UHV-SEM studies of surface processes: Recent progress

Ultramicroscopy North-Holland 149 zyxwvutsrq 11 (1983) 1499 156 Publishing Company UHV-SEM STUDIES J.A. VENABLES, OF SURFACE G.D.T. SPILLER PROCESSES: *, D.J. FATHERS RECENT PROGRESS **, C.J. HARLAND and M. HANBUCKEN zyxwvutsrqp School of M athematical and Phy sical Scrences, University of Sussex, Brighton. Sussex BNI 9Q H, UK Received 3 February 1983; presented at Workshop January 1983 Surface processes on bulk samples have been studied using an ultra high vacuum SEM, equipped with a variety of surface-sensitive techniques. Recent studies of Stranski-Krastanov growth systems and the uses of a recently installed computerised data acquisition system are highlighted. 1. Introduction 2. Techniques available Recently, reviews have been given of the field which is emerging at the interface between electron microscopy and surface science [l-4], the subject of the Workshop on Surface Imaging and Analysis with High Spatial Resolution. At the University of Sussex, our approach has been to use a field-emission gun scanning electron microscope, with a UHV specimen chamber, to study surface processes on a microscopic scale down to around 10 nm resolution. Several related sub-micron techniques have also been developed in the course of this work; these techniques are referenced in section 2. A long-term interest is in adsorption and crystal growth processes on surfaces, and recent results on the layer plus island or Stranski-Krastanov growth mode are highlighted in section 3. A computerised data collection and processing system has been added to this microscope. The actual and potential advantages of such a system for quantitative (dynamic) studies of surface processes is discussed briefly in section 4. The microscope configuration is described in detail elsewhere [5]. Briefly, a field-emission gun plus magnetic lens column delivers an intense spot at l-60 (typically 30) kV onto a bulk sample, which is in a standard stainless steel UHV chamber capable of 2 X lo-” Torr. By adding a magnetic gun lens [6] to the two-lens column [7], the current available has been increased to > 1 PA into 1 pm, with - 0.1 PA into 100 nm and 10 nA into 10 nm [8]. With such currents available, secondary electron videotape recordings can readily be made of dynamic surface processes, such as crystal growth, melting and changes in morphology induced by adsorption [9]. Care must of course be taken to choose systems for study which are not unduly perturbed by the electron beam; in any case intermittent, or before-after, examination of the same area is always possible, and is the normal mode of operation, These current levels are also useful for various forms of microanalysis. In this instrument, we have available a cylindrical mirror analyser (CMA) for Auger electron spectroscopy (AES) [S]; this analyser can also be used to study work function variations (A+) due to adsorbed layers, and diffusion in these layers [lo]. The sample, which is typically a thin plate of dimensions 0.3- 1 mm thick x 5 x (5-10) mm, can * Present address: British Telecom Research Laboratories, Martlesham Heath, Ipswich. Suffolk IP5 7RE, UK. ** Present address: VG Microscopes Ltd., Charlwoods Road, East Grinstead, Sussex RH19 254, UK. 0304-3991/83/0000-0000/$03.00 0 1983 North-Holland 150 J.A. Venahles er ul. / UH V- SEM srudres be processed by standard surface science methods, including heating resistively or by electron bombardment as appropriate, ion bombardment, and in-situ deposition. It can be tilted around its long axis and at glancing angle can be examined by RHEED. An additional crystallographic technique developed is the electron back-scattering pattern (EBSP). This technique is especialy useful for studies of epitaxial relationships, achieving &-0.5” accuracy from < 100 nm sized areas [ 111. of surfucep.~oc~esses The results of a large number of island density measurements for Ag/W(llO), N( R,T) give i in the range 7-10 and E, = 0.50 f 0.03 eV. These results indicate that the small Ag clusters on the surface are relatively unstable and that the Ag binding forces are considerably anisotropic, being much stronger perpendicular to the surface than within the surface layer. Somewhat wavy terraced W(l IO) surfaces are observed after cleaning and annealing, and under certain circumstances nucleation and growth takes place preferentially at terrace edges, as seen in figs. lb-ld. In these cases the Ag crystals tend to be 3. zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA Stranski-Krastanov growth studies elongated along the steps, presumably reflecting anisotropic surface diffusion induced by them. The Of the three modes of crystal growth on surfaces, shapes of the Ag islands are far from the equithe layer plus island, or Stranski-Krastanov, mode librium form, being much flatter and quite irreguis the most intriguing and least understood. We lar. However, the epitaxial orientation as dehave investigated some crystal growth combinatermined from EBSP’s always has the (111) Ag tions which follow this mode using our UHV-SEM face parallel to W( 110) within 0.5”, although there techniques, in order to explore the nucleation and is a small range of azimuthal orientations [I I]. growth mechanisms. These studies have been put Since all Ag atoms above the 2 ML end up in the in the context of other nucleation and growth islands, surface diffusion is extensive. The reason work on the island and layer growth modes in why the islands are not thicker is primarily due to recent review articles [ 121. the difficulty of incorporating adatoms into perFour systems based on silver deposition have fect (111) faces. A model of these effects is given been studied: Ag/Mo(lOO) [ 131, Ag/Si( 111) [ 141, elsewhere [16]. We have also observed decorated Si(100) [20] and W(110) [15], the last having been sub-grain boundaries where the crystal shapes are studied in most detail. In all cases islands form more pyramidal presumably due to the presence of after an intermediate layer is grown on the subone or more threading dislocations. strate, there being 2 ML in the W(110) case. Very similar temperature dependences N(T) Representative micrographs are seen in fig. 1. The were previously observed for Ag/Mo( 100) and layers have been investigated by AES, following Si( 11 I), with E, larger than for W(110) and i Bauer and coworkers, and correlated with the SEM probably larger also. The intermediate layers are observations of islands. In all cases the island quite different in the three cases, but the N(T) density N decreases very strongly, within the comranges are similar, ranging from < lo6 at 500°C to plete condensation regime, as the substrate tem> IO’-10” cm-* at 15O’C [12]. The Mo(100) perature (T) during deposition is raised from typisurface was difficult to anneal to produce comcally 200 to 550°C and increases with deposition pletely flat surfaces, and decoration effects rerate R in the form sulted [ 131. N - RP exp( EJkT). Similar but rather more drastic effects have been seen for the Ag/Si( 100) system [20]. Heating From a suitably developed nucleation theory [ 121 an even slightly carbon-contaminated Si( 100) (2 x the exponent of the power law p is related to the 1) surface produces pits which grow to substancritical nucleus size i, by p = i/(i + 2), and the tially cover the surface on repeated heating to activation energy E, to the cluster binding energy 1100°C. Cleaning at 800°C proved to be preferE, and surface diffusion energy Ed, by able. After deposition of Ag the production of E,=(E,+iE,)/(i+2). holes is markedly accelerated and is the reason J. A. Venahles et al. / UH V- SEM studies of surface processes Fig. 1. Representative observation measured R = 0.06 ML/min, 0 islands on large steps steps in the substrate 151 conditions (8 is the angle of SEM pictures showing the morphology of 4g islands on W (I 10). Deposition from the substrate plane): (a) T = SOO”C, R = 0.3 ML/min, 0 = 15”, islands on “flat” surface; (b) T= 300°C. = lo”, showing islands following small steps in the substrate; (c) T = 500°C. R = 0.3 ML/min, %= 15”, showing towards the edge of the sample: (d) T= 500°C. R = 1 ML/min. 6’ = 40’. showing a variety of island shapes with faintly visible in the background. why substrates could not be indefinitely reused in previous LEED-AES studies [17]. The early stage of growth on clean surfaces is shown in fig. 2, where the SEM picture (fig. 2a) is shown with the micro-Auger spectra (fig. 2b) taken both on and between the islands. About 0.5 ML Ag is contained between the islands for depositions up to at least 10 ML. The RHEED pattern shows the Si( 100) 2 x 1 pattern (fig.2c), with a weak c(4 x 2) pattern sometimes resulting from an 800°C anneal, followed by Ag deposition and removal. Un- like the case of Ag/Si( 111) [14]. the RHEED pattern does not change much with Ag deposition, confirming previous LEED studies [ 171. Even non-visible hole formation has an influence on the island density N(T). In cases where holes subsequently appeared the density was several times higher than the relatively “good surface” case illustrated in fig. 2a, where N = lo6 cm-’ at - 400°C. Multiple Auger spectra, such as fig. 2b, and images can be acquired and processed on a re- 152 J.A. Venables et rrl. / UH V- SEM studies of surfuce processes 51 92 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA b __- .A.___- - __.~__- - - _~ 0 IS00 1000 so0 ELECTRON ENERGY 2000 <EV> Fig. 2. (a) SEM picture of Ag islands on Si(lOO), T = 4OO”C, 5 ML deposit viewed at B = 45”; island density = lo6 cm-‘; (b) Auger spectra of 10 ML deposit at = 400°C, both on and between the islands; (c) RHEED pattern of 0.25 ML deposit showing Si( 100)2 zyxwvutsrqp x 1 pattern. cently installed digital data collection and analysis system, which will be described in more detail elsewhere [ 181. An example using the Ag/Si( 100) islands is shown in fig.3, which also indicates the simple program structure adopted. The picture is retrieved from disk using program RGBRTF in J.A. Venahles et al. / UHV- fig3a, and the particles are counted in fig.3b using program PCOUNT. Further extensions to particle position and size distributions, plus quantitative spectrum analysis, is envisaged. With this system driving the CMA signal, averaged energy-analysed 153 SEM studies zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPO of surface processes pictures can be taken, strutted from 3 pictures Although such energysubstantial background graphic contrast, camp or “Auger” pictt i at neighbouring selected pictures and show mark larisons between Fig. 3. (a) Ag/Si(lOO) island picture displayed on (256)’ x R-bit frame store. (b) Particle counting routine identified particles in black. The inserts in (a) and (b) show the simple terminal commands involved. lres conenergies. contain ed topodifferent on these islands, showing 154 J.A. Venubles et nl. / UH V-SEM energies can be instructive, as shown in fig.4 the case of Pb/W( 110). The Pb/W(llO) system also shows islands expected from previous AES work [19], where intermediate layer was reported as 1 ML thick. our case the islands were prepared by depositing for as the In a studm ofsurf&ce jm~esses thicker layer followed by melting. Fig.4 shows (256)2 x &bit digitally acquired energy-selected pictures of a slightly terraced region of the substrate of which a part was shadowed from the Pb beam by a small obstacle. There is marked contrast across the terraces which indicates discrete Fig. 4. Energy-selected pictures acquired, and displayed at (256)2 x8-bit resolution for Pb/W(I IO): The analysis energies are indicated in the various cases. All images are signal averaged over 64 sweeps (frames) with the exception of the first and last (SEM) image; 1 sweep = 21 s. 30 kV, 7-10 nA. J. A. Vettables et al. / UH V- SEM studies variations in Pb thickness, and this is correlated with the low energy secondary electron peak at 16 V. Contamination by C may have occurred in the shadowed region where W is exposed; none of these features are visible in the SEM image. Such pictures can usefully be supplemented by Auger spectra at particular points of interest for more positive identification. ofsurface zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQ processes 155 also thanks the Deutsche for financial support. Forschungsgemeinschaft References 7 (198 1) 8 1. [‘I J.A. Venables, Ultramicroscopy PI J.A. Venables, in: Chemistry and Physics of Solid Surfaces IV, Eds. R. Vanselow and R. Howe (Springer, Berlin, 1982) p. 123. and G. Honjo, in: Crystals, [31 K. Yagi, K. Takayanagi Growth, Properties and Applications, Vol. 7 (Springer, 4. zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA Conclusions Berlin, 1982) p. 47. (Institute of Physics, 141 A. Howie, in: Proc. EMAG’El The studies described in section 3 give interestLondon, 1981) p. 419. ing and rather direct information about adsorp[51 J.A. Venables, A.P. Janssen. P. Akhter, J. Derrien and C.J. tion, diffusion and binding processes at surfaces in Harland, J. Microscopy 118 (1980) 351. metal-metal and metal-semiconductor systems [61 J.A. Venables and G.D. Archer, in: Electron Microscopy 1980, The Hague (7th European Congr. on Electron Miwhich are not electron beam-sensitive. Extension croscopy Foundation, Leiden, 1980) p. 54. to more beam-sensitive systems such as gas chem5 (1980) [71 J.A. Venables and A.P. Janssen, Ultramicroscopy isorption requires the use of minimum dose tech297. niques. PI J.A. Venables and D.J. Fathers, in: Proc. 10th Intern. This is one of the many reasons why a flexibly Congr. on Electron Microscopy, Hamburg, 1982, Vol. 1, p. 181. organised computer-based data collection and [91 J.J. Mktois, G.D.T. Spiller and J.A. Venables. Phil. Mag. analysis system has been added to the UHV-SEM. A46 (1982) 1015. With this system we can collect a series of SEM [lOI A.P. Janssen, P. Akhter, C.J. Harland and J.A. Venables, and energy-selected pictures, spectra from specific Surface Sci. 93 (1980) 453; P. Akhter and J.A. Venables, Surface Sci. 103 (1981) 301. points on the image, and if necessary diffraction [Ill C.J. Harland, P. Akhter and J.A. Venables, J. Phys. El4 patterns; coupled with electron beam-blanking, the (1981) 175. latest pictures and spectra can be continuously [I21 J.A. Venables and G.D.T. Spiller, Nucleation and Growth viewed and updated while the reaction is proceedof Thin Films, in: Surface Mobilities on Solid Materials, ing very largely in the absence of the influence of Ed. Vu Thien Binh (Plenum-NATO AS1 Series, 1982); the beam. J.A. Venables and G.D.T. Spiller, Rept. Progr. Phys., in press. Reactions we hope to study in the future in[I31 K. Hartig, A.P. Janssen and J.A. Venables, Surface Sci. 74 clude surface diffusion and interdiffusion, and ad(1978) 69. sorption-induced facetting. Preliminary work indi[141 J.A. Venables, J. Derrien and A.P. Janssen, Surface Sci. 95 cating the feasibility of such studies is given in (1980) 411. refs. [9] and [lo]. An approach to computerisation [I51 G.D.T. Spiller, P. Akhter and J.A. Venables, Surface Sci., submitted. is described in ref. [ 181. and J.A. Venables, J. Crystal Growth, 1161 J.M. Bermond submitted. and H. Neddermeyer, Surface Sci. 114 [I71 M. Hanbiicken Acknowledgements (1982) 563. [I81 D.J. Fathers, C.J. Harland and J.A. Venables, to be submitted. We thank the organisers for the opportunity for E. Bauer, H. Poppa and G. Todd, Thin Solid Films 28 I191 J.A. Venables to take part in the workshop. We (1975) 19. thank the SERC for continuing to support this to be PO1 M. Hanbiicken, J.A. Venables and H. Neddermeyer, work, and for salary support for G.D.T. Spiller, submitted. D.J. Fathers and M. Hanbticken. M. Hanbticken