Layer-by-Layer Electrodeposition of Copper in the Presence of o

18846

J. Phys. Chem. B 2005, 109, 18846-18851

Layer-by-Layer Electrodeposition of Copper in the Presence of o-Phenanthroline, Caused by a New Type of Hidden NDR Oscillation with the Effective Electrode Surface Area as the Key Variable Shuji Nakanishi, Sho-ichiro Sakai, Kensaku Nishimura, and Yoshihiro Nakato* DiVision of Chemistry, Graduate School of Engineering Science, Osaka UniVersity, Toyonaka, Osaka 560-8531, Japan ReceiVed: March 16, 2005; In Final Form: May 20, 2005

Electrochemical deposition of copper (Cu) from aqueous acidic Cu2+ solutions with o-phenanthroline (ophen) shows both potential and current oscillations, together with a (partially hidden) N-shaped negative differential resistance (N-NDR), indicating that the oscillations are classified into hidden N-NDR (or HNNDR) oscillations. The color and the surface morphology of Cu deposits oscillate in synchronization with the potential and current oscillations. Microscopic inspection has shown that dense round Cu leaflets, which look gray, grow in the positive side of the potential oscillation or in the high-current state of the current oscillation, whereas thin Cu leaflets, which look black, grow in the opposite-side stages of the potential and current oscillations, thus finally resulting in a layered Cu deposit with the layer thickness of about 5 µm. The appearance of the NDR is explained to be due to adsorption of the reduced form of a [Cu(II)(o-phen)2]2+ complex, which suppresses the Cu electrodeposition. The increase in the effective electrode surface area by growth of thin Cu leaflets, on the other hand, causes a current increase that can hide the NDR. This NDRhiding mechanism is of a new type and the present oscillation is regarded as a new-type of HN-NDR oscillator.

Introduction Electrochemical reactions with nonlinear kinetics show a variety of oscillations, accompanied by self-organized spatiotemporal patterns at electrode surfaces, as summarized in a recent review.1 Of various oscillations, oscillatory electrodeposition is an interesting phenomenon from the point of view of formation of ordered micro- and nanostructures at solid surfaces, because ever-changing spatiotemporal patterns at the deposit surface caused by the oscillatory electrodeposition are recorded in the form of architecture of the deposits. In fact, we recently found that the oscillatory electrodeposition led to very unique structures, such as metal latticeworks of a micrometer size2 and layered structures with the uniform thickness of about 50 nm over a macroscopically wide area of 1 mm × 1 mm.3,4 In addition, the elucidation of the mechanisms enabled us to tune the resultant micro- or nanostructures by modulating the electrochemical oscillations through changes in external parameters such as the applied current density.4,5 Other workers also reported some examples on the ordered-structure formation by oscillatory electrodeposition.6-10 In these studies, however, only the phenomena were reported and the mechanism was not clarified. The lack of the mechanism makes systematic tuning of deposit structures difficult. It is thus of key importance to reveal the mechanism for use of the electrodeposition for microand nanostructuring. The oscillatory electrodeposition is also of much interest from the point of view of exploration of new-type oscillatory mechanisms to get a deep understanding of nonlinear kinetics. According to the literature,11 all reported electrochemical oscillations can be categorized into four classes, depending on the roles of the true electrode potential (or the Helmholtz-layer * Author to whom correspondence should be addressed. Tel: +81-66850-6235; fax: +81-6-6850-6236; e-mail [email protected].

potential, E). Electrochemical oscillations in which the E plays no essential role and is kept essentially constant are called “strictly potentiostatic”-type (or Class I). This type can be regarded as chemical oscillators that contain electrochemical reactions. Electrochemical oscillations in which the E is involved as the essential variable but still not the autocatalytic variable are called S-NDR type (or Class II), because the oscillations in this case arise from an S-shaped negative differential resistance (S-NDR) in the current density (j) versus E curve. When the E is the autocatalytic variable, the oscillations arise from an N-shaped NDR (N-NDR) and are called N-NDR type. Furthermore, of the N-NDR oscillations, the oscillations in which the N-NDR is hidden by a current increase from another process are called hidden N-NDR or HN-NDR type (or Class IV). It is known1,11 that the N-NDR type shows only current oscillations, whereas the HN-NDR type shows both current and potential oscillations. The HN-NDR oscillators are further divided into three or four subcategories,11,12 depending on how the NDR is hidden. The above classification is of great value but includes implicit assumptions that the electrode surface is geometrically uniform, apart from adsorption or passivation layers, and that its morphology does not change during electrochemical reactions. These assumptions do not hold for electrodes with etch pits at the surfaces13 and electrodeposition reactions producing dendrites.2,5,14-16 The effective area of the electrode surface rather often acts as the autocatalytic key variable for the oscillatory electrodeposition reactions.2,5,14-16 In the present paper, we report the mechanisms of current and potential oscillations for copper (Cu) electrodeposition from aqueous acidic Cu2+ solutions with o-phenanthroline. The appearance of the potential oscillation itself was reported before by Schlitte et al.,7 though they did not propose any mechanism. Our detailed experiments have revealed that this oscillation is

10.1021/jp0513871 CCC: $30.25 © 2005 American Chemical Society Published on Web 09/17/2005

Layer-by-Layer Electrodeposition of Copper

Figure 1. Schematic illustration of an electrochemical cell and arrangement of three electrodes, together with an ODM.

classified into an HN-NDR oscillation of a new type, with the effective electrode surface area (Seff) as the NDR-hiding variable.

J. Phys. Chem. B, Vol. 109, No. 40, 2005 18847

Figure 2. (a) Potential-controlled j vs U in 0.5 M CuSO4 + 0.5 M H2SO4 (pH ) 0.7) in the absence (dashed curve) and presence (solid curve) of 2.0 mM o-phen, obtained in the first negative potential scan. The scan rate is 10 mV/s. (b) j vs t at a constant U of -0.15 V, with an external resistance of 15 Ω being inserted. The numbers, 1-4, on the j vs t are used in the Discussion section.

Experimental Section A polycrystalline Au (99.99% in purity) disk of 6 mm in diameter was used as the working electrode (WE), together with a 10 × 10 mm2 Pt plate as the counter electrode (CE) and a Ag|AgCl|sat.KCl electrode as the reference electrode (RE). The arrangement of the three electrodes is schematically shown in Figure 1. All the surfaces of the Au-disk electrode, except one surface exposed to the electrolyte, were covered with epoxy resin for insulating, and the uncovered Au surface was directed toward an optical digital microscope (ODM) or a digital video camera (DVC) placed outside the electrochemical cell to inspect the Au surface during oscillations under in situ conditions. We confirmed that the oscillatory behavior was hardly influenced by the direction of the uncovered Au surface. The Au electrode was polished with diamond slurry and immersed in a hot solution of 1.5 M HNO3 + 1.5 M H2O2 for 10 min before use to remove surface contamination. The electrolyte was 0.5 M H2SO4 + 0.5 M CuSO4 containing 0∼2.0 mM o-phenanthroline (hereafter abbreviated as o-phen). The electrolyte was prepared using special grade chemicals and pure water, the latter of which was obtained by purification of deionized water with a Milli-Q water purification system. Current density (j) versus potential (U) and j versus time (t) were measured with a potentiostat and a potential programmer (Nikko-Keisoku, NPGS-301) and were recorded digitally at 100 Hz with a data-storing system (instruNET, GW Instruments). Inspection of electrodeposits was also made with a highresolution scanning electron microscope (SEM, Hitachi S-5000), apart from the above-mentioned DVC (Canon, DM-IXY DV M) and ODM (Keyence VH-5000). Results Figure 2a shows U-controlled j versus U in 0.5 M CuSO4 + 0.5 M H2SO4 (pH ) 0.7) with and without 1.5 mM o-phen, obtained in the first negative U scan. In the absence of o-phen (dashed curve), the Cu-deposition current starts to flow at about +0.05 V and increases rather monotonically with the negative U shift. On the other hand, in the presence of o-phen (solid curve), the current considerably decreases compared with the case of no o-phen, and an NDR appears at about -0.12 V. Furthermore, when the U is kept constant at the potential region of the NDR, say, -0.15 V, with an external resistance of 15 Ω, a current oscillation is observed, as shown in Figure 2b. No stable current oscillation appears without the external resistance

Figure 3. Potential-controlled j vs U in 0.5 M CuSO4 + 0.5 M Na2SO4 + 1.5 mM o-phen of pH ) 2.0 (solid curve) and pH ) 3.0 (dashed curve), obtained in the first negative scan. The scan rate is 10 mV/s.

in the present system, probably because the ohmic drop, which plays an important role in the N-NDR oscillation, is not sufficiently large.11 The oscillation period is affected by the magnitude of the resistance. Figure 3 shows U-controlled j versus U in 0.5 M CuSO4 + 0.5 M Na2SO4 + 1.5 mM o-phen, in which the pH value is adjusted to 2.0 and 3.0 by addition of small amounts of H2SO4. Comparison of two curves in Figure 3 and the solid curve in Figure 2a indicates that the j increases with increasing the pH in a potential region more positive than the NDR. The negative shift of the potential region of the NDR with increasing pH is mainly due to the increase in the ohmic drop in the electrolyte between the working and the reference electrodes by the increase in the j with the pH, as explained in detail in the Discussion section. Figure 4a shows j-controlled j versus U in 0.5 M CuSO4 + 0.5 M H2SO4 (pH ) 0.7). In the presence of 2.0 mM o-phen, a sudden negative U shift occurs at j ) about 10 mA cm-2. The region of the potential shift corresponds well to the potential region of the NDR in Figure 2a. When the j is kept constant at a value of -10 to -15 mA cm-2, the U oscillates spontaneously (without an external resistance in this case), as shown in Figure 4b. The observation of both the current and potential oscillations is known11,17 to be characteristic of the hidden N-NDR (HNNDR) oscillation, as mentioned earlier, and thus the results of Figures 2 and 4 show that the present oscillations can be classified into the HN-NDR type. The actual observation of the NDR in Figures 2 and 3 can be attributed to the fact that the NDR is only partially hidden because the potential scan is much faster than the rate of the NDR-hiding process. The oscillation period in the present system is about one hundred to several hundred seconds, which are about 102∼103 times longer than that for normal HN-NDR oscillations reported.12,18-22

18848 J. Phys. Chem. B, Vol. 109, No. 40, 2005

Nakanishi et al.

Figure 4. (a) Current-controlled j vs U in 0.5 M CuSO4 + 0.5 M H2SO4 in the absence (dashed curve) and presence (solid curve) of 2.0 mM o-phen, obtained in the first negative current scan. The scan rate is 0.1 mA/s. (b) U vs t at a constant j of -14 mA cm-2.

This fact indicates that the NDR-hiding process is quite slow, in harmony with the above argument. We did real-time (in situ) observation of electrodeposits during the potential oscillation using a DVC. We also inspected the electrodeposit surface at various stages of the potential oscillation with a high-resolution SEM by pulling out the electrode from the electrolyte at each stage. The pictures on the left and the right sides of Figure 5 are the DVC and the SEM images, respectively, where the number of each stage means that the image was taken at the stage of the potential oscillation, marked by the same number in Figure 4b. The DVC images indicate that the electrode surface changes in color from black (stage-1) to gray (stage-2) and inversely from gray (stage3) to black (stage-4), synchronizing with the potential oscillation. On the other hand, the SEM images show that the electrodeposit is composed of leafletlike particles of a few micrometers in size, and the particles change their shape in synchronization with the potential oscillation as follows: At stage-1, just after the U shifted to the positive in the potential oscillation (Figure 4b), a large number of leaflets are produced vertical to the substrate (Figure 5). The leaflets become slightly thick and round at stage2, accompanied by a slight negative U shift. At stage-3, the leaflets get considerably thick and round, accompanied by a steep negative U shift. At stage-4, new leaflets start to grow vertical to the substrate, with a gradual positive U shift. Then, the growth of a large number of new thin leaflets, accompanied by a sudden U shift toward the positive, leads to stage-1 again. Comparison of the DVC and SEM images indicates that the thin leaflets growing vertical to the substrate gives the black image in the DVC, whereas the dense round leaflets gives the gray image in the DVC. Similar DVC and SEM inspection during the current oscillation gave exactly the same results as for the potential oscillation. Namely, the thin leaflets lying vertical to the substrate were formed at the low-current state, whereas the dense round-shaped leaflets were formed at the high-current state. Figure 6 shows a cross-sectional view of a Cu deposit obtained during the potential oscillation of four periods. For this experiment, the sample (a Cu deposit on the Au substrate) was cut vertical to the substrate, and then the cross section was polished with aluminum powder and etched with 0.25 M aqueous [Cu(NH4)2]Cl4 to get a sharp contrast in a layered structure. Figure 6 clearly shows the formation of four layers in the deposit, in accordance with the number of the oscillation periods.

Figure 5. DVC (left) and SEM (right) images taken at various stages of the potential oscillation at j ) 14 mA cm-2. The numbers of stages, 1-4, of Figure 4 correspond to those, 1-4, of Figure 3b. The electrolyte: 0.5 M CuSO4 + 0.5 M H2SO4 + 1.5 mM o-phen.

Figure 6. A cross-sectional view of a Cu deposit obtained during four periods of the potential oscillation in 0.5 M CuSO4 + 0.5 M H2SO4 + 1.5 mM o-phen at j ) -15 mA cm-2.

Figure 7 shows surface SEMs for Cu deposits obtained at potentials where no oscillation is observed, that is, at (a) U ) -200 mV versus Ag/AgCl, which lies considerably more negative than the potential region of the NDR (UNDR), and at (b) U ) -80 mV versus Ag/AgCl, which lies considerably more positive than the UNDR. The deposit at a potential in U < UNDR (Figure 7a) is composed of a large number of thin leaflets, similar to the surface at stage-1 of the potential oscillation (Figure 5), whereas that at a potential in U > UNDR (Figure 7b)

Layer-by-Layer Electrodeposition of Copper

Figure 7. Surface SEMs for Cu deposits obtained at (a) U ) -200 mV and (b) U ) -80 mV vs Ag|AgCl. The electrolyte: 0.5 M CuSO4 + 0.5 M H2SO4 + 1.5 mM o-phen.

J. Phys. Chem. B, Vol. 109, No. 40, 2005 18849 that this oscillation is classified into the HN-NDR type,11,17 as mentioned earlier. It is thus of key importance to clarify the origins of the NDR as well as the process that hides the NDR for elucidating the mechanism of this HN-NDR oscillator. Let us first consider the origin of the NDR. When o-phen is added to the electrolyte, the Cu-deposition current considerably decreases, compared with the case of no o-phen (Figure 2). In addition, this current decrease becomes more prominent as the solution pH gets lower (Figures 2 and 3). The current decrease can be attributed to adsorption of protonated (positively charged) o-phen, o-phen-H+ (pKa ) 4.93), or o-phen-H22+ (pKa ) 1.83), formed in an acidic solution, on the electrode (Cu-deposit) surface, which suppresses the Cu deposition (or Cu2+ reduction) reaction. This argument is supported by in-situ STM experiments23 by Sugimasa et al., who reported that a stable adsorption layer of o-phen was observed on Cu(111) in an acidic solution in a U region from -0.1 to -0.35 V versus Ag|AgCl. The concentration of protonated o-phen will be higher in lower-pH solutions, thus leading to its larger adsorption and hence stronger suppression of the deposition current in the lower-pH solutions, as observed in Figures 2 and 3. The adsorption of protonated o-phen, however, cannot explain the origin of the NDR because the adsorption occurs in a wide U region below about +0.05 V, as estimated from the j decrease in Figures 2 and 3, contrary to the appearance of the NDR in a rather narrow U region. A part of o-phen will exist in the form of a complex with Cu2+ ions, [Cu(II)(o-phen)2]2+, in the acidic Cu2+ solution. It is reported24-26 that electrochemical redox reactions for the complex in a neutral aqueous solution

[Cu(II)(o-phen)2]2+ + e- f [Cu(I)(o-phen)2]+

Figure 8. (a) j vs t when the U was stepped to various values shown in the figure in 0.5 M CuSO4 + 0.5 M H2SO4 + 1.5 mM o-phen. (b) j vs t in the same solution as a, in which the U is initially kept at -200 mV and then stepped to -80 mV at around t ) 360 s.

is composed of dense, round leaflets, similar to the surface at stage-3 of the potential oscillation (Figure 5). This result implies that the U plays a key role in determining the morphology of the deposit surface. We did another experiment for investigating the oscillation mechanism. Figure 8a shows j versus t at varied U. When the U is stepped to a potential in U > UNDR and fixed at this potential, the j remains nearly constant except for the initial change for about 50 s (curves 1-4). On the other hand, when the U is stepped to a potential in U < UNDR and fixed at this potential, the j increases (in the absolute value) gradually with time, apart from the initial rapid decrease (curves 5 and 6). Figure 8b shows the j versus t in another experiment, in which the U is first stepped to -200 mV (in U < UNDR) and then shifted suddenly to -80 mV (in U > UNDR) at t ) around 360 s. While the U is at -200 mV, the j keeps increasing (in the absolute value), whereas the j decreases after the U is shifted to -80 mV. Discussion The copper electrodeposition from aqueous acidic Cu2+ solutions with o-phen shows both the current and the potential oscillations, together with the (partially hidden) NDR, indicating

(1)

are reversible, the equilibrium redox potential, Uredox, being located at about -0.05 V versus Ag/AgCl. It is also reported26 that the reduced form of the complex, [Cu(I)(o-phen)2]+, is strongly adsorbed on a graphite electrode in a U region more negative than the Uredox. Although the reported experiments were performed in a neutral aqueous solution with a graphite electrode, we can expect that similar adsorption also occurs in the present system. Namely, we can expect that the reduced form of the complex is strongly adsorbed on the Cu deposit in a U region more negative than the Uredox, which suppresses the Cu deposition (or Cu2+ reduction). (The concentration of o-phen is much lower than that of Cu2+ in the present system and thus the major part of Cu2+ exists as free Cu2+ ions, which are reduced to Cu metal.) The amount of the reduced form, [Cu(I)(o-phen)2]+, increases steeply with a negative U shift near the Uredox, and thus the surface coverage (θred) for the adsorbed reduced form also increases steeply with the negative U shift. This implies that the Cu-deposition current decreases sharply with the negative U shift near the Uredox, namely, an NDR appears at around Uredox. One may point out that the UNDR in Figures 2 and 3 is somewhat more negative than the reported Uredox value (about -0.05 V vs Ag/AgCl) in a neutral aqueous solution. In addition, the UNDR shifts toward the negative with increasing pH (Figures 2 and 3), though the Uredox for eq 1 should be independent of pH. These deviations between the UNDR and the reported Uredox come from the ohmic drop in the electrolyte between the working and the reference electrodes. A simple correction of the ohmic drop, jSR, where S is the electrode surface area and R is the solution resistance (50 Ω) between the working and the reference electrodes measured by an impedance analysis, has shown that the corrected UNDR’s in all solutions of Figures 2 and 3 are in agreement with the reported Uredox value.

18850 J. Phys. Chem. B, Vol. 109, No. 40, 2005 Next, let us consider what process hides the NDR. This problem has long remained difficult to understand, because any possible electrochemical reaction could not explain the HNNDR behavior in the present system, contrary to the HN-NDR oscillators reported thus far.12,18-22 Our present detailed experiments have finally revealed that the effective surface area, Seff, for the electrode (or the Cu deposit) acts as the NDR-hiding slow variable. The results of Figure 7 indicate that the Cu deposit obtained in U > UNDR has a smooth surface and hence a small Seff, whereas the deposit in U < UNDR is composed of thin leaflets and has a large Seff. This implies that the Cu-deposition current in U < UNDR increases with time by the increase in the Seff because of the growth of thin Cu leaflets, though the deposition current in U > UNDR remains nearly constant with time owing to little change in the Seff. The important point is that the former current increase due to the Seff increase can hide the NDR caused by the adsorption of the reduced complex, [Cu(I)(o-phen)2]+, which also occurs in U < UNDR. This expectation is really confirmed by the experiments of Figure 8. Figure 8a shows that the Cu-deposition current in U < UNDR increases (in the absolute value) with time, except the initial rapid change, though the current in U > UNDR remains nearly constant with time. The observed current increase in U < UNDR clearly indicates that the current increase by the increase in the Seff can really exceed the current decrease due to the adsorption of the reduced complex, [Cu(I)(o-phen)2]+, namely, the current increase by the Seff increase can hide the NDR. Besides, the result of Figure 8b indicates that the increased current due to the increased Seff in U < UNDR decreases again in U > UNDR owing to a change of thin leaflets with large Seff into thick round leaflets with small Seff. It may be helpful to note here that there is no work reporting that o-phen acts as a brightener in metal electrodeposition. Now, let us consider the mechanism of the potential oscillation, which is observed at a constant applied jap, on the basis of the model described thus far. At stage-1 in Figure 4b, just after the U has shifted to the positive and has reached a potential in U > UNDR, a large number of thin leaflets, which were formed in the preceding stage as discussed later, are present at the electrode surface (see stage-1 of Figure 5) with no adsorbed complex, [Cu(I)(o-phen)2]+. The thin leaflets then get thick and round with time at a potential in U > UNDR, as mentioned earlier, accompanied by the gradual decrease in the Seff. The gradual decrease in the Seff causes a gradual negative U shift to keep the constant applied jap. When the leaflets get considerably thick and round and the U approaches UNDR (stage-2), the adsorption of the reduced form of the complex, [Cu(I)(o-phen)2]+, starts to occur, resulting in an increase in the surface coverage (θred) for it. The increase in θred causes a decrease in vacant surface sites for the Cu deposition and thus causes a negative U shift to keep the jap, which in turn leads to a further increase in θred. Here is an autocatalytic process. As the θred increases steeply with the negative U shift near UNDR, the autocatalytic process induces a steep U shift to the negative, and stage-3 is realized. At stage-3, the U lies in the region of U < UNDR (Figure 4b), and thick round Cu leaflets, which are covered with the adsorbed complex, [Cu(I)(o-phen)2]+, exist at the deposit surface (see stage-3 of Figure 5). Under this condition of U < UNDR, thin leaflets grow, accompanied by the increase in the Seff, as mentioned earlier. The increase in the Seff causes a positive U shift to keep the applied jap. When a large number of thin Cu leaflets grow and the U approaches UNDR (stage-4), the adsorbed complex, [Cu(I)(o-phen)2]+, starts to be desorbed from the surface by its oxidation, resulting in a decrease in θred. The

Nakanishi et al. decrease in θred implies the increase in vacant sites, thus causing a positive U shift to keep the constant jap, which in turn causes a further decrease in θred. Here is also an autocatalytic process arising from the same origin as mentioned above. Thus, the U steeply shifts to the positive, and stage-1 is reached again. By this mechanism, one cycle of the potential oscillation adds one layer to the Cu deposit, in agreement with the experiment (Figure 6). The current oscillation under a constant applied Uap can be explained in the same way as the potential oscillation above. In the high-current state (stage-1 of Figure 2b), a large number of thin leaflets, which were produced in the preceding stage as discussed later, are present at the electrode surface with no adsorbed [Cu(I)(o-phen)2]+ complex. Also, at this stage, the true electrode potential (or the Helmholtz-layer potential, E) lies much more positive than the Uap owing to the ohmic drop in the solution, or more precisely because the E is given by E ) Uap - jS R where the current density j is negative for the reduction current. This implies that, even if the Uap is placed in the region of UNDR, the E is more positive than the UNDR, that is, E > UNDR, or more strictly E > ENDR (ENDR refers to the region of E where the NDR is observed). Thus, thin Cu leaflets get thick and round with time, as mentioned earlier, accompanied by a gradual decrease in the Seff. The gradual decrease in the Seff leads to a gradual decrease in the j (see Figure 8b), which induces a gradual decrease in the ohmic drop and thus a gradual negative shift in the E. When the thin Cu leaflets get considerably thick and round and the E approaches the ENDR (stage-2 of Figure 2b), the adsorption of the reduced [Cu(I)(o-phen)2]+ complex starts, resulting in an increase in the θred. An increase in the θred leads to a decrease in the j and a decrease in the ohmic drop in the solution, thus leading to a negative shift in the E, which in turn causes a further increase in the θred. Thus, the E sharply shifts toward the negative by the same autocatalytic process as mentioned earlier, resulting in the appearance of the low-current state (stage-3). In the low-current state with E < ENDR, new thin leaflets grow on the previously produced thick round leaflets, accompanied by an increase in the Seff. The increase in the Seff causes an increase in the j and an increase in the ohmic drop in the solution and then causes a positive E shift (see Figure 8a and b). When a large number of thin leaflets grow and the E approaches the ENDR, the desorption of the adsorbed [Cu(I)(ophen)2]+ complex starts, resulting in a decrease in the θred. The decrease in the θred induces an increase in the j and a positive shift in the E, which in turn leads to a further decrease in the θred. Thus, the system goes into the high-current state again. Finally, we have to discuss why thin leaflets grow in U < UNDR and why the thin leaflets become thick and round in U > UNDR. We previously reported2,5,14-16 that needle- or leafletlike deposits were produced by the autocatalytic crystal-growth mechanism, which works under the diffusion-limited condition. However, in the present system, the current density j for the Cu deposition does not reach the diffusion-limited value even at a potential of the negative end of the potential oscillation. This implies that this mechanism cannot be applied to the present oscillation. An alternative possible explanation may be given by assuming that the reduced [Cu(I)(o-phen)2]+ complex is adsorbed in U < UNDR only on a particular crystal face of Cu, that is, only on the flat side surfaces of the Cu leaflets, as schematically shown in Figure 9. Thus, the Cu growth in this direction is strongly suppressed, or in other words, the Cu growth only occurs in the direction parallel to the leaflet plane (Figure 9), thus resulting in the growth of thin leaflets. A similar

Layer-by-Layer Electrodeposition of Copper

Figure 9. Schematic illustrations of growth modes for the Cu deposit in the presence of o-phen in the potential region of (a) U < UNDR (or E < ENDR) and (b) U > UNDR (or E > ENDR).

mechanism leading to anisotropic crystal growth is reported27-30 for the deposition of metals such as Zn, Cu, and Sn in the presence of organic additives. In U > UNDR, on the other hand, the adsorbed [Cu(I)(o-phen)2]+ complex is desorbed from the surface, and then isotropic crystal growth can occur, resulting in the formation of thick round leaflets. Further studies are, however, necessary to get a definite conclusion. In conclusion, the present work has revealed that the Seff can work as the slow NDR-hiding valuable for both the current and the potential oscillations. The time constant for the Seff change is much larger than that for the normal NDR-hiding valuables reported, thus resulting in the oscillations of the long period of several hundred seconds. Detailed studies on the present newtype oscillations will serve for deep understanding of nonlinear kinetics as well as development of micro- and nanostructuring by oscillatory electrodeposition. Acknowledgment. The authors thank the Core Research for Evolutional Science and Technology (CREST) program of the Japan Science and Technology Agency (JST) for financial support. This work was also partly supported by a Grand in Aid of the Ministry of Education, Culture, Sport, Science, and Technology (MEXT) for scientific research and by the Kao Foundation for Arts and Science. References and Notes (1) Krischer, K. Nonlinear Dynamics in Electrochemical Systems. In AdVances in Electrochemical Science and Engineering; Alkire, R. C., Kolb, D. M., Eds.; Willey-VCH: Weinheim, Germany, 2003; Vol. 8, p 89. (2) Nakanishi, S.; Fukami, K.; Tada, T.; Nakato, Y. J. Am. Chem. Soc. 2004, 126, 9556.

J. Phys. Chem. B, Vol. 109, No. 40, 2005 18851 (3) Sakai, S.-I.; Nakanishi, S.; Fukami, K.; Nakato, Y. Chem. Lett. 2002, 31, 640. (4) Nakanishi, S.; Sakai, S.-I.; Nagai, T.; Nakato, Y. J. Phys. Chem. B 2005, 109, 1750. (5) Tada, T.; Fukami, K.; Nakanishi, S.; Yamasaki, H.; Fukushima, S.; Nagai, T.; Sakai, S.-I.; Nakato, Y. Electrochim. Acta, in press. (6) Switzer, J. A.; Hung, C. J.; Huang, L. Y.; Switzer, E. R.; Kammler, D. R.; Golden, T. D.; Bohannan, E. W. J. Am. Chem. Soc. 1998, 120, 3530. (7) Schlitte, F.; Eichkorn, G.; Fischer, H. Electrochim. Acta 1968, 13, 2063. (8) Krastev, I.; Koper, M. T. M. Physica A 1995, 213, 199. (9) Saliba, R.; Mingotaud, C.; Argoul, R.; Ravaine, S. Electrochem. Commun. 2002, 4, 629. (10) Zhang, M. Z.; Wang, M.; Zhang, Z.; Zhu, J. M.; Peng, R. W.; Wang, N. M. Electrochim. Acta 2004, 49, 2379. (11) Strasser, P.; Eiswirth, M.; Koper, M. T. M. J. Electroanal. Chem. 1999, 478, 50. (12) Mukouyama, Y.; Nakanishi, S.; Chiba, T.; Murakoshi, K.; Nakato, Y. J. Phys. Chem. B 2001, 105, 7246. (13) Mukouyama, Y.; Nakanishi, S.; Konishi, H.; Ikeshima, Y.; Nakato, Y. J. Phys. Chem. B 2001, 105, 10905. (14) Nakanishi, S.; Fukami, K.; Sakai, S.-I. Nakato, Y. Chem. Lett. 2002, 31, 636. (15) Fukami, K.; Nakanishi, S.; Sakai, S.-I.; Nakato, Y. Chem. Lett. 2003, 32, 532. (16) Fukami, K.; Nakanishi, S.; Tada, T.; Yamasaki, H.; Sakai, S.-I.; Fukushima, S.; Nakato, Y. J. Electrochem. Soc. in press. (17) Krischer, K. Principles of Temporal and Spatial Pattern Formation in Electrochemical Systems. In Modern Aspects of Electrochemistry; White, R. W., Bockris, O. M., Conway, R. E., Eds.; Plenum: New York, 1995; Vol. 32, p 1. (18) Strasser, P.; Lu¨bke, M.; Eickes, C.; Eiswirth, M. J. Electoanal. Chem. 1999, 462, 9. (19) Wolf, W.; Krischer, K.; Lu¨bke, M.; Eiswirth, M.; Ertl, G. J. Electroanal. Chem. 1995, 385, 85. (20) Strasser, P.; Lu¨bke, M.; Raspel, F.; Eiswirth, M.; Ertl, G. J. Chem. Phys. 1997, 107, 979. (21) Treindl, L.; Doblhofer, K.; Krischer, K.; Samec, Z. Electrochim. Acta 1999, 44, 3963. (22) Nakanishi, S.; Sakai, S.-I.; Hatou, M.; Mukouyama, Y.; Nakato, Y. J. Phys. Chem. B, 2002, 106, 2287. (23) Sugimasa, M.; Inukai, J.; Itaya, K. J. Electrochem. Soc. 2003, 150, E266. (24) Fung, C.-S.; Wong, K.-Y. J. Electroanal. Chem. 1996, 401, 263. (25) Culjak, I.; Mlakar, M.; Branica, M. Electroanalysis 1995, 7, 64. (26) Lee, C.-W.; Anson, F. C. Inorg. Chem. 1984, 23, 837. (27) Despic _, A. R. Deposition and Dissolution of Metals and Alloys. In ComprehensiVe Treatise of Electrochemistry; Conway, R. E., Bockris, O. M., Yeager, E., Khan, S. U. M., White, R. E., Eds.; Plenum: New York, 1983; Vol. 7, p 451. (28) Carlos, I. A.; Bidoia, E. D.; Pallone, E. M. J. A.; Almeida, M. R. H.; Souza, C. A. C. Surf. Coat. Technol. 2002, 157, 14. (29) Trejo, G.; Ruiz, H.; Borges, R. O.; Meas, Y. J. Appl. Electrochem. 2001, 31, 685. (30) Yu, J.; Chen, Y.; Yang, H.; Huang, Q. J. Electrochem. Soc. 1999, 146, 1789.