2015-GARCIA-PRIETO-Influence of Microstructural Characteristics on Fracture Toughness of Refractory...

J

A

BmSEcbdmt

K

1

dttoptp

tbe

h0

ARTICLE IN PRESS+ModelECS-9946; No. of Pages 16Available online at www.sciencedirect.com

ScienceDirect

Journal of the European Ceramic Society xxx (2015) xxxxxx

Influence of microstructural characteristics on fracture toughness ofrefractory materials

Ana Garca-Prieto a, Manuel Dos Ramos-Lotito a,b, Delia Gutirrez-Campos b,Pilar Pena a, Carmen Baudn a,

a Instituto de Cermica y Vidrio, CSIC, Kelsen 5, Madrid 28049, Spainb Universidad Simn Bolvar, Dept. Ciencia de los Materiales, Valle de Sartenejas, Caracas 1080, Venezuela

Received 28 July 2014; received in revised form 15 December 2014; accepted 16 December 2014

bstract

asic relationships between the microstructure and the texture of refractories and their toughness have been established. A series of commercialaterials has been chosen in order to highlight the influence of microstructural characteristics on fracture behaviour and associated toughness.ilica, silicaalumina and silicaaluminazirconia based shaped refractories and a calcium aluminate cement bonded concrete have been analysed.xtensive microstructural characterisation has been performed using a combination of techniques, including chemical analysis by X-ray fluores-ence, X-ray diffraction, reflected light optical microscopy and scanning electron microscopy with analysis by dispersive energies. Fracture haseen characterised using stable fracture tests of SENB tested in 3 point bending. Stability was reached in displacement and crack mouth openingisplacement controlled tests. Size effect has been analysed by using two different specimen sizes and relative notch lengths. For the range oficrostructures studied, the obtained results have allowed to characterise toughness and establish the relationships toughness-microstructure and

exture. 2015 Elsevier Ltd. All rights reserved.

eywords: Alumina; Castables; Microstructure; Refractories; Toughness

tmiusacag

s

. Introduction

The study of commercial refractories is quite uncommonue to the challenges generated by the impossibility of con-rolling the chemical composition and/or the microstructure ofhe material. Nevertheless, tackling this task has the advantagef avoiding problems linked to reproducing the manufacturingrocess at a laboratory scale, which is not a trivial issue dueo inherent differences between laboratory and refractory plantsrocesses.

The performance of refractories in use is directly related to

Please cite this article in press as: Garca-Prieto A, et al. Influence of microsJ Eur Ceram Soc (2015), http://dx.doi.org/10.1016/j.jeurceramsoc.2014.1

heir microstructure and texture which, in turn, is determinedy the characteristics of the raw materials (chemical and min-ralogical composition and size and shape distribution) and by

Corresponding author. Tel.: +34 917355840.E-mail address: [email protected] (C. Baudn).

iitdimv

ttp://dx.doi.org/10.1016/j.jeurceramsoc.2014.12.020955-2219/ 2015 Elsevier Ltd. All rights reserved.

he processing or installation procedure. Thus, the study of com-ercial products imposes the necessity of characterising them

n a comprehensive manner. It is particularly important for thenderstanding of their behaviour in terms of the basic materialcience relationships between their composition, microstructurend properties. By using several complementary techniques likehemical analysis, mineralogical studies, and microstructuralnd textural analysis, it is possible to obtain useful data forenerating a detailed description of the material.

Refractories are applied in processes involving mechanicaltrains, variable high temperatures and aggressive environmentsncluding corrosion and erosion from solids, liquids and gasesn movement. In particular, the thermal stress fracture of refrac-ory components, due to temperature cycling and/or temperatureifferences through the material, is a widespread problem of

tructural characteristics on fracture toughness of refractory materials.2.020

ndustrial importance. In the same way, mechanical overload asay be originated by impact during the loading of the process

essel, as occurs in electrical arc furnaces, or by deformations
dx.doi.org/10.1016/j.jeurceramsoc.2014.12.020http://www.sciencedirect.com/science/journal/09552219dx.doi.org/10.1016/j.jeurceramsoc.2014.12.020mailto:[email protected]/10.1016/j.jeurceramsoc.2014.12.020

ARTICLE IN PRESS+ModelJECS-9946; No. of Pages 162 urope

otcceptpaaab

sbtftf

we(cmcapmscifi

rtUfottcmftatb

fposa

a

ohbbjw(iivuinbvFfrf

tttwmttsois

tcmLrtdftmuf

inrosadt

A. Garca-Prieto et al. / Journal of the E

f the kiln shell, as observed in the cement industry, can leado fracture. Fracture as a result of thermal or mechanical shocksould lead to a catastrophic failure of the refractories with strongonsequences for the process in which they are used. How-ver, fracture can also result just in the development of a crackattern.1 In this regard, it should be pointed out that the conven-ional characterisation of the mechanical properties of refractoryroducts is done in terms of two properties: crushing strengthnd modulus of rupture. Even though such properties might bedequate for quality control purposes, the results achieved do notllow the evaluation or study of the fracture processes, and moreasic mechanical properties are needed for characterisation.

The extension of fracture due to thermal or mechanicaltrains, i.e.: the damage would be determined by the ratioetween the amount of energy available and the energy neededo create new crack surfaces or specific fracture energy. There-ore, this ratio will characterise the resistance of materialso subcritical crack growth and the proneness to catastrophicailure.2,3

Refractory products are heterogeneous ceramic materialshich fracture exhibits notable deviations from pure lin-

ar elastic.1,410 Several energy-consuming processes aheadprocess zone) and behind (process wake) the crack tip areonsidered to contribute to this behaviour. Microcracking andultiple crack branching are usually observed in the frontal pro-

ess zone, while grain bridging and friction of the crack faces areble to consume energy in the process wake zone. As a result, thearameters that evaluate toughness of refractories are no longeraterial constants but they increase for increasing crack exten-

ion. In the field of advanced ceramics, this fracture pattern isalled rising R-curve behaviour, with R representing toughness;n contrast with the brittle flat R-curve observed for glass or forne grain size ceramic specimens.

The R-curve concept is not used for the characterisation ofefractories because of the experimental difficulties associatedo follow a single crack in such heterogeneous microstructures.nlike the R-curve, the work of fracture, wof, has been success-

ully used to describe the fracture of refractories.6 The advantagef this energy parameter is that it does not require any assump-ions about the constitutive equation of the body with the cracko discuss its propagation.11,12 In terms of energy, the criti-al energy release rate, Gc, of refractories with well designedicrostructures is always significantly lower than the specific

racture energy, GF (2wof). The ratio between the specific frac-ure energy and the energy release rate, GF/Gc, has been defineds a toughness, flexibility or apparent ductility ratio.1 The higherhis relation, the higher is the resistance of the material to damagey thermal or mechanical strains.

An important point to consider when determining the specificracture energy (or the work of fracture) of refractories is theotential influence of the specimen size (size effect) in thebtained values because GF increases with increasing fractureurface until the specimen geometry allows the development of


well developed wake zone.Nakayama, Tattersall and Tappin, and Davidge and Tappin

ccomplished the first studios on the determination of work

cbr

an Ceramic Society xxx (2015) xxxxxx

f fracture in the 60s,1315 and since then, many laboratoriesave conducted such analyses but a standard test has not yeteen established. The concept of work of fracture introducedy Nakayama4,13 is defined as the mean work per unit of pro-ected fracture area required to propagate a crack in a stableay. In Nakayamas test a parallelepiped bar with a triangular

chevron) notch at its centre is loaded in three point bend-ng (3-pb). Using this geometry, refractory specimens fracturen a stable way when a sufficient stiff machine is used. Thealue of work of fracture is determined from the total areander the LoadDisplacement curve recorded during the exper-ment and the size of the projected fracture surface. Chevronotch experiments often imply high variability (>10%)6,7,1618

ecause coarse aggregates at the apex of the chevron notch giveery high values as compared to the average for the material.or example, 1023% variability in wof has been reportedor alumina-spinel castables,18 and for high alumina refracto-ies (45100 wt.% alumina) variability up to 38%7,16 has beenound.

In the 1980s the wedge-splitting fracture test was developedo perform stable fracture tests and patented by Tschegg.19 Thisest is a special form of the so-called compact tension test,he specimen with a groove and notch is split in two halveshile monitoring the load and crack mouth opening displace-ent (CMOD). In this experimental setup, large specimens of

he size of bricks can be tested.5,8,2030 Most data produced usinghe wedge-splitting test are reported for pure magnesia, magne-ia spinel and magnesia carbon refractories,5,8,9,2427 which areut of the scope of this paper. A relatively low number of stud-es provide data for high alumina castables and alumina-basedhaped materials.20,22,23,2830

Ribeiro and Rodrigues22 applied the wedge splitting methodo characterise fracture energy of two high-alumina refractoryastables. Miyaji et al.29 analysed five different castable for-ulations and introduced a figure of merit derived from theoadDisplacement curve to evaluate the thermal shock damage

esistance. More recently, a methodology was presented usinghe wedge-splitting test complemented with images obtaineduring mechanical loading to determine the crack propagationor a pure alumina and alumina with titania and zirconia addi-ives refractory compositions.28,30 Jin et al.27 have proposed a

ethodology to estimate the tensile strength, and Youngs mod-lus of refractories in addition to the specific fracture energyrom wedge splitting test results.

It should be pointed out that variability of data of mechan-cal properties in the refractory literature is most of the timeot reported, and, in many cases, only one data for each mate-ial experimental condition is provided. This is often the casef work of fracture values determined by the splitting testo, it is not possible to discuss in a general way the repeat-bility associated with this technique. Nevertheless, the scarceata available reveal rather high dispersion of wof results; forwo commercial alumina based low cement (2 wt.% alumina)


astables heat-treated at 1100 C, 12 and 18% variability haveeen reported22 and variability between 5 and 22% has beeneported for basic refractories.24
dx.doi.org/10.1016/j.jeurceramsoc.2014.12.020

ARTICLE IN PRESS+ModelJECS-9946; No. of Pages 16urope

sdotort

ntcmwtpitG

p

oaicgeatgolp

vboatcita

msporacsb

hp

im

K

wgrfsF

f

G

wt

fpN

tr

ota

2

scapAa

rwpc

dEQ

sA


Despite the fact that there is an evident interest in the wedgeplitting test for work of fracture determination of refractoriesue to its capability for testing relatively large specimens inrder to average the whole fracture process, it is complicateo establish this method as a laboratory routine one. More-ver, as mentioned before, variations in reported ranges ofesult would mask the effect of microstructural differences onoughness.

From the extensive spectrum of available fracture tough-ess tests, bending of parallelepiped specimens with straighthrough notches (SENB, Single Edge Notch Beam) using asontrol variable the displacement of the load frame (displace-ent control) is a relatively simple way of testing materialsith R-curve fracture, like refractory products. In general, in

he refractory field this method is used to determine toughnessarameters describing the initiation of fracture, critical stressntensity factor in mode I, KIC, and the energy for crack initia-ion, nbt which is a measure of the critical energy release rate

c, Gc = 2nbt.4,15,3133

The advantage of using stable fracture tests of SENB in 3-b is that both parameters for initiation, nbt, and propagation,wof, of fracture can be extracted from a single test. The sizef the specimen can be readily adjusted to ensure that the lig-ment is large enough to encompass the fracture process zone;n this way, the results will be statistically valid. As a term ofomparison, a representative volume 34 times the largest aggre-ate size was determined by Romero and Masad34 and Wagonert al.35 for SENB testing of asphalt concrete. However, therere two main experimental problems to solve with regard tohis test.1 On the one hand, the attainment of stability for thiseometry is more difficult than for the chevron specimens and,n the other hand, straight through notches are more prone toead to the wandering of the propagating crack from the initiallane.

When displacement of the load frame is used as controlariable for SENB in 3-pb, a general requirement to reach sta-ility is to use a high stiffness machine.3639 The crack mouthpening displacement (CMOD) has been proposed and useds control parameter for stable fracture testing under condi-ions that would have led to unstable fracture for displacementontrolled tests. In this regard, detailed procedure and theoret-cal considerations for performing CMOD controlled fractureests of brittle materials have been previously reported by theuthors.3739

In this work six different types of commercial refractoryaterials with distinctive specifications in chemical compo-

ition and microstructural characteristics were tested. Theroducts were evaluated using CMOD and displacementf the load frame as control parameters using equivalentates and two relative notch lengths (0.25 and 0.50). Oncessured the significance of the obtained data, the importanthallenge in the study has been to correlate the micro-tructural features of the refractory materials with the fracture


ehaviour.From the stable fracture tests different toughness parameters

ave been evaluated; the usual terminology used in refractoryractice has been assumed for reporting. The critical stress

awtp

an Ceramic Society xxx (2015) xxxxxx 3

ntensity factor in mode I, KIC, has been calculated from theaximum load according to Guinea et al.40:

IC = 3PL2BW3/2

K() (1)

here P is the applied load, L is the span, B and W are theeometrical parameters of thickness and width of the specimen,espectively. K () is a general shape function which is validor any value of the relative notch length (0 1) andpan-to-depth ratios ( = L/W) larger than 2.5 (2.5 16).40rom KIC and Youngs modulus, the energy for crack initiation,nbt (= Gc/2) was calculated according to the analysis of Irwin

or plane strain conditions41:

C = K2IC(1 2)

E(2)

here KIC is the critical stress intensity factor in mode I, ishe Poissons ratio and E is the Youngs modulus.

To evaluate the fracture process, wof has been calculatedrom the area under the LoadDisplacement curve and therojection of the fracture surface following the procedure ofakayama.4,13

In order to estimate the inelastic energy contribution to frac-ure, the ratio between the specific fracture energy and the energyelease rate, GF/Gc, has been calculated using the experimentalwof and nbt values.

The chemical, structural and microstructural characteristicsf the six studied refractory materials were correlated withheir fracture behaviour using the different toughness parametersbove described.

. Experimental

Six different types of commercial refractory materials weretudied: two aluminasilicazirconia (AZS), a superduty fire-lay, one group 28 insulating firebrick, one standard silica bricknd a high-alumina regular castable heat treated at the use tem-erature (1100 C). They were labelled as follows: AZS1, AZS2,S, ASI, S and C, respectively. Fig. 1 shows the macroscopic

spect of these six refractories.Chemical analysis was carried out with a Philips (Holland) X-

ay fluorescence equipment, model MagiX PW 2424. Samplesere prepared with the standard procedure of forming a fusedellet. Li2B4O7 was added to the ground powder and preparedapsules were heat treated at 1000 C.

Bulk and true densities and apparent and true porosity wereetermined following the procedures described in two standards:N 993-142 and EN 993-2.43 He picnometry was done using auantachrome (USA) apparatus.Determination of crystalline phases was performed on ground

amples in a Bruker (Germany) X-ray diffractometer, model D8dvance, with copper anode (CuK1 = 0.15418 nm) working


t 40 kV and 40 mA. Scans were performed in continous modeith steps of 0.05 at a rate of 153 s per step. The analysis of

he XRD patterns was accomplished using the EVA 6.0 Diffraclus software (Bruker, Germany). The experimental diffraction

ARTICLE IN PRESS+ModelJECS-9946; No. of Pages 164 A. Garca-Prieto et al. / Journal of the European Ceramic Society xxx (2015) xxxxxx

Fig. 1. Macroscopic aspect of the six commercial refractory materials studied. Scanned images of polished cross sections of tested specimens (25 mm 25 mm). (a)A 2. (c) s

pf

t4c

s(idifip

itocmas

m((cGt

m

fvad

(oS(ifcMphtpapta

3

3

luminasilicazirconia brick, AZS1. (b) Aluminasilicazirconia brick, AZSilica brick, S. (f) High-alumina regular castable, C.

atterns were compared to the files of the International Centreor Diffraction Data (ICDD).44

Standard specimens (EN 993-6)45 for mechanical charac-erisation (150 mm 25 mm 25 mm and 200 mm 40 mm 0 mm) were diamond machined from the received bricks andoncrete pieces.

The size distributions of the aggregates were evaluated fromcanned (HP Scanjet 5370 C, USA) images of the lateral surfaces150 mm 25 mm 25 mm) of the specimens used for mechan-cal testing with the Leica Qwin software (UK). The equivalentiameter was calculated from the surface of the particles assum-ng spherical shape. A minimum of 560 particles was analysedor each material. This study was not performed for the insulat-ng brick because its largest microstructural features were theores which all had similar size (1200 m).

Specimens for microstructural evaluation were embeddedn resin in vacuum environment to assure the penetration ofhe resin in the pores. Then, microstructural characterisationf the materials was carried out using a reflected-light opti-al microscopy, RLOM, with a Zeiss (Axiophot, Germany)icroscope and field emission scanning electron FE-SEM with

nalysis by energy dispersive X-ray spectroscopy (EDS) micro-cope (Hytachi S-4700 type I, Japan).

Youngs modulus of the refractory materials was deter-ined from the resonance of the parallelepiped bars

150 mm 25 mm 25 mm) tested in flexure by impactGrindosonic, Belgium). Calculations were performed using theommon value of Poissons ratio for refractory materials (0.17).iven values are the average of 6 determinations and errors are


he standard deviations.Room temperature modulus of rupture (MOR) was deter-

ined by three point bending, 3-pb, (span 125 mm; 0.5 mm/min)

imca

Superduty fireclay brick, AS. (d) Group 28 insulating brick, ASI. (e) Standard

ollowing the procedure of EN 993-6 standard,45 using a uni-ersal testing machine (Instron 1114, USA). Reported valuesre the average of 3 determinations and errors are the standardeviations.

Notches for toughness testing were done using a BuehlerUSA) sawing machine model IsoMet 4000 with diamond discf 300 m width to reach notches with tip radius around 100 m.pecimens with relative notch lengths = a/W = 0.25 and 0.50a = notch length, W = specimen width) were prepared and testedn three point bending using spans of 125 mm and 180 mmor the small and large specimens, respectively. All tests wereonducted in universal testing machine (model EM1/50/FR,icrotest, Spain) with capability of crack mouth opening dis-

lacement (CMOD) recording and controlling. This equipmentas been described elsewhere.37,38 Tests were performed usinghe CMOD and the displacement of the frame load as controlarameters. Rates of 70 m/min and 0.02 mm/min for CMODnd displacement, respectively, were applied. Three tests wereerformed for each testing condition; reported values of theoughness parameters are the average of the three determinationsnd errors are the standard deviations.

. Results

.1. Physico-chemical characterisation

Table 1 shows the complete chemical analyses for the stud-


ed refractories. As expected, major constituents of both AZSaterials (AZS1 and AZS2) are Al2O3, SiO2 and ZrO2. SiO2

ontent is more than double in AZS2 than in AZS1 while Al2O3nd ZrO2 contents are close for both products. Main differences

ARTICLE IN PRESS+ModelJECS-9946; No. of Pages 16A. Garca-Prieto et al. / Journal of the European Ceramic Society xxx (2015) xxxxxx 5

Table 1Chemical analysis of the studied materials (wt.%).

wt% AZS 1 AZS 2 AS ASI S C

Al2O3 68.8 0.4 57.7 0.4 44.6 0.3 48.9 0.3 0.71 0.5 74.2 0.4SiO2 10.8 0.3 22.8 0.2 48.9 0.4 45.7 0.3 95.2 0.3 13.2 0.2Fe2O3 0.20 0.04 0.16 0.04 3.08 0.03 1.09 0.03 0.22 0.04 0.82 0.03K2O 0.038 0.005 0.61 0.05 1.32 0.05 0.16 0.06 0.11 0.05MgO


Fig. 2. Characteristic microstructural features of the aluminasilicazirconia materials. Polished surfaces. (a) AZS1. Porous alumina aggregates (grey), zirconsand (white) and fine alumina (grey) are observed. Reflected light optical microscopy micrograph. (b) AZS1. Detail of a porous aggregate. Reflected light opticalmicroscopy micrograph. (c) AZS2. Electrofused mullite aggregates surrounded by a mullite matrix (grey) with dispersed zircon flour (white). Reflected light opticalmicroscopy micrograph. (d) AZS2. Detail of an electrofused mullite aggregate with alumina particles embedded. Reflected light optical microscopy micrograph.( ey pam

ct

(tcd

3

sTs

osilica refractory, S, has the widest distribution from 5400 mdown to about 100 m. Aggregates for the AS material are

Table 4Parameters of the distributions of aggregate size in the studied materials.

Grain size (m) AZS 1 AZS 2 AS S C

Average 1709 1591 1227 1736 1393Standard deviation 597 694 539 923 732

e) AZS2. Detail of the matrix showing partially decomposed zircon flour (gricroscopy micrograph.

The silica brick (S) is composed of the silica polymorphsristobalite and tridymite with minor amount remnant quartz;races of wollastonite (CaSiO3) are also present in this brick.

The major phase in the conventional high-alumina castableC) is corundum, followed by mullite. The hydraulic phases ofhese types of refractories (CaAl2O4 and CaAl4O7) were alsolearly identified in the XRD pattern. Fluorite (CaF2) was alsoetected in this material.

.2. Microstructure


Condensed information of quantitative data for aggregateize distributions of the five dense materials is presented inable 4. Maximum aggregate sizes for S, C and AZS2 compo-itions are similar and the largest (5000 m). Minimum value

MMN

rticles surrounded by nanometric white zirconia particles). Scanning electron

f this parameter corresponds to the dense fireclay material (AS,4000 m) while for AZS1 it is intermediate (4400 m). The


aximum 4396 5178 3901 5370 5212inimum 634 623 457 91 315o. of particles analysed 643 729 567 563 943


Fig. 3. Characteristic microstructural features of the fireclay-based bricks. Reflected light optical micrographs of polished surfaces. (a) Superduty fireclay brick, AS.Mullite chamote aggregates in a mullite matrix. (b) Detail of the aggregate- matrix interface showing bonding. (c) Group 28 insulating brick, ASI. Spherical poresa

sa

fF

powadpa

dcbs(tgtmcmwa

m

o((clsg

(saC(t

bTRsftpCii

re observed. (d) Detail of the mullite matrix in figure (c).

maller than those of the AZS being the distribution centredround 1200 m.

Micrographs summarising the characteristic microstructuraleatures of the studied refractory materials are shown inigs. 25.

Microstructure of AZS1 is constituted by grey roundedorous aggregates (4400600 m, Table 4, Fig. 2a) in a matrixf fine and medium porous grey particles and dense roundedhite particles (180 m) (Fig. 2b). Significant porosity is

lso observed in the matrix. According to the crystalline phasesetected by XRD (Table 3) and the morphology and colour of thearticles in RLOM, the grey particles were identified as aluminand the dense white particles as zircon sand.

The microstructure of AZS2 (Fig. 2ce) is dominated byense grey aggregates (5000600 m, Table 4) constituted byolumnar grains and significant amounts of glassy phase at theoundaries. Clearer particles are observed in the interior ofome of the aggregates. Such features, together with the XRDTable 3) and chemical analyses (Table 1), allowed identifyinghese aggregates as electrofused mullite, with corundum andlass as secondary phases. Fine (


Fig. 4. Characteristic microstructural features of the standard silica brick, S.Polished surfaces. (a) General aspect of the microstructure. Reflected light opti-cal microscopy micrograph. (b) Detail showing the highly transformed silicagrains (cristobalite) and the tridymite grains embedded in the CaO-rich glassyphase formed in the matrix of the refractory. Scanning electron microscopymicrograph. (c) Reflected light optical micrograph showing the grain boundaryc

sbomfta

Fig. 5. Characteristic microstructural features of the high-alumina regularcastable, C. Polished surfaces. Reflected light optical microscopy micrographs.(a) General aspect of the microstructure. Coarse bauxite aggregates and mediumbrown corundum particles (angular) are observed. (b) Detail showing the sub-s(

3

ataASI. Intermediate range of values was obtained for AZS1, ASand S.

Table 5Youngs modulus (E and E0) and modulus of rupture (MOR) of the studied mate-rials. E tested in flexure by impact. E0 calculated from the MSA model.51 MORdetermined from 3-pb tests. Main crystalline phases detected by X-ray diffrac-tion are also shown: ZS = zircon, A = corundum, M = mullite, m-Z = baddeleyite,Q = quartz, C = cristobalite, CAC = calcium aluminates.

Material E (GPa) E0 (GPa) MOR (MPa) Crystalline phases

AZS 1 29 3 54 9.2 0.6 ZS + AAZS 2 46 2 110 14 2 M + ZS + m-ZAS 26 1 57 8.0 0.9 MASI 4 1 30 1.0 0.2 M

racking.

ubstructure formed by two main phases (Fig. 5b) typical ofauxite aggregates, in agreement with the crystalline phasesbserved by XRD (Table 3). They were formed by corundum,ullite and secondary Fe and Ti containing phases. Medium

raction was constituted of brown corundum. According to


he XRD (Table 3), the matrix was constituted by calciumluminates.

SC

tructure of the bauxite aggregates constituted by mullite (grey) and corundumwhite).

.3. Mechanical characterisation

Youngs modulus (E) and modulus of rupture (MOR) valuesre summarised in Table 5. Both parameters follow the samerend, being the largest for the AZS2 material and the high-lumina castable, C, and the lowest for the insulating material,


17 1 33 8.2 0.5 Q + C + A 50 4 91 13 2 A + CAC + others


Fig. 6. Characteristic Load (P)Displacement (d) curves. The dimensions of the beams tested were 150 mm 25mm 25 mm for (a) and (b) and200mm 40mm 40 mm for (c). The relative notch lengths were 0.25 for (a) and (c) and 0.50 for (b). (a) Group 28 insulating brick, ASI. For the same material,similar curves were obtained for different tests performed under the same conditions. (b) Aluminasilicazirconia brick, AZS2. Similar results were obtained forb rent snc

tp

tmcofmr

iwsbweo

ca

aswtnw

smnvci

oth control parameters. (c) High-alumina regular castable, C. Notice the appaontrol parameter.

Characteristic LoadDisplacement curves recorded duringhe fracture tests are plotted in Fig. 6 and values for the toughnessarameters are summarised in Tables 6 and 7.

A monotonous decrease of load with time was found in allests indicating that stable fracture was reached both in displace-

ent and in CMOD controlled tests. All LoadDisplacementurves showed a linear elastic region followed by a long tailf monotonous decreasing load for increasing displacementrom the maximum load. Just before reaching the maxi-um load, some of the curves showed a moderate non-linear

egion.The only case in which it was not possible to reach stabil-

ty using CMOD control at the experimental rate used in thisork was the combination of the largest specimen size and the

mallest notch ( = 0.25) (Fig. 6c). In this case, an apparent snap-ack was observed in the LoadDisplacement curves, which


ould indicate that the material presented brittle fracture. How-ver, fracture in displacement control was stable and, in fact, thebtained curves were coincident with those obtained in CMOD

lA

ap-back for specimens tested using the crack mouth opening displacement as

ontrol when the LoadDisplacement loop associated with thepparent snap-back present in the latter was relieved.

LoadDisplacement curves obtained for the same materialnd experimental conditions were similar (Fig. 6a) and, con-equently, variability of the calculated toughness parametersas relatively low (10% in most cases, Tables 6 and 7). For

he same material, specimen and span dimensions and relativeotch lengths, curves obtained using both control parametersere similar (Fig. 6b).As shown in Tables 6 and 7, for each material there are no

ignificant differences between the toughness parameters deter-ined using different specimen and span dimensions, relative

otch lengths and control parameters. Therefore, the averagealues obtained for relative notch lengths, = 0.5, in CMODontrolled tests (Table 6) are plotted in Fig. 7 to facilitate compar-son. For all materials, values of the energy for crack initiation,


nbt, are lower than those of the work of fracture, wof. Theargest value of the toughness ratio corresponds to materialZS1, followed by C.


Fig. 7. Average toughness parameters for the studied materials calculated from the values recorded during stable tests ( = 0.50, crack mouth opening displacementcontrolled tests, Table 6). (a) Critical stress intensity factor in mode I, KIC. (b) Work of fracture, wof. (c) Energy for crack initiation, nbt. (d) Toughness ratio,wof/nbt.

v(lftsmb

mbbsorf(pFa

4

4

fios

FHoaca

cnf

Materials AZS2 and C present the maximum KICalues (1 MPa m1/2), followed by AS, AZS1 and S0.50.7 MPa m1/2); the lowest value corresponds to the insu-ating firebrick (ASI) (Fig. 7a). The work of fracture does notollow the same trend. The high-alumina castable (C), presentshe highest value, followed by AZS1. AZS2 and AS presentimilar values, which are higher (40%) than those of the silicaaterial, S. The lowest value is again found for the insulating

rick, ASI (Fig. 7b).Characteristic fracture surfaces are shown in Fig. 8. Both AZS

aterials presented tortuous fracture even though the aggregatesehaved differently. Most aggregates in AZS1 were traversedy the crack (Fig. 8a) whereas most aggregates of AZS2 wereurrounded by the main crack (Fig. 8b). Fracture in the matrixf the dense fireclay and insulating materials (AS and ASI) wasather flat and the main crack traversed the aggregates in theormer (Fig. 8c and d). The fracture surface of the silica materialS, Fig. 8e) was the flattest with the main crack traversing allarticles. The fracture surface of the high-alumina castable (C,


ig. 8f), presented opposite features, as it was tortuous and mostggregates were surrounded by the main crack.

rg

. Discussion

.1. Microstructure

The nature and estimated relative amounts of phases identi-ed by XRD (Table 3) were consistent with the main constituentsf the refractories according to their chemical and micro-tructural analysis (Table 1, Figs. 25).

In the AZS refractories appeared the typical zircon impurities,e2O3 and TiO2 in the form of ilmenite and rutile, and Y2O3 andfO2 in solid solution. The presence of significant impuritiesf Na2O + K2O in AZS1 material is attributed to the porouslumina aggregates obtained by Bayer process. Very little alkaliontent was detected in AZS2 which had electrofused mulliteggregates (Tables 1 and 3, Fig. 2).

Both dense fireclay and insulating materials, AS and ASI,ontained significant amounts of the typical impurities for alumi-osilicate natural raw materials: iron oxides (assumed as Fe2O3or the chemical analyses), TiO2, CaO and MgO. The presence of


elatively large amounts of K2O compared to other alkalis sug-ests that these materials were manufactured with raw materials


F ens (A oup 2c

cF

SttciF

ipa

cudA

capFmAo

mAotttliquid formation in the matrix would be lower (1555 C) dueto the formation of transient liquid phases at the zircon-mulliteinterfaces (temperature of the ZrSiO4Al6Si2O13SiO2 eutectic

ig. 8. Characteristic fracture surfaces. Optical images for tested specimluminasilicazirconia brick, AZS2. (c) Superduty fireclay brick, AS. (d) Gr

astable, C.

ontaining K-feldspar (K,Na,Ca,Ba)(Si,Al)4O8 (Tables 1 and 3,ig. 3).

Material S was a conventional silica brick (9597 wt.%iO2, 2.53.5 wt.% CaO), well converted into cristobalite and

ridymite. Wollastonite (CaSiO3) probably comes from the reac-ion of lime with silica.46 It should be pointed out that a lowontent of CaO (typically 24 wt.%) may be used as bindern silica bricks without loss of refractoriness (Tables 1 and 3,ig. 4).

The common impurities of bauxites and brown corundum,ron oxides (Fe2O3 in the chemical analyses) and TiO2, wereresent in the high-alumina castable, C, due to the nature of theggregates (Tables 1 and 3, Fig. 5).

The phosphorous (assumed as P2O5 for calculations in thehemical analyses) present in most materials is attributed to these of H3PO4 (phosphoric acid), Al(H2PO4)3 (aluminium dihy-rogen phosphate) or sodium polyphosphate as additives. In factlPO4 was identified in AZS1, AS and ASI.As a first approach for analysing the expected phases in

omplex materials as a function of temperature, simplifiedverage compositions considering only the three major com-onents can be plotted in ternary phase equilibrium diagrams.ig. 9 shows simplified average compositions of both AZS


aterials (AZS1 and AZS2) inside the ternary phase diagraml2O3ZrO2SiO2.47 AZS1 lies in the Al2O3 primary fieldf the ternary system while AZS2 is located in the primary

150mm 25 mm 25 mm). (a) Aluminasilicazirconia brick, AZS1. (b)8 insulating brick, ASI. (e) Standard silica brick, S. (f) High-alumina regular

ullite field of the ternary system and in the binary systeml6Si2O13ZrO2. Taking into account the average compositionf these refractories both should form stable liquid phases atemperatures higher than 1750 C (close to the temperatures ofhe ZrO2Al2O3Al6Si2O13 and ZrO2Al6Si2O13 eutectics ofhe ternary and binary systems). However, temperature for first


Fig. 9. Phase equilibrium diagram of the system Al2O3ZrO2SiO2.47

Please cite

this article

in press

as: G

arca-Prieto A

, et

al. Influence

of m

icrostructural characteristics

on fracture

toughness of

refractory m

aterials.J

Eur

Ceram

Soc (2015),

http://dx.doi.org/10.1016/j.jeurceramsoc.2014.12.020

AR

TIC

LE

IN P

RE

SS

+Model

JEC

S-9946;

No.

of Pages

16

12

A.

Garca-P

rieto et

al. /

Journal of

the E

uropean C

eramic

Society xxx

(2015) xxxxxx

Table 6Toughness parameters determined using CMOD controlled tests. KIC is the stress intensity factor in mode I calculated from the maximum load point, in MPa m1/2; nbt is the energy for crack initiation in J/m2; wofis the work of fracture in J/m2 and wof/nbt is the toughness ratio.

150 mm 25 mm 25 mm 200 mm 40 mm 40 mm = 0.25 = 0.50 = 0.25

KIC nbt wof wof/nbt KIC nbt wof wof/nbt KIC nbt wof wof/nbt

AZS1 0.69 0.05 8 3 76 7 9 2 0.57 0.09 6 1 72 10 14 4AZS2 1.0 0.1 10 5 60 7 6.0 0.5 1.0 0.2 10 3 52 4 6 3AS 0.74 0.05 10 1 64 8 6.3 0.2 0.73 0.03 10 1 52 3 5.1 0.1 0.80 0.07 12 2 72 7 6.1 0.5ASI 0.11 0.01 1.4 0.2 8 1 5.7 0.8 0.17 0.04 4 1 9 1 3 1 0.081 0.007 0.8 0.1 50.0 0.6 6.3 0.4S 0.49 0.02 7 0.5 33 4 4.7 0.7C 1.1 0.2 12 4 112 17 9 2 1.2 0.3 14 4 110 13 9 3 1.3 0.14 18 3 130 13 7.3 0.9

Table 7Toughness parameters determined in displacement controlled tests. KIC is the stress intensity factor in mode I calculated from the maximum load point, in MPa m1/2; nbt is the energy for crack initiation in J/m2;wof is the work of fracture in J/m2 and wof/nbt is the toughness ratio.

150 mm 25 mm 25 mm 200 mm 40 mm 40 mm = 0.25 = 0.50 = 0.25

KIC nbt wof wof/nbt KIC nbt wof wof/nbt KIC nbt wof wof/nbt

AZS1 0.61 0.03 6 1 75 12 11 1 0.62 0.03 6.5 0.5 75 14 11 1AZS2 1.1 12 66 5.24 1.05 0.08 11 2 62 4 5.44 0.87AS 0.68 0.05 8 1 65 8 7.52 0.65 0.69 0.04 9 1 69 7 7.7 0.1ASIS 0.72 15 42 2.8 0.53 0.05 7 1 35 2 4.37 0.85C 1.3 0.2 16 4 126 18 7.9 0.5 1.2 0.1 13.5 2.5 136 15 10 0.6

ARTICLE IN PRESS+ModelJECS-9946; No. of Pages 16A. Garca-Prieto et al. / Journal of the Europe

F

poo

stcZcacsA

lratctoa

itcla1amfpsb

t

sltbo

4

rtbmv

tc4sia

eTtbt(rct

baEmaccatctc

actsmst

ig. 10. Phase equilibrium diagram of the system Al2O3SiO2TiO2.49

oint of the ternary system). The presence of impurities as TiO2r iron could lower this temperature (1450 C for iron,48 1450r 1500 C for TiO2).47

Considering the chemical (Table 1), the mineralogical analy-is (Table 3) and the microstructure (Fig. 2a and b) of AZS1,his material was formulated with aggregates of porous cal-ined alumina (73 wt.%) and the source of ZrO2 was a purerSiO4 sand (27 wt.%). No evidence of reaction betweenorundum and zircon was observed. The porosity of the bricknd the little or no reaction of ZrSiO4 with the finest parti-les of Al2O3 present in the matrix, suggest a temperature ofintering lower than 1450 C (the lowest invariant point of thel2O3ZrO2SiO2TiO2 system).47

AZS2 (Tables 1 and 3, Fig. 2ce) was fabricated with70 wt.% of electrofused mullite. Electrofusion originates mul-

ite of composition 2:1, in agreement with the presence of someemnant corundum inside the aggregates (Fig. 2d). From theverage composition and the microstructure of this material andhe composition of the aggregates (mullite + alumina), it can beoncluded that the source of ZrO2 was 30 wt.% of angular par-icles of zircon flour of sizes smaller than 50 m. The presencef glass at the zircon/alumina interfaces (Fig. 2e) would indicate

sintering temperature higher than 1450 C47 (Fig. 9).The simplified average compositions of the fireclay and the

nsulating materials (AS and ASI, Table 1) are displayed insidehe ternary phase diagram Al2O3SiO2TiO2 in Fig. 10.49 Theompositions of both materials lie in the primary field of mul-ite in the binary system Al6Si2O13SiO2 (eutectic at 1595 C)nd in the ternary system Al6Si2O13SiO2Al2TiO5 (eutectic at480 C). Therefore, both materials would form liquid phasest temperatures higher than 1480 C. Taking into account theicrostructures and compositions of these materials, both were

ormulated by mixing a kaolinitic chamote (aggregates) withlastic refractory clay. As the matrix of the fireclay brick had


imilar composition as that of the chamote aggregates, a goodonding between the aggregates and the matrix was observed.

The porosity of the insulating firebrick (ASI) was due tohe addition of pore generators to the fireclay formulation;

u(af


everal processes for pore generation are well reported in theiterature.50 As described above, the average aggregate size inhe dense fireclay material (1200 m, Table 4) was compara-le to the diameters of the main macrostructural characteristicsf the insulating firebrick which are the pores (Fig. 3c).

.2. Mechanical behaviour

In order to correlate the microstructural characteristics of theefractory materials with the Youngs modulus, E, obtained inhe present study, values for the fully dense materials, E0, haveeen calculated using the exponential equation derived from theinimum solid area models (MSA)51 using the experimental E

alues (Table 5) and the true porosities (Table 2).Calculated E0 values for all materials (Table 5) are lower than

hose expected for crystalline bonded zero porosity materialsonstituted by the major crystalline phases detected (E0 220,00, 280, 7090 GPa for mullite,52 corundum,53 zircon53 andilica polymorphs,54 respectively). This is a typical character-stic of refractories due to the weak bonding between the largend inert aggregates and the fine matrix, as discussed below.

E0 for the silica brick is about one half of what could bexpected for a mixture of silica polymorphs (7090 GPa).54his fact can be explained considering the microstructure of

his material, formed by large cristobalite particles surroundedy a transformation zone of tridymite + glass (Fig. 4). The largehermal expansion mismatch between the two crystalline phases 10 and 21 106 C1 for cristobalite and tridymite,espectively)54 would lead to grain boundary cracking duringooling from the fabrication temperature. In fact, in the RLOM,he boundaries present the typical dark colour of cracks (Fig. 4c).

No cracks were observed in the AZS2 brick constitutedy phases with similar thermal expansion coefficients ( 4.1nd 4.5 106 C1, for mullite52 and zircon,55 respectively).ven though there is a significant thermal expansion mis-atch between the major phases present in AZS1 (zircon

nd corundum-alumina, 4.5 55 and 8.4 106 C1 52) noracks were observed in AZS1 specimens (Fig. 2a and b), whichan be explained by the low stiffness of the porous aggregatesnd weakly bonded to the matrix. These two latter characteris-ics will be responsible for the extremely low E0 of AZS1 asompared to AZS2, contrary to what could be expected fromhe major crystalline phases present in the materials (AZS1:orundum + zircon, AZS2: mullite + zircon + baddeleyite).

There are also differences between E0 for the dense fireclaynd the insulating materials, which presented similar chemi-al composition and crystalline phases, which can be attributedo the presence of stiff aggregates in the dense material. Thetiffest constituents have a determining influence on the Youngsodulus values obtained from the response of specimens to the

mall and instantaneous deformations associated with the impactest.56

The relative performances of the materials in terms of mod-


lus of rupture (MOR) and Youngs modulus (E) were similarTable 5). This behaviour is characteristic of refractories withggregate sizes within the typical ranges (30006000 m). Dif-erences in the MOR of such materials are determined by E

ARTICLE IN PRESS+ModelJECS-9946; No. of Pages 161 urope

ber

mdiosStimr

tmafinassbm

atit

(

(

(

ftitamhcnm

oipwgt

4 A. Garca-Prieto et al. / Journal of the E

ecause they present only slightly differences in KIC, and theffect of the critical defect size is masked by the dispersion ofesults associated with the statistical character of strength.

The characteristics of the loaddisplacement curves withonotonous decrease of the load for increasing displacement

uring fracture are those corresponding to materials with ris-ng R-curve during fracture. This material characteristic allowsbtaining stable fracture in displacement control for sufficientlytiff and performing machines, as observed (Fig. 6b and c).uch fracture behaviour is reflected in the fact that values of

he energy for crack initiation, nbt, are lower than those averag-ng the whole fracture process, wof, as occurs for all the studied

aterials (Fig. 7b and c), and is highlighted by the toughnessatio values higher than 1 (Fig. 7d).

The creation or not of the fully developed process zone duringesting depends on the scale of the microstructural features of theaterial tested and the experimental set up. The fact that there

re no significant differences between the values of the work ofracture obtained using different notch sizes for any material,ndicates that, even for the smallest specimens with the largestotches used in this work (specimens of 25 mm 25 mm sectionnd = 0.50), the fracture area is sufficiently large to generate theteady state of fracture. As discussed in the introduction, rathermall specimen sizes relative to the microstructural features haveeen found adequate for SENB in 3-pb testing heterogeneousaterials as asphalt concrete.34,35

In order to highlight the influence of the microstructural char-cteristics on the mechanical parameters summarised in Fig. 7,he six materials have been compared in terms of their mechan-cal behaviour and their microstructural features, according tohe following considerations:

1) AZS1 and AZS2 were two AZS materials with aggregateswith similar size distributions (Table 4) and extremely dif-ferent physicochemical and microstructural characteristicsof the constituents. Aggregates were porous alumina forAZS1 and dense mullite for AZS2. The matrix in AZS1was formed by fine alumina particles and represented a smallpart of the total volume of the material in which a significantamount of medium size zircon particles was present (Fig. 2a-b). In AZS2 there was a high amount of matrix composedby mullite, zircon and zirconia particles (Fig. 2c-e). Aggre-gates were weakly linked to the matrix in AZS1 and stronglybonded to the matrix in AZS2. AZS1 presented KIC and wofvalues similar to those reported for alumina-zircon refracto-ries with similar phase composition32 (KIC = 0.62 MPa m1/2

and wof = 66 J/m2).The fracture path in the materials isdetermined by the characteristics of the aggregates, thematrix and the matrix-aggregate bonds. In this way, the weakporous aggregates present in AZS1 were easily traversed bythe crack whereas the dense ones in AZS2 were surrounded.The dense aggregates with higher strength present in AZS2impeded the initiation of fracture leading to KIC and nbt


values for this material higher than those for AZS1. Themedium size zircon particles in AZS1 were responsible forcrack arrest and deflection during propagation leading tohigher values of work of fracture than that of AZS2. As

iw

p

an Ceramic Society xxx (2015) xxxxxx

a consequence, the toughness ratio was much higher forAZS1.

2) AS and ASI had matrices of similar fireclay composition butvery different microstructural features of similar sizes. Themajor microstructural characteristics of these two materialswere chamote aggregates firmly bonded to the matrix in ASand pores in ASI (Figs. 3ad and 7ad).

AS presented nbt and wof values similar to the clas-sical ones reported by Nakayama for fireclay refractories4

(nbt = 10.6 J/m2, wof = 60 J/m2), however the nbt valuesfor AS were lower than those for mullite-rich laboratorymixes31 (nbt = 1825 J/m2) and lower than those reportedfor similar compositions with improved matrix due to theaddition of medium particles33 (nbt = 44 J/m2).

All toughness parameters, KIC, nbt and wof, of theporous material (ASI) were much lower (1520%) thanthose of the dense one (AS), as expected. The pores in ASIact as stress intensity sinks during fracture which leads toan increased fracture energy as compared to that for crackinitiation. Therefore, the toughness ratio for this materialwas about 60% of that of AS.

3) Composition and microstructure of S and C materialswere totally different. S was formed by particles of equalcomposition whereas C presented a well differentiatedcementitious matrix and sintered alumina aggregates. Thelarge differences in the toughness ratios of these materi-als suggest that the inelastic crack propagation processesof these two refractory types are markedly different,6 ascorresponds to the large differences in the microstructuralfeatures. The crack path was straight, traversing all grains,in the single-composition material S and highly tortuous inthe high-alumina castable C. In this latter, the aggregateswere surrounded by the cracks because they were tougherthan the matrix.

In general, work of fracture values for oxide refractories runrom about 30 J/m2 for the most brittle ones to 100120 J/m2 forhose with well-designed microstructures, as the castable stud-ed here.1 This material (C) presented wof values similar tohose reported for a series of high alumina castables (7090lumina wt.%, 110120 J/m2, 7 115 J/m2)22 with designedicrostructures for thermal shock. Those materials included

igh strength zirconia mullite aggregates which conferred theastables extremely high values and, consequently, lower tough-ess ratios (wof/nbt 24)7 than the value obtained here foraterial C.From the above discussion, it is clear that the characteristics

f the aggregates determine the toughness values for fracturenitiation, KIC and nbt (Fig. 7a and c). For similar crystallinehase composition (Table 3), alumina, the two extreme casesould be that of the castable, with well sintered alumina aggre-ates (Fig. 5a and b) and material AZS1 (Fig. 2a and b), in whichhe aggregates presented high levels of porosity. The insulat-


ng material, in which the aggregates were substituted by pores,ould be the lowest limit for this trend.Differently than in the case of fracture initiation, the

resence of microstructural features capable for crack arrest

ARTICLE IN PRESS+ModelJECS-9946; No. of Pages 16urope

agdbaShwt

ecta

5

pdnttas1t

ttahf

A

t4Sp

R

1

1

1

1

1

1

1

1

1

12

2

2

2

2

2

2

2

2

2

3

3


nd deflection is needed for high resistance to crack propa-ation, wof (Fig. 7b). The microstructure that most clearlyemonstrates this fact is that of material AZS1 (Fig. 2a and), with medium-sized dense particles dispersed in the matrixnd the highest toughness ratio (Fig. 7d). Materials AS and

are examples of the opposite trend. These materials, withomogeneous composition through the microstructure andell bonded microstructural constituents presented the lowest

oughness ratios of the dense materials.The effectiveness of the microstructural elements for tough-

ning is determined, not only by their nature, but also by theharacteristics of their bonding to the matrix. Material C is theypical case of strong aggregates weakly bonded to the matrixnd presents the highest wof of all studied materials (Fig. 7b).

. Conclusions

The capability of stable fracture tests of SENB tested in 3oint bending to characterise toughness of refractories has beenemonstrated. When carefully performed using the high stiff-ess and performing machines nowadays available it is possibleo establish displacement controlled tests as routine laboratoryests for stable fracture. For the typical microstructural char-cteristics of commercial refractories studied here, standardize specimens (150 mm 25 mm 25 mm) tested with span25 mm and relative notch length ( = 0.5) give differentiatedoughness values for different microstructures.

The main microstructural features that influence the resis-ance of materials to initiation of fracture are different fromhose that regulate crack propagation. The characteristics of theggregates determine toughness for crack initiation while, forigh values of work of fracture the presence of microstructuraleatures capable for crack arresting and deflection are needed.

cknowledgements

The authors acknowledge the financial support of CYTEDhrough the network ref. 312RT0453 and project MAT2013-8426-C2-1R; and the supplying of materials by INSERTEC.A. (Spain). Ana Garca-Prieto acknowledges the financial sup-ort of the JAE-CSIC fellowship program JAEPre 2010 00274.

eferences

1. Bradt RC. Fracture of refractories. In: Schacht CA, editor. Refractorieshandbook. New York: Marcel Dekker Inc; 2004. p. 1138.

2. Hasselman DPH. Figures-of-merit for the thermal stress resistance of high-temperature brittle materials: a review. Ceram Int 1978;4(4):14750.

3. Hasselman DPH. Thermal stress resistance of engineering ceramics. MaterSci Eng 1985;71:25164.

4. Nakayama J, Abe H, Bradt RC. Crack stability in the work-of-fracture test:refractory applications. J Am Ceram Soc 1981;64(11):6715.

5. Harmuth H, Rieder K, Krobath M, Tschegg E. Investigation of the nonlinear


fracture behaviour of ordinary ceramic refractory materials. Mater Sci EngA 1996;214:5361.

6. Homeny J, Darroudi T, Bradt RC. J-Integral measurements of the fractureof 50% alumina refractories. J Am Ceram Soc 1980;63(56):32631.

3


7. Salvini VR, Pandolfelli VC, Bradt RC. Extension of Hasselmans thermalshock theory for crack/microstructure interactions in refractories. Ceram Int2012;38(7):536975.

8. Harmuth H, Tscheg EK. A fracture mechanics approach for the developmentof refractory materials with reduced brittleness. Fatig Fract Eng Mater Struct1997;20(11):1585603.

9. Harmuth H, Bradt RC. Investigation of refractory brittleness by fracturemechanical and fractographic methods. Refract Man 2010:610.

0. Palmer GB, Baker G. A load-cycling technique for R-curve behaviour:application to a low cement refractory. Int J Fract 1993;62(3):23344.

1. Bueno S, Baudn C. Mechanical behaviour of structural ceramics. Bol SocEsp Ceram V 2007;46(3):10318.

2. Sakai M, Urashima K, Inagaki M. Energy principle of elastic-plastic fractureand its application to the fracture mechanics of a polycrystalline graphite. JAm Ceram Soc 1983;66(12):86874.

3. Nakayama J. Direct measurement of fracture energies of brittle heteroge-neous materials. J Am Ceram Soc 1965;48(11):5837.

4. Tattersall HG, Tappin G. The work of fracture and its measurement in metals,ceramics and other materials. J Mater Sci 1966;1(3):296301.

5. Davidge RW, Tappin G. The effective surface energy of brittle materials. JMater Sci 1968;3(2):16573.

6. Larson DR, Coppola JA, Hasselman DPH, Bradt RC. Fracture tough-ness and spalling behavior of high-Al2O3 refractories. J Am Ceram Soc1974;57(10):41721.

7. Chlup Z, Boccaccini DN, Leonelli C, Romagnoli M, Boccaccini AR.Fracture behaviour of refractory ceramics after cyclic thermal shock. Ceram-Silikaty 2006;50(4):24550.

8. Ko YC. Influence of the total fines content on the thermal shock damageresistance of Al2O3-spinel castables. Ceram Int 2001;27(5):5017.

9. Tschegg EK, Austria Patent Specification AT 390,328 (1986).0. Brhwiler E, Wittmann FH. The wedge splitting test, a new method

of performing stable fracture mechanics tests. Eng Fract Mec1990;35(1/2/3):11725.

1. Harmuth H. Stability of crack propagation associated with fractureenergy determined by wedge splitting specimen. Theor Appl Fract Mec1995;23(1):1038.

2. Ribeiro S, Rodrigues JA. The influence of microstructure on the maxi-mum load and fracture energy of refractory castables. Ceram Int 2010;36:26374.

3. Ribeiro S, Ribeiro DC, Souza Dias MB, Ribeiro Garcia GC, Bento dosSantos EM. Study of the fracture behavior of mortar and concretes withcrushed rock or pebble aggregates. J Mater Res 2011;14(1):4652.

4. Harmuth H, Manhart C. Fractographic investigations of ordinaryceramic refractory materials with reduced brittleness. Key Eng Mater2009;409:20915.

5. Huilin L, Guifang D. Application of wedge splitting test in refractories.Chinas Refract 2006;15(2):2730.

6. Brochen E, Dannert C. Thermo-mechanical characterisation of magnesia-carbon refractories by means of wedge splitting test under controlledatmosphere at high-temperature. In: Proceedings of 13th biennial worldwideconference on refractories, UNITECR 2013. 2013.

7. Jin S, Gruber D, Harmuth H. Determination of Youngs modulus, fractureenergy and tensile strength of refractories by inverse estimation of a wedgesplitting procedure. Eng Frac Mec 2014;116:22836.

8. Skiera E, Malzbender J, Mnch J, Dudczig S, Aneziris CG, SteinbrechRW. Controlled crack propagation experiments with a novel alumina-basedrefractory. Adv Eng Mater 2012;14(4):24854.

9. Miyaji DY, Otofuji CZ, de Anchieta Rodrigues J. The loaddisplacementcurve of steady crack propagation: an interesting source of information forpredicting the thermal shock damage of refractories. In: Proceedings of 13thbiennial worldwide conference on refractories, UNITECR 2013. 2013.

0. Malzbender J, Skiera E, Mnch J. A simple methodology to visualize crackpropagation for ceramic materials. J Mater Sci 2014;49:4036.

1. Aksel C. The effect of mullite on the mechanical properties and ther-


mal shock behaviour of aluminamullite refractory materials. Ceram Int2003;29:1838.

2. Rendtorff NM, Aglietti E. Mechanical and thermal shock behavior of refrac-tory materials for glass feeders. Mater Sci Eng A 2010;527:38407.
dx.doi.org/10.1016/j.jeurceramsoc.2014.12.020http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0285http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0285http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0285http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0285http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0285http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0285http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0285http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0285http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0285http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0285http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0285http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0285http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0285http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0285http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0285http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0285http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0285http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0285http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0285http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0285http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0285http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0285http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0285http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0285http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0285http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0285http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0285http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0290http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0290http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0290http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0290http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0290http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0290http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0290http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0290http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0290http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0290http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0290http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0290http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0290http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0290http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0290http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0290http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0290http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0290http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0290http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0290http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0290http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0290http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0290http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0290http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0290http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0290http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0295http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0295http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0295http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0295http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0295http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0295http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0295http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0295http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0295http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0295http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0295http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0295http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0295http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0295http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0295http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0295http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0300http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0300http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0300http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0300http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0300http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0300http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0300http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0300http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0300http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0300http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0300http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0300http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0300http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0300http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0300http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0300http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0300http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0300http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0300http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0300http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0300http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0300http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0300http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0300http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0300http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0305http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0305http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0305http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0305http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0305http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0305http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0305http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0305http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0305http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0305http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0305http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0305http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0305http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0305http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0305http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0305http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0305http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0305http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0305http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0305http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0305http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0305http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0305http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0305http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0305http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0305http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0305http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0305http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0305http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0305http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0305http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0305http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0305http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0310http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0310http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0310http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0310http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0310http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0310http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0310http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0310http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0310http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0310http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0310http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0310http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0310http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0310http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0310http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0310http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0310http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0310http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0310http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0310http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0310http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0310http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0310http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0310http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0310http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0310http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0310http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0310http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0310http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0310http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0315http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0315http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0315http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0315http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0315http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0315http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0315http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0315http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0315http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0315http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0315http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0315http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0315http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0315http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0315http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0315http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0315http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0315http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0315http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0315http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0315http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0315http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0315http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0315http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0315http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0315http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0315http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0320http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0320http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0320http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0320http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0320http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0320http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0320http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0320http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0320http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0320http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0320http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0320http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0320http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0320http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0320http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0320http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0320http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0320http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0320http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0320http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0320http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0320http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0320http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0320http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0320http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0320http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0320http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0320http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0320http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0320http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0320http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0320http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0320http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0325http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0325http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0325http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0325http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0325http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0325http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0325http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0325http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0325http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0325http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0325http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0325http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0325http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0325http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0325http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0325http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0325http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0325http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0325http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0325http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0325http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0325http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0325http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0330http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0330http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0330http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0330http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0330http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0330http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0330http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0330http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0330http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0330http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0330http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0330http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0330http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0330http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0330http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0330http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0330http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0330http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0330http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0330http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0330http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0330http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0330http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0330http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0330http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0330http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0330http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0330http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0330http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0330http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0330http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0330http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0330http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0330http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0335http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0335http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0335http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0335http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0335http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0335http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0335http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0335http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0335http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0335http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0335http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0335http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0335http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0335http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0335http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0335http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0335http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0335http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0335http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0335http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0335http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0340http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0340http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0340http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0340http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0340http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0340http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0340http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0340http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0340http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0340http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0340http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0340http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0340http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0340http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0340http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0340http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0340http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0340http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0340http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0340http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0340http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0340http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0340http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0340http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0340http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0340http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0340http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0340http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0340http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0340http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0340http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0340http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0340http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0340http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0345http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0345http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0345http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0345http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0345http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0345http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0345http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0345http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0345http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0345http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0345http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0345http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0345http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0345http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0345http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0345http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0345http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0345http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0345http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0345http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0345http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0345http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0345http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0350http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0350http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0350http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0350http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0350http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0350http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0350http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0350http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0350http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0350http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0350http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0350http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0350http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0350http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0350http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0350http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0350http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0350http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0350http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0350http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0350http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0350http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0350http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0350http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0350http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0350http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0350http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0350http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0355http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0355http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0355http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0355http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0355http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0355http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0355http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0355http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0355http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0355http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0355http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0355http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0355http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0355http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0355http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0355http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0355http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0355http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0355http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0355http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0355http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0355http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0355http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0355http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0355http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0355http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0355http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0355http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0360http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0360http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0360http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0360http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0360http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0360http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0360http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0360http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0360http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0360http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0360http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0360http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0360http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0360http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0360http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0360http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0360http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0360http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0360http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0360http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0360http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0360http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0360http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0360http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0360http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0360http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0360http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0360http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0360http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0360http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0360http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0360http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0365http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0365http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0365http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0365http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0365http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0365http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0365http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0365http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0365http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0365http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0365http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0365http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0365http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0365http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0365http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0365http://refhub.elsevier.com/S0955-2219(14)00677-3/sbref0365http://refhub.elsevier.com/S0955-

2015-GARCIA-PRIETO-Influence of Microstructural Characteristics on Fracture Toughness of Refractory...

Documents

Transcript of 2015-GARCIA-PRIETO-Influence of Microstructural Characteristics on Fracture Toughness of Refractory...