Instrumental methods of chemical analysis, including electron and ion-probes, X-ray fluorescence, atomic absorption, and plasma-source spectrometry, have largely replaced classical wet-chemical analysis except for ‘referee’ analyses, but have not so far proved able to determine the net state of oxidation of a rock or mineral with reasonable accuracy. Mössbauer spectroscopy can indeed give fairly accurate Fe²⁺ and Fe³⁺ estimates in favourable circumstances, but fails when the Fe²⁺: Fe³⁺ ratio is very small or very large or when the iron is present in several different lattice positions with different surroundings; and it cannot arrive at the net state of oxidation when other elements of variable valency are present. The term ‘net state of oxidation’ calls for some comment, and is dealt with in the full paper; for the moment we note as an example that Mössbauer spectroscopy shows that much of the iron in ludlockite is ferric, whereas a ‘ferrous iron’ determination corresponds to an arsenate of lead and ferrous iron, from which it follows that part of the arsenic must be trivalent, a result that could not have been arrived at from either the ‘ferrous iron’ or the Mössbauer data along. Thus in the present state of the art a wet-chemical method for net state of oxidation is a necessity.

In this paper the development of methods from the determination of the net state of oxidation (‘ferrous iron’) is shortly reviewed; each of the methods currently in use has its own field of usefulness and its own specific weaknesses, but all are ill-suited to the determination of small amounts of ferric iron in presence of large amounts of ferrous.

A new technique is described in detail that allows separate determinations of Fe²⁺ and Fe³⁺ in a single small sample (1 to 10 mg or less). It depends on extraction of the Fe³⁺ from a citrated solution of the mineral, buffered to pH c.4, with the Fe²⁺ complexed by o-phenanthroline and the Fe³⁺ by 8-hydroxyquinoline, by a chloroform solution of 8-hydroxyquinoline (‘oxine’); Fe³⁺ in the chloroform extract is determined colorimetrically, as is Fe²⁺ in the aqueous solution after replenishment of the o-phenanthroline (largely extracted by the chloroform).

Acid-soluble minerals are dissolved in 8N H₂SO₄ on the water-bath with complete exclusion of air (HCl has been found to be unsatisfactory; a ferrous salt dissolved in 6N HCl and heated 3 hours on the water-bath with exclusion of air was largely oxidized).

Silicates (including rocks) are dissolved in a 10:1 mixture of 8N H₂SO₄ and 40% HF in screw-capped PTFE vessels, and the special reagent (o-phenanthroline + 8-hydroxyquinoline) is satura-ted with H₃BO₃ to complex excess HF and protect the separator and colorimeter cells.

The iodine monochloride method for the deter-mination of the net state of oxidation in acidsoluble materials has also been adapted for use on silicates by use of M/25 ICI in 6N HCl together with 1/10th its volume of 40% HG, using a PTFE vessel and adding an equal volume of 6N HCl saturated with H₃BO₃ before extraction with CCl₄. This technique has two advantages: air need not be excluded during solution, and the acid layer after extraction of the iodine by CCl₄ can be used for other determinations; but it has two dis-advantages: iodine in CCl₄ has a relatively weak absorption, about 1/8th that of Fe³⁺ o-hydroxyquinolate and 1/20th that of Fe²⁺ o-phenanthroline; and sulphides, including pyrite, are readily oxidized by the ICl as are any organic contaminants. For rocks, the first of these disadvantages is unimportant, since a fairly large sample is normally necessary to ensure proper sampling; the latter is, however, serious, though in some rocks the sulphides could probably be attacked by pure dry chlorine without oxidation of the silicates.

Finally, a procedure is described for the determination of titanium and total iron, or of manganese, titanium, and total iron after a determination of Fe²⁺ by the ICl technique.

The new techniques, here shortly described, are dealt with more fully in the Miniprint section.

Procedure for acid-soluble minerals. Reagents: H₂SO₄, c.8N, de-aerated and stored under N₂ or CO₂.

NaOH, adjusted to a normality 0.9 that of the H₂SO₄.

12% o-phenanthroline in water.

20% sodium acetate in water.

1% 8-hydroxyquinoline (‘oxine’) in CHCl₃.

Dual reagent: 1 g oxine, 12 g o-phenanthroline, 10 g sodium citrate, in 12 ml glacial acetic acid and 88 ml H₂O.

The sample should if possible contain ≮ 10 µg of FeO or Fe₂O₃, whichever there is least of, and not more than 1 mg F₂O₃. It is weighed into a glass-stoppered tube of 4 ml capacity, 3.0 ml H₂SO₄ added, the air displaced by N₂ or CO₂, and the tube heated on the water-bath until solution is complete. The solution is poured into a 25 ml separator containing 3.0 ml NaOH, 2 ml 20% sodium acetate, and 2 ml dual reagent, rinsing the tube with the minimum of water, and the ferric oxine complex extracted with 5 ml (or more) 1% oxine in CHCl₃, shaking vigorously for 30 sec; if the CHCl₃ layer appears black after a few seconds shaking, add a further 5 or 10 ml 1% oxine. Separate, add 1 ml 12 % o-phenanthroline to the aqueous layer and extract with a smaller volume of 1% oxine. For < 100 µg Fe₂O₃ extractions with 5, 2, 1, 1 ml oxine will be satisfactory; for more Fe₂O₃ increase the volumes; the fourth extract should be completely colourless (if the second appears colourless, make only three). Add extra o-phenanthroline after each extraction, including the last.

Bulk the aqueous and CHCl₃ layers appropriately and colorimeter against standards and a blank, the aqueous layer at λ 508 nm for FeO, the CHCl₃ layer at 579 nm for Fe₂O₃ (if Al and Cu are absent, the CHCl₃ layer may be colorimetered at 467 nm with a gain in sensitivity of about 1.5 times). The pH of the aqueous layer should be about 4.0 to 4.5.

Procedure for rocks and minerals insoluble in 8N H₂SO₄. The 4 ml glass tubes are replaced by 7 ml screw-capped polytetrafluorethylene (PTFE) capsules, and 5.0 ml 8N H₂SO₄ and 12 ml 40% HF used for solution. The NaOH is increased to 5.0 ml, and the dual reagent is modified, being saturated with H₃BO₃ a to complex the HF; a saturated solution of H₃BO₃ is used to rinse out the PTFE capsule. Otherwise the procedure is as for acid-soluble minerals.

Modified iodine monochloride method. For rocks and minerals insoluble in 1 : 1 HCl, the procedure of Hey (1974) for acid-soluble minerals is followed, except that: the solvent is a mixture of 9 parts M/25 ICl in 1:1 HCl and 1 part of 40% HF, duly stabilized; solution takes place in a PTFE capsule, which must be stoppered, but displacement of air is unnecessary; and the solution is transferred to the separator with 5 ml 1:1 HCl saturated with H₃BO₃. (Note: The M/12 ICl solution recom-mended in Hey (1974) is unnecessarily strong; 4 ml M/25 ICl are enough to oxidize 5 mg FeO).

Sulphosalicylic acid as a reagent for Fe and Ti. The deep purple colour given by sulphosalicylic acid (phenol-2-carboxylic-4-sulphonic acid) with Fe³⁺ in weakly acid solution is not suitable for determination of Fe³⁺, but in alkaline solution iron (Fe³⁺ or Fe²⁺) reacts to give a yellow colour, peak at λ 425 nm; interference by Mn or Co (which undergo air-oxidation) is suppressed by ascorbic acid, added after the solution is made alkaline, and the only other slightly interfering metal is V, which gives an absorbance about 1/70 that of Fe at λ 425 nm; for the small amounts of V present in rocks, a correction is hardly necessary. Interference by large amounts of acetic acid or citric acid is suppressed by extra sulphosalicylic acid.

In weakly acidic solution, pH 3.6–4.3, sulpho-salicylic acid gives an intense absorption, peak λ 395 nm, with Ti; the purple Fe³⁺ colour can be bleached by ascorbic acid. It is thus practicable to determine total Fe in the aqueous layer after a Fe²⁺ determination by the ICl method, and then Ti in the same solution. If the solution is then evaporated with nitric acid and either sulphuric or perchloric acid, the picric acid formed from the sulphosalicylic acid does not interfere with deter-mination of Mn as MnO4−.