glass transition and softening temperatures. Thermo-physical properties of the glasses (Tg, Ts and TEC) with increasing quantity of Ba3(PO4)2 have been studied. The glasses are characterized for structural features with XRD and solid-state NMR. Seals of the glasses with Crofer-22 APU metallic alloy were made, and the microstructure at the interface and chemical stability of seals at high temperature were investigated.
2.
Experimental
Nominal compositions (mol%) of glass samples are given in Table 1. These were prepared by using melt-quench technique. Al(OH)3, BaCO3, H3BO3, La2O3, NH4H2PO4, SiO2 and SrCO3 of 99.9% purity were mixed, ground and calcined at a maximum temperature of 900 C in a Pt-10% Rh crucible. It is to be noticed that Ba3(PO4)2 is obtained in the glass
Table 1 e Nominal BASP glass composition (mol%). The glasses were formulated by addition of Ba3(PO4)2 to BASP0. Sample
SiO2
BaO
SrO
B2O3
La2O3
Al2O3
P2O5
BASP0 BASP1 BASP2
30 28.8 27.7
30 31.7 33.3
20 19.2 18.5
10 9.6 9.2
5 4.8 4.6
5 4.8 4.6
0 0.9 1.8
composition from the decomposition of BaCO3 and NH4H2PO4. The calcined charge was then melted at 1450e1550 C and held at this temperature for 1 h for homogenization. The melt was then poured on a stainless steel plate. The glass samples were annealed at around 650 C for 4 h. Weight losses were carefully checked during glass preparation, hence actual composition is considered to be similar to nominal composition. Differential thermal analysis (DTA) was carried out from 20 C to 950 C using a TG/DTA instrument (Model Setsys, Setaram, France). DTA measurements were carried out on powdered samples with particle size less than 63 mm. Thermal expansion was measured from 20 C to 900 C using TMA instrument (Model Labsys, Setaram, France). Thermal expansion coefficient was then measured in the linear part of the dilatation curve. The crystalline phases in glass-ceramics samples were identified using powder X-ray diffractometer (Bruker) with CuKa as X-ray source. XRD data were collected on powder samples with particle size less than 63 mm. These measurements were done in the range (2q angle) 10e80 with a step of 0.02 s1 and dwell time of 0.2 s. The sintering behavior of the BASP glasses was studied by means of a hot stage microscope (HSM, Hesse, Germany). For these measurements, the sample powders were pelletized and placed on an Alumina support. This was then placed within a furnace and heated at 10 C$min1. The shape of the sample was monitored using a camera. The projected area of the sample viewed in side elevation was then used to characterize the sintering, deformation, sphere and flow temperatures of the glasses. 11 B, 27Al, 29Si and 31P MAS-NMR spectra were recorded at 18.8, 18.8, 9.4 and 9.4T respectively, on Bruker AVANCE spectrometers, with 4 mm probes at 12.5 kHz spinning speed, except for 29Si for which a 7 mm probe at 5 kHz was used. The Larmor frequencies were 128.4, 104.2, 79.4 and 162.3 MHz for 11 B, 27Al, 29Si and 31P, respectively. For 11B, the pulse duration was 2 ms (p/6), and the recycle delay was 10 s. For 27Al, the pulse duration was 1.5 ms (p/8), and the recycle delay was 2 s. For 29Si, the pulse duration was 1.6 ms (p/5), and the recycle delay was 180 s. For 31P, the pulse duration was 1.6 ms (p/6), and the recycle delay was 120 s. All relaxation delays were chosen long enough to enable relaxation at the field strength that was used. The 11B chemical shifts are given relative to BPO4 at 3.6 ppm, 27Alchemical shifts are relative to AlCl3 at 0 ppm, 29Si chemical shifts are relative to tetramethylsilane (TMS) at 0 ppm, and 31P are relative to 85% H3PO4 at 0 ppm. For studying the adhesion/bonding, sealing with Crofer-22 alloy was carried out at a maximum temperature of 1000 C employing the sandwich geometry. Microstructure near the interface was investigated using Cameca SX 100 electron
11362
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 1 1 3 6 0 e1 1 3 6 9
Table 2 e TMA data for BASP glass samples. TEC (0.2 106 K1)
Sample As prepared
After 6 h at 800 C
After 100 h at 800 C
11.4 12 13
12.2 12 13
12.4 12.5 12.1
BASP0 BASP1 BASP2
Tg (2 C)
Flow temperature (3 C)
650 592 584
1145 1107 1108
probe microanalyser (EPMA) and wavelength dispersive analysis of X-Rays (WDAX) was carried out for determining element diffusion across the interface.
3.
Results
The compositions reported in Table 1 yielded bubble-free glasses, named BASP in the following text. XRD analysis of these glasses has shown absence of any long range order.
3.1. Thermo mechanical analysis (TMA)/differential thermal analysis (DTA) The thermo-physical properties of the BASP glasses are listed in Table 2. The glass transition temperature has decreased from 650 C to 584 C when P2O5 was increased from 0 to 1.8 mol% and BaO from 30 to 33.3 mol%. The TEC values of the glass samples are found to increase from 11.4 to 13 106 C1. DTA plots of glass samples are shown in Fig. 1. The sample BASP0 shows an onset for crystallization (Tcry) at 756 C. Addition of Ba3(PO4)2 does not modify significantly this value, although onset of crystallization is detected at slightly higher temperature (764 C) for BASP1.
3.2.
Hot stage microscopy (HSM)
The normalized areas of the BASP samples as recorded with HSM are plotted as a function of temperature in Fig. 2. The incorporation of Ba3(PO4)2 does not modify the sintering temperature, which remains at 710 C. The
Fig. 1 e DTA plots of BASP glasses.
Fig. 2 e Normalized area versus temperature of BASP glasses obtained from Hot Stage Microscopy (HSM).
maximum shrinkage temperature (Tshri) and flow temperature decrease with addition Ba3(PO4)2. The shape of the normalized area versus temperature curves indicates that the BASP0 and BASP1 glasses soften over a wide temperature range indicating glassy behavior. BASP2 glass exhibit a wide plateau before flow, which is a characteristic of materials exhibiting devitrification upon heating [22]. This indicates the occurrence of significant crystallization in BASP2 glass.
3.3.
X-ray diffraction (XRD)
XRD patterns of the glass samples after heat treatment at 800 C for 6 h are depicted in Fig. 3a. The glass BASP0 shows the formation of Ba2SiO4, Sr2SiO4 and BaAl2Si2O8 as major phases. However, with addition of 1 mol% Ba3(PO4)2, relative intensity of peak corresponding to BaAl2Si2O8 (around 23 ) has decreased. It implies tendency of formation of such a phase reduces when Ba3(PO4)2was added to base composition (BASP0). Overall, the peaks have become sharp, which implies better crystallized samples. Evolution of crystalline phases in BASP0 glass after different heat treatment time is shown in Fig. 3b. XRD peaks have become sharp when sample were heat treated for 500 h. Intensities of peaks corresponding to formation of BaAl2Si2O8 have decreased (for instance at 22.5 ) and while peak at 27 for Sr2SiO4 has increased. It implies that the tendency to form Sr2SiO4 is more prevalent than formation BaAl2Si2O8 when glass is heat treated at 800 C for long time (500 h). XRD plots
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 1 1 3 6 0 e1 1 3 6 9
11363
Fig. 3 e XRD plots of: (a) BASP glasses after heat treatment at 800 C for 6 h; (b) BASP0 glass after heat treatment at 800 C from 6 h to 500 h; (c) BASP glasses after heat treatment at 800 C for 100 & 500 h.
11364
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 1 1 3 6 0 e1 1 3 6 9
for the other glass samples (BASP1&2) after heat treatment at 800 C for 100 to 500 h are shown in Fig. 3c. There was no evolution of additional phase when heat treatment time was increased from 100 h to 500 h.
3.4.
NMR
3.4.1.
29
Si MAS-NMR
Fig. 4 shows the 29Si MAS-NMR spectra of the as-prepared BASP glasses and after 100 h heat treatment at 800 C. The spectra of the as prepared glasses show a broad feature in the w70ew90 ppm range with a Gaussian line shape, characteristic of bond angle and length distributions inherent to the amorphous state. Its chemical shift is centered on w80 ppm, which indicates that Q2 SiO4 units dominate in the glass network [23]. This result is in accordance with the glass composition, which contains a large quantity of network modifying oxides. Some Q3 SiO4 units bonded to Al or B may be present, since chemical shifts of such units shift toward less negative values [24,25]. After 100 h of heat treatment at 800 C, large modifications are observed on the 29Si MAS-NMR spectra (Fig. 4). Two narrow resonances with large intensity are now present at w68 and w89 ppm, which indicate the presence of crystalline phases. According to literature data, they can be assigned to Ba2SiO4, with chemical shift reported at w70 ppm, and to BaAl2Si2O8, with chemical shift reported at 88 ppm [26,27]. The resonance at w68 ppm may also be assigned to Sr2SiO4 or a mixed (Ba,Sr)SiO4, owing to the close values of Ba2þ and Sr2þ field strengths [28]. The weak resonance at w85 ppm may be assigned to (Ba,Sr)Al2Si2O8 mixed Celsian phase.
3.4.2.
11
B MAS-NMR
11
The B NMR spectra of the parent and heat treated BASP glasses are shown in Fig. 5. The spectra are similar for all the compositions. The resonance observed at 16 ppm can be assigned to BO3 and a very weak one at 1 ppm assigned to BO4 units [29]. Despite the use of a high-field spectrometer that enables a high resolution for quadrupolar nuclei like 11B
Fig. 4 e 29Si MAS-NMR spectra of as prepared BASP glasses and after heat treatment at 800 C for 100 h.
Fig. 5 e 11B MAS-NMR spectra of as prepared BASP glasses and after heat treatment at 800 C for 100 h.
(nuclear spin I ¼ 3/2), the spectra are broad and featureless, meaning that all 11B nuclei are present in a very similar environment. After heat treatment for 100 h (Fig. 4), the 11B NMR spectra of glass-ceramics remain quite similar to those of the parent glasses. Nevertheless, the main resonance shows on its top the presence of two singularities around w18 and w17 ppm. They suggest that some boron nuclei are involved in regular crystalline environments [30]. Celsian is indeed known to incorporate many different elements in its structure to form substitution solid solution [27,31]. Concerning boron, we have already reported that NMR gave evidence for the formation of a mixed Al, B phase [32].
3.4.3.
27
Al MAS-NMR
27
The Al NMR spectra for BASP glasses and glass-ceramics are reported in Fig. 6. The spectra of glasses consist of
Fig. 6 e 27Al MAS-NMR spectra of as prepared BASP glasses and after heat treatment at 800 C for 100 h.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 1 1 3 6 0 e1 1 3 6 9
11365
expected. A much narrower resonance is observed for the glass-ceramics, thus revealing that Al is fully involved in the crystalline phases. The spectra consist of a resonance centered at w55 ppm, which is characteristic of AlO4 tetrahedral units in Celsian (Ba,Sr)Al2Si2O8 crystals [26,34]. The broad line shape at the bottom of 27Al resonances of glass-ceramics indicate that some Al remains in the residual glass.
3.4.4.
31
P MAS-NMR
31
Fig. 7 e 31P MAS-NMR spectra of as prepared BASP glasses and after heat treatment at 800 C for 100 h.
a broad resonance centered at w65 ppm, which is characteristic of aluminum in tetrahedral coordination [33]. Owing to the large quantity of charge compensating cations available in the glasses, this tetrahedral coordination was
The P NMR spectra of the glasses and glass-ceramics are shown in Fig. 7. A main broad resonance is observed at w8 ppm, which is assigned to Q0 (orthophosphate units) [35]. This chemical shift is close to that of Q0 units in Ba3(PO4)2 crystal (9 ppm) [36], but different from that of Sr3(PO4)2 crystal (3 ppm) [37]. Hence we expect that the orthophosphate anions are mainly charge balanced by Ba2þ but some Sr2þ cannot be excluded. The resonance becomes somewhat sharper in glass-ceramic samples. Nevertheless, the Ba and Sr atomic distribution in Q0 orthophosphate site may be at the origin of broad line shape, thus we can expect the presence of phosphate crystal in the glass-ceramics samples. Since the P2O5 quantity is small in these glasses, the presence of these phosphate crystals could not be confirmed by XRD analyses.
Fig. 8 e Microstructure at the interface of glass with Crofer-22 APU: (a) BASP0; (b) BASP1; (c) BASP2; (d) EPMA line scan of BASP1.
11366
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 1 1 3 6 0 e1 1 3 6 9
Fig. 9 e Microstructure near the interface between glass and Crofer-22 APU after heat treatment at 800 C: (a) BASP0 (for 100 h); (b) BASP0 (for 500 h); (c) BASP1 (for 100 h); (d) BASP1 (for 500 h); (e)BASP2 (for 100 h); (f) BASP2 (for 500 h).
3.5.
Microstructure and chemical stability
Fig. 8(aec) shows the microstructures at interfaces of BASP glasses with Crofer-22 APU. All the sealing glasses bonded well to the metallic part as no bubbles and cracks were observed at the interface. A limited interdiffusion of Cr, Sr and Ba across the boundary is observed, which is required for good adhesion between metal and glass (Fig. 8d). Fig. 9(aef) shows the EPMA images recorded at the interface of metal/glass joints after heat treatment at 800 C for 100 h and 500 h. The microstructure of BASP0 is composed of coarse crystals as seen from EPMA images, while BASP1and BASP2 show more homogeneous microstructure with smaller crystals. There were no major changes observed in microstructure after heat treatment at 800 C for 500 h. As seen from the images, all the seals were found with good interfaces. EPMA images of seals confirm limited enrichment of Cr at the interface (Fig. 10(aef)) even after heat treatment at 800 C for 500 h. Fig. 11 shows the comparison of Cr diffusion at interface
of BASP1 glass with Crofer-22 APU alloy after different time of heat treatment at 800 C. In BASP1 glass, Cr diffusion of around 3 mm is observed for as prepared seal and it increases marginally to 5 mm after 500 h of heat treatment at 800 C. Mapping of the other elements at the interface of all glassceramics compositions did not reveal any significant differences; hence the images are not reported.
4.
Discussion
Thermo-physical properties of glasses and glass-ceramics are strongly dependent upon the composition [38]. According to continuous random network theory, SiO2 is a network forming oxide and BaO is a network modifying oxide [39]. The addition of barium phosphate leads to decrease in overall silica content, which is expected to decrease the rigidity of the network. Thus the TEC values of BASP glasses increase with addition of barium phosphate, concomitantly Tg values decrease.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 1 1 3 6 0 e1 1 3 6 9
11367
Fig. 10 e Cr mapping at the BASP glass Crofer-22 APU interface after heat treatment at 800 C: (a) BASP0 (for 100 h); (b) BASP0 (for 500 h); (c)BASP1 (for 100 h); (d) BASP1 (for 500 h); (e) BASP2 (for 100 h); (f) BASP2 (for 500 h).
It is known that modifier cations are required for charge compensation for PO3 4 anions, when P2O5 is added to glass. But no evolution in 29Si spectra of the as-prepared glasses is observed for the 3 different compositions (Fig. 4). This was expected, since we formulated the glasses by the simultaneous addition of BaO and P2O5 in proportion 3/1, to obtain an equivalent amount of Ba3(PO4)2. It has indeed been shown that, when Ba3(PO4)2 is added to silicate glasses in small amount, P2O5 does not participate to the silicate glass network and remains as isolated orthophosphate species [40]. As a result, the Al coordination remains primarily tetrahedral and B forms mainlyBO3units. The formation of less rigid planar BO3, units as observed in 11B NMR contributes to keep moderate values of flow temperature and Tg for BASP glasses.
Fig. 11 e Cr diffusion at BASP1 glass e Crofer-22 APU interface after different time of heat treatment at 800 C.
For effective sealing, glass should attain maximum shrinkage before crystallization [13]. All the BASP glasses have shown attainment of maximum shrinkage before crystallization. However, crystallization becomes more important and occurs earlier when Barium phosphate was increased to 2 mol % (Fig. 2). MAS-NMR spectra of the glass-ceramics are in accordance with the phase emergence observed by XRD. Since the incorporation of Ba3(PO4)2 does not affect the network the major phases formed are not much changed. However, there is an increased crystallization and phase stabilization. We observed the formation of Ba2SiO4, Sr2SiO4 and BaAl2Si2O8 phases. As crystallization of B is evident on B MAS-NMR spectra, we suggest the formation of (Ba,Sr)(Al,B)2Si2O8 like phase. Identification of such a phase is not possible due to inherent limitation of XRD. In glass-ceramics, thermo-physical properties depend on the crystalline phase evolved during crystallization. BASP glass-ceramics have high TEC (12 106 C1) despite the crystallization of hexacelsian like phase, which has lower than desired TEC (8 106 C1) [3]. This also suggests the formation of (Ba,Sr)Al2Si2O8; since SrAl2Si2O8 has a higher TEC (11 106 C1). B incorporation in such a phase may further lead to increase in TEC. The decrease in flow and sealing temperatures with addition of Ba3(PO4)2 in BASP glasses helps in forming good bonds with Crofer-22 APU at temperature lower than 1000 C, however we have carried out all the sealing experiment at 1000 C for comparing the results for BASP glasses. All the seals with Crofer-22 APU have shown good adhesion. The seals were found intact after heat treatment at 800 C up to 500 h. It has been reported that the BaO containing glasses interact chemically with chromia-forming alloys forming BaCrO4, which leads to the separation of GCs from the alloy matrix due to thermal expansion mismatch [6]. In the present investigation, we did not observe major enhancement in
11368
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 1 1 3 6 0 e1 1 3 6 9
Cr-diffusion after 500 h at 800 C and no adverse reactions at interface were observed even after 500 h of heat treatment. Therefore, we conclude that incorporation of Ba3(PO4)2 improves the sealing characteristics of BASP glasses/glassceramics. Longer test duration is under investigation to evaluate the effect of Barium phosphate addition on the very long term.
5.
Conclusions
A new way of formulating sealing glasses by the combined addition of P2O5 and BaO as Ba3(PO4)2 has been presented in the present paper. These barium alumino-silicate glasses are studied for structural, thermo-physical properties and sealing properties. The glasses do not show major structural changes with addition of BaO and P2O5 in the form of Ba3(PO4)2. Moreover, this leads to decrease in softening temperatures, thus enabling seal fabrication (with Crofer-22 APU) below 1000 C. Interfaces of seals reveal good bonding with Crofer-22 APU, which did not deteriorate even after heat treatment at 800 C for 500 h. The glasses show the development of Ba2SiO4, BaAl2Si2O8, and Sr2SiO4 crystalline phases upon heat treatment at 800 C, which are not found to be detrimental for high temperature sealing applications.
Acknowledgments The authors thank the IFCPAR for funding this work vide project number 4008-1. The FEDER, Re´gion Nord Pas-de-Calais, Ministe`re de l’Education Nationale de l’Enseignement Supe´rieur et de la Recherche, CNRS, and USTL are acknowledged for funding of NMR spectrometers.
references
[1] Steele BCH, Heinzel A. Materials for fuel-cell technologies. Nature 2001;414:345e52. [2] Pascual MJ, Guillet A, Duran A. Optimization of glasseceramic sealant compositions in the system MgOeBaOeSiO2 for solid oxide fuel cells (SOFC). J Power Sources 2007;169:40e6. [3] Fergus JW. Sealants for solid oxide fuel cells. J Power Sources 2005;147:46e57. [4] Ley KL, Krumpelt M, Kumar R, Meiser JH, Bloom I. Glasseceramic sealants for solid oxide fuel cells: part I. Physical properties. J Mater Res 1996;11:1489e93. [5] Brochu M, Gauntt BD, Shah R, Miyake G, Loehman RE. Comparison between barium and strontium glass composites for sealing SOFCs. J Eur Ceram Soc 2006;26: 3307e13. [6] Batfalsky P, Haanappel AAC, Malzbender J, Menzler NH, Shemel V, Vinke IC. Chemical interaction between glassceramic sealant and interconnect steels in SOFC stacks. J Power Sources 2006;155:128e37. [7] Yang Z, Xia G, Meinhardt KD, Weil KS, Stevenson JW. Chemical stability of glass seal interfaces in intermediate temperature solid oxide fuel cells. J Mater Eng Perform 2004; 13:327e34. [8] Mahapatra MK, Lu K. Glass-based seals for solid oxide fuel and electrolyzer cells e a review. Mater Sci Eng R 2010;67: 65e85.
[9] Yang Z, Meinhardt KD, Stevenson JW. Chemical compatibility of barium-calcium-aluminosilicate-based sealing glasses with the ferritic stainless steel interconnect in SOFCs. J Electrochem Soc 2003;150:A1095e101. [10] Ghosh S, Sharma AD, Kundu P, Mahanty S, Basu RN. Development and characterizations of BaOeCaOeAl2O3eSiO2 glasseceramic sealants for intermediate temperature solid oxide fuel cell application. J Non-Cryst Solids 2008;354: 4081e8. [11] Sohn SB, Choi SY, Kim GH, Song HS, Kim GD. Stable sealing glass for planar solid oxide fuel cell. J Non-Cryst Solids 2002; 297:103e12. [12] Haanappel VAC, Shemet V, VinkeI C, Quadakkers WJ. A novel method to evaluate the suitability of glass sealantealloy combinations under SOFC stack conditions. J Power Sources 2005;141:102e7. [13] Lara C, Pascual MJ, Duran A. Glass-forming ability, sinterability and thermal properties in the systems ROeBaOeSiO2 (R ¼ Mg, Zn). J Non-Cryst Solids 2004;348: 149e55. [14] Lara C, Pascual MJ, Prado MO, Duran A. Sintering of glasses in the system ROeAl2O3eBaOeSiO2 (R ¼ Ca, Mg, Zn) studied by hot-stage microscopy. Solid State Ionics 2004; 170:201e8. [15] Tiwari B, Dixit A, Kothiyal GP. Study of glasses/glassceramics in the SrOeZnOeSiO2 system as high temperature sealant for SOFC applications. Int J Hydrogen Energy 2011;36: 15002e8. [16] Ojha PK, Rath SK, Chongdar TK, Gokhale NM, Kulkarni AR. Physical and thermal behaviour of SreLaeAleBeSi based SOFC glass sealants as function of SrO content and B2O3/SiO2 ratio in the matrix. J Power Sources 2011;196:4594e8. [17] Wang SF, Lu CM, Wu YC, Yang YC, Chiu TW. La2O3eAl2O3eB2O3eSiO2 glasses for solid oxide fuel cell applications. Int J Hydrogen Energy 2011;36:3666e72. [18] Larsen PH, Poulsen FW, Berg RW. The influence of SiO2 addition to 2MgOeAl2O3e3.3P2O5 glass. J Non-Cryst Solids 1999;244:16e24. [19] Donald IW, Metcalfe BL, Gerrard LA. Interfacial reactions in glass-ceramic-to-metal seals. J Am Ceram Soc 2008;91: 715e20. [20] Ananthanarayanan A, Kothiyal GP, Montagne L, Tricot G, Revel B. The effect of P2O5 on the structure, sintering and sealing properties of barium calcium aluminum boro-silicate (BCABS) glasses. Mater Chem Phy 2011;130:880e9. [21] Grussaute H, Montagne L, Palavit G. Structural interpretation of the modification of soda-lime-silica glass properties by P2O5. Glass Sci Technol 2000;73:380e4. [22] Boccaccini AR, Hamann B. Review in situ high-temperature optical microscopy. J Mater Sci 1999;34:5419e36. [23] Goswami M, Kothiyal GP, Montagne L, Delevoye L. MAS-NMR study of lithium zinc silicate glasses and glasseceramics with various ZnO content. J Solid State Chem 2008;181:269e75. [24] Goel A, Tulyaganov DU, Pascual MJ, Shaaban ER, Munoz F, Lu Z, et al. Development and performance of diopside based glasseceramic sealants for solid oxide fuel cells. J Non-Cryst Solids 2010;356:1070e80. [25] Martens R, Muller-Warmuth W. Structural groups and their mixing in borosilicate glasses of various compositionsean NMR study. J Non-Cryst Solids 2000;265:167e75. [26] Mackenzie KJD, Smith ME. Multinuclear solid-state NMR of Inorganic materials. p. 230e242. London: Pergamon; 2001. [27] MacKenzie KJD, Kemmitt T. Evolution of crystalline aluminates from hybrid gel-derived precursors studied by XRD and multinuclear solid-state MAS NMR: I. Celsian, BaAl2Si2O8. Thermochimica Acta 1999;325:5e12. [28] Murdoch JB, Stebbins JF, Carmichael ISE. High-resolution 29 Si NMR study of silicate and aluminosilicate glasses: the
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 1 1 3 6 0 e1 1 3 6 9
[29]
[30]
[31]
[32]
[33]
effect of network-modifying cations. Am Mineral 1985;70: 332e43. Ananthanarayanan A, Kothiyal GP, Montagne L, Revel B. MAS-NMR studies of lithium aluminum silicate (LAS) glasses and glasseceramics having different Li2O/Al2O3 ratio. J Solid State Chem 2010;183:120e7. Stamboulis A, Hill RG, Law RV. Characterization of the structure of calcium alumino-silicate and calcium fluoro-aluminosilicate glasses by magic angle spinning nuclear magnetic resonance (MAS-NMR). J Non-Cryst Solids 2004;333:101e7. Lee KT, Aswath BP. Role of mineralizers on the hexacelsian to celsian transformation in the barium aluminosilicate (BAS) system. Mater Sci Eng A 2003;352:1e7. Ananthanarayanan A, Kothiyal GP, Montagne L, Revel B. MAS-NMR investigations of the crystallization behaviour of lithium aluminum silicate (LAS) glasses containing P2O5 and TiO2 nucleants. J Solid State Chem 2010;183: 1416e22. Bunker BC, Kirkpatrick RJ, Brow RK, Turner GL, Nelson C. Local structure of alkaline-earth boroaluminate crystals and glasses: II, 11B and 27Al MAS NMR spectroscopy of alkalineearth boroaluminate glasses. J Am Ceram Soc 2005;74:1430e8.
11369
[34] Stebbins JF, Kroecker S, Lee SK, Kiczenski TJ. Quantification of five and six coordinated aluminum ions in aluminosilicate and fluoride-containing glasses by high-field, highresolution 27Al NMR. J Non-Cryst Solids 2000;275:2425e32. [35] Grimer AR, Haubenreisser U. High field static and MAS 31P NMR: chemical shift tensor of polycrystalline potassium phosphate P2O5 K2O. Chem Phys Lett 1983;99:487e90. [36] Turner GL, Smith KA, Kircpatrick RJ, Oldfield E. Structure and cation effects on phosphorus-31 NMR chemical shifts and chemical-shift anisotropies of orthophosphates. J Magn Reson 1986;70:408e15. [37] Mudrakovskii IL, Shmachkova VP, Kotsarenko NS, Mastikhin VM. 31P NMR study of IeIV group polycrystalline phosphates. J Phys Chem Solids 1986;47:335e9. [38] Macmillan PW. Glasseceramics. New York: Academic Press; 1977. p. 1e5. [39] Shelby JE. Introduction to glass science and technology. Royal Soc Chem; 1997:73e4. [40] Grussaute H, Montagne L, Palavit G, Bernard JL. Phosphate speciation in Na2OeCaOeP2O5eSiO2 and Na2OeTiO2eP2O5eSiO2 glasses. J Non-Cryst Solids 2000; 263e264:312e7.
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/he
A new formulation of bariumestrontium silicate glasses and glass-ceramics for high-temperature sealant K. Sharma a,b, G.P. Kothiyal a, L. Montagne b,*, F.O. Me´ar b, B. Revel b a
Glass and Advanced Ceramics Division, Bhabha Atomic Research Centre, Mumbai 400085, India Universite´ Lille Nord de France, UCCS e Unite´ de Catalyse et Chimie du Solide e UMR CNRS 8181, Ecole Nationale Supe´rieure de Chimie de Lille, Universite´ des Sciences et Technologies de Lille, BP 108, 59562 Villeneuve d’Ascq Cedex, France b
article info
abstract
Article history:
P2O5 is a common nucleating agent in glass-ceramics. To avoid the negative effect of P2O5
Received 28 January 2012
on sealing glass properties, we combined it with Barium oxide in the glass formulation. The
Received in revised form
effect of Ba3(PO4)2 incorporation on thermo-physical, structural and sealing characteristics
11 April 2012
of a glass of composition (mol%) 30SiO2e20SrOe30BaOe10B2O3e5La2O3e5Al2O3 is reported.
Accepted 29 April 2012
The glass transition temperature decreases from 650 C to 584 C when Ba3(PO4)2 is
Available online 19 June 2012
increased from 0 to 2 mol%. Thermal expansion coefficient of the glass samples are found around 11.4 to 12.0 106 C1, which is close to that of Crofer-22 APU.
Keywords:
29
Si MAS-NMR
measurements indicate that glass network is composed of mainly Q2 units, while
31
P
0
SOFC sealant
forms Q units. XRD shows the formation of Ba2SiO4, Sr2SiO4 and BaAl2Si2O8 crystalline
Thermal properties
phases. There is no further evolution of crystalline phases when heat treated at 800 C for
XRD
500 h. Seals with Crofer-22 plates have been made and interface reveals good bonding. The
MAS-NMR
glass composition formulated by addition of 1 mol% Ba3(PO4)2 has shown good thermal and sealing characteristics, while further increase of Ba3(PO4)2 to 2 mol% leads to increased crystallization tendency. Copyright ª 2012, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.
1.
Introduction
Solid oxide fuel cells (SOFCs) convert chemical energy into electrical energy, which presents higher efficiency and lower environmental impact as compared to conventional power generation systems [1,2]. The SOFC cells operate at 600e900 C for thousands of hours, and, for efficient and effective operation, it requires a hermetic seal for channeling of fuel and oxygen (or air). In addition, sealing material has to meet several rigorous requirements, including chemical and electrical compatibility with the cell components [3]. Glass and glass-ceramics are explored as promising materials for sealing application in SOFC [4]. A major challenge for this kind of sealants is to avoid the formation of
undesirable crystalline phases during working of the cell, for instance Celsian (BaAl2Si2O8) and its polymorph hexacelsian crystalline phases having low thermal expansion coefficients (TEC), which induces stress in the seal and lead to failure of seals [5]. Further, during long-term operation, interfacial reactions of BaO with Crofer-22 APU material lead to formation of BaCrO4 phase which is detrimental for long-term stability of seal [6,7]. In order to develop suitable sealant materials for high temperature applications, several glass and glass-ceramics mainly based on silicates and borates have been studied [8e10]. BaOeCaOeAl2O3eB2O3eSiO2 (BCABS) glass system has been studied for joining dissimilar metal and ceramic SOFC materials [11]. Haanappel et al. [12] studied the behavior of
* Corresponding author. E-mail address: [email protected] (L. Montagne). 0360-3199/$ e see front matter Copyright ª 2012, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.ijhydene.2012.04.142
11361
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 1 1 3 6 0 e1 1 3 6 9
BCABS seals in presence of oxidizing and reducing atmospheres and reported excess corrosion of interconnect. Lara et al. [13,14] studied the ROeAl2O3eBaOeSiO2 (R ¼ Ca, Mg, Zn) glass system and reported their glass forming ability, sinterability, thermal and electrical properties. As mentioned above, significant amount of BaO in glasses leads to degradation of seals due to formation of BaCrO4. Thus, in order to avoid this, SrO based glasses have been studied. Tiwari et al. [15] reported bonding of SrO and ZnO based silicate glasses with Crofer-22 APU. Ojha et al. [16] studied the SrO and La2O3 based Al2O3eB2O3eSiO2 glasses and reported that SrO modified the network and resulted in change in thermo-physical properties. Thermal expansions of coefficients of glasses are repor ted around 9 106 C1. Wang et al. [17] prepared La2O3eAl2O3eB2O3eSiO2 glasses free of alkaline earth metals and reported a decrease in softening temperature of glasses with an addition of La2O3. However, there is still a need to reduce the softening temperature (Ts), so that seals can be prepared at temperatures lower than 1000 C. It is also desired to match the TEC (12 106 C1) with the metallic interconnects of SOFCs. In the present study, SrOeBaO based aluminosilicate glasses with various compositions are prepared by melt quench technique. The system is of particular interest as replacing BaO with SrO mitigates the formation of BaCrO4. We also expect that with this we may induce the formation of SrAl2Si2O8, which has higher TEC than BaAl2Si2O8. Further, originality of our approach lies in the combined addition of BaO and P2O5 to tailor the sealing and thermo physical properties of the sealants. Larsen et al. reported that phosphate based glass exhibit insufficient durability for sealing but addition of silica can increase the stability of phosphate in the glass [18]. Moreover, P2O5 can be useful as nucleating agent to avoid undesirable reaction at the metal glass interface [19]. Finally, addition of P2O5 in alumino-silicate glasses scavenges the modifier cations and increase the network polymerization, hence viscosity and sealing temperature increase, which is detrimental for the sealing application [20]. Thus a series of glasses have indeed been formulated by addition of Ba3(PO4)2 to a base glass composition. According to previous studies [21], it is expected that with this combined addition, the network polymerization will not get affected but we will get a decrease in glass transition and softening temperatures. Thermo-physical properties of the glasses (Tg, Ts and TEC) with increasing quantity of Ba3(PO4)2 have been studied. The glasses are characterized for structural features with XRD and solid-state NMR. Seals of the glasses with Crofer-22 APU metallic alloy were made, and the microstructure at the interface and chemical stability of seals at high temperature were investigated.
2.
Experimental
Nominal compositions (mol%) of glass samples are given in Table 1. These were prepared by using melt-quench technique. Al(OH)3, BaCO3, H3BO3, La2O3, NH4H2PO4, SiO2 and SrCO3 of 99.9% purity were mixed, ground and calcined at a maximum temperature of 900 C in a Pt-10% Rh crucible. It is to be noticed that Ba3(PO4)2 is obtained in the glass
Table 1 e Nominal BASP glass composition (mol%). The glasses were formulated by addition of Ba3(PO4)2 to BASP0. Sample
SiO2
BaO
SrO
B2O3
La2O3
Al2O3
P2O5
BASP0 BASP1 BASP2
30 28.8 27.7
30 31.7 33.3
20 19.2 18.5
10 9.6 9.2
5 4.8 4.6
5 4.8 4.6
0 0.9 1.8
composition from the decomposition of BaCO3 and NH4H2PO4. The calcined charge was then melted at 1450e1550 C and held at this temperature for 1 h for homogenization. The melt was then poured on a stainless steel plate. The glass samples were annealed at around 650 C for 4 h. Weight losses were carefully checked during glass preparation, hence actual composition is considered to be similar to nominal composition. Differential thermal analysis (DTA) was carried out from 20 C to 950 C using a TG/DTA instrument (Model Setsys, Setaram, France). DTA measurements were carried out on powdered samples with particle size less than 63 mm. Thermal expansion was measured from 20 C to 900 C using TMA instrument (Model Labsys, Setaram, France). Thermal expansion coefficient was then measured in the linear part of the dilatation curve. The crystalline phases in glass-ceramics samples were identified using powder X-ray diffractometer (Bruker) with CuKa as X-ray source. XRD data were collected on powder samples with particle size less than 63 mm. These measurements were done in the range (2q angle) 10e80 with a step of 0.02 s1 and dwell time of 0.2 s. The sintering behavior of the BASP glasses was studied by means of a hot stage microscope (HSM, Hesse, Germany). For these measurements, the sample powders were pelletized and placed on an Alumina support. This was then placed within a furnace and heated at 10 C$min1. The shape of the sample was monitored using a camera. The projected area of the sample viewed in side elevation was then used to characterize the sintering, deformation, sphere and flow temperatures of the glasses. 11 B, 27Al, 29Si and 31P MAS-NMR spectra were recorded at 18.8, 18.8, 9.4 and 9.4T respectively, on Bruker AVANCE spectrometers, with 4 mm probes at 12.5 kHz spinning speed, except for 29Si for which a 7 mm probe at 5 kHz was used. The Larmor frequencies were 128.4, 104.2, 79.4 and 162.3 MHz for 11 B, 27Al, 29Si and 31P, respectively. For 11B, the pulse duration was 2 ms (p/6), and the recycle delay was 10 s. For 27Al, the pulse duration was 1.5 ms (p/8), and the recycle delay was 2 s. For 29Si, the pulse duration was 1.6 ms (p/5), and the recycle delay was 180 s. For 31P, the pulse duration was 1.6 ms (p/6), and the recycle delay was 120 s. All relaxation delays were chosen long enough to enable relaxation at the field strength that was used. The 11B chemical shifts are given relative to BPO4 at 3.6 ppm, 27Alchemical shifts are relative to AlCl3 at 0 ppm, 29Si chemical shifts are relative to tetramethylsilane (TMS) at 0 ppm, and 31P are relative to 85% H3PO4 at 0 ppm. For studying the adhesion/bonding, sealing with Crofer-22 alloy was carried out at a maximum temperature of 1000 C employing the sandwich geometry. Microstructure near the interface was investigated using Cameca SX 100 electron
11362
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 1 1 3 6 0 e1 1 3 6 9
Table 2 e TMA data for BASP glass samples. TEC (0.2 106 K1)
Sample As prepared
After 6 h at 800 C
After 100 h at 800 C
11.4 12 13
12.2 12 13
12.4 12.5 12.1
BASP0 BASP1 BASP2
Tg (2 C)
Flow temperature (3 C)
650 592 584
1145 1107 1108
probe microanalyser (EPMA) and wavelength dispersive analysis of X-Rays (WDAX) was carried out for determining element diffusion across the interface.
3.
Results
The compositions reported in Table 1 yielded bubble-free glasses, named BASP in the following text. XRD analysis of these glasses has shown absence of any long range order.
3.1. Thermo mechanical analysis (TMA)/differential thermal analysis (DTA) The thermo-physical properties of the BASP glasses are listed in Table 2. The glass transition temperature has decreased from 650 C to 584 C when P2O5 was increased from 0 to 1.8 mol% and BaO from 30 to 33.3 mol%. The TEC values of the glass samples are found to increase from 11.4 to 13 106 C1. DTA plots of glass samples are shown in Fig. 1. The sample BASP0 shows an onset for crystallization (Tcry) at 756 C. Addition of Ba3(PO4)2 does not modify significantly this value, although onset of crystallization is detected at slightly higher temperature (764 C) for BASP1.
3.2.
Hot stage microscopy (HSM)
The normalized areas of the BASP samples as recorded with HSM are plotted as a function of temperature in Fig. 2. The incorporation of Ba3(PO4)2 does not modify the sintering temperature, which remains at 710 C. The
Fig. 1 e DTA plots of BASP glasses.
Fig. 2 e Normalized area versus temperature of BASP glasses obtained from Hot Stage Microscopy (HSM).
maximum shrinkage temperature (Tshri) and flow temperature decrease with addition Ba3(PO4)2. The shape of the normalized area versus temperature curves indicates that the BASP0 and BASP1 glasses soften over a wide temperature range indicating glassy behavior. BASP2 glass exhibit a wide plateau before flow, which is a characteristic of materials exhibiting devitrification upon heating [22]. This indicates the occurrence of significant crystallization in BASP2 glass.
3.3.
X-ray diffraction (XRD)
XRD patterns of the glass samples after heat treatment at 800 C for 6 h are depicted in Fig. 3a. The glass BASP0 shows the formation of Ba2SiO4, Sr2SiO4 and BaAl2Si2O8 as major phases. However, with addition of 1 mol% Ba3(PO4)2, relative intensity of peak corresponding to BaAl2Si2O8 (around 23 ) has decreased. It implies tendency of formation of such a phase reduces when Ba3(PO4)2was added to base composition (BASP0). Overall, the peaks have become sharp, which implies better crystallized samples. Evolution of crystalline phases in BASP0 glass after different heat treatment time is shown in Fig. 3b. XRD peaks have become sharp when sample were heat treated for 500 h. Intensities of peaks corresponding to formation of BaAl2Si2O8 have decreased (for instance at 22.5 ) and while peak at 27 for Sr2SiO4 has increased. It implies that the tendency to form Sr2SiO4 is more prevalent than formation BaAl2Si2O8 when glass is heat treated at 800 C for long time (500 h). XRD plots
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 1 1 3 6 0 e1 1 3 6 9
11363
Fig. 3 e XRD plots of: (a) BASP glasses after heat treatment at 800 C for 6 h; (b) BASP0 glass after heat treatment at 800 C from 6 h to 500 h; (c) BASP glasses after heat treatment at 800 C for 100 & 500 h.
11364
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 1 1 3 6 0 e1 1 3 6 9
for the other glass samples (BASP1&2) after heat treatment at 800 C for 100 to 500 h are shown in Fig. 3c. There was no evolution of additional phase when heat treatment time was increased from 100 h to 500 h.
3.4.
NMR
3.4.1.
29
Si MAS-NMR
Fig. 4 shows the 29Si MAS-NMR spectra of the as-prepared BASP glasses and after 100 h heat treatment at 800 C. The spectra of the as prepared glasses show a broad feature in the w70ew90 ppm range with a Gaussian line shape, characteristic of bond angle and length distributions inherent to the amorphous state. Its chemical shift is centered on w80 ppm, which indicates that Q2 SiO4 units dominate in the glass network [23]. This result is in accordance with the glass composition, which contains a large quantity of network modifying oxides. Some Q3 SiO4 units bonded to Al or B may be present, since chemical shifts of such units shift toward less negative values [24,25]. After 100 h of heat treatment at 800 C, large modifications are observed on the 29Si MAS-NMR spectra (Fig. 4). Two narrow resonances with large intensity are now present at w68 and w89 ppm, which indicate the presence of crystalline phases. According to literature data, they can be assigned to Ba2SiO4, with chemical shift reported at w70 ppm, and to BaAl2Si2O8, with chemical shift reported at 88 ppm [26,27]. The resonance at w68 ppm may also be assigned to Sr2SiO4 or a mixed (Ba,Sr)SiO4, owing to the close values of Ba2þ and Sr2þ field strengths [28]. The weak resonance at w85 ppm may be assigned to (Ba,Sr)Al2Si2O8 mixed Celsian phase.
3.4.2.
11
B MAS-NMR
11
The B NMR spectra of the parent and heat treated BASP glasses are shown in Fig. 5. The spectra are similar for all the compositions. The resonance observed at 16 ppm can be assigned to BO3 and a very weak one at 1 ppm assigned to BO4 units [29]. Despite the use of a high-field spectrometer that enables a high resolution for quadrupolar nuclei like 11B
Fig. 4 e 29Si MAS-NMR spectra of as prepared BASP glasses and after heat treatment at 800 C for 100 h.
Fig. 5 e 11B MAS-NMR spectra of as prepared BASP glasses and after heat treatment at 800 C for 100 h.
(nuclear spin I ¼ 3/2), the spectra are broad and featureless, meaning that all 11B nuclei are present in a very similar environment. After heat treatment for 100 h (Fig. 4), the 11B NMR spectra of glass-ceramics remain quite similar to those of the parent glasses. Nevertheless, the main resonance shows on its top the presence of two singularities around w18 and w17 ppm. They suggest that some boron nuclei are involved in regular crystalline environments [30]. Celsian is indeed known to incorporate many different elements in its structure to form substitution solid solution [27,31]. Concerning boron, we have already reported that NMR gave evidence for the formation of a mixed Al, B phase [32].
3.4.3.
27
Al MAS-NMR
27
The Al NMR spectra for BASP glasses and glass-ceramics are reported in Fig. 6. The spectra of glasses consist of
Fig. 6 e 27Al MAS-NMR spectra of as prepared BASP glasses and after heat treatment at 800 C for 100 h.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 1 1 3 6 0 e1 1 3 6 9
11365
expected. A much narrower resonance is observed for the glass-ceramics, thus revealing that Al is fully involved in the crystalline phases. The spectra consist of a resonance centered at w55 ppm, which is characteristic of AlO4 tetrahedral units in Celsian (Ba,Sr)Al2Si2O8 crystals [26,34]. The broad line shape at the bottom of 27Al resonances of glass-ceramics indicate that some Al remains in the residual glass.
3.4.4.
31
P MAS-NMR
31
Fig. 7 e 31P MAS-NMR spectra of as prepared BASP glasses and after heat treatment at 800 C for 100 h.
a broad resonance centered at w65 ppm, which is characteristic of aluminum in tetrahedral coordination [33]. Owing to the large quantity of charge compensating cations available in the glasses, this tetrahedral coordination was
The P NMR spectra of the glasses and glass-ceramics are shown in Fig. 7. A main broad resonance is observed at w8 ppm, which is assigned to Q0 (orthophosphate units) [35]. This chemical shift is close to that of Q0 units in Ba3(PO4)2 crystal (9 ppm) [36], but different from that of Sr3(PO4)2 crystal (3 ppm) [37]. Hence we expect that the orthophosphate anions are mainly charge balanced by Ba2þ but some Sr2þ cannot be excluded. The resonance becomes somewhat sharper in glass-ceramic samples. Nevertheless, the Ba and Sr atomic distribution in Q0 orthophosphate site may be at the origin of broad line shape, thus we can expect the presence of phosphate crystal in the glass-ceramics samples. Since the P2O5 quantity is small in these glasses, the presence of these phosphate crystals could not be confirmed by XRD analyses.
Fig. 8 e Microstructure at the interface of glass with Crofer-22 APU: (a) BASP0; (b) BASP1; (c) BASP2; (d) EPMA line scan of BASP1.
11366
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 1 1 3 6 0 e1 1 3 6 9
Fig. 9 e Microstructure near the interface between glass and Crofer-22 APU after heat treatment at 800 C: (a) BASP0 (for 100 h); (b) BASP0 (for 500 h); (c) BASP1 (for 100 h); (d) BASP1 (for 500 h); (e)BASP2 (for 100 h); (f) BASP2 (for 500 h).
3.5.
Microstructure and chemical stability
Fig. 8(aec) shows the microstructures at interfaces of BASP glasses with Crofer-22 APU. All the sealing glasses bonded well to the metallic part as no bubbles and cracks were observed at the interface. A limited interdiffusion of Cr, Sr and Ba across the boundary is observed, which is required for good adhesion between metal and glass (Fig. 8d). Fig. 9(aef) shows the EPMA images recorded at the interface of metal/glass joints after heat treatment at 800 C for 100 h and 500 h. The microstructure of BASP0 is composed of coarse crystals as seen from EPMA images, while BASP1and BASP2 show more homogeneous microstructure with smaller crystals. There were no major changes observed in microstructure after heat treatment at 800 C for 500 h. As seen from the images, all the seals were found with good interfaces. EPMA images of seals confirm limited enrichment of Cr at the interface (Fig. 10(aef)) even after heat treatment at 800 C for 500 h. Fig. 11 shows the comparison of Cr diffusion at interface
of BASP1 glass with Crofer-22 APU alloy after different time of heat treatment at 800 C. In BASP1 glass, Cr diffusion of around 3 mm is observed for as prepared seal and it increases marginally to 5 mm after 500 h of heat treatment at 800 C. Mapping of the other elements at the interface of all glassceramics compositions did not reveal any significant differences; hence the images are not reported.
4.
Discussion
Thermo-physical properties of glasses and glass-ceramics are strongly dependent upon the composition [38]. According to continuous random network theory, SiO2 is a network forming oxide and BaO is a network modifying oxide [39]. The addition of barium phosphate leads to decrease in overall silica content, which is expected to decrease the rigidity of the network. Thus the TEC values of BASP glasses increase with addition of barium phosphate, concomitantly Tg values decrease.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 1 1 3 6 0 e1 1 3 6 9
11367
Fig. 10 e Cr mapping at the BASP glass Crofer-22 APU interface after heat treatment at 800 C: (a) BASP0 (for 100 h); (b) BASP0 (for 500 h); (c)BASP1 (for 100 h); (d) BASP1 (for 500 h); (e) BASP2 (for 100 h); (f) BASP2 (for 500 h).
It is known that modifier cations are required for charge compensation for PO3 4 anions, when P2O5 is added to glass. But no evolution in 29Si spectra of the as-prepared glasses is observed for the 3 different compositions (Fig. 4). This was expected, since we formulated the glasses by the simultaneous addition of BaO and P2O5 in proportion 3/1, to obtain an equivalent amount of Ba3(PO4)2. It has indeed been shown that, when Ba3(PO4)2 is added to silicate glasses in small amount, P2O5 does not participate to the silicate glass network and remains as isolated orthophosphate species [40]. As a result, the Al coordination remains primarily tetrahedral and B forms mainlyBO3units. The formation of less rigid planar BO3, units as observed in 11B NMR contributes to keep moderate values of flow temperature and Tg for BASP glasses.
Fig. 11 e Cr diffusion at BASP1 glass e Crofer-22 APU interface after different time of heat treatment at 800 C.
For effective sealing, glass should attain maximum shrinkage before crystallization [13]. All the BASP glasses have shown attainment of maximum shrinkage before crystallization. However, crystallization becomes more important and occurs earlier when Barium phosphate was increased to 2 mol % (Fig. 2). MAS-NMR spectra of the glass-ceramics are in accordance with the phase emergence observed by XRD. Since the incorporation of Ba3(PO4)2 does not affect the network the major phases formed are not much changed. However, there is an increased crystallization and phase stabilization. We observed the formation of Ba2SiO4, Sr2SiO4 and BaAl2Si2O8 phases. As crystallization of B is evident on B MAS-NMR spectra, we suggest the formation of (Ba,Sr)(Al,B)2Si2O8 like phase. Identification of such a phase is not possible due to inherent limitation of XRD. In glass-ceramics, thermo-physical properties depend on the crystalline phase evolved during crystallization. BASP glass-ceramics have high TEC (12 106 C1) despite the crystallization of hexacelsian like phase, which has lower than desired TEC (8 106 C1) [3]. This also suggests the formation of (Ba,Sr)Al2Si2O8; since SrAl2Si2O8 has a higher TEC (11 106 C1). B incorporation in such a phase may further lead to increase in TEC. The decrease in flow and sealing temperatures with addition of Ba3(PO4)2 in BASP glasses helps in forming good bonds with Crofer-22 APU at temperature lower than 1000 C, however we have carried out all the sealing experiment at 1000 C for comparing the results for BASP glasses. All the seals with Crofer-22 APU have shown good adhesion. The seals were found intact after heat treatment at 800 C up to 500 h. It has been reported that the BaO containing glasses interact chemically with chromia-forming alloys forming BaCrO4, which leads to the separation of GCs from the alloy matrix due to thermal expansion mismatch [6]. In the present investigation, we did not observe major enhancement in
11368
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 1 1 3 6 0 e1 1 3 6 9
Cr-diffusion after 500 h at 800 C and no adverse reactions at interface were observed even after 500 h of heat treatment. Therefore, we conclude that incorporation of Ba3(PO4)2 improves the sealing characteristics of BASP glasses/glassceramics. Longer test duration is under investigation to evaluate the effect of Barium phosphate addition on the very long term.
5.
Conclusions
A new way of formulating sealing glasses by the combined addition of P2O5 and BaO as Ba3(PO4)2 has been presented in the present paper. These barium alumino-silicate glasses are studied for structural, thermo-physical properties and sealing properties. The glasses do not show major structural changes with addition of BaO and P2O5 in the form of Ba3(PO4)2. Moreover, this leads to decrease in softening temperatures, thus enabling seal fabrication (with Crofer-22 APU) below 1000 C. Interfaces of seals reveal good bonding with Crofer-22 APU, which did not deteriorate even after heat treatment at 800 C for 500 h. The glasses show the development of Ba2SiO4, BaAl2Si2O8, and Sr2SiO4 crystalline phases upon heat treatment at 800 C, which are not found to be detrimental for high temperature sealing applications.
Acknowledgments The authors thank the IFCPAR for funding this work vide project number 4008-1. The FEDER, Re´gion Nord Pas-de-Calais, Ministe`re de l’Education Nationale de l’Enseignement Supe´rieur et de la Recherche, CNRS, and USTL are acknowledged for funding of NMR spectrometers.
references
[1] Steele BCH, Heinzel A. Materials for fuel-cell technologies. Nature 2001;414:345e52. [2] Pascual MJ, Guillet A, Duran A. Optimization of glasseceramic sealant compositions in the system MgOeBaOeSiO2 for solid oxide fuel cells (SOFC). J Power Sources 2007;169:40e6. [3] Fergus JW. Sealants for solid oxide fuel cells. J Power Sources 2005;147:46e57. [4] Ley KL, Krumpelt M, Kumar R, Meiser JH, Bloom I. Glasseceramic sealants for solid oxide fuel cells: part I. Physical properties. J Mater Res 1996;11:1489e93. [5] Brochu M, Gauntt BD, Shah R, Miyake G, Loehman RE. Comparison between barium and strontium glass composites for sealing SOFCs. J Eur Ceram Soc 2006;26: 3307e13. [6] Batfalsky P, Haanappel AAC, Malzbender J, Menzler NH, Shemel V, Vinke IC. Chemical interaction between glassceramic sealant and interconnect steels in SOFC stacks. J Power Sources 2006;155:128e37. [7] Yang Z, Xia G, Meinhardt KD, Weil KS, Stevenson JW. Chemical stability of glass seal interfaces in intermediate temperature solid oxide fuel cells. J Mater Eng Perform 2004; 13:327e34. [8] Mahapatra MK, Lu K. Glass-based seals for solid oxide fuel and electrolyzer cells e a review. Mater Sci Eng R 2010;67: 65e85.
[9] Yang Z, Meinhardt KD, Stevenson JW. Chemical compatibility of barium-calcium-aluminosilicate-based sealing glasses with the ferritic stainless steel interconnect in SOFCs. J Electrochem Soc 2003;150:A1095e101. [10] Ghosh S, Sharma AD, Kundu P, Mahanty S, Basu RN. Development and characterizations of BaOeCaOeAl2O3eSiO2 glasseceramic sealants for intermediate temperature solid oxide fuel cell application. J Non-Cryst Solids 2008;354: 4081e8. [11] Sohn SB, Choi SY, Kim GH, Song HS, Kim GD. Stable sealing glass for planar solid oxide fuel cell. J Non-Cryst Solids 2002; 297:103e12. [12] Haanappel VAC, Shemet V, VinkeI C, Quadakkers WJ. A novel method to evaluate the suitability of glass sealantealloy combinations under SOFC stack conditions. J Power Sources 2005;141:102e7. [13] Lara C, Pascual MJ, Duran A. Glass-forming ability, sinterability and thermal properties in the systems ROeBaOeSiO2 (R ¼ Mg, Zn). J Non-Cryst Solids 2004;348: 149e55. [14] Lara C, Pascual MJ, Prado MO, Duran A. Sintering of glasses in the system ROeAl2O3eBaOeSiO2 (R ¼ Ca, Mg, Zn) studied by hot-stage microscopy. Solid State Ionics 2004; 170:201e8. [15] Tiwari B, Dixit A, Kothiyal GP. Study of glasses/glassceramics in the SrOeZnOeSiO2 system as high temperature sealant for SOFC applications. Int J Hydrogen Energy 2011;36: 15002e8. [16] Ojha PK, Rath SK, Chongdar TK, Gokhale NM, Kulkarni AR. Physical and thermal behaviour of SreLaeAleBeSi based SOFC glass sealants as function of SrO content and B2O3/SiO2 ratio in the matrix. J Power Sources 2011;196:4594e8. [17] Wang SF, Lu CM, Wu YC, Yang YC, Chiu TW. La2O3eAl2O3eB2O3eSiO2 glasses for solid oxide fuel cell applications. Int J Hydrogen Energy 2011;36:3666e72. [18] Larsen PH, Poulsen FW, Berg RW. The influence of SiO2 addition to 2MgOeAl2O3e3.3P2O5 glass. J Non-Cryst Solids 1999;244:16e24. [19] Donald IW, Metcalfe BL, Gerrard LA. Interfacial reactions in glass-ceramic-to-metal seals. J Am Ceram Soc 2008;91: 715e20. [20] Ananthanarayanan A, Kothiyal GP, Montagne L, Tricot G, Revel B. The effect of P2O5 on the structure, sintering and sealing properties of barium calcium aluminum boro-silicate (BCABS) glasses. Mater Chem Phy 2011;130:880e9. [21] Grussaute H, Montagne L, Palavit G. Structural interpretation of the modification of soda-lime-silica glass properties by P2O5. Glass Sci Technol 2000;73:380e4. [22] Boccaccini AR, Hamann B. Review in situ high-temperature optical microscopy. J Mater Sci 1999;34:5419e36. [23] Goswami M, Kothiyal GP, Montagne L, Delevoye L. MAS-NMR study of lithium zinc silicate glasses and glasseceramics with various ZnO content. J Solid State Chem 2008;181:269e75. [24] Goel A, Tulyaganov DU, Pascual MJ, Shaaban ER, Munoz F, Lu Z, et al. Development and performance of diopside based glasseceramic sealants for solid oxide fuel cells. J Non-Cryst Solids 2010;356:1070e80. [25] Martens R, Muller-Warmuth W. Structural groups and their mixing in borosilicate glasses of various compositionsean NMR study. J Non-Cryst Solids 2000;265:167e75. [26] Mackenzie KJD, Smith ME. Multinuclear solid-state NMR of Inorganic materials. p. 230e242. London: Pergamon; 2001. [27] MacKenzie KJD, Kemmitt T. Evolution of crystalline aluminates from hybrid gel-derived precursors studied by XRD and multinuclear solid-state MAS NMR: I. Celsian, BaAl2Si2O8. Thermochimica Acta 1999;325:5e12. [28] Murdoch JB, Stebbins JF, Carmichael ISE. High-resolution 29 Si NMR study of silicate and aluminosilicate glasses: the
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 7 ( 2 0 1 2 ) 1 1 3 6 0 e1 1 3 6 9
[29]
[30]
[31]
[32]
[33]
effect of network-modifying cations. Am Mineral 1985;70: 332e43. Ananthanarayanan A, Kothiyal GP, Montagne L, Revel B. MAS-NMR studies of lithium aluminum silicate (LAS) glasses and glasseceramics having different Li2O/Al2O3 ratio. J Solid State Chem 2010;183:120e7. Stamboulis A, Hill RG, Law RV. Characterization of the structure of calcium alumino-silicate and calcium fluoro-aluminosilicate glasses by magic angle spinning nuclear magnetic resonance (MAS-NMR). J Non-Cryst Solids 2004;333:101e7. Lee KT, Aswath BP. Role of mineralizers on the hexacelsian to celsian transformation in the barium aluminosilicate (BAS) system. Mater Sci Eng A 2003;352:1e7. Ananthanarayanan A, Kothiyal GP, Montagne L, Revel B. MAS-NMR investigations of the crystallization behaviour of lithium aluminum silicate (LAS) glasses containing P2O5 and TiO2 nucleants. J Solid State Chem 2010;183: 1416e22. Bunker BC, Kirkpatrick RJ, Brow RK, Turner GL, Nelson C. Local structure of alkaline-earth boroaluminate crystals and glasses: II, 11B and 27Al MAS NMR spectroscopy of alkalineearth boroaluminate glasses. J Am Ceram Soc 2005;74:1430e8.
11369
[34] Stebbins JF, Kroecker S, Lee SK, Kiczenski TJ. Quantification of five and six coordinated aluminum ions in aluminosilicate and fluoride-containing glasses by high-field, highresolution 27Al NMR. J Non-Cryst Solids 2000;275:2425e32. [35] Grimer AR, Haubenreisser U. High field static and MAS 31P NMR: chemical shift tensor of polycrystalline potassium phosphate P2O5 K2O. Chem Phys Lett 1983;99:487e90. [36] Turner GL, Smith KA, Kircpatrick RJ, Oldfield E. Structure and cation effects on phosphorus-31 NMR chemical shifts and chemical-shift anisotropies of orthophosphates. J Magn Reson 1986;70:408e15. [37] Mudrakovskii IL, Shmachkova VP, Kotsarenko NS, Mastikhin VM. 31P NMR study of IeIV group polycrystalline phosphates. J Phys Chem Solids 1986;47:335e9. [38] Macmillan PW. Glasseceramics. New York: Academic Press; 1977. p. 1e5. [39] Shelby JE. Introduction to glass science and technology. Royal Soc Chem; 1997:73e4. [40] Grussaute H, Montagne L, Palavit G, Bernard JL. Phosphate speciation in Na2OeCaOeP2O5eSiO2 and Na2OeTiO2eP2O5eSiO2 glasses. J Non-Cryst Solids 2000; 263e264:312e7.