Surface Modification by Ion Implantation to Improve the Oxidation Resistance of Materials for High Temperature Technology

The intermetallic Titaniumaluminides are expected to have a high potential as material in high temperature technology. Their high specific strength at temperatures above 700°C offers the possibility for manufacturing components of aerospace and automotive industries. With a specific weight of 50% of that of the widely used Ni-based superalloys TiAl is very suitable for fast rotating parts like turbine blades in aircraft engines and land based power stations, exhaust valves or turbocharger rotors. Thus lower mechanical stresses and a reduced fuel consumption are expected. In contrast to these benefits TiAl shows insufficient oxidation resistance at temperatures above 750°C (Rahmel et al., 1995). To reach higher service temperatures a protective alumina scale would be needed. A surface modification avoids any detrimental influence on the excellent mechanical properties of the material. By using the “halogen effect” a dense protective alumina scale was formed after doping the metal surface with small amounts of chlorine (Kumagai et al., 1996; Schütze & Hald, 1997; Donchev et al., 2003; Schumacher et al., 1999a; Schumacher et al., 1999b; Hornauer et al., 1999). The halogen effect can be explained by a thermodynamic model assuming the preferred formation and transport of volatile Alhalides through pores and microcracks within the metal/oxide interface and their conversion into alumina, forming a protective oxide scale on the surface (Donchev et al., 2003). However the alumina scale fails during thermocyclic loading of Cl-implanted TiAlsamples. Based on this model comprehensive calculations have been performed for fluorine and TiAl predicting a positive effect. The results of thermodynamical calculations have to be transformed into F-concentrations. The beam line ion implantation is chosen as a method of F-application because of its accuracy and reproducibility. By varying the implantation parameters optimal conditions have to be determined to meet the region of F-amounts necessary for a positive F-effect. The implanted F-depth profiles can be verified by the non-destructive ion beam analysis method PIGE (Proton Induced Gamma-ray Emission). The alumina scale formed via the F-effect is adherent even under thermocyclic conditions. However technical use is only possible if the fluorine effect can be stabilized for a time of at least 1000 hours. Hence after the alumina scale formation the stability of oxidation protection depends on the behaviour of the implanted fluorine vs.


Introduction
The intermetallic Titaniumaluminides are expected to have a high potential as material in high temperature technology. Their high specific strength at temperatures above 700°C offers the possibility for manufacturing components of aerospace and automotive industries. With a specific weight of 50% of that of the widely used Ni-based superalloys TiAl is very suitable for fast rotating parts like turbine blades in aircraft engines and land based power stations, exhaust valves or turbocharger rotors. Thus lower mechanical stresses and a reduced fuel consumption are expected. In contrast to these benefits TiAl shows insufficient oxidation resistance at temperatures above 750°C (Rahmel et al., 1995). To reach higher service temperatures a protective alumina scale would be needed. A surface modification avoids any detrimental influence on the excellent mechanical properties of the material. By using the "halogen effect" a dense protective alumina scale was formed after doping the metal surface with small amounts of chlorine (Kumagai et al., 1996;Schütze & Hald, 1997;Donchev et al., 2003;Schumacher et al., 1999a;Schumacher et al., 1999b;. The halogen effect can be explained by a thermodynamic model assuming the preferred formation and transport of volatile Alhalides through pores and microcracks within the metal/oxide interface and their conversion into alumina, forming a protective oxide scale on the surface (Donchev et al., 2003). However the alumina scale fails during thermocyclic loading of Cl-implanted TiAlsamples. Based on this model comprehensive calculations have been performed for fluorine and TiAl predicting a positive effect. The results of thermodynamical calculations have to be transformed into F-concentrations. The beam line ion implantation is chosen as a method of F-application because of its accuracy and reproducibility. By varying the implantation parameters optimal conditions have to be determined to meet the region of F-amounts necessary for a positive F-effect. The implanted F-depth profiles can be verified by the non-destructive ion beam analysis method PIGE (Proton Induced Gamma-ray Emission). The alumina scale formed via the F-effect is adherent even under thermocyclic conditions. However technical use is only possible if the fluorine effect can be stabilized for a time of at least 1000 hours. Hence after the alumina scale formation the stability of oxidation protection depends on the behaviour of the implanted fluorine vs.

Surface modification of TiAl by fluorine ion implantation 3.1 Principles of the halogen effect for TiAl
γ-TiAl with an Al-content between 44 and 50 at.% has a very promising prospect in high temperature application ( fig. 1). In contrast to the good mechanical properties at temperatures above 750°C the TiAl-surface is covered with a porous titania scale on top and a porous scale of fast growing mixed titania and alumina underneath ( fig. 2). This scale does not hinder the further inward diffusion of oxygen and nitrogen leading to the continuous formation of this mixed oxide scale. Already after 12 h at 900°C the scale thickness exceeds 10 µm. Finally, these processes will lead to a rapid destruction of the alloy ( fig. 2). The insufficient oxidation resistance of TiAl can be improved by using the halogen effect. The principles of the halogen effect at TiAl are illustrated in fig. 3 for fluorine.     (Donchev et al., 2003).
A preferred formation of Al-fluorides occurs, if their partial pressure is significantly higher than that of the Ti-fluorides formed. Thus, the Al-fluorides migrate to the surface where they disintegrate to F 2 and Al, which is oxidized to alumina due to the increasing oxygen partial pressure. The free gaseous fluorine is able to return into the subsurface zone to form Alfluorides again. This cycle process leads to the formation of a thin protective Al 2 O 3 -scale. Thermodynamic calculations for TiAl with Al-amounts between 40-50 at.% using the software FactSage (FactSage) show the existence of a region with significantly higher partial pressures of Al-fluorides compared to Ti-fluorides as depicted in fig. 4 (Masset & Schütze, 2008). In this case a AlF-partial pressure of at least 10 -8 bar is required to form and to transport a significant amount of Al to the surface (Fergus, 2002). By increasing the total F-partial pressure to p(F) min at first the partial pressure of AlF meets this condition and the window for the F-effect is open. If the total F-partial pressure exceeds the value of p(F) max the partial pressures of competing Tifluorides become significant. Thus the window for the F-effect is closed. The existence of a window for the F-effect at TiAl was found for temperatures between 900 and 1200°C defining a corridor for the F-effect for this temperature range (Donchev et al., 2003). Fig. 4. Partial pressure of several metal fluorides vs. F-total partial pressure for Ti60Al40 at 900°C (calculated with FactSage). The region for a positive F-effect is located between the total F-partial pressures of p(F) min and p(F) max .

F-implantation and non-destructive analysis of the implantation profiles
The corridor for the F-effect found in 3.1 is expressed in terms of fluorine partial pressures. However, for practical applications the fluorine concentration in TiAl is needed. Thus this corridor has to be transformed into a corridor for F-concentrations (in at.%). However there is no equation between F-pressures and F-concentrations to allow a direct calculation. Therefore a screening has to be performed to study the oxide scale structure after oxidation at 900°C with respect to different fluorine amounts within the near-surface region. The beam line ion implantation was chosen because of its accuracy and reproducibility. Based on TRIM-calculations (Ziegler et al., 1995)   The calculation of the F-implantation profiles by using the Monte Carlo-software T-DYN (Biersack, 1999) allows to design the ion implantation prior to the experiment. The calculated www.intechopen.com F-depth profiles between the fluences of 5 x 10 16 F cm -2 and 5 x 10 17 F cm -2 /20 keV are illustrated in fig. 6 revealing the corresponding maximal F-concentrations between 15 and 70 at.%. Due to a sputtering yield of 0.60 the saturation effects occur remarkably above a fluence of 4 x 10 17 F cm -2 . The non-destructive PIGE-technique was used to determine the fluorine concentration depth profiles within the first 1.5 μm of the ion implanted samples before and after oxidation. By using the resonance depth profiling the implantation profiles can be verified as depicted in fig. 7. A CaF 2 -single crystal is used as a standard for the quantification of fluorine in at.%. The F-maximum is shifted into a depth of 35-45 nm in order to obtain a lower loss of fluorine during the heating process. The small deviations at the surface may be caused by sputtering whereas the differences at the end of the ion range may be influenced by secondary ion collisions.

Screening of the implantation parameters
The screening consists of the following steps: 1. Calculation of the F-depth profiles by using Monte Carlo-software T-DYN. 2. F-ion implantation by using the implantation parameters (fluence, ion energy) determined in item 1. 3. Verifying of the F-depth profiles by using PIGE. 4. Oxidation of the implanted samples (24-120h/900°C/air). 5. Measurement of F-depth profiles by using PIGE. 6. Metallographic cross-section preparation. 7. Metallographic inspection of the oxide scale by using SEM/EDX. 8. Definition of the next set of implantation parameters and switch to item 1.
As a criterion for the screening the stability of the F-effect for an oxidation of at least 120h/900°C is required. Fig. 8. Cross-section of implanted γ-TiAl (5 x 10 16 F cm -2 /20 keV). The small pictures show the surface. Left image: After oxidation (12h/900°C/air) the surface is covered with a grey thin alumina scale. Right image: After oxidation (120h/900°C/air) the surface is covered with a grey alumina scale, but a mixed oxide scale with a TiO 2 -scale on top grows partly on the surface. The Ni-coating is from the metallographic preparation.
A fluence of 10 16 F cm -2 /20 keV gives a F-concentration which is too low for the F-effect. Fig. 8 illustrates the initial formation of a thin alumina scale of a TiAl-sample implanted with a fluence of 5 x 10 16 F cm -2 /20 keV. After short oxidation (12h/900°C/air) the implanted surface is covered completely with an alumina scale. However after an oxidation time of 120h the surface is covered partly by a fast growing mixed oxide scale with TiO 2 on top. After implantation of 10 17 F cm -2 /20 keV the alumina scale formed remains stable after oxidation of 120h/900°C/air ( fig. 9). To obtain a higher F-concentration a fluence of 2 x 10 17 F cm -2 /20 keV shows the best results even after long term oxidation ( fig. 10, 11). The samples implanted with fluences of 3 x 10 17 F cm -2 /20 keV and more show no sufficient alumina scale growth.    . Protective alumina scale on a F-implanted γ-TiAl-sample (2 x 10 17 F cm -2 /20 keV) after oxidation (1000h/900°C/air).
As a result of the screening an optimal set of implantation parameters (2 x 10 17 F cm -2 / 20 keV) has been obtained showing the formation of a pure and dense protective alumina scale on the TiAl-surface (Zschau et al., 2002). This fluence corresponds to a maximal initial Famount of about 44 at.% as known from the calculation with T-DYN ( fig. 6).

Fluorine profile and alumina scale formation
The non-destructive measurement of the F-depth profiles of implanted and oxidized TiAlsamples by using PIGE reveals a typical shape which can be divided into 3 characteristic regions as shown in fig. 12 (Zschau et al., 2002). Starting with a negligible F-amount on the surface and within the initially formed alumina scale the F-concentration increases to a maximum value C F max located at the metal/oxide interface. The position of the www.intechopen.com metal/oxide interface can be verified by RBS-measurements revealing the depth profiles of O, Al and Ti ( fig. 13). Within the metal the F-concentration decreases. The shape of the F-profile reveals a distinct F-enrichment in the metal/oxide-interface supporting the suggested mechanism. The alumina scale acts as a diffusion barrier and prevents the Floss via the surface. On the other hand, the fluorine diffuses into the metal. Thus, the behaviour of the F-reservoir at the metal/oxide-interface during long term oxidation determines the stability of the F-effect. Therefore a stable oxidation protection of TiAl is connected to the presence of a F-reservoir at the interface between the formed Al 2 O 3 -scale and the metal. Additionally the fluorine stimulates the repeated formation of the protective alumina scale once it has been damaged locally.

Dependence of the F-profiles during heating
The fluorine profiles implanted into TiAl undergo a rapid alteration during the heating process. Starting as a typical implantation profile determined by the fluence and the energy www.intechopen.com of the F-ions, the maximum F-concentration C F max decreases rapidly after reaching a temperature of 500°C. By using a fluence of 2 x 10 17 F cm -2 /20 keV the F-maxima drop to a value of about 5 at.% after oxidation of 1h/air as depicted in fig. 14. The pronounced F-loss is caused by the evaporation of gaseous Ti-fluorides, especially TiF 4 (Zschau et al., 2005). For temperatures up to 700°C fluorine is present within the near-surface region as a result of the transport to the surface. If the fluorine could escape also at higher temperatures, the Feffect would not be able to work. However at 800°C the oxide growth changes from a mixed Al 2 O 3 /TiO 2 -type to pure Al 2 O 3 as was proven by RBS-measurements. The alumina scale formed inhibits the outward diffusion of fluorine drastically as illustrated by the F-profiles at 900 and 1000°C in fig. 14.

Dependence of the F-concentration during long-term oxidation of TiAl
From a technical point of view the parts manufactured from TiAl have to withstand oxidation for service times of at least 1000 hours under cyclic conditions. Thus the time behaviour of the F-reservoir during long-term oxidation has been investigated by using PIGE after isothermal and cyclic oxidation at 900°C and 1000°C. The time dependence of the F-maximum C F max is chosen to characterize the stability of the F-effect. The formation of a protective alumina scale is confirmed by metallographic preparation and SEM studies. For longer oxidation times the F-maximum drops to 1-2 at.% after 500h/900°C (see fig. 17). By considering the F-profiles after isothermal oxidation up to 500h/900°C/air ( fig. 17) a nearly stable F-concentration can be expected for oxidation times > 500 h.  (1) with a constant term c 0 fits the time behaviour of C F max (2 x 10 17 F cm -2 /20 keV) during isothermal oxidation (4000 h/900 °C/air). Table 1. Parameters of the function (1) describing the C F max -dependence after single Fimplantation (2 x 10 17 F cm -2 / 20 keV) and oxidation of TiAl. The parameters for double implantation are in italics (see section 3.3.4).
In the case of isothermal oxidation at 900°C first results have been published in (Zschau & Schütze, 2008). After extending the oxidation time to 4000 h ( fig. 18) a nearly constant time behaviour of C F max is obtained as depicted in tab. 1. The SEM inspection confirms a dense protective alumina scale on the sample surface revealing that the F-effect is still working. The time dependence of C F max after passing the heating phase can be expressed by an exponential decay function starting at t=12 hours of the following type (1), where t denotes the time, t 1 a time constant, A the F-amount after the alumina scale has been established and c 0 the asymptotic F-concentration: Equation 1 suggests that after formation of a dense alumina scale C F max decreases only slightly approaching a constant value of c 0 . The fit parameters summarized in tab. 1 allow www.intechopen.com the description of the F-reservoir essential for the oxidation protection reaching a constant value of about 1 at.%. By increasing the oxidation temperature to 1000 °C the kinetics of C F max corresponds also to eq. 1. The parameters in tab. 1 indicate that a fluorine content of 0.27 at.% stabilizes the F effect even at 1000 °C at least within 1000 h. Cyclic oxidation (1 hcycles/900 °C/air) was performed up to an oxidation time of 2592 h. Before starting the cyclic process a preoxidation was performed to establish a dense alumina scale on the surface. The time behaviour above 12 h of the F-maximum in fig. 19 has been fitted by eq. 1, whereas the parameters in tab. 1 reveal the presence of about 1 at.% fluorine at the metal/oxide interface. The remaining dose shows a constant value of about 7.5 x 10 16 F cm -2 as can be seen in fig. 20.

www.intechopen.com
The results for cyclic loading at 1000 h/1000 °C indicate the stability of the F-content also under these extreme conditions as summarized in tab. 1. The F-maximum reaches values of about 0.7 at.%.
In agreement with the metallographic inspections in fig. 21 it was demonstrated that an excellent oxidation protection can be obtained already after single fluorine implantation.
The mass gain obtained by TGA-measurements depicted in fig. 22 shows the existence of an alumina kinetics i. e. a slow growing alumina scale also for cyclic oxidation (24 hcycle/1000 h/900 °C/air). The excellent oxidation protection of a TiAl-alloy via the F-effect has been demonstrated by the oxidation of a turbocharger rotor with and without fluorine treatment for 1000 h at 1050°C in air ( fig. 23). While severely corroded blades are observed for the unprotected rotor, the F-treated rotor is intact and is covered with a protective alumina scale (Donchev et al., 2006).

Increase of the F-reservoir by double F-implantation of TiAl
Despite the good results obtained for single step implanted samples in the present study the formation of an increased F-reservoir should be performed. The basic idea is that the nearsurface implanted F -after having undergone a high F-loss during heating -has to stimulate the formation of the alumina scale, whereas the deeper implanted fluorine acts as a reservoir for longer oxidation time scales.
To optimize the set of implantation parameters numerous Monte Carlo -simulations were performed by using the software T-DYN. The influence of the implantation parameters in the case of double implantation is illustrated in fig. 24. From the results two implantation strategies have been suggested. The first implantation is done with the higher energy (55 keV) followed by the second implantation with a lower energy (20 keV). In these cases the implantation profiles show a near-surface peak with a maximum located in a depth of 30-40 nm, followed by a lower second peak in a depth of about 100 -140 nm. The second strategy www.intechopen.com works in the reverse way: The high energy implantation is followed after a low energy implantation. These profiles show a more flat behaviour, so that the difference between the two maxima decreases. This can be explained by the implantation process: During the high energy implantation a part of the ions collides with the near-surface fluorine from the former implantation step leading to a smoother depth profile. In principle, by using more implantation steps a nearly rectangular implantation profile can be obtained. To choose a suitable set of implantation parameters one has to take into account that the near-surface implanted F is reduced during heating. In contrast to this the deeper implanted fluorine should be less affected by the fluorine loss. Therefore the F-reservoir requires a lower F-maximum compared to the near-surface level. Keeping this in mind the first strategy was chosen with 10 17 F cm -2 / 55 keV as first and 10 17 F cm -2 / 20 keV as the second step. The difference between single and double implantation is illustrated by the obtained F-depth profiles in fig. 25 after comparable oxidation times. Both F-maxima are located at the metal/oxide interface. While in the single-step case the F-maximum is in a depth of about 1250 nm, for the two-step sample the F-maximum occurs distinctly closer to the surface. The underlying area -this corresponds to the remaining F-dose -after double implantation is twice as high as that of the single implantation case. The near-surface region without fluorine corresponds to the alumina scale. Once formed the alumina scale acts as a diffusion barrier for fluorine and prevents further F-loss from the surface. From fig. 25 it can be concluded that the alumina growth after two-step implantation is slower compared to the single-step case. This is in agreement with the micrograph showing the cross-section in fig

Diffusion coefficient of fluorine in TiAl
After the growth of the alumina scale has been started, the remaining fluorine must be localized within the metal/oxide interface to act as a reservoir for slow alumina growth during long term oxidation protection of TiAl. Once formed, the alumina scale acts as a diffusion barrier for the remaining F into the alloy preventing a further F-loss to the surface. On the other hand, the time behaviour of the F-profiles during oxidation at 900°C, depicted in fig. 27, suggests a slow diffusion process into the metal. A basic parameter for the stability of this F-reservoir is the diffusion coefficient of F in TiAl at 900°C. There were no data available up to now. Based on the F-depth profiles after oxidation of 12, 120, 250 and 500h at 900°C illustrated in fig. 27 a first determination has been performed (Zschau et al., 2007;Zschau & Schütze, 2008) by using the diffusion equation where c is the fluorine concentration in at./cm 3 , t is the oxidation time, x the distance from the metal/oxide interface into the metal and D the diffusion coefficient in cm 2 /s. The analytical solution is given by the formula (3) (Philibert, 1990) with Q denoting the number of fluorine atoms in at./cm 2 remaining at the metal/oxide interface after oxidation of 12 h/900°C. By getting the logarithmic form of equation (3) (4) which is a linear function with respect to x 2 . By supposing that after 2 hours of oxidation the initial alumina scale has been formed and hinders the F-loss outwards, the diffusion of F into the metal is the dominating process. After the transformation of the depth profiles the fit of the parameter D leads to the results summarized in tab. 2. Table 2. Diffusion coefficients of fluorine in TiAl at 900°C calculated from the fluorine depth profiles of the implanted TiAl-samples. The mean value is D = 1.56 x 10 -15 cm 2 s -1 .

Principles of the halogen effect for Ni-base superalloys
Most technical Ni-base alloys are so-called borderline materials from an oxidation point of view, i. e. their Al-contents of 2-5 wt.% are not sufficient to form a continuous dense and protective aluminium oxide layer on the surface. Unprotected Ni-base borderline alloys form a fast growing mixed oxide scale consisting of oxides and spinels from Ni, Cr, Al and other alloying elements with high metal consumption rates and limited barrier effect against ingress of species from the operating environment. Underneath the fast growing oxide scale internal oxidation takes place leading to discrete aluminium oxide particles which do not form a continuous diffusion barrier (Litz et al., 1989;Nickel et al., 1999;Pint et al., 2006). Additionally nitrides are formed at the metal/oxide interface due to the inward diffusing nitrogen. As one possibility to protect these alloys against oxidation at temperatures above 1000°C Al-rich coatings were applied. Today this method has become a state-of-the-art process for the manufacturing of components for high temperature technology like turbine blades for gas fired power stations and aircraft engines. Typical borderline alloys are IN738 and IN939 whose compositions are given in tab. 3.

Alloy
Ni Wagner`s oxidation theory (Kofstad, 1988) predicts the formation of a dense alumina scale, if the Al-activity is sufficiently high. Based on this a detailed calculation of the minimum (or critical) Al-concentration N c for several Ni-base alloys has been performed (Zschau et al., 2009a). Three groups of alloys were considered according to their Al-content: Chromia formers (group 1), alumina formers (group 2) and bond coats (group 3). The results summarized in fig. 30 reveal a strong dependence of N c on temperature, but only a weak dependence on the chemical composition of the alloy. By considerung the alloys of group 1 the formation of a protective external Al 2 O 3 -scale is possible above the borderline (i. e. the calculated N c -curve), whereas beneath the borderline internal oxidation of Al takes place. The basic idea is to shift the borderline between alumina and chromia formers to lower Alconcentrations which allows the formation of a dense alumina scale at lower Al-contents of the bulk alloys. This can be achieved by increasing the outward mobility of Al on the surface by using the halogen effect. The halogen effect was applied already succesfully for TiAl-alloys as shown in the previous sections.   (Zschau et al., 2009a).
Following the thermodynamic considerations for TiAl the ability of the selective formation of Al-fluorides as the basic condition has to be studied. The fig. 31 shows possible reactions at the metal/oxide interface. Thermodynamical calculations were carried out for the alloys IN738 and IN939 by using the software FactSage. All calculations assume thermodynamic equilibrium and an infinite system. Because no data bases for Nibase alloys and fluorine or chlorine were available up to now, new data bases had to be developed. The calculations were performed for F and Cl and both alloys at temperatures between 900-1200°C. The results for Fluorine and IN738 at 1000°C are summarized in fig.  32, where the minimum F partial pressure p(AlF) min for a significant formation of AlF is denoted. By increasing the total F-partial pressure to p(F) min the partial pressure of gaseous AlF meets the value of p(AlF) min which opens the window for the F-effect. The effect works until the partial pressures of gaseous fluorides of other alloying elements (Ni, Cr, Co, Ti…) reach the value of the minimal partial pressure p(AlF) min too defining the partial pressure p(F) max and leading to the formation of a mixed oxide scale. Hence a positive F-effect is possible within the corridor between p(F) min and p(F) max . The results of the calculations are summarized in fig. 33 showing the behaviour of this corridor within the temperature range of 900-1200°C. From fig. 33 it can be concluded, that the F-effect is possible for IN738 within this temperature range 900-1200°C. The calculation for IN939 shows similar results. For Cl a positive effect is predicted too (Zschau et al., 2009b). However due to the good results obtained for F at TiAl the investigations have been continued with fluorine.

Screening of the implantation parameters for Ni-base superalloys
As already discussed for TiAl the calculated optimal fluorine partial pressures for the positive fluorine effect cannot be transformed into fluorine concentrations by straightforward calculations. Therefore a screening of the fluorine amount near the surface is necessary by using beam line ion implantation. The implantation profiles for fluences between 10 16 and 4 x 10 17 F cm -2 / 38 keV calculated by using the Monte Carlo software T-DYN are shown fig. 34. The energy of 38 keV meets the projected range of about 35 nm as was selected in the TiAl-case. The calculated F-profiles cover the range of fluorine amounts between 2 and 44 at%. For fluences above 4 x 10 17 F cm -2 a saturation effect was found, hence higher fluences were not considered. Following these results implantations were carried out on one side of the samples with fluences between 10 16 and 4 x 10 17 F cm -2 at a constant ion energy of 38 keV. The F-profiles obtained with PIGE (Zschau et al., 2009c) reveal the good agreement between calculated and implanted F-depth profiles ( fig. 35). The minor deviations at the end of the ion range may be caused by secondary ion collisions and the sputter process can explain the small differences near the surface.  After oxidation at 1050°C/air between 24 h and 168 h and metallographic preparation the SEM-micrographs reveal a change of the oxidation mechanism. The non-implanted side (24h/1050°C/air) shows an outer porous Cr 2 O 3 -scale and an inwards growing Al 2 O 3scale ( fig. 36a). The porous chromia rich layer as well as the inward growing rather discontinuous alumina scale do not protect the metal against the inward diffusion of oxygen and nitrogen. Hence also needle-like Ti-nitrides are observed up to a depth of 30-40 μm within the metal. In contrast to this the implanted side (1 x 10 17 F cm -2 / 38 keV) after oxidation of 60h/1050°C/air reveals a completely different behaviour ( fig. 36b): An outer Ni-Al-spinel scale covers a continuous alumina scale of 1-2 μm thickness. No internal nitrides were found. Minor inclusions of alumina are visible showing the tendency for a change from internal to external oxidation.  As a result of the screening the F-effect was found to be working between 5 x 10 16 and 4 x 10 17 F cm -2 . Optimum implantation parameters were determined to be between 5 x 10 16 and 1 x 10 17 F cm -2 . These fluences correspond to maximal F-contents of 10-18 at.%, as can be seen from fig. 34. At distinctly higher fluences the external alumina partly tends to the formation of an internal alumina scale.

Long time stability of the F-effect at Ni-base superalloys
The TGA-measurements illustrated in fig. 37 show the mass gain during long term oxidation. The TGA measurements were carried out at 1050°C for 1000 hours at samples of IN738 implanted with fluences of 5 x 10 16 , 1 x 10 17 and 2 x 10 17 F cm -2 / 38 keV. Within the first 50-100 hours all samples reveal a distinct mass gain. The non-implanted sample is characterized by following a nearly linear mass loss within the whole oxidation time. The sample implanted with 2 x 10 17 F cm -2 shows a parabolic mass gain but it did not reach slow growth kinetics. After implantation with a fluence of 1 x 10 17 F cm -2 the parabolic mass gain reaches a slow growth rate after 400 hours, but after 600 hours a slow mass loss is obtained. After implanting with a fluence of 5 x 10 16 F cm -2 the parabolic behaviour after 200 hours can be characterized by a kinetic constant of k p = 2.51 x 10 -9 g 2 cm -4 h -1 . After 900 hours a small mass loss occurs. Generally the results of the TGA curves can be influenced by several processes as formation of transient oxides, internal oxidation and nitridation, scale spallation and loss of gaseous Cr-species. The high mass gain during the first 100 h may be due to the formation of metastable transient oxides. The mass loss is most probably caused by the spallation of an outer chromia or spinel scale and/or the evaporation of Cr-based gaseous species. In order to reveal at least the final scale structure metallographic crosssections were prepared after the TGA-measurements.

Conclusions
The ion implantation of fluorine combined with the non-destructive Ion Beam Analysis have been applied to investigate the halogen effect for fluorine at γ-TiAl and Ni-base alloys. By suitable surface doping of γ-TiAl (44-50 at.% Al) with fluorine the high temperature oxidation resistance above 750°C has been improved dramatically by stimulating the formation of a dense protective alumina scale on the metal surface. The optimal F- concentrations have been defined. Recently these results have been used to apply other methods of F-application (gas phase, liquid phase, plasma immersion ion implantation) for the treatment of technical components with a complex geometry. Based on the results of the beam line ion implantation the optimal parameters can be found also for each of these process routes. The ion beam analysis, especially PIGE, plays an important role in these investigations. The fluorine effect at TiAl-alloys is close to industrial application in high temperature technology.
The Ni-base superalloys with 2-5 wt.% Al-content form a non-protective complex oxide scale during oxidation at temperatures above 900°C. After surface modification by fluorine ion implantation the oxidation mechanism can be changed dramatically to the growth of a dense protective alumina scale. Optimal F-concentrations for the F-effect for 1 st and 2 nd generation Ni-base superalloys have been determined. Recently basic work is in progress to apply the halogen effect to the 3 rd and 4 th generation superalloys and to improve the scale adherence. Additional studies focuse on the mechanism of the initial oxidation in the presence of the fluorine effect.