Current Opinion in Solid State and Materials Science 10 (2006) 163–172
Electrochemical metal deposition on silicon Yukio H. Ogata b
a,*
, Katsutoshi Kobayashi a, Munekazu Motoyama
b,1
a Kyoto University, Institute of Advanced Energy, Uji, Kyoto 611-0011, Japan Kyoto University, Graduate School of Energy Science, Yoshida-Honmachi, Sakyo, Kyoto 606-8501, Japan
Received 2 February 2007; accepted 9 February 2007
Abstract Electrochemical metal deposition is utilized to fabricate micro- and nano-structures. A variety of the structures have been achieved in metal patterning and structuring and also in the structure formation of silicon itself. The controlling factors and conditions of the size, morphology and distribution have been investigated. The importance of metal deposition by displacement reaction should be recognized. Ó 2007 Published by Elsevier Ltd. Keywords: Electrodeposition of metal; Silicon; Porous silicon; Displacement deposition; Micro- and nano-structure formation
1. Introduction Anodic dissolution of silicon has been intensively studied for many years, whereas the cathodic behavior including metal deposition had not received much attention. However, since application of a wet process for copper wiring was demonstrated by the IBM group in 1997 [1], the importance of the electrochemical metal deposition has been widely recognized. The wet process is now extensively applied to the fabrication of micro- and nano-structures. Silicon is one of the most appropriate substrates for this purpose because an extremely well-defined surface is available, minute structural building is possible on and in the surface, and silicon is easily dissolved with alkaline solution leading to the use of silicon as a template to produce macro- or nano-structures. Local control of deposition becomes very important to achieve the minute structure formation. Understanding of the nucleation process [2] and the metal deposition behavior on silicon [3,4] is indispensable for the realization. *
Corresponding author. Tel.: +81 774 38 3500; fax: +81 774 38 3499. E-mail addresses:
[email protected] (Y.H. Ogata),
[email protected] (K. Kobayashi), Munekazu.Motoyama@t02. mbox.media.kyoto-u.ac.jp (M. Motoyama). 1 Tel./fax: +81 75 753 4719. 1359-0286/$ - see front matter Ó 2007 Published by Elsevier Ltd. doi:10.1016/j.cossms.2007.02.001
The present review focuses on the work reported in 2005 and 2006 with some additional important reports published previously. Copper deposition for wiring for integrated-circuits and printed-wiring boards are not dealt with here because it is a vast field and it should be treated as a separate topic. Many interesting fine structures such as nanorods, nanotubes and nanorings have been reported recently. These structures emerge spontaneously in some cases, and in other cases templates are needed for structure formation. Well-ordered alignment of the individual structures is also required depending upon the purposes. The alignment can be performed by pre-patterning of a silicon surface via lithography or other methods, or restriction of the reaction sites, for example, optically. Now the properties of microor nano-deposits are very important in addition to the bulk properties. Metal deposition by displacement reaction or electroless deposition without reducing agent in electrolyte is often used. Immersion deposition is the simplest electrochemical process and a favorable process to control a very small amount of deposition; furthermore, silicon is the preferred substrate because of the less-nobleness although it must be noted that the surface is subject to oxidation. Here we review recent research in electrochemical metal deposition on silicon focusing on the formation of micro- and nano-structures and related background such as displacement deposition and nucleation-and-growth of deposit
164
Y.H. Ogata et al. / Current Opinion in Solid State and Materials Science 10 (2006) 163–172
nuclei as well as a number of other interesting works, including some of our activities.
2. Characteristics of metal deposition on silicon Metal deposition on silicon has some characteristic features [5]. First, silicon and deposited metal usually show a weak interaction leading to the 3D island growth or Volmer–Weber mechanism [6]. Electrochemical reaction on silicon is generally sluggish. Once the surface receives metal deposition, electrochemical reaction is possible also on the deposit surface, where the reaction rate is faster than that on the uncovered silicon surface. The relative deposition rates on the substrate and deposits determine the morphology of deposits. Second, silicon is a semiconductor and hence charge transfer occurs through the bands: electron transfer via the conduction band or hole injection to the valence band for the cathodic or deposition reaction. Charge transfer via the other levels such as surface states is also possible. The semiconductor characteristics also enable the photo-excita-
tion (Fig. 1). The photo-excitation is effective even after the surface is covered by deposits until absorption of light becomes zero at the metal–silicon substrate interface due to the complete absorption by the deposited layer. A Schottky-like energy barrier formed at the interface hinders the transfer of majority carriers, but the energy bending is favorable to the transfer of photo-excited minority charge carriers. Deposition of metal produces a Schottky barrier when the work function of metal is larger than that of ntype silicon or smaller than that of p-type silicon. Porous alumina and track-etched polycarbonate templates are extensively studied for templates in processes producing metal nanostructures by depositing the desired materials into the pores followed by dissolution of the substrates [7–9]. They are insulating materials, and hence additional procedures, such as introduction of conductive material at the pore bottom in electrodeposition and sensitization by catalyst on the pore wall in electroless deposition, are required for the pore filling to fabricate the structures. Porous semiconductors are not insulating although the conductivity is dependent on the dopant level. The difference makes direct metal deposition on the pore wall possible in electrodeposition on silicon. Furthermore, silicon is less noble and the surface acts as a reducing agent resulting in spontaneous deposition of some metals. These properties are favorable for metal deposition on the one hand, but deteriorate the controllability of the deposition on the other hand. The conductivity and photoactivity provide us promising possibilities of silicon over insulating materials as template for nanostructure fabrication after overcoming the drawbacks. 3. Micro- and nano-structures Silicon is one of the most sophisticated materials accepting micro- or nano-machining as a result of the developed integrated-circuit technology. This technology can be utilized for sophisticated metal deposition on silicon producing 2D and 3D structures. The extreme way is to utilize self-assembly, and many studies have been conducted. 3.1. Position-selective local deposition and metal patterning
Fig. 1. Energy diagrams of silicon/electrolyte interfaces during metal deposition on n-type silicon (a) and p-type silicon (b). On n-type silicon, metal deposition proceeds in a system in which the redox potential is situated in the band gap when a negative overpotential is applied and the potential barrier is reduced. Charge transfer via the valence band (VB) is also possible if the redox potential is situated near the valence band. On ptype silicon, no deposition takes place under polarization in the dark for lack of electrons in the conduction band (CB) when the redox potential is positioned in the bandgap. Electrons are generated by illumination and metal deposition becomes possible under illumination.
Local cathodic reaction takes place on p-type silicon when the substrate is photo-excited locally, and an energy level of a redox system is situated suitably above the valence band (Fig. 1). Local illumination by a laser beam enables position-selective metal deposition (Fig. 2). This method provides a possibility of maskless and/or resistless patterning of metal on silicon. Sheck et al. demonstrated position-selective copper deposition on p-type silicon by focusing a laser spot of 1–2 lm on the surface in acidic CuSO4 solution at open-circuit [*10]. The deposited dot sizes were about 30 lm on the silicon of 5–25 X cm. The blurring results from the high diffusion length of photoexcited electrons: it can reach in the range of sub-millimeters. They applied P-ion implantation to the surface to
Y.H. Ogata et al. / Current Opinion in Solid State and Materials Science 10 (2006) 163–172
165
Fig. 2. Position-selective metal deposition with local illumination on p-type silicon. (a) Photo-generated electrons are consumed for metal deposition under cathodic polarization, while metal deposition is accompanied by oxidation of silicon under open-circuit condition. The generated electrons diffuse outside of the illuminated area because of their large diffusion length causing blurring of dotted patterns. (b) Nickel dot pattern prepared in an acidic solution containing nickel ions under cathodic polarization. A laser beam was suspended for ten seconds at each spot.
introduce defects and enhance the recombination rate and obtained dots of about 10 lm. Sasano et al. also showed position-selective deposition of copper but by photoassisted electrodeposition in their case [11]. The dot size was about 20 lm using a 5 lm laser spot without additional treatment. They used porous silicon instead of flat silicon as a substrate. The rough surface suppresses blurring. They also drew a pattern by moving the substrate installed on an x–y stage. Laser interference lithography (LIL) may be applicable to this method [12,*13]. The method makes one-time exposure for a large area patterning possible. The method is applicable to systems in which the redox potential is far from the valence band of silicon or generally less-noble metal systems [14]. If not, displacement reaction of metal ions takes place spontaneously, and it hinders the patterning. The metal patterning is basically possible only on a p-type semiconductor. However, copper patterning on n-type silicon becomes possible if we use different reducing behavior between porous and bare silicon surfaces. Porous layer patterning that results from anodization is possible on n-type silicon. When the patterned specimen is dipped in acidic solution containing cupric ions, copper is deposited only at the porous part and it makes the patterning. This is caused by higher reducing ability on the porous surface than on the bare surface [15]. The photoassisted patterning is interesting but the minimum patterned size is limited to dimensions of micrometers. The use of short wavelength, ultraviolet or vacuum ultraviolet causes absorption by electrolyte, and the blurring of excited carriers is inevitable even though short wavelength light has been used. Finer 2D structures or patterning can be achieved also by other methods. They are based on the introduction of defects on which the reduction reaction preferentially proceeds [16] and on the physical confinement of reaction sites. The defects can be introduced mechanically, physically, and also electrochemically. Nanoindentation technique with an atomic force microscope (AFM) has been applied to draw palladium lines with sub 100 nm widths on an oxidized silicon surface by electrodeposition [17] and a copper
dot array of 50 nm in the feature size on a hydrogen-terminated silicon surface by displacement deposition [18]. The size of the latter was then optimized and 25–30 nm was achieved along with showing the applicability of gold, silver and cobalt [*19]. The size depends on the type of metal. Meanwhile, Trimmer et al. have shown the possibility of an electrochemical method to introduce localized defects [*20]. A short anodic pulse (100 ns) is applied to highly doped ptype silicon using a tungsten tool electrode with a sharp tip in HF solution, and thereby a local area on the silicon surface is selectively electropolished [21]. The required pulse width depends on the dopant concentration and a shorter pulse is required for specimens with lower dopant levels. This procedure introduces localized defects and the following electrodeposition produces isolated copper islands with 500 nm diameter. The surface after the selective etching gives a variety of surface states such as damaged, flatly etched, or porosified surfaces. Keeping the etched silicon in HF solution dissolves the porous and porous-to-electropolished transition surfaces and the treatment gives uniform copper deposition. If the deposition sites are limited physically, we can achieve position-selective deposition of metal. The electrochemical scanning capillary microscope technique can be applied to the local deposition of metals [*22,23]. A capillary containing electrolyte and reference and counter electrodes is electrochemically connected with a substrate (gold or n-type silicon) through electrolyte in the capillary. A copper line of 200 nm width has been achieved using a capillary with an inner diameter of 150 nm. The small opening of the capillary may limit the high drawing rate or current due to the diffusion limitation. A similar micro-contact of electrolyte was used for partial occlusion via holes with silver electrodeposits [24]. Here a droplet of electrolyte itself without a capillary is used, and the feature size of the remaining hole is greater than that obtained by the capillary method, in the order of micrometers. Natural lithography using porous alumina and colloidal crystals can avoid the complexity of the control in the capillary handling. Ono et al. use an aluminum-sputtered p-type silicon as the substrate [*25]. It is well known that pores in
166
Y.H. Ogata et al. / Current Opinion in Solid State and Materials Science 10 (2006) 163–172
anodically formed porous alumina are well ordered [9]. Anodization of silicon occurs when the porosification of aluminum reaches the interface of silicon. Dissolution of silicon oxides formed at the bottoms of alumina pores exposes a bare silicon surface. They obtained ordered copper dot arrays of 80–90 nm by the subsequent displacement deposition. They also utilized polystyrene beads as selfassembled mask. Dot arrays of copper and silver with 50–100 nm size were obtained [*25,26]. The utilization of a composite substrate has been performed by Chou et al. [*27]. They did not use empty pores but utilized hydrophilic channels of Nafion. Nafion consists of a hydrophobic resin body and hydrophilic channels with ion exchange groups. Displacement deposition of gold and electrodeposition of silver were demonstrated on a Nafion-coated silicon surface. The size is smaller than 100 nm. The deposits distribute randomly but their purpose of the catalyst introduction for fuel cells does not require well-assembled alignment. Microdeposition of metals on silicon is meaningful in its applications regardless of whether it is aligned randomly or orderly. Random deposition is utilized for catalyst, and ordered deposition is utilized for optical devices and discrete magnetic recording media. Preparation of highly ordered dot array alignment without conventional photolithographic technique is a challenging subject. 3.2. Structure formation The extension of patterning to structure formation in metal deposition on silicon is important to apply the electrochemical method to fabricate micro- and nano-structures. Silicon is readily dissolved by alkaline solution, and it is easily oxidized and the oxides are readily dissolved in HF solution. Furthermore, anodization of silicon produces a variety of porous structures (porous silicon) with the pore size ranging from a few nanometers to a few tens of micrometers [4,28]. Metal deposition in these structures and the successive silicon dissolution or the other treatments have been utilized to form 3D structures. Many interesting results have been obtained. Zhang et al. [**29,30] prepared macroporous silicon of 1 lm diameter using a p-type silicon wafer of 10 X cm and then dipped in a NiSO4 solution containing NH4F. The size of silicon pore walls is reduced by repetition of oxidation and etching in HF solution. Displacement reaction of nickel occurs producing silicon oxides, and the oxides are dissolved in the presence of HF. They finally observed the total substitution of nickel for silicon and obtained free-standing nickel pillars of about 200 nm size as one-step process. The complete conversion to nickel does not seem plausible because the displacement deposition of nickel should proceed in the presence of oxidizable species, silicon here; however, they observed the complete conversion by energy dispersive X-ray spectroscopy (EDX). Macroporous silicon prepared from a flat silicon wafer gives random pores. Meanwhile well-ordered pore
arrays are achieved by pre-patterning with photolithography. Sato et al. demonstrated the formation of well-aligned nickel microneedles using a pre-patterned n-type silicon [*31]. It is noted that not only noble metals but also lessnoble metals can be deposited without the assistance of illumination on n-type silicon. Therefore, we find many works on metal deposition using n-type silicon compared to those using p-type silicon. They developed the single solution process: macropores are formed under illumination and then cathodic polarization follows to deposit nickel into the macropores in an ethanoic HF solution containing Ni(OSO2NH2)2. Kobayashi et al. [*32,33] started from a pre-patterned p-type silicon wafer by photolithography. The use of p-type silicon allows the use of photoexcitation effects. The diameter of macropores was 4 lm. They electrodeposited copper in the dark and nickel under illumination; in their case the simple acidic solutions were used without fluorides. Copper deposition can proceed by hole injection via the valence band along with charge transfer via the conduction band, whereas nickel deposition proceeds only by the conduction band process because of its less-noble nature, and hence illumination is indispensable for the deposition. The different situations cause preferential growth from the pore bottom in copper deposition and uniform growth on the entire pore wall in nickel deposition resulting in well-aligned copper microrod arrays and nickel microtube arrays, respectively (Fig. 3). They also found that the pore filling properties are different in noble metals; for example, the pore filling with gold is difficult because of inhomogeneous deposition inside the pores. Nickel nanowires of 70 nm diameter were demonstrated using anodically prepared porous alumina from an aluminum E-beam-deposited silicon substrate [34]. To extend the method to small pores down to tens of nanometers in diameter, further understandings of deposition mode depending upon the type of metal, mass transfer and infiltration of electrolyte inside minute pores are essential. Micro- or nano-structures consisting of silicon itself are also valuable. Galvanic metal deposition is utilized to prepare the structures. Metal-catalyzed chemical vapor deposition (CVD) of silicon is a famous technique to prepare silicon nanowires. The method utilizes difference in liquidus temperatures of silicon and silicon/gold alloy. Yesseri et al. [*35,36] added gold catalysts on p-type silicon using displacement deposition, and the following CVD with a mixture gas of SiH4, HCl and H2 produced silicon nanowires. The oxidation state is an important factor to form the well-aligned nanowire arrays. They also demonstrated the application of this nanowire structure to surface enhanced Raman spectroscopy after silver nanocrystal deposition on silicon nanowires. Peng et al. also tried to prepare silicon nanowires utilizing galvanic metal deposition. Their method utilizes the local cell mechanism instead of the above vapor–liquid–solid growth mechanism. P-type silicon is immersed in 10 mM AgNO3 + 4.6 M HF solution at 50 °C. Silver displacement deposition takes place leading to dendrite formation, by which silicon oxidation and the
Y.H. Ogata et al. / Current Opinion in Solid State and Materials Science 10 (2006) 163–172
167
Fig. 3. Pore filling with metal inside the macropores of p-type silicon. (a) Metal grows from the pore bottom in the dark when the valence band process is possible (noble metals). Preferential growth at the bottom results mainly from the current distribution. Microrods are obtained after dissolving the silicon substrate. (b) Electrodeposition does not take place for less-noble metals in the dark. Illumination generates electrons over the wall area and enables the reduction reaction on the wall leading to the formation of a microtube structure.
dissolution are accompanied. Rinsing with copious water detaches the deposits leaving silicon nanowires of a few nm diameter [*37]. Later they separated the process into electrodeposition of metal and metal-catalyzed etching, and investigated the mechanism [**38,39]. After electrodepositing silver nanoparticles, the silicon is immersed in 10 mM Fe(NO3)3 + 4.6 M HF solution. Deposited silver particles catalyze the reduction of ferric ions and function as local cathode, where electrons to be consumed are supplied by the oxidation reaction of silicon. Bare silicon just beneath the silver particles serves as an effective local anode. HF dissolves the formed oxides and finally leaves silicon nanostructures (Fig. 4). The behavior is independent of the type of dopant. The use of Si(1 0 0) results in nanowires and Si(1 1 1) nanoribbons. This process is based on local dissolution of single-crystalline silicon, and the resultant nanostructures are composed of single crystals. Silver and gold to a lesser extent are good metals for the nanostructure formation, whereas platinum and copper do not give satisfactory results. The silicon nanowires show low reflectivity and they applied this surface to an antireflection layer of solar cells [40]. Boring behavior into silicon by metal-catalyzed local etching was also observed by Tsujino and Matsumura [**41]. When a silicon substrate, on which silver has been deposited by a displacement reaction, is immersed in a 10% HF + 30% H2O2 (10:1 v/v) solution at room temperature, a microporous layer is formed and straight nanoholes
(50 nm in diameter) develop deep (500 lm) under the layer in p-type Si(1 0 0). Similar results were obtained also in n-type silicon. Besides, palladium deposition does not make straight holes and there is no nanohole formation for platinum and copper deposition. This is also caused by the metal-catalyzed etching of silicon. They concluded that the different behavior of metals was attributed to the different catalytic activity for the reduction of H2O2. They also observed inclined boring in Si(1 1 1). They used catalytic etching for the formation of antireflection surface for solar cells [42]. Later they optimized the conditions for the platinum catalyst and observed cone-shaped concavities filled with microporous silicon body when HF concentration was low and helical nanocoils of 100 nm diameter at high HF concentration [*43]. The varieties in the structures are brought about by the balance of local oxidation of silicon and the dissolution with HF. The detailed mechanism should be studied further. The fine structure formation is an advantage in that it proceeds spontaneously. The structures are utilized as a building block for creating useful devices. So far, most of the studies used the chemical characteristics of silicon itself, but the characteristics as semiconductor such as the photoeffects do not seem to be efficiently employed. Well-ordered assembly of individual structures is also an issue to be studied. It is an exciting topic showing a promising future, and development of further studies is expected.
168
Y.H. Ogata et al. / Current Opinion in Solid State and Materials Science 10 (2006) 163–172
Fig. 4. Selective growth of pores. (a) Deposition of metal particles on silicon. (b) Reduction reaction is catalyzed on the particles. Counter reactions accompanying silicon oxidation proceed beneath the particles. (c) Formed oxides are dissolved by HF. (d) Metal particles sink into the silicon substrate. (e) The continuation bores deep holes into silicon leaving nanowires structure.
4. Deposition by displacement reaction In the foregoing sections many researchers utilized displacement metal deposition. The deposition on silicon by displacement reaction has been variously called open-circuit deposition, electroless deposition, spontaneous deposition, galvanic displacement deposition and immersion deposition. In this review ‘‘displacement deposition’’ has been used since we would like to avoid using the most-frequently used electroless deposition. We would like to use it to represent the deposition from solution containing reducing agents that is usually called electroless deposition. Silicon itself serves as the reducing agent in the displacement deposition on silicon. We have used ‘‘immersion plating (deposition)’’ in our works. Displacement metal deposition is the easiest process and suitable for microdeposition that is necessary for the aforementioned micro- or nano-patterning and structure formation when dilute solution is used. The process avoids a drawback originated from inhomogeneous current distribution that is often encountered in electrodeposition. The displacement deposition has been applied to platinum catalyst preparation in a membrane electrode assembly for a miniature fuel cell consisting of a silicon frame [44] and a CoBiFeB soft magnetic underlayer on silicon for perpendicular magnetic recording media [45]. The electrochemistry of displacement deposition is basically simple. The main reactions consist of reduction of metal ions and oxidation of silicon Menþ þ ne ! Me Si þ 2H2 O ! SiO2 þ 4Hþ þ 4e
ð1Þ ð2Þ
SiHx þ 2H2 O ! SiO2 þ ð4 þ xÞHþ þ ð4 þ xÞe
ð3Þ
Silicon is oxidized (Reaction (2)) and the resultant electrons are consumed for the reduction of metal ions (Reac-
tion (1)). The standard potential of Reaction (2), E02 , is estimated as 0.89 V vs. SHE. The equilibrium potential E2 depends on pH but it always shows very negative potential at any pH conditions. Silicon is usually pretreated with HF solution, and thereby the surface is hydrogen-terminated. The potential (Reaction (3), E03 Þ does not seem very different from that of Reaction (2). Displacement deposition of metal is thermodynamically possible in a system in which equilibrium potential E1 is more positive than E2. Table 1 shows the standard electrode potentials for some relating systems. Judging from the negative potential of the Si/SiO2, many metals can be deposited on silicon. However, it is usually not the case in acidic solution; only noble metals can be deposited [47]. Reactions other than metal deposition compete with the deposition reaction. Hydrogen evolution (Reaction (4)) is the typical reaction 2Hþ + 2e ! H2
ð4Þ
Less-noble metals have more negative potential than the E4, and hydrogen evolution reaction disturbs the metal Table 1 Standard electrode potentials of some metal systems in aqueous solution [46] Electrode O2/H2O Pt2+/Pt AuCl 4 =Au Pd2+/Pd Ag+/Ag PtCl2 4 =Pt PdCl2 4 =Pd Cu2+/Cu AgCl/Ag H+/H2 Fe3+/Fe Pb2+/Pb
E0, V vs. SHE 1.229 1.188 1.002 0.915 0.799 0.758 0.60 0.340 0.222 0.000 0.04 0.126
Electrode 2+
Sn /Sn Ni2+/Ni Co2+/Co In3+/In Cd2+/Cd Fe2+/Fe Ga3+/Ga Cr3+/Cr Zn2+/Zn SiO2/Si Mn2+/Mn SiF2 6 =Si
E0, V vs. SHE 0.136 0.257 0.277 0.338 0.403 0.44 0.529 0.74 0.763 0.888 1.18 1.37
Y.H. Ogata et al. / Current Opinion in Solid State and Materials Science 10 (2006) 163–172
deposition. Increase of pH decreases E4 and extends the possibility of deposition of less-noble metals. Reduction of dissolved oxygen is another major cause to hinder metal deposition [48]. The reactions during displacement deposition can proceed through the general corrosion type or the local-cellcorrosion type. Metal deposition and oxidation of silicon take place at the same site in general-corrosion, while the cathodic and anodic sites are separated from each other in local-cell-corrosion. Both processes can be possible but the local-corrosion type process is predominant judging from the formation of developed silver dendrites [*37] and the growth of copper deposit layer as thick as 380 nm (averaged as a uniform layer) [47]. Fig. 5 schematically shows the progress of displacement reaction on silicon. Electrons produced by silicon oxidation migrate in silicon and also in deposited metal and participate in the reduction of metal ions at some sites. Once the surface is completely covered with silicon oxides, no reactant exposing to electrolyte is present and metal deposition ceases for lack of electron supply or available sites for the counter reaction. The oxide formation determines the deposition morphology and distribution. The oxidation of silicon was investigated during copper deposition on porous silicon, and it has been shown that homogeneous oxidation is dominating in the early stages of displacement reaction and then inhomogeneous oxidation becomes significant [49]. The presence of HF and related species causes dissolution of the oxides and affects the metal deposition states greatly [50,51], and are effectively utilized for nanostructure formation as we have already mentioned the examples in the previous sections. Transition of the surface conditions can be monitored by the temporal variation of open-circuit potential (OCP) during displacement deposition (Fig. 6) [48,**52,53]. Niwa et al. investigated the OCP variation during immersion of n-type silicon (8–12 X cm) in 0.1 M NiSO4 + 0.3 M (NH4)2SO4 solution (pH 9.0) [**52]. They explained it based on the results obtained by in situ attenuated total reflection Fourier transform infrared spectroscopy (ATR-
Fig. 5. Schematic illustration of displacement deposition. Reactions during displacement deposition can proceed through the general-corrosion type (a) or the local-cell-corrosion type (b). Metal deposition and silicon oxidation take place at the same site according to the general-corrosion mechanism. Meanwhile, the cathodic and anodic sites are separated from each other according to the local-cell-corrosion mechanism. The electrons produced by silicon oxidation migrate in silicon and also inside the deposited metal resulting in metal growth on deposits and even at the tips of dendrites.
169
Fig. 6. OCP of p-type silicon (porous silicon) vs. time curves during displacement deposition of copper in 10 mM CuSO4 acidic solution. (a) Measured OCP of a copper electrode. (b) OCP of a silicon electrode. The potential shifts in the negative direction, stays for a while, then increases, and finally approaches a steady potential close to the Cu2+/Cu equilibrium potential (a). (c) OCP of a silicon electrode in a solution with 50 mM HF. Oscillation of the potential starts after induction time and lasts longer than several hours.
FTIR). A negative shift of OCP (i) caused by the initially formed silicon suboxide layer is followed by nucleation of nickel with a sudden recovery of the OCP (ii), then a gradual increase of the OCP in the positive direction (iii) where the growth of deposits becomes dominant over the nucleation, and finally the increase and attainment of a steady OCP (iv) that corresponds to the equilibrium potential of Ni2+/Ni. Nickel deposition terminates at stage (iv). A more noble metal, copper, shows a short time for stage (iii) [48]. A thermodynamic model of OCP during electroless deposition (including reducing agent in solution) of nickel on silicon was also proposed [54,55]. The presence of HF in the solution induces oscillation of OCP (Fig. 6c) [48,**56,57]. Potential oscillation of a silicon electrode under highly anodic polarization is well known, but the OCP oscillation was observed for the first time. When p-type silicon or the porous surface is immersed in 10 mM CuSO4 solution (pH 2.7) containing 6–300 mM HF, the OCP oscillates. The oscillation does not occur in the solutions with the higher and lower HF concentrations and also on highly doped silicon [**56]. The behavior is explained as follows. The OCP is a mixed potential during silicon oxidation and copper deposition. When the silicon surface is oxidized as counter reaction of the reduction of cupric ions, the area of electrochemically active bare surface decreases and the OCP shifts in the positive direction. The oxide layer is inhomogeneous and the local breakdown by HF exposes a part of the active surface to the electrolyte, resulting in the recovery of the OCP. The breakdown-and-repair continues and shows a sustained oscillation. The mechanism should be further investigated: why is the local event harmonized showing a regular oscillation or why does the phenomenon depend on the type and concentration of dopant?
170
Y.H. Ogata et al. / Current Opinion in Solid State and Materials Science 10 (2006) 163–172
OCP measurements are important to understand the displacement reaction. The OCP is a mixed potential and is determined by anodic and cathodic reactions participating in the reaction. Among these reactions the potential contribution from silicon is somewhat unclear. The standard potential of silicon oxidation (Reaction (2)) is not definite, for example, and the surface condition changes during deposition (bare, partially or completely oxidized, covered with deposits, or others) and the changes cause the shift in OCP. Understanding the complete picture at the surface during displacement deposition is a subject to be further studied. 5. Control of deposition The demand for micro- and nano-structure formation by metal electrodeposition requires understanding of nucleation and growth in the early stages of deposition. The Scharifker-Hills model is widely used for the 3D multiple nucleation with diffusion controlled growth [58]. Hemispherical diffusion takes place toward randomly distributed nuclei. The diffusion zones develop, overlap, and finally the development leads to the formation of a planar diffusion zone (Fig. 7). The cases of the instantaneous and progressive nucleation have been analyzed. The model fits well with metal deposition on silicon; however, calculated nucleus density based on the current transient according to the model usually deviates from experimentally obtained density by the direct observation with a scanning electron microscope (SEM), for example. The experimentally determined density with SEM is greater than the density according to the model fitting by a few orders. Radisic et al. [**59,60] explained the deviation based on the results obtained with their in situ transmission electron micro-
scope (TEM) observation and data processing [**61] where they dealt with copper electrodeposition on gold instead of silicon. They performed the first direct observation of the nucleation-and-growth process in the early stages of metal deposition and proposed the participation of ad-atom formation and surface diffusion. The additional process can explain the deviation. The concurrent surface model must be verified further but their model shows us the importance of the events taking place outside of the hemispherical diffusion zones. The stochastic simulation of the kinetically controlled metal deposition was also conducted [62]. Mun˜oz and Staikov investigated the influence of oxidation of p-type silicon in cobalt electrodeposition under illumination [*63]. The current-time behavior received a detailed analysis with the AFM observation and capacitance measurements, and they showed that the presence of an oxide layer introduces a barrier for charge transfer leading to grain-like cobalt electrodeposits and the defects in the oxides act as cluster nucleation sites for the deposits. A double-potential step technique is sometimes utilized to control the distribution and size of electrodeposits. The first potential step of high overpotential controls the nucleation or the density of deposits, and the successive potential step of low overpotential increases the size without further nucleation. Kawamura et al. [*64,65] demonstrated that similar control was possible under polarization at a constant potential by changing the illumination conditions on a p-type silicon surface. They succeeded in controlling the deposition in platinum and copper, but the control failed in palladium and gold deposition and they ascribed the different behavior to their fast rates of the displacement reaction. Composition control of metal deposits is important and has been long studied. Alloy deposition of compound semiconductors [66,67], magnetic materials for new recording media [*68], and silicide formation at the interface between silicon and deposited iron [69] have been studied. Further studies are expected for the fine structure control on deposit-substrate interaction, microdiffusion of reactants, and growth of deposits. In addition, understanding of the band structure during electrodeposition is indispensable for metal deposition on silicon. For example, metal deposition takes place both on p-type and n-type silicon surfaces [70], but the comprehensive picture has not been given. 6. Conclusions
Fig. 7. Development of diffusion zones around metal deposits. (a) Hemispherical diffusion zones supply metal ions toward the metal islands. Radisic et al. [**59] propose a possibility of metal deposition at the exclusion surface along with the deposition through the diffusion zones. (b) The diffusion zones develop and overlap each other. (c) The further development approaches a linear diffusion condition.
We reviewed recent progress on electrochemical metal deposition on silicon focusing on the micro- and nanostructure control. The works show promising possibilities of natural lithography for the fine patterning and structure formation. Further studies in the field will be extensively pursued because the galvanic processes have a competitive edge over the physical processes due to the cost-effective operation and facilities. In addition to these efforts, understanding of the deposition mechanism must be further
Y.H. Ogata et al. / Current Opinion in Solid State and Materials Science 10 (2006) 163–172
advanced. Fine control of metal deposits on silicon should be based on comprehensive understandings of the interfacial energy structure along with oxidation conditions of silicon and mass transfer mentioned above. Acknowledgements The authors thank T. Sakka, K. Fukami, Y.L. Kawamura and T. Matsumoto for their discussion and help during the preparation of the manuscript, and acknowledge financial support by a grant-in-aid from the Japan Society for the Promotion of Science under Grant No. 18350107. References The papers of particular interest have been highlighted as: * of special interest; ** of very special interest. [1] Andrecacos PC, Uzoh C, Dukovic JO, Horkans J, Deligianni H. Damascene copper electroplating for chip interconnections. IBM J Res Dev 1998;42:567–74. [2] Hyde ME, Compton RG. A review of the analysis of multiple nucleation with diffusion controlled growth. J Electroanal Chem 2003;549:1–12. [3] Oskam J, Long JG, Natarajan A, Searson PC. Electrochemical deposition of metals onto silicon. J Phys D 1998;31:1927–49. [4] Zhang XG. Electrochemistry of silicon and its oxide. New York: Kluwer Academic/Plenum Publishers; 2001 [chapter 6]. [5] Huang Q, Deligianni H, Romankiw LT. Electrodeposition of gold on silicon nucleation and growth phenomena. J Electrochem Soc 2006;153:C332–6. [6] Budevski E, Staikov G, Lorenz WJ. Electrochemical phase formation and growth. Weinheim: VCH; 1996. p. 294–7. [7] Whitney T, Searson PC, Jiang JS, Chien CL. Fabrication and magnetic properties of arrays of metallic nanowires. Science 1993;261:1316–9. [8] Martin CR. Nanomaterials: a membrane-based synthetic approach. Science 1994;266:1961–6. [9] Masuda H, Fukuda K. Ordered metal nanoholes arrays made by a two-step replication of honeycomb structures of anodic alumina. Science 1995;268:1446–68. [*10] Scheck C, Liu YK, Evans P, Schad R, Bowers A, Zangari G, et al. Photoinduced electrochemical deposition of Cu on p-type Si substrates. Phys Rev B 2004;69:035334. [11] Sasano J, Schmuki P, Sakka T, Ogata HY. Laser-assisted maskless Cu patterning on porous silicon. Electrochem Solid-State Lett 2004;7:G98–G101. [12] Ji R, Lee W, Scholz R, Go¨sele U, Nielsch K. Templated fabrication of nanowire and nanoring arrays based on interference lithography and electrochemical deposition. Adv Mater 2006;18:2593–6. [*13] Koynov S, Brandt MS, Stutzmann M. Metal-induced seeding of macropore arrays in silicon. Adv Mater 2006;18:633–6. [14] Sasano J, Schmuki P, Sakka T, Ogata YH. Maskless patterning of various kinds of metals onto porous silicon. Phys Status Solidi A 2005;202:1571–5. [15] Kurokawa A, Kawamura YL, Sakka T, Ogata YH. Different deposition behavior of copper on silicon and porous silicon surfaces during immersion plating. Phys Status Solidi C, in press. [16] Schmuki P, Erickson LE. Selective high-resolution electrodeposition on semiconductor defect patterns. Phys Rev Lett 2000;85: 2985–8.
171
[17] Santinacci L, Djenizian T, Hildebrand H, Ecoffey S, Mokdad H, Campanella T, et al. Selective palladium electrochemical deposition onto AFM-scratched silicon substrates. Electrochim Acta 2003;48:3123–30. [18] Homma T, Kubo N, Osaka T. Maskless and electroless fabrication of patterned metal nanostructures on silicon wafers by controlling local surface activities. Electrochim Acta 2003;48:3115–22. [*19] Kubo N, Homma T, Hondo Y, Osaka T. Fabrication of patterned nanostructures with various metal species on Si wafer surfaces by maskless and electroless process. Electrochim Acta 2005;51:834–7. [*20] Trimmer AL, Maurer JJ, Schuster R, Zangari G, Hudson JL. Allelectrochemical synthesis of submicrometer Cu structures on electrochemically machined p-Si substrates. Chem Mater 2005; 17:6755–60. [21] Allongue P, Kirchner JV, Trimmer AL, Schuster R. Electrochemical micromachining of p-type silicon. J Phys Chem B 2004; 108:14434–9. [*22] Staemmler L, Suter T, Bo¨hni H. Nanolithography by means of an electrochemical scanning capillary microscope. J Electrochem Soc 2004;151:G734–9. [23] Staemmler L, Suter T, Bo¨hni H. Glass capillaries as a tool in nanoelectrochemical deposition. Electrochem Solid-State Lett 2002;5:C61–3. [24] Polk BJ, Bernard M, Kasianowicz JJ, Misakian M, Gaitan M. Microelectroplating silver on sharp edges toward the fabrication of solid-state nanopores. J Electrochem Soc 2004;151:C559–66. [*25] Ono S, Oide A, Asoh H. Nanopatterning of silicon with use of selforganized porous alumina and colloidal crystals as mask. Electrochim Acta 2007;52:2898–904. [26] Asoh H, Arai F, Ono S. Site-selective chemical etching of silicon using patterned silver catalyst. Electrochem Commun 2007;9: 535–9. [*27] Chou J, McFarland EW, Metiu H. Electrolithographic investigations of the hydrophilic channels in Nafion membranes. J Phys Chem B 2005;109:3252–6. [28] Lehmann V. Electrochemistry of silicon. Weinheim: Wiley-VCH; 2002. [**29] Zhang X, Tu KN, Xie YH, Tung CH. High aspect ratio nickel structures fabricated by electrochemical replication of hydrofluoric acid etched silicon. Electrochem Solid-State Lett 2006;9:C150–2. [30] Zhang X, Tu K-N, Xie Y-H, Tung C-H, Xu S. Single-step fabrication of nickel films with arrayed macropores and nanostructured skeletons. Adv Mater 2006;18:1905–9. [*31] Sato H, Homma T, Mori K, Osaka T, Shoji S. Electrochemical formation process of Si macropore and metal filling for high aspect ratio metal microstructure using single electrolyte system. Electrochemistry 2005;73:275–8. [*32] Kobayashi K, Harraz FA, Izuo S, Sakka T, Ogata YH. Microrod and microtube formation by electrodeposition of metal into ordered macropores prepared in p-type silicon. J Electrochem Soc 2006;153:C218–22. [33] Harraz FA, Kamada K, Kobayashi K, Sakka T, Ogata YH. Random macropore formation in p-type silicon in HF-containing organic solutions. J Electrochem Soc 2005;152:C213–20. [34] Wu MT, Leu IC, Yen JH, Hon MH. Preparation of Ni nanodot and nanowire arrays using porous alumina on silicon as a template without a conductive interlayer. Electrochem Solid-State Lett 2004;7:C61–3. [*35] Yesseri AA, Sharma S, Jung GY, Kamins TI. Electroless deposition of Au nanocrystals on Si(1 1 1) surfaces as catalysts for epitaxial growth of Si nanowires. Electrochem Solid-State Lett 2006;9:C185–8. [36] Yesseri AA, Sharma S, Kamins TI, Li Z, Stanley Williams R. Growth and use of metal nanocrystal assemblies on high-density silicon nanowires formed by chemical vapor deposition. Appl Phys A 2006;82:659–64.
172
Y.H. Ogata et al. / Current Opinion in Solid State and Materials Science 10 (2006) 163–172
[*37] Peng K, Zhu J. Morphological selection of electroless metal deposits on silicon in aqueous fluoride solution. Electrochim Acta 2004;49:2563–8. [**38] Peng K, Hu J, Yan Y, Wu Y, Fang H, Xu Y, et al. Fabrication of single-crystalline silicon nanowires by scratching a silicon surface with catalytic metal particles. Adv Funct Mater 2006;16:387–94. [39] Peng K, Wu Y, Fang H, Zhang X, Xu Y, Zhu J. Uniform, axialorientation of one-dimensional single-crystal silicon nanostructure arrays. Angew Chem Int Ed 2005;44:2737–42. [40] Peng K, Xu Y, Wu Y, Yan Y, Lee S-T, Zhu J. Aligned singlecrystalline Si nanowire arrays for photovoltaic application. Small 2005;1:1062–7. [**41] Tsujino K, Matsumura M. Boring deep cylindrical nanoholes in silicon using silver nanoparticles as a catalyst. Adv Mater 2005;17:1045–7. [42] Tsujino K, Matsumura M, Nishimoto Y. Texturization of multicrystalline silicon wafers for solar cells by chemical treatment using metallic catalyst. Sol Energy Mater Sol Cells 2006;90:100–10. [*43] Tsujino K, Matsumura M. Helical nanoholes bored in silicon by wet chemical etching using platinum nanoparticles as catalyst. Electrochem Solid-State Lett 2005;8:C193–5. [44] Brito-Neto JGA, Shintaro Araki S, Hayase M. Synthesis and characterization of porous platinum layers deposited on highly doped n-type porous silicon by immersion plating. J Electrochem Soc 2006;153:C741–6. [45] Osaka T, Asahi T, Yokoshima T, Kawaji J. Electroless-deposited soft magnetic underlayer on silicon disk substrate for doublelayered perpendicular magnetic recording media. J Magn Magn Mater 2005;287:292–7. [46] Bard AJ, Parsons R, Jordan J, editors. Standard potentials in aqueous solution. New York: Marcel Dekker; 1985. [47] Harraz FA, Tsuboi T, Sasano J, Sakka T, Ogata YH. Metal deposition onto a porous silicon layer by immersion plating from aqueous and nonaqueous solutions. J Electrochem Soc 2002;149:C456–63. [48] Sasano J, Murota R, Yamauchi Y, Sakka T, Ogata YH. Redissolution of copper onto porous silicon in immersion plating. J Electroanal Chem 2003;559:125–30. [49] Hamm D, Sakka T, Ogata YH. Immersion plating of copper onto porous silicon with different thickness. Electrochim Acta 2004;49: 4949–55. [50] Oh Y-J, Cho SM, Chung C-H. An in situ ATR-FTIR study on palladium displacement reaction on hydrogen-terminated silicon surface. J Electrochem Soc 2005;152:C348–55. [51] Harraz FA, Sasano J, Sakka T, Ogata YH. Different behavior in immersion plating of nickel onto porous silicon from acidic and alkaline fluoride media. J Electrochem Soc 2003;150:C277–84. [**52] Niwa D, Homma T, Osaka T. Deposition mechanism of Ni on Si(1 0 0) surfaces in aqueous alkaline solution. J Phys Chem B 2004;108:9900–4. [53] Niwa D, Homma T, Mori K, Osaka T. Deposition behavior of Ni on Si(1 0 0) surfaces in an aqueous alkaline solution. J Electrochem Soc 2005;152:C54–9.
[54] Chang YL, Ye WC, Ma CL, Wang CM. The Thermodynamic model of open-circuit potential for electroless deposition of Ni on silicon. J Electrochem Soc 2006;153:C677–82. [55] Tong H, Wang C. Electroless deposition of Ag and Pd-activated TiN/p-Si(1 0 0) substrate. Appl Phys A 2005;81:137–42. [**56] Ogata YH, Sasano J, Itoh T, Sakka T, Rayo´n E, Pastor E, et al. Electrochemical oscillation of open circuit potential during immersion plating of copper on silicon. J Electrochem Soc 2005; 152:C537–41. [57] Nagai T, Nakanishi S, Mukouyama Y, Ogata YH, Nakato Y. Periodic and chaotic oscillation of the electrochemical potential of p-Si in contact with an aqueous (CuSO4 + HF) solution, caused by electroless Cu deposition. Chaos 2006;16:037106. [58] Scharifker B, Hills G. Theoretical and experimental studies of multiple nucleation. Electrochim Acta 1983;28:879–89. [**59] Radisic A, Ross FM, Searson PC. In situ study of the growth of individual island electrodeposition of copper. J Phys Chem B 2006;110:7862–8. [60] Radisic A, Vereecken PM, Hannon JB, Searson PC, Ross FM. Quantifying electrochemical nucleation and growth of nanoscale clusters using real-time kinetic data. Nano Lett 2006;6:238–42. [**61] Williamson MJ, Tromp RM, Vereecken PM, Hull R, Ross FM. Dynamic microscopy of nanoscale cluster growth at the solid– liquid interface. Nat Mater 2003;2:532–6. [62] Drews TO, Radisic A, Erlebacher J, Braatz RD, Searson PC, Alkire RC. Stochastic simulation of the early stages of kinetically limited electrodeposition. J Electrochem Soc 2006;153:C434–41. [*63] Mun˜oz AG, Staikov G. Electrodeposition of Co on oxide modified p-Si surfaces. Electrochim Acta 2006;51:2836–44. [*64] Kawamura YL, Sakka T, Ogata YH. Photoassisted control of Pt electrodeposition on p-Type Si. J Electrochem Soc 2005;152: C701–5. [65] Kawamura YL, Sakka T, Ogata YH. Illumination-modulated electrodeposition of various kinds of noble metal on p-type silicon. Electrochemistry 2006;74:544–8. [66] Go´mez H, Henrı´quez R, Schrebler R, Co´rdova R, Ramı´rez D, Riveros G, et al. Electrodeposition of CdTe thin films onto nSi(1 0 0): nucleation and growth mechanisms. Electrochim Acta 2005;50:1299–305. [67] Wang C, Wang Q, Chen L, Xu X, Yao Q. Electrodeposition of Sb2Te3 films on Si(1 0 0) and Ag substrates. Electrochem Solid-State Lett 2006;9:C147–9. [*68] Osaka T, Sayama J. A challenge of new materials for next generation’s magnetic recording. Electrochim Acta 2007;52: 2884–90. [69] Zarpellon J, Jurca HF, Klein JJ, Schreiner WH, Mattoso N, Mosca DH. Iron silicide formation from Fe thin-film electrodeposition on hydrogen-terminated Si(1 1 1). J Electrochem Soc 2005;152:C808–12. [70] Harraz FA, Sakka T, Ogata YH. A comparative electrochemical study of iron deposition onto n- and p-type porous silicon prepared from lightly doped substrates. Electrochim Acta 2005;50:5340–8.