- Email: [email protected]
The growth of manganese layers on Si(1 0 0) at room temperature: A photoelectron spectroscopy study
Applied Surface Science 255 (2009) 7642–7646
Contents lists available at ScienceDirect
Applied Surface Science journal homepage: www.elsevier.com/locate/apsusc
The growth of manganese layers on Si(1 0 0) at room temperature: A photoelectron spectroscopy study C.A. Nolph a, E. Vescovo b, P. Reinke a,* a b
University of Virgina, Department of Materials Science and Engineering, 395 McCormick Road, Charlottesville, VA 22904, USA National Synchrotron Light Source, Brookhaven National Laboratory, Upton, NY 11973-5000, USA
A R T I C L E I N F O
A B S T R A C T
Article history: Received 1 September 2008 Received in revised form 13 April 2009 Accepted 13 April 2009 Available online 21 April 2009
The combination of spin-and charge based electronics in future devices requires the magnetic doping of group IV semiconductors, and the formation of ferromagnetic contacts. The doping of Mn with Si is one of the material systems which is discussed in this context. The present study focuses on the growth of Mn on a Si(100)(2x1) surface, and the evolution of the surface was observed as a function of Mn coverage with synchrotron-based photoelectron spectroscopy. The reaction of Mn with the Si(100) surface at room temperature leads the formation of silicide at the boundary between the Si substrate and the Mnoverlayer, presumably with MnSi stoichiometry. The residual sub-oxide reacts with the Mn and therefore incorporates a few percent of Mn-O-Si at the interface. The analysis of the sub-oxide composition indicates that the Si+1 component is the most reactive oxidation state. The overlayer is dominated by Mn, either as Mn-metal or as a Mn-rich silicide phase, and the metallic layer introduces a band bending in Si. As a consequence of our observations, including information from a recent STM study, the formation of ferromagnetic contacts which require ideally a flat and compositionally homogenous overlayer, cannot be achieved through room temperature deposition of Mn on the Si(100) (2x1) surface. The influence of residual oxides and surface defects on the growth process will be further investigated. ß 2009 Published by Elsevier B.V.
Keywords: Manganese Silicon(1 0 0) Silicide Photoelectron spectroscopy Spintronics
1. Introduction The combination of group IV semiconductors with transition metal elements has been an active area of research for decades and was mostly geared towards the development of metal contacts and silicide leads [1,2]. With the advent of spintronics the interest has shifted to the doping of semiconductors with transition metal elements which can lead to the formation of dilute magnetic semiconductors [3–8]. The coupling between the magnetic moment of the transition metal atoms is then mediated by holes, and the magnetism can be tuned by the application of an electric field. The second driving force to study the formation of transition metal overlayers is given by the interest in ferromagnetic contact layers to provide the injection of a spin-polarized current into the semiconductor. Some MnSi compounds exhibit ferromagnetism, and the formation of a Mn-silicide overlayer has therefore been suggested as a route to achieve the spin-injection into group IV semiconductors. The introduction of Mn, which has a half filled d-shell and carries a large magnetic moment, into a Si lattice is highly
* Corresponding author. E-mail address: [email protected] (P. Reinke). 0169-4332/$ – see front matter ß 2009 Published by Elsevier B.V. doi:10.1016/j.apsusc.2009.04.047
coveted, and would allow the seamless combination of spintronics and charge based electronics. However, the doping of Si is hampered by the extremely low solubility of Mn in Si, the preference of Mn for interstitial lattice sites over the electronically active substitutional sites, and the competition with silicide formation [6,9–15]. In order to circumvent these problems we are investigating the feasibility of a surface-driven approach for the incorporation of Mn into a Si matrix. This approach has been suggested in several theoretical studies, which indicate that selected Mn–Si surface structures, and delta-doped layers of Mn embedded in Si might indeed be ferromagnetic [11,16–23]. A set of recent STM studies has been geared towards the investigation of the formation of Mn surface structures on the Si(1 0 0) surface [18–20,24]. The formation of silicide crystallites is observed at elevated temperatures, but 2D surface structures can evolve for lower substrate temperatures (below about 300 8C) [12,14,16,25]. The study presented here explores the bonding at the interface, and the growth of the overlayer during the room temperature deposition of Mn on Si(1 0 0) with photoelectron spectroscopy, which examines the core level and valence band. The PES analysis leads us to conclude that an ultrathin interface between Mn and the Si surface is formed, which is covered by a layer of Mn clusters. The role of the Si-sub-oxide is discussed in detail.
C.A. Nolph et al. / Applied Surface Science 255 (2009) 7642–7646
2. Experimental A Si(1 0 0) wafer was cleaned by repeated annealing cycles reaching temperatures of about 1000 8C, which were achieved by electron bombardment of the sample backside. The presence of the (2 1) reconstruction was confirmed by the presence of sharp reflexes in the LEED pattern, and the cleaning cycles were repeated until no further improvement was observable in the LEED pattern. Mn was subsequently deposited from an electron beam evaporation source (EFM3 Omicron Nanotechnology) and about 1 ML was deposited on the Si(1 0 0) sample. The coverage was determined by the XPS analysis of the relative peak areas of the Mn3p and Si2p core level using the excitation cross sections given by Yeh and Lindau [26]. The Fermi energy position or the position of the valence band maximum are used as energy reference. The photoelectron spectroscopy was performed at beamline U5UA of the National Synchrotron Light Source at Brookhaven National Laboratory [27]. U5UA is an undulator beamline, which is equipped with four gratings for optimum operation in the photon energy range between 15 and 200 eV. The light is linearly polarized in the horizontal plane, and the analyzer is fixed and its axis is at 458 with respect to the light incidence. The angle between sample surface normal and analyzer is modified by rotation of the sample, consequently the angle between sample surface normal and the incoming light is modified at the same time. The geometry is shown in the insert in Fig. 3.
7643
For the fit of the core level peaks we use the program fit [28], which employs the Marquard–Levenberg algorithm for the least square fit procedure [29,30]. The peaks are described by Doniach– Sunjic functions and a Shirley background, and the quality of the fit is judged by the quality of the residuals, and the ability of the fit to describe sharp saddle points and minima of the experimental curve [31,32]. A branching ratio of 0.6 and an energy separation of 0.5 eV was used for the Si2p1/2 and Si2p3/2 components. The smallest number of peaks, which gives a good fit of the experimental data, is used for analysis of the data. 3. Results and discussion Fig. 1a shows the Si2p core level (solid lines) before and after the deposition of Mn for a photon energy of 160 eV and includes the results of the peak fit procedure (dotted lines). An excellent fit can be obtained for the Si(1 0 0) surface prior to Mn deposition by using four peaks. The results for both photon energies (140 and 160 eV) yield the same binding energies within a margin of 0.1 eV. A LEED diffraction pattern, given as an inset in Fig. 1, confirms the presence of the (2 1) reconstruction and sharp, welldefined spots are observed prior to the deposition of Mn. After the deposition of Mn five peaks are required to achieve an equally excellent description of the Si2p peak. The structural analysis of thin films produced by the deposition of Mn on Si(1 0 0) (2 1) was recently performed with STM, and is
Fig. 1. On the left-hand side of the figure the Si2p core level for the Si(1 0 0) (2 1) surface, and after the Si2p after the deposition of Mn are shown. The results of the fit procedure are included and the labels of the peaks are indicated in the figure. The LEED pattern of the Si surface is shown as an inset and confirms the presence of the (2 1) reconstruction. A second inset shows the STM image measured after the deposition of 1.2 ML of Mn on Si, which was described in detail in a recent publication [19]. On the right-hand side, the area of all fit-curves is shown before and after the deposition of Mn. The respective Mn3p peak is shown at the bottom.
7644
C.A. Nolph et al. / Applied Surface Science 255 (2009) 7642–7646
published in [19]. The Si(1 0 0) (2 1) showed a low number of defects and was an excellent surface exhibiting a long-range reconstruction. For coverages below about 0.3 ML the formation of monoatomic Mn-chains are observed, which are oriented perpendicular to the Si-dimer rows. However, for higher coverages exceeding a ML of Mn relatively large, mostly elongated, flat clusters are present on the surface, but the terraces of the underlying Si(1 0 0) surface can still be discerned. An STM image obtained for those coverages is included as an insert in Fig. 1. The clusters still retain a slight preference for an orientation of their long axis to run perpendicular to the dimer rows of the Si(1 0 0) (2 1) reconstruction, which is deduced from their orientation with respect to the terrace edge. In Fig. 1 (right-hand side) the relative peak intensities of the Si2p core level contributions are summarized, and the peak numbers begin with the most intense peak, which is positioned at 98.4 eV for Si(1 0 0) and at 99.1 eV after the deposition of Mn [33– 35]. The contribution positioned at the lowest binding energy is the most intense peak for Si(1 0 0). After the deposition of Mn an additional peak with a contribution of nearly 20% to the total peak area is observed on the low binding energy side of the Si2p peak. The emergence of this additional contribution (labeled A in the figure) is readily apparent upon inspection and leads to a shallower slope of the low binding energy side of the Si2p core level. The width of the Fermi edge/valence band edge is conserved for all measurements, and the changes in the Si2p core level are therefore not caused by a modification of the spectral resolution. The corresponding Mn3p core level, which is also included in Fig. 1, is positioned at 46.9 eV, and cannot be described reliably by a fit procedure due to the presence of multiple overlapping peaks from the spin-orbit and multiplet splitting typical for 3d transition metals [20,36–38]. The position of the main peak is commensurate with either Mn or Mn-silicides, or a mixture of both. The formation of Mn-oxides, which introduces a large chemical shift, can be discerned by the emergence of a shoulder on the high binding energy side of the Mn3p peak, which has been observed in samples stored for more then 12 h in vacuum (not shown). The most intense peak (labeled #1) in the Si(1 0 0) spectrum is positioned at 98.4 eV, and the peaks labeled #2, and #3 are positioned at 99.1 eV, and 100.1 eV with a shift of 0.7 eV, and 1.7 eV with respect to the most intense peak (peak #4 at the highest binding energy of 100.54 eV has a very small contribution and its position is therefore not reliable). These peak positions agree with the presence of sub-oxides as described by Himpsel et al. [35] and the most intense peak can be ascribed to the Si-bulk. The excitation cross section is larger for Sin+ oxidation states than for Si0, and the relative areas therefore overestimate the sub-oxide contributions. We cannot detect a surface core level state, which would be positioned on either side of the Si2p peak, albeit we might be unable to separate the most intense surface core level peak from the first sub-oxide peak [39,40]. The valence band spectra, which were measured with a photon energy of 40 eV (Fig. 2) do not show a surface state, albeit the surface state is usually, even for a superb surface, relatively weak at these photon energies. After the deposition of Mn, the Si-core level peak is shifted to higher binding energies. All Si2p peak fits were performed independent of each other, and the amount of relative peak shift is established by a comparison of the individual fit results, and the positions of all peaks is used to describe the Si2p positions before and after the Mn-deposition. The most intense peak, which is representative of the Si-bulk atoms is shifted by +0.7–99.1 eV (hn = 160 eV), and by 0.4 eV for hn = 140 eV. The Gaussian and Lorentzian full-width at half maximum (FWHM) are 0.5 0.05 and 0.04 0.015 eV, respectively, and are identical for peak #1 before and after the Mn deposition. This supports the assumption that this component can indeed be attributed to the Si-bulk peak. All other peaks
Fig. 2. The VB spectra for the Si(1 0 0) (2 1) surface and the surface after the deposition of Mn are shown as a function of the angle between surface normal and analyzer. The inset illustrates the geometry for an angle of 08. The states below about 0.5 eV binding energy correspond to the Si-bulk states for the Si(1 0 0) (2 1) surface, and the region close to EF after the deposition of the Mn is dominated by the Fermi-edge of the metallic Mn overlayer.
exhibit a larger FWHM, which can also be seen by inspection of the fitcurves in Fig. 1. The high binding energy peaks related to the sub-oxides are shifted by the same amount for the respective photon energies, their peak positions relative to the most bulk peak is preserved (within a margin of 0.15 eV, a reasonable precision for the fit procedure of low intensity peaks). Peak #2 is slightly shifted towards the Si-bulk peak, and the magnitude of the shift is right at the limit of the precision of the fit. This shift for peak #2, which is also greatly reduced in intensity, might indicate a reaction with the overlying Mn-layer and incorporation of oxygen within the Mn silicide-matrix. The appearance of the silicide peak (labeled A) is coupled with a decrease in the intensity of the peak at +0.7 eV (labeled #2). A reaction of Mn with the surface might occur preferentially on the Si+1 sub-oxide sites and therefore account for the decrease in this peak’s intensity, or a redistribution to the higher oxidation states occurs, which indeed exhibit overall a slight increase in intensity (peaks #3 and #4). The Si-sub-oxides [41] are highly reactive and often react by oxidation of the metallic overlayer, leading to the reduction of Si [42–45]. A similar reaction can be expected with Mn, which is easily oxidized, and has several stable oxidation states. The Si2+1O is the most abundant sub-oxide on our surface, and offers a highly active reaction center for the Mn-adatoms. The dramatic decrease in the intensity of the Si2O peak (peak #2) indicates at first sight a preferential interaction with this oxidation state. However, the relative contributions from the other oxidation states are only around 5% of the total peak area and as such are prone to
C.A. Nolph et al. / Applied Surface Science 255 (2009) 7642–7646
a relatively large error in the fit procedure. Whether their contributions change by a relevant percentage can therefore not be stated with certainty. An investigation of Si surfaces with different relative concentrations of the sub-oxides would answer this question and allow for a direct and qualitative comparison of the kinetics of the Mnreaction with Si-oxides. The addition of a Si2p contribution on the low binding energy side is consistent with the formation of a silicide with the composition MnSi [16] or a manganese rich phase, Mn5Si3 [37]. The binding energy shift with respect to bulk Si is indicative of a charge transfer from Si to Mn. The ratio of Mn to Si bound in silicide is 65:35, and has been determined from the peak areas for the silicide peak (label A) and the Mn3p core level after correction for the respective excitation cross section. This leaves us with two alternative interpretations: firstly, the reaction of Mn with the Si surface leads to the formation of a Mn-rich silicide, or secondly, MnSi is formed at the immediate Mn–Si interface and the remaining Mn is present in its elemental state. The position of silicide related Si2p core level agrees with both types of silicides. The rigid shift of all Si-related peaks to higher binding energies after the deposition of Mn is attributed to a shift in the Fermi level position at the surface, and pinning of EF at a new position due to the formation of a metallic overlayer. The downward band bending is introduced as a reaction to the Fermi level shift at the surface compared to the bulk, and extends into the Si-bulk. It introduces a rigid shift of the substrate core levels. This shift which is slightly smaller for hn = 140 eV, is therefore likely due to the increased surface sensitivity as compared to 160 eV photon energy, whose extension into the bulk probes a larger fraction of the total band bending. In Fig. 2 a set of valence band spectra is shown for the original Si(1 0 0) (2 1) surface and the Si(1 0 0) + Mn. The geometry of the setup is included as an inset in Fig. 2. The angle between the analyzer and the sample normal is varied in 28 increments and a representative selection of VB-spectra is shown [33,46]. The direction of rotation corresponds to a measurement in the surface Brillouin zone along the direction from G to X. The spectra of the Si(1 0 0) surface does not show a surface state related peak below 1 eV, which can be attributed to the presence of sub-oxides and the use of a relatively high photon energy, where the intensity of the surface state is relatively low [39,40]. The peaks positioned at 2.1, 3.2, and 11.2 eV are assigned to bulk states of the Si and they show no dispersion with the variation of the angle, albeit their relative intensities are modified. The double peak between 6.0 and 7.0 eV is a superposition of excitation from Si-bulk states, and the presence of surface oxides. This assignment agrees with results from the core level analysis. A further increase in the surface oxygen concentration, which was done in a separate experiment, leads to a VB spectrum, which is dominated by the peak at 6.9 eV. After the deposition of Mn, the spectra changed dramatically and an intense Fermi edge is observed, a low intensity peak is present at 3.2 eV, and two peaks at 6.9 and 8.2 eV are identified. The Si related contribution at 11.2 eV cannot be detected any more; the spectra in the spectral region adjacent to the Fermi edge dominated by the Mn-containing overlayer [16,37,47]. The VBspectra are unchanged upon variation of the sample angle with respect to the analyzer, which agrees with a polycrystalline nature of the Mn-containing surface layer. The intense ledge-like peak extending from the Fermi edge to about 4 eV reflects the contributions from the 3d-states of Mn, which dominate the spectra due to their large cross section. For further analysis of the overlayer we used the variation of the photon energy, which gives access to deeper lying states, and allows a comparison with a-Mn (figure 3). This phase is obtained by the evaporation and condensation of Mn at room temperature, and the energy dependence of the valence band spectra has been
7645
Fig. 3. The VB spectra after the deposition of Mn as a function of the photon energy. An intense Ferme edge is present and confirms the metallicity of the overlayer.
discussed in detail in [37]. The valence band spectra after the Mndeposition for an angle of 08 and variable photon energy are summarized in Fig. 3. The intense Fermi edge, and the peak at 2.7 eV stem from the Mn d-band, which overwhelms any residual contribution of the Si-p bands in this energy range. The peak at 17 eV which is only visible in the 60 eV spectra is related to the M2,3VV Auger band. The VB-spectra overall agree quite well with those reported for a-Mn as a function of photon energy. The peak around 6.5 eV, however, has two overlapping contributions: one from Si bulk states, and a second one which can be assigned to residual surface oxides, and its intensity is reduced considerably with higher photon energy and therefore larger information depth and decreasing cross section. Only the lack of a pronounced minimum at about 2 eV might indicate the presence of another material, and could be related to the contribution from Si-p valence band states, Mn–Si, or Mn–Si–O hybrid bands at the interface. However, most recent calculations of Mn-silicide density of states indicate that the hybridization leads to additional states at energies between 1 and 2 eV [16], which has been shown experimentally for Mn5Si3 single crystals. We therefore suggest as a tentative model that a silicide is formed at the boundary between Si and the Mn-overlayer, while the overlayer itself is dominated by metallic Mn. The STM images from our previous study show that clusters rather then continuous layers cover the Si(1 0 0) surface after Mn deposition, which is in contrast to the layer-growth described for silicides on Si(1 1 1). These clusters would then consist of Mn, while their contact area to the Si surface is a silicide of as yet unknown composition. As a consequence of our observations, the formation of ferromagnetic contacts which require ideally a flat and ferromagnetic overlayer with the desired composition, cannot be achieved through room temperature growth of Mn-silicides. The presence of the antiferromagnetic Mn-metal and the cluster formation are
7646
C.A. Nolph et al. / Applied Surface Science 255 (2009) 7642–7646
detrimental to the formation of a good ferromagnetic contact to the Si(1 0 0). 4. Conclusions In summary, the reaction of Mn with the Si(1 0 0) surface leads at room temperature to the formation of a thin silicide interface, presumably with MnSi stoichiometry. The overlayer is dominated by Mn, either as Mn-metal or as a Mn-rich silicide phase, and thus will not provide the desired ferromagnetic contact-layer which is the prerequisite for the formation of spin-injection layers. In contrast to the Si(1 1 1) 7 7 surface, the overlayer forms clusters rather then well-defined single silicide layers. The presence of surface sub-oxides, which were observed during the XPS study, can conceivably function as surface traps for Mn and might influence the interfacial bonding and extent of Fermi level pinning at the interface. Acknowledgements The authors gratefully acknowledge the support of this work by the Nanoelectronics Research Initiative (NRI) in conjunction with NSF, and the support of this work by the NSF-supplement to the MRSEC ‘‘Center for Nanoscopic Materials Design’’ at the University of Virginia. Use of the National Synchrotron Light Source, Brookhaven National Laboratory, was supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract No. DE-AC02-98CH10886. References [1] W. Speier, E. Vonleuken, J.C. Fuggle, D.D. Sarma, L. Kumar, B. Dauth, K.H.J. Buschow, Photoemission and inverse photoemission of transition-metal silicides, Phys. Rev. B 39 (1989) 6008. [2] J.H. Weaver, A. Franciosi, V.L. Moruzzi, Bonding in metal disilicides Casi2 through Nisi2—experiment and theory, Phys. Rev. B 29 (1984) 3293. [3] K.W. Edmonds, Ferromagnetic nanodevices based on (Ga,Mn)As, Curr. Opin. Solid State Mater. Sci. 10 (2006) 108. [4] C.F. Hirjibehedin, C.P. Lutz, A.J. Heinrich, Spin coupling in engineered atomic structures, Science 312 (2006) 1021. [5] A.M. Nazmul, S. Kobayashi, S. Sugahara, M. Tanaka, Control of ferromagnetism in Mn delta-doped GaAs-based semiconductor heterostructures, Phys. E 21 (2004) 937. [6] A.J.R.d. Silva, A. Fazzio, Stabilization of substitutional Mn in silicon-based semiconductors, Phys. Rev. B 70 (2004) 193205. [7] S.A. Wolf, D.D. Awschalom, R.A. Buhrman, J.M. Daughton, S.v. Molnar, M.L. Roukes, A.Y. Chtchelkanova, D.M. Treger, Spintronics: a spin-based electronics vision for the future, Science 294 (2001) 1488. [8] S. Kuroda, N. Nishizawa, K. Takita, M. Mitome, Y. Bando, K. Osuch, T. Dietl, Origin and control of high-temperature ferromagnetism in semiconductors, Nat. Mater. 6 (2007) 440. [9] X. Luo, S.B. Zhang, S.-H. Wei, Theory of Mn supersaturation in Si and Ge, Phys. Rev. B 70 (2004) 033308. [10] Y.D. Park, A.T. Hanbicki, S.C. Erwin, C.S. Hellberg, J.M. Sullivan, J.E. Mattson, T.F. Ambrose, A. Wilson, G. Spanos, B.T. Jonker, A group-IV ferromagnetic semiconductor: MxGe1 x, Science 295 (2002) 651. [11] M.C. Qian, C.Y. FOng, K. Liu, W.E. Pickett, J.E. Pask, L.H. Yang, Half-metallic digital ferromagnetic heterostructures composed of delta-doped layer of Mn in Si, Phys. Rev. Lett. 96 (2006) 27211. [12] M. Krause, A. Stollenwerk, M. Licurse, V.P. LaBella, Ostwald ripening of manganese silicide islands on Si(0 0 1), J. Vac. Sci. Technol. 24 (2006) 1480. [13] Y.C. Lian, L.J. Chen, Localized epitaxial growth of MnSi1.7 on silicon, Appl. Phys. Lett. 48 (1986) 359. [14] S. Teichert, D.K. Sarkar, S. schwendler, H. Giesler, A. Mogilatenko, M. Falke, G. Beddies, J.-J. Hinneberg, Preparation and properties of MnSi1.7 on Si(1 0 0), Microelectron. Eng. 55 (2001) 227. [15] S. Teichert, S. Schwendler, D.K. Sarkar, A. Mogilatenko, M. Falke, G. Beddies, C. Kleint, H.-J. Hinneberg, Growth of MnSi1.7 on Si(0 0 1) by MBE, J. Cryst. Growth 227–228 (2001) 882.
[16] E. Carleschi, E. Magnano, M. Zangrando, F. Bondino, A. Nicolaou, F. Carbone, D. Van der Marel, F. Parmigiani, Manganese silicide single crystal and films deposited on Si(1 1 1): a comparative spectroscopic study, Surf. Sci. 601 (2007) 4066. [17] M. Hortamani, H. Wu, P. Kratzer, M. Scheffler, Epitaxy of Mn on Si(0 0 1): adsorption, surface diffusion, and magnetic properties studied by density-functional theory, Phys. Rev. B 74 (2006) 205305. [18] M.R. Krause, A.J. Stollenwerk, J. Reed, V.P. LaBella, Electronic structure changes of Si(0 0 1)-(2 1) from subsurface Mn observed by STM, Phys. Rev. B 75 (2007) 205326. [19] H. Liu, P. Reinke, The growth of manganese nanostructures on the Si(1 0 0)(2 1) surface: observation of the sub-monolayer regime with scanning tunneling microscopy, Surface Science 602 (2008) 986. [20] E. Magnano, E. Carleschi, A. Nicolaou, T. Pardini, M. Zangrando, F. Parmigiani, Growth of manganese silicide films by co-deposition of Mn and Si on Si(1 1 1): a spectroscopic and morphological investigation, Surf. Sci. 600 (2006) 3932. [21] G. Profeta, S. Picozzi, A. Continenza, First-stage Mn adsorption on clean Ge(1 1 1), Phys. Rev. B 70 (2004) 155307. [22] H. Wu, M. Hortamani, P. Kratzer, M. Scheffler, First-principles study of ferromagnetism in epitaxial Si–Mn thin films on Si(1 0 0), Phys. Rev. Lett. 92 (2004) 137202. [23] H. Wu, P. Kratzer, M. Scheffler, Density-functional theory study of half-metallic heterostructures: interstitial Mn in Si, Phys. Rev. Lett. 98 (2007) 117202. [24] L. Zhang, D.G. Ivey, Low temperature reaction of thin layers of Mn on Si, J. Mater. Res. 6 (1991) 1518. [25] A. Kumar, M. Tallarida, M. Hausmann, U. Starke, K. Horn, Thin manganese films on Si(1 1 1)-(7 7): electronic structure and strain in silicide formation, J. Phys. D— Appl. Phys. 37 (2004) 1083. [26] J.J. Yeh, I. Lindau, At. Data Nucl. Data Tables 32 (1985) 1. [27] E. Vescovo, H.-J. Kim, Q.-Y. Dong, G. Nintzel, D. Carlson, S. Hulbert, N.V. Smith, U5UA: a new high-resolution undulator beamline at the NSLS for spin-resolved photoemission spectroscopy, Synchrotron Radiat. News 12 (1999) 10. [28] http://escalab.snu.ac.kr/berd/Fitt/fitt.html. [29] K. Levenberg, A method for the solution of certain non-linear problems in leastsquares, Quart. Appl. Math. 2 (1944) 164. [30] D.W. Marquardt, An algorithm for the least-square estimation of nonlinear parameters, SIAM J. Appl. Math. 11 (1963) 431. [31] S. Doniach, M. Sunjic, J. Phys. C 3 (1970) 285. [32] D.A. Shirley, Phys. Rev. B 5 (1972) 4709. [33] U.A. Effner, D. Badt, J. Binder, T. Bertrams, A. Brodde, C. Lunau, H. Neddermeyer, M. Hanbucken, Photoemission and scanning tunneling microscopy on K/Si(1 0 0), Surf. Sci. 277 (1992) 207. [34] Y. Enta, S. Suzuki, S. Kono, Angle-resolved photoemission study of the electronic structure of the Si(0 0 1) (4 2) surface, Phys. Rev. Lett. 65 (1990) 2704. [35] F.J. Himpsel, F.R. Mcfeely, A. Talebibrahimi, J.A. Yarmoff, G. Hollinger, Microscopic structure of the SiO2/Si interface, Phys. Rev. B 38 (1988) 6084. [36] A.V. Borne, R.L. Johnson, B. Sonntag, M. Talkenberg, A. Verweyen, P. Wernet, J. Schulz, K. Tiedtke, C. Gerth, B. Obst, P. Zimmermann, J.E. Hansen, Suppression of the low-spin multiplet components in the 3p photoelectron spectra of atomic and solid 3d metals, Phys. Rev. A 62 (2000) 052703. [37] A. Irizawa, A. Yamasaki, M. Okazaki, S. Kasai, A. Sekiyama, S. Imada, S. Suga, E. Kulatov, H. Ohta, T. Nanba, Bulk-sensitive photoemission of Mn5Si3, Solid State Commun. 124 (2002) 1. [38] M. Oku, S. Suzuki, N. Ohtsu, T. Shishido, K. Wagatsuma, Comparison of intrinsic zero-energy loss and Shirley-type background corrected profiles of XPS spectra for quantitative surface analysis: study of Cr, Mn and Fe oxides, Appl. Surf. Sci. 254 (2008) 5141. [39] E.L. Bullock, R. Gunnella, L. Patthey, T. Abukawa, S. Kono, C.R. Natoli, L.S.O. Johansson, Surface core level photoelectron diffraction from Si-dimers at the Si(0 0 1)2 1 surface, Phys. Rev. Lett. 74 (1995) 2756. [40] E. Landemark, C.J. Karlsson, Y.-C. Chao, R.I.G. Uhrberg, Core level spectroscopy of the clean Si(1 0 0) surface: charge transfer within assymmetric dimer of the 2 1 and c(4 2) reconstructions, Phys. Rev. Lett. 69 (1992) 1588. [41] R.Q. Zhang, T.S. Chu, H.F. Cheung, N. Wang, S.T. Lee, High reactivity of silicon suboxide clusters, Phys. Rev. B 64 (2001) 113304. [42] C. Anandan, X-ray photoelectron spectroscopic study of room-temperature evolution of oxide-covered hydrogenated amorphous silicon/aluminum interface, Appl. Surf. Sci. 89 (1995) 57. [43] N. Benouattas, A. Mosser, A. Bouabellou, Surface morphology and reaction at Cu/Si interface—effect of native silicon suboxide, Appl. Surf. Sci. 252 (2006) 7572. [44] S. IIda, S. Abe, Ti thin-film reaction on SiO2/Si, Appl. Surf. Sci. 78 (1994) 141. [45] S. Kennou, S. Ladas, M.G. Grimaldi, T.A.N. Tan, J.Y. Veuillen, Oxidation of thin erbium and erbium silicide overlayers in contact with silicon oxide films thermally grown on silicon, Appl. Surf. Sci. 102 (1996) 142. [46] L.S.O. Johansson, R.I.G. Uhrberg, P. Martensson, B.V. Hansson, Surface-state band structure of the Si(1 0 0)-2 1 surface studied with polarization-dependent angle resolved photoemission on single domain surfaces, Phys. Rev. B 42 (1990) 1305. [47] H. Sugawara, A. Kakizaki, I. Nagakura, T. Ishii, Extreme ultraviolet photoelectronspectra of alpha-manganese, J. Phys. F Met. Phys. 12 (1982) 2929.
Contents lists available at ScienceDirect
Applied Surface Science journal homepage: www.elsevier.com/locate/apsusc
The growth of manganese layers on Si(1 0 0) at room temperature: A photoelectron spectroscopy study C.A. Nolph a, E. Vescovo b, P. Reinke a,* a b
University of Virgina, Department of Materials Science and Engineering, 395 McCormick Road, Charlottesville, VA 22904, USA National Synchrotron Light Source, Brookhaven National Laboratory, Upton, NY 11973-5000, USA
A R T I C L E I N F O
A B S T R A C T
Article history: Received 1 September 2008 Received in revised form 13 April 2009 Accepted 13 April 2009 Available online 21 April 2009
The combination of spin-and charge based electronics in future devices requires the magnetic doping of group IV semiconductors, and the formation of ferromagnetic contacts. The doping of Mn with Si is one of the material systems which is discussed in this context. The present study focuses on the growth of Mn on a Si(100)(2x1) surface, and the evolution of the surface was observed as a function of Mn coverage with synchrotron-based photoelectron spectroscopy. The reaction of Mn with the Si(100) surface at room temperature leads the formation of silicide at the boundary between the Si substrate and the Mnoverlayer, presumably with MnSi stoichiometry. The residual sub-oxide reacts with the Mn and therefore incorporates a few percent of Mn-O-Si at the interface. The analysis of the sub-oxide composition indicates that the Si+1 component is the most reactive oxidation state. The overlayer is dominated by Mn, either as Mn-metal or as a Mn-rich silicide phase, and the metallic layer introduces a band bending in Si. As a consequence of our observations, including information from a recent STM study, the formation of ferromagnetic contacts which require ideally a flat and compositionally homogenous overlayer, cannot be achieved through room temperature deposition of Mn on the Si(100) (2x1) surface. The influence of residual oxides and surface defects on the growth process will be further investigated. ß 2009 Published by Elsevier B.V.
Keywords: Manganese Silicon(1 0 0) Silicide Photoelectron spectroscopy Spintronics
1. Introduction The combination of group IV semiconductors with transition metal elements has been an active area of research for decades and was mostly geared towards the development of metal contacts and silicide leads [1,2]. With the advent of spintronics the interest has shifted to the doping of semiconductors with transition metal elements which can lead to the formation of dilute magnetic semiconductors [3–8]. The coupling between the magnetic moment of the transition metal atoms is then mediated by holes, and the magnetism can be tuned by the application of an electric field. The second driving force to study the formation of transition metal overlayers is given by the interest in ferromagnetic contact layers to provide the injection of a spin-polarized current into the semiconductor. Some MnSi compounds exhibit ferromagnetism, and the formation of a Mn-silicide overlayer has therefore been suggested as a route to achieve the spin-injection into group IV semiconductors. The introduction of Mn, which has a half filled d-shell and carries a large magnetic moment, into a Si lattice is highly
* Corresponding author. E-mail address: [email protected] (P. Reinke). 0169-4332/$ – see front matter ß 2009 Published by Elsevier B.V. doi:10.1016/j.apsusc.2009.04.047
coveted, and would allow the seamless combination of spintronics and charge based electronics. However, the doping of Si is hampered by the extremely low solubility of Mn in Si, the preference of Mn for interstitial lattice sites over the electronically active substitutional sites, and the competition with silicide formation [6,9–15]. In order to circumvent these problems we are investigating the feasibility of a surface-driven approach for the incorporation of Mn into a Si matrix. This approach has been suggested in several theoretical studies, which indicate that selected Mn–Si surface structures, and delta-doped layers of Mn embedded in Si might indeed be ferromagnetic [11,16–23]. A set of recent STM studies has been geared towards the investigation of the formation of Mn surface structures on the Si(1 0 0) surface [18–20,24]. The formation of silicide crystallites is observed at elevated temperatures, but 2D surface structures can evolve for lower substrate temperatures (below about 300 8C) [12,14,16,25]. The study presented here explores the bonding at the interface, and the growth of the overlayer during the room temperature deposition of Mn on Si(1 0 0) with photoelectron spectroscopy, which examines the core level and valence band. The PES analysis leads us to conclude that an ultrathin interface between Mn and the Si surface is formed, which is covered by a layer of Mn clusters. The role of the Si-sub-oxide is discussed in detail.
C.A. Nolph et al. / Applied Surface Science 255 (2009) 7642–7646
2. Experimental A Si(1 0 0) wafer was cleaned by repeated annealing cycles reaching temperatures of about 1000 8C, which were achieved by electron bombardment of the sample backside. The presence of the (2 1) reconstruction was confirmed by the presence of sharp reflexes in the LEED pattern, and the cleaning cycles were repeated until no further improvement was observable in the LEED pattern. Mn was subsequently deposited from an electron beam evaporation source (EFM3 Omicron Nanotechnology) and about 1 ML was deposited on the Si(1 0 0) sample. The coverage was determined by the XPS analysis of the relative peak areas of the Mn3p and Si2p core level using the excitation cross sections given by Yeh and Lindau [26]. The Fermi energy position or the position of the valence band maximum are used as energy reference. The photoelectron spectroscopy was performed at beamline U5UA of the National Synchrotron Light Source at Brookhaven National Laboratory [27]. U5UA is an undulator beamline, which is equipped with four gratings for optimum operation in the photon energy range between 15 and 200 eV. The light is linearly polarized in the horizontal plane, and the analyzer is fixed and its axis is at 458 with respect to the light incidence. The angle between sample surface normal and analyzer is modified by rotation of the sample, consequently the angle between sample surface normal and the incoming light is modified at the same time. The geometry is shown in the insert in Fig. 3.
7643
For the fit of the core level peaks we use the program fit [28], which employs the Marquard–Levenberg algorithm for the least square fit procedure [29,30]. The peaks are described by Doniach– Sunjic functions and a Shirley background, and the quality of the fit is judged by the quality of the residuals, and the ability of the fit to describe sharp saddle points and minima of the experimental curve [31,32]. A branching ratio of 0.6 and an energy separation of 0.5 eV was used for the Si2p1/2 and Si2p3/2 components. The smallest number of peaks, which gives a good fit of the experimental data, is used for analysis of the data. 3. Results and discussion Fig. 1a shows the Si2p core level (solid lines) before and after the deposition of Mn for a photon energy of 160 eV and includes the results of the peak fit procedure (dotted lines). An excellent fit can be obtained for the Si(1 0 0) surface prior to Mn deposition by using four peaks. The results for both photon energies (140 and 160 eV) yield the same binding energies within a margin of 0.1 eV. A LEED diffraction pattern, given as an inset in Fig. 1, confirms the presence of the (2 1) reconstruction and sharp, welldefined spots are observed prior to the deposition of Mn. After the deposition of Mn five peaks are required to achieve an equally excellent description of the Si2p peak. The structural analysis of thin films produced by the deposition of Mn on Si(1 0 0) (2 1) was recently performed with STM, and is
Fig. 1. On the left-hand side of the figure the Si2p core level for the Si(1 0 0) (2 1) surface, and after the Si2p after the deposition of Mn are shown. The results of the fit procedure are included and the labels of the peaks are indicated in the figure. The LEED pattern of the Si surface is shown as an inset and confirms the presence of the (2 1) reconstruction. A second inset shows the STM image measured after the deposition of 1.2 ML of Mn on Si, which was described in detail in a recent publication [19]. On the right-hand side, the area of all fit-curves is shown before and after the deposition of Mn. The respective Mn3p peak is shown at the bottom.
7644
C.A. Nolph et al. / Applied Surface Science 255 (2009) 7642–7646
published in [19]. The Si(1 0 0) (2 1) showed a low number of defects and was an excellent surface exhibiting a long-range reconstruction. For coverages below about 0.3 ML the formation of monoatomic Mn-chains are observed, which are oriented perpendicular to the Si-dimer rows. However, for higher coverages exceeding a ML of Mn relatively large, mostly elongated, flat clusters are present on the surface, but the terraces of the underlying Si(1 0 0) surface can still be discerned. An STM image obtained for those coverages is included as an insert in Fig. 1. The clusters still retain a slight preference for an orientation of their long axis to run perpendicular to the dimer rows of the Si(1 0 0) (2 1) reconstruction, which is deduced from their orientation with respect to the terrace edge. In Fig. 1 (right-hand side) the relative peak intensities of the Si2p core level contributions are summarized, and the peak numbers begin with the most intense peak, which is positioned at 98.4 eV for Si(1 0 0) and at 99.1 eV after the deposition of Mn [33– 35]. The contribution positioned at the lowest binding energy is the most intense peak for Si(1 0 0). After the deposition of Mn an additional peak with a contribution of nearly 20% to the total peak area is observed on the low binding energy side of the Si2p peak. The emergence of this additional contribution (labeled A in the figure) is readily apparent upon inspection and leads to a shallower slope of the low binding energy side of the Si2p core level. The width of the Fermi edge/valence band edge is conserved for all measurements, and the changes in the Si2p core level are therefore not caused by a modification of the spectral resolution. The corresponding Mn3p core level, which is also included in Fig. 1, is positioned at 46.9 eV, and cannot be described reliably by a fit procedure due to the presence of multiple overlapping peaks from the spin-orbit and multiplet splitting typical for 3d transition metals [20,36–38]. The position of the main peak is commensurate with either Mn or Mn-silicides, or a mixture of both. The formation of Mn-oxides, which introduces a large chemical shift, can be discerned by the emergence of a shoulder on the high binding energy side of the Mn3p peak, which has been observed in samples stored for more then 12 h in vacuum (not shown). The most intense peak (labeled #1) in the Si(1 0 0) spectrum is positioned at 98.4 eV, and the peaks labeled #2, and #3 are positioned at 99.1 eV, and 100.1 eV with a shift of 0.7 eV, and 1.7 eV with respect to the most intense peak (peak #4 at the highest binding energy of 100.54 eV has a very small contribution and its position is therefore not reliable). These peak positions agree with the presence of sub-oxides as described by Himpsel et al. [35] and the most intense peak can be ascribed to the Si-bulk. The excitation cross section is larger for Sin+ oxidation states than for Si0, and the relative areas therefore overestimate the sub-oxide contributions. We cannot detect a surface core level state, which would be positioned on either side of the Si2p peak, albeit we might be unable to separate the most intense surface core level peak from the first sub-oxide peak [39,40]. The valence band spectra, which were measured with a photon energy of 40 eV (Fig. 2) do not show a surface state, albeit the surface state is usually, even for a superb surface, relatively weak at these photon energies. After the deposition of Mn, the Si-core level peak is shifted to higher binding energies. All Si2p peak fits were performed independent of each other, and the amount of relative peak shift is established by a comparison of the individual fit results, and the positions of all peaks is used to describe the Si2p positions before and after the Mn-deposition. The most intense peak, which is representative of the Si-bulk atoms is shifted by +0.7–99.1 eV (hn = 160 eV), and by 0.4 eV for hn = 140 eV. The Gaussian and Lorentzian full-width at half maximum (FWHM) are 0.5 0.05 and 0.04 0.015 eV, respectively, and are identical for peak #1 before and after the Mn deposition. This supports the assumption that this component can indeed be attributed to the Si-bulk peak. All other peaks
Fig. 2. The VB spectra for the Si(1 0 0) (2 1) surface and the surface after the deposition of Mn are shown as a function of the angle between surface normal and analyzer. The inset illustrates the geometry for an angle of 08. The states below about 0.5 eV binding energy correspond to the Si-bulk states for the Si(1 0 0) (2 1) surface, and the region close to EF after the deposition of the Mn is dominated by the Fermi-edge of the metallic Mn overlayer.
exhibit a larger FWHM, which can also be seen by inspection of the fitcurves in Fig. 1. The high binding energy peaks related to the sub-oxides are shifted by the same amount for the respective photon energies, their peak positions relative to the most bulk peak is preserved (within a margin of 0.15 eV, a reasonable precision for the fit procedure of low intensity peaks). Peak #2 is slightly shifted towards the Si-bulk peak, and the magnitude of the shift is right at the limit of the precision of the fit. This shift for peak #2, which is also greatly reduced in intensity, might indicate a reaction with the overlying Mn-layer and incorporation of oxygen within the Mn silicide-matrix. The appearance of the silicide peak (labeled A) is coupled with a decrease in the intensity of the peak at +0.7 eV (labeled #2). A reaction of Mn with the surface might occur preferentially on the Si+1 sub-oxide sites and therefore account for the decrease in this peak’s intensity, or a redistribution to the higher oxidation states occurs, which indeed exhibit overall a slight increase in intensity (peaks #3 and #4). The Si-sub-oxides [41] are highly reactive and often react by oxidation of the metallic overlayer, leading to the reduction of Si [42–45]. A similar reaction can be expected with Mn, which is easily oxidized, and has several stable oxidation states. The Si2+1O is the most abundant sub-oxide on our surface, and offers a highly active reaction center for the Mn-adatoms. The dramatic decrease in the intensity of the Si2O peak (peak #2) indicates at first sight a preferential interaction with this oxidation state. However, the relative contributions from the other oxidation states are only around 5% of the total peak area and as such are prone to
C.A. Nolph et al. / Applied Surface Science 255 (2009) 7642–7646
a relatively large error in the fit procedure. Whether their contributions change by a relevant percentage can therefore not be stated with certainty. An investigation of Si surfaces with different relative concentrations of the sub-oxides would answer this question and allow for a direct and qualitative comparison of the kinetics of the Mnreaction with Si-oxides. The addition of a Si2p contribution on the low binding energy side is consistent with the formation of a silicide with the composition MnSi [16] or a manganese rich phase, Mn5Si3 [37]. The binding energy shift with respect to bulk Si is indicative of a charge transfer from Si to Mn. The ratio of Mn to Si bound in silicide is 65:35, and has been determined from the peak areas for the silicide peak (label A) and the Mn3p core level after correction for the respective excitation cross section. This leaves us with two alternative interpretations: firstly, the reaction of Mn with the Si surface leads to the formation of a Mn-rich silicide, or secondly, MnSi is formed at the immediate Mn–Si interface and the remaining Mn is present in its elemental state. The position of silicide related Si2p core level agrees with both types of silicides. The rigid shift of all Si-related peaks to higher binding energies after the deposition of Mn is attributed to a shift in the Fermi level position at the surface, and pinning of EF at a new position due to the formation of a metallic overlayer. The downward band bending is introduced as a reaction to the Fermi level shift at the surface compared to the bulk, and extends into the Si-bulk. It introduces a rigid shift of the substrate core levels. This shift which is slightly smaller for hn = 140 eV, is therefore likely due to the increased surface sensitivity as compared to 160 eV photon energy, whose extension into the bulk probes a larger fraction of the total band bending. In Fig. 2 a set of valence band spectra is shown for the original Si(1 0 0) (2 1) surface and the Si(1 0 0) + Mn. The geometry of the setup is included as an inset in Fig. 2. The angle between the analyzer and the sample normal is varied in 28 increments and a representative selection of VB-spectra is shown [33,46]. The direction of rotation corresponds to a measurement in the surface Brillouin zone along the direction from G to X. The spectra of the Si(1 0 0) surface does not show a surface state related peak below 1 eV, which can be attributed to the presence of sub-oxides and the use of a relatively high photon energy, where the intensity of the surface state is relatively low [39,40]. The peaks positioned at 2.1, 3.2, and 11.2 eV are assigned to bulk states of the Si and they show no dispersion with the variation of the angle, albeit their relative intensities are modified. The double peak between 6.0 and 7.0 eV is a superposition of excitation from Si-bulk states, and the presence of surface oxides. This assignment agrees with results from the core level analysis. A further increase in the surface oxygen concentration, which was done in a separate experiment, leads to a VB spectrum, which is dominated by the peak at 6.9 eV. After the deposition of Mn, the spectra changed dramatically and an intense Fermi edge is observed, a low intensity peak is present at 3.2 eV, and two peaks at 6.9 and 8.2 eV are identified. The Si related contribution at 11.2 eV cannot be detected any more; the spectra in the spectral region adjacent to the Fermi edge dominated by the Mn-containing overlayer [16,37,47]. The VBspectra are unchanged upon variation of the sample angle with respect to the analyzer, which agrees with a polycrystalline nature of the Mn-containing surface layer. The intense ledge-like peak extending from the Fermi edge to about 4 eV reflects the contributions from the 3d-states of Mn, which dominate the spectra due to their large cross section. For further analysis of the overlayer we used the variation of the photon energy, which gives access to deeper lying states, and allows a comparison with a-Mn (figure 3). This phase is obtained by the evaporation and condensation of Mn at room temperature, and the energy dependence of the valence band spectra has been
7645
Fig. 3. The VB spectra after the deposition of Mn as a function of the photon energy. An intense Ferme edge is present and confirms the metallicity of the overlayer.
discussed in detail in [37]. The valence band spectra after the Mndeposition for an angle of 08 and variable photon energy are summarized in Fig. 3. The intense Fermi edge, and the peak at 2.7 eV stem from the Mn d-band, which overwhelms any residual contribution of the Si-p bands in this energy range. The peak at 17 eV which is only visible in the 60 eV spectra is related to the M2,3VV Auger band. The VB-spectra overall agree quite well with those reported for a-Mn as a function of photon energy. The peak around 6.5 eV, however, has two overlapping contributions: one from Si bulk states, and a second one which can be assigned to residual surface oxides, and its intensity is reduced considerably with higher photon energy and therefore larger information depth and decreasing cross section. Only the lack of a pronounced minimum at about 2 eV might indicate the presence of another material, and could be related to the contribution from Si-p valence band states, Mn–Si, or Mn–Si–O hybrid bands at the interface. However, most recent calculations of Mn-silicide density of states indicate that the hybridization leads to additional states at energies between 1 and 2 eV [16], which has been shown experimentally for Mn5Si3 single crystals. We therefore suggest as a tentative model that a silicide is formed at the boundary between Si and the Mn-overlayer, while the overlayer itself is dominated by metallic Mn. The STM images from our previous study show that clusters rather then continuous layers cover the Si(1 0 0) surface after Mn deposition, which is in contrast to the layer-growth described for silicides on Si(1 1 1). These clusters would then consist of Mn, while their contact area to the Si surface is a silicide of as yet unknown composition. As a consequence of our observations, the formation of ferromagnetic contacts which require ideally a flat and ferromagnetic overlayer with the desired composition, cannot be achieved through room temperature growth of Mn-silicides. The presence of the antiferromagnetic Mn-metal and the cluster formation are
7646
C.A. Nolph et al. / Applied Surface Science 255 (2009) 7642–7646
detrimental to the formation of a good ferromagnetic contact to the Si(1 0 0). 4. Conclusions In summary, the reaction of Mn with the Si(1 0 0) surface leads at room temperature to the formation of a thin silicide interface, presumably with MnSi stoichiometry. The overlayer is dominated by Mn, either as Mn-metal or as a Mn-rich silicide phase, and thus will not provide the desired ferromagnetic contact-layer which is the prerequisite for the formation of spin-injection layers. In contrast to the Si(1 1 1) 7 7 surface, the overlayer forms clusters rather then well-defined single silicide layers. The presence of surface sub-oxides, which were observed during the XPS study, can conceivably function as surface traps for Mn and might influence the interfacial bonding and extent of Fermi level pinning at the interface. Acknowledgements The authors gratefully acknowledge the support of this work by the Nanoelectronics Research Initiative (NRI) in conjunction with NSF, and the support of this work by the NSF-supplement to the MRSEC ‘‘Center for Nanoscopic Materials Design’’ at the University of Virginia. Use of the National Synchrotron Light Source, Brookhaven National Laboratory, was supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract No. DE-AC02-98CH10886. References [1] W. Speier, E. Vonleuken, J.C. Fuggle, D.D. Sarma, L. Kumar, B. Dauth, K.H.J. Buschow, Photoemission and inverse photoemission of transition-metal silicides, Phys. Rev. B 39 (1989) 6008. [2] J.H. Weaver, A. Franciosi, V.L. Moruzzi, Bonding in metal disilicides Casi2 through Nisi2—experiment and theory, Phys. Rev. B 29 (1984) 3293. [3] K.W. Edmonds, Ferromagnetic nanodevices based on (Ga,Mn)As, Curr. Opin. Solid State Mater. Sci. 10 (2006) 108. [4] C.F. Hirjibehedin, C.P. Lutz, A.J. Heinrich, Spin coupling in engineered atomic structures, Science 312 (2006) 1021. [5] A.M. Nazmul, S. Kobayashi, S. Sugahara, M. Tanaka, Control of ferromagnetism in Mn delta-doped GaAs-based semiconductor heterostructures, Phys. E 21 (2004) 937. [6] A.J.R.d. Silva, A. Fazzio, Stabilization of substitutional Mn in silicon-based semiconductors, Phys. Rev. B 70 (2004) 193205. [7] S.A. Wolf, D.D. Awschalom, R.A. Buhrman, J.M. Daughton, S.v. Molnar, M.L. Roukes, A.Y. Chtchelkanova, D.M. Treger, Spintronics: a spin-based electronics vision for the future, Science 294 (2001) 1488. [8] S. Kuroda, N. Nishizawa, K. Takita, M. Mitome, Y. Bando, K. Osuch, T. Dietl, Origin and control of high-temperature ferromagnetism in semiconductors, Nat. Mater. 6 (2007) 440. [9] X. Luo, S.B. Zhang, S.-H. Wei, Theory of Mn supersaturation in Si and Ge, Phys. Rev. B 70 (2004) 033308. [10] Y.D. Park, A.T. Hanbicki, S.C. Erwin, C.S. Hellberg, J.M. Sullivan, J.E. Mattson, T.F. Ambrose, A. Wilson, G. Spanos, B.T. Jonker, A group-IV ferromagnetic semiconductor: MxGe1 x, Science 295 (2002) 651. [11] M.C. Qian, C.Y. FOng, K. Liu, W.E. Pickett, J.E. Pask, L.H. Yang, Half-metallic digital ferromagnetic heterostructures composed of delta-doped layer of Mn in Si, Phys. Rev. Lett. 96 (2006) 27211. [12] M. Krause, A. Stollenwerk, M. Licurse, V.P. LaBella, Ostwald ripening of manganese silicide islands on Si(0 0 1), J. Vac. Sci. Technol. 24 (2006) 1480. [13] Y.C. Lian, L.J. Chen, Localized epitaxial growth of MnSi1.7 on silicon, Appl. Phys. Lett. 48 (1986) 359. [14] S. Teichert, D.K. Sarkar, S. schwendler, H. Giesler, A. Mogilatenko, M. Falke, G. Beddies, J.-J. Hinneberg, Preparation and properties of MnSi1.7 on Si(1 0 0), Microelectron. Eng. 55 (2001) 227. [15] S. Teichert, S. Schwendler, D.K. Sarkar, A. Mogilatenko, M. Falke, G. Beddies, C. Kleint, H.-J. Hinneberg, Growth of MnSi1.7 on Si(0 0 1) by MBE, J. Cryst. Growth 227–228 (2001) 882.
[16] E. Carleschi, E. Magnano, M. Zangrando, F. Bondino, A. Nicolaou, F. Carbone, D. Van der Marel, F. Parmigiani, Manganese silicide single crystal and films deposited on Si(1 1 1): a comparative spectroscopic study, Surf. Sci. 601 (2007) 4066. [17] M. Hortamani, H. Wu, P. Kratzer, M. Scheffler, Epitaxy of Mn on Si(0 0 1): adsorption, surface diffusion, and magnetic properties studied by density-functional theory, Phys. Rev. B 74 (2006) 205305. [18] M.R. Krause, A.J. Stollenwerk, J. Reed, V.P. LaBella, Electronic structure changes of Si(0 0 1)-(2 1) from subsurface Mn observed by STM, Phys. Rev. B 75 (2007) 205326. [19] H. Liu, P. Reinke, The growth of manganese nanostructures on the Si(1 0 0)(2 1) surface: observation of the sub-monolayer regime with scanning tunneling microscopy, Surface Science 602 (2008) 986. [20] E. Magnano, E. Carleschi, A. Nicolaou, T. Pardini, M. Zangrando, F. Parmigiani, Growth of manganese silicide films by co-deposition of Mn and Si on Si(1 1 1): a spectroscopic and morphological investigation, Surf. Sci. 600 (2006) 3932. [21] G. Profeta, S. Picozzi, A. Continenza, First-stage Mn adsorption on clean Ge(1 1 1), Phys. Rev. B 70 (2004) 155307. [22] H. Wu, M. Hortamani, P. Kratzer, M. Scheffler, First-principles study of ferromagnetism in epitaxial Si–Mn thin films on Si(1 0 0), Phys. Rev. Lett. 92 (2004) 137202. [23] H. Wu, P. Kratzer, M. Scheffler, Density-functional theory study of half-metallic heterostructures: interstitial Mn in Si, Phys. Rev. Lett. 98 (2007) 117202. [24] L. Zhang, D.G. Ivey, Low temperature reaction of thin layers of Mn on Si, J. Mater. Res. 6 (1991) 1518. [25] A. Kumar, M. Tallarida, M. Hausmann, U. Starke, K. Horn, Thin manganese films on Si(1 1 1)-(7 7): electronic structure and strain in silicide formation, J. Phys. D— Appl. Phys. 37 (2004) 1083. [26] J.J. Yeh, I. Lindau, At. Data Nucl. Data Tables 32 (1985) 1. [27] E. Vescovo, H.-J. Kim, Q.-Y. Dong, G. Nintzel, D. Carlson, S. Hulbert, N.V. Smith, U5UA: a new high-resolution undulator beamline at the NSLS for spin-resolved photoemission spectroscopy, Synchrotron Radiat. News 12 (1999) 10. [28] http://escalab.snu.ac.kr/berd/Fitt/fitt.html. [29] K. Levenberg, A method for the solution of certain non-linear problems in leastsquares, Quart. Appl. Math. 2 (1944) 164. [30] D.W. Marquardt, An algorithm for the least-square estimation of nonlinear parameters, SIAM J. Appl. Math. 11 (1963) 431. [31] S. Doniach, M. Sunjic, J. Phys. C 3 (1970) 285. [32] D.A. Shirley, Phys. Rev. B 5 (1972) 4709. [33] U.A. Effner, D. Badt, J. Binder, T. Bertrams, A. Brodde, C. Lunau, H. Neddermeyer, M. Hanbucken, Photoemission and scanning tunneling microscopy on K/Si(1 0 0), Surf. Sci. 277 (1992) 207. [34] Y. Enta, S. Suzuki, S. Kono, Angle-resolved photoemission study of the electronic structure of the Si(0 0 1) (4 2) surface, Phys. Rev. Lett. 65 (1990) 2704. [35] F.J. Himpsel, F.R. Mcfeely, A. Talebibrahimi, J.A. Yarmoff, G. Hollinger, Microscopic structure of the SiO2/Si interface, Phys. Rev. B 38 (1988) 6084. [36] A.V. Borne, R.L. Johnson, B. Sonntag, M. Talkenberg, A. Verweyen, P. Wernet, J. Schulz, K. Tiedtke, C. Gerth, B. Obst, P. Zimmermann, J.E. Hansen, Suppression of the low-spin multiplet components in the 3p photoelectron spectra of atomic and solid 3d metals, Phys. Rev. A 62 (2000) 052703. [37] A. Irizawa, A. Yamasaki, M. Okazaki, S. Kasai, A. Sekiyama, S. Imada, S. Suga, E. Kulatov, H. Ohta, T. Nanba, Bulk-sensitive photoemission of Mn5Si3, Solid State Commun. 124 (2002) 1. [38] M. Oku, S. Suzuki, N. Ohtsu, T. Shishido, K. Wagatsuma, Comparison of intrinsic zero-energy loss and Shirley-type background corrected profiles of XPS spectra for quantitative surface analysis: study of Cr, Mn and Fe oxides, Appl. Surf. Sci. 254 (2008) 5141. [39] E.L. Bullock, R. Gunnella, L. Patthey, T. Abukawa, S. Kono, C.R. Natoli, L.S.O. Johansson, Surface core level photoelectron diffraction from Si-dimers at the Si(0 0 1)2 1 surface, Phys. Rev. Lett. 74 (1995) 2756. [40] E. Landemark, C.J. Karlsson, Y.-C. Chao, R.I.G. Uhrberg, Core level spectroscopy of the clean Si(1 0 0) surface: charge transfer within assymmetric dimer of the 2 1 and c(4 2) reconstructions, Phys. Rev. Lett. 69 (1992) 1588. [41] R.Q. Zhang, T.S. Chu, H.F. Cheung, N. Wang, S.T. Lee, High reactivity of silicon suboxide clusters, Phys. Rev. B 64 (2001) 113304. [42] C. Anandan, X-ray photoelectron spectroscopic study of room-temperature evolution of oxide-covered hydrogenated amorphous silicon/aluminum interface, Appl. Surf. Sci. 89 (1995) 57. [43] N. Benouattas, A. Mosser, A. Bouabellou, Surface morphology and reaction at Cu/Si interface—effect of native silicon suboxide, Appl. Surf. Sci. 252 (2006) 7572. [44] S. IIda, S. Abe, Ti thin-film reaction on SiO2/Si, Appl. Surf. Sci. 78 (1994) 141. [45] S. Kennou, S. Ladas, M.G. Grimaldi, T.A.N. Tan, J.Y. Veuillen, Oxidation of thin erbium and erbium silicide overlayers in contact with silicon oxide films thermally grown on silicon, Appl. Surf. Sci. 102 (1996) 142. [46] L.S.O. Johansson, R.I.G. Uhrberg, P. Martensson, B.V. Hansson, Surface-state band structure of the Si(1 0 0)-2 1 surface studied with polarization-dependent angle resolved photoemission on single domain surfaces, Phys. Rev. B 42 (1990) 1305. [47] H. Sugawara, A. Kakizaki, I. Nagakura, T. Ishii, Extreme ultraviolet photoelectronspectra of alpha-manganese, J. Phys. F Met. Phys. 12 (1982) 2929.