Electrochimica Acta 51 (2005) 775786

Principles of electrochemical nanotechnology and their application for materials and systems
J.W. Schultze a, , A. Heidelberg a , C. Rosenkranz a , T. Sch apers b , G. Staikov b,
b a AGEF e.V.-Institut an der Heinrich-Heine-Universit at D usseldorf, D-40225 D usseldorf, Germany Institut f ur Schichten und Grenz achen (ISG) and cni Centre of Nanoelectronic Systems for Information Technology, Forschungszentrum J ulich, D-52425 J ulich, Germany

Received 21 February 2005; accepted 15 April 2005 Available online 19 September 2005

Abstract Electrochemical nano technology (ENT) is a promising eld growing in strong connection with electrochemical microsystem technology (EMT). Some principles of EMT like characterization and localization of reactions can be transferred to ENT. Some other principles like mechanical motion, quantum effects or microuidics differ and have to be separately discussed. Properties of the nanocell, the smallest electrochemical two-electrode cell, are discussed as example of a special tool of ENT. It is useful for the preparation of nano-scaled metal-insulator-electrolyte-structures (MIE-structures). The role of the EMT number for eld-induced localizations is discussed with the Si/SiO2 -system as an example. Lateral metal-insulator-metal-structures (MIM-structures) are prepared in the nanocell with the system Nb/Nb2 O5 /H2 O as an example. 2005 Published by Elsevier Ltd.
Keywords: Electrochemical nanotechnology; Electrochemical microsystem technology; MIE-structures; MIM-structures

1. Introduction The nanoscopic and molecular structure of the solid/liquid interface has been the center of interest for electrochemists for more than 100 years. Helmholtz developed the rst model, and the process of crystallization was described by Kossel and Stranski in an atomistic scale. First pictures of the solid surface with atomic resolution have been obtained by eld electron microscopy and since 1983, by STM. Now we proceed from the understanding to the designing of new atomistic processes and the structure of the solid/liquid interface [1]. As a consequence, electrochemical nano technology (ENT) is a fast developing eld in a wide range of possible denitions, and the question arising is that how can it be described? It is the aim of this paper to describe the principles of ENT at rst from a general point of view in strong con

nection to electrochemical microsystem technologies (EMT) [13]. Then, special aspects of ENT will be discussed for the so-called nanocell as an example. Predictions of the EMT number for nano-scaled localization of electrochemical reactions will be proved by the example of eld-induced oxidation of silicon and niobium. Finally, an example is given for the preparation of nanostructured lateral MIM systems.

2. Electrochemical nano technology 2.1. Systems ENT is used to describe any idea of single systems, which could be fabricated, but there is no real technology for mass production and no market for the products. On the other hand, there are already some well-established products prepared by electrochemical techniques or applied in electrochemical systems. To get an idea of ENT, we will describe some processes, materials and systems in the nm range (Fig. 1) [513]:

Corresponding author. Tel.: +49 2461 61 2064; fax: +49 2461 61 3907. E-mail address: g.staikov@fz-juelich.de (G. Staikov). Deceased.

0013-4686/$ see front matter 2005 Published by Elsevier Ltd. doi:10.1016/j.electacta.2005.04.073

776

J.W. Schultze et al. / Electrochimica Acta 51 (2005) 775786

Fig. 1. Examples of electrochemical nano technology (ENT) (upper part, from left to right): SEM images of Ta goblets on Al [1] and of free-standing Ni-tubes [5]; TEM image of a PtRu catalyst on carbon applied for fuel cells [8]; square cells in Al2 O3 [12]; (lower part, from left to right): soft magnetic material for hard disk drives [9]; SEM image of Ta capacitors [10]; nanowire of LiMoSe3 contacted by Pt [13]; SEM picture of a pore lled with copper by the Damascene process [11].

(i) Nanoporous systems: the rst nanotechnological process yielding an almost periodic or at least regular nanostructure was invented by Fischer [4]: ELOXAL was the trivial name for electrolytically-oxidized aluminum which is now called porous alumina. It consists of pores with a length of some micrometers and a width of some tens nanometer formed by Al2 O3 . It can be colored, sealed for surface protection or used as template for Ni-tubes [5], metallic and semiconducting nanowires [6] or membranes [7]. (ii) Nanomaterials, e.g. clusters of Pt, Pt/Ru or other noble metals are used in catalytic electrode layers, e.g. the membrane electrode arrays of fuel cells [8]. Other nanosized materials (carbon, Ni(OH)2 ) are used in batteries and as supercapacitors. In reading/writing heads of hard diskdrives, nanosized magnetic materials are applied [9]. (iii) Capacitors, e.g. Ta electrolyte capacitors, are prepared by dened passivation and contact layers of MnO2 or PEDOT with a thickness of some tens nanometer [10]. (iv) In micro- and nanoelectronics, nanosized conducting lines of Cu are prepared by Damascene process [11]. (v) Equipment for electrochemical investigations in the nmrange, like all SXM techniques including nanosized electrodes as well as the electronic equipment. 2.2. Denition of ENT The presented examples show that ENT already has large importance in high-tech and high-chem. industry. By analogy to electrochemical microsystem technologies [14], we can dene ENT by the following principles:

(i) Electrochemistry should be involved in the production process (e.g. with magnetic materials) or in the applied system, e.g. membrane/electrode assembly. (ii) The size of the product or system should be in the nmrange, i.e. <1 m. The distinction of the submicron range 0.1 < x < 1 m is possible but not necessary. (iii) Technology means that a reproducible process of manufacturing should exist. In rst approximation, ENT can be limited to realize technical processes, i.e. mass production lines. In a wider sense, proposals and patents could be included. 2.3. Reactions and processes In ENT, electrochemical processes like structure formation, deposition, dissolution and modication can be considered, which are shown in Table 1, reactions (1)(5). Chemical reactions that consist of combined ion transfer reactions (ITR) like oxide precipitation or dissolution (6) are also applied.

3. Principles transferred from EMT 3.1. Description of ENT by scaling down By developing a new eld, new principles and terms could be dened. Following the technological development, however, the scaling down from macroscopic science via EMT [13] to ENT could be possible within a wide range. To check this approach, some principles of EMT and ENT are summarized in Table 2.

J.W. Schultze et al. / Electrochimica Acta 51 (2005) 775786 Table 1 Reactions used in electrochemical nano technologies (ENT) ion transfer reactions (ITR), electron transfer reactions (ETR) Process i : Metal deposition, i+ : etching, ECM Anodic oxidation Anodic oxidation + h Chemical metal deposition Electrodeposition of paint Chemical dissolution Reaction type ITR ITR ITR induced by laser light Cathodic ITR, anodic ETR Precipitation induced by ETR Combined ITR Reaction formula M M + zH2 O MOz + 2zH+ + 2ze M + z H2 O + h MOz + 2zH+ + 2ze Mz+ + ze M CH2 O + H2 O CO2 + 4H+ + 4e Rx-COO + H+ RxCOOH H+ + e (1/2)H2 SiO2 + 6HF [SiF6 ]2 + 6H+ Mz+ + ze

777

Reaction number (1) (2) (3) (4) (5) (6)

Two groups can be distinguished. The rst group of terms and principles can be equally applied (Section 3): characterization of systems, localization of reactions, combination of materials or application of etching techniques. This will be shown in the next sections. On the other hand, some effects described in Section 4 get a new quality due to the change of dimensions by one to four orders of magnitude: Microuidics [15] is a hot topic in EMT (e.g. for turbulences [16] and microreactors [17]) but not in ENT. In ENT however, molecular motions are playing an important role. The motion of systems has to be changed from mechanically driven systems to piezo driven techniques. Quantum effects clearly appear only in nanoscopic dimensions. 3.2. Characterization Micro- and nanosystems can be classied by same terms [18]: (i) Number of systems: single systems are investigated by research, while multiple systems are used in mass production. (ii) Order of multiple systems: natural systems are usually statistically arranged, while technology prefers periodic systems. (iii) The aspect ratio A = depth/width allows for the separation of at (A < 1) and deep systems (A 1). (iv) The conductivity of materials is important for the potential distribution: metals (M), semiconductors (S), insulators (I) or classes of materials (oxides (O), polymers (Pol)) have to be distinguished.

(v) Active systems require functional, variable material properties, while passive systems can be constructed with simpler, static materials. 3.3. Localization of reactions As shown in Fig. 2, electrochemical reactions described in Table 1 can be localized by (i) geometric blocking, e.g. masks, photoresists; (ii) chemical modication of the surface, e.g. SAMs; (iii) localized signals, e.g. by a focussed laser or focussed electric eld, or; (iv) controlled transport reactions, e.g. in the SECM. These methods well known from EMT [19], can be miniaturized for ENT, with some modications. For example, photoresists have to be developed by electron beams instead of light, or a wetting droplet has to be substituted by a condensed water lm [20]. Implantation can be miniaturized down to single ion implantation [21], and SAMs are now often applied in the nanometer-scale. The SECM has been successfully miniaturized down to tens of nanometer, but the microjet has no analogon in the nanoscale [22]. Some localization principles were especially developed for the nanometer-scale with low and high aspect ratios. They will be discussed in Section 4.2. 3.4. Combination of materials Metals, insulators and semiconductors differ in the distribution of potential. The nomenclature M: metal, S: semicon-

Table 2 Comparison between the similar and different principles used in EMT and ENT EC microtechnology Similar principles Characterization Localization of reactions Combination of materials Etching/sacricial layers Fluidics Cell, layer interface Microscopy Mechanics Polycrystalline Molecular motions Cell, layer interface SXM (e.g. STM, AFM), SEM, TEM Piezo Single crystalline, dened Quantum effects EC nanotechnology

Different principles Charging dimensions Image Drive Substrates Quantization

778

J.W. Schultze et al. / Electrochimica Acta 51 (2005) 775786

Fig. 2. Common principles of EMT and ENT: localization of reactions by geometric blocking, chemical modication, localized signals and controlled transport.

ductor, I: insulator, E: electrolyte is taken from electronics. While metal surfaces have equipotential planes, in semiconductors and insulators, large potential drops can be generated. Therefore, MM systems can be generated by traditional galvanic techniques, while MIE or MIM systems require special measures (see Section 5). Due to the eld distribution, lateral and vertical systems have different principles [19]. 3.5. Negative structures and sacricial layers The formation of three-dimensional structures sometimes require the production of negative structures. The etching of sacricial layers is an example often realized in microstructures. Typical sacricial layers consist of dissolvable metals,

e.g. copper, or oxides which can be etched in HF. These techniques [23,24] can be transferred from EMT to ENT. The inverse reactions ((1), (4) and (6)) can be used for dissolution (Table 1). Fig. 3 gives four examples: (i) An inverse opal was formed by the polymerization of methacrylate containing SiO2 -balls following the dissolution of SiO2 in HF. Such structures can be used as templates for catalysts or optical systems [25]. (ii) A free-standing metal contact can be prepared on top of a sacricial layer which is nally etched [26]. Such systems are known for EMT, e.g. sensors in airbags. (iii) Free-standing Ni-tubes prepared in Al2 O3 were nally isolated by etching the Al2 O3 by HF [5].

Fig. 3. Negative structures and sacricial layers as templates for nanostructures (from left to right): production of an inverse opal; a route to generate freestanding metal contacts; production of Ni-tubes in porous Al2 O3 [5]; formation of nano holes in SiO2 [20].

J.W. Schultze et al. / Electrochimica Acta 51 (2005) 775786

779

(iv) Negative structures were prepared on Si by localized oxidation by AFM and following selective etching [20]. 3.6. Characteristic data and measuring techniques Going from microscopic to nanoscopic dimensions, characteristic data will change by orders of magnitude, and measuring methods will have to be adjusted or will be limited to certain regions of optimum sensitivity. This can be explained best by choosing normalized values, e.g. referred to a radius r0 = 1 m. Data of nanostructures which depend on the area like i, q and C can be used as a typical example for comparison of data in the micro- and nanoscale (Fig. 4). (i) If the current density i (A/cm2 ), the charge density q (C/cm2 ) or the double layer capacity c (F/cm2 ) are constant, the absolute current I (A), charge Q (C) and capacity C (F) will increase with increasing radius, i.e. I = I0 C = C0 r r0 r r0
2

106 cm2 are sufcient in yielding grains with a diameter r > 10 m. For a nanosystem, we need at least one nucleus/structure, i.e. the nucleation density has to be strongly increased: N = N0 r r0
2

(2)

(iii) Equivalent considerations can be carried out for other data and measuring techniques with the general dependence X = X0 r r0
n

(3)

with n ranging from 2 to +3. 4. Different principles of EMT and ENT 4.1. General discussion Some principles of EMT are limited to microscopic dimensions and cannot be transferred to the nanoscale, and some effects appear which are not known in EMT. Examples are summarized in the lower part of Table 2. Mechanical motions represent a well-known example, since the mechanical drive by screws has to be substituted by piezo techniques in the case of ENT. Microuidics [1517] represent a eld where the transfer from macroscopic to microscopic dimensions is not possible. With the processes in the Nernstian diffusion layer and its relation to the Prandtl layer the continuum theory is still limited to the micrometer range [28]. In the micrometer-range, molecular motions can only be considered. On the other hand, the double layer is usually limited to molecular dimensions and to a few nanometers even in diluted solutions. Hence, in EMT the discussion of the double layer is separated as an interface phenomenon. In ENT, however, the dimensions can interfere with those of the structure that are shown for the nanocell as an example. Finally, quantum effects are not known in EMT, but they become important in molecular dimensions. 4.2. Special localization principles used for ENT For the purpose of ENT, some principles of reaction localization were developed which were not yet applied in EMT due to lack of advantages. Therefore, they will be summarized in the following and in Fig. 5. (i) For the special problem of electrocatalysis, a galvanic deposition was developed [29]. Noble metal clusters have to be deposited on carbon particles where they are in contact with the electronically-conducting carbon and with the ionically-conducting electrolyte. Since electrons can be delivered via the contacted carbon particles, and metal ions can be deposited only from the ionically-

,
2

Q = Q0

r r0

and (1)

Since measurements of currents are limited to I > fA and charges Q > fC, the direct measurement of Faradaic currents of structure formation is limited to structures with n > 1019 mol substance. For example, the detection of upd-phenomena is possible for structures with A > 104 nm2 or r > 50 nm. Measurements on single atoms like tunnel currents or measurements in Bards single molecule cell [27] are possible only due to the multiplication by various orders of magnitude. Measurements of the double layer capacity are limited by C > 1014 F, i.e. for electrode areas A > 1 m2 . (ii) Some phenomena depend critically on the available area like nucleation density. For technical systems in the macro- and micro-range, typical densities of

Fig. 4. Double logarithmic plot of various parameters x in dependence on the radius r of the nano/micro system: I, current; C, capacitance, A, area, V, volume, N, minimum nucleaction density.

780

J.W. Schultze et al. / Electrochimica Acta 51 (2005) 775786

Fig. 5. Special localization principles in the nanometer scale (upper part, preparation of nanostructures with low aspect ratio A < 1): deposition of noble metals on carbon for usage in fuel cells; deposition of metal cluster by the method of Schindler et al. [30] and Kolb [31]; (lower part, A > 1): electrochemical deposition of polybithiophene into porous Si [18]; lling of trenches with copper by Damascene process [11]; chemical deposition of Ni-tubes in porous Al2 O3 [5].

conducting phase, the galvanic deposition allows the reaction at the desired places only. This process is especially used in fuel cell technology. (ii) A local enrichment in the electrolyte can be realized by the cathodic accumulation of metal atoms on a conducting tip following anodic dissolution. This local enrichment allows localized cathodic deposition of various metals on metallic and semiconducting substrate. The deposited clusters can be investigated in situ directly after the deposition by STM and STS [30]. (iii) The method by Kolb applies a direct transfer of atoms from the STM tip to the substrate and also results in localized deposition on the substrate [31]. For high aspect ratios, physical methods, e.g. PVD, yield a deposition on top of the structure, but not inside. In electrochemical systems, on the other hand, electronic differences, seed layers or rate determining diffusion processes can be applied: (iv) The band structure of porous Si differs at the bottom of pores, where crystalline c-Si with a band gap of 1.14 eV is found and at the walls, that consist of porous porSi, with a band gap of 23 eV. Therefore, the anodic polymerization inside the holes can start from the pore bottom only, resulting in a very uniform and controlled lling of the pores [21]. (v) Diffusion into the deep trenches is slower than on the well-accessible surface. This principle is applied in the damascene process for Cu-deposition in deep trenches. Inhibitors which are slowly incorporated into copper, are depleted at the pore bottom, but form complete layers at the upper surface [8].

(vi) In the insulating pores of Al2 O3 , chemical Nideposition can be induced by seeding with PdCl2 / SnCl2 . Then, the chemical deposition takes place at the whole inner pore surface, resulting in Ni-tubes [2]. (vii) Special problems arise with the localized electric eld. This will be explained in the next section for the special case of the electrochemical nanocell (see Fig. 6).

Fig. 6. (a) Schematic representation of the local oxidation of Nb in the nanocell. (b) Schematic diagram of the potential drop.

J.W. Schultze et al. / Electrochimica Acta 51 (2005) 775786

781

4.3. The nanocell: the smallest electrochemical cell in the world The electrochemical nanocell was developed by Sugimura and Nakagiri [32] and later described in detail by Bloe et al. [20]. It is schematically shown in Fig. 6. It demonstrates two aspects of special interest which should be discussed. (i) It includes some principles not known from macroscopic cells. The nanocell is a two-electrode cell lled by water condensation under BET conditions. The thickness of the lm is adjusted by the partial pressure of water. Due to the small distance between the electrodes, a Helmholtz layer cannot be formed at the solid/liquid interfaces. Protons and hydroxyl ions recombine immediately. Excess charges remain in the M (or S) phases. At positive polarization, an insulating oxide is formed at the surface of M (e.g. Ti, Nb) or S (e.g. Si). All dimensions are in the range of nanometers: the distance between the two electrodes is less than 1 nm. The adsorbed water lm has a volume between 1016 and 1018 cm3 given by the vapor pressure. The electrolyte volume is so small that the mass balance can be maintained only by a continuous supply from the vapor phase [11]. Due to this necessary mass exchange, it is an open system in contrast to the usual closed systems. Further, the ionic concentration in the neutral water is 107 mol/l, which causes very high resistance. In addition, the double layer is absent, since the theoretical thickness of the Gouy layer exceeds the electrode distance by some orders of magnitude. Finally, the eld distribution is inhomogeneous, given by the radial symmetry of the counter electrode. (ii) The nanocell opens new possibilities for nanostructuring. STM and AFM opened a new world of nanoscopic systems. Recent experiments by Kolb unfolded the nanotechnology for MME systems [31]. At rst, it seems to be surprising that most of the electrochemical, nanotechnological experiments were limited to such systems. First experiments by Sugimura et al. demonstrated, however, that nanostructuring can be carried out in MIE and semiconductor-insulator-electrolyte (SIE) systems too [32]. The reason for the limitation of these experiments is given by the electrochemical limitation of the necessary application of the so-called nanocell. 4.4. The EMT number for localization of the electric eld 4.4.1. Denition of EMT number Various possibilities of the localization of electrochemical reactions have been described in Figs. 2 and 5. One could assume that the localization by the STM tip which was successfully demonstrated for MME-systems could be applied for other systems as well. This, however, is not the case

since the eld distribution depends strongly on the resistances of the electrolyte and the transfer reactions. Macroscopic galvanics have an opposite problem: a homogeneous reaction on the whole surface is desired. The Wagner number Wa = Rpol /d describes the principles in dependence on the conductivity , the resistance of electrode polarization Rpol and distance d between the electrodes under constant conditions. For localization in the nano-range, the analogous EMT number EMT = Rpol Rel (4)

was introduced [19] with Rel : electrolyte resistance. In case of EMT 1 localization can be achieved while for EMT > 1 the structure extends laterally and becomes homogeneous (non-structured) for EMT 1. Fig. 7 shows a correspondent double logarithmic plot. In contrast to the Wagner number, resistances Rpol and Rel in the nanocell depend on the polarization time and the distance x from the counter electrode. As can be seen in Figs. 6 and 10, and in the experimental proles in Fig. 11, the thickness of the Nb oxide increases with time and decreases with x. Therefore, Rpol = Rox varies by orders of magnitude. Moreover, the resistance of the electrolyte lm depends on the mass balance, i.e. reaction time, extension of the product, etc. For MME systems (metalmetal-electrolyte) the ion transfer resistance is usually very small, i.e. Rion < 50 /cm2 (io > 1 mA/cm2 ). Then, electrolyte resistances Rel > 100 /cm2 are sufcient which is realized in diluted solutions c < mmol. In the case of MIE and SIE systems localization of the oxide formation in the range of EMT < 1 can only be achieved at very high electrolyte resistances, because of the high resistances of the formed oxides (e.g. (SiO2 ) = 1014 to 1016 cm). The resistance of pure water is still too low to

Fig. 7. Double logarithmic plot of log Rpol vs. Rel to express the inuence of the EMT number on the localization of maskless nanostructuring. The inset shows the resistances that have to be taken into consideration for the local oxidation in the nanocell.

782

J.W. Schultze et al. / Electrochimica Acta 51 (2005) 775786

meet the condition EMT < 1. For such high values, only the reduction of the cross section can help, which is achieved in monomolecular water adsorption lms in the nanocell [20]. 4.4.2. Experimental demonstration of delocalization in the macro- and nanocell To demonstrate the predictions, structuring experiments were carried out with the SIE system Si/SiO2 /H2 O in various electrolytes. Experiments were described by Bloe [33]. As can be derived from potentiostatic oxide formation experiments, Rpol Rion is in the range of (SiO2 ) = 1014 to 1016 cm. Specic resistances of the electrolyte can be varied between 0.1 cm (H2 SO4 ) and 108 cm in pure water. Assuming a reduction of Rel by six orders of magnitude, values of 1014 can be derived using the nanocell. Data for various electrolytes in the macro- and nanocell will be summarized later in Fig. 9. Corresponding experiments were carried out in typical cells. The localization experiments were done on p-Si in various electrolytes, on the one hand, by using a 10 m diameter microelectrode and on the other hand, by local oxidation with AFM. For details of the experimental setup see [20,33]. For all experiments, ultrapure chemicals from Aldrich were used. Results are shown in Fig. 8. By using a 10 m diameter microelectrode local oxidation of Si was carried out in various electrolytes. For the experiments the microelectrode was positioned 1 m above the working electrode and a 1 s

Fig. 8. Delocalization of the anodic oxidation of Si by conducting electrolytes: (a) in the macrocell with a 10 m counter electrode in 103 M H2 SO4 ; (b) in the macrocell with a 10 m counter electrode in tridest. H2 O and (c) in the nanocell.

Fig. 9. Comparison of nanostructures generated in NH3 -atmosphere at pH 11.8 (top image) and in H2 O-atmosphere at pH 7 (bottom image) at an oxidation voltage of 10 V for t = 1 s. The graph on the right-hand side shows the variation of the resistance of the aqueous phase in the macro-and nanocell and the resulting maximum localization of the electrochemical oxidation of Si in dependence on the pH-value.

J.W. Schultze et al. / Electrochimica Acta 51 (2005) 775786

783

pulse at an oxidation voltage of 10 V was applied. In electrolytes with high conductivity, like 103 M H2 SO4 , localization is impossible and the whole surface is covered by a thin oxide layer, which can be seen in Fig. 8a. In millipore water (R = 18 MW) a weak localization can be achieved (Fig. 8b). The obtained oxide structure has a diameter of 600 m, which clearly shows that even the high resistance of the millipore water allows lateral growth of the oxide. A better localization of the obtained oxide structures and especially fabrication of nanoscopic oxide structure cannot be achieved by further reducing the electrolyte content because the dissociation of the water itself is the limiting factor. The only possibility to further increase the resistance is to decrease the electrolyte cross section, which can be achieved in the electrochemical nanocell. In the nanocell a localization of the electrochemical oxide formation on valve metals and Si is possible as can be seen in Fig. 8c. This experiment clearly shows, as predicted by the EMT-concept, that the formation of insulating oxide structures is only possible in the nanocell. 4.4.3. Variation of the electrolyte concentration The question arises, whether the electrolyte concentration can be varied in the nanocell? In principle, this is possible by the application of vaporizing electrolytes. However, the increase of conductivity again limits the localization. When using an ammonia solution (c = 25%) for adjusting the water partial pressure, an instant delocalization of the oxide formation in the nanocell can be seen, which is shown in the AFM images in Fig. 9. Due to the formation of NH4 OH, the pH value is changed into the basic regime. This means that the electrolyte concentration is increased and the resistance Rel decreases. The experiment was also repeated for diluted HCl solution with the same result. These experiments show by analogy to the experiments described above and the EMT concept, that nanostructuring in the nanocell is possible but in a limited way. For maximum localization, the local oxidation has to be carried out in millipore water, which is summarized in the graph in Fig. 9.

Fig. 10. (a) and (b) Schematic diagrams of lateral MIM systems formed in the macrocell (a) and nanocell (b) by total oxidation of Nb; (c) potentiodynamic oxide formation in acetate-buffer at a pH of 6.1: normalized current density i/dU/dt and electrode capacity C measured at 1 kHz; (d) oxide thickness and reciprocal capacity vs. the oxidation potential U.

5. Preparation of lateral MIM-systems in the nanocell 5.1. Total oxidation of an ultrathin Nb-lm In this section we show the fabrication of lateral MIM nanostructures by oxidation in the nanocell. An ultra thin Nb-lm is used as a substrate which can be oxidized to an insulating Nb2 O5 . Since the anodic reaction obeys the high eld law of oxide growth, the thickness of the formed oxide can be adjusted by the choice of the anodization potential. Since the oxide formation factor is 2.6 nm/V, 10 nm thick Nblms can be totally oxidized. This is schematically shown in Fig. 10a for the microcell which yields a homogeneous thickness of Nb2 O5 . It was checked in the microcell and will be applied in the nanocell.

All oxidation experiments were carried out on sputtered, less then 10 nm thick Nb-lms on 300 nm thermally grown SiO2 on p-Si. For the nanoscopic MIM-structures an Nblm thickness of 5 nm was preferred. Microscopic MIMstructures were formed in acetate-buffer with a pH value of 6.1 in the electrochemical microcell [34]. Fig. 10c shows the potentiodynamic oxide formation in acetate-buffer at a pH of 6.1. The constant anodic current starting at 0.5 V proves the formation of the insulating Nb2 O5 following the high eld law. The electrode capacity decreases simultaneously from 10 F/cm2 down to 2 F. Depending on the metal lm thickness, a breakdown of i and C is observed at 4 and 10 V, respectively. At this potential, the lm is completely oxidized, it becomes insulating and loses contact with the surrounding Nb metal electrode. These measurements clearly show that it is possible to completely oxidize the Nb-lm in the microcell and to form microscopic lateral MIM-structures as shown in Fig. 10a. The quantitative analysis of coulometric data in Fig. 10d yields a growth factor k = 2.8 nm/V. The plot of 1/C versus U yields a permittivity D = 44. 5.2. Localization in the nanocell Experiments in the nanocell differ for two reasons. Due to the absence of a reference electrode, a large potential drop in

784

J.W. Schultze et al. / Electrochimica Acta 51 (2005) 775786

Fig. 11. (a) Time dependence of the oxide growth on Nb. Oxidation of a 10 nm thick Nb-layer at 10 V for the times indicated. (b) AFM line scans; (c) inuence of oxidation potential at a lateral sweep rate 0.2 m/s; (d) line scan. Nb-lm thickness d = 5 nm for (c) and (d).

the electrolyte and at the counter electrode has to be taken into account (see Fig. 6). Further, the inhomogeneous distribution of the potential yields a lateral structure similar to a lense (see Fig. 10b). Experiments of local oxidation in the nanocell were carried out with a Digital Instrument Nanoscope E in combination with a Molecular Imaging scan head and a home build potentiostat with a potential range from 10 to 10 V. In addition to this, a Digital Instruments Nanoman system with a potential range from 12 to 12 V was also used for local oxidation. After the oxidation, the surface topography was measured by AFM. As counter electrodes, diamond-doped cantilevers supplied by Nanosensors or Co/Cr coated Si-cantilevers by Nanodevices were applied. All experiments were carried out at a temperature of 23 C and a relative humidity of 40%. For the formation of MIM nanostructures on Nb-lms, the time and potential dependence of local oxidation in the nanocell was investigated. Fig. 11a shows an AFM image of Nb2 O5 dots on the Nb-layer. These dots were generated by applying an oxidation pulse of 10 V for the times indicated in the AFM image. In Fig. 1b at an oxidation time of 100 ms oxide formation cannot yet be identied. But starting at 200 ms pulses, the oxide formation is indicated by sharp peaks. The height and width of the peaks increase simultaneously. The potential dependence of the oxide formation in Fig. 11c and d shows a similar behavior: oxide formation

starts at about 3 V, the peak height and width increasing with potential. At U > 9 V the oxide height remains constant, since the Nb-lm is totally oxidized. As shown in [20], the lateral extension of the oxide is given by the resistance of the electrolyte lm lateral to the surface. Assuming a sharp, point tip, the potential drop in the insulating oxide is highest under the tip. With increasing distance from the tip, an increasing part of the total voltage drops in the electrolyte. Therefore, a smaller potential drop is available for oxide formation, and the thickness of the oxide decreases and reaches zero at 100200 nm distance. In agreement with the high eld law of oxide growth, this distance increases with U and log t. The quantitative analysis of these experiments yield a potential offset of about 3 V for the two-electrode cell due to the potential drop in the electrolyte and at the counter electrode (see Fig. 6). The rst inspection of the peak heights yield an apparent growth factor k = 1.8 nm. Taking into account the consumption of the substrate, a corrected growth factor knano can be determined. From the ratio between the oxide height above the surface z(ox) and the total oxide thickness d(ox), the correction factor can be calculated. z(ox) = 0.64 d (ox) knano = 1.8 (nm/V) k = = 2.8 nm/V = k 0.64 0.64 (5)

(6)

J.W. Schultze et al. / Electrochimica Acta 51 (2005) 775786

785

Fig. 12. Plot of the oxide thickness vs. the oxidation potential for the nanoand the macrocell. The inset shows a model of the grown oxide in order to determine the growth constant.

The corrected growth factor for oxidation in the nanocell is identical to the growth factor obtained for the microcell, which can be seen in Fig. 12. 5.3. Preparation of dened MIM nano structures For the formation of nanoscopic MIM-structures it is essential that complete oxidation can be achieved in the nanocell as in the microcell. In order to investigate this, four 1 m long lines, forming a square Nb-oxide structure with an Nb inclusion in the middle were written. These squares were

written with different oxidation potentials from 6 to 12 V at a sweep rate of 0.2 m/s. After the formation of the squares, a 500 nm long oxide line was written inside of the oxide square on the remaining Nb at an oxidation potential of 10 V and a sweep rate of 0.2 m/s to test if there is still a contact between the Nb inside and outside the square. In the case of complete oxidation of Nb along the lines of the square, it is not possible to locally oxidize the inside of the square. Fig. 13) summarizes the obtained results. At an oxidation potential of 6 V, a 1.5 nm high and 100 nm wide oxide square was formed. When the oxidation potential is increased to 9 V, the formed oxide has a height of 2.5 nm and a width of 200 nm. At 12 V oxidation potential the oxide height is not increased any further but the width increases to 300 nm. Inside the 6 V square an oxide line with a height of 2.6 nm and a width of 200 nm was obtained. This result correlates well with the height and width of an oxide line written on the bare Nb-lm as shown in Fig. 11c. This shows that at an oxidation potential of 6 V no complete oxidation is obtained, and that there is still enough Nb underneath the Nb-oxide to enable local oxidation inside the square structure. When oxidizing inside the 9 V square, the obtained oxide line is only 1.4 nm high and 50 nm wide. This shows that at an oxidation potential of 9 V the oxidation is almost complete, but that there are still conducting channels, so that a thinner oxide line is formed at 10 V inside the square on bare Nb. These channels have a high resistance

Fig. 13. Formation of lateral MIM-structures on Nb-thin lms by local oxidation in the nanocell.

786

J.W. Schultze et al. / Electrochimica Acta 51 (2005) 775786 [2] J.W. Schultze (Ed.), Electrochim. Acta 42 (1997) 2985; T. Osaka (Ed.), Electrochim. Acta 44 (1999) 3603; J.W. Schultze, G. Staikov (Eds.), Electrochim. Acta 47 (2001) 1. [3] J.W. Schultze, T. Osaka, M. Datta (Eds.), Electrochemical Microsystem Technologies, New Trends in Electrochemical Technology, vol. 2, Taylor & Francis, London, New York, 2002. [4] H. Fischer, Z. Metallkunde 29 (1937) 319. [5] W.H. Smyrl, et al., Electrochim. Acta 48 (2003) 3229. [6] C.R. Martin, et al., J. Mater. Chem. 7 (1997) 1075. [7] P. Bocchetta, C. Sunseri, G. Chiavarotti, F. Di Quarto, Electrochim. Acta 48 (2003) 3175. [8] M.-S. L ofer, H. Natter, R. Hempelmann, K. Wippermann, Electrochim. Acta 48 (2003) 3047. [9] T. Osaka, Electrochim. Acta 47 (2001) 23. [10] U. Merker, K. Reuter, K. Wussow, S. Kirchmeyer, Proceedings of the 16th Passive Components Conference on CARTS-Europe 20002, 2002, p. 71. [11] P.C. Andricacos, C. Uzoh, J.O. Dukovic, J. Horkans, H. Deligianni, IBM J. Res. Dev. 42 (1998) 567. [12] H. Asoh, S. Ono, T. Hirose, M. Nakao, H. Masuda, Electrochim. Acta 48 (2003) 3171. [13] A. Heidelberg, Dissertation, Heinrich-Heine-Universit at D usseldorf, 2005. [14] J.W. Schultze, V. Tsakova, Electrochim. Acta 44 (1999) 3605. [15] C. Deslouis, Electrochim. Acta 48 (2003) 3279. [16] T. Haisch, E.J. Mittemeijer, J.W. Schultze, Z. Metallkunde 92 (2001) 417. [17] H. L owe, W. Ehrfeld, J. Schiewe, in: J.W. Schultze, T. Osaka, M. Datta (Eds.), Electrochemical Microsystem Technologies, New Trends in Electrochemical Technology, vol. 2, Taylor & Francis, London, New York, 2002, p. 245. [18] J.W. Schultze, K.G. Jung, Electrochim. Acta 40 (1995) 1369. [19] J.W. Schultze, A. Bressel, Electrochim. Acta 47 (2001) 3. [20] H. Bloe, G. Staikov, J.W. Schultze, Electrochim. Acta 47 (2001) 335. [21] T. Schenkel, A. Persaud, S.J. Park, J. Vac. Sci. Tech. B 20 (2002) 2819. [22] J.V. Macpherson, N. Simjee, P.R. Unwin, Electrochim. Acta 47 (2001) 29. [23] M. K ohler, Atzverfahren f ur die Mikrotechnik, Wiley-VCH, Weinheim, 1998. [24] W. Ehrfeld, Handbuch der Mikrotechnik, Hanser, M unchen, 2002. [25] R. Zentel, et al., J. Appl. Phys. 94 (2003) 1205. [26] T. Fritz, W. Mokwa, U. Schnakenberg, Electrochim. Acta 47 (2001) 55. [27] F.R.F. Fan, J. Kwak, A.J. Bard, J. Am. Chem. Soc. 118 (1996) 9669. [28] K.J. Vetter, Elektrochemische Kinetik, Springer Verlag, Berlin, G ottingen, Heidelberg, 1961. [29] M.S. L ofer, B. Gross, H. Natter, R. Hempelmann, T. Krajewski, J. Divisek, Scripta Mater. 44 (2001) 2253. [30] W. Schindler, P. Hugelmann, M. Hugelmann, F. K artner, J. Electroanal. Chem. 522 (2002) 49. [31] J.C. Ziegler, G.E. Engelmann, D.M. Kolb, Z. Phys. Chem. 208 (1999) 151. [32] H. Sugimura, N. Nakagiri, Jpn. J. Appl. Phys. 34 (1995) 3406. [33] H. Bloe, Dissertation, Heinrich-Heine-Universit at D usseldorf, 2002. [34] M.M. Lohrengel, A. Moehring, M. Pilaski, Electrochim. Acta 47 (2001) 137.

because of their small diameters, thus, less potential is available for the oxidation inside the square. Inside the square that has been written with an oxidation voltage of 12 V no oxide line was obtained. Here the oxidation was complete throughout the whole oxide square, so that no conducting path was available and no oxide could be formed inside the square. From this experiment it can also be concluded that the obtained Nb-oxide does not show a breakthrough up to voltages of 10 V. With the described experiments it could be shown that it is possible to obtain nanoscopic, lateral MIMstructures on ultra thin Nb-lms. However, it needs to be mentioned here that the condition of the AFM tip as well as the thickness uniformity of the Nb thin layer are of crucial importance for these experiments.

6. Conclusions The formation of lateral MIM-structures in microscopic and nanoscopic dimensions on ultra thin Nb-lms is possible. Microscopic MIM-structures can be generated in the electrochemical microcell, because complete oxidation through the Nb-lm can be achieved, which is indicated by the current and capacitance breakdown. The growth of the Nb-oxide occurs via the high eld law and from this the growth factor of 2.8 nm/V can be obtained for microscopic as well as for nanoscopic oxides. The generation of nanoscopic MIM-structures can also be achieved in the electrochemical nanocell. At an oxidation potential of 12 V complete oxidation of the 5 nm thick Nb-lm is obtained. From this it can be concluded that MIM-structures on ultra thin Nb-lms can be generated. The localization of the oxide formation in the nanoscopic range can only be achieved in the nanocell with a water lm as electrolyte, which is formed by the water adsorption layers on the tip of the sample. Any increase in the electrolyte concentration, meaning an increase in the conductivity of the electrolyte, results in the delocalization of the oxide formation, as well as an increase in the cross section of the electrolyte lm (micro electrodes). This behavior can be explained well by the EMT concept and the use of the EMT number.

References
[1] J.W. Schultze, G. Staikov (Eds.), Electrochemistry in Molecular and Microscopic Dimensions, Electrochim. Acta (Special Issue) 48 (2003) 2851.