Introduction

The investigation of the structure of surfaces used to be the specificity of strongly interacting particles such as electrons. With the advent of synchrotron radiation, the surface scattering signal of the much weaker interacting X-rays became strong enough to offer a viable alternative. Indeed, interpretation there becomes much simpler since multiple scattering is sufficiently weak. The extremely good collimation of the ESRF undulator beams provides additional advantages: use of smaller samples and higher intensity. In addition, interface scattering can be observed since X-rays can penetrate thick cover layers while reflections unique to interface structures are singled out. A large variety of problems are now under investigation. In this context it should be mentioned that the IF CRG (Collaborating Research Group) beamline has made a large investment in attaching a complete molecular beam epitaxy laboratory to its beamline. While the major goal of this installation will be fundamental research, one can also see its interest for the investigation of industrial semiconductor devices.

 

Real time study of Ag(100) homoepitaxy with X-ray diffraction profiles

Homoepitaxy of Ag(100) at different temperatures has been studied by collecting the angular distribution of the intensities of X-rays diffracted from the growing film, in real time without interruption of the growth. This has been achieved by means of a CCD camera mounted in the diffractometer of ID3. The profiles of the diffracted intensities provide information on the temporal evolution of the long-and medium-range surface correlations. These correlations are theoretically described with scaling laws similar to those of the critical phenomena. The coarsening exponent describes the temporal evolution, during growth, of the medium-range surface correlations such as the characteristic lengths separating the nucleated islands on the surface. The roughening exponent describes the temporal evolution of the surface roughness. Our measurements allowed both exponents at different temperatures to be determined. In addition, analysis of the results suggests that the energy barrier to descend steps is small (Figure 64).

Publication

J. Alvarez (a), E. Lundgren (a), X. Torrelles (a) and S. Ferrer (a), to be published.

(a) ESRF

 

 

 

Segregation-induced critical phenomena at FeCo(001) surfaces

The influence of the surface upon bulk critical behaviour has been in the focus of scientific interest during the last decade. Since the free surface is the omnipresent natural symmetry break in any real system, the proper understanding of criticality in "semi-infinite" matter is one milestone in the endeavour to master criticality in non-ideal materials. Model systems are continuous order-disorder transitions in binary alloys. Here one current challenge is to understand on a microscopic level whether and how the surface enrichment of one species ("surface segregation") alters the critical behaviour at and near the surface.

A particularly interesting situation emerges at the (001) surface of the bcc-alloy FeCo which undergoes a continuous order-disorder transition as shown in Figure 65. Normal to the (001) direction the CsCl structure is composed of alternating layers belonging to sublattice 1 or 2, occupied respectively by Fe and Co atoms; thus the presence of the (001) surface breaks the symmetry between the two sublattices. As the X-ray scattering study described below will demonstrate, this has crucial consequences upon the surface-related critical behaviour.

Since Fe and Co are neighbours in the periodic table, ordering phenomena in FeCo produce only very weak X-ray scattering signals as mediated by the one-electron X-ray contrast |fFe-fCo|2 between Fe and Co (fFe and fCo being associated to the atomic form factors); thus the X-ray observation of surface-related order fluctuations in FeCo requires synchrotron radiation with the highest brilliance. Typical results of such a study which has been performed at the TROIKA beamline ID10A are shown in Figure 66: the asymptotic Bragg profiles (L-profiles) across the (001) superlattice position as observed closest to Tc are composed out of a Gaussian contribution generated by bulk long-range order and a broader component. The temperature dependence of the Gaussian contribution in Figure 67 can be used as a reference for the bulk order parameter Ibulk ~ t2 providing the bulk critical temperature to be Tc = 919.5 K and the bulk critical exponent to be = 0.307 ± 0.11 (inset in Figure 67), in excellent agreement with the theoretical expectation ( = 0.315).

The interesting observation in Figure 66 is the broader component which is mediated by the (001) surface and survives the bulk disordering: the temperature dependence of the integrated intensity of this surface-related order (Figure 67) now allows one to experimentally discriminate between two competing field-theoretical models implying surface order above the bulk critical temperature.

Model 1 relates a non-zero surface order at T > Tc to enhanced surface couplings and predicts a 2d-phase transition at a higher surface critical temperature leading to the curve denoted "2d-Ising" in Figure 67. This is apparently not observed.

Model 2 assumes that the (001) surface is exposed to a non-zero surface field ("h1 - field") which couples with the surface susceptibility, leading to a smooth order parameter decay in analogy to a ferromagnet in an external magnetic field. The expected temperature dependence of such a surface order denoted "11h1" is in excellent agreement with the data.

The microscopic origin of this surface field is surface segregation which leads to an enrichment of Fe at the FeCo (001) surface. Since the internal interactions favour Fe-Co nearest neighbours, the second (001) layer will be Co enriched and so forth, thereby stabilising a non-zero surface order.

Publication

S. Krimmel (a), W. Donner (a), B. Nickel (a), H. Dosch (a), C. Sutter (b), G. Grübel (b), Phys. Rev. Lett. 78, 3880 (1997).

(a) Institut für Materialwissenschaften, Universität Wuppertal (Germany)
(b) ESRF

 

 

 

In situ X-ray diffraction study of metal electrodeposition

The first stages of the electrodeposition of a metal on foreign substrate will determine the important parameters of the epilayer for further applications, like corrosion protection, metallurgy, electrocatalysis or microelectronics. In recent years, surface X-ray diffraction has been successfully extended to investigate solid-liquid interfaces such as the electrochemical ones where access to the surface structure is drastically limited by the lack of in situ techniques such as electron diffraction. The design of those electrochemical cells was, however, based on the constraints imposed by X-ray techniques, caused mainly by photon absorption or scattering by the electrolyte. The new generation of high energy synchrotron sources such as the ESRF allows work at high X-ray energy (above 20 keV) with a very high photon flux arriving on the sample. Using those major gains, a spectroelectrochemical cell was developed, based rigorously on the electrochemical requirements enabling in situ X-ray diffraction (Figure 68). Such a type of cell allows kinetic studies of surface structure transitions to be performed and to follow the deposition of adsorbates in real time. The Cu deposition on gold single crystal substrate is a model system to understand the different phenomena acting in an electrochemical interface like strong electrical fields and competition between different adsorbates that are known to dramatically influence the growth of a foreigner thin film on a substrate. Using the cell presented in Figure 68, it has been possible to measure, on ID32, at the same time (Figure 69) the intensity evolution of selected diffraction peaks, together with the electrical response of the system (Figure 69b) called "Voltammogram", which gives the charge crossing the interface and provides direct information on the Cu coverage or phase transitions at the gold surface. The diffracted intensities are proportional to the structure factor (Figure 69a) and, at selected positions, enable structural information on the surface and the interface to be obtained. The correlation of both these measurements leads to a full characterisation of the surface/interface structure and of the electrical changes associated with the kinetics. An advantage of electrochemistry over ultra-vacuum deposition is its unique ability to perform a deposition during the cathodic sweep of the potential (Figure 69a-curves in red) and to remove the deposit during the anodic sweep (curves in blue). The influence of thermodynamical parameters in phase transition associated with adsorption/desorption of cations can be investigated in a reversible way. Finally, the deposition of a thicker Cu film has being performed by electrochemistry and followed by X-ray diffraction. Despite the 14% mismatch between Cu and Au (111) lattice, we have built up a two-dimensional 10 nanometers (10-9 m) thick epilayer. The real time monitoring of the thin film during deposition by X-ray diffraction opens new perspectives for optimising the thin film growth by electrochemistry where electron-based techniques cannot be applied. We wish to thank F. Comin (ESRF) for his strong support and we also acknowledge the support of A. Jeanne-Michaud.

Publication

[1] F. Brossard (a), V. H. Etgens (a), A. Tadjeddine (b), Nucl. Inst. and Meth. B, (1997) forthcoming.
[2] V. H. Etgens (a), A. Tadjeddine (b), F. Brossard (a), M. C. M. Alves (c), D. L. Abernathy (d), Y. Borenstein (e), to be published.

(a) Laboratoire de Minéralogie-Cristallographie, Paris (France)
(b) LURE, Orsay (France)
(c) LNLS, Campinas (Brasil)
(d) ESRF
(e) Laboratoire d'Optique des Solides, Paris (France)

 

 

 

Surface and interface magnetism

Nuclear Resonance Scattering probes the magnetic and electronic structures via the hyperfine interaction in a solid. Depth-selective information of this hyperfine interaction, on a nanometer-scale, can be gained by either probe layers at various depths of the sample (i) or by variation of the incidence angle which goes along with a change in the penetration depth of the X-rays in the sample (ii).

(i) By using probe layers of 57Fe, the local magnetism as a function of depth can be studied, for instance close to the interface between a magnetic and a non-magnetic layer. To test the potential of this technique, thin layers of Fe were grown with the technique of MBE (Molecular Beam Epitaxy) on a Ge(100) substrate and covered with 2 nm of Au. 1-10 monolayers (ML) of 57Fe were incorporated at various positions in the Fe layer. The experiment was performed at ID18 where, in order to enhance the sensitivity, a grazing incidence geometry was chosen. The magnetisation was aligned parallel to the linear polarisation of the incoming photon beam, so that only the four m = ±1 transitions contribute to the time spectrum of the reflected intensity.

Samples grown with 1 ML of sulphur as a surfactant show a dominant magnetic pattern if the Ge interface is 40 ML away from the probe layer (Figure 70, bottom panel). The linewidth obtained from the fit is only 20% larger than the natural linewidth, pointing to an excellent structural quality of the layer. A dramatic change is observed when there is only 5 ML natural Fe between the probe layer and the interface (Figure 70, top panel). The parameters of the resulting non-magnetic spectrum are consistent with the formation of a mixed Ge/Fe interlayer. Although not immediately obvious, this non-magnetic component is also present in the other spectrum, with a relative intensity of 7%. Samples grown without sulphur show less interdiffusion but also broader lines. The results show that nuclear resonant scattering is capable of detecting interface magnetism with a sensitivity of 1 ML, the limit determined by count rate.

(ii) Around the critical angle of the (electronic) total reflection angle, c, the penetration depth of photons strongly depends on the glancing angle and decreases to a few nm for 0. Therefore, if the time structure of nuclear resonant photons specularly reflected from a thin film is measured at increasing , the hyperfine interaction is integrally sampled in surface layers of increasing thickness. Taking into account the interference of the radiation reflected from different depth and carrying out a simultaneous evaluation of the spectra corresponding to various glancing angles, one can extract a depth profile of the hyperfine fields. This can be used for identifying different local environments, chemical phases at different depth as well as for revealing their magnetic structure.

Figure 71 shows time spectra taken on an oxidised thin film of 57Fe at two glancing angles. One spectrum was taken within about 20 minutes, in contrast to the typical data collection time of an analogous energy domain Mössbauer experiment of several weeks. The penetration depth that belongs to the upper curve is about 20 nm so that this spectrum stems from the whole thin film. The fast beating is characteristic of an iron oxide of spontaneous magnetisation, probably Fe3O4. The penetration depth of the 14.4 keV photons at the glancing angle of the lower curve is only about 2 nm corresponding to the uppermost 8­10 atomic layers. Here no magnetic interaction is seen; this layer is probably -FeOOH.

Publications

[1] L. Niesen (a), A. Mugarza (b,c), M.F. Rosu (b), R. Coehoorn (d), R.M. Jungblut (d), F. Roozenboom (d), A.Q.R. Baron (e), A.I. Chumakov (e), R. Rüffer (e), to be published
[2] D.L. Nagy (a), L. Bottyán (a), L. Deák (f), V.N. Gittsovich (g), O. Leupold (f), R. Rüffer (e), V.G. Semenov (g), H. Spiering (h), E. Szilágyi (a), to be published.

(a) KFKI Research Institute for Particle and Nuclear Physics, Budapest (Hungary)
(b) Materials Science Centre, University of Groningen, (The Netherlands)
(c) Universidad de Bilbao (Spain)
(d) Philips Research Laboratories, Eindhoven (The Netherlands)
(e) ESRF
(f) University of Hamburg (Germany)
(g) University of St. Petersburg (Russia)
(h) University of Mainz (Germany)

 

 

 

Local structural relaxations in strained thin films

The microscopic deformation suffered by the local structure of semiconductor heterostructures under epitaxial strain has attracted considerable attention lately. The main issue is whether and to what extent the deformations are affected by the magnitude and sign of the applied stresses. A comparison with the classical results on bulk pseudobinary alloys, in which the bond lengths have a strong tendency to remain close to their values in binary compounds, is also of interest.

With this objective, we have performed XAFS measurements in InxGa1-xAs thin films grown on InP(001) in the concentration range x = 0.3 to 0.7. The thickness of all samples was below the critical value for strain relaxation, i.e. all samples were pseudomorphic. The in-plane strain ranges from -1.1% to 1.6% ( from compressive to tensile).

XAFS measurements were performed in the fluorescence mode at the Ga and As edges on the CRG beamline GILDA (BM8), using dynamical sagittal focusing. The analysis of the first shell data has been performed using Fourier filtering techniques; the values of the Ga-As and As-In bond lengths so determined are shown in Figure 72 as a function of concentration along with a linear fit to the data.

With the experimental points we also show the predictions of the Virtual Crystal Approximation, of the model by Cai and Thorpe for unstrained alloys and of a model we have developed for strained layers (SL model). The main assumption of the latter model is that for negligible tetragonal distortion the local structure of the epilayer out of the growth plane is the same as that of a bulk alloy.

The experimental points clearly show a different behaviour as a function of concentration with respect to the unstrained alloys; in fact the Ga-As bond actually has a negative slope. We find an excellent agreement with the SL model. This is the first evidence that first shell bond lengths can change as a function of strain, illustrating the effect of the substrate lattice parameter on the local structure of the epilayer.

Publication

F. Romanato (a), D. DeSalvador (a), M. Natali (a), M. Berti (a), A. Drigo (a), S. Pascarelli (b), F. Boscherini (c), S. Mobilio (d), G. Rossetto (e), A. Camporese (e), G. Torzo (e) and Lamberti (f)

(a) INFM, (Italy)
(b) GILDA CRG, ESRF
(c) INFN, (Italy)
(d) Dipartimento di Fisica, Università Roma III, (Italy)
(e) ICTIMA-CNR, Padova (Italy)
(f) Dipartimento di Chimica, Fisica e dei Materiali, Università di Torino (Italy)

 

 

 

Kinetics of alloying of ultra-thin Co films on Pt(111)

Ultra-thin superficial alloy layers are a most interesting new material. Their elaboration will be governed by local equilibrium properties, known to often differ from what happens in bulk material. Thus, metastable new phases can be prepared, some of which exhibit interesting magnetic properties. This is the case for Pt-Co alloy films, on Pt(111) substrates, which can present an unexpected easy magnetisation axis perpendicular to the film. Obviously several energy terms (size effect, surface tension, structure, preference to hetero-bonding) compete in a subtle way to control the structure and composition of ultra-thin alloy films during their elaboration in a layer-by-layer fashion. It is important to separate their respective contributions to understand how to control these new structures and possibly their new properties.

We have performed, with the 'SUV' diffractometer at the French CRG-IF (BM32) beamline, a detailed study of 'rodlike' intensity distribution during the annealing of ultra-thin deposits of Co on a Pt(111) substrate. Co was deposited at room temperature on the Pt(111) substrate: due to the large lattice mismatch (10%) the Co film grows with an unrelaxed dense (111) stacking leading to a new set of rods, parallel to the substrate ones. Their intensity distribution depends on the surface film morphology and on the composition from the outermost layers down to the buried interface. In situ diffraction measurements were taken during carefully controlled long annealing conditions up to 460 °C, just before complete dissolution. Two situations were investigated: i) a thin deposit of 3 Co monolayers (ML) and ii) a "thick" deposit of 10 Co ML. Real time changes of the deposited film were followed by measuring the position and the intensity of several rods. Drastic changes have been observed according to the initial Co thickness.

After gentle annealing, the thick 10 ML Co deposit turned to a dominant hcp phase before any interdiffusion; this phase exhibited a strong resistance to Pt diffusion and switched rather "abruptly" to a PtCo fcc alloy around 400 °C (bulk transition temperature from Co hcp to Co fcc). The 3 ML Co deposit, too thin to exhibit an hcp phase, changed progressively to an fcc alloy.

Figure 73 evidences the two kinds of kinetics. "Kscans" versus temperature are presented as a map; the Pt rod trace sits at K = 1, the other peak, at K = 1.1 for 30 °C, arises from Co film and it shifts towards K = 1 when more and more Pt is incorporated.

Figure 74 displays a few reflectivity curves recorded during annealing of the 10 ML deposit, to illustrate how to obtain depth profile information; the increasing number of Kiessig fringes indicates a thickening of the film in a layer-wise mode, but with Pt/alloy interface more and more diffuse.

X-ray diffraction proves indeed to be most powerful to follow in real time the formation of alloy metastable states. The final metastable alloy (annealing up to 450 °C) is however the same Pt60Co40 with bulk-like features: fcc structure, Pt surface segregation and partial ordering compatible with the L10 phase.

Publication

M.C. Saint-Lager (a), R. Baudoing-Savois (a), M. De Santis (a), P. Dolle (a), to be published.

(a) Lab. de Cristallographie, CNRS, Grenoble (France)

 

 

 

NiO(111) single crystal surface study

The understanding, at an atomic level, of the mechanisms involved in metal/oxide interfaces during their formation is of great scientitic interest. It is known that the structure, composition and morphology of ceramic surfaces strongly influence their overall properties as well as the properties of the heterostructures one can build on them. Because of their intrinsic insulating character, experimental results are scarce but the theoretical models need them in order to refine the parameters of interest. Grazing incidence X-ray diffraction (GIXD) in ultra-high vacuum conditions being insensitive to charge build-up, it is an ideal tool to investigate the structure of these interfaces. The SUV apparatus on BM32 (CRG-IF) is optimised for such GIXD studies.

The NiO(111) surface exhibits particularly interesting properties: it is insulating at room temperature, highly resistant to corrosion and is an antiferromagnet with the pure spin planes parallel to the surface. This material is thus a good candidate for the elaboration of sensors operating in harsh environments. However no experimental investigation of the single crystal surface is available. In the past, it was believed that the Coulomb interaction may be long-ranged and that polar surfaces like NiO(111) must decompose in macroscopic (100) facets. On the other hand, recent theoretical works by D. Wolf demonstrated that the problem may be overcome by summing over neutral shells of molecules. The resulting surface should then be p(2x2) reconstructed exhibiting small micro-pyramids. In order to get a clear answer we have investigated experimentally the NiO(111) single-crystal surface by GIXD.

After air-annealing at 1300 K for re-crystallisation and without outgassing, strong crystal truncation rods (CTR's) (Figure 75) were measured perpendicularly to the surface. At the out-of-phase conditions (in-between Bragg reflections) the rod is still intense and sharp (Figure 75 - inset), indicating a surface of very small roughness, with large atomically flat terraces only limited by the mosaic spread of the crystal. In-plane measurements (Figures 76a, 76d) along the high symmetry directions show that the surface structure is almost unreconstructed. This simple measurement shows immediately that the surface does not decompose in (100) facets in our preparation conditions. After outgassing at 860 K, the in-plane measurements (Figures 76b, 76e) reveal two new and strong features: extra peaks at each half integer value corresponding to a p(2x2) reconstruction, and Bragg peaks at the exact positions expected for epitaxial relaxed FCC Ni(111) (h = 1.17 and 2.24 in Figure 76b). Out-of-plane scans confirm the 3D character of the metallic Ni. At this stage the surface is reduced and decorated with Ni clusters. The surface was heated again in a partial pressure of oxygen at 860 K. This oxidation leads to vanishing of the metallic Ni clusters. Only the p(2x2) reconstruction remains (Figures 76c, 76f). Macroscopic facetting was never observed in these conditions. The signal on the NiO rods increases (Figure 76) showing that the surface is flatter than in any of the previous situations. We may think that the surface is completely reconstructed and flat at this stage.

From our experiment we have obtained a rather unexpected result: we must conclude that, below 1300 K, this surface is stable, provided no oxygen vacancies are produced. Roughening occurs when the surface undergoes a reduction i.e. oxygen desorption, but this phenomenon is reversible. These results confirm the theoretical predictions and open the possibility of studying epitaxial growth on NiO(111). Moreover our observations may be of use for preparing high quality NiO(111) thin films.

Publication

A. Barbier(a), G.Renaud (a), Submitted to Surf. Sci. Lett. (1997).

(a) DRFMC, CEA-Grenoble (France)

 

 

 

Layering of liquid Ga in contact with diamond(111)

Already in 1891, W. Sutherland noted that the properties of condensed matter change drastically when one goes from a solid, where "each molecule is hemmed in by its neighbours, so it cannot change its position" to a liquid, where "each molecule ceases to be hemmed in and manages to wriggle through amongst its neighbours". Now, what happens microscopically at the interface between a liquid and a solid? One may expect that the atoms or molecules in the liquid adjacent to the solid wall order themselves in layers parallel to the interface. Indeed, this is what we have observed, but the effect extends only over a very limited depth interval.

In an experiment at the undulator beamline ID10A, we have performed an X-ray reflectivity study of liquid gallium in contact with a (111) oriented diamond substrate. We made sure that the solid-liquid interface was atomically clean. Since diamond is essentially transparent to energetic X-ray (17 keV), we could let the incident beam pass right through the diamond and thus reach the interface from the diamond side (see inset of Figure 77a).
The specularly reflected X-ray intensity is found to exhibit a strongly damped oscillation as a function of the value of the perpendicular component Q. This is the signature of a rapidly decaying oscillatory density profile within the liquid. The best fit to the data is obtained for the decaying oscillatory density profile shown in Figure 77b. The oscillation period in the Ga profile equals 3.83 Å. This distance coincides with the distance between two consecutive (001) planes of almost upright oriented Ga2 dimers in solid -Ga, which is the stable solid phase of Ga at low temperature and ambient pressure. The layering extends into the bulk liquid with a 1/e-decay depth of 4 Å. This decay length is approximately equal to that of the Ga bulk pair correlation function, indicating that short-range order in bulk liquid Ga and layering at the interface are closely related, even though the layering period is different. A schematic representation of our model for the layered liquid on top of 2x1 reconstructed diamond surface is shown in Figure 77c.

Our observation may have implications for the understanding of freezing transitions in liquids in contact with the container wall. Whether the layering affects the energetic barrier to heterogeneous freezing, and hence the maximum supercooling interval, is still an open question warranting further investigation.

Publication

W.J. Huisman (a), J.F. Peters (a), M.J. Zwanenburg (a), S.A. Vries (a), T.E. Derry (d), D.L. Abernathy (b) and J.F. van der Veen (c), to be published.

(a) FOM Institute for Atomic and Molecular Physics, Amsterdam (The Netherlands)
(b) ESRF
(c) University of Amsterdam (The Netherlands)
(d) University of Witwatersrand, Johannesburg (South Africa)