Biomaterials
Volume 24, Issue 18 , August 2003, Pages 3027-3038

Evaluation of corrosion on plasma sprayed and anodized titanium implants, both with and without bone cement

L. Reclaru^{, , a}, R. Lerf^b, P. -Y. Eschler^a, A. Blatter^a and J. -M. Meyer<sup>c

a</sup> Groupe PX, PX Tech, Bldv des Eplature 46, La Chaux-de-Fonds 2304, Switzerland
^b PI Precision Implants AG Aarau, Switzerland
^c School of Dentistry, University of Geneva, 19 Rue Barthelemy-Menn, 1205, Geneva, Switzerland

Received 22 February 2003;  accepted 9 March 2003. ; Available online 1 May 2003.

1. Abstract

The corrosion behavior of titanium with vacuum plasma sprayed titanium coatings and with anodized surfaces, both with and without polymeric bone cement were evaluated. Electrochemical extraction tests were carried out with subsequent analysis of the electrolyte by ICP-MS in order to verify our hypothesis of the ionic permeability of the polymer cement.

The complexity of the situation resides in the existence of two interfaces: electrolyte-polymer and polymer-metal.

The surface preparation (treatment of the surface) plays an important role in the corrosion resistance of titanium. The electrochemical magnitudes that were examined reveal that the plasma spray surfaces have the lowest corrosion resistance.

The cement, in spite of having reduced electrical conductivity in comparison to metal, is an ionic transporter, and therefore capable of participating in the corrosion process. In the present study, we observed in fact crevice corrosion at the metal-cement interface. In the case of plasma spray surfaces, a process of diffusion of titanium particles in the electrolyte could accompany the crevice corrosion.

In this study, we have shown that there is a corrosion process at the surface of the titanium through the cement which has as a consequence on the one hand the formation of titanium cations and on the other hand the growth of a passive layer on the titanium.

In conclusion, we identified two principal factors that influence the corrosion process:

• the type of surface treatment for the titanium

• the ionic conductivity of the cement. There is indeed ionic transport through the cement; as evidenced by the presence of titanium in the electrolyte solution (ICP-MS analysis) and chloride at the surface of the titanium sample (EDX analysis).

We show that the polymer cement is an ionic conductor and participates in the corrosion of the embedded titanium. We cannot deduce from our results, however, whether the polymer itself possesses corrosive properties. Long-term experiments will be necessary to study the degradation behavior of the polymer cement.

Author Keywords: Ti VPS; Anodized titanium; Orthopedic implants; Bone cement; Crevice corrosion; Electrochemical extraction test

2. Article Outline

1. Introduction

2. Materials and methods

2.1. Sample preparation

2.2. Measuring techniques

2.2.1. Rotating electrode

2.2.2. Evaluation of the pitting or crevice corrosion (ASTM F746-87)

2.2.3. Extraction tests

2.2.4. Analysis of the solution by ICP-MS (inductively coupled plasma mass spectrometry)

2.2.5. Analysis by EDX of the test sample surface

3. Results

3.1. Open circuit potential (E_oc)

3.2. Polarization resistance (R_p) and corrosion current

3.3. Potentiodynamic polarization curves

3.4. Crevice corrosion simulation

3.5. Extraction tests

4. Discussion

4.1. The corrosion potential (E_corr)

4.2. Polarization resistance (R_p)

4.3. The potentiodynamic polarization curves

4.4. Coulometric zone analysis

4.5. Simulation of crevice corrosion

4.6. Extraction tests

4.7. Theoretical considerations concerning the polymer cement

5. Conclusion

References

3. 1. Introduction

Titanium and titanium alloys are widely used as orthopedic implants because of their favorable mechanical properties and good biocompatibility. However, some specific cases are known, where highly loaded cemented implants made from Ti alloys produced unsatisfactory results in clinical practise [1]. Prior in vitro experiments showed that micromotion alone could not explain debonding and osteolysis reported for such cemented implants [2]. The study of implant-tissue interfaces reveals for certain implants a local coloration of the tissue around the implant due to the diffusion of most probably oxides of titanium. The diffusion of titanium into the tissue is driven by chemical/electrochemical reactions that take place at the implant surface in contact with an electrolyte, such as a body fluid like the bone plasma. This indicates that the polymer cement participates in the corrosion process by ionic conduction.

The degree of the degradation of the implant surface depends on the particular surface treatment employed [3, 4, 5, 6 and 7]. Studies of the corrosion mechanisms associated with different surface treatments therefore is of great importance. Anodization is a common surface treatment for titanium implants. More recently, there was a growing interest in titanium plasma sprayed overcoats as a viable alternative to sintered bead or diffusion-bonded fiber metal surfaces, because the inherent roughness of such coatings is believed to favor the osteo-integration of the bone [8 and 9]. In the present study, we have evaluated the corrosion behavior of titanium with vacuum plasma sprayed titanium coatings (Ti VPS) and with anodized surfaces, both with and without a polymeric bone cement used in orthopedics. Electrochemical extraction tests were carried out with subsequent analysis of the electrolyte by ICP-MS in order to verify our hypothesis of the ionic permeability of the polymer.

4. 2. Materials and methods

The samples are Grade 2 titanium in disk form for the plasma spray surface (Ti VPS) and in bar form for the anodized surface.

Anodisation of the titanium was performed in an electrolyte on the base of phosphoric acid at a voltage of 58.5 V. In vacuum plasma spraying, a high-temperature plasma is expanded through a nozzle, generating a directional high-velocity (several 100 m s⁻¹) plasma jet that is directed onto the substrate (e.g. an implant). The plasma jet partially melts and projects the titanium powder particles injected into the stream onto the substrate. Upon impact, the titanium droplets deform and solidify to form the Ti VPS coating. We produced for our study the same Ti VPS coating as is typically used for implants. Such coatings consist of a dense primer layer (thickness about 50 
m) and a second less dense and rough layer (about 150 
m). The primer layer is obtained by using a fine titanium powder and a high plasma energy while the second layer is realized by using a coarser powder and a lower plasma energy. The substrate temperature does not exceed 400°C during the whole VPS process.

2.1. Sample preparation

The base materials were an anodized band and a Ti VPS coated plate of commercially pure titanium grade 2. Sample disks of 11 mm diameter were prepared by swaging the band and by electro-erosion in water of the plate. Electro-erosion was chosen in order to preserve the integrity of the crumbly plasma sprayed coating.

The anodized samples were cleaned with Deconex 16 NT (cleaner for delicate components) in an ultrasound bath, washed with ultra pure water (electrical conductivity 18 M
 cm) and rinsed with pa ethylic alcohol (Merck). The Ti VSP samples were washed in a solution of ultra pure water and pa ethylic alcohol (Merck).

The sample disks were mounted in PTFE collars adapted to the tests (Fig. 1). For the tests with bone cement, CEMEX^® Acrylic bone cements for orthopedic use (IC 005 US 006 18 044 046 051 052 G&S) was prepared following the instructions furnished by the manufacturer. Sample disks were covered with this material by:

• either manually pressing the cement into the PTFE jacket, rising 2 mm above the sample surface;

• or applying cement disks with a thickness of 0.9 mm, obtained from a bone cement mass pressed between two plates of glass.

(22K)

Fig. 1. Design of samples.

2.2. Measuring techniques

It is important to note that we will express the evaluated corrosion currents as currents (A) rather than current densities (A cm⁻²) throughout the paper. Currents from the different samples, which all have the same size, can be compared to each other, whereas the comparison of current densities would be ambiguous. The ambiguity arises from the fact that all the samples have indeed the same geometrical surface area (close to 1 cm²), but their electrochemically active surface may vary significantly. The highly porous and rough Ti VPS, in particular, exposes clearly a larger active surface as compared to the smooth anodized surface.

2.2.1. Rotating electrode

This measuring technique uses a potentiostatic PAR Model 273 A circuit. The corrosion cell is made of a special glass used for this technique. The counter-electrodes are made of platinum and the reference electrode is saturated with Calomel (SCE). The measuring system is protected by a Faraday cage.

The measurements are carried out at 300 rpm. The samples were subjected to the following measurement cycle:

• immersion in the electrolyte de-aerated with N₂ during 48 h with the recording of the open circuit potential for the last 15 h.

• tracing of a linear polarization curve (±20 mV) to calculate the polarization resistance (R_p). The scanning rate was 0.1 mV s⁻¹.

• tracing of the polarization curves (±150 mV) to calculate Tafel's slope from which the corrosion current (I_corr) is deduced [10]. The scanning rate was 0.1 mV s⁻¹.

• tracing of the global polarization curve from −1000 mV up to +1200 mV SCE at a scanning rate of 0.25 mV s⁻¹.

2.2.2. Evaluation of the pitting or crevice corrosion (ASTM F746-87)

We have also studied the behavior of corrosion at the metal-cement interface using the technique described by ASTM for the test of crevice corrosion. The test environment was an artificial bone fluid (Table 1) (Burks and Peck, French Specification NF S 90-701, ISO 10993-9) at a temperature of 37°C. The choice of this technique is based on the following reasoning: if there is diffusion of the electrolyte to the interface, then the metal surface will be subjected to the phenomenon of localized crevice corrosion. We have stopped the potentiostatic tracing at a level of 775 mV to check at the interface surface for any deterioration and for the presence of ions. Those observations would help to find out if there was in fact an electrical current passing across the interface, and if it was anodic.

Table 1. Chemical composition of artificial bone fluid according to Burks and Peck

2.2.3. Extraction tests

For the evaluation of the quantity of titanium cations diffused through the polymer, the two electrochemical extraction techniques described below were employed. The extraction milieu was a solution of NaCl 9 g l⁻¹ (ASTM F746-1998) in ultra-pure water (electrical resistivity of 18 M
 cm).

• The cyclic voltametry technique: consists of carrying out cyclic sweeps between -1000 mV (cathodic domain) and +1000 mV (anodic domain) SCE. We run 48 sweep cycles which corresponds to a test time of 12 h. We repeated the extraction test with the same sample for another 48 cycles. Therefore the total time of extraction by cyclic voltametry was 24 h.

• The potentiostatic technique (controlled potentiel coulometry): consists of carrying out an excitation to 800 mV SCE during 26 s and then setting a fixed potentiel for 36 min. We have carried out potentiostatic measurements on four fixed potential levels 600, 650, 700 and 750 mV SCE. On each level we carried out 10 measurement cycles, corresponding to a duration of 360 min. So the total extraction test lasted 1440 min (24 h). Using this technique we can easily determine the total quantity of the electrical charge consumed during the experiment, which can easily be related to the concentration of electroactive species in the cell.

2.2.4. Analysis of the solution by ICP-MS (inductively coupled plasma mass spectrometry)

The detection limits of this method are routinely in the order of 5 ppt for most heavy metals. Masses inferior to 100 are slightly more susceptible to interference than higher masses and their detection limits consequently somewhat less favorable. The system used in our measurements to determine the quantity of titanium in the electrolyte was an ICP-MS thermo-Optek PQ2+.

2.2.5. Analysis by EDX of the test sample surface

After the extraction test, we eliminated half of the cement on the disk sample (Fig. 2) to check for the presence of chloride ions on the titanium surface by EDX (energy dispersive X-ray analysis) in an electron microscope.

(35K)

Fig. 2. Titanium grade 2 sample disk (
11 mm) with a layer of bone cement. After the extraction test, half of the cement is eliminated as shown for the examination of the plasma sprayed surface.

5. 3. Results

3.1. Open circuit potential (E_oc)

Metals and alloys in an electrolytic environment generate an electrical potential, which varies in function with time. They stabilize at a stationary value after a long period of immersion. The corrosion potential thus measured allows a relative comparison of the alloy's nobility in the particular environment and to establish a galvanic series.

Titanium with its native oxide shows a particular behavior in an electrolyte. It stays at negative potential values for an extended period of immersion, meaning that the native oxide is non-passive. Hoar [11] and Fraker [12] demonstrated that the passivation of titanium is an aging process and thus depends strongly on the duration of immersion. In the electrochemical study of titanium, in order to have passive oxide layers, a pseudo-stationary regime must be attained. This regime is attained only after 6 days of immersion. Another technique to obtain a pseudo-stationary regime is to perform an anodic polarization at + 800 mV SCE during 30 min and then keeping the test sample immersed permanently in the electrolyte for all of the measurements taken [13]. The thickness of the oxide layer after the passivation is estimated 1.15-1.25 nm according to Auger and coulometric analyses [14]. In our case, we find ourselves with two surface states, as prepared by anodic oxidation and plasma spraying. Therefore, in order to maintain the integrity of the layer, the electrochemical tests were started up after 48 h of immersion. Fig. 3 displays the evolution of the open circuit potential during the last 15 h of immersion for a titanium sample with a plasma sprayed surface as compared to a same sample with additionally a bone cement disk pressed on top of it. The open circuit potential trace of another plasma sprayed sample without cement is also included in Fig. 3 for having an idea about sample-to-sample scattering.

(7K)

Fig. 3. Potential curves in an open circuit measured for the Ti VPS samples, both with and without bone cement, as well as for a natively passivated Ti sample and anodized samples.

3.2. Polarization resistance (R_p) and corrosion current

In the vicinity of the open circuit potential of corrosion (E_oc) we measured the current which results from the variation of the electrical potential imposed (E_oc±20 mV). From the so obtained polarization curves I=f(E), we calculated the polarization resistance (R_p) by determining the slope of the tangent to the curve at I=0. R_p is representative for the degree of protection of the passivation layer at the metal surface. The higher the value of R_p, the lower the corrosion current I_corr, and hence the better the metal will resist corrosion. In our case, we traced the linear polarization curves for the test samples both with and without bone cement. The corrosion currents are calculated from the polarization curves recorded in the Tafel domain. The polarization resistances (R_p) calculated from the linear polarization curves and the corrosion currents (I_corr) derived from the Tafel slopes are presented in Table 2.

Table 2. Electrochemical parameters calculated from the linear polarization curves and Tafel slopes

Experimental parameters are specified in Section 2.2.

3.3. Potentiodynamic polarization curves

Potentiodynamic polarization curves have been systematically traced after 48 h of immersion on both types of surfaces, Ti VPS and those additionally covered with the bone cement. In order to get a more precise image of the corrosion behavior of these types of surfaces, and in order to ensure reproducibility of the measuring technique, we have carried out an important number of measurements. In all, we evaluated 12 test samples (2 samples without cement, 2 samples with pressed cement and 2 with a cement disk, each for anodized and plasma sprayed surfaces). Our observations and comments are based on the totality of the current values from these recorded curves. The polarization curves for the plasma sprayed surface are shown in Fig. 4, those for the anodized surface are presented in Fig. 5.

(6K)

Fig. 4. Potentiodynamic polarization curves measured for the Ti VPS samples: as plasma sprayed, covered with pressed cement (plasma+cemp), and with a cement disk (plasma+cemd).

(6K)

Fig. 5. Potentiodynamic polarization curves measured for the anodized titanium samples: as anodized, covered with pressed cement (anod+cemp), and with a cement disk (anod+cemd).

In Fig. 6 and Fig. 7 are shown, in a comparative manner, the potentiodynamic polarization curves of the various surfaces with and without cement.

(6K)

Fig. 6. Comparison of potentiodynamic polarization curves of Ti VPS (plasma), anodized titanium (anod), and a titanium dental implant.

(7K)

Fig. 7. Comparison of potentiodynamic polarization curves of Ti VPS (plasma) and anodized (anod) titanium, covered with pressed cement (+cemp) or with a cement disk (+cemd).

3.4. Crevice corrosion simulation

The potentiostatic curves, recorded at incremental steps of +50 mV between the open circuit potential up to 775 mV SCE, are shown in Fig. 8 and Fig. 9.

(6K)

Fig. 8. Potentiostatic curves recorded during the crevice corrosion test of Ti VPS covered with a disk of bone cement. The curves were traced at incremental potential steps of 50 mV, starting at the open circuit potential.

(9K)

Fig. 9. Potentiostatic curves recorded during the crevice corrosion test of anodized titanium covered with a disk of bone cement. The curves were traced at incremental potential steps of 50 mV, starting at the open circuit potential.

3.5. Extraction tests

Fig. 10 shows the potentiodynamic curves of the cemented Ti VPS sample obtained by cyclic voltametry. The traces displayed correspond to the first and to the last (# 48) cycles. The currents measured during cycling are of the order of a few hundred nano-amperes.

(6K)

Fig. 10. Potentiodynamic curves of the cemented Ti VPS sample obtained by the cyclical voltametry technique: first cycle and last cycle (#48).

For the second extraction technique employed (the controlled potentiel coulometry), the potentiostatic curves of ten cycles at 650 mV SCE are shown in Fig. 11.

(12K)

Fig. 11. Potentiostatic curves of the cemented Ti VPS sample recorded at a level of 650 mV SCE during 36 min.

In Table 3 are summarized the quantities of electrical charges consumed in the extraction process at each potential level. These quantities were derived from the integration of the potentiostatic curves. The total electrical charge of the extraction test amounts to 6106, 6 
C.

Table 3. Electrical charge (
C) consumed during the test

The electrolyte solutions were analyzed by ICP-spectroscopy after the extraction tests. The concentrations of titanium cations found in the two tests are smaller than 0.5 
g l⁻¹ in either case.

EDX analysis of the sample surfaces before and after the extraction tests reveals the presence of chloride ions after the extraction (Table 4).

Table 4. EDX analysis of the plasma sprayed surface after the extraction test and after removal of the cement layer

6. 4. Discussion

4.1. The corrosion potential (E_corr)

The negative corrosion potentials of the bare surfaces (Table 2) confirm the hypothesis that we are not in pseudo-stationary regimes. However, when covered with cement, whether pressed or in disk form, the potential values jump into the cathodic (positive) domain. On the one hand, this switch into the cathodic domain indicates a certain "nobility" of the cemented surface. On the other hand, the feasibility of the measurement gives us quite a fundamental information, namely that the polymer is permeable to the passage of charge carriers towards the metallic surface.

4.2. Polarization resistance (R_p)

The values calculated for the polarization resistance in Table 2 deserve some comments:

• The value obtained for the bare plasma sprayed surface is very weak compared to the values calculated for the two other types of surfaces. The R_p value, when normalized to the geometric surface, would be even smaller, as the electrochemically active surface (surface BET) of the porous TiVPS is significantly larger than the geometric one. This is therefore a surface state that will not resist corrosion as good as others.

• The values obtained for the titanium dental implant and the anodized titanium surface are situated in the same domain as those we have measured with other titanium implants.

• The R_p values in the presence of cement are very high. For comparison, the R_p values of precious alloys of the same surface area are typically between 800 and 4000 k
[15].

• The differences in the R_p between the bare surfaces and those covered with the polymer are very significant. We therefore cannot reasonably represent it by a linear combination of the R_p of the metallic surface and that of the polymer surface. In our opinion, the R_p values measured in the polymer-metal system are specific to the polymer. The apparent difference between pressed cement (13 000 k
) and cement disks (22 000 k
) is due to physicochemical factors of the polymer, such as microporosity, thickness…, rather than a variability of the polymer. Though the R_p of the polymer layer is important, it remains an electrical conductor, therefore able to transport ions across the electrolyte-polymer interface and thus contributing to some degradation mechanism.

• The measurement of corrosion currents also confirm that the cement is electrically permeable.

4.3. The potentiodynamic polarization curves

The potentiodynamic polarization curves measured (Fig. 4, Fig. 5 and Fig. 7) also confirm the polymer's capacity to transport electrical charges.

If we compare the potentiodynamic polarization curves of the different surfaces (Fig. 6) we notice the relative poor resistance towards corrosion of the plasma sprayed surface. Its corrosion rate, according to the potentiodynamic curves, is of hundred times higher than that of an anodized surface. Such higher values are not in fact surprising when looking at the micrographs in Fig. 12. It is obvious that the highly porous and rough nature of the Ti VPS coating exposes a much larger titanium surface to the electrolyte than the smooth anodized surface. Similar observations were made in other studies [4]. Furthermore, as suggested by the cross sectional micrograph, crevice corrosion at the titanium-Ti VPS interface due to cracks and pores will additionally contribute to the overall corrosion. The comparison of the potentiodynamic curves of the second test sample, measured under the same experimental conditions, confirms our observation ( Fig. 13).

(56K)

Fig. 12. Scanning electron micrographs from the Ti VPS sample: surface (a) and cross section (b).

(5K)

Fig. 13. Potentiodynamic polarization curves measured on Ti VPS (plasma) and anodized (anod) samples.

The presence of the polymer cement at the surface of plasma sprayed titanium displaces the polarization curves into the range of 10⁻⁷-10⁻¹² A, values that we find for the other covered or uncovered surface states. It must be noted that the presence of polymer at the surface masks the nature of the substrates used (plasma sprayed or anodized) (Fig. 7). The potentiodynamic polarization curves are found to be in the same range of currents. The only distinction that we can establish is that the currents measured in the presence of pressed cement are lower in comparison to those measured in the presence of a cement disk. The "plasma+cemd" and "anod+cemd" curves move to the right, into the anodic domain ( Fig. 4 and Fig. 5). This difference is most probably related to different thicknesses of the cement layers. Clearly, the most crucial observation is that the cement layers transport ions to the interface just as well as bare surfaces.

4.4. Coulometric zone analysis

The coulometric analysis consists of dividing the anodic polarization curves into two distinct potential zones: Zone I: E(i=0) to +500 mV and Zone II: +500 to +1200 mV. The surfaces under the polarization curves are integrated for each zone and the obtained values are expressed in mC. In other words, we integrate the amount of current consumed for the electrochemical degradation of the sample in an anodic domain in between Ecorr and +1000 mV. The integrated values are summarized in Table 5.

Table 5. Zonal coulometric analysis by integration of the potentiodynamic polarization curves

Even though the intensity of the electrochemical degradation of an implant varies from one person to another, we consider a metal corrosion in Zone I to present a general risk while Zone II, exceptionally, presents a potential risk.

Table 5 shows clearly that the corrosion behavior of the bare surfaces depends strongly on the surface treatment. The plasma sprayed coating shows the worst behavior. The amount of electrical charge consumed for its surface degradations are of the order of hundreds of millicoulombs.

Covering with cement layers yields a substantial reduction of the amount of the electrical charge to the order of some hundreds of
C. The electrical conductivity hence is strongly diminished by the presence of the cement, although this layer remains permeable to the transport of electrical charges across the cement-metal interface.

4.5. Simulation of crevice corrosion

The potentiostatic curves (Fig. 8 and Fig. 9) also confirm that cement has the capacity to transport ions to the interface. The measured currents are in the range of 100-150 nA. The form of the curves is different. This is not probably in direct relationship to the type of substrate, but rather a consequence of differences in the homogeneity of the polymer mass. A current of some 100 nA measured at a potential of 775 mV SCE, can be considered as a low current. Metallic surfaces, whether precious alloys or steel, generally show up currents of the order of
A at that potential level. It is therefore obvious that the polymer cement limits the transport of electrical charges to the interface. The visual examination of the surfaces after the test reveals traces of humidity at the interface, showing that some of the electrolyte had passed through the polymer layer. EDX analysis of the metallic surfaces (anodized titanium and plasma spray titanium) after the elimination of the cement layer additionally reveals the presence of chlorine. There is therefore a crevice corrosion process occurring at the cement-metal interface. It will be interesting to study in the same way the transport of fluoride ions across the cement layer. Fluoride ions are known to attack titanium much stronger than chloride ions.

4.6. Extraction tests

Based on Faraday's law, which assumes that all the measured electrical charge is used to extricate the Ti²⁺ cations from the surface of the sample through the bone cement into the electrolyte, we expect a theoretical quantity of 1.515 
g of titanium in the solution, corresponding to a concentration of titanium of 75 
g l⁻¹. This theoretical concentration by far exceeds the real quantities measured by ICP-MS (< 0.5 
g l⁻¹, the equivalent of 47 
C). This discrepancy indicates that the electrical charges are mostly consumed in the transport through the cement by other ions (sodium and chloride) and probably also by the growth of the titanium passive layer composed of oxides and/or other chemical species.

4.7. Theoretical considerations concerning the polymer cement

All the various results point to one same conclusion; that the polymer immerged in an electrolytic environment has the capacity to transfer electrical charges. We suppose several different mechanisms:

• The polymer absorbs water into its structure. In this way, the ionic migration will follow the laws of ion transport, driven by an electrical field generated by the potential difference between the measuring electrode and the counter-electrode. The infrared spectroscopic analysis of the polymer reveals small absorption bands at 3439, at 2153 and at 1634 cm⁻¹. These bands correspond indeed to vibrations due to the absorption of water (Fig. 14).

• Examination of the pressed cement surface (Fig. 15) reveals the presence of crevices (cracks) and holes. Micro-crevices and porosity of the polymer cement is probably generated during manufacturing, handling or use. Other factors may also intervene, such as kinetic polymerization, the temperature, the pH, chemical degradation processes, the presence of monomers, dimers, etc. The pores, however, are not very numerous; on a surface of 0.98 cm², we counted between 20 and 30 pitting marks or holes. Transversal cuts (Fig. 16) do not provide any evidence that these holes and cracks would completely run through the entire thickness of the cement. However, numerous closed pores are found, together with some barium sulfate inclusions as evidenced by EDX analysis. The polymer surface, before embedding, was protected by gilding in order to eliminate any diffusion of resin towards the cement (Fig. 16d). We notice in Fig. 16c that the gilding goes around the porosity. This suggests that the porosity is due to the removal of an inclusion, possibly barium sulfate crystals, that existed before the gilding. Another part of these pores probably originate from the cement preparation.

• The fact that we found the metallic surface to be wet after the crevice test leads us to suppose that also an electro-kinetic phenomenon took place, more precisely, a phenomenon of electro-osmosis. The electro-kinetic potential, or the
-potential, expresses the potential difference between the surface of the solid and the mass of the liquid, whereas the electrochemical potential
is the potential difference between the metal mass and the mass of the liquid. The two potentials may either be of the same sign or of opposite signs. The electrolyte may have an influence on the electro-kinetic potential values. This influence is essentially due to cations if the wall of the solid is negatively charged, and to the influence of anions if the wall is positively charged. For the polymers, the electro-kinetic potentials are situated between 30 and 50 mV. Therefore, electrolyte can be transferred in the polymer-metal interface by imposing an electrical field [16]. This could explain the transport of chlorine ions to the titanium surface during the crevice corrosion test ( Fig. 17).

• The cement may also be considered as an organo-mineral solid solution with ionic conduction, in other words, a solid electrolyte with a relatively high rate of ionic conductivity at room temperature. A solid electrolyte is made up of phases, which are formed by mixing and dissolving mineral salts in a polymer. Barium sulfate is used as a component in the manufacturing of solid electrolytes. In comparison with classical electrolytes, the polymer electrolytes have the peculiarity of exhibiting at once cationic and anionic conduction. This explains the possibility of tracing potentiodynamic curves in both, the cathodic and the anodic domain for test samples covered with cement with thicknesses of 2 mm. The amplitude of the passing currents depends upon the intrinsic properties of the phases (structure of the solid electrolyte) and of the electrical potential imposed. As for the cement in question, a better understanding of its "electrical circuit" could be gained by using spectroscopic impedance techniques [17]. An electrochemical insertion technique would probably be useful as well in the present case.

(8K)

Fig. 14. FT-IR (ATR) spectra of cement disks after immersion in the electrolyte.

(35K)

Fig. 15. Electron microscope photograph of a pressed cement surface after crevice corrosion testing.

(96K)

Fig. 16. Transversal section of a titanium sample with a plasma sprayed coating and covered with a cement disk.

(46K)

Fig. 17. EDX analysis of the Ti VPS under cement after crevice corrosion testing. Note the presence of chlorine.

The cement is a mixed conductor. The electrical exchanges at each of the interfaces could therefore show up in one of the following manners (Fig. 18):

• the exchange at the cement-electrolyte interface;

• the transfer into the cement layer;

• the exchange at the cement-titanium interface.

(10K)

Fig. 18. Schematic representation of the exchange process in the metal-cement-electrolyte interfaces.

7. 5. Conclusion

The surface treatment plays a dominant role in the corrosion resistance of titanium. The electrochemical tests performed reveal that the plasma sprayed titanium coatings present the least corrosion resistance among the various surfaces investigated. Crevice corrosion can generate titanium particles that are liberated into the electrolyte. Clinical studies as to the migration of titanium particles from an implant into the body tissues have already been performed [18, 19, 20 and 21].

Even when covered with cement, as shown in this study, there is definitely a corrosion process at work at the titanium underneath, whether it is anodized or plasma sprayed. The cement, in spite of its strongly reduced electrical conductivity in comparison to a metal, is an ionic conductor, and therefore capable of participating in the corrosion process. The corrosion manifests itself in the release of titanium cations and in the growth of a titanium passive layer. Also, crevice corrosion can occur in the metal-cement interface.

In summary, the bone cement does reduce but not suppress the corrosion of the underlying titanium. With its electrolyte-polymer and polymer-metal interfaces, cemented samples constitute a complex system wherein the corrosion process must be attributed to several different causes. Conjectures about any long-term consequences of the observed corrosion remain purely speculative and need further investigation. On the one hand, if we extrapolate the quantity of titanium released over a period of 10 years, we obtain a total release of 1.8 mg. On the other hand, as we have seen, the electrical charge consumed through the cement is not really related to the formation and migration of titanium cations, but rather to the growth of the titanium passive layer.

8. References

1. H.G. Willert, L.G. Broback and G.H. Buchhorn, Crevice corrosion of cemented titanium alloy stems in total hip replacements. Clin Orthop (United States) 333 (1996), pp. 51-75. Abstract-MEDLINE | Abstract-EMBASE   | Full Text via CrossRef

2. Weber W, Rieger W, Clausen J, Schmotzer H, et al., Zirconia-Alumina: An Alternative Bearing for Hip Arthroplasty Bioceramics—Conference, vol. 13, 2001. p. 533-6.

3. C. Aparicio, F.J. Gil, C. Fonseca, M. Barbosa and J. Planell, Corrosion behaviour of commercially pure titanium shot blasted with different materials and sizes of shot particles for dental implant applications. Biomaterials 24 (2003), pp. 263-273. SummaryPlus | Full Text + Links | PDF (541 K)

4. M. Cabrini, A. Cigada, G. Rondelli and B. Vicentini, Effect of different surface finishing and of hydroxyapatite coatings on passive and corrosion current of Ti6Al4V alloy in simulated physiological solution. Biomaterials 18 (1997), pp. 783-787. SummaryPlus | Full Text + Links | PDF (657 K)

5. C.H. Ku, D.P. Pioletti, M. Browne and P.J. Gregson, Effect of different Ti-6Al-4V surface treatments on osteoblasts behaviour. Biomaterials 23 (2002), pp. 1447-1454. SummaryPlus | Full Text + Links | PDF (243 K)

6. Z. Cai, H. Nakajima, M. Woldu, A. Berglund, M. Bergmann and T. Okabe, In vitro corrosion resistance of titanium made using different fabrication methods. Biomaterials 20 (1999), pp. 183-190. SummaryPlus | Full Text + Links | PDF (153 K)

7. G. Chen, X. Wen and N. Zhang, Corrosion resistance and ion dissolution of titanium with different surface microroughness. Biomed Mater Eng 8 (1998), pp. 61-74. Abstract-EMBASE | Abstract-MEDLINE  

8. W.C. Head, T.H. Mallory and R.H. Emerson, The proximal porous coating alternative for primary total hip arthroplasty. Orthopedics 22 (1999), pp. 813-815. Abstract-MEDLINE | Abstract-EMBASE  

9. R.B. Bourne, C.H. Rorabeck, B.C. Burkart and P.G. Kirk, Ingrowth surfaces—plasma spray coating to titanium alloy hip replacements. Clin Orthop Rel Res 298 (1994), pp. 37-46. Abstract-MEDLINE | Abstract-EMBASE  

10. F. Mansfeld, The polarization resistance technique for measuring corrosion currents. In: M.G. Fontana and R.W. Staehle, Editors, Advances in corrosion science and technology vol. 6, Plenum, New York (1976), pp. 164-263.

11. T.P. Hoar and D.C. Mears, Corrosion-resistant alloys in chloride solution: materials for surgical implants. Proc R Soc (London) SCRA294 1439 (1966), pp. 486-510.

12. Fraker AC, Ruff AW, Yeager MP. Corrosion of titanium alloys in physiological solutions. Proceedings of the Second International Titanium Conference, Plenum, NY, 1973. p. 2447-57.

13. Meyer J-M, Reclaru L, Ecuvillon M. Measurement and evaluation of localized corrosion phenomena between titanium implant and dental alloys. In: Kovacs P, Istephanous NS, editors. Proceedings of the Symposium on Compatibility of Biomedical Implants, San Francisco, May 23-25, 1994. Pennington, NJ, USA: The Electrochemical Society Inc., 1994. p. 295-306.

14. B. Grogosgeat, L. Reclaru, M. Lissac and F. Dalard, Measurement and evaluation of galvanic corrosion between titanium TiAl6V4 implants and dental alloys by electrochemical techniques and Auger spectrometry. Biomaterials 20 (1999), pp. 933-941.

15. L. Reclaru and J-.M. Meyer, Zonal coulometric analysis of the corrosion resistance of dental alloys. J Dent 23 (1995), pp. 301-311. SummaryPlus | Full Text + Links | PDF (6153 K)

16. Emschwiller G. Chimie physique, vol. II. Paris, France: Presses Universitaires de France, 1961. p. 903-10.

17. Deportes C, Duclot M, Fabry P, Fouletier J, Hammou A, Kleitz M, Siebert E, Souquet J-L, Electrochimie des Solides. Grenoble, France: Presses Universitaires de Grenoble, 1994. p. 182-6, 300-1, 308-20.

18. E.A. Salvati and F. Doty, Particles of metallic debris in cemented total hip artoplasty. Clin Orthop 293 (1993), pp. 160-170.

19. Y. Nakashima, D.H. Sun and R.L. Maloney Wj, Goodman SB, Schurman DJ, Smith, Introduction of matrix metalloproteinese expression in human macrophages by orthopedic particulate debris in vitro. J Bone Surg Br 80 4 (1998), pp. 694-700. Abstract-MEDLINE | Abstract-EMBASE   | Full Text via CrossRef

20. M. Manlapz, W.J. Maloney and R.L. Smith, In vitro of human fibroblast by retrieved titanium alloy wear debris. J Orthop Res 14 4 (1996), pp. 465-472.

21. A. Han Ch, Johansson CB, Wennerberg A, Albrektsson, Quantitative and qualitative investigations of surface enlarged titanium and titanium alloy implants. Clin Oral Implants Res 9 (1998), pp. 1-10.

Corresponding author. Tel.: [phone]; fax: [phone]

---