Synthesis and characterization of BaAl2Si2O8 using mechanically activated precursor mixtures containing coal fly ash

Synthesis and characterization of BaAl2Si2O8 using mechanically activated precursor mixtures containing coal fly ash

Available online at www.sciencedirect.com Journal of the European Ceramic Society 33 (2013) 3287–3300 Synthesis and characterization of BaAl2Si2O8 u...

2MB Sizes 23 Downloads 97 Views

Available online at www.sciencedirect.com

Journal of the European Ceramic Society 33 (2013) 3287–3300

Synthesis and characterization of BaAl2Si2O8 using mechanically activated precursor mixtures containing coal fly ash C.M. López-Badillo a,b,∗ , J. López-Cuevas a , C.A. Gutiérrez-Chavarría a , J.L. Rodríguez-Galicia a , M.I. Pech-Canul a a

CINVESTAV-IPN, Unidad Saltillo, Parque Industrial Saltillo-Ramos Arizpe, Industria Metalúrgica 1062„ 25900 Ramos Arizpe, Coahuila, Mexico b ITESM Campus Saltillo, Prol. Juan de la Barrera 1241 Ote., 25270 Saltillo, Coahuila, Mexico Received 6 February 2013; received in revised form 1 May 2013; accepted 5 May 2013 Available online 12 June 2013

Abstract BaAl2 Si2 O8 (BAS) was synthesized by solid state reaction, using coal fly ash [CFA, containing (in wt.%) 75–78% silico-aluminous glassy phase, 14.6% mullite, 5.2% quartz, 1.9% magnetite, plus other minor phases] as main raw material. A BaCO3 –CFA–Al2 O3 powder mixture of stoichiometric BAS composition was mechanically activated in an attrition mill for up to 12 h and then sintered at 900–1300 ◦ C. The monoclinic BAS phase (Celsian) was desired because it has better mechanical and thermal properties than the hexagonal BAS polymorph (Hexacelsian), but this tends to form first, remaining frequently metastably at low temperatures; besides, the Hexacelsian to Celsian conversion is sluggish and difficult to occur. The reaction rate, the apparent density and the mechanical properties of the synthesized materials increased with increasing milling time and sintering temperature. The mineralizing effect caused by some of the CFA impurities produced Hexacelsian to Celsian conversions higher than those previously reported for mechanically activated BAS materials. © 2013 Elsevier Ltd. All rights reserved. Keywords: Solid state reaction; Mechanical activation; Thermal analyses; Celsian; Fly ash

1. Introduction Barium aluminosilicate, BaAl2 Si2 O8 (BAS), is highly refractory (melting point of ∼1760 ◦ C),1 and it exists primarily in three different polymorphic forms: Celsian or Monocelsian (monoclinic), Hexacelsian (hexagonal) and ␣-Hexacelsian (orthorhombic). Hexacelsian is thermodynamically stable at temperatures between 1590 ◦ C and the melting point, whereas Celsian is stable at temperatures below 1590 ◦ C.2 However, Hexacelsian can exist as a metastable phase at all temperatures between 1590 ◦ C and room temperature. Besides, at ∼300 ◦ C, Hexacelsian undergoes a rapid and reversible structural transformation into the orthorhombic form,3 which is accompanied by a volume change of ∼3%,4 which usually originates microcracks ∗ Corresponding author at: CINVESTAV-IPN, Unidad Saltillo, Parque Industrial Saltillo-Ramos Arizpe, Industria Metalúrgica 1062, 25900 Ramos Arizpe, Coahuila, Mexico. Tel.: +52 844 4389600; fax: +52 844 4389610. E-mail address: [email protected] (C.M. López-Badillo).

0955-2219/$ – see front matter © 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jeurceramsoc.2013.05.014

in the material, with a corresponding reduction in its mechanical properties. Thus, Hexacelsian is an undesirable phase. In contrast, Celsian is resistant to oxidation, reduction and alkali attack,2,4 and it also shows good thermal shock resistance,5 low thermal expansion coefficient6 and rather low dielectric permittivity.7,8 Celsian has numerous applications as refractory glass–ceramic and in precision molds for the casting of molten metals and alloys. It is also used for high temperature engine components5 and as substrate for electronic applications.9,10 It can be used to produce printed capacitors of reduced dielectric constant11 and it is a promising candidate for advanced radome application5,12 or for missile nose cones.5 Celsian has been commonly synthesized by solid-state reaction13–15 from mixtures of BaCO3 , Al2 O3 and SiO2 reagentgrade powders, or using either kaolin16 or topaz17 as alternative sources of Al2 O3 and SiO2 . Celsian has also been synthesized by the glass–ceramics route18–20 using mixtures of BaCO3 , Al2 O3 and SiO2 , in some cases with additions of MgO.18 Other synthesis routes that have been employed for this are: ion exchange of

3288

C.M. López-Badillo et al. / Journal of the European Ceramic Society 33 (2013) 3287–3300

zeolite precursors,21,22 sol gel,23–25 solid metal oxidation26–28 and hydrothermal processes.27,29 A raw material that has recently been employed for the synthesis of Celsian, as a novel alternative source of Al2 O3 and SiO2 , is coal fly ash (CFA).13,30–32 CFA is a by-product of the combustion of coal in power-generation plants,33 which is collected from the electrostatic precipitators34 and whose main constituents are silico-aluminous glass, mullite (Al6 Si2 O13 ), quartz, free unburned carbon and small amounts of other minerals.35 This use of CFA contributes to the solution of an important environmental problem, since it is estimated that the current annual worldwide production of CFA is around 500 million tones,36 most of which is discarded in open landfills. At the same time, a valuable product is obtained from a very lowcost industrial residue,37 with potentially important economic impact. Hexacelsian tends to be the first phase to form, and its transformation into the monoclinic phase is very sluggish.38 However, the kinetics of this transformation can be enhanced by the addition of mineralizers such as Li2 O, NaF, CaO, TiO2 , MgO, SrF2 , BeO and Cr2 O3 .14,39,40 Mechanical activation is a process that has also been tried in order to promote this transformation, although the results obtained so far have not been very encouraging. It has been found41 that an increased surface area in the milled precursor powders promotes the surface nucleation of Hexacelsian rather than the bulk nucleation of Celsian. These results were obtained using stoichiometric mixtures of BaCO3 , Al2 O3 and SiO2 crystalline reagent-grade powders mechanically activated in a vibratory mill. However, there are reasons to believe that the use of alternative raw materials such as CFA could produce results that are somewhat different to those previously reported in the literature, and this has not been studied in detail so far. The main goal of this work was to investigate the effect of mechanical activation on the synthesis of BAS using a stoichiometric precursor mixture of BaCO3 , Al2 O3 and CFA powders. 2. Experimental procedure 2.1. Raw materials For the preparation of a precursor mixture corresponding to the stoichiometric composition of BAS, the starting materials used were: BaCO3 (Alkem, Mexico, with purity of 99.43 wt.% and mean particle size of 5.3 ␮m), Al2 O3 (HPA, SASOL, USA, with purity of 99.99 wt.% and mean particle size of 1.5 ␮m), and CFA collected at the electrostatic precipitator bags of the “José López Portillo” Mexican power plant, with a mean particle size of 86 ␮m, and whose chemical composition was analyzed by X-Ray Fluorescence (Brucker AXS S4 Pioneer, XRF apparatus).While the BaCO3 and Al2 O3 were used as-received, the CFA was first sieved using a 125 ␮m mesh and then milled for 1 h in a ball mill and subsequently subjected to a manual elimination of a large part of its original content of iron oxides by means of a wet magnetic separation process using a neodymium magnet. This was done in order to minimize the formation of

iron-containing secondary phases of low melting point during the synthesis of BAS. The portion that was repelled by the neodymium magnet due to its high content of diamagnetic minerals, such as SiO2 , was designated as “beneficiated CFA” or simply as “CFA”. This fraction constituted ∼80% (in weight) of the as-received CFA.42 After subjecting it to an additional wet ball milling for 1 h, achieving a mean particle size of 16.9 ␮m, this material was employed for the synthesis of BAS. The precursor mixture was composed by (wt.%): 44.38% of BaCO3 , 43.59% of beneficiated CFA and 12.03% of Al2 O3 . 2.2. Mechanical activation of precursor mixture The precursor mixture was milled in a closed chamber laboratory attrition mill fitted with Teflon-lined propellers and container, and which was operated at 1600 rpm. Milling was carried out in air at room temperature, for 5 min (non-mechanically activated reference mixture), 4 h, 8 h and 12 h, employing 8 mol% MgO-partially stabilized ZrO2 balls with a diameter of 15 mm as milling media, with a ball/load mass ratio of 5:1, using ethanol as dispersion medium. 2.3. Characterization of milled materials 2.3.1. Thermal analysis The TGA/DTA curves were recorded by a simultaneous Thermogravimetric/Differential Thermal Analyzer (Perkin Elmer, Pyris Diamond TGA/DTA apparatus). The analyses were performed using 15 ± 0.09 mg samples and platinum pans, under air atmosphere from ambient temperature to above 1200 ◦ C, at a heating rate of 10 ◦ C/min. The derivative DTG curves were also obtained for the studied materials. 2.3.2. Surface area The specific surface area of the precursor mixtures was calculated from the Brunauer–Emmett–Teller (BET) isotherms determined by the volumetric method using a gas sorption analyzer (Quantachrome, AUTOSORB-1). For this, the samples were degassed at 300 ◦ C and the adsorption of nitrogen was carried out at −195.8 ◦ C. 2.3.3. Particle size distribution The mean particle size of the mechanically activated and reference precursor mixtures were determined by the laser diffraction method using a Coulter LS 100Q particle size analyzer. 2.3.4. X-ray diffraction (XRD) The crystalline phases present in the mechanically activated and reference precursor mixtures were identified by XRD, using a Philips X’Pert 3040 apparatus. The XRD patterns were ˚ in the 2θ range recorded employing Cu K␣ radiation (λ = 1.54 A) of 10–80◦ , using a scan step of 0.03◦ , a voltage of 40 kV, and a current of 30 mA. XRD was also used to estimate the degree of amorphization attained by BaCO3 contained in the milled

C.M. López-Badillo et al. / Journal of the European Ceramic Society 33 (2013) 3287–3300

samples using a method developed by Ohlberg and Strickler,43,44 which is based on Eq. (1):   U0 Ix × 100 A= 1− × (1) I0 Ux where A denotes the degree of amorphization, U and I denote the background and the integrated intensities of the diffraction lines of the reference (0) and mechanically activated (x) samples, respectively, in the corresponding XRD patterns. The average crystallite size (size of the coherently diffracting domains) of BaCO3 contained in the milled powders was also estimated by XRD, using Scherrer’s equation45 : t=

0.9λ BC cos θ

ESEM apparatus, and employing an acceleration voltage of 20 kV. Prior to this, the sintered samples were cross-sectioned and one part of them was mounted in cold-cure epoxy resin. Then, the cut surface was ground using SiC papers with successive grit sizes from 80 grit to 2400 grit, which was followed by polishing to a mirror finish using diamond particles with successive sizes of 3, 1 and ¼ ␮m. Prior to their observation on the SEM, the samples were etched for 15 s in 0.5 wt.% hydrofluoric acid aqueous solution and then graphite-coated using a JEOL JEE-400 vacuum evaporator. The measurement of the grain size and of the aspect ratio (length/width)47 of the BAS grains was carried out on digital micrographs of the sintered materials using the tpsDig software (version 2.16).

(2)

where t is the crystallite size (Å); λ is the X-ray wavelength (Å); BC is the angular width measured at half the maximum intensity (radians), and θ is the angle (radians) that exactly satisfies the Bragg equation for the particular values of λ and the interplanar spacing involved. The K␣1 and K␣2 contributions were deconvoluted using the MDI Jade 6 software, which was also employed to determine the position of the peaks. For this, the peak width was fitted to a pseudo-Voigt mathematical function. 2.4. Solid state reaction The mechanically activated mixtures were uniaxially pressed at 100 MPa into cylinders weighting 1 g and with a diameter of 1.0 cm and a height of ∼0.5 cm. The disks were subsequently heat treated in the temperature range of 900–1300 ◦ C, in steps of 100 ◦ C, with heating rate of 5 ◦ C/min and holding time of 5 h in all cases. 2.5. Characterization of sintered materials

2.5.3. Evaluation of apparent density and mechanical properties The modulus of rupture (MOR) and the Young’s modulus (E) of the materials sintered at 1300 ◦ C were determined from the slope of the elastic region of the stress–strain curves obtained from four-point flexural strength tests carried out according to the ASTM C 1161-02c standard,48 using a MTS QTEST/100 mechanical testing machine with a 500 N load capacity, employing a rate of load application (crosshead speed) of 0.05 mm/s until fracture of the specimens. For these tests, prismatic bars with dimensions of 45 mm (length) × 3 mm (width) × 4 mm (height) were employed, which were cut using a StruersAccutom 5 diamond saw and subsequently polished in the cut surfaces and borders prior to the mechanical tests, in order to avoid possible surface damages. The MOR and the E values of the prismatic bars were calculated using Eqs. (4) and (5),49 respectively: MOR = σ = E=

2.5.1. Phase composition The crystalline phases formed in the sintered samples were identified by XRD, using the same equipment and conditions mentioned before. The Celsian and Hexacelsian phases were quantified with the aid of the XPowder 2004.04 software, which uses non-linear least squares methods and weighting achieved by means of the normalized RIR method described by Chung.46 The Hexacelsian to Celsian conversion fraction (f, %) was calculated using the weight percentages determined by XRD for both phases, together with the following expression: % Celsian f = × 100 % Celsian + % Hexacelsian

3289

3PL 4bd 2

11PL3 64bd 3 D

(4) (5)

where P is the applied load (N), L is the distance (mm) between support points, b is the sample width (mm), d is the sample height (mm), and D is the sample deflection (mm). The apparent density (g/cm3 ) of the materials was determined by the Archimedes’ principle in distilled water, according to the ASTM C 20-00 standard.50 3. Results and discussion 3.1. Characterization of raw materials

(3)

A reaction mechanism was proposed from the obtained XRD and DTG results. Whenever possible, in order to back up the proposed reaction mechanism, free energy computations were performed for the considered reactions using the HSC Chemistry (version 6) software. 2.5.2. Microstructure The microstructure of the sintered samples was analyzed by Scanning Electron Microscopy (SEM) using a Philips XL30

The chemical compositions of the as-received and of the beneficiated CFA are given in Table 1. As it can be seen, both materials have aluminosilicate compositions enriched with Fe and Ca, with moderate loss on ignition (LOI) and limited amounts of MgO, TiO2 , Na2 O, K2 O and of other oxides. Since both materials contain more than 70 wt.% of SiO2 + Al2 O3 + Fe2 O3 as well as a relatively low CaO content, thus, it is determined that the used CFA belongs to class F, according to the ASTM C618-03 standard.51 After the magnetic separation process, the SiO2 /Al2 O3 mass ratio increased

3290

C.M. López-Badillo et al. / Journal of the European Ceramic Society 33 (2013) 3287–3300

Table 1 Chemical composition of CFA (wt.%). CFA

SiO2

Al2 O3

Fe2 O3

CaO

MgO

TiO2

Na2 O

K2 O

Other oxidesa

LOIb

ARc BCd

59.04 61.77

28.58 25.31

5.56 2.94

2.13 3.15

0.78 0.63

0.77 1.04

0.24 0.42

0.21 1.4

0.5 0.33

2.19 3.01

a b c d

MnO2 , P2 O5 , ZrO2 , PbO and SrO. Loss on ignition. As-received CFA. Beneficiated CFA.

from 2.10 for the initial CFA to 2.44 for the beneficiated CFA, with a simultaneous reduction of ∼53 wt.% in the content of Fe2 O3 . The loss on ignition was attributed basically to the oxidation at high temperature of free unburned carbon present in the material. The as-received CFA, as observed on the SEM, was composed mainly by spherical particles with varying sizes, which were either hollow (cenospheres) or had its interior filled with a large number of smaller spherical particles (plenospheres). These particles had high SiO2 and Al2 O3 content, according to the SEM/EDS analyses carried out. A small amount of Ferich and rough spherical particles, which contained also SiO2 and Al2 O3 , and which could correspond to magnetite, as well as a relatively small number of irregular or spherical porous particles of unburned carbon, were also found. According to Long-González et al.13 and Medina et al.,52 this kind of CFA contains ∼75–78.3 wt.% of amorphous phase. Semiquantitative XRD analyses showed that the crystalline portion of this material contained (in wt.%): 5.2% quartz, 14.6%mullite, and 1.9% magnetite. The XRD analysis, Fig. 1, revealed the presence of quartz and mullite, as well as the presence of a large amount of an amorphous phase, in the beneficiated CFA. The XRD patterns of Al2 O3 and BaCO3 , Fig. 1, allowed us to verify that no significant impurities were present in both powders.

3.2. Effect of mechanical activation on the physicochemical properties of the precursor mixtures

Fig. 1. XRD patterns obtained for BaCO3 , alumina, and beneficiated CFA. Key: ( ) BaCO3 , (␣) Al2 O3 , (Q) quartz, and ( ) mullite.

Fig. 2. Specific surface area and mean particle size obtained for the precursor mixtures as a function of milling time.

Once the precursor mixtures were milled, their specific surface area (SSA) and mean particle size were determined and plotted against milling time, Fig. 2. As it can be seen, the SSA increased from a value of 3.64 m2 /g for the reference mixture to a value of 7.1 m2 /g for the material milled for 8 h. However, when milling time was increased to 12 h, the specific surface area decreased to 5.4 m2 /g, which was associated with an agglomeration of the particles, which occurred due to a tendency of the mechanically activated materials to reduce their elevated surface free energy.53 This is consistent with the decrease observed in the particle size during the first 8 h of milling, as well as with the increment noted in it after the subsequent 4 h of mechanical activation. The XRD patterns of the reference and mechanically activated precursor mixtures are shown in Fig. 3. The identified peaks belong to BaCO3 (Witherite, JCPDS card no. 71-2394), ␣-Al2 O3 (Corundum, JCPDS card no. 46-1212), Al6 Si2 O13 (Mullite, JCPDS card no. 15-0776), and SiO2 (␣-quartz, JCPDS card no. 33-1161) crystalline phases. It can be noticed that even after 12 h of mechanical activation the formation of new phases is not detected in the mixtures. However, comparing the reference mixture with the mechanically activated ones, it can be observed that with increasing milling time the XRD patterns show a reduction in the peak intensities as well as a line broadening, which are caused by a refinement of crystallite size.54 It is well-known55 that when milling time is increased the mixture can undergo a partial amorphization, as well as an increase

C.M. López-Badillo et al. / Journal of the European Ceramic Society 33 (2013) 3287–3300

Fig. 3. XRD patterns obtained for reference and mechanically activated precursor mixtures. Key: ( ) BaCO3 , (␣) Al2 O3 , (Q) quartz, and ( ) mullite.

3291

Fig. 5. Amorphization degree (A) of BaCO3 contained in the precursor mixtures as a function of milling time.

3.3. Thermal analysis of the precursor mixtures in the concentration of defects and in the degree of structural disorder.53 The average crystallite size estimated using Scherrer’s equation is shown in Fig. 4 as a function of milling time. For this, the (1 1 1) reflection of BaCO3 contained in the mixtures was used. The results showed that the average crystallite size of BaCO3 was reduced rapidly and significantly since the very beginning of the mechanical activation process, decreasing from ∼106 nm to ∼66 nm after 12 h of milling. These results indicate that as the milling time increased, the crystallite size of BaCO3 decreased and this phase showed a tendency to undergo a gradual amorphization during the milling process.56 The degree of amorphization of BaCO3 contained in the mixtures as a function of milling time is depicted in Fig. 5. As it can be seen, the degree of amorphization increased with increasing time of mechanical activation, attaining a value of ∼70% after 8 h of milling, with no significant variation in it during the next 4 h.

3.3.1. Thermogravimetric analysis (TGA) The TGA curves obtained for the milled and non-milled precursor mixtures, Fig. 6, revealed that all samples started losing weight since the very beginning of the heating stage, which was probably due to the release at low temperatures of water adsorbed on the surface of the particles. Probably the relative importance of this contribution to the total weight loss of the samples increased with increasing milling time, since it was observed that the total weight loss increased from a value of 11.47% for the reference material to a value of 13.11% for the mixture milled for 12 h. It was also observed that the temperature at the end of the weight loss curve, Tf , was shifted towards a lower value with increasing milling time. In this way, Tf decreased from a value of ∼1054 ◦ C for the reference material to a value of ∼906 ◦ C for the mixture milled for 12 h.Since the weight losses detected at high temperatures were basically caused by the release of CO2 during the solid state reactions occurring between BaCO3 and the other components of the mixture, as well as by the thermal decomposition of the remnant BaCO3 , the obtained results mean that all

Fig. 4. Crystallite size of BaCO3 contained in the precursor mixtures as a function of milling time.

Fig. 6. TGA curves obtained for precursor mixtures milled for: (a) 0 h, (b) 4 h, (c) 8 h, and (d) 12 h. Tf = temperature at the end of the weight loss curve.

3292

C.M. López-Badillo et al. / Journal of the European Ceramic Society 33 (2013) 3287–3300

Fig. 7. DTG curves obtained for precursor mixtures milled for: (a) 0 h, (b) 4 h, (c) 8 h, and (d) 12 h. Tf , temperature at the end of the weight loss curve. Indicated peaks are explained in the text. Fig. 8. DTA curves obtained for precursor mixtures milled for: (a) 0 h, (b) 4 h, (c) 8 h, and (d) 12 h. Indicated peaks are explained in the text.

of these reactions took place at lower temperatures, and at faster rates, in the mechanically activated materials, in comparison with the reference precursor mixture. It is well known57,58 that mechanical activation of BaCO3 decreases both the temperature and the activation energy required for the thermal decomposition of this carbonate, increasing at the same time its reactivity at high temperature toward other phases mixed with it. This could be attributed to partial amorphization of the BaCO3 , caused by the high-energy milling, as well as to the high degree of homogeneity, particle and crystallite size refinement and increased SSA achieved in the precursor mixtures during the milling process. Although the mean particle size tended to increase and the SSA tended to decrease in the precursor mixtures milled for more than 8 h, the net effect of all the mentioned factors was an increased reactivity of the materials with increasing milling time. All the TGA curves also showed a change in slope at ∼680 ◦ C, which became less significant with increasing milling time. 3.3.2. Derivative thermogravimetric analysis (DTG) Fig. 7 shows the DTG curves obtained for the reference and the mechanically activated precursor mixtures. In the reference sample, the highest weight losses occurred from ∼450 ◦ C, with the presence of a first peak located at ∼611–640 ◦ C, peak 1, and a second peak with maximum point located at ∼676 ◦ C, peak 2. According to the DTG analysis of pure CFA, the main weight losses of this material take place at the temperature interval of ∼450–765 ◦ C, with a maximum point located at ∼661–679 ◦ C, which can be attributed to the burning of free carbon present in this material. Since the maximum point of the latter peak coincides with that of peak 2 observed in the DTG curve obtained for the reference precursor mixture, this could be attributed, at least in part, to the same cause. In the case of the reference precursor mixture, there is evidence that an incipient formation of barium silicates, as a result of the chemical reaction of BaCO3 with CFA, also takes place in the temperature interval at which peaks 1 and 2 were observed. The formation of this kind of compounds has been reported above ∼500 ◦ C in non-mechanically

activated BaCO3 –SiO2 mixtures with a Ba/Si molar ratio of 5:1, with a maximum point of the corresponding DTG curve located at 687 ◦ C.59 Allameh et al.26 also detected the formation of BaSiO3 and Ba2 SiO4 at 650 ◦ C, using Ba–Al–Al2 O3 –SiO2 solid precursors. In the same DTG curve obtained for the reference precursor mixture, peaks 3 and 4, with maximum points located at ∼806 ◦ C and ∼1000 ◦ C, respectively, are attributed to the reaction of BaCO3 with other components of the mixture, which gives rise to the formation of additional amounts of barium silicates, barium aluminate, and Hexacelsian. This is based on the results obtained from the XRD analysis of samples of the same material treated at 900 ◦ C and 1000 ◦ C. On the other hand, peak 5, with a maximum point located at ∼1037.7 ◦ C, could be attributed to the thermal decomposition of unreacted BaCO3 remaining in the samples, which gives rise to the formation of CO2 and BaO. Since the presence of BaO was never detected in the XRD patterns obtained for the heat-treated samples, thus, it was deduced that as soon as it was formed, this oxide reacted with Al2 O3 to form BaAl2 O4 , which was found even in the samples treated at 1100 ◦ C. However, the formation of additional amounts of barium silicates and/or Hexacelsian by reaction of BaO with other components of the mixture cannot be completely ruled out. These results are consistent with those reported by Maitra and Foger,59 who obtained a DTG curve for a non-mechanically activated BaCO3 –Al2 O3 mixture with a Ba/Al molar ratio of 1:1, in which the presence of a peak located at ∼1060 ◦ C was observed. This peak was attributed to the thermal decomposition of unreacted BaCO3 remaining in the mixture at that temperature. In the case of the mechanically activated mixtures, we observed that peaks 3 and 4 found in the DTG curves were shifted toward lower temperatures, with peak 3 becoming more significant while peak 5 completely disappeared, with increasing milling time. For the reference material, the maximum point of peak 4 was located at ∼1000 ◦ C, while for the mixture mechanically activated for 12 h the corresponding point was located

C.M. López-Badillo et al. / Journal of the European Ceramic Society 33 (2013) 3287–3300

3293

at 886 ◦ C. This could be explained by the increased reactivity achieved by the mechanically activated materials, when compared with the reference mixture. In accordance with the discussion presented in Section 3.3.1, the solid state reactions that occurred at temperatures lower than ∼1000 ◦ C between BaCO3 and the other components of the mixtures, took place at faster rates in the case of the mechanically activated materials, and this completely consumed all the BaCO3 contained in such mixtures. Thus, peak 5 completely disappeared from the corresponding DTG curves since in these cases there was no BaCO3 left that could be thermally decomposed at higher temperatures. In fact, it was observed by XRD that the amount of BaAl2 O4 formed in the mechanically activated samples decreased with increasing milling time and sintering temperature. It is worth mentioning that the corresponding DTG curve showed that the thermal decomposition of non-mechanically activated and pure BaCO3 took place in the temperature range of ∼846–1340 ◦ C, with a maximum point located at ∼1137 ◦ C. However, in the case of our reference material the evolution of CO2 started at much lower temperatures by direct reaction of BaCO3 with other components of the mixtures. This is consistent with the results reported by Boˇskovi´c et al.,41 Maitra and Foger,59 Strobel et al.,60 and Moya Corral and García Verduch,61 among others. However, for the mechanically activated materials, both the direct reaction of BaCO3 with the other components of the mixtures and the thermal decomposition of this carbonate occurred at even lower temperatures due to an enhanced reactivity of the mixtures.

3.3.3. Differential thermal analysis (DTA) The DTA curves obtained for the reference and mechanically activated mixtures are shown in Fig. 8 Endothermic peak 1, located in the temperature range of 669–677 ◦ C, was probably due to the formation of barium silicates, since it was very close to peak 2 observed in Fig. 7. Endothermic peak 2, which was located at 807–814 ◦ C, was caused by the occurrence of the orthorhombic to hexagonal phase transition of BaCO3 .13,59,62–64 It was observed that peak 2 was shifted toward lower temperatures and that its intensity decreased with increasing milling time. Probably, in comparison with the reference material, a smaller amount of unreacted BaCO3 remained in the mechanically activated mixtures on reaching the temperature of occurrence of the referred phase transition. Thus, the intensity of peak 2 decreased for the latter materials because they contained a smaller amount of available BaCO3 that could undergo the mentioned phase transition. Lastly, exothermic peak 3, which was located at ∼906–1054 ◦ C, probably corresponded to the temperature at which the evolution of CO2 from the precursor mixtures ended. It was observed that this temperature decreased with increasing milling time, coinciding with the values and trend showed by Tf determined from the TGA and DTG curves obtained for the reference and mechanically activated materials, as it can be seen in Table 2.

Fig. 9. XRD patterns obtained for precursor mixtures milled for: (a) 0 h, (b) 4 h, (c) 8 h, and (d) 12 h, and then sintered at: (1) 900 ◦ C, (2) 1000 ◦ C, and (3) 1300 ◦ C. Key: (M) Celsian, (H) Hexacelsian, () BaAl2 O4 , (Q) quartz, (␣) Al2 O3 , ( ) BaCO3 , () BaSiO3 , () BaSi2 O5 , and (䊉) Ba2 SiO4 .

3.4. Phase evolution during sintering of the precursor mixtures 3.4.1. Effect of milling time Fig. 9 gives the XRD results obtained for the samples milled for 0–12 h and then sintered at 900 ◦ C, 1000 ◦ C or 1300 ◦ C,

3294

C.M. López-Badillo et al. / Journal of the European Ceramic Society 33 (2013) 3287–3300

Table 2 Temperature of occurrence of main thermal events (◦ C) detected in the milled and non-mechanically activated BaCO3 –CFA–Al2 O3 mixtures (refer to Figs. 6–8). Milling time (h)

0 4 8 12 a b

DTG

TGA

1

2

3

4

5

∼611–640 ∼604–623 ∼598–616 ∼596–604

675.9 675.7 676.4 668

806 806 806 806

1000 920.5 903.7 886

1037.7 – – –

Tf

a

∼1061 ∼962 ∼951 ∼909

DTA Tf

a

1054 959 946 906

1

2b

677 677 674 669

814 811 811 807

3 ∼1054 ∼959 ∼945 ∼906

Tf , temperature at the end of the weight loss curve. Orthorhombic to hexagonal BaCO3 phase transition.

with holding time of 5 h in all cases. Table 3 summarizes the results obtained for all milling times and sintering temperatures employed. It can be seen that at a given sintering temperature, the solid state reaction process became more complete with increasing milling time, which was probably due to an enhanced reaction kinetics achieved under these conditions. This was evidenced by the formation of a larger amount of BaAl2 Si2 O8 , together with the presence of smaller amounts of remnant components of the precursor mixture as well as smaller amounts of transient phases (barium aluminate and barium silicates) formed during sintering, with increasing milling time. In this sense, it is noticeably that practically pure BAS (Hexacelsian) was obtained at 900 ◦ C for 12 h of milling of the precursor mixture. 3.4.2. Effect of sintering temperature For a given milling time, with increasing sintering temperature the reaction kinetics increased, since under these conditions there was a decrement in the amount and number of unreacted components of the precursor mixture, as well as of the formed transient phases (barium aluminate and barium silicates). It was also observed that from a sintering temperature of 1000 ◦ C and a milling time of 4 h, the Hexacelsian to Celsian conversion increased with increasing sintering temperature, although full conversion was never attained in the sintered samples.

after 12 h of milling the smallest crystallite size was achieved by the milled materials (Fig. 4). This could have increased the number of sites available for the surface nucleation of Hexacelsian, and thus, the value of f decreased under these conditions. It is important to mention that while Boˇskovi´c et al.41 obtained practically pure Hexacelsian at 1200 ◦ C/1 h, either with 2 or 4 h of milling of the precursor mixture, we were able to obtain a Hexacelsian to Celsian conversion of 82.2% for the case of the non-mechanically activated materials sintered at 1200 ◦ C/5 h, while with 8 h of milling of the precursor mixture and sintering at the same temperature, a f value of 86.3% was achieved by us. The maximum f value obtained in the present work was 94.2%, which corresponded to the non-mechanically activated material sintered al 1300 ◦ C/5 h. Thus, although in our case a full Hexacelsian to Celsian conversion was never attained in the mechanically activated materials, probably due to the promotion of the surface nucleation of Hexacelsian by the milling process, we were able to obtain a higher Hexacelsian to Celsian conversion than Boˇskovi´c et al.,41 which was more accentuated with increasing sintering temperature, even for the samples milled for the longest milling time employed by us (12 h). This could be attributed mainly to our use of CFA as raw material, which contains important amounts of Na2 O, CaO, TiO2 , and MgO, all of which could act as mineralizers that promote the Hexacelsian to Celsian conversion.14

3.5. Hexacelsian to Celsian conversion According to Fig. 10, the occurrence of the Hexacelsian to Celsian conversion was observed only in the temperature range of 1000–1300 ◦ C, in which the conversion fraction, f, increased with increasing sintering temperature. It can be noticed that from 1000 ◦ C to 1200 ◦ C, f achieved a maximum value at a milling time of 8 h. However, in all these cases f decreased with 12 h of milling. It is also worth mentioning that 1100 or 1200 ◦ C made no significant difference in the values of f achieved by any of the mechanically activated materials. At 1300 ◦ C, a steady decrease was observed in the value of f with increasing milling time. The observation of a maximum value of f occurring for 8 h of milling in the temperature range of 1000–1200 ◦ C, could be related to the fact that at this milling time we obtained the highest surface area and the smallest mean particle size in the mechanically activated precursor mixture (Fig. 2). This could have enhanced the kinetics of formation of Celsian. In contrast,

Fig. 10. Hexacelsian to Celsian conversion (f) obtained in the sintered mixtures as a function of milling time and sintering temperature.

Table 3 Summary of results obtained from the XRD analyses carried out for the samples milled for several times and subsequently sintered at several temperatures. Sintering temperature (◦ C)

Milling time (h) 4

900

Main phases: BaCO3 , BaAl2 Si2 O8 (Hexacelsian), BaAl2 O4 ; Minor phases: Al2 O3 , SiO2 (quartz), BaSiO3 , BaSi2 O5 , Ba2 SiO4

Main phases: BaAl2 Si2 O8 (Hexacelsian), BaCO3 ; Minor phases: BaAl2 O4 ,Al2 O3 , SiO2 (quartz), BaSi2 O5

1000

Main phases: BaAl2 Si2 O8 (Hexacelsian), BaAl2 O4 ; Minor phases: BaSi2 O5 , Ba2 SiO4 , SiO2 (quartz), BaSiO3 ,

Main phases: BaAl2 Si2 O8 (Hexacelsian) Trace phases: SiO2 (quartz), BaAl2 O4 , BaSi2 O5

1100

Main phases: BaAl2 Si2 O8 (Hexacelsian), BaAl2 Si2 O8 (Celsian), BaAl2 O4

Main phases: BaAl2 Si2 O8 (Celsian), BaAl2 Si2 O8 (Hexacelsian)

1200

Main phases: BaAl2 Si2 O8 (Celsian), BaAl2 Si2 O8 (Hexacelsian), BaAl2 O4

Main phases: BaAl2 Si2 O8 (Celsian) BaAl2 Si2 O8 (Hexacelsian)

1300

Main phases: BaAl2 Si2 O8 (Celsian); Minor phases: BaAl2 O4 , BaAl2 Si2 O8 (Hexacelsian)

Main phases: BaAl2 Si2 O8 (Celsian); Minor phases: BaAl2 Si2 O8 (Hexacelsian), SiO2 (quartz)

8

12 Main phases: BaAl2 Si2 O8 (Hexacelsian); Minor phases: BaSi2 O5 Trace phases: Al2 O3

Main phases: BaAl2 Si2 O8 (Hexacelsian); Minor phases: BaAl2 Si2 O8 (Celsian)

Main phases: BaAl2 Si2 O8 (Celsian) BaAl2 Si2 O8 (Hexacelsian); Minor phases: SiO2 (quartz)

C.M. López-Badillo et al. / Journal of the European Ceramic Society 33 (2013) 3287–3300

0

Note: main phases are written in the order of decreasing relative proportions in the sample.

3295

3296

C.M. López-Badillo et al. / Journal of the European Ceramic Society 33 (2013) 3287–3300

3.6. Microstructure The samples sintered at temperatures of 900 ◦ C and 1000 ◦ C lacked sufficient mechanical strength, thus, they could not be prepared for microscopy observations without fracturing them. In general, at 1100 ◦ C, at all milling times employed, a poorly densified microstructure was observed on the SEM (Fig. 11), with the presence of a considerable amount of large pores in it. Partly reacted particles could be observed in the nonmechanically activated material. With increasing milling time, particle size decreased and the sample’s matrix became constituted by needle-like BAS (Hexacelsian + Celsian) grains. In general, grain size was difficult to evaluate in the nonmechanically activated samples due to extensive agglomeration of the grains that took place at the beginning of the densification process. This was more pronounced in the materials sintered at 1100 ◦ C, for which neither the grain size nor the aspect ratio (length/width) of the grains could be evaluated. In the samples sintered at 1300 ◦ C, at all milling times employed, a higher degree of densification was observed, with the presence of needle-like Hexacelsian and Celsian grains in their microstructure, and with an evident diminution in the size and amount of porosity with increasing milling time. In the nonmechanically activated samples sintered at this temperature, the aspect ratio of the needle-like BAS grains varied in the range of 1.10–5.65. On the other hand, in the case of the material milled for 8 h, the aspect ratio and the mean size of the BAS grains were 4.6 and ∼15 ␮m, respectively, while in the material milled for 12 h the corresponding values were 4.7 and ∼11 ␮m. Thus, although the grain size decreased in the materials sintered at 1300 ◦ C with increasing milling time, the aspect ratio of the grains was relatively unaffected by the mechanical activation process. These results suggest that an increased milling time favored the development of a finer microstructure in the sintered materials probably due to the formation of a larger amount of BAS nuclei in them, which in turn was associated with the formation of a larger number of smaller BAS grains. This is consistent with the enhancement of the nucleation process of crystalline phases by mechanical activation previously observed by other researchers.65 These results also suggest that mechanical activation could be a way to control the grain size of the sintered BAS, in such a way that a small enough grain size is achieved in the materials so as to ensure that good physical and mechanical properties are obtained in them, which is a critical condition that must be fulfilled if they are to be used, for instance, for applications as refractory glass–ceramic and high temperature engine components.5

BaSi2 O5 and Ba2 SiO4 ), as well as a large amount of remnant components of the precursor mixture [BaCO3 , SiO2 (quartz) and Al2 O3 ], were detected in the sintered samples, additionally to the presence of a relatively small amount of Hexacelsian. Considering the phase composition of the precursor mixture and the reaction products found in the sintered samples (see Sections 3.2 and 3.4), it is thought that the chemical reactions likely occurring up to 900 ◦ C, without mechanical activation of the precursor mixture, were reactions (1) to (8) given below. 2BaCO3 + SiO2 = Ba2 SiO4 + 2CO2(g) 900◦ C

G



= −24.96 kcal

(1)

BaCO3 + 2SiO2 = BaSi2 O5 + CO2(g) ΔG900

◦C

= −23.454 kcal

(2)

BaCO3 + SiO2 = BaSiO3 + CO2(g) ΔG900

◦C

= −16.876 kcal

(3)

BaO + Al2 O3 = BaAl2 O4 ΔG700

◦C

= −3.990 kcal

(4)

Al6 Si2 O13 + 3BaCO3 + 4SiO2 = 3BaAl2 Si2 O8 + 3CO2(g) ΔG900

◦C

= −102.529 kcal

(5)

[xSiO2 ·yAl2 O3 ]amorphous + (x/2 − y)Al2 O3 + (x/2)BaCO3 = (x/2)BaAl2 Si2 O8 + (x/2)CO2(g)

(6)

[xSiO2 ·yAl2 O3 ]amorphous + (2y − x)SiO2 + yBaCO3 = yBaAl2 Si2 O8 + yCO2(g)

(7)

3.7. Reaction mechanism Since in this work it was intended to obtain phase-pure BaAl2 Si2 O8 , preferably as Celsian, in the sintered samples, it can be said that the solid state reaction process was far from being completed when a precursor mixture without mechanical activation was heat treated at 900 ◦ C/5 h. This is based on the observation that under these conditions the formation of a considerable amount of transient phases (BaAl2 O4 , BaSiO3 ,

[xSiO2 ·yAl2 O3 ]amorphous + yBaCO3 = yBaAl2 Si2 O8 + yCO2(g)

(8)

The formation of BaAl2 O4 by the occurrence of a chemical reaction between BaCO3 and Al2 O3 was not thermodynamically feasible at temperatures of up to 900 ◦ C, reaction (9). Thus, it

C.M. López-Badillo et al. / Journal of the European Ceramic Society 33 (2013) 3287–3300

was hypothesized that BaAl2 O4 was formed in some places of the samples which were locally enriched in Al2 O3 and BaO, reaction (4), due to an imperfect mixing of the initial powders. However, the latter reaction is thermodynamically feasible only at temperatures ≤700 ◦ C. BaCO3 + Al2 O3 = BaAl2 O4 + CO2(g) G900

◦C

= 24.582 kcal

(9)

In reactions (6) to (8), [xSiO2 ·yAl2 O3 ]amorphous represent the silicoaluminous glassy phase contained in the CFA. The Gibbs energy of these reactions could not be calculated due to a lack of pertinent data for the glassy phase. The chemical reaction, or reactions, actually happening among the latter depends on the chemical composition of the glassy phase, which was not determined by us. When the precursor mixture that was milled either for 4 or 8 h was sintered at 900 ◦ C/5 h, the formation of a larger amount of Hexacelsian as well as smaller amounts of BaAl2 O4 and BaSi2 O5 , were observed in the samples, in comparison with the

3297

non-milled material sintered under the same conditions. It was also observed that the milled and sintered materials contained smaller amounts of unreacted BaCO3 , quartz and Al2 O3 . All of this, together with the fact that in the latter case the formation of transient phases such as BaSiO3 and Ba2 SiO4 was not detected, indicated that, after sintering, reactions (1) to (8) were more complete in the milled samples than in the non-mechanically activated material. However, it was also very likely that chemical reactions such as reactions (10) to (17) were responsible for the diminution observed in the proportion of transient phases in the milled and sintered samples. Ba2 SiO4 + 3SiO2 = 2BaSi2 O5 G900

◦C

= −21.94 kcal

(10)

Ba2 SiO4 + SiO2 = 2BaSiO3 ΔG900

◦C

= −8.788 kcal

(11)

Fig. 11. SEM micrographs obtained for precursor mixtures milled for 0 h (a) and (d), 8 h (b) and (e), and 12 h (c) and (f), and then sintered at 1100 ◦ C (a)–(c), or 1300 ◦ C (d)–(f).

3298

C.M. López-Badillo et al. / Journal of the European Ceramic Society 33 (2013) 3287–3300

BaSiO3 + BaCO3 = Ba2 SiO4 + CO2(g) ΔG900

◦C

= −8.088 kcal

(12)

Al6 Si2 O13 + 2BaSiO3 = 2BaAl2 Si2 O8 + Al2 O3 ΔG

900◦ C

= −33.278 kcal

(13)

1.5(Al6 Si2 O13 ) + Ba2 SiO4 = 2BaAl2 Si2 O8 + 2.5Al2 O3 ΔG900

◦C

= −40.081 kcal

(14)

Al6 Si2 O13 + BaCO3 + 2BaSi2 O5 = 3BaAl2 Si2 O8 + CO2(g) ΔG900

◦C

= −55.621 kcal

(15)

BaSi2 O5 + Al2 O3 = BaAl2 Si2 O8 ΔG900

◦C

= −12.045 kcal

(16)

BaAl2 O4 + 2SiO2 = BaAl2 Si2 O8 ΔG900

◦C

= −60.201 kcal

(17)

When the precursor mixture that was milled for 12 h was sintered at 900 ◦ C/5 h, the solid state reaction process was almost completed. In this case, probably many of the previously mentioned chemical reactions took place, in such a way that the sintered sample was almost entirely composed by Hexacelsian, with an additional very small amount of remnant BaSi2 O5 and Al2 O3 . The main difference observed between the samples milled for different times and then sintered at temperatures higher than 900 ◦ C, consisted mainly in the degree of completeness achieved by the solid state reaction process, as well as in the degree of Hexacelsian to Celsian conversion reached by the samples under those conditions. When temperature was increased to 1000 ◦ C, without mechanical activation of the precursor mixture, it was observed that the reaction process was more complete, in comparison with the non-milled materials sintered at 900 ◦ C/5 h. At 1000 ◦ C/5 h, the formation of comparable proportions of Hexacelsian and BaAl2 O4 , as well as small amounts of BaSiO3 , BaSi2 O5 and Ba2 SiO4 , were detected. A very small amount of quartz was additionally found in the sintered samples. With 4 h of milling, practically only Hexacelsian was detected in the samples, with the additional presence in them of traces of BaAl2 O4 , BaSi2 O5 and quartz. Thus, under these conditions the solid state reaction process was nearly completed. Consequently, with either 8 or 12 h of milling, the latter process was already fully completed, with the presence in the samples of only BaAl2 Si2 O8 . Furthermore, the Hexacelsian to Celsian conversion had also already started.

Table 4 Apparent density, MOR and E values obtained for the samples sintered at 1300 ◦ C, with mechanical activation of the precursor mixtures for 0, 8, or 12 h. In all cases the standard deviation is shown. Milling time (h)

ρap (g/cm3 )

MOR (MPa)

E (GPa)

0 8 12

3.18 ± 0.09 3.32 ± 0.02 3.36 ± 0.09

15.16 ± 0.49 60.73 ± 7.32 83.42 ± 4.57

7.57 ± 0.95 30.56 ± 4.04 71.60 ± 1.67

At 1100 ◦ C/5 h, without mechanical activation the presence of Hexacelsian as predominant phase was observed, but with a considerable amount of remaining BaAl2 O4 still found in the samples. However, at this sintering temperature, the milling time of the precursor mixture made no difference with respect to the progress of the solid state reaction process, since this reached full completeness in all cases. Also in all cases, the Hexacelsian to Celsian conversion continued advancing during sintering of the samples. At 1200 ◦ C/5 h, without mechanical activation Celsian predominated, but with a small amount of Hexacelsian and BaAl2 O4 still present in the samples. At this temperature, the reaction process had already finished for all milling times and, again in all cases, the Hexacelsian to Celsian conversion continued advancing during sintering of the samples. However, for milling times of either 8 or 12 h, a small amount of quartz was also found in the sintered samples, which was probably due to precipitation of SiO2 which was in solid solution in the BaAl2 Si2 O8 , which could have happened during the occurrence of the Hexacelsian to Celsian conversion.6 At 1300 ◦ C/5 h, similarly to the observations made for the previous sintering temperature, without mechanical activation the presence of Celsian, BaAl2 O4 and Hexacelsian was detected in the sintered samples. However, in this case the amount formed of the latter two phases was considerably smaller. This indicated that now both the reaction process and the Hexacelsian to Celsian conversion had reached a higher stage of advance in the sintered materials. Again, at this temperature the reaction process had already finished for all milling times employed and the Hexacelsian to Celsian conversion continued its progress. Also for milling times of either 8 or 12 h, a small amount of quartz was detected in the sintered samples, probably due to the same reason already explained. The XRD results obtained and the reaction mechanism proposed in this work for the non-mechanically activated materials, coincided in general with those reported by Long-González et al.13 for similar raw materials and sintering temperatures. 3.8. Apparent density and mechanical properties The results of the apparent density and mechanical properties obtained for the materials sintered at 1300 ◦ C are shown in Table 4. As it can be seen, both the MOR and E values increased significantly with increasing milling time, achieving after 12 h of milling a maximum value of 83.42 MPa for the first property and of 71.60 GPa for the second one. These values are comparable to those reported by Ye et al.23 and by Johnsen,

C.M. López-Badillo et al. / Journal of the European Ceramic Society 33 (2013) 3287–3300

as quoted by Bansal,2 for BAS materials synthesized using the glass–ceramics and the solid state reaction routes, respectively, without mechanical activation. The first authors reported a MOR of ∼100 MPa, which was obtained from three-point flexural strength testing of their materials, while the second author reported values of MOR = 101 ± 16 MPa and E = 101 ± 7 GPa obtained from four-point flexural strength tests. Additionally, Sunget al.66 reported a MOR ∼87 MPa and E ∼69 GPa, both of them obtained from four-point flexural strength tests. These values are also comparable with the ones obtained by us in the present work. Similarly, the apparent density of our materials increased with increasing milling time, achieving a maximum value of 3.36 g/cm3 after 12 h of milling. This value corresponds to 99% of the theoretical density of Celsian (3.39 g/cm3 , according to the JCPDS card no. 381450). This is due to the fact that the Celsian phase predominates in the samples sintered at 1300 ◦ C. 4. Conclusions The use of CFA as a raw material greatly favored the kinetics of formation of Celsian by a solid state reaction process, especially with 8 h of milling of the precursor mixture. However, with 12 h of milling, surface nucleation of Hexacelsian predominated, and as a result, the Hexacelsian to Celsian conversion (f) decreased in the materials. Values of f higher than those previously reported for similar mechanically activated BAS materials were obtained in the present work, which was probably due to a mineralizing effect of Na2 O, CaO, TiO2 and MgO impurities present in the used CFA. Thus, the latter phenomenon allowed us to overcome to a considerable extent the negative effect on the Hexacelsian to Celsian conversion caused by an extensive milling of the precursor mixture. Lastly, both the apparent density and the mechanical properties of the materials sintered at 1300 ◦ C increased with increasing milling time. The values obtained for these properties were comparable to those of BAS synthesized from high-purity raw materials, previously reported by other researchers. Acknowledgments The authors express their gratitude to CONACYT and Cinvestav-Saltillo for the financial support and facilities provided for the development of this work, as well as to the personnel of the “José López Portillo” power plant, Nava, Coahuila, Mexico, for supplying the employed CFA. Many thanks are also due to F. Botello-Rionda for his assistance with the thermal analyses. References 1. Semler CE, Foster WR. Studies in the system BaO–Al2 O3 –SiO2 : IV, the system celsian–silica–alumina. J Am Ceram Soc 1970;53:595–8. 2. Bansal NP. Solid state synthesis and properties of monoclinic celsian. J Mater Sci 1998;33:4711–5. 3. Martinez Caja L, Moya Corral JS, García Verduch A. Polimorfismo de la celsiana. Bol Soc Esp Ceram Vidr 1974;13:441–3. 4. Yoshiki B, Matsumoto K. High-temperature modification of barium feldespar. J Am Ceram Soc 1951;34:283–6.

3299

5. Beall GH. Refractory glass ceramics. US Patent 7867932; 2011. 6. Moya Corral JS, García Verduch A. Solid solution of silica in celsian. Trans J Brit Ceram Soc 1978;77:40–4. 7. Lee I, Covino J. Synthesis of monoclinic celsian. US Patent 6039929; 2000. 8. Hoghooghi B, McKittrick J, Helsel E, Lopez OA. Microstructural development, densification and hot-pressing of celsian ceramics from ion-exchanged zeolite precursors. J Am Ceram Soc 1998;81:845–52. 9. Oobuchi T, Otagiri T, Ide Y. Process for production of ceramic porcelains, ceramic porcelains and electronic components. US Patent Application 20090048089 A1; 2009. 10. Colomban P, Courret H, Romain F, Gouadec G. Sol–gel-prepared pure and lithium-doped hexacelsian polymorphs: an infrared, Raman, and thermal expansion study of the ␤-phase stabilization by frozen short-range disorder. J Am Ceram Soc 2000;83:2974–82. 11. Bacher RJ, Nakayama T. Celsian containing dielectric crossover composition. US Patent 3837869; 1974. 12. Talmy IG, Haught DA. Method of preparation monoclinic BaO. Al2 O3 ·2SiO2 . US Patent 5695725; 1997. 13. Long-González D, López-Cuevas J, Gutierrez-Chavarría CA, Pena P, Baudin C, Turrillas X. Synthesis of monoclinic celsian from coal fly ash by using a one-step solid-state reaction process. Ceram Int 2010;36:661–72. 14. Lee K-T, Aswath PB. Role of mineralizers on the hexacelsian to celsian transformation in the barium aluminosilicate (BAS) system. Mater Sci Eng A 2003;352:1–7. 15. Lee K-T, Aswath PB. Synthesis of hexacelsian barium aluminosilicate by a solid-state process. J Am Ceram Soc 2000;83:2907–12. 16. Kobayashi Y. Transformation kinetics from hexacelsian to celsian for powders having uniform particle size. Ceram Int 2001;27:179–84. 17. Talmy IG, Haught DA. Low temperature synthesis of high purity monoclinic celsian using topaz. US Patent 4994419; 1991. 18. Eichler K, Solow G, Otschik P, Schaffrath W. BAS (BaO, Al2 O3 , SiO2 )glasses for high temperature applications. J Eur Ceram Soc 1999;19: 1101–4. 19. Bansal NP, Drummond III CH. Effect of isostatic pressing on phase formation in BaO·AI2 O3 ·2SiO2 glass. J Mater Sci Lett 1994;13:53–5. 20. Hyatt MJ, Bansal NP. Crystal growth kinetics in BaOAl2 O3 2SiO2 and SrOAl2 O3 2SiO2 glasses. J Mater Sci 1996;31:172–84. 21. Esposito S, Ferone C, Pansini M, Bonaccorsi L, Proverbio E. A comparative study of the thermal transformations of Ba-exchanged zeolites A, X and LSX. J Eur Ceram Soc 2004;24:2689–97. 22. McKittrick J, Hoghooghi B, Lopez OA. Vitrification and crystallization of barium aluminosilicate glass ceramics from zeolite precursors. J Non-Cryst Solids 1996;197:170–8. 23. Ye F, Gu JC, Zhou Y, Iwasa M. Synthesis of BaAl2 Si2 O8 glass–ceramic by a sol–gel method and the fabrication of SiCpl /BaAl2 Si2 O8 composites. J Eur Ceram Soc 2003;23:2203–9. 24. Lee I, Covino J. Sol–gel synthesis of monoclinic phase of barium aluminosilicate ceramics. Mater Res Bull 1994;29:55–62. 25. Lee WE, Chen M, James PF. Crystallization of celsian (BaAl2 Si2 O8 ) glass. J Am Ceram Soc 1995;78:2180–6. 26. Allameh SM, Sandhage KH. Synthesis of celsian (BaAl2 Si2 O8 ) from solid Ba–Al–Al2 O3 –SiO2 precursors: I. XRD and SEM/EDX analyses of phase evolution. J Am Ceram Soc 1997;80:3109–26. 27. Zhang X-D, Sandhage KH, Fraser HL. Synthesis of BaAl2 Si2 O8 from solid Ba–Al–Al2 O3 –SiO2 precursors: II. TEM analyses of phase evolution. J Am Ceram Soc 1998;81:2983–97. 28. Schmutzler HJ, Sandhage KH. Transformation of Ba–Al–Si precursors to celsian by high-temperature oxidation and annealing. Metall Mater Trans B 1995;26:135–50. 29. Colella C, Aiello R. Thermal conversion of synthetic cymrite into hexagonal barium feldspar. Thermochim Acta 1978;27:253–60. 30. Amritphale SS, Anshul A, Chandra N, Ramakrishnan N. Development of celsian ceramics from fly ash useful for X-ray radiation-shielding application. J Eur Ceram Soc 2007;27:4639–47. 31. López-Cuevas J, Long-González D, Gutiérrez-Chavarría CA. Effect of milling time on the physical and mechanical properties of celsian–mullite composites synthesized from coal fly ash. In: MRS Proceedings, vol. 1373. 2012. p. 43–52.

3300

C.M. López-Badillo et al. / Journal of the European Ceramic Society 33 (2013) 3287–3300

32. López-Cuevas J, Long-González D, Gutiérrez-Chavarría CA. Thermal behavior of celsian ceramics synthesized from coal fly ash. In: MRS Proceedings, vol. 1373. 2012. p. 19–24. 33. Seidel A, Sluszny A, Shelef G, Zimmels Y. Self inhibition of aluminum leaching from coal fly ash by sulfuric acid. Chem Eng J 1999;72:195–207. 34. Lee SH, Sakai E, Daimon M, Bang WK. Characterization of fly ash directly collected from electrostatic precipitator. Cem Concr Res 1999;29:1791–7. 35. Singer A, Berkgaut V. Cation exchange properties of hydrothermally treated coal fly ash. Environ Sci Technol 1995;29:1748–53. 36. Ahmaruzzaman M. A review on the utilization of fly ash. Prog Energy Combust Sci 2010;36:327–63. 37. Barnes DI. Novel products from combustion ash – legislative and marketing issues. J Chem Technol Biotechnol 2002;77:229–33. 38. Bahat D. Kinetic study on the hexacelsian–celsian phase transformation. J Mater Sci 1970;5:805–10. 39. Ferone C, Esposito S, Dell’Agli G, Pansini M. Role of Li in the low temperature synthesis of monoclinic celsian from (Ba, Li)-exchanged zeolite-A precursor. Solid State Sci 2005;7:1406–14. 40. Khater GA, Idris MH. Role of TiO2 and ZrO2 on crystallizing phases and microstructure in Li, Ba aluminosilicate glass. Ceram Int 2007;33:233–8. 41. Boˇskovi´c S, Kosanovi´c D, Bahloul-Hourlier D, Thomas P, Kiss SJ. Formation of celsian from mechanically activated BaCO3 –Al2 O3 –SiO2 mixtures. J Alloys Compd 1999;290:230–5. 42. López-Cuevas J, Long-González D, Gutiérrez-Chavarría CA, RodríguezGalicia JL, Pech-Canul MI. Alumina extraction from mexican fly ash. In: MRS Proceedings, vol. 1243. 2009. p. 91–6. 43. Ohlberg SM, Strickler DW. Determination of percent crystallinity of partly devitrified glass by X-ray diffraction. J Am Ceram Soc 1962;45:170–1. 44. Baláˇz P. Extractive Metallurgy of Activated Minerals. 1st ed. Amsterdam: Elsevier Science B.V.; 2000. 45. Cullity BD. Elements of X-Ray Diffraction. 2nd ed. Massachusetts: Addisson-Wesley; 1978. 46. Chung FH. Quantitative interpretation of X-ray diffraction patterns of mixtures, I. Matrix-flushing method for quantitative multicomponent analysis. J Appl Crystallogr 1974;7:519–25. 47. Gantenbein D, Schoelkopf J, Matthews GP, Gane PAC. Determining the size distribution-defined aspect ratio of rod-like particles. Appl Clay Sci 2011;53:538–43. 48. Standard test method for flexural strength of advanced ceramics at ambient temperature. ASTM C 1161-02c. 49. Standard test method for flexural modulus of elasticity of dimension stone. ASTM C 1352-96. 50. Standard test methods for apparent porosity, water absorption, apparent specific gravity, and bulk density of burned refractory brick and shapes by boiling water. ASTM C 20-00.

51. Standard specification for coal fly ash and raw or calcined natural pozzolan for use in concrete. ASTM C 618-03. 52. Medina A, Gamero P, Querol X, Moreno N, De León B, Almanza M, et al. Fly ash from a Mexican mineral coal I: mineralogical and chemical characterization. J Hazard Mater 2010;181:82–90. 53. Ghasri-Khouzani M, Meratian M, Panjepour M. Effect of mechanical activation on structure and thermal decomposition of aluminum sulfate. J Alloys Compd 2009;472:535–9. 54. Suryanarayana C. Mechanical alloying and milling. Prog Mater Sci 2001;46:1–184. 55. Küc¸ük F, Yildiz K. The decomposition kinetics of mechanically activated alunite ore in air atmosphere by thermogravimetry. Thermochim Acta 2006;448:107–10. 56. Pourghahramani P, Forssberg E. Microstructure characterization of mechanically activated hematite using XRD line broadening. Int J Miner Process 2006;79:106–19. 57. Criado JM, Diánez MJ, Morales J. Influence of the mechanical treatment on the structure and the thermal stability of alkaline-earth carbonates. J Mater Sci 2004;39:5189–93. 58. Chen G, Niu D. Mechanical activation of barium aluminate formation from BaCO3 –Al2 O3 mixtures. J Alloys Compd 2006;413: 319–22. 59. Maitra AM, Foger K. Identification of solid solutions and other phases in strontium and barium catalysts containing oxides of magnesium, aluminum or silicon as additive. II. Catalyst characterisation. Appl Catal A 1994;114:83–107. 60. Strobel R, Maciejewski M, Pratsinis SE, Baiker A. Unprecedented formation of metastable monoclinic BaCO3 nanoparticles. Thermochim Acta 2006;445:23–6. 61. Moya Corral JS, García Verduch A. Estudio de la reacción entre el caolín y el carbonato de bario. Bol Soc Esp Ceram Vidr 1976;15:379–81. 62. Arvanitidis I, Sichen D, Seetharaman S. A study of the thermal decomposition of BaCO3 . Metall Mater Trans B 1996;27:409–16. 63. Weinbruch S, Büttner H, Rosenhauer M. The orthorhombic-hexagonal phase transformation in the system BaCO3 –SrCO3 to pressures of 7000 bar. Phys Chem Miner 1992;19:289–97. 64. Earnest CM. Calorimetric heat of transition assignments by microcomputerbased differential thermal analysis: part II. The ␥–␤ (orthorhombic to hexagonal) transition of barium carbonate. Thermochim Acta 1989;137:365–71. 65. Wenbin F, Lianxi H, Wenxiong H, Erde W, Xiaoqing L. Microstructure and properties of a TiAl alloy prepared by mechanical milling and subsequent reactive sintering. Mater Sci Eng A 2005;403:186–90. 66. Sung Y-M, Kim S. Mechanical properties of off-stoichiometric BaO·Al2 O3 ·2SiO2 glass–ceramics. J Mater Sci Lett 1999;18:1657–60.