In situ scanning tunneling microscopy in electrolyte solutions

Progress

Pergamon

in Surface Saence. Vol. 5X. No. 3. pp ill-24X. 199X 0 1998 Elsev~cr Science Ltd. All rights rexned Pnnted I” Great Bntarn 0079-68 16/98 $ IY 00

PII: SOO79-6816(98)00022-7

IN SITU SCANNING TUNNELING MICROSCOPY IN ELECTROLYTE SOLUTIONS KING0

ITAYA

Department of Applied Chemistry, Graduate School of Engineering, Tohoku University, Aoba-yama 04, Sendai 980-8579. Japan (The Itaya Electrocherniscopy Project, ERATO/JST)

Abstract Until recently, there had been only few in situ methods available for the structural determination of an electrode surface in solution at the atomic level. Now, several recent investigations have demonstrated scanning tunneling microscopy (STM) to be a powerful new technique for in situ characterization, with atomic resolution, of surfaces under potentiostatic control. The object of this review is to highlight some of the recent progress made, mainly, but not exclusively, in the author’s laboratory, on in situ STM with atomic resolution. The review is focused on several selected topics, including structures of specifically adsorbed anions, underpotential deposition, adsorption of organic molecules, and electrochemical dissolution of metals and semiconductors. A combination of in situ STM and ex situ ultrahigh vacuum techniques has revealed detailed atomic structures of various adlayers, particularly iodine adlayers on Au, Ag, and Pt electrodes. It was recently demonstrated that aromatic molecules such as benzene adsorbed on Rh, Pt, and Cu can be clearly “seen” while the electrode is immersed in electrolytic solution. The atomic structures of semiconductor surfaces of Si, GaAs, and InP were successfully imaged in solution. Furthermore, it has been established that dynamic processes of electrochemical etching of metals and semiconductors can be followed by in situ STM. The work on semiconductors may well foim the basis of development of a technology for preparing atomically flat substrate surfaces, which are expected to be required by the semiconductor industry of the next generation.

Contents 1. Scope of the Review

123

2. Introduction

123

3. Experimental

1’5

Aspects

A. Principle of In Situ STM in Electrolyte Solution B. Preparation of Well-defined

Electrode Surfaces

125 126 128

4. Selected In Situ STM Studies 121

K. ltaya

122

A. B. C. D.

Structure of Specifically Adsorbed Anions Underpotential Deposition Adsorption of Organic Molecules on Iodine-modified Electrodes Adsorption of Aromatic Molecules on Clean Bare Electrodes

E. Site-selective Anodic Dissolution of Metals F. Electrochemical Dissolution Processes of Semiconductors 5. Concluding Remarks Acknowledgments References

Acronyms ADAM

Angular distribution Auger microscopy

ARUPS

Atomic force microscopy Angle-resolved UV photoemission

AES CE cv DEMS EELS EC ECALE IRAS LEED OCP RE RHE SCE SFG SHG STM sxs UHV UPD uv WE

spectroscopy

Auger electron spectroscopy Counter electrode Cyclic voltammogram Differential electrochemical mass spectrometry Electron energy loss spectroscopy Electrochemical Electrochemical atomic layer epitaxy Infrared absorption spectroscopy Low-energy electron diffraction Open circuit potential Reference electrode Reversible hydrogen electrode Saturated calomel electrode Sum frequency generation Second harmonic generation Scanning tunneling microscopy Surface x-ray scattering Ultrahigh vacuum Underpotential deposition Ultra-violet Working electrode

STM

in Electrolytes

123

1. Scope of Review Of the phenomena which occur at the interface between a solid and a liquid, common examples include the deposition and corrosion of metals, the charging and discharging of storage batteries, and the wet processing of semiconductor devices. Such processes, and many others of similar nature, involve electrochemical oxidation-reduction reactions that take place at solid-electrolyte interfaces. Until recent whilst, there had been few in situ methods available for the structural determination of an electrode surface, in solution, at the atomic level. Atomic level information had previously been acquired only via surface spectroscopic techniques in ultrahigh vacuum (UHV) [l-4]. However, since its invention by Binnig and Rohrer [5], scanning tunneling microscopy (STM) was immediately established as an invaluable and powerful surface analysis technique with atomic resolution in UHV. Belatedly, but assuredly, developments in STM operated at solid-liquid interfaces led to its valuation as arguably the premier technique for atomic-level surface structural investigations of chemical processes taking place at solid-liquid interfaces. It has been demonstrated that in situ STM makes it possible to monitor, under reaction conditions, a wide variety of electrode processes such as the adsorption of inorganic and organic species, the reconstruction of electrode surfaces, the dissolution and deposition of metals and semiconductors. Several review articles on in situ STM and related techniques such as in situ atomic force microscopy (AFM) have been published [6- 111. The review by Gewirth and Niece [ 1I] is the most comprehensive in terms of results obtained on various substrates of metals and semiconductors. The present article describes the current status of in situ STM. Due to publication limitations, the focus is on selected topics, deemed of great importance, and mainly on our own experimental results. Experimental procedures are only briefly described, because detailed reviews on this aspect have already been written [7].

2. Introduction The electrode-electrolyte interface has long been a major interest in electrochemistry ever since Helmholz described a model of the electrical double layer in 1879; this model was then followed by several modified models such as the Gouy and Chapman and the Stem models [ 121. These models dealt with the distribution of solvated cations and anions of supporting electrolytes generally on the solution side of the electrode-electrolyte interface. The effect of the nature of the metal has often been disregarded, and the treatments simply assume the free electron model. A more realistic model, taking into account the atomic structure of the electrode, can be depicted as in Fig. 1 including specifically adsorbed chemical species and the structure of water molecules at the electrode surface. It was first realized in 1980 that the adsorption-desorption reaction of hydrogen on Pt single crystal electrodes is strongly dependent on their crystallographic orientations; this demonstrated the importance

124

K. ltaya

( CD HZ0

electron transfer double layer

e

,4 b adsorbed

ion

Fig. 1. Contemporary atomic-scale model of the electrode-electrolyte interface. of the nature of metal surface at the atomic level [ 13,141. Clavilier first developed an experimental method, the so-calledjlame annealing and quenching method, to expose well-defined electrode surfaces in aqueous electrolyte solutions without the use of UHV techniques [ 131. In addition to conventional electrochemical methods, various experimental techniques, such as inji-a-red absorption spectroscopy (IRAS), Ultra-violet (UV)-visible reflectance spectroscopy, and low-energy electron diffraction (LEED) in UHV have been applied to understand the electrode-electrolyte interface [ 15,161. However, these techniques provide only averaged information, not a local picture on the atomic scale. On the other hand, since Binnig and Rohrer invented STM in 1981 [5], and since Sonnenfeld and Hansma demonstrated in 1986 that STM can be used in electrolyte solutions [ 171, several research groups quickly realized its importance for the study of electrode-electrolyte interfaces. In early 1988, we proposed a new concept for in situ STM with a four-electrode configuration where the electrochemical potentials of the tunneling tip and the substrate can be independently controlled with respect to a common reference electrode [ 181. Other research groups also independently reported similar potentiostatic STM apparatuses in 1988 [19-211. In the past decade, many successful experiments with STM, and its related technique in situ AFM, have been done under electrochemical conditions. The underpotential deposition (UPD) of Cu on Au( 111) was the first system to be intensively investigated by in situ STM and AFM, demonstrating the capability of these techniques not only to resolve individual atoms but also the adlayer structure at the solid-liquid interface. This was then followed by structural investigations of adlayers of anions and organic molecules formed on electrode surfaces. The

STM

in

Electrolytes

125

reconstruction of electrode surfaces has also been investigated at solid-liquid interfaces and compared with observations in UHV. It cannot be overemphasized that in situ STM allowed us not only to determine the surface structure, but also to follow various electrochemical reactions, such as the deposition and dissolution of metals and semiconductors. One of the fundamental goats of interfacial electrochemistry is to reveal the relationship between the surface structure and the surface reactivity of electrodes in the presence of adsorbed chemical species and solvent molecules. This review article also describes a few examples of dynamic processes of electrode reactions.

3. Experimental

Aspects

A. Principle of In Situ STM The review by Siegenthaler describes a detailed comparison between various types of electric circuits to control the electrode potentials of the tunneling tip and the substrate independently using the so-called bipotentiostut [7]. Figure 2 illustrates the apparatus of in situ STM with the four-electrode configuration. Figure 2(b) depicts the electrochemical cell for the four-electrode configuration. Using a bipotentiostat, the electrode potentials of the substrate (WE,) and the tunneling tip (WE,) can be controlled independently with respect to a reference electrode (RE).

The electrochemical current (i,)

flowing through the substrate and the counter electrode (CE) can be monitored from the output of a current follower. The tunneling current (i,) can be measured by the other amplifier.

I

,

XYZ Pie20 Controler

Feedback \’

PZT

Electrochemical STM Cell

Fig. 2. Apparatus of in situ STM with the four-electrode configuration.

126

K.

ltaya

The side wall of the tip must be isolated in order to reduce a background electrochemical current flowing through the tip. Soft glass, organic polymers, and Apiezon wax have been used. Details of tip coating methods have been described by Siegenthaler [7]. B. Preparation of Well-defined Electrode Surfaces As a fundamental basis for all STM studies, electrode-electrolyte interfaces must be prepared reproducibly, and methods must be established to observe these interfaces accurately. Well-defined single crystalline surfaces must be exposed to solution in order to understand surface structure-reactivity relationships on the atomic scale. It is still difficult to elucidate electrochemical (EC) reactions on the atomic scale using polycrystalline electrodes. Efforts have succeeded to produce extremely well-defined, atomically flat surfaces of various electrodes made of noble metals, base metals, semiconductors, and possibly metal oxides without either oxidation or contamination in solution. (i) Flame-annealing and quenching method. A unique and very convenient way to expose welldefined clean Pt into aqueous solution was proposed by Clavilier in 1980, in which mechanically exposed single crystal Pt was annealed in an oxygen flame and quenched in pure water [ 13,141. He also established a method of preparing a single crystal Pt electrode by melting a Pt wire in the flame. This technique was extended by Hamelin for Au [22], by Furuya for Ir [23], and by us for Rh and Pd

[241. Figure 3 shows typical examples of cyclic voltammograms (CVs) of the three low indexed Pt surfaces in a sulfuric acid solution obtained in our laboratory. Although the CVs shown in Fig. 3 were obtained in our laboratory, the CV feature was the same in detail as those reported previously [25,26], I

,

100

-100

E/Vvs.RHE Fig. 3. Cyclic voltammogramsof Pt( 11I), Pt( 1lo), and Pt( 100) in 0.5 M H,SO,. Scan rate; 50 mV/s.

STM

127

in Electrolytes

indicating that the technique gives reproducible results from laboratory to laboratory in the world. defined Ag electrodes could also be prepared by a similar method [27].

Well-

The result shown in Fig. 3 shows that the hydrogen adsorption-desorption reaction is a very structure-sensitive reaction on the Pt surfaces, as indicated by the different shapes and peak positions for the Cvs of the different crystalline faces. Although Clavilier and coworkers quantitatively analyzed the binding sites of hydrogen using systematically prepared stepped surfaces [26], more recent investigations using a CO replacement technique clearly indicate that the charges shown in Fig. 3 include a significant contribution of the adsorption and desorption of sulfate/bisulfate [28]. Nevertheless, a direct evidence to support the existence of well-defined surface in solution was demonstrated by us in 1990 using in situ STM [29]. Figure 4(a) showsour first STM imageof a flameannealedPt( 111) in sulfuric acid solution. The height of eachstepis ca. 0.23 nm in accord with the monatomicstepheight of 0.238 nm on the Pt( 111) surface. The monatomicstepsobserved on the surfaceam usually locatedon nearly parallelstraightlines or form an angleof 60” asexpectedfor the surfacewith threefold symmetry. The terracesseemto be atomically flat. Later it was shown that the terracewascomposedof Pt atomsforming a (1 x 1) structureasshownin Fig. 4(b) [30]. On the upper and lower terraces,the Pt( 11l)-( 1 x 1) structurewas clearly discernedat potentialsnearthe hydrogen evolution reaction. The nearest-neighborspacingand corrugationheight were 0.28 nm and 0.03 nm, respectively. It will be shown in the later sectionthat well-definedRh(111) can be also preparedby the flameannealingmethod [31]. Flame-annealed Au singlecrystals were more frequently used for various studiesincludingthe potentialinducedreconstructioninvestigatedby severalgroups[32,33]. However, it mustbe emphasizedthat the flame-annealingmethodcan be appliedonly to Au, Pt, Rh, Pd, Ir, and possibly Ag.

nm Fig. 4. In situ STM imagesof the flame-annealedPt( 111)in solution. From [29,30]

*O

128

K. ltaya

(ii) LJHV-EC. It is well-established that clean surfaces are exposed in UHV by cycles of Ar-ion bombardment and high temperature annealing. The surface structure and composition axe usually determined by LEED and Auger electron spectroscopy (AES). By using a UHV-EC system. in which a chamber for EC measurements was interfaced to a UHV apparatus, well-defined substrates can be transferred into an EC cell in purified Ar atmosphere. This experimental procedure was successfully apphed to various metals such as Pt, Au, Pd, Rh and so on [2,4]. However, for some metals, such as Ni, the oxidation of the surface took place in the EC chamber before immersion of the electrode into electrolyte solutions due to the presence of trace amounts of oxygen and water vapor [34]. The same difficulty was encountered for Cu electrodes [35]. It is clear that the problem of substrate oxidation of reactive metals occurring during the immersion and emersion processes is still unsolved in the UHV-EC method. (iii) Iodine-CO

replacement

technique.

It is important to note that the iodine adlayers are known

to protect highly sensitive surfaces of metal single crystals from oxidation and contamination in the ambient atmosphere, providing easy preparation and handling of well-defined surfaces in many aspects during EC measurements [2,4]. The iodine/CO replacement is known to be a method for exposing welldefined and clean surfaces of such electrodes as Pt and Rh in solution [36-381. The adsorbed iodine on these surfaces can be replaced by a CO adlayer. Clean surfaces are then exposed in solution by the EC oxidation of CO from the surface. (iv) Electrochemical etching method.

As described above, the flame-annealing and quenching

method can only be applied to limited metals such as Pt, Au, Rh, Pd, and Ir and can not be used for more industrially important less noble metals such as Ni, Co, Fe, and Cu, because they arc heavily oxidized in the flame as well as in air. These metals are also difficult to transfer into electrolyte solutions without oxidation even by using UHV-EC. However, it will be described in detail in a later section that the anodic dissolution of various metals occur only at the step edge under carefully adjusted EC conditions. resulting in atomically flat terracestep structures. Although the etching method has not yet been well recognized as a promising method for exposing well-defined surfaces of various metals, we demonstrate in this article that the layer-bylayer dissolution occurs on various metals, such as Ni, Ag, Co, iodine-modified Pd and Cu, resulting in the formation of atomically flat terrace-step structures.

4. Selected In Situ STM Studies A. Structure of Specifically Adsorbed Anions The adsorption of anions such as iodide [39-471, bromide [30,48,49], cyanide [50,51], and sulfate/bisulfate [.52-571 on electrode surfaces is currently one of the most important subjects in electrochemistry. It is well-known that various EC surface processes, such as the underpotential deposition of hydrogen and metal ions are strongly affected by co-adsorbed anions [2-41. Particularly, structures of the iodine adlayers on Pt, Rh, Pd, Au, and Ag surfaces have been extensively investigated

STM

in Electrolytes

129

using UHV-EC techniques such as LEED [2]. For example, the commensurate (d3 x 43)R30”, (3 x 3), and (47 x d7)R19.1° adlattices were found to form on the well-defined Pt( 11 I)-( 1 x 1) surface, depending on the electrode potential and pH of the solution [47]. More recently, these structures were confirmed by STM in both air [39] and solution [40-431. In contrast with Pt( 11 l), only one phase of the commensurate (43 x 43)R30” structure was observed on Pd( 111) and Rh( 111) surfaces with ex situ LEED [4,58] and in situ STM [59,60]. On the other hand, it has recently been recognized that the iodine adlayer structures are more complicated on Au and Ag surfaces. Although several discrepancies about the iodine adlayer structure on Au( 111) (I-Au( 111)) are found in the literature [46], s&ace x-ray scattering (SXS) studies carried out recently by Ocko et al. revealed structural changes of I-Au( 111) in KI solution [61]. They found an increasing degree of compression, the so-called electrocompression, of the iodine adlattice with increasing iodine coverage and electrode potential. Instead of commensurate structures found on Pt( 11 l), Rh( 11 l), and Pd( 111) as described above, they proposed that the iodine adlayer should be characterized as two distinct series of incommensurate adlattices, a centered rectangular phase and a rotated hexagonal phase [61]. We have recently reported the structures of I-Au( 111) in Kl solution determined by both ex situ LEED and in situ STM [46,62,63], which agree with Ocko et a1.k SXS Similar electrocompression was also found recently on Ag( 111) using the LEED and in situ STM techniques [63,64]. Our results clearly demonstrate that the complimentary use of LEED and in situ STM is a powerful technique for determining atomic structures of the iodine adlayers on single crystal electrodes. (i) Iodine adlayers on Pt(ll1). The objective of this section is to describe the potential results.

dependence of the structure of iodine on Pt( 111). Hubbard and coworkers have extensively investigated the structure of the iodine adlayer formed on Pt( 111) in aqueous iodide solutions using the UHV-EC technique [2,47]. They demonstrated that the adlayer structures, mainly the commensurate (3 x 3) and (47 x 4 7)R19.1°, were formed on the well-defined Pt( 11 l)-( 1 x 1) surface in the double layer potential range, depending on the electrode potential and pH of the solution. For example, in a solution containing 10 mM KClO, and 0.1 mM KI, adjusted to pH 4 with HI, they found the (3 x 3) and (-\/7 x Y’ 7)R19.1” structures in anodic and cathodic potential ranges, respectively (see Fig. 4 in ref. 47). It has long been expected, at least by us, that the transformation between those structures should take place reversibly when the electrode potential was changed in the double-layer potential range. Although previous STM studies revealed various atomic structures on Pt in air and in solution [38-431, no direct in situ STM investigation was carried out for the structural change expected from the results reported by Hubbard and coworkers. However, as described in our recent paper [65], the structural transformation induced by changing the electrode potential did occur, but was surprisingly slow. Nearly perfect (3 x 3) and (47 x d7)R19.1” adlayers could be prepared by immersion of the electrode in a solution and cathodic potentials, which is consistent with the previous result showed both of the structures to co-exist on the Pt( 111) surface, when immersed at potentials in the middle of the double layer potential range. Fig.

S(a), in which a (67

x

47) domain can be clearly

containing iodide ions at anodic [47]. However, in situ STM the clean Pt( 111) electrode was A typical example is shown in

seen at the center of the image,

130

I

0

K. ltaya

I

5

I

10

I

15

I

20

25 nm

Fig. 5. Adlayer structuresof iodine on Pt(ll1). (,a) at 0.45 V. potential was steppedfrom 0.45 to 0. IS V. From [651.

(b) IO min after !tv

surroundedby (3 x 3) domains. The potentialdependentstructural changewas directly probed by a time dependentin situ STM experiment. When the electrodepotential was steppedto the cathodic potentiallimit. the (47 xv’7) structure was expected to appearupon consumingthe (3 x 3) domains accordingto the previouswork [47]. When the electrodepotentialwas steppedafter the acquisitionof the imageshown in Fig. S(a), the secondimage shown in Fig. 5(b) was obtained after 10 min. indicatingthat the interconversionbetweenthe two structureswas very slow. Only a few iodine atoms markedby arrowswere incorporatedinto the (47 x 47) domain. The aboveresult provides direct evidencethat the surfacediffusion of iodine atomsis very slow on Pt( Ill), suggestingthat the iodineatomsareattachedon Pt( 111) through a strong chemicalbond. The initially formed structure was practically insensitiveto changesin the electrodepotential under our experimentalconditions [65]. It may be noted that Lynch and Corn examined the phase transition betweenthe two iodine adlayersas a function of electrodepotentialusing an optical secondhamwnic generation(SHG) method[66]. They observedthe phasetransition at a relatively fast scanrate of 10 mV/s. Our result wasobtainedin a 0.05 M H?SO,+ 1 mM KI solution, while Lynch and Corn useda 10mM NaF + 0.1 mM KI solution whosepH was adjustedto 4 by additionof HCIO,. Although we cannotyet explain the differences,it is anticipatedthat pH may be an importantfactor on the rate of the phasetransition. It is alsopossiblethat strong light illuminationin the SHG experimentincreasesthe surfacediffusion of iodineatoms. Further investigationis neededto elucidatethe dynamic processesin the iodine adlayers, even though the I-I%(111) system was thought to bc well-defined and wellcharacterizedby previousinvestigators. (ii) Iodine on Au( 111) and Ag( 111). The structuresof the iodine adlayersare more complicated on Au( 111) and Ag( 111). Various structureshave beenreportedfor I-Au( 111). McCauley and Bard

STM

in Electrolytes

131

found only the (43 x 43)R30” structure in their STM studies in air [67]. Haiss et al. reported several structures such as (43 x 43)R30”, (5 x 43), and (7 x 7)R21.8” in air and in a nonaqueous solvent [68]. Under a very similar set of experimental conditions, Tao and Lindsay found a potential-dependent transition only from (43 x 1/3)R30” to (3 x 3) [69]. We also previously reported (5 x 43) and (7 x 7)R21.8” structures in pure HClO, solution in the absence of KI [45]. These discrepancies strongly suggest that the structure of I-Au( 111) is sensitive to EC parameters, such as electrode potential. However, recent in situ SXS studies by Ocko er al. revealed a series of I-Au( 111) adlattices [6 11. The adlattice constants varied continuously with the electrode potential in each of the two-dimensional phases designated by Ocko er al., the rectangular (p x 43) phase and the rotated hexagonal phase [61]. (a) In situ STM. We have extensively investigated the same system by the complementary use of in situ STM and ex .sim LEED. Figure 6 shows a CV of a well-defined Au( 111) in 1 mM KI. A pair of anodic-cathodic peaks associated with adsorption-desorption of iodine appeared below 0 V 1~s. Ag/Agl. The small pair of peaks at 0.5 V vs. Ag/AgI is due to a phase transition between the two phases described in detail below. Two different types of images were obtained in the potential range from -0.2 V to 0.54 V. as shown in Fig. 7. The transition between these two phases occurred at ca. 0.5 V, where a small reversible peak

(43 x v!3)R30°

(d

I

-0.4

I

-0.2

I 0

I

0.2

r x $3 r)R(SO”+a)

I

I

0.4

0.6

E / V vs. Ag/Agl Fig. 6. Cyclic voltammogram of Au( I1 1) in 1 mM KI at a scan rate of 50 mV/s. From [46,63].

132

K. ltaya

Fig. 7. In situ STM images at potentials of -0.2 (a), 0.3 (b), and 0.5 V tc), respectively. From [46]. appears in the voltammogram in Fig. 6. As shown in Figs. 7(a) and 7(b), all adatoms were at the same height with atomic corrugation about 0.03-0.04 nm. Figure 7(c), on the other hand, shows an STM image at the potential of 0.5 V. The I adlattice obtained at -0.2 V possessed (43 x 43)R30” symmetry with a characteristic interatomic distance of 0.50 nm as shown in Fig. 7(a). No distortion from the three-fold symmetry was observed. A positive shift of the electrode potential resulted in the formation of a more densely packed adlattice. As illustrated by the unit cell in Fig. 7(b), the spacing of atoms in 63 direction remained unchanged within the experimental error, regardless of the electrode potential between 0 and 0.4 V. Images obtained at more positive potentials contain more atoms in the direction perpendicular to the -\13direction, resulting in a uniaxial compression as shown in the unit cell shown in Fig. 7(b). Accurate determination of the interatomic distance was rather difficult due to distortion caused by a thermal drift. To determine the adlattice constant more precisely, STM images were recorded at two different tunneling currents (ca. 1 nA and 40 nA) consecutively. The first scan imaged the iodine adlayer only, and the second scan only the underlying Au( 1 11)( 1 x 1). The lattice constants were more accurately determined by overlapping these two images [63]. The evaluation of p in c@ x 43R-30”) was based on the nearest iodine-iodine distance. The distance between two nearest iodine atoms in the compressed rows varied from 0.50 nm (43 x 43)R30” in Fig. 7(a)) to 0.43 nm in the potential range between -0.1 V and 0.4 V with an error of ca. 0.02 nm. In the phase observed at potentials more positive than 0.5 V, shown in Fig. 7(c), the atomic adlattices of I possess a true 6-fold symmetry. The nearest I-I distance was smaller than that of (43 x 43)R30” (0.50 nm), and the entire lattice seemed to be rotated by several degrees with respect to the (63 x 43)R30”. This type of adlattice has been denoted as rot-hex by Ocko et al. [61]. The adlattices are furthermore modulated with periodically arranged surface features. These features, namely groups of

STM

in Electrolytes

133

slightly elevated I atoms, are interpreted as Moire patterns resulting from the mismatch between the adlattice and the lattice of Au( 111). The Moire pattern can be analyzed by simulation to determine the adlattice constant [62,63]. The STM images were simulated by computer calculation based on a simple “hard-ball contact model”. Various compression ratios (O-15 % subtracted from the values for (43 x 1/3)R30”) and rotation angles (O-5” from (43 x 1/3)R30”) were tested. The rot-hex I-I distance obtained by this simulation varied from 0.45 nm to 0.43 nm and the rotation angle from 2” to 5”, the variations corresponding to the change of potential from 0.45 V to 0.55 V. (b) Ex situ LEED. Figure 8 shows LEED patterns obtained after emersion from 1 mh4 KI under EC potential control. A genuine (43 x 43)R30” was observed upon emersion at -0.2 V vs. Ag/AgI (a). At the more positive emersion potential of 0.35 V (b), the LEED patterns underwent splitting into “triads” and shifting of the original (43 x 1/3)R30” spots. As the potential was increased, the subspots moved away slightly further from the center and the distances between the split spots increased. Additional weak spots were also seen near the split spots. Quantitative analysis of these LEED patterns has been described in our previous papers [46,62,63]. Figure 9 shows an illustration of a real lattice of the c@ x 43R-30”) structure (a) and a LEED pattern calculated by taking into account the existence of three symmetrical domains (b). Six spots at the comer of the elongated hexagon are due to the LEED spots from one of the equivalent three c(p x 43R-30”) domains. Occasionally the additional weak spots near the three split spots as described above, which

Fig. 8. LEED patterns on Au( 111) obtained at potentials of -0.2 (a) and 0.35 V (b). From [46,63].

134

K. ltaya

0

Au

Fig. 9. Illustration of the real lattice with fundamental unit vectors a,, a, of Au( 11 I)-( 1 x 1) and ~(2.5 x 1/3R-30”) adlattice vectors, b, and b,. a,*, a**, b,*, and b,* are the reciprocal units vectors. From [46]. can be seen in Fig. 8(b), have been interpreted as the electron double scattering via the six fundamental spots nearest to the center spot of the (1 x 1) lattice of the substrate. Large open circles and medium open circles are the fundamental Au(1 ll)-(1 x 1) spots and c@ x 43R-30”) single-scattering spots, respectively. Small filled circles are the double-scattering spots. The variation of the adlattice constants is only along a particular direction of the Au atomic rows. As the largest value of p is equal to 3, at which the adlattice is identical to (213 x 43)R30”,

a value of p

smaller than 3 signifies compression of the (43 x 43)R30” structure only in the direction of the Au( 111) atom row. Hence, the variation of p is referred to as uniaxial compression. As can be seen in Fig. 10, the lattice parameters determined by in situ STM and ex situ LEED are fairly consistent with that obtained by SXS, which is usually believed to give the most accurate values. Nevertheless, our results are consistent with those obtained by SXS, demonstrating that complimentary use of in situ STM and ex situ LEED is a powerful technique for characterizing the atomic structure of iodine on Au( 11 I). It is noteworthy that the c@ x 43R-30”) structure was also observed at potentials more positive than 0.5 V, at which a rotated hexagonal structure had been expected. This observation is probably due to the instability of the adlayer in UHV. Cochran and Farrell reported a similar instability in UHV [70]. They observed similar LEED patterns with the splitting features for a gas phase adsorption on Au( 111).

STM

in Electrolytes

135

0.42

2.4

0.4

2.5 2.6 p

0.38

2.7

z

2 2.8 2

0.36

2.9

0.34

3.0 0.32

-0.2

0

0.2

0.4

E / V vs. AglAgI Fig. 10. Continuous variation of the surface coverage (0,) and parameter p. The solid line by SXS [61]. Solid circles by LEED and squares by STM. From [46,63]. The same phase transition from the c@ x 43-R30”) to the rotated hexagonal structures was found and characterized by in situ STM and ex situ LEED for an iodine adlayer on Ag( 111) in an HI solution [64]. Our result is consistent with that obtained using SXS carried out by Ocko et al. [71]. LEED patterns with six splitting spots were observed at potentials before the bulk formation of AgI, indicating that the rotated hexagonal structure was stable even in UHV [64]. The iodine adlayer on Ag(ll1) was complicated in an alkaline solution, showing several structures including square (43 x 43R-30”) and ((43 x 43)R30” [64], suggesting that there is a remarkable pH dependence of the structure of iodine on Ag(lll). (iii) Sulfate/bisulfate.

The adsorption of sulfate/bisulfate has become an important issue in

electrochemistry of well-defined surfaces voltammograms on a well-defined Pt(ll1)

since Clavilier and coworkers reported anomalous in sulfuric and perchloric acid solutions [ 13,14,26].

Although the anomalous state of adsorption observed on Pt(ll1) in sulfuric acid had long been interpreted as being due to the adsorption-desorption process of hydrogen [2.5,26], the recent result reported by Clavilier and coworkers using a CO replacement technique demonstrated that the so-called butterfly peak observed on Pt( 111) in sulfuric acid solution is possibly due to the adsorption-desorption of sulfate rather than that of hydrogen [28]. It has been demonstrated by several groups that in situ STM can be used to visualize adsorbed sulfate/bisulfate species on Au( 11 l), Pt( 111) and Rh( 111). (a) Au( 111). An ordered structure with a (43 x 47) symmetry was first observed for the adsorbed sulfate/bisulfate on Au( 111) in sulfuric acid by Magnussen et al. [52], who proposed a model structure based on the assumption that the adsorbed species is bisulfate, not sulfate, with a surface coverage of

136

K. ltaya

0.4. More recently, Weaver and coworkers reported STM images with the same symmetry ot’ (~3 x 47) on Au(l11) [53], as that observed by Magnussen et ul.. However. they proposed a possibility ct! incorporation of hydronium cations in the ordered sulfate adlayer by taking into account the rcqult that the surface coverage of sulfate on Au( 111) determined by chronocoulometry and radiochemical assay IS 0.2 [72]. Note that this surface coverage is one half of that for the structure proposed by Magnussen ~‘f al. asdescribedabove. (b) Pt(ll1). Stimming and his coworkers found by in siru STM that adsorbedsulfate ions form the sameadlayerstructureasthat found on Au(l11) [55,56]. Ordereddomainswith (43 x d7) symmetry appearedin the potentialrangebetween0.5 and 0.7 V vs. a reversible hydrogrn electrode (RHE) in 0.05 M H,SO,. As shownin Fig. 3, only the (111) surfaceshowsthe characteristicbutterfly peaksat potentialsslightly negativethan 0.5 V. Their STM observationsconfirmed that the butterfly peaksare dueto the adsorptionanddesorptionof sulfate ions as indicatedwith the CO replacementtechniqueby Clavilier as describedabove [28]. STM imagesobtainedon Pt( 111) were interpretedin terms of the coadsorptionof sulfateanionsandwater. (c) Rh(ll1). Wieckowski andcoworkershave previously investigatedthe voltammetric propertiesof Rh singlecrystal electrodes,showingan interestingcomparisonbetweenthe voltammetricbehaviors of Rh(111) andPt( 111)[58,73-751. A more detailedanalysisof the voltammetry of Rh(111)in perchloric acid solutionwasrecently reportedby Clavilier et al. [76]. It is remarkablethat the electroreductionof perchlorateanionstakesplaceon Rh electrodes[76,77]. It was reportedin our early paperthat atomic STM imagesof Rh( 11l)-( 1 x 1) structurewere obtainedat potentialscorrespondingto the positive and negativeendsof the butterfly peakin perchloric acid solution[60]. On the other hand,the voltammetric behavior of Rh(l11) in solutionscontaining sulfuric acid is very different from that on Pt( 111) as reportedby Zelenay and Wiekowski [75]. The adsorptionof sulfate/bisulfateseemsto occur more strongly on Rh( 111)than on Pt( 111), becausethe hydrogen adsorption-desorption peakson Rh( 111) in perchloric acid are strongly compressedtoward negative potentialsnear the onset of the hydrogen evolution reaction in solutionscontainingsulfate [75]. This behavior strongly encourageus to try to reveal adlayerstructuresof sulfateon Rh(111)with in situ STM. Wieckowski andco-workersreportedon the potentialdependenceof sulfateadsorptionon Rh(11I ) in 0.1 M HClO, in the presenceof H,SO, studiedby using a radiochemicalassaytechnique[75]. It was clearly demonstrated that the desorptionof sulfateoccursin a very narrow potentialrangeof the cathodic peak asdescribedabove. The maximumsurfaceconcentrationof sulfatewas found to dependon the electrodepotentialaswell asthe concentrationof sulfuric acid in 0.1 M HClO,. According to the data reportedby Zelenay andWiekowski (seeFig. 7 in ref. 75), the maximumsurfaceconcentrationof ca. 4 x lOi ions/cm*wasfound in a 0.1 M HClO, + 0.2 mM H,SO, solutionusing35s labeledsulfate. The above valuecanbe translatedto a surfacecoverageof ca. 0.25 by dividing it by the surfacedensity of Rh( 11l)-( 1x1) (1.6 x 10” atoms/cm*). It was reported in our recent paper that an atomically flat terrace-stepstructure was routinely observedby in situ STM on well-orderedRh( 111) in perchloricacid and HF solutions. The ordered terrace-stepstructure was easily observedin HClO,, HF, and even H,SO, solutions [3 1,57,60]. Figure 1l(a) showsa typical large-scaleSTM imageacquiredin a largeareaof 500 x 500nm. Only a

STM

d

160

260

360

400

in Electrolytes

560 nm

137

im

Fig. 11. Topographic STM image (a) and atomic resolution image (b) of a flame-annealed Rh(l11) in HF. From [31]. few step lines were found in the scanned area, and atomically flat terraces usually extended over 100 nm. The well-defined steps were found to be all monatomic and aligned predominantly along the atomic rows of the Rh( 111) substrate. Although the result shown in Fig. 1l(a) indicates that me well-ordered Rh( 111) surface was exposed by the flame-annealing and quenching, the hexagonal close-packed structure with a nearest neighbor space of ca. 0.27 nm was consistently observed at 0.5 and 0.8 V in 0.1 M HClO, and 0.1 and 0.7 V in HF. Figure 1 l(b) shows a typical high-resolution STM image of a relatively large area of 5 x 5 nm acquired at 0.5 V in a HF solution. A hexagonal structure can be readily recognized with a nearest neighbor space of ca. 0.27 nm. The image shown in Fig. 1 l(b) directly indicates that the Rh( 111) surface has the (1 x 1) structure. High-resolution STM imaging conducted on atomically flat terraces at 0.5 V in H,SO, readily discerned atomic features as shown in Fig. 12(a) obtained near the step edges [57]. The image areas include three terraces with monatomic steps. It is clearly seen that parallel atomic rows in each domain are located in the directions forming angles of nearly 60” or 120”. It is also recognized that individual bright spots exist very near the monatomic step. This observation strongly indicates that the entire surface of Rh( 111) is almost completely covered by adsorbed sulfate ions even very near the end of the terraces. Figure 12(b) presents an STM image showing a more detailed internal structure acquired in an area where a single domain appeared on a wide terrace. It can be seen that there are two different parallel rows with a 30” rotation relative to the underlying Rh lattice. One appears as bright spots. The

K. ltaya

138

(b) Bv

0

4

8

16

0

1

2

3

4

5

nm

Fig. 12. High-resolution From 1571.

nm STM images of sulfate adlayer on Rh( 1I 1) obtained in H,SO,+.

observed atomic distance in these bright rows along the A direction is equal to 0.46 nm. The average distance between neighboring bright rows is ca. 0.7-0.73 nm. The interatomic distance of 0.74-0.75 nm observed along the B direction in this particular STM image in Fig. 12(b) is slightly larger than that of the 47aRh (0.707nm), probably due to a small thermal drift during the acquisition of the image. However, it was ascertained that the distance along the B direction is very close to the 47aRh based on the averaging of all atomic images obtained in this study. The angle between the directions marked by arrows A and B in Fig. 12(b) is ca. 72”. The above results strongly indicate that the unit cell can be defined by the so-called (43 x 47) structure.

The rows along the direction marked by the arrow B are

constituted of alternative bright and dark spots in the image shown in Fig. 12(b). The dark spots appeared almost at the center between neighboring bright spots in the direction of B. Magnussen et al. have also found the position for the darker spots equidistantly between two brighter spots along the B direction [52]. However, it is noteworthy that the position of the dark spots with respect to the bright spots in the unit cell shown in Fig. 12(b) depended on the experimental conditions such as tunneling current, bias and data acquisition mode, i. e., constant current or constant height. The dark spots were sometimes found at the positions more close to one of the neighboring bright spots and aligned slightly off the rows of B. We also acquired images with the same structure in the constant-current mode (2- 15 nA) in order to determine corrugation amplitudes. Corrugation amplitudes of 0.03-0.05 nm were observed along the bright rows in the A direction marked by arrow A. The results described above are almost the same as those reported on Au( 111) by Magnussen et al. [52] and other investigators [53,54] and as those observed on Pt(ll1) 155,561. Magnussen et al. proposed a model structure with a unit cell, the so-called (63 x 47), for the adlayer of bisulfate (HSO;)

STM

in Electrolytes

139

on Au( 111). Both bright and dark spots were interpreted as bisulfate ions adsorbed on Au( 111). Therefore, the surfacecoverageof this proposedstructureis 0.40. Note that comprehensivedata are now available for the surface coverage of sulfate/bisulfateadsorbedon Au(ll1). Lipkowski and coworkers employedchronocoulometicand radiochemicalassay techniquesto determinethe Gibbs excessof sulfate on Au( 111) [72]. A good agreementwith the surface concentrationsof sulfate determinedby such independenttechniqueswas reported. They obtaineda value of ca. 2.8 x IO’” ions/cm’ at potentialsnear the onsetof gold oxide formation on Au( 111) where the ordered structure with (43 x d7) symmetry was found by in situ STM [52-541. Note that the above value can be translatedto a surfacecoverageof 0.20 by dividing it by the surfacedensity of Au( 11l)-( 1x1) ( 1.39 x 10” atoms/cm’). Weaverand coworkersproposeda structureof the adlayer of sulfatewith a surface coverageof 0.2 with an identical symmetry of (63 x 47) [53], interpreting that all the bright spots appearingin the STM imagessimilarto thoseshownin Fig. 12 correspondto the sulfate ions adsorbed on Au( 1I 1). They also suggesteda possibility that the adlayer incorporateshydronium cations [53]. The dark spotsdescribedabove have beenassignedto hydronium cationson Au( 111). Consequently, the surfacecoverageof sulfateshouldbe equal to 0.2 accordingto the modelproposedby Weaver and coworkers [53]. According to the coveragevalue of ca. 0.2 obtainedon Rh( 111) by Wiekowski as mentionedabove [75], it is reasonableto expect that only the bright spots in the STM imagesobservedon Rh( 111) correspondto the adsorbedsulfateor bisulfate. If sulfateor bisulfateis assumedto be also trigonally coordinatedon Rh(l1 l), a ball-modelcan be presentedas shown in Fig. 13(a), where the SO,‘. (or HSO,) is positionedat the three-fold hollow sites. It can be seenin Fig. 13(a) that the sulfateions

0

Rh

eH20

(1st layer)

6 H20 (2nd We0

Fig. 13. Model structuresof the sulfateadlayeron Rh( 111) surface:(a) sulfateon the 3fold site; (b) coadsorptionof sulfateandhydrogen-bondedwater chains. From [57].

140

K. ltaya

along the 43 direction form an almost close-packed row. On the other hand, an open space can be found between neighboring rows of the sulfates. As described above, Weaver and coworkers proposed a model where coadsorbed hydronium cations exist along the 43 direction between neighboring rows 01’ sulfates [53]. The dark spots which appeared in the STM images were assigned to be the coadsorbed hydronium cations. Such cation coadsorption was expected to minimize the coulombic repulsion between adjacent SO,*- on Au( 111). Although the coadsorption of hydronium cations is thought to be a factor which explains the nonuniform interatomic distances of the (43 x 47) structure, it is not clear why the adsorbed sulfates have the different spacings along the 43 and 47 directions. In our previous paper, we proposed a new model to explain the nonuniform spacing in the unit cell of (43 x \17). The reader can see in Fig. 13(a) that uncoordinated Rh atoms are arranged in a zig-zag form in the 43 direction between neighboring rows of the adsorbed sulfates. In a new model shown in Fig. 13(b), hydrogen-bonded water chains are simply inserted along the 43 direction between neighboring rows of the sulfates. It is well-known that ordered superlattices of water are formed on the (Ill) surface of fuce-centered cubic (fee) lattices, such as Pt and Rh at low temperatures [78,79]. The (43 x -\j3)R30” symmetry was found by LEED for at least the first water layer on these surfaces. Ross proposed a model of hydrogen-bonded aqueous networks to explain anomalous features of the voltammetry of Pt( 111) [79]. The model shown in Fig. 13(b) includes the adsorbed sulfate/bisulfate and hydrogen-bonded water chains formed along the 43 direction. The model shows only the first water bilayer. Water molecules in the first layer are bonded directly to Rh atoms at the ontop site via the oxygen lone pair. For the ice-like lattices on the fcc( 111) surfaces, two hydrogen bonds form to oxygen lone pairs of two water molecules in the second layer [79]. It is assumed, however, that only a hydrogen bond forms to an oxygen lone pair of a water molecule in the second layer as shown in Fig. 13(b), although it is expected that the water in the second layer forms a hydrogen bond with an oxygen lone pair of sulfate, which is not drawn in Fig. 13(b) for the clarity of the figure. According to the model, it is possible that the dark spots which appeared in the STM image shown in Fig. 12(b) arise from the water molecules in the second layer. Although the model presented here would seem to be equivalent or similar to the model proposed by Weaver et al. [53], our model more confidently explains the feature of the nonuniform interatomic distances in the (43 x 47) structure. Note that it is also possible to insert the hydrogen-bonded water chain with the reversed configuration between the first and second water layers. In this case, the water in the second layer is located almost at the center of the unit cell shown in Fig. 13. It is also noteworthy that the hydrogen-bonded water chains are expected to form hydronium cations in acidic solutions. If the water molecule in the second layer is protonated to form hydronium cations, the model presented here is equivalent to the model proposed by Weaver er al. [53]. The observed position for the dark spots depended on the experimental conditions as described above. Such dependence might have resulted from the two possible structures of the hydrogen-bonded water chains. We have calculated the heights of the adsorbed sulfate and the water molecule in the second layer for the mode1 shown in Fig. 13(b) using a hard-ball contact model, yielding values of 0.35 nm and 0.30 nm, respectively. The outer-sphere of the water molecule in the second layer is slightly lower that that of the sulfate. It is interesting to compare these values with the corrugation heights described

STM in Electrolytes

141

above, although it is true from a theoretical point of view that corrugation heights observed in STM images should arise from electronic factors, such as wave functions of adsorbates rather than spatial structural factors. We believe that a similar coadsorption of water chains and sulfate might occur on Au( 111) and also possibly on Pt( 111). More recently, Ataka and Osawa proposed a new model, in which water molecules bridge the adsorbed sulfates via hydrogen bonding [80]. When the electrode potential was cathodically stepped in increments of 10 mV from 0.5 V, the structure with the (43 x 47) symmetry shown in Fig. 12 was consistently observed over the terraces until the electrode potential reached ca. 0.15 V. The adlayer structute almost abruptly disappeared at potentials near the hydrogen evolution reaction, showing the Rh( 11 l)-( 1 x 1) structure, indicating that the adsorbed sulfate is completely desorbed at the potential of the hydrogen evolution. It should be noted that Wieckowski and coworkers investigated the adlayer structure of sulfate/bisulfate on Au( 11 l), Rh( 11 l), and Pt( 111) using ex situ LEED in vacuum, reporting a structure of (43 x 43)R30” on these substrates [81-831. In UHV, we believe that water molecules evaporate, inducing the change in structure from (43 x 47) to (63 x 43)R30” on these substrates, although a more precise technique, such as SXS, should be used to explain the difference found by LEED.

in situ

STM and ex

situ

Nevertheless, it is now clear that sulfate/bisulfate form adlayers with the same structure and symmetry on at least three different substrates, Au( 11 l), Rh( 11 l), and Pt( 111). Note that Ito’s group also found the (43 x 47) structure for a UPD layer of Cu on Pt( 111) in 0.05M H,SO, in the presence of 1 mM CuSO, [84]. It is noteworthy that the atomic diameters of Au, Rh, and Pt are 0.289, 0.268, and 0.278 nm, respectively. The diameter of Rh is the smallest among the above three metals. In general, structures of many adlayers depend strongly on the diameter of substrates. The appearance of the same (43 x 47) structure on the substrates with different lattice parameters might suggest that the coadsorption of sulfate/l-isulfate and water illustrated in Fig. 13(b) is flexible with respect to the change in lattice parameter of the substrate. Such a flexibility may be due to the existence of water molecules between the sulfatelbisulfate chains with a relatively weak hydrogen bonding. (iv) Cyanide. A detailed structural analysis of the CN adlayer on Pt electrodes was carried out by P-I situ LEED [85-871 as well as in situ infrared spectroscopy (IR) [88-921, sum frequencv ~enerafion (SFG) [93]. It is believed that the adsorbed CN linearly binds to the Pt substrate predominantiy through its C-end. Real space structures of the CN adlayer on Pt( 111) were determined by ex situ LEED 185-871, lvhich showed that two different adlattices consisting of the CN- anion and its protonated form (CNH)

exist

with the structures of (243 x 2d3)R30° and (4 13 x 413)R14”, depending on the electrode potential and pH. The latter structure was found to appear at negative potentials 1861. It was also clearly demonstrated that various cations interact with the CN adlayer on Pt( 111). Among the monocations examined, potassium cations are most strongly attached on the adlayer [87]. On the other hand, in sifu STM was recently employed, for the first time, to examine the CN adlayer on Pt( 111) [SO]. It was reported that six CN functional groups form a hexagonal ring with an additional CN in the center of the ring (see Fig. 14(b)). This structure differs from that suggested on the basis of the ex siru LEED data [8.5], although both structures had the same (243 x 243)R30” symmetry.

K. ltaya

142

:b)

Fig. 14. High-resolution STM image (a) of the hollow CN hexagonal arrangement. Two ball models of the (243 x 243)R30” structure are shown in (b) and (c). From [51]. Our in situ STM observations revealed a new structure for the CN adlayer and the complexation of alkali cations such as Na’ and K’ with the CN adlayer on Pt( 111) [51]. An atomically resolved STM image acquired in an alkaline CN solution containing Na* cations is shown in Fig. 14(a). Figure 14(a) shows an STM image of the hollow hexagonal pattern. This image (5 x 5 nm) was acquired in 0.1 mM NaCN + 0.1 M NaCIO, (pH 9.5) at 0.6 V with a bias voltage of -50 mV and a tunneling current of 20 nA. Under these experimental conditions well-arranged hexagonal rings, aligned in a direction 30” rotated from the close-packed directions of Pt( 111) lattice, are observed. The 0.95+0.02 nm distance between the nearest neighbor hexagonal rings, as measured from their centers, is roughly twice as large as the 1/3 lattice spacing of the Pt (0.2778 nm). This ordered atomic feature can be characterized as the (243 x 2d3)R30° structure, which has the same symmetry as that proposed by the LEED [85-871 and in situ STM studies [50]. ref. 50) which was symmetric atop sites that the center spots cyanide in the center

However, Weaver and co-workers reported an STM image (see Figs. 3 and 4 in interpreted as the (263 x 243)R30”-7CN structure with cyanides bound in surrounded by hexagonal rings of near-top CN. It is important to see in Fig. 14(a) are now essentially invisible in the image, strongly suggesting that there is no of the hexagonal ring.

Based on the results described above, two model structures are presented in Figs. 14(b) and 14(c) to outline atomic arrangements of the adsorbed CN on Pt( 111). The model structure shown in Fig. 14(b) is essentially a replica of the model proposed by Weaver and co-workers [50] in which the center spot

STM

in Electrolytes

143

Fig. 15. High-resolution STM image (a) of the ordered (243 x 243)R30” structure observed in the solution containing K’ ions. Bright spots appear in the center of the CN ring. (b) a model indicates that K’ cations bounded in the center of the CN ring. From [511.

was attributed to the adsorbed CN.

However,

the image shown

in Fig. 14(a) suggests that the

adsorbed CN is not located in the center of the six membered ring. Figure 14(c) is a new model where the adsorbed CN in the center is removed. It is interesting to note that the six-membered ring is similar in structure to crown ethers.

Crown

ethers are known to effectively complex with the alkali metal cations. The configuration of the CN adlayer is such that the C is bound to the Pt electrode with the N facing the solution side. Each nitrogen atom contains a lone-pair of electrons which is expected to act as a binding site similar to the oxygen atoms in crown ethers. Figure 15(a) shows a high-resolution STM image obtained in a solution containing potassium cations. It was surprising to find that the bright spots appear in the center of the six-member ring as illustrated in Fig. 15(b). It was found that K’ cations are more strongly bounded in the center of the CN ring than Na’, because bright spots due to the coordinated Na’ cations were only sparsely observed in the center of the hexagonal ring of CN. On the other hand, the bright spots appear almost uniformly over the entire imaged area as shown in Fig, 15(a). We have also observed a dynamic process of the adsorption-desorption of the coordinated K’ cations by changing the electrode potential. It was found that such an equilibrium reaction resulted in the IR peak frequency-potential curve [9 11. Our STM result described above is probably the first case to describe outer Helmholtz layer, because almost all previous STM studies have elucidated the adlayer structure directly attached on the electrode surface [94].

144

K. ltaya

(v) Other anions. There are not many in situ STM studieson adlayersof chloride and bromide formed on single crystal electrodes. We have investigateda Br adlayer on Pt( 111), revealing the existenceof asymmetricand hexagonal(3 x 3) structuresin the double-layer potential range [30]. Atomic imagesobservedat potentialsslightly negativewith respectto the first characteristicsharppeak were not consistentwith the previously proposedmodel for a (4 x 4) structureby the Hubbard group [95]. Our preliminary study indicatedthat incommensurate-like Br adlayerswere formed at potentials more negativethan the first characteristicpeak. Pt( 111)-(1 x 1) structure was discernedat potentials nearthe hydrogenevolution potential,indicatingthat the adsorbedBr was desorbedat theseconditions. Note that Lucaset al. found incommensurate Br adlayersin the double-layer potential range using an SXS method[49]. Further work is neededto resolvethe abovediscrepancy. The adsorptionof thiocyanateon Pt( 111)and Rh(111)wasalsoinvestigatedby in situ STM. Ex situ STM was usedto examinethe structureof thiocyanateon Au( 111) in air, revealing a squareatomic pattern [96]. On Pt( 111) a (2 x 2) structurewas moreclearly observedin a cathodic potential range [97], which is consistentwith the resultreportedby the Hubbardgroup usingex situ LEED 185,861.In contrast to the existenceof a large number of investigationsof Au and Pt electrodes,few papers addressingthe adsorptionof anionson Rh electrodesarefound in the literature[58]. We have recently found that thiocyanateform (2 x 43) and (2 x 2) structureson Rh(111) in the alkaline and the acidic media,respectively [98]. The adsorptionof halideson Ag electrodesare also of fundamentalinterest, becauseof a strong interactionbetweenAg and halides[27,64,7 1,99,100]. Sneddonand Gewirth characterizedthe halide adsorptionand the bulk growth of Ag-halide using in situ AFM [ 1001. STM cannotbe usedto study processes suchasthe bulk growth of Ag-halide becauseof the inhibition of the tunnelingcurrent of the insulatinglayer. Suggsand Bard investigatedthe adlayersof chloride formed on Cu(ll1) and Cu(lO0) [101,102]. They revealed the (643 x 643)R30” and (42 x d2)R45” structures on Cu(ll1) and Cu( loo), respectively. They alsofound the anodicdissolutionof Cu on Cu(111)and (100) in solutioncontaining chlorideionstook placepreferentiallyat stepsites,resultingin layer-by-layer dissolution,which will be discussedin a later section. It is clear that more comprehensivestudiesare neededto rationalize the relationshipbetween the adlayerstructureandthe structureof electrodesurfaces.In this respect,it shouldbe noted that recentin situ SXS hasclarified detailedpotential-dependent structuralchangeswith high accuracy. Particularly,

Ocko and his coworkersreportedthe adlayerstructuresof Br on Au( 111), Au( IOO), and Ag( 100) [1031051. It was demonstratedthat complimentaryuse of in situ SXS and in situ STM is a promising techniquefor determiningatomic-levelstructuresof electrode-electrolyteinterfaces[ 1031. B. Underpotential Deposition The EC adsorptionof hydrogen and metalson a foreign metalsubstratetaking place in a potential regionpositiveto the thermodynamicallyreversiblepotentialis called underpotential deposition (UPD). Particularly the UPD of hydrogen on single crystal Pt electrodeswas intensively and systematically investigatedby Clavilier and his coworkers [26]. The UPD of a metal, M, on a different metal

STM

in Electrolytes

145

substrate, MS, is expected to occur at potentials more positive than the reversible potential for the bulk deposition of M when an interaction between M and MS is greater than that of MS-MS [ 106-1081. The UPD process is important in EC reactions, such as metal deposition, as the initial step of a series of reactions [109] and also because of electrocatalytic effects induced by adatoms formed by the process [ 1 lo]. Although a large number of UPD systems have been investigated by using conventional EC techniques, such as CV, to evaluate the thermodynamics and kinetics, the structural information of UPD layers was first obtained mainly by Hubbard and his coworkers using the UHV-EC technique [2,4]. With a considerable amount of previous knowledge available on the UPD phenomenon itself, in situ STM was applied, for the first time, to determine the structure of the adlayer of Cu on Au( 111) in sulfuric acid solutions with atomic resolution. (i) Cu UPD on Au(ll1). Magnussen et al. reported the first atomic image of a Cu adlayer on Au( 111) in a sulfuric acid solution [ 1111. They found a (63 x d3)R30” structure after the first UPD peak which transformed into a second phase of (5 x 5) structures. However, the appearance of the (5 x 5) structure was confirmed to be due to chloride contamination [ 1121, suggesting that the UPD process is extremely sensitive to co-adsorbates, which will be discussed further in a later section. Figure 16 shows CVs of an Au( 111) electrode obtained in a pure 0.05 M H,SO, solution in the presence of 1 mh4 CuSO, [ 1131. The main oxidation peak at 1.25 V and the cathodic peak at 0.82 V are due to the oxidation of the surface of Au( 111) and the reduction of the oxide layers, respectively. On the other hand, two different waves for the UPD of Cu are clearly observed in the potential region between 0.35 and 0 V vs. SCE before the beginning of the bulk deposition. Figure 16(b) shows a detailed CV for the UPD observed at a scan rate (v) of 1 mV/s. Two distinctly different processes can

0

0.5

1.0

1.5

v vs. SCE

Fig. 16. Cyclic voltammograms for a Au(ll1) electrode in 0.05 M H,SO, + 1 mM CuSO,. Scan rates; 20 mV/s (a), 1 mV/s (b). From [ 1131.

146

K. ltaya

be seen clearly in the CV shown in Fig. 16(b). The peak current was proportional to the scan rate only up to 5 mV/s, and then approximately proportional to v I’*. suggesting that the UPD of Cu on Au( 1I 1) is a surprisingly slow process. Batina et al. reported a significant effect on the shape of CV obtained in solutions containing chloride and an organic molecule, crystal violet [ 1141. Nevertheless, we concluded that the (43 x \/3)R30” 1s the only observable structure for a long period of time in the potential range between 0.18 and 0. I V in a “pure” H,SO, [ 1131. Figure 17(a) shows a high-resolution STM image obtained in a 0.05 M H,SO, + 1 mM CuSO, solution [ 1131. Although the wide terrace of the Au( 111) surface was almost completely covered by the Cu adlayer with the (43 x 43)R30” structure, several types of phase boundary can be seen in Fig. 17(a). Figure 17(b) shows a model structure of the phase boundary marked by arrow (a), in which two (63 x 1/3)R30” domains are shifted by a half position along the direction indicated by the arrow. In the model structure shown in Fig. 17(b), it was assumed that the solid circles represent Cu atoms. The same (43 x 43)R30” structure was also found by in situ AFM [ 1151. However, a coulometric curve obtained simultaneously with CV shown in Fig. 16(b) showed that the ratio of the charges consumed during the first and second UPD processes was roughly 2: 1, suggesting that the surface coverage of Cu was about 2/3 after the first UPD peak [ 1131. According to the model structure shown in Fig. 17(b), the surface coverage must be l/3 because of the (43 x 43)R30” structure. A similar coulometric curve was previously reported by Kolb [3] who assumed a honeycomb structure with the surface coverage of 2/3, twice as large as that corresponding to the simple (63 x 1/3)R30” structure.

I

I

0

5

1 10 nm

Fig. 17. High-resolution STM image (a) and a model structure (b) of a copper adlattice on Au( 111). From [ 1131.

STM

in Electrolytes

147

This discrepancy was carefully investigated by Shi and Lipkowski using a chronocoulometric technique. They measured the Gibbs excess of Cu adatoms and that of coadsorbed sulfate (S04-’ ) as a function of the electrode potential, and concluded that Cu adatoms are packed to form a honeycomb (63 x 43)R30” structure with the center of each honeycomb cell occupied by a sulfate ion [ 116,117]. Finally, Toney et al. examined the above system using an SXS technique [ 1181, and concluded that the Cu atoms form a honeycomb lattice and are adsorbed on threefold hollow sites with sulfate ions located at the honeycomb centers. They also concluded that three oxygens of each sulfate ion bond to Cu atoms. According to all of the results described above, the most reliable model structure can be presented as shown in Fig. 18, in which the top (a) and perspective (b) views are given. This model structure is essentially the same as that shown by Toney et al. [ 1181. According to the model structure, the corrugation observed by in situ STM and AFM must be considered to be due to the coadsorbed sulfate ions, and not the Cu atoms. The study of the UPD of Cu on Au( 111) was a very important lesson for understanding limitations and strengths of various in situ techniques. It is clear that STM and AFM cannot distinguish chemical species.

(W

Fig. 18. Top (a) and perspective views (b) of the coadsorbed Cu2+ and SOd2-adlayer on Au( 111). Although intensive studies have been carried out for the UPD of Cu on (ii) Cu UPD on Pt(ll1). Au( 111) as described above, Pt is also an important substrate for understanding the UPD process of Cu. We first reported an in situ STM image obtained with Pt( 111) in a 0.05 M H,SO, + 5 mM CuSO, solution in the absence of halide ions, such as chloride [ 1191.

K. ltaya

148

g O--10

-

-20

-

-30

-

-40

-

-5oh

0.4

1 0.5

I 0.6

I 0.7 E/V YS RHE

I 08

I 0.9

ii

lb

nm

Fig. 19. Cyclic voltammogram (a) and high-resolution STM image (b) of a copper adlayer onPt(ll1). From[119]. Figure 19(a) shows a CV obtained with a well-prepared Pt( 111). The first UPD peak at 0.68 V is very narrow with a half-width of ca. 5 mV. The second wave is quite broad with the half-width of 3040 mV. The two stripping peaks appear with nearly equal heights having the half-width of ca. 5 mV [ 1191. It is well known that the shape of CV for the UPD of Cu on Pt( 111) is strongly affected by the presence of halide and organic molecules [ 120- 1231. In our previous paper [119], it was reported that an ordered structure with the (43 x d3)R30” symmetry appeared at a potential half way between the two stripping peaks as shown in Fig. 19(b). However, a coulometric curve This result is consistent with that observed by ex situ LEED [122]. indicated that the coverage of Cu after the first peak was approximately 2/3 [ 1191. From the result for the UPD of Cu on Au( 11 l), which was described above, it can be assumed that the same honeycomb (43 x 43)R30” structure as shown in Fig. 18 forms in the “pure” H,SO, solution in the absence of halide such as chloride. More detailed STM studies have been carried out by Ito et al. [ 124- 1261. However, it is surprising for us that they observed the (43 x d3)R30” structure at potentials far more positive than the UPD peak. A (43 x 47) structure was found in the entire potential region of UPD. The (63 x 43)R30” structure observed before the UPD [125] cannot be interpreted as being due to bisulfate anions adsorbed on Pt( 11 l), because only the (43 x 47) structure was observed for (bi)sulfate on Pt( I1 1) [55,56].

STM

Recently, Ross and coworkers

in Electrolytes

149

have intensively investigated the Cu UPD on Pt( 111) using SXS and

other techniques [ 127-1291. They proposed, based on their SXS analysis, a (43 x 43)R30” structure similar to the structure observed on the Au(ll1) electrode shown in Fig. 18. However, it must be pointed out that they found this structure at potentials more negative than the second peak, i.e., after completion of the UPD reaction. In this potential region, Ito and coworkers found the (43 x 47) structure [ 1261. There are many conflicting reports on the UPD of Cu on Pt( 111) as briefly described above. Further work is needed on this system. In order to evaluate the role of halide in the UPD of Cu on Pt( 1 1 1), it seems necessary to investigate the UPD process on well-defined halide-modified Pt( 111) as the first step. The iodine-modified Pt electrodes (I-Pt) is one of the most promising substrates, because the structure of I is well-defined as described already. Hubbard and coworkers carried out an ex situ LEED work on the UPD of Cu on I-modified Pt( 111) (I-Pt( 111)) [ 1301. They observed two structures during the UPD of Cu on a Pt(l11)(47 x 47)R19.1”-I electrode. A (3 x 3) structure was found as the first stage, and a (10 x 10) structure was proposed after completion of the UPD, because of an appearance of a set of three spots near the position of a genuine (43 x 43)R30”. Ito and coworkers carried out an in situ STM, reporting a (d3 x d3)R30” structure [ 1241. We have recently investigated the UPD of Cu by complimentary use of LEED and in situ STM 1651. We did not find the (3 x 3) structure reported previously [ 1301. Our LEED and in situ STM revealed that the adlayer structure is c(p x 43-R30”), the so-called (p x 43) with a p value of ca. 2.6, which is different from the (43 x 43)R30” and (10 x 10) structures. This structure is similar to that for the iodine adlayer on Au( 111) as described above. Again, it is clear that complimentary use of LEED and in situ STM will be a reliable technique for revealing the role of other halide, such as chloride and bromide, fortheUPDofCuonPt(l11). (iii) Ag UPD on Au(ll1). We investigated the UPD of Ag on Au( 111) on purpose, because the lattice constants of Ag and Au agree to within 0.3 % [ 13 I]. This system can be considered as a model system for the growth of metal on a foreign substrate (Au). We obtained a very characteristic CV in a H,SO, solution as shown in Fig. 20. It can be seen that the UPD of Ag occurs in at least three different steps on Au( 111). The first UPD with a characteristic shape is commenced at ca. 1.2 V. A very sharp peak with a half-width of ca. 5 mV appears as the main peak, accompanying a broad shoulder. A small sharp peak can also be seen after passing the main peak. As indicated in Fig. 20(a), the (43 x -\j3)R30” structure was found in the potential range between 0.85 and 1.1 V. However, this structure disappeared and no well-resolved atomic image was obtained in the second UPD region (II), suggesting that additional Ag adatoms deposited by the second UPD collapse the ordered structure. The surface coverage of Ag at the end of the second UPD was ca. 2/3 as estimated by the coulometry [ 13 11. However, a very clear (1 x 1) structure was observed at the end of the third peak (III), indicating that the first close-packed Ag layer is epitaxially formed in the third UPD process. The same result was reconfirmed in our later paper [ 1321. In a HClO, solution, the UPD of Ag also occurs in three steps as shown in Fig. 20(b), although the first UPD peak is broader than that in H,SO,. A high-resolution STM showed that a (4 x 4) structure is formed after the first UPD peak in Fig. 20(b) [ 1321. The atomic image of the (4 x 4) structure was

150

K. ltaya

40

III

(a)

20 cu k po ..-20

-40

0.6

0.8 E/m.

1.0 1.2 RHE

0.6

0.8 E/w.

1.0 1.2 RHE

Fig. 20. Cyclic voltammograms for Ag UPD on Au( 111) in H,SO, (a) and in HClO, (b). From [131,132]. consistently observed at potentials between 0.85 and 1.1 V, while it disappeared in the second UPD region in the same manner as was observed in H,SO,. When the electrode potential passed the third UPD peak, a clear (1 x 1) structure with an interatomic distance of 0.28 was observed, indicating that the first close-packed Ag adlayer is also epitaxially formed in the HClO, solution. We also investigated the UPD of Ag in a HF solution [ 1331. The structure of the first UPD adlayer was the same, (4 x 4), as that found in the HClO, solution. These results strongly suggest that sulfate or bisulfate anions are strongly coadsorbed on the surface of Au( 111) with the Ag adlayer in H,SO,. The appearance of the same structure in HClO, and HF indicates that neither of the two, ClO; and F-, anions is strongly attached on Ag-Au( 111). On the other hand, Gewirth and coworkers reported entirely different structures using in situ AFM. Although they investigated effects of anions of supporting electrolytes such as H,SO,, HCIO,, acetic acid, and nitric acid on the adlayer of Ag, a (3 x 3) structure was found after the first UPD peak in H,SO, [ 1341. The same structure was also found at potentials slightly positive than the bulk region [ 1341,while we found the (1 x 1) epitaxial layer [ 131,132]. Note that the sharp peaks (III) observed in both H,SO, and HClO, were assigned to the bulk deposition in the previous paper [ 1341. Our results described above are not consistent with those obtained by in situ AFM. Gewirth and coworkers also investigated this system using ex situ LEED, and reported (3 x 3) and (5 x 5) structures [135]. Wieckowski and coworkers investigated the UPD of Ag on Au( 111) in HF to minimize the effect of

STM

in Electrolytes

151

I :l

Ag :

l

Au:

(-J

nm

Fig. 21. Cyclic voltammogram (a), STM image (b), and a model structure (c) for the Ag UPD on an iodine-modified Au( 111). From [45]. specifically adsorbed anions using ex situ LEED. They found several adlayers with structures including (3 x 3), (5 x 5), and (6 x 6) but not the (4 x 4) structure found in our study as described above [ 1361. Lorenz and coworkers investigated the Ag deposition on Au( lOO), revealing only the formation of a well-ordered Au( lOO)-( 1 x l)Ag adlayers [ 1371. We also investigated the UPD of Ag on an iodine-modified Au( 111) to understand the role of coadsorbed anions [45]. Figure 21(a) shows a CV of the UPD obtained in a 0.1 M HClO, + 1 mM AgClO, solution. It was rather surprising that a similar feature was found in the CV obtained with lAu( 111). Three sets of UPD peaks in Fig. 21(a) are seen to be similar to those found on bare Au( 11 I) shown in Fig. 20, although the shapes and heights of the peaks are quite different for the electrodes with and without iodine modification, As described above, the structure of iodine on Au(ll1) is the incommensurate c(p x 43R-30”). We found the so-called (5 x 43) structure at potentials before the UPD, which is equivalent to ~(2.5 x 43R-30”) [46]. Two different structures of (3 x 3) were found after the formation of the first UPD layer [45]. The predominant (3 x 3) structure observed is shown in Fig. 2 l(b). It is seen in Fig. 21(b) that the atomic rows are almost completely parallel to those of Au( 111). An ordered corrugation with an almost uniform height of ca. 0.06 nm is seen along the atomic row labeled A in Fig. 21(b). On the other hand, two different kinds of atoms appear alternately along the row labeled B.

152

K. ltaya

Figure 21(c) shows a depiction of a (3 x 3) structure. In this model, four Ag atoms underneath tin iodine layer are positioned directly on top of the Au atoms making up the four ;c,rner\ i)* tilt: :.‘nl’ ~,‘Ci’ The other four Ag atoms in the unit cell are located at bridge sites. It is asxumed ~irat todtne atom iu,;’ simply sitting at the 3-fold hollow sites in the Ag layer. Therefore, the three iodine atom Ioc~tted WA each comer of the unit cell appear as image features higher than the iodine atoms sitting at the center 01’ the unit cell. The STM data are in agreement with this model. In the other (3 x 3) structure, the fout corners of the unit cell appeared as bright spots with an equal corrugation height. Weaker spots were found in between the two bright spots. This observation suggests that the second type of the I 3 x 3) structure may correspond to a previously proposed (3 x 3) structure for the LJPD of Ag on an i-Ptt 1I i ) electrode[ 1381. It wasalsoconcludedin the samepaper[45] that the structureof iodineatomson a bulk depositedAg was(43 x ?j3)R30”at a very negativepotential. Although this structureshouldbe refined by c(p x 43R30”) as describedin the recentpapers[64,71], the (43 x 43)R30” structure is confirmed to exist at cathodicpotentialsby e.usiruLEED [64] andSXS [71]. (iv) Ag UPD on Pt(ll1). Although numerousinvestigationshave previously beencarried out for the UPD of Ag on bare Pt electrodes[ 139-1411,the CV featuresare complicatedparticularly by the oxidation of Pt substrates. The first UPD usually occurs at potentialsnear the oxidation peak of Pt substrate.Even with a singlecrystal of Pt( 111) El Omar ef al. found severalbroad peaksfor the first UPD [ 1391.However, we clearly demonstratedfor the first time that substrateoxidation doesnot occur on a carefully preparedPt( 111) surface in the potential region of the first UPD of Ag [ 1421. The oxidation of a well-definedPt( 111)takesplaceonly at potentialsmorepositive than 1.2 V vs. RHE in a 1 M H,SO, solution [25,142]. Lesswell-defined Pt(ll1) electrodesusually show a smalloxidation peakat ca. 1.OV, which is dueto the oxidation of Pt atomslocatedat defect sitessuchassteps.

0

i

3

nm

Fig. 22. Cyclic voltarnmogram(a) and STM image (b) for the Ag UPD on Pt( 111) in H,SO,. From [142].

STM

in Electrolytes

153

Figure 22(a) shows a typical CV of a well-defined Pt( 111) surface in 1 M H,SO, in the presence of 1 mM Ag,SO,. It is surprising that two sets of well-defined deposition and stripping peaks are seen prior to the bulk deposition of Ag shown by the dashed line. The redox couple of the second UPD process appears at a potential ca. 50 mV more positive than the equilibrium potential of Ag/Ag’ of 0.63 V vs. RHE. This result is consistent with that reported by El Omar et al. [139]. However, the peak C, observed at 1.07 V for the first UPD is extremely narrow with a half-width of ca. 10 mV. The corresponding anodic peak A, shows an extraordinarily sharp spike at 1.08 V followed by a relatively broad peak. The characteristic shape of CV was not altered by repeating cycles, indicating that the substrate oxidation did not occur in the potential range shown in Fig. 22(a). The characteristic shape of the first UPD peak changed dramatically to broad small peaks even on slightly disordered Pt( 111) as shown in the previous paper [ 1391. In a slightly dilute H,SO, (0.05 M) solution, the stripping of the first UPD gave a single sharp peak without such a sharp spike as seen in Fig. 22(a). Note that the shape of the CV reported by Wieckowski et al. is also different from that shown in Fig. 22(a), although they found a sharp spike during the stripping of the first UPD [143]. It was found by coulometry that the total charges consumed for the first and second steps were both nearly equal to ca. 240 uC/cm2 [ 1421, suggesting that the first and second adlayers are nearly close-packed Ag monolayers. A very clearly ordered structure appeared over the Pt( 111) terraces after the completion of the first UPD layer. Fig. 22(b) shows a typical high-resolution STM image acquired at 0.9 V. The same atomic features were consistently observed in the potential range between 0.6 and 1.0 V. The interatomic distance was found to be ca. 0.28 nm, and the atomic rows were parallel to the [llO] direction of the Pt( 111) substrate. The above result strongly suggests that the first adlayer of Ag is epitaxially formed

4030-

207 IOE < o3 -lO-2o-3o-w

I

0.6

I

0.7

0.8

0.9

1.0

1.1

E /Vvs.RHE

Fig. 23. Cyclic voltammogramfor the Ag UPD on an iodine-modifiedPt( 111). Scanrate; 2 mV/s. From [43].

154

K.

ltaya

on the Pt( 11 I ) surface. On the other hand, we also found an atomic image very similar to that snown in Fig. 22(b) after completion of the second UPD. This result might suggest that Ag( i I I )-t I s I i W;I>alxo formed in the second UPD, although no Moire pattern due to the lattice mismatch between Pt (().27h nm) and Ag (0.289 nm) was found. The interatomic distance of Ag might have gradually changed from 0.278 nm to 0.289 nm with increasing number of the Ag layers. Hubbard and co-workers have intensively carried out ex situ investigations of the LJPD of Ag on iodine-modified Pt( 111) electrodes (I-Pt) using the UHV-EC technique employing LEED and the recently developed angulur distribution Auger microscopy (ADAM) [ 138,144-1463. We have examined this system using in situ STM 143). Figure 23 shows a typical CV observed at the Pt( 1 I I)-( $7x1’7)R19.1 ‘-I electrode in 1.0 M HClO, in the presence of I mM AgCIO, 1431. Three sets of UPD peaks can be seen, before the bulk deposition, at ca. 1.05, 0.79. and 0.63 V IVY. RHE, respectively. Although the CV shown in Fig. 23 is very similar to that reported previously [ 144.1451, the UPD peak heights seen here are much sharper and higher than those reported previously. STM images observed at 1.05 V before the UPD consistently showed the (47 x <7)R19.1”-1 structure. Figure 24(a) shows an STM image observed at 0.95 V after the completion of the first Ag adlayer. It is clearly seen that the four corners of each unit cell appear as bright spots with an equal corrugation height. Weak spots can be seen between neighboring bright spots. One additional spot is located in each unit cell. All these weak spots showed a nearly equal corrugation height. The atomic rows are parallel to the [ 1101 direction.

2

3

Figure 24(b) shows a depiction of the (3x3) structure which

4

5 nm

Fig. 24. High-resolution STM image obtained at 0.95 V (a) and a model structure (b) for the Ag adlayer on an iodine-modified Pt(ll1). From [43].

STM

in Electrolytes

155

was previously proposed by Hubbard and co-workers

based on their LEED analysis [ 138,144]. This model can explain qualitatively the STM image shown in Fig. 24(a). Note that the (3x3) structure was observed in the potential range between 0.95 V and 0.85 V. When the electrode potential was scanned past the second UPD peak, a completely different atomic image was observed in the potential range between 0.75 V and 0.65 V. The atomic rows were rotated by 30” (+ 2”) with respect to the Pt( 111) lattice [43]. The observed interatomic distance was nearly equal to that expected for the (43

x d3)R30” structure. All spots in the images observed at potentials between the second and third UPD peaks appeared with a nearly equal corrugation height of ca. 0.05 nm. No clear long-range ordered pattern appeared in the STM images obtained in the present experimental conditions. Hubbard and coworkers reported LEED patterns for the adlayers after the second UPD peak and also after the bulk deposition of Ag [ 138,144] as well as those for an iodine layer on a bulk single crystal of Ag( 111) 1991. The LEED patterns for all cases showed a triangular splitting. The most recent paper proposed a new model based upon the LEED pattern and Auger distribution data [ 1461. Rotation of the Ag adlattice by 27.2” brings the Ag lattice into registry with the Pt( 111) substrate every 243 Pt-Pt distances and every 9 Pt-Pt distances in the directions of R30”, and R60”, respectively [ 1461. However, we did not see in our STM images any clear long-range ordered patterns expected from the new model. Although a further investigation may be needed for interpretation of the STM image, we anticipate that incommensurate structures such as c@ x -/3-R30”) might be formed after the second UPD, which should result in the triangular splitting in LEED patterns, as described in our previous papers [46,64]. (v) Hg UPD on Au(ll1). Chen and Gewirth investigated the UPD of Hg on Au( 111) using in situ AFM in solutions containing sulfate, nitrate, acetate, and perchlorate ions [ 1471. They found a

$

40

20

30

15

20

IO

10

5

$3

0

IL -10

-5

-20

-10

-30

.15

-40

-20 0.8

1.2

1.0 E/V

vs.RHE

Fig. 25. Cyclic voltammogram for the Hg UPD on Au( 111) in H,SO,. Scan rate; 2 mV/s. From [54].

156

K. ltaya

((3/2)&3 x 2)

(4xv3)

Fig. 26. Typical two STM images obtained at 1.05 V and corresponding model structures. The dashed line in (b) indicates the unit cell of (43 x d19). From 1541. hexagonal overlayer with a 0.58 nm spacing in sulfate, nitrate, and perchlorate electrolytes at potentials just

prior to bulk deposition. In the potential region of bulk deposition, a different hexagonal structure

was found with a 0.58 nm spacing, which was attributed to that of the alloy between Hg and Au [ 1471. We also investigated the UPD of Hg on Au( 111) using in situ STM. It was shown that the EC response was strongly influenced by the crystal orientation of Au single crystal as well as the anions of sulfate and perchlorate [54]. On Au(100) and Au(l IO), CV curves were very similar either in 0.05 M H,SO, or in 0.1 M HCIO,. However, entirely different CVs were observed on Au( 111) in the two electrolyte solutions. Several very sharp peaks were found for the first UPD in H,SO, as shown in Fig. 25, while symmetric, but broad peaks appeared in HClO,. An extraordinarily sharp spike was found at 1.20 V, followed by the first main UPD peak appearing as separated double peaks during the potential scan from 1.3 V in the negative direction. Two small peaks can be seen at 1.1 and 0.9 V in the doublelayer region. In the reverse scan, two corresponding small anodic peaks are seen at 1.03 and 1.12 V, respectively. The stripping of the UPD layer appears as a single sharp peak at 1.19 V, followed by an additional sharp spike at 1.21 V. Our STM results can be summarized as follow [54]: a)The so-called (43 x 47) structure due to the adsorption of sulfate was clearly observed at potentials prior to the first UPD reaction. b)Two different structures were observed on the same terrace after completion of the

STM

in

Electrolytes

157

formation of the first UPD layer. Two images are shown in Fig. 26 with corresponding structural models. In Fig. 26(a), the rectangular unit cell corresponds to a lattice symmetry of ((3/2)43 x 2). On the other hand, two types of parallel rows are seen along the A direction with different corrugation heights, and the darker row exists in the middle of the two neighboring bright rows as shown in Fig. 26(b). The model structure of (4 x d3) is shown in Fig. 26(b). c)In a 0.1 M HCIO,, a (2 x 43) rectangular structure was found. Although we tentatively assumed that each spot corresponds to a Hg atom, it is clear that coadsorbed sulfate or bisulfate must be taken into account in understanding the adlayer structure. Li and Abruna recently investigated the UPD of Hg in a H,SO, solution using the SXS technique [ 1481. They found a (43 x 419) structure in agreement with our STM image shown in Fig. 26(b). The unit cell shown by the dashed line in Fig. 26(b) corresponds to the (43 x 419) structure as defined by Li and Abruna (see Fig. 5 in ref. 148). Our unit cell contains two adsorbates, while their X-ray data indicate three adsorbate units in each unit cell. They proposed a model consisted with a sulfate or bisulfate on top of the mercury. This situation is similar to that in the UPD of Cu on Au( 111) shown in Fig. 18. However, it is difficult to interpret our STM image based on their model. As shown in Fig. 26(b), only one dark row exists in the middle of the neighboringbright rows. The structuremodelproposedon the basisof SXS shouldgive a different appearancein STM. Further work is neededbefore a final conclusioncan be drawn. (vi) UPD of Other Metals. The review article by Gewirth and Niece is useful for find the literature on UPD of various metalson different substrates[ 111. Here we only describea few selectedUPD systemsandrelatedadsorptionprocesses. Stickney and coworkershave beentrying to developa new methodologyfor preparingthin films of semiconductorssuch as CdTe, called electrochemicalutomic luyer epitaxy (ECALE), using UPD processes[ 149-1531. Depositsin ECALE are formed by alternatingthe UPD of atomic layers from separatesolutions. If this techniqueis establishedin solutions,there is a great possibility that it can replace more conventional processessuch as atomic layer epitaxy performed by chemical vapor depositionandmolecularbeamepitaxy in gasphase.CdS monolayerswere alsosuccessfullyprepared on Au( 111) [ 1541. STM plays an important role in characterizingeach atomic layer with atomic resolution[ 150-1541. Catalytic oxygen reductionin the presenceof UPD specieshasbeeninvestigatedparticularly for Pb, Tl, and Bi on Au electrodes[ 1lo]. It is important to understandelectrocatalyticprocesseson an atomic scale. Our previouspaperdescribedthe effect of UPD of Cu on oxygen reduction at Pt( 111) basedon atomic structures of Cu determinedby STM [123]. Chen and Gewirth first determined adlayer structuresof Bi on Au( 111) usingin situ AFM [ 1551,reporting two structureswhich are a (2 x 2) and an incommensuratestructure, respectively. More detailedstructural changeswith electrodepotential have beeninvestigatedby complimentaryuseof SXS andSTM [ 1561.They concludedthat the first and secondBi adlayershave a (2 x 2) structure and a uniaxially incommensuratestructure, @ x -\/3), respectively. The structuresdeterminedby SXS and STM were in agreementwith those determined previously by AFM [ 155,156]. It is interestingto note that the (p x 43) structurehas beenobservedfor iodine on Au( 111)andAg( 111) asdescribedabove [46,61-64,711. In the caseof iodineon Au( 11l),

158

K. itaya

STM images acquired for the (p x 43) structure were always flat with no indication of height difference among the iodine adatoms [46,62]. We assumed that all iodine atoms are located on lines bisector lines, across two-fold bridge sites in the @ x 43) structure. Using a hardball contact model, the height difference was calculated to be 0.008 nm, which could hardly bc detected in STM [62]. However, a clear Moire pattern was observed for Bi on Au( 1 I l), which was attributed to the difference in height between Bi atoms in or near atop sites on Au surface and those that are in or near threefold hollow sites [ 1561. According to the models proposed for the (p x 43) [61,62,156], there is no adatoms in or near atop sites on Au as long as all adatoms are located on the bisector line. A further study is needed on the unit cell for Bi on Au( 111) to explain the Moire pattern observed by STM. For thallium (Tl’), Clavilier et al. compared the UPD of H’ and Tl’ on Pt( 1I I) in the presence of specifically adsorbed sulfate anions using voltammetry [ 1571. Adzic et al. recently carried out an SXS study on structures of Tl’ on Pt( 111) in various electrolyte solutions [ 1581. In a sulfuric acid solution, the (43 x 7/3)R30” structure was found after the first UPD peak. Our preliminary in situ STM also showed a clear (43 x 43)R30” image in the same potential range examined by Adzic et al. [ 1591. The coadsorptionof sulfate or bisulfate during the UPD was shown by infrared adsorptiontechniques [126,160]. An SXS study was also extendedto the UPD of Tl’on Ag(100) [161]. In situ STM was appliedto determineadlayer structuresof Tl’ on Au( 111)in an alkalinesolution[ 1621. The UPD of Pb is alsointensively investigatedon varioussubstrates,suchasAu( 111) [163,164] and Ag singlecrystal electrodes[ 165-1691.Schmidtet al. madean interestingcomparisonbetweenadlayer structuresof Pb on Ag( 100) and Au( 100) [ 1701. A c(2 x 2) structure on Ag( 100) and c(2 x 2) and ~(342 x 42)R45” structureson Au(100) were found in the low and mediumcoveragerange by in situ STM [ 1701. Using in situ SXS and STM techniques,Adzic ef al. investigatedthe UPD of Pb on low index Pt surfaces[171]. A rectangular(3 x 43) structurewas found to form on Pt( 111). It was also noticedthat a different structure,(243 x 263)R30”, coexistednearthe stepedges,suggestingan orderdisorderphasetransition. No orderedstructureswere found on Pt( 100)and Pt( 1IO). It wasreportedthat the UPD layer of Cd on Au( 111) possesses three different structuresdepending on the electrodepotential, all of which showeda bandedmorphology[ 1721. Finally, Bubendorff et al. found an interestingUPD of Ni on Au( 11I), which occurredin a sulfamate solution [ 1731. They concludedthat sulfamateanionsadsorbedon Au( 111) promotethe UPD of Ni. The Ni layer was thought to be built up on the surfacedue to a complexationof Ni*’ by the sulfamate adlayeron Au( 111). We found a similarsituationin the adsorptionof aurouscyanide (AuCN) on the surfaceof Au( 111)from solutionscontainingKAu(CN), [ 1741.This systemmight not be classifiedas UPD in classicaldefinition, but the formation of the complex of AuCN on Au( 111) occurs at potentials positive with respectto the reversiblepotential. Such a complex existing on Au( 111) is an important precursorfor the bulk depositionof Au from KAu(CN),. Uosaki et al. reportedan STM study for the structureof adsorbedhexachloroplatinatecomplex on Au( 111) [ 1751. It is reasonablyexpectedthat suchadsorbedcomplexesareimportantintermediatesfor the bulk depositionof Pt. Although the above mentionedsurface complexesform nearly without charge-transferreactions, these systemsmust be consideredas UPD processes,becausethe simple UPD processesdescribedabove are also strongly affectedby the surfacecomposition,including speciessuchasadsorbedsulfateandhalide.

STM

in Electrolytes

159

C. Adsorption of Organic Molecules on Iodine-modified Electrodes STM has also made it possible directly to determine orientations, packing arrangements, and even internal structures of organic molecules adsorbed both on surfaces in UHV and at solid-liquid interfaces [ 176- 1791. For example, individual molecules and distinguishable molecular shapes of benzene on Pt(ll1) [180], coadsorbed benzene and CO on Rh(ll1) [181], naphthalene on Pt(ll1) [182], and copper phthalocyanine on Cu( 100) [183] have successfully been resolved with STM under UHV conditions. These results have stimulated a large number of STM studies of ordered molecular adlayers in UHV and air as well as at solid-liquid interfaces. A variety of experimental procedures have been reported for the preparation of ordered molecular adlayers on well-defined substrates including single crystals of metals and layered materials, such as highly ordered pyrolytic graphite (HOPG) and MoS, [ 176-1781. Alkylthiols have been intensively investigated on metals such as Au [ 1791, because the -SH group is known to be chemically attached to the Au surface through the formation of a covalent bond between S and Au atoms, producing denselypacked adlayers. On the other hand, it is well-known that simple physical adsorption can also provide ordered adlayers of molecules, such as liquid crystals and n-alkanes on inert substrates such as HOPG and MoSz [ 177,178]. The Langmuir-Blodgett technique has long been applied to many amphiphilic molecules, in which ordered monolayers formed at the water-air interface can be transferred onto various substrates [ 1771. From the EC point of view, the adsorption of organic molecules at electrode-electrolyte interfaces can be considered as one of the most promising approaches not only for the preparation of ordered adlayers, but also for elucidating the role of properties of adsorbed molecules and the nature of electrodeelectrolyte interfaces [2,4,184]. In spite of a large number of reports describing the observation by STM and related techniques, such as AFM of adsorbed organic molecules in UHV, air, and organic liquids, only a few in situ STM studies have been carried out for organic molecules adsorbed at electrode-electrolyte interfaces under EC conditions.

Recently, pioneering studies have shown high-

resolution images of molecules such as DNA bases (adenine, guanine, and cytosine) adsorbed on HOPG and Au( 111) in electrolyte solutions [ 185 1881. In another study, xanthine and its oxidized form [189] and porphyrins [ 1901 were found to form ordered adlayers on HOPG. Further, the orderdisorder transition in a monolayer of 2,2’-bipyridine on Au( 111) was reported as a function of electrode potential [ 1911. More recently, the adsorption of thymine on Au single crystal electrodes has been investigated [ 1921. Although a number of successful in situ experiments using STM and AFM have been performed to determine the structure of organic adlayers, HOPG, similar layered crystals, and Au electrodes have almost exclusively been used as the substrate. Therefore, the role of interaction between organic molecules on one hand and substrates on the other in ordering processes has not yet been fully understood. We have long been interested in finding a more appropriate substrate to investigate the adsorption of organic molecules. Recently, we disclosed a novel property of iodine-modified electrodes for the adsorption of organic molecules [ 193-1981. It was found, for the first time, that a water soluble porphyrine, 5, 10, 15, 20- tetrakis(N-methylpyridinium-4-yl)-21H, 23H-porphyrine (TMPyP), formed highly-ordered molecular arrays via self-ordering on the iodine-modified Au( 111) electrode in HClO,

160

K. ltaya

solution II93]. As described in the preceding section, the iodine adlayers are thought to protect metal surfaces from oxidation and contamination in the ambient atmosphere. providing an easy method ot

preparation of well-defined surfaces [2,4,36-381. However, it should now be recognized that the I-Au (I I 1) electrode is one of the most promising substrates for the investigation of organic molecules in solution. Indeed, WC discovered this electrode, with great generality, to bc suitable on which to form highly-ordered adlayers not only TMPyP but also various other molecules such as crystal violet [ 1941. It was also demonstrated in our recent papers that various iodine-modified metal electrodes such as Ag(lll) [1961, Pt(l 11) 11971, and Pt(100) [I981 can also be employed as a substrate on which to investigate the adsorption of organic molecules. In this section, we briefly summarize in sitar STM of organic molecules adsorbed on the I-modified electrodes. (i) TMPyP on I-Au(ll1). The experimental procedure was rather simple. A well-defined Au( 1I I ) surface prepared by the flame-annealing and quenching method was immersed into 1mM IU solution for several min and then thoroughly rinsed with 0.1M HCIO, solution. The iodine modified electrode. thus prepared, was installed in an EC cell containing a pure HClO, solution for in situ STM measurements. Under potentialcontrol, the iodineadlayerstructurescanbe determinedby in .situSTM asdescribedabove [45,46,62]. After achieving an atomicresolution, a dilute solution of TMPyP was injectedinto the HCIO, solution. After the addition of TMPyP, STM imagesfor the iodine adlayer becameunclear within the first S-10 min, becauseof the adsorptionof TMPyP. and then orderedadlayersbecamevisible, extendingover atomically flat terraces[ 1931. usually

(a)

nm

nm

Fig. 27. High-resolutionSTM imagesof TMPyP on an I-modified Au( 111) in HCIO,. From [ 193,194].

STM

in Electrolytes

161

Molecular orientations, packing arrangements and even internal molecular structures of the adsorbed TMPyP molecules can be seen in high-resolution STM images. Two typical STM images are presented in Figs. 27(a) and 27(b). These high-resolution images directly demonstrate that the flat-lying TMPyP molecule can be recognized in the images as a square with four additional bright spots. The shape of the observed features in the image clearly corresponds to the chemical structure of TMPyP molecule. The characteristic four bright spots located at the four comers of a square correspond to the pyridinium units of TMPyP. The center to center distance between the bright spots was found to bc 1.3 nm measured diagonally, which is nearly equal to the distance between two diagonally located pyridinium units. In addition to the internal structure, the STM image shown in Fig. 27(a) reveals details of the symmetry and the packing arrangement. It can be seen that there are three different molecular rows marked by arrows I, II, and III. In the row marked I, all TMPyP molecules show identical orientation with an intermolecular distance of ca. 1.8 nm. An alternate orientation can be seen along the row II in which every second molecule shows the same orientation. On the other hand, the rotation angle of ca. 45” can be recognized between two neighboring molecules in row III. The well-ordered TMPyP arrays were found to extend over atomically flat terraces as shown in Fig. 27(b), which was obtained in areas involving monatomic step edges. It is clearly seen that the same molecular feature as that seen in Fig. 27(a) extends over the lower and upper terraces. The ordered structure could be seen even on the relatively narrow terrace in the upper part of the image. It is also surprising to find that individual TMPyP molecules exist very near the monatomic step. The result shown in Fig. 27(b) indicates that the entire surface of the I-Au( 111) is almost completely covered by ordered TMPyP molecules even near the end of the terraces. The same structure was consistently observed in a potential range between 0.6 and 1.0 V vs. RHE in 0.1 M HClO,. Figures 28(a) and 28(b) illustrate a structural model showing a top view and a side view of the ordered TMPyP adlayer on the iodine adalyer on Au( 11 l), respectively. We have also investigated the adsorption of TMPyP on a

TMPyP

Fig. 28. Illustrative depictions of the structure of TMPyP adlayer on I-Au( 111): (a) topview, (b) side-view. From [ 1951.

K. ltaya

162

well-defined Au( 111) in the absence of iodine adlayer [ 1951. After achieving an atomic resolution for the Au( 1I I)-( Ix 1) structure in 0.1 M HCIO,, a dilute solution of TMPyP was injected into the soiution in a manner similar to that described above. Although the TMPyP molecules adsorbed directly on Au( I 11) could be seen by STM, the adsorbed molecules did not form ordered adlayers. Disordered adlayers formed on bare Au( 11 I) suggest that strong interactions including chemical bonds between the Au substrate and the organic molecules prevented self-ordering processes from occurring, which must involve surface diffusion of the adsorbed molecules. The surface diffusion of the molecules adsorbed on bare Au was found to be very slow. Relatively weak van der Waals type interactions between the hydrophobic iodine adlayer and the organic molecules could be the key factor promoting self-ordering processes on the I-Au( 111) substrate. Initial stages of adsorption. The in situ STM study alsoallowed us to evaluate dynamic processes of the formation of ordered TMPyP adlayers on I-Au( 111). In order to follow time-dependent processesin the growth of ordered phases,a more dilute solution of TMPyP (ca. I x 10.’ M) was employed. Approximately 15min after addingthe TMPyP solution,flat-lying TMPyP moleculesbegan to be observedwith its characteristicsquareshapeas shown in Fig. 29(a). The imageshown in Fig. 29(b) was acquiredat nearly the samelocation 20 s after Fig. 29(a) was obtained. Many flat-lying TMPyP moleculescan be recognizedas squareswith four bright spotsat the comers of each square. The moleculeswere aligned one-dimensionallyforming molecularchains at the early stage of the formationof the orderedphase. Adjacentmoleculesin eachchain were alignedwith eachother with a side-by-sideconfiguration. Two pairsof pyridinium ringsin the molecule seemedto be attached to thoseof the adjacentmolecules. From Fig. 29, it can be seenthat such chainsgrew relatively rapidly during the time interval of 20 s. The molecularchainsare partially straightover a short range. but they

nm ”

nm

2-O

Fig. 29. STM imagesof TMPyP moleculesasearly stageson I-Au( 111). From [ 1951

STM

in

Electrolytes

163

form long-twisted lines. During the continuous imaging after the acquisition of the image shown in Fig. 29(b), it was found that the relative location and shape of the chains in the images always changed rapidly. The intermolecular distance between two adjacent molecules in each chain varied in the range between 1.7 and 2.2 nm. These results strongly indicate that the adsorbed TMPyP molecules have a relatively large surface mobility on the iodine adlayer, which is in contrast to the result obtained on the bare Au( 111) surface as described above. The formation of the molecular chains seems to involve a weak coordination of counter anions, presumably perchlorate, compensating for the repulsive interaction between the positively charged pyridinium units. Isolated molecules usually appeared as simple spots without exhibiting the characteristic shape of TMPyP, which can be seen in the upper right-hand portion of the image in Fig. 29(a). This result suggests that the isolated molecules undergo not only lateral movements but also rotations in parallel with the surface. These twisted one-dimensional chains described above were found to become straight and form new two-dimensionally ordered arrays in a relatively small domain as a transitional phase. Figure 30(a) shows an example of images acquired in an initially formed small domain. It is surprising that the molecular chains with the same side-by-side configuration grew in the particular direction indicated by the arrow I’. forming nearly perfect straight lines. The orientation of each molecule in all chains is the same, i.e., the side-by-side configuration, and the intermolecular distance in each chain was found to be ca. 1.8 nm. The distance between the nearest-neighbor chains was ca. 1.8 nm. However, the molecular chains were shifted alternately by a half position of each molecule along the direction 1’. This alternation seems to have resulted from the minimization of the repulsive interaction between the pyridinium moieties. A model structure is illustrated in Fig. 30(b) with a rectangular unit cell. A unit cell lattice is outlined with the lattice parameters of a= 3.7f0.2 nm and b= 1.8kO.l nm. The adlayer structure in Fig. 30 corresponds to a surface concentration of about 3.0 x 10” molecules/cm’. which is slightly smaller than that found for the final structure. All molecules were already rotated by nearly 45 ” in the molecular row indicated by the dashed line in Fig. 30(a). In a relatively short period of time, all molecules along many particular rows shifted by a half position and rotated by nearly 45” with respect to the orientation of molecules in the unchanged rows similar to those along the dashed line in Fig. 30(a). Eventually, the adlayer shown in Fig. 27(a) appeared as the final structure thermodynamically most stable phase.

to form

the

The structural transformation in the TMPyP array described above was always found to occur when the domain size of the first stage as shown in Fig. 30(a) was smaller than several tens of nm. Interestingly, once the most stable structure appeared in a domain, it expanded over the entire area of atomically flat terraces with the final packing arrangement. Molecules in the disordered phase are progressively incorporated into the ordered domain with the final internal structure. This twodimensional growth can explain the formation of the highly-ordered TMPyP adlayer extending over the wide area. The phase transition can only be expected to occur when the domain size is less than several tens of nm, because the phase transition requires cooperative rotational and translational motions in each

K. ltaya

164

/

I’

0

io

I-5

nm

Fig. 30. STM image (a) and model structure (b) of a TMPyP array observed as a transitional phase. From [ 1951 ordered phase. The complicated force relationship in the molecular arrays consists mainly of two different interactions: perpendicular interactions between the electrode surface and molecules as a driving force for initial adsorption, and intermolecular interactions for self-organization leading to the formation of the ordered layer. Surface diffusion of adsorbed molecules seems to be a key factor for the ordering process in general. On bare Au( 1 11), TMPyP molecules can not diffuse because of a strong interaction between TMPyP and the Au surface. On the other hand, the relatively weak van der Waals type interaction between the TMPyP molecules and the iodine adlayer on the metal substrate allows the adsorbed molecules to diffuse on the iodine adlayer, resulting in the formation of an ordered molecular phase. In the expected intermolecular interactions, electrostatic interactions between the positively charged pyridinum units might play a significant role in the determination of the adlayer structure. These positive charges should be balanced by the coordination of anions such as perchlorate in solution. Although high-resolution STM images, such as those shown in Figs. 27, 29, and 30, did not show additional corrugations due to perchlorate, perchlorate anions are expected to be located near or between the pyridinium rings. Detailed investigations using different anions are of special interest. It is worthwhile to note that TMPyP formed an ordered structure on an iodine-modified Ag( 111) in an alkaline electrolyte solution containing IU (0.1 mM KI + 10 mM KF + 0.1 mM KOH at pH 10) [196]. In this case, I-, F- and OH- are expected to contribute to the charge neutrality at the TMPyP adlayerelectrolyte interface.

STM

(iii) TMPyP on I-Ag(ll1).

in Electrolytes

165

The I-Ag( 111) electrode is one of the most interesting iodine-modified

electrodes, because the structures of iodine adlayers on Ag( 111) have been well-characterized as described above [64,71], and also because bare Ag electrodes have frequently been used as the substrate for the investigation of the adsorption of organic molecules in solution [ 1841. It was reported in our recent paper that TMPyP molecules were also irreversibly adsorbed and formed highly ordered molecular adlayers on the surface of the I-Ag( 111) [ 1961. Figure 3 I (a) shows a high-resolution STM image acquired in an area of 15 x 15 nm, revealing clear internal molecular structures and molecular orientations in the ordered adlayer. The STM image includes the TMPyP arrays formed on two adjacent terraces separated by a monatomic step. An individual TMPyP molecule can be clearly recognized as a square with four additional bright spots at the corners. As described above, these bright spots can be attributed to the pyridinium units in the TMPyP molecule. The molecules along the [ 1121 direction can be seen to align with the same orientation, resulting in the straight molecular rows.

However, domains with the same molecular orientation are rather narrow, forming long stripes along the [I 121 direction. On the lower terrace, at the right-hand side of the image, striped domains consisting of three molecular rows can be seen. Although all molecules in each domain show the same orientation, the molecules in the adjacent domains were rotated by ca. 45” with respect to those in the middle domain. The structure of the TMPyP adlayer observed on the upper terrace is illustrated by the model shown in Fig. 3 1(b). The model represents two adjacent domains (A,B) in the TMPyP adlayer separated by domain boundaries indicated by the solid lines. In the domains denoted by A, the adsorbed molecules form an almost square lattice. On the other hand, a slightly tilted lattice is formed in domain 8.

nm

15

Fig. 31. High-resolution STM image of TMPyP molecules adsorbed on I-Ag( 111) (a) and a model (b) with two domains denoted as A and B. From [ 1961.

166

K. ltaya

The structure of the TMPyP layer on I-Ag( 111) is clearly different from that on I-Au( I I I ) shown II? Fig. 27. In general, many factors should be taken into account to explain the adlayer structure\ and th difference between those on different substrates. The interactions between adsorbed molecules and iodine adlayers should be different on Ag and Au. Note that the architecture of TMPyP seemed to be primarily determined by the structure of the underlying iodine adlayer. in the alkaline KI solution at pH 10, we observed the continuously varying series of iodine adlayers from the square (4.3 x -Y~R-30”) structure to the triangular (7/3 x 43)R30” structure as described previously (641. II was also found that one of three equivalent 43 directions was always unchanged during the electrocomprcssion 1641. Presumably, the TMPyP molecules align in this particular 43 direction on I-Ag( 11 1) 11961. (iv) TMPyP on I-Pt( 111) and I-Rh(l 11). In order to complete our understanding of the role of iodine adlayer in the formation of ordered molecular adlayers on I-Au( 111) and I-Ag( ill ). we have extended our investigation to other iodine-modified electrodes such as Pt( 1 I 1) and Rh( 11 1) [ 197 1. The commensurate (3 x 3) and (11’7x 67)Rl9.1” structures on Pt(ll1) 12,471 and the ($3 x ;3)R30‘ structure on Rh( 111) 158,601were previously well-characterized by ex situ LEED and in siru STM. In contrast to I-Au( 111) and I-Ag( ill). it was found that TMPyP did not form ordered adlaycrs on IPt( 111) nor on Rh( 111). Figure 32(a) shows an example of the STM images of adsorbed TMPyP molecules on an I-Pt( 111) electrode with the (3 x 3) iodine structure in 0.1 M HCIO, [ 1971. The surface of I-Pt( 11 1) was almost completely covered by flat-lying TMPyP molecules. The individual TMPyP molecules can be recognized as the characteristic square shape. However, it is clear that the adsorbed TMPyP did not form a highly ordered array on the I-Pt( 111) surface, although several domains with a short-range ordering can be seen in the STM

image.

A prolonged imaging in the same

Fig. 32. STM images of TMPyP on I-Pt( 111) (a) and I-Rh( 111) (b) in HClO,.

From [197,1991.

STM

in Electrolytes

167

area showed that the position of each TMPyP molecule was not altered, resulting in the same disordered adlayer. This observation indicates that the surface diffusion of the adsorbed TMPyP molecule on IPt( 111) is rather slow. TMPyP molecules seem to be more strongly adsorbed on I-Pt( 111) than on the other iodine-modified electrodes. Similar disordered adlayers have also been found on the I-Rh( 111) surface as shown in Fig. 32(b) [ 1991. It is clear that TMPyP molecules are flat-lying on I-Rh( 111). Each molecule can be recognized as a bright square with four additional spots, the same as that found on I-Pt( 111). Although the individual TMPyP molecules can be seen clearly, the adlayer is not highly-ordered and the molecules are randomly distributed over the surface. However, it is important to note that highly-ordered domains were produced by limited potential cycles in a cathodic region where the adsorption-desorption on TMPyP was expected to occur [ 1991. STM images obtained in a highly-ordered domain were almost the same as that observed on I-Au( 111) as shown in Fig. 28. The identical unit cell was identified for TMPyP on I-Rh( 111). (v) TMPyP on I-Pt(lOO). It should particularly be emphasized that the iodine layer plays a crucial role in the formation of highly-ordered TMPyP arrays. The relatively weak van der Waals type interaction on the iodine adlayer seems to be a key factor in the formation of ordered molecular arrays of such large molecules. However, the relation between the TMPyP and iodine adlayer structures is not fully understood as described above [ 195,196], because the iodine adlayer structures on Au( 1 I I) and Ag( 1 11) are complicated by a potential dependent compression in the adlayers [61,62,64,7 I]. In evaluating the structural relationship, we were particularly interested in the iodine adlayer on Pt( lOO), which is known to form a (42 x 542)R45” structure, because of its characteristic double-row structure [38,41,42,200-2031. The iodine atoms in two nearest bright rows along the direction of are located near on-top sites of Pt( loo), and in every third row of iodine atoms, which appears as a dark row, they are located on fourfold hollow sites [200,201]. We expected in our recent study that such a unique structure might affect ordering processes and adlayer structures of TMPyP [ 1981. In this section, we first describe the iodine adlayer with the (42 x 542)R45” structure, and then the adsorption of TMPyP on the I-Pt( 100). (a) Iodine adlayer structure. The iodine adlayer was prepared by cooling down an annealed Pt sample in iodine vapor. Relatively large domains with the (42 x 562)R45” structure similar to those reported previously were seen by STM [38,201]. conditions in iodine vapor.

The domain size strongly depended on cooling

Figure 33(a) shows a typical example of the high-resolution STM images obtained on the terraces. It is clearly seen that iodine atoms are aligned along the [ 1001 direction to form the characteristic doublerow structure. Each bright spot corresponds to an iodine atom adsorbed nearly at the on-top position of the substrate Pt atom, while neighboring dark stripes consist of iodine atoms located on fourfold hollow sites, which is consistent with the finding reported previously [38,201]. The nearest distance between iodine atoms along the bright rows was 0.4 nm, and the periodicity perpendicular to the bright iodine rows was 2.0 Z!Z0.1 nm. These results allowed us to define the unit cell with (42 x .542)R45” symmetry, which is shown in Fig. 33(a). A hard ball model illustrated in Fig. 33(b) is essentially the

168

K. ltaya

0

: 1 (near top site)

(-1.

I (four fold site)

.’ .Pt

Fig. 33. High-resolutionSTM imageof I-Pt( 100) (a) and a model of the (42 x 542)R45”

structure(b). From [ 1981. sameasthat proposedpreviously [200]. The dark circles in Fig. 33(b) correspondto the iodine atoms located on fourfold hollow sites of Pt(lOO) surface. Note that the (42 x 562)R45” structure was consistentlyobtainedat leastin the potentialrangebetween0.3 and 0.7 V in a pure HClO, solution, while Baltruschatef ~11. reportedthat the structureof iodine adlayer on Pt( 100) was potential-dependent in a solutioncontainingiodide ions[41]. Although we did not observeother structuressuch as a (42 x v’2)R45 [41] underour experimentalconditions, we did find somedefectssuch as irregularly formed, missingrows of iodine and antiphaseboundaries,similar to those found by Vogel and Baltruschat 12011.The defectsseemedto be formedmainly underiodine dosingconditionswith insufficient supply of iodine vapor. After the atomic resolution shown in Fig. 33(a) was achieved, a TMPyP solutionwasdirectly injectedinto the STM cell. After injectingthe TMPyP solution,the atomic imageof iodine adlayerwas blurred and then disappeared,and several minuteslater, a small ordered domainof adsorbedTMPyP appearedon the surface. The blurredSTM imagewas caused by adsorbed
of TMPyP.

TMPyP moleculeshaving surface mobility. A similar situation was encounteredin the adsorption TMPyP on I-Au( 111) [ 1951and I-Ag( 111) [ 1961at an initial stageas describedpreviously. Small domains of ordered TMPyP adlayers expanded over the entire area of atomically flat terraces, incorporatingrandomlyadsorbedmolecules. Interestingly, it was observedthat the ordereddomainof

STM

in Electrolytes

169

the TMPyP adlayer increased in area in the particular direction of [ 1001,which was parallel to the iodine atomic rows. Figure 34 shows typical examples of STM images acquired after completion of the formation of an ordered adlayer. Each bright spot now corresponds to a flat lying TMPyP molecule, and they make straight molecular rows along the [IOO] direction. Periodically aligned TMPyP molecules were observed not only in atomically flat terraces, but also near the step edges of the substrate. It is also recognized that TMPyP molecules form almost square lattices with the same molecular orientation. Figure 34(b) shows a higher resolution STM image of the TMPyP array. The square shape of the bright spots reflect the molecular structure of TMPyP. It is clearly seen that all TMPyP molecules are aligned with the same orientation and with the side-by-side configuration along the molecular rows in the direction of [ 1001. It is also important to note that a similar side-by-side configuration can be seen along the other [ 1001 direction perpendicular to the direction shown by the arrow sign in Fig. 34. The nearest neighbor distance between TMPyP molecules along the [ 1001direction was ca. 1.9 (+ 0.05) nm, which is almost identical to that found for an intermediate structure of TMPyP with the side-by-side configuration formed on I-Au( 111). It is slightly larger than 0.17 nm, which is the corresponding value for TMPyP molecules adsorbed on I-Ag( 111) [ 1961. It can be seen that the distance between neighboring molecular rows is ca. 1.9 nm, which seems to be the same periodicity as that for the long side of the (42 x 542)R45” unit cell shown in Fig. 33(b), indicating that there is a strong correlation between the adlayers of iodine and that of TMPyP. TMPyP molecules are thought to be adsorbed on specific sites on the iodine adlayer. All TMPyP molecules appear with the same corrugation height of ca. 0.13 nm, indicating that they are located on equivalent sites. Based on the results described above, a model structure of TMPyP on the (42 x 542)R45”-I adlayer was constructed as shown in Fig. 35. In this model, the center of each porphyrin ring is located on top of an iodine atom in the dark row. All pyridinium rings are located on the iodine atoms in the bright iodine rows. The length of the square unit cell (a=b) is calculated to be 1.97 nm, which is very close to the value obtained experimentally as described above. Note that the side-by-side configuration causes a stronger repulsive interaction between the nearest positively charged N-pyridinium groups. A similar side-by-side configuration was observed for TMPyP adsorbed on I-Au( 111) at an intermediate stage as shown in Fig. 30. This configuration was converted to other structures presumably to minimize the repulsive interaction [ 1951. In the present case, the interaction between TMPyP molecules and the iodine adlayer might be weaker than that on I-Au( 111) or I-Ag( 11I), because a part of the porphyrin ring is not directly attached to iodine atoms in the dark row. Under these circumstances, we believe that the ion paring with perchlorate anions plays a crucial role in determining the adlayer structure. Particularly, the side-by-side configuration in the direction b in Fig. 35 can only be explained by the coordination of perchlorate anions. The surface coverage and the surface concentration of TMPyP can be calculated to be 0.02 and 2.77 x lOI molecules/cm’, respectively, based on the model shown in Fig. 35. A larger value of 3.7 x 10” molecules/cm* was obtained for TMPyP on I-Au( 111) as described in our previous paper [ 1941, indicating that the TMPyP adlayer has an open spaced structure on I-Pt( 100).

170

K. ltaya

f

20 nti

[1001

f

[1001

lOnm

Fig. 34. High-resolution STM images of TMPyP on I-Pt( 100) in HCIO,. From [ 1981.

/

/

Fig. 35. A model structure of TMPyP molecules adsorbed on I-Pt( 100). From [ 1981.

STM

in

Electrolytes

171

As described above, the surface of I-Pt( 111) was almost completely covered by flat-lying TMPyP molecules. Each individual TMPyP molecule could be recognized as the characteristic square shape. However, the adsorbed TMPyP molecules did not form an ordered array on the I-Pt( 111) as shown in Fig. 32(a). The surface diffusion of adsorbed TMPyP molecules was rather slow on I-Pt( 111). These results further support that the structure of the iodine layer on Pt( 100) plays an important role in the formation process of the highly-ordered TMPyP adlayer. The result observed on I-Pt( 1001 clearly indicates that the adlayer structure of TMPyP is controlled by the interaction between iodine adlayer and TMPyP. The adlayer structure of TMPyP on I-Pt( 100) is totally different from that observed on I-Au(1 11) or I-Ag(1 11). This difference is primarily due to different atomic arrangements of iodine atoms. TMPyP molecules align not only in the direction parallel to the iodine atomic rows, but also in the neighboring molecular rows with the same side-by-sideconfiguration. Such a highly-ordered molecular arrangement in both directions seems to result from the stabilization of repulsive force between methylpyridinium groups by perchlorate anions in solution. The effect of anions is of special interest in this case. (vi) Other molecules on I-Au(ll1). Here, we briefly describe further evidence that the I-Au(I 1 1) electrode can be employed as an ideal substrate for in situ STM imaging of various adsorbed organic molecules in solution. Organic substances investigated were water soluble cationic molecules purposely selected based on their characteristic shapes: triangular and linear. Hexamethylpararosaniline (crystal violet) and 4.4’-bis(N-methyl-pyridinium)-p-phenylenedivinylene (PPV) were also found to form highly-ordered molecular arrays on top of the iodine monolayer adsorbed on Au( 111) [ 1941. In sitrl STM with near-atomic resolution revealed their orientation, packing arrangement, and internal structure of each molecule. A typical high-resolution STM image of the molecular arrays of crystal violet is shown in Fig. 36. It is also surprising to see that the STM

(a)

image shows a distinctly

lb)

Fig. 36. High-resolution

STM image of crystal violet on I-Au( 1 11). From [ 1941.

K. ltaya

172

characteristic, propeller-shaped feature for each molecule with highly-ordered arrays. Each molecule has three benzene rings located at the apexes of a triangle with an equal distance from the central carbon atom. Three bright spots seem to correspond to the benzene rings. The center of each spot is located ar a distance of ca. 0.35 nm from the center of the triangle. An additional spot can also be seen at the position of the central carbon atom of crystal violet. According to the STM image, it is clear that ali molecules are oriented in the same direction. The unit cell shown in Fig. 36 can be characterized by the lattice parameters: a=0.9 nm, b=l 1 nm and the angle of ca. 7.5”, indicating that the crystal violet adlayer was slightly deformed from three-fold symmetry. The third compound investigated is the highly symmetric cationic PPV molecule with two terminal pyridinium rings connected with the straight phenylene-divinylene core. The image shown in Fig. 37 is a typical STM image of the ordered PPV adlayers formed on the I-Au( 111) surface. The individual flat-lying PPV molecule can be seen as a linearly aligned feature consisting of three bright spots that can be attributed to the three aromatic rings in PPV. Three bright spots aligned in a straight line suggest that the PPV molecules adsorb on the I-Au( 111) surface with a straight configuration.

PPV molecules are

expected to form straight or bent configurations, depending on the relative orientation of the two trans CH=CH double bonds. The STM image shown in Fig. 37 indicates that the two trans CH=CH double bonds are located on the opposite side, forming the straight configuration in the adlayer shown in Fig. 37. It is also clearly seen that the tightly packed arrangement forms long striped domains. In each domain, all molecules show the same orientation as indicated by the model in Fig. 37. The width of each domain along the molecular axis was found to be ca. 2.1 nm from the STM image, which

10 nm Fig. 37. High-resolution

STM image of PPV on I-Au( 111). From [ 1941.

STM

in Electrolytes

173

corresponds to the total molecular length of PPV. It is also interesting to note that a zig-zag arrangement appears alternately in these striped domains. In this section, we described further evidence that the I-Au( 111) electrode can be employed as an ideal substrate for in siru STh4 imaging of adsorbed organic molecules in solution. Organic substances investigated were all water soluble cationic molecules with characteristic shapes: triangular, linear, and square. Molecules of crystal violet, PPV, and TMPyP were all found to form highly-ordered arrays on top of the iodine monolayer adsorbed on Au( 111). In situ STM with near-atomic resolution revealed their orientation, packing arrangement, and even internal structure of each molecule. The novel approach, using the iodine monolayer as an intermediate layer for the adsorption and formation of molecular arrays, has a great potential for investigations of many organic molecules including more complexed molecules such as optical isomers [204,205] and native biological materials. Finally, it is noteworthy that Clavilier et al. reported a self-assembly of methylene blue at sulphurmodified Pt single crystal electrodes using STM in air [206], suggesting that the S-Pt electrode might also be useful for the investigation of adsorption of organic molecules. D. Adsorption

of Aromatric

Molecules

on Clean

Bare

Electrodes

The adsorptionof organic moleculesonto bare electrodesurfacesin electrolyte solutions under potentialcontrol haslong beeninvestigatedfor elucidatingthe role of structureandproperty of adsorbed moleculesin EC reactions[2,4,184]. Although conventionalEC and optical techniques,such as IR, Raman,and SHG spectroscopies have been extensively appliedto the investigationof the molecular adsorptionat electrodesurfacesin solution [ 1841,they usually can provide only averagedinformation on the molecularorientationand packingwithin an adlayer. Ex situ techniquessuchas LEED and AES, usingthe UHV-EC technique,have alsobeenextensively employedfor generatingunderstandingof the relationshipbetweenthe adsorbedmoleculesand the atomic structure of the electrodesurfaces[2,4 ]. More recently, in situ STM has been well-recognizedas an important in situ method for structural investigationof adsorbedchemicalspecieson well-definedelectrodesurfacesin electrolytesolution wi!h atomicresolution. Although smallinorganicspecies,suchas halide, sulfate, cyanide, and thiocyanateadsorbedon the metalelectrodesurfacecan be visualized relatively easily by in situ STM as describedalready, high resolution STM imageshave scarcely been reported for organic moleculesadsorbedat bare metal surfaces.A few recentpapershave reportedin situ STM imageswith relatively high-resolutionof DNA suchasadenine,guanine,and cytosine, and of 2,2’-bipyridine adsorbedon Au( I 11) [ 187.1911. The adsorptionof thymine was also investigatedon Au single crystal electrodes[192]. In these bases,

experiments,HOPG and Au singlecrystalshave beenalmostexclusively usedas,the electrodematerials [ 179,185.1921.

In the previous section, it was demonstratedthat highly-ordered molecular arrays of porphyrine, crystal violet, and a linear aromaticmolecule,PPV, were easily formed on iodine-modifiedAu( 111) rather thanon bareAu, andthey were visualizedin solutionby in situ STM with near-atomicresolution, revealingpacking arrangementsand even internal molecularstructures. Such an extraordinarily high resolutionachievedin solution strongly encouragedus to investigatethe adsorptionof relatively small

174

K. ltaya

organic molecules such as benzene directly attached to the electrode surface in order to understand clectrocatalytic activities of noble metals such as Pt and Rh. On the other hand, voluminous reports describe investigation of the adsorption in IJHV of aromatics, such as benzene and its derivatives on Pt, Rh, Ni, Ir, Ru, and Pd, which were performed by using various surface sensitive techniques such as LEED, AES, and electron energy-loss .spectrosct?py (EELS). The purposeof those investigationswas to evaluategas phasecatalytic reactions, such as hydrogenation, dehydrogenationand dehydrocyclization 12071. In IJHV the (3 x 3) superlatticeof benzeneand CO coadsorbedon Rh( 111) revealeda well-orderedarray of ring-like featuresassociated with adsorbedbenzenemolecules,while CO did not appearin STM images[ 18I]. A more detailed STM study wasrecently reportedby Somorjaiet al. [208]. Nevertheless,we described,for the first time, the adlayerstructuresof benzeneadsorbedon Rh( 111) and Pt( 111) in HF solutions(3I]. High-resolutionSTM imagesallowed us to determinethe packing arrangementand eventhe internalstructureof eachbenzenemoleculein solution. (i) Benzene on Rh(l11) and Pt(ll1) in HF. Figures 38(A) and 38(B) show cyclic vo1tummogrcrm.s (CVs) of Rh( 11I ) and Pt( 111) electrodesin the absenceand presenceof benzenein

0.01 M HF, respectively. In the absenceof benzene,the CV obtainedon the well-defined Rh(111) and Pt( 111)exhibitedseveralhighly reversiblecharacteristicpeaks. The overall shapeof the CV wassimilar to that reportedpreviously for Rh( 111)in HF solutions12091.It was noted that the heightsand widths of thesecharacteristicpeaksdependedon the quality of the surfaceof Rh( 111) preparedby the flameannealingandquenchingmethod. Note that the electrochemistryof Rh(111)in HCIO, iscomplicatedby the EC reductionof perchlorateanions, releasingchloride as a reductionproduct [60,76,77,209]. For this reason,HF solution was usedexclusively for the investigationon the adsorptionof benzeneonto the chloride-freeRh(ll1) surface. After the Rh(111) electrodewas subjectedto the CV measurementin the pure HF solution. the electrodewastransferredinto a 0.01 M HF solutioncontainingca. 1 mM benzene. The CV indicateda featurelessdouble-layerregion between0.3 and 0.7 V as shown in Fig. 38(A). The cathodic current commencingat about0.3 V was consideredto bc due to simultaneouslyoccurring processes,such as the desorptionof adsorbedbenzene,the adsorptionof hydrogen, and the irreversiblehydrogenationof benzeneto cyclohexane. according to the previous studies using d$erential electrochemicul mass spectrometr?; (DEMS) [210,2 1I]. A similar featurelessCV was also obtainedwith a benzene-dosed Rh( 111)electrodein pure 0.01 M HF. The Rh(111)electrodewas immersedin 0.01 M HF containing 1 mM benzenefor 1min at the opencircuit potentialand thentransferredto the pureHF solution. These resultstrongly suggestthat benzeneis chemisorbedand remainedon the surfaceof Rh(ll1). at leastin the potentialrangebetween0.3 and0.7 V. The CV’s of Pt( 111) in 0.01 M HF without and with 1 mM benzeneare shown in Fig. 38(B). In the absenceof benzene,the CV is a typical exampleof a well-definedPt(ll1) electrodein HF in good agreementwith previousmeasurements [2 121. The adsorptionof benzeneon Pt( 111) alsoresultedin a featurelessdoublelayer regionin the potentialrangebetween0.25 and 0.6 V. Almost reversible,sharp

STM

in Electrolytes

175

(B)

E, V h

SCE)

(b)

0

0.5 E, V (VL RHE)

0

0.5

1.0

E, V (VS. RHE)

Fig. 38. Cyclic voltammograms of Rh( I 11) (A) and Pt( 111) (B) without (a) and with (b) 1 mM benzene. From [31]. peaks appeared at 0.08 V just before the hydrogen evolution reaction as seen in Fig. 38(B). The full width at the half maximum of the cathodic peak was ca. 30 mV. These sharp peaks can be considered to be due to the adsorption and desorption of benzene on Pt( 111) associated with those of hydrogen [210,21 I]. (a) Structure on Rh( 111). A well-defined terrace-step structure was easily observed on the wellprepared Rh( 111) face as shown in Fig. 1l(a). The atomic image of Rh( 11 l)-( 1 x 1) was routinely discerned on the terrace in the pure HF solution as shown in Fig. 1l(b), The almost perfectly aligned hexagonal structure can be seen with an interatomic distance of 0.27 nm, indicating that the structure of the Rh( 111) surface is (1 x 1). Identical atomic images were consistently observed in the potential range between 0.1 and 0.75 V. No additional species were found in STM images at potentials corresponding to the butterfly peaks or the hydrogen adsorption and desorption peaks, in agreement with our previous result obtained on Rh( 111) in 0.1 M HClO, [60]. After achieving the atomic resolution shown in Fig. 1l(b), a small amount of 1 mM benzene solution was directly added to the STM cell at 0.45 V. The average concentration of benzene in 0.01 M HF was 10 pM. Immediately after the injection of benzene, completely different patterns appeared in STM images. Figure 39(a) shows an example of the STM images acquired at 0.45 V. It is evident that the atomically flat terraces are now covered by ordered benzene adlayers. An averaged domain size was about 10 x 10 nm. The adsorbed benzene molecules appear to form a square adlattice in each domain.

176

K. ltaya

0

5

IO

15

20

i

i

nm Fig. 39. High-resolution

5

b

nm

STM images of the ~(243 x 3)r.~ct benzene adlayer In HF. From [31].

Furthermore, the molecular rows in a given domain cross each other forming boundaries at an angle of either 60” or 120”. More details of the orientation of benzene in the adlayer are revealed by the higher resolution STM image shown in Fig. 39(b). The acquisition of the STM image was performed specifically under conditions with minimal thermal drift in x and y directions in order to determine the unit cell of the adlayer as accurately as possible. It is seen in Fig. 39(b) that the molecular rows along the direction of arrows A and B cross each other at 90”. and they are always parallel with the close-packed and 11’3 directions of the Rh( 111) substrate, respectively. The intermolecular distances along these directions are not equal to each other and were found to be on the average, 0.8 nm and 0.9 nm. respectively. Based on the orientation of molecular rows and the intermolecular distances, we concluded that the benzene adlayer was composed of rectangular unit cells, namely ~(243 x 3)rect (8=0.17),

as shown in Fig.

39(b). The known lattice spaces of 243 and 3 on Rh( 11 I) (0.268 nm) correspond to 0.93 nm and 0.80 nm, respectively, which are consistent with our experimental values. Surprisingly, the STM image allowed us to determine the internal structure and micro-orientation of each benzene molecule adsorbed on Rh( Ill). It is clear that each spot is split into two bright spots. forming a characteristic shape of dumbbell for each benzene molecule. The STM discerned a 0.0 1 nm corrugation between the valley and ridge of each benzene molecule. It can also be seen in Fig. 39(b) that the orientation of dumbbell-shaped benzene is not the same for all molecules, but depends on their

STM

in Electrolytes

177

positions. The dumbbell-shape of the central benzene molecule in the unit cell shown in Fig. 39(b) is clearly rotated by 60” with respect to the molecules located on the four comers of the unit cell. The molecules on the corners of the unit cell appeared with an identical feature, suggesting that they are situated on an identical binding site. It is also seen that the orientation of these dumbbells is always rotated by 30” with respect to the direction of close-packed rows (arrow A in Fig. 39(b)) of the Rh( 111) substrate. The STM image shown in Fig. 39(b) provides more detailed information on the orientation of molecule in the unit cell as discussed below. The ~(243 x 3)rect structure described above was consistently observed in the potential range between 0.4 and 0.7 V without additional structural transitions. On the other hand, it was found that the adlayer structure changed at negative potentials. A negative potential step from 0.45 to 0.35 V induced a reconstruction in the benzene adlayer from ~(243 x 3)rect symmetry to an ordered hexagonal pattern. The electrode potential of 0.35 V is near the onset potential of the cathodic current as shown in Fig. 38(a). Figure 40 shows a set of STM images acquired in almost the same area in order to reveal the dynamic process of phase transition. It is clearly seen in Fig. 40(a) that a new domain appeared with the hexagonal array of benzene on the upper right comer marked by solid lines, while the ~(243 x 3)rect structureremainedas the main phase. A further cathodicstep to 0.25 V resultedin a predominantly hexagonalphase,while eliminatingthe ~(243 x 3)rect domainsas shown in Fig. 40(b). Such a longrangeorderedhexagonalpatterncould be seenover almostthe entire areaof the terraceat 0.25 V. All benzenemoleculesexhibited the samecorrugationheightof 0.07 nm, similarto that in the ~(243 x 3)rect structure. To reveal the internal molecularstructurein the hexagonalphase,STM imageswere acquiredunder particularly carefully adjustedexperimentalconditionswith minimalthermaldrift. Figure 41(a) shows one of the highestresolutionimagesacquiredon the terraceshown in Fig. 40(b). Comparedwith the

0

7

14

21

28

35

0

5

10

15

20

25 nm

Fig. 40. STM imagesof domainboundariesof ~(243 x 3)rect and (3 x 3) benzeneadlayers (a) and the pure (3 x 3) structure(b) on Rh(ll1). From [31].

K. ltaya

178

(b)

J

0

1

2

3

4

5 nm

0

c

1

2

3

4

5 nm

Fig. 41. High-resolutionSTM imagesof the (3 x 3) structureon Rh( 111). (a) top view. (b) height-shadedplot. From [31]. crystal orientation, [ITO], determinedby the Rh(l1 I)-( 1 x 1) atomic image, it can be seen that all benzenemoleculesare almostperfectly aligned along three close-packeddirections of Rh( 111). The molecularrows crosseachother at an angleof either 60” or 120” within an experimentalerror (+2”). The intermoleculardistancealong theserowswasfound to be 0.8 nm, which correspondsto 3 timesthe lattice parameterof Rh(111). Therefore,we concludethat the hexagonalstructureis (3 x 3)-C& (8 = 0.11) as shown by the unit cell superimposedin Fig. 4 l(a). Moreover, a careful examinationof the imagerevealsthat eachbenzenemoleculeappearsasa set of three spotswith similarintensities. It can alsobe seenthat a cleardip existsin the center of eachtriangle with three lobes. Thesefeaturescan be more clearly seenin the height-shadedsurfaceplot obtainedby applying a mild 2D Fourier transform filter methodasshownin Fig. 41(b). The spacingbetweenthe two lobesin eachmoleculewasfound to be about 0.3 nm. In addition, a weakeradditionalspot with a smallcorrugationof about0.02 nm can be seenin the unit cell. It is important to note that all thesefeaturesof the STM imagefor the (3 x 3) adlayerobservedat 0.25 V are essentiallyidenticalto thosefound for the coadsorbedbenzeneand CO adlayeron Rh( 1I 1) in UHV reportedby Ohtani er ul. [176,181]. They also found the weaker spot, which wasattributed to the coadsorbedCO or artifactscausedby asymmetrictips [ 18l]. When the electrodepotentialwassteppedfurther in the negativedirection, the ordered(3 x 3) domain becameislandswith the sameinternal structure, suggestingthat the desorptionof benzeneoccurred preferentially at edgesof the islandsof ordered(3 x 3) domains12131.Eventually, all adsorbedbenzene moleculesweredesorbedfrom the surfaceat 0.1 V due partially to the hydrogenadsorptionand partially

STM

in Electrolytes

179

to the hydrogenation as expected from the result obtained by DEMS [210,211], and the Rh( 11 l)-( 1 x 1) structure, similar to that shown in Fig. 1l(b), was consistently discerned at 0.1 V. The structural changes described above were reversible. When the electrode potential was stepped back to the positive region, the (3 x 3) and ~(243 x 3)rect phases returned at the potentials described above [3 1,2 131. The structures and registries of chemisorbed benzene on Rh( 111) have been thoroughly scrutinized by the surface sensitive techniques, such as LEED, EELS, and angle-resolved UV photoemission spectroscopy (ARUPS) in UHV [207]. These previous studies revealed various structures for benzene, including well-known structures such as ~(243 x 4)rect and (3 x 3), depending on whether CO was present unintentionally or intentionally in the UHV chambers. Although it has been repeatedly demonstrated that the adlayer structures of benzene on Rh and Pt were greatly affected by the presence of CO in the adlayer, the structure of the pure benzene adlayer has not yet been fully understood. Neuber et al. recently reported that a completely new structure with a (419 x 419)R23.4” symmetry appeared for the pure benzene adsorption on Rh( 111) under cleaner UHV conditions in the absence of CO, and the previously known structures of ~(243 x 4)rect and (3 x 3) were found to appear upon admission of CO [214]. However, it was found in our study [31] that the anodic peak due to the oxidation of CO was hardly detectable in CV even after a prolonged STM experiment for several hours in an au-saturated HF solution. We strongly believe that the adlayer structures found in HF solution described above did not result from contamination with CO. It is extremely important to recognize that in the previous study of the adsorbed benzene on Rh( 111) in UHV [214,215], one of the structures of the pure benzene adlayer was attributed to the ~(243 x 3)rect structure, which was found in solution. This result strongly suggests that the existence of water molecule on top of the benzene adlayer plays a minor role in determining the structure of benzene. Figure 42 (a) shows a proposed model for the ~(243 x 3)rect structure. All of the adsorbed benzene

Fig. 42. Space models of the ~(243 x 3)rect (a) and the (3 x 3) structures (b). From [3 11.

180

K. ltaya

molecules are assumed to be located on the two-fold bridging sites. The benzene molecule at the center of the unit cell also occupies a two-fold site, but it is rotated by 60” from the orientation of the moleculc~ at the corners. Weiss and Eigler reported three distinct types of STM images for isolated benzene molecules located at three-fold hollow, atop, and bridge sites on Pt( 1 I 1) at 4 K [ 1801. They assigned the single bump elongated perpendicularly to the bridge to the bridge-bonded benzene. In Fig. 39(b) each benzene molecule is seen with the dumbbell shape on Rh( 1 11) and elongated perpendicularly to the bridge. It is clear that the direction of each elongated dumbbell is always rotated by approximately 30. with respect to that of the corresponding atomic row of Rh( Ill). These detailed features can be explained by the model structure shown in Fig. 42(a), where two lobes marked by the circles are assumed to be localized to near carbon atoms (1, 2, 6 and 3, 4, 5) bonded across the Rh atoms. The STM image obtained at 0.25 V shown in Fig. 41(a) can be explained by the structural model with the (3 x 3) symmetry illustrated in Fig. 42(b). Although the structure proposed here is basically the same as that proposed previously, based on the LEED, EELS [207] and STM [ 18 1) studies in UHV, for the adlayer of coadsorbed benzene and CO on Rh( 11 l), two CO molecules thought to be located at the three-fold hollow sites in the unit cell are omitted in Fig. 42(b). Each benzene molecule is assumed to bond at the three-fold hollow site. The coadsorption of CO was unlikely to take place in the solution under the present condition, because no oxidation peak was observed as described above. Instead of CO. water molecules or hydronium cations might be coadsorbed near the uncoordinated three-fold hollow sites to stabilize the (3 x 3) structure, their function being similar to that of the coadsorbed CO. The weak small spots seen in Fig. 40 might be due to such coadsorbed water molecules or hydronium cations. According to the previous STM in UHV [ 1811 and the theoretical calculations [2 161, it can be expected that the three spots for each benzene molecule shown in Fig. 41 are located between the Rh atoms as indicated by the small circles in Fig. 42(b). (b) Structure on Pt(lll). The in situ STM imagingof benzeneadlayerson Pt( 111) was carried out in the samemanneras that on Rh( 111). The atomic resolution of Pt( 111) substrateswas first achievedat ca. 0.2 V within the hydrogen adsorptionregionbefore the dosingof benzene. A few drops of 1 mM benzenesolutionwas injectedusually at potentialsbetween0.35 and 0.6 V. The hexagonal I’t( 111) lattice in the STM imagewas instantaneouslyreplacedby more widely spaced, protruding features,which apparentlyare attributableto chemisorbedbenzenemolecules.It wasfound in our study that benzeneadlayerson Pt( 111) mostly appearedaslessorderedphasesas shown in Fig. 43(a) than thoseon Rh( 111). However, somepatchesof orderedstructures,for example,at the upper portion of the STM imagewere unambiguouslyidentified asordereddomains. The intermoleculardistancesand directionsof molecularrows indicatethat the structure of the benzeneadlayer at 0.35 V is ~(243 x 3ked, the sameas that found on Rh(l1 I). The degreeof ordering in the ~(243 x 3)rect array on Pt( 111) is lessthan that on Rh(11l), andthe ordereddomainsize is typically 5 x 5 nm. It is interesting to note that poor orderingin the adsorbedbenzeneadlayeris known to occur on Pt( 111) in the absence of CO in UHV [217]. The appearanceof the disorderedarray on Pt( 111) in HF solution seemsto parallel the poor orderingobservedin UHV.

STM

in Electrolytes

181

More interestingly, in situ STM revealed the reconstruction of benzene adlayer that occurs upon the cathodic potential step from 0.35 to 0.25 V. Figure 43(b) shows a large scale image of an area of 25 x 25 nm obtained at 0.25 V. It is clearly seen that the adlayer is composed of highly ordered arrays with hexagonal rings, although a large number of boundaries exist between the well-ordered domains. All molecules seem to exhibit the same corrugation height of ca. 0.07 nm, suggesting that they are situated on the same binding site. A close-up STM image obtained in an ordered domain is shown in Fig. 44(a), revealing the almost perfect array with hexagonal rings, in which only a translational domain boundary (A) can be seen. Note that some diffuse, much weaker features can also be seen inside the unit cells, presumably due to coadsorbcd water molecules. We made great efforts to determine the adlayer structure as accurately as possible. A precise comparison of the images of Fig. 44(a) and of Pt( 11 l)-( 1 x 1) revealed that the molecular rows are rotated by approximately lo” with respect the Pt rows. The unit cell length corresponding to twice the intermolecular distance along the molecular row averages 1.26 nm. which is nearly equal to 421 times the interatomic distance of Pt (0.2778 nm). It is also evident that the molecular rows seen in Fig. 44(a) form an angle of 60” or 120” within the experimental error. On the basis of these detailed results, we proposed that the benzene adlayer has a (421 x 42l)R10.9” structure (0 = 0.14), whose ball models are illustrated in Fig. 44(b). All benzene molecules are located on equivalent bridge sites in the model shown in Fig. 44(b). Although we proposed other possible models [3 11, the model shown in Fig. 44(b) is thought to be one of the most likely structures for (42 1 x ~21)R10.9”.

0

5

10

20

15

nm

0

lb

1s

20

25

nm

Fig. 43. STM imagesof disorderedbenzeneadlayer obtainedat 0.35 V (a) and highresolutionimageacquiredat 0.25 V (b) on Pt( 111). From [3 11.

K. ltaya

182

(a)

0

2

4

6

8

10

nm Fig.

44. High-resolution STM image of the (421 x 421)R10.9” structure (a) and an

illustrative model(b). From [3 11.

(ii)

Naphthalene

on Rh(ll1)

and Pt( 111).

After the atomic resolutionshown in Fig. 11 was

achievedwith Rh(11l), a saturatednaphthalenesolutionwasaddedto the STM cell at 0.3 V [218]. The averageconcentrationof naphthalenewasca. 0.02 mM. Completelydifferent patternsappearedin STM imageswithin 10min after the injectionof naphthalene.A high resolutionimageacquiredin an ordered domainis shown in Fig. 45(a) [218]. It is clearly seenthat the molecularrows parallel the ~1 lO> direction of the substrateindicatedby the arrows in Fig. 45(a). More importantly, the STM image allowed us to determinethe internal structure and orientation of each naphthalenemolecule. The elongatedfeaturesalongthe longer molecularaxis (C,) were discernedfor eachmolecule. In addition, the imagesof somemoleculesclearly showa two-ring structureexpectedfrom the molecularmodel. It can alsobe seenthat naphthalenemoleculesam perfectly alignedwith a regular micro-orientationalong naphthalenefurther support the predominantadsorbate-substrate interactionfor hydrophobic molecules and theminor role of water molecules. A typical arrangementof naphthalenemoleculeson Pt( 111)was found to be a full monolayerof flatlying naphthalenemolecules[2 181. Although the overall appearanceof the naphthaleneadlayer on Pt( 111) in the large scalewas similar to that on Rh( 11I), close inspectionrevealedthat the adlayer includedmany randomly orientedmolecules.Becausethe molecularrows were nearly parallelto the close-packedPt atomic rows, and becausethe intermoleculardistancewas 3 times that of the Pt substrate,the structureroughly fitted a (3 x 3) symmetry. However, there were many defectsin the molecular rows. Periodical rotation of the moleculesof naphthaleneby 60” is seen within each molecularrow with every third molecule being in the sameorientation. A further magnifiedview in

STM

in Electrolytes

i,

183

i

nn7

i

tl nrn

Fig. 45. High-resolution STM images of naphthalene on Rh( 111). (a) top view. (b) perspective review. From [218]. height-shaded mode is shown in Fig. 45(b), in which the two-ring structure can be more clearly seen. The nearest-neighbor distance of an average of 0.82 nm is equivalent to three times the Rh lattice parameter of 0.268 nm. According to the results described above, the unit cell can be defined as a (343 x 3-\j3)R30° symmetry as shown in Fig. 45(b) [218]. To make a comparison between the STM image and the adlayer structure, the STM image in Fig. 45(b) is reproduced in Figure 46(a) with partially overlaid molecular models of naphthalene. This (343 x 363)R30” structure results in a surface coverage of 0.11. A ball model of this structure is illustrated in Fig. 46(b). It is now clear that all naphthalene molecules align their C2 axes along the close-packed directions of Rh substrate. The molecules aligned along the < 112> direction, which is the so-called 1/3 direction, have the same orientation.

The spacing between two

adjacent molecules along the 43 direction is measured to be 1.4 nm, which is three times the 43 spacing. In the model structure shown in Fig. 46(b), two carbon atoms at the 9- and lo-positions are assumed to be attached directly to a Rh atom. It is noteworthy that this structure is identical to that previously proposed from LEED results by UHV workers [219]. If one recalls the identical results for benzene adsorbed on Rh( 111) in UHV and in HF solution as described above [3 11, the results obtained with naphthalene further support the predominant adsorbate-substrate interaction for hydrohobic molecules and the minor role of water molecules.

184

K. ltaya

0

2

4

6

8 nm

Fig. 46. STM imageof naphthalenewith an overlaid molecularmodel(a) and a ball model

(b). From [218]. We have previously demonstratedthat the surface diffusion of adsorbedmoleculesplayed an important role in the formation of well-ordered adlayers [3 1,195]. We anticipatedthat naphthalene moleculeswere more strongly adsorbedon Pt( 111) than on Rh( 11I), resultingin a lower diffusion rate and thus a disorderedphaseon Pt(l11). Once naphthalenemoleculesland on the surface, they are unlikely to diffuse andeven re-orient themselvesto createan orderedpattern. This point can be further illustratedby a separateexperiment,in which isolatednaphthalenemoleculeson Pt( 111) were clearly discerned[2 l&22 I]. It wasfound that the moleculeswere immobileon Pt( 111) even at a low surface coverage. Identical imageswith the sameorientationwere consistentlyobservedfor at least 15-20 min, indicating that the surfacediffusion of naphthaleneon Pt( 111)is very slow. For this reason,the degree of ordering was not improved by prolongedexposure or potential cycles between0. I and 0.8 V in solutionscontainingnaphthalene. (iii) Other molecules. We have briefly investigated 1,2- and 1,4-naphthoquinone(NQ) as representativederivatives of naphthalenein order to understandthe effect of functional groups on the molecularorganization 12181.Generally speaking,thesequinones, similarly to naphthalene.formed orderedadlayerson Rh(111)and mostly disorderedadlayerswith the flat-lying orientationon Pt( 111). The high-resolution STM images revealed the details of internal molecular structures. It was demonstratedthat 1,2-naphthoquinone formeda well-orderedadlayerwith the identical structure,(343 x 343)R30”, which wasidentical to that found for naphthalene. More interestingly, it was found that an additional bright spot, unseenin the imagefor naphthalene,is seenat the 2-position of each 1,2naphthoquinonemolecule[21X]. This spotexhibits a ca. 0.03 nm highercorrugationwith respectto the naphthalenering, which is likely to be dueto the oxygen at the 2-position.The oxygen at the l-position wasnot clearly seenin the STM images.

STM

185

in Electrolytes

The STM image shown in Fig. 47(a) shows a monolayer of anthracene on Rh(ll1) with clear resolution for each individual molecule. A close-up view of Fig. 47(a) is shown in Fig. 47(b), revealing preferential alignment of the molecules with their long C, axis along the close-packed directions of Rh, which was also observed for naphthalene. The STM image in Fig. 47(b) was filtered by a 2D FI method to remove features of less than 0.2 nm. This result unambiguously discloses the internal molecular structure of anthracene with clear identification of three craters in each molecule. The length and width of a single anthracene molecule are 0.8 and 0.35 nm, respectively, and its corrugation of 0.08 nm is comparable to the values for benzene and naphthalene imaged in UHV or in solution. It was also found that molecules of anthraquinone and 1,4,9,10-anthracenetetrol adsorbed in a manner similar to those of anthracene [218]. To elucidate the effect of molecular structure on the packing arrangement, biphenyl was further investigated on Rh( 11 I) [22 11. In contrast to the planar structure of naphthalene and anthracene, the two aromatic rings of biphenyl are slightly off the co-planar configuration because of the restriction of hydrogen atoms at 2,2’ and 6,6’ positions. However, it was found that biphenyl formed disordered adlayers on Rh( 111) in HF [22 I]. Although STM images disclosed the internal molecular structure of biphenyl with clear identification of two rings in each molecule, the non-planar configuration was not clearly seen [221]. We expected that two rings of biphenyl behave like two benzene molecules, and they prefer to be attached on the bridge sites as benzene molecules do [22 11. The appearance of the disordered adlayer of biphenyl suggests that biphenyl is more strongly attached on Rh(l1 I) than

0

10

20

30 0

1

2

nm Fig. 47. STM images of anthracene on Rh(lll).

3

nm From [218].

186

K. ltaya

naphthalene. It is of special interest to distinguish structures among a series of benzene dertvativcs such a\ phthalic acid. terephthalic acid, and hydroquinons [222]. (iv) Renzene, naphthalene and anthracene on Cu( 111). The bonding and coordination oi organicmoleculeswith metalelectrodesurfacesarethe fundamentalissueboth in electrochemistry[ 1841 andin catalysisin gasphase[207]. The use of other metalsinsteadof Rh and Pt is extremely important to understandthe role of the interactionbetween substrateand adsorbatesin ordering processesof adsorbedorganicmolecules. Here, we briefly describe our recent study of adlayer structures of benzene, naphthaleneand anthracenedirectly attachedon a well-definedCu(l11) in aqueousHCIO, solution [223]. A Cut,11I)-( 1 x 1) structurewas observedon atomicallyflat Cu(111) surfacein a doublelayer potentialrangein the absenceof organic molecules. Although atomically rough surfaces were consistently observed immediatelyafter the electropolishing,it was found that well-defined terrace-stepfeatures gradually developedat potentialsnearthe anodicdissolution,indicatingthat the dissolutionproceededby a layerby-layer modein HCIO,. The anodic dissolutionof various metalswill be discussedin the following section. Figure 48(a) is a typical largescaleSTM imageof a Cu(111) surfaceacquiredat -0.1 V. Monatomic steps(0.22 nm) arc seento run nearly parallel, or at an angleof ca. 60”, to eachother, asexpectedfor surfaceswith a three-fold symmetry. However. it is clear that the stepsinclude many defectssuch as kinksbecausethe steplinesare not atomicallystraight. The terracesareatomically flat with a low defect density, suggestingthat the Cu( I 11)surfacehasa well-definedstructurein HCIO,.

03 s

(a)

.

50nm

’

Fig. 48. Topographic(a) and high-resolution(b) STM imagesof Cu( 111) in 0.1 M HCIO,. From 12231.

STM

in Electrolytes

187

High-resolution

imaging made it possible to observe individual Cu atoms on the terraces. To our knowledge, atomic resolution has not yet been achieved on a bare Cu( 111) in solution [ 1011. Our first atomic image of Cu( 111) surface is shown in Fig. 48(b) [223]. The hexagonal closed-packed structure can be seen with an interatomic distance of ca. 0.26 nm (the diameter of Cu atom is 0.256 nm), indicating that the Cu( 111) surface has a (1 x 1) structure. The arrows point directions of close-packed Cu( 11 1) determined by the crystallographic orientation of the crystal. The corrugation amplitude was quite low, ca. 0.01-0.015 nm, which is significantly smaller than that observed on Au, Pt, Rh or Ag, typically 0.02-0.03 nm. The same (1 x 1) structure was consistently observed in the double layer potential range. No evidence was found for the oxidation of Cu( 111) in HClO,, while it was reported based on in situ AFM experiments that Cu( 100) surface might be covered with adsorbed 0 or OH-. exhibiting a (t/2 x 2/2)R45” structure, even in HClO, [224]. Nevertheless, the image shown in Fig. surface retains a clean and 48(b) demonstrates that, under the present conditions, the Cu(ll1) unreconstructed (1 x 1) structure. (a) Benzene adlayer. After the atomic resolution image of Cu( 11 l)-( 1 x 1) structure shown in Fig. 48(b) was observed, a small amount of benzene solution was directly injected into the EC cell. The average concentration of benzene was ca. 1 mM. The image shown in Fig. 49(a) is an example acquired in an area containing molecular defects. It is clear that each benzene molecule appears as a set of three spots with almost the same corrugation height. The intermolecular distance along the close-packed directions of Cu( 111) is ca. 0.77 nm, which corresponds to three times the lattice distance of Cu( 111). The molecular rows cross each other at an angle of either 60” or 120” within an experimental error. A clear dip exists at the center of each triangle with three lobes. The difference in CorrugatiOn

I

2nm

Fig. 49. High-resolution From [223].

STM image (a) and a model structure (b) of benzene on Cu( 111)

K. ltaya

188

height between the spot and the dip is ca. 0.006 nm. Therefore, we conclude that the benzene adlayer forms a (3 x 3)-C,H, structure with a surface coverage of 0. I I as illustrated m Fig. 49(b). A hcnzenc molecule appeared with only three isolated spots in a triangular configuration. indicating that each benzene molecule is located on the 3-fold hollow site. It is rather surprising that all features observed for benzene adsorbed on Cu( 1 1 I ) is almost identical to that found on Rh( 111) as described above. It is interesting to note that the adsorbed monolayer of benzene on Cu( I 11) is desorbed in UHV in the temperature range between 150-240’ K [225]. At room temperature, all benzene molecules except those adsorbed at defect sites undergo complete desorption from the Cu( 111) surface [225]. These results obtained in UHV are in contrast to the in sir/r STM result. The benzene adlayer on Cu( 111) is stable at room temperature in solution, suggesting that water molecules play an important role of stabilizing the adlayer. (b) Naphthalene adlayer. Figure 50(a) shows a high-resolution STM image of a naphthalene adlayer. A long-rangeorderedadlayer can be clearly seenover a large area. NaphthaIenemolecules uniformly cover the Cu(I1 1) surfacewith a smallnumberof defects. Each spothasan elongatedfeature correspondingto an individual naphthalenemolecule.The molecularrows were found to be parallelto the direction of the underlying Cu( 111) lattice. The STM shows details of the packing arrangementandthe internal structureof the naphthaleneadlayer. In this image,the two-ring structure of the naphthdlenemoleculeappearsas two main spotswith additionaldetailsof the internal structure. The distancebetweenthe two main spots is ca. 0.56 nm as expected from the molecularmodel of naphthalene.It is clear that eachmoleculeobviously bondsto Cu( I 11) with the flat-lying orientation. The naphthalenemoleculesare alignedwith the longer molecularaxis (C,) in the samedirection along the Cu rows. The observeddistancebetweenthe nearestneighbor moleculesis I .OIt 0.02 nm, which

(b)

‘2nm

’

Fig. 50. High-resolutionSTM image(a) and a model (b) of naphthaleneon Cu( 111). Frc3rn12231.

STM

in Electrolytes

189

is nearly equal to four times the Cu lattice of 0.256 nm. All features in the STM images indicate that the structure of naphthalene adlayer can be defined as Cu( 111)(4 x 4)C,,H, as shown in Fig. SO(b). The (363 x 3-\j3)R30° structure found on Rh( 111) results in a surface coverage of 0.11, which is nearly twice as large as that found on Cu( 111). This difference suggests that naphthalene is more strongly attached on Rh( 111) than on Cu( 111). The repulsive interaction between adjacent naphthalene molecules seems to bc a predominant factor favoring the formation of the (4 x 4) structure. In this model, each naphthalene molecule is placed onto two neighboring two-fold sites with its C, axis aligned along of Cu( 111) substrate. (c) Anthracene adlayer. It was surprisingto find an extraordinarily orderedadlayerof anthracene on Cu( 111) in view of the fact that anthraceneformed completelydisorderedadlayerson Rh(111) and Pt( 111) as describedabove. Even in a larger areaof 30 x 30 nm, the highly-orderedmoleculararray wasseenwith molecularresolution. Each moleculeappearedasan elongatedspot. Figure 51(a) shows a high resolutionSTM imageshowingdetailsof an anthraceneadlayer. The imageof each molecule showsdetailsof the internal structureof anthracene. It is alsoclear that the anthracenemoleculesare preferentially alignedwith their long C, axesalongoneof the close-packeddirectionsof Cu(111) lattice asindicatedby the setof arrows. The side-by-sideconfiguration is clearly seenin the imageshown in Fig. 5 1(a). The intermoleculardistancealongthe straightmolecularrows is ca. 1.25+ 0.05 nm, which correspondsto five timesthe lattice parameterof Cu. The distanceon the shorter sideis ca. 1.O f 0.05 nm. The two directionsin the unit cell are parallelto the Cu(ll1) lattice for all moleculesobserved. Thesefeaturesstrongly suggest that the structureof the molecularadlayerobservedis (4 x 5)-C,,H,, with a coverageof 0.05 asshownin Fig. 5 l(b). We tentatively proposethat eachanthracenemolecule occupiesnearly three 2-fold bridgesiteswith its long C2axis alongCu(111)lattice. The center ring of

Fig. 51. High-resolution STM image(a) and a model(b) of the adlayer of anthraceneon Cu(111). From [223].

190

K. ltaya

anthracene might be exactly located at the center of a 2-fold bridge site, whereas two other rmgs are slightly shifted from the center position. A detailed inspection of the STM image shown in Figure 5 1(b) revealed the non-equivalent nature of the three rings of anthracene. In conclusion, Cu electrodes are interesting substrates with which to investigate the adsorption of organic molecules in solution, because of the weak interaction between aromatic hydrocarbons and Cu. Well-ordered adlayers of various molecules can be expected to form on Cu electrodes. Note that Behm and his coworkers recently reported adlayers structures of benzotriazole adsorbed on Cu( 100) [226]. Benzotriazole is a well-known inhibitor against the corrosion of Cu. E. Site-selective

Anodic

Dissolution

of Metals

Atomically well-defined surfaces are absolutely needed for in situ STM investigation of electrochemical reactions with atomic resolution. Au, Pt, Rh, Pd, Ir and possibly Ag are the metals which can be prepared to make atomically flat electrodes using the flame-annealing and quenching technique as described already. However, this technique can not be applied to other fundamentally and practically important metals, such as Ni, Co, Fe, and Cu because of the oxidation in the flame. This section deals with details of the anodic dissolution of various metals proceeding only at step edges under carefully adjusted EC conditions. The layer-by-layer process of anodic dissolution results in the formation of atomically flat terrace-step structures. It is reasonably expected that such an in situ method will become one of the most useful techniques for exposing well-defined surfaces of various metals at least for EC studies with STM. Although anodic dissolution has long been employed for “electropolishing” metals to prepare mirror-like surfaces, not much fundamental knowledge is available about the anodic dissolution on the atomic scale. Here, the anodic dissolution of bare Ni, sulfurmodified Ni. bare Ag, iodine-modified Ag, iodine-modified Pd, and Cu will be described in detail based on iw situ STM observations. The anodic dissolution of Au( Ill), Co(OtXll), and Fe( 110) will also be described briefly. (i) Bare Ni(ll1) Electrode. Numerous EC studies have been performed for Ni metal to understand its anodic dissolution, passivation, and oxide formation, because Ni is one of the most important electrode materials used for various industrial purposes such as electrocatalysis, batteries, and electroplating [227]. However, atomic-scale elucidation of EC oxidation processes on well-defined Ni substrates is still an important subject which remains to be investigated. difficulty in the preparation of well-defined Ni surfaces in solution.

This is probably due to the Chemical and EC etching

techniques have been most frequently used in preparing Ni surfaces for EC studies [227]. Bard and coworkers recently reported on their observation by in situ STM of Ni(iO0) in an alkahne solution, in which the surface was initially prepared by chemical etching, followed by in situ cathodic polarization for removing oxide layers [228]. They presented an atomic STM image with a quasi-hexagonal lattice obtained at an early stage of the oxidation of Ni( lOO), which was attributed to the (I 11) plane of NiO or the (0001) basal plane of Ni(OH),. An attempt to image an oxide-free Ni( 100) surface was unsuccessful probably due to a difficulty of removing spontaneously formed oxide layers in alkaline solutions [228]. An ex situ STM study in air also recently demonstrated the capability of STM to reveal atomic structures of electrochemically formed Ni oxide layers 12291. Although bare Ni( 111) surface

STM

in Electrolytes

191

was anticipated to be unstable in alkaline solutions even under cathodic polarization [228], it was firmly concluded by MacDougall and Cohen that complete removal of Ni oxides was possibleonly in acidic solutionsof pH lessthan ca. 3 [230,231]. On the other hand,Wang er al. madean effort to exposewell-definedNi( 111) in an alkalinesolution using an UHV-EC system, in which a well-definedcleanNi( 111) and a Ni( 111) protectedwith a CO adlayerwere prepared,characterizedand transferredto an EC cell [34]. The c(4 x 2)-CO adlayer on Ni( 111) was expectedto protect its surface from the oxidation. Unfortunately, on Ni( 111) the EC oxidation of CO took placein a basicsolutionat rather positivepotentialswherethe Ni( 111) surfacewas alsoelectrochemicallyoxidized [34]. We carried out irz sifu STM of Ni( 111)in an acidic solutionof NqSO, of pH 3.0, in which Ni oxides werereducedby cathodicpolarization [232]. The Ni singlecrystalsusedin our study were subjectedto thermalannealingin a Hz stream. The surfacesof crystals were polishedmetallographically,finished with 0.25 pm diamondpaste,and sonicatedsuccessivelyin trichloroethylene,acetone,methanoland purewater. Then the crystalswere annealedin a quartz tube suppliedwith H, (1 atm) at 1237K for 3 to 10 hr. The furnace was slowly cooled down to room temperatureunder the H, stream. To avoid oxidation of surfacescausedby oxygen in air, the crystalswereimmersedinto hydrogensaturatedwater underthe H? streamandmountedon an STM cell [232]. Figure 52 showsa CV of Ni( 111)recordedat a scanrate of 20 mV/s in 0.05 M Na$O, (pH 3). The opencircuit potential (OCP) wasfound to be -0.37 V. The first scanwasmadein the negativedirection from the OCP to -0.8 V, where scan direction was reversed. A smallcathodic current can be seen during the cathodicscan,which might be dueto a slow hydrogenevolution reaction. In the subsequent anodicscan,current abruptly increasedat about-0.4 V, formed a peakat -0.05 V, andrapidly decreased with increasingpositive potential. Upon reversingthe scanat 0.25 V, no correspondinganodiccurrent was found, indicating that the Ni(ll1) surface was passivatedby the formation of oxide layers. However, the cathodiccurrent observedin the secondcathodic scanwasclearly largerthan that found in the first cathodic scan, suggestingthat the oxide layer formed at the positive potentials was electrochemicallyreducedto metallicNi. After the reductionof the oxide layer, the anodicdissolution peak appearedagain in the subsequentanodic scan with a similar magnitudeof current density. MacDougall and Cohen reportedthat the oxide film formed on polycrystalline Ni in Na,SO, solutions was composedof NiO with a nearly constantthickness(0.9 - 1.2 nm) in the pH rangeof 2 to 8.4, and that the oxide film could be completelyremovedby a negativescanin solutionsof pH lessthan ca. 3 (230,2311. In the resultshownin Fig. 52, the anodiccurrent, at leaston the rising portion of the peak, is expectedto be due to simpleanodicdissolutionof Ni to form solubleNi*’ speciesin solution. In general,STM imagesacquiredon freshly preparedNi( 111) showed atomically flat terrace-step features,indicatingthat our experimentalproceduresuccessfullyproducedwell-definedsurfacesof Ni. However, we usuallyfound somesmallislandson the atomically flat terracesat OCP, at which potential STM imagesshoweda hexagonalatomic structurewith a lattice constantof 0.29 (M.02) nm. Our STM

192

K. ltaya

1200

I

ii loo0 t z 800 5 t

-200

t I

I

-0.8

I

e

I

I

-0.4 0 ElVvs.SCE

I

I

+0.4

Fig. 52. Cyclic voltammograms on bare Ni( 111) in 0.05M NaSO, (pH 3.0). From [232]. images acquired on these islands seemed to be similar to the image reported previously using an ex-situ STM in air for electrochemically passivated Ni( 111) [229]. Nevertheless, it was found in our study that these islands disappeared when the electrode potential was set at potentials more negative than -0.5 V where the oxide layer was electrochemically reduced, as expected from Fig. 52. Figure 53 shows two examples of in situ STM images with atomic resolution obtained at -0.50 V in 0.05 M Na,SO,. The image of Fig. 53(a) shows that the surface of Ni( 111) in the solution has an atomically flat terrace-step structure.

The monatomic steps shown in Fig. 53(a) are in parallel to the

atomic rows of Ni( 11 l), but sometimes they form long twisting curves. Attention was paid to determining lattice parameters as precisely as possible on Ni( 111) at -0.5 V, because the oxide formed on Ni( 111) also exhibited a hexagonal lattice in STM images with an interatomic distance of ca. 0.29 nm, which is slightly larger than that (0.25 nm) of the Ni( 11l)-( 1 x 1) structure. Figure 53(b) shows an atomic image acquired on a relatively large terrace under such conditions. An almost perfect hexagonal lattice is seen with an angle of 60” (f2”) between the atomic rows. It is also confirmed that the atomic rows are parallel to the close-packed directions of Ni( 111) determined from the orientation of the crystal on the STM stage as shown by an arrow sign in Fig. 53(b). The average value of the nearest-neighbor distances in all atomic directions lies between 0.23 nm and 0.25 nm. We conclude that the surface structure of Ni( 11 I) at -0.5 V is Ni( 11l)-( 1 x 1). This is the first atomic image of bare Ni( I1 l)-( I x 1) obtained in solution. To our knowledge, no STM image of clean Ni( 111) has been presented even in UHV [233], although the structures of Ni( 100) and Ni( 110) have been revealed [234,235].

STM

0

2.5

5.0

Fig. 53. High-resolution

nm

in Electrolytes

7.5 0

193

i.5

5’.0 nm

STM images of Ni( 111) at -0.5 V. From [232].

The Ni( 1 11)( 1 x 1) structure was observed in the potential range between -0.6 and -0.4 V, indicating that no oxidation took place in this range. However, the (1 x 1) structure was also consistently observed even at potentials near the foot of the anodic dissolution. This result indicates that the anodic dissolution occurred with the substrate structure (1 x 1) being maintained at least at the foot of the anodic dissolution peak. We acquired a set of STM images of the same area at -0.3 V where the anodic dissolution took place slowly. The images shown in Fig. 54 are two examples of those acquired with a 50 s interval. The arrangement of atoms as well as the step lines can be clearly discerned in these images. It is seen that the step edge seen at the upper right comer of Fig. 54(a) is retracted in the downward direction with time to expose the underlying Ni(l1 l)-( 1 x 1) plane. No pit formed on the terrace, indicating that the anodic dissolution took place only at the step edges. The result shown in Fig. 54 clearly demonstrates that the anodic dissolution proceeds in the layer-by-layer mode at the onset of the anodic peak. We have also investigated Ni( 100) in the same solution [232], and found that a well-ordered atomic arrangement appeared with a four-fold symmetry. The lattice constant of 0.50 (fO.O1) nm observed in the ordered domains was twice as large as that of Ni( 100). The rows of adatoms were parallel to the substrate Ni( 100) atom rows. The adlattice structure was designated as ~(2 x 2). No Ni( lOO)-( 1 x I) lattice was found in the STM images acquired at -0.5 V. It is well demonstrated by LED [236] and more recently by STM [235,237] in UHV that oxygen adatoms on Ni( 100) form ~(2 x 2) and c(2 x 2)

K. ltaya

194

0

2.5

5.0

7.5 0 nm

2.5

5.0

7.5 nm

Fig. 54. Successively recorded STM images of Ni( I1 I ) at -0.33 V. (b) was acquired aftei 50 s of (a). From [232]. structures. It is noteworthy that the ~(2 x 2) lattice was consistently observed in the potential range between -0.4 and -0.6 V [232]. The ~(2 x 2) oxygen adlayer seems to be a thermodynamically stable phase that is difficult to reduce electrochemically at -0.5 V. We further attempted to acquire an atomic image at a more negative potential of -0.8 V, but failed to achieve atomic resolution because of the hydrogen evolution reaction 12321. When the potential was scanned more positively, the Ni(100) surface was first covered with a hexagonal lattice, then atomically roughened. The layer-by-layer mode observed for the anodic dissolution of Ni(l11)

was not the case for Ni( 100) 12321. The surface of Ni( 100) became rougher with further progress of anodic dissolution. The Ni( 100) surface was expected to be covered partially or fully by oxides depending on the electrode potential even in the active region. Under such conditions, relatively fast dissolution might occur on the Ni( 100)~~(2 x 2) surface. Slow chemical dissolution of the oxide layer might be followed by either me fast dissolution of Ni or the replenishment of the oxide layer. This heterogeneous dissolution seems to be the case of roughening of the Ni( 100) surface brought about by anodic dissolution. (ii) S-modified Ni(lOO). Among numerous EC studies on Ni, Oudar and Marcus previously reported an interesting effect of chemisorbed S on the rate of anodic dissolution of Ni 12381. They found that the presence of a monolayer of S increased the rate of dissolution of Ni substrates. The passive potential region resulting from the formation of Ni oxides significantly shifted toward more positive potentials in H,SO, solutions when the S adlayer was present on the Ni surface. In other words, the corrosion rate of Ni and its alloys was drastically increased by the presence of S adlayers

STM

in

Electrolytes

195

[238]. Potential-pH diagrams for Ni electrodes with S adlayers predicted that the S adlayer is strongly attached on Ni and is very stable in wide potential and pH ranges [239]. We employed in situ STM to investigate the S adlayer structure and the influence of S on the anodic dissolution of Ni(lOO) single crystal electrodes at the atomic scale [240]. The Ni(100) crystals were annealed in a quartz tube under Hz atmosphere (latm.) at 1250 K for 3 hr, the conditions similar to those used for Ni( 111) as described above. The S-modified electrodes were prepared by two different methods after the thermal annealing. One was similar to that previously described by Oudar and Marcus [238]. After annealing in the H, stream at 1250 K, the furnace was cooled down to 750-800 K, and H, gas ( 1 atom.) containing 200 ppm H2S was introduced for 1 min. The cooling was continued to room temperature in pure H, flow. The other was a solution dipping method. The annealed crystals were cooled to room temperature in the quartz tube under H, stream, and then dipped into a hydrogen \aturated 5 mM Na,S solution. The contact with Na,S solution was carried out in the same quartz tube under HZ stream in order to avoid oxidation of clean Ni surfaces by oxygen in air. The S-modified electrodes were rinsed with a HZ saturated Na$O, solution and quickly transferred into an EC cell. Figure 55 shows CVs of bare and S-modified Ni(lOO) electrodes by dashed and solid lines, respectively. OCP for both electrodes were found to be ca. -0.38 V. In the anodic scan from OCP, the anodic current due to the dissolution of Ni increased and reached a peak at ca. -0.05 V on the bare Ni( 100) surface. The anodic current then decreased at more positive potentials because of the passivation by the formation of oxides, which is similar to what was observed with Ni( 111) in Fig. 52. On the other hand, the anodic current surprisingly continued to increase with potential beyond 0 V on the S-modified Ni( 100) electrode in sulfide-free 0.05 M NaSO, (pH 3) as shown by the solid line in Fig. 55. This particular CV was obtained using a S-modified Ni( 100) electrode prepared by the solution dipping method described above. The anodic current was found to be enhanced by the S adlayer consistently during many cycles in the potential range shown in Fig. 55, suggesting that the S monolayer remains on the surface during the anodic dissolution even at such high current densities as 5 tnA/cm‘. A similar remarkable increase in the dissolution current was first observed in a sulfuric acid solution by Oudar and Marcus [238], using S-modified Ni electrodes prepared in a gas phase. An Ni( 100) electrode treated in the gaseous mixture of HZ-HIS was also used in our study, which yielded a CV with almost the same feature as that shown in Fig. 5.5 (solid line), indicating that either method can be employed to fortn a monolayer of S on Ni surfaces. The formation of S adlayers by similar solution dipping methods was previously investigated and characterized on Pt and Pd single crystal surfaces using e.~situ LEED [24 1,242]. It is also noteworthy that the S monolayer was found to be stable even during the hydrogen evolution reaction. The remarkable stability of S monolayers on Ni was recently interpreted based on potential-pH diagrams calculated from standard Gibbs energies of formation of S adlayers [239]. Figure 56 shows a typical example of topographic STM images of the S-modified Ni( 100) electrodes acquired at -0.45 V. The scan area was 200 x 200 nm. All steps observed were mostly monatomic with a height of 0.176 nm. Wide terraces extending over 100 nm were usually observed on wellprepared S-Ni( 100) electrodes as shown in Fig. 56. A preferential orientation of step lines is seen in the

196

K.

-I’

’

-0.8

ltaya

‘

-0.4

0

0.4

E/Vvs.SCE Fig. 55. Cyclic voltarnmograms of S-modified Ni(100) (solid line) and bare Ni(100) (dotted line) in Na$O,. From 12401.

200 nm Fig. 56. Topographic STM image of S-modified Ni( 100) in NqSO,.

From [240].

() WI tr,

STM in Electrolytes

a

197

Will

[oioj

c(2 x 2)

Fig. 57. High-resolution STM image (a) and a model structure (b) of S-modified Ni(100). From [240]. image acquired even before the anodic dissolution. The step lines seem to be nearly in parallel with the [OOl] and (OlO] directions, which are rotated 45” off from the close-packed direction ([Oil]) underlying Ni(lO0).

of the

Figure 57(a) shows an STM image with atomic resolution obtained on an atomically flat terrace. It is cleariy seen that the S adlayer has a structure with four-fold symmetry, i.e., a square lattice, The height of each S atom was typically 0.04 nm. The atomic rows of 5’ are also rotated by 45’ with respect to the close-packed direction of the underlying Ni( 100). The observed lattice constant of ca. 0.35 nm corresponds to d2 times the lattice constant of Ni(lOO), 0.25 nm. This struchlre is designated as ~(2 x 2) as shown jn Fig. 57(b), It is also interesting to note that single atomic defects of S were sometimes observed in well-ordered S adlayers as shown in Fig. 57(a).

Those single atomic defects were

continuously observed at the same relative locations in subsequently acquired images for at least 30 min, indicating that the surface mobility of adsorbed S atoms was very small. Figure 58 shows an STM image of a relatively large area (10 x 10 nm) with atomic resolution, including three atomic layers. It is easily seen that the S adlayer forms the c(2 x 2) structure over terraces. A single domain of the S adlayer was found on each terrace. Phase boundaries were not usually observed on each atomically flat terrace, suggesting that the S adiayet with a Iong-range ordering formed on the WlOO)

surface. The image shown in Fig. 58 also reveals detailed stmctures of

K. ltaya

198

Fig. 58. High-resolution

STM image of S-modified Ni( 100) near step edges. From [240]

(W

0

20 nm

nm

Fig. 59. Successively recorded STM images of S-modified Ni(100). (b) was acquired after 7 s of (a). From [240].

STM

in Electrolytes

199

the step edge. Individual S atoms on the upper edge of the step marked (a) can be clearly seen with the same corrugation amplitude as observed at the center of the terrace, suggesting that the S atoms along the step lines are also located on the four-fold hollow site, similarly to those at the center of the terrace. The step (b) is atomically straight and parallel to the direction of the S atom row, while the step (a) includes kink sites. As described in the following section, the anodic dissolution seems to occut preferentially along the [OlO] direction, starting from the kink sites followed by row-by-row dissolution at the step edges. We pursued an investigation on the effect of adsorbed S on the dissolution at potentials near the onset of anodic current. Soon after the acquisition of the image shown in Fig. 58, the electrode potential was slightly shifted positively to follow the movement of the step line resulting from the anodic dissolution. The c(2 x 2) structure was consistently observed on the terrace even under the anodic dissolution. Figures 59(a) and 59(b) illustrate the time-dependent etching process of S-modified Ni( 100). These images were acquired consecutively with a time interval of 7 s in an area of 20 x 20 nm. All lines seen in the images shown in Fig. 59 were found to represent monatomic steps. We focused our attention to the progressive etching of the top-most layer, marked T,, in the image shown in Fig. 59(a). As described above, the step lines are nearly parallel to the [OOl] or [OlO] direction, but they are not always straight, indicating that some step lines include kink sites as shown in Fig. 58. As a general feature, it is clear that the anodic dissolution takes place preferentially at the step sites. The steps continued to recede to dissolve the terrace (T,). No pit formation was found to occur on the terrace, indicating that the dissolution of Ni on S-modified Ni(100) proceeds by the layer-by-layer mode. In the first 7 s, relatively fast etching took place along the [OlO] direction in the lower part of the terrace T,, splitting the continuous terrace into two domains. The step S, with a staircase shape also disappeared and was converted to a relatively straight step line along the [Ol l] direction as shown in Fig. 59(b). Note that the [Ol I] direction is parallel to the close-packed atomic rows of the underlying Ni atoms. It was interestingly found that the comer marked C, in Fig. 59(a) remained in all subsequently acquired images, while the comer marked C, was rapidly retracted in the first 7 s from Figs. 59(a) to 59(b), forming a step line along [Ol 11. It is also important to note that the upper part of the terrace marked A with the shape of a narrow strip became smaller in width in the first 7 s but remained unchanged in the subsequent images. This observation strongly suggests that the etching rate along the [OOl] direction is significantly slower than that along the [OlO] direction. Such an anisotropic etching process seems to be important in explaining the morphological change in the shape of terraces during the dissolution illustrated in Fig. 60. The anisotropic dissolution process can be further elucidated by the structural models depicted in Fig. 60 [240]. Figure 60(a) shows a model structure of the S adlayer on Ni( 100) including three step lines along [OOl], [OlO], and [Ol l] directions, respectively. The S adlayer was drawn to scale using the interatomic distance (0.25 nm) of Ni and van der Waals diameter (ca. 0.35 nm) of S atom [242]. The adsorbed S atoms on Ni( 100) seem to form an almost close-packed adlayer, because the diameter of S is nearly equal to 42 times the lattice constant of Ni( 100). All S atoms on the upper terrace should be located at the four-fold hollow sites even near the step edges, because the S atoms on the upper terrace

K. ltaya

Fig. 60. Structural models of S-adlayers on Ni( 11 l)-( 1 x I). From [2401. along the step edges appeared as spots with the same corrugation height as that observed at the center of the terrace as shown in Fig. 58. According to the model structure shown in Fig. 60(a), neighboring S atoms along the step edges on the lower terrace are located with different spaces depending on the orientation of the steps. There is almost no space at the step along the [OOl] direction. Each Ni atom located on both the upper and the lower terraces along this step is coordinated by an S atom. On the other hand, the steps along the [OlO] and [Ol l] directions provide larger spaces near the step edge. It is important to note that the step lines along the [OOl] and [OIOJ directions become nonequivalent because of the presence of the S adlayer with the c(2 x 2) structure. The S atomic rows should be shifted by a half space at the step edge along the [OlO] direction as shown in Fig. 60(a). On the other hand, the S rows are continuously aligned in the straight line along [OOl]. Such a difference in the alignment of the atomic rows of S is seen in the STM image in Fig. 58. The S rows at the step edge along [OlO] were indeed shifted by a half space. Figure 60(b) illustrates the atomic arrangement at a comer of the upper terrace. This comer corresponds to that marked C, in Fig. 59(a) which was stable against the anodic dissolution as described above. The nonequivalent step lines along ]OOl ] and ]OlO] can also be seen in Fig. 60(b). We proposed a model to explain the local anisotropic etching process using the local atomic structures shown in Fig. 61. The energy required to remove a Ni atom from the surface can be estimated by a simple bond counting model [243,244]. The single Ni-Ni bond energy (E,,-,i) is 0.365 eV ( l/12 of the sublimation energy of Ni (4.38 eV)). The bond energy between Ni and adsorbed S atoms (I&) can be estimated as 0.456 eV from the free energy of the adsorption of S on Ni( 100) (l/4 of 1.82 eV ]239]) in which the coordination number (4) of the adsorbed S on Ni( 100) is taken into account. The energy required to remove each Ni atom from the step edge along [OlO] equals 6ENIeN,+ En,.,, because one Ni atom at the step edge has six neighboring Ni atoms and one S atom when the S atom

STM in Electrolytes

201

Anodic disolution

(a)

td)

(e>

Fig. 61. Schematic illustration of the anodic dissolution process of S-modified Ni( 100). From [240]. which has been coordinated with the removed Ni atom stays on the upper terrace after the removal. However. if the S atom can slide down to the lower terrace to find a four-fold hollow site during the removal of the Ni atom, the total energy becomesonly 6E,,.,,. The movement of the S atom at the step edge from the upper terrace seems to be a key factor in determining the preferential etching process. The Ni atom such as (g) in Fig. 61(c) at a step site along [OiO], except the one at the comer such as that marked (a) requires the energy of 6E,,.,, + E&S for removal, becausethe S atom cannot find a hollow site on the lower terrace due to its van der Waals diameter. On the other hand, the Ni atom (a) can be more easily removed from the upper terrace, because the S atom attached on Ni (a) slips down to the lower terrace during the removal of the Ni atom. The neighboring Ni atom (b) should be easily removed from the surface during or after the removal of Ni (a), becausethe Ni atom (b) is coordinated only with six neighboring Ni atoms. Therefore, the simultaneous removal of two Ni rows along the [OlO] direction is expected to accompany the movement of the first S row on the upper terrace to the lower one as illustrated in Figs. 61(a) and 61(b). After the dissolution of Ni atoms (a and b), the Ni atoms (c and g) can then be easily removed in the samemanner, and the etching is propagated in the [OlO] direction. The above mechanism explains the row-by-row etching process along the [OlO] direction. The first removal of the Ni atoms such as (c) and (e) needs the energy of 7E,,.,,, even though the S atom can slide down to the lower terrace after the removal of these Ni atoms. This explains the faster etching of Ni

202

K. ltaya

along [OIO] compared with that along [OOl]. The instability of the steps along the [Ol l] directicon ian bc explained as a consequence of the etching along the [OlOJ direction from the step edge along 10 ! i i The row-by-row etching mechanism proposed here also explains the etching profile at the comer ot (. in Fig. SC)(a). The step line along [OOI] involving the comer C2 became one of the steps along 10I 11 a\ shown in Fig. 59(b). This process is expected to occur when the etching starts with the corner atom b) the row-by-row mechanism. The model structure shown in Fig. 60(b) can explain the high stahllity ot the comer marked C, in Fig. 59(a). Both Ni atoms marked by arrows in Fig. 60(b) need the higher energy of 7E,,+,, + E,,., to remove because there is no vacancy to accept the S atom on the lower terrace. The high stability of this type of comers will terminate further etching along the atomic rows in the lolO] direction. The propagation of etching along the rows indicated by the dashed arrow in Fig. 60(b) will be terminated at the comer. (iii) Ag(100). It is now clear that the adlayer plays an important role in the anodic dissolution processes of metals. This sectiondescribesthe anodicdissolutionof an iodine-modifiedAg(100) [245]. A commercially grown Ag( 100) single crystal was metallographically polished and sequentially sonicatedin acetone,methanol,andpure water. The singlecrystal was then placedfor 2 hr in a quartz tube continuouslypurgedwith purified H, and maintainedat 1100 K. The samplewas cooledto room temperatureunder H, steamand subsequentlybrought into contact with ultrapurewater saturatedwith H, and then transferred into an EC cell with a droplet of pure water on it to mitigate surface contamination. For the preparationof the iodinemonolayeron Ag, the cleansurfacemustbe exposedto the KI solution under potential control. Immersionat OCP leadsto the formation of bulk AgI and roughensthe surface (64). It should be mentionedthat several recent studies on I-pretreatedAg electrodesusingSTM [27] andX-ray photoelectronspectroscopy[247] may have beencomplicatedby the presenceof AgI multilayers sincethe l-adlayers in those studieswere preparedby immersionat OCP. In our study, the clean Ag(100) electrodewas immersedin a solution that contained0.1 M HClO, and 1 mM KI at a specific potentialin the double-layerregion. After 5 min, the electrodewas emersedat the samepotential, and immediatelytransferredto the EC cell containing pure HClO, for voltammetry and STM experiments. Figure 62(a)showsa typical large-scaleSTM imageof the bareAg( 100)electrodein 0. 1M HClO, at 0.1 V [245,246]. Atomically flat terracescan be seenthat extend over a few hundred nm. Steps, mostly monatomicin height,arecharacterizedby arc-shaped(ratherthan straight) edges;relatively large islandsarecircular in morphology. A similarsurfacemorphologyhasbeenreportedfor real and quasiperfect Ag( 100)electrodes[ 168,248]. The shapeandlocation of the steplinesobservedin Fig. 62 were not significantly changedin the double-layerpotentialrange,suggestingthe absenceof potential-induced reconstructionon the bare Ag( 100). Figure 62(b) showsa high-resolutionSTM imageacquiredon the atomically flat terrace. It can be clearly seenthat the Ag( 100) surfacehas a square lattice with an interatomicdistanceof ca. 0.29 nm along the ( 110) direction; the corrugation amplitudeof each Ag atomis about0.05 nm. The imageshownin Fig. 62(b) demonstrates that, under the presentconditions, the Ag( 100) surface retainsthe unreconstructed(1 x 1) structure, consistentwith previous results [ 168,170]. Imagesobtainedin the double-layerpotential rangealsoshowedthe (1 x 1) structure; the

STM

50

nm

in Electrolytes

203

100

Fig. 62. Topographic (a) and high-resolution (b) STM images of bare Ag( 100) in HClO,. Arrows indicate the ( 110) direction. From [245]. absence of potential-induced reconstruction is thus demonstrated. (iv) I-modified Ag(lOO). Figures 63(a) and 63(b) show a typical large scale and a high-resolution STM images of the I-Ag(100) surface in HClO,, respectively. It is interesting to note that the monatomic steps now run parallel to the ( 100) direction, rotated by 45” with respect to the atomic rows of the Ag substrate. The change in the step-direction may have occurred upon exposure of the Ag( 100) surface to the RI solution. The interatomic I-I distance was found to be 0.4 nm, which corresponds to 42 times the lattice constant of Ag( 100). Thus, it can be concluded that the I adlayer possesses a ~(2 x 2) structure. Previous UHV-based studies on the chemisorption of iodine from I, vapor showed a ((2 x 2) LEED pattern on Ag( 100) [249,250]. Our in situ STM result is thus in good agreement with the ex situ LEED data. A more important observation is that the ~(2 x 2) structure was consistently found in the double-layer region; at least under the present conditions, no potential-induced structural changes in the iodine adlayer are indicated. It is important to view the potential-independence of the I-Ag( 100) adlayer structure in the context of what has been observed for the I-Ag( 111) adlayer [64,7 11. Our recent work on I-Ag( 111) adlaycr by means of tandem in situ STM and ex situ LEED revealed a continuous compression of the adlayer, from (43 x 43)R30”, via c@ x 43R-30”), into a rotated hexagonal phase in acidic HI solutions [64], when the I coverage was increased as a result of a positive shift in the applied potential. In general, the (Ill) planes of Ag and Au exhibit several incommensurate iodine structures [63]. The single commensurate structure of I on the Ag (100) surface may be explained by the difference in the coordination numbers of the iodine adatoms on the ( 100) and (111) surfaces. Each iodine adatom on Ag (100) is expected to be located on a four-fold hollow site; hence, each iodine atom has a coordination number of four.

204

K. ltaya

Fig. 63. Topographic (a) and high-resolution (b) STM images of I-modified Ag( 100) in HClO,. From [245]. In comparison, an iodine atom on a three-fold hollow-site on Ag( 111) has a coordination number of only three. The exceptional stability of I on Ag( IOO), due to a higher coordination number, may explain the formation of the single commensurate c(2 x 2) structure. As shown in Fig. 63(a), the monatomic step-lines found on the I-Ag(100) are mostly parallel to the atomic rows of iodine atoms in either the [OOl] or [OlO] direction, rotated 45” with respect to the closepacked Ag atomic rows. Figure 64 shows a set of STM images acquired sequentially after a time interval (64 s) at the onset of (low-current) anodic dissolution. It can be seen that although the step-lines remained nearly parallel to the [OOl] and [OlO] directions, they were not always atomically straight. This result indicates the presence of kink sites along the step-lines. For example, whereas the step-line marked S, is almost perfectly aligned along the [OlO] direction, the S, step is not. The initial jaggedness in the latter (SJ step-lines was rendered smooth as a result of the anodic dissolution; staircase-shaped edges were formed as shown in Fig. 64(b). The step-edge marked by the arrow in Fig. 64(a) also includes several kink sites of various orientations. This step-line was likewise converted to a staircaseshaped step-edge (Fig. 64(b)). Although it has become clear that the etching of Ag on I-Ag( 100) takes place exclusively at the step-edges, one can recognize that the etching rate on I-Ag( 100) along the [OlO] direction is faster than along the [OOl] direction. As an illustration of this anisotropic etching, it can be seen that the step-lines marked C, and C, in Fig. 64 remain unchanged, while the steps along the [OlO] direction are progressively retracted. Several other sets of time dependent STM images were obtained in different areas during the anodic dissolution; invariably, similar anisotropic etching processes were found. The anisotropic dissolution found on I-Ag(100) can be rationalized in terms of the surface structural

models shown

in Fig. 65.

These

models

are similar

to

those

shown

in

STM

I

100 nm

in

Electrolytes

I

Fig. 64. Successively recorded STM images of I-modified Ag( 100). (b) was acquired after 64 s of (a). From [245].

Fig. 65. Schematic illustration of the anodic dissolution process of I-modified Ag( 100). From [245].

205

206

K. ltaya

Fig. 60. but include three atomic layers with the steps along either [OlO] or [OOI ] direction; the top-most layer is shown on the lower right-hand section. In the absence of iodine, the step-lines along [OOI ] and [OlO] directions are equivalent, but they are non-equivalent in the presence of the c(2 x 2) adlaycr. In the proposed model, all I atoms on the terraces are located at the four-fold hollow sites, even near the step-edges. As shown in Fig. 65(a), the arrangement of I near the step-edges is dependent on the direction of the step-edges: (i) the steps along the [OlO] direction include larger spaces near the edge compared to those along [OOl]; (ii) the I rows are aligned in a straight line across the three atomic layers along the [OlO] direction; and (iii) in the [OOl] direction, the I rows are shifted by a half-space at each step-edge. These differences in the alighnment of the atomic rows had been clearly discerned by highresolution STM images of S-Ni( 100). The upper-to-lower-terrace diffusion of the first iodine row is driven by the need to minimize the free energy change during Ag dissolution along the step-edges. Because of its open structure, the anodic dissolution along the [OlO] direction is expected to occur more easily than along the [OOl] direction. The Ag atoms marked A and B in Fig. 65(a) are thus the first to be dissolved accompanied by the diffusion of the iodine atom from the upper to the lower terrace. After dissolution of A and B, the same process is expected to continue along the [OlO] direction; this results in the propagation of the anisotropic etching along the direction marked by the arrows in Fig. 65. It is important to note that the different alignments of the atomic rows of iodine do not always extended through several atomic layers. The I adlayer on each terrace has an equal probability to be shifted by a half-space with respect to the adlayers on lower and upper terraces; such shift may result in an alternative alignment as shown in Fig. 65(b). In this case, the anisotropic etching of the top-most layer is expected to proceed along the [OOl] direction as indicated by the arrow. The etching of the second layer must then propagate along the [OlO] direction. It should be noted that this mode of anisotropic etching, rotated by 90” with respect to a particular step-terrace configuration, was observed on much higher or much lower terraces. Nevertheless, the overall features for the anodic dissolution of Ag on I-Ag( 100) can be explained by the model structures shown in Fig. 65. (v) Iodine-modified Pd Electrodes. Recently, Soriaga and coworkers discovered using the UHV-EC that the anodic dissolution of Pd( 111) and (100) single crystal electrodes in pure H,SO, solutions was catalyzed by the presence of monolayers of iodine chemisorbed on these surfaces [25 l2531. Large anodic peaks were found for the dissolution even in non-corrosive electrolyte solutions only when the surfaces of Pd were modified by the iodine adlayer. In this section, we present only the results of our in situ STM studies performed in collaboration with Soriaga, which extends our understanding of the role of adsorbed iodine in the anodic dissolution [59]. Iodine pretreatment was accomplished simply by immersion of the clean electrode, without potential control, in an aqueous 1 mM solution of iodide for 5 minutes. Unadsorbed iodide was removed by rinsing with pure water. The electrochemistry of iodine-free palladium is expected to be dependent upon the crystallographic orientation of the surface in contact with the electrolyte. Such a dependence was well-demonstrated for the oxidation of Pt single crystal electrodes [25]. Figure 66 shows CVs of Pd( 11 l), Pd( 100) and

STM

0

207

in Electrolytes

1.0

0.5

1.5

ElVvsRHE Fig. 66. Cyclic voltammograms of bare Pd(l1 l), Pd(lOO), and Pd( 110) in 0.05 M H,SO,. Scan rate; 20 mV/s. From [59].

400 -

cII 300 'E p *oo. loo-

0-1oo0

0.5 ElVv.s.RHE

1.0

1.5

Fig. 67. Cyclic voltammograms of bare and I-modified Pd( 100) in H,SO,. From [59].

208

K. ltaya

Pd(l10) in 0.05 M H,SO,. All three single-crystal surfaces exhibited sharp anodic oxidation peaks at anodic potentials Eo, that increased in the order: Pd( 110) (E,,, = 0.85 V) < Pd( 1(Ml) iI:,,, = ii,90 VI . Pd( 111) (E,,, = 1. IO V). In contrast, only a broad, undefined peak was obtained at a polycrystalhnc surface [ 25 l]. The structure-sensitivity of E,, almost certainly originates from differences IR surtaceatom arrangements at the three low-index planes [25]. When the Pd electrodes were pretreated with iodine, dramatic changes took place in CVs recorded under conditions identical to those for Fig. 66, as shown in Fig. 67 for an iodine-modified Pd( 100) surface. Two notable features are: (i) the suppression of the surface-oxidation peak at E,,,; and (ii’) the emergence of a new anodic peak (E’,,, = 1.25 V) that yields a peak current nearly an order-of-magnitude larger than that for the I-free surface. If the potential is held at a value just below E’,,,, the current does not decay but remains essentially constant, a behavior diagnostic of a material-limited dissolution process. These results are very similar to those found on S-Ni( 100) as described in a preceding section. Previous work on polycrystalline and single-crystal Pd electrodes, based upon techniques including coulometry, atomic emission spectroscopy, electron diffraction, and surface elemental analysis, has established that the anodic peak is due to the two-electron dissolution reaction : Pdts) = Pd”(,q)

+ 2e

(1)

Similar CV features were observed for the dissolution of I-Pd( 11 l), I-Pd( 100) and I-Pd(ll0) surfaces in 0.05 M H,SO,, depending slightly on the crystallographic orientation [ 591. (a) I-Pd(111). In 0.05 M H,SO, and at potentials within the double-layer region (0.3 V < E < 0.7 V), wide atomically-flat terraces with monatomic steps that intersected one another at an angle of 60“ or 120” were observed at both iodine-free and iodine-pretreated Pd( 111).

Atomic-resolution

images

showed that the iodine adatoms were aligned along the ( 12 1) direction; that is, along a vector rotated 30” with respect to the substrate-atom rows. The distance between nearest-neighbor atoms was determined to be 0.47 * 0.01 nm. The observed structure, Pd( 11 l)-(43 x 43)R30”-1: is identical to that reported earlier based upon LEED measurements j25 11. The potential was scanned to and held at 1.05 V, a potential at which metal dissolution is initiated. Figure 68 presents two STM images, in the same (300 x 300 nm) domain, obtained at the beginning and 25 min after the start of the anodic dissolution, respectively. The progress of dissolution can be monitored via topographical changes in the terrace (designated by a T) and the step-edge (marked with an M) within the rastered domain. It is easy to recognize that, as the low-current dissolution proceeds: (i) the step-edge retracts upward, (ii) the area of the lower terrace increases, and (iii) the (diminished) upper terrace and (enlarged) lower terrace remain pit-free. In addition to these obvious features, one can also note: (i) the gradual dissolution of the islands on the upper left-hand comer in Fig. 68(a), and (ii) the characteristic jagged outline (contour) of the receding upper terrace. The latter observation suggests a preferential dissolution process along either substrate or adsorbate atomic rows.

STM

260

IdO

209

in Electrolytes

300

100

300

ioo

nm

nm

Fig. 68. Successively recordedSTM imagesof I-modified Pd(111). (b) was acquired25 min after acquisitionof (a). From [59].

0

40

nm

80

0

40 nm

80

Fig. 69. SuccessivelyrecordedSTM imagesof I-modified Pd(100) with a time interval of

1 min. From [59].

210

K. ltaya

The images shown in Fig. 68 clearly point to a step-selective layer-by-layer dissolution mechanism. It is noteworthy that high-resolution STM of the post-corrosion terrace rcvcaled an iodine adlattrcc structure identical to that observed before the anodic dissolution; that is. a well-ordered Pd( 1I 1!-I \ 3 r 43)R30”-I adlattice was retained after the corrosion reaction. This observation is in consonance with the data reported earlier on the basis of UHV-EC measurements [2.511. As with the Pd(11I) surface, atomically flat terraceswith monatomicstepswere (b) I-Pd(lOO). observedat both the cleanand iodine-modifiedPd(100) facets in 0.05 M H,SO, in the double-layer region. An orderediodine adlayerstructurewasalsoobservedat flat terraceswith a ~(2 x 2)-I adlattice structure [59], which is consistentwith the structure previously determinedby LIED experiments (2521. Before the anodic dissolution, the surfacemorphology was similar to that observed for the bare Pd(100). Two types of stepswere found in STM images. One is parallelto the ( 1lo} direction, and the other was alignedalong the { 100) direction. At a potential of 0.95 V. a preferentialdissolution arisedfrom differencesin the step-atomconfigurations. The anodic dissolutionwas only taken placeat the steps.parallelto the { 100) direction. It was realizedthat, in contrastto the disorder that ensured from etching at 0.95 V. dissolutionat 1.05 V resultedin an atomically flat surfacetopography. After removal of several layers by the anodic dissolution, the direction of the step line was drastically changed. STM imagesobtainedafter 5 and6 minutesat 1.05 V are shown in Fig. 69: only rectangular terraceswith stepsalong the ( lOO} edgeand its orthogonaldirection are recognizable. All stepswere now only monatomicin height; that is, anodicdissolutionled to diminutionof the stepline parallelto the atomic rows of Pd. Closerinspectionof the STM imagesin Fig. 69 revealsthat dissolutiontranspiredonly at step-edges; corrosion. via pit-formation, was not observed at the terraces (T). More significantly, it can be recognizedthat dissolutionoccurred anisotropically. For example, the “narrowing” (and subsequent disappearance) of the T,, T,, and T, terracesdevelopedalmost entirely along the [OOl]-directed step, with only minimalnarrowingin the directionperpendicularto it. Such anisotropicdissolutionhasbeen observed for sulfur-modified Ni( 100) in acid sulfate solutions and for iodine-modifiedAg( 100) as describedabove. The absenceof pit-corrosion at the terraces, coupled with the selective dissolution at and the continuousretractionof the step-edges,provide evidencethat the corrosionof I-Pd(lO0) proceedsvia a layer-by-layer mechanism. In this regard, Pd(lOO)-c(2 x 2)-I behaves similarly to Pd(l1 l)-(43 x v’3)R30”-I. The presentcaseof anisotropicdissolutionmay be understoodin terms of the schematic modelsshown in Figs. 60, 61, and 65. Removalof Pd atomseasily permitsthe iodine atom to slide down from a four-fold site at the upper terrace onto another hollow site at the lower terrace in a particular direction dependingon the arrangementof iodine.atoms near the step edges. The easeby which iodinemovesfrom a high-symmetrysite from the upperto lower terraceseemsto be alsothe key factor for I-Pd( 100)asthe major driving force in anisotropicdissolution. (c) I-Pd(ll0). In contrastto the other two low-index planes,wide atomicallyflat terraceswere not obtainedwith I-free Pd(1lo)-( 1 x 1) nor with I-Pd( 110) surface[59]. The existenceof a handful of randomlylocatedpits, a few nm in diameterandup to threeatomic layers in depth, was also recognized

STM

in Electrolytes

211

during the anodic dissolution, resulting in a disordered surface. The propagation of pit-corrosion on the upper terrace and inside the rectangular pits leads to progressive deterioration of the I-Pd( 110) surface, a result that was initially indicated by the LEED data [254]. The formation of these pits is simply due to the dissolution of more active Pd atoms on the Pd( 110) terrace-plane. In view of the aggressive reactivity of the I-Pd( 110) surface relative to I-Pd( 111) and I-Pd( lOO), it may just be that pit-corrosion on I-Pd( 110) terraces cannot be completely eliminated. (vi)

Cu electrodes.

Suggs and Bard reported detailed dissolution processesof Cu( 111) and Cu(100) in HCl solutionsusing in situ STM [101,102]. It is well-known that chloride ions am relatively strongly adsorbed on Cu electrode, forming specific structures depending on the crystallographicorientationof Cu singlecrystal electrodes.It was demonstratedin their papersthat the anodicdissolutionof both electrodesproceededby the layer-by-layer modein HCl. We were particularly interestedin the anodicdissolutionof Cu( lOO), on which the adsorbedchloride forms the ~(2 x 2) structure [102]. As describedabove, the anodicdissolutionof Ag(lOO)-c(2 x 2)-I and Pd(100)~~(2x 2)-I is similar to that of Ni( lOO)-c(2x 2)-S. The striking similaritiesobservedfor the threedifferent (100)oriented metal substrates(Ni, Ag, Pd) pretreatedwith different adsorbates(S, I) but having identical~(2 x 2) structures,strongly suggestthat adlayerstructureplays a dominantrole in the anodic dissolution of (lOO)-oriented metal surfaces. In the in situ STM study of the anodic dissolutionof Cu( 100)in HCl by SuggsandBard [ 1021,anisotropicdissolutionalong ( 100) directions was alsofound. Since the Cl-on-Cu(100) adlayerpossesses a c(2 x 2) structure, we anticipatedthat a rationaleidenticalto that proposedfor Ni( 100)c(2 x 2)-S, Ag( 100)x(2 x 2)-I and Pd(100))~(2x 2)-I is operative in the Cl--assisted dissolutionof Cu(100) [255].

d

nm

1;

O

nm

15

Fig. 70. Successivelyrecordedhigh-resolutionSTM imagesof Cu(100) in 0.01 M HCl. From [255].

212

K. ltaya

I

4nm

I

Fig. 71. High-resolutionSTM image (a) and a model of dissolutionof Cu( 100) in HCl. From [255].

Figure 70 shows two high-resolutionimagesacquiredconsecutivelywith an interval of 13 s in the sameareain 0.01 M HCI. Each smallspotcorrespondsto a chlorideion adsorbedon Cu(100) with the 42 x 2) structure. It is surprisingthat the well-ordered~(2 x 2) adlayer can be seenon all terraces. Interestingly,the chlorideatomic rows arenot shifted by a half space,andthey arecontinuously aligned alongthe [OlO] direction throughseverallayers. We focusedour attentionon the progressiveetchingof the terracemarkedT,. A new stepwith four chlorideatomic rowsappearedat the bottom of the terraceT,, andit developedalong the [OOl] direction as can be shown in Fig. 70(b). This step was consistently straight along the [OlO] direction during etching and finally mergedinto the step along the [OlOJ direction, forming a comer illustrated in Fig. 60(b). Figure 71(a) showsa magnifiedimagenearthe stepwith four atomicrows. It is more clearly seen that the atomicrows alongthe (OlO]direction are not shiftedby a half spacealong the [OlO] direction at the stepedgeasshownby the dashedline markedL,. On the other hand, the Cl atomicrows along the direction of [OOl] are indeedshifted by a half-spaceat the step edgeasshown by the line L,. Figure illustratesthe atomicarrangementnear the stepedge. According to the modelsshown in Figs. 60 and 65, the etching shouldproceedby the row-by-row mechanismalong the [OIO] direction as indicatedby the arrow in Fig. 71(b). It is now concludedthat the anodicdissolutionof the four different substrates (Ni, Ag, Pd, Cu) with the (100) orientationwith different adsorbates(S, I, Cl) takesplaceby the same row-by-row mechanism. (vii) Other metals. (a) Au(l II). In early 1989, Trevor et al. demonstratedthat the surface of a roughenedA~(11 1)

becameatomicallyflat at potentialsin the doublelayer region due to the surfacemigrationof Au atoms,

STM

in Electrolytes

213

or the so-called EC annealing [256]. Pits growth and disappearance were explained by the diffusion of Au atoms at the step edge. More importantly, Trevor et al. and we independently discovered that the adsorbed chloride on Au( 111) significantly enhanced step motion [256,2.57]. The anodic dissolution of Au( 111) in a solution containing chloride ions proceeded by the layer-by-layer mechanism [257]. More recently, adsorbate-induced etching of Au( 111) was demonstrated using the organic compound, tetramethylthiourea [258,259]. Such processes might be expected to occur on various metal surfaces, resulting in the formation of well-defined surfaces. (b) Co(OOO1). We recently a solution of Na,SO, at pH electrode in solution [260]. flat terrace step structure at solution of Na,SO,.

carried out an in situ STM study of a Co(OOO1) single-crystal electrode in 3.0 to establish the possibility of exposing a well-defined surface of the It was found that the anodic dissolution of Co(OOO1) produced atomically a potential corresponding to the onset of the anodic current in the acidic

Freshly prepared Co(OOO1) surfaces were usually atomically rough at OCP in

0.05M Na,SO, (pH 3). Atomically rough surfaces were transformed to well-defined terrace-step features at potentials near the onset of the anodic dissolution. After the removal of several atomic layers by anodic dissolution, the electrode potential was set at a value in the relatively narrow double-layer potential range. Figure 72(a) shows a typical in situ STM topographic image of Co(OOO1) in 0.05M Na,SO, (pH 3) [260]. An atomically flat terrace-step structure is seen on the surface with terrace widths of 50-100nm. All steps observed are with a height of 0.2 nm, which is consistent with 0.204 nm high monatomic steps on Co(OOO1). Although only small terraces were usually obtained in the beginning of the anodic

(b)

I

I

I

I

I

I

I

0

100

200

0

1.0

2.0

3.0

nm

nm

Fig. 72. Topographic (a) and high-resolution(b) STM imagesof Co(OOO1)in Na,SO,. From [260].

214

K. ltaya

dissolution, the terrace width increased with time during the anodic dissolution by the layer-by-layei mechanism. Figure 72(b) shows an STM image with atomic resolution obtained on an atomically flat terrace. Nearly perfect hexagonal lattices can be seen clearly with an angle of 60’ (+2”) between the atomic rows. The observed lattice constant of ca. 0.25 nm corresponds to the known lattice constant ot Co(OOOl), 0.2507 nm. It is concluded that the surface structure is designated as Co(OOOl)-( 1 x 1). The STM image shown in Fig. 72(b) is believed to represent the first atomic image of a clean Co@00 1)-( I I 1) obtained in a solution. It is clear that the anodic dissolution will become an important ilz situ technique to expose atomically flat terrace-step structures in solution. This technique has also been applied to Fe( 1IO), revealing the ( I x 1) structure at a potential near the hydrogen evolution reaction in a HCl solution 126 11, which significantly suggests that the anodic dissolution method can be applied to various metals. F. Electrochemical Dissolution Processes of Semiconductors The preparation of clean and stable semiconductor surfaces is the first step in the manufacture of semiconductor devices. The drive towards sub-micron technology for ultra-large scale integrated circuits has focused special attention to wet chemical processes 12623, since high-temperature-based procedures often lead to adverse effects that arise from new and difficult-to-control reaction channels: the pursuit of nanometer-scale technology has likewise necessitated the development of surface characterization methods that enable atomic-scale resolution. Since commercial integrated circuits are still based exclusively on silicon, the wet-chemical processing of Si single-crystal surfaces has been and continues to be widely investigated; much of the interest centers around the nature of the hydrogen termination of the Si surface atoms. The first work to demonstrate that a Si( 111) surface etched in aqueous NH,F is ideally terminated with Si monohydride was based upon IR spectroscopic studies of the Si-H vibrational modes [263]. This original work stimulated further investigations, likewise based upon IR and other spectroscopic techniques, on Si( 100) and Si( 110) substrates [264-2671. STM in vacuum directly established the fact that the NH,F-etched Si(ll1) surface was atomically well-defined with an ideal H-terminated Si( 11 I):H-( 1 x 1) structure [268-2711. While these studies provided critical information on the postetched Si surfaces, it was clear that in situ investigations had to be undertaken if the chemical etching process is to be understood at the atomic level. In response to this need, in situ STM and AFM were adopted to semiconductor-etching studies. Eventually, direct atomic-scale interrogations of the surface structures of Si [268-2801, GaAs [2812831, GaP [284], InSe [285], and InP [286], in the course of the etching reaction, became a reality. Our first in situ STM observation [272] of the Si( 11l):H-( 1 x 1) atomic structure in a non-corrosive solution (aqueous H$OJ spurred investigation of the etching of Si( 111) in corrosive solutions, such as aqueous NH,F [273-2761, NaOH [277,278] and HF [280]. It was soon determined that the etching of Si( 111) was potential-dependent. At potentials markedly negative of the GCP, the etching rate decreased and the dissolution proceeded via a step-selective layerby-layer mechanism [273-2781. At potentials near or more positive than the GCP, the etching rate increased and pit corrosion transpired on the terraces, which resulted in atomically roughened surfaces

STM

in

Electrolytes

215

[277]. The potential-dependence [273-278,287] of the Si-etching process is a technologically relevant issue. Industrial wet-etching processes are usually performed without potential control [262]; hence, the surface chemical reactions that characterize the simple dipping of Si wafers in an etching bath are expected to bear strong resemblance to what are observed from EC etching at the OCP. The ability to control the applied potential during the etching process may serve as an additional critical factor in the preparation of atomically well-defined semiconductor surfaces. We have also demonstrated that in situ STM can be employed to monitor atomic scale features of the etching process [273-2751. For example, we discovered that, in general, multiple H-terminated Si atoms at the kink and step sites were eroded more rapidly than the monohydride-capped atoms ]2732751.

In this section, we first describe detailed etching processes of Si(l1 l), Si( 1 lo), and Si(100) in NH,,F solutions. Atomic images of GaAs and InP surfaces are also briefly discussed. (i) Si(ll1). Figure 73 shows current-potential curves of an n-Si(ll1) electrode in a nitrogensaturated 0.27 M NH,F (pH 5) taken in the dark (solid line) and under illumination (dashed line). The cathodic current observed in the negative going potential scan is attributed to the discharge of protons with subsequent hydrogen evolution at the Si electrode. The dark current reaches a maximum value of 12 l&/cm’ at -0.2 V and decreases slowly to 8 @/cm2 at 0.8 V. This low anodic current markedly increased with an increase in NH,F concentration; for example, the dark currents were 40 and 350 +4!cm2 at 0.8 V in 2 M and 11 M NH,F solutions, respectively. The pH of the NH,F solutions also has slight influence on the magnitude of the anodic current; increasing the pH of 0.27 M NH,F from 4 to 8 roughly doubled the dark current from 3 to 6 @/cm’. The above results are consistent with the previous work 12881. The dark current was attributed to the oxidation of a Si-H bond with the injection of an electron to the conduction band of the n-Si( 111) electrode, forming dangling bonds of Si- and H’ in solution [288]. The surface dangling bond of Si is expected to be attacked further by either H,O or F. The I-V curve, shown as the broken line in Fig. 73, for the n-Si( 111) electrode in 0.27 M NH,F was obtained under white light illumination. The cathodic current observed in the dark at potentials more negative than -0.5 V was not affected by the illumination. On the other hand, the photoanodic current emerges at -0.4 V, peaking at 0.1 V to 350 pA/cm2, and exponentially decaying to a constant current of ca. 100 @/cm’.

These current densities were found to be proportional to the light intensity. Holding

the electrode potential at 0.8 V for 5 min resulted in diminishing of the photocurrent, indicating the formation of the passivating SiO, layer. Although the passivating oxide layer should dissolve in F solutions, the oxidation rate of Si evidently exceeded its dissolution rate under the present condition. The CV results in Fig. 73 strongly suggest that illumination of the Si electrode should be avoided during both voltammetric and STM experiments for investigating the etching process on atomically ordered Si( 111) surfaces. The STM imaging of the n-Si( 111) electrode was performed immediately after the electrode was etched. After the Si electrode was immersed in the NH,F solution, the electrode potential of Si was immediately set to - 1.1 V. Note that the potential effect on the etching rate of Si was reported in NaOH

216

K. ltaya

400 h 1 \ , \ I \ , \

I!

-20

I

3001

‘_

-.----

/ -25

---=---_

,

.------

,

;r

.

7 / 0 9

E/Vvs.SCE Fig. 73. Cyclic voltammograms of n-Si( 111) in 0.27 M NH,F in the dark (solid line) and under illumination (dashed line). From [274]. solutions [277,278]. It was found that both chemical and EC etching mechanisms operate at the OCP, leading to a higher etching rate [277,278]. We also found that applying a cathodic potential of -1.1 V drastically reduced the corrosion rate of Si in NH,F solution. Figure 74 shows a crystallographic orientation (a) and a ball-and-stick model (b), respectively, of a H-terminated Si( 111) with [ 1121 and 11121 oriented steps. One of the most interesting features of a Si( 111) surface is the existence of two structurally different steps where the Si atoms have monohydride and dihydride configurations. They can be exemplified by the steps in the [ 1121 and [niz] directions, respectively (Fig. 74(b)). KY situ STM results have shown that the dihydride bound Si atoms are more reactive than those of the monohydride ones in weakly alkaline HF solutions, resulting in the appearance of the most stable [ 1121 steps [270]. Those experiments were conducted with Si( 111) samples tilted towards [ 1121 and [ 1121 [270].

Furthermore, there are two possible dihydride structures for the Si

atoms at the [ 1121 step. The dihydride axis is either perpendicular or parallel to the (111) plane. It is believed that there is a strong repulsive interaction among the horizontal dihydride structures, so that the perpendicular dihydride, as depicted in Fig. 74(b), is in fact more stable than the horizontal one, as confirmed by an IR spectroscopic study [265]. The horizontal dihydride 12701 might be too reactive to exist on a Si( 111) surface in the presence of etching species of H,O and F. We focused our attention on evaluation of the reactivity difference between the microscopically different steps on Si( 111) 12741. We were able to use in situ STM to locate some areas which contain both types of steps so that the reactivity of these steps can be simultaneously examined under identical conditions. This approach is evidently more advantageous than that of the previous study which used

STM

(a) tot3 view

217

in Electrolytes

(b) side view

Fig. 74. Top view (a) and side view (b) of the Si( 111) surface. From [274]. differently tilted Si( 111) substrates in separated experiments [270]. As the initial surface feature of Si controlled the subsequent etching process, we first recorded an STM image at - 1.1 V to show the initial surface morphology as shown in Fig. 75(a), followed by stepping the electrode potential to a less negative value of - 1.04 V to accelerate the erosion of Si. A series of STM images shown in Figs. 75(b) to 75(f) was acquired successively with a time interval of 13 s [274]. Figure 75(a) shows well-defined double layer steps of 0.32 nm in height and terraces extending more than 25 nm on the (111) surface. The Si( 111) was etched in 11 M NH,F for 3 min at room temperature. The relative heights of terraces are reflected by their brightness in the STM image. The internal atomic structure of terrace (marked T) was readily discerned by a high-resolution STM scan. A well-ordered hexagonal pattern with an interatomic spacing of 0.38 nm was good agreement with the ideal Si( 11 l):H1 x 1 structure [269,272-2741. Consequently, the treatment in 40 % NH,F indeed yielded a long range ordered monohydride-terminated Si( 111) surface with no discemiable vacancy defect in the hexagonal network. The step orientations, as defined by their out notrmals, am shown in Fig. 75(a). It is important to note that both the mono- and dihydride steps were probed by the in situ STM imaging at the same time. The shape of the terraces shown in Fig. 75(a) was termined by the morphology of the step ledges, i.e., the monohydride steps are mostly straight, in strong contrast to the typical zig-zag pattern for the dihydride ones. The small islands (3 nm in diameter, probably impuarities) at the upper edge of the STM image were used as a guide against themal drift during the STM measurement. Their unchanged locations indicate

218

K. ltaya

Cd)

0

w

0

25

50

25

w

50

0

25

50

nm 75. Successively recorded STM images for the etching process of Si( 111) in NH,F. The images were acquired at the time interval of 12.8 s. From [274].

Fig.

STM

in Electrolytes

219

low themal drift. The time-dependent STM results to be presented below demonstrate the important role played by the atomic structure at the steps in controlling the etching rate. Figures 75(b) to 75(f) present the time-dependent etching process of Si( 111) after acquiring the image of Fig. 75(a). During the first 13 s etching of the Si from Fig. 75(a) to 75(b), the width of the upper portion of the terrace T marked by D decreased from 16 to 8 nm, while the lower portion marked by D’ retractedfrom 18to 12.5nm. The relatively fastererosionof the upper half of the [ I 121 dihydride step is thoughtto be due to the higherkink densitywithin the zig-zag pattern stepledge. On the other hand, the monohydridestepin the direction of [l 121seemsto be unchangedfrom Fig. 75(a) to Fig,75(bj. Particularly, the stepswithin the circle markedC in Fig. 75(b) remainedstill in both images,indicating that the ideal monohydride[ 1121and [?I 1] step in the absenceof kink sitesis inactive in the present etchingcondition. The dihydride terminatedstepscontinuedto retract rapidly to dissolvethe terraceT, leaving a smallisolatedisland I in Fig. 75(c). However, it was surprisingto find that the [ 1121step with the dihydride configuration markedby D’ in Fig. 75(c) wasessentiallyunchangedfrom Fig. 75(bj to 75(c). The well-defined step ledge, markedD’ in Fig. 75(c), suggeststhat an ideal dihydride step without kinks is alsostable. After the completeremovalof the islandI in Fig. 75(c), a small,bilayer deep(0.32 nm) pit evolved in Fig. 75(d). This newly formed depressionis apparentlya real pit, not an imaging artifact, becauseit expanded and coalescedwith an adjacent[112] step, as shown in Fig. 75(e). This coalescence introduced many kink sites (marked K) in Fig. 75(e) at the almost inactive [112] step. A close examinationof Fig. 75(e) and 75(f) revealsa very important fact that the etching rate of the ( 1121 monohydridestepis now increasedby the introductionof the kinks into the nearly ideal monohydride step. The [ 1121stepline retractedby ca. 3 nm within 13 s, which correspondsto an etching rate of 14 ntnlmin. This is a rather significant increasefrom a negligiblelevel for the idealmonohydridestepas observed in Figs. 75(a) and 75(b). The suddenincreasewill be discussedfurther in the following section. (a) Etching mechanism. The mechanismof chemicaletching of Si hasbeenproposedto involve the nucleophilicreactionsbetweenthe hydrogenterminatedSi speciesandthe weak nucleophilesof H,O (or possibly the F- ions in the presentstudy) [288]. Preferentialetching of Si at the defect sites such as kinks and stepshasbeententatively attributedto their greaterreactivity and lesssterichindrance. The reactivity differencecan be primarily explainedby the different bondingenergiesfor the mono-, di-, and tri-hydrides structures[277-2791. The steric considerationcan also explain the reactivity difference betweenSi atomson the terrace (monohydride) and at the steps(mono-, di-, or tri-hydrides). This finding is further supportedby the presentSTM study addressingthe local corrosionof Si with different hydride configurations. Furthermore, the in situ STM is alsocapableof revealingthe effect of defect structureson the local corrosionrate. By comparison,this information is difficult to obtain with ex situ methods. The local Si etching processcan be further elucidatedby the relevant structural modelsdepictedin Fig. 76. Both type A andB surfacefeaturesconsistof two intersectingmonohydridesteps.The A type surface is composed of a dihydride kink (D), two monohydride terminated step ledges, and

220

K. ltaya

monohydride terrace. Once the dihydride Si atom at the kink (D) is removed by H,O or F , the two nearest neighboring monohydride Si atoms (M, and M2) are converted to the dihydride structure. The newly formed dihydride Si atoms are further attacked, resulting in a progressive etching along the step ledges. On the other hand, there is no dihydride Si at the intersection of step ledges in B, rendering the etching slow. Type A surface feature was encountered in Fig. 75(e) and 75(f), which was responsible for the sudden increase of etching rate of a monohydride step in the [ 1121 direction. On the other hand, feature B, represented by circle C in Fig. 75(b), was rather stable because of the lack of kink sites. Based on the STM results shown in Fig. 75, it is possible to draw a fii conclusion that A yields a significant etching rate, whereas B is resistive toward the etching reaction. This difference results from the presence of a kink site at the intersection of two step ledges for A. As described above, only the Si atom at the corner of the (111) facet has a dihydride configuration, so that it is expected to remove first to expose dihydride Si atoms. For the Type C surface feature in Fig. 76 where monohydride and dihydride steps meet, the model which features a higher reactivity at a kink still holds. Despite the fact that there is no kink in the model initially, it evolves readily within the dihydride step. The activity of a dihydride step varies and is dominated by its local environment. Specifically, the portion of a dihydride step near the intersection with a monohydride step is more reactive than that far away from the intersection. Formation of a monohydride structure is likely to be the driving force for the formation of a zig-zag pattern, consisting of many segments of monohydride step, outlined by the broken line in C. Note that a tip such as the position designated as K in the zig-zag pattern again represents a highly reactive dihydride kink which

A

B

Fig. 76. Three models of typical surface structures revealed by in situ STM. From [274].

STM

in Electrolytes

221

quickly dissolves the monohydride step. On the other hand, away from the intersection, the dihydride step is relatively inert so that a well-defined step ledge is possible to form. In fact, all of these surface microstructures

have been located by the in situ STM. For example, from Figs. 75(a) to 75(b) the

dihydride stepin the [n2] direction retractedrapidly to evolve a zig-zag patternin the areanear arrow D in Fig. 75(b). This zig-zag patterncan be explainedby the type C structurein Fig. 76. The resultdescribedabovehasrevealeddetailsof the etchingprocessof Si( 111)in dilute NH$ under cathodic potential control. Dissolutionof Si preferentially occurred at the multiple hydrogen bonded kinks of Si. The dihydride stepsseemedto havea higher density of kinks than the monohydridesteps, resultingultimately in different etchingratesof thesesteps. Not only grossscaleSTM imagesbut also atomic resolution STM imageshave revealed details of these etching processes. Erosion of Si proceededmostly in a layer-by-layer fashion, although nucleationof pits started before the complete removalof the uppermostSi layer. (ii) Si(ll0). The aforementionedresults provide compelling evidence that the difference in the chemical reactivity of the monohydrogen-terminatedand dihydrogen-capped Si surface atoms profoundly influencesthe Si-etchingprocess. We extendedour investigationsto the EC etchingof Si( 110)in aqueousNH,F solutions[275]. The Si( 110)substrateis uniquebecausean ideal Si(110)surfaceis expectedto consistof atomiczig-zag rows of monohydride-terminated Si, in the so-calledcoupled monohydride arrangement[266]. Previous IR work on NH,F-etched Si(ll0) suggestedthat the surface consistsof adjacent rows of coupled monohydridesteps[266]. In order to avoid extensive H, bubbleformation, the n-Si( 110)singlecrystalwasimmersedin a 4 M NH,F solution at -1.35 V vs. SCE. A typical STM imageacquiredshortly after immersionat this negativepotentialisshownin Fig. 77(a). It is clear that, prior to extensiveetching,then-Si( 110) surface wasatomically rough, characterizedby a “rolling-hill” structure;the averagecorrugationheight was in the rangeof l-3 nm. The image shown in Fig. 77(a) presumablyreflects the interface that existed betweenSi and SiO,, prior to removal of the latter by the HF pretreatment. It may be recalledthat a similarrolling-hillstructure wasfound for a freshly HF-etchedn-Si( 111)substrate[272]. Extended etching resulted in the developmentof step lines along the cllO> direction and the formationof atomically flat terrace-stepstructures.An exampleis shown in Fig. 77(b), obtainedthirty minutesafter acquisitionof the imagein Fig. 77(a). The imageclearly defines20 nm wide atomically flat terraceswith monatomicstepsof ca. 0.2 nm in height; the terracewidths increasedas the etching duration was lengthened. It is important to note that the shapeof islandsand pits are rectangular, elongatedin the direction. Theseresultsprovide evidencethat EC etching of n-Si( 110) proceeds via a highly anisotropicbut layer-by-layer mechanism. Figure 78(a) shows a high-resolutionSTM imageof an etched n-Si(110) surface in an areawhere both islandsandpits were present. The notablefeaturesin this imageare as follows: (i) zig-zag atomic rows(chains)are perfectly alignedalong the ~110~ direction; (ii) the islandsare constitutedby one or two zig-zag chainsof identical interfacial structure;(iii) the pits are formed by missingrows (not more thantwo) alongthe direction; (iv) the step-edgesaresharplydefined; and(v) only few kink sites

K. ltaya

222

50 nm

Fig. 77. Two typical STM images of Si( 110) in 4 M NH,F.

0

2

4

6

Fig. 78. High-resolution

0

10

0

From [275].

2

nm

STM images of Si( 110) in 4 M NH,F.

nm

From [275].

4

STM

223

in Electrolytes

were found along the step-edges. A top-view atomic-resolution image of the etched surface is shown in Fig. 78(b).

It is easy to

recognize in this figure that the zigzag chains are perfectly aligned along the direction; the distance between every other atom on the chain was measured to be 0.38 nm. In the ~001~ direction, perpendicular to < 1lo>, the distance between the chains was determined to be 0.54 nm. It is noteworthy that, based upon IR experiments, atomically straight and long-range ordered steps had been postulated to exist on Si( 110) surfaces etched in aqueous NH,F solution at the OCP [266]; however, ordering in the direction was not predicted. The present results support the former but not the latter hypothesis since long-range order in both the and directions was observed here. The discrepancy between the two studies arises from the difference in the etching potential; the present work was at a potential ca. 0.5 V negative of the OCP, whereas the earlier investigation was at the OCP. This is easily proven by the fact that, when the potential was scanned from -1.35 V to the OCP, the density of islands and pits was drastically increased which resulted in the deterioration of interfacial order in the direction. When the potential was cycled back to -1.35 V, the well-defined atomically smooth Si( 110) surface was regenerated. An H-terminated Si( 110) surface with a monatomic step along the direction is illustrated by top-view and ball-and-stick models shown in Fig. 79(a) and 79(b). The lattice parameters of the unit cell (Fig. 78(a)) and the height of the monatomic step (Fig. 79(b)) in the models are perfectly consistent with the experimentally derived values (unit-mesh parameters: 0.384 nm and 0.543 nm; step- height: 1.92A). It is thus plausible to concludetbat theEC etching of n-Si( 1 lo), at potentials quite negative of the OCP, yields a well-defined

H-terminated Si(1 lO):H-(1 x 1) structure.

In addition, it may be

(W

[1101 t

I 0:

0: Si (2nd layer) Si (1st layer) a = 0.543 nm, b = 0.384 nm

Fig. ‘79. Top view (a) and ball-and-stick

P011 0: Si

l :H

model (b) of Si( 1lo)-( 1 x 1). From [275].

224

K. ltaya

I 0

1

I

10

20

t 0

I 10

1 nm

1'

Fig. 80. SuccessivelyrecordedSTM imagesof Si( 110)collected at I6 s intervals. From 12751.

possibleto convey the messagethat electrodepotential is an important experimentalparameterin the formation of atomically well-defined n-Si(110)surfaces. STM imagesover a 20 nm x 20 nm areawere acquiredconsecutivelyat pre-selectedtime intervals. Four suchimages,obtainedat l&second intervals at - 1.35V, areshownin Fig. 80(a)to 80(d). Narrow,regularly shapedislandscan be seenin Fig. 80(a) along with wide, irregularly shapedpits; etching-inducedchangescan be monitoredvia alterationsin the shapeand depthsof theseislandsand pits. After the first 16-secondinterval, Fig. 80(a) to 80(b), the islandmarkedA, is almost completely dissolved,whereasthelongerisland AZ becameshorter.After extendeddissolution,thelength(but not the width) of the islandscontinuedto decreasefurther; in Fig. 80(d), almostall of the islandsare no longer observable.On closerinspection,it canbenotedthat the etching of the islandsoccurredat both endsand proceededalong the ~1 lO> direction. The implication is that the dissolutiontranspiresexclusively alongthe direction;etching alongthe perpendicular401> direction does not take placein the absence of kink sites. As discussedbelow, islanddissolutiondevelopsfrom both extremitiesdueto the presenceof the morereactive dihydride Si atomsatboth ends.

STM in Electrolytes

225

With regard to pit-corrosion, the serialized images in Fig. 80 reveal the lateral growth of monatomic pits also along the direction. In addition, the initially irregular shapes became rectangular with step-lines sharply defined along the direction. The evolution of the large, irregularly shaped pit in the lower-right comer of the image in Fig. 80(a), and the merger of the smaller pits B, and B, in Fig. 80(b) into the longer pit C in Fig. 80(c) serve as examples. The gradual alteration of the pit shapes to yield rectangular domains elongated along the direction is indicative of preferential etching along this particular direction. Etching of the pits, formed on the atomically flat terrace at - 1.35 V, was also examined under atomic resolution. Three images, in an area replete with rectangular pits, are shown in Figure 81; the images were obtained at 16-second intervals. The depth of the pits was uniformly found to be 2 A, a value that corresponds to the monatomic step height. It is not easy to determine the atomic features inside the pits because of the narrowness of the pit-widths; indications were that the pits were constituted by a few missing atomic rows. Exclusively, enlargement of the pits occurs along the direction. For example, the pits marked A and B were elongated along the 4 lO> direction and eventually, as can be seen in Fig. 8 l(c), merged to form one larger pit without a concomitant change in the width. This behavior indicates that ~1 lO> directed step-edges within the pits are stable, a circumstance that results in a highly anisotropic pitetching process. The etching rate along the direction is, of course, equal to the rate of disappearance of the Si atoms from the top and bottom extremities of each pit. It was found that the etching rates at both ends of the pit were essentially identical. The etching rate of a pit was smaller than that found for etching of an island, indicating that the islands disappeared more quickly than the pits, which can be seen in Fig. 80. Schematic models are shown in Figs. 82(a) and 82(b), which depict the etching processes of islands and pits, respectively. Since it has already been demonstrated that Si-dihydride sites are more reactive than Si-monohydride, it must be that the Si atoms at both ends of the island are of the dihydride variety;

I 0

I

1 nm I40

nm 140

nm

Fig. 81. Successively recorded STM images of pits formed on Si( 110) with 16 s intervals From [275].

226

K. ltaya

l

:H

@: Si (dihydride) Fig. 82. Ball-and-stick

models of the islands (a) and the pits (b). From [275].

all other surface atoms should be monohydride-capped. In this model, the dihydride Si atoms am first removed; this subsequently leads to the formation of a dihydride site at the next (nearest-neighbor) Si atom (Fig. 82(a)). The latter, in turn, is preferentially etched via nucleophilic attack by H,O and F . The propagation of such sequence of reactions, a process often referred to as a “zipper reaction,” accounts for the complete removal of the islands from the surface. A similar mechanism can be applied to rationalize the unidirectional growth of the pit, Fig. 82(b). The pit-growth apparently results from the etching of the dihydride Si atoms that exist at the ends of an upper atomic row. The difference between the etching rates of the island and pit is not explainable by steric hindrance because, regardless of whether on the island or in the pit, the geometric arrangement of Si atoms nearest a Si-dihydride site is similar. A difference exists, however, in the environment of next-nearest neighbors of Si-dihydride on the second layer; such atoms are gray-shaded in Figs. 82(a) and 82(b). In the case of an island, these atoms are a monohydride-capped Si; in the pit, they are terminated, not by H, but by Si, as in the bulk crystal. It is not unreasonable to expect that such difference can lead to an induction effect on the dihydride Si atom to render it more susceptible to the nucleophilic attack by H20 and F-. Nevertheless, the results obtained on Si(ll0) clearly demonstrate that etching proceeds via preferential attack at the Si-dihydride sites along the ~1 lO> direction [275]. (iii) Si(OO1). From the practical point of view related to the developmentof the LSI technology, it is moredesirableto have an atomic level understandingof the chemicaletchingprocessof Si(OO1)rather than Si( 111)in buffered-HF solutions. Here we describean in situ STM investigationof the chemical etching of Si(OO1)in 0.1 M NH,F (pH 5) [273]. At the initial stageof etching, the Si(OO1)surface showeda nearly ideal (1 x 1) structurefollowed by the appearanceof missingatomic rows along the [ 1lo] direction. Subsequently,progressiveappearanceof ( 1111 microfacetswas clearly seenon the

STM

in Electrolytes

227

Si(OO1) substrate. Because the Si(OO1) surface is roughened by a prolonged etching in a concentrated NH,F solution, we have tried to reveal the initial stage of the etching process using an unetched Si(OO1) surface under cathodic potential control in 0.1 M NH,F. The n-Si(OO1) electrode was mounted onto the STM cell after thorough degreasing with acetone and ethanol, but without the pre-treatment in 40 % NH,F. The potential of the semiconductor was quickly brought to -0.8 V, ca 400 mV more negative than its OCP. Stable STM imaging was readily obtained under this condition. Figure 83(a) shows a typical low resolution STM scan in an area of 125 x 125 nm, acquired at the beginning of the experiment. The step heights in Fig. 83(a) vary between 1.2 and 2.7 nm, although some monatomic steps (step height 0.3 nm) were also found on the terraces. Despite the appearance of the relatively flat terraces in the STM image, a ripple-like surface morphology was always found with higher resolution scans. Ridges along the [ 1 lo] direction are rather straight. A close-up 3D view (Fig. 83(b)) reveals parallel atomic rows along the [ 1lo] direction. The periodicity of 0.39 nm within each row equals the value expected for a Si(OOl)-( 1 x 1) structure, while the spacing between two neighboring rows are multiples of 0.39 nm. The lack of two dimensional ordering, as expected for an ideal (1 x 1) structure, is attributed to missing atomic rows along the [ 1lo] direction and the formation of some (1 x n) structures, where n > 2. These missing atomic rows appear randomly on the relatively flat terrace. It is important to note that these missing rows along the [ 1lo] direction mark the initial stage of [ 111) microfacets formation. It is expected that removal of one atomic row from the uppermost layer of an ideal Si(OO1) surface results in two monohydride terminated Si rows in the second layer. After holding the potential at -0.8 V for 2 hrs, STM imaging revealed the development of a staircase morphology with the height decreasing from the left to the right in the image shown in Fig. 84(a). All stairs are only 4 nm in width but extend over more than 40 nm in length. Consequently, this staircase-like feature is attributed to the formation of ( 1 I 1 ] microfacets. The thermodynamically stable monohydride configuration ultimately drives this facetting process. The height of each step is ca. 1.2 nm. The inset in Fig. 84(a) shows an angle of 54.7” between the (I 11) and (001) planes. The angle observed in the image shown in Fig. 84(a) seems to be consistent with the expected value. A similar angle was reported in the previous paper [289]. Apparently, step bunching occurs on the extensively etched Si(OO1) surface, producing the appearance of stairs with step heights greater than the height of a monolayer. It is remarkable that the STM of the smaller scan area shown in Fig. 84(b) clearly reveals atomic features on the tilted ( 111) microfacets. Hexagonal arrays with an interatomic spacing of 0.39 nm are clearly resolved on rather narrow atomically flat planes. The steps separating these ( 1I1 } microfacets are most certainly double layers. Many different types of kink sites can be clearly seen along the steplines. These poorly although monohydride appearing on the narrow the ( 11 I } microfacets. dissolution of Si atoms

defined step edges were obtained owing to multiple hydrogen bonded Si atoms, Si atoms existed along the [llO] direction as shown in Fig. 84(b). Pits terraces indicate that the etching also occurs by removal of monohydride Si on The formation of pits on Si( 111) was previously explained by a random on the (111) terraces [277,278] and the formation of ( 111) microfacets were

observed as the initial stage of corrosion. Surprisingly, it was possible to acquire an STM atomic image

K. ltaya

228

VlOl

(W

v

(a)

00 0

[ii01

50

100 nm

0

ill01 pi01 L

5.0

nm

10.0

Fig. 83. Initial stages of Si(OO1) in NH,F

nm 50 25 0

Fig. 84. Topographic (a) and high-resolution (h) STM images of the formation of (I 11) facets on Si(OO1). From 12731.

STM

I

in Electrolytes

229

I

I 4.0 nm

Fig. 85. Atomic image of an ideal Si(OOl)-(1 x 1) structure obtained at local area in NH,F. From [273]. of the Si( 11 l):H-( 1 x 1) structure on the( 111) plane, which had a tilt angle of 54.7” with respect to the (001) plane. STM has also revealed a transient state of Si(OOl)-( 1 x 1) structure before the (11 I) facets emerged. (iv) GaAs. It is well-known

that GaAs exhibits various surface-reconstructed

structures such as (4 x

2), (2 x 4) and c(4 x 4) on the (001) surface [290-2931. Surface topographic images reported previously for semiconductors, such as n-TiO, [294], n-ZnO [295], and GaAs [296-3001 have been obtained in aqueous solutions usually without atomic resolution, mainly because of the difficulty in preparing well-defined surfaces. It is, however, becoming more urgent for the semiconductor technology to understand wet chemical etching processes, particularly those of Si and GaAs, with atomic resolution. We acquired the first atomic STM images of GaAs surfaces in a 0.05 M H,SO, solution 128 1,282]. The results clearly demonstrate that the ideal GaAs(OOl)-( 1 x 1) and (11 l)-( 1 x 1) structures exist in a pure H,SO, solution in a cathodic potential region. The samples were n-type Si-doped GaAs (001 ), (11 l)A and (11 l)B wafers, grown by the horizontal Bridgman method. The GaAs(OO1) and (1ll)B samples were treated in 1 M HCl for 10 min at room temperature. The GaAs( 11 l)A sample was etched with a etching solution (1 H,SO,:8H,O,: lH,O by volume). The etching rate of (11 l)A surface has been reported to be the lowest among all low-index planes in this mixed solution [301]. After the etching, the solution was completely replaced with 0.05 M H,SO,. The replacement of the etching solution was carried out repeatedly to exclude HCl in the solution. It is important that the GaAs surface should always be kept submerged in the solution in order to protect it from oxidation and contamination in the ambient atmosphere. Figure 86 shows CVs for the freshly etched GaAs(OOl), (11 l)A, and (11 l)B electrodes in 0.05 M H,SO, in the dark. Proton discharge with hydrogen evolution was observed at the cathodic end of each CV. The hydrogen evolution reaction was previously investigated on GaAs electrodes [302-3041. It

230

K. ltaya

Fig. 86. Cyclic voltammograms of n-GaAs(OOl), n-GaAs( 11 l)A, and n-GaAs(1 1l)B in 0.05 M H,SO, in the dark. Scan rate; 50 mV/s. From [282]. can be seen that the current-potential profiles for the hydrogen evolution reaction are slightly different from each other, suggesting that the surfaces might be terminated by different kinds of atom? depending on their GaAs strongly depends on the crystallographic orientation. On the other hand, it is more significant to note in Fig. 86 that the anodic current for the oxidation of GaAs strongly depends on the crystallographic orientation. The oxidation of GaAs( 11 l)A surface occurs only at potentials more positive than 1.1 V. The oxidation current commences at substantially less positive potentials on the other surfaces. Figure 87(a) shows a typical surface topography of a chemically etched GaAs(OO1) surface acquired in an area of 50 x 50 nm. It can be clearly seen that the surface of the (001) exhibits a well-defined stepterrace structure extending over a large area. Wider terraces are seen to extend over 30 nm. The relative brightness of the terraces in the STM image reflects their heights on the surface, i.e.. the surface ascends from left to right. The rather uniform appearance of the terraces strongly suggests that the (001) surface has a structure well-defined on an atomic scale. It is also clear that the steps intersect each other to fom: an angle of !%I”, as expected for a surface with fourfold symmetry. These steps appearing as straight lines were confirmed as double-layer steps on the (001) surface based on the observed height of 0.28 nm obtained by a cross section analysis. This unique height of 0.28 nm for the steps indicates straightforwardly that the (001) surface prepared by etching in HCl must be either Ga- or As-terminated

STM

in Electrolytes

231

(b)

I 0

I

I

I nm

I

1 50

Fig. 87. Topographic (a) and high-resolution (b) STM images of n-GaAs(OO1) in H,SO,. From [282].

0

nm

I 0

I

I

I

I

nm

Fig. 88. STM images of n-GaAs( 11 l)A in 0.05 M H,SO,. From [282].

1 5.0

232

K. ltaya

(b)

(4

1” *-* ” :,- ..f . *

Jpol gqg

~~~ (001)

040nm

040nm (cl

(d)

w-l L Ill11

[lTOI

(11l)B 0.40 nm

H 040nm

Fig. 89. Top views of ball models for ideal Ga-terminated GaAs(OO1) (a), As-terminated GaAs(OO1) (b), GaAs(l 1l)A (c), and GaAs( 111)B (d). From 12821. (see Fig. 89). According to the crystallographic orientation of the GaAs(OO1) electrode, these steps were found to be parallel to either [ 1 lo] or [ilO] directions. Figure 87(b) shows our first atomic STM image of an atomically flat terrace on a GaAs(OO1) surface. It is clear that the ideal square arrangement expected for the (001) surface with fourfold symmetry is discerned by in situ STM. The observednearestinteratomicdistancesin the ]l lo] and [ilO] directions were found to be 0.4 f 0.02 nm. The atomicimageshown in Fig. 87(b) clearly demonstratesthat the ideal, non-reconstructedGaAs(OOl)-(1 x 1) structureis exposedin H,SO, solution under the cathodic polarization. Note that the ideal(1 x 1) structureseemedto be extendedover the entire region of the terrace,becausepits or even singleatomic defectswere rarely observed. Recent in situ AFM studiesof GaAs(OO1)in solutions containing HCl reported a similar (1 x 1) structure [296,297], which is consistentwith our STM result describedabove. The (111)A surface etched in the mixed solution containingH,O, was also found to have an atomicallyflat terrace-stepstructure in H,SO, solution as shownin Fig. 88(a). All stepsobservedwere double-layerstepswith a height of 0.33 nm. The steps in local areawerestraightand parallelto the close-packedatomic row directionof (111) surface. Figure 88(b) showsa typical atomic STM image,revealing an interatomicdistanceof 0.4 nm with an almost perfect hexagonalarrangement. This result clearly demonstratesthat the ideal GaAs(l1 l)A-( 1 x 1) structure, as shown in Fig. 89(c), is exposedin H,SO, solution. It is reasonablyexpectedthat the uppermostlayer on the (11l)A surfaceconsistsof Ga atoms.

STM

in Electrolytes

233

Finally, it is noteworthy that the GaAs( 11l)B surface prepared by etching in 1 M HCl also showed an atomically flat terrace-step structure in H,SO, with a step height identical to that observed on the (11 l)A surface. Although the average terrace width was typically 5-10 nm, obviously smaller than that found on the A surface, an atomic STM image revealed a hexagonal arrangement of As atoms with the interatomic distance of 0.4 nm. The above results indicate that the GaAs( 11 l)B surface has also the ideal (1 x 1) structure shown in Fig. 89(d) [282]. In summary, it was demonstrated that the well-defined GaAs(OOl), (11 l)A, and (11 l)B surfaces can be prepared by chemical etching in solutions. Atomically flat terrace-step structures were consistently observed by in sifu STM on all three surfaces in H,SO, solution under potential control. Furthermore, we successfully obtained the first atomic STM images which showed that the ideal GaAs(OOl)-( 1 x I), GaAs( 11 l)A-( 1 x l), and GaAs( 11 l)B-( 1 x 1) structures are exposed and persist in H,SO, solution. We tentatively assume that the GaAs(OOl)-( 1 x 1) and GaAs( 11 l)A-( 1 x 1) surfaces are terminated by hydrogen at least under cathodic polarization. (v) InP. In spite of the fact that InP is a very important

material

for both optoelectronic and electronic

device applications, there have been a fewer STM studies on InP than on GaAs. Previous STM studies of InP were mostly carried out in UHV. Atomic images were acquired mainly on cleaved InP(1 IO) surfaces using STM [305] or AFM [306] in UHV. More recently, STM images of a reconstructed InP(OOl)-(2 x 4) surface of epitaxial layers have also been obtained in UHV at an atomic level [307,308]. In situ STM of InP electrode has previously been performed to evaluate the surface roughness of InP(OO1) prepared by different surface treatments [309]. However, no atomic resolution has been achieved in the previous STM observations because of the relatively high surface roughness. We have recently shown the first atomically resolved STM images of InP(OOl), (11 l)A and (11 I)B surfaces in an H,SO, solution under cathodic potential control, which effectively prevented the surfaces from oxidation [286]. These images clearly demonstrate that well-defined InP surfaces can be prepared by chemical etching in an HCl solution. Figure 90 shows high-resolution topographic images of the InP(11 l)A surface. In Fig. 90(a), individual atoms are relatively clearly observed on the atomically flat terraces with a corrugation height of ca. 0.02 nm. Monolayer steps in Fig. 90(a) are also found to be exactly parallel to the close-packed atomic row direction of (111) surface. Figure 90(b) shows an atomically resolved STM image of the ( 11l)A surface, revealing a perfect hexagonal arrangement of In atoms with an interatomic distance of 0.42 nm, as expected for the ideal InP( 11 l)A surface. These images clearly demonstrate that the ideal InP( 11 l)A-( 1 x 1) structure is exposed in the H,SO, solution after the chemical etching. This (1 x 1) structure seemed to extend over the entire region of the terrace; even atomic defects were rarely observed. On the other hand, the chemically etched (11 l)B surface also possessed a well-defined structure in the H,SO, solution. An atomically flat terrace-step structure was observed with a step height identical to that on the A surface, although the average terrace width was typically 5 -10 nm, which was smaller than that on the A surface. The appearance of wider terraces on the (11 l)A surface is reasonable, because the etching rate of (11 l)A [310].

surface in the HCl solution is the slowest of all low-index

planes

K. ltaya

234

0

2

6

4

8

10 v 0

3

4

nm

nm Fig. 90. High-resolution

STM images obtained

on InP( 11 l)A in H,SO,.

From [ 2861.

(a)

I 0

I

I

I

I

I

I

Fig.

91.

High-resolution

I

70

nm STM

5

images of InP(OO1) in H,SO,.

From 12861.

STM

in Electrolytes

235

confirmed by a cross-section analysis to be monolayer steps with a height of 0.29 nm. It can be deduced from this unique height of the steps that the (001) surface etched with HCl solution is terminated by either indium or phosphorus atoms. Figure 91(b) shows a high-resolution STM image acquired on one of the atomically flat terraces shown in Fig. 9 1(a). An almost ideal square arrangement expected for the (001) surface is clearly discerned by in situ STM. The observedinteratomicdistancesin the [ 1lo] and [ 1lo] to 0.42 nm directionswere found to be equalto 0.42 nm. The atomicimagesin Fig. 91(b) clearly demonstratethat the ideal lnP(OOl)-( 1 x 1) structureexists in H,SO, solutionunder cathodicpotentialcontrol. We alsopresumethe InP(OOl)-( 1 x l), (11l)A-( 1 x 1) and (1 I l)B-( 1 x 1) surfacesto be terminatedby hydrogenat leastundercathodicpolarization. In summary,it hasbeendemonstratedthat the well-definedInP(OO1), (111)A and (11l)B surfaces can be preparedby chemicaletching in an HCl solution [286]. In situ STM revealedatomically flat terrace-stepstructureson eachsurfacein H,SO, solutionundera properpotentialcontrol. Furthermore, we successfullyobtainedthe first atomic resolutionSTM imagesof InP surfaces,which showedthat the ideal InP(OOl)-( I x l), (11l)A-( 1 x 1) and (11 l)B-( 1 x 1) structuresare exposedand persistin H,SO, solutionunder EC conditions.

5. Concluding

Remarks

The methodsfor exposingwell-defined electrodesurfacesin solution were reviewed. The flameannealingand quenchingmethodcan be appliedto Au, Pt, Rh, Pd, Ir, and possibly Ag singlecrystal electrodes.The UHV-EC methodcan be usedto Pt, Au, Pd and Rh. For somemetalssuch asNi and Cu, surface oxidation takesplacein the EC chamberbefore immersionof the electrodeinto electrolyte solutions. It was demonstratedthat the EC etching method produced atomically flat terrace-step structuresunder carefully adjustedEC conditions. The methodof anodic dissolutionis expected to becomean important in situ techniquefor exposing well-defined surfacesof various metals and semiconductors. The structuresof specifically adsorbediodine on Pt( 111)andAu( 11l), and briefly on Ag( 111) were discussed,demonstratingthat complementaryuse of in situ STM and ex situ LEED is a powerful combinationto characterizethe atomicstructureof adsorbediodine. The adsorptionof sulfate/bisulfate adsorbedon Au( 111).Pt( 11l), andRh( 111)wasdescribed,with emphasison the fact that the same(43 x 47) structure is formed on thesethree substrates. Our model indicatesthat hydrogen-bondedwater chainsare insertedalong the d3 direction betweenneighboringrows of the adsorbedsulfates. The detailedstructuralanalysisof the CN adlayeron Pt( 111) was carried out by in situ STM, revealingthat six CN groups form a hexagonalring without an additional CN at the center of the ring. The complexationof K’ with the CN adlayerwasalsodiscussed. The UPD of Cu on Au( 111) in H,SO, was discussedin depth. In situ SXS concludedthat the Cu atomsform a honeycomblattice and are adsorbedon threefold hollow siteswith sulfateions locatedat the honeycombcenters. According to the modelstructure,the corrugationobservedby in situ STM and

236

K. ltaya

AFM should be ascribed to the coadsorbed sulfate ions, and not the Cu atoms. The IJPD of Cu on Pt( 111) was also described. The UPD of Ag on bare Au( 111). I-Au( 11 l), bare Pt( 1 I 1). and l-P!{ 1 I I i was extensively discussed. Our result for the UPD of Hg on Au( 111) was compared with that obtained by in situ AFM and SXS. It was shown that, in general, the iodine-modified electrodesare suitable for producing highlyorderedadlayersof variousorganicmolecules.TMPyP formshighly-orderedadlayerson I-Au( 1I l), IAg( 11l), and I-Pt( lOO),however, disorderedstructureswere formed on I-Pt(1 11) and I-Rh( 111). The dynamic processof the formation of orderedTMPyP adlayerswasinvestigatedon I-Au( 111). The adlayerstructureof benzeneon Rh( 111) and Pt( 111) was also describedin detail; it was found to be dependenton the electmdepotential. The (3 x 3) structure found on Rh( 111) in the cathodic potential range is almost identical to that found in UHV for the coadsorbedbenzeneand CO. The molecularshapesof naphthaleneandanthracenecould beclearly discernedby in situ STM. It was also shownthat benzene,naphthalene,andanthraceneform highly-orderedadlayerson Cu( 111). The anodicdissolutionof bare Ni, S-Ni, bare Ag, I-Ag, I-Pd and Cu was describedbasedupon in situ STM observations. The layer-by-layer dissolution occurs on these metals, resulting in the

formationof atomically flat terrace-stepstructures. The anisotropicetching observedon S-Ni( 1OO),IAg(lOO), I-Pd(lOO), and Cu(100) can be explainedin terms of the adlayer structuresnear the step edges. The EC etching processesof Si( 11l), Si( 1lo), and Si( 100) were discussedin relation to the atomic structuresof the step-edges.It was shown that the chemicaletchingproduceswell-definedGaAs and InP single-crystalelectrodes.The atomicstructuresof theseelectrodesin solutioncould be clearly seen. This review clearly demonstratesthat STM allows us not only to determineinterfacial structuresbut alsoto follow EC reactions. It is certainthat in situ STM will continueto be the premiertechniquein the study of the relationshipbetweenthe reactivity andthe structureof electrodesurfaces. Acknowledgments It is my great pleasureto thank my colleagues,T. Yamada,T. Sakuhara,K. Iwasa, S.-L. Yau, Y. Nagatani, S. Tanaka,T. Uchida, G. Muramatsu,P. Muller, L.-J. Wan, J.-D Zhang, M. Kunitake, S. Ando, K. Ogaki, T. Sawaguch,T. Sato, T. Suzuki, Y.-G. Kim, U. Akiba, T. Teshima,H. Yao, A. Mizusawa, K. Kaji, N. Batina, T. Hayashi, J.-H. Ye, T. Natsui, S. Sumiji, K. Hashimoto,M. Akiba, M. Iino, C. Kobayashi, T. Miura, and K. Sawamefor their contributionsto ITAYA-Electrochemiscopy Project under Exploratory Researchfor Advanced Technology Program(ERATO) organizedby Japan Scienceand Technology Corporation (JST) during the period of 1992-1997. The study of iodinemodified Pd electrodeswas performed in collaboration with Prof. M. P. Soriaga (Texas A & M University). I alsogreatfully acknowledgeassistance of Dr. Y. Okinaka, who was the researchadvisor of the project. The author thanksK. Inukai and K. Sashikatafor their help in writing this manuscript. This work was supportedpartially by a Grant-in Aid for Science Researchon Priority Area of “Electrochemistryof OrderedInterfaces” from the Ministry of Education, Science,Sports and Culture, Japan.

STM

in Electrolytes

237

References 1. P. N. Ross and F. T. Wagner, in Advances in Electrochemistry and Electrrochemical Engineering XIII, H. Gerischer and C. W. Tobias (Eds.), Wiley, New York (1984) p. 69. 2. A. T. Hubbard, Chem. Rev. 88, 633 (1988). 3. D. M. Kolb, Z. Phys. Chem. Neue Folge 154, 179 (1987). 4. M. P. Soriaga, Prog. Surf: Sci. 39, 325 (1992). 5. H.-J. Guntherodt and R. Wiesendanger (Eds.), in Scanning Tunneling Microscopy I, SpringerVerlag, Berlin (199 1). 6. P. A. Christensen, Chem. Sot. Rev. 21, 197 (1992). 7. H. Siegenthaler, in Scanning Tunneling Microscopy II, R. Wiesendanger and H.-J. Guntherodt (Eds.), Springer-Verlag, Berlin (1992) p. ‘7. 8. P. N. Ross, in Structure of Electrified Interfaces, J. Lipkowski and P. N. Ross (Eds.), VCH, New York (1993) p. 35. 9. K. Itaya, in New Trends and Approaches in Electrochemical Technology, N. Masuko, T. Osaka and Y. Fukunaka (Eds.), Kodansha, Tokyo, VCH, New York, (1993) p. 181. 10. A. A. Gewirth and H. Siegenthaler (Eds.), in Nanoscale Probes of the Solidniquid

Interface

(NATO AS1 Series-288), Kluwer Academic Publishers, London (1995). 11. A. A. Gewirth and B. K. Niece, Chem. Rev. 97, 1129 (1997). 12. G. Jerkiewicz, in Solid-Liquid Electrochemical Interjaces (ACS Symposium Series 656), G. Jerkiewicz, M. P. Soriaga, K. Uosaki, and A. Wieckowski (Eds.), American Chemical Society, Washington, DC (1997) p. 1. 13. J. Clavilier, R. Faure, G. Guinet, and R. Durand, J. Electroanal. Chem. 107, 205 (1980). 14. J. Clavilier, J. Electroanal. Chem. 107, 211 (1980). 15. M. P. Soriaga (Ed.), in ElectrochemicalSurface Science, Molecularr Phenomenaat Electrode Surjaces(ACS SymposiumSeries378), American ChemicalSociety, Washington,DC (1988). 16. H. D. Abruna (Ed.), in ElectrochemicalIntegaces, Modem Techniquesfor In-Situ Interjixe Characterization,VCH, New York (1991). 17. R. Sonnenfeldand P. K. Hansma,Science232, 211 (1986). 18. K. Itaya and E. Tomita, Surf. Sci. Lett. 201, L507 (1988). 19. P. Lustenberger,H. Rohrer, R. Christoph, and H. Siegenthaler,J. Electrroanal.Chem. 243, 225 (1988). 20. J. Wiechers,T. Twomey, D. M. Kolb, and R. J. Behm, J. Electroanal. Chem.248,45 1 (1988). 21. 0. Lev, F. R. Fan, and A. J. Bard, J. Electrochem. Sot. 135, 783 (1988). 22. A. Hamelin,in Modem Aspectsof Electrochemistry , No. 16, B. E. Conway, R. E. White, andJ. O’M. Bockris (Eds.), PlenumPress,New York (1985) p. 1. 23. S. Motoo and N. Furuya, J. Electroanal. Chem.167, 309 (1984). 24. K. Sashikata,N. Furuya, and K. Itaya, J. Vat. Sci. Technol. B9, 457 (1991). 25. N. Fumya and S. Koide, Su$ Sci. 220, 18 (1989).

238

K. ltaya

26. J. Clavilier, A. Rodes. K. El. Achi. and M. A. Zamakhchari. J. Chim. Pkys PQ. 1291 (1991 i. 27. J. II. Schott and H. S. White, J. Pkys. Chem. 98, 291 (1994). 28. J. M. Feliu, J. M. Orts, R. Gomez, A. Aldaz, and J. Clavilier, J. Electroanul. Chem. 372. 265 (1994). 29. K. Itaya, S. Sugawara,K. Sashikata,and N. Furuya, J. Vuc. Sci. Technol. A8, 5 I5 (1990).

30. 3 1. 32. 33. 34. 35. 36. 37.

S. Tanaka, S.-L. Yau, and K. Itaya, J. Electroanal. Chem.396, 125 (1995). S.-L. Yau, Y.-G. Kim, and K. Itaya, J. Am. Chem.Sot. 118, 7795 (1996). X. Gao and M. J. Weaver, Ber. Bunsenges.Phys. Chem.97, 507 (1993). D. M. Kolb, Prog. Surf: Sci. 51, 109 (1996). K. Wang, G. S. Chottiner, and D. A. Scherson,J. Phys. Chem.97, 10108(1993). J. L. Stickney, C. B. Ehlers, and B. W. Gregory, Langmuir 4, 1368(1988). M. Hourani and A. Wieckowski, J. Electroanal. Chem.227, 259 (1987). P. Zelenay and A. Wieckowski, J. Electrochem. Sot. 139,2552 (1992).

38. C. M. Vitus, S.-C. Chang, B. C. Schardt, and M. J. Weaver, J. Phys. Chem.95, 7559 (1991). 39. B. C. Schardt, S.-L. Yau, and F. RinaIdi, Science243, 1050 (1989). 40. S.-L. Yau, C. M. Vitus, and B. C. Schardt, J. Am. Chem. Sot. 112, 3677 (1990). 4 1. H. Baltruschat, U. Bringemeier, andR. Vogel, Faraday Discuss.94, 317 (1992). 42. R. Vogel, I. Kamphausen,and H. Baltruschat, Ber. Bunsenges.Phys. Chem.96,525 (1992). 43. N. Shinotsuka,K. Sashikata,and K. Itaya, Surf: Sci. 335, 75 (1995). 44. X. Gao and M. J. Weaver, J. Am. Chem.Sot. 114, 8544 (1992). 45. S. Sugita. T. Abe, and K. Itaya, J. Phys. Chem.97, 8780 (1993). 46. T. Yamada. N. Batina, and K. Itaya, J. Phys. Chem.99, 8817 (1995). F. Lu, G. N. Salaita, H. Baltruschat, and A. T. Hubbard, J. Electroanal. Chem.222, 305 (1987). 48. J. M. Orts, R. Gomez, J. M. Feliu, A. Aldaz, and J. Clavilier, Langmuir 13, 3016 (1997). 49. C. A. Lucas, N. M. Markovic, and P. N. Ross,Surf Sci. Lett. 340, L949 (1995). 47.

50. C. Stuhlmann,I. Villegas, and M. J. Weaver, Chem.Phys. Lett. 219, 319 (1994). 5 1. Y.-G. Kim, S.-L. Yau, and K. Itaya, J. Am. Chem. Sot. 118, 393 (1996). 52. 0. M. Magnussen,J. Hagebock,J. Hotlos, and R. J. Behm, Faraday Discuss.94,329

(1992).

53. 54. 55. 56. 57. 58.

G. J. Edens,X. Gao, and M. J. Weaver, J. Electroanal. Chem.375, 357 (1994). J. Inukai, S. Sugita, and K. Itaya, J. Electroanal. Chem.403, 159 (1996). A. M. Funtikov, U. Linke, IJ. Stimming, and R. Vogel, Surf: Sci. Lett. 324, L343 (1995). A. M. Funtikov, U. Stimming, and R. Vogel, J. Electroanal. Chem.428, 147(1997). L.-J. Wan, S.-L. Yau, and K. Itaya, J. Phys. Chem.99, 9507 (1995). M. Hourani, M. Wasberg,C. Rhee,and A. Wieckowski, Croatica Chem.Acta 63, 373 (1990).

59. 60. 61. 62.

K. Sashikata,Y. Matsui, K. Itaya, and M. P. Soriaga,J. Phys. Chem.100, 20027 (1996). L.-J. Wan, S.-L. Yau, G. M. Swain, and K. Itaya, J. Electroanal. Chem.381, 105(1995). B. M. Ocko, G. M. Watson, and J. Wang, J. Phys. Chem.98, 897 (1994). N. Batina, T. Yamada, and K. Itaya, Langmuir 11, 4568 (1995).

STM

in Electrolytes

239

63. T. Yamada, N. Batina, K. Ogaki, S. Okubo, and K. Itaya, in Electrode Processes VI, Proceedings

64. 65. 66. 67.

of the Sixth International Symposium, A. Wieckowski and K. Itaya (Eds.), The Electrochemical Society, Pennington, NJ (1996) p. 43. T. Yamada, K. Ogaki, S. Okubo, and K. Itaya, Sut$ Sci. 369, 321 (1996). J. Inukai, Y. Osawa, M. Wakisaka, K. Sashikata, Y.-G. Kim, and K. Itaya, J. Phys. Chem. 102, 3498 (1998). M. L. Lynch and R. M. Corn, J. Electroanal. Chem.318, 379 (1991). R. L. McCarley and A. J. Bard, J. Phys. Chem.95, 9618 (1991).

68. 69. 70. 71.

W. Haiss,J. K. Sass,X. Gao, and M. J. Weaver, Surf. Sci. Lett. 274, L593 (1992). N. J. Tao and S. M. Lindsay, J. Phys. Chem.96, 5213 (1992). S. A. Co&an and H. H. Farrell, Sur$ Sci. 95, 359 (1980). B. M. Ocko, 0. M. Magnussen,J. X. Wang, R. R. Adzic, andTh. Wandlowski, Physica B 221, 238 (1996). 72. Z. Shi, J. Lipkowski, M. Gamboa,P. Zelenay, and A. Wieckowski, J. Elecroanal.Chem. 366, 317 (1994). 73. M. Wasberg,M. Hourani, and A. Wieckowski, J. Electroanal. Chem.278,425 (1990). 74. P. Zelenay, G. Horanyi, C. K. Rhee, and A. Wieckowski, J. Electroanal.Chem. 300, 499 (1991). 75. P. Zelenay and A. Wieckowski, J. Electrochem.Sot. 139, 2552 (1992). 76. J. Clavilier, M. Wasberg,M. Petit, and L. H. Klein, J. Electroanal. Chem.374, 123 (1994). 77. C. K. Rhee, M. Wasberg, G. Horanyi, and A. Wieckowski, J. Electroanal. Chem. 291, 281 (1990). 78. P. A. Thiel and T. E. Madey, Surf. Sci. Reports7, 211 (1987). 79. P. N. Ross, in ElectrochemicalSurface Science, Molecular-rPhenomenaat Electrode Su$aces (ACS SymposiumSeries378), M. P. Soriaga(Ed.), American ChemicalSociety, Washington, DC (1988) p. 37. 80. K. Ataka and M. Osawa,Langmuir 14,951 (1998). 81. P. Mrozek, M. Han, Y.-E. Sung, and A. Wieckowski, Suti Sci. 319, 21 (1994). 82. Y.-E. Sung, S. Thomas,and A. Wieckowski, J. Phys. Chem.99, 13513(1995). 83. S. Thomas,Y.-E. Sung, H. S. Kim, and A. Wieckowski, J. Phys. Chem.100, 11726(1996). 84. I. Oda, Y. Shingaya, H. Matsumoto, and M. Ito, J. Electroanal. Chem.409, 95 (1996). 85. J. L. Stickney, S. D. Rosasco,G. N. Salaita, and A. T. Hubbard,Langmuir 1, 66 (1985). 86. D. G. Frank, J. Y. Katekaru, S. D. Rosasco,G. N. Salaita, B. C. Schardt, M. P. Soriaga, D. A. Stem, J. L. Stickney, and A. T. Hubbard,Langmuir 1, 587 (1985). 87. S. D. Rosasco,J. L. Stickney, G. N. Salaita, D. G. Frank, J. Y. Katekaru, B. C. Schardt, M. P. Soriaga, D. A. Stern, and A. T. Hubbard,J. Electroanal. Chem.188, 95 (1985). 88. F. Kitamura, M. Takahashi,and M. Ito, Chem.Phys. Lett. 130, 181 (1986). 89. H. Kawashima,Y. Ikezawa, andT. Takamura,J. Electroanal. Chem.317,257 (1991). 90. C. S. Kim and C. Korzeniewski, J. Phys. Chem.97, 9784 (1993).

240

K. ltaya

91. J. Inukai, Y. Morioka, Y.-G. Kim, S.-L. Yau, and K. Itaya, Bull. Chem. Sot. Jpn. 70, 1787 (1997). 92. K. Ashley and S. Pons, Chem.Rev. 88, 673 (1988).

93. P. Guyot-Sionnest and A. Tadjeddine,Chem.Phys. Len. 172, 341 (1990). 94. See“News and Features”,Anal. Chem.68, 236A (1996). 95. G. N. Salaita, D. A. Stern, F. Lu, H. Baltruschat,B. C. Schardt, J. L. Stickney, M. P. Soriaga, D. G. Frank, and A. T. Hubbard, Langmuir 2, 828 (1986). 96. R. L. McCarley, Y.-T. Kim, and A. J. Bard, J. Phys. Chem.97, 2 11 (1993). 97. S.-L. Yau, Y.-G. Kim, and K. Itaya, J. Korean Sot. Anal. Sci. 8, 723 (1995). 98. L.-J. Wan, S.-L. Yau, and K. Itaya, J. Solid State Electrochem. 1, 45 (1997). 99. G. N. Salaita, F. Lu, L. Laguren-Davidson,and A. T. Hubbard, J. Electroamd.Chem. 229, 1 (1987). 100. D. D. Sneddonand A. A. Gewirth, SUI$ Sci. 343, 185 (1995). 101. D. W. Suggsand A. J. Bard, J. Am. Chem.Sot. 116,10725 (1994). 102. D. W. Suggsand A. J. Bard, J. Phys. Chem.99, 8349 (1995). 103. 0. M. Magnussen,B. M. Ocko, J. X. Wang, and R. R. Adzic, J. Phys. Chem. 100, 5500 (1996). 104. T. Wandlowski, J. X. Wang, 0. M. Magnussen,and B. M. Ocko, J. Phys. Chem. 100, 10277 (1996). 105. B. M. Ocko, J. X. Wang, and T. Wandlowski, Phys. Rev. Letts. 79, 1511 (1997). 106. D. M. Kolb, in Advances in Electrochemistryand ElectrochemicalEngineering, Vol. 11, H. Gerischerand C. W. Tobias (Eds.), Wiley, New York (1978) p. 125. 107. B. E. Conway, Prog. SW-&Sci. 16, 1 (1984). 108. G. Kokkinidis, J. Electroanal. Chem.201, 217 (1986). 109. W. J. Lorenz, H. D. Hermann,N. Wuthrich, and F. Hilbert, J. Electrochem.Sot. 121, 1167 (1974). 110. R. R. Adzic, in Advunces in Electrochemistryand ElectrochemicalEngineering,Vol. 13, H. Gerischerand C. W. Tobias (Eds.), Wiley, New York (1985) p. 159. 111. 0. M. Magnussen,J. Hotlos, R. J. Nichols, D. M. Kolb, and R. J. Behm, Phys. Rev. Len. 64, 2929 (1990). 112. 0. M. Magnussen,J. Hotlos, G. Beitel, D. M. Kolb, and R. J. Behm, J. Vat. Sci. Technol. B9, 969 (1991). 113. T. Hachiya, H. Honbo, and K. Itaya, J. Electroanal. Chem.315, 275 (1991). 114. N. Batina, T. Will, and D. M. Kolb, Faraday Discuss.94, 93 (1992). 115. S. Manne, P. K. Hansma,J. Massie, V. B. Elings, and A. A. Gewirth, Science251, 183 (1991). 116. Z. Shi and J. Lipkowski, J. Electroanal. Chem.364,289 (1994). 117. Z. Shi and J. Lipkowski, J. Electroanal. Chem.365, 303 (1994). 118. M. F. Toney, J. N. Howard, J. Richer, G. L. Borges, J. G. Gordon, 0. R. Melroy, D. Yee, and L. B. Sorensen,Phys. Rev. Zett. 75, 4472 (1995).

STM

in Electrolytes

119. K. Sashikata, N. Furuya, and K. Itaya, J. Electroanal. Chem. 316,361 120. J. H. White and H. D. Abruna, J. Phys. Chem.94, 894 (1990).

241

(1991).

121. J. H. White and H. D. Abruna, J. Electroanal. Chem.300, 521 (1991). 122. R. Michaelis, M. S. Zei, R. S. Zhai, and D. M. Kolb, J. Electroanal. Chem.339, 299 (1992). 123. T. Abe, G. M. Swain, K. Sashikata,and K. Itaya, J. Electroanal. Chem.382, 73 (1995). 124. H. Matsumoto, J. Inukai, and M. Ito, J. Electroanal. Chem.379, 223 (1994). 125. Y. Shingaya, H. Matsumoto, H. Ogasawara,and M. Ito, Sur. Sci. 335, 23 (1995). 126. I. Oda, Y. Shingaya, H. Matsumoto, and M. Ito, J. Electroanal. Chem.409,95 (1996). 127. C. A. Lucas, N. M. Markovic, and P. N. Ross,Phys. Rev. B, 56, 3651 (1997). 128. N. M. Markovic, H. A. Gasteiger,C. A. Lucas, I. M. Tidswell, and P. N. Ross,Surf. Sci. 335, 91 (1995). 129. N. M. Markovic, C. A. Lucas, H. A. Gasteiger,and P. N. Ross,Su$ Sci. 372, 239 (1997). 130. J. L. Stickney, S. D. Rosasco,and A. T. Hubbard, J. Electrochem. Sot. 131, 260 (1984). I3 1. T. Hachiya and K. Itaya, Ultramicroscopy 42-44, 445(1992). 132. K. Ogaki and K. Itaya, ElectrochimcaActa 40, 1249(1995). 133. K. Ogaki and K. Itaya, unpublishedresult, (1996). 134. C.-H Chen, S. M. Vesecky, and A. A. Gewirth, J. Am. Chem.Sot. 114,451

(1992).

135. P. Mrozek, Y.-E Sung, M. Han, M. Gamboa-Aldeco,A. Wieckowski, C.-H. Chen, and A. A. Gewirth, ElectrochimcaActa 40, 17(1995). 136. P. Mrozek, Y.-E Sung, and A. Wieckowski, &I$ Sci. 335,44 (1995). 137. S. G. Garcia, D. Salinas,C. Mayer, J. R. Vilche, H.-J. Pauling, S. Vinzelberg, G. Staikov, and W. J. Lorenz, Surf? Sci. 316, 143 (1994). 138. J. L. Stickney, S. D. Rosasco,D. Song, M. P. Soriaga, and A. T. Hubbard, Surf: Sci. 130, 326 (1983).

139. F. El Omar, R. Durand, and R. Faure, J. Electroanal. Chem.160,385

(1984).

140. B. P. Costa,J. Canullo, D. Vasquez Moll, R. C. Salvarezza,M. C. Giordano and A. J. Arvia, J. Electroanal. Chem.244,261 (1988). 141. T. Chierchie, C. Mayer, K. Juttner, and W. J. Lorenz, J. Electroanal. Chem.191, 401 (1985). 142. N. Kimizuka and K. Itaya, Faraday Discuss.94, 117(1992). 143. P. Zelenay, M. Gamboa-Aldeco,G. Horanyi, and A. Wieckowski, J. Electroanal.Chem. 357, 307 (1993).

144. A. T. Hubbard, J. L. Stickney, S. D. Rosasco,M. P. Soriaga, and D. Song, J. Electroanal. Chem. 150, 165 (1983). 145. A. Wieckowski, B. C. Schardt, D. Rosasco,J. L. Stickney, and A. T. Hubbard, Suti Sci. 146, 115 (1984). 146. D. G. Frank, 0. M. R. Chyan, T. Golden, and A. T. Hubbard, J. Phys. Chem. 98, 1895 (1994). 147. C.-H. Chen and A. A. Gewirth, Phys. Rev. Lett. 68, 1571 ( 1992). 148. J. Li and H. D. Abruna, J. Phys. Chem.B, 101, 244 (1997).

K.

242

ltaya

149. B. W. Gregory and J. L. Stickney, J. Electroanal. Chem. 300, 543 (1991). 150. D. W. Suggs and J. L. Stickney, Surj: Sci. 290, 375 (1993). 15 1. T. E. Lister, B. M. Huang, R. D. Herrick II, and J. L. Stickney, J. Vat. Sci. Tech&. B13, 1268 (1995). 152. L. B. Goetting, B. M. Huang, T. E. Lister, and J. L. Stickney, Electrochim. Acta 40, 143 ( 1995). 153. T. E. Lister and J. L. Stickney, J. Phys. Chem. 100, 19568 (1996). 154. U. Demir and C. Shannon, Langmuir 10, 2794 (1994). 155. C.-H. Chen and A. A. Gewirth, J. Am. Chem. Sot. 114, 5439 (1992). 156. C.-H. Chen, K. D. Kepler, A. A. Gewirth, B. M. Ocko, and J. Wang, J. Phys. Chem. 97, 7290 (1993). 157. J. Clavilier, J.-P. Ganon, and M. Petit, J. Electroanal. Chem. 265, 23 1 (1989). 158. R. R. Adzic, J. X. Wang, 0. M. Magnussen, and B. M. Ocko, J. Phys. Chem. 100, 14721 (1996). 159. T. Abe, L.-J. Wan, 160. N. S. Marinkovic, (1995). 161. J. X. Wang, R. R. 162. W. Polewska, J.

and K. Itaya, Proc. 61st Denkikagaku Symp.lSendai, Japan, (1994) p.293. W. R. Fawcett, J. X. Wang, and R. R. Adzic, J. Phys. Chem. 99, 17490 Adzic, 0. M. Magnussen, and B. M. Ocko, Surf: Sci. 344, 111 (1995). X. Wang, B. M. Ocko, and R. R. Adzic, J. Electroanal. Chem. 376, 41

(1994). 163. M. P. Green, K. J. Hanson, R. Carr, and I. Lindau, J. Electrochem. Sot. 137, 3493 (1990). 164. N. J. Tao, J. Pan, Y. Li, P. I. Oden, J. A. DeRose, and S. M. Lindsay, Surf: Sci. Lett. 27 1, L338 (1992). 165. G. Staikov and W. J. Lorenz, in Electrochemical Nanotechnology, W. J. Lorenz and W. Plieth (Eds.), Wiley-VCH,

Weinheim (1998) p. 13.

166. W. Obretenov, U. Schmidt, W. J. Lorenz, G. Staikov, E. Budevski, D. Carnal, U. Muller, H. Siegenthaler, and E. Schmidt, Faraday Discuss. 94, 107 (1992). 167. U. Muller, D. Carnal, H. Siegenthaler, E. Schmidt, W. J. Lorenz, W. Obretenov, U. Schmidt, G. Staikov, and E. Budevski, Phys. Rev. B, 46, 12899 (1992). 168. W. Obretenov. U. Schmidt, W. J. Lorenz, G. Staikov, E. Budevski, D. Carnal, U. Muller, H. Siegenthaler,and E. Schmidt, J. Elecrrochem. Sot. 140, 692 (1993). 169. D. Carnal. P. I. Oden, U. Muller, E. Schmidt, and H. Siegenthaler, Electrochim. Acta 40, 1223 (1995). 170. U. Schmidt, S. Vinzelberg, and G. Staikov, Sue Sci. 348, 261 (1996). 171. R. R. Adzic, J. Wang, C. M. Vitus, and B. M. Ocko, Sur$ Sci. Lett. 293, L876 (1993). 172. J. C. Bondos, A. A. Gewirth, and R. G. Nuzzo, J. Phys. Chem. 100, 8617 (1996). 173. J. L. Bubendorff, L. Cagnon, V. Costa-Kieling, J. P. Bucher, and P. Allongue, SUI$ Sci. Lett. 384, L836 (1997). 174. T. Sawaguchi, T. Yamada, Y.Okinaka, and K. Itaya, J. Phys. Chem. 99, 14149 (1995).

STM

in Electrolytes

243

175. K. Uosaki, S. Ye, Y. Oda, T. Haba, and K. Hamada, Langmuir 13, 594 (1997). 176. S. Chiang, Chem. Rev. 97, 1083 (1997). 177. J. P. Rabe, Ultramicroscopy 42-44, 41 (1992). 178. A. Ikai, SurfI Sci. Reports 26, 261 (1996). 179. G. E. Poirier, Chem. Rev. 97, 1117 ( 1997). 180. P. S. Weiss and D. M. Eigler, Phys. Rev. Lett. 71, 3139 (1993). 181. H. Ohtani, R. J. Wilson, S. Chiang, and C. M. Mate, Phys. Rev. Lett. 60, 2398 (1988). 182. V. M. Hallmark, S. Chiang, and Ch. Wall, J. Vat. Sci. Z’echnol.B 9, 1111(1991).

183. P. H. Lippel, R. J. Wilson, M. D. Miller, Ch. Woll, and S. Chiang, Phys. Rev. Lett. 62, 171 (1989). 184. J. Lipkowski and P. N. Ross(Eds.), In Adsorption of Moleculesat Metal Electrodes,VCH, New York (1992). 185. R. Srinivasan,J. C. Murphy, and R. Fainchtein, J. Electroanal. Chem.312, 293 (1991). 186. R. Srinivasanand P. Gopalan,J. Phys. Chem.97, 8770 (1993). 187. N. J. Tao, J. A. DeRose,and S. M. Lindsay, J. Phys. Chem.97, 910 (1993). 188. N. J. Tao and Z. Shi, J. Phys. Chem.98, 1464(1994). 189. N. J. Tao and Z. Shi, SUI$ Sci. Lett. 321, L149 (1994). 190. N. J. Tao, G. Cardenas,F. Cunha,and Z. Shi, Langmuir 11,4445 ( 1995). 191. F. Cunha and N. J. Tao, Phys. Rev. Lett. 75, 2376 (1995). 192. B. Roelfs, E. Bunge, C. Schroter, T. Solomun, H. Meyer, R. J. Nichols, and H. Baumgartel,J. Phys. Chem.B, 101, 754 (1997). 193. M. Kunitake, N. Batina, and K. Itaya, Langmuir 11, 2337 (1995). 194. M. Batina, M. Kunitake, and K. Itaya, J. Electroanal. Chem.405, 245 (1996). 195. M. Kunitake, U. Akiba, N. Batina, and K. Itaya, Langmuir 13, 1607(1997). 196. K. Ogaki, N. Batina, M. Kunitake, and K. Itaya, J. Phys. Chem.100, 7185 (1996). 197. K. Itaya, B. Batina, M. Kunitake, K. Ogaki, Y.-G. Kim, L.-J. Wan, and T. Yamada,in: SolidLiquid ElectrochemicalIntet$aces (ACS SymposiumSeries656), G. Jerkiewicz, M. P. Soriaga, K. Uosaki, and A. Wieckowski (Eds.), AmericanChemicalSociety, Washington,DC (1997) p. 171. 198. K. Sashikata,T. Sugata, M. Sugimasa,and K. Itaya, Langmuir 14, 2896 (1998). 199. L.-J. Wan, Doctoral Dissertation, Tohoku University (1996). 200. A. Wieckowski, S. D. Rosasco,B. C. Schardt, J. L. Stickney, and A. T. Hubbard, Inorg. Chem.23, 565 (1984). 201. R. Vogel and H. Baltruschat, Su$ Sci. Len. 259, L739 (1991).

202. J. Albers, H. Baltruschat, and I. Kamphausen,J. Electroanal. Chem.395, 99 (1995). 203. J. L, Stickney, S. D. Rosasco,B. C. Schardt, and A. T. Hubbard, J. Phys. Chem. 88, 251 (1984).

204. M. Kunitake, N. Batina, S. Miyano, and K. Itaya, in preparation(1998). 205. F. Stevens,D. J. Dyer, and D. M. Walba, Angew. Chem.Int. Ed. Engl. 35, 900 (1996).

244

K. ltaya

206. J. Clavilier, V. Svetlicic, and V. Zutic, J. Electroanal. Chem. 402, 129 (1996). 207. Ci. A. Somorjai, Introduction to Surjace Chemistry and Catalysis I, John Wiley & Sons Inc., New York (1994). 208. H. A. Yoon, M. Salmeron, G. A. Somorjai, Surf: Sci. 373, 300 (1997). 209. 210. 211. 212. 213.

C. K. Rhee, M. Wasberg, P. Zelenay, and A. Wieckowski, Catal. L&t. 10, 149 (1991). U. Schmiemann, Z. Jusys, and H. Baltruschat, Electrochimca Actu 39,561 (1994). H. Baltruschat and U. Schmiemann, Ber. Bunsenges. Phys. Chem. 97,452 (1993). F. T. Wagner and P. N. Ross, J. Electroanal. Chem. 250, 301 (1988). S.-L. Yau, Y.-G. Kim, and K. Itaya, in Electrode Processes VI, Proceedings of the Sixth International Symposium, A. Wieckowski and K. Itaya (Eds.), The ElectrochemicalSociety,

Pennington,NJ (1996) p. 43. 214. M. Neuber, F. Schneider,C. Zubragel, and M. Neumann,J. Phys. Chem.99, 9160 (1995). 215. C. M. Mate and G. A. Somorjai, Surf: Sci. 160, 542 (1985). 216. P. Sautet and M.-L. Bocquet, Phys. Rev. B, 53,491O (1996). 217. A.Wander, G. Held, R. Q. Hwang, G. S. Blackman, M. L. Xu, P. de Andres, M. A. Van Hove, and G. A. Somorfai, Suti Sci. 249, 21 (1991). 218. S.-L Yau, Y.-G. Kim, and K. Itaya, J. Phys. Chem. B, 101, 3547 (1997). 219. R. F. Lin, R. J. Koestner,M. A. Van Hove, and G. A. Somorjai, Sur$ Sci. 134, 161 (1983). 220. V. M. Hallmark, S. Chiang, J. K. Brown, and Ch. Woll, Phys. Rev. Lett. 66, 48 (1991). 221. S.-L. Yau and K. Itaya, Colloids St&aces A: Physicochem.Eng. Aspects 134,21 222. Y.-G. Kim, S.-L. Yau, M. Yamagishi,and K. Itaya, in preparation(1998). 223. L.-J. Wan and K. Itaya, Langmuir 13, 7173 (1997).

(1998).

224. B. J. Cruickshank, D. D. Sneddon,and A. A. Gewirth, SUI$ Sci. Lett. 281, L308 (1993). 225. M. Xi, M. X. Yang, S. K. Jo, B. E. Bent, and P. Stevens,J. Chem. Phys. 101, 9122 (1994). 226. M. R. Vogt, W. Polewska,0. M. Magnussen,and R. J. Behm, J. Electrochem.Sot. 144, L113 (1997). 227. J. McBreen, in Modem Aspectsof Electrochemistry, No. 21, R. E. White, J. O’M. Bock&

and B. E. Conway (Eds.), PlenumPress,New York (1990) p. 29. 228. S.-L. Yau, F.-R. F. Fan, T. P. Moffat, and A. J. Bard,. J. Phys. Chem.98, 5493 (1994). 229. V. Maurice, H. Talah, and P. Marcus, SUI$ Sci. 304, 98 (1994). 230. 231. 232. 233. 234.

B. MacDougall and M. Cohen,J. Electrochem.Sot. 123, 191 (1976). B. MacDougall and M. Cohen, J. Electrochem.Sot. 123, 1783 (1976). T. Suzuki, T. Yamada, and K. Itaya, J. Phys. Chem.100, 8954 (1996). L. Ruan, I. Stensgaard,F. Besenbacher,and E. Lagsgaard,Phys. Rev. Len. 71, 2963 (1993). F. Besenbacher,I. Stensgaard,L. Ruan, J. K. Norskov, and K. W. Jacobsen,Sue Sci. 272, 334 (1992).

235. E. Kopatzki and R. J. Behm, Suti Sci. 245, 255 (1991). 236. T. Yamada, J. Nakamura,I. Matsuo, I. Toyoshima, and K. Tanaka,Surf: Sci. 207, 323 (1989). 237. E. Kopatzki, S. Gunther, W. Nichtl-Pecher, and R. J. Behm, Surf: Sci. 284, 154 (1993).

STM

in Electrolytes

245

238. J. Oudar and P. Marcus, Appl. Surf: Sci. 3, 48 (1979). 239. P. Marcus and E. Protopopoff, J. Electrochem. Sot. 140, 1571 (1993). 240. S. Ando, T. Suzuki, and K. Itaya, J. Electroanal. Chem.412, 139 (1996). 241. N. Batina, J. W. McCargar, G. N. Salaita,F. Lu, L. Laguren-Davidson,C.-H. Lin, and A. T. Hubbard,Lungmuir 5, 123 (1989). 242. T. Mebrahtu, M. E. Bothwell, J. E. Harris, G. J. Cali, and M. P. Soriaga,J. Electroanal.Chem. 300, 487 (1991).

K. J. Trevor and C. E. D. Chidsey, J. Vuc. Sci. Tech. 139, 964 (1991). 244. W. K. Burton, N. Cabrera,and F. C. Frank, Philos. Trans. R. Sot. London Ser. A, 243, 299 (1951). 245. T. Teshima,K. Ogaki, and K. Itaya, J. Phys. Chem.B, 101, 2046, (1997). 243.

246. K. Itaya, in Interjkcial Electrochemistry,A. Wieckowski (Ed.), Marcel Dekker, New York, in press. 247. M. Kawasakiand H. Ishii, Langmuir 11, 832 (1995). 248. M. Hopfner, W. Obretenov, K. Juttner, W. J. Lorenz, G. Staikov, V. Bostanov, and E. Budevski, Surf: Sci. 248, 225 (1991). 249. E. Zanazzi, M. Maglietta, U. Bardi, andM. Torrini, in Proc. ICSS-4 and ECOSS-3, Cannes,D. A. Degrasand M. Costa(Eds.), (1980) p. 527. 250. U. Bardi and G. Rovida, Sur$ Sci. 128, 145 (1983). 251. J. A. Schimpf, J. R. McBride, and M. P. Soriaga,J. Phys. Chem.97, 105 18 (1993). 252. J. A. Schimpf, J. B. Abreu, and M. P. Soriaga,Langmuir 9, 3331 (1993). 253. M. Soriaga,K. Itaya, and J. K. Stickney, in ElectrochemicalNanotechnology,W. J. Lorenz and W. Plieth (Eds.), Wiley-VCH, Weinheim (1998) p. 267. 254. W. F. Temesghen,J. B. Abreu, R. J. Barriga, E. A. Lafferty, M. P. Soriaga, K. Sashikata,and K. Itaya, Surf. Sci. 385, 336 (1997). 255. K. Kawabe,Doctoral Dissertation,Tohoku University (1996). 256. D. J. Trevor, C. E. D. Chidsey, and D. N. Loiacono, Phys. Rev. Lett. 62, 929 (1989). 257. H. Honbo, S. Sugawara,and K. Itaya, Anal. Chem.62, 2424 (1990). 258. E. Bunge, R.J. Nichols, B. Roelfs, H. Meyer, and H. Baumgartel,Lungmuir 12, 3060 (1996). 259. E. Bunge, S. N. Port, B. Roelft, H. Meyer, H. Baumgartel,D. J. Schiffrin, and R. J. Nichols, Lungmuir 13, 85 (1997). 260. S. Ando, T. Suzuki, and K. Itaya, J. Electroanal. Chem.431, 277 (1997). 261. S. Ando and K. Itaya, in preparation(1998). 262. M. Liehr, M. Heyns, M. Hirose, and H. Parks (Eds.), in Ultra&an SemiconductorProcessing Technology and Su$ace Chemical Cleaning and Passivation, Materials Research Society SymposiumProceedings,Vol. 386, Pennsylvania(1995). 263. G. S. Higashi, Y. J. Chabal, G. W. Trucks, and K. Raghavachari,Appl. Phys. Z&t. 56, 656 (1990). 264.

P. Dumas, Y. J. Chabal, and P. Jakob, Suti Sci. 269/270, 867 (1992).

246

K. ltaya

265. P. Jakob and Y. J. Chabal,J. Chem.Phys. 95, 2897 (1991).

266. 267. 268. 269.

P. Jakob. Y. J. Chabal, K. Kuhnke, and S. B. Christman,.%a: Sci. 302, 49 (1994). S. Watanabe,N. Nakayama,and T. lto, Appl. Phys. Lett. 59, 1458( 1991). R. S. Becker, G. S. Higashi, Y. J. Chabal, and A. J. Becker, Phys. Rev. Left. 65, 1917 ( 1990). ti. S. Higashi, R. S. Becker, Y. J. Chabal, and A. J. Becker, Appl. Phys. Left. 58, 1656 (1991).

270. H. E. Hessel,A. Feltz, M. Reiter, U. Memmert, and R. J. Behm, Chem. Phys. Lett. 186, 275 (1991). U. Neuwald, H. E. Hessel,A. Feltz, U. Memmert, and R. J. Behm, Surf: Sci. Lett. 296, L8 (1993). 272. K. Itaya, R. Sugawara,Y. Morita, and H. Tokumoto, Appl. Phys. Lett. 60, 2534 (1992). 273. S.-L. Yau, K. Kaji, and K. Itaya, Appl. Phys. Lett. 66, 766 (1995).

271.

274. K. Kaji, S.-L. Yau, and K. Itaya, J. Appl. Phys. 78, 5727 ( 1995). 275. J.H. Ye, K. Kaji, and K. Itaya, J. Electrochem. Sot. 143, 4012 (1996). 276. P. Allongue, V. Kieling, andH. Gerischer,ElectorchimcaAcfu 40, 1353(1995). 277. P. Allongue, V. Costa-Kieling, and H. Gerischer,J. Electrochem.Sot. 140, 1009 (1993). 278. P. Allongue, V. Costa-Kieling, and H. Gerischer,J. Electorchem.Sot. 140, 1018(1993). 279. P. Allongue, in Advances in ElectrochemicalScienceand Engineering,Vol. 4, H. Gerischerand C. W. Tobias, (Eds.), Wiley-Interscience, New York (1995) p. 3. 280. S.-L. Yau, F.-R. F. Fan, and A. J. Bard, J. Electrochem. Sot. 139, 2825 (1992). 281. H. Yao, S.-L. Yau, and K. Itaya, Sutf Sci. 335, 166 (1995). 282. H. Yao, S.-L. Yau, and K. ltaya, Appl. Phys. Lett. 68, 1473(1996). 283. K. Uosaki, S. Ye, and N. Sekine, Bull. Chem.Sot. Jpn. 69, 275 (1996). 284. P. Carlsson,B. Holmstrom, H. Kita, and K. Uosaki, Surf: Sci. 237, 280 (1990). 285. K. Uosaki and M. Koimura, J. Electroanal. Chem.357, 301 (1993). 286. H. Yao and K. Itaya, J. Electrochem.Sot. in press(1998). 287. 0. J. Glembocki, R. E. Stahlbush,and M. Tomkiewicz, J. Electrochem.Sot. 132, 145 (1985). 288. H. Gerischerand M. Lubke, Bet-. Bunsenges.Phys. Chem.91, 394 (1987). 289. S.-L. Yau, M. Arendt, A. J. Bard, B. Evans, C. Tsai, J. Sarathy, and J. C. Campbell, J. Electrochem. Sot. 141, 402 (1994). 290. M. D. Pashley, K. W. Haberern.W. Friday, J. M. Woodall, and P. D. Kirchner, Phys. Rev. Lett. 60, 2176 (1988). 291. D. K. Biegelsen,R. D. Bringans, J. E. Northrup, and L.-E. Swartz, Phys. Rev. B, 41, 5701 (1990). 292.

T. Hashizume,Q. K. Xue, J. Zhou, A. Ichimiya, and T. Sakurai, Phys. Rev. Lett. 73, 2208 (1994).

293. K. W. Haberernand M. D. Pashley, Phys. Rev. B, 41, 3226 (1990). 294. K. Itaya and E. Tomita, Chem.Lett. 285 (1989).

295. K. Itaya, E. Tomita, Surf: Sci. Lett. 219, L515 (1989).

STM in Electrolytes

247

296. M. Koinuma and K. Uosaki, Su$ Sci. Lett. 311, L737 (1994). 297. Y. Isikawa, H. Ishii, H. Hasegawa, and T. Fukui, 1. Vuc. Sci. Technol. B12,2713 (1994). 298. R. Sonnenfeld, J. Schneir, B. Drake, P. K. Hansma, and D. E. Aspnes, Appl. Phys. Z&t. 50, 1742 (1987). 299. T. Thundat. L. A. Nagahara, and S. M. Lindsay, J. Vuc. Sci. Technof. A8, 539 (1990). 300. K. Uosaki and M. Koinuma, Faraday Discuss. 94, 361 (1992). 301. S. Iida and K. Ito, J. Electrochem. Sot. 118, 768 (1971). 302. K. D. N. Brummer, J. Electrochem. Sot. 114, 1274 ( 1967). 303. H. Gerischer, N. Muller, and 0. Haas. J. Electroanal. Chem. 119, 41 (1981). 304. J. O’M. Bockris and K. Uosaki, J. Electrochem. Sot. 124, 1348 (1977). 305. Ph. Ebert, G. Cox, U. Poppe, and K. Urban, Ultramicroscopy 42-44,871 (1992). 306. Y. Sugawara, M. Ohta, H. Ueyama, S. Morita, F. Osaka, S. Ohkouchi, M. Suzuki, and S. Mishima, J. Vuc. Sci. Technol. B14, 953 (1996). 307. B. X. Yang, Y. Ishikawa, T. Ozeki, and H. Hasegawa, Jpn. J. Appl. Phys. 35, 1267 (1996). 308. N. Esser, U. Resch-Esser, M. Pristovsek, and W. Richter, Phys. Rev. B, 53, R13257 (1996). 309. Y. Robach, M. Phaner, C. de Villeneuve, and L. Porte, Appl. Phys. Lett. 61,255 1 (1992). 3 10. S. das Neves and M.-A. De Paoli, J. Electrochem. Sot. 140,2599 (1993).