Evaluation of a Microsilica-Based Additives in Al O –Mgo Refractory

PAPER

Evaluation of a Microsilica-Based Additives in Al2O3–MgO Refractory Castables

T. dos Santos Jr., L. F. Krol, A. P. Luz, C. Pagliosa,V. C. Pandolfelli

The first heating cycle of MgO-containing refractory castables is usually considered a challenge by the producers of such products due to the significant mass loss, associated with magnesium hydroxide decomposition between 350–450 °C, that might result in their spalling or even explosion. In order to allow faster and safer drying of this sort of refractory system, a microsilica-based additive (SioxX-Mag) has been developed. Thus, this work focused on investigating the action of this commercial additive in high-alumina castables bonded with 6 mass-% of MgO and prepared with the addition of formic acid. Flowability, hot Young’s modulus, thermogravimetric analyses, cold and hot mechanical measurements were carried out to infer the properties of the compositions with and without the selected SiO2-based product. According to the collected results, the addition of 1 mass-% of the microsilica-containing additive to the designed castables resulted in higher flow and reduced MgO hydration. Consequently, even when subjecting these samples to a very high heating rate (20 °C · min–1), no explosion was detected. When increasing the firing temperature, the interaction among Al2O3–MgO–SiO2 induced the increase of the castables’ mechanical strength up to 900 °C, but liquid phase formation was identified in the tested compositions above 1000 °C, causing the drop of their mechanical properties. Nevertheless, thermodynamic calculations and experimental tests indicated that the formed liquid should progressively react with the castables’ components, giving rise to refractory phases such as mullite and forsterite. Hence, the evaluated microsilica-based additive can be a potential solution to adjust the drying behavior and prevent the explosion of MgO-containing castables.

Although eq. 1 is a simple reaction, the 1 Introduction control of brucite formation in the matrix T. dos Santos Jr., L. F. Krol, A. P. Luz, Besides its high melting point (2800 °C), fraction of MgO-containing castable com- V. C. Pandolfelli good compatibility with basic slags and positions during mixing, curing and/or dry- Federal University of São Carlos great likelihood of reacting with alumina ing stages is a difficult task, as various pa- Materials Engineering Department at high temperatures for spinel (MgAl2O4) rameters (i.e., temperature, water amount São Carlos, SP, 13565-90 generation, magnesium oxide (MgO) is and magnesia granulometry, surface area Brazil pointed out as a potential binder com- and volume) might affect the growing of pound, which can replace the traditional this hydrated phase [5–10]. Consequently, C. Pagliosa calcium aluminate cement, in refractory it is of utmost importance to optimize the RHI Magnesita, Research and Development castable systems comprising low CaO con- nucleation and growing rate of Mg(OH)2 Center tent [1–4]. This binder action is related to during the processing of such refractories, Contagem, MG, 32210-050 MgO hydration reaction with water and in order to inhibit cracking and flaws for- Brazil further brucite [Mg(OH)2] formation, as in- mation in their structure. Many studies re- dicated in eq. 1. Consequently, the brucite ported the use of different additives that Corresponding author: T. dos Santos Jr. crystals tend to fill in pores and voids found allow the adjustment of MgO-water reac- E-mail: [email protected] in the consolidated structure, increasing the tion, which can induce an effective incor- cohesion among the refractory constituents poration of MgO as binder in monolithic Keywords: MgO, hydration, brucite, and the overall mechanical strength level of products [4, 11–15]. Some authors also refractory, SioxX-Mag the samples during their curing and drying stated that, when this hydration reaction is stages. speed up during the early magnesia-water Received: 18.03.2019

contact, Mg(OH)2 crystals may be better Accepted: 01.04.2019 (eq. 1) accommodated in the resulting structure refractories WORLDFORUM 11 (2019) [3] 67 PAPER

as the castables will not be fully hardened silicates (M-S-H gels) with distinct stoichi- fraction. According to the supplier of this yet [3, 16]. ometry, depending on the MgO and SiO2 product [31–33], the addition of 2 mass-% Among the MgO hydrating agents, formic molar ratio (M/S). For instance, Kalousek of SioxX-Mag to MgO castables contain- and acetic acids proved to be good options et al. [26] reported that a gel phase with ing 6 mass-% of SiO2 resulted in improved to control brucite formation in high-alumina chrysotile stoichiometry, Mg3Si2O5(OH)4, flowability, mechanical strength and large compositions, leading to the development can be obtained when M/S is equal to 1,5. pieces were molded without showing of refractories with high green mechanical On the other hand, when M/S is around cracking associated with magnesia hydra- strength [3, 4, 15–17]. However, in general, 0,75, talc phase [Mg3Si4O10(OH)2] might tion. Additionally, the tested compositions the high hydration level of such magnesia- be formed. These amorphous compounds presented higher explosion resistance. bonded products results in a significant limit the hydration of MgO grains, as they Although SioxX-Mag have been designed spalling/explosion trend for the prepared coat the surface of this oxide inhibiting for magnesia-based castables, most likely samples, when they are subjected to their its interaction with water. Consequently, this additive can act minimizing the explo- first heating cycle. As observed in previous cracking and flaws formation in the consoli- sion trend of high-alumina compositions research [15], the detected explosions of dated castables can be minimized with the containing MgO in their matrix fraction. It MgO-bonded castables took place mainly in reduction of brucite content formed in the is also expected that this product may fa- the 350–450 °C temperature range, due to matrix fraction of the microstructure, and vor a better control of magnesia hydration the steam pressure derived from brucite de- the likelihood of further spalling/explosion and improve the castables’ flow and green composition and carboxylates’ desorption. event during the first thermal cycle of these mechanical strength levels. Thus, this work Thus, in order to minimize or prevent the materials is significantly decreased. Moreo- investigated the effect of adding SioxX- formation of cracks or even the explosion of ver, some studies [14, 25] highlight that it is Mag to high-alumina vibratable castables MgO-containing castables, likely solutions more appropriate to induce the generation containing 6 mass-% of fine MgO and pro- consist of inducing the in situ formation of of chrysotile-like gels (M/S ~1,5) in refrac- cessed with formic acid (hydrating agent). hydrotalcite-like phases [18], or incorporat- tory mixtures because the decomposition of ing polymeric fibers [3, 19, 20] or amor- these compounds above 800 °C can give 2 Materials and methods phous silica sources [21] into the designed rise to forsterite (Mg2SiO4). The latter phase Vibratable castables were designed based compositions. Such alternatives will play a helps to improve the castables’ thermome- on Alfred’s particle packing model and us- role by changing the resulting microstruc- chanical properties due to its high refracto- ing distribution modulus q = 0,26 [34]. ture and favoring easier steam withdrawal. riness (melting point ~ 1840 °C). Tab. 1 shows more information about the Silica fume and additives derived from this Refractory users and producers have been raw materials and water contents used raw material are pointed out as efficient constantly interested in finding alterna- during the preparation of such composi- options to control MgO hydration in re- tives to promote fast and save drying of tions. The main constituents of the formu- fractory systems [14, 22–24]. According to ceramic linings. In this sense, new addi- lations were tabular alumina (d ≤ 6 mm, the literature [14, 23, 25], microsilica can tives have been developed to enhance the Almatis/DE), calcined and reactive alumi- be partially dissolved in basic pH aque- permeability of castables, without affecting nas (CT3000SG and CL370, Almatis/DE) ous medium, giving rise to silicilic acid in a greater extent their green mechanical and dead-burnt MgO with d50 = 15 µm – + (HSiO3 · OH3 ). This compound might also strength [21, 27–30]. This is the case of (M30 < 212 µm, RHI Magnesita/BR). The react with brucite phase formed on MgO a silica-based drying agent (SioxX-Mag), castables processed with formic acid con- surface in a further moment, which gen- which was designed for magnesia mono- tained 0,34 mass-% of this additive. A total erates amorphous hydrated magnesium lithics presenting microsilica in their matrix of 1 mass-% of SioxX-Mag (35 mass-%

SiO2, 50 mass-% Al2O3, 15 mass-% oth- ers, Elkem/NO) was incorporated into the Tab. 1 General information of the castable compositions evaluated in this work mixtures prepared with and without formic A B C D acid and, as the microsilica-based product Remarks Only MgO MgO and MgO and MgO, Formic Acid already contained a dispersing agent in its Formic Acid SioxX-Mag and SioxX-Mag composition, no Castament® FS60 (BASF/ Components [mass-%] DE) addition was required processing casta- Tabular alumina 87 87 87 87 bles C and D. The reference compositions Fine aluminas 7 7 7 7 (A and B, without SioxX-Mag) were mixed (CL 370 C e with 4 mass-% of distilled water, whereas CT 3000 SG) the ones containing this microsilica-based M30 < 212 µm 6 6 6 6 additive (C and D) contained 4,5 mass-% Formic acid 0 0,34 0 0,34 of liquid. The compositions were mixed in a rheom- Castament® FS60 0,2 0,2 0 0 eter specially developed for refractory casta- SioxX-Mag 0 0 1,0 1,0 bles. Two processing routes were adopted Water 4 4 4,5 4,5 during the preparation of the mixtures with

68 refractories WORLDFORUM 11 (2019) [3] PAPER

and without formic acid. Distilled water was Tab. 2 Chemical composition of the castable’s matrix fraction directly added to the powdered components Composition B D (alumina, MgO, dispersant) of the acid-free Components [mass-%] compositions, whereas a suspension com- Al O 83,72 83,06 prised by distilled water, formic acid and 2 3

MgO was firstly prepared for the process- SiO2 0,13 1,21 ing of castables containing this hydrating MgO 16,15 15,73 agent. After that, this aqueous suspension Total 100 100 was mixed to the remaining solid fraction of the refractory. Castable samples were molded under vibration, cured at 50 °C for The phase transformations in the castables Moreover, the thermal shock resistance of 24 h and dried at 110 °C for 24 h. microstructure were monitored via in situ the castables were analyzed when they Vibratable flow measurements (ASTM Young’s modulus (E) measurements during were exposed to ΔT ~ 1000 °C. Bar sam- C1445) as a function of time (up to 90 min) drying at 110 °C or along various thermal ples (150 mm x 25 mm x 25 mm) were pre- were carried out after the castables’ mix- cycles up to 1400 °C. The bar resonance viously fired at 1450 °C for 5 h and, after ing step. A conical mold for casting the flow technique (ASTM E1875) was used for that, they were subjected to 10 heating- specimen (diameter of the bottom and top analyzing the stiffening evolution of bar cooling cycles. The damage induced by the opening was ~100 mm and 70 mm, respec- samples (150 mm x 25 mm x 25 mm) and thermal fluctuations was investigated by tively, and height was 50 mm) was filled further details of the applied method can be the evolution of the Young’s modulus of the with the fresh castable mixtures. The mold found elsewhere [36, 37]. prepared compositions. The E analyses were was carefully lifted away from the compo- Cold mechanical strength and apparent carried out according to the bar resonance sitions and the samples were vibrated for porosity of the designed compositions were technique and using the ScanElastic 02 de- 1 min. The castables’ spreading were meas- evaluated after curing at 50 °C·24 h–1, dry- vice (ATCP Engenharia Física/BR). Four sam- ured with a caliper and the final flow was ing at 110 °C·24 h–1 and firing at 600 °C, ples of each refractory were used in these calculated as follow: 900 °C or 1450 °C for 5 h. The Modulus of tests and the obtained results are presented Rupture (MOR, 3-point bending test) of a as the average percentual of the retained (eq. 2) total of 5 bar samples for each tested con- Young’s modulus. dition was determined according to ASTM Thermodynamic simulations were also per- where, D corresponds to the final diameter C133 [38], using a mechanical testing ma- formed in order to predict the likely phases of the sample after 1 min of vibration and chine (MTS 810, MTS/US) and based on that could be formed in the matrix fraction

D0 is the bottom diameter of the conical eq. 5. On the other hand, the apparent po- of the castables in the 600–1500 °C tem- mold (100 mm). The flow measurements rosity level of the prepared specimens was perature range and in an equilibrium condi- were carried out each 15 min after the analyzed as suggested by ASTM C830 [39], tion. These calculations were carried out us- end of the mixing step and up to a total by using kerosene as immersion liquid and ing FactSage™ software (Thermfact/CRTC/ of 90 min. calculating the results as indicated in eq. 6. CA and GTT-Technologies/DE), version 6.4, Cylindrical specimens (40 mm diameter and Fact53 and FTOxid databases. The and 40 mm height) were prepared for the (eq. 5) chemical composition of the matrix frac- thermogravimetric tests. These experiments tion of the designed castables are shown were carried out in a device developed for (eq. 6) in Tab. 2. analyzing the drying behavior of refractory compositions [35] and two heating rates, where, P is the maximum load applied at 3 Results and discussion 5 °C·min–1 and 20 °C·min–1, were employed rupture, L is the span between supports Initially, SioxX-Mag influence on the rheolo- during the evaluation of the designed MgO- (127 mm), and b and d are the width and gy of the designed compositions was evalu- containing castables. The mass loss (TG) depth of the specimen, respectively. Pu, Pi ated via vibratable flow measurements as a and drying rate (DTG) collected for the dried and Ps correspond to wet, suspended and function of time (Fig. 1). As observed, higher –1 (110 °C·24 h ) samples in the 50 °C– dry mass of the samples. For both tests a flow values could be detected for the SiO2- 600 °C temperature range were calculated total of 5 specimens for each evaluated containing castables (C and D) when com- according to eq. 3 and eq. 4. condition were analyzed, and the reported pared to the mixtures dispersed with Casta- values represent the average results with ment® FS 60 (A and B). Nevertheless, the (eq. 3) their standard deviation. latter presented few changes of the flow Hot Modulus of Rupture (HMOR) of bar level with time, whereas C and D showed (eq. 4) samples (similar to the ones used for the a major decrease of this property in the MOR measurements) pre-fired at the same analyzed conditions. When formic acid and where M0 is the initial mass, M is the meas- testing temperature for 5 h were deter- SioxX-Mag were used (castable D), a faster ured mass at time t and Mf is the final mass mined according to ASTM C583 [40] and flow decay was observed after 40 min of of the evaluated sample. using HBST 422 equipment (Netzsch/DE). the end of the mixing step, which suggests refractories WORLDFORUM 11 (2019) [3] 69

Initially, SioxX-Mag influence on the rheology of the designed compositions were evaluated via vibratable flow measurements as a function of time (Fig. 1). As observed, higher flow values could be detected for the SiO2-containing castables (CSuch and D)data when indicate compared that theto selected hydrating agent acted enhancing the cohesion of the the mixtures dispersed with Castament® FS 60 (A and B). Nevertheless,resulting the latter microstructure presented fewand, most likely, a more controlled brucite crystals’ growth took place changes of the flow level with time, whereas C and D showed a majorin the decrease evaluated of this conditions. property On the other hand, when SioxX-Mag was combined with formic in the analyzed conditions. When formic acid and SioxX-Mag wereacid used addition (castable in castableD), a faster D, the final Young’s modulus value measured after 24h was lower than flow decay was observedPAPER after 40 min of the end of the mixing step,the which one suggestsof B samples that a, likelywhich can be related to the limited MgO hydration in this system due to interaction between these two additives might have been responsiblethe for presence such behavior. of the silica-containing additive.

A B C D A B D 200 120

180 100 160

140 80 120

100 60

80 E (G P a)

40 V ibra flow (% ) 60

40 20 20

0 0 0 20 40 60 80 100 0 3 6 9 12 15 18 21 24 T ime (min) Time (hours)

Fig. 1 Vibra flow values of the prepared compositions Fig. 2 In situ Young’s modulus (E) evolution during the drying Fig. 1 Vibra flow values of the prepared compositions as a function of time as a function of time Fig. 2 In situ Young’sprocess at modulus 110 °C·24 (E) h–1 evolution of the samples during containing the drying formic process acid at 110°C/24 h of the previously cured at 50 °C·24 h–1 samples containing formic acid previously cured at 50°C/24 h Besides having the highest initial vibra flow, castable D was selected for further that a likely interaction between these two compared to A material (50 GPa). Such data rate profiles shown in Fig. 3b point out that additives might have been responsible for indicate that the selected hydrating agent samples’ mass loss took place mainly in two characterizations due to the presence of formic acid in its composition, which may result in such behavior. acted enhancingHence, ThetheSioxX cohesionthermograv-Mag ofmay theimetric resultplay results- twodifferent, obtainedimportant temperature when roles using inranges. 5°Calumina.min The-1 - firstbasedas heating one MgO rate,-containing indicated better brucite formationBesides having control the andhighest inhibit initial crack vibra generatioflow, ingn in microstructure the prepared and,samples most [15] likely,. As a more around 350 °C is associated with magnesi- thatcompositions, the microsilica as it- basedcan inhibit additive magnesia contained hydration in sample and limitD partially the interaction inhibit Mg(OH) of this oxide formation with castable D was selected for further charac- controlled brucite crystals’ growth took um hydroxide decomposition, giving rise to 2 castable C did not contain formic acid, this composition was not further characterized. terizations due to the presence of formic acid place induring the hydratingevaluated the previous agent conditions. (formiccuring On and acid). the drying water steps vapor (Fig. as3). reaction The drying product. rate Whereas,profiles shown in Fig. 3b The effectin itsof composition, formic acid which in the may development result in better of theother green hand, mechanical when SioxX-Mag strength was of combinedthe the second one (between 420–450 °C) is brucite formation control and inhibit crack with formicpoint acid out addition that samples’ in castable mass D, lossthe tookrelated place tomainly the carboxylates’in two different desorption, temperature ranges. The refractories duringgeneration drying in at the 110°C prepared is presented samples [15].in Fig. As 2. Accordingfinal Young’s to themodulus Young’s value modulusmeasured after that should be responsible for releasing first one around 350°C100,0 is associated with magnesium hydroxide decomposition, giving rise to castable C did not contain formic acid, this 24 h was lower than the one of B samples, some gas phases B to the furnace D environ(a) - evolution of the cured samples, composition B (containing 0,34 mass-% of formic acid) showed composition was not further characterized. which canwater be related vapor toas thereaction limited product. MgO hy -Whereas,ment the[41–43]. second Both one transformations(between 420-450°C) were is related to the final E result (afterThe effect 24 h) of of formic ~100GPa, acid in whichthe development is twice when dration compared in this to system A material due to (50 the99,5 GPa).presence of more intense for castable B, as this material of the green mechanical strength of the the silica-containingcarboxylates’ additive. desorption, that shouldpresented be responsible higher brucite for r contenteleasing due some to the gas phases to the refractories during drying at 110 °C is pre- The thermogravimetric results, obtained interaction of formic acid with MgO. Hence, furnace environment99,0 [41–43]. Both transformations were more intense for castable B, as this sented in Fig. 2. According to the Young’s when using 5 °C · min–1 as heating9 rate, SioxX-Mag may play two important roles in

Mass loss (%) Hence, SioxXmodulus-Mag evolutionmay play of thetwo curedimportant samples, roles indicatedin aluminamaterial that-based the presented microsilica-based MgO- containinghigher bruciteaddi - contentalumina-based due to theMgO-containing interaction of composi formic- acid with MgO. composition B (containing 0,34 mass-% tive contained in sample D partially inhibit tions, as it can inhibit magnesia hydration compositions, as it can inhibit magnesia hydration and limit the interaction of this oxide98,5 with of formic acid) showed final E result (after Mg(OH)2 formation during the previous and limit the interaction of this oxide with 10 the hydrating24 agent h) of (formic ~100 GPa,acid). which is twice when curing and drying steps (Fig. 3). The drying the hydrating agent (formic acid). 98,0 100 200 300 400 500 600 Temperature of sample (°C ) 0,05 100,0 B D (a) B D (b)

0,04 99,5 ) -1

0,03

99,0

0,02 Mass loss (%) Drying rate (% min 98,5 0,01

98,0 0,00 100 200 300 400 500 600 100 200 300 400 500 600 Temperature of sample (°C ) Temperature of sample (°C ) 0,05 B D (b) Fig. 3 a–b Mass loss (a) and drying rate (b) of castables B and D obtained for thermal treatments carried out with heating rate equal to –1 Fig. 3 Mass–1 loss (a) and drying rate (b)–1 of castables B and D obtained for thermal treatments 5 °C·min0,04. The analysed samples were previously cured at 50 °C·24 h and dried at 110 °C·24 h

) carried out with heating rate equal to 5°C/min. The analysed samples were previously cured at -1

0,03 70 50°C/24 h and dried at 110°C/24refractories h WORLDFORUM 11 (2019) [3]

0,02

Drying rate (% min The microsilica-based additive also induced the samples’ mass loss in a wider 0,01 temperature range (Fig. 3a). Considering that the earlier decomposition of hydrated phases (at

0,00 100 200 300 400 500lower temperatures600 than brucite decomposition and carboxylate desorption) may change the Temperature of sample (°C ) 11

Fig. 3 Mass loss (a) and drying rate (b) of castables B and D obtained for thermal treatments

carried out with heating rate equal to 5°C/min. The analysed samples were previously cured at

50°C/24 h and dried at 110°C/24 h

The microsilica-based additive also induced the samples’ mass loss in a wider

temperature range (Fig. 3a). Considering that the earlier decomposition of hydrated phases (at

lower temperatures than brucite decomposition and carboxylate desorption) may change the

decrease close to 1000°C during the second heating cycle (Fig. 5b), suggesting that a permanent

liquid phase was contained in this refractory as a result of the SioxX-Mag addition to this

system.

100,0 B D (a)

99,5

99,0 decrease close to 1000°C during the second heating cycle (Fig. 5b), suggesting that a permanent * Additionally, castables samples pre-fired at 1450°C for 5 h were also analyzed via hot liquid phase was contained in this refractory as a result of the SioxX-Mag additionMass loss (%) to this Young’s modulus measurements98,5 to confirm whether the liquid phase would still be formed in system. PAPER composition D after previously keeping this material for a longer time at high temperatures.

98,0 According to the obtained results100 (Fig.200 6), almost300 no E change400 could500 be detected600 for sample B, Temperature of sample (°C ) whereas a small softening of refractory D (lower E drop when compared to Fig. 5b) was 0,15 100,0 B D (a) B D (b) observed, indicating that a fraction of the liquid could react with the other constituents of this

0,12 99,5 material and, consequently, give rise to new crystalline compounds. ) -1 * 0,09

99,0

* 150 0,06 (a) B Additionally, castables samples pre-fired at 1450°C for 5 h were also analyzed135 via hot Mass loss (%)

cooling

Drying rate (% . min 120 Young’s modulus measurements98,5 to confirm whether the liquid phase would still be formed in 0,03 105 composition D after previously keeping this material for a longer time at high temperatures.90 98,0 0,00 100 200 300 400 500 600 75

100 200 300 400 500 600 According to the obtained results (Fig. 6), almost no E change could be detected for sample B, heating Temperature of sample (°C ) 60 Temperature of sample (°C ) E (G P a) whereas a small softening0,15 of refractory D (lower E drop when compared to Fig. 5b)45 was Fig. 4 a–b Mass loss (a) and drying B rate (b) of Dcompositions B(b) and D obtained for thermal treatments carried out with heating rate Fig. 4 Mass loss (a)30 and drying rate (b) of compositions B and D obtained for thermal observed, indicating that a fraction–1 of the liquid could react with the other constituents of–1 this –1 equal to 20 °C·min . The analysed samples were previously cured at 50 °C·24 h and dried C ycle 1 at 110 °C·24 h . The asterisk (*) indicates 0,12 15 sample B explosion treatments carried out with heating C ycle 2 rate equal to 20°C/min. The analysed samples were material and, consequently,) give rise to new crystalline compounds. -1 * 0 0 150 300 450 600 750 900 1050 1200 1350 1500 0,09 previously cured at 50°C/24 h and dried at 110°C/24 h. The asterisk (*) indicates sample B Temperature (°C )

150 explosion 150 0,06 (a) B (b) D 135 135 cooling Drying rate (% . min 120 cooling 120 0,03 105 105

90 90 13 0,00 heating 75 100 200 300 400 500 600 75

Temperature of sample (°Cheating ) 60 60 E (G P a) E (G P a) 45 45

Fig. 4 Mass loss (a)30 and drying rate (b) of compositions B and D obtained for 30thermal C ycle 1 C ycle 1 15 15 treatments carried out with heating C ycle 2 rate equal to 20°C/min. The analysed samples were C ycle 2 0 0 previously cured at 50°C/240 150 h and300 dr450ied 600at 110°C/24750 900 h.1050 The1200 asterisk1350 1500(*) indicates sample0 150B 300 450 600 750 900 1050 1200 1350 1500 Temperature (°C ) Temperature (°C )

explosion 150 Fig. 5 a–b Young’s(b) modulus (E) evolution as a function of temperatureD for the dried samples (110°C·24 h–1) of composition B (a) and D (b) 135 cooling Fig. 1 a, b Young’s modulus (E) evolution as a function of temperature for the dried samples 120 (110°C/ 24 h) of composition (a) B and (b) D The microsilica-based105 additive also induced orption led to high steam pore pressure that at 900 °C for both evaluated materials,

the samples’90 mass loss in a wider tempera- exceeded the green mechanical strength13 of which is related to the recuperation of MgO ture range (Fig. 3a). Consideringheating that the the evaluated sample. On the other hand, crystalline structure that takes place at this 14 75 earlier decomposition of hydrated phases the higher explosion resistance of compo- temperature range and leads to pore forma- 60 (at lowerE (G P a) temperatures than brucite decom- sition D can be associated with the lower tion in the resulting microstructure [44–46]. position and45 carboxylate desorption) may brucite content, limited carboxylic acid ad- After that, a continuous Young’s modu- change the30 refractory microstructure and sorption on MgO particles and more pro- lus increase was observed for B refractory C ycle 1 15 even induce easier Csteam ycle 2 withdrawal at nounced mass loss at lower temperatures. above 1000 °C due to its sintering process higher temperatures,0 the spalling likelihood The anti-hydrating effect of SioxX-Mag and the further spinel formation in the ma- 0 150 300 450 600 750 900 1050 1200 1350 1500 of castable D should be decreased due to could also be identified in the E profiles ob- trix region, whereas the same transforma- Temperature (°C ) the combined action of SioxX-Mag and tained during the thermal treatments of the tions could only be verified for D samples

formic acid. In order to investigate the ex- dried samples up to 1400 °C (Fig. 5). The at temperatures higher than 1200 °C and Fig. 1 a, b Young’splosion resistancemodulus (ofE) samples evolution B andas a D, function ad- offirst temperature heating cycle for ofthe the dried castables samples present - during the beginning of the cooling step ditional thermogravimetric measurements ed E drop between 300 °C and 450 °C due down to 1000 °C. The delayed stiffening of (110°C/ 24 h) of composition (a) B and (b) D were carried out using higher heating rate to Mg(OH)2 decomposition. As observed, D the latter castables might be derived from (20 °C·min–1). samples showed lower stiffness decrease as liquid phase generation in the sintered mi- According to Fig. 4, castable B exploded ap- a consequence of the reduced amount14 of crostructure at high temperatures. Moreo-

proximately at 450 °C, as water vapor and brucite in its composition (Fig. 5b). Besides ver, the presence of this liquid phase might other gases derived from carboxylate des- that, a new E decrease could be detected help to relieve the thermomechanical stress

refractories WORLDFORUM 11 (2019) [3] 71

Fig. 7 shows the flexural strength (3-point bending test) and apparent porosity of samples

B and D. Both compositions presented significant mechanical strength (Fig. 7a) increase

between curing (50°C/24) and drying (110°C/24 h) steps, which was followed by free-water

withdrawal and, consequently, an important increment of the overall pore content of such

materials (Fig. 7b). After brucite decomposition and carboxylates desorption during heating up

to 600°C, B samples presented an expressive flexural strength decay from 13 MPa down to 3

MPa. As indicated in Fig. 5b, composition D showed lower E decay up to 600°C, which is in

tune with the less intense mechanical strength decrease observed in Fig. 7a. Besides that, both castablesPAPER fired at 600°C/5 h presented an increase in their apparent porosity levels (Fig. 7b) mainly due to the magnesium hydroxide decomposition.

B D the overall pore content of such materials 150 (Fig. 7b). After brucite decomposition and 135 carboxylates desorption during heating up 120 to 600 °C, B samples presented an expres- 105 sive flexural strength decay from 13 MPa

90 down to 3 MPa. As indicated in Fig. 5b,

75 composition D showed lower E decay up to 600 °C, which is in tune with the less 60 E (G P a) intense mechanical strength decrease ob- 45 served in Fig. 7a. Besides that, both casta- 30 bles fired at 600 °C·5 h–1 presented an H e a ting 15 C ooling increase in their apparent porosity levels

0 (Fig. 7b) mainly due to the magnesium hy- 0 150 300 450 600 750 900 1050 1200 1350 1500 droxide decomposition. Temperature (°C ) An additional modulus of rupture decrease

Fig. 6 Young’s modulus evolution as a function of temperaturesamples of B the did pre-fired not present significant changescould also in theirbe identified mechanical after strength firing and casta apparent- porosity Fig. 6 (1450Young’s °C·5 modulush–1) castable evolution samples as a function of temperature of the pre-fired (1450°C/5 h) ble D at 900 °C·5 h–1 (Fig. 7a), which is results after firing at 900°C (Fig. 7)a. Theresult formation of MgO crystallineof spinel structurephase (MgAl recu-2O4) and strong castableassociated samples with spinel formation (expansive formed in composition D after previously peration and most likely the generation of ceramic bonds (due to sintering process) in both refractory materials fired at 1450°C/5 h led to reaction) [22, 23], leading to a slower stiff- keeping this material for a longer time at forsterite phase, as this transformation is ening effect of the specimens during heat- high temperatures.improved cold According mechanical to the strength ob- valuesfollowed (Fig. by7a) volumetric. shrinkage, increasing.An additional modulus of rupture decrease couldtained alsoresults be (Fig.identified 6), almost after nofiring E change castable ing the overall pores content of the casta- The second thermal cycle of sample B did could be detected for sample B, whereas a ble [21, 47]. On the other hand, samples B D at 900°C/5 h (Fig. 7a), which is a result of MgO crystalline structure recuperation and most not lead to significant E changes (Fig. 5a), small softening of refractory D (lower E drop did not present significant changes in their likely thewhich generation highlights ofthat forsterite the main phasephase, trans as -this whentransformation compared isto Fig.foll owed5b) was by observed,volumetric mechanical strength and apparent porosity

formations took place during the first fir- indicating that a fraction of the liquid50°C / 24h could 110°Cresults / 24h after 600°C firing / 5h at 900°C 900 / 5h °C (Fig. 1450°C 7). / 5hThe shrinkage, increasing the overall pores content of the castable [21,47]. On the other hand, ing cycle and a stable microstructure was react with the other constituents35 of this formation of spinel phase (MgAl2O4) and (a) obtained. Composition D still showed the material and, consequently, give rise to new 15 strong ceramic bonds (due to sintering 30 Young’s modulus decrease close to 1000 °C crystalline compounds. process) in both refractory materials fired samples B didduring not present the second significant heating changes cycle in(Fig. their 5b), mechanical Fig. 7 strengthshows the and flexural apparent strength porosity (3-point at 1450 °C·5 h–1 led to improved cold me- 25 suggesting that a permanent liquid phase bending test) and apparent porosity of sam- chanical strength values (Fig. 7a). results after firing at 900°C (Fig. 7). The formation of spinel phase (MgAl2O4) and strong was contained in this refractory as a result ples B and D. Both compositions20 presented Aiming to identify the influence of the liq- ceramic bondsof (due the SioxX-Magto sintering addition process) to inthis both system. refractory significantmaterials fired mechanical at 1450°C/5 strength h led (Fig. to 7a) in- uid phase formation on the mechanical 15 Additionally, castables samples pre-fired at crease between curing (50 °C·24 h–1) and strength of the designed refractories at improved cold mechanical strength values (Fig. 7a). –1 1450 °C for 5 h were also analyzed via hot drying (110 °C·24 h ) steps, 10which was high temperatures, hot modulus of rupture Young’s modulus measurements to confirm followed by free-water withdrawal and, (HMOR) measurements were carried out whether the liquid phase would still be consequently, an important Mechanical strength (MPincrement a) 5 of using specimens pre-fired at the same test-

0 B D

50°C / 24h 110°C / 24h 600°C / 5h 900°C / 5h 1450°C / 5h C omposition 35 25 (a) (b)

30 20

15 20

15 10

10 prn ooiy( ) Aparent porosity (% 5

Mechanical strength (MP a) 5

0 0 B D B D

C omposition C omposition 25 Fig. 7 a–b Flexural(b) strength (a) and apparent porosity (b) of compositions B and D after curing at 50 °C·24 h–1, drying at 110 °C·24 h–1 or firing at 600 °C, 900 °C or 1450 °C for 5 h Fig. 2 a, b Flexural strength (a) and apparent porosity (b) of compositions B and D after curing 20 at 50°C/24 h, drying at 110°C/24 h or firing at 600°C, 900°C or 1450°C for 5 h 72 WORLDFORUM 11 (2019) [3] 15 refractories

10 16

prn ooiy( ) Aparent porosity (% 5

0 B D C omposition

Fig. 2 a, b Flexural strength (a) and apparent porosity (b) of compositions B and D after curing at 50°C/24 h, drying at 110°C/24 h or firing at 600°C, 900°C or 1450°C for 5 h

70 %, whereas the SiO2-free composition (B) presented better performance, keeping

approximately 90 % of its initial Young’s modulus.

The experimental results of this work indicated that the liquid phase generated in

Aiming to identify the influence of the liquid phase formationcomposition on the Dmechanical might be progressively consumed with time, especially when the samples were strength of the designed refractories at high temperatures, hot moduluskept of at rupture temperatures (HMOR) higher than 1000°C. Hence, in order to understand the phase measurements were carried out using specimens pre-fired at the same testingtransformations temperature that for can 5 induce liquid formation or consumption in the evaluated materials, h. Fig. 8 shows that castable B presented a continuous increase of HMORthermodynamic results between simulations 1100- were carried out in order to predict the likely phases that could be

1400°C, whereas the softening of the microsilica-containing compositionfoun (D)d in resulted the matrix in lower fraction of refractories B and D at high temperatures. Fig. 10 highlights the mechanical strength values above 1200°C, due to the liquid phasephases presencethat may coexistin this in the matrix region of castables B andPAPER D between 600°C and 1500°C microstructure. under thermodynamic equilibrium condition.

B D B D 15 react with alumina to100 induce spinel formation. According to the equilibrium condition, both

12 evaluated systems should present liquid phase formation only above 1575°C. 80

9 60

Alumina Mg Al SiO S pinel Mullite 4 10 2 23

6 100 40 (a) Composition B E retained (% )

HMOR (MP a) 90

80 3 20 70

60 0 50 0 1100 1200 1300 1400

0 2 4 6 8 10 40 Temperature (°C ) C ycle ( T = 1000°C ) 30 react with alumina to induce spinel formation. According to the equilibrium condition, P hases (wt % ) both Fig. 8 Hot Modulus of Rupture (HMOR) of compositions B and D; Fig. 9 Young’s modulus (E) of pre-fired samples (1450 °C·5 h–1) 20 Fig. 8 evaluated Hot modulus systemsthe samples shouldof rupture werepresent previously(HMOR) liquid phase offired compositions formation at the same only Btesting above and D.1575°C. The samplesof castables were B and C retained after thermal shock cycles with Fig. 9 Young’s modulus10 (E) of pre-fired samples (1450°C/5 h) of castables B and C retained temperature for 5 h ΔT = 1000 °C 0 previously fired at the same testing temperature for 5 h after thermal shock cycles with ΔT = 1000°C 600 750 900 1050 1200 1350 1500

Alumina Mg Al SiO S pinel Mullite Temperature (°C ) 4 10 2 23 100 100 Although the in situ Young’s (a) modulus tests pointed out thatComposition B the liquid phase contained in (b) Composition D 90 The predicted90 phases were alumina (Al2O3), mullite (3Al2O3.2SiO2), spinel (MgAl2O4) samples D could be partially80 consumed during their thermal treatment at 1450°C/5 h (Fig.80 6), and a silicate with stoichiometry equal to Mg4Al10Si2O23. Due to the presence of SioxX-Mag in 70 70 the remaining vitreous compound60 in this composition influenced not onlycomposition the hot mechanicalD, a greater60 amount of silicate can be formed in this material and, as the 50 50 strength but also the thermal shock resistance of this refractory. Fig. 9 indicates that a decrease 40 temperature increases,40 Mg4Al10Si2O23 should be decomposed around 1275°C, favoring further

30 P hases (wt % ) 30 of the samples’ stiffness could be observed for composition D after generationtwo thermal of cyclesmulliteP hases (wt % ) ( ΔandT MgO in the resulting microstructure. After that, the latter oxide may 20 20

~1000°C) and a continuous10 reduction of this property was detected up to 10 cycles. The10 E 18

0 0 retained for the SioxX-Mag containing castable at the end of the thermal shock tests was around 600 750 900 1050 1200 1350 1500 600 750 900 1050 1200 1350 1500 Temperature (°C ) Temperature (°C )

100 Fig. 10 a–b Predicted (b) phases to be formed in theComposition D matrix fraction of castables17 (a) B and (b) D according to thermodynamic simulations 90 Fig. 10 a, b Predicted phases to be formed in the matrix fraction of castables (a) B and (b) D and for an equilibrium condition between the temperature range of 600–1500 °C 80

70 according to thermodynamic simulations and for an equilibrium condition between the

ing temperature60 for 5 h. Fig. 8 shows that (ΔT ~ 1000 °C) and a continuous reduction phases that could be found in the matrix temperature range of 600°C-1500°C castable B50 presented a continuous increase of this property was detected up to 10 cy- fraction of refractories B and D at high tem-

of HMOR results between 1100–1400 °C, cles. The E retained for the SioxX-Mag con- peratures. Fig. 10 highlights the phases that 40

whereas 30the softening of the microsilica- taining castable at the end of the thermal may coexist in the matrix region of casta- P hases (wt % )

containing20 composition (D) resulted in shock tests wasConsidering around 70 %,that whereas the evaluated the blesrefractories B and D betweenare heterogeneous 600–1500 °Cmaterials under and contain lower mechanical strength values above SiO -free composition (B) presented better thermodynamic equilibrium condition. 10 2 components with a wide range of particle sizes, the thermodynamic equilibrium for such 1200 °C, 0due to the liquid phase presence performance, keeping approximately 90 % The predicted phases were alumina (Al2O3), in this microstructure. 600 750 900 1050 1200 of its1350 initialsystems Young’s1500 should modulus. be only reached after keepingmullite (3Althem2O at3 ·high 2SiO temperatures2), spinel (MgAl for 2Oa 4very) long time. Although the in situ Young’sTemperature (°C modulus ) The experimental results of this work in- and a silicate with stoichiometry equal to tests pointed out that the liquid phase dicated that the liquid phase generated Mg Al Si O . Due to the presence of SioxX- 4 10 2 23 19 contained in samples D could be partially in composition D might be progressively Mag in composition D, a greater amount of Fig. 10 a, b Predicted phases to be formed in the matrix fraction of castables (a) B and (b) D consumed during their thermal treatment at consumed with time, especially when the silicate can be formed in this material and, –1 according to1450 thermodynamic °C·5 h (Fig. simulations6), the remaining and vitrefor -an equilibriumsamples were condition kept at temperatures between the higher as the temperature increases, Mg4Al10Si2O23 ous compound in this composition influ- than 1000 °C. should be decomposed around 1275 °C, temperature rangeenced ofnot 600°C only -the1500°C hot mechanical strength Hence, in order to understand the phase favoring further generation of mullite and but also the thermal shock resistance of this transformations that can induce liquid for- MgO in the resulting microstructure. After refractory. Fig. 9 indicates that a decrease mation or consumption in the evaluated that, the latter oxide may react with alu- Consideringof the samples’that the stiffnessevaluated could refractories be observed are heterogeneousmaterials, thermodynamic materials and simulations contain were mina to induce spinel formation. According for composition D after two thermal cycles carried out in order to predict the likely to the equilibrium condition, both evaluated components with a wide range of particle sizes, the thermodynamic equilibrium for such

systems should be only reached after keeping them at high temperatures for a very long time. refractories WORLDFORUM 11 (2019) [3] 73

PAPER

systems should present liquid phase forma- Another benefit of SiO2 presence in the castables.J. Am. Ceram. Soc. 97 (2014) 1233– tion only above 1575 °C. analyzed refractories is the fact that sam- 1241 doi:10.1111/jace.12873 Considering that the evaluated refractories ples with higher mechanical strength levels [4] Luz, A.P.; Consoni, L.B.; Pagliosa, C.; Aneziris, are heterogeneous materials and contain could be obtained at intermediate tempera- Ch.G.; Pandolfelli, V.C.: MgO fumes as a poten- components with a wide range of particle tures (600 °C and 900 °C). Nevertheless, tial binder for in situ spinel containing refractory sizes, the thermodynamic equilibrium for liquid phase formation affected the stiff- castables. Ceram. Int. (2018) doi:10.1016/j. such systems should be only reached after ness and hot modulus of rupture results for ceramint.2018.05.201 keeping them at high temperatures for a these castables above 1000 °C. According [5] Thomas, J.J.; Musso, S.; Prestini, I.: Kinetics and very long time. Consequently, Al2O3–MgO– to the experimental tests and thermody- activation energy of magnesium oxide hydra-

SiO2 castables containing microsilica tend namic simulations, this liquid can be pro- tion. J. Am. Ceram. Soc. 97 (2014) 275–282 to give rise to metastable microstructures gressively consumed when the SioxX-Mag- doi:10.1111/jace.12661 [22], comprised by micro-regions present- containing refractories are kept at high [6] Amaral, L.F.; Oliveira, I.R.; Salomão, R.; Frol- ing distinct chemical composition from temperatures and for long periods of time. lini, E.; Pandolfelli, V.C.: Temperature and the overall refractory product. Some stud- Thus, the softening of the samples might be common-ion effect on magnesium oxide (MgO) ies presented in the literature pointed out minimized with the decrease of the liquid hydration. Ceram. Int. 36 (2010) 1047–1054 forsterite (2MgO · SiO2) formation already phase content in the resulting microstruc- doi:10.1016/j.ceramint.2009.12.009 at 815 °C in mixtures containing MgO:SiO2 ture. [7] Birchal, V.S.S.; Rocha, S.D.F.; Ciminelli, V.S.T.: molar ratio higher than 1,25 [14, 25, 26]. Besides favoring an improvement of the The effect of magnesite calcination condi- The cold mechanical strength and appar- flowability, mechanical strength and drying tions on magnesia hydration. Miner. Eng. ent porosity results of composition D, ob- behavior of the Al2O3–MgO castables, the 13 (2000) 1629–1633 doi:10.1016/S0892- tained after firing at 600 °C and 900 °C, selected microsilica-based compound can 6875(00)00146-1 suggest that forsterite might also have been also induce the generation of phases with [8] Rocha, S.D.; Mansur, M.B.; Ciminelli, V.S.: Kinet- formed in this refractory. The generation of high refractoriness, such as forsterite and ics and mechanistic analysis of caustic magne- this phase is desired due to its high melt- mullite. In this sense, a suitable temperature sia hydration. J. Chem. Technol. Biotechnol. 79 ing point (1840 °C), which is higher than and time must be selected during the firing (2004) 816–821 doi:10.1002/jctb.1038 the decomposition temperature of the other step of the castables. Therefore, SioxX-Mag [9] Birchal, V.S.; Rocha, S.D.F.; Mansur, M.B.; silicate compound predicted via thermody- product might also be a promising additive Ciminelli, V.S.T.: A Simplified mechanistic analy- namic simulations. to high-alumina compositions containing sis of the hydration of magnesia Can. J. Chem. MgO in their matrix-fraction. Eng. 79 (2001) 507–511 4 Conclusions [10] Salomão, R.; Arruda, C.C.; Souza, A.D.V.; This work evaluated the influence of add- Acknowledgements Fernandes, L.: Novel insights into MgO hy- ing a microsilica-based additive (SioxX- This study was financed in part by the Co- droxylation: Effects of testing temperature,

.volume and solid load. Ceram. Int ׳Mag) to Al2O3–MgO castables prepared ordenação de Aperfeiçoamento de Pessoal samples with and without formic acid (hydrating de Nível Superior – Brasil (CAPES) – Finance 40 (2014) 14809–14815 doi:10.1016/j.cera- agent) in order to minimize their spalling/ Code 001. The authors would like to thank mint.2014.06.074 explosion trend during heating. According FIRE (Federation for International Refractory [11] Amaral, L.F.; Oliveira, I.R.; Bonadia, P.; Salomão, to the obtained results, this SiO2-contain- Research and Education) and RHI-Magnesi- R.; Pandolfelli, V.C.: Chelants to inhibit magnesia ing additive (which already presented a ta for supporting this work. The authors are (MgO) hydration Ceram. Int. 37 (2011) 1537– dispersant compound in its composition) also grateful to Almatis and Elkem for sup- 1542 doi:10.1016/j.ceramint.2011.01.030 led to the development of compositions plying the raw materials used in this study [12] Matabola, K.P.; van der Merwe, E.M.; Strydom, with better flowability along the 30 min and F. G. Pinola for her assistance in the C.A.; Labuschagne, F.J.W.: The influence of hy- after the end of their mixing process. The experimental tests. drating agents on the hydration of industrial interaction of SiO2 and MgO during the magnesium oxide, J. Chem. Technol. Biotechnol. mixing, curing and drying steps of such References 85 (2010) 1569–1574. doi:10.1002/jctb.2467 castables led to lower amounts of brucite [1] Salomão, R.; Bittencourt, L.R.M.; Pandolfelli, [13] E.M. van der Merwe, C.A. Strydom, Hydration of formation. Consequently, no explosion V.C.: A novel magnesia based binder (MBB) medium reactive magnesium oxide using hydra- was detected even when subjecting those for refractory castables. Interceram. 58 (2009) tion agents. J. Therm. Anal. Calorim. 84 (2006) samples to high heating rate (20 °C·min–1) 21–24 467–471 doi:10.1007/s10973-005-7291-6 and the most significant mass loss of the [2] Souza, T.M.; Braulio, M.A.L.; Luz, A.P.; Bonadia, [14] Souza, T.M.; Luz, A.P.; Santos, T.; Gimenes, SioxX-Mag containing samples took place P.; Pandolfelli, V.C.: Systemic analysis of MgO D.C.; Miglioli, M.M.; Correa, A.M. Pandolfelli, mainly at temperatures lower than 350 °C. hydration effects on alumina – magnesia refrac- V.C.: Phosphate chemical binder as an anti-

On the other hand, the reference materials tory castables. Ceram. Int. 38 (2012) 3969– hydration additive for Al2O3–MgO refractory

(SiO2-free castables), prepared with the ad- 3976. doi:10.1016/j.ceramint.2012.01.051 castables. Ceram. Int. 40 (2014) 1503–1512 dition of a polymeric dispersant, exploded [3] Souza, T.M.; Luz, A.P.; Braulio, M.A.L.; Pa- doi:10.1016/j.ceramint.2013.07.035 during heating and under the same testing gliosa, C.; Pandolfelli, V.C.: Acetic acid role on [15] Santos Jr., T.; Santos, J.; Luz, A.P.; Pagliosa, C.; conditions. magnesia hydration for cement-free refractory Pandolfelli, V.C.: Kinetic control of MgO hydra-

74 refractories WORLDFORUM 11 (2019) [3] PAPER

tion in refractory castables by using carboxylic tion products in the system magnesia-silica- fractory castables via hot elastic modulus meas- acids. J. Eur. Ceram. Soc. 38 (2018) 2152–2163 water. J. Am. Ceram. Soc. 37 (1954) 38–42 urements. Ceram. Int. 39 (2013) 6189–6197 doi:10.1016/j.jeurceramsoc.2017.11.046 doi:10.1111/j.1151-2916.1954.tb14001.x doi:10.1016/j.ceramint.2013.01.038 [16] Santos, T.; Luz, A.P.; Pagliosa, C.; Pandolfelli, [27] Bjorn, M.; Peng, H.; Ming, L.: Cement free MgO [38] ASTM International: Standard test methods

V.C.: Mg(OH)2 nucleation and growth param- castables: part I: Flow, setting and slaking, in: for cold crushing strength and modulus of eters applicable for the development of MgO- Proc. UNITECR 2013, 2013, p. 881–886 rupture of refractories (C133-97), 2003, 1–6 based refractory castables. J. Am. Ceram. Soc. [28] Zetterström, C.; Auvray, J.-M.; C. Wöhrmeyer, C.; doi:10.1520/C0133-97R08E01.2 99 (2016) 461–469 doi:10.1111/jace.14019 Parr, Ch.: Enhanced permeability for a rapid dry- [39] ASTM International: Standard test methods for [17] Souza, T.M.; Braulio, M.A.L.; Pandolfelli, V.C.: out of refractory castables, in: UNITECR 2015, apparent porosity, liquid absorption , apparent Novel Technological Route to Overcome the 14th Bienn. Worldw. Congr., Viena, Austria, specific gravity, and bulk density of refractory Challenging Magnesia Hydration of Cement- 2015, p. 1–4 shapes by vacuum pressure (C380-00), 2013, Free Alumina Castables. refractories WORLDFO- [29] Auvray, J.-M.; Zetterström, C.; Wöhrmeyer, C.: 1–5 doi:10.1520/C0830-06R11.2 RUM 5 (2013) [1] 93–98 The impact of permeability for a rapid dry-out of [40] ASTM Standard: Test Method for Modulus of [18] dos Santos, T.; Pinola, F.G.; Luz, A.P.; Pagliosa, refractory castables, in: Proc. Unified Int. Tech. Rupture of Refractory Materials at Elevated

C.; Pandolfelli, V.C.: Al2O3–MgO refractory Conf. Refract., Wien, 2015: 1–5. doi:101:1- Temperatures (C583-15), 2015, 1–6 doi:http:// castables with enhanced explosion resistance 201506294612 dx.doi.org/10.1520/C0583-15 due to in situ formation of phases with lamel- [30] Luz, A.P.; Moreira, M.H.; Wohrmeyer, C.; Parr, [41] Domen, K.; Yamamoto, H.; Watanabe, N.; Wada, lar structure. Ceram. Int. 44 (2018) 8048–8056 Ch.; Pandolfelli, V.C.: Drying behavior op- A.; Hirose, C.: Sum-frequency generation and doi:10.1016/j.ceramint.2018.01.246 timization of dense refractory castables by temperature-programed desorption studies of [19] Salomão, R.; Pandolfelli, V.C.: The particle size adding a permeability enhancing active com- formic acid on MgO(001) surfaces. Appl. Phys. distribution effect on the drying efficiency of pound. Ceram. Int. (2019) doi:10.1016/j.cera- A Mater. Sci. Process. 60 (1995) 131–135 polymeric fibers containing castables. Ceram. mint.2019.01.242 doi:10.1007/BF01538237 Int. 34 (2008) 173–180 [31] Peng, H.; Luo, M.; Myhre, B.: New additive [42] Peng, X.D.; Barteau, M.A.: Dehydration of [20] Salomão, R.; Pandolfelli, V.C.: Polypropylene fib- packages for self-flowing high-alumina and carboxylic acids on the MgO(100) surface. ers and their effects on processing refractory MgO based refractory castables, in: Proc. ALA- Catal. Letters. 7 (1991) 395–402 doi:10.1007/ castables. Int. J. Appl. Ceram. Technol. 4 (2007) FAR Congr. 2012, Cancun, Mexico, 2012, p. BF00764930 496–502 1–13, www.elkem.com/Global/ESM/support/ [43] Yamamoto, H.; Watanabe, N.; Wada, A.; Domen, [21] Peng, H.; Myhre, B.; Ming, L.:Cement free MgO technical-papers/refractories/82- New-additive- K.; Hirose, C.: Adsorption and decomposition of Castables Part II: Strength and Explosion Resist- packages-for-self-flowing-high-alumina-and- formic acid on MgO(001) surface as investigat- ance, in: Proc. UNITECR 2013, 2013, p. 887– MgO-based-refractory-castables.pdf ed by temperature programmed desorption and 892 [32] Peng, H.; Myhre, B.: New additive package for sum-frequency generation spectroscopy: Recur- [22] Sandberg, B.; Myhre, B.; Holm, J.L.:Castables in microsilica gel-bonded MgO castables, in: Proc. rence induced defect sites. J. Chem. Phys. 106

th the system MgO–Al2O3–SiO2, in: Proc. Unified 56 Int. Colloq. Refract., Aachen, Germany, (1997) 4734 doi:10.1063/1.473470 Int. Tech. Conf. Refract., Kyoto, Japan, 1995, p. 2013, p. 1–6 [44] Ball, M.C.; Taylor, H.F.W.: The Dehydration of 3–11 [33] Peng, H.; Myhre, B.: Microsilica-gel bonded Brucite. Mineral. Mag. 32 (1961) 754–766 [23] Salomão, R.; Pandolfelli, V.C.: Microsilica addi- self-flowing MgO castables using a new addi- doi:10.1180/minmag.1961.032.253.02

tion as an antihydration technique for magne- tive package, in: Proc. IREFCON, Kolkata, India, [45] Green, J.: Calcination of precipitated Mg(OH)2 sia-containing refractory castables. Am. Ceram. 2014, p. 1–7 to active MgO in the production of refrac- Soc. Bull. 86 (2007) 9301–9309 [34] Dinger, D.R.; Funk, J.E.: Particle packing, part II: tory and chemical grade MgO. J. Mater. Sci. 18 [24] Szczerba, J.; Prorok, R.; Czapka, Z.; Madej, D.; Review of packing of polydisperse particle sys- (1983) 637–651 doi:10.1007/BF00745561 Sniezek, E.; Jastrzebska, I.: Influence of microsil- tems. Interceram 41 (1992) 95–97 [46] Wang, J.; Novaro, O.; Bokhimi, X.; López, T.; ica on mechanical properties of basic castables, [35] Innocentini, M.D.M.; Cardoso, A.; Akyioshi, Gómez, R.; Navarrete, J.; Llanos, M.-E.; López- in: D. Goski, J. Smith (Eds.), UNITECR 2013 Proc. M.M.; Pandolfelli, V.C.: Drying stages during the Salinas, E.: Characterizations of the thermal Unified Int. Tech. Conf. Refract., Wiley, Victoria, heating of high-alumina, ultra-low-cement re- decomposition of brucite prepared by sol-gel Canada, 2013, p. 927–932 fractory castables. J. Am. Ceram. Soc. 86 (2003) technique for synthesis of nanocrystalline MgO. [25] Szczerba, J.; Prorok, R.; S´niez˙ek, E.; Madej, D.; 1146–1148 Mater. Lett. 35 (1998) 317–323. doi:10.1016/ Mas´lona, K.: Influence of time and temperature [36] ASTM International: Standard test method for S0167-577X(97)00273-5 on ageing and phases synthesis in the MgO– dynamic Young’s modulus, shear modulus and [47] Hundere, A.; Myhre, B.; Odegard, C.; Sandberg,

SiO2–H2O system. Thermochim. Acta. 567 Poisson’s ratio by sonic resonance (E1875-13), B.; Zhou, N.; Zhang, S.; Bi, Z.; Chen, Z.: Mag-

(2013) 57–64. doi:10.1016/j.tca.2013.01.018 2013, 1–9 doi:10.1520/E1875-08.2 nesium-Silicate-Hydrate bonded MgO–Al2O3 [26] Kalousek, G.L.; Mui, D.: Studies on formation [37] Luz, A.P.; Santos, T.; Medeiros, J.; Pandolfelli, castables, in: Proc. Annu. Conf. Metall., Quebec, and recrystallization of intermediate reac- V.C.: Thermal shock damage evaluation of re- Canada, 1999, p. 2–7.

refractories WORLDFORUM 11 (2019) [3] 75