A Method for the Recovery of Alumina and other Valuable Constituents from Clay Citation Hess, Ben Ewart (1925) A Method for the Recovery of Alumina and other Valuable Constituents from Clay. Bachelor's thesis, California Institute of Technology. doi:10.7907/vqd3-jm20. https://resolver.caltech.edu/CaltechTHESIS:08062024-224911379 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): Lacey, William Noble Thesis Committee: Unknown, Unknown Defense Date: 1 January 1925 Record Number: CaltechTHESIS:08062024-224911379 Persistent URL: https://resolver.caltech.edu/CaltechTHESIS:08062024-224911379 DOI: 10.7907/vqd3-jm20 Default Usage Policy: No commercial reproduction, distribution, display or performance rights in this work are provided. ID Code: 16615 Collection: CaltechTHESIS Deposited By: Benjamin Perez Deposited On: 08 Aug 2024 00:02 Last Modified: 08 Aug 2024 00:03 Thesis Files PDF - Final Version See Usage Policy. 7MB Repository Staff Only: item control page

A !fflHOD FOR THE RECOVERY OF .ALUMTI'lA MID OTHER V.ALU.AJ3LE CONSTITtm-i"TS FROM CLAY Thesi• by Ben E. Hess In partial fulfillment of the re~~irement• for the Degree of Bachelor of Science in Chemical Engineering. CALIFORNIA INSTITUTE OF TECHNOLOGY Pasadena, California 1925 -l- Whenever there is a cheap ·raw material available in large quantitiea, and it contain• chemicals which are in themselvea quite valuable, -the natural tendency ie to attempt to work out a proceH tor their recovery. Attempt• to recover alumin\lm and other valu- able pro·d ucte from clay~ feldspar, -or similar material• would come in this category. of processes. The product• whose recovery has been attempted are aluminum and potassium eultatee and chemically pure silica. The raw , material• WOJ.Pked on were usually either feldspar or clay. There are essentially three factor• that determine the economic feasibility of a process. These a.re the coat and difficulty ot obtaining raw ma.teriala, the cost of the recovery operations, and the value and market for the products. Theee factor• as they re- late to this paper will be taken up in the order given. Feldapar being not easily obtainable all work with which this paper has to deal was done on clay, so clay only will be considered. 'lhe cost and difficulty involved in obtaining clay in practically any amount will be quite low. In this respect it might well be classed with air or water. It may then be said that clay fulfills admirably the first requirement. The cost of recovery operations being unknown, it is not possible to state the degree to which this condition mey be fulfillecl. The third requirement is almost as easily aatief'ied as the firat. The use of altmiin'UD1, both in compounds and aa a metal, in -2- almoet _all the ramifications of modern life shows that -there will al.ways be a market tor it. As muoh as it ia used now, if the price could be lowere4 even a relatively small amount, its field of uae would be greatly increased because it ia unique among the metal• in embodying certain desirable characteristic•, viz. high tensile 1trength, low_density,,high electrical and heat conductivity, and high resistance to atmospheric corrosion. The use of potassium compounds in the industrie•, especially chemical induatriee, 1• too well known to any-one having any knowledge of ehemiatey to need emphasi•. Chemically pure silica ia not marketable in quantltiee at all comparable to al,mitnum or potassium compound•, but doe• find several important applicationa, vis. silica ware, preparation of chemically pure compounds of ailicon, etc. That the1e product, are all valuable is shown by approximate market prices for 1925. $0.02 .05 Si02 per lb. .15 (c.p. ppt.) Thua it is seen that the only thing neceeBuy to make the reaoveey of aluminum, eta. from clq feasible economically 1s to find P" a ta11'17 1imple proce••· The attempt to work out such a prooeH 1• the subject of this paper. heated to a high temperature with a baeic material, it 11 deoompo1ed and the alumimmi is rendered acid-soluble. Other minor complex com- pound• are also decomposed by the high temperature basic treatment and the metallic constituents rendered soluble. The basic material may be either an alkali sulfide or carbonate or other eufficientlr basic compound. The reaation temperature is about al0°C or higher. The slag obtained from thie treatment is treated with an excea1 of sulfuric acid, the metal• form aoluble sulfates and the ailica 11 preoi-pitated in a very tinely divided form. The silica is then reoovered b7 filtration and the metallic sulfates by evaporation and fractional crystallization. The processe1 for the recovery of valuable product, from els, etc. that have been proposed or attempted a.re al.most innumerable and .so diversified that only a few representative ones will be reviewed here. l. L. Grabau treated the clq, Ybich was a variety as free as possible from iron compounds, with dilute hydrofluoric acid. and thus obtained. a solution or alumin-um fluoride. The silica was volatilized. aa eilioon tetrafluoride. 1 2. A Sobrero roasted the clq, a variety as free as possible from iron and calcium oompoun~; then mixed the powdered mas• with conoentrated sulfuric acid, and heated it to a.bout 70°0. .Al:uminuin sulfate was formed, which was extracted with water and the solu- tion was then treated with potassium sulfate for preperation of al-om2. 1Mellor, Vol. V, p. 257• 2 Mellor, Vol. V, P• 258. -4- 3. F. Lauer roasted the aluminiferous material with iron py- rite,, which reagted with it forming alumim.un sulfate. thus obtained 18 leached with water. The mixture The aluminum sulfate goea into solution, while the basic iron sulfates are left ae slime. The sulfate in solution is converted into potash alwn and recovered as auch3. 4. J. Heibling made briquettes from a mixture of the ~lq with amroonium and potassium sulfates. After calcina.tion at 27;• - 280°0, extraction with water gave al~. 4 5. A. H. Cowle1 heated briquettes made from clay, sodium chloride, and charcoal in a stream of air and steam. The carbon keeps the mass porous and allow1 the gases to readily penetrate to the interior of the mase. A sodium al"Uminosilicate is formed, 2li820.2S102 .A1 2 03; this is mixed ·'"1th lime and heated in a rotary kiln: 2N~O. 2S102 • .Al 2 03 + 4CaO = 2Ca.zSiO~ + 2N8-20. Al 2 03. The sod- ium alumina.ta is soluble in water, and the calcium silicate can be used in the manufacture of hydraulic cement5. 6. P. Miguet heated a mixture of clq, lime, scrap iron, and a reducing agent such as carbon in an electric furnace. Silica calcium aluminate, and ferroailicon were formed: The lighter calci-um aluminate floata on the regulus, and is tapped ~llor, Vol. V, p. 258. 4Mellor, V l V 0 • ' p. 258 • 5Mellor, Vol. V, p. 25s. -5- off, cooled, erushed, and leached with a solution of sodium carbonate, whereby sodium alwninate and calcium carbonate are formed: CaO.Al 2 0:, + N92C03 = CaC03 + N220.A1 2 03. Aluminum hydroxide ie ob- tained from the solution by bubbling carbon dioxide through it. The ferroeilicon is a marketable by-product6. 7. E. Hart has proposed. the following methods of recovery of ahwinum, potassium, and silica from feldspar or clq. The raw material is fused with potassium sulfate and carbon. The slag ob- tained from thie treatment, after being finely pulverized, i• leached with a sulfuric acid solution. The precipitated silica ie re- covered by filtration, the alwninum, potassium and any other metallic constituents going into solution as sulfatas. On partial evaporation and then cooling alum crystallizes out whereas iron, etc. remain in 10lution. The last method, outlined by E. Hart, is essentially the method \'vorked on by the writer. The raw material worked with was a clay used by the California. Portland Cement Company, and trade named "Fleming Clay". The first step was the analysis of the clq for alumina, potash: and silica. l. The methods used were: Silica. - A method used by ceraent companies which 1e briefly this. A sample of the clay intimately mixed with sodium carbonate is fused in a nickel crucible. ground and dissolved in hot water. 7 ·Mellor, Vol. V, p. 258. The fused mix is then Perchloric acid is added and the - 6- solution is evaporated to fuining. ed out, The precipitated silica ie filter- dehydrated by igniting, and. weighed. 2. Alumina.. - The solution obtained from the silica. filtra- tion was analysed for alumina by the method of precipitating aluminum phosphates a.s given in fvfahin and Carr; .Agricultural Analysis. p. 259. 3. Pote.eh. - Potash was determined by the perchlorate method given in Mahin and Carr; Agricultural Analyeie, p . . 247. The resul t1 of the analysis were: Sample I Sample II 8102 62.~ 62.3% Al203 16.5% 15.0% I~O 3.6% 3.5% .An average analyeia of the clay from the deposit was obtained from the cement compaey and it ran as follows: S102 59.0S% Al203 lS.28% Fe203 4. 70% CaO 4.04% MgO 2.11% so, 2.00% ~-0 2.4l+i NazO 2.l.!0% It.? 0 and CO2· 4-95% J.s can be seen there is a fairly large discrepancy in the two sets of values. Some of the differences can be explained by the fact that the cement compazl1 's values are an a.vara.ge. The large diecrepency in the case of alumina was due to the fact that some sulfur came down with the aluminum phosphate precipitate and thi• necessitated an ignition in Gooch crucible• that had not been previoualy ignited. The discrepancy in the case of silica 1• probably due to an actual difference in cornpos1 tion of the cl~. The determination of potash ie a rather complex and uncertain procedure, so the discrepancy here can be assumed to be due to errors in technique. Since the analysis of the clay was a minor part of the problem and the time available for work on the prob- 1~ waa limited, it was decided to spend no more time on clq ' analyei1. The values given by the cement company a.re the valuet used in calculations throughout this paper. Fi.lsion Process The chemical reactions that are thrrught to take place in the fusion are: (3) Si02 + I~S + H20 = ~Si03 + H2S, (The .Al 203 and Si02 are of course largely in the form of the compound Al203.2Si02.2H20,) The H2 S prodv.ced in eqv..ations (2) and (3) is not given off but reacts with the iron or other substances present in the clay, or is 7rviellor, 'Vol. II, p. 622. burned to water and so2 • (4) Fe 2 03 + 3H2 S = 2FeS + 3H2 0 + S (4-a) S + 02 = S02 ( 5) 2H2·S + 3¢l2 = 2R2C + 2S02. The fusion was carried on in clay crucibles of small size, height 3 inches, inside diameter at top 2.5 inche•, fitted with clay covers. The crucible after the charge had been put in waa heated in a. fire-clay cru.ci'ble furnace provided with a compressed-air gas burner for the heating element. The charge, which consisted of clay, potassium sulfate, and carbon, was prepared as follow•. The ingredient& 0::ere ground. s,ep- arately to a fine powder, about ninety percent thro~gh eighty mesh. The correct proportions of each (calculations for these proportions follow) were then placed in a mortar and thoroughly ground together so as to provide intimate contact on the reacting surfaces. In spite of the fact that the mass ~.ises the reaction is essentially a reaction between solidi, because even at high temperatures the charge, alt:hough fused, is very vi scou• and hence there can be little mixing. The carbon monoxicle and other gases given off during the heating process create a certain amount of stirring action, but its effect is probably small, because of the high viscosity of the fused mix. -9- Calculations for Char&! Clay - Si02 = 59.08% Al203 = 18. 28%. Molecular weighte: Si0 2 = 60. 3 Al 2 03 = 102.2 l½SO~ = 174. 3 Per unit weight of clay ~so ... required is: for A1 2 0:, = ( .1828)(174. 3/102.2) = • 312. for Si0 2• = (. 59os)(174. 3/60. 3) = 1._106. = 2.0lS Total K2SO~ per unit clq Carbon per 'Unit clay= (2.018)(4)(12) /(174.3) = .552. The size crucible used permitted a charge containing about 24 grams of clay. Theoretical Charg_e Clay - 24 grams J½SO~ - 48 " Carbon - 13 " Calculating K:zSO~ only for the decomposition of the compound JJ. 2 03. 2Si02 • 2H20 and not for the excess Si02 in the els,. Theoretical Charge Clay - 24. grams K,z·SO~ - l S. Carbon - " " To insure complete decomposition would of course require the theoretical charge first calculated, whereas to merely decompose the al umino-sil icate in the clay would require the charge calou.la.ted in the second case. -10- Since the potass'ium sulfate is the most expensive ingredient in the charge, it is desirable to use a minimum of this ma.tu.ial. If the free silica is reacted on by the pota.seium sulfide produced, in preference to the alumino-silicate, the amount of potassh,m sulfate required will be that required. for complete d.ecompcsi tion, whereas if this is not true the amount will be less. If the reverse 1& true, i.e. if the alumino-silicate is reacted on preferentially, the aznc:unt re(Fired for compbte recovery of the alumin'I.Wl will be that calculated in the second case. Since 1 t was not known which substance would be reacted on :f'irat, the rather arbitrary charge given below was used 8.9 a starting point. No Data.. Clay - 24 grams K2-so... - 27 grams Carbon - 8 grams Crucible split during fusion and charge ran out. 2nd Run .Ingredients: Clay - 24 grams I½SP ...- 27 Carbon - 8 11 " Carbon used Y(at. ~ finely powdered charcoal. Treatment: Heated to a bright red color and held at this temperature for 30 minutee. Slag gro 1ind, then leached with exceaa of dilute H2 SO~ (4 N.) • and filtered. Data obtained: The Si02 precipitated by the acid could not be filtered until it was dehydrated by boiling in acid solution for - 11- eome time (approximately twenty minute• in 4 N.acid). The S10~ wae contaminated by a great many in!purities such ae carbon, etc. 3rd~ Ingredients: Same as in 2nd run. Treatment: Heated to a yellowish red color for 20 minutes. All gas evolution had previously ceased. Procedure and calculations for the recovery of al'Ulnin"l:Un in thia and eubseg_uent runs was as follows: .After the crucible was removed from the furnace and cooled the slag was as completely removed from the crucible as possible. This was about ninety percent. In each case the amount wa.a somewhat different so it waa necessary to e1tirr~te the amount unrecovered for each run. mesh. The recovered slag was weighed and grm nd to a.bout eighty A fraction of it was then weighed out (usually about one- fourth), leached with boiling 4 N. sulfuric acid for thirty minutes, filtered, and diluted in a vol"Umetric flask to 500 cc. Two 50 cc. samples were rgmoved with a pipette}poured into beakers and heated to boiling with about l cc. dilute nitric acid, \W.1ich was added to oxidize the ferrous iron to ferric. It was found that it is very dif- ficult to filter an al\lrninum iron hydroxide mixture that contains ferrous hydroxide. As aoon as the oxidation occurred, this wa1 in- dicated by the solution becoming yellow, the solution was cooled and dilv.te t..1:Hl.j.OH was added until all iron and aluminum was precip itated. It was then bolled again for about five minute& or until the odor of ammonia was faint, the precipitate filtered out onto a quantitative filter, thoroughly washed, ignited, and weighed. The recovery of iron was assumed in all cases to be the same ae that of aluminum, because it is relatively simple to determine the amount of iron plus aluminum present, whereas it is cy.iite difficult to determine al1.'Uninum alone. l %yield =(weight p:pt. )(weight slag)(:f'raction sol. taken for analysis) xlCO (fraction slag taken)(% sla.g recovered)(% Fe+Al)(weight clay) Simplified: %yield= (weight ppt.)(ratio total slag to aro•t. used)(~atio total sol. to am 1 t. used in analysie)_ _ weight Fe + Al in clay Data obtained: For this run percent yield calculated as indicated above was 55 %, Ingredients: Same as in 2nd run. Treatment: Heated to a. yellowish color for 20 minutes. Data obtained: In an attempt t o pour. the fused charge out of the crucible it was found that a. portion of the charge was very tluid, whereas the other portion was little more than plastic. This indicated the possibility of the formation of two phases a.t thia temperature and considerable qualitative work was done on the two portions of the charge to determine the difference in properties and the distribution of iron and aluminum. Subseq,uent runs showed, however, that there was not really a two phase system, because a perfectly homogeneous slag could be obtained by a treatment for the same length of time at a higher tert!Peratu.re or a longer treatment at the same temperature. The effect was probably due to the fact that the charge had not had xl00 - 13sufficient time for complete reaction. No analysis was made to obtain the percent yield of aluminum. Ingredients: 1.5 times those in 2nd run. Treatment: Heated to bright yellow color for 45 minutes. Data obtained: The a.pParent two phase• observed in the 4th run were still perceptible in this run, even though the heating per1dd .was over twice as long, but the appearance and other characteristics of the two portions were different than those obtained in the 4th run, and the charge was much more nearly homogeneoue. The yield of aluminum and iron was 601.,. Ingredients: Clay K2 SOli- - Carbon- 24 gram• 27 " 10 " Carbon used in this and subsequent run, unles1 otherwise noted was a good grade of soft coal powdered to finer than 80 mesh. 10 grams of coal yields about S grams of carbon so the amount of carbon for_the charge is essentially unchanged. Treatment: Heated to a bright red color for 10 minutes. Data obtained: Aa in the 4th run the charge consisted very markedly of two portions. The two portions were treated the same as in the 4th run alid the results were practically identical to those obtained in the 4th · run. It should be noted that in the run the temperature was lower and the time of heating shorter than in the 4th run, whioh indicates that to obtain a homogeneous slag a higher temperature or longer time of heating is required. No analysis was ma.de to obtain the percent yield of alumin'Ul'Il. _n h Run sarr~ as in 6th run. Ingredients: Treatment: Heated for 30 minutes to very bright almost dazzl- ing white color by adding a little powdered coal along with the a.ir blast to the furnace burner. Data obtained: The slag was perfectly homogeneous as has al ready been mentioned under 4th run, indicating, that there wam not two phases present in equilibri'lffl, but rather an imperfect mixing of the fused mass. The very high temperature obtained split the crucible and part of the charge was lost so an analysis for recovery of alurainum was not possible. 8th Run Ingredients: same as in 6th run, except that a somewhat larger amount of ccal was added because it was not so finely powdered as in previous runs. Treatment: Heated to a good white color, some\l'{ha.t lower than t hat obtained in the 7th run, for 90 minutes. Data obtained: Slag was perfectly homogeneous, very ha.rd, and extremely difficult to remove from the crucible. There was a layer of slag fused to the crucible that could not be very well rernoved, and as a consequence it was not possible to get an analysis of the recovery of al'Ulnimun with any degree of accuracy at all. To get an idea of the recovery of potassium a determination of soluble aalts waa made. The entire charge and the portion of the crucible that the charge had stuck to was leached out with boiling H2 S014- (4 N.) solution. All 1ol1lble eubstancea, potassi·f.ll'l'l sulfate was the one of primary interest, were thus brought into solution. The solution was diluted to a. definite volume in a. volumetric flask and a definite portion was removed with a pipette. The iron and al~~inum were precipitated out with NH*0H solution and then removed by filtration. The solution was evaporated to drynesa, ig- nited untn ·. a11 (mL.,) 2 S014- was driven off, and weighed. %recovery of soluble salts assuming it to be all ~S011, ( weight___sal t . recovered) = (weight ealt in charge)(:fraction of sol. taken far analyeis) x 100 For this run the recovery calculated as indicated above was 39•% Clay - Ingredients: grams 'l'j:J N82"S011, - 24 " Carbon - 12 " Treatment: Heated to good white color for 30 minute,. Data. obtained: Recovery of alu:oinum and iron \vae 77%Reoovery of soluble salte waa 73% 10th ]am Same as in 9th run. Ingradienta: Treatment: Same as in 9th run. Data obtained: The following runs were made to determine the time req_uired for complete leaching. -16- No. Time of Leaching Table! Temp. of Leaching l 60 min. 20°0 2 Just brought to boil. 3 2 min. Boiling (l00°C) 67.% 4 10 min. Boiling (100°c) 67•% Recovery of Al and Fe 41.% 67.'j'o In determining (l) after leaching and filtering off the residue, it wae necessary to boil the solution and refilter because the silica was not sufficiently dehydrated by the leaching process to be filterable. Ingrediente: Treatment: Same as in 9th run. Heated to reddish yellow color for 45 minutes. Data obtained: Tests for time required for leaching were made a.a in 10th run. No. Time of leaching l Just brought to Temp. of leaching Recovf!ry of Al and Fe Silica not dehydrated so sol. was not boil. analysed. 2 10 min. Boiling (100°0) 50% 3 30 min. Boiling <+00°C) 51% 12th Run Ingredients: Cl~ - 50 grams N~S0 14 - 24 It Carbon - 18 • It ·J:reatment: Heated to a yellow color for 30 minutes. -17- Data obtained: Recovery of alumimnn and iron wa.a 46%. Recover.1 of soluble salts was 35%• Note. Just prior to this run it waa disooverad that, i f a charge of coal were coked in the crucible before it was used for the heat treatment of the charge, the slag would not stick to the oruoible. Thie made it possible to recover all the slag and eliminated the somewhat uncertain estimation of the amount of slag recovered. Ingredi enta: Clay - -;-p gram, 24 Carbon - 10 N~SO~ - Treatxnent: " " Heated to reddish yellow for 15 minutes. Data. obtained.: The slag from this run was used in attempting to work out the third method for the reGovery of aluminum from solution. 1500° Powdered coal used instead of charcoal Powde red coal 8 9 1100° (9) -Same ae Same as (9) excapt 18 grams coal instead of 10 ll 12 13000 15000 Same as (9) 10 Na2 SO~ for K2 SO~ ~ grams clay instead of 24 1300° " 5 1500° 1100° None 3 Temp. of run ± 100° Ingredients Variation from Standard Run recovered Fe recovered 7. 71 5. 87 5.2s 67 . 51. l.i6. n II 30 45 30 " If II 11 35 73. s. 87 77. ff 30 If 39. ti -- 90 -- " 3.03 grams 3.31 55. recovered 60. 45 " 20 min. Length of run --%Al and Fe A-nount .Al and %of soluble sal t1 Table III l I--' I O<l -19- Recovery ~f Aluminum from the Leaching Solution. 1st Method. Aluminum sulfate and potassium sulfate can be recovered from a. solution quite easily and completel y if they a.re in equi-molal concentrations. Alum (Al 2 (so .... ) 3 .K2 so...,.24H2-o) .is very soluble in hot water (400 parte alum per 100 parts water at 100°0) and only slightly soluble in cold water (5 parts per 100 at 0°C). By concentrating the solution to saturation at 100° and then slowly cooling to 0° or othar suitable low temperature it is poseible to obtain fine alum crystals and obtain an almost complete recovery. If sodium salts • _be are present in considerable amounts the alum crystals will/contaminated with Ma.S014,. lOH 0, gla.eeri te (3K so .....·N8-2S0...,) and sodium alum8. 2 2 2nd Method. (Unsuccessful) Since iron is present in the solution in considerable a."IlOunts, any method for the recovery of aluminum must provide for the separation of it from the iron. The method solution 1s boiled with calcium oxide. attempted was this. The Aluminum is amphoteric. whereas iron is not, so in the basic solution the iron would be precipitated aa hydroxide and the aluminum would et~ in as aluminate. The iron hydroxide is then filtered off, and the solution neutralized until only weakly basic whereupon aluminum hydroxide would be precipitated and could be filtered off. The method was unsuccessful because the calcium hydroxide solution is not sufficiently basic to form aluminates. :Both al'U!tlinum and iron are completely precipitated by thia hydroxide. gJ. Ind. and Eng. Chem. v. 10, p. 34s (1918). -20- 3rd Method.. There was not sufficient time to do any quantitative work on this method but some qualitative tests indicate that it has possibilities. The elag was leached with a sodium hydroxide solution (about 2 N.)9, This dissolved out the aluminum leaving the iron as a. sul- fide and hydroxide. The residues were filtered off. The hydroxide solution was then diluted to about .2 N and a fair amount of the aluminura was precipitated, and filtered off. appearance. It had a very pure The dilute hydroxide solution could be reused by con- centrating it to about 2 N. by evaporation. Aa aoon aa the impur- itieG accumulate in euch amounts ae to interfere the solution could be neutralized with sulfuric aoid. This would precipi ta.ta the soluble eilicatee as silica and it could be obtained in a pure form by merely filtering. In Table III (page 18) the more important data obtained has been tabulated so as to show the relations of certain effects. On the whole the interpretation of these data is rather difficult, but they do indicate a few things. l. The best yielde were obtained intle runs where the temper- ature waa highe1t. 2. Considerable amounts of potassium were lost during the fusion process, probably both by evaporation and. mechanically. carbon monoxide and other gases fonned in the reaction no doubt carry along some of the charge. 9Noyes 1 "Qual. Chem. Anal.", 8th Edition, p. 95. The -21- 3. Aleo when uaing sodii.lm considerable amounts were lost, although much less than potassium. Since the amount carried mech- anically should be about the same in either case, this wo~ld indicate that the sodium is not lost so readily by evaporation. 4. When coal is used as a source of· carbon, the loss of 'charge mechanically is considerable. This is due to the large vol- ume of gas given off when the coal cokes. Since this coking occur, a.t a t:amperature below the fusion temperature of the cb.arge, the escaping gase1 varry off the finely pulverized charge. The excep- tionally low yields in run (12) may be explained by this fact, because here an excess of coal wae used. The recovery of aluminum in (9) indicates that the optimum 5. amount of clay par 'lll'li t of sulfate and carbon has not been reached yet. To find the optimum amount further runs sho1:lld be made using large excesses of clay. Conclusion l. Unless the aluminum is recovered a.a outlined in Method III, (page 20), the recovery of silica is impossible becauae of the large amount of insoluble impurities in the slag. 2. The use of potassium sulfate as a raw material is uneconom- ical for these reasona: (a) over half of the sulfate is lost in the fusion process "oy evaporation and mechanically; (b) the amount of potassium in the clay, and hence the maximum amount that can be recovered, is small; and (c) the sulfate is by far the most costly ingredient. -22- 3• If scdi'UJn sulfate is used. instead of pot aseium sulfate as a raw material, the recovery of the potassium in the clay is cut of the question. 4. Since potassium cannot be recovered, and it is doubtful if ailica can, if the process is to pay , the value of the aluminun salt recovered rmlst pay for the procese. 5. Unless some method is discovered radically different from the one worked on by the writer, the value of the aluminum Bal t recovered will never be able to pay for the process. The high temperatui,e1 re- quired for fair yields, the great d.ifficul ty involved in the separar- t ion of the al urnimim from the iron ( the only promising method requi ring the use of the comparatively expansive raagent, sodi-um hydroxide), the nUlnerous operations including grinding, leaching, evaporation, etc.: these costs coupled with the fairly low value of the aluminum sulfate, which would probably be the final product, make the process, from an economical standpoint, quite impossible. The writer wishes at this time to exprass his indebtedness to Professor William N. Lace;r f. () r hh, gviding suggestions in the performance of the experimente.l work, and for the helpful adv:ioe and criticieme in the preparation of this manuscript. Thanks are due the California. Portland cement Company, who furnished the s~nple of Fleming clay used in theee experi~ents. Si..utimaq The experimental work carried. out shows that the recovery of al uminuu1 from Fleming cl 8¥ is perfectly :possible and. that by additional experimental wo rk it would be p'ossi ble to obtain fairly good yields; but that the recovery of silica. and potassium ~~uld be very difficult if not impossible. While the aluminum recovery ie chemically pos s! ble, the cost of res.gents and. operations makes the process impossible from an economic standpoint.