Sunteți pe pagina 1din 13

Int J Fract DOI 10.

1007/s10704-008-9251-1 ORIGINAL PAPER

Scaling properties of slow fracture in glass: from deterministic to irregular topography


Moiss Hinojosa Edgar Reyes-Melo Claudia Guerra Virgilio Gonzlez Ubaldo Ortiz

Received: 4 September 2007 / Accepted: 17 July 2008 Springer Science+Business Media B.V. 2008

Abstract In this work we discuss the morphology and self-afne properties of the slowfracture surfaces of soda-lime glass obtained by a bending process under the effect of applied water vapor. The fractographic analysis showed the presence of secondary cracks in the mirror zone, whereas in the misthackle region step-like morphologies were observed and over them we found ne undulations. The self-afne analysis, performed by two methods, showed the existence of two different statistical distributions for the roughness exponent, . At the beginning of the mirror zone = 0.5, in the misthackle region we detected the same value for ne length scales, whereas at large length scales we observed = 0.8. This scenario may be described by a qualitative model in which the deterministic mirrormisthackle pattern coexists with an irregular topography, the two observed regimes are thus characterized by two different roughness exponents, with the 0.5 value dominating at low-speed/ne-scales and the 0.8 value governing the high-speed/large-scales regimes. Keywords Fracture surfaces Glass Self-afnity Roughness exponent

1 Introduction The study of the mechanisms of crack propagation in all kind of materials requires taking into account the scaling behaviour of the fracture process, which manifest at all relevant scales,

M. Hinojosa (B E. Reyes-Melo C. Guerra V. Gonzlez U. Ortiz ) Facultad de Ingeniera Mecnica y Elctrica, Universidad Autnoma de Nuevo Len, Ave. Universidad, S/N, Ciudad Universitaria, C.P. 66450, San Nicols de los Garza, NL, Mexico e-mail: hinojosa@gama.me.uanl.mx Present Address: C. Guerra Fracture Group: DSM/DRECAM/SPCSI, CEA Saclay, 91191 Gif-sur-Yvette, France e-mail: claudia-maribel.guerra-amaro@cea.fr E. Reyes-Melo e-mail: mreyes@gama.me.uanl.mx

123

M. Hinojosa et al.

from the atomic to the macroscopic levels. At very small scales, of the order of atomic size, the fracture process is dened as the sequential breaking of chemical bonds. However once the rst bond breaks, it is not clear what sequence of events occurs during the crack front propagation, producing fracture surfaces. In addition the exact nature of the bond rupture process has not been identied. Examining surfaces of broken samples by different experimental techniques, it has been reported (Mandelbrot et al. 1984; Bouchaud et al. 1990; Mly et al. 1992; Bouchaud 1997) that fracture surfaces manifest statistical invariance through an afne transformation: (x, y, z) bx, by, b z , where is the Hurst or roughness exponent which quanties the roughness of the surfaces. This parameter has been evaluated by a variety of methods (Schmittbuhl et al. 1995; Gonzlez et al. 2002; ystein Haaving Bakke and Hansen 2007) knowing that for a self-afne surface the typical height h (r ) scales as r . In this context fracture surfaces are self-afne objects characterized in rst approximation by two parameters: the Hurst exponent ( ) and the correlation length ( ). When 1, the roughness decreases, and the surface has a very at morphology, and when 0 the surface roughness increases. On the other hand, is the characteristic length limiting the self-afne behaviour. In 1990 Bouchaud et al. observed that the value of the roughness exponent is surprisingly universal = 0.8 weakly dependent on the nature of the material and on the failure mode. This scenario was then tested in a large numbers of materials all of them fractured under conditions of rapid crack propagation and then analysed by several methods (Mly et al. 1992; Hinojosa et al. 2000; Hinojosa and Aldaco 2002; Hinojosa et al. 2004; Guerrero et al. 2002; Nio et al. 2006). The results found were in agreement with this conjecture. However, in 1994, Millman et al. observed a roughness exponent = 0.4 on fracture surfaces of tungsten single crystal analysed by scanning tunnelling microscopy at nanometer scale, which seemed to question the universal value for the roughness exponent. After these results, some workers (Daguier et al. 1997; Bouchaud 1997, 2003) analysed surfaces of two brittle materials: Ti3 -Al based super 2 intermetallic alloy and soda-lime silica glass. The rst material was fractured in fatigue test and the second one in stress corrosion at velocities around 2 109 8.3 106 m/s. In these experiments they observed two roughness exponents, one at small scales characterized by a value of = 0.5 and at large scales the universal roughness exponents = 0.8, in both materials. In addition, the crossover c was associated to the velocity of crack front propagation; c diverges when the crack velocity vanishes (quasi-static fracture process) and the self-afne regime with 0.8 roughness exponent could not be observed. On the other hand when the velocity of the crack front corresponds to a rapid fracture process, the crossover length gets to very small values and the value = 0.5 is not detected, as it is shown in Fig. 1. The roughness exponent = 0.5 was observed recently on fracture surfaces of soda-lime glass (Guerra-Amaro et al. 2007) broken in bending test under low velocities of propagation. Recent experiments in silica glass, aluminum alloy, mortar and wood at different length scales have shown that the fracture surfaces exhibit anisotropic scaling properties similar to the Family-Vicseck scaling (Ponson et al. 2006a, b). These surfaces are characterized by three critical exponents independent to some extent of the considered material, the loading condition and the crack growth velocity. The roughness exponent was found with a value of = 0.75 along the crack propagation direction, the growth exponent = 0.6 along the crack propagation and nally the dynamic exponent z = 1.25. In addition, several models were derived aiming at understanding the origin of the scaling properties of the fracture surfaces. In 1993 Bouchaud et al. proposed that the fracture surfaces are the trace left behind by a crack front moving through randomly distributed microstructural obstacles, the dynamics of which is described by a phenomenological

123

Scaling properties of slow fracture in glass Fig. 1 Scheme of the scaling behavior of a fracture surface exhibiting two roughness exponents

nonlinear Langevin equation. Ramanathan et al. in 1997a used linear elastic fracture mechanics (LEFM) to derive a linear nonlocal Langevin equation within elastostatic and elastodynamic approximation. Some years after, in 2003, Hansen and Schmittbuhl (2003) proposed that the origin of the universal scaling properties involves a damage coalescence process described by a stress-weighted percolation phenomenon in a self-generated quadratic damage gradient. More recently, in 2006 Bonamy et al.(2006), using LEFM proposed a description of the crack roughness development as an elastic string with no local interactions creeping in 2D random media, the spatial coordinate along the crack globally grows playing the role of time. In this model Bonamy et al. predicts two series of three critical exponents, below the size of the process zone: = 0.75, = 0.6 and z = 1.25, above the size of the process zone = 0.4, = 0.5 and z = 0.8. Though different workers (Bouchaud 1997, 2003; Daguier et al. 1997) have suggested that these roughness exponents are universal, this has been a subject of controversy (Milman et al. 1994; Balankin 1997). It is also true that the evidence accumulated through the years shows that there are deviations from these values which are statistically significant, this has been observed particularly in polymers (Guerrero et al. 2002; Hinojosa et al. 2004, 2005), in this case the deviations from the roughness exponent = 0.8, have been linked to rupture of secondary bonds resulting in crazing and viscoelastic effects. Recently, it was suggested (Hinojosa et al. 2005) that the self-afne character manifest itself at subnanometric scales and that the origin of the behavior should be linked to self-afne properties of the electronic structure. Unfortunately, no theoretical model can reproduce quantitatively the scaling properties of fracture surfaces observed experimentally. The model proposed by Bouchaud et al. (1993) predicts the existence of two characteristic roughness exponents with a crossover length diverging as the crack velocity vanishes. On the other hand, the models proposed by Ramanathan et al. (1997b) and Bonamy et al. (2006) predict a roughness exponent independent of the crack growth velocity. In both cases, no clear relationship between the microstructure and the roughness exponent has been obtained. On the other hand, it was shown that, for slow fracture process, the crack front progresses in a very jerky intermittent manner (Mly et al. 2006; Marchenko et al. 2006) because the crack front is just able to free itself from the pinning microstructural obstacles. For the particular case of silicates glass, a slow fracture process can be induced using water or water vapor as a chemical agent for sub-critical crack growth. In the absence of water, cracks either do not grow or require a higher stress than in the presence of water (Wiederhorn 1967; Zhu et al.

123

M. Hinojosa et al.

2005; Guerra-Amaro et al. 2007). Wiederhorn (1967) was the rst to quantify the sub-critical growth of macroscopic cracks; in addition he established several characteristics of the effect of water on crack motion in glass. Three stages are commonly distinguished: Stage Iultra-slow crack growthIn this regime the crack growth velocity is dictated by the kinetic of the chemical reaction between the water molecules and the stretched Si-O bonds. The crack growth velocity increases exponentially with the energy release rate. Stage IIintermediate velocity regimeIn this region, the crack growth velocity is limited by water transport up to the crack tip. In this regime, the crack velocity exhibits a plateau and does not depend on the load anymore. Stage IIIhigh velocity regimeThe crack velocity increases very rapidly with the energy release rate, and is independent of the presence of water.

Regardless of the presence or absence of water, the sequential breaking of chemical bonds produces fracture surfaces that exhibit a characteristic pattern that is traditionally described as having three distinctive zones in the vicinity of the fracture origin (Mecholsky et al. 1978): a smooth morphology surrounding the fracture origin called the mirror zone. A small band of rougher surface surrounding the mirror region, known as the mist zone which is followed by the hackle zone, composed of large irregularly oriented facets which are separated by steps aligned parallel to the main direction of crack propagation. In some cases, when the fractured sample has very great dimensions, beyond the hackle zone appears another more irregular region called the branching or bifurcation zone. Other morphological features that can be observed are for example the Wallner lines (Wallner 1939; Bonamy and Ravi-Chandar 2005). It should be emphasized that of all these morphological features conform a deterministic geometrical description of the fracture surface within the frame of traditional glass fractography. Nevertheless, when these regions are analyzed at the nanometer scale using AFM they reveal clearly self-afne character spreading over many length-scales. With the above discussed scenario in mind, in this work we analyze the self-afne properties of soda-lime glass fracture surfaces obtained in conditions of slow crack propagation. Our aim is to discuss the coexistence of the roughness exponents = 0.5 and = 0.8 in different regions of the fracture pattern.

2 Experimental In this work we studied a soda-lime glass with the typical dimensions of a slide glass (75 30 1) mm3 . A transverse notch of an approximate depth of 1.5 m was made previously. In order to promote slow propagation of the initial crack front we used a three-point bending system applying water vapor to the notch. The external load was gradually increased each 24 h during a time interval of 4 days. Some of the fracture surfaces were gold sputtered using a Sputter Vacuum-LLC device for SEM-analysis. The SEM-images were obtained by secondary and backscattered electron techniques at several magnications using a SEM LEO Stereoscan 440. For the quantitative analysis we used a Park Scientic-AFM in contact mode, in air at room temperature. The scan frequencies used ranged from 0.4 to 4Hz, and the scan sizes varied from 500 nm 500 nm to 5 m 5 m, all of them of 512 512 points. In order to characterize the scaling properties of these AFM-images, they were analyzed by two methods, the variable bandwidth method or covariance analysis (Schmittbuhl et al. 1995), as implemented by Gonzlez (Gonzlez et al. 2002), and the heightheight correlation

123

Scaling properties of slow fracture in glass

function (Ponson et al. 2006a; Bonamy et al. 2006). In the rst method, for each AFM-image 250 proles out of the 512 are analyzed. This guarantees the accuracy of the calculated value of the roughness exponent, as indicated in (Gonzlez et al. 2002). In the covariance analysis, a height prole of length L is divided into windows or bands of width r indexed by the position of the rst point of the band, i = 1. The standard deviation of the heights, (i), is computed on each window and then averaged over all the possible windows varying the origin at xed r in accordance with the following equation: W (r ) = 1 Nd
Nd i=1

(i) r =

(1)

where Nd is the number of points. The function W (r ) vs r is obtained in a logarithmic plot for each prole then the average of the 250 curves is calculated and analyzed. Using the derivative of this function (Gonzlez et al. 2002), the average roughness exponent is calculated performing a linear regression over the linear interval, then the values for each prole are determined over the same interval and the statistical distribution of the 250 proles is obtained. In the second method, the 1D heightheight correlation function is expected to satisfy: h( x) = [h(x + x) h(x)]2
1/2

3 Results and discussion The fractographic analysis shows that the most part of the area of the fracture surface corresponds to the mirror zone (Fig. 2a). It can then be inferred that the fracture surface is the result of a slow crack propagation process (Mecholsky et al. 1978; Clari et al. 2003). We can observe that the mist and hackle zones (Fig. 2c) only occupy a small portion of the total area of the fracture surface. In addition, the boundary between the mist and hackle zones is

Fig. 2 Backscattered SEM images of the fracture surface, (a) SEM-image showing the different zones: 1. Mirror region, close to the notch 2. Middle of the mirror zone and 3. Misthackle region, (b) and (c) Images at 500 magnications on the middle of the mirror and on the misthackle regions respectively

123

M. Hinojosa et al.

very diffuse; in consequence in the following we refer to this small zone as the misthackle zone. In Fig. 2a the large area corresponding to the mirror zone can be explained considering that when small amounts of energy are delivered to a crack front, the sequential breaking of chemical bonds is very slow and corresponds to the region I as dened by Wiederhorn, where water vapor produces a corrosive attack at the crack tip and the crack velocity depends exponentially on the load. Under these conditions the crack propagation is very slow and at a macroscopic level we observe that the crack front plunges forwards, cleanly slicing material in two, leading to an apparently perfectly at surface. In considering the total time for fracture process (4 days), it was estimated that the crack front velocity in this zone was of the order of some nanometers per second, in accordance with experimental values reported for slow fracture process in glass materials (Wiederhorn 1967; Clari et al. 2003). On the other hand, the presence of small areas with a misthackle morphology is a clear evidence that in this particular region the crack propagation was not a slow process, because once a critical energy ux is exceeded, the crack front can no longer handle the responsibility, and becomes unstable generating increasingly ramied zones of damaged material, in this case crack propagation is faster and corresponds to the region III dened by Wiederhorn, where the crack velocity is independent of the presence of water vapor. In addition, we have detected the presence of secondary cracks (Fig. 2b) by SEM-images observing that those located far from the notch are larger than those near to the notch. It is important to remark here that the presence of secondary cracks on the fracture surface can be associated to the effect of water vapor, let us point out that in the literature there are not many reports of secondary cracks. Figure 3 is a typical 3D-image obtained by AFM in the mirror zone in a region located less than 100 m from the notch. We can see that the mirror zone is not as at as it was suggested by SEM-images. The topographic irregularities detected by AFM in the mirror zone can be considered as a qualitative aspect of the self-afne behavior of these fracture surfaces, taking into account that these irregularities are consistent in AFM-images obtained at smaller scan sizes, less that 2 2 m. On the other hand, for a region located into the misthackle zone (Fig. 4) the AFM-images reveal the presence of step-like morphologies with a characteristic length of around 1 m that are associated with local high speed propagation; in addition we observed again the presence of secondary cracks that are smaller than those detected on the SEM-images. Note also the

Fig. 3 3D-image obtained by AFM in the mirror zone

123

Scaling properties of slow fracture in glass Fig. 4 Typical 3D AFM-image of the misthackle region showing steps as well as secondary cracks (arrow marks). Note also the presence of ne undulations over the surface of the steps at the left corner of the image

presence of ne undulations over the surface of the steps, with a characteristic separation of about 200300 nm. We can speculate that these undulations correspond to Wallner lines due to perturbations caused by stress waves (Bonamy and Ravi-Chandar 2005). In general, the topographic irregularities detected in this region are larger that those observed in the mirror zone, this implies that when the velocity of the crack front increases, the size of the topographic irregularities also increases, given that the velocity of the crack front in the misthackle zone is greater that the velocity in the mirror zone. The surface of the step-like morphologies shown in Fig. 4 deserves special attention, because at the nanometer scale in this particular region each step has a surface morphology which is qualitatively very similar to that of the mirror zone. Thus, one can expect that the self-afne behavior for small regions within one step will be similar to the self-afne behavior of the mirror zone, that is to say that the self-afne behavior will be similar regardless of the difference in crack velocity between the mirror zone (low speed) and the misthackle region (high speed). However, one can also expect that this behavior will be perturbed if several steps are taken into account to obtain a large AFM-image. For the self-afne analysis, three representative regions were explored by AFM: a rst region on the mirror zone located less than 100 m from the notch (region 1 in Fig. 2a), a second region located at the middle of the mirror zone (region 2 in Fig. 2a), and nally a region corresponding to the misthackle zone (region 3 in Fig. 2a). Several AFM-images were obtained for each region. At the beginning of the mirror zone (Fig. 5), the results for the two methods reveal a single self-afne regime with an average value for the roughness exponent of around 0.5 (0.53 and 0.48 for the variable bandwidth method and the height height correlation function, respectively). This result is the conrmation of the existence of an average roughness exponent = 0.5 for slow fracture. A possible correlation length can be established with a value of 200 nm, note also that the heightheight correlation method is apparently more able to detect the correlation length. It is noteworthy to point out that the value reported for Fig. 5 is in agreement with the spacing of the ne undulations discussed above in relation to Fig. 4. In a previous work (Guerra-Amaro et al. 2007), we have reported a roughness exponent of around 0.5 at the beginning of the mirror zone for fracture surfaces obtained under slow crack propagation but without the presence of water vapor. The second region analyzed was located at the middle of the mirror zone (Fig. 2b). The results given on Fig. 6 show that the average roughness exponent, computed using the two methods for all the AFM-images, has values of 0.57 and 0.63 for the variable bandwidth method and the heightheight correlation function, respectively. Again, it is possible to

123

M. Hinojosa et al.

Fig. 5 (a) AFM-image at beginning of the mirror zone, close to the notch, (b) self-afne analyses computed using the variable bandwidth method, (c) statistical distribution of the roughness exponent computed from the variable bandwidth method, (d) loglog representation of 1D heightheight correlation function and (e) statistical distribution from the 1D heightheight correlation function

Fig. 6 (a) AFM-image at the middle of the mirror zones, (b) self-afne analyses computed using the variable bandwidth method, (c) statistical distribution of the roughness exponent computed from the variable bandwidth method, (d) loglog representation of 1D heightheight correlation function and (e) statistical distribution from the 1D heightheight correlation function

estimate a correlation length at around 100200 nm, particularly using the heightheight correlation function. Finally we discuss the self-afne results for the AFM-images obtained in the misthackle zone. All of the AFM-images show larger topographic features compared to those on the mirror zone and in general it was possible to observe step-like morphologies (see Figs. 7a and 8a). These morphologies are related to an increase in the velocity of crack propagation. The energy provided to the crack tip by the external load is used to increase the velocity of crack propagation, and when the velocity has reached its maximum value, the energy is

123

Scaling properties of slow fracture in glass

Fig. 7 (a) AFM-image in the misthackle zone, (b) self-afne analyses computed using the variable bandwidth method, (c) statistical distribution of the roughness exponent computed from the variable bandwidth method, (d) Loglog representation of 1D heightheight correlation function and (e) statistical distribution from the 1D heightheight correlation function

dissipated by propagation on different planes and the number of these planes increases rapidly giving rise to a very irregular surface, which is the misthackle zone. The qualitative analysis of the topography observed in the different AFM images at different scan sizes for this particular region shows than when the scan is made over the surface of a single step, the topographic irregularities are very similar to those obtained in the initial region of the mirror zone (compare Figs. 5a and 8c). The observed morphological scaling behavior can be described in a qualitative way by saying that when one analyzes large regions, one observes the steps, which can be described as a deterministic feature, whereas when one observes small regions one detects the irregular character of the topography, thus one can consider, at least qualitatively, that there is a transition from a deterministic to a irregular topography when the scale of observation is reduced. A rst general self-afnity analysis of this zone (Fig. 7), using images including two or more steps, results in an average roughness exponent of 0.79 and 0.84 for the variable bandwidth method and the heightheight correlation function, respectively. These values can be considered compatible with the reported universal value for high speed/large scales in the mirror region. At this point, it is noteworthy to point out that all of the above results, which include both the mirror and the misthackle zone, are not dependent on the method used, thus all of the above result indicate the coexistence of two statistically different values of the roughness exponents on the fracture surfaces analyzed. In order to explore the effect of having different morphologies (steps at large scales, irregularities at nanometer scale, undulations, secondary cracks), a more detailed analysis on the misthackle region was also performed (Figs. 8 and 9). First, for the analysis of Fig. 8a, which again includes several steps, the variable bandwidth method gives a value of 0.73 whereas the heightheight correlation method results in a value of 0.81, these corroborates the tendence towards a value of around 0.8 when two or more steps are taken into account. When the analysis is performed over one entire step, Fig. 8b, the roughness exponent exhibit values of 0.64 and 0.68 for the variable bandwidth method and the heightheight correlation function, respectively, note that these values are similar to those obtained at the middle of the mirror zone, in agreement with the similarity in morphology and may correspond to a crossover.

123

M. Hinojosa et al.

Fig. 8 AFM-images on the steps from the misthackle zone (a) 44 m2 , (b) 11 m2 (c) 500500 nm2 , (d) Self-afne analyses of the AFM-images computed using the variable bandwidth method (e) Self-afne analyses of the AFM-images computed using the 1D heightheight correlation function

Finally, when only a small region at the center of one step is analyzed, the roughness exponent has values of 0.5 and 0.6 for the variable bandwidth method and the heightheight correlation function, respectively, again these values are clearly compatible with those observed at the beginning of the mirror zone. Figure 9 shows that two different statistical distributions of the values of the roughness exponents are clearly discernible, accompanied by a distribution possibly corresponding to a crossover. Thus, it is safe to say that in this region one can nd two different roughness exponents with average values around 0.5 and 0.8. Moreover, note that the morphological similarity at small length scales between the zones at the middle of the mirror zone and at the middle of the steps in the misthackle region is accompanied by similar values of the roughness exponents. Thus, taken together, the overall results suggest the coexistence of two

123

Scaling properties of slow fracture in glass

Fig. 9 Statistical distributions of the roughness exponent of the images and 8a, (), 8b ( ) and 8c (+a) obtained using (a) the variable bandwidth method and (b) using the 1D heightheight correlation function Fig. 10 Scheme of the qualitative effect of the crack front velocity on the self-afne behavior of the fracture surface

different statistical distributions for the roughness exponents of the mirror and the mist hackle region. In the misthackle region and for length scales smaller than approximately 30 nm, an average roughness exponent very close to 0.5 is observed, whereas for length scales larger than approximately 100 nm the average value of the roughness exponent is very close to 0.8. For intermediate length scales a crossover is apparently detected. One possible interpretation of our experimental results is summarized in Fig. 10, in which the mirrormisthackle pattern can be regarded as a deterministic large-scale description of the fracture surface; we have a transition to a regime with irregular topography when we analyze any of the different regions at ne scales. In the mirror zone, at regions close to the notch, the 0.5 value manifest itself and it is prone to deviations as one approaches the misthackle zone. In the misthackle zone one can observe deterministic features such as steps and undulations, with a transition in the apparent morphology at ne scales. When the proles contain several steps, the 0.8 value dominates whereas for small scales within one single step the 0.5 exponent is observed.

4 Conclusions We have obtained and analyzed the mirror and the misthackle zones of the slow-fracture surface of a soda-lime glass broken in bending under the action of water vapor. In the mirror zone, the fractographic analysis shows the presence of secondary cracks presumably promoted by the presence of water vapor whereas in the misthackle zone we detect steps of about 1 m average spacing and over them it was possible to detect ne undulations. The self-afne analysis shows the existence of two well dened independent statistical distribution for the roughness exponents. At the beginning of the mirror zone we detected the regime

123

M. Hinojosa et al.

characterized by a roughness exponent of about 0.5 average value. The same value is detected in the misthackle zone at the center of the characteristic steps. At large scales in this same region one detects the regime characterized by the value around 0.8, reported as universal for rapid fracture. These statistical distribution are accompanied by what seems to be a crossover where the roughness exponent uctuates about an average value lying somewhere between 0.57 and 0.68; a similar behavior is detected at the middle of the mirror zone. This can be described by a qualitative scenario in which the traditional deterministic mirrormisthackle zones are combined with irregular or self-afne morphology at ne scales.
Acknowledgements Authors acknowledge the nancial support from the Consejo Nacional de Ciencia y Tecnologa (CONACYT) and the Programa de Apoyo a la Investigacin Cientca y Tecnolgica (PAICYT) of the Universidad Autnoma de Nuevo Len. The help from F. Snchez, A. Gonzlez and M. Lara is gratefully acknowledged, we also thank D. Bonamy and E. Bouchaud for a critical reading of the manuscript. We are grateful to F. Clari and C. Roundtree for many enlightening discussions.

References
Balankin A (1997) Physics of fracture and mechanics of self-afne cracks. Eng Fract Mech 57:135203 Bonamy D, Ravi-Chandar K (2005) Dynamic crack response to a localized shear pulse perturbation in brittle amorphous materials: on crack surface roughening. Int J Fract 134:122 Bonamy D, Ponson L, Prades S, Bouchaud E, Guillot C (2006) Scaling exponents for fracture surfaces in homogeneous glass and glassy ceramics. Phys Rev Lett 97:135504 Bouchaud E (1997) Scaling properties of cracks. J Phys Condens Matter 9:43194344 (and references therein) Bouchaud E (2003) The morphology of fracture surfaces: a tool for understanding crack propagation in complex materials. Surface Rev Lett 10(5):797814 (and references therein) Bouchaud E, Lapasset G, Plans J (1990) Fractal dimension of a fractured surfaces: a universal value? Europhys Lett 13(1):7379 Bouchaud JP, Bouchaud E, Lapasset G, Plans J (1993) Models of fractal cracks. Phys Rev Lett 71(14):2240 2243 Clari F, Prades S, Bonamy D, Ferrero K, Bouchaud E, Guillot C, Marlire C (2003) Glass breaks like metal, but at the nanometer scale. Phys Rev Lett 90(7):075504-1075504-4 Daguier P, Nghiem B, Bouchaud E, Creuzet F (1997) Pinning and depinning of crack fronts in heterogeneous materials. Phys Rev Lett 78(6):10621065 Gonzlez VA, Chacn O, Hinojosa M, Guerrero C (2002) Statistical assessment of self-afne applied to short proles. Fractals 10(3):373286 Guerra-Amaro C, Hinojosa M, Reyes-Melo E, Gonzlez-Gonzlez V (2007) Self-afne quasi-static fracture of soda-lime glass. Mater Sci Forum 560:4146 Guerrero C, Reyes E, Gonzlez V (2002) Fracture surface of plastic materials: the roughness exponent. Polymer 43:66836693 Hansen A, Schmittbuhl J (2003) Origin of the universal roughness exponent of brittle fracture surfaces: stressweighted percolation in the damage zone. Phys Rev Lett 90(4):045504-1045504-4 Hinojosa M, Aldaco J (2002) Self-afne fracture surface parameters and their relationship with microstructure in a cast aluminum alloy. J Mater Res 17:12761282 Hinojosa M, Aldaco J, Ortiz U, Gonzlez V (2000) Roughness exponent of the fracture surface of an Al-Si alloy. Aluminum Trans 3(1):5357 Hinojosa M, Gonzlez V, Snchez J, Ortiz U (2004) Scaling properties of the fracture surfaces of a crystalline polymer. Polymer 45:48294836 Hinojosa M, Aldaco J, Rodriguez R, Ortiz U (2005) Roughness exponents, microstructure, correlation length, and the possible origin of selfafne. In: Fracture. Material Research Society Symposium Proceedings, vol 882E, pp EE3.1.1EE3.1.6 Mly KJ, Hansen A, Hinrichsen EL, Roux S (1992) Experimental measurements of the roughness of brittle cracks. Phys Rev Lett 68(2):213215 Mly KJ, Santucci S, Schmittbuhl J, Toussaint R (2006) Local waiting time uctuations along a randomly pinned crack front. Phys Rev Lett 96:045501

123

Scaling properties of slow fracture in glass Mandelbrot BB, Passoja D, Paullay A (1984) Fractal character of fracture surfaces of metals. Nature 308:721 722 Marchenko O, Fichou D, Bonamy D, Bouchaud E (2006) Time resolved observation of fracture events in mica crystal using scanning tunneling microscope. Appl Phys Lett 89:093124 Mecholsky JJ, Freiman W, Rice A (1978) Fractography analysis of ceramics. Fractography in failure analysis ASTM STP 645, pp 367379 Milman VY, Stelmashenki NA, Blumenfeld R (1994) Fracture surfaces: a critical review of fractal studies and novel morphological analysis of scanning tunnelling microscopy measurements. Prog Mater Sci 38:425474 Nio J, Hinojosa M, Gonzlez V (2006) Microstructure and self-afne fracture surface parameters in steel. Mater Sci Forum 509:4348 ystein Haaving Bakke J, Hansen A (2007) Accuracy of roughness exponent measurement methods. Phys Rev E 76:031136 Ponson L, Bonamy D, Auradou H, Mourot G, Morel S, Bouchaud E, Guillot C, Hulin JP (2006a) Anisotropic self-afne properties of experimental fracture surfaces. Int J Fract 140:2737 Ponson L, Bonamy D, Bouchaud E (2006b) Two-dimensional scaling properties of experimental fracture surfaces. Phys Rev Lett 96:035506 Ramanathan S, Ertas D, Fisher DS (1997a) Quasistatic crack propagation in heterogeneous media. Phys Rev Lett 79(5):873876 Ramanathan S, Ertas D, Fisher DS (1997b) Dynamics and instabilities of planar tensile cracks in heterogeneous media. Phys Rev Lett 79(5):877880 Schmittbuhl J, Vilotte JP, Roux S (1995) Reliability of self-afne measurements. Phys Rev E 51(1):131147 Wallner H (1939) Linenstrukturen an Bruchchen. Z Phys 114:368378 Wiederhorn SM (1967) Inuence of water vapor on crack propagation in soda-lime glass. J Am Ceram Soc 50:407414 Zhu T, Li J, Lin X, Yip S (2005) Stress-dependent molecular pathways of silicawater reaction. J Mech Phys Solids 53:15971623

123

S-ar putea să vă placă și