Nuclear Instruments and Methods in Physics Research B 210 (2003) 531–536 www.elsevier.com/locate/nimb
Evidence for intrinsic weak ferromagnetism in a C60 polymer by PIXE and MFM D. Spemann
a,*
€hne b, T. Makarova c, , K.-H. Han b, R. Ho P. Esquinazi b, T. Butz a
a
b
Nuclear Solid State Physics, University of Leipzig, Linn estrasse 5, D-04103 Leipzig, Germany Superconductivity and Magnetism, University of Leipzig, Linn estrasse 5, D-04103 Leipzig, Germany c Department of Experimental Physics, Ume a University, S-90187 Ume a, Sweden
Abstract In this study a C60 polymer has been characterized for the first time with respect to impurity content and ferromagnetic properties by laterally resolved particle induced X-ray emission (PIXE) and magnetic force microscopy (MFM) in order to prove the existence of intrinsic ferromagnetism in this material. In the sample studied the main ferromagnetic impurity found was iron with an average concentration of 175 16 lg/g within the sample volume probed by the ion beam. However, the Fe distribution is very inhomogeneous and characterized by micrometer-large impurity grains of almost pure iron surrounded by an almost pure carbon matrix. With MFM, the ferromagnetic properties have been investigated both in pure and contaminated regions of the sample as determined by PIXE. We found that 30% of the area of pure regions (concentration of magnetic impurities <1 lg/g) is characterized by ferromagnetic domains, which are partly correlated to grains and grain boundaries. These magnetic domains are of intrinsic origin and not related to impurities. In contaminated regions of the sample the ferromagnetic properties are dominated by the impurities as expected from the high impurity concentrations. The combination of PIXE and MFM allowed us to separate between the intrinsic and extrinsic magnetic regions and to directly prove that intrinsic ferromagnetism exists in a C60 polymer. Ó 2003 Elsevier B.V. All rights reserved. PACS: 07.78.+s; 07.79.Pk; 61.72.Ss; 75.50 Keywords: Intrinsic ferromagnetism; C60 polymer; Impurity concentrations; MFM; PIXE
1. Introduction
* Corresponding author. Tel.: +49-341-97-32706; fax: +49341-97-32497. E-mail address:
[email protected] (D. Spemann). URL: http://www.uni-leipzig.de/~nfp.
The recent discovery of weak ferromagnetism in two-dimensionally polymerised highly oriented rhombohedral C60 polymers with a Curie temperature of 500 K has attracted much attention in the scientific community [1,2]. This finding is unusual since the constituent molecules have no magnetic moments. The cause of ferromagnetism (FM) is
0168-583X/$ - see front matter Ó 2003 Elsevier B.V. All rights reserved. doi:10.1016/S0168-583X(03)01092-9
532
D. Spemann et al. / Nucl. Instr. and Meth. in Phys. Res. B 210 (2003) 531–536
not yet known and further studies are necessary. This work has been motivated by two questions: What might cause the magnetic behaviour in a structure made solely by carbon? This question may be related to the ferromagnetic behaviour found in highly oriented pyrolytic graphite [3] in which it is speculated that defects contribute to the measured signal. The second question is: Do magnetic impurities contribute to the observed ferromagnetism? Although there are several experimental hints that speak against a simple contribution due to ferromagnetic impurities, no systematic studies have been realized so far in C60 . In this work we have investigated a C60 polymer by laterally resolved particle induced X-ray emission (PIXE) and magnetic force microscopy in order to check whether there is a correlation between the impurity content and the ferromagnetic properties or not.
in size and a thickness of 1 mm was embedded in transoptic resin and polished using diamond powder with 1 lm grain size. 2.2. PIXE analysis at LIPSION The impurity content of the C60 polymer was determined by PIXE using the 2.25 MeV proton microbeam at the high-energy ion nanoprobe LIPSION. The beam diameters were 0.8 and 1.3 lm depending on the desired beam current of 33 and 200 pA, respectively. The accumulated beam charge was calculated from the RBS yield assuming carbon bulk as matrix. Data analysis and quantitative elemental imaging were performed using the GeoPIXE II 2.7 program [4]. Under these experimental conditions the minimum detection limit for the metallic impurities including iron was 0.1–0.3 lg/g. 2.3. Magnetic force microscopy
2. Experimental 2.1. Sample preparation The undoped polymerized C60 phase investigated was produced in a toroid-type high-pressure apparatus at the Institute of High Pressure Physics in Troitsk (Russia). As starting material high purity, twice-sublimed C60 powder of small crystal size, with a nominal amount of metallic impurities of 22 lg/g supplied by Term USA was used. We analysed pristine C60 powder from this company using PIXE and determined the iron concentration to be 1.2 0.2 lg/g. The concentrations of the other metallic impurities were below 1 lg/g. Cylindrical samples, prepared by cold pressing of the initial powder, were wrapped in a Nb foil and placed in a boron nitride cage which screened the samples from the graphite heater. The samples were compressed at 2.5 GPa, heated to 850 °C and held isothermally, then quickly cooled (quenched) to conserve the phase obtained. The sample studied in this work shows a magnetization M ¼ 150 5 A/m at 1 T and 300 K and a coercive field HC ¼ ð19:0 0:5Þ 103 A/m. The Curie temperature was determined to be TC ¼ 500 2 K. Prior to analysis the sample with 1.4 mm 2.1 mm
Magnetic force microscopy (MFM) is a valuable tool for studying magnetic microstructures with high resolution [5–7]. This method provides the internal structure of domain walls and the spin distributions within domains as well as the general features of domain structures [8–10]. Magnetic force gradient and sample topography images were obtained simultaneously with a Nanoscope III scanning probe microscope from Digital Instruments. The microscope was operated in ‘‘tapping/liftTM ’’ scanning mode, to separate short-range topographic effects from long-range magnetic signal. The scanning probes were batch fabricated silicon cantilevers with pyramidal tips coated with a magnetic CoCr film alloy. Prior to measurement, the probe was exposed to a 3 kOe magnet which aligned its magnetization normal to the sample surface (i.e. z-direction), making the MFM sensitive to the second derivative of the z component of the sample stray field. The influence of the tip magnetization on the MFM images was negligible in all measurements as could be verified by taking images with various tip-sample separations (10–100 nm). More details about the experimental setup and MFM measurements can be found in [11].
D. Spemann et al. / Nucl. Instr. and Meth. in Phys. Res. B 210 (2003) 531–536
3. Results and discussion
533
typically by far less extended in size than the sampling depth. However, the concentration ratios
3.1. Impurity distributions and concentrations The main impurities found in the C60 polymer were silicon, calcium, and iron. Silicon and calcium are mainly concentrated on the outer surface of the sample, whereas iron as the main ferromagnetic impurity was predominantly found in the volume. As can be seen from Fig. 1 the Fe distribution is very inhomogeneous and characterized by micrometer-large grains of almost pure iron surrounded by an almost pure carbon matrix. Large regions of the sample appear pure within the information depth of PIXE which is 36 lm for Fe in carbon bulk (corresponding to 90% of the total yield). Fig. 2 shows the elemental maps of Cr, Mn and Fe in a contaminated region of the sample together with an optical micrograph of the analysed area. The contaminations consist of irregularly shaped grains with sizes of several microns containing Fe, Cr, Mn and Ni (not shown in Fig. 2) which are contained in the bulk of the sample. In fact, only the grain indicated by the arrow in Fig. 2 was optically visible. This grain was probably contained within the bulk and laid open by the sample polishing procedure. The features in the optical micrograph do not appear in the elemental maps in general, so that surface contamination due to sample polishing can be excluded. The elemental concentrations extracted from the PIXE measurements are averaged over the sampling depth of the analysing proton beam. They do not necessarily reflect the absolute composition of the impurity grains as the grains are
Fig. 1. Distribution of the main ferromagnetic impurity found in the sample: Iron. As can be seen, Fe is located in small spots at very high concentrations surrounded by large areas of almost pure carbon. The dashed line indicates the outer surface of the sample. The dotted line encircles the ‘‘pure’’ region used for the MFM measurements.
Fig. 2. Distribution of Cr, Mn and Fe in a contaminated region of the sample together with an optical micrograph from the same scan area. The impurities are located in irregularly shaped grains. Only the grain indicated by the black arrow is optically visible. All other grains are located below the sample surface.
534
D. Spemann et al. / Nucl. Instr. and Meth. in Phys. Res. B 210 (2003) 531–536
of the individual elements in the grains can be determined and were found to be similar within different grains. The typical composition in wt.% is: 97.7% Fe, 1.0% Ni, 0.7% Cr and 0.6% Mn assuming pure iron as matrix for the grains. Table 1 shows the elemental concentrations for the transoptic resin, the whole sample area shown in Fig. 1 excluding the outer surface of the sample along the dashed line, and the ‘‘pure’’ region indicated by the dotted line in Fig. 1. In general, the concentrations from sample areas have been calculated assuming carbon as matrix. Since Cr, Mn, Fe and Ni are mainly contained within grains their concentrations have been additionally calculated assuming iron as matrix. For these elements the average of the two concentration values obtained for carbon and iron matrix, respectively, has been used in Table 1 and the error values have been adjusted accordingly. From the concentration values in Table 1 it is evident, that a contamination of the sample with ferromagnetic impurities originating from the transoptic resin can be excluded. However, the average Fe concentration of 175 16 lg/g in the sample is significant higher than in the C60 powder, indicating that most of the Fe found has been incorporated during some sample preparation steps.
3.2. MFM measurements MFM measurements were performed in both pure and contaminated regions of the sample as determined from PIXE analysis using a scan size of 20 lm 20 lm for all images and a scan height (tip-sample distance) of 50 nm. Since MFM is by far more surface sensitive than PIXE, ferromagnetic signals from the ‘‘pure’’ region of the sample must be of intrinsic origin as the concentration of the ferromagnetic impurities is below 1 lg/g in this region. At these low concentrations only a paramagnetic contribution due to superparamagnetism can be expected from impurities. On the other hand, it is expected that part of the magnetic signal in contaminated areas of the sample as shown in Fig. 2 must be attributed to magnetic impurities due to their high local concentrations. Using MFM it was found that on 70% of the sample area in pure regions magnetic domains can hardly be resolved whereas on the other 30% magnetic domains of two different types have been detected. This shows that the sample consists of magnetic and non-magnetic parts. Fig. 3 shows two MFM images together with the corresponding sample topography from a magnetic part of the ‘‘pure’’ region indicated in Fig. 1. In these regions the observed contrast in the MFM images corre-
Table 1 Elemental concentrations Si S Cl K Ca Ti Cr Mn Fe Co Ni Cu Zn Nb
Transoptic resin
Whole sample area excluding outer surface
‘‘Pure’’ region in Fig. 1
107 9 5.5 0.5 4.0 0.4 <0.60 <0.40 0.3 0.1 <0.20 <0.19 0.6 0.1 <0.18 <0.21 0.6 0.1 <0.27 <2.4
678 45 29.9 1.8 11.6 1.3 5.2 0.7 13.0 0.7 0.5 0.2 0.6 0.2 1.1 0.2 175 16 0.4 0.2 6.7 3.5 27.2 1.7 1.0 0.2 4.8 1.0
629 40 12.4 0.9 7.0 0.7 2.5 0.3 4.9 0.4 <0.30 <0.19 <0.19 0.8 0.1 <0.17 0.6 0.1 15.9 0.8 0.7 0.1 31
All impurity concentrations are given in lg/g. The concentrations from sample areas have been calculated assuming carbon as matrix. Since Cr, Mn, Fe and Ni are mainly contained within grains their concentrations have been additionally calculated assuming iron as matrix. The average of the two concentration values obtained is listed and the error values have been adjusted accordingly.
D. Spemann et al. / Nucl. Instr. and Meth. in Phys. Res. B 210 (2003) 531–536
535
Fig. 3. MFM and topography images taken from the ‘‘pure’’ region (concentrations of ferromagnetic impurities <1 lg/g) indicated in Fig. 1 (scan size: 20 lm 20 lm). The images shown are examples for the two different types of magnetic domains found in 30% of the sample area in this region. A detailed description is given in the text.
sponds to the up (dark) and down (light) normal components of the magnetizations. The first type is characterized by MFM images which are different from the topography images and domain magnetizations oblique to the sample surface (see left panel of Fig. 3). For the second type both topography and MFM images are characterized by corrugated patterns as shown in the right panel of Fig. 3. Here the largest change of the magnetic force gradient obtained by MFM is always located at the edge or boundary of grains (arrows in right panel of Fig. 3), indicating a correlation to the topography. The domain magnetization seems to be oriented approximately normal to the grain surface. The magnetic domains are aligned almost parallel to each other. In contaminated regions of the sample, where iron is the main impurity, there are localized dark and bright areas in MFM images (see Fig. 4). Since the tip of the MFM is magnetized perpendicular to the sample surface, these areas correspond to outof-plane magnetized domains, whose interaction
with the tip is attractive or repulsive according to the up or down direction of their magnetic moments. The magnitude of the force gradients in contaminated regions is at least 10 times larger than that in pure regions. The spot-like features in the MFM image in the left of Fig. 4 are similar to the distribution of impurity grains. Therefore it is very likely that these features are correlated to impurity grains. No magnetic domain images have been found at or near the contaminated regions. As can be seen from the MFM images taken from pure regions of the sample, ferromagnetism is of intrinsic origin in C60 polymers. This is not only evident in view of the very low impurity concentrations in these regions, but also from the shape and size of the observed magnetic structures which differ significantly from the shape and size of the impurity grains. We may expect that the magnetic impurities contribute to the magnetism of the whole sample. However, for almost pure iron clusters a Curie temperature near TC ¼ 1000 K is expected. The measured TC ¼ 500 K for the
Fig. 4. MFM and topography images taken from a contaminated region of the sample (scan size: 20 lm 20 lm). Left: topography and MFM images from impurity grains that intersect the sample surface. Right: MFM image shows a ferromagnetic contribution from impurities which are located below the sample surface.
536
D. Spemann et al. / Nucl. Instr. and Meth. in Phys. Res. B 210 (2003) 531–536
dominating part of the magnetization and the strongly reduced ferromagnetic hysteresis above 500 K [11] are further arguments that the magnetism is independent from the impurity clusters. Additionally, magnetization measurements done in different samples do not show any correlation with Fe-impurity concentration [12]. More detailed information and discussion about the MFM measurements can be found in [11].
4. Conclusions In this study a C60 polymer has been characterized for the first time with respect to impurity content and ferromagnetic properties by laterally resolved PIXE and MFM. From the PIXE measurements both pure (concentrations of ferromagnetic impurities <1 lg/g) and contaminated areas of the sample could be selected for the MFM measurements. It could be proved directly, that intrinsic ferromagnetism exists in C60 polymer. Acknowledgements The travel support by the Federal Ministry of Education and Research of the FRG within the funding scheme ‘‘Innovative regionale Wachstumskerne’’ under grant 03WKI09 is gratefully
acknowledged. Part of this work is supported by the Deutsche Forschungsgemeinschaft within grant no. DFG Es 86/6-3. T.M. is grateful for the support of the Swedish Research Council.
References [1] T.L. Makarova, B. Sundqvist, R. H€ ohne, P. Esquinazi, Y. Kopelevich, P. Scharff, V. Davydov, L.S. Kashevarova, A.V. Rakhmanina, Nature (London) 413 (2001) 716. [2] R.A. Wood, M.H. Lewis, M.R. Lees, S.M. Bennington, M.G. Cain, N. Kitamura, J. Phys.: Condens. Matter 14 (2002) L385. [3] P. Esquinazi, A. Setzer, R. H€ ohne, C. Semmelhack, Y. Kopelevich, D. Spemann, T. Butz, B. Kohlstrunk, M. L€ osche, Phys. Rev. B 66 (2002) 024429. [4] C.G. Ryan, Nucl. Instr. and Meth. B 181 (2001) 170. [5] J.J. Saenz, N. Garcıa, P. Gr€ utter, E. Meyer, H. Heinzelmann, R. Wiesendanger, L. Rosenthaler, H.R. Hidber, H.-J. G€ untherodt, J. Appl. Phys. 62 (1987) 4293. [6] Y. Martin, H.K. Wickramasinghe, Appl. Phys. Lett. 50 (1987) 1455. [7] C. Schonenberger, S.F. Alvarado, Z. Phys. B 80 (1990) 373. [8] W. Williams, V. Hoffmann, F. Heider, T. G€ oddenhenrich, C. Heiden, Geophys. J. Int. 111 (1992) 417. [9] R.B. Proksch, S. Foss, E.D. Dahlberg, IEEE. Magn. MAG-30 (1994) 4467. [10] S. Foss, R. Proksch, E.D. Dahlberg, B. Moskowitz, B. Walsh, Appl. Phys. Lett. 69 (1996) 3426. [11] K.-H. Han, D. Spemann, R. H€ ohne, A. Setzer, T. Makarova, P. Esquinazi, T. Butz, Carbon, accepted for publication. [12] R. H€ ohne, P. Esquinazi, Adv. Mater. 14 (2002) 753.