546 ROONEY: THE DETERMINATION OF TRACE AMOUNTS OF The Determination of Trace Amounts [Vol. 83 of Aluminium in Cast Iron BY R. C. ROONEY (The British Cast Iron Research Association, Bordesley Hall, Alvechurch, Birmingham) A method is described for determining trace amounts of aluminium in cast iron. After preliminary separation of the major constituents by extrac- tion as their diethyldithiocarbamate complexes into chloroform, the aluminium is selectively extracted as cupferrate at pH 4.5. The final determination is made by the polarographic procedure of Willard and Dean. The determination of the acid-insoluble aluminium fraction is described, and the purity of reagents is discussed. The method is also applicable to steel. RECENT work at the British Cast Iron Research Association and elsewhere has shown that small amounts of aluminium can have marked effects on the properties of cast iron, particu- larly with regard to pinholing1 inoculation2 and annealabilit~.~ ,4J5 In order to investigate these effects, it has become necessary to develop analytical methods for determining aluminium at concentrations well below 0.01 per cent.It was also considered desirable t o develop a method to cover the range 0.01 to 0.2 per cent.; this latter figure represents the useful lower limit of the fluoride volumetric method previously developed in these laboratories.6 Colorimetric procedures for iron and steel, in which lake-forming reagents, such as Eriochrome cyanine and aluminon, are used have been reported,',* s9 but these reagents have disadvantages. Close control of conditions is necessary if reproducible results are to be obtained, and the reagents are subject to interferences, especially from iron. Measurement of the colour of aluminium 8-hydroxyquinolat e in chloroform has been used,l0+ but this reagent is again subject to interference by iron and many other constituentsOct., 19581 ALUMINIUM IN CAST IRON 547 of cast iron. Other methods suggested include turbidimetric measurements with cupferron as reagent12J3 and fluorimetric measurements with reagents such as Pontachrome blue- black R.14 The polarographic method of Willard and Dean15 is very successful for steels containing more aluminium than titanium and vanadium, but in cast iron these elements are usually present in far greater amounts than the aluminium. The mercury-cathode separation used by these workers does not separate aluminium from titanium and vanadium, and, when microgram amounts of aluminium are involved, there is interference from the residual amounts of nickel, chromium, lead, etc.in the electrolyte. The chromatographic separation used by Bishop16 suffers from the same defects, as separation from titanium, vanadium, lead, cerium, zirconium, chromium, nickel, cobalt and several other possible micro and semi-micro constituents of cast iron is either not effected by this procedure or only partly effected. The small-scale mercury-cathode electrolysis recommended by Bishop after elution of the aluminium is effective for removing some of these elements, but not all. It is obvious that, in order to determine aluminium satisfactorily by any of the colori- metric procedures at the microgram level, it must be completely separated from all traces of the other constituents of the cast iron.In practice, it was found to be impossible to obtain the aluminium completely free from other metals, especially in a laboratory in which other analysis was being carried out. It was found that air-borne contamination of the solutions with microgram amounts of iron was particularly difficult to prevent, and led to the presence of iron in the final solutions, although it could be shown that, at the intermediate stages, the solutions were completely iron-free. The polarographic procedure is more tolerant to many metals in small amounts, and, if a polarograph with good resolution is used, interfering elements, such as iron, titanium and nickel, can be tolerated so long as their concentration can be reduced below that of the aluminium.Because of this, the polarographic method of determination was used; a cathode- ray polarograph gave the necessary sensitivity and resolution. POLAROGRAPHY- The polarographic determination of aluminium with Solochrome violet RS was investi- gated by using the conditions recommended by Willard and Dean. When 20.0 ml of a 0.05 per cent. aqueous solution of Solochrome violet RS and a final volume of 50 ml were used, a straight-line graph was obtained over the range 50 to 200pg of aluminium. At 40pg, measurement of the aluminium peak became somewhat difficult, owing to the previous reduction of the dye-stuff.A second graph was plotted, for which 5.0 ml of dye solution were used, and it satisfactorily covered the range 10 to 60 p g ; a third graph, for which 1.0 ml of reagent was used, covered the range 1 to 15 pg. The blank value of the cell reagents was originally 4 pg, but, after a considerable amount of work (see “Purity of Reagents,” p. 548) this was reduced to 1 p g ; work below this level was deferred until better reagents are available. These graphs had different slopes, the factors being 1 pA = 24, 14 and 12 pg, respectively. This effect is ascribed to increases in the viscosity leading to lower diffusion coefficients in the solutions containing large amounts of dye-stuff. The range of aluminium that could be determined was further extended by using one-fifth of the volume of all reagents in a final volume of 10 ml.These conditions were used in conjunction with the graph covering the range 1 to 15 pg to give a range of 0-2 to 3.0 pg. Because of the difficulties that were encountered with regard to blank values, this range was rarely used. The effect of various elements under the conditions used for the polarographic deter- mination was next examined. By using 100 pg of each element and 20 ml of dye solution, most of the elements likely to be present with the aluminium were examined. Interference was found to be caused by ferric and ferrous iron, titanium, vanadium, nickel, cobalt, zir- conium and chromium. This agrees with the work of Perkins and Reynolds,l’ who also found interferences from large amounts of cadmium, molybdenum, thorium, antimony, lead, copper and tin.As many of these elements are likely to be present in cast iron at concentra- tions equal to or greater than that of the aluminium, any separation procedure must remove them. SEPARATION OF ALUMINIUM- most likely to be useful was one in which the smallest amounts of reagents were used. EXPERIMENTAL It became obvious duringTthe early part of this work that the separation procedure Great648 ROONEY: THE DETERMINATION OF TRACE AMOUNTS OF [Vol. 83 difficulty was encounteredwith aluminium in thereagents, aluminium in filter-paper, aluminium leached from the glassware and many other sources of contamination. Rosotte’s methodla was used as a starting point for the work, as she had already reported the interference of residual iron after separation at a mercury cathode, together with that of titanium and vanadium, and had overcome this.No mention was made, however, of residual amounts of elements, such as manganese and chromium, which are difficult to deposit completely in mercury. Rosotte removed the residual iron, together with titanium and vanadium, by extraction of the cupferrates at pH 0.3, and, after destroying the excess of cupferron, adjusted the pH to 3.9 and added more cupferron. The turbidity caused by the finely divided aluminium cupferrate was used to determine its concentration, but it seemed preferable to extract this aluminium cupferrate in order to use the more satisfactory method of determination. First, the pH conditions for extraction were examined by using 250 pg of aluminium and 1.0 ml of a 1 per cent.solution of cupferron. The results, corrected for the blank value, were as follows- pH .. .. . . 0.5 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 Aluminium found, pg .. Nil 0.35 155 196 240 246 228 209 10.7 From these results it is apparent that a preliminary extraction with cupferron would be satisfactory, provided that the pH is kept below 1.0, and that, for efficient extraction of the aluminium, the pH must lie between 4 and 5. This is readily achieved by the use of an acetate - acid buffer system. Under these conditions, extractions were made over a range of aluminium concentrations (the extractions of 0.1 and 1.0 pg were made at a later stage with lower blank values). The results, corrected for the blank values, were as follows- Extraction, yo .. .. Nil 0.14 62 79 98 99 91 84 4 Aluminium added, pg . . . . 0.1 0.1 1.0 1.0 10.0 50 100 200 Aluminium found, pg . . . . 0.1 0.1 1.2 0.96 9.8 61 96 205 Recovery, % . . .. .. 100 100 120 96 98 102 98 103 The recoveries were considered to be satisfactory. Extraction of 10 pg of aluminium was also successfully accomplished in the presence of 10 mg of phosphorus pentoxide; phosphate in solutions of high-phosphorus irons will therefore not interfere. As 1 pg of aluminium can be determined readily, it was considered likely that aluminium contents of 0.001 to 0-2 per cent. could be determined easily by using a 0-1-g sample of iron: this corresponds to the range of the graphs, 1.0 to 200 pg. For cast irons containing very little nickel, chromium or cobalt, no preliminary separation of these elements should be necessary, so that a direct cupferron separation of iron, titanium, vanadium, etc., could be made, followed by adjustment of pH and extraction of the aluminium.This procedure was used and a blank value of 14 pg was obtained, equivalent to 0.014 per cent. The procedure was applied to a further series of samples after a preliminary separation at a mercury cathode; a blank value of 11 pg was obtained. It became apparent that an investigation of the purity of the reagents was necessary. PURITY OF REAGENTS- The reagents used in the experimental work were examined by various concentration and extraction procedures in order to determine their aluminium contents; some alternative reagents were also examined.The results are shown in Table I. It was found that ashless filter-papers could contribute 2 to 3 pg of aluminium; silica and graphite residues were therefore removed by centrifugation in all the subsequent work. Samples of lead-free perchloric acid, which had been redistilled, were obtained from the British Drug Houses Limited and from Hopkin and Williams Limited. The aluminium contents were found to be of the order of 0.05 to 0.1 p.p.m., which indicates the difficulty associated with the further purification of reagents. It was decided to use lead-free acids when possible and to attempt to purify the cupferron and sodium acetate. Many methods for both reagents were used, including selective solvent extraction with a number of reagents and solvents, ion exchange under various conditions with both anion and cation-exchange resins, recrystallisation and synthesis of the reagents under closely controlled conditions. However, the aluminium content of the reagents wasOct., 19581 ALUMINIUM I N CAST IRON 549 never significantly decreased, and it is suggested that this is probably due to the aluminium in these neutral reagents being present as aged aluminium hydroxide, which would be comparatively unreactive.TABLE I ALUMINIUM CONTENT OF REAGENTS Reagent Grade Perchloric acid . . . . Sulphuric acid . . . . Hydrochloric acid . . .. Nitric acid . . .. .. Hydrogen peroxide, 100-volume Sodium acetate . . . . Cupferron .. .. Sodium diethyldithiocarbamate . . Chloroform . . .. .. a . . . .. .. .. .. Aluminium content, p.p.m. 3.0 1.0 0.65 0.83 0.032 0*088 4.4 <0.001 <0*001 0.25 ..“Lead free, for foodstuffs ana1vsis”t 0.007 Perchloric acid . , Hydrochloric acid . . Nitric acid . . .. * . - ‘ 1 0.034 * Samples of reagents of similar grade were examined and found to have similar aluminium contents. t When aged in glass bottles, the aluminium content rises appreciably. In view of the high aluminium content of cupferron, it was decided to use sodium diethyl- dithiocarbamate for the preliminary separation of iron. This would also remove copper, nickel, cobalt, manganese and vanadium,lS and recent work in our laboratory has shown that the bulk of the titanium is also extracted.lS Any residual titanium will be detected and can be separated. No more attempts were made to purify sodium acetate further; each batch used was tested for aluminium and any very impure batches were rejected.APPARATUS- Oelschlager20 reported that microgram amounts of aluminium could be leached from glassware by a variety of reagents. This work was confirmed, and it was found that a blank value of 6 pg when glass apparatus was used could be reduced to 1.8 pg if silica or polythene apparatus was substituted whenever possible. Silica and polythene apparatus was there- fore used when possible in the subsequent work. APPLICATION OF THE METHOD TO CAST IRON ACID-SOLUBLE ALUMINIUM- Methods were devised for determining aluminium in cast iron and were checked by investigating the recovery of aluminium added to samples of pure iron. The results are shown in Table 11.An outline of the procedures used, which are described fully under “Method,” p. 550, is as follows- Procedure A-Iron extracted as diethyldithiocarbamate from a 0.1-g sample. Blank value = 2.1 pg = 0.0021 per cent. of aluminium. Procedure B-Iron extracted with isobutyl acetate from a 5-g sample. Blank value = 22 pg E 0.00044 per cent. of aluminium. Procedure C-Electrolysis - dissolution procedure on a 5-g sample. Blank value = 1.2 pg = 0.000024 per cent. of aluminium, I t is apparent that the useful lower limit of the method is set by the blank value. In our laboratory, for a figure to be regarded as reliable, it must not be lower after correction than the blank value alone, i.e., the “signal-to-noise” ratio must not be less than 1 to 1. In order to obtain the lowest possible blank value for the lowest aluminium contents, the sample must be dissolved in a very small amount of acid and the bulk of the iron separated by using a minimum of reagents. This problem was solved in an elegant manner by Chirnside, Cluley and Proffit21 in their method for the determination of boron in nickel strip.The solid sample is used as anode in a mercury-cathode electrolysis cell, a small amount of acid is added and dissolution and deposition proceed simultaneously. By using a high current density and feeding the sample into the electrolyte as it dissolves, 5 g of iron can be dissolved550 ROONEY: THE DETERMINATION OF TRACE AMOUNTS OF [Vol. 83 in about 2 hours in only 1-0 ml of perchloric acid. The extraction of the elements remaining in solution (mainly titanium and vanadium) also requires small amounts of reagents.TABLE 11 RECOVERY OF ALUMINIUM ADDED TO SAMPLES OF PURE IRON Aluminium PLg Nil 10 Procedure Basis material added, -[ 1:: A Commercially pure iron . . 200 50 ‘.{ ;:: B Spectrographically pure iron r- 1.0 5.0 C Spectrographically pure iron “i 10 50 Aluminium found (corrected for blank value), Recovery, Pg % 0.9’ - 9.8 OR 52 1 105 1 199 49 98 101 200 101 100 0.98 98 5.0 100 9.9 99 50 100 * This figure is lower than the blank value and should be neglected. If solid samples or the electrolysis apparatus are not available, the lower limit can be extended below 0.002 to about 0.0004 per cent. by using a 5-g sample and extracting the iron with isobutyl acetate. The 11 per cent. loss of aluminium reported by Werz and Neu- berger,22 who extracted the iron with diethyl ether, is not confirmed, and it is suggested that this discrepancy is caused by the relatively high solubility of diethyl ether in hydrochloric acid compared with the almost complete insolubility of isobutyl acetate.For large numbers of samples with aluminium contents greater than 0.0004 per cent., procedure B is often more convenient than procedure C. The interfering elements left in solution in all procedures are most likely to be titanium, cerium, zirconium, etc., if present, and chromium. Of these, titanium, cerium, etc., will be extracted quantitatively with the aluminium as cupferrates, and, if the pH is greater than 4.2 to 4.3, some chromium may also be extracted. The presence of most of these elements with the aluminium will be obvious, however, and they can be removed fairly simply.ACID-INSOLUBLE ALUMINIUM- Because the aluminium present can be partly extracted from filter-papers by acid solutions, it was decided to determine the acid-insoluble fraction on a different sample. This would allow the use of a large sample weight, as the aluminium in the acid used to dissolve the sample would pass through the filter-paper and be discarded; the large sample weight would minimise the effect of the blank values contributed by reagents such as sodium carbon- ate and hydrofluoric and sulphuric acids, which are difficult to purify. It has been found with several samples that the results for acid-insoluble aluminium are erratic; one particularly bad example gave results ranging from 0.00016 to 0.00052 per cent.As other samples gave reproducible results, it is thought possible that there is a tendency to marked heterogeneity in the distribution of the insoluble aluminium throughout some samples. The methods have been applied to a number of samples; representative results are shown in Table 111. I t is considered that these results, with the exception of the insoluble fractions in the samples of grey iron and iron containing 4 per cent. of chromium, show satisfactory reproducibility. METHOD APPARATUS- Normal volumetric glassware is satisfactory, but solutions should not be allowed to remain in contact with it for longer than necessary; pipettes should not be left standing in bottles of reagents. Polythene storage bottles should be filled with concentrated hydrochloric acid and set aside for 48 hours; they should then be washed well with water before use.All beakers should be of silica, with silica or polythene covers.551 Oct., 19581 ALUMINIUM I N CAST IRON TABLE 111 DETERMINATION OF ACID-SOLUBLE AND ACID-INSOLUBLE ALUMINIUM I N SAMPLES OF IRON Acid-soluble aluminium Acid-insoluble aluminium Sample Procedure found, yo found, yo 0.010, 0*011, 0*010 0.0022, 0.0022 0.0022, 0.0022, 0.0020 0*00076, 0.00078 0.0075, 0*0080, 0.0082 0.0013, 0.0012 0.00092, 0*00096, 0.00095 0.00016, 0.00032, 0.00018, 0.00052, 0.00040 Pig iron , . .. .. A Grey iron . . * . * . B Iron containing 4 per cent. of Spectrographically pure iron C 0*00003, 0.00004, 0.00003 Not determined chromium .. .. c 0.00038, 0.00040, 0.00038 0.00045, 0.00072, 040058 For polarography, a cathode-ray polarograph was used to attain maximum sensitivity. The utility of the method will be limited by the instrument available. For procedure C, a 50-ml tall platinum crucible is recommended for the electrolysis, but satisfactory results can be obtained by using pure nickel or stainless-steel crucibles, provided that they do not give rise to high blank values. REAGENTS- Perchloric acid, 5 N-Dilute 500 ml of perchloric acid, sp.gr. 1.54, to 1 litre with distilled water (see “Water” below). For aluminium contents greater than about 0.01 per cent., AnalaR or equivalent grade perchloric acid should be satisfactory; for below 0.01 per cent. of aluminium, redistilled or “lead free for foodstuffs analysis” perchloric acid should be used.It is advisable to dilute the lead-free acid as soon as received and to store it in a polythene bottle. Hydrochloric acid, sp.gr. 1.18-For aluminium contents greater than about 0.01 per cent., AnalaR or equivalent grade is satisfactory. For lower aluminium contents, the “lead free for foodstuffs analysis” acid should be used, preferably stored in polythene. Nitric acid, sp.gr. 1-42.-Use the grades of reagent as described for hydrochloric acid, but do not store in polythene. Chloro form-AnalaR or equivalent grade is satisfactory. Acetic acid, glacial-AnalaR or equivalent grade is usually satisfactory. For very low aluminium contents, it may be necessary to redistil the acid. Sodium acetate solution, 2 M-Dissolve 272 g of the purest hydrated sodium acetate available in water and dilute to 1 litre.Sodium diethyldithiocarbamate solution, 20 per cent.-Dissolve 20 g of AnalaR or equivalent grade sodium diethyldithiocarbamate in water and dilute to 100 ml. This solution must be freshly prepared. Cupferron solution, 1 per cent.-Dissolve 1 g of the purest available cupferron in 100 ml of water and spin in a centrifuge at 10-cm radius and 2000 r.p.m. for 3 minutes. Use the supernatant liquid. This solution must be freshly prepared. isoButyl acetate-Analytical-reagent grade is usually satisfactory. If high blank values are obtained, the reagent can be purified by shaking it for about 10 minutes with half its own volume of 5 N hydrochloric acid. Solochrome violet RS solution, 0.05 per cent.-Dissolve 0.5 g of the pure dye-stuff in 1 litre of distilled water and store in a polythene bottle.Water-For the earlier stages of the method, distilled or de-ionised water is satisfactory, but the solutions used in the base electrolyte must be made up with distilled water. De- ionised water can give rise to spurious polarographic waves and false figures.= Standard aluminium solution A-Dissolve exactly 2.5 g of high-purity aluminium in hydrochloric acid, add 50 ml of perchloric acid and evaporate to the appearance of fumes. Cool, and dilute to 500 ml. This solution is stable indefinitely. 1 ml = 0.005 g of aluminium. Distilled water must be used for this solution. Standard aluminium solution B-Dilute 2.0 ml of solution A to 1 litre with redistilled water.This solution should be freshly prepared as required. 1 ml = 10 pg of aluminium.552 ROONEY: THE DETERMINATION OF TRACE AMOUNTS OF [Vol. 83 This solution should be freshly prepared as required. PREPARATION OF CALIBRATION GRAPHS- Graph for 40 to 200 pg of aluminizcm-In a series of 50-ml calibrated flasks place 0, 4, 8, 12, 16 and 20-ml portions of standard aluminium solution B, and add 1 drop of methyl red indicator solution. Make just alkaline by adding N sodium hydroxide, and then add 1.0 ml of 5 N perchloric acid. Add 5.0 ml of 2 M sodium acetate solution, 20 ml of 0.05 per cent. Solochrome violet RS solution and dilute to the mark. Immerse the flasks in a water bath at 55" to 70" C for 5 minutes, and then cool a.nd record polarograms between -0.50 and -0.85 volt against the internal pool anode; the aluminium peak occurs at about -0.72 volt.Plot a graph of microamperes against micrograms of aluminium per 50 ml, subtracting the blank value from all readings in order to make the line pass through the origin. Graph for 10 to 60% of aluminium-Proceed as for the preparation of the graph for 40 to 200 pg of aluminium, but use 0, 1, 2, 3, 4, 5 and 6-ml portions of standard aluminium solution B and 5.0 ml of Solochrome violet RS solution. Immediately after diluting to the mark, transfer the solutions to 50-ml stoppered polythene bottles for the period of heating. Graph for 1 to 15 pg of aluminium-Proceed as for the preparation of the graph for 10 to 60 pg of aluminium, but use 0,1,3,5,7,10,12 and 15-ml portions of standard aluminium solution C and 1.0 ml of Solochrome violet RS solution.These solutions will need to be de-oxygenated very thoroughly. PROCEDURE FOR DETERMINING ACID-SOLUBLE ALUMINIUM- A . Aluminium contents from 0.002 to 0.2 per cent.-Weigh 1.0 g of sample into a 150-ml squat beaker and dissolve without heating in 20 ml of diluted hydrochloric acid (1 + 1). When dissolved, cool, and dilute to the mark in a 100-ml calibrated flask. Transfer im- mediately to a polythene centrifuge tube and spin in a centrifuge at 10-cm radius and 2000 r.p.m. for 2 to 3 minutes. By pipette, place 10 ml of the supernatant liquid (= 0.10 g of sample) in a 150-ml conical separating funnel and add 5.0ml of acetic acid, 10ml of sodium acetate solution and 20 ml of sodium diethyldithiocarbamate solution.Shake the separating funnel for 10 to 20 seconds, add 30ml of chloroform, and shake vigorously for 30 seconds. Allow the two layers to separate, rinse the stopper and neck of the funnel with chloroform from a polythene wash-bottle in order to remove particles of precipitate adhering to them, and run off the chloroform layer. Add a further 10 ml of chloroform, shake again, and allow the layers to separate. Add a few drops of sodium diethyldithio- carbamate solution to test for completeness of precipitation; there should be a white pre- cipitate. If the precipitate is coloured, add 5 ml of sodium diethyldithiocarbamate solution, and shake to extract the precipitate in the chloroform layer. Again test for completeness of precipitation; repeat until a white precipitate is obtained when sodium diethyldithiocar- bamate solution is added.Shake, allow the two layers to separate, and then run off the chloroform layer. Extract the aqueous layer with 10-ml portions of chloroform, washing the neck and stopper of the funnel with chloroform after each extraction until the chloroform layer is perfectly colourless (25 ml of sodium diethyldithiocarbamate solution and 50 to 60 ml of chloroform should be sufficient). Discard the chloroform extracts and wash the stem of the funnel with chloroform, removing any stubborn particles with a "spill" of filter-paper. To the aqueous solution in the separating funnel, add 1.0 ml of cupferron solution and set aside for 1 to 2 minutes. Add 15 ml of chloroform, shake vigorously for 30 seconds, and allow the two layers to separate. Note whether the chloroform extract is perfectly colourless or coloured.Run the chloroform layer into a 50-ml beaker, and repeat the extraction with 10 and then 5ml of chloroform, adding these extracts to the 50-ml beaker. Evaporate the chloroform extracts to dryness and add 1.0 ml of nitric acid and 2.0 ml of 5 N perchloric acid. Keep the beaker covered with a well fitting lid in order to minimise loss of perchloric acid, and evaporate to fumes of perchloric acid; continue heating until all organic matter has been destroyed. If the chloroform extract is green owing to the co- extraction of a small amount of chromium from high-chromium irons, the residue in the beaker will probably be orange at this stage, owing to the oxidation of chromium.If chromium is present, add a further 2.0 ml of 5 N perchloric acid, evaporate to the appearance of fumes, and remove the lid. Add 0.5 to 1.0ml of hydrochloric acid to volatilise the Standard aluminium solution C-Dilute 20 ml of solution B to 200 ml with water. 1 ml = 1 pg of aluminium.Oct., 19581 ALUMINIUM I N CAST IRON 553 chromium as chromyl chloride, replace the lid, and again evaporate to the appearance of fumes. If sufficient chromium is left to colour the residue, add a further 0.5 ml of hydro- chloric acid. Continue this procedure, adding more perchloric acid if necessary to maintain the volume, until no orange colour is obtained on further heating to fumes. One addition of hydrochloric acid is usually sufficient. When all organic matter and chromium have been removed, remove the lid and evaporate to dryness.If the chloroform extract is colourless or pale green, proceed as follows. Dissolve the residue in 1.0 ml of 5 N perchloric acid and transfer the solution to a 50-ml calibrated flask. Add 5.0 ml of 2 M sodium acetate and 1.0, 5-0 or 20.0 ml of Solo- chrome violet RS solution, according to the amount of aluminium present. Dilute to the mark with distilled water, and, if 1.0 or 5.0ml of dye solution were used, transfer the solution to a screw-top polythene bottle. Complete the determination as for the preparation of the calibration graphs and read the aluminium contents corresponding to the step or peak heights from the appropriate curve. If the chloroform extract is yellow or brown, owing to the presence of titanium or iron, proceed as follows. Dissolve the residue in 1.0 ml of 5 N perchloric acid and transfer the solution to a 150-ml conical separating funnel.Adjust the volume to 50 to 60 ml and add 1.0 ml of cupferron solution. Add 10ml of chloroform, shake for 30 seconds, and then allow the two layers to separate. Run off the chloroform layer, and wash the aqueous layer with a further 10 ml of chloroform. Continue to shake with 10-ml portions of chloroform until the extract is colourless; two 10-ml portions are usually sufficient. Add 10 ml of 2 M sodium acetate solution, shake, and add 1.0ml of cupferron solution. Extract the aluminium as before, and complete the determination as described. A blank determination must be carried out by treating the reagents alone in the manner described for the sample. If a batch of samples is processed, one blank determination is sufficient, but all samples must then be processed similarly, even though some may contain no chromium or titanium.Unless these samples are processed identically with any that do contain chromium or titanium, and with the blank determination, the correction applied for the blank value will no longer be accurate. B. Aluminium contents from 0.0004 to 0.004 per cent. in plain cast irons only-Weigh 5 g of sample into a 400-ml beaker and dissolve it in 40 ml of hydrochloric acid and 10 ml of nitric acid without heating. When dissolved, cool, and transfer the solution to a centrifuge tube, washing with concentrated hydrochloric acid from a polythene wash-bottle.Spin in a centrifuge at 10-cm radius and 2000 r.p.m. for 3 minutes, and then transfer the super- natant liquid to a 250-ml conical separating funnel. Add 150 ml of isobutyl acetate, and shake for 30 seconds. Allow the two layers to separate and run the lower (acid) layer into a 150-ml beaker. Evaporate to dryness, add 5.0 ml of nitric acid and 4.0 ml of 5 N perchloric acid, and then evaporate to fumes of perchloric acid. When all organic matter has been destroyed, cool, and transfer the solution to a 150-ml separating funnel. Add 15 ml of 2 M sodium acetate and 20 ml of 20 per cent. sodium diethyldithiocarbamate solution. Extract the interfering elements and then the aluminium, and complete the determination as for procedure A .C. Aluminium contents less than 0.0004 per cent. in plain cast irons or less than 0.005 per cent. in alloy cast iron-Solid samples must be used for this determination and they should be pencil shaped. Weigh the sample and calculate the approximate length that will corre- spond to the desired weight of sample. Set up a cell as described by Chirnside, Cluley and Proffit,21 with a 50-ml platinum crucible containing a 20-ml pool of mercury. The crucible is connected to the cathode lead, and 20 ml of water and 2.0 ml of 5 N perchloric acid are added. Connect the sample to the anode lead by means of a crocodile clip, and electrolyse at 3 to 5 amperes. Lower the sample into the solution as it dissolves, and maintain the volume of electrolyte at about 20ml by the addition of water.It is advantageous to cool the crucible in running water if possible and to stir the mercury pool with a bone spatula from time to time. When the calculated length of sample has dissolved, remove, wash and dry the sample, and re-weigh it in order to obtain the weight dissolved. With a flat spiral of platinum wire as anode, continue to electrolyse until the electrolyte gives no test for iron. Disconnect the leads and decant the solution from the mercury into a polythene centrifuge tube, washing with water. If this operation is carried out quickly, the amount of iron that will re-dissolve is very small. Spin in a centrifuge at 10-cm radius and 2000 r.p.m. for 2554 ROONEY [Vol. 83 minutes in order to remove any insoluble residue of silica, graphite, etc., and transfer the supernatant liquid to a 150-ml squat beaker.Add 2 or 3 drops of 100-volume hydrogen peroxide, and boil for a few minutes to destroy excess of peroxide. Cool, and transfer to a 150-ml conical separating funnel. Add 1.0 ml of 1 per cent. cupferron solution, shake, and extract the precipitate in 10 ml of chloroform. Add a further 1.0 ml of cupferron solution, and, if a coloured precipitate forms, shake once more to extract it in the chloroform layer. Continue until the precipitate obtained is white; this should not require more than 3 or 4 ml. Shake for 30 seconds, allow the two layers to separate, and run off the chloroform layer. Extract the aqueous layer with further 10-ml portions of chloroform until the chloroform layer is colourless; add 20 ml of 2 M sodium acetate and then 1 ml of 1 per cent.cupferron solution. Extract the aluminium and complete the determination as for procedure A . PROCEDURE FOR DETERMINING ACID-INSOLUBLE ALUMINIUM- Weigh a 10-g sample into a 400-ml beaker and dissolve it in 100 ml of diluted hydrochloric acid (1 + 1). Boil, and filter through a No. 541 filter-paper of the smallest convenient size, washing well with hot dilute hydrochloric acid (1 + 19). Dry and ignite in a platinum crucible at 700" to 800" C until all carbonaceous matter has been destroyed. Treat the residue with 2 or 3 ml of hydrofuoric acid and 10 drops of dilute sulphuric acid (1 + 4), and then heat on a radiation bath until all the silica has been volatilised and white fumes of sulphur trioxide appear.Add a further 10 drops of dilute sulphuric acid (1 + 4) and again evaporate to dryness. Add 5 ml of hydrochloric acid, 1 or 2 drops of nitric acid and evaporate just to dryness. Dissolve the residue in 1.0 ml of hydrochloric acid, and transfer the solution to a 150-ml separating funnel. Add 5.0 ml of acetic acid and 10 ml of 2 M sodium acetate, shake, and then add 10 ml of sodium diethyldithiocarbamate solution. Complete the determination as for pro- cedure A (see Note). A blank determination must be carried out on the reagents, including the filter-paper and the hydrochloric acid used to dissolve the sample. NOTE-If the acid-insoluble aluminium content is in excess of 0.002 per cent., 1.0 ml of cupferron solution may be insufficient for precipitation.This will be shown by the formation of a coagulated precipitate of aluminium cupferrate rather than a milky suspension; the most satisfactory procedure then is to extract the aluminium in the chloroform layer and add a further 1.0 ml of cupferron solution, If a precipitate forms, extract this in the chloroform and continue until no further precipitate forms. The determination should then be carried out on an aliquot of the final solution containing less than 200 pg of aluminium; the size of aliquot to be taken can be approximately determined from the relationship between cupferron and aluminium, i.e., 1 ml of 1 per cent. cupferron solution is approximately equivalent to 500 p g of aluminium. I thank the Director and Council of the British Cast Iron Research Association for Evaporate to dryness, but do not bake. Fuse the residue with 1.0 g of sodium carbonate and leach the melt with hot water. permission to publish this paper. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 11. 18. 19. 20. 21. 22. 23. REFERENCES Dawson, J. V., and Smith, L. W. L., B.C.I.R.A. J. Res. and Dev., 1956, 6, 226. Dawson, J. V., Ibid., 1957, 7 , 2. Khoroshev, I. I., Dokl. Akad. Nauk SSSR, 1954, 94, 221. Palmer, S. W., Proc. Inst. Brit. Foundrym., 1946/1947, 40, 64~. Taub, A,, Foundry, 1955, 83, 131. Clarke, W. E., and Rooney, R. C., B.C.I.R.A. J . Res. and Dev., 1957, 6, 666. Wacykiewicz, K.. Prace Inst. Minist. Hutn., 1955, 7, 35. Short, H. G., Analyst, 1950, 75, 420. Jean, M., Anal. Chim. Acta, 1954, 10, 526. Wiberley, S. E., and Bassett, L. G., Anal. Cham., 1949, 21, 609. Kassner, J. L., and Ozier, M. A., Ibid., 1951, 23, 1453. Rocquet, P., Rev. Mdtall., 1943, 40, 276. Rosotte, R., Chim. Anal., 1956, 38, 250. Weissler, A., and White, C. E., Ind. Eng. Chem., Anal. Ed., 1946, 18, 530. Willard, H. H., and Dean, J . A., Anal. Chem., 1950, 22, 1264. Bishop, J. R., Analyst, 1956, 81, 291. Perkins, M., and Reynolds, G. F., Anal. Chim. Acta, 1958, 18, 616. Bode, H., 2. anal. Chem., 1955, 144, 165. Rooney, R. C., Anal. Chim. Acta, in the press. Oelschlager, W., Z . anal. Chem., 1957, 154, 329. Chirnside, R. C., Cluley, H. J., and Proffit, P. M. C., Analyst, 1957, 82, 18. Werz, W., and Neuberger, A., Arch. Eisenhicttenw., 1955, 26, 205. Ferrett, D. J., Milner, G. W. C., and Smales, A. A., Analyst, 1954, 79, 731. Received December 13th, 1957