The Reduction of Tungstic Oxide to Tungsten Citation Diack, Samuel Latta (1925) The Reduction of Tungstic Oxide to Tungsten. Bachelor's thesis, California Institute of Technology. doi:10.7907/MY81-ET78. https://resolver.caltech.edu/CaltechETD:etd-12102004-142411 Abstract No abstract. Item Type: Thesis (Bachelor's thesis) Subject Keywords: Chemical Engineering Degree Grantor: California Institute of Technology Division: Chemistry and Chemical Engineering Major Option: Chemical Engineering Thesis Availability: Public (worldwide access) Research Advisor(s): Unknown, Unknown Thesis Committee: None, None Defense Date: 1 January 1925 Record Number: CaltechETD:etd-12102004-142411 Persistent URL: https://resolver.caltech.edu/CaltechETD:etd-12102004-142411 DOI: 10.7907/MY81-ET78 Default Usage Policy: No commercial reproduction, distribution, display or performance rights in this work are provided. ID Code: 4930 Collection: CaltechTHESIS Deposited By: Imported from ETD-db Deposited On: 10 Dec 2004 Last Modified: 03 Oct 2019 23:39 Thesis Files Preview PDF (Diack_sl_1925.pdf) - Final Version See Usage Policy. 931kB Repository Staff Only: item control page

THE REDUCTION of TUNGSTIC OXIDE to TUNGSTEN TRES TIS by Samuel L. Diack In partial fulfillment of the requirements for the degree of Bachelor of Science in Chemical Engineering CALIFORNIA INSTITUTE OF TECHNOLOGY Pasadena, California 1925 THE. REDUCTION OF TUNGSTIC ACID TO TUNGSTEN The purpose of this investigation is to study some methods of reducing Tungstic oxide (WOz) to metallic tung= sten from the standpoint of commercial applicability. Previous Work: The manufacture of tungstic oxide has been investigated by Wolochow Undergraduate Thesis (C.1I.T.) 1924, and a meth= od developed for its production from scheelite. The con= centrated mineral is fused with soda ash and the mixture lixiviated with water. The soluble sedium tungstate which is thus formed is treated with HCl and the insoluble tung stic oxide precipitated. Wolochow worked out a method for the purification of this crude oxide and the purified pro« duct was used as the raw material for this series of experi- ments. There seems to be but two practical methods available at the present time for the reduction of the oxide. (1) Reduction by solid reagents such as carbon in the form of charcoal. For instance: 2WO0z3+ 30 = 3COp+2W {2) Reduction by means of gaseous reagents such as Hydrogen or volatile hydrocarbons such as would be found in illumining gas. For instance: WO0z + 3Hp = W + SHgO Tungsten melts at 3350 degrees C. and is the highest melt- ing element known today, Because of this extremely high melting point it is impossible to have a reduction reaction Ze take place in a molten mix. The reduction must take place between the particles of the oxide and the molecules or the particles of the reducing agent, whether the reduc«- ing agent is gaseous or solid. In either case it can be geen that to effect a complete reduction the oxide must be ground to extreme fineness. In cases where the solid re- ducing agent is used, not only must it be ground very fine» ly but the charge must be mixed intimately also. In this way only can complete reduction be hoped for, There are commercial processes in use today which repre= sent the two methods above named : A powdered charcoal method is in wide use in Europe and hydrogen has found gen= eral application in this country. In the former method the WOs is finely ground and mixed intimately with finely powdered charcoal. The mixture is placed in a closed clay crucible and heated to a temperature from 900 - 1000 deg. C. According to reports Chem, © Met. 23,696, (1920), the carbon and oxide mixture varies from an excess of 10% carbon to 100%. After roasting, the charge is cooled out of contact with air (usually under charcoal) to prevent the reoxidation of the heated metal. When it is cool it is finely ground and washed with water to remove the excess carbon. This process depends on the great difference in specific gravities of the metallic tungsten and the reducing agent. It has been stated, Chem. ®, Met. 23,696, (1920), that by this method a pro= duct containing 98.5% tungsten metal can be obtained. The carbon content is reduced to .3 to .8%. This metallic product is suitable for work where only moderate purity is required - such as the manufactur of high speed steels. The hydrogen reduction method undoubtedly yields a much purer metal -- there being none of the reducing agent left in the final product. This renders a metal of well over 99% purity which is well adapted for the manufacture of e~ lectric light filaments. This hydrogen method is used by the largest manufacturers of tungsten metal in the country = The Fansteel Products Co., and the General Electric Company. In the present investigation both methods were studied on a Laboratory scale. — THE CARBON REDUCTION METHOD The main objects in trying out this method of reduction were to determine the purity of the product with respect to the carbon content and to determine the efficiency of re= duction on a small scale. The methods of analysis used were comparatively simple. It was required to determine: (1) The amount of unreduced ‘03 (2) The amount of carbon in the product after washing. 4. (3) The amount of the loss of tungsten due to the washing.for the removal of the carbon. In the 1st and 2nd experiments, determinations were checked in duplicate as far as possibile, The method for removing the unreduced W03 consisted in boiling the reduc~ ed charge in dilute (about 6N) NHgOH, The solution was filtered after being washed thoroughiy with water, and the residue was used for No. 3 determination. The filtrate was evaporated somewhat to enable more convenient handling and to remove the excess NH4,OH before acidifing. While the solution of ammonium tungstate was still boiling, as excess of HCI was added and the heating continued until the yellow "Oz precipitate came down complstely. This was filtered off on a fine paper, ignited and weighed, The second series of runs used a Gooch filter which ob« viated the possibility of re-reducing the W metal by the paper. The WO thus obtained represented the unreduced oxide that was dissolved by the ammonium hydroxide. It was found later, that this did not accurately represent the total unreduced WO.. present in the charge. It has been found that WO; that has been strongly ignited, does not readily dissolve in NH40H. The addition of a small con- centration of NacCOg renders this solution complete. All previous runs were checked over, using this method of dis~ solving the unreduced oxide. 5 In nearly all experiments there was aifficulty in com pletely removing the unreduced oxide from the charge. In every case the metal was treated until no further oxide could be dissolved. In some instances this’ took five!.treat- ments with ammonia and Naz00s,. The only way in which a satisfactory precipitate of Wos could be produced, that is onewhich would stick well to a filter,, to concentrate the solution to a volume of about 100 cc. and acidify strongly with HCI while the solution was kept boiling vigorously. The determination of the second quantity listed above (i.e., the amount of carbon) was somewhat simpler as far as technique was concerned, An ordinary combustion train for the determination of carbon in a metal was set up, us= ing KOH in a Geissler bulb for the absorption of the lib~ erated COe, Pure oxygen was used to save time and in=« sure complete combustion. A temperature around 800 deg. C. seemed to be all that was necessary to produce a good oxidation of the metal. Weighed samples were placed in the combustion tube, the oxygen turned on and the tube heated. The run was continued until a constant weight was recorded on the absorbing bulb. Precautions were taken to insure the absence of any CO2:in the gas enter= ing the apparatus. Carbon content found in this manner was expressed in terms of the percentage of the fused charge. The determination of the tungsten loss on washing (the third determination) was more difficult as it was hard to get a standardized procedure which would reduce all determinations to a comparable basis. On account of this inaccuracy that must be encountered, a number of runs were made in order to get an idea of the loss. The sample aft= er being treated for the removal of the unreduced oxide was washed with water and the washings were filtered through the same filter that was used to remove the solid matter in the unreduced oxide determination. In all experiments the washings were continued until the mixture came clear after settling for three minutes, In most cases this took three to five treatments. It can be easily seen wherein lies the difficulty in making such a procedure as the above a standard one. There may be differences in the fineness of the roasted charge although care was taken to grind it as fine as possible. If ground too fine too fine a great= er number of washings become necessary owing to the siower rate of settling of the particles. The filter on which the lost tungsten and the excess car- bon were deposited was dried, ifmited, and weighed. The tungsten loss was reported as % of original metal. In the first and third determinations there was a possibility 7. of re~reducing the oxide to the metal. In every experi- ment however, the ignition was long enough and in the pre= sence of an excess of air so that re-oxidation after the reduction would take place and the residue would be the WOz. It can be seen that some error would also be intro=- duced in the tungsten loss determinations due to the fact that the excess carbon from the run is present. This has been compensated for by calculating the amount of ash that would be present from the excess carbon. The ash content of the chartoal was previously determined. Any inaccuracy that may arise from this source will be on the safe side for a commercial process. . 8) *onzd OF pasng “ano PSETTFap 3q OF DSH POYVO-LOTOO UMOUg 18°96 os g¢ O-8 *T *Zap OOOT*PUSPTAS AUTSA OToTzsued OF TTeyewW pooy “3no PeTT Fup eq Of pey pue pesny eszeug 08°66 23°0 Sv £Om- O° ST - § ‘Tf LON* SE eT *2Z38p OOOT = 2T ‘rors usy} aefLeud Zonpoud -eTqTfonzo JO w0440q 04 pesng *paey ATATeI peyeg T4°86 62°T 98 *T fon O*ST "28D OOOCT etqtonuzs oF uotsng LUBTTS -Besseu uefketT UT 3no eueo eSueug SOTOTzAVd OF TTS eH 92°96 89°? Of o- ett fom 0f3t 21 cO00T 006 OTF UBsyy BUTT TBysAdos eucw AOTUazpyUT i peysisu MoU UOTVEPTXO sovjung 4°T8 63°ST Sg 9- COM o@ Co «. RR ei et TU *9000T= SUTASBTS ST} “1IT UOTYBPTXO eodBsuNs gO yuUNOWS ZUSTIS ¥ 6°88 9L°FT 03 9- eT ZoK, O°ST OT SyLeuSY uoTyONpsy % °T WHOM qVonpoIrT Ut Sony “ON STQCL (°*SUTW) OWE L TYLNGNI HEX a es TBUD On uny 9 The above results derived from the first determination show that the reaction can be carried out practically to completion at the end of a 45 minute heating. It will be noticed however, that at the longer periods of heating the fusion becomes more bothersome. In all charges that were run over thirty minutes there was a fusion that rendered the recovery of a good part of the charge irrecoverable, It is evident then that in order to have an efficient com- mercial process we cannot use clay crucibles, as they low~ er the yield and are apt to go to pieces under a high tem= peratursé. As the runs grew longer there was observed an increas= ing preponderance of metallic particles and a smaller a= mount of the black powder that resulted from the shorter runs « This might furnish a good idea of the progrees of the reaction thought no accurate estimate could be made. In the last runs where the reaction was nearly complete there was Little of the black mixture to be seen, and the particles seemed larger sizes. There is some discrepancy in the first two values of the % of WOz in the product. It might be caused by a differ~ ence in surface oxidation or by a lump of Oz accidentally passing through the mortar uncrushed. 10. — are *SOTOT} ou7s*s ~ied OFT Tazow meg $8°O 7 OOOT ce | “0N-928 L *eTtTqTonus 9=9°S UO SZBTR OFTTSI SH! £6°O ¢ 006 oO: Lom-oz 9 7Tsphod — a suTs*yoOBTq {sets O=G°S “Ty ued OTTTeyeu oN —89°0 t 008 so | fom=-se g wotpyonpea [ u eeeees pene sood 04 oenp s0T[090 O0-S°S entqtfk{peq pesny 9L°O T QOOT os | £ou-s2 4 SFT -[BY7OU- SOsse Te} 0-8°S \-sfkio e84et - uotsng T38°O g OOTT OF SOn-SS 3g “O00T _ | an Soto tyasa “ _ OTTTezou mez - aep g-8 °F “Kod yOeTq *dwey Moy Te*T % “a 006 SP ZOM-0F T Syieieg “poag uty “sZayzysea fo “sqaey ” ~ euty eFaeyg # uny & TIadvi ll. The values of the carbon content, which were determined according to the second method of analysis already listed, show conclusively that the carbon content is not merely a function of the number of washings, but that it depends on other factors as well, If it were the case that the number of washings alone determined the purity of the product with re= spect to carbon, we would expect a decrease in the percentage carbon with an increase in the number of washings. It is probable that a great deal depends on the extent of the reduc~ tion of the charge and consequently the state of aggregation of the metal, The effect of increasing the carbon in the entering charge may be seen plainly by comparing the first run with one of the others. Too great an extent of carbon in the origi- nal mixture makes it hard to remove from the product. There is therefore a ratio of carbon to WOz in the original mix= ture which will give the most economical results from the point of view of carbon content and efficiency of the reduc~ tione 12. *sSuTysem WOus Cox Jo *4dd ~g° 4ustts Ard, *&THOTND D3T329S | evBty, _3 Sv eof Tt eT ‘1eded f84TTz uO AOSTqQ STlItT 2-8°T *STyotTnb £1oA petites ge ? S¢ SOn=ST ST : 3-8°T uoqueo ss90xe yon yoy ge oe COM=ST rau ° —_ 3-8 °T Aqaots £100 pet33eg 06° L SZ LOM—ST tt *isded 164 [ —_ : 0-8°T “TES uo seToTysed OFT TH On ee°T 9 0g Cou-ZT| OT , O-8°T A£THOTND psti3es 86°0 g St LOm=8T 6 O-G°S geded sey 1Tts uo .yo8Tg 1s°o s 0¢ LON=SS 8 o-S°S 92°O' v GS SON-SS 4 ; 5-373 AtTmotTs ATUL p9Tt7179E¢g ¥P°O +4 o¢ Som-3°s 9 — *PeaAcuied. SUT . “USBM USYM AwEeTS YOU UOTYNTOS Jeo°g *ATMOTS peT339S SOTOTpueg T3°t Tt o¢ Com-e°s v *X{TMOTS PST475S 0-0°S *BUTISBTS OTI4TT OT*T g se | fom=9°3 g — *Syrewey “48OT ene ysey # = SWTL., esteyg . y UNY *¢ FTidvh 13, The results above show that it is perfectly possible to run this process and keep the metallic losses well under 1%. Little can be gathered from this data besides this fact, as the figures themselves show no reguler variation either with the number of washings, the time of heating, or with the con= stituents of the charges: CONCLUSIONS: The results obtained show that it would be possible to put such a process as the carbon reduction method in commercial use and obtain tungsten metal economically, but not necessarily profitably. The reaction would be made to take place almost completely if the charge was heated for 55 minutes at a temperature above 1000 deg. C,. The experiments also show that it would be impossible to use the ordinary clay crucible under these conditions, as their reactions to sustain-= ed heat causea loss of the product. Future experimental work only will show the best material for the reduction vessel, There are numerous possibilities to choose from although all of them are far more expensive than the clay crucibles. if this reaction were to be put into commercial use it would be necessary of course to use larger apparatus. A larger com=« tustion chamber and larger crucible would be necessary and al- so provision would have to be made for the removal of the fused charge without allowing it to come into contact with the air, 14. On a large scale this would be necessary to prevent the par- tial reoxidation of the hot metal. An arrangement could be effected whereby the crucible is removed directly to a bed of charcoal and buried therein till cool, As to the carbon content of the product, the results are so consistently low that undoubtedly a commercial apparatus could be designed ta keep the carbon content as low as it was in the experimental runs. Ordinary settling tanks might be satisfactory for this purpose, Experiments would have to be run however on the commercial apparatus to determine the fine= ness of the grinding and the time of settling. Such determin= ations could not be made satisfactorily a laboratory scale as they would depend a great deal on the shape and the size of the apparatus used. The determinations of the metal loss in the removal of the carbon show that it is a very economical method of remove al. It should be kept down to 1% or less if the proper con= citions of grinding and settling are maintained. If ina commercial plant it was found that the washing loss was becom= ing too high the washings could be filtered and dried and the coke returned to the charge. In this way the loss could be reduced materially although the cost of such a filtering and drying might be excessive in comparison with the value record« Ode In any case it ought to be perfectly possible to. procuce 15. a product of sufficient purity for alloy steels. The following is an estimate of the cost: 100 #WO3 @ $.604(Wolochow cost of production) $60.40 15# powdered charcoal @ $.06(current price) 290 Total cost of materials $81.30 Assuming a 97% yield (including reduction losses and washing losses) we would get 76.8% of tungsten metal. This would have a value (current prices) of .96 per #, Value of product $73.00 The difference between cost and value of product given above is $11.70. Out of this must come the production costs and overhead expenses. | With the comparatively small difference in cost and value of product given above it is possible that after sub~ tracting the cost of production there would be little or no profit remaining. The increase in value of the materials represents only 19% of their cost. This is the profit that would be derived if there were no production costs. To put this process on a profitable basis, would re~ quire a very efficient plant and an axtermely low overhead, It would be almost necessary to produce the tungstic oxide at the same plant in order to cut down the handling charges. I would not recommend the carbon method as being profitable on a large scale under the existing prices. A rise in the price of tungsten metal would perhaps render the process work- able. 16. THE HYDROGEN METHOD This method is the one used to a large extent in this country by the Fansteel Company, and the General Electric Company. The reaction is carried out in an electric furnace ata temperature of 1100 - 1200 degree C. The oxide is placed in shallow layers on nickel dishes and placed in the quartz tube of the furnace. Hydrogen is preheated by pess« ing it through a jacket surrounding this tube and then passed over the charge. The excess He which does not react with the oxide is dried by passing through CaCI2 and recirculated through the apparatus. The run is continued by the Fan= steel Company for seven hours, at the end of which time the furnace is cooled and the charge removed. The product re=# — sulting from this method is a fine grey powder of remarkable purity. EXPERIMENTAL WORK A number of shorter-runs were made to determine the ex- tent of the reduction of the oxide. The apparatus used was an electric muffle of the ordinary type equipped with a silica tube » Several runs were made using a nickel container: for the oxide but in every case the nickel crumbled and fell in pieces on removal of the charge. It was so mixed with the tungsten that recovery of the latter was impossible. Follow~ a ing this, humber of runs were made using a porcelain combustion 17. boat. Two of these were filled with Woz and placed in the furnaces Hydrogen was passed through to sweep out any air that was in the apparatus and the current turned on, A tem perature of 1100 deg. was used for the runs as the heating elements would not stand a higher temperature. The results for a few of the more representative runs are given in Table 4. ‘In all cases the procedure was the same. TABLE NO. 4 Run No. Time ZReduction Remarks 1 1% hrs.| 50.8% Front boat showed 59.5% af reduction 2 1 hr. 42.4% Grey powder ~ little cak- 3 2 hrs. 62.7% Good metallic lustre 4 3 hrs. 97.9% °° Fine homogenous grey pow= der. Metallic lustre. The product resulting from the above method was free from carbon and contained nearly 100% tungsten. These re= sults show that a complete reduction could be effected at this temperature and under these conditions in less than sev-= en hours. Of course in this reaction a great deal depends on the method of placing the charge in the furnace. The oxide should be in as thin a layer as possible so as to ex~ pose the greatest surface. The charge shrinks markedly L8. during the reduction and the final volume is less than half of the original. I would not recommend the placing of the oxide more than ¢ inch deep, as the time for reduction will be increased far out of proportion to the amount of product. It would be desirable to find some material that would stand this high temperature well and yet have a minimum amount of breakage. Some high temperature alloys such as a ferro=- molybdimum would make perhaps the most satisfactory containers for the charge. Porcelain is not practical for this purpose | as the breakage is too high. Where hydrogen is available in a large quantity it might be profitable to operate this process. The advantages over the carbon method are obvious. There is no treatment necess~ ary to remove the excess reducing agent and there is therefore less handling. The heating costs would run about the same as those of the carbon, but would vary widely according to the location. Several runs were also made using illuminating gas as the reducing agent but no satisfactory results were obtained e The reduction to metal did not proceed to any marked extent at the end of four hours and the product was contaminated with carbon resulting from the cracking of the hydrocarbons in the illuminating Base There was, however, a large proportion of one of the lower oxides of tungsten formed which gave the pro- duct a deep blue color when water was added. The blue 19. particles were so fine as to be unfilterable, Whether or not a gas having a higher proportion of hy= drogen to carbon would effect this reduction without the for- mation of carbon remains to be seen, but as long as there are plants putting out an excess of hydrogen (i.e. an electroly~ tic oxygen plants) there is little need of using illuminating gas for a reducing agent. Suggestions for Future Work. In connection with this problem there has arisen in my mind several possibilities, as far as future work is concerned, Some of them may be carried out on a laboratory scals while others deal with the large scale production. The possibility of using some other gas such as natural or illuminating gas for a reducing agent seems possible should the cost of hydrogen become prohibitive. To be sure, the re-~ sults obtained in this work Were using one:: of these agents (illuminating gas) did not meet with very marked success. However there still remains the possibility of trying a method of pre-cracking the gas (which is really made up of a wide variety of hydrocarbons) into simpler compounds having a smaller proportion of carbon to hydrogen. Undoubtedly some sort of a preheating apparatus could be designed which could effect this cracking and remove the free carbon formed. This also suggests the possibility of lampblack as a by~ 20. product in the manufacture of tungsten. A natural gas containing a large percentage of methane, CH4, might be suitable, although the known properties of me~- thane seem to show that reduction would be next to impossible since methane is a very stable compound. Heating with strong oxidizing agents has little or no effect on its decomposition even at high temperatures. The use of carbon monoxide is another possibility --= pos- gibly there is another use for blast furnace or producer gas here. At any rate I think it would be worth while to test out the possibilities of a pure CO or a CH4 mixture. Little or no idea of the behavior of CO with VO3 can be gained from the reduction with carbon, as the latter reaction is evident- Ay ~ one where all the oxygen for the COzg comes originally from the oxide. In the reduction of many substances it has been found that the reduction is really due to the CO in the reaction and not directly to the carbon. This case is found to be true in the blast furnace production of pig iron. The use of catalyists has never been touched on in con=- nection with this reaction. Little is known of the adsorb- ing properties of either metallic tungsten or of tungstic oxide, and therefore little is known of the reducing reaction in the presence of cataljsts. 21. The formation of the blue oxide by the reduction with illuminating gas suggests the possibility of a new product. The blue was of a deep clear shade and was formed in a collo= dial form. If a suitable adsorbing agent could be found for this compound there might be a future in using it as a dye or a pigment for special purposes. Its deposition on the sides of a test tube giving a clear, even, color may indi- cate a possible use in making blue stained glass. A gradu= al shading into the yellow W6g could be produced by careful heating in the presence of a limited amount of air. It would be interesting to know the action of some of of the powdered metals such as aluminum on tungstic oxide. There is a possibility there of a new method of reduction which would give a satisfactory product for most purposese 226 BIBLIOGRAPHY Chem & Met. Vol. 22. 9, (1920) Engineering News Chem. & Met. " i U.S.Patents Chem. & Met. tt n British Patent Chem. & Met. British Patent Chem, & fet. J. T. E. C, Zeitschrift Fur GATLO TR FC TLL Es U.S.Patent British Patent British Patent Chem, & Met. bye PR byes se Chee a PLES UE 104,432, (1917) 22, 774, (1920) 16, 40, (1917) 1, 202, 534 = , 202,535, Oct. 1916 16, 225, (1917) 20, 595, (1919) 1, 297,000 = 1918 23, 696, (1920) 155,600 = (1921) 2, 7, 3, (1920) 11, 201-204 (1919) See also 1919 Jour. Sci.Chem. Ind, 88, (1914), 74 = 87 1,175, 693 Mar. 14, 1916. 17,603 Apr. 3, 1912. 23,475 Oct. 14, 1912 24, 374,(1921) I also wish to express my indebtedness to Dr. W. W. Lacey for his helpful suggestions. IGIXO WLSIMIL WOH NILSIMIL SO NOLMG&xHd LONPOMd PPYUSINA uYOS LHAWVHD MOTI wokua PADINT L. < = 4 enplssq I SDUTUSDAA YUOL BUIYSOY a “pe anPisaqd I SBI Y SOMA YUO 1 BUIYSOAA , [ saps “ef09 5 BIOUINY A } 42aXIW . | . AALOM /20D1/0Y PPAPFPMOY EOM