- Email: [email protected]
Carbon ion implantation into pure aluminium at low fluences
Surface & Coatings Technology 192 (2005) 317 – 322 www.elsevier.com/locate/surfcoat
Carbon ion implantation into pure aluminium at low fluences C.E. Foerstera,*, T. Fitzb, T. Dekorsyb, F. Prokertb, U. Kreihigb, A. Mqcklichb, E. Richterb, W. Mfllerb a Depto. de Fı´sica-UEPG, Ponta Grossa, Pr 84030-900, Brazil Institute of Ion Beam Physics and Materials Research, Forschungszentrum Rossendorf, D-01314 Dresden, Germany
b
Received 2 December 2003; accepted in revised form 9 August 2004 Available online 27 September 2004
Abstract This work presents selected results from carbon ion implantation into pure Al matrix. The carbon ions were implanted with an ion energy of 25 keV and fluences of 11021 and 21021 C+/m2 at room temperature (RT) and elevated temperature of 400 8C. Elastic recoil detection analysis (ERDA), grazing incidence X-ray diffraction analysis (GIXRD), Raman spectroscopy and high resolution electron microscopy (HRTEM) show the formation of embedded Al4C3 precipitates with carbon concentrations below its stoichiometric level. At RT ion implantation, the Al4C3 precipitates have an average grain size in the order of 2–4 nm. For carbon ion implantation at 400 8C, the precipitates grow up to approximately 20 nm in diameter and are randomly distributed in the implanted region. The carbon excess, not bound in the Al4C3 precipitates, forms highly disordered C–C clusters of approximately 0.2–0.4 nm in size. Implantation at a temperature of 400 8C reduces drastically the carbon clusters content due to the growth of the Al4C3 precipitates. D 2004 Elsevier B.V. All rights reserved. PACS: 61.82.Bg; 78.30.Am; 81.65.Lp; 82.80.Yc Keywords: Carburising; Ion implantation; Raman scattering; Aluminium carbides
1. Introduction Aluminium is a material with wide potential for industrial application due to its high strength to weight ratio, good corrosion resistance and formability. However, even in alloyed form, it presents poor mechanical and tribological properties, i.e. low hardness and wear resistance in comparison to the most common iron alloyed materials. It is well known that processes like nitriding, oxidizing and carburising can be used to improve the mechanical and tribological surface properties of metals. In recent years, nitriding has been intensively investigated as a promising method for surface modification of aluminium, improving the mechanical surface properties by the formation of an AlN layer [1–10]. However, the surface of the aluminium nitride layers is often morphological rough containing deep
* Corresponding author. Tel.: +55 42 220 3044; fax: +55 42 220 3042. E-mail address: [email protected] (C.E. Foerster). 0257-8972/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.surfcoat.2004.08.176
grooves and cracks [11,12], and therefore, it is not satisfactory for industrial application. Carburising of iron and steels is also well studied and often applied as an alternative to nitriding [13]. Consequently, in similarity to these materials, carburising can also be an alternative to improve aluminium mechanical and tribological surface performance. However, hitherto there are only a few publications on this subject. Recently, Uglov et al. [14] reported a study on the structural and phase composition changes in aluminium induced by carbon ion implantation. The experiments were done at 20 keV and carbon fluences ranging from 41020 to 41021 C+/m2 at RT condition. The authors observed that the carbon implantation leads to the formation of Al4C3 precipitates after a fluence of 21021 C+/m2. However, for the higher used fluence (41021 C+/m2), the carbon concentration was 1.5 times greater than the stoichiometric level in the Al4C3 phase. Despite the experimental difficulty to observe carbon clusters, as a result of the carbon excess, the authors suggested that besides Al4C3 synthesis, carbon
318
C.E. Foerster et al. / Surface & Coatings Technology 192 (2005) 317–322
cluster formation also takes place in the near surface region. Other authors also worked with the formation of Al–C with different purposes. Ham et al. [15] performed plasma immersion ion implantation (PIII) in aluminium using methane plasma at 40 kV with an estimated fluence of 41021 C+/m2. This process resulted in the formation of an aluminium carbide interlayer that enhances the adhesion between an aluminium substrate and a post-deposited diamond-like carbon (DLC) coating. Fariaut et al. [16,17] reported formation of Al4C3 surface layers grown on aluminium substrate using excimer laser irradiation. During this process, an excimer laser beam is focused onto the surface modified in a cell containing methane gas. As result, the vapour plasma expands from the surface and a shockwave dissociates and ionizes the gas that reacts with the aluminium surface. It was observed by the authors that the surface wear resistance increased after the treatment. Ning et al. [18] relate synthesis and structural properties of Al–C–N–O thin films produced by RF reactive diode sputtering of aluminium target in plasma of N2, O2 and CH4 gas mixtures. The aim of the work was the production of a hard composite material by selecting appropriate components, their proportions and the results showed the formation of C–N, Al–N and Al–C bonds. As it can be seen, only limited information on the topic of aluminium carburising and Al–C bonds formation can be found in the literature. Therefore, there is the necessity of a systematic study, independent of the used surface-modifying process (glow discharge, ion implantation, plasma immersion ion implantation, etc.), to understand the mechanism involved in the Al–C bond formation. As an instance, for aluminium carburising by ion implantation, parameters such as ion energy, fluence, current density and substrate temperature, as well as their influence on the surface properties, need to be better understood and correlated. With the objective to contribute to the aluminium carburising, the present study reports selected results of carbon ion implantation into aluminium substrate. Complementary surface analysis methods were used in order to correlate the substrate temperature and the ion fluence dependence on the Al4C3 synthesis and the carbon excess distribution in the near surface region.
samples were also implanted at the same implantation parameters but at a higher substrate temperature of 400 8C by using additional heating. The pressure in the vacuum chamber during the implantation was less than 1104 Pa. Elemental depth profiles were obtained by ERDA using 35 MeV Cl7+ ions at a scattering angle of 308 supplied from the Rossendorf 5 MeV Tandem accelerator. GIXRD with Cu Ka radiation at an incidence angle of 0.58 was used to identify the phases formed in the nearsurface region due to the carbon implantation. At this condition, the estimated X-ray (1/e)-penetration depth amounts to approximately 700 nm. Grain size estimation was performed by use of the Scherrer formula. In order to observe the formed chemical bonding, the samples were analyzed by Micro-Raman spectroscopy (Jobin-Yvon LabRam HR) using an incident wavelength of 532.14 nm in the range from 500 to 2200 cm1. To estimate the penetration depth of the incident Raman wavelength, ellipsometry measurements were performed to assure that the resulting spectra are representative in respect to the modified region. HRTEM was applied to observe the precipitate size and their distribution in the near surface region. For this purpose, cross-section samples have been prepared. The analysis was carried out using a Philips CM300 electron microscope, operated at 300 kV.
3. Results and discussion Fig. 1 shows the elemental profiles of carbon and oxygen obtained by ERDA at fluences of 11021 C+/m2 (Fig. 1a)) and 21021 C+/m2 (Fig. 1b)), respectively. The substrate temperature during the implantation is indicated in the figures. As expected, the carbon profiles have a typical Gaussian shape and it should be noted that the obtained profiles are slightly distorted by the depth resolution of the
2. Experimental procedure Aluminium samples in foil shape and thickness of 2103 m were mechanically polished to mirror finish surface and then cleaned by ultrasound. The implantation was carried out by the Rossendorf 200 keV Danphysic implanter fed with CO2 gas with 12C+ at ion energy of 25 keV and current density of 3 AA/cm2. The samples were implanted at fluences of 11021 and 21021 C+/m2. Under these implantation conditions, the substrate heating was limited to about 50 8C by a backside sample cooling. Some
Fig. 1. ERDA profiles for C and O obtained after carbon implantation into Al at fluences of (a) 11021 C+/m2 and (b) 21021 C+/m2 at RT and 400 8C. (4) Carbon, RT; (E) carbon, 400 8C; (o) oxygen, RT; ( ) oxygen, 400 8C.
.
C.E. Foerster et al. / Surface & Coatings Technology 192 (2005) 317–322
ERDA (about 20 nm at surface and increases towards the depth). SRIM simulation for carbon 25 keV implanted in a pure Al matrix results in a mean projected range of 73 nm and a straggling of 28 nm, without considering the presence of a natural aluminium oxide layer. In Fig. 1, it can be seen independently of the fluence that the carbon profiles after implantation at RT or 400 8C are approximately the same (both reaching a depth of about 150 nm). This fact indicates that up to this temperature the carbon atoms do not diffuse significantly into the bulk. For each nominal fluence and independent of the temperature, the areas under the profiles (retained carbon quantity) are approximately the same by considering the experimental error. At the carbon distribution peak, the maximum concentration correspond to c12 and c18 at.% for the lower and higher fluences, respectively. For both cases, the carbon concentration is lower than the stoichiometric carbon level (43 at.%) in Al4C3 phase [19]. Fig. 1 also shows an oxide layer that is always present at the sample surface. Its thickness is limited to about 100 nm (maximum oxygen peak around 30 nm in depth) and does not show any fluence dependence. In contrast to this, by increasing the substrate temperature, a slight decrease in the layer thickness and maximum concentration of oxygen is observed. Most probably, at a substrate temperature of 400 8C and under carbon irradiation, the oxygen atoms become mobile and then they can diffuse into the bulk or out from it into the vacuum. The oxygen inward diffusion to the bulk can be hindered from the carbon layer and therefore the oxygen atoms diffuse preferably to the vacuum, which leads to the observed decrease in the oxygen concentration. Furthermore, in Fig. 1(a) and (b), a clear correlation between the oxygen and carbon depth distribution can be seen too. For the case of carbon ion implantation at RT, the surface oxide layers are thicker and the carbon profiles are slightly shifted to the bulk in comparison to these obtained at 400 8C. Another possible explanation in respect to the preferential oxygen diffusion to the surface at 400 8C carbon implantation can be due to a preferential reactive oxygen sputtering by CO2 formation, which would reduce the oxygen content. However, this assumption needs future investigations. Fig. 2(a) and (b) shows GIXRD patterns after carbon implantation with 11021 and 21021 C+/m2 at RT and 400 8C as substrate temperatures. For clarity, the GIXRD patterns obtained from the samples treated at 400 8C are shifted in respect to these at RT. The Bragg’s peaks clearly indicate diffraction from the rhombohedral Al4C3 phase and fcc Al matrix. Despite the small grazing incidence angle of 0.58, reflections from the aluminium substrate are unavoidable because the thickness of the modified region is small (c150 nm). After carbon implantation at RT and for both fluences, only very weak and broad peaks around 2Hc318 ((101), (012) and (009)) and 2Hc558 ((0015)) are detectable. The carbon implantation at RT generates precipitates, which probably have a small size, and consequently, its size estimation was not possible. On the other hand, for carbon
319
Fig. 2. GIXRD pattern of aluminium samples implanted at (a) 11021 C+/m2 and (b) 21021 C+/m2 at RT and 400 8C. The latter ones are shifted for clarity. The open triangles (4) show the peak positions of the rhombohedral Al4C3 structure and the solid triangles (z) indicate the peak positions of the fcc Al structure. The intensity is plotted on a logarithmic scale.
implantation at 400 8C, the intensity of the peaks increases showing that the formation of the Al4C3 phase is favoured at high temperature implantation. However, the peaks are broad, indicating the formation of crystallites with a fine size. Using the Scherrer formula, from the peak width the average precipitate grain size was estimated to be 19F6 nm for the samples implanted at 400 8C. In Fig. 3(a), it is shown a HRTEM image from the sample implanted at 21021 C+/m2 (RT). In this figure, three grains are indicated, all of about 5 nm in size that can be probably inferred to Al4C3 precipitates. The difficulty to affirm its presence occurs because the interplanar distance for these grains is about 0.28–0.29 nm. Al4C3 structure have interspacing d(101)=0.287 nm and d(102)=0.282 nm), but Al structure also has an interspacing in the order d(110)=0.286 nm. That is also the reason why for very small grains, which were found in the case of low fluence, no reliable identification of the crystal symmetry could be
320
C.E. Foerster et al. / Surface & Coatings Technology 192 (2005) 317–322
Fig. 3. HRTEM images for samples implanted at (a) RT and (b) 400 8C with the fluence of 21021 C+/m2 showing typical grains sizes. Image 3 (b) shows Al4C3 crystallite in (100) projection.
obtained. That means it could not be decided if the grains are really precipitates. The effect of the substrate temperature at 400 8C during the carbon implantation is shown in Fig 3(b). The image shows an Al3C4 crystallite in the (100) projection. In comparison to RT implantations, the precipitates have now a bigger size (about 20 nm in diameter), which is in agreement with the GIXRD results (Fig. 2). Some more information can be obtained as shown in Fig. 4 by a bright field image taken from sample implanted with 21021 C+/m2 at 400 8C. This picture indicates an Al4C3 precipitate in first plane that grows toward to Al bulk and its length is in the order of the implanted carbon range (about 150 nm). Precipitates with lower size were observed too, in which the distribution in function of the depth is in agreement with ERDA profiles (Fig. 1) as well as in agreement with the average grain size estimated by GIXRD. Of course, lower precipitate sizes were also observed, but the precipitate distribution in function of the depth is in agreement with ERDA profiles (Fig. 1) and as well as in agreement with the GIRX size.
The fact that the Al3C4 precipitates are present close to the near surface region and also toward to the Al substrate (Fig. 4) justifies the use of Raman analysis in the present study. For the sample implanted at the lower carbon fluence, the Raman light wavelength penetrates 22 nm in depth (increasing the carbon fluence also increases the light depth penetration). In this region, probably there is simultaneously the presence of Al3C4 precipitates as observed by HRTEM analysis and C–C clusters due the carbon, which is not used to form precipitates. The Raman spectra shown in Fig. 5(a) and (b), for both carbon fluences at RT and 400 8C, reveal a very interesting behavior in two distinct energy ranges. Firstly, in the high energy region (1000–1600 cm1), typical for the sp2 C–C bonds, and secondly in the low-energy region (600–900 cm1), from which one may obtain information about Al–C and Al–O bonds. The higher energy region corresponds to highly disordered C–C sp2 bonds, which are referred in the literature as D (1350 cm1) and G (1580 cm1) peaks [20]. The G peak involves the inplane bond-stretching mode of pairs of sp2-bonded C atoms (this Raman active mode does not require the presence of sixfold rings and occurs at all sp2 sites). The D peak is forbidden in perfect graphite and becomes active only in the presence of disorder (it represents the C breathing mode and in reality it can lie in a very broad range around 1350 cm1). Competitive factors like clustering of the sp2 C–C phase and bond disorder change the shape, position and relative intensity of the G and D peaks. As mentioned above, further information can be extracted from the lower energy region representing the C–Al bond-stretching vibration mode. Infrared spectroscopy has shown that these modes are in the region around 760 cm1, but if a natural oxide layer is present, then additional peaks at 850 and 570 cm1 appear, representing Al–O bonds [18]. It can be seen in Fig. 5 that the D and G peaks change drastically their shape and intensity as a function of the carbon implantation conditions, which allows obtaining some qualitative information. Here, it is not our objective to deconvolute this broad region corresponding to the D and G
Fig. 4. Bright field image for sample implanted at 21021 C+/m2 (400 8C).
C.E. Foerster et al. / Surface & Coatings Technology 192 (2005) 317–322
Fig. 5. Raman spectra for samples implanted with (a) 11021 C+/m2 and (b) 21021 C+/m2. Implantation temperature is indicated in figures. The positions of D, G, Al–O and Al–C peaks are indicated as vertical dashed lines [21].
peaks, because some contradictions exist between authors in respect to the superposition of the sp2 C–C clusters [20,21]. For RT carbon implantation at 11021 C+/m2, the ratio I(D)/ I(G) is approximately the same, indicating the presence of small and highly disordered C–C clusters. Increasing the dose to 21021 C+/m2, the G peak intensity is predominant over D peak, indicating now an increasing cluster size and also less disorder in the C–C bonds. However, the G peak shifts to lower energy. In similarity to a:C films [18], this can be inferred by a stress component due to a high density of carbon clusters surrounded by a highly damaged Al matrix in addition to Al4C3 precipitates. The branches corresponding to Al–C and Al–O bonds show a weak intensity. They are very broad and mainly for the sample implanted with the higher fluence, shifted to lower values in comparison to literature values. These broadening and shifting can also be associated to very small and/or disordered Al4C3 precipitates in addition to the damage (stress) produced by the carbon irradiation in the surround-
321
ing Al matrix. Carbon implantation at 400 8C produces a drastic change in both energy ranges of the Raman spectra (Fig. 5), if compared to the RT implantation cases. The intensity of D and G peaks decreases at the expense of increased intensity of the peaks at the lower energy region. The Al–C peak is, despite its broadening and shifting to lower energy, well-resolved indicating the presence of aluminium carbide precipitates with a large size. This is consistent with the XRD patterns at 400 8C where the Al3C4 Bragg’s peaks are clearly present (Fig. 2) and also to the results from HRTEM analysis (Figs. 3(b) and 4). Furthermore, Fig. 5 also reveals that the carbon implantation at RT results in the formation of carbon clusters as assumed in a previous study by Uglov et al. [14]. Using the relation proposed by Ferrari and Robertson [20], to estimate carbon clusters size for a:C structures, the ratio I(D)/I(G) in our investigations gives carbon size clusters in the order of 0.2– 0.4 nm for both carbon fluences at RT. As briefly discussed above, the peaks D and G change its position according to the carbon fluence. For the lower dose of 11021 C+/m2, they obey approximately the same intensity in the correspondent energy branch indicating disordered C–C clusters. By increasing the fluence to 21021 C+/m2, a strong G component is present showing more ordered graphitic bond. The branch corresponding to Al4C3 (Al–C bond) is broad. This can be due to the disorder and the small size of precipitates as corroborated in the GIXRD patterns by the width of the Al4C3 peaks. Carbon implantation at 400 8C leads to a strong decrease in the intensity of the D and G branches and appearance of a broad branch in the low energy range (Fig. 5). This fact confirms that the high substrate temperature during the carbon implantation into aluminium favours the formation of Al–C bonds producing bigger-size Al4C3 precipitates as observed by GIXRD (Fig. 2) and HRTEM (Fig. 4). Most probably, the growing mechanism occurs due to the presence of an Al3C4 seed phase that is supplied with carbon from by the carbon clusters.
4. Conclusions Carbon implantation at RT and fluences up to 21021 C /m2 results in the formation of Al4C3 precipitates with average sizes from 2 to 4 nm. These precipitates are randomly oriented with respect to the aluminium matrix from near surface region until the carbon implanted range. The excess of carbon, which is not used to form the aluminium carbide phase, creates highly disordered C–C clusters with a very small size (V 0.5 nm). On the other side, carbon implantation at evaluated temperature (400 8C) favours the growth of Al4C3 precipitates up to a mean size of 20 nm in diameter and length in the order of the implanted carbon range. The C–C clusters presence is reduced if implantations are performed at high temperature, indicating that the growth of the Al3C4 precipitates takes +
322
C.E. Foerster et al. / Surface & Coatings Technology 192 (2005) 317–322
place at its expense. In respect to the oxygen reduction content at near surface region, this fact could be due to a preferential reactive oxygen sputtering at the temperature of 400 8C by producing CO2, or by a hindering mechanism due to the aluminium carbide presence. However, these suppositions need future investigations.
Acknowledgment We like to acknowledge the Brazilian agency CAPES for financial support of the sabbatical license of Dr. C.E. Foerster.
References [1] K.C. Walter, R.A. Dodd, J.R. Conrad, Nucl. Instrum. Methods Phys. Res., B 106 (1995) 522. [2] J.H. Booske, W.L. Zhang, W.X. Wang, K. Mente, N. Zjaba, C. Baum, J.L. Shohet, J. Mater. Res. 12 (5) (1997) 1356. [3] S. Simson, T. Reier, W. Schultze, Chr. Buchal, Surf. Coat. Technol. 83 (1996) 49. ¨ sterle, I. Dfrfel, I. Urban, T. Reier, J.W. Schultze, Surf. Coat. [4] W. O Technol. 102 (1998) 168. [5] H.K. Sanghera, J.L. Sullivan, S.O. Saied, Appl. Surf. Sci. 141 (1999) 57. [6] H.R. Stock, C. Jarms, F. Seidel, J.E. Dfring, Surf. Coat. Technol. 94– 95 (1997) 247.
[7] E.I. Meletis, S. Yan, J. Vac. Sci. Technol., A 9 (4) (1991) 2279. [8] N. Renevier, T. Czerwiec, P. Collignon, H. Michel, Surf. Coat. Technol. 98 (1998) 1400. [9] S. Kumar, P.K. Ghosh, J. Appl. Phys. 26 (1993) 1419. [10] D. Manova, S. M7ndel, B. Rauschenbach, Nucl. Instrum. Methods, B 178 (2001) 291. [11] H.-Y. Chen, H.-R. Stock, P. Mayr, Surf. Coat. Technol. 64 (1994) 139. [12] H.R. Stock, C. Jarms, F. Seidel, J.E. Dfring, Surf. Coat. Technol. 94–95 (1997) 247. [13] A. Melber, F. Schnatbaum, F. Preisser, K.-T. Rie, G. Laudin, HTM 48 (3) (1993) 183. [14] V.V. Uglov, N.N. Cherenda, A.L. Danilyuk, B. Rauschenbach, Surf. Coat. Technol. 128–129 (2000) 358. [15] A.H. Hamdi, X. Qiu, G.W. Malaczynski, A.A. Elmoursi, S. Simko, M.C. Militello, M.P. Balogh, B.P. Wood, K.C. Walter, M.A. Nastasi, Surf. Coat. Technol. 103–104 (1998) 395. [16] F. Fariaut, C. Boulmer-Leborgne, E. Le Menn, T. Sauvage, C. Andreazza, P. Andreazza, C. Langlade, Surf. Coat. Technol. 146–147 (2001) 324. [17] F. Fariaut, C. Boulmer-Leborgne, E. Le Menn, T. Sauvage, C. Andreazza-Vignolle, P. Andreazza, C. Langlade, Appl. Surf. Sci. 186 (2002) 105. [18] J. Ning, S. Xu, K.N. Ostrikov, Chai Jianwei, Li Yinang, Koh Mei Ling, S. Lee, Thin Solid Films 385 (2001) 55. [19] H.A. Wriedt, in: T. Massalski (Ed.), Binary Alloy Phase Diagrams, vol. 1, 1990, p. 177. [20] A.C. Ferrari, J. Robertson, Phys. Rev., B 61 (20) (2000) 14095. [21] T. Werninghaus, Micro-Raman Spectroscopy Investigation of Hard Coatings, Doctorate thesis, Technical University Chemnitz-Zwikau, 1997.
Carbon ion implantation into pure aluminium at low fluences C.E. Foerstera,*, T. Fitzb, T. Dekorsyb, F. Prokertb, U. Kreihigb, A. Mqcklichb, E. Richterb, W. Mfllerb a Depto. de Fı´sica-UEPG, Ponta Grossa, Pr 84030-900, Brazil Institute of Ion Beam Physics and Materials Research, Forschungszentrum Rossendorf, D-01314 Dresden, Germany
b
Received 2 December 2003; accepted in revised form 9 August 2004 Available online 27 September 2004
Abstract This work presents selected results from carbon ion implantation into pure Al matrix. The carbon ions were implanted with an ion energy of 25 keV and fluences of 11021 and 21021 C+/m2 at room temperature (RT) and elevated temperature of 400 8C. Elastic recoil detection analysis (ERDA), grazing incidence X-ray diffraction analysis (GIXRD), Raman spectroscopy and high resolution electron microscopy (HRTEM) show the formation of embedded Al4C3 precipitates with carbon concentrations below its stoichiometric level. At RT ion implantation, the Al4C3 precipitates have an average grain size in the order of 2–4 nm. For carbon ion implantation at 400 8C, the precipitates grow up to approximately 20 nm in diameter and are randomly distributed in the implanted region. The carbon excess, not bound in the Al4C3 precipitates, forms highly disordered C–C clusters of approximately 0.2–0.4 nm in size. Implantation at a temperature of 400 8C reduces drastically the carbon clusters content due to the growth of the Al4C3 precipitates. D 2004 Elsevier B.V. All rights reserved. PACS: 61.82.Bg; 78.30.Am; 81.65.Lp; 82.80.Yc Keywords: Carburising; Ion implantation; Raman scattering; Aluminium carbides
1. Introduction Aluminium is a material with wide potential for industrial application due to its high strength to weight ratio, good corrosion resistance and formability. However, even in alloyed form, it presents poor mechanical and tribological properties, i.e. low hardness and wear resistance in comparison to the most common iron alloyed materials. It is well known that processes like nitriding, oxidizing and carburising can be used to improve the mechanical and tribological surface properties of metals. In recent years, nitriding has been intensively investigated as a promising method for surface modification of aluminium, improving the mechanical surface properties by the formation of an AlN layer [1–10]. However, the surface of the aluminium nitride layers is often morphological rough containing deep
* Corresponding author. Tel.: +55 42 220 3044; fax: +55 42 220 3042. E-mail address: [email protected] (C.E. Foerster). 0257-8972/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.surfcoat.2004.08.176
grooves and cracks [11,12], and therefore, it is not satisfactory for industrial application. Carburising of iron and steels is also well studied and often applied as an alternative to nitriding [13]. Consequently, in similarity to these materials, carburising can also be an alternative to improve aluminium mechanical and tribological surface performance. However, hitherto there are only a few publications on this subject. Recently, Uglov et al. [14] reported a study on the structural and phase composition changes in aluminium induced by carbon ion implantation. The experiments were done at 20 keV and carbon fluences ranging from 41020 to 41021 C+/m2 at RT condition. The authors observed that the carbon implantation leads to the formation of Al4C3 precipitates after a fluence of 21021 C+/m2. However, for the higher used fluence (41021 C+/m2), the carbon concentration was 1.5 times greater than the stoichiometric level in the Al4C3 phase. Despite the experimental difficulty to observe carbon clusters, as a result of the carbon excess, the authors suggested that besides Al4C3 synthesis, carbon
318
C.E. Foerster et al. / Surface & Coatings Technology 192 (2005) 317–322
cluster formation also takes place in the near surface region. Other authors also worked with the formation of Al–C with different purposes. Ham et al. [15] performed plasma immersion ion implantation (PIII) in aluminium using methane plasma at 40 kV with an estimated fluence of 41021 C+/m2. This process resulted in the formation of an aluminium carbide interlayer that enhances the adhesion between an aluminium substrate and a post-deposited diamond-like carbon (DLC) coating. Fariaut et al. [16,17] reported formation of Al4C3 surface layers grown on aluminium substrate using excimer laser irradiation. During this process, an excimer laser beam is focused onto the surface modified in a cell containing methane gas. As result, the vapour plasma expands from the surface and a shockwave dissociates and ionizes the gas that reacts with the aluminium surface. It was observed by the authors that the surface wear resistance increased after the treatment. Ning et al. [18] relate synthesis and structural properties of Al–C–N–O thin films produced by RF reactive diode sputtering of aluminium target in plasma of N2, O2 and CH4 gas mixtures. The aim of the work was the production of a hard composite material by selecting appropriate components, their proportions and the results showed the formation of C–N, Al–N and Al–C bonds. As it can be seen, only limited information on the topic of aluminium carburising and Al–C bonds formation can be found in the literature. Therefore, there is the necessity of a systematic study, independent of the used surface-modifying process (glow discharge, ion implantation, plasma immersion ion implantation, etc.), to understand the mechanism involved in the Al–C bond formation. As an instance, for aluminium carburising by ion implantation, parameters such as ion energy, fluence, current density and substrate temperature, as well as their influence on the surface properties, need to be better understood and correlated. With the objective to contribute to the aluminium carburising, the present study reports selected results of carbon ion implantation into aluminium substrate. Complementary surface analysis methods were used in order to correlate the substrate temperature and the ion fluence dependence on the Al4C3 synthesis and the carbon excess distribution in the near surface region.
samples were also implanted at the same implantation parameters but at a higher substrate temperature of 400 8C by using additional heating. The pressure in the vacuum chamber during the implantation was less than 1104 Pa. Elemental depth profiles were obtained by ERDA using 35 MeV Cl7+ ions at a scattering angle of 308 supplied from the Rossendorf 5 MeV Tandem accelerator. GIXRD with Cu Ka radiation at an incidence angle of 0.58 was used to identify the phases formed in the nearsurface region due to the carbon implantation. At this condition, the estimated X-ray (1/e)-penetration depth amounts to approximately 700 nm. Grain size estimation was performed by use of the Scherrer formula. In order to observe the formed chemical bonding, the samples were analyzed by Micro-Raman spectroscopy (Jobin-Yvon LabRam HR) using an incident wavelength of 532.14 nm in the range from 500 to 2200 cm1. To estimate the penetration depth of the incident Raman wavelength, ellipsometry measurements were performed to assure that the resulting spectra are representative in respect to the modified region. HRTEM was applied to observe the precipitate size and their distribution in the near surface region. For this purpose, cross-section samples have been prepared. The analysis was carried out using a Philips CM300 electron microscope, operated at 300 kV.
3. Results and discussion Fig. 1 shows the elemental profiles of carbon and oxygen obtained by ERDA at fluences of 11021 C+/m2 (Fig. 1a)) and 21021 C+/m2 (Fig. 1b)), respectively. The substrate temperature during the implantation is indicated in the figures. As expected, the carbon profiles have a typical Gaussian shape and it should be noted that the obtained profiles are slightly distorted by the depth resolution of the
2. Experimental procedure Aluminium samples in foil shape and thickness of 2103 m were mechanically polished to mirror finish surface and then cleaned by ultrasound. The implantation was carried out by the Rossendorf 200 keV Danphysic implanter fed with CO2 gas with 12C+ at ion energy of 25 keV and current density of 3 AA/cm2. The samples were implanted at fluences of 11021 and 21021 C+/m2. Under these implantation conditions, the substrate heating was limited to about 50 8C by a backside sample cooling. Some
Fig. 1. ERDA profiles for C and O obtained after carbon implantation into Al at fluences of (a) 11021 C+/m2 and (b) 21021 C+/m2 at RT and 400 8C. (4) Carbon, RT; (E) carbon, 400 8C; (o) oxygen, RT; ( ) oxygen, 400 8C.
.
C.E. Foerster et al. / Surface & Coatings Technology 192 (2005) 317–322
ERDA (about 20 nm at surface and increases towards the depth). SRIM simulation for carbon 25 keV implanted in a pure Al matrix results in a mean projected range of 73 nm and a straggling of 28 nm, without considering the presence of a natural aluminium oxide layer. In Fig. 1, it can be seen independently of the fluence that the carbon profiles after implantation at RT or 400 8C are approximately the same (both reaching a depth of about 150 nm). This fact indicates that up to this temperature the carbon atoms do not diffuse significantly into the bulk. For each nominal fluence and independent of the temperature, the areas under the profiles (retained carbon quantity) are approximately the same by considering the experimental error. At the carbon distribution peak, the maximum concentration correspond to c12 and c18 at.% for the lower and higher fluences, respectively. For both cases, the carbon concentration is lower than the stoichiometric carbon level (43 at.%) in Al4C3 phase [19]. Fig. 1 also shows an oxide layer that is always present at the sample surface. Its thickness is limited to about 100 nm (maximum oxygen peak around 30 nm in depth) and does not show any fluence dependence. In contrast to this, by increasing the substrate temperature, a slight decrease in the layer thickness and maximum concentration of oxygen is observed. Most probably, at a substrate temperature of 400 8C and under carbon irradiation, the oxygen atoms become mobile and then they can diffuse into the bulk or out from it into the vacuum. The oxygen inward diffusion to the bulk can be hindered from the carbon layer and therefore the oxygen atoms diffuse preferably to the vacuum, which leads to the observed decrease in the oxygen concentration. Furthermore, in Fig. 1(a) and (b), a clear correlation between the oxygen and carbon depth distribution can be seen too. For the case of carbon ion implantation at RT, the surface oxide layers are thicker and the carbon profiles are slightly shifted to the bulk in comparison to these obtained at 400 8C. Another possible explanation in respect to the preferential oxygen diffusion to the surface at 400 8C carbon implantation can be due to a preferential reactive oxygen sputtering by CO2 formation, which would reduce the oxygen content. However, this assumption needs future investigations. Fig. 2(a) and (b) shows GIXRD patterns after carbon implantation with 11021 and 21021 C+/m2 at RT and 400 8C as substrate temperatures. For clarity, the GIXRD patterns obtained from the samples treated at 400 8C are shifted in respect to these at RT. The Bragg’s peaks clearly indicate diffraction from the rhombohedral Al4C3 phase and fcc Al matrix. Despite the small grazing incidence angle of 0.58, reflections from the aluminium substrate are unavoidable because the thickness of the modified region is small (c150 nm). After carbon implantation at RT and for both fluences, only very weak and broad peaks around 2Hc318 ((101), (012) and (009)) and 2Hc558 ((0015)) are detectable. The carbon implantation at RT generates precipitates, which probably have a small size, and consequently, its size estimation was not possible. On the other hand, for carbon
319
Fig. 2. GIXRD pattern of aluminium samples implanted at (a) 11021 C+/m2 and (b) 21021 C+/m2 at RT and 400 8C. The latter ones are shifted for clarity. The open triangles (4) show the peak positions of the rhombohedral Al4C3 structure and the solid triangles (z) indicate the peak positions of the fcc Al structure. The intensity is plotted on a logarithmic scale.
implantation at 400 8C, the intensity of the peaks increases showing that the formation of the Al4C3 phase is favoured at high temperature implantation. However, the peaks are broad, indicating the formation of crystallites with a fine size. Using the Scherrer formula, from the peak width the average precipitate grain size was estimated to be 19F6 nm for the samples implanted at 400 8C. In Fig. 3(a), it is shown a HRTEM image from the sample implanted at 21021 C+/m2 (RT). In this figure, three grains are indicated, all of about 5 nm in size that can be probably inferred to Al4C3 precipitates. The difficulty to affirm its presence occurs because the interplanar distance for these grains is about 0.28–0.29 nm. Al4C3 structure have interspacing d(101)=0.287 nm and d(102)=0.282 nm), but Al structure also has an interspacing in the order d(110)=0.286 nm. That is also the reason why for very small grains, which were found in the case of low fluence, no reliable identification of the crystal symmetry could be
320
C.E. Foerster et al. / Surface & Coatings Technology 192 (2005) 317–322
Fig. 3. HRTEM images for samples implanted at (a) RT and (b) 400 8C with the fluence of 21021 C+/m2 showing typical grains sizes. Image 3 (b) shows Al4C3 crystallite in (100) projection.
obtained. That means it could not be decided if the grains are really precipitates. The effect of the substrate temperature at 400 8C during the carbon implantation is shown in Fig 3(b). The image shows an Al3C4 crystallite in the (100) projection. In comparison to RT implantations, the precipitates have now a bigger size (about 20 nm in diameter), which is in agreement with the GIXRD results (Fig. 2). Some more information can be obtained as shown in Fig. 4 by a bright field image taken from sample implanted with 21021 C+/m2 at 400 8C. This picture indicates an Al4C3 precipitate in first plane that grows toward to Al bulk and its length is in the order of the implanted carbon range (about 150 nm). Precipitates with lower size were observed too, in which the distribution in function of the depth is in agreement with ERDA profiles (Fig. 1) as well as in agreement with the average grain size estimated by GIXRD. Of course, lower precipitate sizes were also observed, but the precipitate distribution in function of the depth is in agreement with ERDA profiles (Fig. 1) and as well as in agreement with the GIRX size.
The fact that the Al3C4 precipitates are present close to the near surface region and also toward to the Al substrate (Fig. 4) justifies the use of Raman analysis in the present study. For the sample implanted at the lower carbon fluence, the Raman light wavelength penetrates 22 nm in depth (increasing the carbon fluence also increases the light depth penetration). In this region, probably there is simultaneously the presence of Al3C4 precipitates as observed by HRTEM analysis and C–C clusters due the carbon, which is not used to form precipitates. The Raman spectra shown in Fig. 5(a) and (b), for both carbon fluences at RT and 400 8C, reveal a very interesting behavior in two distinct energy ranges. Firstly, in the high energy region (1000–1600 cm1), typical for the sp2 C–C bonds, and secondly in the low-energy region (600–900 cm1), from which one may obtain information about Al–C and Al–O bonds. The higher energy region corresponds to highly disordered C–C sp2 bonds, which are referred in the literature as D (1350 cm1) and G (1580 cm1) peaks [20]. The G peak involves the inplane bond-stretching mode of pairs of sp2-bonded C atoms (this Raman active mode does not require the presence of sixfold rings and occurs at all sp2 sites). The D peak is forbidden in perfect graphite and becomes active only in the presence of disorder (it represents the C breathing mode and in reality it can lie in a very broad range around 1350 cm1). Competitive factors like clustering of the sp2 C–C phase and bond disorder change the shape, position and relative intensity of the G and D peaks. As mentioned above, further information can be extracted from the lower energy region representing the C–Al bond-stretching vibration mode. Infrared spectroscopy has shown that these modes are in the region around 760 cm1, but if a natural oxide layer is present, then additional peaks at 850 and 570 cm1 appear, representing Al–O bonds [18]. It can be seen in Fig. 5 that the D and G peaks change drastically their shape and intensity as a function of the carbon implantation conditions, which allows obtaining some qualitative information. Here, it is not our objective to deconvolute this broad region corresponding to the D and G
Fig. 4. Bright field image for sample implanted at 21021 C+/m2 (400 8C).
C.E. Foerster et al. / Surface & Coatings Technology 192 (2005) 317–322
Fig. 5. Raman spectra for samples implanted with (a) 11021 C+/m2 and (b) 21021 C+/m2. Implantation temperature is indicated in figures. The positions of D, G, Al–O and Al–C peaks are indicated as vertical dashed lines [21].
peaks, because some contradictions exist between authors in respect to the superposition of the sp2 C–C clusters [20,21]. For RT carbon implantation at 11021 C+/m2, the ratio I(D)/ I(G) is approximately the same, indicating the presence of small and highly disordered C–C clusters. Increasing the dose to 21021 C+/m2, the G peak intensity is predominant over D peak, indicating now an increasing cluster size and also less disorder in the C–C bonds. However, the G peak shifts to lower energy. In similarity to a:C films [18], this can be inferred by a stress component due to a high density of carbon clusters surrounded by a highly damaged Al matrix in addition to Al4C3 precipitates. The branches corresponding to Al–C and Al–O bonds show a weak intensity. They are very broad and mainly for the sample implanted with the higher fluence, shifted to lower values in comparison to literature values. These broadening and shifting can also be associated to very small and/or disordered Al4C3 precipitates in addition to the damage (stress) produced by the carbon irradiation in the surround-
321
ing Al matrix. Carbon implantation at 400 8C produces a drastic change in both energy ranges of the Raman spectra (Fig. 5), if compared to the RT implantation cases. The intensity of D and G peaks decreases at the expense of increased intensity of the peaks at the lower energy region. The Al–C peak is, despite its broadening and shifting to lower energy, well-resolved indicating the presence of aluminium carbide precipitates with a large size. This is consistent with the XRD patterns at 400 8C where the Al3C4 Bragg’s peaks are clearly present (Fig. 2) and also to the results from HRTEM analysis (Figs. 3(b) and 4). Furthermore, Fig. 5 also reveals that the carbon implantation at RT results in the formation of carbon clusters as assumed in a previous study by Uglov et al. [14]. Using the relation proposed by Ferrari and Robertson [20], to estimate carbon clusters size for a:C structures, the ratio I(D)/I(G) in our investigations gives carbon size clusters in the order of 0.2– 0.4 nm for both carbon fluences at RT. As briefly discussed above, the peaks D and G change its position according to the carbon fluence. For the lower dose of 11021 C+/m2, they obey approximately the same intensity in the correspondent energy branch indicating disordered C–C clusters. By increasing the fluence to 21021 C+/m2, a strong G component is present showing more ordered graphitic bond. The branch corresponding to Al4C3 (Al–C bond) is broad. This can be due to the disorder and the small size of precipitates as corroborated in the GIXRD patterns by the width of the Al4C3 peaks. Carbon implantation at 400 8C leads to a strong decrease in the intensity of the D and G branches and appearance of a broad branch in the low energy range (Fig. 5). This fact confirms that the high substrate temperature during the carbon implantation into aluminium favours the formation of Al–C bonds producing bigger-size Al4C3 precipitates as observed by GIXRD (Fig. 2) and HRTEM (Fig. 4). Most probably, the growing mechanism occurs due to the presence of an Al3C4 seed phase that is supplied with carbon from by the carbon clusters.
4. Conclusions Carbon implantation at RT and fluences up to 21021 C /m2 results in the formation of Al4C3 precipitates with average sizes from 2 to 4 nm. These precipitates are randomly oriented with respect to the aluminium matrix from near surface region until the carbon implanted range. The excess of carbon, which is not used to form the aluminium carbide phase, creates highly disordered C–C clusters with a very small size (V 0.5 nm). On the other side, carbon implantation at evaluated temperature (400 8C) favours the growth of Al4C3 precipitates up to a mean size of 20 nm in diameter and length in the order of the implanted carbon range. The C–C clusters presence is reduced if implantations are performed at high temperature, indicating that the growth of the Al3C4 precipitates takes +
322
C.E. Foerster et al. / Surface & Coatings Technology 192 (2005) 317–322
place at its expense. In respect to the oxygen reduction content at near surface region, this fact could be due to a preferential reactive oxygen sputtering at the temperature of 400 8C by producing CO2, or by a hindering mechanism due to the aluminium carbide presence. However, these suppositions need future investigations.
Acknowledgment We like to acknowledge the Brazilian agency CAPES for financial support of the sabbatical license of Dr. C.E. Foerster.
References [1] K.C. Walter, R.A. Dodd, J.R. Conrad, Nucl. Instrum. Methods Phys. Res., B 106 (1995) 522. [2] J.H. Booske, W.L. Zhang, W.X. Wang, K. Mente, N. Zjaba, C. Baum, J.L. Shohet, J. Mater. Res. 12 (5) (1997) 1356. [3] S. Simson, T. Reier, W. Schultze, Chr. Buchal, Surf. Coat. Technol. 83 (1996) 49. ¨ sterle, I. Dfrfel, I. Urban, T. Reier, J.W. Schultze, Surf. Coat. [4] W. O Technol. 102 (1998) 168. [5] H.K. Sanghera, J.L. Sullivan, S.O. Saied, Appl. Surf. Sci. 141 (1999) 57. [6] H.R. Stock, C. Jarms, F. Seidel, J.E. Dfring, Surf. Coat. Technol. 94– 95 (1997) 247.
[7] E.I. Meletis, S. Yan, J. Vac. Sci. Technol., A 9 (4) (1991) 2279. [8] N. Renevier, T. Czerwiec, P. Collignon, H. Michel, Surf. Coat. Technol. 98 (1998) 1400. [9] S. Kumar, P.K. Ghosh, J. Appl. Phys. 26 (1993) 1419. [10] D. Manova, S. M7ndel, B. Rauschenbach, Nucl. Instrum. Methods, B 178 (2001) 291. [11] H.-Y. Chen, H.-R. Stock, P. Mayr, Surf. Coat. Technol. 64 (1994) 139. [12] H.R. Stock, C. Jarms, F. Seidel, J.E. Dfring, Surf. Coat. Technol. 94–95 (1997) 247. [13] A. Melber, F. Schnatbaum, F. Preisser, K.-T. Rie, G. Laudin, HTM 48 (3) (1993) 183. [14] V.V. Uglov, N.N. Cherenda, A.L. Danilyuk, B. Rauschenbach, Surf. Coat. Technol. 128–129 (2000) 358. [15] A.H. Hamdi, X. Qiu, G.W. Malaczynski, A.A. Elmoursi, S. Simko, M.C. Militello, M.P. Balogh, B.P. Wood, K.C. Walter, M.A. Nastasi, Surf. Coat. Technol. 103–104 (1998) 395. [16] F. Fariaut, C. Boulmer-Leborgne, E. Le Menn, T. Sauvage, C. Andreazza, P. Andreazza, C. Langlade, Surf. Coat. Technol. 146–147 (2001) 324. [17] F. Fariaut, C. Boulmer-Leborgne, E. Le Menn, T. Sauvage, C. Andreazza-Vignolle, P. Andreazza, C. Langlade, Appl. Surf. Sci. 186 (2002) 105. [18] J. Ning, S. Xu, K.N. Ostrikov, Chai Jianwei, Li Yinang, Koh Mei Ling, S. Lee, Thin Solid Films 385 (2001) 55. [19] H.A. Wriedt, in: T. Massalski (Ed.), Binary Alloy Phase Diagrams, vol. 1, 1990, p. 177. [20] A.C. Ferrari, J. Robertson, Phys. Rev., B 61 (20) (2000) 14095. [21] T. Werninghaus, Micro-Raman Spectroscopy Investigation of Hard Coatings, Doctorate thesis, Technical University Chemnitz-Zwikau, 1997.