Patent Application
20070214857
Solid electrolytic capacitors are made of valve metals, which are metals such as tantalum, aluminum, niobium, vanadium and the like. For high reliability devices, tantalum is the preferred metal, and efforts to improve the performance of capacitors made of tantalum are highly desired. Miniaturization is one of the main technology drivers in the electronics industry. For capacitors, miniaturization is achieved by increasing volumetric efficiency, which is the normalized capacitance per volume or CV/cm3 or normalized capacitance per gram or CV/g. The capacitance (C) of a dielectric is given by:
C=ε0·ε·A/d
where ε0 is the permeability in vacuum, ε is the dielectric constant of the anodic oxide layer, and A and d are the surface area and thickness of the oxide respectively. Since ε0, is a physical constant and ε is a material property which is fixed by the dielectric constant of the valve metal, the only parameters that can be manipulated to enhance volumetric efficiency are area (A) and thickness (d). For practical purposes, the thickness of the anodic oxide film is set by reliability considerations. For a given voltage rating, a thinner anodic oxide layer will provide less resistance to dielectric breakdown leading to lower reliability. Thus, the only feasible means to improve volumetric efficiency is to increase the available surface area by increasing the specific surface area of the valve metal substrate on which the anodic oxide layer is formed.
The specific surface area depends on the morphology of the substrate on which the dielectric film is produced. For tantalum powder, considerable development has pushed the technology to exceptionally high CV/g levels. However, as reported in Y. Pozdeev-Freeman, “How Far Can We Go With High CV Capacitors”, T.I.C. Bulletin, No. 122, June 2005, pp 4-8, these high CV powders suffer from extremely rapid fall-off in CV/g with increasing formation voltages. As a consequence these powders are generally useful for only low voltage applications, and there is still a need for the development of higher CV/g tantalum substrates for solid capacitors rated in range of 35 to 50V range. A significant achievement in tantalum powder technology has been the development of powder having flake morphology. See J. Koenitzer, S. Krause, L. Mann, S. Yuan, T. Izumi and Y. Noguchi, “Tantalum Flakes—Powders for High Reliability Electrolytic Capacitor Applications”, presented at the International Symposium Tantalum and Niobium World, October 2006; J. A. Fife, “Improvements to Volumetric Efficiency”, T.I.C. Bulletin, No. 81, March 1995, pp. 5-8. Because of their structure, flakes have a higher surface to volume ratio than nodular powders. The flat surfaces can provide more contact area between particles resulting in better inter-particle bonding. Also the reduced curvature of flakes lowers the stresses in the oxide layer particularly at higher formation voltages where the oxide is thicker. These last two characteristics help achieve lower DC leakage. Fibers, particularly flat fibers which are essentially two-dimensional structures, should have similar properties to flakes.
The potential advantages of fibers have been known for many years, and several approaches were proposed for making fibers suitable for capacitors. As far back as 1972, Douglas patented a method for making fibers (U.S. Pat. No. 3,681,063), and capacitors from these fibers (U.S. Pat. Nos. 3,742,369 and 3,827,865). The basic approach involved sintering tantalum powder into a porous compact and impregnating the compact with a softer metal, such as copper, nickel, or aluminum that does not react with tantalum. Impregnation was accomplished by melt infiltration of the second metal. The solidified composite structure was drawn or rolled to elongate the tantalum particles to produce fibers. The matrix was removed by etching in a suitable acid resulting in a porous structure of elongated fibers.
Fife in U.S. Pat. No. 4,502,884 describes a method for making loose fibers from tantalum powder and capacitor anodes from these fibers. In this approach, tantalum powder was mixed with a second metal powder using sufficient powder so that the second metal forms the matrix surrounding the tantalum particles. The blend was compacted into a billet and the billet subsequently drawn to elongate the tantalum powder particles. The matrix material was removed by leaching in acid to release the tantalum fibers. Fife also described a method of making anodes by forming the fibers into a felt or mat structure (see U.S. Pat. No. 5,306,462). Fife emphasized the need to have short fibers approximately 400 μm in length and to randomly orient the fibers in order to preserve maximum surface area on sintering. While the benefits of using fibers are known in the art, what has not been fully appreciated is that since flat fibers or ribbon type fiber have greater surface area than round fibers of equivalent cross-sectional area, the thinner the fiber thickness the greater is the surface area as is shown in
In my previous U.S. Pat. Nos. 5,034,857 and 5,869,196 1 describe approaches intended for making continuous fibers. My earlier patented processes involved assembling a composite billet of solid tantalum rods in a soft metal matrix, and then drawing the rod to wire to reduce the size of the tantalum. Copper is the preferred matrix material since it is very ductile, has virtually no solubility in tantalum, and has deformation characteristics that are compatible with tantalum. At a suitable size, the wire was cut into short lengths and bundled together in a secondary billet making a multifilament composite. The composite billet was further reduced by extrusion, drawing, or rolling, or a combination of these methods. The process was repeated a number of times to achieve very high reductions, and produce very fine tantalum fibers.
A variation of the above processes is to draw the tantalum-copper composite until the fibers are a few microns in diameter, then flatten the fibers by rolling to produce a highly aspected, high surface area fiber that is a micron or less in thickness. The flattened fibers thus formed are thin ribbons that have many of the dimensional attributes of flakes and provide higher surface area per weight of metal than round fibers. A further advantage of making continuous fibers is that it avoids the inherent complexities of handling and pressing short fibers. The continuous lengths of fiber can be twisted or braided to form a fiber strip that holds the loose filaments together. Anodes can be stamped directly from the strip, thus eliminating the need to press powders. Since the fibers can be readily processed into strip, relatively thin sections can be made from which to stamp anodes.
The present invention provides an improvement over prior art methods for making electrolytic capacitors. More particularly, the present invention provides an improved method for making capacitor anodes by producing filamentary valve metal fibers by a co-reduction of valve metal filaments within a copper matrix by a combination of drawing and rolling. The copper matrix is then removed leaving valve metal fibers in the form of continuous flat, ribbon-like fibers that have a relatively high aspect ratio of width to thickness, typically of at least about 10 to 1, and as a result relatively high surface area. By producing the fibers in a bundled continuous strip form, they can be made into thin anodes without pressing, thus maintaining the high surface area through subsequent anode sintering and formation.
Further features and advantages of the present invention will be seen from the following detailed description of the invention, taken into conjunction with the accompanying drawings, wherein:
Tantalum fibers were produced as fine filamentary ribbons, which were made by a process of extrusion, drawing, and rolling of a multifilament composite following the teachings of my prior U.S. Pat. No. 5,034,857. The overall process is as follows:
In a preferred embodiment of the present invention, the process begins with pure tantalum rod or high purity tantalum rod having a small amount of impurities, e.g. Fe, Ni, Cr, Cu, Nb, Mo, Si, Ti, W. C or O. A plurality of tantalum rods 12 are assembled substantially parallel to one another, in a copper can. A copper nose and tail are welded onto the can to form a primary billet, and the billet is then evacuated and sealed. The can is then hot extruded to bond the copper to the tantalum and cold drawn to make a copper clad tantalum wire bundle following the teachings of my prior U.S. Pat. No. 5,034,857. Bonding of the copper cladding to the tantalum is essential to prevent oxidation or other contamination of the tantalum during subsequent processing. The resulting copper clad tantalum wire bundle was cut to length, and bundled and restacked into a second copper container, a nose and tail are welded in place, and the secondary billet is evacuated and sealed as before. The secondary sealed billet is optionally prepared for extrusion by hot or cold isostatic pressing in order to collapse any void space within the billet and to promote filament uniformity. After isostatic pressing, the secondary billet is machined to fit the extrusion liner, and the billet is then extruded and drawn followed by rolling to a final preferred thickness of less than 1 μm and preferably less than 0.5 μm.
After rolling to final thickness, the tantalum fibers are immersed in an etching solution such as nitric acid and water to leach the copper.
CV/g and DC leakage are characteristics that depend on the quality and morphology of the fibers. Capacitance is largely a function of surface area, but also depends on the packaging of the filaments in the anode body. To achieve high CV/g it is necessary to create a high amount of useful surface area, eliminate the very thin fiber segments that would be consumed during anodization, and package the fibers to maintain an open pore structure that does not close-off surface area during formation. DC leakage is largely related to surface chemistry, but is also affected by the regularity and uniformity of the oxide structure. To achieve low leakage, it is necessary to have a uniform amorphous oxide without irregularities or discontinuities caused by inclusion protruding through the oxide film or crystallization promoted by impurities at the metal oxide interface. An additional factor that can have a detrimental effect on leakage is inadequately formed neck structures which bond the particles of fibers together in a single network structure. Poorly formed necks will result in local hot spots due to highly resistive junctions causing a breakdown in the oxide particularly at higher formation voltages.
The present invention results in part from the realization that for a fixed volume of material, ribbon type fibers have more surface area than round fibers when the ribbon type fibers have a thickness equivalent to the diameter of the round fibers. Thus it is possible to produce higher surface area fibers by flattening round fibers. This greatly facilitates the production of high surface area fibers, since it is difficult to make very fine, submicron fibers by wire drawing without producing fibers that have highly irregular cross-sections. When used to make anodes for electrolytic capacitors, valve metal fibers having non-uniform cross sections lead to lower CV/g performance. We have discovered that flattening fibers by rolling produces a more uniform surface structure that results in more useable surface area and thus produces a capacitor that has higher volumetric efficiency.
The starting material was a rod 12 of high purity tantalum. The rod was vacuum encapsulated in a copper can 14, extruded and cold drawn to make a copper clad Ta wire. The wire was cut to length, bundled and restacked into a second copper container, and further reduced by drawing followed by rolling to final thickness. After rolling, the resulting Ta fibers were removed from the matrix by leaching the copper with nitric acid. While the drawing and rolling parameters can be varied to produce a wide range of fiber sizes and shapes, the particular deformation sequence and reduction scheduled used in this example resulted in fibers that were approximately 0.5 to 1 μm thick and 35-50 μm wide, and had a B.E.T. surface area greater than 0.300 m2/g.
To make anodes, the fibers were twisted and cut into pieces weighing approximately 50 mg and a tantalum lead wire attached by spot welding. The dimensions of the anodes were approximately 0.3×4×8 mm. Since the fibers are continuous, the length of the fibers forming the anode is equivalent to one of the planar dimensions of the anode. The anodes were sintered under vacuum of greater than 10−3 Pa (7.5×10−6 torr) for 10 minutes or 50 minutes at temperatures of either 1300° C. or 1500° C. The fibers received no other chemical or thermal treatment. The sintered anodes were anodized in a solution of 0.10 V/V % phosphoric acid at 80° C. and a current of 100 mA per gram. Samples were anodized to formation voltages of 100 V. 140 V and 180 V. Capacitance and leakage current were measured in a wet cell of 15 W/W % H2SO4. DC leakage current was measured at a potential of 70% of the formation voltage.
Capacitance values for 100 V. 140 V and 180 V formations are given in Table 1 and
DC leakage values are given in Table 2. As can be seen leakage decreases with increasing sintering temperature and sintering time. The values for the two 1500° C. sintering treatments are shown in
It is thus seen that tantalum fibers produced by a composite co-reduction process in accordance with the present invention have properties suitable for use in forming capacitor anodes particularly those used for higher voltage ratings. Like powders, the fibers can be produced in different sizes depending on the intended application voltage. However, unlike powders, the fibers are organized in a continuous filament structure which improves handling and packaging of the fibers into anodes. Further improvements in CV/g can be realized by producing a more uniform fiber structure, while improvements in DC leakage may be achieved by performing a deoxidation treatment or through optimization of the sintering cycle.
While the invention has been described in detail in connection with the formation of tantalum fibers for solid electrolytic capacitors, the invention also advantageously may be employed with other valve metals commonly used for forming solid electrolytic capacitors, in particular niobium, aluminum, vanadium and like metals and alloys thereof. Yet other changes may be made without departing from the spirit and scope of the invention.
This application claims priority from U.S. Provisional Application Ser. No. 60/783,329, filed Mar. 17, 2006.
| Number | Date | Country | |
|---|---|---|---|
| 60783329 | Mar 2006 | US |
