Você está na página 1de 11

Available online at www.sciencedirect.

com

Acta Materialia 59 (2011) 33733383 www.elsevier.com/locate/actamat

Synchrotron X-ray diraction measurements of internal stresses during loading of steel-based metal matrix composites reinforced with TiB2 particles
D.H. Bacon, L. Edwards 1, J.E. Moatt, M.E. Fitzpatrick
Materials Engineering, The Open University, Walton Hall, Milton Keynes MK7 6AA, UK Received 3 February 2011; accepted 7 February 2011 Available online 5 March 2011

Abstract High-energy synchrotron X-ray diraction was used to measure the internal strain evolution in the matrix and reinforcement of steel-based metal matrix composites reinforced with particulate titanium diboride (TiB2). Two systems were studied: a 316L matrix with 25% TiB2 by volume and a W1.4418 matrix with 10% reinforcement. In situ loading experiments were performed, where the materials were loaded uniaxially in the X-ray beam. The results show the strain partitioning between the phases in the elastic regime, and the evolution of the strain partitioning once plasticity occurs. The results are compared with results from Eshelby modelling, and very good agreement is seen between the measured and modelled response for elastic loading of the material. Heat treatment of the 316-based material did not aect the elastic internal strain response. 2011 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
Keywords: Synchrotron radiation; Metal matrix composites (MMCs); Particulate-reinforced composites; Internal stresses

1. Introduction Metal matrix composites (MMCs) are an important class of engineering material, by virtue of their enhanced specic properties and their ability to be tailored to suit many dierent applications. There has been signicant research into aluminium-based MMCs, because of the inherently low density of the aluminium matrix. Steel-based systems are less common, because steel already has a high elastic modulus. However, there is scope to use ceramic reinforcement in steels to increase the elastic modulus further and to improve wear resistance. In addition, the high toughness of steel means that the reduction in toughness which is associated with MMCs as compared to the unreinforced alloys is less critical than in some alu Corresponding author. Tel.: +44 1908 653100; fax: +44 1908 653858.

E-mail address: m.e.tzpatrick@open.ac.uk (M.E. Fitzpatrick). Present address: Australian Nuclear Science and Technology Organisation, PMB1, Menai, NSW 2234, Australia.
1

minium-based systems. Furthermore, ceramic reinforcements such as TiB2 have a lower density than iron, so their addition gives the additional benet of reducing the density of the nal material, as well as increasing its stiness. As a result, steel-based MMCs are of increasing interest for many aerospace and automotive applications due to the initial high strength and toughness of the unreinforced matrix. There have been numerous investigations of the internal strains in aluminium alloys reinforced with a variety of ceramic particulate [17], and most of these studies have used neutron diraction. However, little or no work has been conducted on the internal strains of steel-based composite systems. In the present work, steel composite systems reinforced with titanium diboride were studied. Boron is strongly neutron-absorbing, and therefore highenergy synchrotron X-ray diraction was applied for studying the internal strains in the material. The load bearing capacity of an MMC is dictated by the load transfer occurring between the compliant, soft matrix

1359-6454/$36.00 2011 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.actamat.2011.02.012

3374

D.H. Bacon et al. / Acta Materialia 59 (2011) 33733383

to the sti, hard reinforcement [8]. The load partitioning ratio between matrix and reinforcement remains constant as long as both phases are deforming elastically. Upon plastic deformation of the matrix, the partitioning ratio changes and the reinforcement phase carries a greater proportion of the load. Thus, the composite becomes more ecient, as a greater proportion of the load is being carried by the stier reinforcing phase. However, with the increasing levels of stress placed on the reinforcement (as the matrix continues to plastically deform), that phase may fracture or debond from the matrix, thus reducing the load carried by the reinforcement and the overall load bearing characteristics of the composite, which often leads to macroscopic failure. Experimental measurements of the load partitioning between phases of an MMC can thus give a wealth of information on the micromechanical evolution of composites during deformation. In this present work synchrotron X-ray diraction has been used to measure the internal lattice strains of both matrix and reinforcement phases within an MMC subjected to uniaxial tensile loading to failure. This enables the load transfer to be assessed for both elastic and plastic deformation of the matrix. The measured results are compared to the expected load transfer between the phases predicted by the Eshelby equivalent inclusion method. A secondary goal was to investigate the eects of diering heat treatments on the load partitioning of the composite. 2. Experimental procedure 2.1. Materials Two materials were studied: a 316L stainless steel reinforced with 25% of titanium diboride (TiB2) particles, and a W1.4418 martensitic stainless steel reinforced with 10% TiB2. The reinforcement particles had a nominal diameter of 5 lm. The material was manufactured by Aerospace Metal Composites Ltd. (Farnborough, UK) by a mechanical alloying, powder metallurgy route. The material was studied as-hot isostatic pressed (HIPed), without any secondary processing. Fig. 1 shows an optical micrograph of the 316L composite. It can be seen that the reinforcement has been well distributed, and there are few areas of reinforcing particle clusters or large areas of unreinforced matrix. The as-received billet was heat treated for 3 h at 1040 C, followed by a cold water quench. In this state the material has an ultimate tensile strength (UTS) of 1100 MPa and an elastic modulus of 242 GPa. The billet was then cut in half and the material was then given a tempering treatment, of either 1 h at 480 C followed by an air cool or 4 h at 550 C followed by an air cool. The W1.4118 composite was also treated at 1040 C and then tempered at 550 C. This material has a UTS of 1280 MPa and an elastic modulus of 220 MPa. Dog-bone-shaped, at tensile test specimens were fabricated by electric discharge machining from the three

Fig. 1. Optical micrographs of the 316L-based MMC.

tempered blocks, with a gauge length measuring 1 5 25 mm3 and broader ends for gripping the samples. Uniaxial tensile tests were performed at room temperature using a loading rig in situ on the X-ray diractometer. A clip-gauge extensometer was attached to the specimens to measure the macroscopic strain, and the stress was calculated from the recorded loads. The loading was conducted in strain control (using strain measurements from the extensometer) at a strain rate of 2 105 s1. 2.2. Strain measurements Synchrotron X-ray diraction measurements were performed on the ID11 beamline at the European Synchrotron Radiation Facility (ESRF) (Grenoble, France). The X-ray diraction experimental setup is shown in Fig. 2. The specimen was irradiated by a parallel beam of 50 keV 0photons ), with (corresponding to a wavelength of k 0.245 A 2 approximate dimensions 0.1 0.1 mm , positioned in the centre of the gauge section along a direction parallel to the sample thickness. A reference powder sample (LaB6) was axed to the specimen to monitor any drift in the diffraction peaks as a result of changes in the beam energy prole. LaB6 was chosen because its diraction rings do not overlap with those of the Fe or TiB2. The diraction pattern from the sample was recorded using a 16-bit CCD camera with a 1242 1152 pixels sensor. The camera was positioned 1 m from the specimen at an angle of 2h = 8 from the incident beam. This distance allows for high strain resolution, recording diraction rings between 5.5 and 10. However, only a segment of the diffraction rings was captured, and therefore it was only possible to measure one strain direction at a time (Fig. 3a). Each test was therefore performed twice: once

D.H. Bacon et al. / Acta Materialia 59 (2011) 33733383

3375

Fig. 2. Schematic representation of the experimental setup on ID11, ESRF.

with a sample loaded to failure with the detector oriented to measure the longitudinal strains and once with a sample loaded to failure with the detector oriented to measure the transverse strains. Good reproducibility was found between the tests, as shown in Fig. 4. It was assumed that the two transverse strain components, which arise from Poisson eects, would be identical. Each CCD image was then caked, where an angular sum of the intensity data is eected to produce a mean diffraction pattern from the segment of the diraction rings recorded in the CCD image. Owing to the high resolution of the CCD, very sharp diraction peaks can be obtained (Fig. 3b). The diraction peaks were then tted using a Gaussian prole. The internal strain for each reection was then calculated using: e d d 0 d0 1

where d0 is the unstressed lattice parameter. Although an attempt was made to determine d0 from powder samples of the matrix and reinforcement materials, the diculty of positioning the powder container reproducibly at the same position as the test samples meant that there were geometry-based shifts in the peak positions which rendered obtaining d0 values by this method impossible for this experimental setup. Hence the d0 values were set as the measured lattice parameters before loading of the sample, which allows for monitoring of the stress evolution from the applied load but neglects any pre-existing thermal residual stress in the material. It is probable that there are thermal mist stresses in the material which are tensile in the matrix and compressive in the reinforcement, owing to the dierence in coecient of thermal expansion between the phases. 3. Modelling the phase stresses The partitioning of an applied stress between the matrix and reinforcement phases can be calculated by using a

model such as Eshelbys equivalent inclusion approach [810]. The measured strain   and the principal stress calcu are composed of three components. This lated from it r discussion will focus on the analysis of the measured (and modelled) stresses. There is a macrostress component rmacro , which is the same in both phases and which in this case arises from the applied loading. Secondly, there is mE an elastic stiness mismatch contribution hri , which expresses the extent of load transfer of the macrostress from the relatively compliant matrix to the stier reinforcement. Thirdly, there is a shape mist contribution: stress-free shape mists (i.e. mists that are independent of any applied load) between matrix and reinforcement can occur in a number of ways, such as plastic deformation of the matrix, phase transformations and thermal expansion mists caused by changes in temperature. In these experiments it has not been possible to determine the shape mists as in previous studies [2,5,11] because, as described earlier, no absolute stress-free reference lattice spacing was obtained. The elastic mismatch contribution expresses the redistribution of the macroscopic load arising from the mismatch in elastic constants: it is therefore directly related to the macroscopic term, and can be written as [8]: hrii r Birmacro r
mE

where Bi is a tensor which depends on the mean reinforcement particle shape (primarily the aspect ratio) and volume fraction, and on the elastic constants of the two phases. The components of B may be calculated from theory using a model such as Eshelbys equivalent homogeneous inclusion approach [9], and in this study the calculations were performed using a program developed by Clyne et al. [8,12]. It was assumed that the reinforcing particles are spherical; this is a reasonable assumption, given that the particles have a low aspect ratio, and as the billet was not subjected to any secondary processing operations after fabrication there is no reason to expect any preferred alignment.

3376

D.H. Bacon et al. / Acta Materialia 59 (2011) 33733383

Fig. 3. (a) CCD image of the diraction spectrum from the 316/25% TiB2 composite, for measurement of the longitudinal strains. (b) Diraction spectrum produced following caking of the CCD image.

The coecients B that were used are shown in Table 1 for particles with an aspect ratio (AR) = 1 (spheres); for comparison, values are also provided for aspect ratios of 1.5 and 2.0, representing ellipsoidal particles with a preferred alignment. In Table 1, the coecients Bst,m and Bst,p describe the dependence of the mean phase elastic mis-

match stresses in the matrix and particles respectively in the s direction to the component of the macrostress in r is the stress the t direction. For example, B13,p rmacro 3 transferred to the particles in the 1-direction due to the macrostress component in direction 3. The volume fractions listed in the table, namely, 10%, 17% and 25%,

D.H. Bacon et al. / Acta Materialia 59 (2011) 33733383

3377

Table 1 Coecients of Bst,m and Bst,p for three dierent reinforcement volume fractions and aspect ratios for the FeTiB2 system. Volume Fraction 10% B11 B22 = B33 B21 = B31 B12 = B13 B23 = B32 17% B11 B22 = B33 B21 = B31 B12 = B13 B23 = B32 25% B11 B22 = B33 B21 = B31 B12 = B13 B23 = B32 AR = 1 Fe 0.042 0.042 0.014 0.014 0.014 TiB2 0.382 0.382 0.124 0.124 0.124 AR = 1.5 Fe 0.061 0.036 0.011 0.021 0.012 TiB2 0.546 0.321 0.095 0.188 0.107 AR = 2.0 Fe 0.075 0.032 0.009 0.031 0.013 TiB2 0.678 0.286 0.082 0.280 0.121

0.070 0.070 0.022 0.022 0.022

0.340 0.340 0.109 0.109 0.109

0.098 0.059 0.017 0.034 0.019

0.480 0.287 0.082 0.164 0.095

0.121 0.052 0.015 0.050 0.022

0.592 0.255 0.071 0.242 0.109

Fig. 4. Macroscopic stressstrain response of the two samples tested of the 316/25% TiB2 composite in the 480 C temper. There is excellent correlation between the mechanical response in the two tests.

0.099 0.099 0.031 0.031 0.031

0.296 0.296 0.093 0.093 0.093

0.138 0.084 0.023 0.046 0.027

0.413 0.250 0.070 0.139 0.082

0.168 0.074 0.020 0.068 0.032

0.504 0.222 0.060 0.204 0.096

correspond approximately to weight fractions of 6%, 10% and 16% respectively. 4. Results and discussion 4.1. Strains due to external loads 4.1.1. 316L/25% TiB2 The lattice strain evolutions for both heat treatments are qualitatively similar, as shown in Fig. 5. As the transverse data show greater scatter, they are presented for the 480 C heat treatment only. The raw errors in the calculated strains are less than 2 le, owing to the accuracy with which the diraction peaks are obtained. Hence the error bars would be smaller than the data points on the graph and are not plotted. The initial response is linear for the iron reections up to an applied stress of 600 MPa. The Fe(1 1 1) response is identical for the two samples, with a small dierence in the slope of the Fe(2 0 0) reection. Beyond 600 MPa, the applied stress vs. lattice strain curves for the Fe reections turn upwards, indicating that comparatively less stress is carried by the matrix as the applied stress is increased. This can be attributed to the onset of plasticity in the matrix, which will increase the elastic mismatch with the reinforcement that will not deform plastically. Since plasticity is occurring in the elastically less sti phase, the degree of mismatch increases, leading to increased load transfer to the particles. During initial loading, the transverse strains increase linearly with applied stresses as expected if the reinforcement phase is elastic, but the strains are negative due to Poissons contraction. This is true up to the point when plasticity was identied to begin within the matrix. At this point for both

The elastic properties of the 316L material were used in making these calculations; there would be minor variations for other matrix alloys depending on the precise values of elastic modulus and Poissons ratio. The reinforcement is aligned, when there is any alignment, along the 1direction. Components for the normal stress relations only are presented for simplicity.

composites, the slope of the stressstrain trajectory of the TiB2 phase changes sign, and the particle transverse strains become increasingly tensile. This eect can be assigned to the change in the eective bulk Poissons ratio of the steel from 0.3 to 0.5 due to conservation of volume upon plasticity [13]. Thus, the Poissons ratio mismatch between the plastic matrix (m = 0.5) and the elastic reinforcement (m = 0.11) leads to tensile strains in the TiB2 in the transverse directions during continued loading to failure. 4.1.2. W1.4418/10% TiB2 The lattice strain measurements for this MMC are shown in Fig. 6a and b. Owing to the smaller volume fraction of reinforcement, only one usable peak could be obtained from the TiB2; however, given that in the 25% TiB2 MMC the response of the two TiB2 peaks was identical, this is not a cause for concern. It also explains the higher scatter in the TiB2 response compared to that for the 316L-based composite with the higher volume fraction of reinforcement. The initial response for all peaks is linear up to an applied stress of 1 GPa with both phases deforming elastically. Beyond this point the Fe matrix starts to deform plastically, again indicated by the strain response turning upwards as the matrix carries less additional stress even though the macroscopic applied stress is increasing. This can again be attributed to the onset of plasticity in the

3378

D.H. Bacon et al. / Acta Materialia 59 (2011) 33733383

Fig. 5. Loading response of the 316L/25% TiB2 material: (a) longitudinal response, 480 C heat treatment; (b) transverse response, 480 C heat treatment; (c) longitudinal response, 550 C heat treatment.

D.H. Bacon et al. / Acta Materialia 59 (2011) 33733383

3379

Fig. 6. Internal strains in the W1.4418/10% TiB2 composite: (a) longitudinal strains; (b) transverse strains.

matrix, which will increase the mismatch with the elastic reinforcement. The yield point of the iron is higher in this MMC compared to that in the 316L MMC as the base matrix has a higher strength as well as a slightly higher elastic modulus (200 vs. 190 GPa). There is also very little plastic ow before the onset of failure of the material, which is indicated by the rapid unloading of the matrix above 1100 MPa. The transverse strain response is dierent from that for the 25%-reinforced MMC, as the particles maintain a negative strain throughout loading. This is a consequence of several factors: the material has a much higher yield stress than the 316L, although its strain-to-failure is higher (Fig. 7); therefore the matrix carries a higher elastic strain before the onset of plasticity, meaning that the overall Poisson strain mismatch in the reinforcement is much larger before plasticity occurs. Furthermore, the smaller volume fraction carries a proportionately higher strain when plasticity does occur (the nal strain in the reinforcement at

failure is greater than that in the 25% TiB2 material) and the internal Poisson strain in the reinforcement balances the Poisson loading of the plastic matrix. 4.2. Internal stresses The measured longitudinal and transverse strains can be used to calculate the axial r1 stress and transverse r2 stress in each phase of the composite. Assuming that the two transverse strain components are equal, e2 = e3: r1 2le1 ke1 2e2 r2 r3 2le2 ke1 2e2 s constants, calculated as: where k and l are Lame k mE 1 m1 2m E l 2 1 m 5 6 3 4

3380

D.H. Bacon et al. / Acta Materialia 59 (2011) 33733383

Fig. 7. Macroscopic stressstrain response of the two samples tested of the W1.4418-matrix composite.

where E is the Youngs modulus and m is the Poissons ratio of each phase of the material. For the TiB2, E = 565 GPa and m = 0.11; for both of the steels m = 0.3, and E = 190 and 200 GPa, respectively, for the 316L and W1.4418 matrices. For the steel, the individual peaks show dierent strain responses, which is indicative of the elastic anisotropy of the lattice: the (1 1 1) direction is the stiest austenitic direction and the (2 0 0) is the least sti. Hence, if the bulk material elastic modulus is used in the calculation of stress, there will be an error introduced in the calculated strains for each reection. In order to account for this, the stresses derived from the response of the austenite peaks were calculated from the component strains using planespecic elastic constants calculated using a Kroner-based model [14]. This gave values of E111 = 246 GPa and E200 = 147 GPa, with corresponding Poisson ratio values of 0.24 and 0.32. 4.2.1. 316L/25% TiB2, 480 C temper The calculated internal stresses for the 316-based MMC are shown in Fig. 8. The two heat treatments showed similar responses in the internal stresses. As strain values from two separate tests were used in calculation of the stresses, the longitudinal and transverse strain responses were tted using a Matlab smoothing routine, and the resulting data t was used in the calculation of the stress response. This allowed for calculation of a smooth stress/load response and overcame the problem that longitudinal and transverse strains will not have been determined at precisely the same load in each test. In the plots selected points only are displayed for overall clarity. Fig. 8a shows the calculated longitudinal stresses and Fig. 8b the transverse stresses. It can be seen clearly that,

for the longitudinal stresses, there is transfer of load from the matrix to the reinforcement during the elastic regime below around 600 MPa. The two titanium diboride reections show very similar behaviour, as do the steel reections in the elastic region: note that the good agreement between the Fe(1 1 1) and Fe(2 0 0) response indicates that the approach of using plane-specic elastic constants is appropriate. With reference to Table 1, it can be seen that, for uniaxial loading, an applied macrostress of 400 MPa would be expected to partition to give 518 MPa in the reinforcing particles: for 25% volume fraction and a particle aspect ratio of 1, B11,p = 0.296, so from Eq. (2) the elastic mE mist stress hrip 0:296 400 MPa 118 MPa, giving a total stress of 400 + 118 = 510 MPa. Likewise, for the matrix, B11,m = 0.099, giving an elastic mist of 40 MPa and a total stress of 360 MPa in the steel matrix. The measurements show that, at 400 MPa, there are stresses of around 350 MPa in the matrix and 570 MPa in the reinforcement; these results are summarized in Table 2. Hence there is excellent agreement for the matrix stresses, and acceptable agreement for the reinforcement. The dierence may be a consequence of small errors in the conversion from strain to stress or may reect some preferred alignment of the particles within the material, although this is unlikely, given that the test samples were extracted from the test billet following HIPing without any secondary processing having been applied. Once the matrix becomes plastic, there is no increase in stress in the steel, and the reinforcement bears an increasing proportion of the applied stress. It is notable that there is actually some unloading of the Fe(1 1 1) grain family as plasticity increases. The transverse stresses are relatively low, and although Eshelby modelling can be used to predict the bulk response, the response of individual reections will not necessarily be in good agreement, as the response of a particular plane in the Poisson direction is dependent on which plane is aligned with the load axis [15]. Following the onset of plasticity, there is a tensile stress generated in the reinforcement, whilst the dierent matrix reections show opposing behaviour. 4.2.2. W1.4418/10% TiB2 Internal stress calculations for the W1.4418 material are shown in Fig. 9. Plane-specic elastic constants were again used in the determination of the Fe stresses, with values of E110 = 212 GPa and E200 = 165 GPa. This material has a higher yield strength than the 316-based material, at around 1000 MPa (Fig. 7). This is reected in the internal stress response, which is linear to a much higher applied stress, although there is perhaps some evidence of deviation from linearity below 1000 MPa in the internal stresses. The longitudinal response in Fig. 9a again clearly shows the transfer of load from the matrix to the reinforcement. As a consequence of the lower volume fraction in this material, there is less load shedding from the matrix, but the stress magnitude in the reinforcing particles is higher.

D.H. Bacon et al. / Acta Materialia 59 (2011) 33733383

3381

Fig. 8. (a) Longitudinal and (b) transverse stresses in the Fe and TiB2 phases of the 316/25% TiB2 composite in the 480 C temper. Table 2 Comparisons of measured internal stresses at and predicted internal stresses from the Eshelby model at 400 MPa for the 316based MMC. Longitudinal (MPa) Measured value 316L matrix TiB2 reinforcement 350 570 Eshelby model 360 520 Transverse (MPa) Measured value 30 10 Eshelby model 12 36

In comparison with the Eshelby modelling, there is very good agreement, as shown in Table 3. By linearly tting the elastic response, the experimentally determined stress at 1000 MPa applied in the reinforcement is 1400 MPa, compared to a predicted response of 1380 MPa (Table 1), whilst for the matrix the predicted stress is 960 MPa and the measured is around 980 MPa. The transverse response is again more dicult to interpret. However, the TiB2 does show the expected com-

pressive stress, though the average response of the two iron peaks is signicantly more compressive than predicted. 5. Conclusions The synchrotron X-ray diraction technique has been used to study the internal stress response of two steel-based metal matrix composites reinforced with titanium diboride particles. Samples of the materials were loaded uniaxially

3382

D.H. Bacon et al. / Acta Materialia 59 (2011) 33733383

Fig. 9. (a) Longitudinal and (b) transverse stresses in the Fe and TiB2 phases of the W1.4418/10% TiB2 composite.

Table 3 Comparisons of measured internal stresses at and predicted internal stresses from the Eshelby model at 1000 MPa for the W1.4418 matrix MMC. Longitudinal (MPa) Measured value W1.4418 matrix TiB2 reinforcement 980 1400 Eshelby model 960 1380 Transverse (MPa) Measured value 95 100 Eshelby model 14 124

The transverse matrix stress is the average of the two diraction peak values.

in tension in situ in the X-ray beam. The results show a transfer of load from the matrix to the reinforcement during the elastic regime, and additional load transfer once the matrix becomes plastic. Heat treatment of the material alters the strength but does not aect the elastic load transfer. The elastic response of the composite has been predicted using an Eshelby-based modelling approach. For the longi-

tudinal stress response, parallel to the loading axis, there is excellent agreement between the experimentally determined internal phase stresses and those predicted using the Eshelby approach. The agreement with the transverse response is relatively poor, although this is not unexpected. The load transfer is consistent throughout elastic loading, and following plasticity there is increasing load

D.H. Bacon et al. / Acta Materialia 59 (2011) 33733383

3383

transfer to the reinforcement and some unloading of certain grain families within the matrix. This indicates that there is good interfacial cohesion between the matrix and the reinforcement during loading of the material, and the results show no evidence of particle decohesion or cracking during tensile deformation. Acknowledgements We are grateful to AMC Ltd. for the supply of the materials studied. We acknowledge the European Synchrotron Radiation Facility for the provision of beamtime and we would like to thank Dr M. Moret for assistance in using beamline ID11; Professor Mark Daymond is thanked for help with the experimental execution, and Dr David Dye for help with the Kroner modelling. D.H.B. was supported by a Doctoral Training Award from the UK Engineering and Physical Sciences Research Council. M.E.F. is supported by a grant through The Open University from The Lloyds Register Educational Trust, an independent charity working to achieve advances in transportation, science, engineering and technology education, training and research worldwide for the benet of all.

References
[1] Allen AJ, Bourke MAM, Dawes S, Hutchings MT, Withers PJ. Acta Metall Mater 1992;40:2361. [2] Fitzpatrick ME, Hutchings MT, Withers PJ. Acta Mater 1997;45:4867. [3] Shi N, Bourke MAM, Roberts JA, Allison JE. Metall Mater Trans 1997;28A:2741. [4] Levy-Tubiana R, Baczmanski A, Lodini A. Mater Sci Eng 2003;A341:74. [5] Dutta M, Bruno G, Edwards L, Fitzpatrick ME. Acta Mater 2004;52:3881. [6] Daymond MR, Fitzpatrick ME. Metall Mater Trans 2006;37A:1977. [7] Fitzpatrick ME, Hutchings MT, Withers PJ. Physica B 1995;213:790. [8] Clyne TW, Withers PJ. An introduction to metal matrix composites. Cambridge: Cambridge University Press; 1993. [9] Eshelby JD. Proc Roy Soc 1957;A241:376. [10] Fitzpatrick ME. In: Fitzpatrick ME, Lodini A, editors. Analysis of residual stress using neutron and synchrotron radiation. London: Taylor & Francis; 2003. p. 263. [11] Fitzpatrick ME, Hutchings MT, Withers PJ. Acta Mater 1999;47:585. [12] Warwick CM, Clyne TW. <http://www.msm.cam.ac.uk/mmc/publications/soft.html>. [accessed April, 2008]. stu [13] Balch D, U ndag E, Dunand D. Metall Mater Trans A 2003;34:1787. [14] Lodini A. In: Fitzpatrick ME, Lodini A, editors. Analysis of residual stress using neutron and synchrotron radiation. London: Taylor & Francis; 2003. p. 47. [15] Daymond MR. Rev Mineral Geochem 2006;63:427.

Você também pode gostar