Conference-NSS-6-Spectromicroscopy with Low-Energy Electrons : LEEM and XPEEM Studies at the Nanoscale

The multi-technique approach in the spectroscopic photoemission and low-energy electron microscope (SPELEEM) has proven to be a powerful tool in studies of crystalline, chemical and magnetic structures at surfaces. The energy filtering offered by this instrument enables a variety of complementary analytical characterization methods. Here, we give a summary of the recent studies with the SPELEEM microscope installed at the Nanospectroscopy beamline of the Elettra synchrotron laboratory. The examples cover topics such as surface corrugation in free-standing graphene layers, spin-reorientation transition in thin Fe films on W(110), and stressinduced adsorbate patterns on single crystal surfaces. Moreover, the SPELEEM capabilities of imaging inelastically scattered electrons are demonstrated. [DOI: 10.1380/ejssnt.2011.72]


I. INTRODUCTION
X-ray photoemission electron microscopy (XPEEM) has become widespread at third generation synchrotron radiation facilities.It is capable of chemical and magnetic characterization at mesoscopic length scales by implementing laterally-resolved x-ray absorption spectroscopy and exploiting circular and linear dichroism techniques.With the addition of an energy-filter, XPEEM also allows x-ray photoemission spectromicroscopy [1].In some cases, an XPEEM setup is combined with low-energy electron microscopy (LEEM), which can be used to monitor the surface structure and its dynamical evolution with high spatial resolution [2,3].Such an instrument has been proven to be a powerful surface science facility with structural, chemical and magnetic sensitivity [1,4,5].
Here, we give the highlights of the recent studies with the spectroscopic photoemission and low-energy electron microscope (SPELEEM) in operation at the Nanospectroscopy beamline of the Elettra synchrotron laboratory (Trieste) [5,6].The examples we have chosen include electronic and structural properties of free-standing graphene layers, microscopic nature of the spin-reorientation transition in an Fe thin film, and stress-induced pattern formation on single crystal surfaces.The first example demonstrates the use of microprobe diffraction (i.e.low-energy electron diffraction and angle-resolved photoemission) in investigating electronic and geometric structure of multilayer graphene in a layer-resolved manner.The second makes use of the polarization of the photon source to probe the magnetic structure at the nanometer scale, whereas the last example illustrates the possibilities by combining imaging and diffraction to study a given system at all length scales.In the last section, in addition to the already well-established methods accessible by the SPELEEM instrument, we summarize the performance of the microscope utilizing inelastically scattered electrons.

II. SPELEEM INSTRUMENT
The spectroscopic photoemission and low-energy electron microscope (SPELEEM) is a multimethod electron microscope based on the cathode immersion lens [4].It can image surfaces, interfaces and thin films at a lateral resolution of few tens nm using photoemitted or elastically/inelastically backscattered electrons from the specimen [6].
A simplified scheme of the microscope layout is presented in Fig. 1.As seen in the sketch, the sample can be probed by using either soft x-ray photons or low-energy electrons.In XPEEM mode, the photon beam is provided by the undulator beamline (Nanospectroscopy, Elettra), at which the microscope is currently installed [7].The source is made of two Sasaki-type Apple II undulators, which can produce circular, linear horizontal and linear vertical polarizations.This is essential to studies making use of circular and linear dichroism in photoemission and absorption.The available photon energies range from slightly below 40 eV up to 1000 eV.The undulator radiation is monochromatized by two varied line space plane gratings.After the monochromator, the x-ray beam is focused into a 25 µm × 3 µm (hor.×ver.)spot on the sample with a 16 • grazing incidence angle.The photon flux on the sample is in the range 10 12 -10 13 photons/s, depending on the chosen photon energy.In LEEM operation, the low-energy electrons are produced by an LaB 6 source.On the sample surface, the electron beam has a circular shape of 85 µm diameter.The incident electron current varies in the range from a few to about 100 nA, and the energy spread of the monochromatic electron beam is 0.5 eV.
The low-energy electron beam (either scattered or photoemitted) leaving the sample is strongly accelerated by the electric field applied between the sample and the objective lens (18 keV), which produces a magnified image (×40) of the sample in the image plane at the beam separator.The subsequent transfer optics (labeled TL, FL, IL in Fig. 1) can either be used to change the magnification, or to image the back-focal plane of the objective lens, thus enabling diffraction imaging.The angular acceptance of the microscope in imaging mode is defined by a circular pinhole (the contrast aperture labeled CA) placed at a diffraction plane at the position of FL.In LEEM, the contrast aperture is used in order to filter out primary and secondary diffracted beams, which is the basis of the bright-and dark-field operation modes.In XPEEM, the main function of the contrast aperture is to limit the angular acceptance, thereby reducing spherical aberrations and improving spatial resolution.
A unique feature of the SPELEEM is its capability of energy filtering, which is enabled by the hemispherical electron energy analyzer with a pass energy of about 908 eV.The band-pass filter is implemented by the energy slit (labeled ES in Fig. 1) placed at the dispersive plane of the analyzer.The two slits available in the microscope correspond to resolutions of 0.3 and 0.6 eV [5].
In addition to the LEEM and XPEEM imaging modes, the SPELEEM enables also microprobe diffraction and spectroscopy.In that case, the probe area is selected using the field limiting aperture (labeled FLA), which is inserted at the achromatic image plane in the middle of the beam separator.Microprobe spectroscopy is implemented by imaging the dispersive plane of the energy analyzer directly onto the detector, providing the snapshot of a 10 eV-wide spectral window.The best energy resolution of the SPELEEM, 0.2 eV, is attained in microprobe spectroscopy [5].In the microprobe diffraction mode, the diffraction pattern is recorded by imaging the back-focal plane of the objective lens.In the case of incident x-rays (giving rise to photoelectron diffraction or angle-resolved photoemission), the energy slit is placed to select a particular photoelectron energy.Relevant to the microprobe operations in general, the minimum FLA size corresponds to a selected area of 2 µm diameter.
The performance of the SPELEEM installed at the Nanospectroscopy beamline of the Elettra synchrotron light source has been described in detail previously [5].In real-space imaging mode, the microscope reaches a lateral resolution of about 10 nm in LEEM and 30 nm in XPEEM.The ultimate resolution is limited by the spherical and chromatic aberrations of the cathode immersion lens and the electron-energy analyzer.High x-ray fluxes result in additional image blur, due to electron-electron stochastic interactions occurring in the microscope image column [8].

III. CORRUGATION IN SUPPORTED AND SELF-STANDING GRAPHENE LAYERS
Graphene, a single graphite layer, has been attracting an enormous attention due to its exotic properties as a truly two-dimensional crystal [9].In particular, its mechanical and transport characteristics make it a target for a wide range of technological applications.Due to the reduced stiffness, single layer graphene membranes are extremely flexible, which results in crystal deformations and corrugations.It is still debated whether such corrugations are due to extrinsic factors, such as interactions with the substrate and impurities, or to intrinsic fluctuations deemed crucial in stabilizing the two-dimensional order.
In the following we summarize our recent work on the morphology and crystal structure of exfoliated graphene aimed at identifying the driving forces for the lattice deformations [10].Our approach takes full advantage of the multi-method capability offered by a SPELEEM: real space imaging of the sample morphology over large surface areas (up to several tens of microns); micro-probe http://www.sssj.org/ejssnt(J-Stage: http://www.jstage.jst.go.jp/browse/ejssnt/)FIG.4: The microprobe ARPES measurements on thick graphite and single layer graphene samples at 90 eV photon energy.The horizontal dashed line marks the contribution of SiO2 substrate (reproduced from Ref. [12] with permission).low energy electron diffraction (µ-LEED) for local characterization of the crystal structure; micro-probe angleresolved photoemission (ARPES), for accessing the electronic structure.
In our study, we compared the surface roughness of suspended and supported portions of the very same graphene flake.Specially prepared samples were used for this purpose.Cylindrical pits were etched in the SiO 2 /Si support to a depth of 500 nm, so that large areas of the graphene film could be effectively suspended.To avoid charging the 300 nm SiO 2 substrate under photon and electron irradiation, the graphene flakes were grounded by Au/Cr stripes evaporated through a metal shadow mask.Figure 2 illustrates the sample layout as imaged by an optical microscope.
An important aspect of LEEM is that it allows an exact measure of the thickness of each graphene flake by monitoring the modulations in electron reflectivity at low electron energy due to quantum electron confinement in the film (see Fig. 3(a)).Due to the intrinsic sensitivity of the line shape of diffracted electron beams to local height variations, micro-probe LEED was instead used to probe quantitatively lattice distortions at length scales well below the horizontal correlation length (few tens of nm in single-and double-layer graphene), which cannot be accessed by direct imaging in LEEM.
The most prominent feature that we observed in all samples was the broadening of the diffraction profile with increasing perpendicular electron momentum transfer.This broadening showed markedly different characteristics in the case of SiO 2 -supported and suspended samples as seen in Fig. 3(b).The differences in surface morphology depending on film thickness and interaction with the substrate were quantified using diffraction lineshape analysis based on the kinematic approach by Yang et al. [11].In particular, the momentum transfer dependence of the diffraction width, measured in Fig. 3(b), gives an estimate of the roughness exponent, an important parameter that describes the short range behavior of the height-height correlation function of the surface.
In spite of the weak graphene-substrate interactions the supported layers appear to be strongly affected by the presence of the SiO 2 substrate underneath.The effect of the substrate decreases with increasing film thickness, which is explained by the increased stiffness of thicker films.The smoothening of the SiO 2 -supported graphene films with increasing thickness can be followed in the progressive sharpening of the diffraction spots as displayed in Fig. 3(b).
In order to understand better the effect of the substrate http://www.sssj.org/ejssnt(J-Stage: http://www.jstage.jst.go.jp/browse/ejssnt/) e-Journal of Surface Science and Nanotechnology and to identify the contributions intrinsic to graphene itself, we have repeated the measurements on regions suspended over substrate pits.As seen in Fig. 3(b), the suspended films are considerably smoother in the shortrange, with the suspended bilayer graphene showing sharp diffraction spots comparable to those of thick graphite films.Qualitatively, this marked change in roughness from single to bilayer graphene is in stark contrast with the progressive smoothening seen in the supported regions caused by the propagation of elastic disturbances.
The partially intrinsic nature of the corrugations in single layer suspended graphene is manifest from the weak temperature dependence of the roughness exponent (not shown) [10].On the other hand, the extrinsic factors contributing to the single-layer corrugation (i.e. the presence of adsorbates and impurities, the effect of boundary conditions in the micron-sized suspended regions, etc) are found to play an important role in determining the film morphology.
The relevance of the structural corrugations to the electronic structure have been investigated by performing layer-resolved micro-ARPES measurements on the samples described above.A comparison between graphite and single layer graphene is given in Fig. 4. The linear dispersion at the Dirac cones is revealed by fitting of the momentum distribution curves.The Fermi velocity of single layer graphene, measured around the Dirac point, was found to be very close to that of graphite and in agreement with calculations [12].On the other hand, the above-mentioned short-range corrugations observed in single layer SiO 2 -supported graphene deeply influences the ARPES measurements, resulting in an energydependent angular broadening of all spectral features.

IV. IN-PLANE SPIN REORIENTATION IN FE FILMS
The properties of Fe films grown on W(110) constitute a classical example of thin-film magnetism, in which http://www.sssj.org/ejssnt(J-Stage: http://www.jstage.jst.go.jp/browse/ejssnt/) the subtle competition between bulk and surface/interface contributions to the magnetic anisotropies result in an inplane to in-plane spin-reorientation transition (SRT) at about 80 Å film thickness [13][14][15][16].This transition between the two in-plane spin orientations can be induced either by changing thickness or temperature.In spite of the considerable information on Fe/W(110), the nature of the transition has not been known at the microscopic scale.Therefore, we have investigated the magnetic structure of Fe films around the reorientation thickness using x-ray magnetic circular dichroism imaging (XMCD-PEEM) [17].
XMCD-PEEM makes use of the magnetic contribution to the x-ray absorption cross-section as the contrast mechanism for imaging domains in ferromagnetic materials.This magnetic signal, otherwise negligible, is enhanced greatly for photon energies near dipole-allowed absorption thresholds, and can be extracted by a polarization analysis [18].On the other hand, x-ray absorption imaging and spectroscopy can be implemented in an XPEEM setup by imaging with the secondary photoelectrons while varying the photon energy.Therefore, a similar polarization analysis results in dichroic images with magnetic contrast [19].Tunability of the photon energy and control over the photon polarization are prerequisites for such studies, thus making the availability of synchrotron sources essential.
In the case of Fe/W(110) films the XMCD-PEEM images were acquired at the Fe L 3 -edge.The absorption and XMCD spectra are shown in Fig. 5(a), which is measured on the region with magnetic axis aligned with the x-ray beam ([1 10] direction for this measurement).The region with magnetization perpendicular to the x-rays give no dichroic signal.Therefore, the XMCD-PEEM image of a wedge-shaped Fe film at around the SRT thickness, displayed also in Fig. 5(a), results in a strong contrast between the two spin-orientations.Interestingly, the boundary at the transition appears sharp with a triangular shape.A simple micromagnetic model explains the particular boundary shape to be a result of the long-range dipolar interactions, which favors certain boundary orientations and forces the observed triangular distortion for the unfavorable orientations [17].The formation of these magnetic charge walls are in full display for the transition around a cone-shaped wedge displayed in Fig. 5(b).
The thickness and temperature dependent XMCD-PEEM measurements show that the transition takes place abruptly, as opposed to a continuous rotation of the spin axis.This is in line with the magnetic anisotropy parameters [14], which predict a first-order transition as a function of thickness.Importantly, the addition of an Au buffer layer at the interface has been recently shown to induce a continuous rotation of spins at the SRT in Fe/W(110) [20].This provides a nice example of how the nature of a transition can be completely altered by modifying the surface/interface anisotropy parameters.

V. STRESS-INDUCED ADSORBATE STRIPES
It is known that a competition between long-range and short-range interactions can stabilize periodic adsorbate patterns on crystalline surfaces [21].In the last decade, there have been several LEEM studies of such patterns (examples can be found in the review Ref. [3]).As a method, LEEM is highly suitable to studies on spontaneously-formed mesoscopic surface patterns, as it spans mesoscopic length scales from 10 nm to several microns with a temporal resolution fitting the fluctuations inherent in such systems [22].
A pronounced temperature dependence appears to be a common ingredient in most systems showing pattern formation under thermal equilibrium.In spite that the entropic contribution to the pattern free-energy was correctly identified as the driving force for the temperature dependence [23], a clear understanding of the consequences of disorder has been missing in the literature.This problem has been the central issue in our recent discovery of high-temperature Pd stripes on the W(110) surface [24].
Submonolayer Pd/W(110) forms well-ordered monolayer stripes along the substrate [1 10] direction at temperatures above 1100 K, as shown in the inset of Fig. 6(a).As mentioned above, the stripe period has a strong temperature dependence, and it changes from about 200 nm down at 1100 K to below 60 nm at 1170 K, at which point the stripe order disappears as observed in LEEM images.
On the other hand, diffraction measurements show that the order within the Pd lattice at the atomic scale persists up to temperatures above the stripe disordering.In terms of accessing atomic scale order using LEED, Pd/W(110) is a favorable case due to the particular structure of the Pd monolayer.Submonolayer Pd is known to be pseudomorphic to tungsten only along [001], whereas along [1 10] it has a slightly larger period compared to that of W(110) surface giving rise to superstructure diffraction spots [25].The corresponding LEED pattern is displayed in the inset of Fig. 6(b).Therefore, the order within the Pd lattice can be monitored via the superstructure spots.The intensity of the Pd diffraction spot as a function of temperature is given in Fig. 6(b) with a clear disordering transition at about 1210 K.As already mentioned, this transition temperature is considerably higher than that at which the mesoscopic order disappears as measured in LEEM.
The combination of the information at the mesoscopic scale (from LEEM) and at the atomic scale (from LEED) can be understood by considering a pattern-forming Ising lattice [24].For the Ising system, given an order-disorder transition, the fluctuation length scale diverges at the critical temperature.However, the pattern disorders at a lower temperature, when the fluctuations reach the characteristic feature size of the pattern.The result is an intermediate phase with short-range order at the atomic scale but no long-range pattern.
In further studies, we have shown that the pattern properties can be drastically modified by manipulating the stress-anisotropy.In the particular example above, this was accomplished by introducing oxygen as a second species to the Pd/W(110) adsorbate system.Via the oxygen content, we have shown that it is possible to control the pattern anisotropy and stripe period, at the extreme case resulting in a 90 • rotation of the stripes [26].

VI. IMAGING WITH INELASTIC ELECTRONS
Electron energy-loss spectroscopy (EELS) is one of the most established methods in studying surfaces [27].As a natural extension, EELS in a microscopy setting has been in use since several decades [28,29].In variation from the laterally-averaging method, energy-loss electron microscopy most commonly utilizes incident electrons of high energy (several tens of keV) and operates at a considerably lower energy resolution.Nevertheless, this combination of microscopy and spectroscopy gives the ability to perform imaging with chemical sensitivity via innershell ionization edges, along with the possibility to probe electronic excitations (such as plasmonic excitations) in a spatially-resolved manner [30].
In spite that the electron energy-loss (spectro)microscopy with high-energy electrons is a wellestablished technique for the analytical characterization of surfaces and interfaces, a similar methodology is pronouncedly unexploited using LEEM.This is in large part due to the requirement of an energy filter, which is lacking in most of the LEEM instruments in operation.Furthermore, those few instruments, which can boast an electron energy analyzer, operate with x-ray and ultraviolet light sources and predominantly utilize photoemission spectroscopy.As such, the only notable study focusing on electron energy-loss using a LEEM came in the recent years, demonstrating 50 nm lateral resolution [31].In this section, we summarize the performance of the Elettra SPELEEM instrument in imaging with inelastically scattered electrons.
As the benchmark system, we choose the surface plasmon excitation on the Ag surface.This, not only gives a favorable case with a well-resolved energy-loss feature at about 4 eV loss, but also provides a direct comparison with the performance of a similar instrument reported in the literature [31].In order to test the lateral resolution, we have prepared small Ag(111) islands by Ag growth on an oxygen-covered W(110) surface.Oxygen has a wellstudied monolayer structure on W(110) [32].Moreover, most transition metals, including Ag [33], do not wet oxygen covered tungsten surface.Therefore, the growth of Ag on a fully oxygen covered surface results in island formation as shown in Fig. 7(b).The surface was prepared by first dosing oxygen at high temperature, followed by 2 ML Ag deposition at room temperature [34].After a short post-annealing at about 200 C • , Ag islands developed with a mean lateral size of about 50 nm.Based on considerations of the Ag area coverage, we estimate the average island height to be about 1 nm.
Figure 7(a) shows the Ag plasmon peak measured by scanning the inelastic tail of the electron spectrum.The spectrum was obtained by placing the exit slit of the energy analyzer on the elastic peak, and subsequently scanning the energy losses across the exit slit by changing the analyzer bias voltage.The image acquired at the plasmon peak (about 3.8 eV energy loss) is displayed in Fig. 7(c).
The EELS image appears almost as sharp as the LEEM image in Fig. 7(b), displaying a lateral resolution of about 18 nm.This value is slightly higher than the best lateral resolution obtained in LEEM (about 10 nm in our instrument).We attribute this loss of lateral resolution partially to sample drift during measurements due to much longer exposures that are needed in EELS imaging (7 minutes for the ELS-LEEM image in Fig. 7(c), compared to seconds for the regular LEEM image).
Related to the limits on the lateral resolution, the angular distribution of the inelastic electrons are much wider in comparison to the sharp LEED spots from a wellordered surface.The difference can be viewed in Fig. 8, which shows the sharp LEED pattern dominated by the (1×1×12) oxygen order along with the diffuse energy-loss-LEED.Therefore, requirement for a small contrast aperture to improve lateral resolution is much more important in ELS-LEEM compared to LEEM, because LEEM utilizes the (00) elastic beam, which usually has a smaller angular spread than the acceptance of the contrast aperture.Therefore, future developments in correction of spherical aberrations is expected to considerably benefit imaging with inelastic electrons.

VII. CONCLUSION
The examples in the preceding sections demonstrate the capabilities of the multi-technique approach used in the SPELEEM instrument.The combination of imaging (LEEM, XPEEM) with microprobe diffraction (LEED, ARPES) and with the tunability of x-ray energy and polarization (as in XMCD-PEEM) creates a unique surface science facility with chemical, magnetic, and structural sensitivity.Furthermore, there are still unexploited techniques within the SPELEEM (such as ELS-LEEM), which may provide new opportunities in the near future.

FIG. 1 :
FIG. 1: SPELEEM sketch.The electron-optical path is traced by the blue line.The energy slit, the contrast and field-limiting apertures, marked in red, are respectively placed at energy dispersive, diffraction and image planes.The objective (obj), transfer (TL), field (FL), inner (IL), projector-1 (P1), accelerating (acc) and retarding (ret) lenses are noted in the figure.The details of the imaging column, along with the inner lens of the analyzer and projector lenses after the analyzer are omitted for simplicity.

FIG. 2 :
FIG.2: Optical microscope image of the graphene flakes placed on a SiO2(300 nm)/Si substrate.The bright lines are the metal grounding wires to avoid charging under irradiation.The flakes appear as light gray patches.The regular grid of bright dots correspond to the cylindrical pits etched into the substrate.A sketch of graphene placed partially on these pits is given in the bottom as a profile view.The gray texture underneath the graphene layer (dark black line) represents the roughness of the substrate surface.

Volume 9 (
FIG. 3: a) (upper panel) LEEM image showing graphene flakes of different thickness as marked on the figure.(lower panel) Quantum oscillations in electron reflectivity as a function of number of graphene layers.Single layer shows no features as expected.b) The HWHM of diffraction spot profiles as a function of momentum transfer along the surface normal.The layer-resolved plots are given for SiO2-supported (left) and suspended (right) films (reproduced from Ref. [10] with permission).

FIG. 5 :
FIG. 5: XMCD-PEEM investigation of the in-plane SRT in Fe/W(110).a) On the left the absorption spectra are shown at the Fe L3-edge acquired with circular polarization.The beam direction is parallel to the magnetic axis along [1 10].On the right, the XMCD-PEEM image is displayed around the SRT thickness on a wedge-shaped film sketched below.The image size is 20 µm.The thickness gradient of the wedge is 1 Å/µm.b) On a cone-shaped wedge, all possible boundary orientations are realized.The image size is about 30 µm.The spin orientations are given by arrows.

FIG. 6
FIG. 6: a) Temperature dependence of the logarithm of the Pd/W(110) stripe period as measured in LEEM.The inset shows the monolayer Pd stripes at 1130 K. Image size is 1 µm.b) Intensity of the Pd diffraction spot as a function of temperature.The Pd spot is marked on the LEED pattern in the inset acquired at 35 eV electron energy.The vertical strip between 1170 K and 1210 K highlights the temperature range corresponding to the intermediate phase.

FIG. 7 :
FIG. 7: Imaging at the Ag surface plasmon excitation.a) Energy-loss spectrum showing the plasmon loss peak at about 3.8 eV.The vertical band denotes the width (0.3 eV) and the position of the energy slit used to acquire the energy-loss image.b) LEEM image (at the elastic peak) at an energy of 18 eV.Ag islands appear bright.c) ELS-LEEM image at an energy-loss of 3.8 eV and a primary energy of 58 eV.

FIG. 8 :
FIG.8: Low-energy electron diffraction from Ag/O/W(110) using (a) elastic electrons, (b) inelastic electrons at energy loss 3.8 eV.Incident electron energy is 58 eV.The superstructure in (a) is due to the oxygen order.The diffuse ring in panel (b) stems from the surface plasmon dispersion of silver with a characteristic dependence on energy loss[31].