J. Phys. Chem. B 2006, 110, 21109-21117
21109
The Influence of Poly(ethylene oxide) and Illumination on the Copper Electrodeposition Process onto n-Si(100) Eduardo C. Mun˜ oz,† Ricardo S. Schrebler,*,† Paula K. Cury,† Claudio A. Sua´ rez,† Ricardo A. Co´ rdova,† Carlos H. Go´ mez,† Ricardo E. Marotti,‡ and Enrique A. Dalchiele‡ Instituto de Quı´mica, Facultad de Ciencias, Pontificia UniVersidad Cato´ lica de Valparaı´so, Casilla 4059, Valparaı´so, Chile, and Instituto de Fı´sica, Facultad de Ingenierı´a, UniVersidad de La Repu´ blica, Herrera y Reissig 565, C.C. 30, 11000 MonteVideo, Uruguay ReceiVed: May 26, 2006; In Final Form: August 14, 2006
In this study, we examined the influence of illumination and the presence of poly(ethylene oxide) (PEO) as an additive for the copper electrodeposition process onto n-Si(100). The study was carried out by means of cyclic voltammetry (CV) and the potential steps method, from which the corresponding nucleation and growth mechanism (NGM) were determined. Likewise, a morphologic analysis of the deposits obtained at different potential values by means of atomic force microscopy (AFM) was carried out. In a first stage, Mott-Schottky measurements so as to characterize the energetics of the semiconductor/electrolyte interface were made. Also, parallel capacity measurements were carried out in order to determine the surface state density of the substrate. It was found that when PEO concentration is increased, the number of these surface states decreases. The CV results indicated that the presence of PEO inhibits the photoelectrochemical reaction of oxide formation on the surface of the semiconductor. This allows a decrease in the overpotential associated with the electrodeposition process. The analysis of the j/t transients shows that the NGM corresponds to progressive three-dimensional (3D) diffusional controlled (PN3DDiff), which was confirmed by the AFM technique. Neither illumination nor the presence of PEO changes the mechanisms. Their influence is in that they diminish the size of the nuclei and the speed with which these are formed, which produces a more homogeneous electrodeposit.
1. Introduction In recent decades, semiconductor materials have been intensively researched, principally because their applications in energy conversion devices in the microelectronic field, and in the photocatalysis and photoelectrocatalysis area, when the surfaces of these materials are modified superficially by metallic elements. In the last-mentioned cases, much of this research has been related to the study of the formation of metallic phases on semiconductor substrates by means of the electrodeposition technique. Likewise, the electric properties of these heterostructures have been evaluated. Different authors1-17 have studied copper electrodeposition on silicon. From these studies, it has been possible to establish that copper electrodeposition on a n-Si semiconductor is influenced by different factors, which include the surface energetic condition of the semiconductor and how it is affected by the electric potentials applied to the semiconductor/electrolyte interface. Another phenomenon that has been observed on monocrystalline silicon is that copper electrodeposition occurs first on substrate defects (kinks, steps, edges) and afterward the process extends to the less active sites (terraces). These features have been observed by means of cyclic voltammetry measurements, which show a splitting of the signal associated with copper electrodeposition, and this has also been confirmed by STM * Corresponding author. E-mail:
[email protected]. Telephone/Fax: +5632-273422. † Pontificia Universidad Cato ´ lica de Valparaı´so. ‡ Universidad de La Repu ´ blica.
measurements.17 In addition, the process of copper electrodeposition on both sites of the n-Si substrate appears at a relatively high overpotential. This is in line with the idea that the adsorption free energy of the copper adatom on this substrate is low,3,18-19 which it is directly related to the kinetic aspects of the electrodeposition process. Regarding the nucleation and growth processes of copper electrodeposition on n-Si, it has been found by means of dimensionless plots that they correspond to a progressive nucleation of three-dimensional (3D) islands (PN3DDiff), where growth is controlled by diffusion (Vollmer-Weber mechanism).2-7,11 This behavior indicates that metal-metal interaction is stronger than the substrate-metal interaction. The copper nucleation and growth process onto silicon has been confirmed by means of various techniques such as scanning electronic microscopy (SEM), atomic force microscopy (AFM), and scanning tunneling microscopy (STM).1,5,7,11,12,16,20,21 Furthermore, this is well-known that the morphology reached by copper electrodeposits is influenced by the presence of additives in the electrolytic bath. It has been reported that, for copper electrodeposition on Si, the use of additives such as polyethylenglycol (PEG), thiourea, and bis(3-sulfopropyl) bisulfide improve the homogeneity and texture of the deposits.11,15,22-24 However, the action of the polymeric additive in nucleation and growth kinetics has not been quantitatively evaluated. In the particular case of polyethers, there are some contradictions in the literature concerning the action mechanism of these additives.15,22,25-26 In fact, these contradictions are related to adsorption phenomena of this additive. Some authors established that the adsorption takes place through complex formation
10.1021/jp063246k CCC: $33.50 © 2006 American Chemical Society Published on Web 09/30/2006
21110 J. Phys. Chem. B, Vol. 110, No. 42, 2006
Mun˜oz et al.
between metal-polyether.15 On the other hand other authors have postulated that polyether molecules are directly adsorbed onto the electrodic surface.22,25,26 Another aspect that has not been studied very much is the illumination effect during the nucleation and growth processes of the metallic phase onto semiconductor substrates such as silicon.21 The aim of this study is to investigate the influence of illumination and of the poly(ethylene oxide) (PEO) additive on the nucleation and growth mechanism of copper onto n-Si(100) in a sulfuric acid media. The study was performed by cyclic voltammetry (CV) and step potential methods (SP). In addition, the morphology of the copper deposits was followed by atomic force microscopy (AFM). Also, a discussion about the nucleus density obtained by both methods was realized. 2. Experimental Section Electrodeposition was performed on monocrystalline n-Si (100) with a resistivity between 1 and 5.5 Ω cm (ND = 1 × 1015 cm-3), P-doped, and with polished/etched surfaces (Int. Wafer Service, CA). The silicon wafer was cut into squares (1.0 × 1.0 cm2) that were first degreased in boiling acetone for 10 min then sequentially cleaned ultrasonically for 10 min in acetone, ethanol, and finally, in water. The water was distilled and deionazed (Millipore) and had a resistivity of 18 MΩ cm. Then the electrodes were treated for 10 min with a 1:1 H2SO4/ H2O2 mixture heated to 80 °C to remove any trace of heavy metals and organic species. Afterward, the oxide film was removed by etching with 2 M HF solution for 2 min and thoroughly rinsed with ultrapure water. The ohmic contact was made with InGa eutectic on the etched face of the samples, and the electrodes were mounted on a Teflon holder. The silicon area exposed to the solution was 0.16 cm2. Before the electrochemical experiments, the electrode surface was again etched for 2 min in 2 M HF solution. After this procedure, an atomically smooth and hydrogen-terminated surface is obtained. For each measurement, a new electrode of n-Si(100) was used because it is well-known that copper can diffuse toward the inside of the silicon.3,7,8,11,12,14,27-31 A platinum wire was used as a counter electrode, and a mercury/mercury sulfate electrode (MSE) (Hg/Hg2SO4, K2SO4 (saturated), 0.640 V vs NHE) was used as a reference electrode. It was positioned close to the silicon samples using a Luggin capillary. All the potentials reported in this study refer to this reference electrode. The electrolytic solutions were prepared from analytical grade reagents with the following compositions: 5 mM CuSO4‚5H2O + 0.1 M H2SO4 + x mg/L [PEO] (0 e x e 100 mg/L). The molecular weight of the additive was 1 000 000. Cyclic voltammetry experiments were carried out at room temperature and at a scan rate of 0.010 V s-1 according to the perturbation programs indicated in the corresponding figures. For the chronoamperometric measurements, the following E/t program was applied to the electrode: the potential was first held for 5 s at a potential value where copper electrodeposition was not observed and then stepped to the electrodeposition potential Ed. The Ed values varied between -0.750 and -1.050 V. All the measurements were carried out in darkness and under illumination. For the latter, a xenon lamp of 75 W (Oriel Instruments 6263) mounted in a lamp holder (Oriel 66902) was connected to the cell through a water filter (Oriel 61945) and a 1 m length optical fiber (Oriel 77578). A power supply of 40200 W (Oriel 68907) was used to generate the arc in the lamp. The illumination power was quantified inside the cell by means
Figure 1. Potentiodynamic j/E profiles of a n-Si(100) electrode in 0.1 M H2SO4, in darkness (a) and under illumination (b). Without copper ions (s) and with 5 mM CuSO4 (- - -). Inset shows the applied E/t perturbation program. Scan rate, 0.010 V/s.
of an energy radiant meter (Oriel 70260). The samples were always illuminated with a constant intensity of 2 mW cm-2. A pure argon stream was passed through the solution for 30 min before measurements and over the solution when the experiments were under way. The electrochemical measurements (cyclic voltammetry and chronoamperometric) were made using a Princeton Applied Research (PAR) model 273 A equipment. The electrochemical impedance spectroscopy (EIS) measurements were made using a Zahner model IM6e equipment. All samples used for ex situ AFM were prepared in the electrochemical cell. The AFM images were obtained with a Digital Instruments Nanoscope IIIa series, employed in tapping mode at a scan rate of 0.02 µm/s. 3. Results and Discussion 3.1. Cyclic Voltammetry Analysis of the n-Si(100) in Darkness and under Illumination, in the Absence of Poly(ethylene oxide). Figure 1 shows the potentiodynamic j/E profiles for a n-Si(100) electrode in 0.1 M sulfuric acid in darkness (Figure 1a) and under illumination (Figure 1b), without copper ions (solid curve) and with 5 mM CuSO4 (dotted curve). These profiles were measured with the E/t perturbation program shown in the inset of this figure. At potentials more negative than Ei, and in the absence of copper ions, the only feature in evidence is the hydrogen evolution reaction (HER) beginning at -1.1 V. In the presence of copper ions and under the two conditions (darkness and illumination), a complex cathodic peak is developed previous to the HER, which presents at least two current contributions marked as C1 and C2. In darkness, the C1 peak is located at -0.97 V, and under illumination it shifts slightly toward more negative potential values (-1.00 V). The C2 peak is located at -1.07 V and its value remains constant under illumination or in darkness. These two contributions correspond to copper electrodeposited onto the semiconductor substrate on sites having different energy.
Influence of PEO/Light on Copper Deposition onto n-Si
J. Phys. Chem. B, Vol. 110, No. 42, 2006 21111
Figure 3. Band diagrams in different situations for the system n-Si(100)/electrolyte: (a) flat bands; (b) equilibrium with 0.1 M H2SO4 + 5 mM CuSO4 electrolyte; (c) cathodic polarization at -0.70 V (start accumulation layer); (d) cathodic polarization at -0.95 V (accumulation layer); (e) n-type silicon covered with a copper film (Schottky barrier formation); (f) depletion layer under illumination at -0.40 V (copper dissolution).
Figure 2. Mott-Schottky plots of n-Si(100) in 0.1 M H2SO4 obtained from measurements done at (a) 100 kHz and (b) 10 kHz. The corresponding flat band potential values are indicated.
Furthermore, in the presence of copper ion, the HER onset is located at more positive potentials, probably due to the fact that the reaction is occurring on previously electrodeposited copper. Therefore, the current values are 1 order of magnitude higher than the values on naked n-Si(100). In the reverse scan, a hysteresis loop can be seen, which is associated with an increase in the electrode area due to the effect of the copper electrodeposition process. In darkness, as was expected, at potentials more positive than -0.5 V and under the two conditions (presence and absence of copper ions), no anodic processes are observed. On the other hand, when the electrode is illuminated and in the absence of copper, a typical anodic photocurrent from -0.78 V, corresponding to the oxidation of an n-type semiconductor, can be seen. In the presence of copper, the corresponding stripping peaks labeled A1 and A2 can be seen. It is important to notice that the anodic profile remains complex where A1 and A2 are associated with the oxidation of the copper atoms previously deposited on silicon sites with higher and lower activity, respectively. The electrodeposition of copper onto n-Si(100) substrates can be explained by means of the band diagram (BD) of the semiconductor/electrolyte interface. For this reason, the position of the flat band potential (VFB) and the majority carrier density value (ND) from the corresponding Mott-Schottky plots were obtained. Moreover, by using these values, the location of the valence (VB) and conduction (CB) bands were estimated. Figure 2 shows the Mott-Schottky plots for the n-Si(100)/ 0.1 M H2SO4 system obtained at 100 kHz (Figure 2a) and 10 kHz (Figure 2b). In Figure 2a, it can be seen that there is linear behavior starting from -0.6 V, while in Figure 2b, this behavior begins at -0.5 V. In the latter case, it can be seen that there is a change in the slope, which might be due to the presence of surface states in the semiconductor substrate, and indeed, this has been suggested by other authors.7,32 The extrapolation of the linear regions in these plots allowed the evaluation of the flat band potential (VFB). The VFB values found at 100 and 10
kHz were -0.660 and -0.665 V, respectively, which is in agreement with the values reported in the literature.3,7,26,31,33-35 Furthermore, from the slope values of the Mott-Schottky plots and using the eq 1, a majority carrier density value ND ) 3 × 1015 cm-3 was obtained.
1 2 kT ) ‚V2 e‚‚ ‚N e C 0 D
(
)
(1)
where C corresponds to the space charge layer capacitance; e, , and 0 are the electron charge, the dielectric constant of the semiconductior, and the vacuum permittivity, respectively. ND is the majority carrier density; kT has the usual significance, and V is the applied potential. From these parameters, the band diagram for the n-Si(100)/ Cu2+ interface in darkness was drawn (see Figure 3). Figure 3a shows the flat band condition, which has been calculated from the corresponding flat band potential value and from the majority carrier density. Figure 3b shows the BD corresponding to the equilibrium condition, which was constructed by considering that the open circuit potential (Eoc) of the n-Si(100)/5 mM CuSO4 + 0.1 M H2SO4 interface corresponds to -0.480 V. In this situation the system is under a depletion layer. The thickness of the space charge region (W), determined from the MottSchottky equation, yields a value of 183 nm. In the potential region where the metal electrodeposition begins ([Eoc + η] < -0.800 V), the system reaches an accumulation situation as shown in Figure 3c. For a potential of -0.95 V, the space charge region thickness was approximately 258 nm (Figure 3d). Under these conditions, the majority carriers (electrons) present in the conduction band accumulate on the silicon surface. Regarding the energetic distribution of the redox couple in solution, the accumulated electrons are transferred to the acceptor levels (WOX) of the Cu2+/Cu system. In this way, the reduction process observed is explained. For potential values
