Distribution of carbon in polycrystalline copper surfaces treated by methane plasma immersion ion implantation

Distribution of carbon in polycrystalline copper surfaces treated by methane plasma immersion ion implantation

Nuclear Instruments and Methods in Physics Research B 267 (2009) 1531–1535 Contents lists available at ScienceDirect Nuclear Instruments and Methods...

3MB Sizes 0 Downloads 12 Views

Nuclear Instruments and Methods in Physics Research B 267 (2009) 1531–1535

Contents lists available at ScienceDirect

Nuclear Instruments and Methods in Physics Research B journal homepage: www.elsevier.com/locate/nimb

Distribution of carbon in polycrystalline copper surfaces treated by methane plasma immersion ion implantation S. Flege a,*, G. Kraft a, K. Baba b, R. Hatada a,b, W. Ensinger a a b

Technische Universität Darmstadt, Department of Materials Science, Darmstadt, Germany Industrial Technology Center of Nagasaki, Applied Technology Division, Omura, Nagasaki, Japan

a r t i c l e

i n f o

Article history: Available online 3 February 2009 PACS: 52.77.Dq 68.49.Sf 82.80.Ms

a b s t r a c t Copper substrates with rather large grains (dimensions of some mm) treated by pulsed methane plasma immersion ion implantation exhibit a laterally inhomogeneous carbon distribution. Areas with high carbon X-ray intensity in electron probe microanalysis coincide with thicker carbon containing layers as seen in sputter depth profiling via secondary ion mass spectrometry. The distribution of carbon within the surface depends not only on the treatment time and pulse repetition rate but also on the orientation of the individual copper grains. Ó 2009 Elsevier B.V. All rights reserved.

Keywords: Plasma immersion ion implantation SIMS EPMA

1. Introduction In plasma immersion ion implantation (PIII) the surface of a substrate material is altered by deposition or incorporation of the species existent in the generated plasma. While the type of alteration can be controlled via the instrumental setup – film deposition can be avoided when the sample is pulse biased with a high voltage which ignites the plasma – the extent of incorporation into the surface is influenced by a variety of parameters [1]. Using a methane atmosphere, the plasma consists of charged hydrocarbons and fragments thereof [2] which are accelerated towards the sample due to the negative sample bias of some kV to some ten kV. As a result, there is on the one hand implantation into the surface; simultaneous sputtering of the surface on the other hand, however, cannot be neglected as the energy range used is precisely suitable for surface removal and is thus common in sputter depth profiling. While the implantation leads to an enrichment of carbon within the surface, the sputtering partly removes it at the same time. There are several factors that influence the sputter rate – and also the implantation depth – like mass, energy and angle of incidence of the impinging particles, as well as the mass, binding energy and crystallographic orientation of the substrate material itself [3]. This means that with flat samples of a polycrystalline material a variation of the resulting depth of the carbon containing layer can be expected depending on crystal * Corresponding author. Tel.: +49 6151166376; fax: +49 6151166378. E-mail address: fl[email protected] (S. Flege). 0168-583X/$ - see front matter Ó 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.nimb.2009.01.126

orientation. For this effect to be readily observable the sample needs to exhibit a high sputter rate that depends strongly on the crystal orientation. It must not bind the carbon tightly inside the surface, i.e. the material has to be a non-carbide former. And the grain size must not be too small. Thus, copper was chosen as the substrate material. Usually, a non-uniform manipulation of the surface is to be avoided as the purpose of the alteration of the surface is an improvement in regard to tribological properties or corrosion resistance [4,5]. In fact, several other materials (carbide formers with lower sputter rates) that were treated under the same conditions as the copper samples here show laterally homogeneous carbon incorporations [6–8]. 2. Experimental Polished polycrystalline copper samples with a size of 1  1 cm2 were fixed onto a sample holder inside the PIII vacuum chamber. The base pressure was 104 Pa, the added methane increased this up to about 1 Pa. The sample holder was pulse biased with a high voltage of 20 kV, the pulse length was 10, 5 and 2.5 ls with a repetition rate of 1, 2 and 4 kHz, respectively and process times were 0.5, 1 and 2 h. In order to estimate the grain size, one copper substrate was exemplarily etch-polished for 1 min with an iron(III)-nitrate solution (2.5 g in 25 ml water and 25 ml alcohol). Then the distribution of the grain boundaries was observed by using an optical microscope.


S. Flege et al. / Nuclear Instruments and Methods in Physics Research B 267 (2009) 1531–1535

The investigation of the lateral distribution of carbon and copper was done by electron probe microanalysis (EPMA) with wavelength dispersive detection (WDX) (instrument: Cameca SX 50, 10 kV accelerating voltage, 40 nA beam current). Mapping of the element distribution was performed via stage-scan with 512  512 pixels, a step size of 20 lm and a measurement time of 200 ms per pixel. The depth distribution of several elements was recorded by secondary ion mass spectrometry (SIMS) (instrument: Cameca ims 5f, 8 keV Oþ 2 primary ions, 20 nA beam current). The measurement area for the positively charged secondary ions was a circle with 60 lm diameter out of a rastered area of 150  150 lm2. Timedepth conversion was done using the end depths of the generated craters as measured with a profilometer (Dektak IIA).

covering of this area from the fastening mechanism inside the preparation chamber.) The same distribution can be recognized in the image of the secondary electrons from EPMA (Fig. 3, color bar adjusted for good contrast), with all the areas that were dark grey in the optical image also appearing dark in this image. This is due to the enrichment of carbon in these areas, the probability of secondary electron emission being proportional to the atomic number and the surface sensitivity of secondary electron emission. In the lateral distribution of the characteristic copper X-rays (Fig. 4) there is hardly any variation because of the considerably higher emission depth of the X-rays. The respective carbon image as recorded by EPMA (Fig. 5(c)) however shows a clear distinction that is naturally the opposite of Fig. 3 with areas of high carbon content indicated by

3. Results and discussion The grain distribution can be seen in Fig. 1 from the etch-polished substrate; the grains are mostly elongated with a length up to several mm. Even with the naked eye some inhomogeneity on the surface can be detected after PIII preparation in the form of discolorations towards grey, see Fig. 2 for the sample with the preparation parameters 1 kHz and 2 h. (The semi circle at the lower edge is due to the

Fig. 3. Secondary electron image by EPMA of the sample shown in Fig. 2.

Fig. 1. Microscope image of a small section of one copper substrate after etchpolishing.

Fig. 2. Optical image of one sample after PIII preparation.

Fig. 4. Distribution of copper by EPMA.

S. Flege et al. / Nuclear Instruments and Methods in Physics Research B 267 (2009) 1531–1535

a high brightness. The surface sensitivity here stems from the fact that the carbon is only present close to the surface (see text below). There is also a dependence of the carbon intensity on the process time as well as on the repetition rate. While doubled process time equals approximately twice the average intensity in the EPMA measurements, the dependence on the repetition rate is not linear because of the tendency of the plasma source towards shorter pulses when using higher repetition rates due to some inherent power limitations; nevertheless the dependence is also evident. In Fig. 5 the carbon images for the same product of process time and repetition rate, i.e. 0.5 h/4 kHz, 1 h/2 kHz and 2 h/1 kHz are shown with all three color bars representing the same range of intensity values. While the first two images show roughly the same variation in intensity, only the sample with the longest process time exhibits the highest absolute intensities. (The areas with high intensity at the right and the bottom of the images are due to the


adhesive carbon discs used for fastening the samples for the EPMA measurements, the small bright spots scattered over the sample are probably due to some voids within the surface as a result of the polishing process leading to preferential enrichment of carbon not least because of surface contaminations.) Several spots on the different samples underwent depth profiling analysis; the samples with 0.5 and 1 h process time were only probed at two locations, one that appeared bright in the EPMA image and one dark one, whereas the samples with 2 h were investigated on up to seven positions associated with different intensity levels in the EPMA images. Generally, the copper shows a profile like the one given in Fig. 6: after some signal variation at the beginning – that is primarily influenced by the oxygen level introduced by the primary ions – the signal reaches a constant value after several hundred seconds corresponding to several ten nm. The carbon signal (in the form

Fig. 5. Distribution of carbon by EPMA, same color bar for all samples, process time (h)/repetition rate (kHz) (a) 0.5/4 (b) 1/2 (c) 2/1 (A–E denote spots chosen for depth profiling).


S. Flege et al. / Nuclear Instruments and Methods in Physics Research B 267 (2009) 1531–1535

Fig. 6. SIMS depth profile of spot B in Fig. 5(c).

Fig. 7. Carbon signals from the depth profiles of spots A–E in Fig. 5(c).

of 12C, and the combined signals 12C1H and 63Cu12C) either decreases from the beginning or shows an initial plateau and falls off after that depending on the particular measurement spot. There is a correlation of the carbon X-ray intensities in the EPMA images and the thickness of the carbon containing layers: in Fig. 7 five carbon signals are shown that were compiled from the depth profiles in the spots denoted A–E in Fig. 5(c) (sample with 1 kHz and 2 h). As the average brightness level of the respective areas increases (in the order A – representing the unaltered surface that was not exposed to the plasma –, E, D, C, B), so does the thickness, from 22 nm in spot E up to 40 nm in spot B (taking the drop of the carbon signal down to 10% of its maximum value as a delimiter for the thickness and interpolating the thickness linearly from the end depth value of the crater). The hydrogen signal (1H) shows the same trend as the carbon signal for thin carbon layers but develops

a tendency towards a pronounced step on the trailing edge with thicker layers.

4. Conclusions Polycrystalline copper samples exhibit non-homogeneous carbon distributions after PIII preparation. This is evident by discoloration as well as in the EPMA images of the carbon distribution. Since the sizes of the neighboring areas with even brightness levels match the ones of the grains, it can be concluded that certain properties that depend on the crystal orientation are responsible for this, namely the sputter rate. Areas with a high sputter rate exhibit thinner carbon containing layers as much of the implanted carbon is removed again. This effect is striking with this set of

S. Flege et al. / Nuclear Instruments and Methods in Physics Research B 267 (2009) 1531–1535

samples because of the large grain sizes and the high sputter rate of copper. While this correlation of grain orientation and resulting thickness of the carbon containing layer appears valid, a proof would actually be possible by visualizing the orientations of individual neighboring grains by electron backscatter diffraction (EBSD) mappings – such measurements are planned. References [1] A. Anders, Handbook of Plasma Immersion Ion Implantation and Deposition, Wiley-VCH, Berlin, 2000.


[2] W. Ensinger, in: R. Wei (Ed.), Plasma Surface Engineering Research and its Practical Applications, Research Signpost, Trivandrum, 2008, p. 135. [3] R. Behrisch (Ed.), Sputtering by Particle Bombardment I, Springer, Berlin, 1981. [4] B. Bhushan, B.K. Gupta (Eds.), Surface treatment by ion beams (Chapter 12), Handbook of Tribology, McGraw-Hill, New York, 1991, p. 12.1. [5] Y. Pauleau (Ed.), Corrosion and wear resistant coatings formed by ion beam techniques (Chapter 18), Materials Surface Processing by Directed Energy Techniques, Elsevier, 2006, p. 595ff. [6] K. Baba, R. Hatada, S. Flege, G. Kraft, W. Ensinger, Nucl. Instr. and Meth. B 257 (2007) 746. [7] K. Baba, R. Hatada, S. Flege, G. Kraft, W. Ensinger, Surf. Coat. Technol., accepted for publication. [8] G. Kraft, S. Flege, K. Baba, R. Hatada, W. Ensinger, Physica Status Solidi a 205 (2008) 985.