Nuclear Instruments and Methods in Physics Research B 190 (2002) 840­845 www.elsevier.com/locate/nimb Hyperfine interaction studies with monolayer depth resolution using ultra-low energy radioactive ion beams A. Vantomme *, B. Degroote, S. Degroote, K. Vanormelingen, J. Meersschaut, B. Croonenborghs, S.M. Van Eek, H. Pattyn, M. Rots, G. Langouche Instituut voor Kern- en Stralingsfysica, Katholieke Universiteit Leuven, Celestijnenlaan 200 D, B-3001Leuven, Belgium Abstract A variety of nuclear techniques rely on the incorporation of radioactive atoms to investigate the microscopic structural, electronic and magnetic properties of a material. In the past, ion implantation has been utilized to introduce these radioactive probes, resulting in a depth distribution of typically several hundreds of A A, and damaging the sample. Both implantation-related deficiencies are incompatible with the ever shrinking sizes relevant in nanostructures. This problem can be circumvented by using ultra-low energy ion beams ­ of the order of 5 eV, i.e. below the displacement energy of the substrate atoms. Consequently, the radioactive probes are ``deposited'' on top of the sample, without generating damage to the substrate. Since the implantation chamber is in vacuo connected with the molecular beam epitaxy deposition chamber, the probe layer can be introduced at any stage during the sample growth (from surface to interface)­­with monolayer depth resolution. As an example, we discuss the ultra-low energy ion deposition of 111In in Cr, followed by analysis with perturbed angular correlation spectroscopy. The aim of the study is to explore the magnetic ordering of Cr thin films. Ó 2002 Elsevier Science B.V. All rights reserved. PACS: 75.70.Ak; 68.35. p; 31.30.G; 61.18.F Keywords: Low energy ion deposition; Nuclear solid state physics; Soft landing; Nanotechnology; Cr spin density wave 1. Introduction copy, provide microscopic information on the vi- cinity of a radioactive probe nucleus. Due to their For several decades, nuclear methods have high sensitivity (in many cases an amount of provided a powerful tool in the study of the 1012­1014 atoms, a small fraction of a monolayer, structural, electronic and magnetic properties of is sufficient), these techniques have shown to yield materials [1,2]. Nuclear orientation either by low very valuable information in thin film studies, temperature or perturbed angular correlation where the amount of material is often too small to (PAC) techniques as well as M oossbauer spectros- apply conventional techniques [3]. Because ther- mal diffusion cannot result in control of the depth profile, these radioactive tracers are mainly intro- * duced by ion implantation, typically using energies Corresponding author. Tel.: +32-16-32-75-14; fax: +32-16- of the order of several tens to several hundreds 32-79-85. E-mail address: andre.vantomme@fys.kuleuven.ac.be (A. of keV. Two major drawbacks of this tech- Vantomme). nique, however, are the depth distribution of the 0168-583X/02/$ - see front matter Ó 2002 Elsevier Science B.V. All rights reserved. PII: S0168-583X(01)01208-3 A. Vantomme et al. / Nucl. Instr. and Meth. in Phys. Res. B 190 (2002) 840­845 841 implanted species (i.e. the roughly Gaussian im- tive In probes on Cr for the study of the magnetic plantation profile) and the damage caused by the properties of Cr thin films. deposited energy. For instance, implantation of In in Cr, with an energy of 50 keV, results in a dis- tribution located 135 A A below the sample surface, 2. LEID: the technique and the setup with a FWHM of 130 A A; and the formation of approximately 570 vacancies per incoming ion. A schematic overview of the LEID setup is Driven by nanotechnology, both for the study of shown in Fig. 1. An isotopically pure ion beam surfaces and interfaces, there has been an in- with an energy of typically 50 keV is delivered by creasing need for monolayer (ML) sensitivity the Leuven isotope separator. A proper mass res- and consequently, for other ways of introducing olution is obtained by bending the ion trajectory the radioactive probes. Some groups use indirect over an angle of 55° in the 1.5 m radius analysis evaporation techniques, e.g. thermally evaporating magnet, thereby separating the 2.5 mm FWHM the probes from a Cu [4] or Mo foil [5] target, in 111In beam over 13.5 mm from its neighbouring which the radioactive probes are initially intro- masses. Subsequently, the ion beam can be swept, duced by diffusion or implantation respectively. yielding a typical beam spot of 4 4 mm2 on the Alternatively, one can combine the mass selectivity target, and finally entering into an electrostatic and controllability of ion implantation with the deceleration lens. This deceleration stage consists extremely low energies of deposition, i.e. low en- of five metal electrode rings with increasing dia- ergy ion deposition (LEID) or soft landing [6­8]. meter and electrostatic potential along the path of LEID has been used in the past for the study of the ions. The specific configuration was chosen surface phenomena (e.g. diffusion, incorporation, based on simulations of the ion path depending on nucleation,. . .) [9,10]. In the present work, we the number, shape and voltage of the electrodes present the use of low-energy radioactive ion using the program SIMION [11]. The high voltage beams for the study of thin film properties with supply to the sample holder is derived directly ML depth resolution and we will show that LEID from the source section of the isotope separator, to can be put forward as an ideal tool for depth de- which a variable negative voltage V2 ¼ 0­200 V is pendent studies of structural, electrical and mag- superimposed. An additional permanent offset netic properties in state of the art materials. In V1 ¼ 50 V is used, shifting the zero point for de- order to illustrate the general concept, we will position to V2 ¼ 50 V. This approach enables us describe the deposition trace amounts of radioac- to check that no ions can reach the sample for Fig. 1. Schematic diagram of the LEID set-up. The deposition energy is selected by adjusting the output voltage of the power supply V2. 842 A. Vantomme et al. / Nucl. Instr. and Meth. in Phys. Res. B 190 (2002) 840­845 facilities. Details on the experimental setup can be found in [7]. 3. Low energy ion deposition of In on Cr Below the N eeel temperature (TN ¼ 311 K), bulk Cr exhibits a so called spin density wave (SDW), i.e. an antiferromagnet of which the magnitude of the magnetic moments vary in a sinusoidal way, with a period of approximately 21 unit cells ( 61 A A). This incommensurate SDW is charac- terized by a spin vector ~ SS and a wave vector ~ Q Q, and is labelled longitudinal or transversal for ~ SSk~ Q Q Fig. 2. Ion current through the grid as a function of output and ~ SS ? ~ Q Q respectively. In bulk Cr, the polarisa- voltage V2. The zero-point energy is set at approximately 50 V tion of the SDW changes from longitudinal below using power supply V1. to transversal above TSF ¼ 123 K, the spin flip temperature. Commensurate antiferromagnetic or- `negative' deposition energies. A firm calibration dering (i.e. no magnitude variation of the magnetic of the deposition energy is obtained by measuring moments) has been reported for Cr alloys or in the current transmitted through a grid, put at the strained Cr lattices. The magnetic phase diagram sample position (Fig. 2). This current is measured for Cr thin films is found to be more complicated on a Faraday cup mounted on the transfer stick [12]. PAC experiments using conventional im- behind the grid, and is normalised to the current plantation of the 111In probes have shown that the measured when no retarding field is applied. Below magnetic ordering in low dimensional chromium approximately 50 V (i.e. Edep ¼ 0 eV), no ions pass, depends on the growth conditions. For instance, a whereas at a higher voltage, the ion current grad- shift of TN and TSF have been reported, as well as ually increases. This gradual increase is related to the collapse of the SDW for film thicknesses below the mesh size of the grid, and is not caused by a the SDW period [13]. spread of the ion energy [7]. In the past, PAC experiments on Cr thin films So far, LEID has been used to study the be- have been performed using high energy ion im- haviour of individual atoms gently deposited onto plantation. It has been shown that the technique the surface of a crystal. Surface diffusion mecha- is very sensitive to the orientation of the spin. nisms, nucleation of deposited ions, incorporation However, until now, PAC is not directly sensitive of ions in terraces, the formation of quasi-one-di- to the direction of the ~ Q Q-vector. In order to derive mensional wires, etc. have been investigated by the polarisation of the SDW consistently from a combining conventional and hyperfine interaction PAC experiment one should be able to probe the techniques [3,9,10]. To incorporate LEID in thin variation of the magnetic hyperfine field from layer film and multilayer studies, a variety of sample to layer. Here we present a first attempt to reach preparation and characterization facilities should that goal. be available in the soft landing chamber, or via an in vacuo coupling. The LEID setup of the Ion and 3.1. Deposition of stable 115In ions Molecular Beam Laboratory at KU Leuven is in vacuo connected with two thermal deposition Before depositing extremely low concentrations units, a scanning tunneling microscope (STM), of radioactive 111In probes, extensive tests were Auger electron spectroscopy (AES), low energy performed with 5 eV stable 115In ion beams onto a electron diffraction, M oossbauer spectroscopy and variety of substrates, using fluences of the order of Rutherford backscattering (RBS) spectrometry 1015 at./cm2 or more. First, the beam profile as well A. Vantomme et al. / Nucl. Instr. and Meth. in Phys. Res. B 190 (2002) 840­845 843 as the uniformity of the implanted region was in- film was covered with 2 1013 111In/cm2 at an vestigated by RBS mapping. After deconvoluting energy of 5 eV, using a beam current of approxi- the He beam profile used for RBS (approximately mately 150 pA. Subsequently, a second 100 A A 1 1 mm2), a homogeneous In profile was ob- thick Cr film was deposited, after which the sample served in a region of about 4 mm wide, i.e. equal to was covered with a Ag capping layer. Hence, a 200 the sweeping area of the ion beam. Channeling A A thick Cr layer is obtained, halfway decorated measurements indicated that the implantation pro- with 111In probes as schematically illustrated in cess did not induce any noticeable damage to the Fig. 3(c) and (d). The lateral uniformity of the substrate, as expected for these extremely low en- deposited probes was confirmed by measuring the ergies. The absence of implantation induced cra- activity as a function of position across the sample. ters on the surface, and the presence of small aggregates deposited onto the substrate, were confirmed by in vacuo STM. No traces of any 3.3. Perturbed angular correlation experiment contamination could be observed within the sensi- tivity of AES. The PAC measurement was performed at room temperature, in a four-detector set-up and with the 3.2. Deposition of radioactive 111In ions sample normal in the plane of the detectors and oriented at 45° relative to the start detectors. The In order to study the polarisation of the SDW PAC spectrum generally reflects a more or less in Cr thin films, it is essential to confine the ra- periodic oscillation with a frequency directly re- dioactive probes to 1 ML, as illustrated in Fig. 3(c) lated to the magnetic hyperfine field at the probe and (d). To achieve optimal crystallinity, a 100 A A nucleus. Furthermore, the spectrum can contain thick Cr film was grown onto an epitaxial two harmonics, their relative presence being a di- hMgOi=Fe (75 A A)/Ag (300 A A) substrate/seed-layer rect measure for the orientation of the hyperfine structure. The roughness of the Cr layer could field (~ SS-vector). Several possible models were potentially impose a limit to the depth sensitivity considered when analysing the PAC time spectrum of the study. Therefore, the sample surface was (Fig. 4). The shape of the spectrum excludes the examined with in vacuo STM. From a scanned presence of a commensurate SDW phase [3], which area of 400 400 nm2, an rms roughness of ap- would result in a periodic oscillation of the am- proximately 0.4­0.5 nm was deduced, i.e. negligi- plitude with little damping. Alternatively, an in- ble compared to the SDW wavelength. This Cr commensurate SDW should be considered for Fig. 3. Schematic configuration of a transversal SDW with its wave vector in the plane (a,c) or out of the plane (b,d) of the Cr layer. The confinement of the radioactive probes in case of soft landing (c,d) results in monolayer depth sensitivity, which cannot be obtained when using conventional ion implantation (a,b). 844 A. Vantomme et al. / Nucl. Instr. and Meth. in Phys. Res. B 190 (2002) 840­845 Fig. 4. PAC time spectrum of a 200 A A thick Cr layer, which was halfway decorated with 111In probes, using ion beam deposition at 5 eV. The solid line is a fit assuming a longitudinal SDW. which the spins (~ SS-vector) are in-plane with the Cr- 4. Conclusion film, a feature that is readily derived from this PAC experiment. Distinction between either a LEID allows to deposit, with ultra low energies transversal SDW (~ Q Q out-of-plane) or a longitudi- ­ down to a few eV, small amounts of radioactive nal SDW (~ Q Q in-plane) is made by a fit procedure tracers in a controlled and non-destructive man- with the magnitude of the frequency (hyperfine ner. Based on the present and existing PAC studies field) as the free parameter. If the SDW were of the Cr magnetism, we illustrated that LEID propagating along the surface normal (~ Q Q out- gives way to the so far experimentally almost in- of-plane, Fig. 3(d)), obviously, there would be no accessible observation monolayer resolved hyper- information on the position of the probe layer with fine fields. Consequently, for other nanostructures, respect to the (anti)nodes of the SDW. To account the technique of soft landing of radioactive probes for this, a possible phase shift is entered in the can become an essential tool for the study of the analysis. Moreover, the depth distribution of the microscopic properties of magnetism or the mag- In probes, as inferred from STM (see above), was netic moment profile, as well as the phase forma- taken into account. For the model consisting of a tion at interfaces and near surfaces. These data are SDW propagating along the surface normal, the excellent testing grounds for ab initio model cal- experimental spectrum could be reproduced for a culations which recently became very reliable for phase shift of approximately p/4, but assuming electron density distributions. It is expected that a hyperfine field substantially different from the this LEID facility will form the basis for hyperfine values known in literature [3]. Much more con- interaction studies of structural, electrical and sistent results were obtained using the model as- magnetic properties of thin films, nanostructures suming an in-plane ~ Q Q vector, i.e. a longitudinal and their interfaces, where ML depth sensitivity is SDW, yielding a hyperfine field in agreement with crucial. the literature. Although neither of the two alternatives could Acknowledgements be ruled out completely as yet, measurements after In depositions at various depths in the Cr layers The authors want to acknowledge financial will allow full characterization of the SDW po- support by the FWO (Fund for Scientific Re- larization. search, Flanders, Belgium) and the Inter Univer- A. Vantomme et al. / Nucl. Instr. and Meth. in Phys. Res. B 190 (2002) 840­845 845 sity Attraction Pole IUAP P4/10. J.M. and B.D. [6] J. Dekoster, B. Degroote, H. Pattyn, G. Langouche, A. are postdoctoral researchers for the National Sci- Vantomme, S. Degroote, Appl. Phys. Lett. 75 (1999) ence Foundation (FWO ­ Vlaanderen). 938. [7] B. Degroote, Ph.D. Thesis, Katholieke Universiteit Leu- ven, 2001, available at www.fys.kuleuven.ac.be/iks/nvsf/ References nvsf.htm. [8] C.R. Laurens, L. Venema, G.J. Kemerink, L. Niesen, Nucl. [1] G. Schatz, A. Weidinger, Nuclear Condensed Matter Instr. and Meth. B 129 (1997) 429. Physics, Wiley, West Sussex, 1996. [9] C.R. Laurens, M.F. Rosu, F. Pleiter, L. Niesen, Phys. Rev. [2] G. Langouche (Ed.), Hyperfine Interaction of Defects in Lett. 78 (1997) 4075. Semiconductors, Elsevier, Amsterdam, 1992. [10] B. Degroote, A. Vantomme, H. Pattyn, K. Vanormelingen, [3] J. Meersschaut, J. Dekoster, S. Demuynck, S. Cottenier, B. submitted to Phys. Rev. B. Swinnen, M. Rots, Phys. Rev. B 57 (1998) R5575. [11] D.A. Dahl, D.E. Delmore, A.D. Appelhans, Rev. Sci. [4] T. Klas, R. Fink, G. Krausch, R. Platzer, J. Voigt, R. Instr. 61 (1990) 607. Wesche, G. Schatz, Surf. Sci. 216 (1989) 270. [12] P. B oodeker, A. Schreyer, H. Zabel, Phys. Rev. B 59 (1999) [5] ISOLDE Collaboration, E. Kugler, D. Fiander, B. Jonson, 9408. H. Haas, A. Przewloka, H.L. Ravn, D.J. Simon, K. [13] J. Meersschaut, J. Dekoster, R. Schad, P. Beli een, M. Rots, Zimmer, Nucl. Instr. and Meth. B 70 (1992) 41. Phys. Rev. Lett. 75 (1995) 1638.