On the fatigue behavior of an AISI 316L stainless steel coated with a PVD TiN deposit

Surface and Coatings Technology 182(2004) 276–286

0257-8972/04/$ - see front matter� 2003 Elsevier B.V. All rights reserved.doi:10.1016/j.surfcoat.2003.07.003

On the fatigue behavior of an AISI 316L stainless steel coated with aPVD TiN deposit

E.S. Puchi-Cabrera *, F. Matınez , I. Herrera , J.A. Berrıos , S. Dixit , D. Bhata, a a a b c´ ´

School of Metallurgical Engineering and Materials Science, Faculty of Engineering, Universidad Central de Venezuela,a

Apartado Postal 47885, Los Chaguaramos, Caracas 1045, VenezuelaUES, Dayton, OH 45432-1894, USAb

Department of Mechanical Engineering, University of Arkansas, Fayetteville, AR 72701, USAc

Received 25 April 2003; accepted in revised form 31 July 2003

Abstract

The effect of a TiN coating on the fatigue properties of an AISI 316L stainless steel has been investigated. The coating wasapproximately 1.4-mm thick and was deposited by means of filtered cathodic arc deposition. It has been determined that theapplication of such a coating to the steel substrate gives rise to a significant increase in both fatigue life and fatigue limit, incomparison with the uncoated steel. The increase in fatigue life has been quantified in terms of the computed values of theBasquin parameters of the different materials tested. Thus, it has been shown that, depending on the maximum alternating stressapplied to the material, the fatigue life of the steel can be increased between 400 and 2119%, whereas the fatigue limit wasobserved to increase by 22%, that is to say, 60 MPa above the fatigue limit of the uncoated substrate, which in the present workwas found to be of approximately 381 MPa. From the microscopic point of view, it has been observed that the coating remainswell adhered to the substrate both in tension and during fatigue testing at low maximum alternating stresses(480 MPa). However,during fatigue testing at elevated maximum alternating stresses(510 MPa) the coating was observed to delaminate from thesubstrate. Also, it has been determined that the fatigue fracture of the substrate-coating composite is dominated by the fracture ofthe TiN coating since fatigue cracks have been observed to form first at the surface of the coating and subsequently to propagatetowards the substrate. It has been concluded that the increase in fatigue properties of the coated substrate is associated mainlywith the compressive residual stresses present in the coating and to the good adhesion of the coating to the substrate observed inmost of the maximum alternating stress range explored in this work.� 2003 Elsevier B.V. All rights reserved.

Keywords: Fatigue properties; AISI 316L stainless steel; TiN coating; Filtered cathodic arc deposition

1. Introduction

It is well known that, in general, stainless steels arecharacterized as having relatively poor wear and gallingresistance, although they offer an excellent generalcorrosion resistance in the atmosphere and in manyaqueous media and oxidizing acids including phosphor-ic, acetic and sulfurous acids, as well as chloridesolutions w1x. Therefore, in order to improve theirtribological performance without compromising their

*Corresponding author. Tel.:q58-212-6628927; fax:q58-212-7539017.

E-mail addresses: [email protected](E.S. Puchi-Cabrera),[email protected](S. Dixit), [email protected](D. Bhat).

corrosion resistance, a number of surface treatmentshave been developed, which include the application ofthin hard coatings of different oxides, nitrides andcarbides on the surface of the parts and components ofinterest. Although the deposition of such coatings willimprove the wear and galling resistance of the coatedmaterial, the question arises as to what extent the fatigueproperties of the coating–substrate system are compro-mised by the presence of the coating when the compo-nent is subjected to cyclic loading and what would bethe fracture mechanism.In a previous investigation conducted by Berrıos et´

al. w2x, the effect of different sub stoichiometric TiNx

PVD coatings, specifically TiN , TiN and TiN ,0.55 0.65 0.75

277E.S. Puchi-Cabrera et al. / Surface and Coatings Technology 182 (2004) 276–286

on the fatigue properties of a 316L stainless steel wasinvestigated. Such coatings were in the range of approx-imately 3–3.60mm in thickness and were depositedindustrially by means of PVD closed field unbalancedmagnetron sputtering(UMS). In relation to this work,it was determined that the application of such coatingsto the steel substrate gave rise to a significant increasein the yield strength of the composite material, evaluatedin tension and also of both fatigue life and fatigue limit,in comparison with the uncoated steel. The increase infatigue life was quantified in terms of the computedvalues of the Basquin parameters of the different mate-rials tested. It was also shown that, depending on thecomposition of the coating, the fatigue life of the steelcould be increased by between 566 and 1677% and thefatigue limit could also be increased by between 9.1and 10.8%. Such increase in fatigue properties for thecoated material was explained from the microscopicpoint of view, in terms of the observation that thecoating remained well adhered to the substrate eitherduring tensile and fatigue testing, and that the fatiguefracture of the substrate–coating composite was domi-nated by the fracture of the TiN coatings, since fatiguex

cracks were observed to form first within the coatingand subsequently to propagate towards the substrate.This work allowed the conclusion that the increase infatigue properties of the coated substrate was associatedwith the intrinsic higher mechanical properties of thedeposits with respect to those of the substrate and to theapparent good adhesion of the deposits.The results reported by Berrıos et al.w2x are partially´

consistent with those described earlier by Parameswaranet al. w3x who carried out a study in order to evaluatethe fatigue properties under rotating bending conditions,of a 17-4 PH stainless steel coated with TiN, depositedby reactive ion plating(PVD), under various combina-tions of electron beam current, bias voltage, filamentcurrent and nitrogen flow. Parameswaran et al.w3x foundout, among other things, that the fatigue life of thecoated 17-4 PH stainless steel was comparable with orbetter than those of the uncoated substrate, and thatduring fatigue testing, the coatings cracked in a brittlemanner.Also, Herr et al.w4x had analyzed the fatigue perform-

ance and tribological properties of a SAE 52100 steelsubstrate coated with TiN coatings, deposited by r.f.magnetron sputtering, under different conditions of biasvoltage and deposition time. By conducting fatigue testsunder alternating bending conditions and determiningthe fatigue limit by means of the staircase methodw5x,they concluded that neither the variation of the biasvoltage, nor the thickness of the coating have a remark-able influence on the fatigue strength of the samplesand that the fatigue strength of the coated substratecould show an increase of the order of 10% in compar-ison with the uncoated substrate.

The results reported by Berrıos et al.w2x also agree´partially with those published by Ferreira et al.w6x whoinvestigated the fatigue behavior of a 42Cr 4Mo steelcoated by PVD with W, WN, WTi and WTiN coatings.These researches concluded that the fatigue life of thecoated specimens only increases in the region of lowstress amplitudes and for the WTi coating. The micro-scopic analysis conducted on several coated samplesallowed these researches the conclusion that the coatingsare too brittle to accommodate the substrate metaldeformation, which leads to the fracture of the filmduring the early stages of the fatigue process. Thus, alarge number of cracks are formed on the surface of thecoating, which grow into the substrate or run along thecoating–substrate interface.Thus, the present investigation has been conducted in

order to evaluate the change in fatigue properties,including fatigue strength and limit, of a 316L stainlesssteel coated also with a TiN coating deposited by filteredcathodic arc deposition(FCA), in comparison with theproperties of the uncoated substrate. Also, we haveaimed to study more closely the fracture process of thecoated materials and compare it with that reported byFerreira et al.w6x and Berrıos et al.w2x. It is believed´that FCA deposition avoids the formation of largemacro-particles, which are present when the metal plas-ma flux is not filtered effectively and, therefore, it is ofinterest to determine whether the absence of such par-ticles has any influence on the fatigue properties andfatigue fracture mechanisms of the coated material.

2. Experimental techniques

The substrate material employed in this investigationwas a 316L stainless steel with the following composi-tion (wt.%): 0.022 C, 0.36 Si, 1.50 Mn, 0.032 P, 0.028S, 16.80 Cr, 2.17 Mo, 11.25 Ni and Fe bal. Among thewide range of engineering applications of this alloy arecomponents for chemical plants, petrochemical refinersand for fossil fuel power stations, parts of nuclearreactors, equipment for critical applications at cryogenictemperatures and biomedical implants for the treatmentof human fractures.All the specimens employed in the present study were

obtained from bars of approximately 12.7-mm diameterand 6-m length from which 6 tensile and 68 fatiguespecimens were machined following the ASTM A 370and ASTM 606 standards. Also, a number of smallcylindrical samples of approximately 10 mm weremachined for characterizing the chemical compositionof the coating, residual stresses and absolute hardness.The diagrams of both the tensile and fatigue samplesare shown in Fig. 1 in which all dimensions are givenin mm.The specimens were machined carefully in order to

minimize the introduction of residual stresses during the

278 E.S. Puchi-Cabrera et al. / Surface and Coatings Technology 182 (2004) 276–286

Fig. 1. Diagrams of the tensile and fatigue specimens employed inthis study. All the dimensions are in mm.

operation, by means of a continuous reduction of thedepth of cut of the material. A horizontal turret lath atlow speeds was employed in order to carry out theturning operation. The elimination of the circumferentialnotches and the final mechanical polishing to a ‘mirror-like’ finish were conducted by grounding the specimenswith successive SiC papers grit 100–1200. In this way,the surface roughness within the gage length of thesamples was maintained below approximately 0.2mm.The deposition process was carried out at UES

Arcomat, Inc., Dayton, Ohio, USA by means of FCAprocess. The defect free Ti ions were ejected out fromthe 908 bent filter associated with the FCA system alongwith the Ti ions from the enhanced direct arc source inpresence of reactive nitrogen gas, to deposit the TiNcoating. The latter was deposited at a temperature of400 8C under an arc current of 70 A for filter arcsources and 50 A for direct arc sources, a bias voltageof 60 V and planetary rotation at 8 rev.ymin.Prior to the deposition of the coating, argon ion

sputter cleaning was carried out at a pressure of8=10 Pa employing a voltage bias ofy400 V for 5y2

min. Subsequently, a Ti buffer layer was depositedemploying only filtered arc sources, at an argon pressureof 6=10 Pa, bias voltage ofy60 V for 6 min. They2

TiN coating was finally deposited using both filter anddirect arc sources, at a nitrogen pressure of 6=10 Pay2

and a bias voltage ofy60 V for 30 min.Both the chemical composition and thickness of the

coating were evaluated by means of secondary neutronmass spectroscopy(SNMS) at MATS, Warrington, UK.However, ball cratering(Calotest, CSEM) and imageanalysis(LECO 500) techniques were also employed in

order to corroborate the coating thickness. The evalua-tion of the residual stresses present in the coating wasconducted at Thin Films Solutions, Durham, UK. TheXRD analysis was carried out using a fully automatedSiemens D500 powder diffractometer, employing CuKa radiation (wavelengths0.15 nm) and a secondarymonochromator. The samples were scanned using a stepsize of 0.058 (2u) and a count time of 20 s per step inthe range 118–1308 (2u). Positive and negative tiltswere used for the stress data. The(422) reflection ofTiN at an approximate angle of 123.88 (2u) was usedfor the stress analysis, this being the highest angle peaknot suffering interference from substrate peaks. Theelastic constants of TiN employed in the calculationwere: Es424 GPa andns0.20 w7x.The evaluation of the mechanical properties of both

the uncoated and coated specimens was conducted bymeans of tensile tests, which were carried out on acomputer-controlled servohydraulic machine(Instron8502, Canton, MA, USA) at a cross head speed of 3mmymin, employing at least three samples for eachcondition. Fatigue tests were carried out under rotatingbending conditions(Rsy1), employing a FatigueDynamics (Walled Lake, MI, USA) RBF-200 equip-ment, at a frequency of 50 Hz(3000 rev.ymin). Thetests were all carried out in air at RT(23 8C) and thebending moment(M , N mm) applied to the specimensB

was determined as a function of the maximum alternat-ing stress(S, MPa) and the diameter(d, mm) of thesample by means of the relationship:

p 3M s S d , N mm (1)B 32

The uncoated substrate was tested at maximum alter-nating stresses in the range of 400–460 MPa, whereasthe coated samples were tested at stresses in the rangeof 480–510 MPa. The evaluation of the fatigue strengthof the uncoated and the coated samples was carried outemploying 24 samples, which fulfills the number ofspecimens required in S–N testing for reliability dataaccording to the ASTM standard 739(12–24 samples)and allows a replication greater than 80%. It is importantto emphasize that, in order to conduct a meaningfulcomparison of the fatigue results between the uncoatedand coated conditions, all the samples were mechanicallyprepared in a similar manner such that all of them hada similar mirror-like polished surface before testing. Thestaircase method with a step of 20 MPa was employedfor the determination of the fatigue limit of the materialin each condition. According to the ASTM standard E-468, infinite life was specified at a number of 5=106

cycles and at least 10 samples of each condition wereused for this purpose.SEM techniques (Hitachi S-2400, Japan) were

employed in order to study the fracture surfaces of

279E.S. Puchi-Cabrera et al. / Surface and Coatings Technology 182 (2004) 276–286

Fig. 2. Typical SNMS depth profile of the coated specimens.

selected coated samples tested at the lowest and highestmaximum alternating stresses. Special attention was paidto the site initiation of fatigue cracks and the differentstages undergone by these during their subsequent prop-agation. The observations were conducted at a potentialof 20 kV, both on the plane of fracture and along cross-sections normal to it, in an attempt to analyze thesequence followed during the fracture process of thecoating–substrate system.

3. Experimental results and discussion

3.1. Characteristics of the deposit

The analysis of the cross-section of some of thecoated samples indicated the deposition of a uniformcoating with a mean thickness of approximately 1.4mmand a coating–substrate interface initially free of cracks.In general, the adhesion of the coating to the substratewas found to be satisfactory, although during fatiguetesting at elevated stresses(510 MPa), as will bediscussed later, delamination of the coating from thesubstrate was observed to occur to some extent, partic-ularly along the lateral wall of the tested samples, nearthe crack initiation site. As mentioned before, the coatingthickness was also evaluated by means of the ballcratering technique. Accordingly, six measurements werecarried out from which the coating thickness was foundto range between approximately 1.1–1.7mm. However,Fig. 2 illustrates the corresponding SNMS depth profileindicating that up to a thickness depth of approximately1.4 mm the composition of the coating, as expected, isdominated mainly by the presence of titanium andnitrogen, followed by carbon to a much lesser extent.The measurement of the residual stresses in the

coating indicated the presence of compressive stressesof the order of approximately 7080"760 MPa and alattice parameter of approximately 0.4276 nm. Suchresidual stresses are significantly higher than those found

by Berrıos et al.w2x in the sub stoichiometric TiNx´coatings deposited by PVD magnetron sputtering thatwere analyzed in their work. These stresses were of theorder of approximately 5700"380 MPa. Since it is wellknown that compressive residual stresses play a funda-mental role in improving the fatigue behavior of mate-rials, it is expected that in the present case an increasein fatigue properties of the coated material in comparisonwith the uncoated substrate is also bound to be observed.However, the comparison between both deposition meth-ods in terms of the increase in fatigue properties of thecoated material must be carried out taking into consid-eration the fatigue properties of the correspondingsubstrate.

3.2. Evaluation of mechanical properties

The tensile tests conducted on the 316L stainless steelsubstrate samples indicated a yield strengthR s0.2%( )

590"48 MPa and a tensile strength,R s699"11max

MPa, whereas for the coated samples the tensile prop-erties were not observed to change significantly incomparison with the uncoated samples. In this case, itwas determined thatR s600"7 MPa andR s0.2% max( )

720"5 MPa. These results also differ from those report-ed by Berrıos et al.w2x regarding the sub stoichiometric´TiN coatings obtained by PVD magnetron sputtering inx

the sense that in such an investigation the coatings wereobserved to give rise to a significant increase in thetensile properties of the coated material in comparisonwith the uncoated substrate, which depended on thenitrogen content of the coating. For the uncoated sub-strate, Berrıos et al.w2x reported values of 489"5 MPa´and 661"3 MPa, for the yield and tensile strength,respectively, whereas for the coated samples theseresearches reported values in the range of approximately535–538 MPa and 668–676 MPa for both properties,respectively. Berrıos et al.w2x explained such an incre-´ment in terms of both the high strength of the TiNx

coatings and their excellent adherence to the substrate,and were able to carry out a crude estimate of the yieldstrength of their coatings by assuming that for the coatedmaterial this property can be represented by a simplelaw of mixtures and taking into account the coatingthickness for the computation of the fraction of thecoating in the mixture. Thus, Berrıos et al.w2x reported´that for their coatings the strength could vary betweenapproximately 25–33 GPa. Since in the present case theyield strength of both the uncoated and coated substratesare within the S.D. of each other, it is not possible toconduct a similar estimation for the strength of thecoating.As mentioned before, the fatigue tests of both the

uncoated and coated specimens were conducted at max-imum alternating stresses that corresponded to a fractionof their yield strength. Thus, the substrate specimens

280 E.S. Puchi-Cabrera et al. / Surface and Coatings Technology 182 (2004) 276–286

Table 1Mean number of cycles to failure(N ) vs. maximum alternating stress(S) for the uncoated specimensf

Stress, MPa Cycles to fracture Mean S.D.

400 207 300 210 600 235 400 236 400 433 100 722 100 340 817 187 391420 136 400 147 700 158 000 200 200 211 100 216 100 178 250 31 855440 82 100 109 200 125 500 127 300 145 300 150 400 123 300 22 861460 57 000 58 000 60 800 68 400 75 900 78 500 66 433 8475

Table 2Mean number of cycles to failure(N ) vs. maximum alternating stress(S) for the coated specimensf

Stress, MPa Cycles to fracture Mean S.D.

480 196 700 198 600 202 700 284 800 403 200 443 000 288 167 100 756490 148 300 169 200 169 800 192 500 219 900 453 400 225 517 104 316500 93 900 98 100 117 900 127 800 147 600 177 500 127 133 28 835510 67 000 104 000 107 200 109 100 119 200 152 600 109 850 25 143

Table 3Experimental results for determining the fatigue limit of the substratematerial

Sample Maximum alternating stress, MPaNumber of cycles

1 395 290 5002 390 286 4003 385 585 1004 380 639 7005 375 5 000 0006 380 5 000 0007 385 386 2008 380 5 000 0009 385 447 10010 380 5 000 000Fatigue limit 381S.D. 2

Table 4Experimental results for determining the fatigue limit of the coatedsamples

Sample Maximum alternating stress, MPaNumber of cycles

1 455 5 000 0002 460 2 637 0003 455 5 000 0004 460 5 000 0005 465 5 000 0006 470 5 000 0007 475 272 0008 470 292 5009 460 428 00010 455 5 000 000Fatigue limit 464S.D. 2

Fig. 3. Mean number of cycles prior to fracture(N ) as function off

the maximum alternating stress applied to the material(S) for theuncoated and coated specimens. The dotted lines indicate the fatiguebehavior of the sub stoichiometric TiN coating reported by Berrıos0.75 ´et al. w2x.

were tested at stresses in the range 400–460 MPawhereas the coated samples were tested at stresses inthe range of 480–510 MPa. The difference in the rangeof maximum alternating stresses was due to the differ-ence in the fatigue behavior between the coated anduncoated samples that was observed during testing.

Under cyclic loading the coated samples were observedto perform much better than the uncoated specimens,which was very likely to be associated with the effectof the compressive residual stresses present in the TiNcoating. Thus, the determination of the stress–life curvefor the coated samples required the application of highermaximum alternating stresses.The number of cycles prior to fracture(N ) as af

function of the maximum alternating stress applied tothe material(S), for the uncoated and coated samplesare presented in Tables 1 and 2, whereas the dataemployed in the estimation of the fatigue limit are givenin Tables 3 and 4. Fig. 3 illustrates these results in agraphical manner from which a number of important

281E.S. Puchi-Cabrera et al. / Surface and Coatings Technology 182 (2004) 276–286

Table 5Parameters involved in the Basquin relationship for the conditionstested

Condition A, MPa m

Substrate 1046.2 0.075TiN 762.6 0.037

Fig. 4. Change in the percentage of increase in fatigue life for the TiNcoatings obtained from different deposition processes, as a functionof the maximum alternating stress.

observations should be pointed out. However, it isimportant to emphasize first that in the analysis of thefatigue behavior of both uncoated and coated specimensat least 6 tests were conducted at each alternating stressin order to fulfill the reliability conditions prescribed inthe ASTM standard E-739 for this kind of testing.The first striking observation regarding the above

figure is the significant difference that can be seenbetween the fatigue behavior of the uncoated and coatedspecimens. The location of the fatigue curve for thecoated samples indicates that the TiN coating gives riseto a significant increase in fatigue properties of thecoated substrate in comparison with the uncoated steel.However, due to the difference in the slope of thefatigue strength curves of both materials, it is clear thatthe magnitude in the increase in fatigue life dependsstrongly on the maximum alternating stress at which thecoated material is subjected.Therefore, in order to quantify the change in the

increase in fatigue life as a function of the maximumalternating stress, the two constants involved in thesimple parametric relationship that associates the numberof cycles to fracture(N ) with the maximum alternatingf

stress applied to the material(S) had to be determinedfrom the experimental data.

ymS s A N (2)f

The above equation is similar to that advanced earlierby Basquinw8x for the description of this kind of data,where A, the fatigue strength coefficient andm, thefatigue strength exponent, represent constants thatdepend on both material properties and testing condi-tions. The values of the parametersA and m arepresented in Table 5 for both the uncoated and coatedspecimens. Such values are of utmost importance bothfor the evaluation of the fatigue life and design purposesunder considerations of high cycle fatigue of any com-ponent made of this kind of steel that could be coatedwith TiN deposited by FCA for improving some of itsproperties, such as corrosion, galling and wear resis-tance. Once the parametersA and m have been deter-mined, the increase in fatigue life can be simplycomputed as given below, together with Eq.(2) and thevalues reported in Table 5 for the computation of thenumber of cycles to fracture for each condition.

TiN Subst.N y Nf f% increases =100 (3)Subst.Nf

Fig. 4 illustrates the change in the % of increase infatigue life for the coated specimens in comparison withthe uncoated ones where it is clearly seen that theimprovement in fatigue life can vary between approxi-mately 400–2119% when the maximum alternatingstress changes from 510 to 460 MPa. Above approxi-mately 463 MPa, such an increase in fatigue life issomewhat less than that reported by Berrıos et al.w2x´for their TiN coating, for which the increase evaluated0.75

in the same maximum alternating stress range wouldspan between approximately 782–2037%. Fig. 4 alsoincludes for comparison the curve that would beobtained from the data reported by Berrıos et al.w2x for´the TiN coating taking into consideration its respec-0.75

tive substrate.Regarding the fatigue limit, it was observed that this

property increased by approximately 22% in comparisonwith the uncoated substrate, that is to say 60 MPa abovethe fatigue limit found for the latter. Such an increaseis twice as high as that reported by Berrıos et al.w2x for´

282 E.S. Puchi-Cabrera et al. / Surface and Coatings Technology 182 (2004) 276–286

Fig. 5. (a) General fracture surface of a sample tested at a maximumalternating stress of 480 MPa. The fracture surface is observed to beflat. A single crack located at A dominated the fracture process.(b)Magnified view of the crack initiation site shown in(a). The crackpropagation direction is indicated by the arrows and some flat facets(F) typical of the early stages of crack propagation are also visible.

the TiN coating, of approximately 11%. In our view,0.75

the difference in fatigue properties displayed by thespecimens coated with a TiN coating deposited by FCAcompared to those samples coated with the TiN depos-x

ited by UMS, particularly at low maximum alternatingstresses, of the order of the fatigue limit, could berelated mainly to the higher compressive residual stress-es present in the former.Berrıos et al.w2x showed that their coatings had an´

excellent adhesion to the substrate material and also avery high yield strength, which combined with com-pressive residual stresses of the order of 5.7 GPa gaverise to a significant increase in fatigue properties. Onthe contrary, the present TiN coating did not give riseto any significant increase in the yield strength of thesubstrate and it was observed to fracture and delaminatefrom the steel surface when tested under certain alter-nating stresses. However, it presented higher residualstresses than the coatings deposited by UMS, whichwould be the dominating factor in explaining the supe-rior fatigue performance displayed by these samples atlow maximum alternating stresses.The present results indicate clearly that the fatigue

fracture of the coated material is controlled by thenucleation of a fatigue crack at the surface of the coatingand its subsequent propagation throughout it, until itreaches the substrate surface and continues its propaga-tion along the cross-section of the specimen. On thecontrary, if fatigue cracks were nucleated at the coating–substrate interface, given the higher fracture toughnessof the TiN coating, such cracks would tend to propagatedirectly into the substrate material without giving riseto any increment in fatigue life in comparison with theuncoated substrate. This explanation would also beconsistent with previous findingsw3,4,6x which reportedan increase in the fatigue properties of other steelscoated with coatings deposited by PVD processes.

3.3. Evaluation of the fracture surfaces of the samples

The fracture surface of some of the coated specimenstested at 480 and 510 MPa that failed at a number ofcycles closest to the mean, were examined in detail bymeans of SEM techniques. This study was carried outmainly with the purpose of investigating the microstruc-tural features that characterize the crack initiation sitesand to clarify the role of the coating in the fractureprocess of the specimen. Also, the cross-section of someof the fracture specimens were analyzed in order tocorroborate the hypothesis that was put forward in theprevious section regarding the fracture sequence of thecoated samples.Fig. 5a illustrates the general fracture surface of a

sample tested at 480 MPa. The fracture surface wasobserved to be flat and the fracture process dominatedby a single crack whose location, A, is indicated by thearrow. The origin of the crack on the photomicrograph

is clearly revealed by the convergence of the fracturelines that emanate from such a point. A magnificationof the site A in this figure, as shown in Fig. 5b, allowsa better appreciation of the different features that char-acterize the neighborhood of the crack initiation site.The crack propagation direction is indicated by thearrows and some flat facets(F), typical of the earlystages of crack propagation, are also visible. In thispicture, the crack nucleation site is located also at A.However, Fig. 6a–d illustrate a different view of the

same specimen presented in the two previous pictures,which allow a better analysis not only of the crack

283E.S. Puchi-Cabrera et al. / Surface and Coatings Technology 182 (2004) 276–286

Fig. 6. (a) Different view of the same specimen presented in Fig. 5. The crack nucleation site is located at the center of C. On the lateral wall(L), a number of circumferential cracks parallel to the main fatigue crack can be seen.(b) Circumferential crack depicted as G. Small nodulespresumably associated with the deposition process can also be revealed.(c) Magnified view of the crack initiation site.(d) Magnified view ofthe circumferential crack pointed out in(b) as G. The nodules mentioned above are also clearly visible.

nucleation site but also of several important featuresthat could be observed on the lateral wall of the sample.In Fig. 6a the crack nucleation site is located at thecenter of C and on the lateral wall(L) a number ofcircumferential cracks parallel to the main fatigue crackcan be seen. When this region is magnified, as shownin Fig. 6b, small nodules presumably associated withthe deposition process can also be revealed. Fig. 6cillustrates a magnification of the crack initiation sitedepicted in Fig. 6a, whereas Fig. 6d shows a magnifi-cation of the circumferential crack pointed out in Fig.6b as G and some nodules which are also clearly visible.Finally, regarding this sample, Fig. 7 illustrates a

composite photomicrograph of the crack initiation siteviewed on the fracture surface where it can be noticedthe absence of secondary cracks along the coating–substrate interface when the specimens are tested at lowmaximum alternating stresses. The white arrows on thepicture indicate the direction of crack propagation andmany flat facets(F) are seen between the fracturemarkings.Fig. 8a, however, shows the general fracture surface

of a sample tested at 510 MPa, where again the fracture

process has been dominated by the propagation of asingle crack. In this case, the fracture surface wasobserved to be somewhat more irregular than thatcorresponding to specimen tested at 480 MPa. The SEMobservations indicated that under the present conditionsthe crack followed a wavy path and that the area sweptby the crack during its propagation was less, in com-parison with the previous sample, giving rise to a largerductile fracture zone, as expected. The crack initiationsite was located within the area designated as A, whichhas been magnified in Fig. 8b. Here the convergence ofthe fracture markings to the crack initiation site is clearlyseen. The white arrows indicate the direction of crackpropagation.The analysis of the lateral wall of the specimen

revealed that under these alternating stress conditions,which achieved approximately 0.85R , part of the0.2%( )

coating was observed to delaminate from the substrate(S), as shown in Fig. 9a. Such a delamination of thecoating seems to begin at the site initiation crack,indicating that once it had reached the coating–substrateinterface, it probably bifurcated to follow two differentpaths: one along the interface, which after the intersec-

284 E.S. Puchi-Cabrera et al. / Surface and Coatings Technology 182 (2004) 276–286

Fig. 7. Composite photomicrograph of the crack initiation site viewedon the fracture surface. The absence of secondary cracks along thecoating–substrate interface when the specimens are tested at low alter-nating stresses can be noticed. The white arrows on the picture indi-cate the direction of crack propagation. Flat facets(F) are seenbetween the fracture markings.

Fig. 8. (a) General fracture surface of a sample tested at a maximumalternating stress of 510 MPa. The fracture process has been domi-nated by the propagation of a single crack. The fracture surface isobserved to be somewhat more irregular than that corresponding tospecimen tested at 480 MPa. The SEM observations indicate thatunder the present conditions the crack followed a wavy path. Also,the area swept by the crack during its propagation was less, in com-parison with the previous sample. As expected, the ductile fracturezone is larger that at 480 MPa.(b) Magnified view of the crackinitiation site located within the area A in Fig. 8a. The convergenceof the fracture markings to the crack initiation site is clearly seen.The white arrows indicate the direction of crack propagation.

tion with other circumferential cracks produced thedelamination of the coating and another towards thesubstrate material, which gave rise to the final fractureof the specimen, in agreement with the mechanismproposed by Ferreira et al.w6x. Fig. 9b illustrates amagnified view of the area designated as C in Fig. 9a.Here, the substrate material shows a lighter tone thanthose locations where the coating still remains bondedto the steel substrate. Fig. 9c represents a furthermagnification of the area depicted as D in Fig. 9b wherethe uncoated substrate and small blocks of the coatingcan be clearly seen.Finally, regarding the analysis of this sample, Fig. 10

shows a magnified view of the crack initiation zone.The fracture markings clearly irradiate from the originof the crack and the propagation occurred along thedirection of the white arrows. The fracture surface isobserved to be much more irregular than that presentedin Fig. 7 due to the elevated alternating stresses appliedto the specimen. Fig. 11, however, illustrates a cross-section of a sample in which three cracks, parallel tothe main crack that gave rise to the fracture of thespecimen, can be seen at an early stage of development.It is observed that the crack at B has advanced throughthe entire coating thickness and is ready to start propa-gating towards the substrate material. On the contrary,the crack at C is observed to propagate along thecoating–substrate interface and the eventual intersectionwith crack B would give rise to the delamination of partof the coating, as shown in previous photomicrographs,as suggested by Ferreira et al.w6x.

4. Conclusions

Coating a 316L stainless steel with a TiN coatingdeposited by FCA gives rise to a substantial increase inthe fatigue properties in comparison with the uncoated

285E.S. Puchi-Cabrera et al. / Surface and Coatings Technology 182 (2004) 276–286

Fig. 9. (a) Lateral wall of the specimen. Under the present alternating stress conditions part of the coating is observed to delaminate from thesubstrate. The delamination of the coating seems to begin at the site initiation crack.(b) Magnified view of the area designated as C in Fig. 9a.The substrate material shows a lighter tone than those locations where the coating still remains bonded to the steel substrate.

Fig. 10. Magnified view of the crack initiation zone. The fracturemarkings clearly irradiate from the origin of the crack. The propa-gation occurred along the direction of the white arrows. The fracturesurface is observed to be much more irregular than that presented inFig. 7 due to the elevated alternating stresses applied to the specimen.

substrate. The increase in fatigue life has been deter-mined to be significantly dependent on the alternatingstress applied to the material. At low maximum alter-nating stresses(460 MPa) the increase in fatigue lifecan be up to 2119%, whereas at elevated maximumalternating stresses(510 MPa) an increase of up to400% can be achieved. In terms of the fatigue limit, ithas been observed that it increases up to approximately22%, that is to say, 60 MPa above the fatigue limitcomputed for the uncoated substrate. The examinationof the fracture surfaces of some selected samples indi-cated that under elevated alternating stresses the coatingtends to delaminate from the substrate. This observationand the fact that the presence of the coating did notgive rise to any significant increase in the yield strengthof the coated material, leads to the conclusion that theincrease in fatigue properties is due mainly to thecompressive residual stresses present in the coating,which were observed to be of the order of 7.08 GPa.The stress–life curves determined in the present studyas well as the microscopic examination of the cross-sections of some of the coated samples that were tested,indicate that fatigue cracks are nucleated at the surfaceof the samples and then propagate either towards thesubstrate or along the coating–substrate interface, asreported in the literature. Hence, it is also concludedthat the deposition of this kind of coating in order to

improve the corrosion, galling and wear resistance ofthe 316L stainless steel substrate does not compromiseat all the fatigue properties of the coated componentand, on the contrary, it gives rise to a significant increase

286 E.S. Puchi-Cabrera et al. / Surface and Coatings Technology 182 (2004) 276–286

Fig. 11. Cross-section of a sample tested at a maximum alternatingstress of 510 MPa. Three cracks, parallel to the main crack that gaverise to the fracture of the specimen, can be seen at an early stage ofdevelopment. The crack at B has advanced through the entire coatingthickness and is ready to start propagating towards the substrate mate-rial. The crack at C is observed to propagate along the coating–sub-strate interface. The eventual intersection with crack B would giverise to the delamination of part of the coating, as shown in previousphotomicrographs.

of the latter, particularly at low maximum alternatingstresses.

Acknowledgments

This investigation has been conducted with the finan-cial support of the Venezuelan National Fund for Sci-ence, Technology and Innovation(FONACIT) throughthe projects LAB-97000644 and S1-2000000642 and theScientific and Humanistic Development Council of theUniversidad Central de Venezuela(CDCH-UCV)through the project PG-08-17-4595-2000. J.A. Berrıos´is deeply grateful to the School of Mechanical Engi-neering, Faculty of Engineering and Architecture of theUniversity of El Salvador.

References

w1x J.R. Davis(Ed.), Stainless Steels, ASM International, MetalsPark, Ohio, USA, 1994.

w2x J.A. Berrıos, D. Teer, E.S. Puchi-Cabrera, Surf. Coat. Technol.´148 (2001) 179–190.

w3x V.R. Parameswaran, J.P. Immarigeon, D. Nagy, Surf. Coat.Technol. 52(1992) 251.

w4x W. Herr, B. Matthes, E. Broszeit, K.H. Kloos, Surf. Coat.Technol. 57(1993) 43.

w5x N.E. Frost, K.J. Marsh, L.P. Pook, Metal Fatigue, DoverPublications Inc, Cambridge, 1999.

w6x J.A.M. Ferreira, J.D.M. Costa, V. Lapa, Int. J. Fatigue 19(1997) 293.

w7x M. Desmaison-Brut, L. Themelin, F. Valin, M. Boncoeur, Euro-Ceramics 3(1989) 258.

w8x O.H. Basquin, Proc. ASTM 10(2) (1910) 625.