Magneto-acoustic emission for the characterisation of ferritic stainless steel microstructural state

ARTICLE IN PRESS

Journal of Magnetism and Magnetic Materials 271 (2004) 381–389

*Corresp

+353-0-21

E-mail

0304-8853/

doi:10.1016

Magneto-acoustic emission for the characterisation of ferriticstainless steel microstructural state

D. O’Sullivana,*, M. Cotterella, S. Cassidya, D.A. Tannerb, I. M!esz!arosc

a Department of Mechanical and Manufacturing Engineering, Cork Institute of Technology, Room A243, Bishopstown, Cork, Irelandb Department of Materials Science and Technology, University of Limerick, Limerick, Ireland

c Department of Materials Science and Engineering, Budapest University of Technology and Economics, H-1111 Goldmann ter 3, V2/159,

Budapest, Hungary

Received 17 June 2003; received in revised form 12 August 2003

Abstract

The role of residual stresses in the failure of metallic components and the need to determine such stresses is well

recognised. Magneto-acoustic emission (MAE) is a relatively new non-destructive detection technique and its working

principle is based on Barkhausen discontinuities or noise and magnetostriction when a ferromagnetic material is

subjected to a varying magnetic field. MAE is being used to characterise the stress state of a ferritic stainless steel (AISI

430). Other stress measurement techniques; X-ray diffraction (XRD), magnetic Barkhausen noise (MBN) have also

been used to support the results achieved using MAE. A new measurement parameter has been developed for

microstructural characterisation called MAE absolute energy and has proved to be a useful quantitative method in

MAE waveform measurement.

r 2003 Elsevier B.V. All rights reserved.

PACS: 07.55.–w; 07.64.+z; 46.60.�c; 75.80.+q

Keywords: Magneto-acoustic emission; Ferritic stainless steel; Residual stress; Barkhausen noise; X-ray diffraction

1. Introduction

High stress corrosion resistance and relativelylow production costs are the main reasons for theincreasing applications of AISI 430 ferritic stain-less steels [1]. But problems such as loss of ductilityand toughness when exposed to elevated tempera-tures, for instance during welding, is a mainconcern [2]. The heat of welding leads to grain

onding author. Tel.: +353-0-21-4326507; fax:

-4326627.

address: [email protected] (D. O’Sullivan).

$ - see front matter r 2003 Elsevier B.V. All rights reserve

/j.jmmm.2003.10.004

coarsening in the heat-effected zone and in theweld metal of ferritic stainless steels because theysolidify directly from the liquid to the ferrite phasewithout any intermediate phase transformation[2–3]. The use of a non-destructive evaluationtechnique to characterise the post-welding micro-structural state of a material could be of use indetermining the optimal levels of heat input andwelding speeds.

For characterisation, ferritic stainless steel (AISI430) samples were plastically deformed and heattreated to various degrees and measured usingBarkhausen sensing techniques. In a ferromagnetic

d.

ARTICLE IN PRESS

D. O’Sullivan et al. / Journal of Magnetism and Magnetic Materials 271 (2004) 381–389382

material, Barkhausen noise (BN) is generated bythe discontinuous movement of irreversible do-mains walls. This movement can be induced byapplying a time varying magnetic field across thesample. This noise can be detected in the form ofacoustic noise (electrical energy) or in the form ofvoltage pulses which are induced in a coil placednear the surface of the material. MagneticBarkhausen noise (MBN) results from the rever-sible and irreversible displacement of 180� andnon-180� domain walls, or by abrupt rotation ofdomain magnetisation vectors at higher magneticfields. Acoustic Barkhausen noise, measured bymagneto-acoustic emission (MAE), is only causedby the discontinuous motion of non-180� walls, orthe irreversible rotation of domains through anglesother than 180� [4–6]. No stresses are generated asa result of 180� wall motion or rotation becausethe strain along a particular axis is independent ofthe direction of magnetic moments if they lie alongthe axis [7]. It is the acoustic pulses given out bythe generation of stresses from the movement ofnon-180� walls that are detected and analysed. Nostresses are generated as a result of 180� wallmotion or rotation because the strain along aparticular axis is independent of the direction ofmagnetic moments if they lie along the axis.

Barkhausen noise can also be affected bypinning sites which act as barriers to domainmovement. Pinning sites are attributed to variousphysical parameters such as grain boundaries,voids, inclusions and dislocations.

Amongst the wide range of techniques availablefor NDT evaluation of stresses, MBN measure-ment has become very popular for stress analysis.Macroscopic magnetic properties such as differ-ential permeability are well known to be stressdependent [8]. The stress dependence of differen-tial permeability carries through to micro-scaleeffects to cause MBN also to be stress-dependent[9–10].

But the significance of MAE for the detection ofstresses in ferromagnetic materials has receivedlittle attention. MAE has not yet been used forindustrial purposes, and it is hoped that the resultspresented here will help to further advance thistechnique for its use as a non-destructive evalua-tion of stress.

The main difference between the two Barkhau-sen techniques is the measurement depth. Thedepth of the measurement depends on the fre-quency of the excitation, but typically for MBN isin the range of 0.1–1 mm (surface measurement)and for MAE between 0 and 10 mm (bulkmeasurement) depending on the permeability ofthe material.

2. Experimental method

Flat specimens have been cut from a sheet ofAISI 430 which has been annealed at 815�C. Theas-received chemical composition of the materialincluded 14.19% Cr, 0.9% Mn, 84.91% Fe. Alltensile samples have been prepared in accordanceto Ref. [11]. These samples were deformed tovarious degrees of strain within the plastic range(i.e. from the yield point of the material to theultimate tensile strength point.). Standardised testspecimen dimensions (120 mm� 10 mm� 1.5 mm)were used as outlined in Ref. [11], and strain wasmeasured using an extensometer with a gaugelength of 50 mm which was calibrated according to[11]. Tensile tests were completed on seven samplesto determine yield point and ultimate tensilestrength of the specimens.

Samples plastically strained to 17.5% weresubjected to annealing heat treatment for thepurpose of stress relief. Heat treatments werecompleted for 100–1100�C with a soaking time of30 min. Samples removed from the furnace wereair-cooled.

Seven specimens were elongated at each plasticstrain percentage. Five measurements of eachmeasurement technique were completed on eachspecimen across the surface. The standard devia-tion of the results from each specimen was thencalculated using conventional formulae.

2.1. Barkhausen measurement

A description of the Barkhausen techniquesused for experimentation is given. In this investi-gation, the magnetic field strength used for sampleexcitation was larger than the maximum magneticsaturation level of the material, even at high levels

ARTICLE IN PRESS

Fig. 1. Absolute energy measurement (a) waveform, (b)

absolute waveform and (c) MAE energy calculation.

D. O’Sullivan et al. / Journal of Magnetism and Magnetic Materials 271 (2004) 381–389 383

of deformation. Because of this, the Barkhausennoise generated is mostly a result of movingmagnetic domains that have enough energy tofree themselves from their pinning sites.

2.1.1. Magnetic Barkhausen noise (MBN)

The MBN utilised a sinusoidal (10 Hz) excitingmagnetic field produced by a function generatorand a power amplifier. The applied measuringhead contained a U-shaped magnetising coil and apick-up coil, which is perpendicular to the surfaceof the specimen. The signal of the pick-up coil wasprocessed by a 0.3–38 kHz band pass filter andamplified with a gain of 50. A National Instru-ments data acquisition card (PCI-6023) with asampling rate of 200 kS/s was utilised. LabVIEWwas used for processing the noise signal. Theapplied sampling frequency was 100 kHz and themaximum magnetic field induction of the excita-tion was 20 G. The root mean square (RMS) valueof the noise signal was determined and was used tocharacterise the microstructural state of thematerial. To ensure that a high level of signal-to-noise ratio was achieved, the background MBNwas measured and subtracted from actual samplemeasurements.

2.1.2. Magneto-acoustic emission (MAE)

A U-shaped electromagnet was used to apply atime varying magnetic field (sinusoid at 10 Hz)across the samples to achieve magnetic domainrotation. The acoustic emission sensor wascoupled to the surface of the specimens usingVaseline BP white petroleum jelly. The couplantinsures good surface contact between the twosurfaces on a microscopic level and aids thetransmittal of acoustic waves between the twosurfaces. The MAE sensing head was applied tothe surface of the specimens, where the AE(acoustic emission) sensor had a spring back forceof 15 N and the electromagnet had an applicationforce of 25 N. A total preamplifier gain of 60 dB(� 1000) was used for signal amplification beforeit was measured with a digital oscilloscope. AllMAE waveforms were collected using a TektronixTDS 220 digital storage oscilloscope, which has asampling rate of 1 GS/s. The magnetic fieldstrength used for the excitation was 35 kA/m. This

magnetic field strength was greater than themaximum sample material magnetic saturationlevel. This ensured high signal-to-noise ratios. TheMAE (absolute energy) background noise wasmeasured as 0.001 V s.

2.1.2.1. Absolute energy. A method to determinethe area of the waveform was derived called theMAE absolute energy. Acoustic emission (AE) canbe defined as the transient elastic waves resultingfrom localised internal micro-displacements takingplace in a solid. The AE consists of bursts ofenergy, which can vary in amplitude over severalorders of magnitude. When this occurs within amaterial, the piezoelectric AE sensor is excited bythe stress waves impinging on its face. Thismechanical signal is converted to an electricalsignal by the piezoelectric device. AE energy is themeasured energy of an electrical waveform, whichis the result of a mechanical excitation of apiezoelectric sensor. Thus since the MAE wave-form is an energy signal, it is characterised by thearea under the waveform over a fixed period.

The first step in the absolute energy calculationprocess is the acquisition of the AE signal by thesystem (Fig. 1(a)). The signal waveform is thenrectified (Fig. 1(b)) or made absolute so that allnegative peaks become positive. Each voltage

ARTICLE IN PRESS

Fig. 2. (a) Both Y cartesian co-ordinates are of same polarity and (b) consecutive points of opposite polarity.


received is evaluated and the total absolute energyfor that signal is determined (Fig. 1(c)). This canbe seen as an accurate method of evaluating theenergy of a signal as the energy is directly relatedto the piezoelectric crystal.

The area of the MAE signal was calculated bytaking each consecutive pair of data points andcalculating the area between these points usinggeometrical techniques from the algorithm devel-oped given below.

When both points Y1 and Y2 are of the samepolarity (either both +ve or both �ve) (Fig. 2(a)),the area was calculated using the following:

Area ¼ ðX2 � X1ÞðY1 þ Y2Þ

2

� �: ð1Þ

When the points Y1 or Y2 are of oppositepolarity (Fig. 2(b)), because the absolute area hasto be calculated, the negative Y value has to bemade absolute to become a positive value and thenthe area between the two points was calculatedusing the following:

Area ¼Y1

2

� �ððX2 � X1Þ � bÞ þ

Y2

2

� �ðbÞ; ð2Þ

where

b ¼ðX2 � X1ÞðY2ÞðY1 þ Y2Þ

: ð3Þ

As shown in Fig. 2(b), b is the length of the baseof the second triangle and a the length of the baseof the first triangle. By using either of these twomethods to calculate the area between the pointsthe total area under the graph can be calculated.An executable MS DOS program was developed

and used to calculate the absolute energy. Therelative unit of MAE (absolute energy) is voltage(V )� time (s).

2.1.3. Magneto-acoustic emission signal

A typical waveform output from the MAEsensor can be seen below. The excitation signal isshown as the sinusoidal waveform, and theresultant MAE waveform is also shown. As shownin Figs. 3(a) and (b), the MAE signal reduces withincreased material deformation.

2.2. Microhardness

Vickers hardness (HV ) measurements werecompleted using a Buehler MMT-3 digital micro-hardness tester. All tests were completed using aload of 2000 gf (19614 mN) according to [12].

2.3. X-ray diffraction

Uniaxial residual stress magnitudes were deter-mined using the local strain X-ray diffractiontechnique (XRD) as outlined in literature [13–16].Residual stress was determined at the centre of onesurface of tensile samples strained to 17.5% one ofwhich was in the as strained condition, one heattreated at 700�C and the other heat treated at1100�C, using a Philips X’Pert X-ray diffracto-meter.

Scan parameters were controlled using PhilipsX’Pert Data Collector (V1.2a) software with 2yvalues (2y is the angle between source anddiffracted X-ray beam) chosen to encompassthe Cu-Ka doublet for the {2 2 2} planes:

ARTICLE IN PRESS

Fig. 3. Shows the resultant MAE signal from (a) no deformation and (b) 17.5% plastically deformed AISI 430 samples. The sinusoid is

of the same scale on both plots.


135�p2yp138:5�: A minimum of 12 scans wereperformed for each measurement using evenlydistributed C angles within the range 0pCp60�

(C is the angle between the surface normal and thebisector of source and diffracted X-ray beam).Only positive tilting was used as negative tiltingcan lead to poor quality peaks when using the o-diffraction geometry [14].

The resulting spectra were analysed usingPhilips X’Pert Stress Software [17]. The Sin2ðCÞtechnique was used with a 1

2S2 value of

5.21� 10�12 (m2/N) taken from literature for the{2 2 2} plane [13]. The Pearson VII technique wasused to calculate the peak position on thediffracted intensity plots [18]. The measurementstaken indicated that the lattice spacing d versusSin2ðCÞ plots were linear, confirming that textureand stress gradient did not affect the calculatedstress magnitudes. Shearing stresses were assumednegligible when compared with normal stressmagnitudes. The errors quoted are fit errors ofthe d versus Sin2ðCÞ plots calculated by thesoftware. This software allowed a misalignmentcalculation to be performed confirming that thespecimen height and machine alignment werecorrect even though specimen height has not beenfound to have a large effect on determined stressmagnitudes [14].

2.4. Microstructural analysis

For microstructural examination, the etching ofthe ferritic AISI 430 stainless steel was carried out

using swab etching with a solution of HCL, HNO3

and H2O with a ratio of 1:1:1. The etching timevaried between 4 and 8 min. The revealed micro-structure is shown in Figs. 7(b)–(d). The grain sizeG was calculated according to ASTM [19] and wasused as a measure and to analyse the deformedmicrostructure, post elongation.

3. Results

The dependence of MAE (absolute energy) andMBN (RMS) on the deformation of AISI 430ferritic stainless steel is shown in Figs. 4 and 6.

It can be seen at the lowest level of plastic strain0% (as-received condition), where no permanentdeformation of the material has occurred, theMAE and MBN are at their highest (Fig. 4). Withincreased straining, both the MAE and MBNcontinuously reduce.

It is known that as plastic deformation in-creases, the dislocation density increases, and thata change in the arrangement of dislocations in thespecimen occurs as indicated by transmissionelectron microscopy [20]. The magnetic domainwalls get pinned or held in place by dislocationinteractions. Because of increased domain-walldislocation interactions at increased plastic defor-mation, the magnetising force is not high enoughto set them free and reduced domain movementresults in decreased BN [21]. This also results inincreased hardening of the material with increasedstrain.

ARTICLE IN PRESS

1/MBN = 7E-05(HV) - 0.0086

R2 = 0.9609

1/MAE = 4.71(HV) - 509

R2 = 0.9946

150

200

250

300

350

400

450

140 150 160 170 180 190 200 210

Hardness (HV)

1/M

AE

(ab

solu

te e

ner

gy)

0

0.001

0.002

0.003

0.004

0.005

0.006

0.007

1/M

BN

(RM

S)

MAE

MBN

Fig. 5. The relationship between hardness and the inverse of

MAE and MBN for AISI 430.

Fig. 4. Dependence of MAE, MBN and hardness on percentage plastic strain for AISI 430.


MBN is seen to be significantly more affected byinitial increases in tensile elongation, from 0% to2.2% plastic strain, than the MAE (i.e. MBNchange in magnitude over 0.56 times greater thanMAE). It is known for the elastic deformation of amaterial, that the 180� domain walls grow at theexpense of the 90� domain walls, thus leading togreater MBN activity. But once the plastic regimeis reached, the macroscopic elastic strain remainsapproximately constant, with plastic deformationoccurring through slip processes [22]. Because ofthe difference in the change in magnitude betweenMAE and MBN, it is therefore believed thatthe dislocation interactions between domain wallsact more strongly with 180� domain walls, thannon-180� domain walls. This has also beenobserved by Buttle et al. [23] for an annealed pureiron sample.

Fig. 5 shows that the inverse of MAE (absoluteenergy) and MBN (RMS) are linearly propor-tional to hardness, i.e. an increase plastic strainwith a corresponding increase in hardness, resultsin a decrease in MAE and MBN. The plotdemonstrates a good correlation between MAEand hardness, suggesting that the MAE techniquecould be used for assessing the hardness ofmaterial, up to a range of 7–8 mm in ferriticstainless steel. The reason why MAE is moreclosely related to hardness than MBN is because of

the depth of measurement that it can achieve.MBN is only a near-surface measurement techni-que (0.1 mm), thus is effected more by surfacestresses.

Fig. 6 shows the effect of annealing on hardness,MAE and MBN. An examination of the plotshows that the BN has low-level energy from100�C to about 500�C where the material is in itshardest state. This would correspond to the highdislocation density levels within the material andassociated domain-wall interactions. The slightincrease in the MBN measurement at 300�C wasdue to superfluous material distortion broughtabout during material preparation.

ARTICLE IN PRESS

Fig. 6. MAE, MBN and hardness as a function of heat treatment temperature for AISI 430.


Above this temperature, the BN increases. Theannealing of the material can be seen to begin at600�C from the hardness measurements. Therecovery process of the dislocations in which theybegin to rearrange themselves occurs here. As aresult, the small change in hardness can beattributed to this rearrangement and a smallreduction in the density of dislocations. The stressrelief caused at these high temperatures affects the90� domain walls more and hence allows theirmovement, more than that of 180� domain walls.Thus annealing initially increases the MAE morethan the MBN. But it is between 700�C and 800�Cthat the material is fully annealed. The annihila-tion of dislocations at the annealing temperatureallows the movements of all 90� domain walls,hence greater increase in magnitude at thistemperature by MBN compared to MAE. This iswhere the minimum hardness and maximumductility, toughness, and corrosion resistance ofthe material occurs.

The increase in hardness above 800�C occursbecause of austenite–martensite embrittlement.Because these samples were cooled in air, theybecome brittle due to small volumes of austenitetransforming to martensite [24]. This can be seenfrom the microstructural examination in Fig. 7(d).The martensite phase has a high dislocation

density which increases the barriers to domainwall movement, thus reducing the Barkhausennoise.

At elevated temperatures, 1000–1100�C, thedislocation recovery process continues and de-creased material strength and coercivity areobserved with corresponding increased MAE andMBN. The reason why MBN increases more thanMAE at 1100�C can be explained by the residualstress measurements completed on the samples byX-ray driffraction, see Fig. 7(a). Increased com-pressive stresses are found on the material surfaceat 1100�C. This can be a result of the increase inthe martensitic phase, which is larger in volumethan that of austenite [24]. This compressive stressthus increases the volume of the 180� domain wallson the surface of the material, allowing for moremovement and hence magnetic Barkhausen activ-ity. Since XRD is only a surface measurementdevice, it is impossible to known how the stressesare affected in the bulk of the material, hence theMAE is also seen to increase, most likely due tothe dislocation recovery process.

At 100�C (prior to annealing) the residualstresses were found to be tensile and relativelyhigh. And as expected, a decrease in the stresses atthe annealing temperature (700�C) was observedi.e. post annealing.

ARTICLE IN PRESS

Fig. 7. (a) XRD residual stress results on heat-treated samples, (b) deformed microstructure (ASTM G ¼ 7:23), (c) annealed (ASTM

G ¼ 6:04) and (d) austenite–martensite embrittlement (ASTM G ¼ 5:64). The martensite, which is the black structure, can be seen to

form within the ferritic grain structure.


Throughout experimentation, the results fromboth Barkhausen techniques are seen to comparefavourably.

4. Conclusion

Magneto-acoustic emission (MAE) was used tocharacterise the microstructural state of ferriticstainless steel. It was found that:

* A new measurement parameter has been devel-oped for microstructural characterisation calledMAE absolute energy based on measuring thearea under the magneto-acoustic emissionwaveform.

* MAE absolute energy was found to be linearlyinversely proportional to hardness.

* In both the plastically strained and heat-treatedsamples, the MBN and MAE measurements areseen to be comparable.

* X-ray diffraction residual stress measurementsand microstructural observations have beencompleted on the samples and they validateBarkhausen noise measurements.

* MAE and MBN can be used as an effectivemeans for ferritic stainless steel microstructuralcharacterisation.

Acknowledgements

The authors gratefully acknowledge financialassistance for this project provided by EnterpriseIreland under the International CollaborationFund, and to the assistance from the Universityof Limerick and the Budapest University ofTechnology and Economics.

References

[1] W.S. Brand*ao, V.T.L. Buono, P.V. Marques, P.J.

Modenesi, Welding Int. 6 (1992) 713.

[2] V.T.L. Buono, B.M. Gonzalez, M.S. Andrade, Scr. Mater.

38 (2) (1998) 185.

[3] T. Mohandas, G. Madhusudhan Reddy, M. Naveed, J.

Mater. Process. Technol. 94 (1999) 133.

[4] H. Kusanagi, H. Kimura, H.S. Sasaki, J. Appl. Phys. 50

(1976) 2985.

[5] S. Tyagi, J. Steinberg, A.E. Lord Jr., P.M. Anderson,

Phys. Status Solidi 64 (1984) 443.

ARTICLE IN PRESS


[6] M.M. Kwan, K. Ono, M. Shibata, J. Acoust. Emission 3

(1984) 144.

[7] D.C. Jiles, Introduction to Magnetism and Magnetic

Materials, Chapman & Hall, London, 1991.

[8] D.C. Jiles, D.L. Atherton, J. Phys. D 17 (6) (1984) 1265.

[9] M.J. Sablik, J. Appl. Phys. 74 (1993) 5899.

[10] M.J. Sablik, B. Augustyniak, J. Appl. Phys. 79 (1996) 963.

[11] BS EN 10 002-1:1990, Tensile testing of metallic materials.

[12] ASTM E 384-99, Standard test method for microindenta-

tion hardness of materials.

[13] V. Hauk, Adv. Sur. Teat. Technol. Appl. Eff. 4 (1987)

251–302.

[14] M.E. Fitzpatrick, A.T. Fry, P. Holdway, F.A. Kandil, L.J.

Shackleton, L. Suominen, Determination of residual

stresses by X-ray diffraction, Measurement, Practice

Guide, 2002, NPL.

[15] P.S. Prevey, in: K. Mills (Ed.), Metals Handbook, Vol. 10,

American Society for Metals, Cleveland, OH, 1986,

pp. 380–392.

[16] I.C. Noyan, J.B. Cohen, Residual Stress Measurement

by Diffraction and Interpretation, Springer, New York,

1987.

[17] Philips X’Pert Stress. 1.0a, 2001, Philips Analytical B.V.

[18] M.M. Hall Jr., V.G. Veeraraghavan, H. Rubin, P.G.

Winchell, J. Appl. Crystallogr. 10 (1977) 66.

[19] ASTM E112-96 standards, Standard Test Methods for

Determining Average Grain Size.

[20] L. Burkhardt, R.E. Beissner, G.A. Matzkanin, J.D. King,

Mater. Evaluat. 40 (1981) 669.

[21] D.J. Buttle, C.B. Scruby, G.A.D. Briggs, E.A. Little,

J.P. Jakubovics, Philos. Trans. R. Soc London A 320

(1986) 363.

[22] C.-G. Stefanita, D.L. Atherton, L. Clapham, Acta Mater.

48 (2000) 3545.

[23] D.J. Buttle, C.B. Scruby, G.A.D. Briggs, J.P. Jakubovics,

Philos. Mag. A 55 (6) (1987) 717.

[24] ASM Handbook, Metallography and Microstructure,

Vol. 9, 1988.

Documents

Magneto-acoustic emission for the characterisation of ferritic stainless steel microstructural state