Abstract
The growth of crystalline ruthenium oxide square nanorods was considered on numerous substrate materials. The nanorods were found to grow easily on insulating substrates, while their growth on electrically conducting and grounded substrates was inhibited. The transfer of electrons from the plasma discharge to the developing nanorods caused the nanorods to be negatively charged and obtain a floating potential relative to ground. The electrical charging of the nanorod played a key role in their development.
1. Introduction
The growth of nanomaterial structures in general has attracted a great deal of interest and excitement over the past decade and a half with recent attention to the growth of one-dimensional nanorods [1–4]. Special interest has been given to RuO2 nanostructures because of their attractive chemical and physical properties [5]. Ruthenium dioxide is a rutile-type tetragonal oxide that exhibits low electrical resistivity (~40 μΩ-cm) and high catalytic activity [6]. As a result of these properties, nanostructured RuO2 materials are of great interest to those working on field emission displays [7] and catalyst in heterogeneous electrochemical reactors including fuel cells and electrolyzers [8]. Nanorods (NRs) of RuO2 have been successfully produced by MOCVD [9] as well as reactive sputtering [5, 10]. The reactive sputtering technique is attractive as it can be easily incorporated as a single-step process in the fabrication of integrated circuits. Also possible is the use of sputtering to deposit large area coatings on a variety of nonplanar surfaces such as automobile parts. In previous papers [11, 12], it was proposed that Ru hyperoxides, formed in the reactive sputtering process, are the high-vapor-pressure species that permit the synthesis of RuO2 nanorods, at a low substrate temperature. The species RuO4 has been proposed as an ideal precursor for the CVD growth of Ru-containing thin films [13, 14]. This high-vapor-pressure precursor is credited with having the surface mobility required to supply chemical reactions that produce highly faceted nanostructures. In this paper, a high plasma density, reactive sputtering process was used to grow RuO2 NRs on a variety of substrate surfaces, with special attention paid to the electrical conductivity of the substrate material. The substrate materials used included a Si wafer substrate piece with various preexisting thin film coatings and metal sheets such as stainless steel and Ti. The nanorods have previously been found to be single crystal by X-ray diffraction analysis but randomly oriented relative to one another as displayed by the SEM pattern shown in Figure 1. The main difference in the growth process used by our group, and that used in other papers involving a sputter process, is that our process utilized a high plasma density (1010 cm−3 versus 108 cm−3) [13, 15–17] and a lower mole fraction of oxygen in the reactor ambient gas mixture (5% versus 50%) [12].
2. Experimental
The RuO2 nanorod materials were formed by self-assembly through a reactive sputtering process onto a heated substrate surface. The process used to grow these nanomaterials on a Si substrate was described in earlier papers [11, 16, 17]. The RuO2 nanorod samples were prepared by reactive sputtering from a Ru metal target operated at an RF frequency of 13.56 MHz and a power level of 50 W. In addition to the RF-driven discharge, an electron cyclotron resonant (ECR) plasma was generated above the sputter target/substrate assembly increasing the plasma density to ~1010 cm−3. The ambient in the reactor was obtained by flowing 100 sccm of a 5/95% O2/Ar gas mixture, throttled to a pressure of 15 mTorr. The samples were heated radiatively from behind with a resistive BN-coated pyrolytic graphite heater to a temperature of 460°C. The resulting nanorod samples were then removed and observed with a JEOL JSM 6060 SEM.
Multiple substrates were used in this paper, including a Si wafer, an Al evaporation-coated Si wafer, a polished Ti sheet, and a Si3N4-coated Si wafer. Strongly conducting substrate materials included a stainless steel sheet, a stainless steel sheet coated with RuO2, AZO on BSG (aluminum-doped zinc oxide on borosilicate glass), and ITO on BSG (indium tin oxide on borosilicate glass).
Electrical characterization was performed while the nanorod-coated substrates were submerged in an electrolyte solution (KOH in water) to ensure adequate contact to the nanorods. All measurements were performed at room temperature. A more detailed description of the apparatus and procedure was published earlier [18].
3. Results and Discussion
The deposited RuO2 nanorods were randomly spaced and oriented with an average length of 1 μm for the growth conditions used. Characterization of the nanorods films by XRD suggests that the dominant growth phase is in the (200) direction, as evident by the peak at 40° in Figure 2.
The substrates in all successful NR growth cases were insulating and amorphous, and no crystalline insulating substrates were tried. The preference for the growth of the (200) phase is not obtained from the crystal orientation of the substrate, but the preferred growth from that phase is normal to the substrate. As the nanorod grows in the axial direction, only one crystal orientation can survive as a consequence of the high aspect ratio of the nanorods. Previous papers have shown a preferential growth of selected phases on insulating crystalline substrates over that on amorphous insulating substrates [19]. The current paper considers the growth of RuO2 nanorods on insulating or conducting substrates.
In an earlier communication, the length was described to be strongly dependent on substrate temperature used during the growth process [11]. In that earlier paper, the RuO2 nanorods were grown on Si substrates that had previously been coated with Au nanodots, which were assumed at that time to be required as nucleation sites for the subsequent growth of the RuO2 NRs. Although no mechanism describing the involvement of the Au nanodots was given, the paper left the impression that these were required for RuO2 nanorod growth. In other papers [5], and explicitly in the present paper, it has been found that Au nanodots are not required and will not serve as the nucleation sites for the growth of RuO2 NRs. In our previous communication [11] the following was stated:
Once formed the RuO4 molecules migrate across the wafer surface and nucleate to form RuO2 clusters. The clusters grow very rapidly into square nanorods with a length of 1.4 μm and width of 30 nm at a substrate temperature of 600°C. This process occurs on a very short timescale, and may be regarded as nearly instantaneous.
We have remained baffled by what appeared to be an almost instantaneous formation of the RuO2 NRs. In the earlier communication [11], it was also stated that the growth proceeds by the following:
Some combination of thermal diffusion of surface species and an electrostatic trapping mechanism that collects sputtered Ru ions from the plasma beam.
Although adatom migration across the substrate surface is still assumed to contribute to the nutrient feed of Ru species to the developing NR, the electrostatic trapping or adatom contribution directly from the discharge was poorly defined and left as speculation.
Experimental evidence gathered in this paper now permits us to state that electrostatic trapping contributes directly to the formation of RuO2 NRs. It has been observed that dense nanorod formation originates on poorly connected fragments of foreign matter (FM) that clings onto the substrate. An SEM image of such an NR-decorated piece of FM is shown in Figure 3. It is safe to assume that pieces of FM are weakly bound to the substrate surface and therefore represent at best poor electrical connection to that substrate. Furthermore, movement of adatom species across the substrate surface and onto the object of FM is unlikely and made difficult by the required movement from the substrate surface up onto the FM particle. An electrostatic mediated collection of Ru species from the gas phase discharge is the only possibility.
Through a trial-and-error process, it was found that RuO2 NRs grow easily on the following substrates: (a) Si wafer, (b) Al evaporation-coated Si, (c) polished Ti sheet, and (d) a Si3N4-coated Si wafer. SEM images of the RuO2 NRs grown on these substrates are shown in Figure 4.
(a)
(b)
(c)
(d)
All of these substrates were highly polished, and it was initially assumed that this was affecting the surface mobility of the Ru adatoms that contributed to RuO2 NR formation. The goal was to extend the above process to the growth of RuO2 NRs on a stainless steel surface. Stainless steel substrates were tried, with both rough and smooth surface finishes, without effect [20]. On further consideration, the difference between the SS substrate and the group of substrates that gave successful results was the electrical resistance of the surface. In the case of the Si, Al, and Ti substrates, each of these elements reacts strongly with the oxygen contained in the air ambient and creates a surface of SiO2, Al2O3, and TiO2. The next step was to further verify that RuO2 NRs could not be grown on conductive surfaces. A sampling of four different substrate materials unaffected by the oxygen-containing ambient was tried: (a) stainless steel, (b) stainless steel coated with RuO2, (c) AZO on BSG (aluminum-doped zinc oxide on borosilicate glass), and (d) ITO on BSG (indium tin oxide on borosilicate glass). During the growth process, the top surface of the substrate was electrically grounded with a conducting stainless steel retaining clip. Each of these samples recorded a sheet resistance of <1 × 10−6 Ω/cm2. SEM images of these four substrate surfaces following a similar process sequence used to grow the RuO2 NRs shown in Figure 4 are shown in Figure 5. There is a very low density of RuO2 NRs present on these conductive surfaces. The few RuO2 NRs that are present could be the result of insulating defects on these surfaces, perhaps resulting from FM or structural damage in the substrate surface.
(a)
(b)
(c)
(d)
The growth of the RuO2 NRs on the Si, Al, and Ti surfaces can be explained as a consequence of the native oxide that exists on these substrate surfaces. Further confirmation that an insulating layer exists between the conducting RuO2 NRs and the Si, Al-coated Si, and/or Ti substrates was obtained from an electrical resistance measurement of these three systems. Electrical contact to the top of the RuO2 NR film was made with an electrolyte solution (KOH in water). The measured V-J curves for the RuO2 NRs on these three different substrate materials are shown in Figure 6. The full curve in Figure 6 was used previously in a paper into the use of RuO2 NR-coated electrodes as cathodes in the electrolysis of an aqueous solution of KOH [18].
The production of hydrogen at the cathode does not begin till a voltage less than −0.5 V is applied to the cathode. The portion of the V-J curve less than this value results from the electron and ion transport across the electrolyte and is shown in Figure 7.
The voltage drop measured from the solution to the substrate electrode is the result of resistive losses: (1) across the substrate, (2) at the oxide interface between the RuO2 NR and the substrate surface, (3) across the RuO2 NR length, and (4) at the interface between the RuO2 NR and the liquid electrolyte. A circuit diagram representing this proposed model is sketched in Figure 8.
The total resistance for one NR, , submerged in an electrolytic solution can be expressed with the following relationship:
The total resistance measured between the electrode substrate and the solution can be calculated from the slope of the V-J curves shown in Figure 7. The measured values are summarized in Table 1.
The resistance measured for the electrode fabricated with the Si substrate is higher than the other two, and this is a result of the higher resistivity of the Si substrate relative to Al and Ti substrates, which are assumed to be negligible. In the case of the Si electrode, with a resistivity of 50 Ω cm, there is a voltage drop across the substrate portion of the electrode, calculated to be 700 Ω. The nanorods for all samples are approximately the same size and shape and are assumed to be of similar consistency. Therefore, the resistances for a nanorod, , should be the same in all cases and should be small with the resistivity of ~40 Ω cm.
The relationship between the resistance measured experimentally and for an individual NR can be expressed as where is the surface density of RuO2 NRs per cm2. Using the above assumptions, the sum of measured resistances at the base of the nanorod connecting to the substrate and at the interface with the electrolyte solution for the three samples is estimated to be as shown in Table 2.
All measured resistance values are reported per cm2 of substrate or electrode area. It is assumed that an insulating film on the above substrates, that permitted the growth of the RuO2 NRs, is the characteristic native oxide layers commonly associated with these materials. Much is known about the native oxide layers on these material substrates, including thickness () and resistivity () [21–23]. These characteristic values are summarized in Table 3.
Assuming these characteristic thickness and resistivity values for what was proposed as the nucleation sites for the RuO2 NRs, the resulting resistance for conduction through a single nanorod would be the product of and . It is not possible herein to assign a specific portion of the total resistance value to either the contact resistance with the substrate or the electrolyte solution, but it is possible to assign an upper limit on either of these values. Assuming that the Ω, then the values given in Table 2 are equal to the resistance at the interface between the RuO2 NR and its substrate, per cm2 of electrode area. The equivalent circuit for the NR-coated electrode is a parallel combination of resistance values assigned to a single NR, . The results of this analysis are organized in Table 4. The calculated surface density of nanorods by this analysis ia acceptable for the Si and Al/Si electrodes, but is an underestimation of those observed on the Ti substrate. The initial assumption that Ω may be an oversimplification and may in fact be larger than . Further investigation into this discrepancy is required.
The plasma discharge in the ECR-enhanced sputter process that is used in the current investigation was the subject of an earlier study that used a Langmuir probe to characterize the physical nature of the plasma stream [24]. The electron temperature of the Ar discharge was found to be less than 2 eV. This value is physically credible and can be used to calculate other parameters that can describe the discharge used. It was proposed above that the growth of the RuO2 NRs occurs as a result of an electrostatic trapping phenomenon. A more complete description of this mechanism requires a proposed nucleation site that is electrically insulated from the grounded substrate. The nucleation site can consist of an island of insulating material on what can be a conducting substrate. In the case of the Si, Al/Si, and Ti substrates, this was accomplished by the formation of SiO2, Al2O3, and TiO2 nucleation sites. The RuO2 NR will nucleate at this site and will be assisted by the fact that the conducting RuO2 mass acts as a body that will attain the floating potential of the plasma discharge. A plasma discharge is a collection of an equal number of electrons and ions. The mass of the electrons is significantly less than that of the ions, and therefore, they travel at significantly higher velocities and will escape the plasma and charge any body in its vicinity. The conducting body as a result of this high-mobility electron flux will develop a negative bias with respect to ground, which is known as the floating potential, . An expression relating the electron and ion temperatures and masses is shown as follows [24]: where = floating potential (volts), = electron temperature (eV), = ion temperature (eV), = ion mass (kg), and = electron mass (kg).
The for the current system and operating conditions can be estimated to be equal to 7 to 8, or approximately 14 V [23, 24].
This result suggests that the RuO2 NR nucleation site will be maintained at this floating potential, in the neighborhood of 14 V relative to ground, and this body will then electrically attract ions from the discharge to this site. In theory, the flux of ions will be equivalent to the flux of electrons to this nucleation site, also known as ambipolar diffusion, see the following [24]: where = electron and ion flux, respectively = electron and ion diffusion coefficient, respectively , and = electron and ion mobilities, respectively .
These ionic species will represent the nutrient flux, which will feed the growth of the nanorods. It is the crystallographic orientation of the developing RuO2 NR which will determine its shape and direction of growth. The electric field and/or electrical potential in the vicinity of the RuO2 NR acts as an electrostatic trap for ions in the ECR plasma stream. This physical process has been represented in the sketch in Figure 9. The diameter of the RuO2 NR during its initial stages of growth is very narrow and acts as an effective electrostatic trap. This can account for the observation that the RuO2 NRs appear to nucleate instantaneously. Continued growth is a much slower process that involves surface migration across the substrate surface and up the NR surface.
4. Conclusions
The nucleation and growth of RuO2 nanorods by the plasma-enhanced sputtering process require the existence of an electrically insulated substrate surface. The growth of the RuO2 NRs on insulating surfaces was attributed to the collection of ionic Ru-containing species produced in the sputter process. The developing RuO2 NR becomes negatively charged as a result of the escaping higher-mobility electrons from the gaseous discharge. The negatively charged RuO2 NR then collects positively charged ionic Ru-containing species from the sputter discharge. This process has been observed to occur in less than a minute.
Acknowledgments
This work has been partially supported by NASA through funding awarded to the Vermont Space Grant Consortium in NASA grant NNX10AK67H, as well as the US Navy, Office of Naval Research.