You are on page 1of 5

Materials Transactions, Vol. 47, No. 3 (2006) pp.

607 to 611 Special Issue on Shape Memory Alloys and Their Applications #2006 The Japan Institute of Metals

Acoustic Emission as a Probe of the Kinetics of the Martensitic Transformation in a Shape Memory Alloy
Stefanus Matheus Cornelis van Bohemen1;2 , Jilt Sietsma1 , Roumen Petrov3 , Marcel Joseph Marie Hermans1 and Ian Malcom Richardson1;2
1 Department of Materials Science and Engineering, Delft University of Technology, Rotterdamseweg 137, 2628 AL Delft, The Netherlands 2 Netherlands Institute for Metals Research, Mekelweg 2, 2628 CD Delft, The Netherlands 3 Department of Metallurgy and Materials Science, Ghent University, Technologiepark 903, B-9052 Zwijnaarde (Gent), Belgium

The kinetics of the martensitic transformation in a CuAlMn shape memory alloy (SMA) has been studied using the acoustic emission (AE) technique. It is demonstrated that the volume fraction of martensite as a function of time and temperature can be derived from the measured AE power. The fraction data obtained can be described by the Koistinen and Marburger (Acta Metall. 7 (1959) 59) equation with high accuracy, which indicates that the nucleation of martensite takes place heterogeneously and that the average volume of martensite crystals is constant over the extent of the transformation. The martensite-start temperature determined from the measured AE data is in good agreement with the value found by differential scanning calorimetry (DSC). Furthermore, the results of AE experiments on the SMA are compared with optical Confocal Laser Scanning Microscopy (CLSM) observations of the surface of the SMA. The observations show that both small and large martensite plates are formed both at the beginning and at the end of the transformation, which is in agreement with the assumption of a constant average volume of martensite crystals used in the Koistinen and Marburger model. (Received August 30, 2005; Accepted November 7, 2005; Published March 15, 2006) Keywords: acoustic emission, martensitic transformation, shape memory alloy, optical confocal laser scanning microscopy, differential scanning calorimetry

1.

Introduction

Martensitic transformations can occur in many metals and alloys provided the conditions are such that diffusioncontrolled transformations are prevented.1) The transition is governed by thermodynamics and kinetics. The underlying thermodynamics determine the available driving force for transformation; the kinetics of a martensitic transformation depend solely on nucleation, because the growth of a martensitic crystal usually occurs rapidly. It is well known that the mechanism of growth is displacable. How the phase nucleates, however, is even today not completely understood. This is mainly due to the great speed of formation, which makes the martensitic transformation a difficult process to study experimentally. During cooling, the transformation starts at a certain temperature, which is designated the martensite-start temperature Ms , and proceeds only upon further cooling below this temperature.1) In order to gain more insight into the kinetics of the martensitic transformation, it is necessary to investigate quantitatively how the transformation progresses upon continuous cooling. Characterization of the volume fraction martensite can be realized by electrical resistivity,2) quantitative metallography,3) dilatometry,4) differential scanning calorimetry5,6) and acoustic emission.2,7) About forty years ago, Koistinen and Marburger presented an empirical equation that describes the volume fraction of residual austenite as a function of temperature below the martensite-start temperature.8) This equation, named the KM relation, was obtained by fitting their (and other investigators’) data from X-ray measurements on plain carbon steels. Some years later, Magee9) showed that the KM relation can be derived theoretically from basic principles, taking into account certain assumptions regarding the nucleation and

growth of martensite. The most important assumption of this theory is that the average volume of the martensite crystals is constant over the extent of the transformation. On this point the KM relation is different from a theory formulated by Fisher,10) which assumes that the average volume decreases strongly as the transformation proceeds. The derivation of the KM relation and the partitioning model of Fisher are discussed in detail in the next section. Acoustic emission is the term used for the elastic waves generated by abrupt changes in the internal stress field of a material.11) These waves propagate through the material and cause surface displacements, which can be converted into an electrical signal using a transducer. The detected signals carry information of the underlying processes occurring at the source. Some examples of acoustic emission sources are plastic deformation (dislocation movement), crack growth and martensitic phase transformations. In the present paper the acoustic emission generated during martensite formation in a shape memory alloy has been studied. The measurements were carried out under continuous cooling conditions. The transformation temperatures derived from the AE data are compared with the results obtained by means of DSC, and the accuracy of the AE technique is evaluated. In addition, Confocal Laser Scanning Microscopy (CLSM) was employed to obtain further insight into the evolution of martensite crystals as the transformation progresses. The objective of this study is to obtain fundamental knowledge about the transformation, such as nucleation rates and the average volume of martensite crystals. It is shown that the measured power of acoustic waves generated during the martensitic transformation yields information about the volume fraction of martensite as a function of undercooling below the Ms temperature. This allows a direct comparison of experimental data with theoretical predictions.

J. which is known as the geometrical partitioning effect. expressing the athermal character of the transformation. The nucleation of martensite is believed to take place at structural imperfections in the parent phase and these nuclei are stimulated to grow into martensite crystals at different degrees of undercooling below Ms . Magee9) showed that this empirical equation (the KM relation) can be derived from first principles assuming that in a temperature decrease dT . which is initially fully austenitic.85 mass%Cu. C.9. Richardson 2. the material was annealed for 10 h at 900 C in a protective atmosphere (Ar) followed by water quenching. M. From this derivation it follows that the positive parameter C1 (the rate constant) can be written as C1 ¼ ÀC2 d ð ÁG ! Þ dT 0 ð2Þ with C2 a positive constant expressing the proportionality between the increase in driving force and the consequent increase in density of activated nucleation sites. The decrease of the average volume of martensite crystals can be rationalized by considering that the initial martensite crystals divide the austenite grain into progressively smaller compartments. the probability of nucleation is greater in The CuAlMn SMA studied has a chemical composition of 80. and C1 is a constant. can be described by f ¼ 1 À expðÀC1 ðMs À T ÞÞ ð1Þ the vicinity of a previously formed crystal (dislocation debris) than it is for random nucleation in untransformed regions.608 S. which is in favour of the Magee model. van Bohemen.1) Since growth can be very fast. each nucleation event directly leads to the formation of a certain volume of the new phase. calculations of the energy barriers to nucleation indicate that the nucleation must take place heterogeneously on pre-existing embryos.10) which assumes that  decreases strongly as the transformation progresses. Petrov. Hermans and I.12. it is often observed that martensitic crystals tend to occur in clusters. The Ms temperature is determined by thermodynamics and the rate constant C1 describes the progress of the transformation for a certain undercooling. M. According to the above equation the progress of the transformation from austenite to martensite is quantitatively described by the Ms temperature and the value of C1 . This fraction is defined as the volume of martensite divided by the volume of austenite that exists in the sample prior to the formation of martensite. About forty years ago. R. which can be shown to be an adequate approximation by thermodynamic calculations. M. The alloy undergoes a thermo-elastic martensitic transformation both during cooling and during heating. During cooling below Ms the high temperature .12) 3. 13.9) Although the exact nature of the nucleation sites is not completely understood.13) If the nucleation is assumed to be heterogeneous (autocatalysis) it can be shown that the average volume of martensite crystals is independent of the volume fraction of martensite. they have different energy barriers to activation. Thus the volume fraction of martensite varies only with the degree of undercooling. and  the average volume per newly formed crystal. however. On the other hand. This model assumes that nucleation occurs homogeneously as a result of random fluctuations. M. The overall transformation rate is governed by nucleation and growth. the number dN of new martensite crystals (plates or laths) that form per unit volume of austenite is proportional to the increase in driving force 0 ÁG ! that is caused by the temperature decrease. As pointed out earlier.26 mass%Mn. After casting. dÁG ! =dT is also assumed constant. which is assumed 0 constant over the extent of the transformation. it was postulated by Koistinen and Marburger8) that the evolution of martensite formation in a sample. J. Theory Theories of the kinetics of the athermal martensitic transformation attempt to describe the progress of martensite formation in a sample.68 mass%Al and 5.13) This effect is known as strain-induced autocatalysis. the assumption that  is constant is in contraction with the Fisher model. Sietsma. Experimental where f is the volume fraction of martensite in the sample at a temperature T below Ms .

-phase (BCC) transforms to .

0 -martensite (M18R).14) Upon heating an inverse transformation to the .

The sample was reheated to a temperature of 90 C. which is above the Af temperature of the alloy.26 mV. Subsequently the sample was cooled to À20 C using a Dewar filled with liquid nitrogen. The rms voltage was measured with a resolution of 0. (6) thermocouple. (4) pre-amplifier (60 dB). The mean square voltage U 2 corrected for the background noise (AE power) is defined by Z 1 tþ 2 0 0 2 U 2 ðtÞ ¼ Up ðt Þdt À Un ð3Þ  t where Up ðtÞ is the voltage output at the transducer and  a time constant chosen as  ¼ 0:5 s. The temperature data of the specimen measured with a K-type thermocouple were (6) (1) (2) (3) Fig. which was cut of an as-cast and annealed block. 1(a) schematic drawing of the experimental set-up is given. (2) waveguide. During cooling. 1 Schematic diagram of the experimental set-up used for AE measurements during cooling of the CuAlMn shape memory alloy: (1) specimen.-phase occurs in the temperature range As to Af . The AE experiments were conducted using a disc-shaped sample with a diameter of 15 mm and a thickness of 1 mm. the root mean square (rms) voltage of the continuous acoustic emission signals was recorded using a PAC (Physical Acoustics Corporation) AE system. using an air heater.24–0. In Fig. (3) transducer. (5) AE analyzing system. A waveguide was spot welded onto the sample to transport the acoustic waves to the sensor and to protect the sensor from excessive heating and/or cooling. (5) (4) . which resulted in a cooling rate of approximately 1 C/s.02 mV relative to a background noise level of 0.

With the integrated value U 2 dt. 2(b). The analysis of the data was started from the maximum in the AE signal. 2(a). The acquisition rate for temperature and AE power measurements was two data points per second. the proportionality factor k can be calculated as k ¼ 25 Â 103 V2 s/m3 . It can be seen that lnð1 À f Þ varies linearly with temperature over the entire extent of the transformation. Subsequently. the AE power can be converted to the volume transformation rate dV =dt. It should be noticed that f ðtÞ is independent of k and Vm . The SMA sample for the optical CLSM-observations. the sample surface was ground and polished at a high temperature by using hot water. The data obtained by the DSC measurements are used for comparison and evaluation of the accuracy of the AE technique.15. (b) the transformation rate and (c) the volume fraction of martensite plotted against time during the martensitic transformation in the shape memory alloy. was placed in a petri-dish and by filling the petri-dish with ice-water the martensitic transformation was induced. 3. 4. Results and Discussion The martensitic transformation in the CuAlMn SMA was investigated using the three techniques described above. 2(c)]. Two thermal cycles between 150 and À100 C at a constant heating and cooling rate of 10 C/min were performed using a NETZSCH DSC 404C differential scanning calorimeter. with a diameter of 15 mm and a thickness of 10 mm. In . in the case when thermal gradients are negligible the rise time of the signal is even shorter. where the mean square voltage of acoustic emission is plotted as a function of time. the fraction curve obtained is compared with the kinetics predicted by the Koistinen–Marburger model. Although the grinding scratches could not be removed completely. The fraction of martensite f at a certain time can now be calculated as the area under the peak up to that specific time divided by the total area under the peak [Fig. which in combination with eq. (4) can be written as df 1 dV U2 ¼ ¼Z dt V m dt U 2 dt ð5Þ with V the volume R of martensite in the sample. f and À lnð1 À f Þ are plotted against temperature in Fig. the SMA was investigated by differential scanning calorimetry (DSC) to determine the transformation characteristic.1 AE measurements It can be understood from Fig. the polishing was found adequate to reveal the surface upheavals due to the transformation. It has been shown in a previous study that the AE energy generated between the start and the finish of the transformation is proportional to the volume of martensite formed with proportionality factor k. The change of the martensite Thus d f =dt / U 2 and the calculated values for the transformation rate are plotted in Fig. the AE power is proportional to the volume transformation rate dV =dt.16) In other words. Finally. To facilitate comparison with the KM relation. In addition. that the martensite formation is reflected by a peak in the AE data. Prior to the measurements. U2 ¼ k dV dt ð4Þ Fig.Acoustic Emission as a Probe of the Kinetics of the Martensitic Transformation in a Shape Memory Alloy 609 imported into the Mistras 2001 system. the backgroundcorrected AE power [eq. 2 The values of (a) the AE power. First it will be demonstrated how the volume fraction of martensite can be directly evaluated from the measured AE power. This results in a broadening of the peak. the results of the DSC measurements and the optical microscopy observations are discussed and compared with the results obtained by the AE technique. When the transformation starts. 4. the values of d f =dt. Optical Confocal Laser Scanning Microscopy (CLSM) insitu observations of the surface of the SMA were performed to get further insight into the growth of martensite plates. The transformation is assumed to reach completion ( f ! 1) where U 2 becomes equal to zero. at the maximum in the signal the thermal gradients are believed to have diminished due to the release of latent heat in the regions where transformations first began. Probably. volume fraction f is the normalized volume transformation rate. (3)] rapidly increases to a maximum value and then tails off to zero as the transformation proceeds. Using this value of k. The few data points during the rise time of the signal may be attributed to the thermal gradients in the sample leading to localized transformations as the sample approaches the martensite start temperature. and the value for the final volume of martensite (Vm ¼ 177 mm3 ). with the integration running over the entire transformation range.

Hermans and I. The results obtained for the shape memory alloy are in good agreement with results obtained in a previous study using low-alloy steels. Fig. Petrov. Sietsma. 3 The progress of the martensitic transformation in the SMA as a function of undercooling: (a) the change in volume fraction.610 S. J. 4 The heat flow measured by DSC during cooling of the CuAlMn Shape memory alloy. The solid line is the fit of the data to the KM equation. the solid line through the data represents the leastsquares fit to the KM equation [eq. R. The solid line through the data is the least-squares fit to the KM equation. J. M. van Bohemen. 5 The volume fraction of martensite as a function of temperature derived from the DSC measurements. Richardson Fig.7) 4.2 DSC measurements Figure 4 shows the DSC heat flow curve measured during cooling. Fig. (b) the volume fraction and (c) the volume fraction data plotted on a logarithmic scale. M. The extracted fit parameters are: Ms ¼ 27 C and C1 ¼ 0:21 KÀ1 . The high accuracy of the fit of the data to the KM equation implies that 0 C1 (¼ C2 dÁG ! =dT ) is constant over the extent of the transformation. which strongly indicates that Magee’s assumptions9) are justified. M. 3(c). M. which reflects the formation of . Fig. (1)]. A peak in the cooling curve is seen. C.

the thermocouple calibration) of both experimental set-ups. represented by the solid line in Fig. The small difference in start-temperature may be related to the differences in the temperature control system (e. which revealed interesting information about the growth of martensite plates. 5. This thickening effect makes the interpretation in terms of nucleation and growth complex. it follows that Ms ¼ 22 C and C1 ¼ 0:16 KÀ1 . 5. This is in good agreement with the results obtained by the AE technique. A partitioning effect and a progressive .g. This is not surprisingly since both techniques measure a part of the energy released during the martensitic transformation. A thickening of these plates is observed as the transformation proceeds. Nevertheless. it is not clear whether a thickening plate may be treated to form in a single nucleation and growth step. Analogous to the analysis of the AE data the fraction f can be evaluated from the area under the DSC peak. and for some plates this thickening clearly occurs in a discontinuous manner.0 -martensite. six characteristic micrographs are shown in Fig. 6. The peak in the DSC data is quite similar to the peak observed in the AE data. 4. The values of f and À lnð1 À f Þ are plotted against temperature in Fig. optical Confocal Laser Scanning Microscopy (CLSM) observation (and video recording) of the surface of the SMA was carried out in order to gain a better insight into the evolution of martensite crystals as the transformation progresses. The transformation starts with the formation of thin plates of martensite with distinct orientations. During cooling of the SMA sample the confocal microscope acquired 1 image per second.3 Optical CLSM observations In addition to the AE measurements. the observations show that small and large plates are formed both at the beginning and at the end of the transformation. From the least-squares fit of the fraction data to the KM equation.

Fisher. G. Trans. A. London. M. Easterling: Phase transformations in metals and alloys (Chapman & Hall. K. M. Sci. (1989) p. 7) S. London. B. van Humbeeck. Bhadeshia: Mater. M. H. J. van Bohemen. J. Bhadeshia: Bainite in steels (The Institute of Materials. Phys. 51 (2003) 4183–4196. L. Eng. 11) H. M. 2001) pp. Turnbull: AIME Trans. 13) H. Ortin and A. 3 (1980) 41–64. G. E. London. M. it can be concluded that the acoustic emission measurements in conjunction with the optical CLSM observations during the martensitic transformation in the studied CuAlMn shape memory alloy have resulted in a better understanding of the nucleation and average volume of martensite crystals. Richardson: Acta Mater. This is in good agreement with the results of the AE measurements on the SMA discussed earlier. J. C. Phys. Mwumba: Metals Forum 4 (1981) 164–174. In particular it was shown or found that: – The experimental data can be fitted to the KM equation with high accuracy over the entire extent of the transformation. den Ouden and I. E. van Bohemen. Wadley. It is important to notice that this transformation behaviour is different from the evolution of martensite crystals in the Fisher model. J. Chandrasekaran. W. Koistinen and R. 5. 4) S. A 129 (1990) 257–272. 10) J. Planes: Acta Metall. Magee: Phase Transformations (American Society of Metals. M. M. McMurtrie and C. D. M. 6) J. N. 9) C. This is based on the fact that the volume fraction of martensite as a function of time during cooling can be derived directly from the measured AE power. Lyon. Rodriguez: Ph. Delaey. M. which is in good agreement with the Magee model. den Ouden: J. 3) D. D. Hermans. R. Andrade and M. K. Phys. The transformation characteristics derived from the AE data are in good agreement with the results obtained by the DSC technique. OH 1970) pp. J. Conclusions CuAlMn shape memory alloy give valuable information about the sizes of martensite crystals formed during transformation. 382–437. Metals Park. It was shown that both small and large plates are formed both at the beginning and at the end of the transformation. P. 115–156. Magee: Metall. K. Hermans and I. 1995) pp. C. 2) G. 16) S. Hollomon and D. 12) R. REFERENCES 1) D. Speich and R. 36 (1988) 1873–1881. Porter and K. 35 (2002) 1889–1894. decrease of  as proposed by Fisher10) is clearly not observed.Acoustic Emission as a Probe of the Kinetics of the Martensitic Transformation in a Shape Memory Alloy 611 Fig. Janssen. H. K. M. M. S. van Bohemen. 34 (2001) 3312–3317. C. D Thesis. Marburger: Acta Metall. Fisher: Acoustic Emission ASTM STP 505 (1972) 140–151. N. Hermans and G. D. H. 2001) pp. H. Richardson: J. G. A. According to Magee this means that the average plate volume  is constant over the extent of the transformation. M. C. – The optical CLSM observations of the surface of a . 76–98. Khan and H. I. Speake: Int. Honeycombe and H. Sietsma. order number 89ISAL0081. 1 (1970) 3185–3191. L. C. 7 (1959) 59–60. 204. It has been demonstrated that acoustic emission is a very suitable technique to study the kinetics of the martensitic transformation in a shape memory alloy. 185–187. A. 14) L. – The start-temperature of the martensitic transformation in the SMA can be determined by means of acoustic emission with high precision. In general. M. D: Appl. Metals Rev. 15) S. Phys. M. Scruby and J. 185 (1949) 691–699. Bhadeshia: Steels (Edward Arnold. which predicts that  strongly decreases as the transformation proceeds. J. D: Appl. J. 5) P. 6 Observation of the change in microstructure during cooling of the SMA sample over a period of 30 s. 8) D.