《复合材料 Composites》课程教学资源（学习资料）第五章 陶瓷基复合材料_Atomic Force Microscopy Study and Qualitative Analysis of Martensite

J.Am. Ceran.Sor,88[1261-1267(2005 Ol:10.llll1551-2916.2005.0017 urna Atomic Force Microscopy Study and qualitative Analysis of Martensite Relief in Zirconia Sylvain Deville, Jerome Chevalier, and Hassan El Attaoui ational Institute of Applied Science, Materials Department, Associate Research Unit 5510, 69621 Villeurbanne Cedex, A recent re S. Deville and J. Chevalier. J. Am. Ceram Soc., 8612 2225(2003)] has shown the new pe ties offered the object of extensive studies over the last 20 years, and its by atomic force microscopy(AFM) to investigate martensitic martensitic features were provided mainly by transmission elec transformation-induced relief in zirconia. In this paper, we stud- tron d a interre Howey the surface relief resulting from martensitic te- of the martensitic features might be tragonal to monoclinic phase transformation in yttria and ceria- found in the literature due to experimental difficulties, and little doped zirconia by AFM. AFM appears as a very powerful tool interest has been attached to these materials so far. Recent to investigate martensite relief with great precision. The ph n imaging resolution has shown that AFM could nomenological theory of martensitic crystallography could be provide very precise measurements of martensitic surface relief successfully applied to explain all the observed features. The on zirconia samples. In this paper, aFm has been used to char- formation conditions of martensite are discussed as well as wavs acterize surface relief resulting from martensitic transformation of accommodating locally the transformation strain, i.e., self- in ceria- and yttria-doped zirconia, at a scale that was never ccommodating variant pairs and microcracking. Variant reached before. These qualitative observations brought new growth sequences are observed. These observations bring new sights into the transformation initiation and propagation sights and explanations on the transformation initiation and II. Experimental Procedure Ceria-stabilized zirconia( Ce-TZP) materials were processed by T ensite transformation model, developed by Bain,' lassical powder mixing processing route, using Zirconia Sales as now been the object of almost a century of investiga- Ltd. powders(Guilford Survey, U.K. ) with uniaxial pressi tions. Its relevance to different types of materials, running fr and sintering at 1550%C for 2 h. Yttria-stabilized zirconia metals to ceramics. has drawn a lot of attention and studies TZP)samples were processed using (3 mol% Y2O3)-TZP pow- the characteristics and macroscopic features of the transforma ders(Tosoh, Tokyo, Japan), and also uniaxial pressing and si tion are now well predicted and understood, th ber of tering for 2 h at 1500C. Residual porosity was negligible quantitative reports of surface relief changes Samples were polished with standard diamond-based product hase martensitic transformation is still limited te whichg surprising considering the scale at which the transformation is order to form grain boundary thermal grooves. The effect of a tical methods and scanning electron microscopy, although both slight thermal etching on the aging behavior has been investi- methods provide a limited spatial resolution, and 3D quantita ated, and it was shown it did not modify either the trans- formation mechanism or its kinetics. Thermal etching was tive informations are not accessible. Quite fortunately, the recent performed to study the location of the transformation with re- development of scanning tur to gr atomic force microscopy(AFM)provides the scientific com- both thermally etched and unetched munity with powerful tools to investigate phenomena charac AFM experiments were carried out with a D3100 nanoscope terized by relief variations at a nanometer scale. Great progress from Digital Instruments Inc ( Santa Barbara, CA), using oxide- as accomplished in the last few years and a few reports might sharpened silicon nitride probes in the contact mode, with an be found on steel-based materials. The absence of specifi formation is accompanied by a large strain(4 vol% and 16% conductive materials make it very attractive to study martensitic shear), surface relief is modified by the formation of monoclinic transformation in ceramics in particular se. The vertical resolution (down to a few tenths of nano- The martensitic transformation of zirconia corresponds to the meters)of AFM allows following very precisely the transfor- tetragonal-to-monoclinic (t-m) phase transformation. Zirco- nia, when doped with yttria (Y2O3) or ceria(Ceo,), is indeed Two types of images have been obtained from AFM exper- retained in its metastable tetragonal struct after iments. The first one is the height image(see Figs. 1, 2, or 10) Upon the action of mechanical stresses or hydrothermal solic- where the height of every point of the scanned surface is mea- itations-ll(i.e, water vapor at 140 C), zirconia might trans- ured. This allows relief 3D imaging, making the image analysi form to its stable monoclinic structure. This transformation, at and interpretation easy The second type is the so-called deriv ed image(see Figs. 6, ll, or 12), where the contrast originates from the rate of relief variation. i.e. all the surfaces with the J. Drennan-contributing editor ame orientation, related to the probe scanning path, will appear with the same contrast. These types of images are very con venient to discern planes with a constant angle, such as the Manuscript No 10893. Received March 3. 2004; approved August 31, 2004 ones forming the sides of the self-accommodating variant pair (SAMVP)of martensite. Either of these two types of image presented here. The vertical and lateral scales of AFM images 126l

Atomic Force Microscopy Study and Qualitative Analysis of Martensite Relief in Zirconia Sylvain Deville,w Je´roˆme Chevalier, and Hassan El Attaoui National Institute of Applied Science, Materials Department, Associate Research Unit 5510, 69621 Villeurbanne Cedex, France A recent report [S. Deville and J. Chevalier, J. Am. Ceram. Soc., 86[12], 2225 (2003)] has shown the new possibilities offered by atomic force microscopy (AFM) to investigate martensitic transformation-induced relief in zirconia. In this paper, we stud￾ied qualitatively the surface relief resulting from martensitic te￾tragonal to monoclinic phase transformation in yttria and ceria￾doped zirconia by AFM. AFM appears as a very powerful tool to investigate martensite relief with great precision. The phe￾nomenological theory of martensitic crystallography could be successfully applied to explain all the observed features. The formation conditions of martensite are discussed, as well as ways of accommodating locally the transformation strain, i.e., self￾accommodating variant pairs and microcracking. Variant growth sequences are observed. These observations bring new insights and explanations on the transformation initiation and propagation sequences. I. Introduction THE martensite transformation model, developed by Bain,1 has now been the object of almost a century of investiga￾tions. Its relevance to different types of materials, running from metals to ceramics, has drawn a lot of attention and studies. If the characteristics and macroscopic features of the transforma￾tion are now well predicted and understood, the number of quantitative reports of surface relief changes resulting from phase martensitic transformation is still limited, which is not surprising considering the scale at which the transformation is occurring. Martensite relief has been investigated mainly by op￾tical methods and scanning electron microscopy, although both methods provide a limited spatial resolution, and 3D quantita￾tive informations are not accessible. Quite fortunately, the recent development of scanning tunneling microscopy (STM) and atomic force microscopy (AFM)2,3 provides the scientific com￾munity with powerful tools to investigate phenomena charac￾terized by relief variations at a nanometer scale. Great progress was accomplished in the last few years and a few reports might be found on steel-based materials.4–6 The absence of specific sample preparation and the possibility of observing bulk non￾conductive materials make it very attractive to study martensitic transformation in ceramics in particular. The martensitic transformation of zirconia corresponds to the tetragonal-to-monoclinic (t–m) phase transformation.7,8 Zirco￾nia, when doped with yttria (Y2O3) or ceria (CeO2), is indeed retained in its metastable tetragonal structure after sintering. Upon the action of mechanical stresses or hydrothermal solic￾itations9–11 (i.e., water vapor at 1401C), zirconia might trans￾form to its stable monoclinic structure. This transformation, at the origin of the transformation toughening effect,12,13 has been the object of extensive studies over the last 20 years, and its martensitic nature is now widely recognized.14 Evidences of martensitic features were provided mainly by transmission elec￾tron microscopy and optical interferometry.15 However, very few quantitative reports16–18 of the martensitic features might be found in the literature due to experimental difficulties, and little interest has been attached to these materials so far. Recent progress19 in imaging resolution has shown that AFM could provide very precise measurements of martensitic surface relief on zirconia samples. In this paper, AFM has been used to char￾acterize surface relief resulting from martensitic transformation in ceria- and yttria-doped zirconia, at a scale that was never reached before. These qualitative observations brought new insights into the transformation initiation and propagation mechanisms. II. Experimental Procedure Ceria-stabilized zirconia (Ce-TZP) materials were processed by classical powder mixing processing route, using Zirconia Sales Ltd. powders (Guilford Survey, U.K.), with uniaxial pressing and sintering at 15501C for 2 h. Yttria-stabilized zirconia (Y￾TZP) samples were processed using (3 mol% Y2O3)-TZP pow￾ders (Tosoh, Tokyo, Japan), and also uniaxial pressing and sin￾tering for 2 h at 15001C. Residual porosity was negligible. Samples were polished with standard diamond-based products. Some samples were thermally etched for 12 min at 13501C, in order to form grain boundary thermal grooves. The effect of a slight thermal etching on the aging behavior has been investi￾gated, and it was shown it did not modify either the trans￾formation mechanism or its kinetics. Thermal etching was performed to study the location of the transformation with re￾spect to grain boundaries. Experiments were performed using both thermally etched and unetched samples. AFM experiments were carried out with a D3100 nanoscope from Digital Instruments Inc. (Santa Barbara, CA), using oxide￾sharpened silicon nitride probes in the contact mode, with an average scanning speed of 5 mm/s. Since the t–m phase trans￾formation is accompanied by a large strain (4 vol% and 16% shear), surface relief is modified by the formation of monoclinic phase. The vertical resolution (down to a few tenths of nano￾meters) of AFM allows following very precisely the transfor￾mation. Two types of images have been obtained from AFM exper￾iments. The first one is the height image (see Figs. 1, 2, or 10), where the height of every point of the scanned surface is mea￾sured. This allows relief 3D imaging, making the image analysis and interpretation easy. The second type is the so-called deriv￾ated image (see Figs. 6, 11, or 12), where the contrast originates from the rate of relief variation, i.e., all the surfaces with the same orientation, related to the probe scanning path, will appear with the same contrast. These types of images are very con￾venient to discern planes with a constant angle, such as the ones forming the sides of the self-accommodating variant pair (SAMVP) of martensite. Either of these two types of images are presented here. The vertical and lateral scales of AFM images 1261 Journal J. Am. Ceram. Soc., 88 [5] 1261–1267 (2005) DOI: 10.1111/j.1551-2916.2005.00174.x J. Drennan—contributing editor w Author to whom correspondence should be addressed. e-mail: sylvain.deville@ insa-lyon.fr Manuscript No. 10893. Received March 3, 2004; approved August 31, 2004

1262 Journal of the American Ceramic Society-Deville et al Vol. 88. No 5 One of the most characteristic features of ptmc is the 150.000mm/div formation of SAMVPs. Their apparition is related in a very straightforward to the effects of local shear and stresses resulting from the transformation. In the PtmC, single martensite plates might appear, leading to an N-like shape of surface relief. If two plates are growing back to back or close enough, their habit planes might join themselves, and the overall surface adopts a triangular shape. It was nought 4 that the formation of SAMVP would be present ly in the case where the transforming region was isolated and surrounded by untransformable material, e.g., zirconia rains in an alumina matrix, or tetragonal precipitates in Mgo-partially stabilized zirconia. However, strain considera- tions must also be taken into account in the formation of the SAMVP. When two variants are growing back to back, their shape strain directions are opposite. Considering the very large shear strain(16%)and volume increase(4%)accompanying the t-m phase transformation, very large stresses appear in the sur- rounding zones of transformed material. These stresses migl concentrate and build up to eventually stop the transformation 1.5 or they might also trigger the transformation of another neigh- boring system, providing certain crystallography relation Image of polishing scratches shear(arrows) after martensitic ships are respected. It is nevertheless worth noticing that the ormation in Ce-TZP formation of SAMVP results in a very large reduction of long range overall shear strain, since shear in the variants of a pair is opposite and equal. This configuration is therefore very favorable from an energetic point of view. In the case of zi 6 e always different here so as to exaggerate the relief and have it conia, SAMVP are present all over the surface(Fig. 1).The orientation of the pairs will be discussed later on, but it is al Aging treatments were conducted in autoclave at 140C. in favorable to accommodate stresses induced by the transfor- water vapor atmosphere, with a 2 bar pressure, in order to mation. The different ways of accommodating stresses are discussed in the next section lI Results and discussion It is worth noticing that the first transformed zones present all (1) Martensite Fundamental Features the same relief after transformation, suggesting that their orien- The phenomenological theory of martensitic tation relationships are very similar. The junction planes of these variants are all perpendicular to the surface(see Figs. 3, 4, or 5) S) phase transformation, i.e., the habit plane of so that the overall long-range lateral stress is almost totally sup- does not exhibit either strain or rotation. Ex pressed. Height variations are not restricted by the surface. es for this were provided here by the observation 三 that these systems are the easiest to transform. When the trans- scratches at the samples surface. Upon transformation,a formation is propagating to the surrounding zones of the surface scratch lying at the surface of transformed grains remains con- due to higher stresses in the surrounding zones of the trans- tinuous and straight. The direction of the scratch is slightly modified. according to the overall shear of the surface, but the activated and transformed. The relief change is consequently scratch remains unbroken. This simple observation is shown in modified. Figure 6 provides an example of a very different Fig. 1. Two pairs of self-accommodating martensitic variants pairs lying side by side might be observed, and scratches running across the surface are observed. It is quite clear that these scratches still remained continuous and straight, thus providin strong evidences supporting the IPS phase transformation model 0.500m/dfv ig. c self a the mabida ting mard si i yari na powr in cea tl is The hig crs the entire drating martensitic variant pairs in 3y-ziP

are always different here so as to exaggerate the relief and have it be seen more clearly. Aging treatments were conducted in autoclave at 1401C, in water vapor atmosphere, with a 2 bar pressure, in order to induce phase transformation at the surface of the samples with time. III. Results and Discussion (1) Martensite Fundamental Features The phenomenological theory of martensitic crystallog￾raphy20,21 relies on the model of an invariant plane strain (IPS) phase transformation, i.e., the habit plane of martensite does not exhibit either strain or rotation. Experimental evidenc￾es for this were provided here by the observation of polishing scratches at the sample’s surface. Upon transformation, a scratch lying at the surface of transformed grains remains con￾tinuous and straight. The direction of the scratch is slightly modified, according to the overall shear of the surface, but the scratch remains unbroken. This simple observation is shown in Fig. 1. Two pairs of self-accommodating martensitic variants pairs lying side by side might be observed, and scratches running across the surface are observed. It is quite clear that these scratches still remained continuous and straight, thus providing strong evidences supporting the IPS phase transformation model. One of the most characteristic features of PTMC is the formation of SAMVPs. Their apparition is related in a very straightforward way to the microscopic and macroscopic effects of local shear and stresses resulting from the transformation. In the PTMC, single martensite plates might appear, leading to an N-like shape of surface relief. If two plates are growing back to back or close enough, their habit planes might join themselves, and the overall surface adopts a triangular shape. It was thought14 that the formation of SAMVP would be present only in the case where the transforming region was isolated and surrounded by untransformable material, e.g., zirconia grains in an alumina matrix, or tetragonal precipitates in MgO–partially stabilized zirconia. However, strain considera￾tions must also be taken into account in the formation of the SAMVP. When two variants are growing back to back, their shape strain directions are opposite. Considering the very large shear strain (16%) and volume increase (4%) accompanying the t–m phase transformation, very large stresses appear in the sur￾rounding zones of transformed material. These stresses might concentrate and build up to eventually stop the transformation, or they might also trigger the transformation of another neigh￾boring system, providing certain crystallography relation￾ships are respected. It is nevertheless worth noticing that the formation of SAMVP results in a very large reduction of long￾range overall shear strain, since shear in the variants of a pair is opposite and equal. This configuration is therefore very favorable from an energetic point of view. In the case of zir￾conia, SAMVP are present all over the surface (Fig. 1). The orientation of the pairs will be discussed later on, but it is al￾ready worth noticing that this mechanism seems to be more favorable to accommodate stresses induced by the transfor￾mation. The different ways of accommodating stresses are discussed in the next section. (2) Martensite Formation It is worth noticing that the first transformed zones present all the same relief after transformation, suggesting that their orien￾tation relationships are very similar. The junction planes of these variants are all perpendicular to the surface (see Figs. 3, 4, or 5), so that the overall long-range lateral stress is almost totally sup￾pressed. Height variations are not restricted by the surface, so that these systems are the easiest to transform. When the trans￾formation is propagating to the surrounding zones of the surface due to higher stresses in the surrounding zones of the trans￾formed regions, different crystallographical systems may be activated and transformed. The relief change is consequently modified. Figure 6 provides an example of a very different Fig. 2. Self-accommodating martensitic variant pair in Ce-TZP. The intersection of the habit plane and the surface (arrow) is clearly visible. Fig. 1. Image of polishing scratches shear (arrows) after martensitic transformation in Ce-TZP. Fig. 3. Self-accommodating martensitic variant pairs in 3Y-ZTP run￾ning across the entire grain. 1262 Journal of the American Ceramic Society—Deville et al. Vol. 88, No. 5

May 2005 AFM Study of zircon 63 X0.5 z 80, 000 nm/div 1.25 0.75 025 0.50 Self accommodating martensitic variant pairs in 3Y-TZP with ig. 4. Self-accommodating martensitic variant pairs with a more com- Most parallel to the surface, leading to a rippled atial arrangement in 3Y-TZP. The untransformed grain on the eformation strain being accommodated by slippin rather than a stable cubic phase grain. Grain boundary thermal grooves are The spatial distribution of the SAMvP at the surface is more complex. Several situations indeed might be observed. Either the SAMVP might run through the entire grain, as shown in Fig 3 orientation relationship to the surface, with two possible expla- or they might also stop at the middle of the grain, and a system nations. The first explanation could be related to a junction plane almost parallel to the surface, leading to a rippled surface, with much smaller height variations. However, higher residual stresses are expected in the surrounding areas in this case. the other possibility is the accommodation of deformation strain by slipping rather than twinning. It is not possible to conclude on his particular point without further local crystallographic 1.0 Fig. 5. Self-accommodating martensitic variant pairs arrangement in Fig. 7. Progressive transformation of a grain in 3Y-TZP. The first var Ce-TZP. Untransformed parts can be seen in between the pairs

orientation relationship to the surface, with two possible expla￾nations. The first explanation could be related to a junction plane almost parallel to the surface, leading to a rippled surface, with much smaller height variations. However, higher residual stresses are expected in the surrounding areas in this case. The other possibility is the accommodation of deformation strain by slipping rather than twinning. It is not possible to conclude on this particular point without further local crystallographic information. The spatial distribution of the SAMVP at the surface is more complex. Several situations indeed might be observed. Either the SAMVP might run through the entire grain, as shown in Fig. 3, or they might also stop at the middle of the grain, and a system Fig. 4. Self-accommodating martensitic variant pairs with a more com￾plex spatial arrangement in 3Y-TZP. The untransformed grain on the left is a stable cubic phase grain. Grain boundary thermal grooves are clearly visible. Fig. 5. Self-accommodating martensitic variant pairs arrangement in Ce-TZP. Untransformed parts can be seen in between the pairs. Fig. 6. Self accommodating martensitic variant pairs in 3Y-TZP with either a junction plane almost parallel to the surface, leading to a rippled surface, or with the deformation strain being accommodated by slipping rather than twinning. Fig. 7. Progressive transformation of a grain in 3Y-TZP. The first var￾iants (arrow) appeared at a grain boundary triple junction. May 2005 AFM Study of Zirconia Martensite 1263

1264 Journal of the American Ceramic Society-Deville et al Vol. 88. No. 5 with a different orientation is activated, as shown in Fig. 4. Some more complex structures between these two situations might be found As far as the location of the variants is concerned. afm al- lows the observation of very interesting features. The transfor mation was never initiated away from the boundaries. i.e habit in the middle of a grain, as this would be energetically too highly unfavorable. SAMVP almost always appear at grains triple unction at first. It can be seen(Fig. 7) that the transformation was indeed initiated at the triple junction before propagating to the rest of the grain. This might be interpreted by taking into account residual stress effects. It was shown that residual stresses resulting from material processing concentrate at grains triple junction. As compared to other regions, these sites will act as preferential nucleation sites The top shape of sAMVP is of prime interest Differences are bserved between Ce-tzp and y-TZP. In the case of y-TZP the junction of variant parts of a pair is always very sharp(Figs 3 and 4). In the case of Ce-TZP, some large, flat untransformed zones might be observed(Figs. 2, 5, and 8)at the junction. Since the surface of these zones seems to be unmodified. it is reason- able to suggest they indeed are not transformed, inasmuch by the PTmc. In fact, it n that the formation of SAMVP was a seq uen Scheme of the process leading to the formation of untrar process. The variants did not form all at once, and even the d triangular zones in Ce-TZP. t and m denotes the tetragonal and formation of a single variant is a sequential process. If the two linac phases variants of the pair grow back to back, with a symmetric relief with respect to the surface, a remaining part of tetragonal phase of triangular shape is left in between, as schematically shown in high that the transformation cannot proceed anymore. - The Fig 9. When the variants are growing, stresses might add up in observed differences between Ce-TZP and Y-TZP could be ex these untransformed zones(depending on the crystallograp plained by the differences in grain size and in crystallographic relationships)until everything is transformed(Figs. 3 and 4) arameters However, it is also possible that stresses, if present, become The sequential growth of SAMVP is illustrated in Fig. 10 The same zone of the grain was observed at two different stages of the transformation, and it might be seen that the variant pair did indeed grow in height(about 10 nm) and in length. The two triangles indicate the end of the junction plane, and the distance in between shows ase of about 50 nm in length. This another clear evidence that even a single pair is formed via a sequential progression Once some variants are formed. a large amount of stresses is accumulated, and the system will try to reduce its overall energy Several strategies are possible for this. The first one is to trigger the transformation of a neigh boring system, as previously men- tioned. This is indeed the main mechanism observed for stress relaxation and accommodation in zirconia. However. the two systems must satisfy certain crystallographic relationships 1, 000 umm/div 150.000m/dfv for the transformation to proceed. In the more favorable case, some coherency is found between two adjacent grains and SAMVP running from one grain to the other one might be ob- served, as shown in Fig. 11. Although the grain boundary ther mal groove disturbs the surface homogeneity of the pairs, the relationship between the SAMp of the two adjacent grains is bvious. This is, however, a very rare occurrence in these ma- terials, and very few transgranular martensite laths might be observed In a more general manner, if no specific crystallographic co respondences are found, the grain will have to accommodate all the strain when the transformation is propagating by mod- fying the spatial arrangement of SAMVP this is particularly obvious(Figs. 5 and 12)in Ce-TZP. A system of large SAMVP s usually formed in the grain, occupying almost its entire sur- ace. To transform the remaining parts of the grain and accom date the strain at the same time, some smaller systems of SAMVP could be formed around. so that the resulting stresses d strains are much lower than those of larger SAMVP.A lot of these small pairs might be observed along the grain boundaries Finally, if no correspondence is found between two grains or the combination of the ind the apparition of very large shear esses might lead to the formation of microcracking

with a different orientation is activated, as shown in Fig. 4. Some more complex structures between these two situations might be found. As far as the location of the variants is concerned, AFM al￾lows the observation of very interesting features. The transfor￾mation was never initiated away from the grain boundaries, i.e., in the middle of a grain, as this would be energetically too highly unfavorable. SAMVP almost always appear at grains triple junction at first. It can be seen (Fig. 7) that the transformation was indeed initiated at the triple junction before propagating to the rest of the grain. This might be interpreted by taking into account residual stress effects. It was shown22 that residual stresses resulting from material processing concentrate at grains’ triple junction. As compared to other regions, these sites will act as preferential nucleation sites. The top shape of SAMVP is of prime interest. Differences are observed between Ce-TZP and Y-TZP. In the case of Y-TZP, the junction of variant parts of a pair is always very sharp (Figs. 3 and 4). In the case of Ce-TZP, some large, flat untransformed zones might be observed (Figs. 2, 5, and 8) at the junction. Since the surface of these zones seems to be unmodified, it is reason￾able to suggest they indeed are not transformed, inasmuch as this effect may be explained by the PTMC. In fact, it was shown that the formation of SAMVP was a sequential process. The variants did not form all at once, and even the formation of a single variant is a sequential process. If the two variants of the pair grow back to back, with a symmetric relief with respect to the surface, a remaining part of tetragonal phase of triangular shape is left in between, as schematically shown in Fig. 9. When the variants are growing, stresses might add up in these untransformed zones (depending on the crystallographic relationships) until everything is transformed (Figs. 3 and 4). However, it is also possible that stresses, if present, become so high that the transformation cannot proceed anymore.23,24 The observed differences between Ce-TZP and Y-TZP could be ex￾plained by the differences in grain size and in crystallographic parameters. The sequential growth of SAMVP is illustrated in Fig. 10. The same zone of the grain was observed at two different stages of the transformation, and it might be seen that the variant pair did indeed grow in height (about 10 nm) and in length. The two triangles indicate the end of the junction plane, and the distance in between shows an increase of about 50 nm in length. This is another clear evidence that even a single pair is formed via a sequential progression. Once some variants are formed, a large amount of stresses is accumulated, and the system will try to reduce its overall energy. Several strategies are possible for this. The first one is to trigger the transformation of a neighboring system, as previously men￾tioned. This is indeed the main mechanism observed for stress relaxation and accommodation in zirconia. However, the two systems must satisfy certain crystallographic relationships for the transformation to proceed. In the more favorable case, some coherency is found between two adjacent grains and SAMVP running from one grain to the other one might be ob￾served, as shown in Fig. 11. Although the grain boundary ther￾mal groove disturbs the surface homogeneity of the pairs, the relationship between the SAMVP of the two adjacent grains is obvious. This is, however, a very rare occurrence in these ma￾terials, and very few transgranular martensite laths might be observed. In a more general manner, if no specific crystallographic cor￾respondences are found, the grain will have to accommodate all the strain when the transformation is propagating by mod￾ifying the spatial arrangement of SAMVP; this is particularly obvious (Figs. 5 and 12) in Ce-TZP. A system of large SAMVP is usually formed in the grain, occupying almost its entire sur￾face. To transform the remaining parts of the grain and accom￾modate the strain at the same time, some smaller systems of SAMVP could be formed around, so that the resulting stresses and strains are much lower than those of larger SAMVP. A lot of these small pairs might be observed along the grain boundaries. Finally, if no correspondence is found between two grains or two parts of a single grain, the combination of the very limited plasticity of zirconia, and the apparition of very large shear strains and stresses might lead to the formation of microcracking t m m habit plane habit plane t t t t t m m Fig. 9. Scheme of the process leading to the formation of untrans￾formed triangular zones in Ce-TZP. t and m denotes the tetragonal and monoclinic phases. Fig. 8. Triangular untransformed zones in self-accommodating mar￾tensitic variant pairs in Ce-TZP (left). A detailed zone is shown (right). 1264 Journal of the American Ceramic Society—Deville et al. Vol. 88, No. 5

May 2005 A FM Study of Zirconia Martensite 1265 35.01nm Section Analysis 35.0 0.20 0.40 0.80 Fig 10. Sequential growth of a variant pair in 3Y-TZP. The corresponding relief profile along the line is plotted on the righ at the end of SAMVP. This is illustrated in Fig. 13. Thermal boundary thermal groove. The S-shaped microcrack is running etching was not performed on this sample, so that there is no across the entire micrograph. Its shape also eliminates the pos- risk for the observed microcrack to be mistaken with a grain sibility of this artifact being a residual polishing scratch In al the phenomenological models developed to explain the t-m phase transformation of zirconia, the formation of micro- 2. 00 cracks and macrocracks as a consequence of the transformation plays a major role in the propagation of the transformation However, these cracks were never clearly observed in the sur rounding zones of transformed regions. The observations reported here therefore provide strong evidence supporting these one of hips with the System Crystallography the great improvements of using AFM as compared with conventional observation methods is that it can provide 3D measurements at a nanometer scale The lateral resolution(as low as 0. I nm) and vertical resolution(0.01 nm) provide very eliable quantitative measurements of surface relief characteris- tics. For example, a precision as low as 0. 2 might be reached when measuring angles between planes, providing the image was quired in good condition principally a probe with a ra- Is of curvature- as low as possible, typically 10 nm for the best probes used here. 3D information is nevertheless not ecessary to measure angle relationships between SAMvP junc tion planes. In all of the observations, junction planes were 2.00 lways found to be either parallel or perpendicular. In regard to the original crystallographic structure(tetragonal), this is in Fig. 11. Transgranular self-accommodating martensitic variant pairs excellent agreement with the theory, since two planes of the te (arrows)in 3Y-TZP. Note the middle part of the grain has not yet tragonal cell are crystallographically equivalent, so that they transformed

at the end of SAMVP. This is illustrated in Fig. 13. Thermal etching was not performed on this sample, so that there is no risk for the observed microcrack to be mistaken with a grain boundary thermal groove. The S-shaped microcrack is running across the entire micrograph. Its shape also eliminates the pos￾sibility of this artifact being a residual polishing scratch. In all the phenomenological models developed25,26 to explain the t–m phase transformation of zirconia, the formation of micro￾cracks and macrocracks as a consequence of the transformation plays a major role in the propagation of the transformation. However, these cracks were never clearly observed in the sur￾rounding zones of transformed regions. The observations reported here therefore provide strong evidence supporting these models. (3) Relationships with the System Crystallography One of the great improvements of using AFM as compared with conventional observation methods is that it can provide 3D measurements at a nanometer scale. The lateral resolution (as low as 0.1 nm) and vertical resolution (0.01 nm) provide very reliable quantitative measurements of surface relief characteris￾tics. For example, a precision as low as 0.21 might be reached when measuring angles between planes, providing the image was acquired in good conditions, i.e., principally a probe with a ra￾dius of curvature27 as low as possible, typically 10 nm for the best probes used here. 3D information is nevertheless not necessary to measure angle relationships between SAMVP junc￾tion planes. In all of the observations, junction planes were always found to be either parallel or perpendicular. In regard ‘to the original crystallographic structure (tetragonal), this is in excellent agreement with the theory, since two planes of the te￾tragonal cell are crystallographically equivalent, so that they might equally transform. Fig. 10. Sequential growth of a variant pair in 3Y-TZP. The corresponding relief profile along the line is plotted on the right. Fig. 11. Transgranular self-accommodating martensitic variant pairs (arrows) in 3Y-TZP. Note the middle part of the grain has not yet transformed. May 2005 AFM Study of Zirconia Martensite 1265

1266 Journal of the American Ceramic Society-Deville et al. Vol. 88. No. 5 untransf omed zohe 3.00 2,00 0.5um ig. 13. Self-accommodating martensitic variant pairs in Ce-TZP nowing the formation of microcracks(arrows) in the surroundings of a transformed zone tion on the local crystallography (i.e, crystallographic orienta- tion of the surface)might be obtained by AFM; it must be combined with different techniques. The lack of quantitative reports should vanish quite rapidly in the next few y carrying out AFM experiments Acknowledgments The at ke to thank the CLA debted to Prof Guenin and Prof. M 0.25 discussions on the subject. References 0.250.500.751.001,25 C. Bain and N. Y Dunkirk. Trans. AIME. 70, 25-45(1924). Binnig H. Rohrer. C. Gerber, and E. Weibel, Fig 12. Self-accommodating martensitic variant pairs system in Ce. TZP, showing the formation of smaller variants along the grain bound- G. Binnig H. Rohrer, C. Gerber, and E. Weibel, ""Scanning Tunneling Mi- croscopy. Surf. Sci., 126, 236-44(1983). ary to accommodate the strain locally Z.G. Yang H.S. NH- C Alloy by "Surface relief Microscopy and Phenomenological Theory of Martensitic Crystallography, Phrs.Rev.B.52[ll1787982(1995) FM has been used here for the first time to investigate pre- sely and qualitatively the surface characteristics of martensitic transformation in zirconia. It was shown that all the features GT. Davenport, L. Zhou, and J. Trivisonno. "Ultrasonic and Atomic Force observed here could be explained by the PTMC. The formation by Temperature and Uniaxia of SAMVP was observed. The different ways of accommodating 妞 oys, Plys8pB locally the strain (i.e, SAMVP formation), system of small SAMVP formation. and microcracking were observed and A. H Heuer Evans, " Transformation-Toughening discussed. As far as saMvP formation conditions are con- erned, it was shown that grains triple junction appear as pref "T. Sato and M. Shimada."Transformat rential nucleation sites. Differences in martensitic surface relief ZrO Polycrystals by Annealing in Water, J. Am. Ceram. Soc., 68[6]356-9 between Ce-TZP and Y-TZP could be explained by differences F. F. Lange, G. L. Dunlop, and B I. Davis, " Degradation in grain size and crystallography Transformation Toughened ZrOrY-O3 Materials at 250"C, "J. Anm Ceram. Soc. With its unique lateral and vertical resolution, the possibility of observing bulk samples(as compared with thin foils used H. Tsubakino. M. Hamamoto, and R. Nozato " Tetragonal to Monoclinic for transmission electron microscopy) and ease of image inter pretation because of the absence of a specific interaction be- C. Garvie. R. H. Hannink. and R. T. Pascoe. "Ceramic Steel? ""Nature tween the probe and the surface of zirconia, AFM appears as a H. Heuer, "Review-Transformation Toughening in Ce unique and extremely powerful tool to investigate martensitic transformation. However it is worth noticing that no informa oc,63S]24-8(1980

IV. Conclusions AFM has been used here for the first time to investigate pre￾cisely and qualitatively the surface characteristics of martensitic transformation in zirconia. It was shown that all the features observed here could be explained by the PTMC. The formation of SAMVP was observed. The different ways of accommodating locally the strain (i.e., SAMVP formation), system of small SAMVP formation, and microcracking were observed and discussed. As far as SAMVP formation conditions are con￾cerned, it was shown that grains’ triple junction appear as pref￾erential nucleation sites. Differences in martensitic surface relief between Ce-TZP and Y-TZP could be explained by differences in grain size and crystallography. With its unique lateral and vertical resolution, the possibility of observing bulk samples (as compared with thin foils used for transmission electron microscopy) and ease of image inter￾pretation because of the absence of a specific interaction be￾tween the probe and the surface of zirconia, AFM appears as a unique and extremely powerful tool to investigate martensitic transformation. However, it is worth noticing that no informa￾tion on the local crystallography (i.e., crystallographic orienta￾tion of the surface) might be obtained by AFM; it must be combined with different techniques. The lack of quantitative reports should vanish quite rapidly in the next few years by carrying out AFM experiments. Acknowledgments The authors would like to thank the CLAMS for using the nanoscope. The authors are also indebted to Prof. Guenin and Prof. Morin for their very fertile discussions on the subject. References 1 E. C. Bain and N. Y. Dunkirk. Trans. AIME, 70, 25–45 (1924). 2 G. Binnig, H. Rohrer, C. Gerber, and E. Weibel, ‘‘Surface Studies by Scanning Tunneling Microscopy,’’ Phys. Rev. Lett., 49, 57–61 (1982). 3 G. Binnig, H. Rohrer, C. Gerber, and E. Weibel, ‘‘Scanning Tunneling Mi￾croscopy,’’ Surf. Sci., 126, 236–44 (1983). 4 Z. G. Yang, H. S. Fang, J. J. Wang, C. M. Li, and Y. K. Zheng, ‘‘Surface Relief Accompanying Martensitic Transitions in an Fe–Ni–C Alloy by Atomic-Force Microscopy and Phenomenological Theory of Martensitic Crystallography,’’ Phys. Rev. B, 52 [11] 7879–82 (1995). 5 N. Bergeon, S. Kajiwara, and T. Kikuchi, ‘‘Atomic Force Microscopy Study of Stress-Induced Martensite Formation and Its Reverse Transformation in a Thermomechanically Treated Fe–Mn–Si–Cr–Ni Alloy,’’ Acta Mater., 48, 4053– 64 (2000). 6 T. Davenport, L. Zhou, and J. Trivisonno, ‘‘Ultrasonic and Atomic Force Studies of the Martensitic Transformation Induced by Temperature and Uniaxial Stress in NiAl Alloys,’’ Phys. Rev. B, 59 [5] 3421–6 (1999). 7 G. M. Wolten, ‘‘Diffusionless Phase Transformation in Zirconia and Hafnia,’’ J. Am. Ceram. Soc., 46, 418–22 (1963). 8 D. L. Porter, A. H. Heuer, and A. G. Evans, ‘‘Transformation-Toughening in Partially Stabilized Zirconia (PSZ),’’ Acta Metall., 27 [10] 1649–54 (1979). 9 T. Sato and M. Shimada, ‘‘Transformation of Yttria-Doped Tertragonal ZrO2 Polycrystals by Annealing in Water,’’ J. Am. Ceram. Soc., 68 [6] 356–9 (1985). 10F. F. Lange, G. L. Dunlop, and B. I. Davis, ‘‘Degradation During Aging of Transformation Toughened ZrO2–Y2O3 Materials at 2501C,’’ J. Am. Ceram. Soc., 69, 237–40 (1986). 11H. Tsubakino, M. Hamamoto, and R. Nozato, ‘‘Tetragonal to Monoclinic Phase Transformation During Thermal Cycling and Isothermal Aging in Yttria– Partially Stabilized Zirconia,’’ J. Mater. Sci., 26, 5521–6 (1991). 12R. C. Garvie, R. H. Hannink, and R. T. Pascoe, ‘‘Ceramic Steel?,’’ Nature (London), 258, 703–4 (1975). 13A. G. Evans and A. H. Heuer, ‘‘Review–Transformation Toughening in Ce￾ramics: Martensitic Transformation in Crack-Tip Stress Fields,’’ J. Am. Ceram. Soc., 63 [5] 241–8 (1980). Fig. 13. Self-accommodating martensitic variant pairs in Ce-TZP, showing the formation of microcracks (arrows) in the surroundings of a transformed zone. Fig. 12. Self-accommodating martensitic variant pairs system in Ce￾TZP, showing the formation of smaller variants along the grain bound￾ary to accommodate the strain locally. 1266 Journal of the American Ceramic Society—Deville et al. Vol. 88, No. 5

May 2005 A FM Study of Zirconia Martensite 1267 4P. M. Kelly and L. R. Francis Rose, "The Martensitic Transformation 2M. S. Wechsler, D. S. Lieberman, and T A. Read, " On the Theory of For- -ts Role in Transformation toughening og. Mater. Sci., 463 mation of Martensite, "J. Met. 197. 1503-13(1953). ISM. Hayakawa and M. Oka, "Structural Study on the Tetragonal to Mono- ormations.Acta Metall, 2. 129-37(1954). -L Gremillard. T. Epicier, J. Chevalier, and G. Fantozzi.""Micro- structural Study of Silica-Doped Zirconia Ceramics, Acta Mater, 48. 4647-52 I6M. Yamamoto, T Fujisawa, T Saburi, M. Hayakawa, M. Oka. T Kurumi- (2000 1时出+20mmYm er.Sci.Lett,9.803-6(1990) 24M. Rhile. L. T. Ma. W. Wunderlich. and A. G. Evans. "TEM Studies H. Tsubakino. Y. Kuroda and M. Niibe "Surface Relief Associated with Phases and Phases Stabilities of Zirconia Ceramics, Physica B. 150, 86-98 Martensite in Zirconia 3 mol% Y tria Ceramics Observed by Atomic H. Zheng, H. S Fang, H. Z Shi. X F. Wang and H.M. Mater Sci. Left, 6, 465-7(1987) "The Study of Martensitic Transformation and Nanoscale Surface Relief in zir- Mater.Sci.Let,21,415-8(2002) 二记 eville and J. Chevalier. ""Martensitic Relief observation by Atomic Force Microscopy in Y tria Stabilized Zirconia. "J. Am. Ceram. Soc., 86[12] (UD. L. Sedin and k. L. Rowen, "Infuence or Tip Sin on AFM Roughness 2225-7(2003) Measurements,App. Surf. Sci. 182, 40-8(2001

14P. M. Kelly and L. R. Francis Rose, ‘‘The Martensitic Transformation in Ceramics—Its Role in Transformation Toughening,’’ Prog. Mater. Sci., 463, 463–557 (2002). 15M. Hayakawa and M. Oka, ‘‘Structural Study on the Tetragonal to Mono￾clinic Transformation in Arc-Melted ZrO2–2 mol.% Y2O,’’ Acta Metall., 37 [8] 2229–35 (1989). 16M. Yamamoto, T. Fujisawa, T. Saburi, M. Hayakawa, M. Oka, T. Kurumi￾zawa, and K. Kusao, ‘‘Scanning Tunneling Microscope Study of Surface Relief Induced by the Tetragonal-to-Monoclinic Transformation in a Zirconia–Yttria Alloy,’’ Ultramicroscopy, 42, 1422–7 (1992). 17H. Tsubakino, Y. Kuroda, and M. Niibe, ‘‘Surface Relief Associated with Isothermal Martensite in Zirconia 3 mol% Yttria Ceramics Observed by Atomic Force Microscopy,’’ J. Am. Ceram. Soc., 82 [10] 2921–3 (1999). 18X. Y. Chen, X. H. Zheng, H. S. Fang, H. Z. Shi, X. F. Wang, and H. M. Chen, ‘‘The Study of Martensitic Transformation and Nanoscale Surface Relief in Zir￾conia,’’ J. Mater. Sci. Lett., 21, 415–8 (2002). 19S. Deville and J. Chevalier, ‘‘Martensitic Relief Observation by Atomic Force Microscopy in Yttria Stabilized Zirconia,’’ J. Am. Ceram. Soc., 86 [12] 2225–7 (2003). 20M. S. Wechsler, D. S. Lieberman, and T. A. Read, ‘‘On the Theory of For￾mation of Martensite,’’ J. Met., 197, 1503–13 (1953). 21J. S. Bowles and J. K. Mackenzie, ‘‘The Crystallography of Martensite Trans￾formations,’’ Acta Metall., 2, 129–37 (1954). 22L Gremillard, T. Epicier, J. Chevalier, and G. Fantozzi, ‘‘Micro￾structural Study of Silica-Doped Zirconia Ceramics,’’ Acta Mater., 48, 4647–52 (2000). 23C. A. Bateman and M. R. Notis, ‘‘Tranverse Twins in Magnesia Partially Stabilized Zirconia (Mg-PSZ),’’ J. Mater. Sci. Lett., 9, 803–6 (1990). 24M. Rhu¨le, L. T. Ma, W. Wunderlich, and A. G. Evans, ‘‘TEM Studies on Phases and Phases Stabilities of Zirconia Ceramics,’’ Physica B, 150, 86–98 (1988). 25M. Yoshimura, T. Noma, K. Kawabata, and S. Somiya, ‘‘Role of H2O on the Degradation Process of Y-TZP,’’ J. Mater. Sci. Lett., 6, 465–7 (1987). 26T. Sato and M. Shimada, ‘‘Transformation of Yttria-Doped Tetragonal ZrO2 Polycrystals by Annealing in Water,’’ J. Am. Ceram. Soc., 68 [6] 356–9 (1985). 27D. L. Sedin and K. L. Rowen, ‘‘Influence of Tip Size on AFM Roughness Measurements,’’ Appl. Surf. Sci., 182, 40–8 (2001). & May 2005 AFM Study of Zirconia Martensite 1267

Copyright of journal of the American Ceramic Society is the property of Blackwell Publishing Limited and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission However, users may print, download, or email articles for individual use