October 1996 ELSEVIER Materials Letters 28(1996)401-407 HREM study of the tetragonal/ monoclinic interfacial fine structure in Zro associated with the martensitic transformation Fangfang Xu Shulin Wen China Received 22 January 1996: 19 March 1996: accepted 25 March 1996 Abstract lycrystals (3Y-TZP)were chosen for the study of the interfacial fine structures resulting from ic transformation of Zro2. HREM observations revealed for the first time that there exists a transition area in-between the transformed martensitic phase(m-ZrO2 )and the parent tetragonal phase(t-zrO2).The dynamic process of this transformation induced by electron beams of variable intensity and its progression suggests that there may exist two different mechanisms with regard to nucleation and grain growth. The former is governed by the thermal stress produced by electron irradiation, whereas the latter is controlled by electron irradiation. Finally, a comparison of the nature of the twins formed during various transformation conditions was made Keywords: Tetragonal Zrc?: Monoclinic Zro,: Martensitic transformation: Interfacial fine structure; Electron irradiation; HREM: I wins 1. Introduction Using the metallurgical theory of martensitic transformations and applying it to the martensitic Apart from the stress generated during rapid transformation occurring in ceramic materials has quenching, extraneous stress generated during the given many important results. The phenomenological fracture of the material will induce the martensitic theory of martensitic transformation permits the de- transformation of t-Zr0,, and absorbing more fra termination of the habit plane of the transformed ture energy resulting in toughening the material. This region, its orientation relationship to the parent crys- is of great significance in enhancing the mechanical tal and the magnitude and direction of the shape properties through the martensitic transformation. strain. The only required input data are the lattice This consideration attracted our attention to reveal parameters of the two phases, the correspondence the details of the microstructural change and crystal- between them and the plane and direction of the lographic relationships accompanying the transfor- invariant lattice shear [1-5]. The theory also gives mation by use of high resolution electron microscopy three possible, simple correspondences depending on (hREM) which monoclinic axis is derived from the unique c axis of the tetragonal parent phase [6]. The tetragonal c axis can become the a, b or c axis of the mono- Corresponding author clinic phase. Hence the three correspondences-A 00167-577X/96/S12.00 Copyright o 1996 Elsevier Science B V. All rights reserved PHs0167577X(96)00093-6
& *H __ __ !!!B ELSEVIER October 1996 Materials Letters 28 (1996) 401-407 HREM study of the tetragonal/monoclinic interfacial fine structure in ZrO, associated with the martensitic transformation Fangfang Xu * , Shulin Wen Shanghai Institute of Ceramics, Chinese Academy of Science, 200050 Shanghai, China Received 22 January 1996; revised 19 March 1996; accepted 25 March 1996 Abstract Y,O,, 3 vol%, stabilized tetragonal zirconia polycrystals (3Y-TZP) were chosen for the study of the interfacial fine structures resulting from the martensitic transformation of ZrO,. HREM observations revealed for the first time that there exists a transition area in-between the transformed martensitic phase (m-ZrO,) and the parent tetragonal phase (t-Z@). The dynamic process of this transformation induced by electron beams of variable intensity and its progression suggests that there may exist two different mechanisms with regard to nucleation and grain growth. The former is governed by the thermal stress produced by electron irradiation, whereas the latter is controlled by electron irradiation. Finally, a comparison of the nature of the twins formed during various transformation conditions was made. Keywords: Tetragonal ZrC,; Monoclinic ZrO,; Martensitic transformation; Interfacial fine structure; Electron irradiation; HREM; Twins 1. Introduction Apart from the stress generated during rapid quenching, extraneous stress generated during the fracture of the material will induce the martensitic transformation of t-ZrO,, and absorbing more fracture energy resulting in toughening the material. This is of great significance in enhancing the mechanical properties through the martensitic transformation. This consideration attracted our attention to reveal the details of the microstructural change and crystallographic relationships accompanying the transformation by use of high resolution electron microscopy (HREM). _ Corresponding author. Using the metallurgical theory of martensitic transformations and applying it to the martensitic transformation occurring in ceramic materials has given many important results. The phenomenological theory of martensitic transformation permits the determination of the habit plane of the transformed region, its orientation relationship to the parent crystal and the magnitude and direction of the shape strain. The only required input data are the lattice parameters of the two phases, the correspondence between them and the plane and direction of the invariant lattice shear [l-51. The theory also gives three possible, simple correspondences depending on which monoclinic axis is derived from the unique c axis of the tetragonal parent phase [6]. The tetragonal c axis can become the a, b or c axis of the monoclinic phase. Hence the three correspondences - A, 00167-577X/96/$12.00 Copyright 0 1996 Elsevier Science B.V. All rights reserved. PII SOl67-577X(96)00093-6
F. Xu, S. Wen/ Materials Letters 28(1996)401-407 B or C. Theoretically, correspondence A is the least specimen was perforated. Then the ultra-thin plate likely, as the match between the lattice parameters is was sputtered with amorphous carbon to the thick the worst and the sum of the transformation strains is ness of about several decades of nanometers the highest. Experimental observations support the The XRD analysis was performed on a D/max-ra existence of correspondences B and C. They are with a copper target and an operating voltage of 40 The HREM examination was performed on a [010lao01l o10la101 JEM-200CX with an operating voltage of 200 kV 0019°to[100]1 0019to[0011 using a top entry double-tilt specimen holder ie.(001)K100) ie.(001)元nK001) An MS (microscanning spectrophotometer, Carl Zeiss K-100)was employed so as to precisely mea [O]n9°to[ool [0Ol9°to[001 sure the interplanar distances on the lattice micro (100)K(10 e.(100)mK100 [01O]—o011 010] 0011-|100 o0l|01 3. Results and discussion Since the phenomenological theory can only de- The addition of 3 vol%Y,0, can make the scribe the correspondence in position between atoms t-ZrO2 stable at room temperature when cooled from in the two structures, it tells nothing about the actual the hot-pressed environment. On the other hand paths chosen by the atoms during the transformation. preparation of the ultra-thin discs for the TEM obser- In our study, HREM was applied to determine such vations, including cutting, polishing and ion-thin paths. We followed the dynamics of the transforma- ning, may have induced the partial martensitic trans tion, obtained micrographs of the interfaces of the formation in the sample. XRD indicated that about transforming bands and revealed the fine structure on 20% of the tetragonal phase had transformed the atomic level of the interphases Fig. la shows the typical morphology of the These d there were hardly any glassy phases in the grain boundaries. During long time(about 45 min)irradia 2. Experimental tion with high intensity electron beams, thermal stress was produced and induced the transformation of the The materials used in this study were 3Y-TZP partially stabilized t-ZrO2. Fig. Ia-lc give the se- ceramics prepared by a wet chemical method. The quence of the dynamic process of transformation, raw materials were chemically pure YCl, and and these pictures were taken under weak electron ZrOCl,. The resulting powder material was hot beams. Though the grains and interfaces were homo- pressed at a pressure of 25 MPa and temperature of geneously irradiated by the electron beams, nucle 1650°C ation of the monoclinic phase took place exclusively The sintered material was then cut into bars with at the grain interfaces(indicated by an arrow in Fig dimensions of 5. x 2.5 X 35 mm. and then me- la) indicating that the thermal stress produced chanically polished. These bars were then examined through irradiation of electron beams was the main y XRD impetus for the nucleation of the transformation as Specimens for HREM studies were prepared by the interfaces were the sites of the stress concentra cutting thin disc sections from the hot-pressed bars. tion. As ZrO, is electrically insulating, we suggest The thin sections were mechanically polished to a that there are two interaction mechanisms with re thickness of less than 30 um. The diameter of the gard to the impact of the electrons on the sample discs was 2.3 mm. Finally, the discs were ion milled First, as the electrons impact the sample, they cannot using a Gatan-600 ion beam thinner at a voltage of 4 diffuse out of the material, their kinetic energy is kv and an incidence beam angle of 15 until the transferred to the sample and a thermal effect was
402 F. Xu, S. Wen /Materials Letters 28 (19961401-407 B or C. Theoretically, correspondence A is the least likely, as the match between the lattice parameters is the worst and the sum of the transformation strains is the highest. Experimental observations support the existence of correspondences B and C. They are B-l C-l ~1001,1l[0101, ~0101,11[0011, [001],9” to [loo], i.e. (001),11(100), B-2 [ 100],9” to [OlO], i.e. (100),11(010), ~0101,11DN, [0011,11mN, ml,ll[1ool, bm,ll[w, [001],9” to [OOl], i.e. (001),11(001), c-2 [100],9” to [loo], i.e. (lOO),II(lOO), D101,11[0101, Kw,11m1, Since the phenomenological theory can only describe the correspondence in position between atoms in the two structures, it tells nothing about the actual paths chosen by the atoms during the transformation. In our study, HREM was applied to determine such paths. We followed the dynamics of the transformation, obtained micrographs of the interfaces of the transforming bands and revealed the fine structure on the atomic level of the interphases. 2. Experimental The materials used in this study were 3Y-TZP ceramics prepared by a wet chemical method. The raw materials were chemically pure YCl, and ZrOCl 2. The resulting powder material was hot pressed at a pressure of 25 MPa and temperature of 1650°C. The sintered material was then cut into bars with dimensions of 5.0 X 2.5 X 35 mm3, and then mechanically polished. These bars were then examined by XRD. Specimens for HREM studies were prepared by cutting thin disc sections from the hot-pressed bars. The thin sections were mechanically polished to a thickness of less than 30 pm. The diameter of the discs was 2.3 mm. Finally, the discs were ion milled using a Gatan-600 ion beam thinner at a voltage of 4 kV and an incidence beam angle of 15” until the specimen was perforated. Then the ultra-thin plate was sputtered with amorphous carbon to the thickness of about several decades of nanometers. The XRD analysis was performed on a D/max-ra with a copper target and an operating voltage of 40 kV. The HREM examination was performed on a JEM-200CX with an operating voltage of 200 kV using a top entry double-tilt specimen holder. An MS (microscanning spectrophotometer, Carl Zeiss K-100) was employed so as to precisely measure the interplanar distances on the lattice micrographs. 3. Results and discussion The addition of 3 ~01% Y203 can make the t-ZrO, stable at room temperature when cooled from the hot-pressed environment. On the other hand, preparation of the ultra-thin discs for the TEM observations, including cutting, polishing and ion-thinning, may have induced the partial martensitic transformation in the sample. XRD indicated that about 20% of the tetragonal phase had transformed. Fig. la shows the typical morphology of the sample. These tetragonal grains are linked closely and there were hardly any glassy phases in the grain boundaries. During long time (about 45 min) irradiation with high intensity electron beams, thermal stress was produced and induced the transformation of the partially stabilized t-ZrO,. Fig. la- lc give the sequence of the dynamic process of transformation, and these pictures were taken under weak electron beams. Though the grains and interfaces were homogeneously irradiated by the electron beams, nucleation of the monoclinic phase took place exclusively at the grain interfaces (indicated by an arrow in Fig. la) indicating that the thermal stress produced through irradiation of electron beams was the main impetus for the nucleation of the transformation as the interfaces were the sites of the stress concentration. As ZrO, is electrically insulating, we suggest that there are two interaction mechanisms with regard to the impact of the electrons on the sample. First, as the electrons impact the sample, they cannot diffuse out of the material, their kinetic energy is transferred to the sample and a thermal effect was
F.Xu, S. en/ Materials Letters 28(1996)401-407 then produced and consequently a thermal stress arises. The second is the irradiation of the electron beam. For the case of high magnification of the electron microscope the electron flux could reach 5.6X 10 electrons/a upon the specimen accord g to the formula /=b2'tj. Here I is the beam current and j the current density [7]. So, the effect of electron impact on the sample atoms is great Fig. 2. Twin-related transformed bands with theit diffraction From Fig. l, it can be seen that after the first stage, that is nucleation at the grain boundary, the transformation preferentially took place in a large grain(C)even though the electron beam had been focused in the region with two small grains(A and B)on the left. a transforming band of about 0. 1 pm wide proceeded in a fixed direction indicated by the numbers I and 2. After it reached another grain boundary(about 5 s), the band returned within the original grain with a definite angle. As the band turned back parallel to the initial monoclinic lath(m-lath)and the martensite bars exhibited a Z-shape structure At the second stage of the transformation process, that is wth of t forming band immediately stopped proceeding. Thi proved that the irradiation rather than a thermody namic mechanism dominated the martensite growth The micrograph and electron diffraction pattern in Fig. 2 confirmed the twin's symmetric relationship of the two neighboring bands. Fig. 3 gives the t/ interfacial fine structure and its analysis and mea- surement. Fig. 3a is the lattice image of the region at lath. According to th solution of the diffraction pattern and the stereo- graphic projection(Fig. 4), in situ orientation rela- by electron beam showing a configura. tionship between the parent tetragonal phase and the tion of Z-shaped transformed band transformed monoclinic phase is B-2, i.e
F. Xu, S. Wen / Materials Letters 28 (1996) 401-407 403 then produced and consequently a thermal stress arises. The second is the irradiation of the electron beam. For the case of high magnification of the electron microscope, the electron flux could reach 5.6 X IO5 electrons/A’ upon the specimen according to the formula I = b2 mj. Here I is the beam current and j the current density [7]. So, the effect of electron impact on the sample atoms is great. Fig. 2. Twin-related transformed bands with theit diffraction pattern. Fig. 1. Dynamic process (a + b --) c) of the martensitic transformation of t-ZrO, induced by electron beam showing a configuration of Z-shaped transformed bands. From Fig. 1, it can be seen that after the first stage, that is nucleation at the grain boundary, the transformation preferentially took place in a large grain (C) even though the electron beam had been focused in the region with two small grains (A and B) on the left. A transforming band of about 0.1 km wide proceeded in a fixed direction indicated by the numbers 1 and 2. After it reached another grain boundary (about 5 s), the band returned within the original grain with a definite angle. As the band turned back again at the boundary, it proceeded parallel to the initial monoclinic lath (m-lath) and the continuous martensite bars exhibited a Z-shaped structure. At the second stage of the transformation process, that is the grain growth of the m-phase, once the electron beam intensity was decreased, the transforming band immediately stopped proceeding. This proved that the irradiation rather than a thermodynamic mechanism dominated the martensite growth. The micrograph and electron diffraction pattern in Fig. 2 confirmed the twin’s symmetric relationship of the two neighboring bands. Fig. 3 gives the t/m interfacial fine structure and its analysis and measurement. Fig. 3a is the lattice image of the region at the front of the transforming lath. According to the solution of the diffraction pattern and the stereographic projection (Fig. 41, in situ orientation relationship between the parent tetragonal phase and the transformed monoclinic phase is B-2, i.e
FXu, S Wen/Materials Letters 28(1996)401-407 solaRo 0·。●·●●s●● ·00··9● ● Lower laver o Middle Layer o Lower Layer Dislocations Tetragonal Trans Monoclinic
404 F. Xu, S. Wen /Materials Letters 28 (1996) 401-407 ! Q 0 0 0 tipper LIyel 0 Middle Layer %r l Upper L1qcr 0 M iddlc Lsyer I 0.5 I hnna I Mo~ro~inic ZrO: lottIm
F. Xu, S. Wen/ Materials Letters 28(1996)401-407 (MS), quantitative analysis of the interphase of the transformation in Fig. 3a confirmed the existence of the transition rcgion and consequently gavc more letails of it, such as dislocations and some regularity (11) of the atomic displacement. In Fig. 3a. we chose a (1x region at the interface including about 10 planes parallel to(010) or(100)m. The spectrophotometer then scanned normal to(010), plane from one end to measuring the thickness of each interpla Then after unitization using b/2 of the tetragonal jolo phase as the unit, we determined the lattice displace (not ment along the direction normal to the(010), plane as shown in Fig. 3c. It can be seen tha (010), planes were compressed in the transition re gion. The distance between two neighboring planes (100(001)a reduced to 0.84 of the original b/2 lattice distance, gether with some dislocations. Moreover, the the relationship between t-and m-lattice inin situ investigation (200)m plane in the transformed monoclinic area did not line up evenly as it should be in the theoretical (100)ml010) [011m I[1011: [100]8.6 to [0101; atomic structure model, whereas these lattice plane [01OJm l [o01 and [oo1] I[ 1001, were distributed as that a wide interplane and Theoretical calculations predicted the habit plane narrow interplane were alternately arrayed, indicat- lying between the (130)and(140) planes of the ing incomplete transformation and a gradient of the trigonal phase [8,9. In Fig. 3a, we can define the atomic displacement during the transformation habit plane indicated as a dotted line which lies 12 According to the definition of a martensitic trans to the(010), or (100)m plane. As the angles on formation, atomic movement would not exceed one (101) plane between (130), and(010),(140), and atomic distance. But in the t-m transformation of (010), are 1315and 9.94 respectively, the theoreti- Zro,, induced by electron beams, no twins were cal solution matches well the experimental observa- produced because a local deformation was produced by slip, and the atomic movement exceeded such a When analyzed closely, it can be shown that there limit. Furthermore, the atoms further away from the xists a transition region at the t/m interfaces, and nucleation site moved a longer distance during the nore often, higher densities of dislocations, whereas transformation. This is indicated in the structure hardly no dislocations are present in the transformed model in Fig. 3b. Therefore, it is not difficult to region, The tetragonal lattice was compressed due to understand that the long-distance movement of atoms the neighboring volume expansion occurring during requires the continuous support of extraneous forces. the transformation. That is to say, the front lattice 1. e. irradiation of electron beams, otherwise the vas subject to compression followed by dilation in transformation would easily be halted the process of the transformation. Fig. 3b gives the The morphologies of the transformation induced approximate atomic displacement of such a transition by electron beams are almost the same as those region. The discovery of the existence of a transition shown in Fig. 2 and 3. The transformation did not area at the t/m ZrO 2 interface has not otohotometer symmetrical relationship of twins, As twins are initi- een reported form twins but transforming bands which exhibit the before. By using a microscanning spectrol 3“ ogether with the illustration of the corresponding atomic structure model(b)showing a an MS analysis which confirms the existence of such a transition area
F. Xu, S. Wen /Materials Letters 28 (1996) 401-407 405 Fig. 4. Stereographic analysis indicating correspondence B-2 for the relationship between t-and m-lattice in in situ investigation. (100),11(010),; [Oll],ll[lOl],; [100],8.6” to [OlO],; [0101,1l[0011, and [0011,11[1001,. Theoretical calculations predicted the habit plane lying between the (130) and (140) planes of the tetragonal phase [8,9]. In Fig. 3a, we can define the habit plane indicated as a dotted line which lies 12” to the (OlO), or (loo),,, plane. As the angles on (101), plane between (1301, and (OlO),, (1401, and (01 O), are 13.15” and 9.94” respectively, the theoretical solution matches well the experimental observations. When analyzed closely, it can be shown that there exists a transition region at the t/m interfaces, and more often, higher densities of dislocations, whereas hardly no dislocations are present in the transformed region. The tetragonal lattice was compressed due to the neighboring volume expansion occurring during the transformation. That is to say, the front lattice was subject to compression followed by dilation in the process of the transformation. Fig. 3b gives the approximate atomic displacement of such a transition region. The discovery of the existence of a transition area at the t/m ZrO, interface has not been reported before. By using a microscanning spectrophotometer (MS), quantitative analysis of the interphase of the transformation in Fig. 3a confirmed the existence of the transition region and consequently gave more details of it, such as dislocations and some regularity of the atomic displacement. In Fig. 3a, we chose a region at the interface including about 10 planes parallel to (OlO), or (1001,. The spectrophotometer then scanned normal to (OlO), plane from one end to another, measuring the thickness of each interplane. Then after unitization using b/2 of the tetragonal phase as the unit, we determined the lattice displacement along the direction normal to the (OlO), plane as shown in Fig. 3c. It can be seen that arrays of (OlO), planes were compressed in the transition region. The distance between two neighboring planes reduced to 0.84 of the original b/2 lattice distance, together with some dislocations. Moreover, the (2001, plane in the transformed monoclinic area did not line up evenly as it should be in the theoretical atomic structure model, whereas these lattice planes were distributed as that a wide interplane and a narrow interplane were alternately arrayed, indicating incomplete transformation and a gradient of the atomic displacement during the transformation. According to the definition of a martensitic transformation, atomic movement would not exceed one atomic distance. But in the t-m transformation of ZrO,, induced by electron beams, no twins were produced because a local deformation was produced by slip, and the atomic movement exceeded such a limit. Furthermore, the atoms further away from the nucleation site moved a longer distance during the transformation. This is indicated in the structure model in Fig. 3b. Therefore, it is not difficult to understand that the long-distance movement of atoms requires the continuous support of extraneous forces, i.e. irradiation of electron beams, otherwise the transformation would easily be halted. The morphologies of the transformation induced by electron beams are almost the same as those shown in Fig. 2 and 3. The transformation did not form twins but transforming bands which exhibit the symmetrical relationship of twins. As twins are initiFig. 3. Lattice image (a) of the twin-related variants together with the illustration of the corresponding atomic structure model (b) showing a transition area at t/m interphase. (c) is the result of an MS analysis which confirms the existence of such a transition area
F. Xu, S Wen/ Materials Letters 28(1996)401-407 Fig. 5. HREM image showing microtwins in the transformed m-Zro, lattice revealing that each pair of twins was initiated by a dislocation ated from the area with high concentration of stress strain as the growing velocity of the twins is ex polycrystalline materials, usually the interf tremely large, nearly close to the spreading velocity region is the site which accommodates most stress), of a shock wave. Therefore, it is understandable that the critical shear stress required for the formation of the concentration of stress in the sample due to the twins is much greater than that required for the slip irradiation of electron beams is comparatively weak Fig. 6. Lattice image of multiple twins produced under conditions other than electron beam irradiation
406 F. Xu, S. Wen/Materials Letters 28 (1996) 401-407 Fig 5. HREM image showing microtwins in the transformed m-ZrO? lattice revealing that each pair of twins was initiated by a disloca Ition. ate d from the area with high concentration of stress (in polycrystalline materials, usually the interface reg ;ion is the site which accommodates most stress), the critical shear stress required for the formation of twi ns is much greater than that required for the slip strain as the growing velocity of the twins is extremely large, nearly close to the spreading velc xity of a shock wave. Therefore, it is understandable that the concentration of stress in the sample due to I the irradiation of electron beams is comparatively w eak. Fig. 6. Lattice image of multiple twins produced under conditions other than electron beam irradiation
F. Xu, S. Wen/Materials Letters 28(1996)401-407 eading to a slow transformation, hence no twins are 4 Conclusions produced. This is because the formation of twins is rcquircd to reduce the pure point system strain and (1) HREM observations revealed for the first time accommodate the quick change of volume and shape the fine interfacial microstructures of the t/m ZrO, during transformation. Thus an control the after the martensitic transformation. There exists a transformation only by adjusting the intensity of th transition area in-between the transformed phase (m- electron beam, and this is why the observation of the 7r0)and the parent phase(t-ZrO, ). The tetragonal d rocess of the martensitic transformat lattice is subject to a compression followed by a Zro, could be realized dilation during the transformation Fig. 5 shows the lattice image of microti 2)The complete transformation process is gov formed d nsformation were not emed by a different mechanism in each stage. The produced by electron beams. It can be seen that at nucleation stage is govemed by thermal stress pro each end of each pair of twin laths, there exists an duced with irradiation of electron beams, whereas edge dislocation of similar direction and on the same the grain growth is controlled by clectron irradiation (3)A comparison was made of the factors that le made each side of the slip plane endure different kinds of stress. The atom bonds over the slip plane multiple twins and twin-related lathe ng microtwins the formation of twins. includin are broken indicating that this site is subject to compressive stress, whereas lattices under the slip plane are dilated and are subject to tensile stress. The micrograph shows that microtwins were rendered at Acknowledgements the region subject to tensile stress. In the case of the ZrO, martensitic transformation in order to keep the The authors wish to acknowledge Dr. Zhang co-lattice condition, sliding made it easy to produce Yufeng for providing the samples for this study and edge dislocations periodically with the same direc Chen Yangguang for the results of Ms analysis tion, and an array of dislocations would inevitably cause the stress concentration. As the stress required for the formation of twins in such a deformed and twisted crystal is much lower than that for a perfect References crystal, microtwins are much easier to form in the area adjacent to the dislocations. Moreover, only a [1 M.S. Wechsler, D.S. Lieberman and T.A.Read, Trans.AIME tensile stress can induce the formation of microtwins 197(1953)1530 in the case of ZrO, ceramics [2] D.S. Lieberman, M.S. Wechsler and T.A. Read, J. Appl. Phys For comparison, Fig. 6 shows the multiple twins 26(1955)473 normally produced in the course of fracture during [3] J.S. Bowles and JK Mackenzie. Acta Metall 2(1954)129 [4]JK Mackenzie and J.S. Bowles, Acta Metall. 2(1954)138 the preparation of discs, with much thicker laths [5] J.S. Bowles and J.K. Mackenzie, Acta Metall. 2(1954)224 compared with the inicrotwins but thinner than the [6] P. M. Kelly, Mater. Sci. Furunn 56-58(1990)335 twin-related laths formed by irradiation of electron 7] S.L. Wen and C.F. He, Nucl. Instr. Methods Phys. Res. 219 beams.The diffraction pattern(shown in the inset in he upper left hand corner) indicates the high density [8]M. Hayakawa, N. Kuntani and M. Oka. Acta Metall. 37 of dislocations and steps which were shown in the [9]M. Hayakawa and M Oka, Mater. Sci. Forum 56-58(1990) area of interfaces
F. Xu, S. Wen /Materials Letters 2X C 1996) 401-407 407 leading to a slow transformation, hence no twins are produced. This is because the formation of twins is required to reduce the pure point system strain and to accommodate the quick change of volume and shape during transformation. Thus, we can control the transformation only by adjusting the intensity of the electron beam, and this is why the observation of the dynamic process of the martensitic transformation of ZrO, could be realized. Fig. 5 shows the lattice image of microtwins formed during the transformation which were not produced by electron beams. It can be seen that at each end of each pair of twin laths, there exists an edge dislocation of similar direction and on the same slip plane. This slip plane is the habit plane. Shear made each side of the slip plane endure different kinds of stress. The atom bonds over the slip plane are broken indicating that this site is subject to compressive stress, whereas lattices under the slip plane are dilated and are subject to tensile stress. The micrograph shows that microtwins were rendered at the region subject to tensile stress. In the case of the ZrO, martensitic transformation, in order to keep the co-lattice condition, sliding made it easy to produce edge dislocations periodically with the same direction, and an array of dislocations would inevitably cause the stress concentration. As the stress required for the formation of twins in such a deformed and twisted crystal is much lower than that for a perfect crystal, microtwins are much easier to form in the area adjacent to the dislocations. Moreover, only a tensile stress can induce the formation of microtwins in the case of ZrO, ceramics. For comparison, Fig. 6 shows the multiple twins normally produced in the course of fracture during the preparation of discs, with much thicker laths compared with the microtwins but thinner than the twin-related laths formed by irradiation of electron beams. The diffraction pattern (shown in the inset in the upper left hand corner) indicates the high density of dislocations and steps which were shown in the area of interfaces. 4. Conclusions (1) HREM observations revealed for the first time the fine interfacial microstructures of the t/m ZrO, after the martensitic transformation, There exists a transition area in-between the transformed phase (mZrO,) and the parent phase (t-ZrO,). The tetragonal lattice is subject to a compression followed by a dilation during the transformation. (2) The complete transformation process is governed by a different mechanism in each stage. The nucleation stage is governed by thermal stress produced with irradiation of electron beams, whereas the grain growth is controlled by electron irradiation. (3) A comparison was made of the factors that led to the formation of twins, including microtwins, multiple twins and twin-related laths. Acknowledgements The authors wish to acknowledge Dr. Zhang Yufeng for providing the samples for this study and Chen Yangguang for the results of MS analysis. References [I] MS. Wechsler, D.S. Lieberman and T.A. Read, Trans. AIME 197 (1953) 1530. [2] D.S. Lieberman, MS. Wechsler and T.A. Read, J. Appl. Phys. 26 ( 1955) 473. (31 J.S. Bowles and J.K. Mackenzie, Acta Metall. 2 (1954) 129. [4] J.K. Mackenzie and J.S. Bowles, Acta Metall. 2 (1954) 138. [5] J.S. Bowles and J.K. Mackenzie, Acta Metall. 2 (1954) 224. [6] P.M. Kelly, Mater. Sci. Forum 56-58 (1990) 335. [7] S.L. Wen and CF. He, Nucl. Instr. Methods Phys. Res. 219 (1984) 333. [8] M. Hayakawa, N. Kuntani and M. Oka. Acta Metall. 37 (1989) 2223. [9] M. Hayakawa and M. Oka, Mater. Sci. Forum 56-58 (1990) 383