dt test on mulllite

8/2/2019 DT Test on Mulllite

1/5

Journal of the European Ceramic Society 17 (1997) 85-890 1996 Elsevier Science LimitedPrinted in Great Britain. All rights reservedPII: SO955-2219(96)00077-5 0955-2219/97/%17.00

Crack Propa,gation Behaviour in Mullite at HighTemperatures by Double-Torsion TechniqueH. Rhanim,pb C. Olagnon, G. Fantozzi & R. TorrecillasLaboratoire GEMPPM URA CNRS 341, bat. 502, INSA de Lyon, 20 bd A. Einstein,6962 1 Villeurbanne Cedex, FranceFaculte des Sciences dE1 Jadida, Departement de Physique, BP 20, El Jadida, MoroccoInstituto National de1 Carbon, La Corredoria, s/n. Apartado 73, 33080 Oviedo, Spain(Received 20 November 1995; revised version received 25 March 1996; accepted 2 April 1996)

AbstractThe double-torsion technique has been used t o stud ysubcriti cal crack grow th in mullit e at room tem per-at ure (RT) and above. The crack v elocity wasevaluated by measuring a change in crack lengthduring a prescribed tim e under a const ant appliedload. Threshold values of st ress int ensity factor,below w hich no propagat ion occurs, of 1.5 and1.6 A4Pa m w ere found at RT and high t empera-ture, respectiv ely. Considering a pow er law , mea-sured stress exponent s of 24, 35, 18 and 10 at RT ,1100, 1200 and 1300C w ere obt ained. A change ofcrack grow th m echanisms is observed around12OOC, w ith cleavage crack grow th occurringbelow 1200C and viscous int ergranular crackgrow th dominat ing at higher t emperat ures. 0 1996Elsevier Science Limited.

1 IntroductionRecently, dense polycrystalline mullite with a littleor nearly no glassy phase has become available bysintering of fine mullite powders. This materialshows the advantages of a relatively high flexuralstrength at high temperature and high creep resis-tance in addition to its low thermal expansion.2,3These characteristics makes mullite a promisingmaterial for high-temperature structural use.However, the crack growth behaviour of thismaterial has not been extensively investigated.

Most mechanisms of slow crack growth (SCG)in ceramics, such as the environmentally con-trolled stress corrosion process, were examined atroom temperature (RT) and only sparse SCG dataare available at high temperatures.4 Hot-pressedsilicon nitride (HPSN), however, is an example ofwell-documented materi,al that has largely been

investigated at elevated temperatures.5,6 By consid-ering the power relation: V = BK,, Evans andWiederhorn found that crack growth at lowvelocities is characterized by a constant slope m = 10and that crack velocity increases with tempera-ture. An activation energy value of about 920 kJmall was found. The mechanism of crack growthin this region was attributed to viscous flow. Morerecently, Lange7 and Kingery et al8 investigatedthe crack growth mechanisms of HPSN in an inertatmosphere. They suggested that SCG in polycrys-talline ceramics at high temperatures dependsprincipally on grain boundary phase as well as onenvironment. On the other hand, Knight andPage have studied SCG of porous ceramics suchas CaO-stabilized ZrO, using the double-torsiontechnique at temperatures in the range lOlO-1400C. They observed three distinct propagationstages. At low K, the behaviour can be describedby I = BK,, with m decreasing with increasingtemperature. In the second stage the velocity isconstant with increasing K,, and finally for Kr nearKr,-, V increases rapidly. The dominant mecha-nism of SCG appeared to be interparticle sliding(leading to pore rearrangement, pore linking andcrack extension).

In this paper the results of crack propagationtests conducted on a mullite ceramic at severaltemperatures using a double-torsion technique arepresented. The crack velocity has been measuredas a function of the applied stress intensity factorand the results are discussed in terms of crackgrowth mechanisms.

2 Experimental ProcedureSeveral methods can be applied to measure thecrack velocity via the double-torsion specimen.

85


2/5

86 H. Rhanim et al.The load relaxation method is the most frequentlyused because it is relatively rapid (one specimen issufficient in principle). When associated with cali-bration of the compliance with crack length it pre-sents the advantage of (at least theoretically)permitting a record of the full V-K, curve with asingle specimen, even in a closed furnace. But thismethod is sensitive to any temperature variationand it is therefore restricted to relatively highvelocities (~10~~ m ~8). In addition, it cannot beapplied when there is some plasticity. For thesereasons the constant load method, which is verysimple, less sensitive to temperature variation andtherefore relevant for lower velocities, has beenselected here. The main drawback is that the pro-cedure for recording a full V-K, diagram is time-consuming since one test gives only one point inthe V-K, diagram.2.1 MaterialThe mullite used in this work was made by fTMA(Spain), from a Baikowski powder (CA 193 CR,France). Green compacts of mullite obtained bycold-isostatic pressing under 200 MPa were sin-tered for 2 h at 1750C. A post heat treatment at1450C for 5 h was subsequently carried out toreduce the glassy phase content.

The density of the material measured byArchimedes method is about 95.5%. The micro-structure analysed by scanning electron microscopy(SEM) on polished and thermally etched surfaces(Fig. I) revealed two distributions of grain sizes:equiaxed grains with size of 14 pm and elongatedgrains with long axis lengths of about 7 pm. Sev-eral pores at multiple grain junctions can alsobe observed. Transmission electron micrographsrevealed the presence of glassy phase at three-grain junctions but since high resolution observa-tions were not conducted, no intergranular glassyfilm has been observed.

Fig. 1. SEM micrograph of thermally etched surface of mullite.

2.2 Preparation of specimensThe specimen dimensions were 20 X 40 X 2 mm3.A 10 mm long notch was sawn in the specimenafter grinding, polishing and annealing. A pre-crack was initiated at RT from the notch tip byloading the specimen at very low speed, in orderto obtain a sharp crack. It is important to notethat no groove has been used to guide the crack.

A constant applied load on a load-controlledtesting machine (Schenck Treble, RM-T 25) wasmaintained for prescribed times and the crackvelocity V = da/At was calculated by measuringthe change in crack length by optical microscopyafter each test. Crack growth measurements wereconducted at RT, 1100, 1200 and 1300C. Thecrack length was measured by optical microscopyand the resolution damin was estimated as 10 pm.For relatively high crack velocities (>lO- m ss),the crack increments were in the range of about100 pm to 1 mm, and the minimum prescribedtimes were a few minutes. The maximum error onthe crack velocity measurement was thereforeabout 10%. For lower crack growth, the errormight be slightly higher. For representation ofthe threshold, when no crack advance wasobserved, an upper velocity limit was calculated asAalAt~

2.3 Analysis methodFor this double-torsion geometry, determinationof the stress intensity factor is based on the factthat the loading point displacement is propor-tional to the crack length and the applied loadp.S,lO

&=p w,I ld(1 + v)e2 cCl(elW)Wwhere W is the width of the specimen, W,,, isthe moment arm and @(e/W) is a calibratingfactor.

From eqn (1) it can be noted that the stressintensity factor K, for the double-torsion configu-ration depends only on the applied load and noton the crack length.

3 Results and Discussion3.1 Room temperatureA typical V-K1 diagram for mullite obtained atRT is shown in Fig. 2. Only the first propagationstage is observed in the crack velocity range inves-tigated. It should be noted that this stage has beenshown to be the most important for lifetime pre-diction.12 In this region the crack velocity rangedfrom lo- to 10e5 m s-l and the stress intensityfactor KI ranged from 1.5 to 2.3 MPa ml.


3/5

Crack propagation in mullit e at high tem peratureV(m/s)

K, (MPafi)Fig. 2. V-K, diagram for mullite at room temperature.

This diagram clearly shows a threshold stressintensity factor K,, below which no crack propa-gation was observed. This is very important andsuggests that, below this value, specimens of mulliteshould present unlimited lifetime at room temper-ature.

On the other hand, the V-K, curve may befitted by a power law: V = BK,. The values of B,m and K,, are given in Table 1.3.2 High temperaturesFigure 3 shows the different measurements obtainedat 1100C represented on a single diagram. Theprocedure was such that several crack propaga-tions were conducted on the same specimen(represented by the same symbol in Fig. 3). It canbe observed that for most of the fourth, fifth andsixth tests on the same specimen (see Fig. 3), thecrack length does not change after testing evenwhen higher stress intensity factors are applied.This phenomenon has lbeen attributed to crackhealing, which increases the resistance to crackgrowth. This behaviour was also observed byKubota et ~1.~ in their study on the fatiguebehaviour of mullite at 1200C. They found thatcrack growth under an applied stress was inter-rupted by crack healing;. In order to avoid thisproblem, a precrack at room temperature wasmade before each test at 1200 and 1300C.

The different V-K1 curves measured at hightemperatures are shown in Fig. 4. They exhibitsimilar characteristics to that obtained at RT with

1o-l2\,41 1 I I 1 I 11.6 1.6 2 2.2KI (MPafi)

Fig. 3. V-K, diagram for mullite at 1100C. The differentsymbols represent different specimens. The number near thesymbol represents the ith test conducted on a given specimenwithout an RT precrack. Arrows represent upper crack velocitylimit when no crack advance was observed.

a threshold stress intensity factor shifted to highervalues of Ki. It should be noted that the thresholdvalues are the same for the different temperaturesinvestigated. However, for 1200 and 1300C onecan observe the presence of a second region wherethe crack velocity increases very weakly withincreasing K, until K,,, where fast failure occurred.Table 1 gives the SCG parameters for the differenttemperatures.One can note that the SCG exponent mincreases from 23.6 to 34.7 when the temperatureincreases from RT to 1100C. However, the oppo-site tendency is observed for temperatures above1100C. Effectively, the exponent m falls from34.7 to 18 at 1100 and 12OOC, respectively. Thischange in SCG behaviour suggests a change in themechanism of crack growth around 1200C.Fields and Fuller14 have proposed a representationof crack velocity derived from creep maps, whereconstant velocity curves are plotted on a stressintensity factor-temperature diagram. This pre-sents the advantage of outlining the propagationmechanisms. The results have been represented onsuch a plot where the toughness has also beenincluded (Fig. 5). In the present case, it showsclearly that for a constant value of Ki the increasein temperature between 1100 and 1200C inducesa decrease of crack velocity, whilst above 1200Cthe crack velocity increases with increasing tem-perature. This can be understood by analysing the

Table 1. SCG parameters at different temperaturesTemperature(C)RT110012001300

K, (MPa ml)

1.511.651.66164

ml Bl m2 B2

23.6 34 x 10-1334.7 1.7 x lo-618 8.6 x 10-u 2.1 7.6 x 1O-710.1 5.5 x 10-s 2.2 8.8 x 10-s


4/5

88 H. Rhanim et al.velocity (rr&)

l o 3 I 1 , I I I, n I

Kkl.?OO~CK/c I1 0 - 1 01 0 " ' , I I ,I1 . 5 2 2 . 5 3 3 . 5

KI ( MP a d WFig. 4. V-K, diagram for mullite at different temperatures.

(4K, ( MP a d i $

2 . 5 -,,;Iy-f 5+t5--+ij900 1 0 0 0 1 1 0 0 1 2 0 0 1 3 0 0 1 4 0 0

T e mp e r a t u r e ( " C )Fig. 5. K,-T diagram for mullite at different crack velocities.

(b)different fracture surfaces by SEM, as shown inFig. 6. One can clearly note a change in SCGmechanisms; at 1100C [Fig. 6(a)] only cleavagecrack growth can be observed. At 12OOC,Fig. 6(b)shows two mechanisms of crack growth: cleavageof the elongated grains and intergranular modenear the equiaxed grains. On the other hand, at1300C we observe slow crack growth regionswith a distinct appearance, characterized by brightregions. A detailed analysis of the SCG regionshows that the nature and mode of fracture arecompletely intergranular [Fig. 6(c)]. This is attrib-uted to the decrease of the intergranular glassyphase viscosity at high temperature. The propaga-tion therefore arises by viscous intergranular crackgrowth.

These double-torsion results measured on macro-cracks have been compared with data obtainedunder static fatigue of bend bars without artificialflaws conducted on the same material, at 1200 and1300C (experimental details are given in Ref. 15).The results in terms of lifetime as a function ofthe applied stress are represented in Fig. 7. Somesimulations of those static fatigue experiments

Fig. 6. SEM micrographs of the fracture surface of a double-torsion specimen tested at (a) llOOC, (b) 1200C and (c)1300C.

have been conducted by integration of the double-torsion V-K, laws (taking into account the differ-ent stages). The results (Fig. 7) show that a fairagreement is obtained between the experiments


5/5

Crack propagation in mullit e at high temperature

T i me t o f a i l u r e s )I I I I 1 1 1 ,

l o 3 -l o * - b1 0 - - DT--- S F I

60 90 1 0 0 1 1 0 1 2 0 1 4 0 1 6 0S t r e s s ( MP a )

Fig. 7. Time to failure obtained by creep experiments andcompared with the simulation.

and the simulation, suggesting that the double-torsion method allows crack growth behaviour tobe predicted correctly.

4 ConclusionThe behaviour of slow crack propagation in mul-lite ceramic was investigated at room temperature,1100, 1200 and 13OOC, sing a double-torsion tech-nique at constant load. A threshold value of stressintensity factor, below which no propagationoccurs, was observed for all temperatures: it wasabout 1.5 and 1.6 MPa m12 at room temperatureand 1100C respectively, and was approximatelyconstant at higher temperatures. Considering apower law, the measured stress exponents of 24,35, 18 and 10 at RT, 1100, 1200 and 1300C wereobtained. A change of crack growth mechanismsis observed around 1200C: cleavage crack growthbelow 1200C and viscous intergranular crackgrowth at higher temperatures.

Dokko, P. C., Pask, J. A. & Mazdiyasni, K. S., Hightemperature mechanical properties of mullite under com-pression. J. Am. Ceram. Sot., 60(34) (1977) 150-155.Okamoto, Y., Creep deformation of polycrystallinemullite. J. Eur. Ceram. Sot ., 6 (1990) 161-168.Penty, R. A. & Hasselman, D. P. H., Creep kinetics ofhigh purity, ultrafine grain polycrystalline mullite. Mater.Res . Bull., 7 (1972) 1117-1124.Lawn, B., Fracture of Brittle Solids, 2nd Edn. CambridgeUniversity Press, 1993, 378 pp.

References1.

2.3.

4.5.

6.

Wiederhorn, S. M., Subcritical crack growth in ceramicsIn Fracture Mechanics of Ceramics, Vol. 2, Mi crostructure,Material and Applications, eds R. C. Bradt, D. P. H.Hasselman & F. F. Lange. Plenum Press, New York,1974, pp. 613-646.

I.

8.9.

10.

11

12.

13.

14.

15.

Evans, E. G. & Wiederhorn, S. M., Crack propagationand failure prediction in silicon nitride at elevated tem-perature. J. Mater. Sci., 9 (1974) 27&278.Lange, F. F., High temperature strength behavior of hotpressed S&N,: evidence for subcritical crack growth.J. Am. Ceram. Sot., 57 (1974) 84-87.Kingery, W. D., Bowen, H. K. & Uhlmann, D. R., Zntro-duction to Ceramics. Wiley and Sons, New York, 1976.Knight, J. C. & Page, T. F., Mechanical properties ofhighly porous ceramics. Part II: slow crack growth andcreep. Br. Ceram. Trans. J., 85 (1986) 66-75.Evans, E. G., A method for evaluating the time depen-dent failure characteristics of brittle materials and itsapplication to polycrystalline alumina. J. Mater. Sci.,7(10) (1972) 1137-1146.Plekta, B. J., Fuller, E. R. & Koepke, B. G., An evalua-tion of double torsion testing-experimental. In FractureMechanics Applied to Brittle Materials, Proc. 11th Symp.Fracture Mechanics, Part II, ed. S. W. Freiman. ASTMSTP 678. American Society for Testing and Materials,Philadelphia, PA, 1979, pp. 19-38.Sudreau, F., Olagnon, C. & Fantozzi, G., Lifetime pre-diction of ceramics: importance of the test method.Ceram. Zn t., 20 (1994) 125-135.Kubota, Y., Ashizuka, M. & Ishida, E., Static fatiguefracture toughness of mullite ceramics at 12OOC.J. Ceram. Sot. Jpn., 102(9) (1994) 805-809.Fields, R. I. & Fuller, Jr, E. R., Crack growth mecha-nism maps. In Adv ances in Fracture Research, Vol. 3, ed.D. Francois. 1981, pp. 1313-1322.Rhanim, H., Olagnon, C., Fantozzi, G. & Torrecillas, R.,Experimental characterisation of high temperature creepresistance of mullite. Ceram. Znt., submitted.

dt test on mulllite

Documents