APPLIED PHYSICS LETTERS VOLUME 75, NUMBER 7 16 AUGUST 1999 Step decoration during deposition of Co on Ag 001... by ultralow energy ion beams J. Dekoster,a) B. Degroote, H. Pattyn, G. Langouche, A. Vantomme, and S. Degroote Instituut voor Kern- en Stralingsfysica, Celestijnenlaan 200D, B-3001 Leuven, Belgium Received 6 April 1999; accepted for publication 22 June 1999 A possibility for decorating atomic steps on single-crystal surfaces by using ultralow energy ion beams is reported. Isotopically pure ion beams are produced by a mass separator and subsequently decelerated by an electrostatic lens. The lens was designed to allow sweeping of the ion beam in order to obtain a uniform deposition over a large area. The preferred sites of single Co atoms on Ag are investigated with in situ scanning tunneling microscopy measurements. A clear indication is found that by increasing the energy of the deposited Co to several electron volts, an enhanced Co decoration of the Ag steps is induced. This technology opens perspectives for an increasing number of elements which can form self-organized nanostructures such as atomic wires on vicinal crystal surfaces. © 1999 American Institute of Physics. S0003-6951 99 03533-0 In thin film growth a variety of deposition parameters encounter because of the relatively high energies which are such as the growth temperature, the deposition rate, and the used in surface science several kilo-electron-volts to a few energy of deposited species have a crucial influence on the mega-electron-volts as compared to the surface energy a resulting physical properties of the thin film. Nowadays, it is few electron volts . Depending on the energy of the ions, straightforward to vary the temperature of the substrate over processes such as channeling along an array of atoms of the a wide range with commercially available systems. Also the crystal, scattering from the surface, implantation, or sputter- deposition rate can easily be varied over several orders of ing events can be observed.6 In ion beam deposition tech- magnitude, e.g., by modifying the dissipated power of an niques such as ion beam sputtering the energy of the depos- electron beam gun and control of the evaporation rate with a ited elements exhibits a distribution. This energy distribution quadrupole mass spectrometer setup. One of the growth pa- shows a pronounced maximum at an energy which is equal rameters which is less well understood is the energy of the to one-half that of the surface binding energy. The average deposited particles. Varying the energy of the deposited spe- energy of the sputtered particles can be up to an order of cies in a controlled manner is a very tedious job but can be magnitude larger since the distribution is asymmetric with done by using ion beams for which the accelerating voltage higher population on the high-energy side of the maximum.7 in principle can be varied.1,2 In other words, the atoms are deposited with an energy on Here we describe a low energy ion beam deposition sys- the order of a few up to a few tens of electron volts. To tem and report its first results. Using in situ scanning tunnel- investigate what is at the origin of the difference between ing microscopy we evidence the possibility to deposit ions at films produced by, e.g., MBE and ion beam sputtering, a a sufficiently low energy so that the surface of the single- deposition method which allows a controlled energy varia- crystal substrate does not experience noticeable damage dur- tion is needed. The deposition method that we have used is ing deposition. Moreover we observed step decoration by Co capable of accurately controlling the energy in the region on Ag 001 using ultralow energy ion beams several elec- which is of interest for ion beam sputtering and thus will tron volts . Using thermal deposition of Co on Ag with mo- allow for a detailed investigation of the energy dependence lecular beam epitaxy MBE hence with typical deposition of the sputter deposition process. energies on the order of a few tens of milli-electron-volts in A 50 keV 59Co ion beam is produced by the Leuven Ion similar growth conditions no such decoration effects can be Separator and is decelerated by an electrostatic lens which is observed, indicating the importance of the energy of the de- mounted in a differentially pumped ultrahigh vacuum UHV posited species for the decoration process. The problem of a section of the beamline. A Faraday cup and beam profile small sample size when using a focused ion beam can be monitor are installed to monitor the current and shape of the overcome by sweeping the ion beam over larger areas. We beam entering the lens. Deflection plates can be used to steer indicate that the method has potential applications for the and/or sweep the beam. In Fig. 1, a schematic view of the discovery of new substrate/film combinations which exhibit deceleration stage is shown. It consists of seven ring-shaped electrodes with increasing diameter toward the last electrode. the phenomenon of step decoration and which can thus lead The sample, which is incorporated in the latter one, can be to fabrication of quasi-one-dimensional wires. To our knowl- inserted by a sample transfer stick which is also used as a edge step decoration has only been obtained previously for Faraday cup behind the lens. The high decelerating potential Co on Cu 111 , Fe on Cu 111 , and Fe on W 110 .3­5 of 50 kV is derived from the high voltage terminal and When an ion hits a surface it is usually not a gentle supplied to the last electrode and the sample . The different electrodes are connected over 500 M resistors to create a a Electronic mail: johan.dekoster@fys.kuleuven.ac.be potential which decreases stepwise toward the ground. The 0003-6951/99/75(7)/938/3/$15.00 938 © 1999 American Institute of Physics Downloaded 20 Mar 2001 to 148.6.169.65. Redistribution subject to AIP copyright, see http://ojps.aip.org/aplo/aplcpyrts.html Appl. Phys. Lett., Vol. 75, No. 7, 16 August 1999 Dekoster et al. 939 FIG. 1. Cross-sectional view of the ion beam deceleration optics. The volt- age can be varied between a few volts and 1 kV and allows one to modify the incident particle energy. applied high voltage is directly derived from the ion source FIG. 2. Lateral Co thickness profile for a layer deposited with 10 eV on a anode potential reduced by an additional voltage which al- Ag surface as determined with Rutherford backscattering spectroscopy. lows the ions to reach the sample with an energy variable from 1 keV to 1 eV. Hence, the ions are accelerated toward croscopy (UHV 2 10 10 mbar) to further characterize the the entrance of the lens and subsequently decelerated in the surface quality. The Ag surfaces are concluded to be ex- lens to reach the sample with the desired energy. The lens tremely clean since no traces of any impurities could be ob- itself was designed using the SIMION program, which calcu- served using Auger electron spectroscopy. The clean surface lates both the electrostatic field and the ion trajectories.8 The shows monoatomic steps 2 Å high between terraces of sev- simulations were performed for a 50 keV incoming 59Co eral hundreds of angstroms average width. The atomic steps beam which was decelerated to energies ranging from 50 of the MBE grown Ag film have a varying distance and down to 1 eV. Two parameters are found to be crucial for orientation. This can be due to the presence of small angle optimal beam transport. First, a nearly perfect axial passage grain boundaries. The mosaic spread of such films as mea- of the ion beam through the lens is important to avoid a sured by x-ray diffraction rocking curves is typically 0.5°. strong increase of the beam spot size at the target. We used a All images are taken with the sample at ambient temperature. beam with diameter of about 4 mm in a lens with electrodes Ion beam depositions were performed in an UHV chamber varying in diameter from 20 up to 50 mm. Second, we ob- (UHV 1 10 10 mbar). A layer of about 1 monolayer ML served that the behavior of the ion beam inside the lens is Co was deposited with a 10 eV ion beam. To protect the extremely sensitive to the position of its focal point. An op- sample from oxidation a 20 Å Ag epilayer was MBE depos- timal focus position is situated inside the lens a few centi- ited on top of this Co layer immediately after the ion beam meters in front of the sample position. Experimentally opti- deposition. mum beam settings are obtained by using a Faraday cup The Ag capping layer also avoids sputtering of the Co behind the lens while no high voltage was applied. This re- layer during subsequent Rutherford backscattering spectrom- sults in an axial beam with a focus at the second Faraday etry RBS measurements, which were used to determine the cup. Second, high voltage is applied to the lens and the focus lateral homogeneity of the deposited layer. The measure- of the beam is shifted toward the lens while optimizing the ments were performed with a 2 MeV 4He ion beam, and a beam in the second Faraday cup. The ion current and thus scattering angle of 172°. Figure 2 shows a lateral thickness the deposition rate is determined by integration of the signal profile of the deposit. Within the experimental error, the pro- at the beam profile monitor. We typically used a current of file is nearly uniform over a region of about 4 mm, which is 150 nA, which gives a deposition rate of about 0.002 Å/s. Ag substrates were prepared in a Riber MDS32 molecu- lar beam epitaxy facility. A 1000-Å-thick Ag single crystal was deposited on polished MgO 001 held at ambient tem- perature. The MgO substrate was chemically cleaned in iso- propanol and annealed to 600 °C for 30 min. To enable smooth and epitaxial growth of Ag on MgO 001 a 50 Å Cr buffer layer was first grown on MgO at 175 °C. The epitaxial relations are Ag 100 001 //Cr 110 001 //MgO 100 001 . The reflection high-energy electron diffraction RHEED pattern after growth reveals the presence of some remaining roughness. Annealing to 200 °C in the ultrahigh vacuum (UHV 5 10 11 mbar) and subsequent gently cooling down allows the quality of the Ag substrate to further improve until a sharper RHEED pattern was observed. The Ag substrate was then transferred in vacuo from the growth chamber to adjacent surface analysis chambers with FIG. 3. 2 MeV 4He random and aligned Rutherford backscattering spec- facilities for Auger spectroscopy and scanning tunneling mi- trum of 1 ML Co deposited with 10 eV on Ag 001 /Cr 001 /MgO 001 . Downloaded 20 Mar 2001 to 148.6.169.65. Redistribution subject to AIP copyright, see http://ojps.aip.org/aplo/aplcpyrts.html 940 Appl. Phys. Lett., Vol. 75, No. 7, 16 August 1999 Dekoster et al. underlying substrate. This is, e.g., the case for Co on the upper side of a step where the underlying Ag step atom has a larger lateral freedom. The importance of the energy of the incident Co par- ticles for the step decoration process is determined by com- paring the ion beam deposited Co layers with thermally de- posited Co. The right-hand panel of Fig. 4 shows a STM topograph of 0.1 ML Co deposited at a rate of 0.02 Å/s on Ag 001 by MBE. The complete absence of step decoration is striking. Due to the low surface mobility of thermally de- posited Co atoms on Ag, the Co islands are on average larger in number and smaller in size than for an ion beam deposited layer.Finally, we indicate that this deposition method allows us to investigate the magnetic properties of Co nanostruc- tures formed at steps on vicinal crystal surfaces by emission Mo¨ssbauer spectroscopy measurements. Using this ultralow FIG. 4. STM image and height profile along the white arrow of 0.1 ML Co energy ion beam deposition, 57Co can be incorporated in, deposited on Ag 001 with energy of 15 eV left panel and deposited with e.g., atomic wires. The sensitivity of emission Mo¨ssbauer thermal energy right panel . In the ion beam deposited layer the upper Ag spectroscopy is several orders of magnitude better than ab- steps are decorated with Co islands of various heights. Co islands nucleate also near a screw dislocation along the step direction both in the unslipped sorber Mo¨ssbauer spectroscopy and will allow the investiga- u and the slipped s region. tion of magnetic ordering and magnetic anisotropy of atomic wires. exactly the length over which the ion beam was swept during In conclusion, we have shown with STM that Co atoms deposition. Due to the finite width of the ion beam the Co deposited on Ag by ultralow energy ion beam deposition coverage does not abruptly drop to zero but exhibits a with a deposition energy of 15 eV do not damage the Ag gradual decrease on the sides. surface. Due to their excess kinetic energy compared to ther- The crystalline quality of the layers is determined prior mally deposited atoms, the Co ions have sufficient surface to and after the Co deposition from channeling measure- mobility to reach substrate regions with a low nucleation ments. We find no evidence of any damage to the Ag epil- barrier, resulting in a strong preference for decorating Ag ayer see Fig. 3 . The minimum yield in the Ag layer is equal steps on the upper side. We have shown that the growth to 12%. There is nearly no He channeling observed in the Co morphology of submonolayer films of Co on Ag can be signal see inset Fig. 3 . Due to the large lattice mismatch changed drastically by increasing the Co energy by several between Ag and either fcc or hcp Co, and due to their dif- orders of magnitude with respect to thermal deposition. This ference in the surface free energy, the Co layer does not clearly shows the importance of the energy of the deposited grow pseudomorphically in a layer by layer mode. This species. Using this method one can accurately control the probably results in highly strained, distorted islands which energy of the deposited material and this will allow a de- show poor registry with the Ag substrate and thus do not tailed analysis of the energy dependence of the deposition show a clear channeling effect. process. In the left-hand panel of Fig. 4 a scanning tunneling This work was supported by the Belgian Fund for Sci- microscopy STM topography image for a nominal cover- entific Research, Flanders FWO , Concerted Action GOA , age of 0.1 ML Co deposited at ambient temperature using a and the Inter-University Attraction Pole IUAP P4/10 . J. D. 59Co ion beam with an energy equal to 15 eV is shown. One and A.V. are postdoctoral researchers of the F.W.O. clearly observes two types of nucleation sites: decoration of steps on the upper side of the terraces and islands in the terraces. For this nominal Co coverage of 0.1 ML, the islands do not have a uniform height distribution-confirming that 1 A. W. Klein, Science 275, 1440 1997 . 2 the Co islands do not grow layer by layer. At the upper side C. R. Laurens, M. F. Rosu, F. Pleiter, and L. Niesen, Phys. Rev. Lett. 78, 4075 1997 . of the steps, the islands are mostly 1 ML high while those 3 H. J. Elmers, J. Hauschild, H. Hoche, U. Gradman, H. Bethge, D. Heuer, formed on the terraces are higher see Fig. 4, left-hand and U. Ko¨hler, Phys. Rev. Lett. 73, 898 1994 . panel . Co islands also nucleate along the step both in the 4 J. Shen, R. Skomski, M. Klaua, H. Jenniches, S. Sundar Manoharan, and slipped and the unslipped part of the crystal near a screw J. Kirschner, Phys. Rev. B 56, 2340 1997 . 5 J. de La Figuera, J. E. Prieto, C. Ocal, and R. Miranda, Surf. Sci. 307-309, dislocation. Seemingly the stress field of the dislocation is 538 1994 . sufficient to act as a nucleation center for Co island growth. 6 Fundamentals of Surface and Thin Film Analysis, edited by C. Feldman At a deposition energy of 15 eV, the Co islands preferably and J. W. Mayer Prentice­Hall, Englewood Cliffs, NJ, 1986 . 7 nucleate in substrate regions where the strain caused by the M. W. Tomson, Phys. Rep. 69, 335 1981 . 8 D. A. Dahl, J. E. Delmore, and A. D. Appelmans, Rev. Sci. Instrum. 61, island can be easily accommodated by a relaxation of the 607 1990 . Downloaded 20 Mar 2001 to 148.6.169.65. Redistribution subject to AIP copyright, see http://ojps.aip.org/aplo/aplcpyrts.html