NOTE ON THE ACTION OF ALUMINIUM POWDER ON SILICA AND BORIC ANHYDRIDE. BY FRANK E. WESTON, B.Sc., AND H. RUSSELL ELLIS, B.Sc. ( A Paper read before the Faraday Society, Tuesday, October 29, 1907, Mr. N. T. M. WILSMORE, M.Sc., in the Chair.) The statement of Goldschmidt that boric anhydride and silica are reduced by aluminium to boron and silicon respectively has been called in question by other workers.::: Having observed that silica in the form of kieselguhr and titanium dioxide in the form of rutile are readily reduced by aluminium, and as we required quantities of boron trichloride, it was thought worth while to examine Goldschmidt's reaction with respect to boric anhydride and silica respectively. ACTION OF ALUMINIUM POWDER ON BORIC ANHYDRIDE. Aluminium powder, the finest obtainable from the British Aluminium Company, was found to react with boric anhydride more or less vigorously, according to the conditions :- (I) Using B,O, which had passed through a sieve of 14,400 meshes to the square inch- (a) The reaction took place instantly in the cold, using a fuse of BaO, and Mg ribbon, spreading throughout the whole mass, and throwing off incandescent particles, when the mixture was made in the pro- portion of the molecular weights Al, and B,O,.(b) The reaction only took place on heating the mixture, which was contained in a Hessian crucible, in a muffle furnace to dull redness, when the mixture was in the ratio of Al, and 2 B203. (2) Using B,O, that had passed through a sieve of 3,600 meshes to the square inch, it was found that the reaction would only take place after the mixture had been heated- (a) Using molecular quantities and heating to about 300' C., the reaction started when a magnesium fuse was used and proceeded through- out the mass.(b) Using quantities represented by Al, and 2 B20,, the action started and proceeded throughout the mass when heated for some time to dull redness in the muffle furnace. The products of the reaction are difficult to isolate owing to the in- solubility of the Also3 produced in the reaction and also to the presence of one or more borides of aluminium. However, it was shown that free boron is produced in the reaction in which excess of B,03 is used, with more or less borides of aluminium ; whilst in the reaction between molecular quantities the borides of aluminium probably predominate.The authors would not recommend this reaction either as a method of See Proceedings of Farnday Society, vol. iii. parts 5 and 6. 170NOTE ON THE ACTION OF ALUMINIUM POWDER 171 obtaining boron, owing to the difficulties of separating it from the other products of the reaction, or for the preparation in quantity of BCl,, since the G1,0, reacts on the BCl, formed with the production of AlCl, and B,O,, and hence the yield of BC1, is far from the amount of RCl, actually first produced. Experimental. Experimertfs, Set 1.-35 grams of B,O, (finest used) and 27 grams of Al-ie., in proportions of Al, and B,O,-were intimately mixed and placed in a Hessian crucible ; the reaction was started in the cold with a fuse of Mg ribbon and '2 gram BaO,, and spread rapidly throughout the mass, throwing off incan- descent particles.When cold, the crucible contained a black porous mass, easily removable from crucible and homogeneous in composition, and easily powdered in an iron mortar ; this powder was finally ground in an agate mortar until it passed through the 14,400 mesh sieve-a residue which did not pass through the sieve scratched the agate mortar. The whole of the powder was extracted with boiling dilute HC1-hydrogen was evolved to a small extent at first ; it was filtered, and the filtrate on evaporation to dryness gave a residue of AlCl,, FeCI, (traces), and H,BO, ; the insoluble residue was again extracted with hot concentrated HCl until filtrate gave no H,BO, and only a trace of AlCl, and FeC1,.This insoluble residue still left gave the alcohol and sulphuric acid reaction for boron, It is probable that the AlCI, came from a little unused Al, the borides of aluminium and the A1,0, which is only very slowly attacked by HC1. The iron came originally from the Al, which Contained a little iron and paraffin wax, or stearin. Exantiriafiorz of' Residue Insoluble in HCI (A).-In order to obtain some idea of the amount of boron or aluniiniuni borides it was decided to convert the same into BC1, by the action of chlorine. The residue was dried, placed in a piece of dry combustion tube, fitted on the one hand to a cylinder of liquid C1 and on the other hand to a tube placed in a freezing mixture 0 of ice and salt which registered a temperature of -19.C. throughout all the experiments.On heating the residue and passing a slow current of chlorine, the mass became red hot, but did not appear to burn rapidly. AlCl, with traces of FeCI, condensed in the cool part of the tube and BCl, condensed (I) On removing from freezing mixture it was seen to be slightly yellow in colour, and immediately commenced to boil ; keeping tube in freezing mixture, a few drops of dry Hg were run in and the tube shaken ; the liquid became colourless and on connecting to another tube placed in a freezing mixture, it was distilled into the same by simply withdrawing the first tube from its freezing mixture. (2) It reacted with water violently, producing H,BO,. (3) It reacted with anhydrous alcohol-free ether, producing a white crystallised solid with considerable evolution of heat (cf. SnCl,, TiCl,, CCl,, and SiClJ ; this reaction is being further investigated. (4) It reacted energetically with Grignard's reagent C,H,MgI, evolving great heat ; this reaction is also being further investigated.( 5 ) Its B.P. was found by Siwoloboffs' method to be 17' C. Hence liquid was BCl,.173 NOTE ON THE ACTION OF ALUMINIUM POWDER EXAMINATION OF RESIDUE LEFT IN TUBE AFTER HEATING I N C1. The residue was black, but contained more white particles than original powder. It was extracted with boiling water, filtered, and filtrate evaporated to dryness, leaving a residue containing AlCl, (with trace of FeC1,) and H,BO,. The part insoluble in water gave no reaction for boron by the alcohol and sulphuric acid test, showing that all the boron or boride present in original had been attacked.The presence of H,BO, in the soluble part is accounted for by the fact that BCl, reacts on heated A1,0,, producing B,O, and AlCl, ; thus AlCl, condensed in cool part of tube comes from the ALO, as well as the borides of aluminium present in the original. ATTEMPTS TO REMOVE A1,0, FROM RESIDUE A. Mefhod I.-the residue A was digested with commercial HF in a platinum dish for one hour, filtered and washed ; the filtrate, on evaporating to dryness, left a residue of AlF, (traces of FeF,) but no boron, the boron or boride if attacked having escaped as HF,. The insoluble residue was extracted eight times with HF, when the filtrate on evaporation left an extremely small residue. On drying this residue and treating with C1 as before combination took place with incandescence and the action was over very quickly.BC1, was produced atid AlCl, (with traces of FeCl,) condensed in the tube as before, but quantity was relatively smaller than before. Method 2.-The residue A was fused in a platinum dish with twenty times its weight of fused borax in a muHe furnace, the top of the crucible being tightly covered with asbestos wool. The heating was continued for 13 hours at a temperature of about goo°C. The melt was allowed to cool and then extracted with water and dilute HCl. The resulting inass was quite black, and had decreased in weight from 8.4 grams to 3.3 grams. On treating this mass with C1 as before, it took fire, and the incandescence rapidly passed throughout the mass, thc products being the same as before.N.B.-In both cases the residue left after passing C1 was quite white and contained B,O,, which on removing with boiling water left a perfectly white residue of A120,. Hence both of these treatments removed the greater part of the A1,0,, and from the colour of the residue after treatment it appears that borides of aluminium were present. The yields of BC1, and AlCl, from residue A were compared before and after treatment. I. Before Treatwent- Residue A taken Amount of residue A attacked = I*@ ,, Weight of BC1, obtained = 3.9 ,, = 13-8 gins. J 9 A1C13 9 , = 8.3 1 , I.c., for I grain attacked z grams BC1, are produced and 4'23 grams AlCl,. 2. After Treatmeizf wifh Borax- Amount of substance taken Weight of BC1, produced = 1-44 ,, 1, MC1, ,, = 1.25 ,, I.e., for I gram attacked are produced 2-99 grams BC1, and 2.55 grams AlC1,.Experiment, Set 2.-Reaction between Al, and z B,O,. 16 grains B,O, and 6-2 gram A1 were mixed, and as reaction did not take place in the cold even = 1,485 gm. 7, ,, attacked =0'485 ,,ON SILICA AND BORIC ANHYDRIDE 173 with a fuse of Mg and BaO,, the mixture was heated in a muffle furnace. The contents of the crucible on cooling presented a hollow appearance, and on extracting with water repeatedly gave a large deposit of HaB03. The mass was distinctly brown in colour. On drying carefully at a gentle heat white fumes were seen to be evolved, and on again extracting with water, H,BO, was obtained, showing that the free boron was undergoing slow combustion. The residue after well washing again with water was dried in ~ C I I O over sulphuric acid.On treating I portion with fused borax in the molten state it immediately took fire, and so the borax was allowed to partially cool beforc adding the impure 13, well covered with asbestos wool, and treated as in previous ex per im e n t . The crude boron now obtained burnt vigorously in C1, producing the same products as before, but the BCI, now greatly predominated, viz.- Amount of substance taken =0*5j gm. ,) substance attacked = 0.06 ,, ,, BCl, produced =0-49 ,, - -0.158 ,, 9 ) Ale13 ,) I.c., I gram of substance attacked produces 8.12 grams BC1, and 2.6 granis AlCl,. From the foregoing experiments it will be seen that the action of A1 on B,O, produces : (I) in molecular quantities Al,O, and B with excess of borides : ( 2 ) in quantities of 2B,O, to A1,0,, boron is the chief product.Exferinzenf, Sei 3.-An attempt was made to see if the amount of boron formed in the second reaction could be estimated. It has already been pointed out that some of the AlCI, produced in the action of C1 on the crude B was formed by the action of the BCl, first formed on the Al,O, in the mixture. 2BC1, + A1,0, = 2AlC1, + B,O,, it is seen that if the amount of BZO, formed in this action, and which is found in the residue after passing the C1, can be estimated, it is possible to calculate (a) the amount of BCI, used up in its formation and ( b ) the amount of AlC1, formed at the same time. The B,O, was therefore estimated in the residue by ascertaining the weight of residue after passing CI, extracting the B,O, with boiling water, and again weighing the dry residue, and hence calculating the B,O, by difference. The apparatus used was as in previous experiments, the crude boron being weighed in a porcelain boat, which could be withdrawn after the passage of the C1 and again weighed.Plugs of glass wool were inserted in the combustion tube, which was about 30 cms. long ; the chlorine was dried by passing through strong sulphuric acid, A fresh quantity of crude boron was made, but was not treated with fused borax as in experiment, Set 2 . 32 gms, B,O, and 12-3 gms. A1 were mixed in a Hessian crucible, covered with asbestos wool and well dried MgO ; the crucible was placed in the hottest part of the muffle and heated to redness for I hour; it was then cooled, the MgO removed from the top by blowing, and the contents of the crucible taken out.The mass came out whole, and on breaking brown streaks were seen interspersed with black and white specks. The mass, after powdering, was treated with water and dilute HCl, practically no effer- vescence taking place, showing that all the A1 was used up ; the aqueous HC1 extract consisted of H,BO, solution ; after thoroughly washing with hot Now, taking the action to he-174 NOTE ON THE ACTION OF ALUMINIUM POWDER water, it was washed with alcohol and dried in a current of dry CO;* at as low a temperature as possible. Action of Heat oiz this Residue.-A small quantity of the substance heated in a crucible lid in the open air rapidly changed colour to a grayish white ; on extracting this with water and evaporating filtrate to dryness H,BO, was obtained, whilst the insoluble portion was still brownish in appearance, showing that the boron had superficially oxidised.The following quantitative operation was carried out :- Weight of substance = 2,991 gins. ,, B,O, found= ~2829 ,, ,, BCl, = 1.88 ,, 9 ) AIC1, = 1'2 ,, But 35 gms. B,O, = t117-5 gms. HC1,. Therefore '2829 ,, = t I 17 x 2829 = -9s gms. BCl,. Also 35 grams. ,, = t 133.5 gms. AlC1,. Therefore ,2829 ,, = f 133-5 x 2829 = 1-08 gm. AlCI,. 35 35 Therefore total weight of BCI, produced by primary reaction = 1-88 + -95 = 2-83 gms, and weight of AlCl, produced by primary reaction = 1.2 - 1.08 = '12 gins. Hence the boron is chiefly present in the form of free boron. The only borides of aluminium that have been described are AlB, and AlB,, (Hampe).Now the ratio of the weight of BCI, to the weight of AlCl, produced by action of C1 on AIB, is as 235 : 133.5, i.e., less than 2 : I ; similarly for AlB,, the ratio is as 1410 : 133.3, i.e., nearly 11 : I . Hence if in above experiment the BC1, had been produced entirely from these borides, the ratio of weights of BC1, and ACl, could not have been greater than I I : I , whilst i n the actual experiment the ratio is as 283 : 12, i.e., nearly 24 : I ; hence the boron greatly predominates over the borides. ACTION OF ALUMINIUM ON SILICA. Vigouroux states that crystalline silicon can be obtained by heating finely powdered quartz with A1 in excess in the electric furnace (R.and S., p. 876, vol. I ) ; Minet (" Production of A1 and its Industrial Use," p. 208) states that if two molecules of silica and four molecules of A1 are thoroughly mixed and carefully heated to 800' C. the reduction from silica to silicon takes place. In our experiments three kinds of silica were used, e.g., kieselguhr, pre- cipitated silica, and silver sand. The aluminium powdcr was the same as used in the experiments with B,O,. It was found that when mixtures of aluminium and silica were made in proportions given by +I1 and 3SiO, the reaction could not be started in the cold by means of a fuse of BaO, and Mg; but on heating to dull redness the reaction commenced and spread throughout the whole mass more or less quickly. Using quantities in the proportions given by 8A1 + 3SiO,, and using a fuse of BaO, and Mg ribbon, the reaction started at once in the cold with kiesel- guhr and precipitated SiO,, but only on heating in the muffle to red heat in the case of the sand.* It was observed that when CO, was passed over the partially dried substance at a temperature of from IOOO to I ~ o O , a small, white sublimate was formed which proved to be ammonium carbonate. This was probably formed by the action of steam on boron nitride (HCl having removed any aluminium boride), viz. :- zBN + 3H,O=B,O, + zNH, and 2NH, + CO, + H,O= (NH,) 2C0,.ON SILICA AND BORIC ANHYDRIDE 175 Silicon was produced in all cases, this being proved by conversion into SiCl, on passing C1 over the products of reduction. No attempt was made to ascertain the quantity of silicon produced, but tests were carried out to ascertain the yield of SiCl, obtained in each case.AlCl, was also formed, owing to the action of SiC1, on the Also,, but the quantities of AICl, were relatively smaller than in the experiments with boron. EXPERIMENTAL. Aluminium Powder and Kieselguhr. I. I n Projorlion of 4 AZ fo 3 Si0,.-1o grams kieselguhr and 6 grams A1 powder were intimately mixed and placed in a Hessian crucible and attempts made to ignite by means of a fuse of BaO, and Mg. No reaction resulted. On placing in a muffle and heating to dull redness the action started and proceeded steadily throughout the mass, and at the surface the Si produced took fire. The mass, on cooling, was extracted with HCl (I -I), when a slow evolution of H took place, but much less than is produced with the product from Mg and SiO,.The residue contained brown masses, and on thoroughly drying the well powdered mass was a dark grey colour. The action of C1 on the residue was carried out in a similar apparatus to that used for Cl and B. Weight of substance taken = 2.72 gms. Loss of weight of substance= -36 ,, Weight of SiCl, produced = -62 ,, Small quantities of AlC1, and FeC1, produced. C1 passed slowly for one hour. 2. In Proportion of 8 A1 and 3 SiO,.-15*64 grams kieselguhr and 19.57 grams A1 were intimately mixed and a portion placed in a Hessian crucible and ignited whilst cold by means of a fuse of BaO, and hlg'; the reaction took place at once and spread rapidly throughout the mass.The other portion, in a Hessian crucible, was placed in a hot muffle, and the reaction commenced in a few minutes and spread throughout the mass. On cooling, both lots were powdered, extracted with HC1 (I - I), filtered, washed, and dried, The action of C1 gave- Weight of substance taken = xo'oo gms. LOSS of weight of substance = '95 ,, Weight of SiCl, produced = 2-47 ,, AlC1, and FeCl, (small quantity). C1 passed steadily for one hour. Aluminium Powder and Amorphous Silicon. I . In Proportion of 4 A1 and 3 SiO,.-s.8 grams A1 and 9.7 grams SO,. This mixture would not react in cold with fuse of BaO, and Mg, but on heating with a blower burner the fuse started the action, which proceeded quietly through the mass. On heating residue as before and passing C1, the following result was obtained :- Weight of substance taken = ro'6 gms.'7 ,9 Weight of SiCl, = 2.72 ,, - Loss in weight - C1 passed slowly for one hour,176 NOTE ON THE ACTION OF ALUMINIUM POWDER 2. 111 Proportion of 8 A1 and 3 SiO,.-3 grams SiO, and 9'7 grams Al. This Residue mixture reacted rapidly in the cold with a fuse of BaO, and Mg. treated as in previous experiments. Weight of substance Weight of SiCl, - '20 ,, = 2'00 gms. Loss in weight of substance = '12 ,, - ,, AlCl, and FeC1, = '20 ,, (about). C1 passed for ten minutes. Alunzinirrvi Powder nud Sand. The sand was extracted with aqua regia till no further action took place, well washed, dried, and sifted through a sieve (3,600 meshes to square inch). I . In Proportion 4 A1 and 3 SiO,.-9-7 grams SiO, and 5-8 grams sand.This mixture would not react with a fuse, cold or hot ; reaction took place on heating to redness in a muffle. The resulting mass was treated as before. A greater evolution of hydrogen took place on heating with HC1 than in case of kieselguhr and amorphous silicon. Weight of substance = 3-00 gms. - ,, SiCl, '20 ), ,, AlC1, = '20 ,,(about). Cl passed for thirty minutes. 2. I n Proportion of' 8 A1 and 3 SiO,.-9.7 grams SiO, and 11.6 grams Al. This mixture also would not react with a fuse, but reaction took place on heating in a muffle to dull redness. On grinding in a mortar the gritty nature of the sand had disappeared. Violent evolution of H was obtained on treating with HCl. Weight of substance = 1-00 gms. - ,, SiCl, - '20 ,, - ,, A1C1, - very small. C1 passed for fifteen minutes. N.B.-The aluminium powder used in all these experiments had been treated for some time at a temperature of 13o0-~5o0 in order to free it partially from the adherent paraffin wax. All the forms of silica were previously heated to redness in a muffle. Tlz ermock emical Data. The Heat of Formation of Al,O, = 390,000 calories ,, ,, 9 , B,O,=317,ooo 9 , 7 9 > I ,, SiO,=219,000 ,, Hcnce 2Al+ H,O? = A1,0, + 2B + ( ~ ~ o , o o o - ~ i ~ , o o o ) calories Also + (qA1 + 3SiO,) = + (2A1,0, + 6Si) + 3 (~8o,ooo-6~~,ooo) calories That is, according to the principle of greatest heat development, B,O, and SiO, should be decomposed by Al, and B,O, should be decomposed more readily than A1,0,. This was found to be the case experimentally when the conditions of esperimenting were the same in each case. = A1,0, + 2B + 73,000 calories. = 3 (2A1,0, + 6Si) + 61,500 calories. CHEMICAL LABORATORY, THE POLTTECHS IC, REGENT STREET, W,