Government Contract The United States Government has rights in the portions of this application related to fuel cells, pursuant to Contract No. DMI-9660598 awarded by the National Science Foundation.

Field of the Invention The present invention relates to reduced grain boundary (herein referred to as"RGB") crystalline thin films, and more particularly to reduced grain boundary crystalline thin films for use as, inter alia, electrolytes in solid oxide fuel cells (SOFC), gas separation membranes, batteries, diffusion barriers, dielectric material in electronic, photonic, radio frequency (RF), and pyroelectric devices.

Background of the Invention While the ability to deposit epitaxial coatings on substrates has been demonstrated in the past, the methods to produce these coatings have several disadvantages such as being relatively slow as well as requiring expensive vacuum and other extensive equipment.

Furthermore, the use of epitaxial coatings has been limited to only a few applications in the past. Significant efforts were spent achieving more perfect substrates with all grains aligned or even true single crystals, and then growing epitaxial single crystal coatings. What was not understood was that many of these coatings would function adequately if not just as well, even if the coatings comprised reduced grain boundaries without being single crystal. With the advent of combustion chemical vapor deposition (CCVD), a new, quicker and less expensive method of producing RGB coatings has been realized. In addition to the cost savings and reduced environmental impact achieved by using the CCVD method, the use of reduced grain boundary coatings has applications to several fields.

A very promising chemical vapor deposition process is the combustion chemical vapor deposition (CCVD) processes described in U. S. Patent Nos. 5,652,021; 5,858,465; and 5,863,604, and issued to Hunt et al. These patents, which are hereby incorporated by reference, disclose methods and apparatus for CCVD of films and coatings wherein a reagent

and a carrier medium are mixed together to form a reagent mixture. The mixture is then ignited to create a flame or the mixture is fed to a plasma torch. The energy of the flame or torch vaporizes the reagent mixture and heats the substrate as well. These CCVD techniques have enabled a broad range of new applications and provided new types of coatings, with novel compositions and improved properties. In addition to these three patents, U. S. Patent No. 5,997,956, also issued to Hunt et al. describes a further method of CVD involving the use of a thermal spray with near supercritical and supercritical fluid solutions. The coating processes disclosed in this patent are also useful for forming the reduced grain boundary coatings of the present invention, and this patent is incorporated by reference as well. For oxygen-sensitive materials, the coatings may be applied by controlled atmosphere chemical vapor deposition techniques as described in U. S. Patent Application No. 09/067,975, filed April 29,1998 and entitled APPARATUS AND PROCESS FOR CONTROLLED ATMOSPHERE CHEMICAL VAPOR DEPOSITION (CACVD), the teachings of which are incorporated herein by reference. The RGB films may also be deposited by conventional techniques know in the art, such as a variety of sol gel, chemical vapor deposition and physical vapor deposition processes.

U. S. Patent Nos. 5,739,086, issued on April 14,1998 and 5,741,377 issued on April 21,1998, both to Goyal et al. teach textured articles having a rolled and annealed, biaxially textured metal substrate and an epitaxial coating deposited thereon. U. S. Patent No.

5,523,587 issued on June 4,1996 to Kwo is drawn toward a method for low temperature growth of epitaxial (single crystal) silicon, and devices produced using this method. U. S.

Patent No. 5,968,877, issued on October 19,1999 to Budai et al. discloses high To YBCO superconductor deposited on a biaxially textured Ni substrate. International Patent Application, Publication No. WO 99/15333, published on April 1,1999 to Fritzemeier et al. is drawn to superconducting articles with epitaxial layers. U. S. Patent No. 5,741,406, issued on April 21,1998 to Barnett et al. discloses solid oxide fuel cells having dense yttria- stabilized zirconia (YSZ) electrolyte films and a method of depositing these electrolyte films.

None of the above references and patents, taken either singly or in combination, is seen to describe the instant invention as claimed.

Summary of the Invention The developed CCVD processes described herein have proved advantageous for forming many types of thin films, thick films and other coatings. Thin films being defined herein as those films having a thickness of less than 10 microns. When used to deposit reduced grain boundary (RGB) films on lattice matching substrates, the CCVD processes produce very dense crystalline coatings. The coatings can have multiple orientations because: the substrate is polycrystalline; and/or the coating has multiple lattice matches with the substrate; and/or different deposition conditions can have different preferred epitaxial relationships. The main objective is to reduce the high energy grain boundaries which can be conduits for unwanted impurities, diffusion, undesired physical properties, and reduce stability and life. These reduced grain boundary coatings can provide superior coatings in several fields such as solid oxide fuel cells (SOFC), gas separation systems and microelectronics. Several different CCVD processes have been developed as disclosed in the above-identified U. S. Patent Nos. 5,652,021; 5,858,465; 5,863,604, and 5,997,956 and in U. S. Patent Application No. 09/067,975. It should be noted that one or more of these CCVD processes may be used to produce the reduced grain boundary coatings of the present invention, depending on the requirements of the application. These requirements include, but are not limited to, speed of deposition, temperature limitations and oxidation reduction or enhancement. While CCVD has shown excellent capability for depositing these RGB films, other deposition processes may be used, as discussed above.

One way to produce some of the various RGB coatings of the present invention, a rolled, or otherwise textured metal substrate is coated with the desired material using CCVD.

Another method involves first depositing an ion implanted template layer of a material that will orient, such as MgO or YSZ. Yet another form is to select a polycrystalline substrate to which there are multiple orientations to which the coating will nucleate from such as substrates including SrTiO3, LaAl03, CeO, MgO, YSZ or alumina, for perovskite coatings such as BST, PZT, and LSC. None of the examples are meant to be limiting to the scope of the invention, but are intended to serve as instructional examples of which additional materials and combinations can be found in the literature. Although other materials with a similar lattice parameter can be used depending on the thin film, nickel is the preferred material. Nickel is preferred as it is relatively inexpensive, has a high melting temperature, is

easily roll textured and can be etched using inexpensive and commercially available materials such as nitric acid, a 50%/50% mixture of acetic acid and nitric acid, or Ce (NH4) 2 (N03) 6. Tn addition, Nickel's lattice parameter facilitate alignment with the lattice of many other materials in the formation of an RGB film. Thus Nickel or Nickel containing alloys are well suited as a substrate for many applications. However, it will be understood that the methods described herein may employ other substrate materials, depending upon the application, such as where cost is less of an issue, to develop RGB films according to the invention. For example, highly conductive metals such as copper and silver may be preferred substrates for radio frequency (RF) and high frequency electronics. It is also to be noted that the substrate used in the deposition process is not necessarily (and oftentimes is not) a component of the final product. After depositing the RGB layer (s) on the textured substrate, the RGB enabling substrate is often etched away, leaving the deposited layer or layers having a preferred RGB.

Should additional layers be required, they may be deposited prior to or after etching away the textured metal substrate. Of course, for some applications, it may be desirable to leave the original textured substrate as part of the final product. The details of the specific construction and coating process of each RGB thin film are further described below, with respect to their particular application.

Nickel and MgO have similar lattice structure. Thus substantially any coating that can be deposited on nickel with reduced grain boundary can be deposited on MgO and vice versa.

Other coating and substrate materials (and lattices) can be found in the Handbook of Chemistry and Physics, Powder diffraction File System, or searched on a materials, web-site, data base such as www. matweb. com. Multiple materials can be grown RGB if there is some compatibility of lattices. This will yield improved performance and longer life to the system.

Certain conductive oxides such as RGB ZnO and RGB doped ZnO are useful for forming electrodes. Another important material for forming electrodes is reduced grain boundary LiCol-xNixO3 where X can range from 0 to 1 but is often advantageously greater than 0 and less than 1. Such material provides good conductors and active electrodes for lithium batteries and other batteries, as the reduced grain boundaries provide numerous benefits such as higher conductivity, greater diffusion barrier and longer life. Such material can be deposited on Nickel, Nickel alloys or MgO. Nickel is one of the preferred electrical conductors, however the coating can be detached from the Nickel and transferred. The LiCol-

XNiX03 can then act as an RGB template for the deposition of other materials.

Another type of RGB coating described herein involves electrolytes and electrodes for use in solid oxide fuel cells (SOFC). Fuel cells are a fundamentally novel way of generating electrical power from a variety of fuels. It has long been recognized that the successful development and commercialization of fuel cells will offer significant environmental advantage and greatly reduce global pollution. The key advantage of a fuel cell is the high- energy conversion efficiency. Without the intermediate thermal energy, the fuel cell directly converts the electrochemical energy to electrical energy. Other advantages include simpler construction, high efficiency at part load, potential for co-generation, and much lower production pollutants. Currently, the commercial market for fuel cells is limited to spacecraft and other specialty uses, due to the expense of the required chemical precursors and deposition equipment. A fabrication process to deposit low-cost and high quality SOFC thin films without complex post-deposition treatment/sintering is essential for the widespread commercialization of fuel cells. Furthermore, for widespread use, the performance/efficiency of SOFCs must be improved. By providing RGB crystal thin films, it has been demonstrated that CCVD processes are capable of producing microstructures and crystallinity needed in both the electrolytes and electrodes for improving SOFC performance. In addition, with its open-atmosphere capabilities, the CCVD process enables continuous depositions of these layers with capital cost reduction and with improved operational costs compared with the other CVD processes, and hence offers significant commercial advantages for the SOFC industry.

Current processes for fabricating oxide electrolytes are limited by the availability of desired powder stoichiometry and characteristics, the film thickness (a result of both raw material availability and process limitation), and the densification for pin hole-free and ideal grain boundary/interface microstructure. These limitations thus hinder the development and possible improvements of SOFC performance and efficiency. Recently, it has been shown that by fully densifying (but still standard ceramic processing texture) and greatly reducing the YSZ electrolyte thickness to 4-5 um, the power density of YSZ-based fuel cell can be increased to 1.6 W/cm2. Further, with an RGB electrolyte, increased conductivity was measured. This indicates that a technique to fabricate an RGB electrolyte layer results in a

very thin and dense electrolyte layer without relying on the availability of raw materials greatly improves the possibility of the commercialization of SOFCs.

While for many applications, such as superconductors, single orientation is critical, it is to be appreciated that for many applications the coating need not be single crystal or single orientation. Even where coating orientation need not be flawless, reduced grain boundaries may go a long way to providing the enhanced performance benefits of epitaxial versus normal polycrystalline deposition. In non-epitaxial or normal non-vapor deposited polycrystalline films, the grain size is a few times smaller than the thickness of the film. With vapor deposited coatings, the equiaxed grain structures are smaller than the film's thickness, and collumnor grains are narrower than the film's thickness. Reduced grain boundary layers exhibit desirable properties, such as aiding pyroelectric effect, reducing leakage current and loss and slightly enhancing capacitance in capacitor applications. By reduced grain boundary layers is meant herein at least columnar, i. e., well defined grains extending from the bottom to the top of the layer, also know as Type II structure, with the grains also being equiaxed in Type III structures having a 1: 1 width to height ratio. Higher width to height ratios are even more desirable. A 5: 1 width to height ratio is preferred. More preferred are ratios of 10: 1, 20: 1 and even higher up to and including epitaxial but not single crystal. The height to width relationships are more clearly defined in the detailed description below. The substrate crystals can have single or multiple orientations and the coating grains can have single or multiple in-plane orientations within each substrate grain. The RGB coatings of the present invention not only reduce the cost of substrates while allowing the formation and use of larger substrates as compared to epitaxial single crystal coatings, but also increase the strength of the coatings well beyond that of more amorphous materials (due to the decrease in grain boundaries), thus RGB is very preferred for commercialization.

The CCVD processes produce dense and RGB electrolyte thin films, porous and adherent electrodes, and may be combined with other coating techniques to construct a superior, solid oxide fuel cell with reduced manufacturing time and costs. By operating in the open atmosphere with inexpensive precursors, CCVD provides continuous coating of RGB electrolyte layers, which will increase the ionic conductivity at lower temperatures. The CCVD-based process provides for better fuel cell performance by minimizing grain boundary/interface resistance and reducing polarizations. In addition, the CCVD system

capitalization requirement is reduced at least ten times compared to a vacuum-based system, and the throughput is far greater than with other technologies. The CCVD process generally uses solutions with all the necessary elemental constituents dissolved in a solvent that also acts as the combustible fuel. The ease of controlling and changing dopants and stoichiometries enables the deposition of multi-layered and multi-compound films in one process, which further increases throughput and reduces production costs. Depositions can be performed at atmospheric pressure and temperature in an exhaust hood, a clean room or outdoors.

Gas separation membranes can also be formed in accordance with the present invention. One application of gas separation membranes is in the production of oxygen.

Mixed-conducting oxide membranes could produce oxygen with lower costs than the present commercial means of oxygen production, namely cryogenics, pressure swing adsorption (PSA) and polymeric membranes. Developing these oxygen semipermeable membranes requires the fabrication of a hybrid membrane comprising a thin film of mixed-conducting oxides supported on a porous substrate, preferably of the same material. CCVD can be used to deposit high-quality, dense, gas-tight, pinhole free RGB layers of mixed-conducting oxides onto a continuous RGB enabling substrate such as Ni and then transfer to porous ceramic substrates. Alternatively, a selectively etchable RGB enabling material, such as Ni, can be sintered in with the final desired porous support, but initially forming a dense substrate. After depositing the RGB on the enabling material, this material is selectively etched away leaving the RGB thin film layer covering the desired porous substrate. Such hybrid membranes will offer both high permselectivity and high transport rates for oxygen, thereby enabling its widespread use as a semipermeable membrane for commercial oxygen production. By reducing the costs of producing pure oxygen, the membranes of the present invention can greatly enhance several industries. Natural gas, comprising mostly methane, can be converted to clean burning transportation fuels, by using pure oxygen for conversion of methane to syngas. Commercial production of such fuels has been hindered by the high cost of oxygen production, which to date is primarily from expensive and energy-intensive cryogenic processing. An alternative is the use of dense ceramic membranes, based on mixed- conducting perovskite ceramic oxides that exhibit high ionic and electronic conductivity for oxygen. These types of membranes have become of great interest as a potentially economical,

clean and efficient means of producing pure oxygen by separation from air or other oxygen- containing gas mixtures. Other applications of oxygen-separating membranes range from small-scale oxygen pumps for medical applications to large-scale usage in combustion processes, e. g. coal gasification. Another application of mixed-conducting oxide membranes is to be found in the field of chemical processing, including the partial oxidation of light hydrocarbons, such as natural gas to value-added products including ethane-ethene mixtures, syngas production, waste reduction and recovery.

The desired perovskite structure (ABO3) consists of a cubic array of corner-sharing BO6 octahedra, where B is a transition metal cation. The A-site ion, interstitial between the BO6 octahedra, may be occupied by either an alkali, an alkaline earth, or a rare earth ion. In many cases, the BO6 octahedra are distorted, tilted due to the presence of A ion, which is generally larger than the B cation. The onset of electronic conductivity mainly depends on the nature of the B-site cation. The total electrical conductivity can be either predominantly ionic as in the acceptor-doped rare earth aluminates or predominantly electronic, as in the late transition metal containing perovskites. Some perovskite oxides (AB03) containing a transition metal at B sites are found to be good mixed-conducting materials, exhibiting both ionic and electronic conductivities. Due to their high electronic and ionic conductivity, these materials can be used as oxygen semipermeable membranes without electrodes and external circuits. Oxygen conductive ceramic materials include yttria-stabilized zirconia doped with either titania and ceria. Conducted on acceptor-doped perovskite oxides include those with the generic formula: Lal-xAxCol-yByQ3-s, wherein A=Sr, Ba or Ca and B=Fe, Cu or Ni).

Another application of gas separation membranes is in the production of hydrogen.

Generation of pure hydrogen is required for various large-scale industrial applications such as hydrodesulfurization and hydrotreating processes in refineries. The increase of hydrogen to carbon ratios in transportation fuels and the burgeoning field of hydrogen fuel cells are expected to greatly increase the demand for hydrogen gas. Hydrogen can be extracted from raw fuels such as natural gas as well as from process streams such as the catalytic gasification of coal, non-oxidative methane conversion and steam reforming. Membrane-based separation processes are cost-effective and environmentally friendly alternatives for separating hydrogen when compared with other more energy intensive processes such as distillation and pressure

swing adsorption. To be useful in industrial processes these separation membranes must be stable at high temperatures (up to 1000 °C) and pressures (up to 600 psi). Industrial membranes must also have reasonably high tolerances to corrosion or poisoning by the effluent streams commonly encountered in industrial processes.

Ultrahigh purity hydrogen can be produced from gaseous mixtures containing hydrogen by the use of three types of membranes including polymeric membranes, inorganic (non-metal porous or nonporous) membranes and dense (metal) membranes. Polymeric membranes suffer from limited selectivity and resistance to high temperatures and reactive chemicals that may be present in typical feed streams. Porous inorganic membranes exhibit very high hydrogen permeability, but suffer from brittleness and very low hydrogen selectivity. Inorganic membranes derived from proton conducting solid-oxide ceramic materials exhibit lower permeability, are brittle, and require an electric current to be applied through electrodes at each surface of the membrane to induce the proton conduction.

Examples of proton conducting membranes that do not require the application of electric fields include the perovskites such as yttria-stabilized strontium zirconate and yttria-stabilized strontium cerate.

Palladium and palladium alloy membranes have generated much interest in their applications to membrane reactors. Membrane reactors function as both separator and reactor in a single unit. These membranes are superior to polymeric membranes and to inorganic (non-metal) membranes in that they exhibit excellent selectivity for hydrogen and can be operated at high temperatures (up to 1000 °C). Palladium-based membranes are easily poisoned by sulfur containing gas contaminants such as hydrogen sulfide. To overcome the deficiencies associated with sulfur-deactivation, palladium metal is coated with a thin layer of yttria-stabilized strontium zirconate, which is not affected by sulfur containing compounds.

Another important factor that comes from the Present invention of forming RGB coatings is the lower energy grain boundaries to the substrate and between the grains within the coating. This can yield higher adhesion and lower susceptibility to re-crystallization. Re- crystallization can lead to change in physical properties and dimensions as well as increased diffusion or reaction with adjacent materials and surface impurities. A RGB coating may have multiple orientations, but these orientations are predominantly low energy (good lattice matching) between at least half of the grains. It is preferred that at least 75%, and even more

preferred that at least 90% of the grains have most of the interfaces with the substrate and within the RGB coating layer (s) be low energy (good lattice translation). A low energy grain boundary is one in which there is a regular alignment (periodic) of adjacent atoms between the two grains. Higher energy grain boundaries have a separation by unordered atoms (not periodic). For such for traditional processing high-energy grain boundaries, the uncontrolled diffusion through is much higher and the stability is much lower than for the low energy grain boundaries formed by the present invention.

The RGB of the present invention does rely on epitaxy to be formed, but not full single crystal coatings. As previously stated the processes, size, cost and yield of single crystal substrates can reduce or prohibit products to be made. Further the processing speed, cost and yield on the single crystal coating can reduce or prohibit products to be made. The properties of the RGB coating are significantly better than traditional coating processing, but when choosing the correct substrate the cost are similar to only moderately higher. The cost of single crystal coatings though is substantially higher.

A wide range of RGB materials in electronics that can benefit are conductors, insulators, and semiconductors. To enable the required advanced capability in applications such as printed wiring boards, solar cells, displays, and electronics in packaging for general merchandise (wireless scanning and tracking), the cost must be low and the scale of production greatly increase. With few grain boundaries the coatings can be thinner while yielding similar or superior performance, which reduces cost, size and weight. By depositing these materials on a selectively removable substrate such as Ni the coating material can be attached to a polymer, (or other organic materials) such as an epoxy, and then the Ni partially or fully removed. This yields a high quality coating in a medium that does not allow such a coating by any process directly. In this way a high quality transparent conductive oxide could be formed on a clear lightweight polymer for displays, shielding, photovoltaics, static control etc. The etched Ni can be readily recycled and used again to further reduce costs.

Semiconductor thin film electronics with layers of conductors with low and high dielectric materials can be made and incorporated directly in polymer packaging for food, beverage, consumer and industrial products. This transferability is key to many opportunities High-energy grain boundaries also disrupt the flow of electrons and photons causing resistance and scattering. Therefore it is desired to not have grain boundaries at all so that the

highest conductivities and transmission is possible, but as has been discussed this limits size and increases costs. In high temperature superconductors there is orders of magnitude changes going from single orientation to multiple orientations (over 90% decrease in critical current). For regular conductors and photon transmission medium this effect is a few percent between single crystal and RGB, but there can be over 50% increase in resistance or scattering with standard polycrystalline materials compared to RGB. The benefit to cost differential is much better for the present RGB in these regular conductors and photon transmission medium.

Photonic materials of particular interest include both non-electrooptic transmission layer materials as well as electrooptic transmission layer material such as LiNbOs, BaTi03 and PLZT. These materials are currently used only in true single crystals. Multi-orientation RGB will show significant improvement over current polycrystalline materials, but most preferred for photon transmission application it is an aspect of using this as predominantly single orientation RBG to minimize scattering, polarization, and dispersion. These active (variable index of refraction) materials are well know in the art for their various applications in active optical components, devices and systems.

Another practical application of RGB thin films is as dielectric material in capacitors.

Capacitors are used in almost every electronic product, as they are one of the key passive components in electronic circuits. They are utilized according to their capacitance values which are dependent on their physical structure and dielectric permittivity. The current invention utilizes the CCVD process to deposit low-loss, RGB perovskite films to produce dielectric layers with extremely low leakage and/or high permittivity. This produces capacitors with a greater capacitance to size ratio, resulting in a reduction in the required size and weight of individual capacitors and the overall circuit. This advantage can be utilized in discrete components as well as embedded devices. Palladium, platinum, nickel, silver, copper, and other metals can be textured to yield a predominant orientation which is amenable to RGB growth.

Once a capacitor is formed, its thickness and dimension are fixed and therefore, the only adjustable factor for modifying capacitance is its permittivity. The RGB dielectrics of the current invention, allow the application of a dc bias to adjust the capacitance of capacitors made of high-permittivity, ferroelectric materials. These electrically tunable capacitors can

be used in a wide variety of applications. For example, a signal filter circuit, normally composed of a capacitor and a resistor, will be able to function as a multiple frequency filter by electrically adjusting the permittivity and hence, the capacitance. Capacitors embedded in printed circuit boards can also take advantages of this multiple frequency capability. In wireless communication, electrical adjustment of permittivity will induce phase shifts that are highly desired for phase array radar. In addition, these tunable devices can also be used as filters and oscillators for telecommunication at variable frequencies. Prior art capacitors are not able to achieve the desired properties using a practical material processing approach.

Thin disks of dielectric materials, sliced from bulk materials, require a voltage in the order of hundreds of volts to achieve practical adjustability. In addition previously deposited dielectric films have been too lossy to render practical utilization of this effect. Nickel capacitors and resonators can be detached or embedded.

In the field of pyroelectrics, the energy-production cycle of a ferroelectric thin film material (FTFM) relies on a temperature dependent hysteresis loop in the charge versus voltage diagram. By controlling both the temperature and the electric field across the FTFM, the hysteresis loop can be traversed in a direction that leads to the generation of energy rather than the more usual dissipation of energy. The active element of the energy conversion circuit relies on a capacitor with the FTFM serving as the dielectric. An extremely thin FTFM layer is advantageous as the capacitance density would be large, the thermal cycle time would be small, and the working voltage would be reduced when compared to thicker dielectric layers. Generally, a thin dielectric film presents processing difficulties, such as short circuits, fatigue (aging) and increased leakage caused by grain boundaries. The RGB dielectric thin film coatings of the present invention overcomes these difficulties by providing extremely thin film dielectrics (<500nm) that are fully dense with minimal leakage current.

Also to be widely used, the finished product must be low-cost and of large area, thus ruling out single crystal coatings. This is often the case in the many applications that can benefit from RGB coatings.

Brief Description of the Drawings Figure 1 is a pole figure showing a double orientation of SrTi03 on textured Ni.

Figure 2 is a plot of the conductivity vs. temperature for two samples A, B, and YSZ from a reference.

Figure 3 depicts pictorially one fuel cell having an epitaxial layer according to the invention.

Figure 4 depicts pictorially one process for forming the fuel cell of Figure 3.

Figure 5 depicts an SEM micrograph of a layer of LSM that has been deposited on a layer of YSZ on sapphire showing a porous and columnar microstructure.

Figure 6 is a schematic side view of prior art crystalline coating showing the grain structure thereof.

Figure 7 is a schematic side view of a reduced grain boundary coating of the present invention showing the decrease in grain boundaries.

Detailed Description of Certain Illustrated Embodiments The present invention may be understood more readily by reference to the following detailed description of the preferred embodiments of the invention and the Figures. While the RGB thin films of the present invention can be deposited using a number of suitable techniques, the preferred method of deposition is CCVD as described in the above-mentioned U. S. Patent No. 5,997,956. Furthermore when a reducing atmosphere is required to avoid oxidation of the substrate or for other reasons, it may be desired to use the controlled atmosphere chemical vapor deposition CACVD technique as taught in co-pending U. S.

Patent Application No. 09/067,975. Other deposition techniques that provide a reducing atmosphere or a vacuum environment may also be used, however as is well documented in the CCVD patents, the numerous advantages of the CCVD and CACVD processes make them the preferred methods.

One deposition method for growing RGB thin films on nickel is CACVD to avoid oxidation of the nickel during the deposition process. Figures 1 through 3C of U. S. Patent No. 5,997,956 illustrate CCVD apparatus that may be used to deposit the epitaxial thin films of the present invention, and reference to these figures should be made with respect to the examples described below. It should be noted that these apparatus are only an example and other types of devices, including devices for other methods of deposition, may be used to produce the thin films described herein. As discussed therein chemical vapor deposition and powder formation may be achieved using thermal spray with fluid solutions including sub- supercritical and supercritical fluid solutions, to produce very fine atomization, nebulization, vaporization or gasification. The dissolved chemical precursors need not have high vapor

pressure, but high vapor pressure precursors can work well or better than lower vapor pressure precursors. By heating the solution fluid just prior to or at the end of the nozzle or restriction tube, the available time for precursor chemical reaction or dissolution prior to atomization is understood to be minimized. This method can be used to deposit coatings from various metalorganics and inorganic precursors.

The above described type of device was employed to form RGB thin films suitable for use as electrolytes in solid oxide fuel cells (SOFC), gas separation membranes, pyroelectric and dielectric material in electronic devices. By using CCVD, CACVD or any other suitable deposition process, RGB films having pore-free, or substantially pore-free, dense structure can be formed. For purposes of describing the RGB thin films of the invention, certain examples are provided below, however, it will be understood that the examples below are merely representative of the films of the invention, and are not to be deemed limiting, or an exhaustive list of such films.

Example 1 In the formation of fuel cells RGB, yttrium stabilized zirconium (YSZ) is deposited using CCVD or CACVD on a textured, preferably large grained, Ni substrate to form a YSZ electrolyte. On top of the electrolyte layer, a cathode such as LaSrMnO (LSM) is deposited.

The cathode can be deposited completely by CCVD or CACVD, or an interface layer from 50-200nm can be deposited using the CCVD or CACVD process and other deposition processes such as sol-gel, porous, or ceramic processing can be used to deposit the bulk of the cathode. After deposition of the cathode, the Ni can be etched off of the electrolyte, as the cathode will provide the necessary support layer. The anode is then deposited on the electrolyte to complete the fuel cell stack.

It is important to note that several different variations of the above method can be used to form the fuel cell. Additionally other electrolytes may be used such as BaCe03, and any textured or enabling crystal substrate can be used. Any suitable electrode materials may be used as well. The RGB crystal electrolyte provides increased performance over finer grained polycrystalline electrolytes by minimizing grain boundary effects. The etching step can be avoided by directly depositing the electrolyte to an electrode. For example, NiO and YSZ powders can be pressed into pellets and sintered to yield dense discs for the anode. The electrolyte is then directly RGB deposited onto these dense discs using CCVD or CACVD.

Heat treatment of the coated discs in a reducing atmosphere reduces the NiO to Ni and creates porosity in the anode. Cathode (e. g. Ag) paste is then applied to the electrolyte by screen printing or brushing and heat treatment is used to remove the organic portion of the paste leaving a porous metal layer for the cathode. The anode discs can be dip coated with additional anode material prior to CCVD coating of the electrolyte.

Fuel cells can be constructed with electrolyte thin film and porous electrodes on each side of the electrolyte layer. The process will most likely involve not only the CCVD technique but also other coating/forming methods. In general, the electrolyte material, say YSZ, will be deposited onto a smooth lattice-compatible, large-grained (grain sizes >10pm, preferably >50, um, the larger the better, even >lmm) substrate. A dense, pore/grain-boundary free or grain-boundary-reduced, YSZ film of 0.5 to five micron neighborhood is then deposited. Consequently, the cathode, say LSM, will be applied onto the YSZ layer via CCVD or by screen printing etc. The substrate will then be removed by etching, reducing or simply melting away depending on the property of the substrate. Finally, the anode, perhaps Ni-YSZ, is formed on the exposed YSZ surface by the CCVD process or by depositing porous YSZ via the CCVD technique and filtrating with Ni (or NiO) slurry. Details in this process development also involve the use obi-layer electrolyte (ceria/YSZ), or creating a rougher/high surface area microstructure of YSZ on top of the dense YSZ layer Figure 3.

Porous LSM layer can be deposited first by growing a thin LSM layer (in nanometer range) on YSZ via CCVD and then build up the thickness by other coating method, in order to further reduce interface resistance. A strong LSM layer is needed because after the removal of the substrate it acts as a supporting layer during the processing. After the removal of the substrate, porous Ni-YSZ layer of 10-30 um thickness will be deposited onto the exposed surface. This anode layer will be thin to maximize efficiency and minimize the polarization, resistance and increase gas diffusion.

It is generally recognized that increasing the TPB (three-phase-boundary between the gas, electrode and electrolyte) line length should increase the reaction rate in fuel cells. One can increase the TPB line length by using a porous composite electrode (cermet e. g. Ni-YSZ, Ru-YSZ). The electrode is required to be porous and electronically conductive in order to facilitate the transport of gaseous species or ions to the electrolyte while allowing for the collection of electrons from the associated electrochemical reactions at the interface. It is also

preferred because the contiguous electrode layer has fine pore sizes near the thin electrolyte to stabilize and support.

The use of CCVD to increase conductivity by reducing film thickness and using RGB crystalline thin films as the electrolyte in SOFCs has been accomplished. To test the feasibility of producing these electrolytes, YSZ films of 480 and 410 nm thickness were deposited onto sapphire substrates via CCVD at-1050° C. The YSZ films were deposited onto sapphire substrates at a deposition rate as high as 10.8 pm/hr at a 1400° C flame temperature and at 2.5pm/hour at a 1150°C deposition temperature. At these flame temperatures, the substrate temperature is about 900-1100° C and below 800°C, respectively.

The samples were tested for impedance spectroscopy. Fully stabilized zirconia with 8 mole percent (m/o) yttria (8YSZ) will be used as the main electrolyte material for deposition studies and process development. It is by far the most common electrolyte in SOFCs for its adequate level of oxygen-ion conductivity and its stability in both oxidizing and reducing environments. The use of fully stabilized phase is preferred to yield maximum conductivity and avoid the problems of phase transformation associated with partially stabilized zirconia.

After the electrodes consisting of Heraeus C-1000 silver paste and silver lead wires were fired onto the samples at 800° C for 10 minutes, these two samples were placed in a furnace with lead wires coming outside for applying a test signal. The frequency was swept from 2x 107 Hz to 0.1 Hz while acquiring the impedance spectra at 600,675 and 730° C. The resistance associated with the YSZ films was then converted to conductivity, and normalized for a standard geometry. The calculated conductivity values are illustrated in Figure 2 along with the data from Reference 15. The figure is a Plot of the Conductivity (in Q'cm') vs.

Temperature (1000/T, K1) for Samples A, B, and YSZ from Reference 15. The resulting high conductivity for CCVD deposited YSZ films can be explained by the reduction of grain boundaries in the film. A reduction of grain boundaries greatly reduces the interface and boundary resistance, and, hence, increases the conductivity. Sources of deviation, which are minor, are the conduction across the surface of the YSZ film instead of through the film, which should be minor since sapphire is not conductive, and the inaccurate measurement of the sample geometry. One should note that the configuration used here for testing can yield data on the YSZ film but the conductivity data is not directly applicable to fuel cell application due to the substrate. After examination by SEM, the films appeared to be dense

and uniform with no observable grain boundary or pores. The XRD pattern of the YSZ films on sapphire showed a preferred orientation of (111). An XRD pole figure pattern of the (111) peak was acquired for a YSZ film deposited onto a-axis sapphire via CCVD. The intensities are less than one, which is less than 0.5% of the maximum, except at the four 45 degree (p locations (90° to each other), and a very minor peak at the origin of the pole figure. This pole figure indicates the high degree of epitaxy in the YSZ film.

Oxide electrolyte materials including yttria-stabilized zirconia (8 m/o yttria, 8YSZ), ceria (Ce02), yttrium-doped ceria (YDC), yttria-stabilized bismuth oxide (YSB) and samarium-doped ceria (SDC), have been deposited on substrates using CCVD processing.

Strontium doped lanthanum manganate (LSM), a commonly used cathode, has also been deposited by the CCVD technique. Deposition parameters were adjusted to achieve porous, columnar structure for LSM films. Yttria-stabilized zirconia-based solid oxide fuel cells were constructed using Ni-YSZ cermet as the anode, YSZ as the electrolyte and silver as the cathode. Ceria films were deposited on both fused silica and sapphire substrates. Highly crystalline films of cerianite phase were produced at as low as 700° C flame temperature. The XRD patterns indicated a (200) preferred orientation and the degree of epitaxy increased as the deposition temperature increased. YDC, YSB and SDC thin films were also deposited onto sapphire substrates to exploit the feasibility of the CCVD process in producing ceria- and bismuth-based electrolytes for SOFCs. Highly crystalline and dense films were produced using CCVD. By adjusting the precursor solution concentration, deposition temperature, and atomization settings, adherent and columnar LSM cathode layers were deposited on sapphire substrates. The film thickness of the deposited layers included layers of approximately 420 nm. SEM micrographs of YSZ thin films deposited by the CCVD processing at 1350° C showed dense films with little or no grain boundaries, however white spheres indicative of foreign particles provided for contrast were identified. The films had an undulating structure achieved by changing degree of atomization, and a columnar structure was also achieved.

Additionally, YSZ (111) XRD pole figures indicated strong epitaxy on a-plane sapphire; and Ce02 (111) XRD pole figures indicated strong epitaxy on (100) lanthanum aluminate. Both of these materials have been RGB formed on nickel.

The main candidate electrode materials will be Ni-YSZ and La doped strontium manganate (LSM). The materials will be either deposited onto electrolyte layer via CCVD or in combination with other methods. An example of porous LSM is shown in Figure 3, while figure 4 pictorially describes the method for forming the porous LSM.. Another example for combining coating techniques could involve an initial deposition of a porous layer of electrolyte material over the dense electrolyte using the CCVD process. After depositing the porous layer, the next step is to infiltrate the electrode material into the pores to form a contiguous porous electrode while maintaining a long TPB line length.

Example 2 In this example, LSM was deposited on a-plane sapphire using CCVD. The precursor solution comprised 0.21 g Mn-2ethylhexanoate (2eh) (6 wt% Mn), 1.96 g La-2eh (2 wt% La), 0.97 g Sr-2eh (1.25 wt% Sr), toluene to 10 ml and 60 g propane. This solution was fed at a rate of 3 ml/min. for a total deposition time of 30 min. 2.42 amps of current were supplied to the needle with 3500 ml/min. of tip oxygen. The flame temperature was maintained at 1200- 1400 degrees C. In Figure 5, the SEM micrograph of LSM on sapphire shows a porous and columnar microstructure. The porosity of the electrode layer must be sufficient for the transport of gaseous species or ions to the electrolyte while allowing for the collection of electrons from the associated electrochemical reactions at the interface.

Example 3 In this example, YSB was deposited on a-plane sapphire using CCVD. The precursor solution comprised 2.88 g Bi-2eh (8.5 wt% Bi in xylene and further diluted with toluene to 2 wt% Bi), 0.08 g Y-2eh (diluted with toluene to 0.69 wt% Y). This solution was added to toluene for a total volume of 10 ml, and then added to 60 g propane. This solution was fed at a rate of 3 ml/min. for a total deposition time of 29 min. 2.50 amps of current were supplied to the needle with 3300 ml/min. of tip oxygen. The flame temperature was maintained at 1200 degrees C. Tip oxygen was 60 psi (with no hydrogen or argon).

Example 4 In this example, YDC was deposited on a-plane sapphire using CCVD. The precursor solution comprised 1.17 g Ce-2eh (12 wt% Ce in 2-ethylhexanoic acid and further diluted with toluene to 1.8 wt% Ce), 0.22 g Y-2eh (diluted with toluene to 0.69 wt% Y). This solution was added to toluene for a total volume of 14 ml, and then added to 51 g propane.

This solution was fed at a rate of 3 ml/min. for a total deposition time of 21 min. 2.76 amps of current were supplied to the needle with 3500 ml/min. of tip oxygen. The flame temperature was maintained at 1350 degrees C. Tip oxygen was 60 psi (with no hydrogen or argon).

In the field of gas separation membranes, specifically for hydrogen selective membranes, hydrogen conducting perovskite ceramic oxide coatings such as SrZrO3 and SrCeO3 are deposited onto palladium. These films are dense, homogeneous, pinhole free and sub-micron thick. Hydrogen transport rates through these composite membranes at 300 °C are approximately 70 GPU's.

Example 5 In this example, SrZr03 was deposited on Pd using CCVD. The precursor solution comprised 2.19 g Sr-2eh (10 wt% Sr in 2-ethylhexanoic acid and further diluted with toluene to 1.5 wt% Sr), 0.912 g Zr-2eh (diluted with mineral spirits to 6 wt% Zr). This solution was added to 160 ml ISP (isopropyl alcohol or IPA), and then added to 75 g propane. This solution was fed at a rate of 3.13 ml/min. for a total deposition time of 38 min. 3.0 amps of current were supplied to the needle with 5930 ml/min. of tip oxygen (80 psi) and 1200 ml/min. of pilot hydrogen. The flame temperature was maintained at 1150 degrees C.

RGB dielectrics for capacitors can be formed from a wide range of dielectric materials. Using the CCVD process, RGB up to single phase, epitaxial (determined from XRD) SrTiO3, and BST films were successfully coated on (100) MgO single crystal substrates. Examples are described below. Further examples discuss RGB deposition of these materials on nickel substrates.

Example 6 In this example, SrTiO3 coatings were deposited onto MgO using the CCVD process.

The solution of the SrTi03 precursor contained 0.0452 wt% of Sr in the form of strontium 2- ethylhexanoate, 0.0449 wt% Ti (di-i-propoxide) bis (acetylacetonate), 13.3329 wt% toluene, 0.5828 wt% isoproponal, and 85. 9919 wt% propane. The constant flow rate for the solution was at 2.0 ml/min and for the tip oxygen 4000 ml/min at 80 psi. The deposition temperature as measured at the substrate front surface varied from 900 to 1100°C.

Example 7 In this example, BST coatings with an end composition of about Bao. sSro sTiO3 were deposited onto MgO using the CCVD process. The solution of the Bao. sSro. stro3 precursor contained 0.0146 wt% of Sr in the form of strontium 2-ethylhexanoate, 0.0420 wt% of Ba in the form of barium 2-ethylhexanoate, 0.0311 wt% Ti (di-i-propoxide) bis (acetylacetonate), 13.3774 wt% toluene, 0.0104 wt% isoproponal, 0.5023 wt% 1-butanol, and 86.0404 wt% propane. The constant flow rate for the solution was at 2.0 ml/min and for the tip oxygen 4000 ml/min at 80 psi. The deposition temperature as measured at the substrate front surface varied from 900 to 1100°C.

Example 8 In this example, BST coatings with an end composition of about Ba06Sr04TiO3 were deposited onto MgO using the CCVD process. The solution of the Bao 6Sro 4TiO3 precursor contained 0.0143 wt% of Sr in the form of strontium 2-ethylhexanoate, 0.0615 wt% of Ba in the form of barium 2-ethylhexanoate, 0.0355 wt% Ti (di-i-propoxide) bis (acetylacetonate), 12.6049 wt% toluene, 0.0118 wt% isoproponal, 1.5333 wt% 1-butanol, and 85. 7412 wt% propane. The constant flow rate for the solution was at 2.0 ml/min and for the tip oxygen 4000 ml/min at 80 psi. The temperature as measured at the substrate front surface varied from 900 to 1100°C.

Example 9 In this example, epitaxial BST coatings with an end composition of about Bao. 6Sro. 4TiO3 were deposited onto a Ni substrate using the CCVD process. The solution of the Bao. 6Sro. 4TiO3 precursor contained 0.0143 wt% of Sr in the form of strontium 2- ethylhexanoate, 0.0615 wt% of Ba in the form of barium 2-ethylhexanoate, 0.0355 wt% Ti (di-i-propoxide) bis (acetylacetonate), 12.6049 wt% toluene, 0.0118 wt% isoproponal, 1.5333 wt% 1-butanol, and 85.7412 wt% propane. The constant flow rate for the solution was at 2.0 ml/min and for the tip oxygen 4000 ml/min at 80 psi. The temperature as measured at the substrate front surface varied from 900 to 1100°C.

Example 10 In this example, SrTiO3 coatings were deposited onto Ni using the CCVD process.

The solution of the SrTi03 precursor contained 0.0108 wt% of Sr in the form of strontium 2- ethylhexanoate, 0.00604 wt% Ti (di-i-propoxide) bis (acetylacetonate), 11.8 wt% methanol

and 88. 1 wt% propane. The constant flow rate for the solution was at 2.0 ml/min, for the tip oxygen 2300 ml/min at 80 psi, for the pilot hydrogen 1925 ml/min at 15 psi, for shield argon 22. 5 L/min at 50 psi, and for shield hydrogen 1900 ml/min at 15 psi. The deposition temperature as measured at the substrate front surface varied from 950 to 1000°C. In figure 1, the pole figure showing the double orientation of the coating is shown.

Example 11 In this example, BST coatings with an end composition of about Bao. s7Sro. 43TiO3 were deposited onto nickel using the CCVD process. The solution of the Ba057Sr043TiO3 precursor contained 0.0029 wt% of Sr in the form of strontium 2-ethylhexanoate, 0.0061 wt% of Ba in the form of barium 2-ethylhexanoate, 0.0037 wt% Ti (di-i-propoxide) bis (acetylacetonate), 10.5887 wt% toluene, and 89.3986 wt% propane. The constant flow rate for the solution was at 5.0 ml/min and for the tip oxygen 4800 ml/min at 80 psi. The deposition temperature as measured at the substrate front surface varied from 925 to 950°C.

Example 12 In this example, BST, RGB coatings 400nm thick, with an end composition of about Bao. s7Sro. 43Ti03, were deposited onto polycrystalline alumina using the CCVD process. The solution of the Bao. 57Sro. 43Ti03 precursor contained 0.0029 wt% of Sr in the form of strontium 2-ethylhexanoate, 0.0061 wt% of Ba in the form of barium 2-ethylhexanoate, 0.0037 wt% Ti (di-i-propoxide) bis (acetylacetonate), 10.5887 wt% toluene, and 89.3986 wt% propane. The constant flow rate for the solution was at 3.0 ml/min and for the tip oxygen 4000 ml/min at 80 psi. The temperature as measured at the substrate front surface varied from 1090 to 1125°C. Patterned electrodes two microns thick were processed onto the coating to form a gap capacitor 1 mm wide with a 7 micron gap; electrical measurements were then performed. The structure had a capacitance of 0.5 pF, a dielectric loss tangent of 0.017, and a tunability of 20% of the capacitance with a 40V bias voltage applied. This result is just slightly lower than our results on single crystal coatings. An RGB form was confirmed through X-ray diffraction using a small (0.2mm) collimator and no sample motion. The gathered data on the area detector screen shows spots of limited grain orientations rather than a continuous peak that would be indicative of the finer grained material produced otherwise.

Figures 5 and 6 illustrate the structural differences between the reduced grain boundaries of the coatings of the present invention as opposed to the grain boundaries of prior art coatings. In figure 6, small grains 600 as well as larger grains 602 are illustrated. Even the large grains, however, have a larger height (on average) than their width. In the inventive coating shown in figure 7, while some of the grains 700 have a larger height than their width, the majority of the grains (such as 702) extend from the top to the bottom of the coating, and have a larger width than height. As previously discussed, a width to height ratios of 5: 1 is preferred, and even more preferred are ratios of 10: 1, 20 : 1 and even higher. The grain boundaries tend to be regions of higher energy level bonds, resulting in lower strength structures. Thus, by reducing the grain boundaries using the coatings of the present invention, the bonds tend to be at lower energy levels and the overall strength of the structure is increased.