Available online at SciVerse ScienceDirect

J. Mater. Sci. Technol., 2014, 30(3), 217e222

A Comparative Study of Various Flow Instability Criteria in Processing Map

of Superalloy GH4742

Ge Zhou1), Hua Ding1)*, Furong Cao1), Beijiang Zhang2)

1) College of Materials and Metallurgy, Northeastern University, Shenyang 110819, China2) High Temperature Materials Division, Central Iron and Steel Research Institute, Beijing 100081, China

[Manuscript received September 9, 2012, in revised form November 27, 2012, Available online 18 July 2013]

* Corresp24 2390cn (H. D1005-03JournalLimited.http://dx

Hot compression tests were conducted on a Gleeble-1500D thermal simulating tester. Based on the deformationbehavior and microstructural evolution of superalloy GH4742, different types of instability criteria of Prasad,Gegel, Malas, Murty and Semiatin were compared, and the physical significance of parameters was analyzed.Meanwhile, the processing maps with different instability criteria were obtained. It was shown that instabilitydid not occur when average power dissipation rate was larger than 50% in the temperature range of 1020e1130 �C, corresponding to the strain rate range of 5 � 10�4e3.2 � 10�3 s�1. The domain is appropriate forthe processing deformation of superalloy GH4742.

KEY WORDS: GH4742 superalloy; Hot compression tests; Flow instability criterion; Processing map

1. Introduction

GH4742 is a typical g0-hardened nickel-based superalloy. Inthe GH4742 superalloy, total content of solution strengtheningelements such as the Co, Mo, Cr is 28.63% (wt%) and totalcontent of aging strengthening elements such as the Al, Ti, Nbis 7.63% (wt%). This alloy possesses high thermal strengtheningproperties and integrative properties, and has long-term workingstability and excellent service performance. However, the in-crease of strengthening elements will lead to the increase ofrecrystallization temperature. Very narrow available deformationtemperature range, high resistance to deformation and poorductility are the main obstacles that restrict hot workability ofthe GH4742 superalloy which usually is difficult in deforma-tion[1e4].Processing map is a powerful tool in the design and optimi-

zation of metallic forming process. Processing map not onlydescribes the deformation mechanisms of specific microstruc-tures in deterministic regions but also describes the instabilityflow regions that should be avoided during forming process. Inthe meantime, optimized forming temperature and strain rates

onding author. Prof., Ph.D.; Tel.: þ86 24 83687746; Fax: þ866316; E-mail addresses: hding@263.net, dingh@smm.neu.edu.ing).02/$e see front matter Copyright� 2013, The editorial office ofof Materials Science & Technology. Published by ElsevierAll rights reserved..doi.org/10.1016/j.jmst.2013.07.008

can be obtained by using the processing map. Therefore, theprocessing maps have been used in more than 200 alloys[5e10].In the present work, based on the concepts of dynamic materialmodel (DMM) proposed by Prasad[11], Gegel[12], Malas[13],Murty[14] and Semiatin[15], of which the Prasad’s criterion isderived carefully by the maximum entropy generation rateprinciple and big plastic deformation. However, when this cri-terion is put into application, it is needed that the establishedconstitutive equation during high temperature compressiondeformation meets the three-order derivability of the strain rate _3.Malas’s and Gegel’s criterion is derived on the basis of ther-modynamics theorem and its theoretical basis is strict. However,as compared with the requirements of Gegel’s criterion in whichthe strain rate sensitivity exponent (m) value is a constantparameter, Malas’s criterion needs not considering m value as aconstant. Entire derivation process of Murty’s criterion does notinvolve the problem that whether or not m value is a constant, theapplication scope of this criterion is the most expansive. How-ever, during calculating power dissipation rate (h) value, thedefinition formula of h value must be used to solve and thecalculation process is cockamamie. Semiatin’s criterion is anempirical formula derived on the basis of microstructure obser-vation of titanium and its alloys. But as compared with otherseveral criterions, this criterion does not take strict theoreticalderivation as the basis.In the present study, a comparison of five plastic instability

criteria (including Prasad, Murty, Gegel, Malas and Semiatin)was carried out based on the hot compression experiment data ofGH4742 superalloy which have important significances for the

218 G. Zhou et al.: J. Mater. Sci. Technol., 2014, 30(3), 217e222

determination and optimization of process parameters duringpractical production and processing of GH4742 superalloy.

2. Experimental

The chemical composition (wt%) of the GH4742 superalloy inthis investigation was as follows: 0.053 C, 13.83 Cr, 9.96 Co,4.84 Mo, 2.67 Al, 2.52 Ti, 2.44 Nb, 0.0036 B, 0.005 V, 0.20 Fe,0.002 La, 0.0038 Ce, 0.001 S, 0.005 P and balanced Ni. Hotcompression tests were performed on a Gleeble-1500D thermalsimulating tester for GH4742 superalloy specimen whosedimension is F8 mm � 12 mm. The test temperature was in therange from 900 to 1150 �C with a temperature interval of 50 �C.The strain rate was in the range from 5 � 10�4e10 s�1. Theheating rate was 10 �C/s and the holding time was 3 min, andthen the specimens were deformed to a true strain of 0.6 and thespecimens were immediately water-cooled to room temperatureafter deformation ends.The deformed specimens were sectioned parallel to the

compression axis and the cutting surfaces were prepared formicrostructure observation. The metallographic specimens wereetched by using a solution of CuSO4 (1.5 g) þ HCl(40 ml) þ C2H5OH (20 ml). The microstructures of the speci-mens were investigated by optical microscopy (OM) withOLYMPUS GX51 microscope.

3. Result and Discussion

Fig. 1 Power dissipation rate (h) map at different strain rates ð_3Þ andtemperatures with a strain of 0.6.

3.1. Processing map theory based on DMM

According to the theory of dissipative structures, Prasadet al.[16] believed that the energy of input system, P, can bedivided into dissipative magnitude, G, and dissipative coordi-nation magnitude, J. Its mathematical definition is as follows:

P ¼ s_3¼ Gþ J ¼Z _3

0

sd_3þZs

0

_3ds (1)

where G is the energy consumed by the plastic deformation ofmaterials among which a majority of energy is turned intothermal energy and small amount of energy is stored in crystaldefect energy. J is the energy dissipated during the microstructureevolution of material deformation. The proportion of two energiesis determined by the strain rate sensitivity exponent, m, offorming component under definite stress:

m ¼ vJ=vG ¼ _3vs=sv_3¼ vlns=vln_3 (2)

The physical meaning of partition rate of system energy can beelucidated clearly from the viewpoint of atomic movement. Thedissipation of material energy can be divided into potential en-ergy and kinetic energy. Potential energy is related to the relativepositions among atoms. The variation of microstructure willresult in the variation of atomic potential energy and hencecorresponds to the dissipative coordination magnitude, J. Kineticenergy is related to the movement of atoms, i.e., the movementof dislocations. Kinetic energy conversion is dissipated in theform of thermal energy and hence corresponds to dissipativemagnitude, G. Differential calculus of dissipative coordinationmagnitude J is expressed as:

dJ ¼ _3ds (3)

Presuming that material conforms to constitutive relation:

s ¼ C _3 (4)

Then, J is expressed as:

J ¼Z _3

0

sd_3¼ m=ðmþ 1Þs_3 (5)

When m ¼ 1, material is in an ideal linear dissipation state.Dissipative coordination magnitude J reaches the maximumvalue Jmax, i.e.

Jmax ¼ s_3=2 (6)

A dimensionless parameter value h which is the powerdissipation rate can be obtained by formulas (5) and (6). Itsphysical meaning is to elucidate the proportion relation of energydissipated by microstructure evolution to linear dissipation en-ergy during material forming. Its value is:

h ¼ J=Jmax¼ 2m=ðmþ 1Þ (7)

The flow stresses at different strain rates and temperatureswith a strain of 0.6 were obtained by thermal simulationcompression test and the power dissipation rate map was ob-tained, as shown in Fig. 1.It can be seen from Fig. 1 that when hot forming is performed

in region I (temperature of 1080e1150 �C and strain rate of5 � 10�4e10�3 s�1), region II (temperature of 1020e1080 �Cand strain rate of 5 � 10�4e3.2 � 10�3 s�1), the mean value ofpower dissipation rate of GH4742 superalloy is 54%. Themaximum value is 56%.

3.2. Analysis and application of different instability criterions forGH4742 superalloy

3.2.1. Gegel’s instability criterion. Based on the second law ofthermodynamics, Gegel et al.[11] found that flow instability isrelated to the temperature sensitivity parameter, S. The definitionof S is as follows:

Fig. 2 Processing map of Gegel’s instability criterion for GH4742superalloy.

Fig. 3 Processing map of Malas’s instability criterion for GH4742superalloy.

Fig. 4 Processing map of Prasad’s instability criterion for GH4742superalloy.

G. Zhou et al.: J. Mater. Sci. Technol., 2014, 30(3), 217e222 219

S ¼�1.T

�0B@vlns=v

�1.T

�1CA ¼ �vlns=vlnT (8)

Meanwhile, Gegel used Lyaponov function L(h,s) andbelieved that the flow stress curve is up-convex when stable flowoccurs in the material. Flow stress decreases with increasingtemperature. h decreases with increasing _3. S decreases withincreasing _3.

vS=vln_3¼ �vðvlnsÞ=ðvlnT$vln_3Þ¼ �vm=vlnT>00vm=vlnT < 0 ð9Þ

Thus, Gegel’s instability criterion becomes:

vh=vln_3>0; vm=vlnT < 0 (10)

The processing map on Gegel’s instability criterion of theGH4742 superalloy generated at the strain of 0.6 is shown inFig. 2. The map is clearly classified into two domains, e.g. thestability deformation domain and instability deformationdomain, the latter being the shaded part in Fig. 2.

3.2.2. Malas’s instability criterion. When Malas et al.[13]

investigated Tie49.5Ale2.5Nbe1.1Mn alloy, they used Lya-ponov function L(h,s) and meantime replaced h with m. Theyproposed Malas’s instability criterion on the basis of Gegel’scriterion:

vm=vln_3>0; vm=vlnT < 0 (11)

The processing map on Malas’s instability criterion of theGH4742 superalloy generated at the strain of 0.6 is shown inFig. 3.

3.2.3. Prasad’s instability criterion. At present, when DMMmethod is used to solve hot working problem, Gegel’s andMalas’s instability criterions have been rarely used. Manyscholars used the instability criterion established by Prasad[17].This kind of criterion takes the extremum principle of irrevers-ible thermodynamics of big plastic flow and satisfies thefollowing relation when flow instability occurs:

vD=vR < D=R (12)

As dissipation coordination magnitude is relevant to themicrostructure evolution of metallurgical process, Prasadreplaced D with J and got:

vJ=v_3< J=_3 (13)

vlnJ=vln_3< 1 (14)

Take logarithm on both sides of formula (5) and seek localderivation for ln_3, one gets:

vlnJ=vln_3¼ vlnðm=ðmþ 1ÞÞ=vln_3þ vlns=vln_3þ 1 (15)

Integrating formulas (5), (14) and (15), one gets Prasad’sinstability criterion:

xð_3Þ ¼ vlnðm=ðmþ 1ÞÞ=vln_3þ m < 0 (16)

The processing map on Prasad’s instability criterion of theGH4742 superalloy generated at the strain of 0.6 is shown inFig. 4.

3.2.4. Murty’s instability criterion. It is believed that m valuein formula (4) of Prasad’s criterion is constant. But Indianscholar Murty et al.[18e23] and Italian scholar Spigarelli et al.[24]

believed that for pure metal and alloy with low alloying ele-ments, m could be considered simply a constant value; while m isnot constant for complex alloy system. Based on this situation,

Fig. 6 Processing map of Semiatin’s instability criterion for GH4742superalloy.

220 G. Zhou et al.: J. Mater. Sci. Technol., 2014, 30(3), 217e222

Murty et al.[18] derived an instability region criterion that issuitable for any stress and strain rate curve. According to thedefinition of J and h:

J ¼Zs

0

_3ds0vJ=v_3¼ _3vs=v_3¼ svlns=vln_3¼ ms (17)

h ¼ J=Jmax ¼ 2J=ðs_3Þ0J=_3¼ hs=2 (18)

According to formula (13), one gets Murty’s instability cri-terion:

2m < h (19)

The processing map on Murty’s instability criterion of theGH4742 superalloy generated at the strain of 0.6 is shown inFig. 5.

3.2.5. Semiatin’s instability criterion. Semiatin and Jonas[15]

put forwards the relationship of flow softening with materialparameter a, a ¼�g/m, where g is flow softening rate g¼ vlns/v 3. The Semiatin’s flow localization criterion becomes:

a > 5 (20)

The processing map on Semiatin’s instability criterion of theGH4742 superalloy generated at the strain of 0.6 is shown inFig. 6.

3.3. Analysis of the hot forming properties for GH4742superalloy

The processing map on different instability criterions of theGH4742 superalloy generated at the strain of 0.6 is shown inFig. 7.I, II, III, IV and V five regions in Fig. 7 represent the flow

instability domains under Gegel, Malas, Prasad, Murty andSemiatin five different instability criterions, respectively. It canbe seen from Fig. 7 that the superposition of five flow instabilitycriterions appears in upper left corner and lower right cornerdomains which are in the temperature range from 900 to 950 �Cover a strain rate range from 1 to 10 s�1; and in the temperaturerange from 1130 to 1150 �C over a strain rate range from

Fig. 5 Processing map of Murty’s instability criterion for GH4742superalloy.

5 � 10�4 to 1.8 � 10�2 s�1, indicating that working instabilityphenomenon appears in above forming domains and should beavoided during practical forming process. Flow instability phe-nomenon probably appears in the instability domains with lesssuperposition of flow instability criterions. The domains withhigh power dissipation rate and no flow instability are appro-priate for the forming.It can be seen from Fig. 7 that the instability domains obtained

by different instability criterions are different. Intersection andsupplement occur among the instability domains. Therefore,when hot forming workability of GH4742 superalloy isanalyzed, different instability criterions, had better be consideredintegratively. The instability domains of hot forming can bejudged correctively by using integrative criterion.Gegel’s criterion is derived on the basis of thermodynamics

theorem and its theoretical basis is strict. Malas’s criterion isobtained by replacing h with m on the basis of Gegel’s criterion.It can be seen from Fig. 7 that the instability regions of these twoinstability criterion are basically the same. However, ascompared with the requirements of Gegel’s criterion that m valueis a constant parameter, Malas’s criterion needs not consideringm value as a constant. Therefore, Malas’s criterion is moreexpansive than Gegel’s criterion. Prasad’s criterion is derivedcarefully by the maximum entropy generation rate principle andbig plastic deformation, as compared with other criterions, thescope of instability region is the smallest, as shown in Fig. 7. Asentire derivation process of Murty’s criterion does not involve

Fig. 7 Instability maps through different instability criterions forGH4742 superalloy: IdGegel; IIdMalas; IIIdPrasad;IVdMurty; VdSemiatin.

Fig. 8 Microstructures of deformed specimens in different domains: (a) _3¼ 10 s�1, T ¼ 900 �C; (b) _3¼ 5 � 10�3 s�1, T ¼ 1150 �C; (c)_3¼ 5 � 10�2 s�1, T ¼ 1050 �C.

G. Zhou et al.: J. Mater. Sci. Technol., 2014, 30(3), 217e222 221

the problem that whether or not m value is a constant, theapplication scope of this criterion is the most expansive. How-ever, during calculating h value, the definition formula of h valuemust be used for solving and the calculation process is cocka-mamie. Semiatin’s criterion is an empirical formula derived onthe basis of microstructure observation of titanium and its alloys.But as compared with other several criterions, this criterion doesnot take strict theoretical derivation as the basis, thus its scope ofapplication is restricted greatly.

3.4. Microstructures of GH4742 alloy during hot deformation

The microstructural evolution during deformation of GH4742superalloy is shown in Fig. 8. Fig. 8(a) shows the metallographat a temperature of 900 �C and a strain rate of 10 s�1. It can beseen that obvious cracks appear between g and g0 phase, indi-cating that flow instability will appear at cracks during defor-mation. Fig. 8(b) shows the metallograph at a temperature of1150 �C and at a strain rate of 5 � 10�3 s�1. It can be seen thatcracks appear at triple junction of grain boundaries. The reason isthat when deformation temperature is high and strain rate is low,it is difficult for grains at grain boundary to slide during defor-mation. Thus stress concentration is formed at triple junction ofgrain boundaries. Dynamic recrystallization cannot proceed fullyat 900 �C and hence the stress concentration formed at triplejunction of grain boundary cannot be relaxed under high strainrate. Therefore, cracks at triple junction of grain boundary appearat high strain rate and hence flow instability appears. This phe-nomenon corresponds to the upper left corner and lower rightcorner in Fig. 7 which is the superposition of five flow instabilitycriterions. Fig. 8(c) shows the metallograph at a temperature of1050 �C and at a strain rate of 5 � 10�2 s�1. It can be seen thatfine equiaxed grains exist and dynamic recrystallization occurs in

Fig. 9 Variation of flow stress (s) with rate ð_3Þ at different temperaturesfor GH4742 superalloy.

this condition. The progression of dynamic recrystallization re-laxes the stress concentration effectively at triple junction ofgrain boundary and adiabatic shear crack does not appear. Themicrostructure in Fig. 8(c) corresponds to the region (tempera-ture of 1020e1130 �C and strain rate of 5 � 10�4e3.2 � 10�3 s�1) and the peak value of power dissipation rate ofGH4742 superalloy in Fig. 7. Thus GH4742 alloy has excellenthot forming performance under this deformation condition.

3.5. Kinetic analysis of GH4742 superalloy during hotdeformation

During hot deformation, it is well accepted that the relation-ship between the steady-state flow stress (s), strain rate (_3) andtemperature (T) is generally expressed in the form of an Arrhe-nius type rate equation:

_3¼ Asnexp

��Q

RT

�(21)

where A is a constant, s is the flow stress, _3is the strain rate, n isthe hardening exponent (n ¼ 1/m, m is the strain rate sensitivity),Q is the deformation activation energy, R is the gas constant, T isthe deformation temperature. On the basis of the stress exponentand activation energy values, the deformation mechanisms formicrostructure development during deformation are identified.For this model, steady-state flow stress (s) data at differenttemperatures (T) and strain rates (_3) at a true strain of 0.6 havebeen used. A plot of lns vs ln_3 for different temperatures isshown in Fig. 9. The value of n is the inverse of the slope ofthe line. The Arrhenius plot obtained between lns and 1/T fordifferent strain rates is shown in Fig. 10. The slope of thecurves and the corresponding activation energy values were

Fig. 10 Arrhenius plot showing the variation of flow stress (s) withtemperature at different strain rates for GH4742 superalloy.

Table 1 Values of activation energies (Q) during high deformation ofGH4742 superalloy

Strain,3

Strainrate,_3(s�1)

900 �C 950 �C 1000 �C 1050 �C 1100 �C 1150 �C

0.6 5 � 10�4 812.94 609.40 580.15 554.56 530.18 564.310.6 10�3 781.63 585.93 557.80 533.20 509.76 542.570.6 5 � 10�3 778.76 583.78 555.76 531.24 507.89 540.580.6 5 � 10�2 771.25 578.15 550.40 526.11 502.99 535.370.6 10�1 732.94 549.43 523.05 499.98 478.00 508.770.6 1 668.75 501.31 477.45 456.19 436.14 464.220.6 10 533.93 400.24 381.03 364.22 348.21 370.63

222 G. Zhou et al.: J. Mater. Sci. Technol., 2014, 30(3), 217e222

obtained from the figure and the activation energy are reported inTable 1. Based on these results, when hot forming is performedin region I (temperature of 1080e1125 �C and strain rate of5 � 10�4e10�3 s�1) and region II (temperature of 1020e1080 �C and strain rate of 5 � 10�4e3.2 � 10�3 s�1), themean value of deformation activation energy for GH4742superalloy is lesser, thus the alloy has excellent formingperformance. The superposition of flow instability criterions,the mean value of deformation activation energy is bigger,thus the alloy has poor forming performance. The resultcorresponds to comparative study of various flow instabilitycriteria in processing map of GH4742 superalloy in Fig. 7.

4. Conclusions

(1) According to power dissipation rate maps integratingseveral different instability criterions, appropriate formingdomains of GH4742 superalloy are in the temperature rangeof 950e975 �C and strain rate range of 5 � 10�4e10�1 s�2,in the temperature range of 975e1125 �C and strain raterange of 5 � 10�4e10 s�1 and in the temperature rangeof 1125e1150 �C and strain rate range of 1e10 s�1. Average power dissipation rate in the domain (temperaturerange of 1020e1130 �C corresponding to the strain raterange of 5 � 10�4e3.2 � 10�3 s�1) is larger than 50%.

(2) Flow instability phenomenon appears in low temperatureand high strain rate domain and high temperature and lowstrain rate domain for the forming of GH4742. The reason isthat adiabatic shear crack appears due to the stress concentration at triple junction of grain boundaries and excessivelyhigh temperature. Dynamic recrystallization takes place inappropriate domains and relaxes the stress concentration attriple junction of grain boundaries. Thus the alloy has excellent forming performance in this region.

(3) Analysis and comparison of different instability criterionsof GH4742 superalloy reveal that the instability domainsof this alloy under different instability criterions are diffe-rent. Intersection and supplement occur among the

instability domains. Based on the analysis and comparisonfor above five types of instability criterions, computationalprocess, and cope of application, it is recommended to useMURTY instability criterion.

AcknowledgmentThis work was supported by the National Key Basic

Research Program of China (No. 2010CB631203).

REFERENCES

[1] X.D. Lu, Q. Deng, J.H. Du, J.L. Qu, J.Y. Zhuang, Z.Y. Zhong, J.Alloy. Compd. 477 (2009) 100e103.

[2] X.D. Lu, J.H. Du, Q. Deng, Z.Y. Zhong, J. Alloy. Compd. 486(2009) 195e198.

[3] R.J. Mitchell, M. Preuss, M.C. Hardy, S. Tin, Mater. Sci. Eng. A423 (2006) 282e291.

[4] H. Monajati, F. Zarandi, M. Jahazi, S. Yue, Scripta Mater. 52 (2005)771e776.

[5] Y. Sun, W.D. Zeng, Y.Q. Zhao, X.M. Zhang, X. Ma, Y.F. Han,Trans. Nonferrous Met. Soc. China 21 (2011) 159e169.

[6] S.H. Huang, Z.D. Zhao, Z.X. Xia, H.Y. Cai, F. Kang, C.K. Hu,D.Y. Shu, Rare Metal Mat. Eng. 39 (2010) 848e852 (in Chinese).

[7] L.M. Yan, J. Shen, Z.B. Li, J.P. Li, X.D. Yan, B.P. Mao, Trans.Nonferrous Met. Soc. China 20 (2010) 1296e1301.

[8] L.Y. Zeng, G.J. Yang, P. Ge, X.N. Mao, Y.Q. Zhao, L. Zhou, RareMetal Mat. Eng. 39 (2010) 1505e1508 (in Chinese).

[9] L. Li, D.J. Song, Trans. Nonferrous Met. Soc. China 20 (2010)s738es742.

[10] F.T. Kong, S.Z. Zhang, Y.Y. Chen, Trans. Nonferrous Met. Soc.China 20 (2010) s233es236.

[11] S.M. Doraivelu, H.L. Gegel, J.S. Gunasekera, J.C. Malas, J.T.Morgan, J.F. Thomas Jr., Int. J. Mech. Sci. 26 (1984) 527e535.

[12] H.L. Gegel, E.W. Collings, Scripta Mater. 7 (1973) 437e443.[13] J.C. Malas, V. Seetharaman, JOM 44 (1992) 8e13.[14] S.V.S. Murty, R.B. Nageswara, B. Kashyap, B.P. Kashyap, Int.

Mater. Rev. 45 (2000) 15e26.[15] S.L. Semiatin, V. Seetharaman, I. Weiss, Mater. Sci. Eng. A 243

(1998) 1e24.[16] K.P. Rao, Y.V.R.K. Prasad, K. Sivaram, Mater. Lett. 10 (1990)

66e70.[17] Y.V.R.K. Prasad, T. Seshachryulu, Int. Mater. Rev. 43 (1988)

243e258.[18] S.V.S.N. Murty, B.N. Rao, Mater. Sci. Eng. A 15 (1998) 76e82.[19] S.V.S.N. Murty, B.N. Rao, J. Mater. Sci. Lett. 21 (1999)

1757e1758.[20] S.V.S.N. Murty, B.N. Rao, J. Mater. Process. Technol. 104 (2000)

103e109.[21] S.V.S.N. Murty, B.N. Rao, Scand. J. Metall. 29 (2000) 148e150.[22] S.V.S.N. Murty, B.N. Rao, B.P. Kashyap, Compos. Sci. Technol. 63

(2003) 119e135.[23] S.V.S.N. Murty, B.N. Rao, B.P. Kashyap, J. Mater. Process. Tech-

nol. 166 (2005) 268e278.[24] S. Spigarelli, E. Cerri, P. Cavaliere, E. Evangelista, Mater. Sci. Eng.

A 327 (2002) 144e154.