PHYSICAL REVIEW B VOLUME 59, NUMBER 11 15 MARCH 1999-I Nanoscale Fe islands on MgO 001... produced by molecular-beam epitaxy S. M. Jordan, R. Schad, A. M. Keen, M. Bischoff, D. S. Schmool,* and H. van Kempen Research Institute for Materials, University of Nijmegen, Toernooiveld 1, NL-6525 ED Nijmegen, The Netherlands Received 3 August 1998 We report that a 10-nm thick Fe film grown at an elevated temperature on MgO 001 forms isolated islands of diameter between 10 and 100 nm. Increasing the deposition temperature causes the islands to decrease in diameter. The resulting films are electrically insulating but show electrical transport properties that vary strongly with growth temperature when capped with 2 nm Au. Films grown at a temperature of 743 K showed a giant magnetoresistance of 0.7% when measured at room temperature. S0163-1829 99 01012-7 I. INTRODUCTION at the desired temperature to within 10 K throughout the deposition. The pressure during growth was 1 10 9 mbar The magnetic and transport properties of granules of mag- or lower. After growth of the Fe layer, the sample was netic material suspended within a nonmagnetic metal matrix cooled to 370 K, and a 2-nm layer of Au was deposited. This have received much interest on account of the discovery of thickness was felt to provide sufficient coverage to protect the giant magnetoresistance GMR effect1,2 in these granu- the Fe from oxidation without obscuring the structure of the lar systems.3 This effect arises from spin-dependent scatter- iron. AES showed strong Au peaks, with slight Fe and O ing of electrons at the ferromagnet/normal-metal interface, peaks visible. and depends critically on the material's physical structure.4 The physical mechanisms governing this effect are of funda- III. SCANNING PROBE MICROSCOPE INVESTIGATION mental importance. However, both measuring and control- ling the properties of an inhomogeneous system are difficult, We examined the completed Au covered films with both and few studies report the correlation of transport and struc- AFM and scanning tunneling microscopy STM . The AFM tural properties.5 measurements were made ex situ in the ``tapping mode'' Here we report a method by which a granular system can using a commercial Si tip of radius 10 nm and internal angle be formed. When Fe is grown at temperatures above 700 K, 30°.12 Micrographs are presented in Figs. 1 and 2 for growth a sharp transition occurs, resulting in the formation of dis- temperatures of 793 and 953 K, respectively. Both AFM continuous islands instead of the continuous films previously images have been ``flattened'' by subtracting each scan reported.6,7 Their size distribution can be controlled simply line's average slope. by changing the deposition temperature, the islands becom- The image of the sample grown at the lower temperature ing smaller as the temperature increases. The structure of the shows large up to 100 nm major axis islands of widely film can then be measured directly using a scanning probe varying shapes and sizes. Channels of typically 2-nm depth microscope. Capping the film with Au allows electrical con- can be observed between the islands, which possibly contain duction to take place, with GMR occurring due to the pres- smaller islands that are not well resolved due to the curvature ence of ferromagnet/normal-metal interfaces. This system of the tip. At this temperature both AFM and STM give can thus provide an unprecedented opportunity to correlate similar results. structural, magnetic, and transport properties. II. SAMPLE PREPARATION Commercially obtained substrates8 were cleaned ultra- sonically, first in acetone and then propan-2-ol to remove surface contamination. Atomic force microscopy AFM showed large terraces of atomic flatness extending over as much as 200 nm. After introduction into the molecular- beam-epitaxy MBE chamber they were heated by electron bombardment of the sample holder to 1300 K and held at that temperature for 1 min. This procedure is sufficient to desorb any adsorbed hydrogen.9­11 After cooling, Auger- electron spectra AES were taken, which showed a KLL carbon peak equivalent to 6% of 1 monolayer. Fe was then deposited from a Knudsen cell temperature 1680 K at a rate of 0.13 nm per minute, the rate being FIG. 1. STM top and AFM bottom micrograph of 10 nm Fe measured before growth by a quartz-crystal thickness meter. grown at 793 K. Black-white contrast approximately 20 nm. Images In all cases, 10 nm of Fe was deposited. The sample was held are 1000 500 nm. 0163-1829/99/59 11 /7350 4 /$15.00 PRB 59 7350 ©1999 The American Physical Society PRB 59 BRIEF REPORTS 7351 TABLE I. Average island areas using two different measure- ments. The values Ac were found by counting the number of islands in a square image of reasonable size. The second and third columns used direct measurements of each island area. The figures in brack- ets represent the area converted into a radius in nm . The resistiv- ities of the completed samples are given. A value of indicates that the resistance was too great to measure with standard techniques. Growth temp. K Ac (nm)2 A¯ (nm)2 nm ( cm) 693 ­continuous­ 7.7 743 5900 (44) 3640 (34) 0.81 5.7 102 793 5600 (42) 3100 (31) 0.90 7.0 104 843 4900 (40) 3000 (31) 0.82 7.3 105 893 4400 (37) 2900 (30) 1.10 3.4 106 943 2200 (26) 1480 (22) 0.71 FIG. 2. STM top and AFM bottom micrograph of 10 nm Fe 993 1700 (23) 1150 (19) 0.96 grown at 953 K. Black-white contrast approximately 30 nm. Images are 500 250 nm. within a square image of sufficient size to contain approxi- When the deposition temperature is increased to 953 K, mately 600 islands. The average radius was then calculated the islands approximately halve in area, and the depth of the using the formula channels between them increases. The AFM and STM im- ages are different here, the former showing pyramids instead of round islands. This is due to the convolution of the pyra- r¯ Ac midal AFM tip and the surface.13 The STM has greater spa- , 1 tial resolution on account of tunneling occurring from a single atom on the tip, but the overall aspect ratio of our which assumes each island to be circular and that the islands mechanically prepared Pt-Ir tips is unknown. STM is practi- cover the surface with a packing fraction of 0.87. The second cally difficult on this system due to the rapid appearance of method was to trace the outline of each island manually and double tip effects; thus, we will base the analysis to follow then measure the areas using commercial software. The radii on the AFM images. were also calculated using Eq. 1 . Table I gives the results. Height profiles of two AFM images are shown in Fig. 3 These two methods do give differing results, but they rely on with the tip profile superposed. At the higher growth tem- different premises. The first assumes that each island is cir- perature both the increased depth and greater steepness of the cular and ignores the fact that the spaces between the islands channels between the islands is clear. The true structure of will be larger than that between perfect circles. The second the edges of the islands is not reflected by the height profiles, relies on manual measurement and tends to underestimate since the curvatures of the tip radius 10 nm and sample are the island areas. The actual value will be between the two comparable. The true steepness is the sum of the two curva- numbers. tures, provided that the tip touches only one part of the Figure 4 shows the histograms of island areas for two sample at once.14 growth temperatures. Superposed is the log-normal distribu- A property of great interest for correlation with physical tion properties is the average area of the islands. This was as- sessed in two ways. First, an estimate of the average island area Ac was measured by counting the number of islands FIG. 4. Island area distribution for two different temperatures. FIG. 3. Height profiles of the AFM images in Figs. 1 and 2. The The lines show the log-normal distribution for the means and stan- tip profile 10-nm radius of curvature is superposed. dard deviations in Table I. 7352 BRIEF REPORTS PRB 59 Fe/Au interfaces. As the applied field increases, the magne- tization vectors of the Fe become aligned, and less scattering occurs leading to a lower resistance. GMR may also occur in the films grown at higher temperature but this measurement is practically difficult due to the magnitude of and the strong temperature dependence. If we assume that maximum resistance occurs at zero net magnetization,17 then Fig. 5 gives the coercivity of our sample as 500 Oe. It is also clear that the saturation field of the sample is high: the resistivity is still changing as H in- creases above 10 kOe. Kreuzer et al.18 report that system of Fe ``dots'' 50 nm in diameter and periodicity 320 nm clearly saturates at a field of 1 kOe. FIG. 5. Magnetoresistance curves with H and the current normal Wang and Xiao et al.19 have studied the temperature de- to each other, but both in the plane of the film. pendence of the GMR effect in the Co20Ag80 granular sys- tem. They found that reducing the temperature from 300 to ln A/A¯ 2 100 K increased the magnetoresistance ratio by 50%. They N A N0exp . 2 claim that this change is due to a reduced magnetization at 2 2 finite temperature due to spin-wave excitations. However, This function has no physical significance,15 however, it fits they also found a change in the form of the field dependence the experimental data reasonably well. The mean and stan- of the resistance, indicating that the magnetic properties of dard deviation were found using the sample change strongly with temperature. Our results do not show any change in the form of the field dependence. N ln A¯ ln An /N 3 n 0 V. DISCUSSION and We can only explain the rapid change in with growth N N temperature as a change in the properties of the Au/Fe inter- 2 ln 2An /N ln A /N 2. 4 face. The geometry of the Fe islands does not change suffi- n 0 n 0 ciently to cause an increase of this order; thus the steepening These results show that the measured island area drops by a sides of the islands must prevent the Au from making con- factor of 3 as the temperature increases by 250 K, but the tact. Au and Fe are immiscible, and the angle between the overall distribution of sizes remains the same. substrate and the Fe may make a difference to the surface tension required to connect the two. The Au atoms were incident at an angle of 15° to the normal during deposition, IV. ELECTRICAL TRANSPORT PROPERTIES and since the slope of the sides of the islands is comparable The electrical resisitivities of these films are not only to this Fig. 3 , shadowing may play a role. A thicker Au much greater than that of a continuous film but depend very layer should lower and allow a direct correlation to be strongly on deposition temperature. Table I shows that an made between the magnetoresistance behavior and the distri- increase in of almost two orders of magnitude of an island bution of island sizes. A possibility to tailor GMR properties 743 K over a continuous 693 K film. Raising the growth in a simple way is also very valuable. temperature further causes to increase still further until the We have yet to explain why islands form; any theory must film becomes essentially an insulator. support not only the existence of the transition between is- The resistivities of the samples grown at 793, 843, and landed and continuous films at 700 K, but the fact that the 893 K all showed a negative temperature coefficient. All islands become smaller with increasing temperature. Several approximately followed the law authors7,20,21 have studied islanded Fe films on MgO, but at coverages in the monolayer regime. It also may be possible 0exp /Tn , 5 that many other metal/insulator systems have this behavior if the deposition temperature is sufficient. suggested by Sheng,16 with n 0.5. This gives approximately a factor of 10 decrease in when the T is increased from 200 to 300 K. This temperature dependence indicates that the ACKNOWLEDGMENTS islands are electrically isolated and conduction occurs through thermally generated carriers. This work has been financially supported by the Dutch Figure 5 shows the magnetoresistance of the sample Foundation for the Fundamental Research of Matter FOM , grown at 743 K for two measurement temperatures with the which is, in turn, financially supported by the Dutch Organi- measurement current and the applied field normal to each zation for Scientific Research NWO . R.S. would like to other but in the plane of the film. The same behavior is seen acknowledge the assistance of the Brite-Euram program of when the current and field are parallel. Reducing the tem- the European Community Contract No. BRE 2-CT93-0569 . perature to 100 K approximately doubled the GMR ratio. D.S.S. was funded by the TMR program of the European This behavior occurs due to spin-dependent scattering at the Union. PRB 59 BRIEF REPORTS 7353 *Present address: Laboratoire de Magne´tisme et d'Optique, Univer- J. Chem. Phys. 105, 1295 1996 . site´ de Versailles-Saint Quentin ­ CNRS, Ba timent Fermat, 45 10 M.A. Karolewski and R.G. Cavell, Surf. Sci. 271, 128 1992 . Avenue des Etats-Unis, 78035 Versailles, France. 11 O. Robach, G. Renaud, and A. Barbier, Surf. Sci. 401, 227 Corresponding author. FAX: 3652190. Electronic address: 1998 . hvk@sci.kun.nl 12 Digital Instruments Inc., 520 E. Montecito St., CA 93103. 1 G. Binasch, P. Gru¨nberg, F. Saurenbach, and W. Zinn, Phys. Rev. 13 F. Atamny and A. Baiker, Surf. Sci. 323, L314 1995 . B 39, 4828 1989 . 14 D. Keller, Surf. Sci. 253, 353 1991 . 2 M.N. Baibich et al., Phys. Rev. Lett. 61, 2472 1988 . 15 J.A. Blackman, B.L. Evans, and A.I. Maaroof, Phys. Rev. B 49, 3 S.S.P. Parkin, R.F.C. Farrow, T.A. Rabedeau, R.F. Marks, G.R. 13 863 1994 . Harp, Q. Lam, C. Chappert, M.F. Toney, R. Savoy, and R. 16 P. Sheng, in Electrical Transport and Optical Properties of Inho- Geiss, Europhys. Lett. 22, 455 1993 . mogeneous Media, edited by J.C. Garland and D.B.Tanner, AIP 4 B. Dieny, S.R. Teixeira, B. Rodmacq, C. Cowache, S. Auffret, O. Conf. Proc. 40 AIP, New York, 1978 , Chap. II, pp. 143­153. Redon, and J. Pierre, J. Magn. Magn. Mater. 130, 197 1994 . 17 J.Q. Xiao, J.S. Jiang, and C.L. Chien, Phys. Rev. Lett. 68, 3749 5 K. Ounadjela, S.M. Thompson, J.F. Gregg, A. Azizi, M. Gester, 1992 . and J.P. Deville, Phys. Rev. B 54, 12 252 1996 . 18 S. Kreuzer, K. Pru¨gl, G. Bayreuther, and D. Weiss, Thin Solid 6 S.M. Jordan, J.F. Lawler, R. Schad, and H. van Kempen, J. Appl. Films 318, 219 1998 . Phys. to be published . 19 J.Q. Wang and G. Xiao, Phys. Rev. B 50, 3423 1994 . 7 B.M. Lairson, A.P. Payne, S. Brennan, N.M. Rensing, B.J. 20 Y. Park, S. Adenwalla, G.P. Felcher, and S.D. Bader, Phys. Rev. Daniels, and B.M. Clemens, J. Appl. Phys. 78, 4449 1995 . B 52, 12 779 1995 . 8 Crystal GmbH, Ostendstrasse 2-14, 12459 Berlin, Germany. 21 C. Liu, Y. Park, and S.D. Bader, J. Magn. Magn. Mater. 111, 9 M.J. Stirniman, C. Huang, R.S. Smith, S.A. Joyce, and B.D. Kay, L225 1992 .