Prsd Ro KALVALA*，Jved AKRAMMno MISRADmodrm RAMACHANDRAN，Jnki Rm GABBITA

aUniversity of Utah，Salt Lake City，USA

bSSN College of Engineering，Kalavakkam，India

cIndian Institute of Technology Madras，Chennai，India

Low temperature friction stir welding of P91 steel

Prasad Rao KALVALAa，*，Javed AKRAMa，Mano MISRAa，Damodaram RAMACHANDRANb，Janaki Ram GABBITAc

aUniversity of Utah，Salt Lake City，USA

bSSN College of Engineering，Kalavakkam，India

cIndian Institute of Technology Madras，Chennai，India

Bead-on-plate friction stir welds were made on P91 alloy with low and high rotational speeds （100 and 1000 RPM）to study their effects on weld microstructural changes and impression creep behavior.Temperatures experienced by the stir zone were recorded at the weld tool tip.Different zones of welds were characterized for their microstructural changes，hardness and creep behavior （by impression creep tests）.The results were compared with submerged arc fusion weld.Studies revealed that the stir zone temperature with 100 RPM was well below Ac1temperature of P91 steel while it was above Ac3with 1000 RPM.The results suggest that the microstructural degradation in P91 welds can be controlled by low temperature friction stir welding technique.

Friction stir welding；P91；Low temperature；Martensite；Carbides；Impression creep

1.Introduction

Modif i ed 9Cr-1Mo steels （P91）have been widely used for thermal power plant applications in view of their excellent creep strength.However，their weld counterparts are found to prematurely fail in their heat affected zone （HAZ）generally known asType IV cracking ［1］.Type IV cracking is found to be located in the f i ne grained heat affected zone （FGHAZ）and is related to the lower creep strength of the FGHAZ compared to the base material ［2，3］.The degradation of martensite lath subgrains into equiaxed subgrains is regarded as one of the major factors in reducing the creep strength of the FGHAZ ［4］. This is attributed to the high temperature excursion of the FGHAZ region above Ac1/Ac3during welding.Due to short dwell times at the elevated temperature，carbides will not dissolve.This results in the formation of martensite （on quenching）with lean carbon and a degraded lath subgrain structure［4］.Analytical results ［5］showed that the stress triaxiality in the FGHAZ will accentuate Type IV cracking.

One of the major technical challenges is to avoid the formation of f i ne grains in the HAZ of P91 welds by limiting the peaktemperatures in the HAZ well belowAc1/Ac3（857/914 °C）［6］.It is not possible to avoid FGHAZ in welds fabricated by conventional fusion welding processes because the melting temperatures in the fusion zone reach temperatures above 1500 °C.If the temperature in the weld metal can be controlled well below Ac1（857 °C），then the HAZ will not experience temperatures above Ac1.Such control is possible in friction stir welding（FSW）where weld temperatures and microstructural changes can be controlled.In the current investigation，experiments were conducted to control the stir zone peak temperature well belowAc1by controlling the weld parameters.The main emphasis of this work was mainly on the microstructural characterization，substantiated by some preliminary impression creep results.Detailed studies on impression creep studies are underway as a follow-up investigation to this work.

2.Experimental

Fig.1.Friction stir welding of P91 in progress.

The chemical composition （wt%）of P91 sheet used is as follows:Cr-8.91；Mo-0.98；C-0.09；Mn-0.42；Si-0.31；V-0.21；Nb-0.07；Fe-rest.Bead-on-plate friction stir welds with argon gas shielding （Fig.1）were made on a 3 mm thick P91 steel（Normalized and tempered）.The welding experiments were performed at Mega Stir Technologies LLC，Provo.A convex scrolled shouldertooldesign madefrom a grade of polycrystalline cubic boron nitride （PCBN）weld tool was used with a small shoulder diameter and a tapered pin.The welding involved 3 stages: （1）Plunging stage:Tool rotational speed: 800 RPM；Plunge depth:2.5 mm with tool feed rate of 76 mm/ min in Z-direction.（2）Dwell stage:Plunge depth is increased from 2.5 mm to 2.8 mm in Z-direction.Once plunge depth reached 2.8 mm，the RPM was reduced to 100 from 800，the tool’s dwell time （with no tool movement）was set at 5 sec with 2500 N axial force.（3）Weld stage:After 5 sec dwell time，the tool started traversing with 100 RPM，55 mm/min traverse speed with 2500 N force.Friction stir welding experiment was repeated on a similar sheet of P91 with 1000 RPM rotational speed，keeping all other conditions same.The temperature of the stir zone was measured at the tip of the tool by inserting K-type thermocouple.For comparative purpose，a submerged arc weld （SAW）（3.75 kJ/mm heat input，post weld heat treated at 760 °C/16 hrs）was included in the study.The specimens were cut，polished and etched in the cross sectional direction of the weld to include stir （weld）zone，HAZ and base metal portions.The polished samples were etched using solution containing 1 g Picric acid+5 mL HCl and 100 mL Ethanol.Preliminary impression creep tests were conducted to assess the relative creep behavior of different welds.Flat specimens with dimensions 20 × 10 × 3 mm were used for impression creep test with the following test conditions:Test temperature: 650 °C；Indenter:Tungsten carbide，1 mm diameter，cylindrical，f l at bottom；Punching stress:280 MPa；Vacuum level:10-3Torr and test time:about 100 hrs （until attaining steady state）. During impression creep testing，the displacement （i.e.，depth of impression）was continuously monitored （at 10 minute intervals）as a function of test time.Testing was terminated after going well into the secondary creep regime.

3.Results and discussion

3.1.Temperature prof i les

Typical temperatures recorded with 100 and 1000 RPM are shown in Fig.2.It can be seen that the peak temperature at the tip of the tool was about 560 °C with 100 RPM and 775 °C with 1000 RPM.The temperature was more or less stable with welding time for 100 RPM weld whereas for 100 RPM weld，it increased gradually.

3.2.Microstructures

Fig.2.Temperature prof i les recorded with 100 and 1000 RPM.

SEM micrographs were taken for different zones of the welds.The base metal consisted of typical martensitic lath structure with carbides distributed along grain boundaries and laths （Fig.3）.When cooled from the austenizing temperature， P91 steel exhibits a lath martensitic structure with a high dislocation density.Post weld tempering results in two kinds of precipitates: （1）M23C6（M=Cr，F(xiàn)e，Mo）carbides located at prior austenite grain boundaries and at other（packet，block，and martensite lath）boundaries，and （2）f i nely dispersed MX-type（M=V，Nb and X=C，N）carbonitrides within laths ［7］.

3.3.FSW weld with 100 RPM

SEM microstructure of stir zone of weld made with 100 RPM （Fig.4（a））showed f i ne grains along with f i ne carbides distributed in the matrix.This can help impede dislocation movement and improve creep resistance within the stir zone.As the peak temperature experienced by the stir zone was well below theAc1（857 °C），it is expected that the carbides will be intact and will not dissolve into the matrix.Although M23C6particles were present at the prior austenite grain boundaries in the base metal，they were found to be fragmented and uniformly distributed in the matrix due to the severe plastic deformation of friction stir welding （Fig.4（a））.The carbide precipitates were too small to be analyzed using SEM-EDS.The tempered martensite structure was found to be preserved with lath features ref i ned.The MX precipitates are expected to be in the matrix undissolved as their dissolution temperature is above 1250 °C［8］.Stir zone microstructure of 100 RPM weld indicates that it is possible to preserve the lath martensite structure without the dissolution of carbides.

Fig.3.Base metal microstructure.

Fig.4.Friction stir weld-100 RPM microstructure.

HAZ microstructure of weld made with 100 RPM （Fig.4（b））was similar to the base metal microstructure （Fig.4（a））and the weld heat generated with 100 RPM appeared to have not caused signif i cant change to its microstructure.This was substantiated by the hardness and microstructural results.The HAZs immediately close to the stir zone （both on advance and retreating sides）did not show signif i cant change in hardness compared to unwelded base metal （Fig.5）.Similarly，the microstructure of these regions showed no signif i cant change compared to the base metal（Fig.6）.These results indicate that the temperatures experienced by the stir zone as well as HAZ of weld made with 100 RPM were below Ac1temperature.The temperature recorded at the tool tip also showed the same trend （Fig.2）.

Fig.5.Hardness prof i le of friction stir weld-100 RPM.

3.4.FSW weld with 1000 RPM

The SEM microstructure of stir zone of weld made with 1000 RPM （Fig.7（a））showed as-quenched martensitic needles. It is very clear from this micrograph that there are no carbide particles in the matrix.HAZ microstructure of weld made with 1000 RPM was found to be similar to the stir zone showing as-quenched martensitic structure with no carbides in the matrix （Fig.7（b））.

For the martensite to exhibit as-quenched lath structure with no carbides in the matrix，P91 should have experienced temperatures in the ausenite region （temperature above Ac3）.This indicates that the temperatures experienced by the stir zone as well as HAZ of the weld made with 1000 RPM were above Ac3temperature.The temperature recorded at the tool tip showed the same trend （Fig.2）for welds made with 1000 RPM.As the reported Ac3for P91 is 914 °C ［6］，it appears that the temperature difference between the tip of the tool and actual stir zone temperature could be approximately 200 °C.

Fig.6.Friction stir weld-100 RPM-Microstructure of HAZ/stir zone.

Fig.7.Friction stir weld-1000 RPM microstructure.

3.5.Submerged arc weld

Fig.8.Submerged arc weld-cross section.

In SAW P91 welds （Fig.8），the following weld zones were observed: （a）coarse grain HAZ （Fig.9（a））；（b）intercritical region；（c）f i ne grain HAZ （Fig.9（b））；and （d）over tempered region.Coarse grain HAZ formed at 0-1.0 mm away from the fusion line which experiences temperatures well above the Ac3temperature during fusion welding.In the coarse grain HAZ，most carbides dissolve completely ［9］.This removes obstacles for austenite grain growth and favors the formation of coarse austenite.Upon cooling，the coarse austenite forms readily into martensite.Fine grain HAZ forms at temperatures just above the Ac3temperature where the α → γ phase transformation is almost complete.At temperatures between 1000 °C and 1135 °C even the large M23C6carbides dissolve ［10］.MX particles do not dissolve in austenite as their dissolution temperatures are even higher.MX is stable even up to a temperature of 1200 °C ［11］.During the PWHT and creep，pronounced recovery and recrystallization of the matrix occurs in this zone resulting in equiaxed grains without a lath structure.Coarsening and agglomeration of M23C6during PWHT and creep were also suggested as main factors reducing the creep rupture strength of the FGHAZ leading to type IV fracture in weldments of P91 steel without W ［12，13］.No lath martensite structure was observed in the f i ne grain HAZ of fusion welds ［4］.The disappearance of a lath martensite structure was caused by the instability of the f i ne grain HAZ microstructure.During the α → γ phase transformation，M23C6carbide particles do not fully dissolve into austenite.This results in a relatively small amount of carbon atoms available for diffusion into the matrix and accordingly the martensite formed on cooling will be lean in carbon content and will be unstable.The lath subgrain structure with a high density of free dislocations plays an important role in enhancing the creep rupture strength of a martensitic steel ［4］. Therefore，the degradation of the lath subgrain structure into equiaxed subgrains （and the absence of lath martensite）can be regarded as one of the main factors reducing the creep strength in the FGHAZ signif i cantly leading to type IV cracking failures.

Fig.9.Submerged arc weld.

Fig.10.Impression creep graphs.

3.6.The impression creep studies

The depth of impression vs.time plots is shown in Fig.10:（a）base metal，（b）SAW f i ne grain HAZ，and （c）friction stir weld HAZ with 100 RPM.A linear f i t for the steady state part of the curve was made，from which the impression rate （which is the slope，dh/dt）was calculated.According to Sastry ［14］，the impression rate divided by the diameter of the indenter gives the creep rate.In the present case，since the diameter of the indenter is 1 mm，the impression rate is equal to the creep rate.The results showed that the f i ne grain HAZ of SAW sample showed higher impression creep rate （Fig.10（b））compared to the base metal（Fig.10（a））.The impression creep rate of HAZ portion of 100 RPM friction stir weld （Fig.10（c））was almost similar to that of the base metal（Fig.10（a））.The impression creep test for 100 RPM weld HAZ demonstrated that the low temperature FSW technique is useful in controlling the HAZ damage in P91 steel.Detailed studies are underway to study the impression creep behavior of other welds to understand the effect of weld parameters.The studies are also taken up on the effect of post weld tempering temperature and time on different welds.This approach can be applied for various other engineering alloys to control the weld metal as well as HAZ degradation which will be useful to improve their service life.

4.Conclusions

In summary，the microstructural degradation in the heat affected zone of P91 welds needs to be controlled in order to use these materials to the full designed life.Low temperature friction stir welding technique of P91 is proposed for this purpose.Temperature measurements，microstructural characterization and the impression creep tests conf i rmed the benef icial effects of the proposed technique.

Acknowledgement

We would like to thank Russell Steel，MegastirTechnologies LLC，Provo，Utah for making the welds and Murray Mahoney，Consultant，Midway，Utah for discussions.

［1］Lundin CD，Liu P，Cui Y.A literature review on characteristics of high temperature ferritic Cr-Mo steels and weldments，WRC bulletin 454. New York，NY，USA:Welding Research Council，Inc.；2000.

［2］Sakthivel T，Vasudevan M，Laha K，Parameswaran P，Chandravathi KS，PanneerSelvi S，et al.Creep rupture behavior of 9Cr-1.8W-0.5Mo-VNb（ASME grade 92）ferritic steel weld joint.Mater Sci Eng A 2014；591:111-20.

［3］Zhao L，Jing H，Xu L.Investigation on mechanism of type IV cracking in P92 steel at 650°C.J Mater Res 2011；26:934-43.

［4］Xue W，Qian-gang P，Yao-yao R，Wei S，Hui-qiang Z，Hong L. Microstructure and type IV cracking behavior of HAZ in P92 steel weldment.Mater Sci Eng A 2012；552:493-501.

［5］Zhao L，Jing H，Xu L，An J，Xiao G.Numerical investigation of factors affecting creep damage accumulation in ASME P92 steel welded joint. Mater Des 2012；34:566-75.

［6］Vuherer T，Dundˉer M，Milovic′LJ，Zrilic′M，Samard?ic′I.Microstructural investigation of the heat-affected zone of simulated welded joint of P91 steel.Metalurgija 2013；52:317-20.

［7］Jara DR.9-12% Crheatresistantsteels:alloy design，TEM characterisation of microstructure evolution and creep response at 650°C［Ph.D.thesis］.Bochum，Chile:Ruhr-Universit?t Bochum；2011.

［8］Suzuki K，Kumai S，Toda Y，Kushima H，Kimura K.Two-phase separation of primary MX carbonitride during tempering in creep resistant 9Cr1MoVNb steel.ISIJ Int 2003；43（7）:1089-94.

［9］Singh K，Jaipal Reddy G，Vidyasagar DV，Thyagarajan V.Creep and creep damage assessment in P91 weld joints，ECCC Creep Conference，12-14 September 2005，London，825-36.

［10］Halid J.Metallurgy and Creep properties of new 9-12%Cr steels.Steels Res 1996；67:369-74.

［11］Yoshino M，Mishima Y，Toda Y，Kushima H，Sawada K，Kimura K. Inf l uence of normalizing heat treatment on precipitation behaviour in modif i ed 9Cr-1Mo steel，ECCP creep conference，Creep&Fracture in HighTemperature Components，12-14 September 2005，London，153-64.

［12］Albert SK，Matsui M，Hongo H，Watanabe T，Kubo K，Tabuchi M.Creep rupture properties of HAZs of a high Cr steel simulated by a weld simulator.Int J Pressure Vessel Piping 2004；81:221-34.

［13］Otoguro Y，Matsubara M，Itoh I，Nakazawa T.Creep rupture strength of heat affected zone for 9Cr ferritic heat resisting steels.Nucl Eng Des 2000；196:51-61.

［14］Sastry DH.Impression creep technique-an overview.Mater Sci Eng A 2005；409:67-75.

Received 26 October 2015；accepted 10 November 2015 Available online 15 December 2015

*Corresponding author.Tel.:+14157228105.

E-mail addresses:jyothipr@gmail.com；prasad.kalvala@utah.edu （P.R. KALVALA）.

http://dx.doi.org/10.1016/j.dt.2015.11.003

2214-9147/? 2016 China Ordnance Society.Production and hosting by Elsevier B.V.This is an open access article under the CC BY-NC-ND license （http:// creativecommons.org/licenses/by-nc-nd/4.0/）.

? 2016 China Ordnance Society.Production and hosting by Elsevier B.V.This is an open access article under the CC BY-NC-ND license（http://creativecommons.org/licenses/by-nc-nd/4.0/）.