9Cr Steels

Trans. Indian Inst. Met.

Vol.56, No. 5, October 2003, pp. 483-497TP 1856

1. INTRODUCTION

9Cr-1Mo ferritic steel and its modified versions arecurrently favoured structural materials for steamgenerator (SG) applications as they offer usefulcombination of high temperature mechanicalproperties and corrosion/oxidation resistance. Theadvent of new generation power plants have increasedthe operating steam temperatures and pressures toachieve higher efficiency and better environmentalprotection. This has led to the development ofmodified versions of 9 % Cr steels with excellentcombination of creep strength and ductility. Thesesteels include plain 9Cr-1Mo steel initially modified

by the addition of strong carbide forming elementssuch as Nb and V (9Cr-1Mo-V-Nb designated asT91 or P91), and further modified by the addition ofW (9Cr-0.5Mo-1.8W-V-Nb designated as T92 orP92) 2. The P92 steel is the same as Japanese NF616steel 3. Modified 9Cr-1Mo steel has been selectedfor all the steam generator components (shell, tubeand tube plate) of Indian prototype fast breeder reactor(PFBR) 4. The selection of modified 9Cr-1Mo steelfor PFBR steam generator applications is based onimportant considerations such as high temperaturemechanical properties including creep, low cyclefatigue (LCF) and creep-fatigue interaction, resistanceto loss of carbon to liquid sodium and consequent

CHARACTERISATION OF MICROSTRUCTURE IN 9 %CHROMIUM FERRITIC STEELS USING ULTRASONIC

MEASUREMENTS

Anish Kumar, B.K. Choudhary, K. Laha, T. Jayakumar,K. Bhanu Sankara Rao and Baldev Raj

Metallurgy and Materials Group, Indira Gandhi Centre for Atomic ResearchKalpakkam - 603102, India

(Received 1 August 2003 ; in revised form 18 September 2003)

ABSTRACT

Ultrasonic measurements have been performed for comprehensive microstructural characterisation in 9 % Crferritic steels. The variation in grain size, the effect of solution annealing temperature on microstructuredeveloped, the variation in microstructure of the weld joints, the effect of post weld heat treatment andprecipitation behaviour during thermal ageing and creep testing have been successfully evaluated by ultrasonicmeasurements. The ultrasonic velocity has been correlated with the hardness and the volume fractions of ferriteand martensite. For the first time, it has been shown that ultrasonic velocity measurements can be used forthe identification of Ac1 and Ac3 critical temperatures. Ultrasonic attenuation and a new ultrasonic parameter,spectral peak ratio (SPR), have been found useful for the assessment of grain size and identification of variousmetallurgical features affecting the grain size such as dissolution of carbides and formation of d-ferrite. Thebenefits of SPR over conventional attenuation measurements have been demonstrated. The study revealed thata single ultrasonic parameter cannot characterise the solution annealing behaviour completely, and henceultrasonic velocity, attenuation and spectral analysis of the first backwall echo need to be used in a complementarymanner for comprehensive characterisation of the solution annealing behaviour. The ultrasonic velocity measurementsperformed on different solution-annealed specimens have been used for the imaging of weld metal and heataffected zone, obtaining the weld profile and also for assessment of the post weld heat treatment. The studyrevealed that the ultrasonic velocity could also be used effectively for investigating the microstructural degradationoccurring due to thermal ageing and creep. The present study has provided a better insight into the ultrasonicwave/microstructure interaction and established new methodologies for non-destructive assessment of variousmicrostructures in 9% Chromium ferritic steels useful for practical situations.

TP 1

484

TRANS. INDIAN INST. MET., VOL. 56, NO. 5, OCTOBER 2003

reduction in strength, resistance to wastage in case ofsmall leaks leading to sodium-water reaction andresistance to stress corrosion cracking in sodium andwater media 4. A major programme is underway atIndira Gandhi Centre for Atomic Research to developand evaluate the steel indigenously. 9 % Cr steels arerecommended for use in the normalised and temperedcondition having fully tempered. The components ofsteam generators operate at high temperatures forlong duration and undergo ageing inducedmicrostructural degradation, which in turn influencesthe mechanical properties. Therefore, it is necessaryto characterise the microstructure using non-destructive evaluation (NDE) methods for qualitycontrol of fabricated and heat treated componentsand for evaluating their performance during hightemperature service.

At elevated temperatures, the performance of ferriticCr-Mo steel weldment is considered to be a lifelimiting factor and a high percentage of failures hasbeen reported to be weld related. In 9 % chromiumsteels, the microstructure of the weld joint consistsof predominantly martensite in the weld metalfollowed by coarse gain martensite with d-ferrite,coarse grain martensite, fine grain martensite andinter-critical structure in the heat affected zone(HAZ) 5. In the as-welded condition, weld metalexhibits very high hardness followed by a hardnessgradient in the HAZ (rapidly decreasing hardness upto the hardness level of base metal). A proper postweld heat treatment (PWHT) is carried out to temperthe weld and HAZ microstructure thus enabling therelease of undesirable tensile residual stresses in theweld region and also to improve the toughness andhigh temperature mechanical properties of theweldment. In order to ensure the desiredmicrostructure and mechanical properties after PWHT,a suitable non-destructive evaluation based techniqueis essential. Ultrasonic technique offers good promisefor the characterisation of microstructures. Further,many a time, it becomes necessary to identify theweld interfaces and the profile for fixing the requiredbeam angle and scan ranges for ultrasonic inspectionand for unambiguous defect location and interpretationof the ultrasonic results. Hence, identification of asuitable NDE parameter is necessary for knowing theweld profile before carrying out automated ultrasonictesting for detection and quantitative characterisationof defects in the weldments.

The velocity of ultrasonic wave is influenced by theelastic moduli and the density of the material, whichin turn are mainly governed by the amount of variousphases present and the damage to the material. Theultrasonic attenuation, which is the sum of absorptionand scattering, is mainly dependent upon the dampingcapacity and the scattering from the grain boundary,wedge cracks (if present) etc. in the material. Bothultrasonic velocity and attenuation have been appliedfor the characterisation of microstructures and forthe evaluation of material properties to ensure theabsence of unacceptable discontinuities and presenceof desired microstructure with acceptable propertiesduring fabrication and heat treatments 6,7. Ultrasonicin-service inspection is carried out to detect anyunacceptable degradation in microstructure andformation and extension of defects in a componentduring service 8. Ultrasonic parameters, such asvelocity and attenuation, have been correlated withthe microstructural features evolved during heattreatments in ferritic steels 9,10, superalloys 11,aluminum alloys 12 and many other materials. Thistechnique has also been used successfully for theevaluation of yield strength 13, fracture toughness 14,grain size 15 and volume fraction of second phases 16.Dubois et al. 17 have used on-line laser based ultrasonicattenuation measurement for monitoring phasetransformation in steels. They observed that theultrasonic attenuation increases initially with increasein temperature followed by a decrease in theintercritical region (between Ac1 to Ac3). Above Ac3,the attenuation increases sharply with temperature dueto increase in grain size. Such a variation in theultrasonic attenuation with increasing solution-annealing temperature indicated that a single ultrasonicparameter might not be sufficient to characterise theseveral microstructural features in ferritic steels. Inaddition to attenuation, other parameters are neededto determine the amount of different phases inintercritical region (between Ac1 to Ac3) and toidentify the various critical temperatures moreunambiguously. It was realised that a multi-parameterapproach can be more appropriate in such conditions.

In Nickel base superalloy Inconel 625, ultrasonicvelocity measurements have been correlated withprecipitation of carbides and various intermetallicphases such as g�, Ni2(Cr, Mo) and d 11. Linearcorrelations have been established between ultrasonic

485

ANISH KUMAR et al. : CHARACTERISATION OF MICROSTRUCTURE BY ULTRASONIC MEASUREMENTS

velocity and yield stress for precipitation of variousphases 11. It has been observed that the dissolution ofNi2(Cr, Mo) phase and precipitation of d-phase havemore effect on ultrasonic velocity as compared tothat on yield stress, whereas dissolution andprecipitation of g� has more influence on yield stressthan on ultrasonic velocity. In order to explain theinfluence of coherent precipitates on ultrasonicvelocity, Fouquet et al. 18 have pointed out that inpresence of precipitates, the material can be consideredas a composite of three components - matrix, matrix-precipitate interface and precipitate. Since ultrasonicvelocity is dependent upon the Young�s modulus anddensity of the material 6, it is dependent upon theYoung�s modulus and density of these threecomponents. In incoherent matrix-precipitateinterface, the loss of coherency leads to decrease inthe Young�s modulus of the interface and thereforereduces the effective ultrasonic velocity of thematerial. The formation of coherent precipitates leadsto an increase in the ultrasonic velocity and theircoarsening or dissolution results in a decrease in thevelocity.

Ultrasonic velocity and attenuation measurements inaustenitic stainless steel weldment indicated that theattenuation measurement is useful for studying thevariation in grain size across the weldment 19,20. Ithas also been pointed out that in austenitic stainlesssteel weldments, ultrasonic velocity is mainly affectedby the grain orientation and not by the microstructure,i.e. the amount of ferrite in the weld metal 20.

Ultrasonic spectroscopy is a further development ofthe pulse-echo technique, which uses broadbandultrasound and analyses the spectra of the echo pulses.In an analogy with light, the specimen is insonifiedwith white sound. The defects or microstructure alterthe �colour� of the wave travelled through the materialand the change in the colour of the sound gives theinformation about the material. Ultrasonicspectroscopy has been used for the characterisationof defects and microstructural features 21. It has beenreported that, in stainless steel, as the grain sizeincreases, the peak frequency and the full width athalf maximum of the autopower spectrum of the firstback wall echo decrease 22. The shift in the peakfrequency has also been used for evaluation ofstructural variations induced by tensile deformationin SUS304 stainless steel 23. The spectral peak

frequency in SUS304 steel was found to increasewith increase in the tensile elongation, which wasattributed to the formation and growth of martensitestructures due to tensile deformation resulting insmaller crystalline grains, thus reducing theattenuation due to ultrasonic scattering 23. Gerickehas illustrated the use of ultrasonic spectroscopy forgrain size determination 24. The frequencycharacteristic of the transducer was found to showtwo well defined humps centered at frequencies ofabout 3.5 and 6 MHz, respectively. The correspondingfrequency responses of steels with varying grain sizesshowed that, as the mean grain size increased, theratio of the heights of the two frequency humpschanged, the higher frequency being preferentiallyattenuated by coarser grained material 24.

The present paper deals with the application of variousultrasonic parameters for comprehensivecharacterisation of microstructures in 9 % Cr ferriticsteels. An attempt has been made to correlate thevarious ultrasonic parameters with microstructure interms of the variations in grain size andmicrostructural constituents with solution annealingtemperature ranging from a-ferrite (1073 K) to g + dphase regions (1673 K), imaging of various regionsof weld joint such as weld metal and HAZ andassessment of post weld heat treatment in modified9Cr-1Mo steel. The paper also presents the variationin ultrasonic velocity with the degradation occurringdue to thermal ageing and creep in 9Cr-1Mo ferriticsteel.

2. EXPERIMENTAL

2.1 Test Specimens, Microstructural examinationsand Hardness Testing

The chemical compositions of both the plain and themodified 9Cr-1Mo steels investigated in the presentstudy are given in Table I. For investigating thesolution annealing behaviour of modified 9Cr-1Mosteel, various specimens of 12 x 12 x 60 mm sizewere soaked for a duration of 5 min. at differenttemperatures ranging from a-ferrite (1073 K) tog + d ferrite phase region (1623 K) followed by oilquenching. These treatments were intended to obtaindifferent microstructures and grain sizes in thespecimens simulating the different regions in the heataffected zone of the weldment. Also, the range of

486


temperatures of these treatments included the solutionannealing (austenitising) treatment generally given tothe ferritic steels. Surface grinding of these sampleswas carried out to obtain the specimens of 10.5 ±0.2 mm thickness with plane parallelism to anaccuracy of better than ± 5 mm. Metallographicexamination was carried out to reveal themicrostructures in different specimens prepared byusual metallographic technique and by etching withVillella�s reagent. After completing all the ultrasonicmeasurements, in order to reveal the prior austenitegrains, these samples were given further heat treatmentat 1033 K for 1 h and aged for 2000 h at 923 K. Thelinear intercept method as per ASTM standard E112-88 was used to find out the average grain size. Thehardness of these specimens was also measured usingVicker�s hardness tester at a test load of 10 kg.

For imaging of weldment and assessment of postweld heat treatment in modified 9Cr-1Mo steel, weldjoint specimens of 60 mm length and 12 mm widthwere machined from 12 mm thick weld pad in theas-welded condition. Some of these specimens weregiven a PWHT of 1033 K for 1 h. Surface grindingof these specimens was carried out to make thesurfaces plane parallel with an accuracy better than± 5 mm. Metallographic examinations were carriedout to reveal the various regions of the weld joints.The Vicker�s hardness measurements were carriedout in the weld metal and HAZ at distances of regularintervals from the weld interface.

In order to investigate the precipitation behaviourand microstructural degradation in 9Cr-1Mo steel,the as-received quenched and tempered (Q + T: 1223

K/5 h water quenched followed by 1023 K/8 h aircooled) specimen blanks of 12 mm diameter and 60mm length were machined in the thickness directionfrom a 300 mm thick forging. These blanks werethermally aged (TA) at 793 and 873 K for durationsranging from 10 to 5000 h. For long ageing times,the samples were taken from the unstressed shoulderregions of the creep specimens tested at 793 and873 K. For investigating the influence of creepdeformation, specimens were obtained from the gaugelength region of creep specimens tested at 873 K forrupture lives ranging from 11 to 15175 h. The detailsof creep tests are described elsewhere 25. The Vicker�smicrohardness measurements were carried out at 0.490N load on all the specimens. For ultrasonicmeasurements, surface grinding of these specimenswas carried out to obtain a constant thickness of 5mm and plane parallelism to an accuracy of ± 2mm. Transmission electron microscopic (TEM)studies were performed on the carbon replicas usinganalytical transmission electron microscope (ATEM),Philips CM 200. The energy dispersive analytical X-ray (EDAX) spectra and selected area diffraction(SAD) patterns were taken for the identification ofvarious precipitates.

2.2 Ultrasonic Measurements

The experimental setup used for the ultrasonicmeasurements is shown in Fig.1. 100 MHz broadband pulser-receiver (M/s. Accutron, USA) and 500MHz digitizing oscilloscope (Tektronix TDS524) wereused for carrying out the ultrasonic measurements.For these measurements, the rf signals were digitized

Table ICHEMICAL COMPOSITION (WT %) OF 9CR-1MO FERRITIC STEELS STUDIED IN THE PRESENT INVESTIGATION.

Material Element (Wt. %)

Cr Mo C Si Mn V Nb S P Fe

Mod. 9Cr-1Mo steel 8.72 0.90 0.096 0.32 0.46 0.22 0.08 0.006 0.012 Balance

Mod. 9Cr-1Mo steel

Weld metal 8.63 1.02 0.098 0.35 0.58 0.20 0.06 0.008 0.011 Balance

9Cr-1Mo steel 9.27 1.05 0.07 0.75 0.63 - - 0.001 0.02 Balance

487


at 500 MHz and the gated backwall echoes (1024 nsduration) from the oscilloscope were transferred tothe personal computer with the help of GPIBinterfacing and LabVIEW software. Ultrasonicvelocities were measured using 5 MHz shear and 15MHz longitudinal beam transducers. Cross correlationtechnique has been used for precise velocitymeasurements. The accuracy in time of flightmeasurement was better than ± 1 ns. The scatter inthe measurement of longitudinal and shear wavevelocities is ± 3 m/s and ± 2 m/s, respectively.Ultrasonic attenuation measurements were made using15 MHz longitudinal wave transducer with fixedperspex delay line of 10 mm. Spectral analysis hasbeen carried out on the rf signal corresponding to thefirst backwall echo obtained by 20 MHz longitudinalbeam transducer. Two distinct peaks have been foundin the autopower spectrum of the first backwall echoesfor all the specimens, when 20 MHz transducer isused. The ratio of these two peaks has been used asan ultrasonic parameter. Specific softwares weredeveloped for automatic on-line measurement ofprecise time of flight and attenuation of ultrasonicwaves and also for the peak ratio measurements fromthe autopower spectrum.

Ultrasonic velocity measurements have been carriedout across the weld-line in two perpendicular sections

(Fig. 2) using 15 MHz longitudinal wave transducerof 4 mm diameter. To study the feasibility of thedetection of various regions in the weldment,ultrasonic velocity measurements have been carriedout across the weld line at three different depths,such as near crown, middle and near root as shownin Fig. 2a (Scan I, Scan II and Scan III respectively),as ultrasonic waves always see the similarmicrostructure throughout the specimen in the wavepropagation direction. But during practical conditions,this surface would not be accessible for themeasurements to be carried out. The only accessiblesurface would be the top surface and hence to evaluatethe weld profile under practical conditions, ultrasonicvelocity measurements have also been carried outfrom the top surface (Fig. 2b). The plot of time offlight (TOF) against scanning distance across the weldis plotted on-line using a PC based TOF measurementsystem, US-TOF, supplied by M/s. FraunhoferInstitute for NDT (IZFP), Germany. At the scanspeed used in the present study, three data pointswere recorded for each millimetre length of thespecimen. The scatter in the ultrasonic velocitymeasurements is less than ± 2.5 m/s.

3. RESULTS AND DISCUSSION

3.1 Solution Annealing Behaviour

Typical photomicrographs for the specimens withvarious microstructures and grain sizes obtained byemploying solution annealing temperatures rangingfrom a-ferrite region to g + d region are shown inFig. 3. It can be seen that in the specimens heattreated at temperatures below lower criticaltemperature (Ac1 ~ 1100 K), the microstructure iscomposed of only ferrite with laths and carbides(tempered martensite) (Fig. 3a). With increase in

Fig. 1 : Experimental setup for ultrasonic velocitymeasurement.

Fig. 2 : Schematic of the ultrasonic scanning (a) for feasibility study and (b) with practical utility.

(a) (b)

488


solution annealing temperature above Ac1, partialaustenitisation during soaking resulted in themicrostructure with ferrite and freshly formedmartensite (Fig. 3b). Further increase in temperaturein the ferrite + austenite region (between Ac1 andAc3) provided higher amount of austenite duringsoaking and higher volume fraction of martensiteafter quenching. The amount of martensite increasedtill the upper critical temperature (Ac3 ~ 1180 K)

and only lath martensite with fine prior austeniticgrains was observed at 1223 K (Fig. 3c). Withincrease in temperature above Ac3 a marginal increasein grain size was observed up to ~ 1373 K (Fig. 3d).Beyond this, a rapid increase in the prior austeniticgrain size was observed (Fig. 3e). The marginalincrease in the grain size can be attributed to thepresence of fine vanadium and niobium carbides(V4C3 and NbC), which restrict the grain growth

(a)

Fig. 3 : Microstructures of the specimens solution annealed at (a) 1073 K (b) 1148 K (c) 1273 K (d) 1373 K (e) 1473 Kand (f) 1623 K for 5 Minutes, followed by oil quenching.

(b) (c)

(e) (f)(d)

489


process during austenitisation in the temperature range1223-1373 K 26. The dissolution of these carbides attemperatures higher than 1373 K results in a rapidgrowth in the prior austenite grain size. At about1490 K (just above Ac4), d-ferrite starts forming atthe prior austenite grain boundaries, which restrictsthe grain growth resulting in relatively finer prioraustenitic grain size and duplex martensite + d-ferritemicrostructure (Fig. 3f).

The variations in ultrasonic longitudinal and shearwave velocities with soaking temperature are shownin Figs. 4(a) and (b), respectively. The variation inhardness with soaking temperature is also shown inthese figures. In the specimens heat treated attemperatures below Ac1 i.e., 1100 K, both theultrasonic longitudinal (Vl » 5995 m/s) and shear(Vs » 3310 m/s) wave velocities were the highest.With increase in temperature above Ac1 (1100 K), arapid decrease in both the velocities till Ac3temperature i.e., » 1180 K was observed. Like sharpchanges in the ultrasonic velocities, the hardness valueincreased rapidly with increasing temperature betweenAc1 and Ac3 temperature. The sharp decrease in theultrasonic velocities and a rapid increase in thehardness in the intercritical region are attributed tothe increase in the amount of martensite with increasein the soaking temperature. Above Ac3 temperaturei.e., 1180 K, the velocities remained almost constant(Vl » 5950 m/s and Vs » 3270 m/s) up to about 1323K. Beyond this, the ultrasonic velocities were foundto decrease marginally. The decrease in the ultrasonicvelocities above 1323 K can be attributed to theincrease in the grain size27. The difference in thelongitudinal wave velocity between ferrite andmartensite is 0.75 % (Fig. 4a), whereas this differencefor shear wave velocity is observed as 1.21 %(Fig. 4b). These observations suggest that the shearwave velocity could be more reliable than longitudinalwave velocity for the same accuracy in the time offlight measurement.

The variations in the ultrasonic attenuation and grainsize with soaking temperature are shown in Fig. 5.It can be seen that both the ultrasonic attenuation andgrain size exhibited similar trend in their variationswith soaking temperature. The specimens heat treatedbelow Ac1 temperature exhibited nearly a constantattenuation of » 0.25 dB/mm. With increasingtemperature above Ac1, a decrease in the attenuation

(a)

(b)

Fig. 4 : Variation in hardness and ultrasonic (a) longitudinaland (b) shear wave velocities with soakingtemperature.

Fig. 5 : Variation in ultrasonic attenuation and grain sizewith soaking temperature.

490


was observed and this decrease in attenuation isascribed to increase in the amount of martensite(having low attenuation) with increasing temperature.Ultrasonic attenuation was found to be minimum(» 0.07 dB/mm) in the specimens quenched from thetemperatures just above the Ac3 i.e., in which theprior austenite grain size was also found to beminimum. Attenuation values exhibited a marginalincrease in the specimens heat treated above 1323 Kfollowed by a rapid increase at temperatures above1373 K. Both marginal and rapid increase inattenuation are attributed to the respective increase inthe grain size. At temperatures above ~ 1473 K, theultrasonic attenuation again decreased with increasingtemperature due to decrease in grain size as aconsequence of d-ferrite formation.

The variation in the attenuation with the soakingtemperature and grain size can be explained with thehelp of the attenuation theory. Ultrasonic waves in apolycrystalline material are attenuated by structuralboundaries and grains. The total attenuationcoefficient a may be expressed as

a = aa + as , (1)

where aa is related to the losses due tothermoelasticity, magnetic hysteresis and dislocationdamping, i.e., ultrasonic absorption, and as takesinto account the losses due to the scattering of theultrasonic waves. The attenuation is dominated byRayleigh scattering, when the wavelength of theultrasonic waves (l) is greater than the grain size d(the condition that is valid in this study i.e.,l = 400 mm and the values of d are in the range 20-130 mm). The ultrasonic attenuation coefficient dueto the Rayleigh scattering (as) in polycrystallinematerials can be expressed as

as =S.d3.f4, (2)

where S is a scattering factor that depends on theelastic properties of the material (including soundvelocity), d is the average grain size in the specimenand f is the frequency.

It can be seen in Fig. 5 that, for the same prioraustenite grain size, the attenuation is found to dependon microstructure. For example, the line dAdBdCrepresents constant grain size value of ~ 50 mmobtained at three soaking temperatures of

TA = 1096 K, TB = 1382 K and TC = 1625 K. Thecorresponding microstructures for these conditions are(I) ferrite, (II) martensite and (III) martensite withd-ferrite, respectively. The attenuation valuescorresponding to these soaking temperatures are aA(0.22 dB/mm), aB (0.1 dB/mm) and aC (0.337 dB/mm), respectively. Papadakis 28 reported that thescattering is proportional to prior austenite grainvolume and is much weaker in martensite than inpearlite and its associated phases i.e., ferrite andcementite. Our results are in agreement with this,i.e., the martensite structure exhibits the lowestattenuation. Even in the presence of martensite, thespecimen with martensite and d-ferrite exhibits higherattenuation than that in the specimen with ferrite andcarbide. This is attributed to the presence of coarserlaths in the ferrite (Fig. 3a), which makes the ferritemore isotropic than d-ferrite. Besides this, thepresence of two phases makes the sample withmartensite and d-ferrite most anisotropic and hence itexhibits the highest attenuation for the same grainsize.

Figures 6a and 6b show typically the changes in thefirst backwall echoes and the autopower spectra withthe change in the grain size in the specimens withmartensitic structure. Two distinct peaks centeredaround 7.0 MHz and 17.5 MHz have been found inthe autopower spectrum of the first backwall echoesin all the specimens, when 20 MHz transducer isused. The relative heights of the two peaks changewith the change in the grain size. The spectral peakratio (SPR) is found to increase with increase ingrain size. Figure 7 shows the variation in SPR withsoaking temperature exhibiting a similar behaviouras shown by ultrasonic attenuation in the variation ofattenuation with soaking temperature (Fig. 5). TheSPR decreased with increase in soaking temperaturein the intercritical region (1100 to 1180 K) becauseof the decrease in the grain size. Beyond this, SPRincreased with increase in the soaking temperaturedue to increase in grain size. The SPR increasedrapidly above 1373 K due to the sharp increase ingrain size. Like attenuation, the SPR decreased withincrease in soaking temperature above 1473 K due todecrease in grain size.

The change in SPR with grain size can be explainedon the basis of ultrasonic attenuation theory (eq. 2)and the spectral analysis of the ultrasonic waves.

491


When the ultrasonic wave is attenuated in the material,it can be assumed that the oscillatory amplitude decaysexponentially as the propagation length of theultrasonic wave increases. If ultrasonic wave isconsidered as the spectral distribution, which has anoscillatory pulse of one or two cycles with anexponential decay envelope, the amplitude of theultrasonic wave g(t) is expressed as 29

g(t) =A.exp(-at).sin2p(f0t-f) (f/f0<t<(2+f)/f0)(3)

where a is the attenuation coefficient, t is time, f0is the carrier frequency, f is phase and A is themaximum amplitude of the pulse wave. The spectrumpower distribution |G(f)| is derived by Fouriertransform 23

( )

( )

( )G f

f

f

f f

f

f f

d

d

d

2

0

0

0

0

2 2

0

2

12

22

4µ

+-æ

èç

ö

ø÷ -

-æ

èç

ö

ø÷

-

+ -

exp

exp cos

a

a p

a p

(4)

From the ultrasonic scattering model (eq. 2), theattenuation increases with increase in grain size andfrequency. As the grain size increases, a and also thefrequency dependency of a increase 15. Therefore,as the prior austenitic grain size increases, moreattenuation takes place (Fig. 6) and the ratio of thetwo peak heights in the autopower spectrum changes(Fig. 8b) due to the increased attenuation of higherfrequency components.

Figures 8a and 8b show the effect of the couplingcondition on the first backwall echo and its autopowerspectrum, respectively, for the specimen solutionannealed at 1573 K. It can be seen that when thecoupling condition was not good, the amplitude ofthe first backwall echo and its autopower spectrumdecreases. In the autopower spectrum, the amplitudeof both the peaks decreased maintaining the peakratio almost constant and therefore the peak ratioremains independent of variations in the couplantcondition. Similar behaviour has been found in allthe other specimens used in this study. Theindependence of ultrasonic spectral parameter oncouplant condition is in agreement with our earlierwork on type 316 austenitic stainless steel 22. These

(a)

(b)

Fig. 6 : Variations in (a) the first backwall echo and (b) theautopower spectrum with grain size in specimenswith martensitic structure.

Fig. 7 : Variation in spectral peak ratio and grain size withsoaking temperature

492


observations clearly indicate that if these spectralparameters are used for grain size measurement thenthe error involved in the measurements due to variationin the couplant condition can be minimised for on-line and/or when a large number of measurementsare to be made by moving the specimen/transducer.This has an important practical significance for gainsize measurement by ultrasonic measurements. Inaddition to this, since only first back wall echo isrequired for the grain size measurements, thisapproach can be used for grain size measurementseven in thicker and highly attenuating materials, whereobtaining multiple back wall echoes with good signalto noise ratio is very difficult.

3.2 Weld Imaging and Evaluation of PWHT

The microstructures of the ferritic steel weldmentconsist of weld metal and HAZ microstructures apartfrom that of base metal. The microstructures in theHAZ are similar to that of the base metal specimenssolution annealed at different temperatures rangingfrom a-ferrite (1073 K) to g + d-ferrite phase region(1673 K), which have been characterised and discussedin the previous section. In this section, an attempthas been made to characterise the different regimesof the weldment using the information obtained onsolution annealing material. The step-wiseinvestigation involved feasibility study, imaging ofweldment and assessment of PWHT. The practicalsignificance of this work is discussed.

3.2.1 Feasibility Study

Figure 9 shows the variation in ultrasonic velocitywith scanning distance across the weld-line at threedifferent depths, such as near crown, center and nearroot (indicated in Fig. 2a and Fig. 9 as Scan I, ScanII and Scan III) of the weldment in as-weldedcondition. Ultrasonic velocity is found to be constant(~ 6005 m/s) in the parent base metal and itdecreases as the HAZ is approached. Ultrasonicvelocity decreases rapidly up to the weld interfacefollowed by a gradual decrease to a minimum

Fig. 8 : Effect of different coupling conditions on (a) thefirst backwall echoes and (b) their autopower spectra(note that the heights of both the peaks gets changedby the change in coupling condition, but the spectralpeak ratio remains independent of the couplantcondition.

(a)

(b)

Fig. 9 : Variation in ultrasonic velocity with scanning distanceacross the weld-line at three different depths. (WM� Weld metal, HAZ � Heat affected zone).

493


(~ 5925 m/s) at the centre of the weld and anincrease in the other side of weld in a symmetricmanner. The difference in the width of the weld atdifferent depths can be seen clearly in the velocityplot. The decrease in ultrasonic velocity in the weldmetal is attributed to the presence of martensiticstructure with lower elastic modulus compared tothat of the ferrite. In addition to this, the presenceof residual stress would also have some contributionon the change in the ultrasonic velocity. However,the expected influence of residual stress on ultrasonicvelocity is much less (~20 m/s) 30 as compared tothe large change in the velocity (80 m/s) observed inthe present investigation. Further, the present studyhas been carried out on the specimens cut from theweld-pad, and hence majority of the residual stresswould have got released. Even after the PWHT(discussed in next paragraph), considerable differencein the velocity exists between the weld and the parentmetal (~30 m/s), which further indicates that in9Cr-1Mo ferritic steel weldments, the variation inultrasonic velocity around the weld is predominantlygoverned by microstructure rather than by residualstress. Ultrasonic velocity has been found to beminimum (~ 5925 m/s) at the weld centre near thecrown (Scan I in Fig. 2a). The minimum velocity atthe weld centre increases towards the root of theweld (~ 5941 m/s in Scan III in Fig. 2a). In theweld metal, the higher ultrasonic velocity at the rootcompared to that in the crown is attributed to marginaltempering of the martensitic structure in the rootregion due to thermal cycle during the subsequentweld passes.

The variations in ultrasonic velocity and hardnesswith scanning distance across the weld-line at thecentre of the weld (Scan II in Fig. 2a) in the as-welded and after PWHT conditions are shown inFig. 10. After PWHT, ultrasonic velocity increasesfrom 5930 m/s to 5970 m/s in the weld metal,showing the possibility for the assessment of PWHTand its adequacy using ultrasonic velocitymeasurements. Even though ultrasonic velocityincreases after PWHT, it does not reach the velocityof that of the parent base metal. This can be ascribedto the fact that the heat treatment for 1 h at 1033 Kgiven on the weldment is not sufficient for completetempering of the microstructure. This is furthersupported by the microhardness measurements made

in the as-welded and tempered specimens (Fig. 10).In the as-welded specimen, the maximum hardnesswas found at the centre of the weld (550 VHN). Thehardness of the parent metal is 225 VHN. Even afterPWHT, hardness at the centre of the weld is 300VHN, i.e. 75 VHN more than the parent metal (Fig.10). In the parent metal, a slight decrease in ultrasonicvelocity is observed after PWHT. It can be seen inFig. 4a that the minimum ultrasonic velocity in theHAZ could be 5950 m/s only and hence this valuecan be used for identifying the boundary between theHAZ and the weld metal. In the weld metal, ultrasonicvelocity is further lower due to martensitic structurewith high dislocation density. This is also supportedby the hardness measurements, which shows that themaximum hardness in the specimens simulating theweld microstucture is 450 VHN only (Fig. 4) ascompared to 500 VHN at the centre of weld metal inthe as-welded specimen (Fig. 10).

3.2.2 Ultrasonic Velocity Based Imaging of WeldProfile

Ultrasonic velocity measurements carried out acrossthe weld line at three different depths (near crown,centre and near root as shown in Fig. 2a) of theweldment in as-welded condition have been used toimage the weld structure. Figure 11 shows theultrasonic velocity based image of the weldment inas- welded condition. Ultrasonic velocity in the parentmetal has been taken as greater than 6000 m/s. In theintercritical region, it has been taken as less than6000 m/s but greater than 5950 m/s, as discussedearlier in the case of simulated weld specimens. Theweld metal is defined as having velocities less than

Fig. 10: Variation in ultrasonic velocity and hardness withscanning distance in as welded and after PWHTconditions.

494


5950 m/s. The ultrasonic velocity image of theweldment is found to almost replicate the actualphotograph of the macro-etched weldment, as shownin Fig. 12. The slight difference in the weld profile(near crown and root) is because the velocity scannearest to the crown and root could be done only 2mm away from the edges, as the diameter of thetransducer used was 4 mm and hence the weld profiledetails very close to the crown and root could not berevealed.

3.2.3 Study with Practical Significance

The variation in ultrasonic velocity with scanningdistance across the weld-line in as-welded and PWHTconditions are shown in Fig. 13. The measurementswere performed from the top surface of the weldmentas depicted in Fig. 2b (this surface would be accessiblefor measurements on any weldment of a component).As discussed earlier, with increased fraction ofmartensitic phase in the direction of wavepropagation, ultrasonic velocity should decrease. Thiscan be seen in Fig. 13 that the ultrasonic velocitydecreases as the measurement location is slowly movedfrom parent metal region to centre of the weld i.e.,with increased fraction of weld metal (martensiticstructure) in the direction of wave propagation.

Comparison of ultrasonic velocity plot in Fig. 13with the photomacrograph of the weldment (Fig. 12)reveals that the ultrasonic velocity profile almostreplicates the weld profile and hence suchmeasurements can be employed effectively to depictthe weld profile in the ferritic steel weldments.Further, it can be seen clearly in Fig. 13 that theultrasonic velocity increases (5910 m/s to 5955 m/s)in the weld region after PWHT due to the temperingof the martensitic structure. This suggests that thePWHT can be monitored on actual weldment usingultrasonic velocity measurements. This study has thepractical significance even during the repair weldingand subsequent PWHT, as the weld profile can bedetermined and also the adequacy of the PWHT withproper microstructure can be assessed non-destructively. This study has been carried out on thespecimens of 12 mm thickness with plane parallelismof ± 5 mm. For practical applications of thedetermination of weld profile, the weld bead shouldbe flushed properly and the thickness variations shouldbe as less as possible. Whereas, for the assessment ofPWHT (where the difference in velocity in as-weldedand PWHT conditions is measured), the variation inthickness can be taken care of by carrying out themeasurements at the same location before and afterPWHT.

3.3 Microstructural Degradation during ThermalAgeing and Creep

The variation in ultrasonic velocity in 9Cr-1Mo steelwith ageing time at 793 and 873 K and creep exposureat 873 K are shown in Fig. 14. Based on the variationin ultrasonic velocity with exposure time, the wholeageing period has been divided in four differentregimes; regime-I, regime-II, regime-III and regime-IV for better clarity in the discussion. In Fig. 14, thevarious regimes are shown for thermal ageing at873 K. Regime-I consists of increase in ultrasonicvelocity with ageing time at shorter durations (up to10 h at 873 K and 100 h at 793 K). At intermediatedurations, ultrasonic velocity decreased with ageingtime (up to 1000 h at 873 K and 2200 h at 793 K)in regime-II followed by increase in ultrasonicvelocity (up to 5000 h at 873 K and 10850 h at793 K) in regime-III. At long durations, continuousdecrease in ultrasonic velocity with ageing time(beyond 5000 h at 873 K) was observed in regime-IV. The variation in hardness with ageing time also

Fig. 11: Ultrasonic velocity based imaging of the weld profile.

Fig. 12: Macrophotograph of the etched weldment.

Fig. 13: Variation in ultrasonic longitudinal wave velocitywith scanning distance across the weld-line in as-welded and PWHT weldments.

495


exhibited four different regimes similar to that ofultrasonic velocity (Fig. 15).

TEM micrographs for specimens thermally aged fordifferent durations are shown in Fig. 16. The selectedarea diffraction (SAD) patterns for the indicatedprecipitates are shown in insets. Q+T as-receivedspecimen exhibited presence of precipitates at prioraustenite grain and lath boundaries and in the intralathmatrix regions. Figure 16a shows globular precipitatesin the interlath and acicular precipitates in the intralathregions. The analysis of SAD pattern and EDXspectrum revealed that the intralath precipitates (shownby a needle in Fig. 16a) are mainly of Cr2X typewith hexagonal close packed structure. This chromiumrich precipitate has been identified as Cr2N 31-33. Theglobular precipitates at the grain and lath boundariesare mainly of M23C6 [(Cr Fe)23C6] type with facecentred cubic structure.

The initial increase in the ultrasonic velocity in regime-I can be ascribed to the further formation of fineCr2N precipitates in the intralath regions at shortdurations of ageing compared to that in Q+Tcondition (Fig. 16 a and b). This increasedprecipitation at short duration of ageing results fromthe ageing of specimens at lower temperatures of 793and 873 K than the 1023 K used for temperingtreatment. The formation of fine precipitates inspecimen aged at 873 K saturates early compared to

Fig. 14: Variation in ultrasonic velocity with ageing time at793 K and 873 K and creep deformation at 873 K.Various regimes shown for ageing at 873 K.

Fig. 15: Variation in hardness with ageing time at 793 K and873 K. Various regimes shown for ageing at 873 K.

Fig. 16: Transmission electron micrographs for the specimens(a) quenched and tempered, thermally aged at 873 Kfor (b) 10 h, (c) 1000 h, (d) 793 K for 10850 h and(e) 873 K for 15175 h.

496


results from microstructural degradation 25. The creepenhanced precipitation can be seen by the shift in theregimes to lower time as well as higher ultrasonicvelocity in creep exposed condition compared to thatin thermal ageing. The study demonstrated that theultrasonic velocity is sensitive to microstructure ofthe steel and the use of this technique can be extendedto assess the microstructural degradation and itsinfluence on the state of health of a component duringservice, by judicious selection of time periods betweenany two consecutive inspections. Further, the variationin the thickness of the actual component can be takencare of by carrying out the ultrasonic velocitymeasurements at the same place during service or byusing the ratio of ultrasonic longitudinal and shearwave velocities 36, if feasible. This ratio is same asthe transit time ratio for these two wave modes andhence the thickness is not required to be known atthe measurement location.

4. CONCLUSIONS

The study revealed that ultrasonic parameters can besuccessfully used for comprehensive characterisationof the solution annealing behaviour of ferritic steels.The ultrasonic velocity can be used to determine Ac1and Ac3 temperatures, hardness and the volumefraction of martensite and ferrite in the specimensheat treated between Ac1 and Ac3 temperatures.Ultrasonic attenuation and a new ultrasonic parameter,spectral peak ratio (SPR), are found to be sensitiveto grain size and these can be used to determine grainsize, temperature to the onset of dissolution of NbCand V4C3 and the formation of d-ferrite i.e., the Ac4temperature. The independence of spectral peak ratioon couplant condition has an important practicalsignificance, that the error involved for the on-linegrain size measurement by ultrasonic measurementscan be reduced by exploiting this parameter. Ultrasonicvelocity can also be used for the imaging of weldprofile in ferritic steels and for the evaluation of theadequacy of PWHT. These results suggest that theultrasonic parameters can be used effectively as aquality control tool for the initial and post fabricationheat treatment of the components. Further, theultrasonic velocity has been found to be very sensitiveto the microstructural degradation in 9Cr-1Mo ferriticsteel occurring due to thermal ageing and creepexposures and therefore, it has potential for lifeassessment applications of these steels.

that in specimen aged at 793 K resulting in lowerduration of regime-I at 873 K. The gradual decreasein ultrasonic velocity in regime-II is attributed to thecoarsening and consequent decrease in number densityof Cr2N precipitates. The reversal in trend in thevelocity in regime-III is attributed to the occurrenceof secondary precipitation in the steel (Fig. 16c),which also counter acts the effect of coarsening ofCr2N precipitates. The secondary precipitates (asindicated by needle in Fig. 16c) are identified asFe2Mo (Laves phase) with hexagonal structure. Withfurther ageing in this regime, both ultrasonic velocityand hardness increase due to the increased amount ofFe2Mo precipitation (Fig. 16d). The sharp reductionin the ultrasonic velocity and hardness observed inregime-IV at 873 K can be ascribed to excessiveageing at long durations resulting in coarsening ofprecipitates (Fig. 16e). The effect of ageingtemperature on precipitation kinetics is exhibited byshifts in the regimes of ultrasonic velocity to lowerageing time with increasing temperature. Further,the absence of regime-IV at lower ageing temperatureof 793 K could have resulted from the fact that thisregime of rapid decrease in the velocity at longdurations may occur at higher ageing time at 793 Kthan that used in this investigation. The possibility ofthe use of ultrasonic velocity to characterize thethermal ageing behaviour in 9Cr-1Mo ferritic steelwas also attempted in our centre a few years ago 34.

The variation in ultrasonic velocity with creepexposure at 873 K also exhibited a behaviour similarto that observed for thermally aged specimens(Fig. 14) and suggested that the creep damage in9Cr-1Mo ferritic steel is mainly due to themicrostructural degradation, and not due to cavitationdamage which is expected to decrease the velocitycontinuously with increasing amount of damage. Thisis in agreement with the detailed investigationperformed on the creep behaviour of 9Cr-1Mo steel25. Analysis of creep data in terms of creep rate-rupture life relationships indicated high creep damagetolerance factor 25, 35 and metallographic investigationexhibited absence of typical creep damage in the formof wedge cracks and r-type cavities in the steel. Thesteel retained its high creep ductility and the failuremode remained transgranular characterised by dimplesresulting from microvoid coalescence. All theseobservations suggested that 9Cr-1Mo steel is not proneto the typical creep damage and the creep failure

497


ACKNOWLEDGEMENTS

We are thankful to Dr. S. L. Mannan, AssociateDirector, MDG and Mr. P. Kalyanasundaram, Head,Div. for PIE and NDT Development, for theircooperation. We are also thankful to Prof. K. K.Ray, IIT Kharagpur for useful discussions.

REFERENCES1. Standard specification for seamless ferritic, austenitic alloy-

steel boiler, superheater, and heat exchanger tubes, ASTMstandards, A213, (1991).

2. Sikka V K, in Proc AIME Topical Conf Ferritic Alloysfor Use in Nuclear Energy Technologies, Snowbird,Utah (Eds.) Davies J W, and Michael D J, (1983) p317.

3. Nippon Steel Technical Report No. 50, July 1991.

4. Mathew M D, Choudhary B K, and Albert S K, Selectionof materials for PFBR nuclear steam supply systemcomponents, IGCAR Report - PFBR/MDG/2002/001,Kalpakkam.

5. Laha K, Chandravathi K S, Bhanu Sankara Rao K,Mannan, S L, Sastry D H, Trans Indian Inst Metals53 (2000) 217.

6. Nondestructive Testing Handbook, Second edition,Ultrasonic Testing, (Ed) Paul Mclntire, ASTM, USA,7 (1991) 836.

7. Shah B K, Anish Kumar, Bandyopadhyay M,Bandyopadhyay A K, and Kulkarni P G, Insight 41(11)(1999) pp. 707-709.

8. Mignogma R B, Duke J C Jr. and Green R E, MaterialsEvaluation 38 (1980) 37.

9. Grayeli N, and Shyne J C, in Review of Prog.Quantitative NDE, Plenum Press, NY, (Eds)Thompson R B, and Chimenti D E, 4b (1985) p 927.

10. Papadakis E P, Metall Trans A, 1 (1970) pp. 1053-1057.

11. Anish Kumar, Shankar, Vani, Bhanu Sankara Rao K,Jayakumar T, and Baldev Raj Philos. Mag. A, 82(2002) 2529.

12. Rosen M, Ives L, Ridder S, Biancaniello F, andMehrabian R, Mat. Sci. & Eng., 74 (1985) 1.

13. Klinman R, and Stephencon E T, Materials Evaluation,39 (1981) 1116.

14. Vary A, Materials Evaluation 46 (1988), 642.

15. Klinman R, Webster G R, Marsh F J, and StephensonE T, Materials Evaluation 38 (1980) 26.

16. Kutty T R G, Cndrasekharan K N, Pannakal J P,Ghoshal S K, and De P K, NDT & E Intern 20 (1987)359.

17. Dubois M, Moreau A, Militzerand M, and Bussiere JF, Scripta Mater 39 (1999), 333.

18. Fouquet F, Merle P, Kohen M, Merlin J, and GobinP F, Acta Metall 27 (1979) 315.

19. Crook M J, and Jordan G R, NDT&E International 23(1990) 221.

20. Wang W, Rokhlin S I, Lippold J C, and Adler L,Materials Evaluation 47 (1989), 1388.

21. Fitting D W, and Adler L, Ultrasonic Spectral Analysisfor NDE (1981) Plenum Press, New York.

22. Anish Kumar, Jayakumar T, Palanichamy P, and BaldevRaj, Scripta Materialia 40 (1999) 333.

23. Honjoh K, Jpn J Appl Phys 33 (1994) 1554.

24. Gericke O R, Paper number 1 in The Future ofUltrasonic Spectroscopy, edited by P.M. Reynolds,British Non-Ferrous Metals Research Association,London (1971).

25. Choudhary B K, Saroja S, Bhanu Sankara Rao K, andMannan S L, Mater Metall Trans A30 (1999) 2825.

26. Orr J, Burton D, and Rasche C, �The Sensitivity ofMicrostructure and Mechanical Properties of Steel 91 ToInitial Heat Treatments�, in The Manufacture andProperties of Steel 91 for the Power Plant and ProcessIndustries, 5th November 1992, VDEh, Dusseldorf.

27. Palanichamy P, Josheph A, Jayakumar T, and BaldevRaj, NDT & E Intern 28 (1995) 179.

28. Papadakis E. P, Physical Acoustics: Principles andMethods, Academic Press, New York, 4b (1981) 269.

29. Simpson W A Jr, J Acoustical Society of America 56(1974) 1776.

30. Josheph A, Palanichamy P, Jayakumar T, and BaldevRaj, Sc. and Tech. of Welding and Joining 3(6) (1998)267.

31. Sanderson S J, Metal Sci., 12 (1978) 220.

32. Wall M, Edwards B C, Hudson J A, in Proc. SpecialistMeeting �Mechanical Properties of Structural MaterialsIncluding Environmental Effects, Chester, England (1983),Report IWGFR-49, 2 (1984) p 545.

33. Titchmarsh J M, Wall M, and Edwards B C, in Proc.Analytical Electron Microscopy, (Eds). Williams D W,and Joy D C, San Francisco Press, San Francisco,USA, (1984), p. 247.

34. Bhattacharya D K, Characterization of microstructure insteels by magnetic and ultrasonic techniques (Ph. D.Thesis), Indian Institute of Science, Bangalore. 1995.

35. Phaniraj C, Choudhary B K, Bhanu Sankara Rao K,and Baldev Raj, Scripta Materialia, 48 (2003) 1313.

36. Anish Kumar, Jayakumar T, Baldev Raj and Ray K K,Insight 45 (7) (2003) 484.

Documents

9Cr Steels