490 CORBETT AND GLTERIN: I)ETERJII?iATION O F [,4nalyst, 1701. 9 1 The Determination of Aluminium in Iron and Steel BY J. A. CORBETT (Plzj~sicnl ;lIetallu~,qy Secttoit, Conawotiwealth Scieiztifc atad Idatstrial Resravcli Oiyatzisatioiz, 4 Icstvalia) Various colorimetric reagents have been examined for their sensitivity in a standard method for determining aluminium in ferrous metals. Inter- fering elements are removed by a mercury-cathode separation followed by cupferron - chloroforni extraction. In the method adopted, aluminium is determined by measuring the optical density of its complex with -Alizarin red S - calcium reagent. The method has been tested with a wide range of steels. THIS project was undertaken at the request of the Committee on Sampling and Analysis of Ferrous Metals, Standards Association of Australia, as part of its programme of developing standard methods for the analysis of steels.After a survey of published methods the authors concluded that none was completely satisfactory as a standard method applicable to all types of steels. As a result, the investigation described here was undertaken. The determination of aluminium at low levels is particularly susceptible to errors arising from the introduction of extraneous aluminium, and a high degree of analytical skill is required to prevent contamination by minute traces of this ubiquitous element. In addition to this avoidable random introduction of aluminium, there is an unavoidable pick-up of aluminium from impurities in reagents and from dissolution of aluminium from glassware.This need not be a serious problem because, in the hands of an experienced analyst, the blank from these sources can be quite reproducible. The method of analysis should be designed to keep the magnitude of this aluminium pick-up within reasonable limits. As so many analytical reagents can be a source of minute traces of aluminium, a method should in general require only small additions of reagents, restricting, in particular, the use of alkaline reagents and avoiding their contact with glass. The storage of all reagents in polythene containers reduces the risk of contamination. I t should be emphasised, however, that it is the reproduci- bility of the blank rather than its magnitude that imposes the lower limit to the range of aluminium that can be determined, and, in fact, the blank may be greater than the net amount of aluminium in the sample aliquot at very low levels of aluminium.I t is unlikely, howe\rer, that satisfactory reproducibility will be attained with very high aluminium pick-up. P R E: L I hi I N A K Y s E PA K A T I o N s There are no known colorimetric reagents that are specific for aluminium. On the contrary, with all reagents there is only a rather restricted Iist of elements that do not interfere with the colour-forming reaction, and an efficient separation of most of the elements occurring in steels will be necessary. Several methods' 9 * have been described, in which interfering elements form complexes instead of separating. Such methods involve the use o f large amounts of reagents and are applicable only to certain types of steels.For a standard method to deal with all classes of steels and irons, an efficient separation of most of the elements present will be necessary. Blair, Power, Griffiths and M.'ood3 have discussed previously published separations of aluminium from iron which they have classified under the headings : precipitation ; solvent extraction; chromatography; and mercury-cathode electrolysis. They selected the last as the most suitable technique for their method for determining trace amounts of aluminium. \Then used with a c1.c. supply of sufficient power output, electrolysis in a perchloric acid medium over a mercury cathode provides a rapid and elegant method o f removing many alloying elements, including manganese, nickel, chromium and molybdenum, as well as the iron.In view of the requirement to keep reagent additions to a minimum this electrolytic technique is the obvious choice for a preliminary separation.August, 19661 ALChlINIUJl IK IROK AND STEEL 49 1 AIERCURY-CATHODE ELECTROLYSIS- The design of a mercury-cathode apparatus for rapid electrolysis should allow the use of high currents and reasonably low electrolyte temperatures, both factors increase the hydrogen o\-er-voltage on mercury, and contribute to current efficiency. Metals amalgamating with the mercury tend to lower the hydrogen over-voltage so that the use of contaminated mercury may result in imperfect separations. The use of a magnet beneath the cell to attract deposited ferro-magnetic metals below the surface of the mercury can thus improve cell efficiency.In this investigation we used a water-cooled Rfelaven cell operating at 10 amps and a magnetic mercury-cathode cell (similar to the design of Center, Overbeck and Chase4), operating at 12 to 15 amps. Both designs have proved satisfactory for this work, as no doubt would other types of cells designed to operate at these current levels. Electrolysis of 0-5-g samples of plain carbon steels were completed in less than 15 minutes, but molybdenum-bearing stainless steels may require up to 1 hour for completion. Blair3 and his colleagues showed that traces of chloride interfere with the electrolysis, and that repeated fuming with perchloric acid is needed to remove such traces when the composition of the steel, in particular its chromium content, necessitates the use of hydro- chloric acid in the attack of the sample.We have found that if the solution is boiled to allow perchloric oxidation of chromium to its higher valency state, double fuming at this high temperature will remove the chloride. During electrolysis there is a tendency for elements to be deposited in the order of their electrode potentials, although some simultaneous deposition does occur. With stainless steels it is found that complete deposition of nickel and then iron will occur before that of chromium. Molybdenum, if present, will be the last of these alloying elements to be completeljr deposited. Spot tests should be used to test for the completeness of deposition of iron, chromium and molybdenum (see Notes).The spot test for chromi~m,~ in which one drop of solution is used, will detect the presence of 50 pg of chromium in the electrolyte. Tt has not been found necessary to include a spot test for manganese. With steels con- taining up to 2 per cent. of manganese, less than 0-2 mg of manganese remains in the elec- trolyte when deposition of iron is complete. With 12 per cent. manganese steels the presence of manganese in the electrolyte becomes obvious because of the anodic oxidation to perman- ganate or oxides of manganese. When electrolysis is continued until the electrolyte becomes colourless, up to 1.3 mg of manganese remains in the electrolyte. However, even this amount of manganese is below interference level with the principal colorimetric reagents for aluminium.FINAL SEPARATION Of the elements that have been mentioned in the literature as steel constituents, those that would be present in solution after electrolysis are titanium, vanadium, zirconium, beryllium and phosphorus. There could also be traces of iron, manganese, chromium or molybdenum, at levels below the limits of detection of the spot tests. Various methods are available to isolate aluminium from these elements. Sodium hj-droxide was used by Scholes and Smith,6 Hill' and Studlar and Eichleri to precipitate, as hydroxides, some of the elements listed above. The remainder were complesed with hydrogen peroxide to prevcnt interference with the colorimetric reagent used. The probabilitj- of aluminium pick-up from alkaline reagents has been mentioned.A further danger with this technique is the possibility of loss of aluminium as aluminium phosphate when a high phosphorus alloy is encountered. Claassen, Bastings and Visser' used a series of solvent extraction separations with 8-hydroxyquinoline and chloroform, with complexing reagents and different values of yH. A knowledge of the sample composition is necessary, and with some steels a further separation with cupferron is required. A cupferron - solvent extraction seemed preferable to the above techniques and has been widely used for this type of separation. I t was decided to investigate its application tothe present method. CUPFEKKON PRECIPITATION - CHLOROFORM EXTRACTION- Blair, Power, Griffiths and Wood3 have been concerned to remove residual traces of iron which cause severe interference with the Eriochrome cyanine R reaction with aluminium.They made a small addition of cupferron with chloroform extraction of cupferrates to rernox-e492 CORBETT AXD GUERIN : DETERMINATION OF [Analyst, Vol. 91 traces of iron, titanium and vanadium. As an additional precaution they added hydrogen peroxide to ensure that any excess of titanium or vanadium was in the non-interfering oxidised state. If titanium is present in greater than trace amounts it will react with all of the added cupferron and will be extracted, leaving iron present in the aqueous layer. Those workers’ restriction on the amount of cupferron apparently arose from their desire to keep the total aluminium pick-up as low as possible. We preferred to ensure the complete removal of iron, titanium, zirconium, molyb- denum and vanadium at this stage by repeating the addition of cupferron and chloroform extraction until the chloroform extract was colourless.Hydrogen peroxide, which interferes with some colorimetric reactions of aluminium, can now be omitted, so the choice of colori- metric reagent is widened. The acidity of the aqueous phase during chloroform extraction of cupferrates should be controlled. Slight losses of aluminium by extraction occur if the pH is above 0.4, and the extraction of iron is retarded in solutions that are too strongly acid.* Our experiments have indicated that a satisfactory separation can be effected in molar acid solution, so an appropriate addition of acid is made before the separation. The repeated cupferron - chloroform extraction necessary with some alloy steels may result in a slightly increased, but reproducible, pick-up of aluminium. Provided the blank determination for such steels is given identical treatment no error is introduced by this technique.It should be noted that any traces of manganese and chromium in the electrolyte after mercury-cathode electrolysis will not be removed by the cupferron - chloroform extraction. The only other elements likely to be present in irons and steels, and which are not removed by the two separations described or by a preliminary filtration of the insoluble matter, are the aluminium, and also magnesium, beryllium and phosphorus. Assuming that the sample weight is limited to 0 5 g , and the aliquot taken for colour development is not greater than one-fifth of the electrolyte, the levels at which elements could be present in the aliquots are as follows- There is a possibility of error in this technique.Beryllium . . . . . . 200 pg for 0.2 per cent. beryllium alloy Chromium . . . . . . 10 pg assuming the spot test is effective Magnesium . . . . . . 200pg for 0.2 per cent. alloy Manganese . . . . . . 300 pg for 12 per cent. manganese steel Phosphorus . . . . . . 1000 pg for 1 per cent. phosphorus alloy COLORIMETRIC REAGENTS A Unicam spectrophotometer SP600 was used for all colorimetric work described. After studies of the published colorimetric reagents used for aluminium, five reagents were considered to merit detailed investigation. These were- Alizarin red S - calcium reagent. Arsenazo.Eriochrome cyanine R (Solochrome cyanine RS), 8-H ydrox y quinoline. S tilbazo. The investigation of each of these reagents has included: (a) the determination of the optimum wavelength for photometric measurement, from a study of graphs of optical density against wavelength for the reagent and the aluminium complex; (b) conformity with Beer’s law ; (c) calculation of sensitivity, which has been expressed : (i) as the extinction coefficient E with respect to 1 gram atom of aluminium per litre of solution (as advocated by International Union of Pure and Applied Chemistry), and (ii) as the concentration range of aluminium corresponding to the optical density range 0 to 1.0 in the cell size recommended for the particular reagent; (a) interference studies, for the most part confined to those elements which could be present following the two major separations.Elements given as not inter- fering have been tested at least to the levels listed above. The results of the above work are shown in Table I.Alizarin red S - calcium reagent Optimum wavelength 490 mp for colour measurement pH for colour development 4.4 to 4-65 TABLE I RESULTS OF INVESTIGATIONS ON COLORIMETRIC REAGENTS Eriochrome cyanine R Arsenazo (Solochrome cyanine RS) 8-Hydroxyquinoline Stilbazo 580 mp 532 mp 392 mp 520 mp 6.1 t o 6-11 Approximately 6- 1 Extraction Approximately 6.8 a t pH 4.9 t o 5.0 Conformity to Beer’s Conforms Conforms Slight but consistent Conforms Deviations suggesting the the existence of more (0 t o 100 pg of aluminium per 100 ml) law. (Range tested (0 t o 80 pg of aluniinium (0 to 100 pg of aluminium deviations detected shown in brackets.) per 100 ml) per 100 ml ) (0 to 60 pg of aluminium per 100 ml) than one compound (0 to 300 pg of aluminium per 100 ml) E a t optimum wavelength 1.8 x 104 1.2 x 104 6-75 x 104 6.7 x lo3 3.8 x 104 Concentration range 0 t o 80 pg of aluminium per 100 ml in 2-cm cells 0 to 100 pg of aluminium per 100 ml in 2-cm cells 0 to 60 pg of aluminium pcr 100 ml in 0-5-cm cells 0 to 100 pg of aluminium per 100 ml in 4-cm cells 0 t o 70 pg of aluminium per 100 ml in l-Cm Cells and cell size Interferences : Effect on optical density a t optimum wavelength- { Slight increase 40 pg Be = lpg A1 None Slight increase Drastic increase Drastic increase ( a ) Beryllium { 0 * 7 p g R e = l p g A l ‘ (2 pg Be = 1 pg A1 (b) Other elements None None.Higher levels of None. Higher levels of None None { 4 0 p g H e - l p g A l listed on page 492 chromium cause increase chromium cause decrease a t levels specified Notes on reagent The reaction of sodium The orange coloured Acidified (nitrated) alizarin sulphonate aqueous solution forms solutions of Merck’s with aluminium in the a cherry-red complex Eriochrome cyanine R presence of calcium with a1urnini~m.l~ (The and Gurr’s Solochrome ions has been used in reagent used was from cyanine RS were found analysis of rocks, slags Tokyo Kasei Kogyo to give identical re- and coal ash.l0~l1~l* Co. Ltd., Japan.) actions. This reagent has been widely used for aluminium deter- minations in steel^.^^^*^ Aluminium hydroxy- quinolinate is extracted from the aqueous phase with chloroform vielding a pale yellow analyses.16J7 solution.This reaction has been used in steel14 and cast-iron16 analyses. The reaction with alu- minium to form a reddish brown complex has been used in steel G 0 Z v1 4 M M r bP CD W494 C‘ORBETT AND GVERIK DETER311KL4TIOK OF [Analyst, lrol. 91 PRACTICABILITY- In addition to the investigations outlined above we have introduced the concept of the “practicability” of a colorimetric reagent and have considered the five reagents in this respect. Under this term we have included rapidity of development of the coloured complex and its stability, the stability of the reagent solution and the effect on the optical densit?.of slight changes in conditions such as pH and buffer concentration. All of these factors play a part when the reproducibility of a method is determined experimentally. Another factor is the optical density of the reagent. If the reagent absorbs appreciably at the optimum wavelength it is not practicable to use a large excess of reagent and there is a tendency for the calibration graph to deviate from linearity towards the upper end, unless the equilibrium constant for the complex formation is high. Absorption by the reagent reduces the slope of the calibration graph, and also, in the type of instrument used here, limits the cell size that can be used without unduly opening the slit. CHOICE OF COLORIMETRIC REAGEXT- Each of the 5 colorimetric reagents investigated in detail has sufficient sensitivitj- for determining aluminium in steels and irons.We have found that the lower limit of detection of aluminium is determined by the reproducibility of aluminium pick-up in the blank and sample rather than by the sensitivity of the colorimetric reagents tested. I t would seem, then, that the choice of reagent for this standard method should depend primarily on the reproducibility of the colour-forming reaction rather than its sensitivity. One criterion for a standard method is that it should give satisfactory results in the hands of a competent analyst. On this basis all of the above reagents can be considered as satisfactory as it is possible to obtain accurate reproducible results for aluminium with each reagent, provided the relevant conditions are sufficiently closely controlled.The factors affecting repro- ducibility are different for each reagent, and with some of the reagents are difficult or irksome to control in practice. I t was therefore considered that reproducibility tests, made under the ideal conditions for each reagent, would not provide a realistic basis for comparison. The factors affecting reproducibility, that we have discussed in the section on “Practicability,” provide results for a more realistic appraisal of the reproducibility which could be expected under laboratory conditions, and it was consideration of these factors that guided our final choice of reagent for the standard method. Close control of pH and buffer concentration should present no difficulty provided that the analyst takes the elementary precaution of preparing a sufficient volume of buffer solution to deal with all samples, blanks and standards in the batch of analyses.Close control over time of standing is, however, irksome, when measurements are being made against a reference solution whose density is also time-dependent. Strict adherence to Beer’s law is desirable as it obviates the need for close plotting of the calibration graph with each batch of analyses. With Eriochrome cyanine R close control over time of standing is essential because the optical densities of both the aluminium complex and reference solutions decrease on standing. There are slight deviations from Beer’s law. Stilbazo has similar defects, the deviations from Beer’s law being slightly more severe.The lower sensitivity of the 8-hydroxycluinoline complex is compensated by the low absorption of the reference solution which permits the use of larger cells and smaller dilutions for absorption measurements. The colour development in\wlving a chloroform extraction is not as simple as with the other reagents, and the coloured solution is light sensitive and subject to the disadvantage of a volatile liquid. Aqueous solutions of arsenazo are stable, and the optical density of the aluminium complex remains constant in the interval from 1 to 5 hours after d o u r development. The reagent itself has a low absorption at the wavelength used. I t suffers from severe interference from beryllium. This is not regarded as a serious defect; however, this element is found in a few special stainless alloys and experimental batches of steels. Aqueous solutions of Alizarin red S are stable for about 2 weeks and the optical density of the calcium - aluminium complex remains constant in the interval from 1 to 4 hours after colour development.Absorption by the excess reagent is not severe. We are of the opinion that arsenazo and Alizarin red S - calcium reagent are the two most suitable reagents for use in a standard method for determining aluminium. The latterAugust, 19661 ALUMINIUM I N IRON AND STEEL 495 has the slight advantage that interference from beryllium is almost negligible. In deciding to recommend the Alizarin red S - calcium reagent to the Australian Committee on Analysis of Ferrous Metals we were further influenced by the wealth of experience in its use by analytical chemists in the analysis of non-metallic materials.APPLICATION OF ALIZARIN RED s - CALCIUM REAGENT TO THE METHOD- The aluminium, before colour development, is in dilute perchloric acid solution. The pH for colour development may be in the range 4-4 to 4.65, but it must be kept uniform to within +0-02 units. The acid solution can be brought directly to the required pH by adding a high concentration of buffer solution, but this addition was found to cause a marked decrease in the optical density of the aluminium complex. As with the other colorimetric reagents for aluminium the buffer concentration must be kept reasonably low for efficient colour de- velopment. With the resulting low buffer capacity it is necessary that before colour develop- ment the solution is neutralised with sodium hydroxide followed by a small measured excess of acid to re-dissolve the aluminium.The addition of 10 ml of a buffer of M sodium acetate - M acetic acid (pH 4.75) brings the final pH within the required limits. As sodium hydroxide is a likely source of trace aluminium, the amount required to neutralise the solution has been kept to a minimum by evaporating the perchloric acid solution almost to dryness before neutralisation. The addition of 7 mg of Alizarin red S provides sufficient excess of reagent to ensure linear calibration over the range 0 to 80 pg of aluminium. Ideally, 0-5-g portions of high purity iron should be used in the blank and standard. As “pure” irons, commercially available, contain traces of aluminium, this addition is not recommended.The time required to complete the determination of aluminium in a single steel sample is approximately 8 hours. METHOD The method is applicable to all steels and irons, and has a range from 0.002 to 10 per cent. of aluminium. APPARATUS- 10 amps and should be water-cooled. solution at 490 mp may be used. Mercury-cathode cell-The cell should be designed to operate at a current of at least Spectro$hotometer-Any instrument suitable for measuring the optical density of a REAGENTS- All reagents should be of the highest purity obtainable and distilled water should be used throughout. Certain types of analytical-grade reagents are unsuitable because of the presence of either aluminium or other impurities.All solutions should be stored in polyethylene or polypropylene containers. Cupferron, 2 per cent. w/v-Dissolve 2 g of cupferron in 50 ml of water and dilute the solution to 100 ml. This solution should be colourless and must be prepared each day. Sodium hydroxide, 2 M-Dissolve 80 g of sodium hydroxide pellets in 700 ml of water in a polyethylene container, cool the solution and dilute it to 1 litre. Hydrochloric acid, 0.2 M-Dilute 18 ml of hydrochloric acid (sp.gr. 1.18) to 1 litre. PhenolphthaZein indicator-Dissolve 0.1 g of phenolphthalein in 50 ml of ethanol and dilute the solution to 100ml with water. Calcium chloride-Dissolve 14 g of calcium carbonate in 50 ml of hydrochloric acid (50 per cent. v/v). Boil the solution for 2 minutes. Cool and dilute to 1 litre.Bufler solution-Dissolve 140 g of hydrated sodium acetate (CH3COOWa.3H,O) in water. Add 60 ml of glacial acetic acid and dilute to 1 litre. Alizarin red S solution, 0.14 per cent. w/v-Dissolve 140 mg of Alizarin red S in 75 ml of water and dilute the solution to 100ml. Chromium spot-test, solution A-Dissolve 10 g of sodium hydroxide pellets in 50 ml of water, cool the solution and add to it 50 ml of hydrogen peroxide (6 per cent.). Chromium spot-test, solution 23-Dissolve 0.5 g of diphenylcarbazide in 50 ml of glycerol. (This solution is satisfactory for several days.) To 5 ml of the glycerol solution add 5 ml of Filter if necessary.496 CORBETT AND GUERIN: DETERMINATION OF [Anahst, VOl. 91 sulphuric acid (25 per cent. v/v) and 5 ml of glacial acetic acid.(This solution is satisfactory for 1 to 2 hours.) Iron and molybdenzm spot-test solution C-Dissolve 5 g of tin(I1) thiocyanate (SaSCS.H,O) in 50 ml of water and dilute the solution to 100 ml. Iron and molybdenum spot-test solzition D-Dissolve 32 g of stannous chloride l(SnC1,.2H,O) in 40 ml of hydrochloric acid (sp.gr. 1-18) and dilute the solution to 100ml with water . PROCEDURE- Transfer 0-5 g of sample to a 100-ml beaker, add to it 10 ml of nitric acid (50 per cent. v/v) and allow it to digest until the solvent action ceases (Sote 1). Add 5 ml of perchloric acid (60 per cent.) and evaporate to fumes of perchloric acid. Allow the mixture to fume for 1 minute with the cover removed (Note 2). Cool the residue, add to it 10 ml of water, heat to dissolve soluble salts and filter the mixture through a small filter-paper.Wash the filter- paper with hot water and reserve the filtrate (A) (Note 3). Transfer the paper to a platinum crucible, char, then ignite it at a temperature not exceeding 1000" C. Cool, and moisten the residue with 5 or 6 drops of dilute sulphuric acid (20 per cent. v/v), add to the residue 2 ml of hydrofluoric acid and evaporate to dryness. Heat the residue to 800" C for several minutes, then fuse it with 0-5 g of sodium hydrogen sulphate. Cool the mixture, add 10 ml of water and dissolve the solid by heating (Kote 4). If the total aluminium is required add this extract to filtrate (A). If separate results are required for acid-soluble and acid-insoluble aluminium, treat the extract as described in Xote 5.Transfer the solution to the mercury-cathode cell with a minimum amount of water. The volume of electrolyte should not exceed 70 ml. Electrolyse at 10 to 15 amps, washing down the cover and inside of the cell with water after 30 minutes. Continue the electrolysis until deposition is complete, i.e., until spot tests indicate that iron or, if present, chromium and molybdenum have been removed from the electrolyte (Note 6). Remove the electrolyte and filter it immediately into a 100-ml standard flask (Kote 7) with the minimum volume of water for washing, and make the solution up to the mark. Transfer by pipette a 20-ml aliquot (Xote 7) into a 200-ml separating funnel. Add to the solution 2 ml of hydrochloric acid (50 per cent. v/v) and mix. Introduce 1 ml of cupferron solution (2 per cent.w/v), shake the mixture, and allow it to stand for 5 minutes. Add 15ml of chloroform, shake the solutions for 30 seconds, allow the two phases to separate and run the chloroform layer into a beaker. Extract the aqueous layer with a further 10 ml of chloroform, then run off the chloroform layer. Add 1 ml of cupferron (2 per cent. w/v) to the aqueous portion, mix, and allow the solutions to stand for 5 minutes. Add 10 ml of chloroform and shake the solutions for 30 seconds, allow the layers to separate, note whether the chloroform layer is coloured (Note 8) and run off the chloroform layer. Run the aqueous layer into a 100-ml beaker. Evaporate the aqueous portion to about 5 ml, add 1 ml of nitric acid (50 per cent. v/v) and evaporate to fumes of perchloric acid.Continue the evaporation until no free liquid is visible although fumes of perchloric acid are still being emitted. If drops of perchloric acid remain on the beaker wall, carefully wash down with water and repeat the evaporation. Cool, add 10 ml of water and warm to dissolve salts. Cool, add two drops of phenolphthalein solution and add sodium hydroxide (2 M) from a polythene wash-bottle until the colour just changes to pink. KO more than 2 to 4 drops of sodium hydroxide should be required. Titrate with 0-2 M hydrochloric acid until the solution becomes colourless and add 1.0 ml in excess. Add, in order, with a burette and shaking the solution after each addition, 2 ml of calcium chloride solution 10 ml of buffer solution and 5 ml of Alizarin red S solution (0.14 per cent.w/v); dilute the solution to the mark with water. Allow the solution to stand for 1 hour, transfer it to a 2-cm cell and measure the optical density a t 490mp against the reference solution. (See under Calibration.) Transfer the solution to a 100-ml calibrated flask and dilute to 50 ml. REAGENT BLANK- Each sample must be accompanied by a reagent blank solution. The treatment of the blank must be identical with that of the sample throughout the method. Measure the optical density a t 490 mp against the reference solution. (See under Calibration.)August, 19661 ALUMINIUM I N IRON AKD STEEL 497 CALIBRATION- Aluminium solution-Dissolve 1.757 g of aluminium potassium sulphate (A1,(S0,),K2S0,.24H,0), in distilled water, add 1 ml of sulphuric acid (sp.gr.1.84) and dilute to 1 litre. Dilute 100 ml of the above solution to 1 litre with water. 1 ml = 10 pg of aluminium. Calibration procedure-To 50 ml of water in a 100-ml calibrated flask add 1.0ml of 0.2 M hydrochloric acid, and proceed with colour development as described under “Pro- cedure.” Allow the solution to stand for 1 hour. Each sample must be accompanied by a standard. This is prepared by measuring with a burette a 35-ml portion (Note 9) of the diluted aluminium standard solution into a 100-ml beaker, and proceeding through all steps of the method. The treatment of the standard must be identical with that of the sample. Measure the optical density a t 490mp against the reference solution and deduct the reagent blank to give the optical density due to 70 pg of aluminium in the aliquot.CALCULATIONS- Deduct the reagent-blank optical density from the test-solution optical density and convert to weight of aluminium in the aliquot taken, by reference to the calibration standard. Hence calculate the percentage of aluminium in the sample. This is the reference solution. NOTES- An addition of 5 ml of hydrochloric acid (sp.gr. 1.18) is suitable. Evaporate to fumes of pcrchloric acid, and with the cover on the beaker boil the solution vigorously for 30 seconds to oxidise the chromium. Remove the cover and allow the mixture to fume freely for another 30 seconds. Cool, wash down the sides of the beaker with water and repeat this evaporation to fumes, boiling and fuming, to remove the last traces of chloride. 3.If tungsten is present remove the filtrate and wash the paper with 10 ml of ammonia solution (50 per cent. v/v), then with water. 4. The fused residue from some complex alloys may not completely dissolve in water. If i t does not, transfer the extract to a 100-ml beaker and boil i t to dissolve the residue as far as possible. Add the solution and the insoluble residue to the main filtrate A. 5. For the determination of acid-insoluble aluminium, add to the extract, filtered if necessary, 1 ml of perchloric acid (60 per cent.), 2 ml of hydrochloric acid (50 per cent. v/v) and continue with the cupferron extraction as described under “Procedure.” 6. In each of the following spot tests, 1 drop of the electrolyte is placed in a small porcelain crucible or in the depression of a porcelain spot plate.I ~ o n - ~ i d d 1 drop of nitric acid (50 per cent. v/v) and 1 drop of the sodium thiocyanate solution C. A red colouration indicates the presence of iron. Chromium--Add 1 drop of solution A , mix and then add two drops of solution B, and mix. The resulting solution should be acid. An immediate purple colouration indicates the presence of chromium. MoZybdenum--hdd 1 drop of solution C and 2 drops of solution D. ,4 pink to red coloration indicates the presence of molybdenum. 7 . The aliquot must not contain more than 7 0 pg of aluminium. When necessary the sample weight, the dilution, or the aliquot can be adjusted to conform to this limit. If the aliquot is less than 20 ml i t should be diluted to 20 in1 with water.The aliquot should be approximately 0.5 molar with respect to perchloric acid. If necessary, make an appropriate addition of perchloric acid (60 per cent.). 8. If this chloroform extract is coloured, repeat the addition of 1 ml of cupferron (2 per cent. w/v) and the extraction \vith chloroform, until the chloroform extract is colourlcss. 9. The aliquot for colour development should contain 70 pg of aluminium. With standards accompanying steels containing more than 0.07 per cent. of aluminium a proportionately larger initial amount of aluminium will be necessary. 1. The addition of hydrochloric acid is necessary to effect dissolution of some alloy steels. 2. If hydrochloric acid has been used special care must be taken to remove it. RESULTS Tests were made on a group of X.B.S.and B.C.S. steels with aluminium contents ranging from 0.002 to 6.98 per cent. The results are shown in Table IT. These results show that aluminium can be determined with satisfactory accuracy. Statistical tests at low levels of aluminium indicated that the lower limit of detection of the method should be not greater than 0.001 per cent. of aluminium. At this lower limit a pick-up of 6 pg of aluminium in the blank-determination aliquot was sufficiently reproducible to enable an additional 1 pg of aluminium in the sample aliquot to be detected.498 CORBETT AND GUERIN [Anazyst, VOl. 91 TABLE I1 RESULTS OF TESTS ON N.B.S. AND B.C.S. STEELS, ALUMINIUM CONTENT Certificate value, percentage of Sample aluminium N.B.S. 106 . . . .1.06 N.B.S. 55e f . . . 0.002 N.B.S. l7Oa . . . . 0.046 (low alloy steel) (ingot iron) (open hearth steel) (high speed steel) (mild steel) (magnet alloy) (1 8/ 12 stainless with niobium, molybdenum) * B.C.S. 241/1 . . .. B.C.S. 271 .. . . 0.008 B.C.S. 2337 . . . . 6.98 B.C.S. 246 ,. .. t 0.02 TO 6.98 PER CENT. Laboratory 1 Laboratory 2 Mean result 1.060 0.0032 0.0483 0.0229 0.0083 7.05 0.0027 No. of deter- mina tions 5 5 3 5 5 5 3 I Standard deviation 0.009 0*0009 0.0006 0.0019 0*0002 0.041 0*0002 Mean result 1.064 0-001 3 - 0.0246 0*0080 7.07 0.0026 No. of deter- Standard minations deviation 5 0.01 1 5 0.0005 0.00 13 5 5 0*0005 5 0.059 - 2 * No certificate value is given for B.C.S. 241/1. Scholes and Smith6 found 0.024 per cent. of aluminium. 7 Certificate and experimental values for B.C.S. 233 are for soluble aluminium. $ No certificate value is given for B.C.S. 246. Further reproducibility tests are now being carried out by a panel of fellow members of the Ferrous Analysis Committee in order to determine the 95 per cent. confidence limits for incorporation in the proposed standard. The authors wish to acknowledge helpful discussions a t meetings of the Committee on Sampling and Analysis of Ferrous Metals, Standard Association of Australia. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. REFERENCES Claassen, -4., Bastings, L., and Visser, J., Analytica Chim. Acta, 1954, 10, 373. Hill, U., Analyt. Chem., 1959, 31, 429. Blair, D., Power, K., Griffiths, D. L., and Wood, J. H., Talanta, 1960, 7, 80. Center, E. J., Overbeck, R. C., and Chase, D. L., Analyt. Chem., 1951, 23, 1134. Evans, B. S., and Higgs, D. G., Analyst, 1945, 70, 75. Scholes, P. H., and Smith, V. D., Iron Steel Inst., 1962, 200, 729. Studlar, K., and Eichler, V., Chemist Analyst, 1962, 51, 68. Short, H. G., Analyst, 1950, 75, 420. Parker, C. A., and Goddard, A. P., AnaZytica Chim. Acta, 1950, 4, 517. Shapiro, L., and Brannock, W. W., Bull. U.S. Geol. Surv., 1956, No. 1036-C. Archer, K., Flint, D., and Jordan, J., Fuel, Lond., 1958, 37, 421. British Standard 1016 : Part 14, 1963. Kuznetsov, V. I., and Golubstova, R. €3.. Zav. Lab., 1955, 21, 1422. Sandell, E. B., “Colorimetric Determination of Traces of Metals,” Third Edition, Interscience Rooney, R. C., B.C.I.R.A. J. Res. and Dev., 1958, 7, 436. Jean, M., Analytica Chim. Acta, 1954, 10, 526. Wetlesen, C. U., Ibid., 1962, 26, 191. Publishers Inc., New York, 1959, 243. Received May 26th, 1965