Acoustic emission in materials research--a review.
Acoustic emission (AE) is used widely in non-destructive evaluation/structural health monitoring to detect the presence of dangerous flaws and to locate the positions of such flaws in structures of various kinds. It is also indispensable for the investigation of dynamic behavior of materials. During the second half of the 20th century, AE from materials, especially AE from deformation and fracture, was studied extensively, cf., Journal of Acoustic Emission (1982-) and conference proceedings of IAES (Japan), ICAE (US), EWGAE (Europe) and AECM (US/Europe). Two recent books are also valuable in understanding the current status of this broad field [1, 2].
From these past studies, we have gained substantial knowledge on the origins of AE signals and can utilize it effectively in most NDE applications. However, our knowledge on the science of AE is hardly complete because of the limitation of experimental and theoretical resources at the time these studies were made. This article builds on reviews by the present author [3, 4], examines new methods of AE signal analysis and attempts to summarize the progress on subjects related to materials research.
Methods of AE Signal Analysis
When one records an AE signal, it contains the information on its source, the propagation medium (sample or structure), AE sensor and electrical instrument used. These can be expressed by the characteristic or transfer functions of the source, propagation medium, sensor and electronics ([H.sub.s], [H.sub.m], [H.sub.t], and [H.sub.e]). In the frequency domain, the transfer function of the AE signal, [H.sub.AE], is given by the product of the four transfer functions above:
[H.sub.AE] = [H.sub.s] x [H.sub.m] x [H.sub.t] x [H.sub.e]. (1)
This can be performed easier as summation in the logarithmic scale and Fig. 1 shows an example from Sause [5; Fig. 2.20], who modeled a signal from a fiber composite. While AE source is given as a smooth function decreasing with frequency (-6 dB at 1.8 MHz), the medium has the major effect on the final signal detected. In this example system, even a large change in [H.sub.s] (e.g., -6 dB at 900 kHz) is hardly distinct.
[FIGURE 1 OMITTED]
[FIGURE 2 OMITTED]
Although the frequency spectra are often adequate in NDE applications of AE, it is essential to examine the time-domain features of AE sources and signals in order to understand the physical basis of AE. From complete transfer functions (inclusive of phase behavior), we can inverse-Fourier transform H's and gain their time-domain equivalent (denoted as h's here), i.e., waveforms. In the time domain, however, we need to utilize convolution integral (or deconvolution) instead of frequency-domain multiplication (or division); see  for details, but a convolution integral of [h.sub.1] and [h.sub.2] yielding [h.sub.3] is written as (2) below.
[h.sub.3] = [h.sub.1] [dot encircle] [h.sub.2] = [[integral].sup.t.sub.0] [h.sub.1](s)[h.sub.2](t - s)ds 0 < t < [infinity] (2)
Implementing convolution and deconvolution procedures with typically large AE signal files requires considerable expertise in software development. Sause  recently changed this situation by making AWARE++ software available as freeware. (http://www.physik.uni-augsburg.de/ exp2/downloads/aware.html)
Applying the above principles requires careful consideration and many simplifying assumptions. Initial successful work to determine [H.sub.AE] was developed in the 1980's and is collectively known as "deconvolution analysis". This method evaluated AE from crack advances on compact tension samples for fracture toughness tests. See [3, 7, 8]. Outcome was the knowledge of Hs, or specifically, the crack dynamics and crack volume. Wadley and Scruby  examined low temperature fracture of iron (intergranular) and mild steel (cleavage) using a special specimen shape, called Yobell. Figure 2 shows two cases for brittle fracture, giving the displacement (measured by a capacitive sensor) and deduced crack volume vs. t. In these studies, the crack plane is known and a calibrated sensor with Ht was at the epicenter position. Theoretical Green's function for [H.sub.m] was used. The key step is to get [H.sub.s] from [H.sub.AE] via inverse (deconvolution) processing using known [H.sub.m], [H.sub.t], and He. However, this is prone to instability due to the superimposed noise in [H.sub.AE] . This approach is also restricted severely because [H.sub.m] is unavailable for most test geometries and was seldom used since around 1990.
[FIGURE 3 OMITTED]
[FIGURE 4 OMITTED]
First principle elastodynamic calculations of AE waveforms are possible for simple geometries. Ohtsu and Ono [9a,b] formulated displacement solutions assuming smooth-rising source functions of varying rise time, that is, [h.sub.s] [dot encircle] [h.sub.m]. Here, sources are described in terms of dislocation motion, which can also be given using a moment tensor, commonly used in geophysics. Such modeling or simulation process is most useful in obtaining details of a source by seeking a good match with experimental waveform, i.e., source parameters are used as variables to get the best fit of calculated and observed waveforms, thus, avoiding the deconvolution procedure. Takemoto and coworkers [10a,b, 11] have developed this forward-processing scheme and used it in various AE analyses in composite materials under loading. For example, Fig. 3 shows four types of failure, displacement signals observed and matched simulated waveforms for a unidirectional glass-fiber composite. This led to signal classification as a function of loading. From the source function, crack velocity estimates were also obtained. See Fig. 4. This source function analysis proved its utility in many areas.
A useful development from the moment-tensor representation of source motion is "moment tensor analysis" or moment tensor inversion. This method relies on the first arriving P-waves at multiple sensor locations and obtains the tensile and shear components of the source motion. This approach (available as "PAC special moment tensor" software package: http://www.acoustic emission.com/index.aspx) is most suited for large civil engineering structures, and has recently been reviewed elsewhere [in Chaps. 5 and 8, ref. 2]. In concrete and rock testing, the source types of AE during crack propagation have been identified. In most other materials testing, however, sample sizes are too small to use this method.
The use of 3D finite element analysis (FEA) is the next step in characterizing AE signals. Hamstad and coworkers have conducted a wide range of calculations, mainly directing their attention to AE signals generated and propagated on a plate. Starting from the modeling of Lamb waves generated by pencil-lead fracture in thin and thick plates [12a,b], FEA results were compared with modal superposition method [13a], and effects of source rise time, monopole vs. dipole sources, plate thickness and plate width were examined [13b, 14, 15]. FEA method is especially valuable for modeling interior AE sources as these cannot be simulated experimentally. See also http://www.engr.du.edu/profile/Marvin.htm. Figure 5 shows a comparison between experimental and computed out-of-plane displacement from a pencil-lead break source at distance of 203 mm [12b]. 50-kHz high-pass filter was applied numerically.
[FIGURE 5 OMITTED]
Another subject of critical importance is AE sensors and their transfer functions, [H.sub.t]. Circular capacitive sensors used in  have good displacement responses to plane waves, but develop a high-frequency limit for spherical waves and have low sensitivity . Commonly used piezoelectric disc sensors have complex responses due to cross coupling of radial and thickness vibration modes, especially when incident waves are spherical or Rayleigh or Lamb waves. We have examined the sensor characteristics using laser interferometer as reference standard [16a,b]. Sensor responses depend on the wave types, i.e., plane waves, and spherical waves with different radius of curvature. The sensor responses were determined by the excitation of input waves using an ultrasonic transmitter or pulse-laser in conjunction with a transfer block. In addition, it was noted that the ultrasonic transmitter generated radial waves that excited the radial mode of sensor under test. Displacement due to the input waves was obtained using laser interferometer, with which the deconvolution of sensor output yields the transfer function. Because of the source-sensor distance dependence of the transfer function, [H.sub.t], its use in equation (1) is much more complex. It is further anticipated that the incident angle of waves also changes sensor responses. An example of source-sensor distance dependence for PAC-WD sensor is shown in Fig. 6, where upper graph is for a plane-wave input and bottom one is for a 25-mm radius spherical wave input [16a].
[FIGURE 6 OMITTED]
This behavior of AE sensors based on common piezoelectric discs and rings makes it impractical to broaden the usage of Ht beyond specialized research. It is still essential to develop suitable calibration methods that respond to AE waves encountered in laboratory and field applications. In particular, one must take into account radial mode excitation in commonly used face-to-face sensor calibration. A new method, the bar-wave based approach of sensor evaluation, is recently demonstrated [16b]. We utilized bar waves with a long flat metal bar. The waves are excited using an ultrasonic transmitter and displacements of the bar are characterized with a laser interferometer as before. Frequency from 20 kHz to 1 MHz is used for characterization purpose. Using FFT, the power spectral density of AE sensors was obtained. Figure 7 shows the bar-wave calibration of PAC-R15 sensor thus obtained. Note the excitation of the radial component of this sensor at 150 kHz.
[FIGURE 7 OMITTED]
The most advanced scheme for understanding the nature of AE signals relies on full-scale modeling of AE sources, propagation and sensing. Sause  conducted a detailed study of composite failure covering the entire sequence from the modeling of fracture, wave propagation and piezoelectric sensing, all utilizing finite element analysis. His analysis of fracture takes into account the finite extent of AE sources and the inhomogeneous elastic properties in its vicinity. In the planar specimen geometries employed, the source model excites Lamb waves with symmetric and antisymmetric modes, incorporating established procedures for Lamb wave modeling (see Hamstad above). A 3D model of the used sensor, PAC-WD including piezoelectric conversion was developed, simulating frequency dependent sensitivity of this sensor, but with a computationally more efficient sensor model. He demonstrated that the shape and frequency content of the simulated sensor signal are critically dependent on the aperture of the PZT-elements of the sensor.
Simulations of various source mechanisms result in excitations exhibiting different ratios of [S.sub.o]- and [A.sub.o]-modes in line with earlier work of Gorman and Prosser . Sause's results also show that a distinction of matrix-cracking and fiber-breakage at the same source position is possible based on different intensity ratios of symmetric and antisymmetric Lamb wave modes. Strong antisymmetric modes contribute to a strong lower frequency contribution, while strong symmetric modes correspond to high frequency intensity. These characteristics for matrix cracking (lower frequencies) and fiber breakage (higher frequencies) have been known based on experimental observations [17-19]. The different geometries, represented by homogenized plies with rotated elastic properties, significantly influence the simulated AE signals, yet a distinct dependency of the signal characteristics on the source mechanism is still retained. However, at larger (>50 mm) propagation distances, false attribution of failure mechanisms tends to increase due to Lamb wave dispersion. Figure 8 gives an example of fiber-breakage signals simulated for two stacking sequences [Fig. 3.26, ref. 5].
[FIGURE 8 OMITTED]
Developing own codes for full-scale FEA modeling is not for non-specialists, but it is noted that a software package, such as PZFlex , deals with sensor and propagation analysis. Fracture and vibration FEA packages are common. Thus, we hope to see further AE analysis development in the direction Sause superbly pioneered in his doctoral dissertation.
Wavelet transform (WT) has changed how we examine AE signals in the last 15 years. While this author was aware of WT's potential from an article in Dr. Dobb's Journal (April 1992) as digital equivalent of audio spectrograms (developed at Bell Labs and used for AE analysis of rocket motorcases at Aerojet by Green et al. in 1964 ), early books of Daubechies and Kaiser were of little help for self-implementation. In 1995, Yasuhisa Hayashi (Shizuoka University, Japan) completed a program for WT computation, the source code of which was published in 1996  with his former colleagues at Aoyama Gakuin University (AGU). Its uses were initially limited to Takemoto group , but became worldwide once collaboration between AGU and Jochen Vallen (Vallen Systeme, Germany) produced a freeware called, AGU-Vallen Wavelet in the past decade. This software now includes Choi-Williams transform calculation as well and has become an indispensable tool of AE workers everywhere. The utility of AGU-Vallen Wavelet is demonstrated broadly, so we show an example of another WT software, applied to AE signals from an SiC-SiC ceramic matrix composite. Here, projection of WT coefficients (left figure) and the FFT of the entire signal (right) are shown in Fig. 9. Hayashi and Wakayama  extracted the first and second major frequencies from the projected WT diagrams and correlated these to initiation and propagation of micro-damages, such as cracks in SiC matrix, crack propagation along thickness and fiber buckling. Different time segments of a WT diagram are projected and the major frequencies are obtained as shown. (In a WT diagram, the center frequency of the most (or second most) intense peak is defined as the first (or second) major frequency, but such peaks are obscured in a regular FFT spectrum. This feature of WT projection display is now available in AGU-Vallen Wavelet as well.
[FIGURE 9 OMITTED]
A new parameter for quantifying the unpredictability of fluctuations in time-series signals (i.e., AE waveforms) has recently appeared. Pincus  introduced a feature called approximate entropy, or ApEn, which expresses the likelihood of a similar pattern not being repeated subsequently. ApEn is small when many repetitive patterns exist in AE signals, while less predictable signals have higher values of ApEn. This was initially used in physiology and other biomedical fields (cf., http://www.physionet.org/physiotools/ApEn/; a link to an ApEn software is given), but its applications to AE signal analysis have been published [25, 26]. The utility of ApEn for AE analysis needs to be established for various conditions, but it is clearly shown in Fig. 10  that ApEn quantifies "noisyness" and that such a parameter was previously unavailable. Bearing defect size is shown to correlate well with ApEn in Fig. 11.
[FIGURE 10 OMITTED]
[FIGURE 11 OMITTED]
Another feature of AE signals worth examining is time statistic or temporal distribution of AE events. Manthei and Eisenblatter  reviewed works starting from Mogi's paper , giving indications of two types of statistics (Poisson or independent and Polya or prior-event dependent processes). This parameter may provide a useful angle in AE analysis.
While most AE signals can be described as burst events, certain phenomena produce "continuous" AE signals. These resemble white (broadband) noise or pink (frequency limited) noise and often found during plastic deformation due to dislocation glide, fracture/decohesion of inclusions, fluid/gas leaks and magnetization. Only their rms voltage levels provided a means of characterization. Ono et al.  evaluated continuous AE signals from a steel and found their fractal dimension to increase toward random (from 1.4 to 1.65, with 2.0 being white noise) as yielding is reached. Amplitude distribution also indicated Gaussian behavior. However, Lyapunov exponent and correlation dimension showed no apparent correlation. They also simulated Gaussian behavior from a laser-pulse input on a typical round-bar tensile sample, forming the basis of continuous AE signals via superposition. The origin of continuous AE signals with Gaussian distribution can be traced to the small source size and the irregular reflections and refractions of sample geometry. This produces spreading waves that are randomized through bouncing at the shoulders and enlarged grip regions.
Vinogradov and coworkers systematically examined AE from plastic deformation, including single crystal studies and successfully extracted a useful parameter in connection to dislocation theory of glide, i.e., median frequency, [f.sub.m] [30-32]. It is defined as 
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (3)
where G(f) is the power spectral density function of signal segment being considered. They also use the integrated G(f) over the detected frequency range as AE energy, E. Figure 12 gives plots of [f.sub.m] and E with the stress-strain curve of a copper single crystal . E values rose with strain during Stage I, followed by steady decrease during Stage II. The median frequency was ~250 kHz in Stage I (with a slight increasing trend), but [f.sub.m] increased to 550 kHz at 40% strain. This finding is in common with previous results for single crystals and polycrystals tested in tension where the AE power spectral density shifts to higher frequencies with increasing strain [e.g., 34]. This trend is viewed in terms of reduced glide distance of dislocations, and resultant increase in the number of glide loops. In a dilute Cu-Ge alloy  and in Fe , however, [f.sub.m] is essentially unchanged with strain and other rationales are needed. (cf., Fig. 13) Work on Ge-alloying of Cu  shows the reduction in the rms voltage of AE, attributed to the raised frictional stress due to solution hardening (in turn, increasing [f.sub.m] as more dislocations are needed to account for the reduced dislocation velocity). Another alloying effect appeared in the power density spectra, even though the experimental setups are basically identical, but twinning becomes active with higher Ge content. Work on Fe  demonstrates the presence of two types of strain instability and their consequences on AE behavior during yielding and necking. Recent activity in simulating work hardening with phase-field theory [cf., ref. 34] resulted in some reasonable predictions, such as increasing dislocation density with stress squared and increasing fractal dimensions . However, some tried to introduce dislocation avalanches, invoking burst AE observed in ice deformed near melting point . This has no physical basis. Earlier, such bursts were definitively linked to micro-crack observation . It is probably wishful thinking that these computational physicists examine complex reality of deformation and AE results.
[FIGURE 12 OMITTED]
[FIGURE 13 OMITTED]
AE in Materials Research
Deformation and Fracture
In the last decade, AE studies of plastic deformation have been limited with the exception of hcp alloys of Mg (and Ti and Zr to lesser extent). These materials deform via mechanical twinning, which has previously received much less attention than research on dislocation glide. For Mg alloys, see [37-41]; Ti alloy  and Zr alloy . Most of the AE work on Mg alloys unfortunately used just AE counts, so only indications of AE activities are monitored.
The most recent paper  used modern AE methods (getting PSD, [f.sub.m] and [V.sub.rms] digitally) and obtained corresponding in situ measurement of deformation twins with a third-generation ENGIN-X stress-strain diffractometer at a spallation neutron source in the UK. They used annealed ZM20 alloys (2Zn-0.5Mn) with fine (FG=17 [micro]m) and coarse (CG=114 [micro]m) grain sizes. Stress-strain curves and AE ([V.sub.rms]) are shown in Fig. 14 for both grain sizes in compression (a,b) and tension (c,d). Beside the usual high AE activities at yield, AE reflects the nature of the deformation twinning, which is more active in compression, and both exhibit tension-compression asymmetry. Observed work-hardening behavior is significantly different in tension and compression, long attributed to twinning. Neutron diffraction provided the twinning volume and residual stress during interrupted testing while keeping stress/strain as samples were deformed in the diffractometer. They found that most of the twin nucleation occurred at the onset of plasticity at strains <1.5% in both FG and CG samples, whereas plastic deformation at higher strains is governed by twin growth and dislocation glide. Grain size affected the twin developments with strain; in FG, twinning begins cooperatively in many parent grains simultaneously: in CG, twins increased progressively over a larger strain range. AE results show the instability of the microstructure upon stress relaxation at strains <2% and a gradual stabilization of the microstructure at higher strains. This seems to be due to the effect of an increasing number of twins, which act as barriers to the dislocation's movement. This, the authors noted, is only feasible to determine from AE study.
Lou et al.  conducted tension-compression cycling experiments and reported untwining (or reversal of twinning) activities when stress direction is reversed. Chmelik et al. [39,40] showed increased twinning at higher temperatures (Fig. 15) and little effect on AE of having ultra-fine grain sizes.
[FIGURE 14 OMITTED]
[FIGURE 15 OMITTED]
Taniyama et al.  utilized advance AE instrumentation to identify the deformation mode of AE events observed. They used a Grade 1 Ti (commercially pure Ti) plate with a spherical indentation, mounting four small sensors (PAC Pico) on the edges and cooled it to 200 K. At below 230 K (esp. <210 K), twinning started from the area around the indentation. Figure 16 shows the waveforms recorded from one such event. The initial P-wave arrivals have two positive and two negative polarity results. This fact (and Ch5 waveform from a displacement sensor at the epicenter position under the plate) implies the source of this AE event is of Mode-2 shear motion. This agrees with the expected twinning. From this figure, the rise time of P-waves is -us and grain size is several hundred [micro]m , thus implying the twin velocity of several hundred m/s. Further examination of twin dynamics should be worthwhile in Ti and other hcp alloys at various temperatures.
[FIGURE 16 OMITTED]
Yoshida et al. [44, 45] examined AE behavior of gamma-prime or [Ni.sub.3]Al in single crystals or in directionally solidified (DS) condition, some with B additions. With B additions, AE at yield increased, but AE in later part of work-hardening stage decreased. In DS samples, AE from the work-hardening stage was much more active due to the presence of grain boundaries. The authors suggested frictional drag from B and dissociation of APB obstacles for the two types of AE, but no concrete evidence is available. Since this material (also superalloys that contain high [gamma]' content) is used at elevated temperature, AE at their peak strength condition at >1000 K should be most interesting. No such study has been done to date.
Fukaura et al. [46a,b,c] used AE to evaluate the stress, at which constituent carbide particles fracture in ultra-high strength steels (JIS SDK11 and its modification, M-SKD11) in connection to the fatigue strength study. AE showed that carbide fracture stress increased from 1.1 to 1.8 GPa with the modification of the composition (reducing C and Cr and increasing Mo and V).
Reed and Walter  correlated AE serrated flow (due to twinning, but labeled as Portevin-Le Chatelier or PLC effect) in a Mg alloy. Sophisticated models have been published to account for the physics of PLC [47, 48]. AE is used to justify some of the assumptions. It remains to be seen if the models are valid in explaining various observations. Another theory  provides the lattice dynamics analysis of plastic deformation. Ravi-Chandler  reviewed dynamic fracture models and Minozzi et al.  used a lattice model for dynamic fracture in a disordered medium to correlate AE to internal damage evolution. They find that the cumulative AE amplitudes are directly related by a power law to the cumulative damage, namely, the distribution of the AE burst energies follows a power law and its exponent being related to the exponent describing failure avalanches in quasi-static models.
Waveform simulation procedure [10, 11] was applied to fatigue tests of high strength (SUJ2: 1C-1.5Cr) steel. Using an experimental transfer function with source rise time and volume that best match observed waveforms, Yonezu et al.  obtained AE source parameters (crack volume and speed). Figure 17 shows that AE begins at the start of da/dN-Stage II and the crack is transgranular. AE rises sharply above 8 MPa x [square root of m] (near Stage III) in this steel and the crack becomes intergranular. Microscopic source analysis indicated that fatigue crack progresses along a few grain boundaries accompanied by strong AE signals. As shown in Fig. 18, crack volumes are typically several times that of a single grain IG crack.
[FIGURE 17 OMITTED]
[FIGURE 18 OMITTED]
Another work  examined torsional fatigue of stainless steel cylinders. AE behavior was correlated to microscopic fracture surface observations. Fatigue of Ti alloys was examined by classical AE method . Soboyejo et al.  introduced mechanistic models for Ti fatigue life prediction using AE parameters. They claim that AE counts are related directly to the crack driving force, [DELTA]K, and the measured AE parameters are also related directly to fatigue fracture mechanisms. No evidence is given in support, however, and readers should use caution. A thesis  used NIST sensors to get Rayleigh waves from fracture toughness samples undergoing crack growth. Only wavelet spectrograms are shown and meaningful conclusions cannot be drawn. Another guided wave AE work was given at EWGAE27. Using a 2D FE model, Lee et al.  capably showed that emissions from fatigue crack growth in a plate are Lamb waves and that for crack growth of up to 86% of the plate thickness, the [A.sub.0]-mode signal had larger amplitude than the [S.sub.0]-mode signal. Above that, the [A.sub.0]-mode amplitude drops steeply and the [S.sub.0] becomes larger. See Fig. 19 for this behavior.
[FIGURE 19 OMITTED]
Aggelis et al.  used AE parameters commonly used in concrete studies for A1 fatigue evaluation. Changes in RA values, for instance, increased sharply prior to final fracture, signaling failure mode shift. Shiwa et al.  examined ultrasonic fatigue using AE and nonlinearlity parameters. On application side,  reported fatigue crack detection work on full-size steel girders for a bridge.
AE from martensitic transformation has been studied since the 1930's and has attracted continued attention in connection with shape memory effects. Van Bohemaen  recently established that bainitic transformation in steel is also displacive type, like martensitic one. For many years, it has been argued that bainite also is a displacive transformation because of surface reliefs observed, but definitive proof has eluded researchers. See [62, 63]. Here, AE from martensitic and bainitic transformation was characterized with C45 (0.45C medium carbon steel) and 16MnSi (0.16C-1.7Mn-1.6Si) steels using a spot welding machine and wideband AE sensing (PAC WD; 4 MHz sampling). AE parameters obtained include [V.sub.rms], mean frequency, defined by (4) as
[bar.f] = [integral] G(f) f df / [integral] G(f) df) (4)
calculated from waveforms recorded (G = PSD), and peak amplitude. Figure 20 shows results from C45 steel: a) Averaged [V.sub.rms] of AE signals generated during bainite and martensite formation in steel after spot welding (indicated by W). First peak (B) is bainite formation, and second peak (M) is martensite formation. b) The mean frequency is higher during bainite formation. c) Peak signal amplitude is shown. The observed difference in the mean frequency has been attributed to the characteristics of the (jerky) interface movement of the two transformations: the mean life time of events is shorter for bainite, and the mean free path is less for bainite if the mean velocity is similar in both.
[FIGURE 20 OMITTED]
Yoshida and coworkers have examined AE from Cu-Al-Ni single crystals (one of shape memory alloys or SMA) [64a,b], showing systematic variations between two types (super-elastic and thermoelastic) of martensitic phase changes under stressing. Event rise times were ~1 [micro]s in most cases. Inamura et al.  reported pre-martensitic transformation phenomena of [alpha]'-martensitic transformation in 304 stainless steel just above the martensite start temperature. This corresponds to the formation of fine [alpha]'-martensite particles. A model was presented for martensitic transformation due to thermal cycling, monitored by AE in Cu-Zn-Al . This model exhibits the repetitive bursts of energy under successive thermal cycles in a small temperature interval as observed in AE experiments. It is interesting to note the existence of thermal emissions, paralleling AE .
AE from ferroelectrics is an active field of study. Dul'kin et al. [68a] examined using AE multiferoic ceramic solid solution of relaxor ferroelectric PFW with ferroelectric PbTi[O.sub.3] (PT) together with dielectric measurement. AE response appeared at Burns temperature, at ferroelectric phase transition temperatures, and at Neel antiferromagnetic phase transitions. Based on detected AE responses, they modified existing phase diagram for PFW-PT system. The same group [68b] studied a PZT-based ferroelectric and found that the AE peak intensity is generally proportional to the temperature derivative of the dielectric permittivity, in agreement with a model proposed for a thermally cycled small elastic dipole. Prabakar and Rao  suggested a sudden appearance of burst AE under an applied dc field during temperature decrease cycle indicates the structural phase transition from cubic to tetragonal in soft PZT ceramics (PZT-5A, 5J and 5H). Phase transition temperature is dependent on applied field, which produced many AE signals in the ferroelectric phase. These resulted in two clusters in pattern recognition analysis indicating two different source mechanisms, piezoelectric strain (at <2 kV/cm) and strain due to domain alignment (2-10 kV/cm). Perez-Reche et al.  conducted AE study of a giant magnetocaloric material, [Gd.sub.5][Si.sub.2][Ge.sub.2], during the paramagnetic-monoclinic [left and right arrow] ferromagnetic- orthorhombic magneto-structural phase transition. The detected AE signals showed that avalanches between metastable states control the transition kinetics.
Coating, Film, Friction and Wear
AE study of plasma-sprayed thermal barrier coating (TBC) started in 1979 with indentation and thermal resistance evaluation [71a,b], but soon extended to that of the thermal cycling performance for TBC in gas turbine applications . AE has recently been used in finding the nature of thermal cycling damages; [73,74] used AE with 4-point bending of thermal fatigued samples, trying to evaluate effects of thermally grown oxide (TGO), a critical concern without suitable NDE methods available. Park et al.  reported that, with longer thermal fatigue, TGO layer grew and AE was generated earlier. However, with more cracks and micro-porosity, AE events, coefficients, energy and amplitudes decreased. The same authors also showed changes in laser-induced AE signals as a function of TGO thickness. Such changes may be from the structural changes in the TGO, but can be a useful NDE method if confirmed. On this subject, a few doctoral dissertations [75a-c] appeared, but their AE approach is conventional.
Ma and Takemoto  quantitatively evaluated thermal shock behavior of plasma-sprayed TBC. Bond coat of this TBC was NiCoCrAlTaY (AMDRY997) and top coat, Y-stabilized zirconia, PSZ (Metco204NS). Two of the samples were also PSZ only, thin and thick. Laser provided heating source and waveform simulation analysis of AE signals (discussed earlier) was used to determine the fracture dynamics of TBC. They used an off-epicenter displacement-sensitive sensor to detect AE signals. These are compared with calculated waveforms corresponding to five types of cracking sources, schematically shown in Fig. 21 together with expected waveforms. In Fig. 22, simulated and detected waveforms of type A, C and E cracks are compared side-by-side. For the simulation, crack volume (1~7 x [10.sup.-15] [m.sup.3]) and rise time (1~3 [micro]s) were adjusted to best match the observed signals. They found that vertical Mode-I fracture to be the major fracture mode under low heat-flux laser power conditions. Lateral or interfacial fracture (Mode-I or II) was prevalent under high heat-flux conditions. Delamination (or decohesion) was responsible for the eventual coating failure in thin (or thick) TBC. A bond coat is beneficial for improving the thermal shock resistance, as expected.
AE-instrumented scratch tests have been used for adhesion testing of coatings and films for some time. Piotrkowski et al. [77, 78] examined this established technique using wavelet transform (WT) analysis. WT of AE signals obtained from scratch tests on hot dip galvanized steel samples with different corrosion levels was determined. Depending on damage mechanisms, wavelet power was distributed in different frequency bands, which were automatically obtained by searching for the relative minima of the wavelet entropy of signals and appropriate clustering methods. Correlation between the different mechanisms was corroborated by bispectrum analysis (BA). Coating damage was evaluated via the evolution of the wavelet power in a specific frequency band, corresponding to the fracture of the intermetallic phase of the galvanized coating. Damage increased with the level of corrosion, but adherence was not dramatically affected. The application of two signal-processing techniques, WT and BA, contributed to the consistency of the results.
[FIGURE 21 OMITTED]
AE is used in examining other films and coatings. Ikeda et al.  evaluated polycrystalline diamond films with AE to detect ring cracks in indentation test. This led to finding the static strength of 3.6~4.0 GPa. The strength under cyclic loading was found as 1.4~3.2 GPa, distinctly smaller than the static value. AE method with laser excitation can measure the adhesive strength of coating, used for CVD diamond . The same method was used for PSZ and 316L stainless steel coating by plasma spray .
Sause [5, 82] applied his FEM modeling to the analysis of coating failure on composite plates. AE from coating failure consists of crack initiation, crack growth and delamination and is distinguishable from signals arising from failure of CFRP substrate. Experiment was compared to FEM models of AE signals and pattern recognition analysis was applied to separate different types of AE signals.
[FIGURE 22 OMITTED]
A final note of interest on film/coating is anodized Al under strain. R.G. Liptai's 1963 dissertation [83a] dealt with anodized surfaces as barrier to dislocation glide. A recent work [83b] again showed even thin films (2-20 [micro]m thick) contributes to AE in the elastic range.
Friction and wear have benefitted from the use of AE for a long time. A book on "AE in Friction"  is out recently. Early work on AE was on monitoring magnetic hard disk  and this still continues with  studying the contact behavior of sub-ambient pressure tri-pad sliders during start/stop and constant speed operations and finding contact forces to be a few mN.  studied lubricated sliding contacts with AE detecting the presence of wear-reducing additives and the predominant wear process.  evaluated tribological behavior with AE of alumina, silicon nitride and 52100 steel under dry sliding with a ball-on-flat-disc. Correlations exist between the wear rate and the integrated AE [V.sub.rms] signal, but strong influence of the sliding speed on the wear rate was found.  conducted wear-life tests of different diamond-like carbon (DLC) coatings on a steel substrate (100Cr6) under slip-rolling friction. AE turned out to be an excellent tool for an in situ estimation of the end of the wear life of the DLC coatings. Under lubricated conditions, if a high AE activity was detected, the respective coating failed later in the tests. When no high AE activity was recorded, the respective coatings did not reach the failure criterion and were labeled 'resistant' to wear.  provided a review of the application of AE techniques for monitoring forming and grinding processes. AE generated during the forming processes is useful for detecting die wear and cracking, friction properties, state of lubrication, galling, stick-slip, etc.
Corrosion and Stress Corrosion
Use of AE in corrosion and stress corrosion cracking (SCC) studies started early . (Note "cracking" in SCC is most often not from mechanical stress, but stress-assisted rapid chemical dissolution, but we will use the conventional term.) Okada et al.  showed SCC of 304 austenitic and 25Cr-5Ni-3Mo ferritic stainless steels in boiling Mg[Cl.sub.2] produced no AE because active-path corrosion (APC) is the cause. From our present knowledge, this is transgranular (TG) type SCC with the use of 143[degrees]C-boiling Mg[Cl.sub.2]. In contrast, SCC of high strength martensitic stainless steel (13Cr-Ni-Mo) in 3%NaCl and in boiling Mg[Cl.sub.2] is from hydrogen embrittlement and accompanied by AE. Hydrogen evolution in corrosive reaction is detectable with AE [92,93a], has steep amplitude dependence  and shows resonating waveforms [93b]. Suezawa, et al.  also measured corrosion potential in AE-SCC study of 304 stainless steel. (It is interesting to note they found a shift to higher frequency in deformation-induced AE of 304, as discussed earlier [30, 32].)
Early corrosion research was reviewed by Yuyama et al.  Most of the early SCC work failed to control corrosion potential, however, making it difficult to assess electrochemical conditions. Takemoto [95, 96] reviewed more recent AE work on SCC centering on stainless steels. General corrosion contributes to increased AE rms voltage levels and can be distinguished from localized corrosion (pitting, crevice, exfoliation, SCC, etc) . Corrosion AE studies are complemented well by the use of corrosion potential fluctuation (CPF)  or electrochemical noise (EN) . CPF and EN are identical parameters. Pitting in Al alloys generates AE by the release of hydrogen micro-bubbles [99a,b]. Initiation of pits is quiet, but growth stage shows two types of AE signals [99a]. Jones and Friesel [99c] found no bubbles in pitting of 304 stainless steel and suggested that rupture of an oxide or salt cap during pit growth is a possible source of the AE with only 1-3% of the pits, or pits larger than 10 [micro]m being active. Yuyama  attributed 304 stainless steel crevice corrosion to hydrogen bubble evolution. Crevice corrosion in spot-welded 304 stainless steel in 3.5% NaCl at noble potential emitted two types of AE, one from rust cracking, giving broadband signals and the other from hydrogen bubbles with narrow band AE . AE is shown to detect the early development of exfoliation corrosion in Al alloys . Differences in T6 and T7 conditions are evident and Al-2024 is more resistant than Al-7449. Again, hydrogen bubble release is the active AE source. A few other AE mechanisms were suggested.
Taking advantage of the source function analysis method [10, 11], Hayashi and Takemoto  showed the high-speed nature of molten chloride attack on stressed 304 stainless steel at 600[degrees]C to be as high as 67 m/s and discovered fall-off of grains and Mode I and II AE sources. Similarly, Fujimoto et al.  examined APC type chloride-SCC of 304 austenitic stainless steel and showed no AE from TG-SCC produced in 35% Mg[Cl.sub.2] at 98[degrees]C, but detected AE from IG-SCC, also APC type, in sensitized 304 (caused by 38% Mg[Cl.sub.2] at 95[degrees]C, increasing-stress test) with fast rise time of 0.17 to 1.0 ps. They used Lamb waves for the analysis. This APC type IG-SCC is the active one, operative in high-temperature, high-pressure water. This is also caused by dilute fluoride and tetrathionic acid (H2S4O6) solutions at room temperature . Yonezu et al.  demonstrated the cause of this AE to come from grain fall-off by local corrosion of Cr-depleted grain boundaries. Here, three types of AE were identified with the source function analysis method and CPF: AE from falling-off of grains, fracture of corrosion product and hydrogen gas evolution. Figure 23 shows examples of surface and cross section of 304 stainless steel that underwent IG-SCC in MgCb . Most recently, Kovac et al.  reported on AE from IG-SCC, attributing ductile fracture ligament as the cause of this AE. In austenitic steels, AE from ductile fracture is known to be weak, so their suggested mechanism is not persuasive.
[FIGURE 23 OMITTED]
SCC behavior of a dual phase stainless steel (DP3) in MgCb showed APC-type SCC in 42% Mg[Cl.sub.2] solution without primary AE [106a]. When this APC-SCC occurs, solution in narrow grooves or cracks is changed to strong acidic solution due to hydrolysis reaction of chromium chloride with water, and the hydrogen gas evolution takes place in the crack. Though they did not confirm hydrogen gas evolution visually, this is consistent with the expected small size of hydrogen bubbles. For this material, hydrogen assisted cracking in 35% MgCh solution occurred with AE . SCC due to ammonia in brass, caused in contact with Mattsson's solution (NACE G37-73) was APC-type and IG-SCC, giving AE as above [106b].
Aluminum alloys, especially 7000-series with Zn, suffer from SCC in moist air and from salt water. This SCC is due to hydrogen effects [107, 108] and is controlled by the Cu content of grain boundary precipitates (GBP). SCC rate in NaCl solutions is determined by the rate of anodic dissolution/de-alloying of GBP, which controls the rate of cathodic hydrogen charging of grain-boundary regions just ahead of crack tips. For SCC in moist air, anodic dissolution is unlikely as matching crack-arrest markings accompanying AE are found. Hydrogen generation occurs during crack arrest periods, hydrogen diffusing in along boundaries ahead of crack-tips for 0.1-0.3 [micro]m, causing rapid crack-growth due to hydrogen-induced weakening of atomic bonds with film rupture and/or the hydrogen concentration at a critical level, and crack-arrest and blunting result [108-110].
Studies of hydrogen embrittlement (HE) of high-strength steels were of critical importance to space technology in early days of AE . Recent work applied the source function analysis and deduced details of source mechanisms and their size (source volume) as a function of sensitization in 329 dual phase stainless steel . This steel possesses hardened ferrite phase and suffered HE ahead of main crack, while austenite phase exhibited ductile fracture, while the steels sensitized at 750[degrees] and 800[degrees]C showed HE in the ferrite and sigma phases. Hydrogen charging effects on sensitized 304 [113a,b] are consistent with grain boundary separation due to hydrogen gas accumulation being the cause of observed IG failure . Hydrogen charging in Ti behaves differently . Micro-fractures of titanium hydride produced by hydrogen charging into tensile loaded titanium caused AE of the Mode-I fracture. Crack opening time was 0.1-0.4 ps, and crack volume was 0.3-1.9 x [10.sup.-15] [m.sup.3]. Finally, AE is used to evaluate the delayed fracture susceptibility of high-strength steels . Elastically loaded strip (Almen strip: SAE J4542) and slow strain-rate tests (with immersion in aggressive solutions) are used successfully to detect the threshold strain to induce the subsurface micro-cracks in the steels. This test method is applicable to evaluate shot-peening effects and trapped and diffusible hydrogen.
This paper reviewed recent progress in methods of signal analysis used in acoustic emission as applied to materials research. The achievements in understanding of AE from materials during deformation and fracture, phase transformation, coating, film, friction and wear, and corrosion and stress corrosion were examined systematically (but not exhaustively). There are many important areas of AE research that this review could not cover, but it is evident that AE has advanced substantially and contributes to various areas of materials research with increased depth. This makes it mandatory for every AE worker to thoroughly survey available literature before embarking on a new study. All current and future researchers should take advantage of accumulated body of knowledge, which has become much more accessible via the Internet. They can also benefit from the availability of new analytical tools and vastly advanced instrumentation. It is hoped that they successfully resolve many of outstanding issues in the near future.
 T. Kishi, M. Ohtsu and S. Yuyama, eds., Acoustic Emission--Beyond the Millenium, Elsevier, 213 p., 2000.
 C.U. Grosse and M. Ohtsu, eds., Acoustic Emission Testing, Basics for Research Applications in Civil Engineering, Springer, 414 p., 2008.
 K. Ono, New goals for acoustic emission in materials research, Acoustic Emission--Beyond the Millenium, 2000, Elsevier, pp. 57-76.
 K. Ono, Current understanding of mechanisms of acoustic emission, J. Strain Analysis for Engineering Design, 40 (1), 2005, 1-15.
 M. Sause, Identification of failure mechanisms in hybrid materials utilizing pattern recognition techniques applied to acoustic emission signals, Augsburg Univ. doctoral dissertation, Mensch und Buch Verlag, 2010, 305 p.  Convolution: http://en.wikipedia.org/wiki/Convolution.
 H. N. G. Wadley and C. B. Scruby, Proc. Intl Conf. on Acoustic Emission, Dunhart, 1981, pp. 125-153.
 M. Enoki and T. Kishi, Acoustic Emission--Beyond the Millenium, 2000, Elsevier, pp. 1-18.
 a) M. Ohtsu and K. Ono, J. Acoustic Emission, 3, 1984, 27-40. b) J. Acoustic Emission, 5, 1986, 124-133.
 a) H. Suzuki, M. Takemoto and K. Ono, J. Acoustic Emission, 11, 1993, 117; b) J. Acoustic Emission, 14, 1996, 35.
 M. Takemoto, H. Nishino and K. Ono, Acoustic Emission - Beyond the Millenium, 2000, Elsevier, pp. 35-56.
 a) J. Gary and M. Hamstad, J. Acoustic Emission, 12 (3-4), 1994, 157-170;
b) M.A. Hamstad, J. Gary, A. O'Gallagher, J. Acoustic Emission, 14 (2), 1996, 103-114.
 a) D. Guo, A. K. Mal and M.A. Hamstad, J. Acoustic Emission, 16, 1998, S222-S232. b) M.A. Hamstad, J. Gary, A. O'Gallagher, J. Acoustic Emission, 16, 1998, S251-S259.
 W. H. Prosser, M.A. Hamstad, J. Gary and A. O'Gallagher, J. Acoustic Emission, 17 (1-2), 1999, 37-47.
 M. A. Hamstad, A. O'Gallagher and J. Gary, J. Acoustic Emission, 19, 2001, 258-274.
 a) K. Ono, H. Cho, T. Matsuo, J. Acoustic Emission, 26, 2008, 72-90. b) 28, 2010, 256-277.
 M. R. Gorman, W. H. Prosser, J. Acoustic Emission, 9, 1991, 283-288.
 K. Ono and Q. Huang, "Pattern Recognition Analysis of Acoustic Emission Signals", Progress in Acoustic Emission VII, eds. T. Kishi et al., JSNDI, Tokyo, 1994, pp. 69-78.
 J. Bohse, J. Acoustic Emission, 22, 2004, 208-223.
 PZFlex: http://www.pzflex.com/about.aspx.
 A. Green, C.S. Lockman, R.K. Steele, Modern Plastics, 41, 1964, 137-139, 178, 180.
 H. Suzuki, T. Kinjo, Y. Hayashi, M. Takemoto and K. Ono with Appendix by Y. Hayashi, J. Acoustic Emission, 14, 1996, 69-84.
 T. Hayashi and S. Wakayama, J. Acoustic Emission, 26, 2008, 160-171.
 S.M. Pincus, Proc. National Academy of Sciences of the USA, 88 (6), 1991, 2297-2301.
 R. Yan and R.X. Gao, Mechanical Systems and Signal Processing, 21, 2007, 824-839.
 Li Lin and Fulei Chu, Nondestructive Testing and Evaluation, 26 (2), 2011, 119-128.
 G. Manthei and J. Eisenblatter, ref. 2, pages 284-286.
 K. Mogi, Bull. Earthquake Res. Inst., 40, 1962, 125-173.
 K. Ono, H. Cho and M. Takuma, J. Acoustic Emission, 23, 2005, 206-214.
 A. Vinogradov, V. Patlan, S. Hashimoto, Phil. Mag. A, 81, 2001, 1427.
 A. Vinogradov, D. L. Merson, V. Patlan, S. Hashimoto, Materials Science and Engineering, A341, 2003, 57-73.
 A. Razarev and A. Vinogradov, J. Acoustic Emission, 27, 2009, 144-156.
 D. Rouby, P. Fleishmann and C. Duvergier, Phil. Mag., 37, 1983, 689-705.
 a) M. Koslowski, R. LeSar, R. Thomson, Phys. Rev. Letters, 93, 2004, 265503; b) M. Koslowski, Phil. Mag., 87, 2007, 1175-1184.
 T. Richeton, J. Weiss, F. Louchet, Acta Materialia 53, 2005, 4463-4471.
 D. Bowles and W. St. Lawrence, Proc. Western Snow Conf. 1977, April 18-21, Albuquerque, NM, pp. 88-94; N. Sinha, Proc. 1982 SESA/JSME Conf., Pt II, 1982, pp. 767-772.
 O. Muransky, M.R. Barnett, D.G. Carr, S.C. Vogel, E.C. Oliver, Acta Materialia, 58, 2010, 1503-1517.
 X.Y. Lou, M. Li, R.K. Boger, S.R. Agnew, R.H. Wagoner, Intl J. Plasticity 23, 2007, 44-86.
 F. Chmelik, F. Moll, J. Kiehn, K. Mathis, P. Lukac, K-U. Kainer and T. G. Langdon, J. Acoustic Emission, 20, 2002, 108-128.
 M. Janecek and F. Chmelik, Magnesium Alloys--Design, Processing and Properties, ed. F. Czerwinski, InTech, 2011, pp. 43-68.
 J. M. Reed, M. E. Walter, Materials Science and Engineering, A359, 2003, 1-10.
 Y. Taniyama, T. Matsuo, H. Cho, M. Takemoto and G. Nakayama, Proc. IAES18, JSNDI, pp. 295-302.
 P.K. Chaurasia, C.K. Mukhopadhyay, S. Murugan, P. Muralidharan, K. Chandran, V. Ganesan, T. Jayakumar, P.V. Kumar, J. Nuclear Materials, 322, 2003, 217-227.
 K. Yoshida, T. Saito, A. Zhu, H. Takagi and K. Sakamaki, Intermetallics, 6, 1998, 1-5.
 K. Yoshida, Y. Masui, T. Nagamachi and H. Nishino, J. Acoustic Emission, 23, 2005, 189-195.
 a) K. Fukaura, Y. Yokoyama, D. Yokoi, N. Tsujii, and K. Ono, Met. Mater. Trans., 35A,
2004, 1289-1300; b) K. Fukaura and K. Ono, J. Acoustic Emission, 19, 2001, 91-99; H. Cho, K. Fukaura and K. Ono, J. Acoustic Emission, 21, 2003, 112-119.
 M.A. Lebyodkin, T.A. Lebedkina, F. Chmelik, T.T. Lamark, Y. Estrin, C. Fressengeas, and J. Weiss, Phys. Rev. B 79, 2009, 174114-9.
 J. Kumar and G. Ananthakrishna, Phys. Rev. Lett. 106, 2011, 106001-4.
 A. Trochidis and B. Polyzos, J. Appl. Phys., 78, 1995, 170-173.
 K. Ravi-Chandler, Intl J. Fracture, 90, 1998, 83-102.
 M. Minozzi, G. Caldarelli, L. Pietronero and S. Zapperi, Eur. Phys. J., B36, 2003, 203-207.
 A. Yonezu, T. Ogawa and M. Takemoto, Proc. IAES18, JSNDI, 2006, pp. 463-470.
 Z. Shi, J. Jarzynski, S. Bair, S. Hurlebaus and L.J. Jacobs, J. Mech. Engineering Sciences, 214, Part C, 2000, 1141-1149.
 D.H. Korn, P. Ducheyne and J. Awerbuch, J. Mat. Sci., 27, 1992, 3133-3142.
 W.O. Soboyejo, B. Rabeeh, Y. Li, A.B.O. Soboyejo and S. Rokhlin, Fatigue & Fracture of Engineering Materials & Structures, 21, 1998, 557-568.
 M.R. Horne, PhD dissertation, Virginia Polytechnic Institute and State University, 2003.
 C.K. Lee, J.J. Scholey, P.D. Wilcox, M.R. Wisnom, M.I. Friswell and B.W. Drinkwater, Advanced Materials Research, 13-14, 2006, 23-28.
 D. G. Aggelis, E. Z. Kordatos, T. E. Matikas, Proc. IAES20, 2010, pp. 73-78.
 M. Shiwa, Y. Furuya, H. Yamawaki, K. Ito and M. Enoki, Proc. IAES20, 2010, pp. 319-324.
 J. McKeefry and C. Shield, Acoustic Emission Monitoring Of Fatigue Cracks In Steel Bridge Girders, MN-DOT report, MN/RC-1999-36, 1999, 138 p.
 S.M.C. Van Bohemen, J. Sietsma, M.J.M. Hermans, I.M. Richardson, Phil. Mag., 85 (16), 2005, 1791-1804.
 R.W. Cahn and P. Haasen, Physical Metallurgy, 4th ed., N. Holland, 1996, p. 1469.
 H.K.D.H. Bhadeshia, Bainite in Steels, Inst. Materials, 2001, 460 p.
 a) T. Yasuda, S. Kondo, H. Nishino and K. Yoshida, J. Acoustic Emission, 26, 2006, 270-278; b) T. Yasuda, B. Pang, H. Nishino and K. Yoshida, Materials Transactions, 52 (3) 2011, 397-405.
 T. Inamura, S. Nagano, M. Shimojo, K. Takashima and Y. Higo, J. Acoustic Emission, 19, 2001, 85-90.
 S. Sreekala1 and G. Ananthakrishna, Phys. Rev. Lett., 90, 2003, 135501-1.
 L. Carrillo and J. Ortin, Phys. Rev. B56, 1997, 11508-11517.
 a) E. Dul'kin, E. Mojaev, M. Roth, S. Kamba, and P. M. Vilarinho, J. Appl. Phys., 103, 2008, 083542; b) E. Dul'kin, E. Mojaev, M. Roth, O. Khamman and X. Tan, Appl. Phys. Lett., 95, 2009, 252903.
 K. Prabakar and S. P. Mallikarjun Rao, Ferroelectrics Letters Section, 32 (5-6), 2005, 99-110.
 F-J. Perez-Reche, F. Casanova, E. Vives, L. Manosa, A. Planes, J. Marcos, X. Batlle and A. Labarta, Phys. Rev. B73, 2006, 014110.
 a) S. Safai, H. Herman and K. Ono, Am. Ceram. Soc. Bull., 58 (6), 1979, 624; b) Proc. 9th. Int. Thermal Spraying Conf., The Hague, 19-23 May, Netherlands Inst. Lasteniek, 1980, pp. 129-132.
 C.C. Berndt, J. Engineering for Gas Turbines and Power, 107, 1985, 142-146.
 J. Voyer, F. Gitzhofer, M.I. Boulos, J. Dionne, Thermal Spray 1996: Practical Solutions for Engineering Problems, ASM International, 1996, pp. 303-310.
 J.H. Park, J.S. Kim, K.H. Lee, J. Materials Processing Technology 187-188, 2007, 537-541.
 a) S. Chang, Purdue University, 2005; b) N.H. Faisal, Heriot-Watt University, 2009; c) S. Nekkanty, Ohio State University, 2009.
 X.Q. Ma and M. Takemoto, Materials Science and Engineering A, 308, 2001, 101-110.
 R. Piotrkowski, A. Gallego, E. Castro, M.T. Garcia-Hernandez, J.E. Ruzzante, NDT&E International, 38, 2005, 260-267.
 R. Piotrkowski, E. Castro, A. Gallego, Mechanical Systems and Signal Processing, 23, 2009, 432-445.
 R. Ikeda, M. Hayashi, A. Yonezu, H. Cho, T. Ogawa and M. Takemoto, Proc. IAES18, JSNDI, 2006, pp. 455-462.
 R. Ikeda, H. Cho, A. Sawabe and M. Takemoto, Diamond and Related Materials, 14, 2005, 631-636.
 T. Uchiyama, R. Ikeda, H. Cho and M. Takemoto, Proc. IAES18, JSNDI, 2006, pp. 303-462.
 M.G.R. Sause, D. Schultheifi and S. Horn, J. Acoustic Emission, 26, 2008, 1-13.
 a) R.G. Liptai. An Investigation of the Acoustic Emission Phenomenon. PhD dissertaion, Michigan State University, East Lansing, Michigan. 1963. b) J.M. Goetz, Honors thesis, Ohio State University, 2005.
 V. Baranov, E. Kudryavtsev, G. Sarychev, V. Schavelin, Acoustic Emission in Friction, Elsevier, 2007, 232 p.
 M-K. Tse, J. Leifer and A.F. Lewis, J. Acoustic Emission, 4, 1985, S127-S131.
 A. G. Khurshudov and F. E. Talke, J. Tribol. 120, 1998, 54-59.
 R.J. Boness and S.L. McBride, Wear. 149, 1991, 41-53.
 H.S. Benabdallah, J. Materials Science, 34, 1999, 4995-5004.
 M. Lohr, D. Spaltmann, S. Binkowski, E. Santner, M. Woydt, Wear 260, 2006, 469-478.
 T. Jayakumar, C.K. Mukhopadhyay, S. Venugopal, S.L. Mannan, Baldev Raj, J. Materials Processing Technology 159, 2005, 48-61.
 H. Okada, K-I. Yukawa and H. Tamura, Corrosion. 30, 1974, 253-255.
 Y. Suezawa, H. Takemoto, and T. Tanaka, Boshoku Gijutsu (J. Japan Soc. Corrosion Engr.), 24, 1975, 239-248.
 a) O. Takano and K. Ono, Proc. IAES2, 1974, p. VI-31; b) T. Morikawa and K. Ono, Plating & Surface Finishing, 89 (12), 2002, 50-54.
 S. Yuyama, T. Kishi and Y. Hisamatsu, J. Acoustic Emission, 2, 71-93.
 M. Takemoto, H. Cho, K. Ono, Proc. 26th European Conference of AE, DGfZP, Berlin, 2004, pp. 617-629.
 M. Takemoto, Proc. ICAE6, AEWG, 2007, pp. 415-424.
 M. Winkelmans, U. Muller and H.-J. Bassler, Materials and Corrosion, 58 (12), 2007, 907-915.
 R.C. Newman, K. Sieradzki, Scripta Metall. Mater. 17, 1983, 621-624.
 a) H. Idrissi, J. Derenne, H. Mazille, J. Acoustic Emission, 18, 2000, 299-306; b) S. Krakowiak and K. Darowicki, J. Solid State Electrochem, 13, 2009, 1653-1657; c) R.H. Jones and M.A. Friesel, Corrosion NACE, 48 (9), 1992, 751-758.
 H. Cho and M. Takemoto, Proc. IAES18, JSNDI, 2006, pp. 287-294.
 F. Bellenger, H. Mazille, H. Idrissi, NDT&E International, 35 (6), 2002, 385-92.
 K. Hayashi and M. Takemoto, Proc. IAES13, JSNDI, 1996, pp. 76-81.
 S. Fujimoto, M. Takemoto and K. Ono, J. Acoustic Emission, 19, 2001, 63-74.
 A. Yonezu, H. Cho and M. Takemoto, Advanced Material Research, 13, 2006, 243-250.
 J. Kovac, C. Alaux, T.J. Marrow, E. Govekar, A. Legat, Corrosion Science, 52, 2010, 2015-2025.
 a) K. Kagayama, T. Ogawa, A. Yonezu, H. Cho and M. Takemoto, J. Acoustic Emission, 24, 2006, 127-138; b) F. Uchida, H. Nishino, M. Takemoto and K. Ono, J. Acoustic Emission, 19, 2001, 75-84.
 N.J.H. Holroyd, in Environment-Induced Cracking of Metals, R.P. Gangloff and M.B. Ives, eds., NACE, Houston, 1989, pp. 311-345.
 S.P. Lynch, S.P. Knight, N. Birbilis, B.C. Muddle, in Effects of hydrogen on materials, Proc. 2008 Intl Hydrogen Conference, B. Somerday et al. eds., ASM International, 2008, pp. 243-250.
 G.M. Scamans, Metall. Trans., 11A, 1980, 846-850.
 P. Martin, J.L. Dickson, J.-P. Ballon, Materials Science and Engineering, 69 (1), 1985, L9-L13.
 H.L. Dunegan and A.S. Tetelman, Engineering Fracture Mechanics, 2, 1971, 387-390.
 K. Kurose, H. Cho, T. Ogawa and M. Takemoto, Proc. IAES19, JSNDI, 2008, pp. 121-127.
 a) S.H. Carpenter and D.R. Smith, Jr., Metall. Trans., 21A. 1990, 1933-1939; b) S.H. Carpenter, K. Ono and D. Armentrout, J. Acoustic Emission, 24, 2006, 119-126.
 M. Takemoto, M. Nakamura, S. Masano and S. Ueno, J. Acoustic Emission, 27, 2009, 241-253.
Department of Materials Science and Engineering, University of California, Los Angeles, CA 90095-1595
This paper is an edited version of a paper published in the Proceedings of World Conference on Acoustic Emission, 2011, Beijing, China, held in Beijing, August 24-26, 2011. It was presented via video recording as a keynote lecture.
|Printer friendly Cite/link Email Feedback|
|Publication:||Journal of Acoustic Emission|
|Date:||Jan 1, 2011|
|Previous Article:||Research on gas pipeline leak location method based on empirical mode decomposition.|
|Next Article:||Pattern recognition study of acoustic emission signals of aircraft fatigue cracking based on waveform analysis and artificial neural networks.|