Home

Search

Collections

Journals

About

Contact us

My IOPscience

Origin of anomalous magnetite properties in crystallographic matched heterostructures: Fe3O4(111)/MgAl2O4(111)

This content has been downloaded from IOPscience. Please scroll down to see the full text. 2013 J. Phys.: Condens. Matter 25 485004 (http://iopscience.iop.org/0953-8984/25/48/485004) View the table of contents for this issue, or go to the journal homepage for more

Download details: IP Address: 137.120.4.50 This content was downloaded on 28/06/2014 at 11:11

Please note that terms and conditions apply.

IOP PUBLISHING

JOURNAL OF PHYSICS: CONDENSED MATTER

J. Phys.: Condens. Matter 25 (2013) 485004 (8pp)

doi:10.1088/0953-8984/25/48/485004

Origin of anomalous magnetite properties in crystallographic matched heterostructures: Fe3O4(111)/MgAl2O4(111) D Gilks1 , L Lari1,2 , J Naughton1 , O Cespedes3 , Z Cai4 , A Gerber5 , S M Thompson1 , K Ziemer4 and V K Lazarov1 1

Department of Physics, University of York, Heslington, York YO10 5DD, UK York-JEOL Nanocentre, University of York, Heslington, York YO10 5BR, UK 3 School of Physics and Astronomy, University of Leeds, Leeds LS2 9JT, UK 4 Department of Chemical Engineering, Northeastern University, 360 Huntington Avenue, Boston, MA 02115-5000, USA 5 School of Physics and Astronomy, Tel Aviv University, Tel Aviv 69978, Israel 2

E-mail: [email protected]

Received 29 July 2013, in final form 30 September 2013 Published 31 October 2013 Online at stacks.iop.org/JPhysCM/25/485004 Abstract Magnetite films grown on crystallographically matched substrates such as MgAl2 O4 are not expected to show anomalous properties such as negative magnetoresistance and high saturation fields. By atomic resolution imaging using scanning transmission electron microscopy we show direct evidence of anti-phase domain boundaries (APB) present in these ¯ shifts determine the atomic structure of heterostructures. Experimentally identified 1/4h101i the observed APBs. The dominant non-bulk superexchange interactions are between 180◦ octahedral-Fe/O/octahedral-Fe sites which provide strong antiferromagnetic coupling across the defect interface resulting in non-bulk magnetic and magnetotransport properties. (Some figures may appear in colour only in the online journal)

1. Introduction

in thin films is essential. Despite significant effort to prepare thin films of magnetite with bulk like properties in the majority of cases Fe3 O4 thin films exhibit non-bulk anomalous properties. The most significant of which are negative magnetoresistance (MR) [11, 12] and exceptionally high saturation fields [13], neither of these are seen in bulk crystals of Fe3 O4 . Growth defects such as anti-phase domain boundaries (APBs) have been identified as the main cause of such anomalous thin films behaviour due to the creation of strong antiferromagnetic (AFM) coupling at the APBs. Therefore magnetite performance when used as an electrode in devices would be critically dependent and influenced by the presence/absence of APBs. The APBs in heterostructures are due to crystallographically and lattice mismatched atomic structures, e.g. Fe3 O4 –MgO (spinel-FCC) [11, 14], Fe3 O4 –Al2 O3 (spinel–corundum) [12, 15], Fe3 O4 –GaAs (spinel–zinc

Thin film magnetite (Fe3 O4 ) has attracted significant research interest in recent years as a potential material platform for spintronic device applications [1–8]. Fe3 O4 exhibits ferrimagnetic ordering and a high Curie temperature (858 K). In addition ab initio electronic calculations predict Fe3 O4 to be half-metallic with 100% spin-polarisation at the Fermi level [9, 10]. In Fe3 O4 , minority states are present at the Fermi level but a band gap of 0.5 eV exists in the majority states. These properties are important for spin injection electrodes and other spintronics devices such as magnetic tunnelling junctions where highly spin-polarized materials are required above room temperature. Device applications require fabrication of thin films and heterostructures. Since the desirable attributes of Fe3 O4 are bulk properties, retaining these properties of Fe3 O4 0953-8984/13/485004+08$33.00

1

c 2013 IOP Publishing Ltd Printed in the UK & the USA

J. Phys.: Condens. Matter 25 (2013) 485004

D Gilks et al

blend) [16, 17]. Therefore, Fe3 O4 /MgAl2 O4 heterostructures are of special interest because both the film and the substrate have the same crystal symmetry and close lattice constants, with mismatch of ∼3%. However, regardless of crystal symmetry and lattice constants with one to one ratio, thin films of magnetite grown on MgAl2 O4 (001) still show anomalous properties [18]. This unexpected behaviour of Fe3 O4 films on MgAl2 O4 was attributed to interface originated APBs. However high resolution phase contrast bright field STEM imaging on Fe3 O4 (111)/MgAl2 O4 (111) interface does not appear to support misfit dislocation origin of APBs formation [19]. In this work we perform atomic resolution Z-contrast STEM in order to find the origin of the anomalous behaviour of magnetite films when grown on symmetry matched substrates. This approach would clearly provide answers on the atomic structure of the APBs and their unique identification in the films. In this work we directly show evidence of the presence of APBs in Fe3 O4 . Based on experimental data the atomic structure of APBs is determined. The APBs break the symmetry of the Fe octahedral sublattice which results in a significant number of non-bulk superexchange interactions (SEIs) with antiferromagnetic character across the boundary that give rise to the anomalous film properties.

¯ zone Figure 1. HRTEM image of Fe3 O4 film and substrate in [110] axis.

performed using a double aberration corrected field emission JEOL FS-2200 JEM TEM/STEM and a JEOL-2011 TEM both operating at 200 kV. Multislice image simulations [21] have been produced to correlate the experimental (S)TEM atomic column intensities with proposed model APBs atomic structure.

3. Results 2. Methods

Figure 1 shows a representative HRTEM image that gives an ¯ overview of the Fe3 O4 /MgAl2 O4 heterostructure in the [110] viewing direction. The film has uniform thickness (30 nm) and distinctive lattice fringes between the substrate and the film demonstrate the single crystal nature of the film and the abrupt film/substrate interface. SAD patterns from the interfacial region of the sample in ¯ and [110] ¯ zone axis are shown in figures 2(a) and the [112] (b) respectively. The shared crystal symmetry of the film and substrate can be seen through the superimposed film/substrate reflections in both diffraction patterns. The radial separation of diffraction spots corresponds to the 3% difference between the film and substrate lattice constants. These diffraction patterns demonstrate the single crystal nature of the film and show no evidence of any secondary phases. From the SAD the epitaxial cube to cube relationship between film and substrate is found to be: Fe3 O4 (111) k MgAl2 O4 (111) and ¯ k MgAl2 O4 (110). ¯ Fe3 O4 (110) The stoichiometry of the film is confirmed by in situ XPS (figure 3) taken directly after deposition. This data shows the characteristic Fe 2p peak of magnetite with 2p(1/2) and 2p(3/2) clearly resolved without the presence of any additional peaks that would indicate the presence of secondary phases [22]. This result indicates the single-phase structure of the film with stoichiometric Fe3 O4 composition, in contrast to previous reports that have found significant phase separation in polar Fe3 O4 (111) films [23]. As expected due to the low penetration depth of XPS no signal from the substrate was detected. Further confirmation of the films stoichiometry was provided by ex situ infra-red Raman spectroscopy (figure 4). The higher penetration depth of this technique compared to XPS gives a contribution to the data from the substrate. The shared

Fe3 O4 films of 10, 30 and 50 nm thickness have been grown on (111) oriented MgAl2 O4 single crystal substrates by molecular beam epitaxy (MBE) methods. Fe and atomic O have been simultaneously deposited by Knudsen cell and a RF-assisted plasma source respectively. MBE base pressure was less than 2 × 10−10 mbar, and the deposition has been ˚ min−1 in a partial pressure of conducted at a rate of ∼1.2 A −6 atomic oxygen of 5×10 mbar. During growth, the substrate temperature has been held at 350 ◦ C. Film growth was monitored in real time with reflective high energy electron diffraction and characterized postdeposition by using in situ x-ray photoelectron spectroscopy (XPS) with Al Kα source and ex situ infra-red Raman spectroscopy. In-plane magnetization versus applied field (MH) magnetic hysteresis curves, have been measured using superconducting quantum interference device magnetometer in applied magnetic fields up to 9 T, with applied field along ¯ direction. Magnetoresistance (MR) measurements the [110] have been carried out in longitudinal in-plane geometry along ¯ direction using four point probe contact methods at the [110] room temperature in applied fields up to 10 kOe. Structural film characterization has been performed using high resolution transmission electron microscopy (HRTEM), selected area diffraction (SAD), and high angle annular dark field scanning transmission electron microscopy, (HAADFSTEM). Cross-sectional TEM microscopy specimens were ¯ and [110] ¯ zone axes, by conproduced in orthogonal [112] ventional methods that include mechanical thinning/polishing followed by low angle Ar ion milling in order to achieve electron transparency [20]. Electron microscopy has been 2

J. Phys.: Condens. Matter 25 (2013) 485004

D Gilks et al

¯ zone axis and (b) the [110] ¯ zone axis. Figure 2. SAD patterns from a cross-section of the Fe3 O4 /MgAl2 O4 interfacial region in (a) the [112]

Figure 3. In situ XPS showing the Fe 2p(1/2) and Fe 2p(3/2) peaks characteristic peaks of Fe3 O4 film. 735 eV.

structure of both film and substrate is shown by the shared vibrational frequency at 668 cm−1 . However, the reduction of the 311 cm−1 peak representative of MgAl2 O4 [24] and the enhancement of the 540 cm−1 peak representative of Fe3 O4 [25, 26] with increasing film thickness show the transition to a pure Fe3 O4 signal in the thicker films. The absence of resonant peaks at 245 cm−1 and 500 cm−1 representative of α-Fe2 O3 and γ -Fe2 O3 respectively and the absence of a shoulder at 652 cm−1 show the good stoichiometry of these films without the inclusion of any secondary phases [27, 28]. Next we present the MH and MR results (figures 5 and 6). The in-plane magnetization curves at 150 K clearly show that even at fields of 8 T none of the films are saturated, while their coercivity is ∼500 mT. Furthermore, the room temperature in-plane MR data shows that at applied fields of 8 kOe these films have negative MR in a range between −0.7 and −1%. Both MH and MR data of the films show the non-bulk like behaviour, consistent with other reports on thin film Fe3 O4 properties [11, 13]. In order to understand this anomalous behaviour and correlate structure to properties in these films atomic

Figure 4. Ex situ Raman spectra of Fe3 O4 at 193, 306, 538 and 668 cm−1 . Measurements were taken to 1900 cm−1 but no further structure is shown in the curves beyond 900 cm−1 .

resolution HAADF-STEM measurements were undertaken. Figure 7 shows a HAADF image of the interfacial region ¯ zone axis. This between Fe3 O4 and MgAl2 O4 in the [112] image clearly demonstrates the crystal symmetry matching of film and substrate as well as the atomically and chemically abrupt nature of this interface. The misfit dislocations at the interface formed due to the film/lattice mismatch are clearly ¯ atomic planes. These outlined by Bragg filtering of h440i misfit dislocations lead to relief of the film strain arising as a result of film–substrate lattice mismatch. A similar analysis ¯ zone axis (90◦ apart from [112] ¯ view) from the in [110] Fe3 O4 /MgAl2 O4 interface is shown in figure 8. The interface ¯ dislocation here is outlined by Bragg filtering of the (222) planes. These arrays of interfacial misfit dislocations clearly do not change translational symmetry of the film in the surrounding area of the dislocation core. This demonstrates 3

J. Phys.: Condens. Matter 25 (2013) 485004

D Gilks et al

Figure 7. (a) HAADF-STEM image of the interfacial region ¯ zone axis (a). The between Fe3 O4 an MgAl2 O4 in the [112] matched crystal symmetry across this interface can be seen in the continued structural motif from film to substrate. (b) Bragg filtered ¯ image showing dislocation cores (c) fast Fourier transform of h440i image (a), circled reflections are used to create the Bragg image in (b). Dislocation cores are outlined in both the original and filtered image with circles.

Figure 5. In-plane magnetization curves from thin film Fe3 O4 . This data has been taken at 150 K using a cold stage VSM. At 8 T magnetization has not saturated.

Figure 6. In-plane MR Fe3 O4 films by four point probe techniques.

that no APBs are nucleated as a result of misfit dislocations. The presence of misfit dislocations at the interface does not enforce a break in the bulk symmetry of the film apart from the dislocation core vicinity. In order to verify the presence of APBs in the films, detailed analysis of the atomic structure throughout the film has been performed. The result of this analysis is illustrated in figure 9 which shows an atomic resolution HAADF image, taken from the middle region of the film. An APB running in the film growth direction is clearly seen, the rough outline of the boundary region is presented by white dashed lines. The presence of the APB has been identified through changes in the atomic column intensities of the ‘3FeB ’ atomic

Figure 8. (a) HAADF-STEM image of the interface between ¯ zone axis. (b) Fast Fourier Fe3 O4 and MgAl2 O4 in the [110] ¯ reflections used for creating the transform of the image in (a), (222) Bragg image in (c) are outlined. (c) Bragg filtered image with a single dislocation core. Dislocation cores are outlined in both the original and filtered image with circles.

¯ viewing direction the iron atomic columns planes. In the [110] occupancy of the ‘3FeB ’ layers alternates between double and single occupancy (see the inset in figure 9(a)). 4

J. Phys.: Condens. Matter 25 (2013) 485004

D Gilks et al

¯ zone axis. (a) The extended defect region is Figure 9. HAADF-STEM image of an APB defect within the Fe3 O4 film, imaged in the [110] identified by the disruption of the rhombohedral motif, outlined in the inset upper right corner with the Fe atomic columns denoted as the following: large (small) red circles represent double (single) occupied FeB columns and green circles represent FeA atomic columns. (b) Atomic model of the APB boundary outlined in (a), the colour code is the same as in (a) with O atoms as blue. (c) Calculated HAADF map ¯ shift between coalesced Fe3 O4 grain results in loss of the alternating contrast of Fe columns in ‘3FeB ’ of the model in (b). The 1/4h011i layers indicated by dashed yellow lines in (a). The match between the calculated image and experimental image is illustrated by the inset of the model image inside the middle of APB region in (a).

As the contrast in the HAADF imaging is ∼ Z 2 ; the single to double occupancy of the neighbouring Fe atomic columns in the 3FeB layer can be used to determine the conservation of translational symmetry. For bulk like structures the ‘3FeB ’ ¯ projection should alternate between high layer in the [110] and low brightness atomic columns. Indeed this is clearly illustrated in the areas left and right of outlined region. However in the middle region (within the dashed lines) this characteristic pattern of well-ordered magnetite film is lost. Instead of alternating FeB site intensities, in the boundary region neighbouring sites on the ‘3FeB ’ planes are all equally ¯ bright. This effect is explained by introducing a 1/4h011i type shift of the magnetite unit cell between the left and right grain in the field of view. This type of shift is schematically shown in figure 9(b). Based on this shift atomic models of the APB have been constructed from which HAADF maps were calculated using the multislice methods [21]. The calculated HAADF map is shown in figure 9(c). As seen from the simulation, the pure geometrical shift between the right and left grain with an overlapping interface region of ∼3 nm results in equal intensity of the FeB atomic columns within the APB region. This relative shift produces a break in the translational symmetry and hence forms an APB. The APB of shown is not edge on with respect to the beam direction; as a result instead of a sharp single atomic plane APB an overlapping region between neighbouring grains is seen ¯ projection. across several nanometres in this [110]

we have identified the presence of APBs in these films which are not related to interface dislocation formation. Next we discuss how the atomic structure of these APBs leads to non-bulk superexchange interactions (SEIs) which are ultimately responsible for the anomalous macroscopic behaviour of these films. The dominant SEI in bulk Fe3 O4 that provides the AFM coupling between the A and B sublattices is the FeA –O–FeB with an exchange integral value of JAB = −22 K. The other two important interactions are the JBB = 3 K, and JAA = −11 K [29, 13]. The magnitude and sign of SEIs depends of the atomic distances and angles of O mediated bonding e.g. the FeB –O–FeB interaction can vary from FM to AFM by varying the bond angle from 90◦ towards 180◦ . The effect of the APB defects is in the change of the local atomic co-ordination of Fe3 O4 , such as altered cation distances and bond angles of SEIs in the region of an APB. These non-bulk SEIs can be dominant and can change the magnetic and magnetotransport behaviour of the films. Based on the experimental evidence we consider only in-plane APB shift vectors orthogonal to the [111] growth direction. These planes demonstrate hexagonal symmetry as expected of (111) planes, characterized by six major ¯ type planes crystallographic directions. There are three h110i ¯ and three h112i type planes. The shift vectors will be confined to a set of vectors that do not perturb the underlying O sublattice imposing APB shift vectors to be a linear combination of shifts along these major crystallographic directions. The observed half shift in the positioning of the ¯ rhombohedral structural motif (figure 9) along the [112] direction (left–right in the diagram) is achieved by a shift ¯ 1/4ah011i, ¯ ¯ or 1/4ah101i, ¯ vector of 1/4ah011i, 1/4ah101i as shown in figure 10. We would like to note that linear

4. Discussion The results presented clearly show that these Fe3 O4 films demonstrate anomalous M–H and MR properties despite the good structural ordering and stoichiometry determined by the mesoscopic film analysis methods. Furthermore, 5

J. Phys.: Condens. Matter 25 (2013) 485004

D Gilks et al

¯ plane with 1/4ah011i ¯ shift vectors. Bulk Fe3 O4 bonds are given for comparison. Table 1. Summary of interfacial bonds for APB on (112) Bulk like bonds across the APB are in italics. Strongly antiferromagnetic FeB –O–FeB bonds are in bold. Boundary plane

Shift vector

Bond angle

Bond type

Fe1 –O ˚ distance (A)

Fe2 –O ˚ distance (A)

Fe1 –Fe2 ˚ distance (A)

¯ [112]

Bulk

90 125.3

FeB –O–FeB FeA –O–FeB

2.09 1.81

2.09 2.09

2.96 3.46

71 90

¯ [112]

¯ (011) ¯ (011), ¯ ¯ (101), (101)

54.7 70.52 90 109.5 125.3 180

FeA –O–FeB FeA –O–FeA FeB –O–FeB FeA –O–FeA FeA –O–FeB FeB –O–FeB

1.81 1.81 2.09 1.81 1.81 2.09

2.09 1.81 2.09 1.81 2.09 2.09

1.81 2.09 2.96 2.95 3.46 4.18

48 8 118 8 100 23

No.

and angles across the boundary which can be used to give an insight into the nature of the SEIs across the boundary. In table 1, the angles of SEIs for FeB –O–FeB , FeA –O– FeA , and FeA –O–FeB are summarized with the distance between the Fe cations and the Fe–O bond lengths. Compared to bulk interactions the most important difference at the APB is the formation of 180◦ FeB –O–FeB bonds. The SEIs of such bonds are strongly antiferromagnetic. In addition weak low angle (54.7◦ ) FeA –O–FeB AFM bonds are created as well as weak low angle (70.5◦ ) and (109.5◦ ) FeA –O–FeA bonds. From the summarized data the most important SEIs across the APB are the 180◦ FeB –O–FeB bonds due to their strength and number. While in bulk all FeB –O–FeB interactions are weakly FM, the AFM nature of a significant portion of the FeB –O–FeB at the APB most likely provides regions where pining of magnetic domain walls occur. This pining, due to the strong AFM interactions, would make it very hard to saturate magnetite films [13, 30, 31] as illustrated by M–H measurements presented in figure 5. Perturbation of the FeB –O–FeB interaction also has a significant effect on the magnetotransport. The conductivity in Fe3 O4 is limited to electron hoping mechanisms on the B sublattice [32]. The hoping probability is proportional to ∼cos(θ/2), implying that 180◦ FeB –O–FeB interactions would not contribute to the film conductivity. Also it does explain the increase of the conductivity when H field is applied [11]. This APB bonds analysis has been repeated by ¯ 1/4[101], ¯ and 1/4[101] ¯ introducing 1/4[011], shift vectors. As indicated in the table 1 the results are identical for the four different shift vectors. These results provide an explanation of the anomalous magnetic and magnetotransport properties measured in Fe3 O4 /MgAl2 O4 films. Atomistic model of the APB presented in this work (confirmed by the good fit between the simulated and experimental APB atomic structure) are a good approximation to the real APB atomic structure.

Figure 10. Schematic diagram of four translational vectors, ¯ 1/4[011], ¯ 1/4[101] ¯ and 1/4[101] ¯ 1/4[011], that give rise to APB formation, with Fe positions altered and unperturbed O sublattice. ¯ plane of the APB. The dashed line corresponds to the (112)

¯ shifts could result in a shift vector combination of 1/4ah011i ¯ direction. However, this would not break the along the [112] structural motif of Fe3 O4 so it would not be visible in the projection. From figure 9 we determine the in-plane shift vector to ¯ type. The boundary of this defect is on (112) ¯ be of 1/4ah011i planes orthogonal to the imaging direction. The ∼3 nm lateral extent of this defect outlined by white dashed lines is due to the 3D nature of film growth. When coalescence between grains occurs the formed APBs are extended onto several ¯ planes. For the boundary shown in figure 9 the region (112) ¯ planes that define the APB is ∼3 nm. Assuming of (112) the specimen thickness of ∼30 nm, the observed APB is 5.7◦ mistilted with respect to incident electron beam. In order to quantify the number of non-bulk bonds that ¯ shifts simple arise due to APBs as a result of 1/4h011i geometrical models of the APB have been constructed based ¯ atomic plane boundary. An overview of the on a single (112) APB model is detailed in the appendix which shows atomic stacking of the APB cell with corresponding atomic positions of Fe and O. This model allows us to find the atomic distances

5. Conclusion Atomic resolution HAADF-STEM images reveal that Fe3 O4 films show the presence of APBs despite being grown on substrates with the same crystallographic symmetry. The APB properties are independent of the specific in-plane shift vector. 6

J. Phys.: Condens. Matter 25 (2013) 485004

D Gilks et al

¯ APB. Each Figure A.1. Schematic diagram of the twelve layers required to make a full structural repeat unit of (a) Fe3 O4 and (b) 1/4h011i tri-layer block is projected on the x–y plane, while the z positions are indicated by the size of the symbols. Small filled symbols denote the bottom layer, small unfilled symbols represent the central and large unfilled symbols represent the top layers of the three layer blocks.

Simple atomistic modelling based on possible interface configurations confirm the existence of 180◦ FeB –O–FeB bonds that give rise to an AFM nature of the APB. It appears that strongly antiferromagnetic FeB –O–FeB bonds across the boundary are the main superexchange interaction responsible for the anomalous magnetic and magnetotransport properties of these Fe3 O4 films. The APBs found in the film are not driven by strain relief in the film, since no direct evidence is found to link the misfit dislocation network at the film–substrate interface to APB formation.

sented by blue squares, FeB represented by red circles, FeA with green triangles. The repeat unit of Fe3 O4 along the [111] direction consist of 12 atomic planes with 6 atomic O planes on which all the sites are occupied. O planes separate the alternating ‘3FeB ’ and ‘FeA FeB FeA ’ planes. The full stacking sequence can be written as follows: i: [3FeB /4O/FeA FeB FeA ]/ii: [4O/3FeB /4O]/iii: [FeA FeB FeA /4O/3FeB ]/iv: [4O/FeA FeB FeA /4O]. Graphical presentation of the stacking is shown in figure A.1, for simplicity the stacking is shown in four blocks (i, ii, iii and iv) each containing three atomic planes projected on the x–y plane. For the APB model, an interfacial plane has been chosen, as shown by the dashed line through figure A.1(b). All ¯ atoms below this interface have been translated by 1/4[011] which shifts the positions of the FeA FeB FeA sites and the FeB vacant sites with respect to untranslated part of the cell. This translation does not affect O sites due to their high symmetry. However, as seen from figure A.1(b)-i, this shift brings FeA and FeB sites into non-bulk configurations which ultimately determine non-bulk SEIs across the APB.

Acknowledgment This work has been supported by the EPSRC grant EP/K013114/1.

Appendix. The structure of Fe3 O4 and APBs Here we present schematics of the atomic planes at ¯ APB studied in this work. O is reprethe 1/4h011i 7

J. Phys.: Condens. Matter 25 (2013) 485004

D Gilks et al

References

[17] Watts S M, Nakajima K, van Dijken S and Coey J M D 2004 J. Appl. Phys. 95 7465 [18] Luysberg M, Sofin R G S, Arora S K and Shvets I V 2009 Phys. Rev. B 80 02411 [19] Gilks D, Lari L, Cai Z, Cespedes O, Gerber A, Thompson S, Ziemer K and Lazarov V K 2013 J. Appl. Phys. 113 17B107 [20] Lari L, Lea S, Feeser C, Wessels B W and Lazarov V K 2012 J. Appl. Phys. 111 07C311 [21] Kirkland E J, Loane R F and Silcox J 1987 Ultramicroscopy 23 77–96 [22] Lazarov V K, Chambers S A and Gajdardziska-Josifovska M 2003 Phys. Rev. Lett. 90 216108 [23] Lazarov V K, Weinert M, Chambers S A and Gajdardziska-Josifovska M 2005 Phys. Rev. B 72 195401 [24] Cynn H, Sharma S K, Conney T F and Nicol M 1992 Phys. Rev. B 45 500 [25] Gupta R, Sood A K, Metcalf P and Honig J M 2002 Phys. Rev. B 65 104430 [26] L¨ubbe M, Gigler A M, Stark R W and Moritz W 2010 Surf. Sci. 604 679–85 [27] Zoppi A, Lofrumento C, Castellucci E M and Sciau Ph 2008 J. Raman Spectrosc. 39 40–6 [28] Chamritski I and Burns G 2005 J. Phys. Chem. B 109 4965–8 [29] De Grave E et al 1993 Phys. Rev. B 47 5881 [30] Margulies D T, Parker F T, Spada F E, Goldman R S, Li J, Sinclair R and Berkowitz A E 1996 Phys. Rev. B 53 14 [31] Hibma T, Voogt F C, Niesen L, van der Heijden P A A, de Jonge W J M, Donkers J J T M and van der Zaag P J 1999 J. Appl. Phys. 85 5291 [32] Ihle D and Lorenz B 1986 J. Phys. C: Solid State Phys. 19 5239–51

[1] Alexe M, Ziese M, Hesse D, Esquinazi P, Yamauchi K, Fukushima T, Picozzi S and Gosele U 2009 Adv. Mater. 21 4452 [2] Hu G and Suzuki Y 2002 Phys. Rev. Lett. 89 276601 [3] Bataille A M, Moussey J-B, Paumier F, Gota S, Guittet M-J and Gautier Soyer M 2005 Appl. Phys. Lett. 86 012509 [4] Kado T, Saito H and Ando K 2007 J. Appl. Phys. 101 09J511 [5] Alldredge L M B, Chopdekar R V, Nelson-Cheeseman B B and Suzuki Y 2006 Appl. Phys. Lett. 89 182504 [6] Kado T 2008 Appl. Phys. Lett. 92 092502 [7] Fonin M, Dedkov Yu S, Pentcheva R, R¨udiger U and G¨untherodt G 2007 J. Phys.: Condens. Matter 19 315217 [8] Pratt A, Kurahashi M, Sun X, Gilks D and Yamauchi Y 2012 Phys. Rev. B 85 180409 [9] Yanase A and Siratori K 1984 J. Phys. Soc. Japan 53 312 [10] Zhang Z and Satpathy S 1991 Phys. Rev. B 44 13319 [11] Eerenstein W, Palstra T T M, Saxena S S and Hibma T 2002 Phys. Rev. Lett. 88 247204 [12] Ramos A V, Moussy J-B, Guittet M-J, Bataille A M, Gautier-Soyer M, Viret M, Gatel C, Bayle-Guillemaud P and Snoeck E 2006 J. Appl. Phys. 100 103902 [13] Margulies D T, Parker F T, Rudee M L, Spada F E, Chapman J N, Aitchison P R and Berkowitz A E 1997 Phys. Rev. Lett. 79 5162 [14] Ziese M and Blythe J 2000 J. Phys.: Condens. Matter 12 13 [15] Moussy J-B et al 2004 Phys. Rev. B 70 174448 [16] Prakash R, Choudhary R J, Sharath Chandra L S, Lakshmi N and Phase D M 2007 J. Phys.: Condens. Matter 19 486212

8