International Journal of Fatigue 33 (2011) 1477–1489

Contents lists available at ScienceDirect

International Journal of Fatigue

journal homepage: www.elsevier .com/locate / i j fa t igue

Influence of laser polishing on the high cycle fatigue strength of mediumcarbon AISI 1045 steel

R. Avilés a, J. Albizuri a,⇑, A. Lamikiz a, E. Ukar a, A. Avilés a,b

a Department of Mechanical Engineering, ETSI, Faculty of Engineering, University of the Basque Country, Alameda de Urquijo s/n, 48013 Bilbao, Spainb Advanced Design & Analysis, IDOM Engineering & Consultancy, Avda. Lehendakari Aguirre, n. 3, 48014 Bilbao, Spain

a r t i c l e i n f o a b s t r a c t

Article history:Received 8 April 2011Received in revised form 20 May 2011Accepted 5 June 2011Available online 12 June 2011

Keywords:Fatigue strengthLaser polishingMedium carbon steelFatigue designLife prediction

0142-1123/$ - see front matter � 2011 Elsevier Ltd. Adoi:10.1016/j.ijfatigue.2011.06.004

⇑ Corresponding author.E-mail address: joseba.albizuri@ehu.es (J. Albizuri)

Laser polishing is a surface treatment which is becoming a feasible alternative to mechanical processessuch as conventional polishing and grinding, in a number of applications including moulds and dies.Being a relatively new process the knowledge of its effects on the fatigue behaviour is very poor. Thisresearch deals with the influence of laser polishing in the absence of an inert gas on the high cycle fatigue(HCF) of AISI 1045 steel. The experimental part of the work includes tests on tensile properties, surfaceroughness, residual stresses together with fatigue tests of mirror and laser polished specimens. The fati-gue behaviour is slightly related to the surface roughness but mostly to the microscopic properties of thesurface resulting from the melting and evaporation of material and the remainder of the heat affectedzone (HAZ). A theoretical study, based on the experimental results, is addressed to obtain quantitativevalues of the fatigue strength useful for the analysis and design of laser polished components. It will alsobe demonstrated that laser polishing improves the HCF behaviour of AISI 1045 steel components whosesurface roughness before laser polishing is higher than a threshold value.

� 2011 Elsevier Ltd. All rights reserved.

1. Introduction

Laser surface treatments can be considered as unconventionaland emerging manufacturing processes. Industrial applications oflaser surface treatments are relatively new and limited to somespecial cases. However, applications are growing continuouslydue to the advances in automation and quality of the processedsurfaces. One of the most relevant laser surface treatment pro-cesses is that of laser hardening. Laser hardening is based on theuse of a high intensity laser radiation source that rapidly heatsthe surface of the steel to the austenitic region. Due to the highrates of heat transfer, temperature gradients are set-up which re-sult in rapid cooling by conduction. This causes the transformationfrom austenite to martensite without the need for externalquenching. This process is actually being used in the manufactureof automotive parts and in the die and mould industry and, being athermal treatment, the knowledge of the mechanical properties,including improved fatigue behaviour, is good [1–3].

There are other laser surface treatment processes that have notbeen so widely used in industry and which are still subject to con-siderable research effort. This is the case for laser shot peening,which is based on the generation of shock waves by high energyradiation pulses. The result is the introduction of compressive

ll rights reserved.

.

residual stresses in the surface improving the fatigue resistance ofcomponents [4–6]. Laser shot peening can induce a deeper com-pressive residual stress layer and improve fatigue strength moreeffectively than shot peening [7]. Other processes, such as laser tex-turing, are based on the selective elimination of material by short,high power energy pulses (108–109 W/cm2) vaporising part of thematerial and melting another part which solidifies on the edgesand bottom of the crater left by the pulse [8]. The results of this pro-cess show high resolution texturing, but due to its low productivityrates and relatively high cost, it is only used for the production ofhigh added value parts, such as medical implants or photovoltaiccells. In laser-surface alloying or LSA the laser melts a shallow re-gion of the base metal and then additive is added and alloyed withthe molten base metal. It seems that if the additives are the rightones the fatigue resistance is almost unaffected by this process [9].

Finally, the laser polishing process is based on the melting andlater solidification of a micro layer of material using a laser beamas the heating source in order to obtain a smooth surface topogra-phy. The idea is to reduce the height of the surface projections bymelting and evaporation of material. This process has been appliedfor more than ten years to the polishing of non-metallic materials,such as diamond coatings, optical lenses and silicon wafers [10,11].There are also some studies on the application of the laser polish-ing process on metallic surfaces, showing good results from theeconomic and surface quality points of view [12–15]. Roughnessreduction rates of up to 80% can be obtained with final average

Nomenclature

HAZ heat affected zoneN number of cyclesrut tensile strength (Rm)ryp yield strengthr0.2 0.2% strain tensile strength (R0.2)min minimum valuemax maximum valuem mean valuer amplitudel denoting lasera, b, c, d, e measurement points in the X-ray diffraction testsDr stress range rmax � rmin

R fatigue ratio rmin/rmax

R0, R0.1, R�1 fatigue tests with R = 0, R = 0.1 and R = �1Ra average roughnessRath threshold average roughness for 106 cyclesrn fatigue limit at 106 cycles in the fully reversed axial

loading R�1 testr0n R�1 alternating bending fatigue limit of the materialr6, rl6 fatigue strength at 106 cycles in the fully reversed axial

loading R�1 tests, mirror finish and laser polished spec-imens

r06, r0l6 R�1 alternating bending fatigue strength at 106 cycles,mirror finish and laser polished specimens

Mss mean stress sensitivityrmax3, rmax6 fatigue strengths (maximum stress) at 103 and

106 cycles for the mirror finished specimens in a R0,1

axial testrlmax3, rlmax6 fatigue strengths (maximum stress) at 103 and

106 cycles for laser polished specimens in a R0,1 axialtest

cs modifying factors of the fatigue limit owing to surfaceroughness

cl3, cl6 laser polishing modifying factors of the 103 and 106

cycles fatigue strength (for calculations purposes)HCF, LCF high-cycle fatigue, low-cycle fatiguea3, al3 reduction factors of the tensile strength for 103 cycles,

mirror and laser polished materiala6, al6 reduction factors of the tensile strength for 106 cycles,

mirror and laser polished material

1478 R. Avilés et al. / International Journal of Fatigue 33 (2011) 1477–1489

roughness values (Ra) below 1 lm. Laser polishing is used to pro-duce a relatively smooth surface finish in previously machinedcomponents. However, this process cannot achieve a perfectlysmooth surface, and in many cases it cannot produce the degreeof smoothness of a mirror finish. There are also some characteristicsurface effects caused by this process, including surface pores,inclusions, and also micro-cracking, a consequence of the melting,evaporation and solidification stages [15]. Despite these surfaceproperties, the process is still useful for moulds and dies in appli-cations not needing a very low surface roughness.

The finishing operation of large metallic surfaces is one of themost important operations for the die and mould industry. Usuallythe final polishing operation is carried out manually taking morethan 20% of the total manufacturing time of the entire mould orstamping die. Moreover, it must be performed by highly qualifiedworkers, so this means high production cost and long lead times.In the die and mould industry, a skilled hand polisher needs typi-cally more than 20 min for 1 cm2 while it is estimated that laserpolishing machines will need about one minute. However, handpolishers can reach average roughness values Ra as low as0.005 lm, whereas current laser polishing techniques with indus-trial lasers, cannot provide this degree of smoothness. Neverthe-less, the roughness level that laser polishing can reach is enoughfor many industrial applications.

In the last years there has been renewed effort into the automa-tion of the polishing process; some studies propose a combinationof abrasive techniques with robotic manipulators [16] or the appli-cation of electron beam radiation [17,18] in order to polish the sur-face by melting a microscopic layer of the surface. Nevertheless,these solutions present some limitations. For instance, the use ofrobotic manipulators requires programming and setting up eachcase, limiting its application when the geometry of the surface tobe polished is complex. On the other hand, polishing by means ofelectron beam radiation implies a small working area and is onlysuitable for small parts. Moreover, the parts have to be placed ina vacuum chamber and the total cost of the equipment may betoo high for most die and mould makers.

Therefore, laser polishing is becoming a suitable alternative tohand polishing since many workshops are already using laser sys-tems for welding, repairing or hardening different components.

However, there is almost no data concerning the influence of thisprocess on the mechanical properties and in particular in the fati-gue behaviour of laser polished components. There are two mainparameters in the estimation of the profitability of laser polishingwhen compared with other surface finish processes; one is the sav-ing in the final cost of the product, where laser polishing may haveadvantage, but the other is the expected life and here is where fa-tigue must be considered.

2. Scope of this research and work plan

Laser polishing is a process intended to improve the surface fin-ish of machined components, but not necessarily the fatigue prop-erties. In fact, the melting of a surface layer can produce gasabsorption, inclusions and micro-cracking, all of which may affectthe fatigue strength [15]. Also, laser polishing can be performedusing an inert gas such as Ar to protect the melted surface or with-out inert gas, as is the case in this research. In this case the meltedsurface reacts with atmospheric oxygen producing a very thin layerof Fe oxides. It is important to point out that the evaporation andthe melting of a micro layer presents an important difference be-tween laser polishing and other treatments used specifically to im-prove the fatigue behaviour, as for instance laser hardening. In thissense it can be stated that laser polishing is a finish process sharingsome features with laser beam welding, which is also based on themelting of material [19]. Laser polishing in this work is done with-out inert gas, and is the last finish process.

The aim of this paper is to obtain quantitative data on thebehaviour of laser polished components in the HCF zone. Thesedata will be presented in the form of reduction coefficients andquantitative values which can be used by designers to estimatethe fatigue life of laser polished components. The work presentedhere focuses on medium carbon AISI 1045 steel.

A set of 70 fatigue specimens were manufactured with a surfaceRa roughness of about 0.04–0.08 lm which can be considered amirror surface finish. A subset of 45 specimens was chosen at ran-dom and laser polished with a HAZ of 100–150 lm, giving a surfaceRa roughness of 0.30–0.35 lm. Four specimens were used in a sta-tic stress–strain test, confirming that the static mechanical proper-

Table 1Laser polishing tests conditions.

TestRun

Test partdiameter(mm)

N(rpm)

X axisfeedrate(m/min)

Laserspotwidth(mm)

Laserpower(W)

Energydensity(J/cm2)

Results

1 8.00 240 6.03 0.90 150.19 166.00 Onlyheating

2 264.20 292.00 Somemeltedmaterialareas.

3 325.72 360.00 Optimummeltedmateriallayer

4 506.68 560.00 Highmeltedmateriallayer

5 No inert gas

R. Avilés et al. / International Journal of Fatigue 33 (2011) 1477–1489 1479

ties remained practically unchanged after the laser polishing. The25 mirror polished and 28 laser polished specimens were used inR0,1 constant amplitude axial fatigue tests to obtain the fatigue lifecurves in the range of 104–106 cycles, which amply covers the ser-vice live planned for dies and moulds. The rest of the specimenswere used for other experimental tests. The optimum parametersof the laser polishing were chosen following the procedure pre-sented in reference [15]. The experimental part of the work also in-cludes a chemical analysis of the steel, tensile properties, surfaceroughness of mirror and laser polished specimens, micro-hardnessmeasurements in the HAZ, and residual stresses in the surface andin the depth of the specimens.

The fatigue S–N curve of the mirror polished specimens is usedas the reference for the theoretical derivation of the coefficientsassociated with the laser polishing. The effect of the surface rough-ness will be separated from that of the laser treatment itself.Depending on the surface roughness of a component, the laser pol-ishing can improve or reduce its fatigue strength. If the initial sur-face roughness is higher than a threshold Rath value the fatiguestrength of the component will be improved and reduced if theroughness is lower. The values of Rath will be also obtained forthe AISI 1045 steel in the range 105–106 cycles and over 106 cycles.

3. Laser polishing treatment; experimental set-up

The laser polishing tests were carried out with a high powerdiode laser head Rofin Sinar DL015S. The laser head can achievea power, on continuous mode, of about 1.5 kW. Internally, the laserconsists of two diode stacks that generate a mixed wavelength of808–940 nm. The resulting laser beam presents a top-hat energydistribution with a 0.9 � 0.9 mm2 square spot shape. The laserhead is installed on a conventional three axis machining centrewith an additional rotary axis. The complete experimental set-upused in this research is shown in Fig. 1.

Laser polishing test conditions were obtained from previouswork [14,15]. From these tests, it was demonstrated that the re-quired overlapping between two consecutive laser tracks to polisha complete surface is nearly 20% of the laser beam width. The laserbeam was focused on the test part while the X axis and the rotary Aaxis were moving continuously. Therefore, the resulting laser trackon the surface is a helix trajectory along the complete treated sur-face. In this case, the overlapping is defined by the helix pitch,which has been set to 0.2 mm (20% of laser width). Different en-ergy densities were tested in order to adjust the conditions tothe material. The optimum value was set to 360 J/cm2. Table 1shows the different tests carried out to find the optimum laser pol-ishing parameters.

Fig. 1. Experimental set-up for the laser polishing of the fatigue test specimens.

Today the main use for laser polishing is the surface finishing oflarge metallic surfaces; therefore the optimisation of the parame-ters is based on the quality and cost of the surface finish but noton the mechanical properties of the HAZ or of the component.

4. Experimental results

4.1. Mechanical and chemical properties of the material

The work presented in this paper is the first step of a broaderresearch project on the effects that the laser polishing has on dif-ferent steels. The material chosen is an annealed medium carbonsteel AISI 1045, subjected afterwards to a normalizing heat treat-ment. This ductile steel is extensively used in industry and abun-dant bibliography exists on its mechanical properties andbehaviour. Therefore it was considered as the most suitable steelfor this first study.

The acquisition of the steel, the chemical and spectroscopicanalyses, the manufacturing of the specimens, as well as the staticand axial fatigue tests were done at the ‘‘Metallurgy Research CentreAZTERLAN’’ (www.azterlan.es). The X-ray diffraction measurementsof surface residual stresses were carried out at the laboratories‘‘IDEKO Ik4 Research Alliance’’ (www.ideko.es) and the in-depthresidual stresses at ‘‘ASM Accurate Stress Measurements(www.acstress.com)’’. The laser polishing of a part of the fatiguespecimens, the assessment of the alternating bending fatigue limitand the measurement of the roughness were carried out in theworkshop facilities of the Department of Mechanical Engineering.The depth of the laser-treatment affected layer (Heat AffectedZone, HAZ) is approximately 100–150 lm.

The chemical composition of the steel is given in Table 2.The material was supplied as 18 mm diameter bars from which

four tensile and 57 fatigue specimens were manufactured accord-ing to ASTM standards. All the fatigue specimens were polished toa mirror-like surface finish. To obtain the static mechanical proper-ties of the material, the four tensile specimens of 10 mm diameter

Table 2Chemical composition of AISI 1045 (at.%).

C Si Mn P S Cr Mo Ni Cu Fe

0.46 0.32 0.70 0.013 0.018 0.23 0.02 0.08 0.21 Balance

1480 R. Avilés et al. / International Journal of Fatigue 33 (2011) 1477–1489

were tested; the average tensile strength rut was found to be732 MPa and the yield strength ryp (or r0.2) 466 MPa. Four fatiguespecimens with laser polishing were also used in a static tensile testand, as was expected, the static strength properties remained prac-tically unchanged. The average values of (r0.2, rut) are (466 MPa,732 MPa) for the untreated material and (471 MPa, 715 MPa) forthe laser polished samples. Therefore the same value of rut is as-sumed for untreated and laser polished samples. A small increaseof dispersion was observed in the laser polished material, mostprobably due to the type of specimens used in this tensile test.

The raw material is a low alloy steel that present a classicalgrain structure, ferrite and perlite, which is modified under lasertreatment conditions. In Fig. 2, the metallographic structure of a la-ser treated test specimen is shown. In the HAZ the structure ismainly composed of martensite with variable quantities of ferrite.The austenization time is not sufficient to allow the migration ofcarbon throughout the former ferritic zones, so the layer is not

Fig. 2. (a) Metallographic structure of the laser treated zone and spectroscopic re

fully martensitic. In the outer surface the spectrum shows the pres-ence of oxygen (see Fig. 2) in the form of Fe oxides.

It was observed that the process creates a 100–150 lm thick-ness HAZ (120 lm in Fig. 2). The micro-hardness measurementsof Fig. 3 show that this area presents a harder and more brittlestructure due to an uncontrolled hardening of the treated layer;this effect is also commented on in references [14,15]. The micromelting and evaporation of the base material causes a HAZ thatis not perfectly homogeneous because of the base material struc-ture; the properties of this HAZ cannot be considered as defectsbut as typical features of this laser treatment.

4.2. Fatigue test planning and results

Moulds and dies work basically with pulsating stress withadded thermal stresses, so the R0,1 tests are well suited for this kindof application. Moreover, this paper deals with the fatigue behav-

sults for the outer surface. (b) Inclusions of manganese sulphur in the HAZ.

Fig. 3. Hardness in the HAZ and surface zones.

R. Avilés et al. / International Journal of Fatigue 33 (2011) 1477–1489 1481

iour at laboratory temperatures, although future tests are plannedat higher temperatures, with different fatigue ratios to account forthermal stresses and with other steels. Accordingly, using an axialfatigue test machine for this research was the more appropriatechoice. On the other hand, the life of moulds and dies is in the or-der of hundreds of thousand cycles, thus a 106 upper life limit forthe tests is large enough for the purposes of this work.

The constant amplitude axial fatigue tests were carried out in aMTS 810 servo-hydraulic machine with load control, at a frequencyof 20 Hz and a stress fatigue ratio R = 0, 1 (R0.1). The dimensions ofthe specimens [20] used in the axial fatigue tests are shown inFig. 4a; the L1–L2 zone in the centre is where the laser polishingwas applied. This zone presents a glossy dark grey layer as can

Fig. 4. (a) Fatigue specimen (dimension

be seen in Fig. 4b, which is caused by the oxides. This thin layerhas not been removed before the fatigue tests because the laserpolishing is, in principle, intended to be the last process finish ofindustrial components. Both untreated and laser polishedspecimens were used in the fatigue tests to analyze the effect ofthe laser polishing on the fatigue behaviour.

A total of 25 mirror finish fatigue specimens were tested at fivestress levels: 545, 515, 500, 485 and 470 MPa. Two specimens didnot break after 106 cycles at 470 MPa. Twenty-eight other speci-mens were laser polished and tested at six stress levels, 515, 500,485, 460, 440 and 425 MPa (three at 425 MPa and five for all theother cases). One specimen did not break at 440 MPa after 106 -cycles as well as three specimens tested at 425 MPa. All the laser

s in mm). (b) A treated specimen.

1482 R. Avilés et al. / International Journal of Fatigue 33 (2011) 1477–1489

polished specimens broke in the central region, as shown by theexample in Fig. 5 which broke after 2.75 � 105 cycles at 440 MPa.The melted layer (darker), the zones of slow and fast crack growthand the final ductile failure area can be easily identified in thephoto Fig. 5c. It is observed that the dark grey layer in the testedspecimens becomes matt and almost disappears in the fracturezone. The results of the tests are plotted in the log rmax–log N dia-gram of Fig. 6. The laser polished specimens show a lower fatigueresistance than the mirror finished ones, the R0,1 fatigue resistanceat 106 cycles is rmax6 = 461 MPa for the untreated specimens andrlmax6 = 420 MPa for the laser polished ones.

The statistical scattering in the results of the treated specimensis similar to those for the mirror finish samples, proving the robust-ness of the laser treatment process.

However, even though it is not essential for the purposes of thisresearch, a rotary bending machine Zwick/Roell Mod. UBM 200tcwas used to assess the 106 fatigue limit of the untreated material.

Fig. 6. Experimental R0,1 fatigue results of mirror and laser polished specimens.

4.3. Measurements of the surface roughness

These measurements were carried out using a contact profilom-eter Taylor Hobson Talysurf Series 2 in the axial and tangential

Fig. 5. (a) Fatigue fracture of a laser polished specimen. (b and c) Fracture zone.

R. Avilés et al. / International Journal of Fatigue 33 (2011) 1477–1489 1483

direction. The surface topographic maps are plotted in Fig. 7 for themirror and laser polished surfaces.

The roughness average Ra for the mirror polished specimens isin the range 0.0383 lm to 0.0801 lm (along the AA, Fig. 7), and0.0308 lm to 0.0437 lm (along BB, Fig. 7) for the laser polishedones. Fig. 8 shows measured profiles in the axial direction for bothtypes of specimen.

The surface of a laser polished specimen is shown in Fig. 9 pre-senting the classical shape of a melted and later solidified surface.However, the most important features of the surface affecting thefatigue behaviour are the micro-cracks and the inclusions of man-ganese sulphur that are caused during the solidification. Also thethin layer of oxides does not have a uniform depth, behaving alsoas a source of inclusions in the HAZ. For instance, Fig. 9b showsan enlargement of a typical Y shape crack of about 40 lm inthe zone marked with a rectangle in Fig. 9a. In addition severalinclusions can be observed in the zone. These flaws are the maincause of the reduction in the fatigue life observed in Fig. 6 whencompared with the mirror finished specimens. They are not theresult of a defective laser polishing but are inherent features ofthis kind of process with the parameters used in this work. Beingan emerging technique, future developments must pay attentionto better surface properties if enhanced fatigue behaviour isdesired.

Fig. 10 shows an 8000� magnification of a micro-crack in theouter surface of a new specimen.

Fig. 7. Topographic maps of the mirror

4.4. Residual stresses

Apart from the surface flaws described previously, the residualstresses could have also influenced the fatigue behaviour, mainly ifthey were tensile in the surface zone.

The residual stress measurements in the surface were carriedout using an X-ray diffractometer XSTRESS 3000 from Stresstech.The measurements were carried out at five points on the speci-mens and in three directions from each point: 0� (axial), 45�and 90�. Points a, b and c are in the laser polished zone and dand e in the untreated zone, as shown in Fig. 11. Six specimenswere used in the surface residual stress measurements, three mir-ror finished and three laser polished. One of each group was be-fore the fatigue test, one unbroken after 106 cycles, and onebroken.

The X-ray beam penetrates to a depth of a few lm below thesurface, but the residual stresses caused by the manufacturing pro-cesses of the specimens have considerable variations over depthsof this order due to the subsurface stress gradient, in this case alsodue to the thin layer of oxides. Therefore, surface residual stressmeasurements are subjected to significant error [21].

Fig. 12 shows the axial (0�) surface residual stress component inthe mirror and laser polished specimens. The measurements have aband of uncertainty preventing the direct use of these data forquantitative purposes or to set general conclusions, but the resid-ual stresses in the surface are clearly compressive.

finish and laser polished surfaces.

Fig. 8. Surface roughness of the mirror finish and laser polished specimens.

Fig. 9. (a) Photo of the laser polished surface. (b) Micro-cracks and inclusions in theouter surface layer.

Fig. 10. 8000� magnification of an outer surface micro-crack.

Fig. 11. Measurement points in the new and broken specimens.

1484 R. Avilés et al. / International Journal of Fatigue 33 (2011) 1477–1489

As far as the authors know, a detailed study of the in-depthresidual stresses caused by laser polishing has not yet been pub-lished. In this work two new laser polished specimens were usedin the tests with an iXRD instrument and 1 mm round aperture,using electropolishing to a depth of about 0.4 mm. The resultsshown in Table 3 and in Fig. 13 are in good agreement with the sur-face measurements of Fig. 12. The residual stresses remain com-pressive to a depth of about 0.1 mm and reach the maximumtensile value at 0.17 mm. Accordingly, the residual stresses canbe dismissed as the main cause of the observed reduction in the fa-tigue strength.

5. Quantitative analysis of the tests

As usual in practical applications, the 50% failure probabilityrmax–N lines of Fig. 14 have been extrapolated to 103 cycles, lead-ing to rmax3 = 684 MPa fatigue strength with fatigue ratio R0,1 forthe mirror finished specimens and rlmax3 = 563 MPa for the laserpolished ones, as given by points P3 and Q3 in Fig. 14. It must bepointed out that the stresses at P6 and Q6 are not fatigue limits, justthe fatigue strengths at 106 cycles. As was noted in Section 4.2, themain interest in this research is the fatigue behaviour in the finitelife zone, in the range of hundreds of thousand of cycles.

The fatigue tests were performed with a fatigue ratio R0,1. Thusthe relationship between the mean and amplitude of the stress is:

Fig. 12. Surface residual stress in axial direction along the specimens.

Table 3Residual stress versus depth in the centre of two new laser polished specimens; axialdirection.

Sample 1 Sample 2

Depth (mm) Stress (MPa) Depth (mm) Stress (MPa)

0.000 �337 ± 10 0.000 �332 ± 160.017 �322 ± 14 0.005 �344 ± 80.034 �343 ± 13 0.023 �326 ± 130.055 �239 ± 12 0.063 �328 ± 110.095 76 ± 16 0.093 �230 ± 70.163 270 ± 12 0.173 296 ± 120.334 197 ± 9 0.362 201 ± 10

Fig. 13. Axial residual stress versus depth

R. Avilés et al. / International Journal of Fatigue 33 (2011) 1477–1489 1485

rm

rr¼ 11

9ð1Þ

As was described in Sections 2 and 4.1, the laser polishing hasbeen applied to standard specimens whose original surface finish(almost mirror finish) is achieved by means of mechanical pro-cesses that cause moderate residual compressive stresses in thesurface. The beneficial effect of these stresses is implicitly includedin the fatigue test results and in the Goodman lines AP6 and AP3.Besides, for the steel used in this research, the sensitivity of the fa-tigue properties to moderate compressive residual stresses issmall.

for two new laser polished specimens.

Fig. 14. R0,1 fatigue-life lines extrapolated to 103 cycles.

Fig. 15. Goodman criterion for the fatigue tests.

Table 4Values of the threshold average roughness.

Life Rath (lm)

5 � 104–105 25105–106 15>106 5

1486 R. Avilés et al. / International Journal of Fatigue 33 (2011) 1477–1489

Assuming the Goodman [22] criterion is adequate to considerthe effect of the mean stresses in the fatigue life, Fig. 15 showsthe Haigh [23] diagram with the 106 and 103 standard failure linescommonly used for design and analysis purposes. It is generally ac-cepted in the literature [24–27] that the results of the reverse pureaxial loading tests (fatigue limit rn) are different from the rotating-beam tests (fatigue limit r0n) and accepting that the relationshipbetween the 106 cycle strengths r6 and r06 is approximately thesame as the one between the fatigue limit values, this can be writ-ten as:

rn ¼ 0:9r0n ) r6 � 0:9r06 ð2Þ

More discussion exists about the value to be used for 103 cycles,where the effects of the plasticity, eccentricity of the axial loads,and other errors, have a significant influence. In this paper a fatiguestrength a3rut will be used (see Fig. 15), where the value of a3 isquoted as being between 0.75 and 0.9 in the machine design bibli-ography [24–27]. The actual value of the parameter a3 for our casewill be afterwards adjusted on the basis of the fatigue test results.

In the HCF zone it is easy to obtain the relationship between thefatigue strength (maximum stress) in the axial R0,1 tests and theprediction of the fatigue strength (stress amplitude) that a R�1

bending test would provide. Using the AP6 line of the Goodman cri-

terion and Eq. (2), for 106 cycles the following equation can bewritten (see Fig. 15):

r06 ¼rmax6rut

2rut � 1:1rmax6ð3Þ

In the case of the mirror finished specimens, with rut = 732 MPaand rmax6 = 461 MPa, this results a 106 cycle fatigue strengthr06 ¼ 353 MPa.

Eq. (3) can be modified for the laser polished specimens to giveAQ6, where rl6 � 0:9r0l6 and rlmax6 = 420 MPa:

r0l6 ¼rlmax6rut

2rut � 1:1rlmax6ð4Þ

from which we obtain r0l6 ¼ 307MPa; and r0l6=r06 ¼ 0:87.Therefore, the laser polishing reduces the 106 cycles alternating

bending fatigue strength to 87%, as the laser-affected layer is thin(100–150 lm), but is in the crack initiation phase where this sur-face treatment has more influence [28–31]. This reduction can beattributed to two causes, the surface roughness and the effects inthe HAZ.

According to the roughness measurements presented in Section4.3 and the well known relationship between the surface rough-ness Ra with the tensile strength and the modifying factor cs ofthe endurance limit r0n [32], this factor at 106 cycles is 0.99 forthe mirror specimens used in the tests and 0.965 for the laser pol-ished ones.

For design purposes, it is advisable to separate the modifyingfactors or reduction coefficients of the two causes mentionedabove. Therefore:

r0l6 ¼ cscl6r06 ð5Þ

where cl6 is the modifying factor of the 106 fatigue strength due tothe laser polishing of AISI 1045 steel to be used in the Marin [33]equation. The product cscl6 is 0.87, and since cs = 0.99 the reductioncoefficient directly attributable to the laser polishing process iscl6 = 0.89, which must be used together with a surface finish factorwhich cannot be over 0.965 (0.97 for practical engineering pur-poses). As has been previously noted, laser polishing is based onmelting material to a depth larger than the asperities in the surface,in the range 100–150 lm in this study, thus the resulting surfaceroughness is not dependent on the previous roughness of the com-ponent. Therefore it can be stated that laser polishing will improvethe fatigue behaviour in the zone over 106 cycles when the averageroughness of the machined surface leads to a coefficient cs lowerthan 0.87. Then, according to Johnson [32], for steel with 732 MPatensile strength the threshold average roughness Rath is in the range4–5 lm, say 5 lm for practical purposes. The same reasoning andsimple calculations lead to cs equal to 0.725 and a threshold averageroughness of 13 lm for a life between 105 and 106 cycles, with15 lm a practical reference value. Table 4 shows values of Rath fordifferent life zones.

The line P3Z6 in Fig. 14 represents the estimated fatiguestrength of untreated specimens with the same roughness as thelaser polished ones. Thus the surface roughness of the laser pol-ished specimens is not the main cause of the reduction in the fati-gue strength when compared with the strength of the mirrorpolished specimens. The R0,1 fatigue strength at 106 cycles of the

Table 5Modifying factors of the fatigue strength for 100 lm HAZ laser polishing.

Life R�1 bending

cl al cs

103 0.69 0.6 1106 0.89 0.42 60.97

R. Avilés et al. / International Journal of Fatigue 33 (2011) 1477–1489 1487

untreated specimens with the same roughness as the laser pol-ished ones would be 454 MPa (point Z6) instead of the 461 MPa ob-tained for the mirror finished and the 420 MPa of the laser polishedsamples.

On the other hand, the fact that the mirror specimens have asurface finish modifying factor of 0.99 indicates that an estimationof the endurance limit in the rotating-beam test R�1 (r0n � r06=0:99)is 357 MPa. Also, in 2009, Fatemi et al. [34] reported longer life foralternating bending fatigue strength than predicted with axial testsfor medium carbon steels, so a slightly higher r0n could be ex-pected. However, being unnecessary for this research, the alternat-ing bending fatigue limit of the untreated material has beenobtained in a Zwick/Roell UBM 200tc testing machine leading tor0n � 360 MPa.

The mean stress sensitivity (slope of the Goodman straight line)Mss or ratio r06=rut is equal to 0.48, in good agreement with the wellknown recommended 0.5 endurance ratio for this kind of steel[24–27,35]:

Mss ¼r0nrut� r06

0:99rut¼ 0:49 ð6Þ

In 1987 Manson and Muralidharan [36] presented a non-classi-cal method for using axial fatigue information to estimate alternat-ing bending fatigue life in the LCF zone. However, it will be shownthat for this steel and for the purposes of this paper, the classicalapproach based on the Goodman criterion provides sufficientagreement between the experimental and predicted results. Ashas been obtained before, the extrapolation of the fatigue test re-sults down to 103 cycles for the mirror polished specimens givesa rmax3 value of 684 MPa. From the AP3 Goodman line for 103 cy-cles and R0,1 test, as shown in Fig. 15, the predicted a3 is:

Fig. 16. (a) R�1 bending fatigue-life lin

a3 ¼0:9rmax3

2rut � 1:1rmax3ð7Þ

Eq. (7) gives a3 = 0.865 (0.87 for practical purposes) for thissteel, in accordance with the recommended values (0.75–0.9)noted previously.

The results of the tests on the laser polished specimens showthat the fatigue strength at 103 cycles is lower than that of the mir-ror finished specimens. The laser polishing modifying factor is nowcl3 and Eq. (7) can be transformed for the AQ3 line into:

al3 ¼ a3cl3 ¼0:9rlmax3

2rut � 1:1rlmax3ð8Þ

Since the value of rlmax3 is 563 MPa, this results al3 = 0.6 andcl3 = 0.69. As the surface finish has little or no influence at 103 cy-cles, then the 0.69 value is the modifying factor to be used for thislifetime.

For the AISI 1045 steel, the fatigue–lifetime curve of laser pol-ished specimens can also be described in terms of the tensilestrength in terms of al3 and al6:

al3 ¼ cl3a3

al6 ¼ cscl6a6

�ð9Þ

then:

r03 ¼ a3rut ¼ 0:87rut

r06 ¼ a6rut ¼ 0:48rut

�)

r0l3 ¼ al3rut ¼ 0:6rut

r0l6 ¼ al6rut ¼ 0:42rut

�ð10Þ

Table 5 shows the modifying factors to be applied to the 103 and106 bending R�1 fatigue strengths of the raw material (cl3 and cl6),using reference data obtained on mirror finished specimens, andon the tensile strength (al3 and al6). These values can be useful inpreliminary fatigue calculations using the Basquin [24–27,36–38]fatigue–lifetime curve of the laser polished component being ana-lyzed. It is important to point out that usually the modifying coef-ficients are applied in the Marin equation [33] on the alternatingbending fatigue limit r0n, while they are used here to modify the103 and 106 alternating bending fatigue strengths to obtain the fa-tigue–lifetime line of laser polished components in the finite-life-time region.

The X-ray diffraction tests show that the residual stresses re-main compressive in the surface and subsurface zone to a depth

es. (b) R�rmax relation, axial load.

1488 R. Avilés et al. / International Journal of Fatigue 33 (2011) 1477–1489

of 0.1 mm after the laser polishing. Thus the reduction of the fati-gue resistance cannot be attributed to residual stresses resultingfrom the melting and solidification process. Nevertheless, simplecalculations based on effective residual stresses [28,35] show thatthe effect of the laser polishing on the fatigue resistance is equiv-alent to a situation with surface tensile residual stresses of about100 MPa.

Fig. 15a shows the N3N6 alternating bending fatigue-life R�1 lineof the laser polished AISI 1045 specimens together with the M3M6

line for the mirror finished specimens. Fig. 16b presents the rela-tionship between the fatigue ratio R and axial rmax/rut. Observethe 0.9 factor in the M6, N6 values in Fig. 16b. with respect to theones in Fig. 16a. The dotted lines are those recommended in the lit-erature [24–27] for this kind of steel before laser polishing. On theother hand, the number of cycles that moulds and dies are usuallyused is in the grey zones of Fig. 16.

In summary, it has been shown that laser polishing affects theHAZ by changing the surface residual stresses and modifying themetallurgical properties [14,15] and as a result changing the fati-gue behaviour of the treated specimens, as has been quantifiedhere for the AISI 1045 steel. Future work is necessary to have a bet-ter knowledge of the residual stress gradient, crack growth andmetallurgical changes in this and other steels.

6. Conclusions

The laser polishing creates a HAZ of about 100–150 lm depthwith the parameters used in this work. The melting, evaporationand fast solidification of the micro-layer changes the metallurgicalstructure and generates a surface layer whose features have a sig-nificant influence on the fatigue strength. The direct contributionof the surface roughness is small when compared with the restof the effects of the HAZ. The residual stresses in the surface andsubsurface zones are compressive to a depth of 0.1 mm.

The outer surface of the laser polished specimens presents Feoxides, micro-cracks and inclusions. There are also inclusions ofmanganese sulphur in the HAZ. These flaws are produced duringthe complex melting and solidification process of the surface layerand are inherent features of the laser polishing process with theparameters used in this work and the main cause of the reductionin the fatigue strength when compared with the results of the un-treated mirror polished specimens.

The laser polishing parameters were chosen using an optimisa-tion criterion that is based on the roughness of the surface and onthe cost of the process, but not the fatigue behaviour of the laserpolished components. The results presented in this work will beuseful to develop new criteria for the selection of the optimumparameters, including the fatigue life as a constraint.

For practical engineering purposes it is customary to separatethe quantitative influence of the surface roughness from the othereffects of the laser polishing, deriving specific modifying factors forthe R�1 bending fatigue resistance. The modifying factors to beused in the Marin formula for the AISI 1045 steel are 0.89 for106 cycles and 0.69 for 103 cycles, with the surface roughness coef-ficient considered separately. The R�1 bending fatigue strength ofthe laser polished specimens at 103 cycles can be estimated to be60% of the tensile strength and about 40% at 106 cycles. These mod-ifying factors can be useful for total lifetime methods, which arebased on the Basquin fatigue–lifetime curves. From another pointof view, it has been shown that the quantitative effect of the laserpolishing used in this work is equivalent to the addition of a meanstress of about +100 MPa to the mirror specimens.

The laser polishing is used as a surface finish process intendedto reduce the roughness of machined components. On the onehand, this process improves the fatigue life through the reduction

of the surface roughness, but on the other hand it causes some sur-face flaws, harming the fatigue strength. However, depending onthe surface roughness of the component before laser polishing,the fatigue behaviour after the treatment is improved or reduced.It has been demonstrated that laser polishing, with the parametersused in this work, improves the fatigue behaviour of AISI 1045steel for a lifetime over 106 cycles if the initial average roughnessRa is higher than a threshold value Rath equal to 5 lm, while forthe 105 to 106 cycles life range the threshold Rath is 15 lm.

The results and conclusions presented in this paper can be use-ful for design and analysis engineers in total life fatigue calcula-tions of laser polished components. However, more research isrequired to investigate the behaviour of other steels and the influ-ence of other laser polishing parameters.

Acknowledgments

The authors wish to acknowledge the financial support receivedfrom the Department of Research and Universities of the BasqueGovernment for the research project IT432-10. The authors alsoacknowledge the Ministry of Science and Innovation of Spain forfinancial support provided through the ‘‘SURFACER’’ projectDPI2010-20317-C02-01 and the DPI2009-07900.

References

[1] Singh HB, Copley SM, Bass M. Fatigue resistance of laser heat-treated 1045carbon steel. Metall Trans A 1981;12A:138–40.

[2] Kennedy E, Byrne G, Collins DN. A review of the use of high power diode lasersin surface hardening. J Mater Process Technol 2004;155–156:1855–60.

[3] Heitkemper M, Bohne C, Pyzalla A, Fischer A. Fatigue and fracture behaviour ofa laser surface heat treated martensitic high-nitrogen tool steel. Int J Fatigue2003;25(2):101–6.

[4] Hammersley G, Hackel LA, Harris F. Surface prestressing to improve fatiguestrength of components by laser shot peening. Opt Laser Eng 2000;34:327–37.

[5] Montross CS, Wei T, Ye L, Clark G, Mai Y. Laser shock processing and its effectson microstructure and properties of metal alloys: a review. Int J Fatigue2002;24(19):1021–36.

[6] Mahagaonkar SB, Brahmankar PK, Seemikeri CY. Effect on fatigue performanceof shot peened components: an analysis using DOE technique. Int J Fatigue2009;31(4):693–702.

[7] Gao Y-K. Improvement of fatigue property in 7050–T7451 aluminium alloy bylaser peening and shot peening. Mater Sci Eng A 2011;528:3823–8.

[8] Etsion I, Sher E. Improved fuel efficiency with laser textured piston rings. TribolInt 2009;42:542–7.

[9] McDaniels RL, White SA, Liaw K, Chen L, McCay MH, Liaw PK. Effects of a lasersurface processing induced heat-affected zone on the fatigue behaviour of AISI4340 steel. Mater Sci Eng A 2008;485:500–7.

[10] Erdemir A, Halter M, Fenske GR, Krauss A, Gruen DM, Pimenov SM, et al.Durability and tribological performance of smooth diamond films produced byAr-C60 microwave plasmas and by laser-polishing. Surf Coat Technol1997;94–95:537–42.

[11] Wang HY, Bourell DL, Beaman JJ. Laser polishing of silica slotted rods. Mater SciTechnol 2003;19:382–7.

[12] Shao TM, Hua M, Tam HY, Cheung EHM. An approach to modelling of laserpolishing of metals. Surf Coat Technol 2005;197(1):77–84.

[13] Dobrev T, Pham DT, Dimov SS. Laser polishing. UK: Elsevier, ManufacturingEngineering Centre, Cardiff University; 2006.

[14] Lamikiz A, Sánchez JA, López de Lacalle LN, del Pozo D, Etayo JM, López JM.Laser polishing techniques for roughness improvement on metallic surfaces.Int J Nanomanufactur 2007;1(4):490–8.

[15] Ukar E, Lamikiz A, López de Lacalle LN, del Pozo D, Arana JL. Laser polishing oftool steel with CO2 laser and high-power diode laser. Int J Machine ToolsManuf 2010;50(1):115–25.

[16] Ryuh BS, Park SM, Pennock GR. An automatic tool changer and integratedsoftware for a die polishing station. Mech Mach Theory 2006;4:415–32.

[17] Uno Y et al. High-efficiency finishing process for metal mold by large-areaelectron beam irradiation. Precis Eng 2005;29:449–55.

[18] Gao Y-K. Surface modification of TA2 pure titanium by low energy high currentpulsed electron beam treatments. Appl Surf Sci 2011. doi:10.1016/j.apsusc.2011.03.00.

[19] Sonsino CM, Kueppers M, Eibl M, Zhang G. Fatigue strength of laser beamwelded thin steel structures under multiaxial loading. Int J Fatigue2006;28:657–62.

[20] ASTM E466-07. Standard practice for conducting constant amplitude axialfatigue tests of metallic materials. Annual book of ASTM standards, vol. 03.01.American Society for Automotive Engineers; 1993.

R. Avilés et al. / International Journal of Fatigue 33 (2011) 1477–1489 1489

[21] Prevéy PS. X-ray diffraction residual stress techniques. Metals handbook. AmSoc Met 1986;10:380–92.

[22] Goodman J. Mechanics applied to engineering. London (UK): Longman-Green;1899.

[23] Haigh BP. Report British Assoc 1915;85:163–70.[24] Juvinall RC, Marshek KM. Fundamentals of machine component design. 3rd ed.

Wiley; 2000.[25] Norton RL. Machine design, an integrated approach, 3rd ed. Pearson

International Edition; 2006.[26] Shigley JE, Mischke CR, Budynas RG. Mechanical engineering design, 7th ed.

McGraw Hill; 2004.[27] Avilés R. Análisis de fatiga en máquinas. Editorial Thomson-Paraninfo: Spain;

2005.[28] Webster GA, Ezeilo AN. Residual stress distributions and their influence on

fatigue lifetimes. Int J Fatigue 2001;23(1):S375–83.[29] Suresh S. Fatigue of materials. Cambridge solid state science series. Cambridge

University Press; 1991.

[30] Champoux RL, Underwood JH, Kapp JA, editors. Analytical and experimentalmethods for residual stress effects in fatigue. ASTM STP 1004; 1988.

[31] Lados DA, Apelian D, Donald JK. Fracture mechanics analysis for residual stressand crack closure corrections. Int J Fatigue 2007;29(4):687–94.

[32] Johnson RC. Machine design, vol. 45 (11). USA: Penton Publishing; 1967. p.108.

[33] Marin J. Mechanical behaviour of materials. Prentice-Hall; 1962.[34] Fatemi A, McKelvey S, McCutcheon S. Comparison of axial and bending fatigue

of a medium carbon steel including specimen geometry and residual stresseffects. SAE document number: 2009-01-0422; 2009.

[35] Totten G, Howes M. Inoue T. Handbook of residual stress and deformation ofsteel. ASM International; 2002.

[36] Manson SS, Muralidharan U. Fatigue life prediction in bending from axialfatigue information. Fatigue Fract Eng Mater Struct 1987;9(5):357–72.

[37] Basquin OH. Proceed ASTM 1910;10:625–30.[38] Kun F, Carmona HA, Andrade JS, Jr., Herrmann HJ. Universality behind

Basquin’s law of fatigue. Phys Rev Lett 2008;100:094301.