Background of the Invention

This invention relates to anodes for electrolytic cells for the production of aluminum, and specifically to a method to reduce anode stud corrosion which will result in a reduction in anode voltage losses, labor required to reset anode studs and stud maintenance costs and an improvement in anode and cell performance. A commonly utilized electrolytic cell for the manufacture of aluminum is of the classic Hall-Heroult ..design, utilizing carbon anodes and a substant ally flat carbon-lined bottom which functions as part of the cathodic system. The electrolyte used in the pro¬ duction of aluminum by electrolytic reduction of alumina consists primarily of molten cryolite with dissolved alumina, and may contain other material such as fluorspar, aluminum fluoride, and other metal fluoride salts. Molten aluminum resulting from the reduction of alumina is most frequently permitted to accumulate in the bottom of the receptacle forming the electrolytic cell, as a molten metal pad or pool over the carbon-lined bottom, thus acting as a liquid metal cathode. Carbon anodes extending into the receptacle from above,, and con¬ tacting the molten electrolyte, are adjusted relative to the liquid metal cathode. Clark collector bars, fre¬ quently of steel, are often embedded in the carbon-lined cell bottom, completing the connection to the cathodic system. Similarly the commonly utilized carbon anodes are physically and electrically connected to anode studs, most often of steel, which are suitably raised and lowered as necessitated by the oxidization of the carbon anode and the necessary renewal thereof.

The electrolyte contained in the electrolytic cell forms a solid crust where exposed to the cooler atmosphere above the electrolyte, which in t_ ^,,r n is covered with a layer of alumina for periodical enrichment

OMPI

of the electrolyte and thermal insulation of the bath in the electrolyte pot. The anodes, consisting of carbon, penetrate the alumina layer and the crust, extending into the electrolyte, for conduction of the electric current which maintains the electrolysis. The crust, and the aluminum oxide deposited thereon, nor¬ mally do not form a. gas-type seal around the circum¬ ference of each anode, due to rising gases and motion of the molten electrolyte. In addition, the crust is periodically broken for enrichment of the electrolyte with alumina.

The gases released from the electrolytic process, primarily a mixture of gaseous fluorides, carbon dioxide, CO2, and carbon monoxide, CO, penetrate the carbon anode through cracks and open porosity within the carbon anode. These gases can react with chemical components within the anode to form a corrosive gas such as CO + S - COS, carbonyl sulfide. The anode gas and/or gaseous products are corrosive to the anode studs supporting the carbon anodes and providing electrical connection thereto. The temperatures within the anode can range from 100°C or greater at the top of the anode to the temperature of the electrolyte 900 to 1000°C at anode lower surface. Thus the anode stud, normally an unpro- tected steel surface, is subjected to highly corrosive gases at temperatures which expedite corrosion and deterioration of such materials.

Although considerable effort has been expended to protect other components of -the electrolytic cell, such as the electrodes themselves, we are not aware of any satisfactory method to reduce anode stud corrosion. It is known that corrosion of the anode studs in a vertical stud Soderberg aluminum reduction cell contri¬ butes directly to increased stud maintenance cost and power consumption, as well as reduced metal quality and cell performance. The anode studs corrode, forming a

scale containing various forms of iron sulfide and iron carbide. It has been shown that one corrosive agent involved is carbonγl sulfide, COS, which forms in a reaction between CO and the sulfur in the anode carbonaceous materials. During stud pulling, i.e., removal of the stud for resetting to a greater dis¬ tance from the anode face, pieces of the scale remain in the anode, which in time are transferred qualitatively into the metal. It has also been observed that the iron content in the aluminum metal produced was a direct function of the sulfur content of the anode materials in Soderberg cell operation. It has previously been re¬ ported that low alloy steels corrode significantly less than ordinary carbon steels when exposed to a sulfur- bearing anode mass, although later results disputed this reported improvement. A steel stud coated with aluminum appears to be protected from attack by a fluoride-free sulfur-bearing anode mass. However, with the introduction of amounts of volatile fluoride, known to be present in actual anode gases, such aluminum coat¬ ing and stud material were heavily corroded.

Formation of a poor electrically conducting iron sulfide film on an anode stud increases the cell voltage loss in the anode and consequently increases the energy required to produce aluminum. The increased stud to carbon contact resistance produces local non-uniformity in the anode current distribution, which can initiate and/or enhance the formation of anode spikes, which can short-circuit through the metal pad causing severe local heating within the anode. Thus, it is desirable to prevent the formation of this scale or film. It is noted that such short-circuiting can result in the melting of several inches of iron from the stud tip, and the forma¬ tion of small metal globules of iron in the anode carbon. Development of a cost-effective, corrosion-resistant, electric-conductive stud coating would clearly help to reduce cell energy requirements, improve metal quality,

C PI

and permit the use of low-cost, high sulfur-content anode carbonaceous materials. In addition, anode stud maintenance would be reduced, consequently simplifying the stud resetting process, hence further reducing operating costs.

Summary of the Invention

It is an object of the present invention to provide a corrosion-resistant coating on the steel anode stud, which coating is electrically conductive and resistant to high temperature air oxidation and thermal shock. It is a further object of the present invention to provide a method for application of a protective coating to anode studs and other metal components of an electrolytic reduction cell. Such coatings help reduce the energy requirement for producing aluminum metal, reduce anode stud maintenance costs, reduce iron contamination in the aluminum metal, and maintain a more uniform anode current distribution, which could lead to improved cell current efficiency and simplify the stud resetting process. These and other objects of the invention are attained by application to the steel anode studs of a coating com¬ prising titanium diboride (TiB2) and/or similar materials, such as zirconium diboride, titanium carbide and zirconium carbide. Additives may be used to produce other desired coating qualities, such as molybdenum disilicide to improve resistance to thermal oxidation. Sintering aids such as rhodium or iridiu may be used to help reduce the porosity and improve the ^' strength of the coating.

Description of Preferred Embodiments The present invention relates specifically to the application of a corrosion-resistant coating to a VSS cell anode stud. However, this concept may also be applied to the studs of a horizontal anode stud cell, the metal holders of prebaked anodes, and other metal cell parts subject to corrosion. A suitable corrosion

0---

resistant coating has potential application wherever corrosion occurs and/or improved electrical contact is desired.

Conventionally, the anode studs utilized in a VSS cell comprise a low carbon steel material. It has been found, by experimentation, that when a corrosion resis¬ tant coating is applied to a conventional steel anode stud, improved results are obtained when a stainless steel sub-coating is also used. This prior coating re- duces thermal stresses and improves the bonding between the corrosion-resistant coating and the base metal. The stainless steel sub-coat may be applied in any conven¬ tional manner, such as by plasma spray, vapor deposi¬ tion, electric arc, flame spray, etc. Suitable other materials for utilization as the sub-coat or bond coat include chromium based alloys, such as chrome! , nickel containing stainless steel, such as Inconel, and other alloys which tend to reduce thermal stresses and improve the bonding between the outer coatings and the stud sub- strate.

It has been found that the corrosion resistant coating may be effectively utilized over the entire stud, or over the lower-most portion of the stud. Further, thickness of the corrosion-resistant coating material may be varied from 2 mils to approximately 100 mils. However, it is noted that a non-porous or impervious coating is most desirable. It is also noted that the coating may have a homogeneous composition and density, or have a controlled composition with a density gradient from outer-most surface to the portion in contact with the bond coating.

Suitable coating materials have been found to be titantium diboride, zirconium diboride, titantium diboride- molybdenum disilicide, and zirconium di bori de-πolybdeπu disilicide. Other materials found useful include titantium carbide, zirconium carbide, molybdenum disili¬ cide, and mixtures of these materials with any of the

metal oxides associated with non-consummabl e anodes in the patent literature. The top protective coating may be applied in any conventional manner, such as by plasma spray, vapor deposition, electric arc, flame 5 spray, etc.

It has been found that mixture of TiB2 ⁺ MoSι ^' 2 is the preferred coating material of the materials listed when applied using a plasma spray process.

Example I - Bond Coat Composition

10 Chemical corrosion tests were conducted to deter¬ mine which of the potential bond coats exhibited the best resistance to corrosion by sulfur at elevated temperatures. The sulfur corrosion resistance of the bond coat is not critical to the success of the coating

15 system; however, it is desirable for the bond coat to be corrosion resistant in case the top coat has or deve ops defects. The following illustrates that corrosion of 309 stainless steel is significantly less than other metals tested when baked for 150 hours at 1000°C in

20 anode paste.

Metal wt % Cr wt % Ni mils corroded Carbon steel 0 0 25 Nitro 60 18 8 25

333 B stainless steel 44 24 27 25 330 stainless steel 34 23 22 309 stainless steel 25 12 <1

Example II - Typical Plasma Spray Coating Procedure

A 309 stainless bond coat and TiB ₂ » ZrB ₂ , and TiB ₂ ' MoSι ^' 2 top coats were applied to 1/4 in., 1/2 in. and 1 in. 30 diameter low carbon steel test rods and tapered 4-5 in. diameter steel VSS stud tips, using a plasma spray process The coated test rods were used for laboratory tests while the coated stud tips were used in a pilot test using

production VSS aluminum reduction cells ( lOO amp line current). A micrometer was used to determine coating thickness.

Sample preparation consisted of degreasing with methyl-ethyl-ketone followed by grit blasting with 54 mesh grit (Al 2O3 ) _

The 309 stainless steel bond coat was applied utili¬ zing a plasma spray technique employing 400-800 amps with an argon plus 5 volume % H2 plasma gas, utilizing 309 stainless steel, -200 to +325 mesh, to achieve the desired coating thickness, typically 2-10 mils, pre¬ ferably 8-10 mils. The substrate was preheated to 150°C and the spray rate and cooling air/inert gas flow were adjusted such that a substrate temperature of 95-370°C was maintained, with a 95-150°C range preferred. Bond strength tests were used to help select the preferred operating parameters.

The operational parameters for the corrosion-resis¬ tant top coat, such as TiB2 _> involve the use of an argon plus 5 volume % H2 plasma gas operating at 400-800 amps utilizing an appropriate spray rate and air/inert gas cooling to maintain a sample temperature in the range 95°C to 370°C, with a preferred sample temperature less than 200°C. Successful coatings of each of the corrosion resistant materials over the bond coating were achieved. Preferred coating thickness is about 10 mils although a range of from about 2 to 20 mils is acceptable.

Example III - Carbon to TiB?- Contact Resistance

A hot-pressed bar of titantium diboride, 1/2" square, was baked for 24 hours at 875°C in anode paste, a mixture of coke and pitch obtained from the Martin Marietta VSS Aluminum Reduction Plant located at The Dalles, Oregon. Small pieces of baked anode carbon remained attached to the TiB2 when the baked sample was broken apart. The resistance of a carbon to Ti B2 to carbon

section of the test sample was compared to that for an equal length and cross section of pure anode carbon. The resistance for both measurements were 0.1 + 0.1 ohms. Accordingly, there is qualitatively no measurable contact resistance between the hot-pressed T B2 and baked anode carbon.

Example IV - TiB? /Stain! ess Steel /Substrate Resistance

A titanium diboride coating over stainless steel on a steel substrate was subjected to contact resistance measurement. The resistance of the steel rod was mea¬ sured utilizing the same procedure, absent the coating materials. The difference between the measured resistance for the coated and uncoated steel rod was halved to yield total resistance for the coating and associated interfaces. It was found that the typical total measured resistance for a 10 mil TiB2 coating plus a 2 mil stainless steel bond plus the TiB2/stainl ess steel /substrate steel inter¬ faces is about 4 micro ohms per square centimeter of coating surface area. In a VSS anode, the current density through the stud coating would approximate 1 amp per cm^, resulting in an estimated 4 x !0~ ⁶ volt drop across the stud coating. Such a low voltage drop is insignif cant compared to the 100 to 300 mV drop across the uncoated stud/carbon interface experienced commercially.

Example V - Thermal Shock Tests

Coated test rods were rapidly cycled between 900°C and 100°C to test thermal stress properties of the various coatings. In each cycle, the sample was heated in a 900°C furnace for 15 minutes in a nitrogen atmosphere, then allowed to cool in air for 10 minutes. With no stainless steel bond coat, the TiB2 coating started to crack after 10 heat cycles. The T B2 coating with a stainless steel bond coat exhibited no evidence of cracking after 14 heat cycles. The ZrB2 coating, with a stainless steel bond coat, had no cracks after 9 heat cycles. It is to be noted

that the small radius of curvature and faster cool-down rate of the test samples makes this thermal stress test more severe than would be experienced in real commercial anode operation. Further, there is a 2-3 week annealing time in a vertical stud anode to help relieve thermal stress, which annealing time is not present in the labor¬ atory test.

Example VI - Corrosion Resistance to Vertical Stud Soder- berg Anode Environment A test reactor was used to simulate the corrosive environment within a VSS anode. The anode environment reactor comprised a tube furnace surrounding a stainless steel reactor tube, into which were placed pitch coke plus 1 wt.% Atmolite (NaAlF.^), and carbon, with the coated portion of the test anode submerged in the carbon. Elec¬ trical connections were made to a constant current power supply and the tube furnace was thermally insulated. The Atmolite was added to the pitch coke to provide trace amounts of volatile fluoride, which is normally found in anode gases, since Atmolite is the compound which normal¬ ly vaporizes from a cryolite bath. Carbonyl sulfide (COS) was forced through the system to simulate bath fume penetration of the VSS Anode, at a concentration of about 50 times that found in typical vertical stud anode operational gases. Hence, the laboratory corrosion test represented an accelerated test condition.

Photographs of test rods before and after the 4-hour corrosion test indicate typical scale thickness of the uncoated section of the test rod to be from 100 to 200 mils. X-ray diffraction analysis identified FeS, Fe and S as the major components of the corrosion scale. In each 4-hour corrosion test, the diameter of the corroded steel test rod, not including the scale, was typically reduced by about 50 mils, which represents a 35 wt.% loss of the metal rod, in uncoated sections. However, the coated sections of the test rods showed no increase

in diameter following the corrosion tests for rods coated with either TiB ₂ , ZrB ₂ or TiB ₂ ^* 10 wt.% MoSi ₂ - In several tests the coated rod was polarized anodically to give a current density through the coating similar to that for a stud in a VSS anode cell (1.0 amp/cm ² ). The TiB coating has a slightly more metallic appearance following the corrosion test with current than following the tests without current. The ZrB ₂ and TiB2 ^* 10 wt.% MoSι ^* 2 coatings were dimensional ly unaffected during the corrosion test, although both coatings developed a white-grey surface discoloration, with ZrB2 being more discolored. There was no sign of spalling or cracking of the coatings as a result of the corrosion test.

Example VII - Coating Resistance Following Corrosion Tests Qualitative coating resistance measurements were made before and after corrosion tests to determine if the test had s gnif cantly changed the coating's electrical proper¬ ties. The relative measurement for each coating was made by clipping two clip- leads of a digital ohm-meter 1" apart on the coated section of the test rod. Prior to the cor¬ rosion test, the observed resistance for the TiB2_ ZrB2 , and TiB ₂ ^* 10 wt.% oSi coatings were 0.5 + 0.1 ohms. The TiB2 and TiB2 ^* 10 wt.% MoSi2 coatings showed no increase in resistance following the 4-hour corrosion tests. How- ever, the ZrB2 coating resistance increased by a factor of 20 to 50. The resistance of a T B2 coating which had been partially oxidized prior to the corrosion test dropped from around 2,000 ohms to 0.-8 ohms following a 2.5-hour corrosion test.

Example VIII - Surface Oxidation Test

A simulated cool-down of the stud tip after pulling was achieved by the controlled removal of the test sample from a vertical tube furnace. For each controlled cool- down cycle, the sample was first held at 900°C for 15 min- utes in a nitrogen atmosphere, then with air owini

the furnace, the sample was slowly withdrawn from the fur¬ nace such that the sample temperature dropped from 900°C to 500°C in 8 minutes, at which point the sample was removed from the furnace and allowed to air-cool for an additional 7 minutes. The oxidation results are illus¬ trated in Table 1. The relative coating resistances were measured as described in Example VII, and the percent increase in resistance is given by the formula:

(observed resistance) - (initial resistance) x 100 = % increase

(initial resistance)

The air oxidation of the TiB2 coating is improved by the addition of MoSι ^* 2« However, the MoSι ^* addition must be kept to a minimum to avoid a degradation of the coating thermal shock resistance. Tests have indicated that the MoSι ^" 2 addition to the TiB2 coating material should be in the 0-10 wt.% range, although higher oSι ^' 2 concen¬ tration may be acceptable. The preferred range for the MoSι ^' 2 concentration is 5-10 wt.% for preventing air oxida¬ tion of the coating.

TABLE 1

STUD COATINGS OXIDATION TESTS

Bond Top Number of Increase 1n Observati ons Coat Coat Cycles Coating Resistance

SS TiB ₂ 270% Slight white-gray surface dis¬ coloration.

3 1200% White-gray surface. 5 3900% White-gray surface

SS TiB2+10% 0% ' No observed change.

MoSi ₂

140% Slight white-grey surface dis¬ coloration.

900% White-grey surface with scattered dark spots of original surface color; change in coating diameter

Example IX - Pilot Test of Coated Stud Tips

The lower 24 in. portion of 10 VSS studs (about 5 in. diameter) were coated with a 309 stainless steel bond coat and a corrosion resistant top coat utilizing a plasma spray process. The 309 stainless steel bond coats ranged from 7 to 9 mils in thickness. The top coats (3 to 5 mils thick) tested were composed of TiB plus MoSi2- The oSi2 content in the top coat ranged from 5 to 10 weight percent. The coated studs were monitored for four consecutive two-week stud cycles in production VSS anodes. Normal potroom precedures were used in setting and pulling the test studs. The studs were not cleaned between each two-week stud cycle. The pilot test data demonstrated the following benefits of coated studs: 1. The coating prevents corrosion of the steel stud in a VSS anode.

2. Where coating flaws existed, there was no under¬ cutting of the good coating.

3. Compared to an average uncoated stud in an equivalent anode location, the average coated stud carried 15-45% more current which indicates that the average electrical resistance in the anode area associated with a coated stud is reduced by 13 to 41%. An average 20% reduction in overall anode resistance is indicated when coated studs are used in the entire anode. For an VSS anode using all coated studs, the average anode voltage drop -would be decreased by 0.10 volts which would save approximately 0.16 Kwh per pound of aluminum produced in a typical lOO amp VSS aluminum producing cell.

4. The current variations between coated studs at a given location in different VSS anodes was less than that for corresponding uncoated studs.

5. When pulled, unlike the uncoated studs, the coated studs did not require cleaning to remove scale and other debris before being reset in the anode. 6. A coating life of 4 to 6 months is anticipated before recoating is required. 7. The stud crane operator estimated force required to pull the coated studs was equivalent to that for a normal coated stud. The examples have illustrated that the corrosion re¬ sistance of the materials utilized exceeds that of any coating or monolithic stud material used previously. Plasma spraying these coatings on the tips of the VSS cells anode studs represents a simple, convenient, and economical method to improve stud life and reduce anode voltages, anode current non-uniformities, iron contamina¬ tion of the metal pad, stud resetting costs and stud maintenance cost. Improved corrosion resistance will permit use of more readily available, lower cost, higher sulfur content carbon materials in a VSS cell anode. Although the corrosion resistance coating of the present invention have been applied by plasma spray tech¬ niques, it is clear to one of ordinary skill in the art that other alternative methods of application would also be acceptable, such as vapor deposition, electro-deposi¬ tion, flame spraying, chemical deposition, sintering, and conceivably press fitting of a formed sheet material. The area to be coated may range from a few inches of the stud tip to the entire stud,-'while coating thickness may range from 2 mil to 100 mils. The corrosion resistant material may be composed of titanium diboride, zirconium diboride, titanium carbide, zirconium carbide or any refractory metal boride or carbide or a mixture of these materials. Additives may be added to obtain additional desired coating properties. A bond coat may be required to help bond the outer corrosion resistant coat to the stud.

OMPI

./> \v ^* ι?o