OCt., 1917
T H E J O U R N A L OF I N D U S T R I A L A N D ENGINEERING CHEMISTRY'
THE SEPARATION OF ALUMINUM FROM IRON BY MEANS OF ETHER By SAMUEL PALKIN Received June 16. 1917
Various methods' for t h e separation of aluminum from iron have been proposed. None of these, however, have found extensive use in analysis. As t h e method of separation herein proposed is based on t h e use of organic solvents, only such processes as have a direct bearing on this method will be discussed in this paper. It had been noted by Gladysz2 t h a t hydrous aluminum chloride was very slightly soluble in strong hydrochloric acid, while ferric chloride was readily soluble in t h a t medium. Gooch a n d Havens3 utilized t h e observation of Gladysz, b u t introduced t h e use of ether t o reduce t h e solubility of aluminum chloride. They precipitated this salt from a mixture of equal parts of concentrated hydrochloric acid a n d ether saturated with hydrochloric acid gas a t I 5 O C. It appears t h a t t h e method of Gooch a n d Havens is t h e only one on record which utilizes a n y organic solvent (one-half aqueous hydrochloric acid a n d onehalf ether). Frankforter4 in studying t h e action of aluminum chloride on aliphatic ethers found t h a t t h e presence of moisture caused precipitation of a compound of this salt with water and hydrochloric acid from a n ether solution of t h e anhydrous aluminum chloride. This precipitate, however, was not constant in composition. H e utilized this reaction as a qualitative test for water in ether. T h e method proposed in this paper, like t h e Gooch a n d Havens method, depends on t h e insolubility of hydrated aluminum chloride in solvents as contrasted with t h a t of iron, b u t may be more closely compared with t h e qualitative observation of Frankforter, as i t neither uses a mixture of water and ether, nor requires saturation of t h e medium with hydrochloric acid gas. I t is primarily t h e latter feature in t h e Gooch a n d Havens method which is troublesome a n d objectionable for ordinary laboratory procedure, as in each determination, hydrochloric acid gas must be passed through t h e solution a t low temperature ( I S " C.) t o saturation (a point rather difficult t o determine with certainty) a n d t h e precipitate washed with ether-water mixture, itself saturated with HC1 gas. I n t h e proposed method, t h e medium is entirely organic solvent, except for traces of moisture introduced b y t h e solvents. The essentials of t h e method are as follows: The dried mixed chlorides of aluminum and iro? are taken u p in a small amount of hydrochloric acid alcohol solution and evaporated t o crystallization of t h e salts. T h e residue is then again acidified with alcoholic HC1. Ether (U. S. P.), which contains a trace of water, is gradually added, and t h e aluminum 1
W. W. Scott, "Standard Methods of Chemical Analysis," 1917,
pp. 4-5. 2 J
4
Ber., 16 (1883). 447. A m . J . S c i , [4] 11 (1896), 416. J. A m . Chem. SOC.,87 (1915). 2560-7.
951
is precipitated as a hydrated chloride, varying i n composition, while t h e iron chloride remains in solution. EXPERIMENT A L E F F E C T O F MOISTURE-It was observed in preliminary experiments t h a t anhydrous aluminum chloride was readily soluble in absolute alcohol and absolute ether, b u t t h a t t h e presence of a trace of moisture, introduced as such or by means of wet alcohol or wet ether, caused a n immediate precipitate. An excessive amount of moisture, however, invariably caused t h e formation of a second layer of a viscous gelatinous mass. It was necessary, therefore, t o determine t h e limits of water tolerance and t h e best means of introducing t h a t water. A series of experiments was performed in which water was introduced in varying amounts using, at first, strong aqueous HC1, as t h a t acid was found necessary t o keep t h e iron in solution. This manner of introducing t h e water was found impracticable as t h e amounts of HC1 necessary caused, in many cases, a n excessive amount of water t o be introduced. An alcoholic solution of HC1 was, therefore, prepared a n d water was introduced b y means of t h e ether used. The following facts were observed: I-When a n insufficient amount of water was present as obtained b y dissolving t h e salt in a small quantity of alcohol a n d using largely absolute ether, t h e precipitate came down in a n exceedingly finely divided condition, colloidal or semi-colloidal, practically impossible of filtration. When filtered, t h e precipitate was exceedingly hygroscopic, behaving somewhat like finely powdered (dried) calcium chloride: t h e physical s t a t e of t h e precipitate was such t h a t a clean separation from t h e iron would be impossible even if t h e precipitation were complete. It was, in fact, found t h a t u p t o a n alcoholic content of about 4 per cent, t h e precipitation was complete, when t h e suspension was filtered several times. a-When a n excessive amount of water was present, which depended on t h e amount of salt present t o take care of i t a n d t h e amount of water introduced with t h e ether, i t was found t h a t a gelatinous layer of t h e salt was formed. EFFECT O F ALc0HoL-h excessive amount of alcohol had a tendency similar t o t h a t observed in t h e case of insufficient water, namely, t o precipitate t h e aluminum chloride in a very finely divided state. I n addition, t h e precipitation is incomplete when alcohol is present .beyond certain limits, depending upon t h e amount of water present. E F F E C T O F H Y D R O C H L O R I C ACID-This reagent was introduced in t h e form of t h e alcoholic solution. Aside from t h e fact t h a t HC1 was required t o keep t h e iron in solution, i t was found t h a t a slight acidity was essential for two reasons: ( I ) t o make t h e precipitation complete, ( 2 ) so t o modify t h e physical state of t h e precipitate as t o render it easy t o filter. On t h e other hand, excessive amounts of acid seemed t o tend toward formation of gelatinous salt layer as in the case of excessive water.
T H E J O U R N A L OF I N D U S T R I A L A N D ENGINEERING C H E M I S T R Y
QS 2
A large number of experiments were performed t o observe t h e effect of these various factors. Table I contains a few which show in a quantitative way t h e results obtained.
;,
TABLE I-PRECIPITATION OF ALUMINUM CXLORIDEFROM MIXTURESOF 'VARYING PROPORTION OF ALCOHOL, HCl, ETHER (U. S. P.) A N D ABSOLUTE ETHER,A N D IN THE PRESENCE OF FERRIC CHLORIDE Mo. USED Absolute Alcoholic Ether Ether Condition Residues 7 Alcohol HCI W.S. P. Absolute of as AlzOa AlCla FeCli Cc. 35% Cc. Cc. Filtrate Mg. 100 2 50 Clear 0.2 2 50 Clear 0.2 2 100 'drops 0.1 0 . 5 cc. 3 100 1.5 50 Clear 1 cc. 1 50 '.. .. Clear 0.0 4 100 2 drops 2 0.1 50 . . Clear 5 50 50 Clear 0.1 .. 6 50 ... 1.5 0 . 5 cc. 25 Clear 0.3 ? 10 25 ... 1.5 0 . 5 cc. Clear 0.3 2.5 cc. 250 .. '8 500 ... 2.5 Clear 0.2 25 9 10 25 . . . 1 . 5 0.5 cc. Clear 40 10 500 ... 1.5(a) 0.5 cc. 40 0.2(b) 40 Clear 40 ... i.5(a) 0 . 5 cc. 11 500 1 cc. 1 Clear 12 10 500 50 1 13 10 1 cc. 90 50 . . Clear 1 14 10 1 cc. 5Q 50 . . Clear (a) Approximate. ( b ) Total residue on bath. No. I
...
i
.*.
...
As indicated in t h e outline of t h e method given above, t h e chlorides of aluminum and iron, after driving off -early all of t h e moisture in an oven, are taken up in alcoholic HC1 and absolute alcohol, evaporated t o a viscous semi-crystalline mass and then definite amounts of HCI, alcohol and ether are added. It was deemed Advisable t o see how much, approximately, HC1 and alcohol remained after evaporating t o a viscous mass. An experiment was performed using 0.5 g. anhydrous aluminum chloride (the maximum amount of comshined salts permitted in t h e proposed method) adding absolute alcohol and HC1-alcohol, heating until salts completely dissolved and evaporating t o viscous mass. ,The total was weighed before and after, and t h e differ.ence found t o be about 0.62 g., t h a t is, both HC1 and alcohol. In the actual method, 0 . 5 cc. alcoholic HC1 is then added. The total volume could not exceed I . 5 cc. in t h a t case (to which is then added t h e requisite quantities of ether). It is evident, as illus%ratedin Table I, t h a t the amount of aluminum chloride remaining in solution is practically negligible. This fact is borne out by later experiments on actual separation of known quantities of t h e salts as shown in Table II. iI-R&foLTS AlCL used Cc. solution I . . . . 50A 2. ... 5 A 3 . . . . 25A 4.... S A S . . . . 25 4-A 6 . . . . 50 A-A '1. ... S A 8.... 2 5 A 9 . . . . 50A 10 . . . . 2 A 11 .... 5 0 A 12 .... SOB 93 . . . . 25 B 1 4 . . . 25 B 15. . . . 25 B 1 6 . . . 25 C 17 . . . . 2 5 C 18 . . . . 25 C 19..'.. 2 5 C 20. ... 2 c 21 . . . . 2c 22 .... 50D1 2 3 . . . . 50 Dt 2 4 . . . . 50 Dt
TABLE
No.
.
.
OFJTAINSD BY APPLICATION OF PROPOSED METHOD Amount GRAMSAh08 GRAMSFez08 FeCL used Found Calculated Found Calculated 20 mgs. 0.2068 0.2069 0.0204 0.0207 700 mgs. 0.1036 0.1035 350 mgs. 0.0207 0.0207 TOO mzs. .... 0.1007 0.1004 300 mis. .... .... 0.2004 0.2007 I O mgs. .... .... 0.0212 0.020'1 50 cc. I 0.2029 0.2035 0.1041 0.1035 0.1014 0.1017 25 cc. I 0.2070 0.2069 0.0208 0.0204 5 cc. I 0.0085 0.0083 Spilled I .... .... 0.2071 0.2069 2 cc. I 0.0081 0.0082 0.1995 0.2001 0.0478 0.0482 10 cc. I1 0.1201 0.1205 0.0999 0.1001 25 cc. I1 0.1001 0.1001 Spilled 2 cc. I1 .... 0.1201 0.1205 0.1004 0.1001 25 cc. I1 0.1027 0.1224 0.1227 25 cc. I11 0.1026 Spilled 0.1227 25 cc. 111 0.1027 0.1027 0.1226 0.1227 25 cc. I11 0.1030 0.1027 0.1250 0.1251 0.1029 0.1027 25 cc. IV 0.0101 0.0098 0.0084 0.0082 25 cc. IV 0.2501 0.2552 0.0086 0.0082 2 cc. I V 0.2010 0.2008 .... 0.0100 2 cc. IV 0.2008 0.2008 2 cc. I V 0.0102 0.0100 0.2009 1.2008 0.0102 0.0100 2 cc. IV
....
I n the experiments for t h e quantitative separation of aluminum from iron, alcoholic solutions were made up of C. P. (Kahlbaum's) aluminum chloride and C. P. ferric chloride, respectively, in which 5 0 cc. were ap-
Vol. 9 , No.
IO
proximately equivalent t o 0 . 5 g. of salt. Aliquots were pipetted a t definite temperature both for determination of metal content of solution (blank) and for actual separation by t h e proposed method. I n t h e determination of iron, subsequent t o the separation from aluminum by t h e proposed method, experiment has shown t h a t t h e iron residue offers considerable difficulty of oxidation even with nitric and sulfuric acids, due t o the formation of resinous condensation products of which the organic solvent is the source. Precipitation of t h e hydroxide is therefore not recommended. A satisfactory procedure was found in conversion of t h e iron chloride t o sulfate and direct ignition. The method was submitted t o Messrs. C. D. Wright, H. E . Buckbinder, E. K. Nelson and W. F. Kunke, of t h e Bureau of Chemistry, for cooperation. Residues of mixtures of iron and aluminum chloride obtained by evaporation of definite volumes of solution were submitted t o each individual. The results obtained appear in Table 111. TABLEIII-cO6PERATIVE
RESULTSBY BUREAUOF CHEMISTRY WORKERS SOLUTION TAKEN Grams Grams AlCL FeCla -AlnOa--FesOaANALYST Cc. Cc. Found Calc. Found Calc. C. D. Wright.. . . . . . 2 B 25 I1 0.0078 0.0080 . . . 0.2411 25 B 25 I1 0.1007 0.1001 .... 0.1205 SOB 25 I1 0.1998 0.2001 0,0098 H. E. Buckbinder.. . 25 C 25 I11 0.1031 0.1027 0: i i 2 9 0.1227 E. K. Nelson.. . . . . . 25 D 25 IV 0 1009 0.1014 0.1251 0.1251 W. F. Runke ....... 25 C 25 I V 0 1023 0.1027 0.1246 0.1251
.
PROPOSED METHOD
The chlorides of iron and aluminum (not over 0 . 5 g. total) are evaporated t o dryness on t h e steam b a t h in a 150 cc. Erlenmeyer flask (preferably lipped). The residue is stirred occasionally with a glass rod t o allow better exposure of t h e salts t o drying. The rod is allowed t o remain in t h e flask. The residue is then further dried in an oven a t about 120' C. for one-half hour, stirring occasionally as before. When dry, t h e residue is moistened with 0 . 5 cc. t o I cc. absolute alcoholic HC1 solution' ( 2 5 t o 35 per cent HCl), warmed on t h e steam b a t h and stirred t o convert any oxides t h a t have been formed t o chlorides; about 3 or 4 cc. absolute alcohol are added and t h e flask is heated on t h e steam bath until all salts have gone into solution.2 When this has been accomplished t h e flask is allowed t o remain on t h e steam bath until salts begin t o crystallize out and t h e resulting residue is quite viscous but not solid. About 0 . 5 cc. alcoholic hydrochloric acid is now added, t h e flask again warmed and the residue stirred t o have t h e mass uniformly impregnated with t h e HC1. The flask is then removed and 3 0 cc. U. S. P. ether (sp. gr. 7 1 3 t o 716 a t 2 5 O C . ) are added gradually with stirring. The hydrated aluminum chloride should come down as a uniform, granular, white precipitate, leaving t h e supernatant liquid clear. Forty cc. absolute ether are then added, 1 The alcoholic HC1 is prepared by passing HCl gas through absolute alcohol, the gas being conveniently generated by the action of concentrated HBO, on anhydrous calcium chloride. * When AlCla is present t o the extent of nearly I/% g.. and very little iron is present, it is sometimes very difficult to get all salts in solution. With small amounts of iron present the last traces of AlCla need not pass into solution.
OCt., 1917
T H E J O U R N A L OF I N D U S T R I A L A N D ENGINEERING C H E M I S T R Y
with stirring, and t h e flask allowed t o stand. T h e solution, when t h e precipitate has settled, is filtered through a Gooch crucible, using a bell jar arrangement, into a 2 5 0 t o 300 cc. Erlenmeyer flask. T h e crucible is fitted preferably with a small circular filter paper (cut t o size) instead of asbestos. T h e flask should be washed with wash-ether (100parts absolute ether a n d 2 parts alcoholic HC1) w h i l e f i l t e r i n g so as not t o allow a n y ferric chloride t o dry on t h e white precipitate or on t h e crucible. T h e usual precautions of washing t h e flask, crucible a n d funnel are taken t o insure complete transfer of iron. It is not necessary, however, t o transfer all of t h e aluminum chloride precipitate t o t h e Gooch crucible. T h e aluminum precipitate is removed with t h e paper from t h e crucible b y tapping i t into a 2 5 cc. beaker a n d then washing with water t o remove adhering particles. The original flask is also washed t o transfer completely a n y adhering aluminum precipitate. The aluminum chloride solution is diluted t o approximately I O O cc., about 5 g. ammonium nit r a t e are dissolved in i t , a n d made just alkaline‘ with ammonia (using methyl red as indicator, preferably), boiled, filtered a n d t h e precipitate is washed in t h e usual way. The precipitate is ignited in a covered crucible2 and weighed as A1203. T h e ether solution of iron is distilled or evaporated t o remove t h e ether. T h e residue is transferred with a little water a n d HC1 t o a weighed platinum dish, using as little water as possible, a n d evaporated t o dryness on t h e s t e a m b a t h , moistened with I cc. concentrated H2S04, warmed on t h e steam b a t h gently t o expel most of t h e HC1, a n d t h e n slowly over a flame by placing t h e dish on a triangle which in t u r n is p u t on a n asbestos gauze (to prevent too rapid heating), a n d heated until all t h e ferric chloride is converted t o sulfate a n d no further fuming of sulfuric acid takes place. The dish is then heated over a free flame a n d finally over a blast t o entirely convert t h e sulfate t o ferric oxide. This is t h e n cooled a n d weighed as okide in t h e usual manner.
ascertaining t h e presence of aniline, t h e calcium hypochlorite test is employed. T h e quantitative determination is carried out b y precipitating t h e aniline as tribromoaniline with bromine water. The precipit a t e is caught on a tared filter paper, dried in v a c u o , a n d weighed. I n this connection, i t occurred t o t h e writer t h a t i t might be possible t o work out, for estimating t h e aniline, a colorimetric method based on t h e qualitative test with hypochlorite. If this were possible, i t certainly would be desirable, for not only would considerable time and effort be saved b u t probably a considerably smaller sample t h a n I O liters would be sufficient a n d quantitative measurements could also be applied in cases where t h e amount of aniline in t h e samples available, or t h a t can be conveniently collected, is too small for t h e gravimetric determination. A detailed study of t h e hypochlorite test for aniline was therefore undertaken. THE HYPOCHLORITE TEST FOR ANILINE
The hypochlorite test for aniline is described by various authors in rather indefinite language. Thus Blythl in describing this test simply states t h a t a n aqueous solution of aniline or its salts is colored blue by “ a little” chloride of lime or hypochlorite of soda. Heffter,2 although stating t h a t a n excess of t h e reagent should be avoided, does not give any information as t o t h e proper strength of t h e hypochlorite solution or what amount of i t t o use for a given volume of t h e solution t o be tested. All t h e information given is contained in t h e statement t h a t on adding t o t h e aqueous solution of aniline, chloride of lime or sodium hypochlorite solution, a purplish violet coloration appears which later changes t o a dirty red. T h a t t h e proportion of hypochlorite t o aniline a n d t h e degree of their concentration are, however, important factors and hence should be taken into consideration in applying this test, appears conclusively proven b y t h e following results:
BUREAUOF CHEMISTRY DEPARTMENT OF AGRICULTURE WASEINGTON. D. C.
A METHOD FOR THE COLORIMETRIC ESTIMATION OF SMALL AMOUNTS OF ANILINE By
ELIASELVOVE
Received August 27, 1917
I n t h e sanitary examination of air in industrial establishments where aniline is employed, t h e detection a n d estimation of aniline vapors is quite important. A method for this purpose has been worked out b y Hebert a n d Heim.3 According t o these authors, t h e aniline vapors are collected by bubbling I O liters of t h e air through a suitable absorption bulb containing I O cc. of water acidified with sulfuric acid. For 1 W Blum. “Determination of Aluminum as Oxide,” Scientific Papers of the Bureau of Standards No. 286. 2 The difficulties involved in the accurate determination of alumina are not generally known and the reader is referred to the exhaustive work by Doctor Blum OD this subject. 8 Rev. chim. ind., 11, 338-340; from Chem. Abs., 6 (1911), 791.
953
EFFECT OF VARYING THE PROPORTION OF HYPOCHLORITE
T h a t t h e proportion of hypochlorite used for a given amount of aniline is a n important factor in t h e test, even when t h e cdncentration of t h e aniline remains t h e same, and t h a t in certain cases i t may even cause a change in t h e result from ’negative t o positive, is shown by the results obtained in t h e following experiment. T o each of two small tubes there was added 0 . 5 cc. of a 0 . 0 2 per cent aqueous solution of aniline. No. I was then mixed with 0 . 5 cc. of a calcium hypochlorite solution, t h e available chlorine of which was about I per cent; No. 2 was similarly treated with 0.j cc. of a calcium hypochlorite solution, t h e available chlorine of which was only about 0 .I per cent. Immediately after mixing, No. I developed t h e characteristic purplish color b u t No. 2 appeareds practically colorless. 1
“Poisons: Their Effects and Detection,” 4th Ed., p. 285.
8
On standing a little while, however, a slight yellowish color developed.
* Neuberg. “Der Ham.” 1 (1911), 831.
