Final report - Leibniz Gemeinschaft
Transcription
Final report - Leibniz Gemeinschaft
Final report Space-resolved nanomechanical properties of functional surfaces experimenent and simulation Leibniz-Institute: Leibniz-Institut für Oberflächenmodifizierung Reference number: SAW-2011-IOM-2 Project period: 01.05.2011-30.11.2014 Contact partner: Prof. Dr. Stefan G. Mayr 1 Space-resolved nanomechanical properties of functional surfaces -experimenent and simulation Executive Summary The aim of the project was the space-resolved investigation of mechanical properties of surfaces, thin films and nanostructures. The work was split into three parts according to the three participating work groups. Within the group of Prof Grundmann core shell ZnO-BiFeO 3 heterostructures were deposited using pulsedlaser-deposition (PLD) to demonstrate the realizability of ferroelectric heterostructures. The mechanical characterization was done by measuring the magnetoelectric coupling of BaTiO 3 -BiFeO 3 multilayer structures. The magnetoelectric coefficient was thereby measured as function of PLD oxygen pressure during growth and temperature. Within the group of Prof Rauschenbach hexagonal GaN films were grown on 6H-SiC(0001) substrates by nitrogen ion beam assisted molecular beam epitaxy (IBA-MBE). Using contact-resonance atomic force microscopy (CR-AFM) the indentation modulus was measured using CRAFM for three different layer thicknesses at three different temperatures. Thereby it was shown that the indentation modulus is significantly higher for material samples with a thickness of only a few contact radii. This effect is attributed to the stiffer SiC substrate. Additionally it was shown that the sample is significantly softer at a temperature of 600 °C compared to the substrate temperatures of 700 °C and 730 °C. This effect can be traced back to the number of defects in the sample. Furthermore the indentation modulus of Substoichiometric nitrogen ion implanted silicon was measured. While after implantation the indentation modulus decreases, implantation with fluences > 1 × 1016 N ions cm--2 and subsequent annealing led to an increase of the indentation modulus above values of crystalline silicon. X-ray photoelectron spectroscopy showed the presence of Si-N bonds that can explain the stiffening effect of the procedure. Within the work group of Prof. Mayr, a combined approach of systematic CR-AFM measurements and finite element analysis was performed to study the behavior of the tip modulus as fit parameter within a multireference sample analysis. The strong deviation found between tip modulus and indentation modulus of the tip's material is essentially explained by its shape. Finite element analysis unraveled the relationship between opening angle of the tip apex and the effective tip modulus. Using DFT simulations the elastic constants of silicon and strontium titanate thin films was studied and used to design a FEM layer model to study the influence of surface elasticity on the indentation modulus. Additionally an analytical model is proposed describing the reduction of the indentation modulus with contact radius agreeing well with the simulation results. Finally, mechanical properties around GaN crystal steps were studied using molecular dynamics simulations. A reduction of the microscopic elastic constants was measured within a lateral distance of 1 nm to the step edge. The effects of surface presence and breakdown of half-space symmetry were discussed separately by employing additional finite element simulations. 2 Final report for SAW project Space-resolved nanomechanical properties of functional surfaces -experimenent and simulation 1. Questions at the beginning and aim of project (see project application) Surfaces, nanoscaled objects as well as nanostructured solids show up different mechanical properties compared to their bulk counterparts. Although the main origin lies in the existence of interfaces or boundary surfaces, i.e. a reduced dimensionality, the actual underlying physical principles are not well understood up to now. The objective of this research project is the investigation of local mechanical properties during nanoscaled deforming of surfaces, thin films and freestanding nanostructures. Using representative examples of a few uniquely defined metals and semiconductors, we want to enlighten the underlying materialphysical scenarios. Therefore we want to deploy systematic experiments and computer simulations (ab initio, classical molecular dynamics and finite elements). The gained knowledge should be used for the development of new concepts optimizing mechanical properties of functional surfaces and nanostructures. 2. Development of realized work including deviations from initial concept project partner Universität Leipzig, Institut für Experimentelle Physik II, Abt. Halbleiterphysik: Preparation of surfaces, thin films and nanostructures coworkers in the project: Prof. Dr. M. Grundmann, Dr. Helena Franke, M.Sc. Peter Schwinkendorf, Prof. Dr. M. Lorenz Semiconductor Physics Group at University Leipzig has experience for decades in growth of thin oxide films, multilayers with single layer thickness in nm-range [UL1 to UL11], and nanostructures [UL12 to UL16] by pulsed laser deposition (PLD). With that, the initial conditions for the envisaged preparation work can be considered as excellent. However, we realized during the first attempts to grow BaTiO3based nanowire arrays that the established ZnO growth processes cannot be adopted to BaTiO3, see results below. ZnO exhibits a strongly anisotropic hexagonal wurtzite structure while BaTiO3 is tetragonal (nearly cubic) and almost isotropic. Therefore, we have grown core-shell nano heterostructures consisting of a ZnO core and a BaTiO 3 shell. Because of the dimensions in the nm-range, the corresponding nanomechanical characterization of the core-shell structures is not trivial. Therefore we worked in parallel on planar multilayer structures of BaTiO 3 and BiFeO 3 . In such structures, information about the nanomechanical networking of both crystalline phases can be gained by measurement of the magnetoelectric coupling, i.e. the strain-coupled interaction of the magnetic (BiFeO 3 ) and ferroelectric component (BaTiO 3 ). This strain-coupling on a microscopic length scale is a current hot research topic because of the application relevance of multiferroic and magnetoelectric composites for future memory concepts. 3 project partner Leibniz-Institut für Oberflächenmodifizierung – work group Prof. Rauschenbach Characterization of surfaces, thin films and nanostructures Beteiligte Mitarbeiter: Prof. Dr. B. Rauschenbach, M. Sarmanova Contact resonance AFM For the mechanical characterization we used a self-implemented CR-AFM based upon a commercial programmable AFM (MFP-3D, Asylum Research), which allowed us to perform a quantitative analysis of the samples based upon a multi-reference samples approach [IOM-1, IOM-2]. The AFM probes used for CR-AFM imaging were PPP-NCLR (NanoSensors, Switzerland) with spring constants in the range between 19 and 33 N/m. Accurate values of the spring constants were determined by an integrated thermal-noise calibration method of the employed AFM. The applied tip loads varied between 225 and 900 nN, depending on individual probe stiffness and setup configuration. Representative 2nd contact resonance frequency values for investigated samples were acquired over the squared area on each sample. The Quantitative measurements were conducted by point mapping over 1×1 μm2. 20 uniformly distributed points were taken for each square area. The multireference samples approach was employed for the quantitative stiffness analysis. The Qualitative results were obtained by continuous scanning over 100×100 nm². Three types of reference samples were used: amorphous fused silica (FS, indentation modulus M s = 75 GPa), (100) oriented silicon (Si, M s = 165 GPa) and sapphire (M s = 433 GPa). After the CR-AFM measurements, every individual probe was visualized from the top and side by means of scanning electron microscopy (SEM). The obtained images were subsequently digitized for finite element analysis of the cantilever dynamics, yielding relationship between contact resonance frequencies and contact stiffness of the tip-sample system k. The latter describes the interactions between the tip and the sample. Indentation moduli of the implanted samples were determined by fitting the equation k m ∗ E ∗ = Eref ( ) k ref using experimental data points for both reference samples. Here, the index ref relates to fused silica and Si reference samples, respectively. E* is the reduced elastic modulus of the tip-sample system defined from the indentation moduli of the tip M tip and the indentation modulus of the sample M sample : 1 1 1 = + ∗ Mtip M𝑠𝑎𝑚𝑝𝑙𝑒 E Materials GaN films The investigated samples were epitaxial c-plane (0001) gallium nitride (GaN) films. They were grown on 6H-SiC(0001) substrates by nitrogen ion beam assisted molecular beam epitaxy (IBA-MBE). In situ analysis by reflection high energy electron diffraction (RHEED) was performed for controlling the 4 deposition process. The alterable processing parameters in this study were the SiC substrate temperature during deposition and the deposition time. Three sets of films were produced. Every set corresponds to certain substrate temperature during deposition: 630°C, 700°C or 730°C. Ga remains on the surface were transformed to homoepitaxial GaN by a post-ion nitridation process step at substrate temperature of 700°C. Films with three characteristic thicknesses were grown for every substrate temperature: “thin” films with the thickness of several tens of nm, “middle-thickness” films with the thickness of around 100 nm and “thick” films with the thickness higher than 200 nm. The crystalline structure of the GaN films was studied ex situ by x-ray diffraction (XRD). Sub-stoichiometric nitrogen ion implanted silicon Commercial (100)-oriented silicon samples with a thickness of 410 µm were used as substrate material. The samples were implanted at room temperature with 100 keV N+ ions at fluences between 1 × 1015 and 1 × 1017 cm-2. Afterwards, the implanted samples were annealed at 800°C in three half-hour steps in Ar gas flow between 0.1 and 0.2 l/min, employing a heating rate of 100°C/min. The depth distribution of nitrogen was determined by time-of-flight secondary ion mass spectrometry (ToF-SIMS) with 2 keV O+ ions for sputtering and 15 keV Ga+ ions for creating the secondary ions for analysis. For the nitrogen content, monitoring the secondary ion intensity of Si 2 N+ was found to be the most effective. Depth profiles were recorded after the first (0.5 h) and the third annealing step (1.5 h). The structure of a Si sample implanted with the highest fluence of 1 × 1017 cm2 and annealed for 1.5 h was investigated by glancing angle X-ray diffraction measurements (XRD). Chemical bonds states were studied by means of x-ray photoelectron spectroscopy (XPS). Depth profiles for the signals of O 1s, N 1s and Si 2p were measured. Quantitative analysis of the spectra and simulation of the chemical states of the elements in the case of insufficient energetic resolution was performed using commercial XPS data analysis software. Particularly, summing of functions representing individual peaks corresponding to SiN x , SiO 2 and Si (including a second shifted peak due to band bending) gives a curve that represents the measured Si 2p peak. Here, a depth step was corresponding to approximately 20 nm. The depth calibration was in all cases realized by optical confocal microscopy with 10 nm resolution for determining the final crater depth. The measurement of complex contact stiffnesses and Q-factors of real surfaces, amorphous thin films/glasses was too much of challenge, but first measurements of glassy metals were already done and suggest that surface changes slightly during the measurement. Project partner Leibniz-Institut für Oberflächenmodifizierung – work group Prof. Mayr Modelling of surfaces, thin films and nanostructures coworkers in the project: Prof. Dr. S. G. Mayr, M. Jakob, J. Buchwald CR-AFM With the help of the performed quantitative studies [IOM-1, IOM-2], we were able to make an essential contribution to the enhancement and better understanding of the method using finite element analysis of the tip-sample contact [IOM-3], which explains the occurrance of the low tip modulus compared to the indentation modulus of the material. 5 Thin films To investigate the mechanical behavior of surfaces, we first applied density functional theory (DFT) calculations to ultrathin films, i.e. films up to 5 lattice constants thickness. Utilizing the multiscale approach, the results were compared to atomistic simulations and interpreted employing a continuum model that contains surface elasticity from the beforehand calculated surface elastic constants. These constants were implemented building up a layer model that reflects the surface elasticity. Given that local and surface sensitive properties are nearly exclusively accessible by scanning probe techniques we used FEM simulations to study the effect of surface elasticity on indentation response. From the results, especially, the contact radius and force (of a spherical tip) dependence of the indentation modulus, we found an analytical model that describes the reduction of the indentation modulus at a surface. Nanostructures Besides smooth surfaces we also looked at the mechanical behavior of a step of a few lattice parameters in height. This was done by evaluating a microscopic expression of the elastic constants, derived from the fluctuation-dissipation theorem and also by the indentation modulus derived down to the atomic scale. But also for such a local expression surroundings like the break-down of half space symmetry plays a non-negligible role. Therefore additional finite element simulations, in which surface stresses and elasticity are omitted, were done and compared to atomistic simulations. 3. Presentation of achieved results and discussion, future applications UNI – work group Prof. Grundmann (a) Growth of ferroelectric BaTiO 3 nanostructures BaTiO 3 nanostructures were grown in the PLD Q-chamber (Fig. 1) at temperatures in between 500°C and 950°C, Ar-pressures in between 1 mbar and 200 mbar, and a flow rate of 10 sccm, and target-tosubstrate distance from 3 to 4 cm. Pressures around 1 mbar correspond to the PLD film mode (used for ZnO films), while higher pressures around 100 mbar induce a vapour-liquid-solid (VLS) growth process (reported for GaAs nanowires), Fig. 1. Scheme of PLD Q-chamber for growth of nanostructures: Focused laser pulses (a) provided by the excimer laser hit the target (b) and excite the plasma (c). The plasma expands in off-axis direction and is scattered at the background gas to condensate at the substrate (d). The following growth processes were testes in detail: • at as-received substrates: fused silica (SiO 2 ), SrTiO 3 (100) • at substrates with Ti and TiO x nucleation layer 6 • • • at Si substrates with Ti nucleation layer plus Au nanoparticles as additional catalyst (Fig. 2) at preprepared BaTiO 3 film with Au nucleation nanoparticles (from gold colloide solution) at different prepared layer stacks: BaTiO 3 , ZnO template from W-chamber, ZnO-wire template, GaN-wire template • at substrate with Au catalyst from colloide solution. There were a lot of experimental efforts to extend the well developed state of knowledge on growth of ZnO-based nanostructures in the Grundmann group at University Leipzig, see for example [UL1, UL8, UL12-16], to ferroelectric materials such as for example the cubic BaTiO 3 . For that, the well established high-pressure pulsed laser deposition [UL13, UL15] was adapted to BaTiO 3 . Figures 2 to 4 show results of these first attempts to grow BaTiO 3 nanostructures. Fig. 2. Scanning electron microscopy (SEM) images of typical BaTiO 3 nanostructures (sample Q1417) on SiO 2 (fused silica) with gold nucleation seeds. The scale bars bottom right correspond to 300 µm (left, 60 x) and 30 µm (right, 600 x). Au (200) Au (111) 64 BTO (002), (200) 100 BTO (111) 144 BTO (101), (110) BTO (001), (100) Intensity (counts) Fig. 3. Spectrum of EDX element analysis (left) and SEM image (right) of a BaTiO 3 nanostructure (Q1417), see Fig. 2. The EDX analysis evidenced the relevant elements Ba und Ti (of BaTiO 3 ), Au of the nucleation layer, and Si of SiO 2 substrate. The scale bar in the SEM image corresponds to 3 µm. 36 16 4 0 15 20 25 30 35 40 45 50 55 60 65 2Theta-Omega (°) Fig. 4. X-ray diffraction pattern (2 θ-ω scan with Cu K α ) of the BaTiO 3 nanostructures of Fig. 2 and 3 (same sample). Beside the reflections of gold the BaTiO 3 -peaks are visible, which are broadened due to the reduced coherence length in the nanostructures. Indexing of peaks was done according to JCPDS 2.0, Au 04-0784 and BaTiO 3 83-1880. Figures 1 and 2 show that the shape of the BaTiO 3 nanostructures is up to now not regular, although the transfer of chemical elements in the high-pressure PLD process appears stoichiometric. Most 7 probably, the almost isotropic tetragonal/cubic crystalline structure of BaTiO 3 with isotropic growth velocities in all three crystallographic directions is the reason for its different growth behavior in relation to the wurtzite ZnO. The hexagonal ZnO structure with highly anisotropic growth velocities in a- and c-direction induces the preferential c-axis oriented growth of ZnO-based nanostructures with high aspect ratio [UL13, UL15]. Because of the missing anisotropy in BaTiO 3 , we have investigated ZnO nanowires as core which are covered with nm-thin BaTiO 3 films, see following section. (b) Core-shell ZnO-BaTiO 3 nano heterostructures Another possibility to grow BaTiO 3 nanostructures are core-shell structures made of ZnO nanowires (core) deposited with a thin BaTiO 3 film (shell). Such structures with hexagonal core (ZnO) and cubic wrapping (YSZ/Al 2 O 3 Braggspiegel) have been successfully demonstrated in the work group. Therefore, we consider this route as practicable for the realization of ferroelectric nanostructures. (a) (b) Fig. 5. (a) Deposited GaN nanowire. The unregular surface structures are BaTiO 3 . (b) Horizontal cross section of a covered ZnO nanowire. In the center is the ZnO core, which is covered by BaTiO 3 , and subsequently by gold and platinum. (a) (b) Fig. 6. (a) Cross section of a BaTiO 3 -covered ZnO wire (diameter ca. 200 nm), with designation of the materials. From center to outer edge of the structure we see the almost hexagonal ZnO wire, a BaTiO 3 layer, and gold plus platinum. (b) ZnO-BaTiO 3 core-shell structure with designation of individual diameter and thickness of ZnO wire and BaTiO 3 film, respectively. 8 (a) (b) Fig. 7. (a) Vertical cross section through a core-shell structure. (b) Obviously, at the top side of the ZnO wire remains a hollow space by covering with the BaTiO 3 shell. The dimensions of the hollow space are given in (b). Furthermore, we have deposited in the PLD G-chamber existing nanowire samples of ZnO and GaN. Homogeneously covered core-shell structures were the result (Fig. 5-7). Vertical growth of BaTiO 3 could be conditionally observed. Figures 5 to 7 show as final result the successful growth and microscopic demonstration of ZnOBaTiO 3 core-shell nano heterostructures by means of a combination of high- and low-pressure PLD techniques (unpublished). (c) BaTiO 3 -BiFeO 3 multilayers with magnetoelectric coupling The above in section (b) demonstrated ZnO-BaTiO 3 core-shell structures could not be further characterized by nanomechanical and ferroelectric investigations in the course of the project because of limited experimental possibilities for nano-sized samples. Therefore, the activities were further developed towards multiferroic 2-2 nanocomposite thin films. (a) (b) Fig. 8. (a) Transmission electron microscopy image (dark field STEM, (110) cross section) of a multilayer with 15 layer pairs BaTiO 3 /BiFeO 3 at MgO(100), grown with 0.01 mbar oxygen pressure (G4178). (b) XRD reciprocal space map around the SrTiO 3 (001) reflection. The multilayer shows clearly resolved fringe satelite peaks because of the layer stack. The satellite peaks are broadened horizontally due to the tilt mosaizity of the films. The images were taken from [UL17]. 2-2 nanocomposites are multilayer structures consisting of two different crystalline phases (here in our case BaTiO 3 and BiFeO 3 ), that are both arranged two-dimensionally (called 2-2 composite) on top of each other. Fig. 8 shows a typical cross section taken by transmission electron microscopy (TEM, 9 kindly measured by G. Wagner and O. Oeckler from Faculty of Chemistry of Universität Leipzig) of such a heterostructure with nano dimensions. The single layer thickness amounts here only 14 to 23 nm, and at higher PLD growth pressure only about 6.1 nm (BTO) and 7.7 nm (BFO), for more details see [UL2 and UL17]. 1x10-3 50 BaTiO3/BiFeO3 Multilayer p(O2) = 0.01 mbar 0 0 -25 Polarization (10-2 C/m2) 25 Current (µA) 5x10-4 20 -20 -50 (a) -26 -13 0 13 Electric field (106 V/m) 26 µ0M(T) T = 300 K 10-4 0 -10-4 0 -0.3 BTO BFO BTO67/BFO33 BTO33/BFO67 15xBTO/BFO -5x10-4 -1x10-3 (b) -4 T = 300 K 0.0 0.3 -3 -2 -1 0 1 2 Magnetic field µ0H (T) 3 4 Fig. 9. Multiferroic properties of BaTiO 3 /BiFeO 3 multilayers: (a) Typical dynamic current-voltage loop (green) mit indications of broad ferroelectric switching peaks, and ferroelectric hysteresis loop P(E) (blue), measured at 2 kHz. (b) Magnetic hysteresis loop of a 15 x BTO/BFO multilayer (red), in comparison with the indicated composite and single layer samples. The multilayer (red curve) shows highest saturation magnetization. Figures taken from [UL2]. Fig. 9 demonstrates the multiferroic functionality of the BaTiO 3 -BiFeO 3 multilayers at room temperature (300 K), i.e. the samples are simultaneously ferroelectric (Fig. 9 a) and weakly ferromagnetic (Fig. 9 b). Of particular interest for future application in memory devices is however the magnetoelectric coupling in the samples, i.e. if the ferroelectric polarization can be influenced by means of a magnetic field. Experimentally available is the dynamically measured magnetoelectric coefficient αME = δE/δH demonstrated in Fig. 10, see [UL2] and references therein for details of measurement. The generally accepted model of magnetoelectric coupling is based on the magnetostrictive and piezoelectric effect. A magnetic field induces strain of the magntostrictive phase of the composite, which is transferred mechanically via strain coupling to the piezoelectric phase and results in changed ferroelectric polarization. The demonstrated ME coefficient of the multilayer samples at room temperature can be controlled via the PLD growth pressure, as shown in Fig. 10 (a). The detailed temperature dependence of ME coefficient as given exemplary for two different multilayer samples in Fig. 10 (b) is up to now not understood completely. The structural phase transitions of BaTiO 3 may play a role here. This is subject of current and future investigations to design multiferroic composites with even higher magnetoelectric coupling suitable for applications [UL17]. 10 20 15 10 5 (a) 20 ME coefficient (V/cmOe) ME coefficient (V/cmOe) 25 0.01 p(O2)(mbar) 0.1 18 16 14 12 10 (b) 8 0 24 22 20 18 0.1 mbar 16 0.25 mbar 14 12 10 50 100 150 200 250 300 Temperature (K) Fig. 10. Magnetoelectric coefficient of BaTiO 3 /BiFeO 3 ×15 multilayers in dependence on (a) the PLD oxygen partial pressure during growth, and (b) on the temperature for two selected growth pressures as indicated. The figures are adapted from [UL17]. IOM – work group Prof. Rauschenbach GaN films The AFM images in Fig. 1 represent typical surfaces of epitaxial GaN films. Surfaces of all specimens mostly consist of the terrace-step structure which is characteristic for the preferred two-dimensional growth. From RHEED and XRD measurements it is known that the produced GaN films consist mainly of the thermodynamically stable hexagonal GaN polytype. There is no big qualitative difference between the topography of films prepared with different substrate temperature conditions. Nevertheless, the surface of GaN films prepared on the SiC substrate at the temperature of 630°C exhibits more pores and consists of smaller crystallites. This is induced by a lower adatom mobility during growth at the lower substrate temperature. Fig. 11. Topography of the “middle-thick” GaN films (~100 nm). Substrate temperature: a) 630°C, b) 700°C and c) 730°C. AFM images. Dealing with the low-thickness films demands accurate determination of the contact area, because the problem of the substrate influence on the measured elasticity value is of great interest in this case. Stable measurements were performed with the same probe at all 11 samples, including GaN films and reference Si and FS specimens. The probe was visualized directly after CR-AFM measurements by SEM (Fig. 12). The tip apex contour can be regarded as flat punch with a diameter of 25±5 nm. The corresponding contact radius ac between the tip and the surface is 13±3 nm. 11 Accurate measured material volume can be determined from the distribution of the stress induced in the film by the tip. For utilized experimental conditions stress decays at the depth over 400 nm. Fig. 12. SEM image of the probe after CR-AFM measurements. Elasticity values calculated from the measured contact resonance are plotted in Fig. 13. In the whole, indentation moduli are quite low and correspond to the values typical for bulk GaN (E=181 GPa). The typical measurement inaccuracy is in the range of 5-10%. According to the stress distribution the account of the substrate contribution into the measured indentation moduli is important. But for the thick films measured elasticity values are free of the substrate influence within the inaccuracy. There is no qualitative difference between films prepared with the substrate temperatures 700°C and 730°C. At the same time, the elasticity of the films which were produced with the substrate heated to 630°C is significantly lower. This could be explained by the presence of high amount of defects inside the films and smaller crystallites. Fig. 13. Indentation moduli of GaN films measured at an applied load of 300 nN. A thickness dependence of the measured indentation moduli could be attributed to the influence of the stiff SiC substrate. The percentage contributions into the indentation moduli of the GaN films with different thicknesses are summarized in Table 1. The elasticity of thick films could be measured directly, while for the thinner ones it is necessary to take into account the influence of the substrate. Young’s modulus for the SiC substrate was measured also by CR-AFM to be 430±40 GPa. This means that the measured values of indentation moduli are high due to the influence of the stiff material beneath the GaN film. 12 Table 1. Portion of the response from the GaN films with different thicknesses when measuring indentation moduli. A tip with the shape of a flat punch and a diameter of 25 nm is assumed. Thickness, nm 31 43 51 95 100 230 238 250 Portion, % 65.4 74.9 79.0 89.9 90.6 97.7 97.9 98.1 Sub-stoichiometric nitrogen ion implanted silicon In general, keV ion implantation is accompanied by sputter erosion that leads to surface roughening due to its stochastic nature. In the present experiments, the surface topography of 100 keV N-ion implanted Si surfaces was studied by using AFM before and after thermal annealing (Fig. 14). From the lack of roughening we surmise that the surface damage/erosion due to nitrogen ion implantation and subsequent annealing was negligible. Fig. 14. AFM surface topographs of the Si samples after the last annealing step; the calculated root mean square roughnesses (RMS) are also given. According to the SRIM simulation and calibrated SIMS measurements the implanted layer extends up to a depth of about 400 nm with an N concentration maximum at 270 nm. The achieved nitrogen concentrations were significantly below a level of 5.8 × 1022 cm-3 (or 53.7 at.% N) that is necessary for stoichiometric Si 3 N 4 . Since the annealing temperature was chosen relatively low, no diffusion of atoms inside the silicon nitride layer is expected. The total nitrogen content did not change, but some surface oxide layer was formed during the last annealing procedures. XRD measurement indicates existence of crystallites with different crystal orientations. 13 st Fig. 15. Secondary ion mass spectrometer (SIMS) depth profiles for the Si samples after the 1 annealing step (solid lines) and ion depth profiles obtained by the SRIM modeling (dashed lines). To estimate the material volume measured by CR-AFM, the nanometer–resolved SEM was performed for visualization of the tip apex of the every individual probe. Significant blunting was observed for all used probes. The contours of the tip apexes can be regarded as flat punches with a diameter of several tens of nm. Indentation moduli as a function of the annealing duration are plotted in Fig. 16. The level of the indentation moduli for crystalline Si is shown for comparison. As-implanted samples are softer (stiffness between 110 and 120 GPa) than the original crystalline silicon material because of the implantation induced amorphization. It is also shown, that the stiffness decreases with increasing ion fluence. An annealing procedure for 0.5 h at 800°C partially recovers defects in the damaged material layer. Consequently, the indentation modulus increases in all implanted samples, where the samples with low-fluence implantation increased up to the value, close to the initial level of stiffness (165 GPa for crystalline silicon). In contrast, the high-fluence implanted samples required 1 h of annealing to obtain this recovering level. Moreover, the high-fluence implanted and annealed samples exhibit higher stiffness than crystalline silicon and the indentation moduli reached values up to 180 GPa. This result indicates that some nitrogen-silicon bonds can be formed after implantation with fluences > 1 × 1016 N ions cm--2 and annealing. Fig. 16. Calculated indentation moduli as a function of annealing time. Also shown is an exemplary SEM micrograph of a blunted tip apex. 14 XPS measurements allowed the detection of Si-N bonds inside the implanted layers. Detailed analysis was performed for the sample with the highest nitrogen ion fluence. The nitrogen peak N 1s at 397.2 eV indicates the presence of Si-N chemical bonds in the material. The total atomic content of the N atoms bonded to Si atoms at the distribution maximum was found to be 10% which is almost equal to the total atomic content of nitrogen in the maximum (see Fig. 15). Si 2p doublets with different energy as result of the curve fitting procedure are indicative for Si atoms connected to N atoms. The binding energy of 102.4 eV is attributed to bonding of Si with four N atoms (as in stoichiometric silicon nitride) and the doublet peak at an energy of 100.7 eV characterizes Si atoms connected to less than four N atoms (Fig. 17). At the same time, the concentration of the Si atoms with four Si-N bonds was less than 1% that does not exceed the noise error of the measurements. Therefore, all N atoms were connected to Si atoms, but they did not form stoichiometric silicon nitride. Returning to the long term annealing, the observed subsequent abrupt decrease of the indentation moduli down to values between 100 and 130 GPa can be caused by strong oxidation during the 3rd annealing step. The thickness of the oxide layer for implanted and annealed sample was measured with high resolution by XPS. Thick oxide layer was detected at the sample surfaces with a thickness between 30 and 60 nm. This indicates that oxidation has a strong influence on the stiffness values obtained by CR-AFM, because near-surface oxide layers give a substantial contribution to the measured values. Fig. 17. XPS spectra of the Si sample with the highest nitrogen ion fluence after the last annealing step (signal acquired at the distribution maximum). IOM – work group Prof. Mayr CR-AFM For the quantitative determination of the elastic properties, we did some multireference sample studies [IOM-1, IOM-3] including finite element analysis of the contact model adjusting shape-index n and tip modulus Mtip. The FEM simulations led to good predictions of the dispersion relations of the first two eigenmodes. From this we could conclude that the small tip modulus, which was already determined similar by others, can be ascribed essentially to physical processes and the applied contact model and has hardly something to do with the beam bending model [IOM-3]. Further FEM simulations utilizing a conical indenter with experimentally measured values of the contact radius and opening angle and literature values of the elasticity tensor led to a good prediction of the tip 15 modulus. It has been shown that the tip modulus is not just a function of the elastic constants, but also of the opening angle and converges to the indentation modulus for big opening angles (Fig. 18). Fig. 18. a) Sideview of two DLC coated cantilever tips b) radialsymetric FEM-model of the simulated tip and related pressure distribution c) Relative behavior of the tip modulus related to the indentation modulus of Si for different opening angles an tips (from [IOM-3]). Thin films Besides the question of the low tip modulus it was not clear up to now how surface elasticity caused by surface stresses affects the contact stiffness as well as the indentation modulus especially for such a surface sensitive techniques like CR-AFM. Therefore, we simulated ultra-thin films of different film thicknesses, of just a few lattice constants for two materials (silicon and strontiumtitanate). Within the approach of continuum theory, the reduction of the elastic constants of a thin film caused by implemented surface stresses as a function of its film thickness L can be described by 𝐶𝑖𝑗𝑘𝑙 (L) = C𝑏𝑢𝑙𝑘 𝑖𝑗𝑘𝑙 + 2 𝑑𝑖𝑗𝑘𝑙 L because the force which is acting on a volume element is shrinking with increasing film thickness. a) b) Fig. 19. Elastic constants a) C xxxx and C zzzz and b) C xxyy and C xxzz, for Si. 16 a) b) Fig. 20. Elastic Constants a) C yzyz and C xyxy for Si and b) C xxxx and C zzzz for SrTiO 3 . Within the DFT simulations we were able to reproduce this qualitative behavior especially in the diagonal elements of the elasticity tensor (Fig.: 19 a), 20 a), 20 b) and 21 a)) while the non-diagonal elements vary widely (Fig. 19 b) and 21 a)). Greater deviations from that law are only observed for the smallest thicknesses, i.e. for silicon, one lattice constant and three lattice constants for strontiumtitanate. These deviations are no properties of the single surface, but are explained by the interactions of both surfaces. From the DFT calculated surface elastic constants d ijkl , we were able to build up a continuum model, where the surface elasticity is implemented by layers of the same film thickness differing (reduced) elastic constants. Further FEM simulations using different tip radii/forces which led to different contact radii gave us an estimation from which contact radii, surface elasticity becomes important and has to be taken into account. In Fig. 21 b) we see that a significant change for both materials is observed already at a contact radius rc of 5 nm. From the simulation results we were able to devise an analytical model that describes the indentation modulus as a function of its contact radius assuming each layer can be described by an effective indentation modulus Mi=M inf -c/ai, where c denotes the reduction, i the number of layers and a the thickness of each layer. Assuming further that the stress field increases along each direction linearly with the contact radius and each layer contributes equally, we get the following expression for the indentation modulus as function of the number of contributing layers (which is proportional to the contact radius r c ): 𝑛 1 1 𝑐� 𝑛 𝑖=1 𝑀𝑖𝑛𝑓 − 𝑎𝑖 𝑀(𝑛) = �� −1 =� −1 𝑛 1 𝑐 1 + 2 � 𝑐 � 𝑀𝑖𝑛𝑓 𝑀𝑖𝑛𝑓 𝑎𝑛 𝑖=1 𝑖 − 𝑎𝑀𝑖𝑛𝑓 at which the last term in the denominator can be approximated by a harmonic series which we represent by is asymptotic expansion: −1 𝑐 1 1 + 2 �𝑙𝑛(𝑛) + 𝛾 + +. . . �� 𝑀(𝑛) ≈ � 2𝑛 𝑀𝑖𝑛𝑓 𝑀𝑖𝑛𝑓 𝑎𝑛 where γ is the Euler-Mascheroni number. In Fig. 21 b), we fitted the simulation data by the expansion up to the third term of the expansion taking c as a free fitting parameter. 17 a) b) Fig. 21. a) Elastic Constants C xxyy C xxzz, C yzyz and C xyxy for SrTiO 3 b) FEM-Simulation of rel. Indentationsmoduls as a function of Contact radius r c . Nano structures Another interesting point is the dissolvability of smallest nanostructures at surfaces using CR-AFM. There we can find similar effects as with thin films, where additional stresses change the elastic behavior. For the investigation of the local elastic properties, we analyzed the stress fluctuations of the surface around a step of a few lattice constants height (Fig. 22). Fig. 22. yy- component of Stress fluctuation (multiplied by the quadratic Wigner-Seitz volume) of the upper Ga atoms along y-axis. The stress fluctuations can be connected to elastic constants using the fluctuation dissipation theorem: 𝐵𝑜𝑟𝑛 �− 𝐶 𝑙𝑜𝑐 = �𝐶𝑖𝑗𝑘𝑙 2𝑁𝑘𝑏 𝑇 𝑉 ��𝜎𝑖𝑗 𝜎𝑘𝑙 � − �𝜎𝑖𝑗 �⟨𝜎𝑘𝑙 ⟩� + �𝛿𝑖𝑘 𝛿𝑗𝑙 − 𝛿𝑖𝑙 𝛿𝑗𝑘 � 𝑘𝑏 𝑇 𝑉 At which the first term (Bornterm) only differs for atoms with different coordination numbers and the last term (kinetic term) can be neglected for small temperatures. The stress fluctuations were compared to a local defined indentation modulus (Fig. 23 a)) which is extracted from the forces acting on a specific atom and the two-dimensional Wigner-Seitz volume, which one get from the projection of the voronoi cells of the surface atoms to the surface plane. The same simulation was 18 also done for a bigger contact radius, realized force acting on all surface atoms within a certain radius (Fig. 23 b)). Here we see an effective softening, when the indenter is approaching the step edge (y=9). To measure this effect one would need a lateral resolution of < 1 nm. At the bottom of the step edge, we don't see this effect, although one would expect that surface stresses also affect the mechanical properties at the bottom. This can be understood by comparing at the local modulus which is more sensitive to surface effects (Fig. 23 a)) and FEM-simulations (Fig. 24), where surface stresses are neglected. a) b) 2 Fig. 23. MD-Simulation of the local indentation modulus with contact area a) A C =0.09 nm and b) A C =1.23 nm² on top and below the GaN step edge. The local indentation modulus is notably sensitive to local stresses, as the effective contact area is minimal. Therefore it agrees pretty good with the stress fluctuation, so it decreases both at the bottom and on top of the step edge, while the FEM simulations show a clear dependence of the step height, and even an increase of the elastic constants at the bottom of the step (Fig. 24). From these results one can conclude that both effects cancel in the MD simulation of the flat punch indentation with the bigger radius at the bottom of the step. Also on top we have an overlap of both effects, where the geometrical effect dominates the stress effect, which is seen by the range, but also by the difference between different step heights. 19 2 Fig. 24. FEM simulation of the indentation modulus (A C =1.23 nm ) on top and below the GaN step edge (without consideration of surface stresses.) 4. Statement about commercialization Application demonstrators for future magnetoelectric memory and sensor devices are possible in case of further successful work on the multilayers according to section (c) above. 5. Cooperation partners Katholieke Universiteit (KU) Leuven, Belgien, Instituut voor Kern- en Stralingsfysica, Celestijnenlaan 200D, B-3001 Leuven, Dr. Vera Lazenka, Prof. André Vantomme, Prof. Kristiaan Temst (measurement of ME coefficients); Prof. Dr. H. Karl, Institute of Physics, University of Augsburg (nitrogen implantation). 6. Graduation thesis PhD thesis Peter Schwinkendorf, submission expected for 2015 PhD thesis Alexander Melvin Jakob (2014) PhD thesis Jörg Buchwald, submission expected for 2015 PhD thesis Marina Sarmanova, submission expected for 2014 7. List of publications (bold with direct relation to the SAW project) [UL1] M. Lorenz, M S Ramachandra Rao (guest eds.), Special issue "25 years of pulsed laser deposition" J. Phys. D: Appl. Phys. 47, 030301 (2014) [UL2] M. Lorenz, V. Lazenka, P. Schwinkendorf, F. Bern, M. Ziese, H. Modarresi, A. Volodin, M. J Van Bael, K. Temst, A. Vantomme and M. Grundmann, Multiferroic BaTiO3–BiFeO3 composite thin films and multilayers: strain engineering and magnetoelectric coupling, J. Phys. D: Appl. Phys. 47, 135303 (2014) [UL3] M. Lorenz, A. de Pablos-Martin, C. Patzig, M. Stölzel, K. Brachwitz, H. Hochmuth, M. Grundmann and T. Höche, Highly textured fresnoite thin films synthesized in situ by pulsed laser deposition with CO2 laser direct heating, J. Phys. D: Appl. Phys. 47, 034013 (2014) [UL4] M. Jenderka, J. Barzola-Quiquia, Z. Zhang, H. Frenzel, M. Grundmann, and M. Lorenz, Mott variable-range hopping and weak antilocalization effect in heteroepitaxial Na2IrO3 thin films, Phys. Rev. B 88, 045111 (2013) [UL5] K. Brachwitz, T. Böntgen, M. Lorenz, and M. Grundmann, On the transition point of thermally activated conduction of spinel-type MFe2O4 ferrite thin films (M=Zn, Co, Ni), Appl. Phys. Lett. 102, 172104 (2013) [UL6] M. Stölzel, A. Müller, G. Benndorf, M. Brandt, M. Lorenz, and M. Grundmann, Determination of unscreened exciton states in polar ZnO/(Mg,Zn)O quantum wells with strong quantum-confined Stark effect, Phys. Rev. B 88, 045315 (2013) [UL7] M. Grundmann, T. Böntgen, M. Lorenz, Occurrence of Rotation Domains in Heteroepitaxy, Phys. Rev. Lett. 105, 146102 (2010). 20 [UL8] M. Lorenz, Pulsed Laser Deposition of ZnO-based Thin Films, chapter 7 in: K. Ellmer, A. Klein, B. Rech (eds.), "Transparent Conductive Zinc Oxide. Basics and Applications in Thin Film Solar Cells" (Springer Series in Materials Science Vol. 104, Berlin, 2008) p. 303-358. [UL9] E. M. Kaidashev, M. Lorenz, H. von Wenckstern, A. Rahm, H.-C. Semmelhack, K.-H. Han, G. Benndorf, C. Bundesmann, H. Hochmuth, M. Grundmann, High electron mobility of epitaxial ZnO thin films on c-plane sapphire grown by multistep pulsed-laser deposition, Appl. Phys. Lett. 82, 3901 (2003). [UL10] M. Lorenz, H. Hochmuth, D. Natusch, H. Börner, K. Kreher, W. Schmitz, Large-area double-side pulsed laser deposition of YBCO thin films on 3-inch sapphire wafers, Appl. Phys. Lett. 68, 3332 – 34 (1996). [UL11] http://www.uni-leipzig.de/~grundm/publ.php [UL12] Thomas Nobis, Evgeni M. Kaidashev, Andreas Rahm, Michael Lorenz, Marius Grundmann Whispering gallery modes in nano-sized dielectric resonators with hexagonal cross section Phys. Rev. Lett. 93, 103903 (4 pages) (2004) [UL13] M. Lorenz, E.M. Kaidashev, A. Rahm, Th. Nobis, J. Lenzner, G. Wagner, D. Spemann, H. Hochmuth, M. Grundmann, MgxZn1-xO (0≤x<0.2) nanowire arrays on sapphire grown by highpressure pulsed-laser deposition, Appl. Phys. Lett. 86, 143113 (3 pages) (2005) [UL14] M. Grundmann, Architecture of nano- and microdimensional building blocks, phys. stat. sol. (b) 247, 1257-1264 (2010) [UL15] M. Lorenz, A. Rahm, B. Cao, J. Zúñiga-Pérez, E.M. Kaidashev, N. Zhakarov, G. Wagner, T. Nobis, C. Czekalla, G. Zimmermann, M. Grundmann, Self-organized growth of ZnO-based nano- and microstructures, phys. stat. sol. (b) 247, 1265-1281 (2010) [UL16] C. Czekalla, T. Nobis, A. Rahm, B. Cao, J. Zúñiga-Pérez, C. Sturm, R. Schmidt-Grund, M. Lorenz, M. Grundmann, Whispering gallery modes in ZnO nano- and microwires phys. stat. sol. (b) 247, 12821293 (2010). [UL17] M. Lorenz, G. Wagner, V. Lazenka, P. Schwinkendorf, H. Modarresi, M. J. Van Bael, A. Vantomme, K. Temst, O. Oeckler, M. Grundmann, Correlation of magnetoelectric coupling in multiferroic BaTiO3-BiFeO3 superlattices with oxygen vacancies and antiphase octahedral rotations, revision submitted to Appl. Phys. Lett (2014). [IOM-1] A. M. Jakob, M. Müller, B. Rauschenbach and S. G. Mayr, Nanoscale mechanical surface properties of single crystalline martensitic Ni–Mn–Ga ferromagnetic shape memory alloys, New J. Phys. 14 (2012) 033029. [IOM-2] A. M. Jakob, M. Hennes, M. Müller, D. Spemann, S. G. Mayr, Coupling of Micromagnetic and Structural Properties Across the Martensite and Curie Temperatures in Miniaturized Ni-Mn-Ga Ferromagnetic Shape Memory Alloys , Adv. Func. Mat. 23 (2013) 4694–4702 21 [IOM-3] A.M. Jakob, J. Buchwald, B. Rauschenbach and S. G. Mayr, Nanoscale-resolved elasticity: contact mechanics for quantitative contact resonance atomic force microscopy, Nanoscale, 6 (2014) 6898-6910 [IOM-5] J. Buchwald and S.G. Mayr, Influence of surface stresses on indentation response, submitted [IOM-6] M. Sarmanova, H. Karl, S. Mändl, D. Hirsch, S.G. Mayr and B. Rauschenbach, Elastic properties of sub-stoiciometric nitrogen ion implanted silicon, submitted 22