Biomaterials 27 (2006) 1

Fatigue and cyclic deformation behap

a,b

iser

g a

; ac

6 S

ar

s w

ma

he

ison with other metals, titanium alloys exhibit an surface. Thus, the in situ detection and characterization

to fatigue life and crack growth [11,12], whereby themeasurement of the free corrosion potential allowed thedetection of surface damage in axial or rotating bending

ARTICLE IN PRESS

eier@mv.uni-kl.de (D. Eier).1Until 15/05/05: Institute of Materials Science and Engineering,

University of Kaiserslautern, Gottlieb-Daimler-Str., 67663 Kaiserslautern,

Germany.

fatigue tests of passivating metals, such as stainless steels ortitanium alloys in saline solutions [13,14]. In addition,current measurements are often used to study local

0142-9612/$ - see front matter r 2005 Elsevier Ltd. All rights reserved.

doi:10.1016/j.biomaterials.2005.08.012

From 01/06/05: Swiss Federal Laboratories for Materials Testing and

Research (EMPA), Department of Advanced Materials and Surfaces,

Ueberlandstrasse 129, 8600 Duebendorf, Switzerland.advantageous combination of excellent biocompatibilityand high mechanical properties, which qualies them asmaterials for load-bearing implants. Their excellent bio-compatibility is assumed to be due to the formation of adense and stable TiO2 layer, which rebuilds spontaneouslyafter being damaged, even in solutions with low oxygencontents [14]. In order to improve the bioadhesion and the

of mechanically induced surface damages such as fatiguecracks is of major interest to satisfactorily predict the longterm in vivo behaviour of implant materials.Despite their physiological and mechanical relevance a

possible inuence of the highly complex in vivo loadingconditions on the cyclic deformation behaviour of thesealloys has hardly been considered yet. In particular, workon the fatigue behaviour of titanium implant alloys withclinically relevant surface conditions is rather scarce.Former work dealt mainly with different aspects limited

Corresponding author. Tel.: +49 631 205 2411; fax: +49 631 205 2137.E-mail addresses: christian.leinenbach@empa.ch (C. Leinenbach),physiological media indicated surface damages such as slip bands, intrusions and extrusions or nally microcracks. Microstructural

changes on the specimen surfaces were examined by scanning electron microscopy (SEM).

r 2005 Elsevier Ltd. All rights reserved.

Keywords: Titanium alloys; Surface modication; Titanium oxide; Surface roughness; Corrosion fatigue

1. Introduction

The fatigue strength, the low specic weight and thecorrosion resistance of titanium alloys are importantproperties for their use as aerospace materials as well asfor medical applications as implant materials. In compar-

corrosion behaviour, titanium implants are often surfacemodied, for example by surface roughening, oxidation orcoating techniques [57]. These surface treatments caninuence the mechanical properties of the implant alloy[810]. The integration behaviour can be inuenced bystructural and morphological changes of the implanthysteresis measurements. In addition, the change of the free corrosion potential and the corrosion current during testing in simulatedalloys in simulated

Christian LeinenbachaInstitute of Materials Science and Engineering, University of Ka

bSwiss Federal Laboratories for Materials Testin

Received 27 April 2005

Available online

Abstract

In this investigation, the cyclic deformation behaviour of the bin

axial stress-controlled constant amplitude and load increase tests a

relevant surface treatments (polishing, corundum grit blasting, ther

All tests were realized in oxygen-saturated Ringers solution. T2001208

viour of surface-modied titaniumhysiological media

,1, Dietmar Eiera,

slautern, Gottlieb-Daimler-Str., 67663 Kaiserslautern, Germany

nd Research (EMPA), Duebendorf, Switzerland

cepted 15 August 2005

eptember 2005

y titanium alloys Ti-6Al-4V and Ti-6Al-7Nb was characterized in

ell as in rotating bending tests. The inuence of different clinically

l and anodic oxidizing) on the fatigue behaviour was investigated.

cyclic deformation behaviour was characterized by mechanical

www.elsevier.com/locate/biomaterials

corrosion effects on metallic materials such as crevice orpitting corrosion [1517]. In some cases, potentiostaticallycontrolled current measurements were used to gaininformation on fatigue-induced surface damages on stain-less steels in corrosive media [18,19]. Load increase tests incombination with mechanical hysteresis measurementsallow to determine the endurance limit of a material witha very limited number of specimens [20]. In this work, thecyclic deformation behaviour of the implant alloys Ti-6Al-4V and Ti-6Al-7Nb with different clinically relevantsurface conditions was characterized in rotating bendingtests as well as in axial constant amplitude tests and loadincrease tests in Ringers solution.

2. Materials and methods

2.1. Materials

Cylindrical specimens were machined from hot and cold drawn rods of

ARTICLE IN PRESSC. Leinenbach, D. Eifler / Biomathe binary titanium implant alloys Ti-6Al-4V and Ti-6Al-7Nb, which

corresponded to the requirements of DIN ISO 5832 parts 3 and 11. Two

batches of each alloy with slightly different mechanical properties supplied

by different manufacturers, designated (C) and (E), respectively, were

considered. The chemical composition and the monotonic properties are

listed in Tables 1 and 2. Both alloys show very high yield and tensile

strengths, Rp0.2 and Rm, and ductility, A5. The (C)-batches were used for

axial fatigue specimens. Rotating bending specimens were taken from the

(E)-batches.

2.2. Surface preparation

In order to compare different surface conditions, the specimens were

mechanically polished (pol), corundum grit blasted (cgb) or thermally (to)

and anodically (ao) oxidized. The mechanically polished surfaces were

nally polished with 0.25mm SiO2 suspension, whereby surface roughnessvalues of Ra 0:03mm were achieved. The machining and mechanicalpolishing-generated surface-compressive residual stresses of approxi-

mately 406MPa (Ti-6Al-4V) and 340MPa (Ti-6Al-7Nb), respectively,as determined with the sin2c-method (cf. Table 3). Grit-blasted surfaces

Table 1

Chemical composition of the investigated Ti-6Al-4V and Ti-6Al-7Nb

alloys

[wt.%] C N Al O Fe V/Nb H

Ti-6Al-4V (C) 0.012 0.005 5.88 0.11 0.12 3.97/ 0.0019

Ti-6Al-4V (E) 0.02 0.01 6.04 0.105 0.18 4.02/ 0.0047

Ti-6Al-7Nb (C) 0.01 0.008 5.85 0.185 0.125 /6.90 0.003

Ti-6Al-7Nb (E) 0.01 0.004 6.03 0.175 0.13 /7.17 0.0022

Table 2

Monotonic properties of the investigated Ti-6Al-4V and Ti-6Al-7Nb

alloys

Alloy Rp0.2 (MPa) Rm (MPa) A5 (%)

Ti-6Al-4V (C) 880 960 18.5

Ti-6Al-4V (E) 865 1020 15

Ti-6Al-7Nb (C) 875 967 16.5Ti-6Al-7Nb (E) 882 1070 16.6on Ti-6Al-4V specimens were produced using an air shot peening device

and Al2O3 (corundum) shot with a mean particle size between 0.5 and

1mm. Oxide lms were generated on Ti-6Al-7Nb specimens. Thermal

oxidation was performed on the polished specimens in a laboratory

furnace in air at 675 1C for 180min. Anodically oxidized specimens wereproduced after polishing under galvanostatic conditions in 1M H2SO4 at

room temperature. During the oxidation, the constant current density of

10mA/cm2 resulted in a permanently increasing potential. The power

supply was immediately switched off at a nal potential of 40V.

2.3. Experimental techniques and loading parameters

2.3.1. Surface characterization

The chemical composition at the specimen surfaces was characterized

by Auger electron spectroscopy (AES, Perkin Elmer SAM 600). Element

depth proles could be determined by solid state sputtering using an Ar+-

ion gun with a maximum accelerating voltage of 5 kV. Fourier

transformation infrared spectroscopy (FTIR) was used to characterize

the structure of oxidized specimens (IFS 66 v/s, Bruker). The measure-

ments were carried out in the reection mode at an angle of 701 with p-polarized light. A Philips XPert X-ray diffractometer (Philips Analytical

B.V.) was used to measure surface residual stresses on specimens with

different surface conditions. All measurements were performed using Cr-

Ka radiation (l 0:229nm). The surface residual stresses were determinedby the sin2c-method. The elastic lattice strains were measured for the(1013) plane (2y 118:521) in the range 1161p2yp1221 in the positiveand negative c-direction between +601 and 601.

2.3.2. Fatigue tests

The axial fatigue tests were performed on a Schenck servo-hydraulic

testing system. Data were recorded by means of a software based on

LabView programmed at the Institute of Materials Science and Engineer-

ing, University of Kaiserslautern. The axial stress-controlled constant

amplitude tests were performed at a frequency of 5Hz and a load ratio of

R 1. Furthermore, the cyclic deformation behaviour was investigatedin stepwise load increase tests. The initial stress amplitude of 300MPa was

increased by 25MPa every 20,000 cycles over a period of 1000 cycles.

Single-step rotating bending tests at a frequency of 10Hz were performed

on a Schenck PUP Simplex rotating bending machine. All fatigue tests

were performed in oxygen-saturated Ringers solution at 37 1C. At thebeginning of the tests the pH-value was set to 7.4.

The cyclic deformation behaviour was characterized by the evolution of

the plastic strain amplitude versus the number of cycles. Fatigue-induced

surface changes were detected and characterized by corrosion potential

and current measurements. The free corrosion potential is measured

between the working electrode (WE, specimen) and the reference electrode

(RE, Argenthal Ag/AgCl, +207mV to standard hydrogen electrode

(SHE)). To determine the free corrosion current, the WE is connected with

a counter electrode (CE) via a low noise multimeter. The CE is a specimen

consisting of the same material with the same surface condition as the

working electrode. Consequently, both electrodes are on the same

electrochemical potential versus the RE while the surface of the loaded

specimen is undamaged. This experimental setup allows the simultaneous

detection of the corrosion current and potential in open circuit conditions.

The principles of the electrochemical techniques have been described

elsewhere in greater detail [13,14]. Fig. 1 shows schematically the

experimental setup for axial fatigue tests. All testing parameters are

summarized in Table 4.

3. Results

3.1. Surface conditions

3.1.1. Corundum grit blasted specimens

terials 27 (2006) 12001208 1201The SEM micrograph in Fig. 2(a) shows the coarsesurface structure of a grit-blasted specimen. Due to their

ARTICLE IN PRESS

rundum grit blasted, ao: anodically oxidized, to: thermally oxidized)

Ti-6Al-7Nb pol Ti-6Al-7Nb ao Ti-6Al-7Nb to

0.040 0.065 0.121

390 (755) 287 (720) +95 (732)

Table 4

Testing parameters

Test type Testing parameters Medium

Constant amplitude

testNmax 2 106 Oxygen-saturated

Ringers solution

f 5HzStepwise load increase

test

sa;min 300MPa Oxygen-saturatedRingers solution

Dsa 25MPa inDN 103DNstep 2 104f 5Hz

Rotating bending

testsNmax 2 106 Oxygen-saturated

Ringers solution

f 10Hz

materials 27 (2006) 12001208Table 3

Surface residual stresses and mean roughness values (pol: polished, cgb: co

Ti-6Al-4V pol Ti-6Al-4V cgb

Ra (mm) 0.040 5.210sres (MPa) 406 (734) 160 (726)

C. Leinenbach, D. Eifler / Bio1202high kinetic energy, Al2O3 particles fracture at the surfaceand are partly embedded in the material (Fig. 2(b)). Theachieved surface roughness values were comparable to datagiven by implant manufacturers in the range ofRa 426mm. Additional residual stress measurementsrevealed that the compressive residual stresses causedby grinding and polishing were signicantly reducedfrom 400 to 160MPa by the grit-blasting process (cf.Table 3).

3.1.2. Thermally and anodically oxidized specimens

Fig. 3 shows scanning electron micrographs of specimensurfaces after oxidation. The oxide lms produced bythermal oxidation (a) result in a very ne-grained crystal-line microstructure, while the anodically oxidized surfaces(b) exhibit a nanoporous structure. The porosity issignicantly higher in the Al-rich a-phase than in the Nb-rich b-phase. The microstructure below the surface oxidelayer was not affected by the oxidation. In particular, theformation of the so-called a-case, a hard and brittle phasewith high amounts of oxygen and nitrogen, which isdisadvantageous for the mechanical properties, could notbe observed.Both processes resulted in oxide layers between

80100 nm. The thickness was determined by the light blueinterference colour, which is characteristic for oxide lmsof approximately 8090 nm [6]. The oxide layer thickness

Fig. 1. Schematic drawing of the experimental setup for axial fatigue tests.

Fig. 2. SEM micrograph of corundum grit-blasted Ti-6Al-4V (C) (a),

cross section (b).

ARTICLE IN PRESS

Fig. 3. SEM micrographs of thermally (a) and anodically oxidized (b) Ti-

6Al-7Nb (C).

Fig. 4. AES-depth proles of a thermal and an anodic oxide layer on Ti-

6Al-7Nb.

C. Leinenbach, D. Eifler / Biomawas veried using AES. In Fig. 4, the concentration of thealloying elements Ti, Al and O as a function of the depthfrom the surface is shown for both a thermally oxidizedand an anodically oxidized Ti-6Al-7Nb specimen. In AES-depth proles, the element concentration is usually plottedagainst the sputter time. In this work, the sputter time hasbeen transferred to a Ta2O5-depth equivalent, which hasbeen determined on a Ta2O5 target with a well-denedthickness and a similar sputter behaviour as TiO2. On bothspecimens, the oxide concentration decreases at a depth ofapproximately 90 nm, while the titanium as well asaluminium concentrations increase. The titanium andoxide distributions appear more homogeneous in theanodic oxide layer. A signicant amount of aluminiumwas detected near the surface of the thermally oxidizedspecimen, possibly indicating the formation of Al2O3beside TiO2 and possibly other titanium oxides.With FTIR-absorption spectra it could be proved that

the thermal oxide lms consist mainly of crystalline TiO2 inthe Rutile modication with some amount of TiO2 in theAnatas structure (cf. Fig. 5) [2123]. The FTIR spectra ofthe anodically oxidized specimens revealed oxide lms,which are mainly amorphous with small amounts ofAnatas and Rutile [24]. Surface-compressive residualstresses were measured on anodically oxidized specimens

Fig. 5. FTIR-spectra of thermally and anodically oxidized Ti-6Al-7Nb.

terials 27 (2006) 12001208 1203while tensile residual stresses were detected on thermallyoxidized specimens. In comparison with polished speci-mens a slight decrease of the compressive residual stresseswas found for anodized Ti-6Al-7Nb. The pronounceddecrease of the compressive residual stresses after thermaloxidation might be partly explained with thermally inducedstress relaxation [25,26]. The presence of tensile residualstresses could be a result of the different thermal expansioncoefcients of the bulk material and the oxide layer.

3.2. Influence of the surface condition on the fatigue

behaviour

3.2.1. Corundum grit-blasted specimens

In Fig. 6 the S,N-curves of grit-blasted and polished Ti-6Al-4V are presented for axial fatigue tests (a) and rotatingbending tests (b). In comparison with polished Ti-6Al-4V,

ARTICLE IN PRESSmaC. Leinenbach, D. Eifler / Bio1204the fatigue life of grit-blasted Ti-6Al-4V is signicantlyreduced. The data were approximated by linear regressionbut not evaluated by statistical methods. The endurancelimit for axial loading could be estimated to 475MPa. Thiscorresponds to a decrease of more than 150MPa comparedto polished specimens. For rotating bending, the endurancelimit could be estimated to 540MPa. The decrease is lesspronounced than in axial fatigue tests, but the slightlydifferent material compositions and microstructures ofaxial fatigue and rotating bending specimens have to betaken into account. With regard to the results of thesurface residual stress measurements it can be stated thatno increase of the surface compressive residual stresses canbe detected after the cgb process, which is usually achievedin shot peening treatments. The surface roughening iscaused by abrasion. It is thus assumed that the fatiguebehaviour of Ti-6Al-4V is essentially determined by thesurface topography. High mechanical stresses are causedby the surface grooves or remaining corundum particles(cf. Fig. 2).

3.2.2. Thermally and anodically oxidized specimens

Fig. 7 shows the S,N-curves of thermally and anodicallyoxidized Ti-6Al-7Nb compared to the S,N-data forpolished specimens. For axial loading (Fig. 7a)), thermaloxidation in general reduced the number of cycles to failurein comparison with polished specimens, whereas for

Fig. 6. S,N-curves for polished and grit blasted Ti-6Al-4V in oxygen-

saturated Ringers solution, (a) axial loading, (b) rotating bending.terials 27 (2006) 12001208anodically oxidized specimens signicantly higher numbersof cycles could be determined at stress amplitudes below650MPa. A similar number of cycles to failure to those forpolished specimens were measured at higher stress ampli-tudes. The endurance limits are affected by the surfacetreatment and could be estimated to 500MPa (to) and530MPa (ao) for the different surface conditions. Forrotating bending (Fig. 7b), a signicant decrease of theendurance limit of the thermally oxidized specimens fromapproximately 675 to 640MPa and shorter fatigue lives athigher stress amplitudes compared to polished ones wereobserved. Anodically oxidized specimens, however, showeda higher endurance limit (E700MPa) and longer fatiguelives.

3.3. Characterization of fatigue-induced surface damage in

stepwise load increase tests

Figs. 8 and 9 show the results of stepwise load increasetests on Ti-6Al-4V and Ti-6Al-7Nb specimens with thedifferent surface treatments. The development of the plasticstrain amplitude, ea,p, is plotted versus the number ofcycles. Very low values of ea,p in the range of0.20.3 103 were measured at stress amplitudes up to750MPa for all specimens. For the polished specimen ofTi-6Al-4V, the plastic strain amplitude started to increase

Fig. 7. S,N-curves for polished, thermally and anodically oxidized Ti-6Al-

7Nb in oxygen-saturated Ringers solution, (a) axial loading, (b) rotating

bending.

ARTICLE IN PRESSomaFig. 8. Plastic strain amplitude versus the number of cycles in stepwise

load-increase tests (Ti-6Al-4V, Ringer).

C. Leinenbach, D. Eifler / Biat a stress amplitude of 700MPa. The corundum grit-blasted specimen failed at 600MPa without any change ofthe plastic strain amplitude. A similar behaviour wasobserved for thermally oxidized Ti-6Al-7Nb, which failedat a stress amplitude of 500MPa, whereas the anodicallyoxidized specimen showed pronounced cyclic softening atstress amplitudes of more than 700MPa.In addition to the mechanical hysteresis measurements

the changes of the free corrosion potential as well as thecorrosion current were measured. While hysteresis mea-surements are generally used for the characterization of thedegree of damage in the bulk material, fatigue-inducedsurface damages can be detected more precisely with theseelectrochemical techniques. The tests were started afterpassivation of the implant material in Ringers solution inthe load-free state. Surface deformations such as slipbands, intrusions and extrusions or microcracks lead to anoxide lm rupture and a local loss of passivity. As aconsequence, the free corrosion potential is shifted towardsmore negative values. Since the working electrode isconnected to the undamaged counter electrode, thepotential difference between the two electrodes induces a

Fig. 9. Plastic strain amplitude versus the number of cycles in stepwise

load-increase tests (Ti-6Al-7Nb, Ringer).corrosion current according to Ohms law. In oxygencontaining media, a new oxide lm will be createdimmediately due to repassivation. Thus, the ion ux intothe solution is reduced resulting in a drop of the corrosioncurrent and an increasing potential. The overall potentialand current changes over the time depend on the rates ofboth activation and repassivation.In Fig. 10, the corrosion current and potential as a

function of the number of cycles are presented for the cgbspecimen tested in a stepwise load increase experiment. Inboth signals, distinct peaks appear at the beginning of eachstep. With increasing stress amplitude, the initial decreaseof the potential as well as the increase of the currentbecomes more pronounced. A slow recovery of bothsignals can be observed during the 2 104 cycles withconstant amplitudes. Both signals indicate that the originalsurface oxide layer is ruptured if the stress amplitudeandthus the strain amplitude is increased, even at low-stressamplitudes. As a result, metallic ions are released if the

Fig. 10. Change of the corrosion potential and corrosion current versus

the cycle number in a load increase test of grit-blasted Ti-6Al-4V (C) in

Ringers solution.

terials 27 (2006) 12001208 1205specimen is strained. It can be assumed that cracks initiateat remaining corundum particles or in sharp grooves wherelarge stress concentrations exist. In the following, theformation rate of fresh surface decreases, resulting in adecreasing current and an increasing potential in the singlesteps. This leads to the assumption that the crack growthrate decreases as well. Pronounced crack propagationstarts if a critical stress amplitude is reached. SEMinvestigations on grit-blasted specimens after fatiguetesting revealed the formation of microcracks in thevicinity of corundum particles, as can be seen in Fig. 11.In Fig. 12 the development of the open circuit potential

and the corrosion current over the number of loadingcycles in a stepwise load increase test is shown for ananodically oxidized and a thermally oxidized specimen. Forthe anodically oxidized specimen, similar potential andcurrent curves were measured as for the polished ones. Thecorrosion potential and current remain almost constant upto a stress amplitude of 675MPa. The slight increase can bereferred to surface charging effects. The continuously

ARTICLE IN PRESSmaC. Leinenbach, D. Eifler / Bio1206decreasing potential and the increasing current at a stressamplitude of 700MPa indicate the initiation and propaga-tion of a fatigue crack. In contrast, numerous peaks both inthe potential and in the current signal of the thermallyoxidized specimen were detected throughout the whole test.This indicates that the surface oxide layer is obviouslydamaged by many cracks, which form spontaneously atmany sites.SEM investigations on axial and rotating bending

specimens were performed in order to determine thedamage characteristics of thermally and anodically oxi-dized specimens. The tests were interrupted after thepotential and current-indicated surface damage. Onanodically oxidized specimens, plastic deformation markssimilar to slip bands as well as delamination of the oxidelm in some areas at relatively high cycle numbers could be

Fig. 12. Change of the corrosion potential and corrosion current versus

the cycle number in a load-increase test of anodically and thermally

oxidized Ti-6Al-7Nb (C) in Ringers solution.

Fig. 11. SEM-micro cracks on grit-blasted Ti-6Al-4V in the vicinity of a

corundum particle (stepwise load increase test, N N f , z loadingdirection).observed. (cf. Fig. 13a and b). On thermally oxidizedspecimens, small, narrow cracks perpendicular to theloading direction which are characteristic for brittlematerials such as ceramics could be observed on bothaxial fatigue and rotating bending specimens (cf. Fig. 13cand d). This leads to the conclusion that the amorphousoxide lm produced by anodic oxidation is more ductilethan the highly crystalline thermal oxide lm, which isobviously not able to deform in the same manner as thebase material and ruptures even at relatively low stressamplitudes. These results underline the high performanceof current and potential measurements to provide verydetailed information about microstructural effects at thesurface. By applying the electrochemical techniques instepwise load increase tests, it is possible to characterize thesurface damage behaviour in corrosive environment withina rather short time.

4. Discussion

Referring to the results of the single step and rotatingbending tests, it can be concluded that corundum gritblasting leads to a pronounced decrease of the fatigue lifeof titanium alloys. It can be stated that cracks initiatemainly in the vicinity of remained corundum particleswhere a pronounced stress concentration occurs. Inaddition, it can be assumed that the particle impact inducesa local residual stress eld, which further enhances thecrack formation and the initial crack growth. Subse-quently, the crack propagates and leaves the critical regionin the surrounding of the particle and the plastic zone atthe crack tip then controls the further crack growth. Thus,the crack propagation rate as well as the activation rate offresh titanium surface is reduced and, as a result, thecorrosion current decreases. Pronounced crack propaga-tion starts if the critical stress intensity is reached at thecrack tip or if several small microcracks coalesce and forma larger crack. Although, no quantitative crack growthmeasurements were performed in this work. Even if thecracks do not necessarily grow to a critical crack length,which nally leads to failure, the continuous surfaceactivation during cyclic loading results in a high electro-chemical activity in comparison with polished or oxidizedspecimens. In addition, overloads can lead to a pronouncedincrease of the crack propagation rate. Thus, the risk of asudden fatigue failure is higher in the case of corundum gritblasted implants than for polished or oxidized ones.The large differences between the number of cycles to

failure of thermally and anodically oxidized Ti-6Al-7Nbspecimens lead to the assumption that despite the smallmicrostructural differences between the two alloy batchesthe fatigue behaviour is signicantly affected by the oxidelm structure and topography. It can be assumed that theamorphous oxide lms have a higher toughness andductility than the highly crystalline lms. These ndings

terials 27 (2006) 12001208are consistent with the results of scratch tests performedearlier, where it has been shown that thermal oxide lms

ARTICLE IN PRESS

ctio

a

omashow a rather brittle behaviour whereas oxide lmsproduced by anodic oxidation exhibit a more ductiledeformation behaviour [27]. The fatigue behaviour seemsto be less inuenced by the surface residual stresses, sincepolished and anodically oxidized specimens showed similar

Fig. 13. SEM-characteristic surface damages on Ti-6Al-7Nb, z loading direN 1:85 105, (c) to, axial, sa 700MPa, N 500, (d) to, rot. bend., s

C. Leinenbach, D. Eifler / Bicompressive residual stresses but different fatigue lives.

5. Conclusions

In the present investigation, the cyclic deformationbehaviour and the formation of fatigue-induced damageof the binary titanium alloys Ti-6Al-4V and Ti-6Al-7Nbwas characterized by mechanical hysteresis, corrosionpotential and current measurements in Ringers solution.The inuence of the surface conditions on the cyclicdeformation behaviour of the titanium alloys was investi-gated. The fatigue strength was decreased by corundumgrit blasting and thermal oxidation. In contrast, anodicallyoxidized specimens showed a higher fatigue strength andlonger fatigue lives than polished ones. Current andpotential measurements are a powerful method to char-acterize the damage behaviour of passivated metals undercomplex mechanical and electrochemical loading. More-over, stepwise load increase tests proved to be very usefulto estimate the fatigue behaviour as well as the damagebehaviour of titanium surfaces under cyclic loading withina relatively short time.

Acknowledgments

The authors greatly acknowledge the nancial support ofthe Deutsche Forschungsgemeinschaft DFG. They furtherthank R. Hanneforth of Stahlwerk Ergste and M. Windlerof Zimmer Holding, Winterthur, who supplied the testmaterials.

References

n; (a) ao, axial, sa 700MPa, N 2300, (b) ao, rot. bend., sa 740MPa720MPa, N 3:2 104.

terials 27 (2006) 12001208 1207[1] Effah EAB, Bianco PD, Ducheyne P. Crystal structure of the surface

oxide layer on titanium and its changes arising from immersion. J

Biomed Mater Res 1995;29:7380.

[2] Poilleau J, Devilliers D, Garrido F, Durrand-Vidal S, Mahe E.

Structure and composition of passive titanium oxide lms. Mater Sci

Eng 1997;47(3):23543.

[3] Milosev I, Metikos-Hukovic M, Strehlblow HH. Passive lm on

orthopedic TiAlV alloy formed in physiological solution. Biomater-

ials 2000;21(20):210313.

[4] Sittig C, Hahner G, Marti A, Textor M, Spencer ND, Hauert R. The

implant material Ti6Al7Nb: surface, microstructure, composition

and properties. J Mater Sci: Mater Med 1999;10(4):1918.

[5] Serruys Y, Sakout T, Gorse D. Anodic oxidation of titanium in 1M

H2SO4, studied by Ruther-ford backscattering. Surf Sci

1993;282:27987.

[6] Velten D, Biehl V, Aubertin F, Valeske B, Possart W, Breme J.

Preparation of TiO2 layers on cp-Ti and Ti-6Al-4V by thermal and

anodic oxidation and by solgel coating techniques and their

characterisation. J Biomed Mater Res 2002;59:1828.

[7] Frauchiger VM, Schlottig F, Gasser B, Textor M. Anodic plasma-

chemical treatment of CP titanium surfaces for biomedical applica-

tions. Biomaterials 2004;25(4):593606.

[8] Guleryuz H, Cimenoglu H. Effect of thermal oxidation on corrosion

and corrosion-wear behaviour of a Ti-6Al-4V alloy. Biomaterials

2004;25:332533.

[9] Li SJ, Niinomi M, Akahori T, Kasuga T, Yang R, Hao YL. Fatigue

characteristics of bioactive glass-ceramic-coated Ti-29Nb-13Ta-4.6Zr

for biomedical application. Biomaterials 2004;25:336978.

[10] Baleani M, Viceoconti M, Toni A. The effect of sandblasting

treatment on endurance properties of titanium alloy hip prostheses.

Artif Org 2000;24(4):2969.

[11] Lutjering G. Inuence of processing on microstructure and mechan-

ical properties of (a+b) titanium alloys. Mater Sci Eng A1998;243:3245.

[12] Niinomi M, Akahori T, Fukunaga K, Eylon D. Fatigue crack

initiation and fatigue life prediction of Ti-6Al-4V ELI. In: Boyer RR,

Eylon D, Lutjering G, editors. Fatigue Behaviour of Titanium Alloys.

Warrendale (PA): The Minerals, Metals & Materials Society; 1999.

p. 30714.

[13] Leinenbach C, Fleck C, Eier D. The cyclic deformation behaviour

and fatigue induced surface damage of the implant alloy TiAl6Nb7 in

simulated physiological media. Int J Fatigue 2004;26:85764.

[14] Leinenbach C, Eier D. Characterization of the cyclic deformation

behaviour and fatigue crack initiation on titanium in physiological

media by electrochemical techniques. Z Metallkd 2004;95(6):53541.

[15] Legat A, Dolecek V. Corrosion monitoring system based on

measurement and analysis of electrochemical noise. Corrosion

1995;51(4):295300.

[16] Hickling J, Taylor DF, Andresen PL. Use of electrochemical noise to

detect stress corrosion crack initiation in simulated BWR environ-

ments. Mater Corros 1998;49:6518.

[17] Matsch S, Bohni H. Localized corrosion and electrochemical noise

analysis. Mater Corros 2001;52:4308.

[18] Li YF, Farrington GC, Laird C. Cyclic responseelectrochemical

Interaction in mono and polycrystalline AISI 316L stainless steel in

H2SO4 solution, part IIII. Acta Metall Mater 1993;41:693737.

[19] Li YF, Laird C. Development of the electrochemical fatigue sensor.

In: Lutjering G, Nowack H, editors. Fatigue 96: proceedings of the

sixth international fatigue congress, vol. III. Oxford: Pergamon; 1996.

p. 19738.

[20] Schwilling B, Fleck C, Eier D. Cyclic deformation behaviour of

(a+b) titanium alloys under complex mechanical and physiological

loading conditions. Z Metallkd 2002;7:6206.

[21] Phillippi CM, Lyon SR. Longitudinal-optical phonons in TiO2(rutile) thin-lm spectra. Phys Rev 1971;B3(6):20867.

[22] Thibault S. Infrared reection-absorption spectra of oxide lms on

titanium. Thin Solid Films 1976;35:L335.

[23] Villegas MA, de Pablos A, Fernandez Navarro JM. Structural

and microstructural study of glasses in the LiOTiO2SiO2-system.

J Mater Sci 1995;30:9959.

[24] Morsi MM, El-Shennawi AWA. Some physical properties of silicate

glasses containing TiO2 in relation to their structure. Phys Chem

Glasses 1984;25(3):648.

[25] Hanagarth H, Vohringer O, Macherauch E. Fatigue behaviour of

shot peened TiAl6V4 in the temperature range 20 1CoTo450 1C. In:Wohlfarth H, Kopp R, Vohringer O, editors. Proceedings ICSP 3.

Oberursel: DGM; 1988. p. 47784.

[26] Hirsch T, Vohringer O, Macherauch E. Optimizing of the bending

fatigue behaviour of TiAl6V4 by shot peening. In: Lutjering G,

Zwicker U, Bunk W, editors. Proceedings fth interna-

tional conference on Titanium, vol 3. Oberursel: DGM; 1985.

p. 216370.

[27] Leinenbach C, Fleck C, Eier D. Characterization of mechanically

induced surface damage on the implant alloy TiAl6Nb7 in quasi-

physiological media. Mat-wiss u Werkstofftech 2002;33:4427.

ARTICLE IN PRESSC. Leinenbach, D. Eifler / Biomaterials 27 (2006) 120012081208

Fatigue and cyclic deformation behaviour of surface-modified titanium alloys in simulated physiological mediaIntroductionMaterials and methodsMaterialsSurface preparationExperimental techniques and loading parametersSurface characterizationFatigue tests

ResultsSurface conditionsCorundum grit blasted specimensThermally and anodically oxidized specimens

Influence of the surface condition on the fatigue behaviourCorundum grit-blasted specimensThermally and anodically oxidized specimens

Characterization of fatigue-induced surface damage in stepwise load increase tests

DiscussionConclusionsAcknowledgmentsReferences