Contents lists available at ScienceDirect

Journal of Non-Crystalline Solids

journal homepage: www.elsevier.com/locate/jnoncrysol

Sintering and crystallization of SrO-CaO-B2O3-SiO2 glass-ceramics withdifferent TiO2 contents

Laís D. Silvaa,b,⁎, Alisson M. Rodriguesb, Ana Candida M. Rodriguesb, María J. Pascualc,Alicia Duránc, Aluísio A. Cabrala,d

a Postgraduate Program in Materials Engineering, Federal Institute of Maranhão – IFMA, 65030-001 São Luis, MA, Brazilb Center of Research, Technology, and Education in Vitreous Materials (CeRTEV), Department of Materials Engineering (DEMa), Postgraduate Program in MaterialsScience and Engineering (PPGCEM), Federal University of São Carlos (UFSCar), 13565-905 São Carlos, SP, Brazilc Instituto de Cerámica y Vidrio (CSIC), C/Kelsen 5, Campus de Cantoblanco, 28049 Madrid, Spaind Department of Physics – DEFIS, Federal Institute of Maranhão – IFMA, 65030-001 São Luís, MA, Brazil

A R T I C L E I N F O

Keywords:Glass-ceramicSealantSolid oxide fuel cellSinteringCrystallization kineticsThermal expansion coefficient

A B S T R A C T

Glass and glass-ceramics (GCs) of the CaO-SrO-B2O3-TiO2-SiO2 system are promising candidates as sealants forsolid oxide fuel cells (SOFCs), mainly because of the possibility of the formation of Sr2SiO4 and Sr(TiO3) ascrystalline phases, which minimizes the interfacial reaction between the interconnection elements and thesealant. Four glass compositions of this system, with different TiO2 content, were obtained. The crystallinephases formed after crystallization were identified by X-ray diffraction (XRD). Combining Differential ScanningCalorimetry (DSC) and Optical Microscopy (OM), it was found that the overall activation energy for crystal-lization did not change with TiO2 content. The predominant crystallization mechanism is controlled by thesurface, and the growth of the crystalline layer is governed by diffusion. The sintering behavior was investigatedby DSC and Hot Stage Microscopy (HSM). All glass compositions exhibited a single-stage shrinkage behavior, butthose with a larger amount of TiO2 showed high sinterability. The measured coefficient of thermal expansion(CTE) varied from 10.2 × 10−6 K−1 (100–550 °C) in the glass to 13.9 × 10−6 K−1 (100–550 °C) in the GCs,which is in good agreement with the typical CTE of SOFC components. These consistent values of the CTE andgood sintering behavior allow us to propose that the glass composition with a higher percentage of titanium is apromising candidate for further experimentation and development as a sealant material.

1. Introduction

Solid oxide fuel cells (SOFCs) are electrochemical devices thatconvert chemical energy into electrical energy with high efficiency athigh temperatures. Each cell consists of an electrolyte integrated with aporous anode and cathode [1,2], whereas an SOFC assemblage com-prises individual cells connected with the interconnect components,which transport the electronic current between the unit cells and theexternal circuit. The interconnection components are exposed to redu-cing or oxidizing atmospheres because the fuel (hydrogen or hydro-carbons) is oxidized at the anode, while the oxidant (air or oxygen) isreduced at the cathode. Among the various existing SOFC designs, in-termediate temperature solid oxide fuel cells (IT-SOFCs) operate attemperatures between 600 and 1000 °C and allow the use of ferriticstainless steels as interconnect components [3–5]. Nevertheless, thedevelopment of a hermetic seal is important because it can prevent

mixing of fuel and oxidant into the fuel cell, and it keeps the cellcomponents together [6–10].

Therefore, the development of a sealant for SOFC applications mustmeet several crucial requirements, e.g., (i) the sealants must have acoefficient of thermal expansion (CTE) similar to that of the cell's othercomponents (i.e., cathode, anode and interconnection components) ingeneral in the range of (9–13) × 10−6 K−1[11,12]; (ii) good me-chanical properties [13]; (iii) the material must be an electrical in-sulator (electrical resistance> 2 kΩ cm), to avoid parallel currents[14]; and (iv) be chemically compatible with other fuel cell components[6], while minimizing thermal stresses during high-temperature op-erations [9,15]. These requirements constitute a major challenge.

Several studies have shown that glass and glass-ceramic (GC) ma-terials are promising candidates for sealant applications, mainly due tothe possibility of tailoring the chemical composition [6,9,16]. In thecase of an IT-SOFC, the interfacial reaction between glass-ceramic

http://dx.doi.org/10.1016/j.jnoncrysol.2017.07.021Received 5 May 2017; Received in revised form 14 July 2017; Accepted 25 July 2017

⁎ Corresponding author at: Center of Research, Technology, and Education in Vitreous Materials (CeRTEV), Department of Materials Engineering (DEMa), Postgraduate Program inMaterials Science and Engineering (PPGCEM), Federal University of São Carlos (UFSCar), 13565-905 São Carlos, SP, Brazil.

E-mail address: laisds_@ppgcem.ufscar.br (L.D. Silva).

Journal of Non-Crystalline Solids xxx (xxxx) xxx–xxx

0022-3093/ © 2017 Elsevier B.V. All rights reserved.

Please cite this article as: Silva, L.D., Journal of Non-Crystalline Solids (2017), http://dx.doi.org/10.1016/j.jnoncrysol.2017.07.021

sealants and ferritic interconnects results in the formation of chromatephases, e.g., BaCrO4 and SrCrO4, leading to a weak adherence and apossible physical separation between the sealant and the interconnectalloy [4,6,17–21].

Zhang et al. [20] and Chen et al. [21] recently demonstrated thatglasses and glass-ceramics of the CaO·SrO·B2O3·SiO2 (CSrBS) systemdoped with TiO2 (2–8 mol%) are potential candidates for sealant ap-plications due to the precipitation of Sr2SiO4 and Sr(TiO3), whichminimizes the interfacial reaction between the sealant and the com-ponent alloys of the IT-SOFC. Additionally, its CTE values closely matchthose of the interconnect elements (e.g., 9.8 × 10−6 K−1, GC, to11.3 × 10−6 K−1, glass). However, the sinterability and crystallizationkinetics of these glasses were not yet investigated. The knowledge ofglass sinterability and crystallization kinetics are crucial for predictingthe sealing ability of the glass powders to ceramic or metallic surfaces[22], since grained glass samples devitrify more easily than bulk glasssamples and the corresponding glass-ceramic materials can be obtainedthrough a process of simultaneous amorphous state sintering andcrystallization. Moreover, the effect of TiO2 on the crystallization ki-netics was not yet investigated.

Other very important point of view to be analyzed is that the glasssealant must be resistant against the crystallization during the cellsoperation to avoid thermal and mechanical stresses [21]. In fact, theseauthors analyzed the glass stability against crystallization duringheating using the relation Tx − Tg, where Tx and Tg are the onset of thecrystallization peak and the glass transition temperatures, respectively.Nevertheless, Nascimento et al. [23] pointed out that the glass stabilityis more precisely evaluated by the parameters that also take into ac-count the melting or liquidus temperature, such as KH = (Tx − Tg) /(Tm − Tx), KW = (Tx − Tg) / Tm and KLL = Tx / (Tg + Tm), where Tmmeans melting or liquidus temperature [24].

Based on the above, the purposes of this work are: (i) to investigatethe effect of TiO2 in the sinterability of glasses doped with differentTiO2 contents; (ii) to assess the crystalline phases precipitated; (iii) toidentify the predominant crystallization mechanisms of these glasses byDifferential Scanning Calorimetry (DSC); and (iv) to analyze the effectof TiO2 on the glass stability and the crystallization mechanism. Inorder to do this, four glass compositions of the CSrBS system dopedwith different TiO2 contents (1, 2, 3 and 4 mol%) were prepared. Thecrystalline phases of the corresponding GCs were then identified inorder to confirm the presence of Sr2SiO4 and Sr(TiO3) phases. Also, theCTE values of each parent glass and corresponding GCs were de-termined to ascertain their compatibility with the interconnect ele-ments of the IT-SOFCs.

2. Experimental procedures

2.1. Glass synthesis

Four glass compositions of the (26 − 0.25x)CaO (26 − 0.25x)SrO(4 − 0.05x)B2O3 xTiO2 (44 − 0.45x)SiO2 (with x = 1, 2, 3 and 4 mol%) system were obtained and referred as T1, T2, T3 and T4, respectively.The reagents grade alkaline earth carbonates, boric acid, SiO2, and TiO2

were mixed and placed in a platinum crucible for melting at 1500 °C for1 h. Then, each liquid was quenched in air onto a steel plate and re-melted three times to prevent the formation of bubbles and streaks. Inthe last re-melting, the liquids were pressed between two stainless steelplates (splat-cooling) and annealed at 50 °C below the correspondingglass transition temperature (Tg-50 °C) to avoid residual stresses. TheChemical composition was determined using a MagiX Super Q version3.0 Phillips X-ray fluorescence spectrometer (Philips, The Netherlands),equipped with IQ+ analytical software used for semi-quantitativeanalysis. Calibration curves were prepared from standards of certifiedcomposition of natural and synthetic calcium phosphates and calciumsilicates. The sample preparation was performed by melting of 0.3 g ofeach sample and 5.5 g of Li2B4O7 in a Pt/Au crucible and casting into

pearl form in a special controlled furnace Perl'X3 (Philips). The nominaland experimental compositions in mol% are presented in Table 1.

2.2. DSC analysis

DSC experiments were performed to infer the glass-stability and thenon-isothermal crystallization kinetics of each glass. All experimentswere performed in an NETZSCH STA 449C thermal analyzer usingplatinum crucibles under air atmosphere. For this purpose, monolithicglass samples (3.0 × 3.0 × 2.5 mm) of each composition were heatedfrom room temperature until the corresponding liquidus or meltingtemperature using different rates (ϕ = 5, 10, 15 and 20 °C/min). Fromthese DSC data, the onset values of Tg, Tx and Tm were estimated and theglass stability of each glass was calculated using the equations proposedby Hrüby (KH), Weinberg, (KW) and Lu and Liu (KLL): KH = (Tx − Tg) /(Tm − Tx), KW = (Tx − Tg) / Tm and KLL = Tx / (Tg + Tm) [24].

To estimate the sinterability parameter for each vitreous composi-tion, DSC experiments were also performed for powdered glass samples(< 43 μm) at the same heating rate (10 °C/min) and temperature range(room temperature until 1400 °C).

The Avrami coefficient, n, and the activation energies for overallcrystallization, E, were calculated from the Ozawa [25–27] and Matu-sita & Sakka equations [26,28] given by Eqs. (1) and (2), respectively:

− = − −n χϕ

ln( ln(1 ))ln T (1)

⎜ ⎟⎛⎝

⎞⎠

= +ϕT

mERT

ctelnn

P P2

(2)

In these relations, χ is the crystallized fraction at a given tempera-ture, m is the crystal growth dimensionality, while Tp and R are thecrystallization peak temperature and the gas constant, respectively.

It should be remembered that the DSC data must be analyzed as-suming two different scenarios: (1) that crystal growth occurs at a fixednumber of nuclei or surface crystallization; (2) that further nuclei areformed during the DSC measurement. For both cases, theMatusita & Sakka equation (Eq. (2)) can be used to estimate the acti-vation energy for crystallization, which includes the nucleation andcrystal growth phenomena. In the former, one has n = m. When there isa significant nucleation, n = m + 1.

2.3. Dilatometric behavior

The CTE and softening temperature (Td) were measured for eachbulk glass sample (11 × 5× 3 mm) and the corresponding glass-cera-mics. To obtain glass-ceramics for CTE measurements, cylindrical pel-lets with 2.5 cm in diameter of glass powder (< 43 μm) were uniaxiallypressured and partially crystallized upon a single heat treatment at850 °C with different dwelling times (20, 60 and 100 h). CTE was

Table 1Nominal and experimental compositions (mol%) of the T1, T2, T3 and T4 glasses.

Glasses CaO SrO B2O3 TiO2 SiO2 Impurities

Nominal compositionT1 25.75 25.75 3.95 1.00 43.55 –T2 25.50 25.50 3.90 2.00 43.10 –T3 25.25 25.25 3.85 3.00 42.65 –T4 25.00 25.00 3.80 4.00 42.20 –

Chemical analysisa

T1 24.71 (5) 21.95 (1) 3.36 (3) 1.03 (4) 46.72 (9) 2.21 (1)T2 24.06 (3) 24.21 (8) 3.83 (2) 1.91 (4) 45.97 (6) n.da

T3 24.35 (9) 23.46 (2) 3.71 (7) 3.02 (3) 45.46 (9) n.da

T4 24.59 (9) 21.45 (5) 3.58 (7) 4.27 (8) 43.88 (5) 2.20 (2)

a Uncertainties of the last digit in parentheses. n.d.: below the limit of determination ofthe analytical method.

L.D. Silva et al. Journal of Non-Crystalline Solids xxx (xxxx) xxx–xxx

2

measured on a sample cut from this pellet. These properties were cal-culated in the temperature range between 100 and 550 °C. The ex-periments were conducted by dilatometry (NETZSCH DIL 402 C), understatic air and with a heating rate of 5 °C/min.

2.4. X-ray analysis

Partially crystallized samples prepared after heat treatment at850 °C for 20, 60, 100 (same heat treatment used for CTE measure-ments) and 382 h, were milled, sieved (< 43 μm) and characterized byXRD (Siemens Ultima IV diffractometer). XRD diffractograms werecollected in the range 10° ≤ 2θ ≤ 90°, with a step size of 0.02° and acounting time of 1 s. All experimental data were collected at roomtemperature. The crystalline phases were identified using the SearchMatch software.

2.5. Sintering behavior

To investigate the sintering behavior, cylindrical pellets of the T1,T2, T3 and T4 glasses, with height and diameter of 3 mm, were preparedby cold-pressing the glass powder (< 43 μm). Each glass compact wasthen placed on a 10 × 15 × 1 mm alumina (≈99.5 wt% Al2O3) sup-port of a side-view Hot Stage Microscope (HSM) (Hesse Instruments,EM 201, HR18) equipped with image analysis system and a Leica 1750/15 electrical furnace. Samples were heated up to 1400 °C at 10 °C/minand photographs were taken each 10 °C, or when the area contractionwas higher than 5%. Following the same procedures described by Laraet al. [29], the decreasing percentage of height, width, and area of eachsample were recorded, as well as the starting and maximum sinteringtemperatures, TFS and TMS, respectively, and the A/Ao ratio, where Aand Ao are the pellet's final and initial area.

The concurrence between glass sintering and crystallization kineticscan be inferred by the sinterability parameter (SC), which is given bythe difference between the onset crystallization (Tx) and the maximumsintering (TMS) temperatures (Eq. (3)). The larger is this difference;more independent are the crystallization and sintering processes. Theideal sinterability for a good glass-ceramic material for SOFCs should bebetween 25 and 60 °C [29–31].

= −S T TC X MS (3)

2.6. Isothermal crystallization

To assess the crystallization mechanism, some as received bulk glasssamples of T4 were heat treated at 850 °C for different periods of times.Then, they were embedded in epoxy resin, ground with silicon carbide(200, 300, 400, 500, 600 and 1200 mesh), polished with dilute CeO2,and etched with a solution of 0.6 HF/0.2 HCl (vol%) for 10 s. Themicrographs were obtained with the help of a Leica microscope DMRXmodel (Leica Microsystems, Wetzlar, Germany) in the reflected lightmode. The crystal growth rate, U, was estimated from the expressionU=dL/dt, where L and t are the crystalline layer thickness (in the caseof surface crystallization) and the time of heat treatment.

3. Results

Fig. 1 shows the DSC traces obtained from T1, T2, T3 and T4 bulkregular glass samples (3.0 × 3.0 × 2.5 mm) when heated at 10 °C/minfrom room temperature until the corresponding melting peak. The va-lues of Tg, Tx, TP and Tm were estimated as indicated in Fig. 1, andsummarized in Table 2.

Based on the Tg, Tx and Tm values obtained in Fig. 1, the evolution ofKH, KLL and KW parameters with the TiO2 content is presented in Fig. 2.Table 2 also presents the corresponding KH, KW and KLL values.

Fig. 3 shows the XRD patterns collected from fully crystallized T1,T2, T3 and T4 glass samples. The main crystalline phases identified were

CaSiO3 (JCPDS Card no. 43-1460), Ca2SiO4 (JCPDS Card no. 87-1261),Sr2SiO4 (JCPDS Card no. 38-0271) and Sr(TiO3) (JCPDS Card no. 86-0179).

Fig. 4 shows the variation in the relative area (from Hot Stage Mi-croscope HSM) and heat flow (from DSC) both obtained at the sametemperature range and heating rate (10 °C/min). Table 3 summarizesthe values of the temperatures of Tg, TX, TFS and TMS, as well as the ratioA/Ao (A is the pellet's final area) for each glass compact. Table 3 alsoshows the “sinterability” parameter, Sc, which was calculated con-sidering the TX and TFS data of Table 2. Furthermore, the softeningtemperature (Td) and thermal expansion coefficient data are also listed.

Fig. 5 shows part of the DSC curves obtained for the bulk glasssamples (3.0 × 3.0 × 2.5 mm) heated at different heating rates (ϕ= 5,

Fig. 1. DSC curves obtained at 10 °C/min from bulk glass samples for (a) T1; (b) T2; (c) T3;and (d) T4 glasses.

Table 2Tg, Tx (onset), TP, Tm, KH, KLL and KW data obtained for the CSrBS bulk glass samplesdoped with different TiO2 contents. Tg, Tx and Tm were taken as onset values.

TiO2, mol% 1 2 3 4

Tg (°C), ± 3 689 683 686 690Tx (°C), ± 5 903 912 922 917TP (°C),± 2 973 981 982 995Tm (°C), ± 3 1314 1290 1286 1266KH 0.521 0.600 0.630 0.638KLL 0.461 0.470 0.473 0.475KW 0.135 0.145 0.147 0.145

Fig. 2. KLL, KW and KH parameters versus TiO2 concentration (mol%). The dashed line isjust a guide to the eyes.

L.D. Silva et al. Journal of Non-Crystalline Solids xxx (xxxx) xxx–xxx

3

10, 15 and 20 °C/min).Based on Fig. 5, the temperature dependence of the crystallized

fraction (χ) was calculated from integration of each crystallization peak(figure not shown here) at 930 and 940 °C, using sigmoidal baselines.Fig. 6 shows the Ozawa plots of each glass composition. Taking intoaccount the slopes of the corresponding straight lines, the values of theAvrami index were calculated, and are presented in Table 4.

Fig. 7 shows the Matusita & Sakka plots obtained for the T1; T2; T3;and T4 glasses. The corresponding values of overall activation energy, E,were calculated from the slopes of these graphs and are presented inTable 4.

4. Discussion

Fig. 1 shows that the values of Tg of all glass compositions are inside

Fig. 3. XRD spectra obtained for the GCs of the T1, T2, T3 and T4glasses, which were crystallized at 850 °C for 382 h: (◯) CaSiO3

(43-1460), (△) Ca2SiO4 (87-1261), (✩) Sr2SiO4 (38-0271), (◇)SrTiO3 (86-0179) and (♣) non-identified.

600 700 800 900 1000 11000.6

0.7

0.8

0.9

1.0

1.1 Area DSC

T, °C

Are

a, A

/A0

TFS

TP

Tc

TMS

-0.2

0.0

0.2

0.4

a)

DS

C,

V/m

g

600 700 800 900 1000 1100

0.7

0.8

0.9

1.0

1.1

TMS

TFS

Tc

TP

Area DSC

T, °C

Are

a, A

/A0

-0.2

0.0

0.2

0.4

c)

DS

C,

V/m

g

600 700 800 900 1000 11000.6

0.7

0.8

0.9

1.0

1.1 Area DSC

T, °C

Are

a, A

/A0

-0.3

-0.2

-0.1

0.0

0.1

0.2

0.3

b)D

SC,

V/m

g

Tc T

MS

TP

TFS

600 700 800 900 1000 11000.6

0.7

0.8

0.9

1.0

1.1

d)

Area DSC

T, °C

Are

a, A

/A0

TMS

TC

TFS

TP

-0.2

0.0

0.2

0.4

DSC

, V

/mg

Fig. 4. HSM (Hot Stage Microscope) and DSC (DifferentialScanning Calorimetry) curves obtained for the powderedglasses a) T1, b) T2, c) T3 and d) T4 (< 43 μm). The mea-surements were conducted at the same heating rate (10 °C/min) and atmosphere (static air).

Table 3Experimental data of Tg, TX, TFS and TMS (glass powder - 43 μm), SC, A/Ao, Td, and CTEobtained for the T1, T2, T3 and T4 glasses and corresponding GCs (850 °C, 100 h).

TiO2, mol% 1 2 3 4

Tg (°C), ± 3 688 680 684 686TX (°C),± 5 812 809 833 849TFS (°C),± 5 746 706 739 738TMS (°C)± 5 820 804 813 819SC (ºC)± 5 −8.5 4.45 19.4 30.2A/Ao ± 0,02 0.70 0.73 0.67 0.70Td (°C) ± 5 727 724 722 728CTE (10−6 °C−1),± 0,1 (glass) 13.5 10.2 10.9 10.7CTE (10−6 °C−1),± 0,1 (glass-ceramics – 100 h) 13.9 11.1 10.8 11.0

L.D. Silva et al. Journal of Non-Crystalline Solids xxx (xxxx) xxx–xxx

4

the temperature range of the SOFCs operation (600–1000 °C [9,11,12]).On increasing TiO2 concentration, it is noteworthy that, except for T1,Tg tends to increase, which means that the glass network strength in-creases.

Combining the dependence of Tg (Fig. 1) and the GS parameters(Fig. 2) with the increasing TiO2 concentration, one can conclude that

the glass network strength increases and the tendency for crystallizationdiminishes. The KH parameter was the most sensitive to changes in ti-tanium concentration. It is also noticeable that the glass composition T4is the most resistant against the devitrification.

As demonstrated by Fig. 3, different TiO2 content in the SrO-CaO-B2O3-SiO2 glass system favors the precipitation of phases like CaSiO3

(JCPDS Card no 43-1460), Ca2SiO4 (JCPDS Card no 87-1261), Sr2SiO4

(JCPDS Card no 38-0271) and Sr(TiO3) (JCPDS Card no 86-0179). Thepeaks marked with the symbol (♣) belong to an unknown crystallinephase. Despite the very long time used here to crystallize completelythe glass samples (382 h at 850 °C), the crystalline phases precipitatedhere, as well the non-identified peaks, are in excellent agreement withZhang et al. [20] and Chen et al. [21] results. It should also be re-membered that the presence of CaSiO3 and some Sr-containing (Sr(TiO3) and Sr2SiO4) crystalline phases are required for the high per-formance of glass sealants, since the former contribute to the increasingof the thermal expansion coefficient [6] while the latter with the

500 600 700 800 900 1000 1100

-0.1

0.0

0.1

0.2

0.3

0.4

0.5

a

DS

C,

V/m

g

T, °C

5 °C/min

10 °C/min

15 °C/min

20 °C/min

500 600 700 800 900 1000 1100-0.1

0.0

0.1

0.2

0.3 b

DSC

, V

/mg

T, °C

5 °C/min 10 °C/min 15 °C/min 20 °C/min

500 600 700 800 900 1000 1100-0.1

0.0

0.1

0.2

0.3 c

DSC

, V

/mg

T, °C

5 °C/min 10 °C/min 15 °C/min 20 °C/min

500 600 700 800 900 1000 1100

0.0

0.1

0.2

0.3

0.4

d 5 °C/min

10 °C/min

20 °C/min

15 °C/minD

SC

, V

/mg

T, °C

Fig. 5. DSC curves obtained for the bulk glass samples: (a)T1; (b) T2; (c) T3; and (d) T4. The experiments were con-ducted at different heating rates (5, 10, 15 and 20 °C/min).

Fig. 6. Ozawa plots obtained for the T1; T2; T3; and T4 bulk glass samples, respectively. χis the crystallized fraction at 930 °C, 940 °C, 940 °C and 940 °C. The heating rates usedwere: ϕ= 5, 10, 15 and 20 °C/min.

Table 4Mean values of the Avrami's index and the activation energy obtained for the T1, T2, T3and T4 glasses, using the Ozawa and Matusita & Sakka models, respectively.

Glass n E, kJ/mol

T1 1.17 ± 0.02 (900–930 °C) 402 ± 14T2 1.25 ± 0.10 (940–960 °C) 381 ± 34T3 1.32 ± 0.06 (940–960 °C) 385 ± 28T4 1.62 ± 0.05 (940–960 °C) 384 ± 26

Fig. 7. Matusita & Sakka plots (TP is the temperature of the crystallization peak) obtainedfor the T1; T2; T3; and T4 bulk glass samples. The heating rates (ϕ) used were 5, 10, 15 and20 °C/min.

L.D. Silva et al. Journal of Non-Crystalline Solids xxx (xxxx) xxx–xxx

5

improvement of the chemical compatibility between sealants and themetallic component of IT-SOFCs [20].

According to Table 3, the experimental CTE values (from 100 to550 °C) measured for each glass composition and the correspondingglass-ceramics are very close, where the CTE of the former varied from(10.2 ± 0.1) × 10−6 to (13.5 ± 0.1) × 10−6 K−1, while those ofGC ranged from (10.7 ± 0.1) × 10−6 to (13.4 ± 0.1) × 10−6 K−1.These values are in an excellent agreement with the experimental CTEdata measured for some interconnect components in the same tem-perature range, such as a ferritic stainless steel 430SS are11.4 × 10−6 K−1[32], and 12.5 × 10−6 K−1 to Crofer 22 APU steelplates [33]. Hence, this indicates that both glasses and glass-ceramics ofthe CSrBS system are promising candidates to be applied as sealants forIT-SOFCs.

Furthermore, it was also observed that different dwelling times (20,60 and 100 h) at 850 °C did not change significantly the CTE values. Forthis reason, only the CTE data of the GC obtained at 100 h are presentedin Table 3.

Based on experimental data, Geasee et al. [34] demonstrated thatthe glass transition temperatures and the coefficient of thermal ex-pansion are strongly correlated properties. Therefore they should betaken into account to the selection of glass sealant compositions. Fouryears later, Fergus [6] extended the Tg vs. CTE plot adding othercompositions, which is reproduced here in Fig. 8, where the rectangulararea corresponds to the most promising glass compositions for sealantsapplication. As one can see from Fig. 8, the glasses T3 and T4 are insidethe rectangular area. Combining with the precipitation of the Sr2SiO4

and SrTiO3 crystalline phases (Fig. 3), it seems that these glasses are themost promising candidates for sealants in IT-SOFCs, which allows theuse of ferritic stainless steels as interconnect components [20,21].

Table 3 shows that the temperatures of first (TFS) and maximumshrinkage (TMS) varied from 706 to 746 °C and from 804 to 821 °C,respectively. These temperatures did not change monotonically withTiO2 content. The values of A/Ao changed from 0.67 to 0.73, whichcorresponds to 95–98% of the theoretical density of each glass powdercompact [11,35,36].

From Fig. 4, the HSM curves exhibit a single stage shrinkage be-havior, and the DSC scans showed a single exothermic crystallizationpeak (in agreement with Figs. 1 and 5). It is noteworthy that only T4(Fig. 4d) presented complete densification before reaching the crystal-lization event, which means the highest densification. One can also seethat the sinterability increases with the TiO2 content (Table 3), and T4has the highest SC value (≅30 °C). Therefore, although the T3 and T4

glass match the criteria proposed by Geasee et al. [34], only T4 pre-sented SC values suitable for sealing applications [29–31,37].

According to Table 4 and Fig. 6, n varied between 1 and 1.5. Inprinciple, this means that these glasses crystallize mainly from thesurface or in bulk from a fixed number of nuclei. In both cases, thecrystal growth can be controlled by interface or diffusion [38]. In thefirst one, the crystal dimensions should increase linearly with time,while in the condition of diffusion-controlled growth, it should increasewith the square root of time (~ t ). One must remember that non-stoichiometric compositions favor the crystallization by diffusion con-trol. In fact, Fig. 9 shows that the T4 glass has a linear L vs. t de-pendence, and no crystal in the volume was revealed in the corre-sponding micrograph (inset of Fig. 9). In addition, Fig. 10 shows that,even upon 20 h of heat treatment at 850 °C, at least three crystallinephases are precipitated: CaSiO3, Sr2SiO4 and SrTiO3. It must be em-phasized that even at lower temperatures and shorter times than thoseused here, diffraction lines corresponding to these crystals were alreadydetected, as demonstrated by Zhang et al. [20].

Also, assuming the experimental error, an almost constant value of Ewas obtained (≈380 kJ/mol) for all TiO2 contents, as one can see fromFig. 7 and Table 4. Taking into account the complexity of the crystalline

Fig. 8. Glass transition temperature dependence of thecoefficient of thermal expansion for several glass composi-tions to be used as SOFC sealants including experimentaldata obtained in this work. The original figure was ex-tracted from reference [6]. The dashed line is just a guide tothe eyes.

Fig. 9. Layer thickness as a function of square root of time of treatment at 850 °C for theT4 glass.

L.D. Silva et al. Journal of Non-Crystalline Solids xxx (xxxx) xxx–xxx

6

phases, it should be expected that the activation energies were affectedby the changing of the composition of glass/crystal interface during theheating of the glass samples. Nevertheless, it is well known that theactivation energies are calculated taking into account the temperatureof the crystallization peak, which is related to the overall crystal-lization. In this paper, the DSC curves of each glass composition (Fig. 5)at different heating rates recorded a single exothermic peak, whosewidth can indicate the precipitation of more than one crystalline phase.This result is in agreement with the XRD spectra presented at Fig.10.Therefore, the DSC experiments presented here did not allow to sepa-rate the contribution of each crystalline phase to the activation energy.

According to the Ozawa and Matusita & Sakka models (Eq. (1) and(2), respectively), the values of n and E are related to the shift of Tp withthe heating rate. This kinetic effect is related to the time required toachieve the overall crystallization, i.e., as the heating rate increases,less time is spent during the heating, and more residual glass remains,and Tp shifts to highest values. Hence, a possible changing of thecrystallization mechanism (n) should be reinforced by variations in theactivation energies, as demonstrated by Marques et al. [39], who in-vestigated the influence of the particle size on the crystallization me-chanism of an LS2 glass.

In our case, we investigated the crystallization behavior of differentTiO2% containing SrO-CaO-B2O3-SiO2 glass system with the same par-ticle size (bulk). Although Tp has shifted with the TiO2 content, almostthe same values of E were obtained, which means that n should also beconstant. This result means that introducing x TiO2, at least for1 < x < 4 (mol%), was not effective to change the crystallizationmechanism.

Hence, based on these three pieces of evidence, we suggest that thecrystallization of the SrO-CaO-B2O3-SiO2 glasses with different TiO2

contents is mainly from the surface, and the growth of the crystallinelayer is governed by diffusion.

5. Conclusions

The thermal properties, sintering and crystallization kinetics wereinvestigated for the CaO·SrO·B2O3·SiO2 (CSrBS) system doped withvarious TiO2 content (1–4 mol%).

The Sr2SiO4, Sr(TiO3), Ca2SiO4 and CaSiO3 crystalline phases wereprecipitated in all glass-ceramics. These SrO-rich phases are particularlyimportant since they minimize the interfacial reaction between thesealant and the component alloys of the IT-SOFCs.

As the Tg and the GS parameters of each glass increase with the

increasing TiO2, one can conclude that the glass network strength in-creases and the tendency for crystallization diminishes. The glass T4 isfound to be the most resistant against the devitrification.

All glasses exhibit a single stage shrinkage behavior and a singleexothermic crystallization peak. From the HSM curves, it was demon-strated that only T4 presents complete densification before reaching thecrystallization event, which means the highest densification.

Based on DSC experiments, one can conclude that the Avramicoefficient, as well as the activation energy of each glass did not changewith the TiO2 molar percentage.

Combining XRD and OM techniques, it was demonstrated that thecrystallization mechanism of these bulk CSrBS glass samples doped withTiO2 (1–4 mol%) was mainly from the surface, and the growth of thecrystalline layer is governed by diffusion.

Although, T2, T3 and T4 glasses present values of Tg and CTE, ade-quate for applications in intermediate temperature solid oxide fuelcells, only T4 glass presents the value of sintering parameter (SC) sui-table for sealing applications.

Acknowledgments

The authors are indebted to the following Brazilian research fundingagencies: CNPq (National Council for Scientific and TechnologicalDevelopment) – Process # 2013/304452 and 2013/488182, CAPES(Federal Agency for the Support and Improvement of Higher Education)for granting a master's scholarship to the first author, CAPES grantnumber PNPD20131474-33001014004P9, FAPEMA (MaranhãoFoundation for Scientific Research and Development), Process # 2014/00260 and 2014/01781, FAPESP Process # CEPID2013/07793-6, andPPGCEM/UFSCar - CAPES Process # AUXPE no. 0880/2016.

References

[1] R.W. Lashway, Fuel cells: the next evolution, MRS Bull. 30 (2005) 581–586, http://dx.doi.org/10.1557/mrs2005.164.

[2] A.A. Reddy, D.U. Tulyaganov, M.J. Pascual, V.V. Kharton, E.V. Tsipis,V.A. Kolotygin, et al., SrO-containing diopside glass-ceramic sealants for solid oxidefuel cells: Mechanical reliability and thermal shock resistance, Fuel Cells 13 (2013)689–694, http://dx.doi.org/10.1002/fuce.201200237.

[3] P. Kofstad, R. Bredesen, High temperature corrosion in SOFC environments, SolidState Ionics 52 (1992) 69–75, http://dx.doi.org/10.1016/0167-2738(92)90092-4.

[4] Z. Yang, K.S. Weil, D.M. Paxton, J.W. Stevenson, Selection and evaluation of heat-resistant alloys for SOFC interconnect applications, J. Electrochem. Soc. 150 (2003)A1188, http://dx.doi.org/10.1149/1.1595659.

[5] D. Simwonis, H. Thülen, F. Dias, A. Naoumidis, D. Stöver, Properties of Ni/YSZporous cermets for SOFC anode substrates prepared by tape casting and coat-mix®process, J. Mater. Process. Technol. 92–93 (1999) 107–111, http://dx.doi.org/10.1016/S0924-0136(99)00214-9.

[6] J.W. Fergus, Sealants for solid oxide fuel cells, J. Power Sources 147 (2005) 46–57,http://dx.doi.org/10.1016/j.jpowsour.2005.05.002.

[7] P.A. Lessing, A review of sealing technologies applicable to solid oxide electrolysiscells, J. Mater. Sci. 42 (2007) 3465–3476, http://dx.doi.org/10.1007/s10853-006-0409-9.

[8] R.K. Brow, M.L. Schmitt, A survey of energy and environmental applications ofglass, J. Eur. Ceram. Soc. 29 (2009) 1193–1201, http://dx.doi.org/10.1016/j.jeurceramsoc.2008.08.011.

[9] N.P. Bansal, E.A. Gamble, Crystallization kinetics of a solid oxide fuel cell seal glassby differential thermal analysis, J. Power Sources 147 (2005) 107–115, http://dx.doi.org/10.1016/j.jpowsour.2005.01.010.

[10] M. Peksen, Numerical thermomechanical modelling of solid oxide fuel cells, Prog.Energy Combust. Sci. 48 (2015) 1–20, http://dx.doi.org/10.1016/j.pecs.2014.12.001.

[11] M.J. Pascual, V.V. Kharton, E. Tsipis, A.A. Yaremchenko, C. Lara, A. Durán, et al.,Transport properties of sealants for high-temperature electrochemical applications:RO–BaO–SiO2 (R=Mg, Zn) glass–ceramics, J. Eur. Ceram. Soc. 26 (2006)3315–3324, http://dx.doi.org/10.1016/j.jeurceramsoc.2005.11.002.

[12] K.L. Ley, M. Krumpelt, R. Kumar, J.H. Meiser, I. Bloom, Glass-ceramic sealants forsolid oxide fuel cells: part I. Physical properties, J. Mater. Res. 11 (1996)1489–1493, http://dx.doi.org/10.1557/JMR.1996.0185.

[13] J. Puig, F. Ansart, P. Lenormand, L. Antoine, J. Dailly, Sol–gel synthesis andcharacterization of barium (magnesium) aluminosilicate glass sealants for solidoxide fuel cells, J. Non-Cryst. Solids 357 (2011) 3490–3494, http://dx.doi.org/10.1016/j.jnoncrysol.2011.06.025.

[14] T. Zhang, S. Wang, Seals for planar solid oxide fuel cells: the state of the art, in:S.P. Jiang, P.K. Shen (Eds.), Nanostructured Adv. Mater. Fuel Cells, 1st ed, CRC

Fig. 10. XRD pattern obtained for the GCs of the T4 glass, which were crystallized at850 °C for 20, 60 and 100 h: (◯) CaSiO3 (43-1460), (✩) Sr2SiO4 (38-0271), (◇) SrTiO3

(86-0179) and (♣) non-identified.

L.D. Silva et al. Journal of Non-Crystalline Solids xxx (xxxx) xxx–xxx

7

Press, Boca Ratom, 2013, p. 600, , http://dx.doi.org/10.1201/b16107.[15] R.N. Singh, Self-repairable glass seals for solid oxide fuel cells, J. Mater. Res. 27

(2012) 2055–2061, http://dx.doi.org/10.1557/jmr.2012.188.[16] F. Smeacetto, M. Salvo, F.D. D'Hérin Bytner, P. Leone, M. Ferraris, New glass and

glass–ceramic sealants for planar solid oxide fuel cells, J. Eur. Ceram. Soc. 30(2010) 933–940, http://dx.doi.org/10.1016/j.jeurceramsoc.2009.09.033.

[17] M.K. Mahapatra, K. Lu, Interfacial study of Crofer 22 APU interconnect-SABS-0 sealglass for solid oxide fuel/electrolyzer cells, J. Mater. Sci. 44 (2009) 5569–5578,http://dx.doi.org/10.1007/s10853-009-3781-4.

[18] F. Smeacetto, M. Salvo, M. Ferraris, V. Casalegno, P. Asinari, A. Chrysanthou,Characterization and performance of glass–ceramic sealant to join metallic inter-connects to YSZ and anode-supported-electrolyte in planar SOFCs, J. Eur. Ceram.Soc. 28 (2008) 2521–2527, http://dx.doi.org/10.1016/j.jeurceramsoc.2008.03.035.

[19] L. Peng, Q. Zhu, Thermal cycle stability of BaO–B2O3–SiO2 sealing glass, J. PowerSources 194 (2009) 880–885, http://dx.doi.org/10.1016/j.jpowsour.2009.06.018.

[20] T. Zhang, Q. Zou, F. Zeng, S. Wang, D. Tang, H. Yang, Improving the chemicalcompatibility of sealing glass for solid oxide fuel cells: blocking the reactive speciesby controlled crystallization, J. Power Sources 216 (2012) 1–4, http://dx.doi.org/10.1016/j.jpowsour.2012.05.034.

[21] J. Chen, Q. Zou, F. Zeng, S. Wang, D. Tang, H. Yang, et al., Tailoring the sealingproperties of TiO2–CaO–SrO–B2O3–SiO2 glass-ceramic seals: thermal properties,chemical compatibility and electrical property, J. Power Sources 241 (2013)578–582, http://dx.doi.org/10.1016/j.jpowsour.2013.04.134.

[22] A. Goel, A.A. Reddy, M.J. Pascual, L. Gremillard, A. Malchere, J.M.F. Ferreira,Sintering behavior of lanthanide-containing glass-ceramic sealants for solid oxidefuel cells, J. Mater. Chem. 22 (2012) 10042, http://dx.doi.org/10.1039/c2jm16300d.

[23] M.L.F. Nascimento, L.A. Souza, E.B. Ferreira, E.D. Zanotto, Can glass stabilityparameters infer glass forming ability? J. Non-Cryst. Solids 351 (2005) 3296–3308,http://dx.doi.org/10.1016/j.jnoncrysol.2005.08.013.

[24] T.V.R. Marques, A.A. Cabral, Influence of the heating rates on the correlation be-tween glass-forming ability (GFA) and glass stability (GS) parameters, J. Non-Cryst.Solids 390 (2014) 70–76, http://dx.doi.org/10.1016/j.jnoncrysol.2014.02.019.

[25] T. Ozawa, Kinetics of non-isothermal crystallization, Polymer (Guildf). 12 (1971)150–158, http://dx.doi.org/10.1016/0032-3861(71)90041-3.

[26] A.M. Rodrigues, J.M. Rivas Mercury, V.S. Leal, A.A. Cabral, Isothermal and non-isothermal crystallization of a fresnoite glass, J. Non-Cryst. Solids 362 (2013)114–119, http://dx.doi.org/10.1016/j.jnoncrysol.2012.11.026.

[27] A.R. Molla, A.M. Rodrigues, S.P. Singh, R.F. Lancelotti, E.D. Zanotto,A.C.M. Rodrigues, et al., Crystallization, mechanical, and optical properties oftransparent, nanocrystalline gahnite glass-ceramics, J. Am. Ceram. Soc. 100 (2017)

1963–1975, http://dx.doi.org/10.1111/jace.14753.[28] K. Matusita, S. Sakka, Kinetic study of crystallization of glass by differential thermal

analysis—criterion on application of Kissinger plot, J. Non-Cryst. Solids 38–39(1980) 741–746, http://dx.doi.org/10.1016/0022-3093(80)90525-6.

[29] C. Lara, M.J. Pascual, A. Durán, Glass-forming ability, sinterability and thermalproperties in the systems RO-BaO-SiO2 (R =Mg, Zn), J. Non-Cryst. Solids 348(2004) 149–155, http://dx.doi.org/10.1016/j.jnoncrysol.2004.08.140.

[30] A. Goel, M.J. Pascual, J.M.F. Ferreira, Stable glass-ceramic sealants for solid oxidefuel cells: influence of Bi2O3 doping, Int. J. Hydrog. Energy 35 (2010) 6911–6923,http://dx.doi.org/10.1016/j.ijhydene.2010.04.106.

[31] D.U. Tulyaganov, A.A. Reddy, V.V. Kharton, J.M.F. Ferreira, Aluminosilicate-basedsealants for SOFCs and other electrochemical applications - a brief review, J. PowerSources 242 (2013) 486–502, http://dx.doi.org/10.1016/j.jpowsour.2013.05.099Review.

[32] Long Products Stainless Steel Grade Sheet, North Am. Stainl., 2010, http://www.northamericanstainless.com/wp-content/uploads/2010/10/Grade-4301.pdf (ac-cessed September 4, 2015).

[33] L. Luo, Y. Lin, Z. Huang, Y. Wu, L. Sun, L. Cheng, et al., Application ofBaO–CaO–Al2O3–B2O3–SiO2 glass–ceramic seals in large size planar IT-SOFC,Ceram. Int. 41 (2015) 9239–9243, http://dx.doi.org/10.1016/j.ceramint.2015.01.091.

[34] P. Geasee, T. Schwickert, U. Diekmann, R. Conradt, Glasses from the system RO-R2O3-SiO2 as sealants of high chromium steel components in the planar SOFC,Ceram. Mater. Components Engines, Wiley-VCH Verlag GmbH, Weinheim,Germany, 2007, pp. 57–62, , http://dx.doi.org/10.1002/9783527612765.ch10.

[35] A. Goel, D.U. Tulyaganov, M.J. Pascual, E.R. Shaaban, F. Muñoz, Z. Lü, et al.,Development and performance of diopside based glass-ceramic sealants for solidoxide fuel cells, J. Non-Cryst. Solids 356 (2010) 1070–1080, http://dx.doi.org/10.1016/j.jnoncrysol.2010.01.012.

[36] A.A. Reddy, D.U. Tulyaganov, M.J. Pascual, V.V. Kharton, E.V. Tsipis,V.A. Kolotygin, et al., Diopside-Ba disilicate glass-ceramic sealants for SOFCs:Enhanced adhesion and thermal stability by Sr for Ca substitution, Int. J. Hydrog.Energy 38 (2013) 3073–3086, http://dx.doi.org/10.1016/j.ijhydene.2012.12.074.

[37] S.T. Reis, M.J. Pascual, R.K. Brow, C.S. Ray, T. Zhang, Crystallization and proces-sing of SOFC sealing glasses, J. Non-Cryst. Solids 356 (2010) 3009–3012, http://dx.doi.org/10.1016/j.jnoncrysol.2010.02.028.

[38] J.W. Christian, Formal theory of transformation kinetics, Theory Transform. Met.Alloy, 3rd ed., Pergamon Press, Oxford, 2002, pp. 529–552, , http://dx.doi.org/10.1016/B978-008044019-4/50016-7.

[39] L.E. Marques, A.M.C. Costa, M.C. Crovace, A.C.M. Rodrigues, A.A. Cabral, Influenceof particle size on nonisothermal crystallization in a lithium disilicate glass, J. Am.Ceram. Soc. 98 (2015) 774–780, http://dx.doi.org/10.1111/jace.13380.

L.D. Silva et al. Journal of Non-Crystalline Solids xxx (xxxx) xxx–xxx

8