RHEED Investigations of Copper Deposition on Gold in the Underpotential Region HEINZ GERISCHER BY HANS0.BECKMANN DIETERM. KOLB*AND GUNTER LEHMPFUHL Fritz-Haber-Institut der Max-Planck-Gesellschaft Faradayweg 4-6 1 Berlin 33 W. Germany Received 31st August 1977 The underpotential deposition of a Cu monolayer on single crystal gold electrodes of (111) and (100) orientation is investigated by cyclic voltammetry and after removal of the electrode from the electrochemical cell and subsequent transfer into a vacuum chamber by reflected high energy elec- tron diffraction (RHEED). It is found that underpotential deposition of about 2/3 of a Cu mono- layer causes a (d3'x 4%R30 superstructure to appear in RHEED. After stripping a faint superstructure of the same type is still visible in the RHEED pattern.Deposition and stripping of bulk Cu also rearranges the Au electrode surface in a characteristic way producing an additional (2 x 2) superstructure. No such structure is seen when the electrode potential is cycled only in the double layer charging region of Au and in the oxygen adsorption region. The observed RHEED patterns are discussed in terms of surface rearrangements during Cu deposition due to the formation of ordered surface and bulk alloys. The underpotential deposition of metals on foreign metal substrates that is to say the deposition of metal atoms from their ions in solution up to a monolayer at elec- trode potentials positive to the respective Nernst potential is often the initial step in metal deposition reactions.This monolayer formation not only represents an in- teresting case of chemisorption on metal surfaces but is also very important for the further growth of the deposit. Numerous electrochemical experiments have been performed in recent years to study adsorption isotherms adsorption-desorption kinetics and relative binding energies. More recently measurements on single crystal surfaces have revealed more detail in the adsorption i~otherms,~-~ which thus allow some conclusions to be drawn on the structure of the adsorbate layer and its depend- ence on the electrode surface struct~re.~?~ However none of these in-situ electro-chemical experiments give a direct insight at a molecular level such as the electronic properties of the adsorbate and its geometric arrangement.This information seems only to be accessible by photoemission and electron scattering experiments such as UPS XPS Auger LEED and RHEED. The severe drawback of these highly surface- sensitive methods for electrochemical systems is their need for a vacuum which means that the electrode surfaces cannot be looked at under real electrochemical operating conditions. On the other hand underpotentially formed metal monolayers are certainly among the most promising systems to be studied under vacuum conditions because their strong bonding to the substrate guarantees that the character of the adsorbate is not changed significantly by removal from the electrochemical environ- ment. First experimental results with XPS on the electronic structure of under- potential deposits seem to support this view.6* Up to now however the metal * To whom correspondence should be addressed.RHEED INVESTIGATIONS OF COPPER DEPOSITION ON GOLD adsorbates have not been investigated by structure sensitive methods like LEED or RHEED which also allow one to test the quality of single crystal surfaces used in electrochemical experiments and its influence on adsorbtion behaviour.8 One of the open problems in underpotential deposition of metal monolayers which still has to be solved is the question of whether or not the adsorbate forms regular arrays at certain coverages. To answer this question the adsorbate covered electrode has to be transferred from the electrochemical cell into a high vacuum chamber and investigated by electron scattering techniques.Although LEED seemed to be the obvious choice we considered RHEED as more promising for the following reason. Since LEED requires an ultra-high vacuum system the experimental efforts would be substantially higher than with RHEED which can be performed at Torr. For LEED the cleaning of the electrode surface plays a crucial role and usually no LEED pattern is observed before the surface is ion-bombarded or heated. Such a treatment however would inevitably destroy the monolayer. No such problems are usually encountered with RHEED because of the much higher electron energy used while this technique is still highly surface sensitive by virtue of working at grazing angles of incidence. EXPERIMENTAL Gold single crystal discs were cut by spark erosion from melt-grown single crystal rods after orientation by Laue back scattering.The discs of -2 mm thickness and 10 mm in diameter with (111) or (100) surfaces were mechanically polished with abrasives of succes- sively finer grades electrochemically polished in a cyanide bath9 for 10-50 min at current densities up to 2 A cm-2 and annealed at 850 "Cfor 4 11in ahelium atmosphere. The quality of the single crystal surface was finally tested by RHEED. The electrochemical experiments were performed in a standard type electrochemical cell with separated compartments for the calomel reference and the Pt counter electrode. For measurements with single crystal electrodes the dipping technique described by Schultze et aL4 was used.The crystals were glued with conducting silver paste onto a special holder which also fitted into the RHEED apparatus. The electrode holder was lowered until the electrode dipped into the electrolyte and then partially withdrawn so that the electrode was above the electrolyte level with only the desired single crystal surface in contact with the electrolyte because of surface tension. The base electrolyte was either 0.5 mol dm-3 HzS04 or 1 mol dm-3 HC104 and Cu2+-ions were added from or 1 mol dm-3 stock solutions of CuS04. All chemicals were of p.a. grade and dissolved in triply distilled water. All potentials are quoted against the saturated calomel electrode (SCE). Reflected electron diffraction experiments'O were conducted in a conventional diffraction camera under ordinary high vacuum conditions of Torr.The crystal was mounted on a special specimen stagell allowing arotation of 360"about an axis perpendicular to the crystal surface and a second rotation of -& 12" about an axis in the crystal surface to tilt the crystal with respect to the direction of the incident electron beam. In order to reduce contamination effects the specimen was surrounded by an anticontamination shield which was cooled down to liquid nitrogen temperature. The diameter of the electron beam was w 70pm at the speci- men. Under these conditions the crystal area contributing to the diffraction pattern was at grazing incidence conditions (angle between crystal surface and incident electron beam in the order of 2") x 70 x 2000 pm.The diffraction pattern was recorded photographically. RESULTS Fig. 1 shows cyclic current potential curves for Au(l11) and Au(100) electrodes in 1 mol dm-3 HClO, containing 5 x low4mol dm-3 CuSO,. The crystallographic orientation of the electrode surfaces and their quality have been tested by RHEED H. 0. BECKMANN H. GERISCHER D. M. KOLB AND G. LEHMPFUHL 53 after surface preparation. In the potential region between 0.0 and 0.6 V the Cu mono- layer is formed or stripped depending on the direction of the potential scan while beyond +1.O V oxygen adsorption on gold takes place the adsorbate being reduced again on the cathodic scan around +0.95 V. With a few minor exceptions these curves resemble closely those reported by Schultze et L~Z.,~ demonstrating the marked influence of the substrate structure on oxygen adsorption and the metal deposition re- action.Tn fig. 2 the potential scale for the Cu monolayer adsorption and desorption 100 l~ I -80 -60 ht -& 40 a 3 + 20 -0 0.5 1 .o 1.5 USCE /v FIG. 1.-Cyclic current-potential curves for Au(l11) (-) and Au(100) (--) in 1 in01 dmP3 HC104+ 5 x mol dm-3 CuS04. Scan rate = 10 mV s-'. has been expanded to reveal more clearly the details in the adsorption/desorption isotherms. Moreover a lower scan rate had been chosen. From the result shown in fig. 2 it is noted that the current against potential curves for adsorption and desorption of Cu on Au(100) indicate a slow but rather reversible process while deposition and stripping of the Cu monolayer on Au(ll1) clearly occurs at somewhat different peak potentials suggesting some type of phase transition.It is interesting to note by com- parison of fig. 1 and 2 that the adsorption peak for Cu on Au(100) is split into two when increasing the scan rate from 1mV s-l to 10 or even 5 mV s-l. Further increase of the scan rate shifts the second peak still more cathodic while the first one remains unaffected at least up to a sweep speed of 100 mV s-l. The charge connected with the various desorption peaks is in good agreement with data reported by Schultze et aL4 The charge for a complete monolayer was found to be 360 ,uC cm-2 for Cu on Au(100) and 440 ,uC cm-2 for Cu on Au(ll1). Both num- bers agree very nicely with those for a (1 x 1) structure on Au and sustain the assump- tion of an epitaxial monolayer.The charges corresponding to the four peaks in the RHEED INVESTIGATIONS OF COPPER DEPOSITION ON GOLD desorption spectrum of Cu on Au(ll1) are 50,220 and 170 pC cmV2 for peak No. 1 2 and 3 and 4together respectively (fig. 2). This means that at a desorption poten- tial close to peak No. 3 the surface is covered roughly by 2/3 of a monolayer. It is also found that the half width of the most cathodic desorption peak for Cu on Au( 1 1 1) (peak No. 4 in fig. 2) is markedly dependent on the scan rate increasing from S112= 8 6 4 e4 '2 5 4 3 Go -2 -2 -6 0 0.2 0.4'' 0.6 0.8 'SCE FIG.2.-Cyclic current-potential curves for Au(l11) (-) and Au(100) (--) in 1 mol dm-3 HC104+ 5 x mol dm-3 CuS04.Scanrate = 1 mV s-l. The desorption peaks for Cu on Au( 11 1) are numbered. 10 mV at 1 mV s-l to close to 100 mV at 200 mV s-l whereas the more anodic de- sorption peak (No. 2 in fig. 2) has a half width NN 30-40 mV nearly independent of the scan rate (fig. 3). Although in the beginning the main effort of this work was focused on the study of the bare Au surface before and after Cu deposition to look for " finger prints " which the deposit had left on the surface we also tried to investigate adsorbate covered surfaces. Therefore before transferring the electrode from the electro- chemical cell into the vacuum chamber several experiments were performed to find out how much of a deposited layer would remain on the surface during the transfer process.The electrode was dipped into a cell with a Cu2+ containing electrolyte the potential held at a value close to E (~0.0 V against SCE) to allow monolayer deposition then the electrode was removed from the cell with potentiostatic control operating. The electrode was rinsed carefully with deaerated triply distilled FIG.4.-RHEED pattern from a clean Au(ll1) surface for the [211] azimuth (60 keV). The indices of the reflections for this azimuth are shown in fig. 5(6). [Toface page 55 H. 0. BECKMANN H. GERISCHER D. M. KOLB AND G. LEHMPFUHL 55 water dried in a N2stream and dipped into a copper-free solution in a second electro- chemical cell at the very same potential.The subsequently recorded stripping curve revealed that at least half of a monolayer had still been present on the surface. We assume that diffusion of Cu into Au is one of the main reasons for the observed loss. Similar experiments with Cu and Ag on Pt showed that under carefully chosen con- ditions nearly a complete monolayer could be transferred from one cell to the other. A small cathodic current was observed in all cases when the electrodes were dipped into the second cell at the deposition potential indicating that a minor fraction of the adsorbate had been oxidized during the transfer. > E \ N 50 I '0 25 1 1 10 100 1000 scan rate / mV s-1 FIG.3.-Half width &12 of the desorption peaks in the current-potential curves for Cu on Au(111) peak No.2; (-0-0-0-)peak No. 4. as a function of scan rate. (-O-O-O-) The RHEED experiments were performed mainly with Au(1 1 1) surfaces which were prepared as described in the experimental section. Kikuchi lines in the diffrac- tion patterns showed the good single crystallinity of the electrodes.10 A reflection diffraction pattern in the [?ill] azimuth with Bragg reflections and Kikuchi lines is shown in fig. 4 for a clean (1 11) surface of an electropolished and annealed crystal. The Bragg reflections are elongated into streaks perpendicular to the shadow edge [see also fig. 5(6)]indicating a smooth crystal surface. Due to the mean inner poten- tial of the crystal the streaks are shifted towards the shadow edge.The sharp spots in fig. 4 are produced by electrons which passed through steps on the surface. The same diffraction pattern was observed after dipping the electrode into the electro- chemical cell containing the metal ion free base electrolyte and cycling the electrode potential in the double layer region. Even when the potential was driven a few times into the oxygen adsorption region the diffraction pattern did not change. The same was true for an Au(1 11) electrode in Cu2+ ion containing electrolyte provided that Cu deposition was carefully avoided during the potential cycling. Subsequently the potential was scanned into the region where the Cu monolayer is formed carefully avoiding the bulk Cu deposition and after reversal of the potential scan at various points the electrode was removed from the cell while the potential RHEED INVESTIGATIONS OF COPPER DEPOSITION ON GOLD was held in the double layer region at -+0.9 V.When the scan had been reversed after about half monolayer coverage a faint superstructure was seen. The same superstructure became more pronounced when the electrode was removed from the cell just at the potential corresponding to 2/3 of a monolayer (at No. 3 in fig. 2). This diffraction pattern is shown in fig. 5. The horizontal distance between the super- structure streaks is 1/3 of the distance between 000 and 022 as indicated in fig. 5(b). These additional streaks correspond to a (43 x 4%R30 superstructure arising from a 1/3 or 2/3 coverage at the ~urface.~ The fact that the superstructure is visible even after stripping the Cu adsorbate indicates that the deposited Cu leaves a rearranged surface.The superstructure always disappears during the observation with an in- tense electron beam obviously because of local annealing by the beam. Reversal of the potential after completion of the monolayer caused the same superstructure described above to appear regardless of whether the surface is bare of Cu or still covered with the full Cu monolayer. When the potential was cycled into the bulk Cu deposition region to record the usual current against potential curves for bulk and monolayer formation of Cu a drastic effect was observed on the Au(ll1) surface after removal from the electrolyte at a potential positive enough to guarantee a Cu free surface.Besides the (djx 4% R30 superstructure a (2 x 2) structure and sometimes a more complicated not yet identified superstructure was observed together with a Debye-Scherrer ring pattern. This is shown in fig. 6. (The superstructures after bulk Cu deposition were visible with and without a Cu monolayer on the surface. However since their contrast was better with the Cu monolayer we show these diffraction patterns.) It may be interesting to note that the superstructures could be changed but not removed by electropolishing which only made the Debye-Scherrer rings disappear. Annealing removed the superstructures and restored the initial state of the surface. DISCUSSION The RHEED iiivestigations have shown that the deposition of Cu onto Au(ll1) re- arranges the gold surface even in the underpotential region.The adsorbate covered substrate shows a (d3x 16)R30 superstructure when 2/3 of a monolayer is de- posited underpotentially. A possible model explaining such a superstructure is shown in fig. 7. The electrochemically determined coverage 9 = 2/3 allows us to de-cide which atoms are Cu atoms in this geometric arrangement (fig. 7). This seems to be the first direct evidence for the occurrence of ordered arrays in electrolytically formed metal deposits. Surprisingly the superstructure is also seen although it is very weak when the adsorbate is stripped and the bare Au surface is investigated by RHEED. This means that the adsorption process rearranges the Au surface e.g. by place exchange between CLKand Au leaving a finger print of the adsorbate geo- metry at.the surface after removal. This also explains the high kinetic barrier often found for underpotential deposition which would not be understandable if a mere adsorption process took place. The fact that an Au(ll1) surface either covered with one monolayer of Cu or after stripping still shows the 1/3 streaks in the RHEED pattern rather than a (I x 1) structure is more difficult to explain. Obviously completing the monolayer after deposition of the 2/3 coverage is just a filling up of vacant surface sites without further rearrangement. However it is not yet clear whether this is only observed after at least some potential cycling into the bulk deposition regime which definitively re- arranges the Au surface.Further experiments are necessary to substantiate this find-ing. suDer structure (444 1462 1 -d -a shadow edge ii3 111 131 e022 0 e022 000 lbl FIG. S(a).-RHEED pattern with superstructure streaks from a Au(ll1) surface covered with approximately2/3 of a Cu monolayer. [21I] azimuth. (b) Indices of the Bragg reflections for the [Zl I] azimuth. Streaks due to a (djx dgR30 superstructure are indicated. [Toface page 56 FIG.6.-RHEED patterns from a Au(ll1) surface covered with a Cu monolayer after repetitive bulk Cu deposition and stripping. Besides the clear (43x 2/3R30 superstructure as shown in fig. 6(a), Debye-Scherrer rings are visible. Fig. 6(6) shows a (2 x 2) superstructure. [Tofuce page 57 H.0. BECKMANN H. GERISCHER D. M. KOLB AND G. LEHMPFUHL 57 FIG.7.-Model of the Au(ll1) surface leading to a (45x 47)R 30 superstructurein the diffraction pattern. The surface layer is composed of two different kinds of atoms. The experimental results also show that deposition of bulk Cu at potentials nega- tive of E and subsequent stripping rearranges the surface strongly. The bare Au( 11 1) surface after such a treatment shows an additional (2 x 2) superstructure which could be assigned to the formation of ordered domains of an Au,Cu alloy. Dissolving the Cu from this alloy should leave vacant sites in the surface corresponding to 1/4 of a monolayer coverage. In addition removal of Cu from the alloy causes a breakdown of the gold lattice leading to the formation of small randomly oriented Au crystallites on the surface.From the width of the resulting Debye-Scherrer rings an average size of ~30 to 50A in diameter can be estimated depending on the number of preceding cycles. Since the amount of bulk Cu deposited and stripped during each potential cycle is only of the order of a few monolayers the observed clustering shows that the Au atoms which are left on the surface after decomposition of the alloy are rather mobile. It is an interesting phenomenon that electropolishing could not re- move these superstructures despite a substantial etching while the Au crystallites did dissolve. All the above results indicate that the structure of the initially perfect Au(ll1) surface is altered during the very first potential cycle especially when allowing bulk deposition to occur.The recorded current against potential cycles therefore represent curves from reconstructed surfaces. Such a reconstruction seems also to occur at the Au(100) surface which so far we have found more difficult to prepare well enough for RHEED. CONCLUSION RHEED measurements very conveniently allow one to study the geometric struc- ture of metal adsorbate layers. This can be done either by investigating the bare substrate surface and looking for “ finger prints ” which the adsorbate layer has left or by studying directly the adsorbate covered surface. In the latter case problems arising from the transfer of the electrode from the electrochemical cell into the vacuum chamber have to be overcome.For Cu on Au(ll1) the more anodic of the two pro- nounced peaks in the desorption spectrum can be attributed to the desorption of an ordered array with a (43x dgR30 structure and a coverage of 213 of a monolayer. Stripping the Cu adsorbate obviously leaves a rearranged Au surface which indicates a more complicated deposition process for the monolayer than simple adsorption. RHEED INVESTIGATIONS OF COPPER DEPOSITION ON GOLD Presumably a place exchange occurs which would also explain the high kinetic barrier often found in underpotential deposition. Bulk Cu deposition immediately leads to alloying probably to the formation of Au,Cu and dissolution of Cu leaves a strongly rearranged surface together with small randomly oriented Au crystals.This observa- tion has an interesting consequence for surface preparation since monolayer and bulk metal deposition on gold obviously may be used to "prepare '' various rearranged surfaces. See e.g. D. M. Kolb in Advances of Electrochemistry and Electrochemical Engineering ed. H. Gerischer and Ch. Tobias (Wiley-Interscience New York 1978) vol. 1 I. A. Bewick and B. Thomas J. Electroanalyt. Chem. 1975 65 911. J. W. Schultze and D. Dickertmann Surface Sci. 1976 54,489. D. Dickertmann F. D. Koppitz and J. W. Schultze Electrochim. Acta 1976,21,967. K. Juttner G. Staikov W. J. Lorenz and E. Schmidt J. Electroanalyt. Chem. 1977 SO 67. J. S. Hammond and N. Winograd J. Electroanalyt. Chem. 1977,80 123. J. S. Hammond and N. Winograd J.Elecfrochem. SOC.,1977,124,826. W. E. O'Grady M. Y. C. Woo P. L. Hagans and E. Yeager J. Vacuum Sci. Technol. 1977 14 365. W J. McG. Tegart The Electrolytic and Chemical Polishiizg of Metals in Industry and Research (Pergamon New York 1959). lo H Raether in Handbuch der Physik (Springer Berlin 1957) vol. 32 p. 443. R.Didszuhn 2.angew. Phys. 1970 30 226.