260 Analyst, April, 1971, Vol. 96, $9. 260-273 Determination of Eleven Metals in Small Samples of Blood by Sequential Solvent Extraction and Atomic-absorption Spectrophotometry BY H. T. DELVES, G. SHEPHERD AND P. VINTER (Department of Chemical Pathology, The Hospital for Sick Children, Great Ormond Street, and the Institute of Child Health, University of London, 30 Guilford Street, London W.C.1) A method is described for the determination of eleven metals in a 1-ml solution of an oxidised blood sample. The metals iron, copper, bismuth, zinc, cadmium, lead, cobalt, nickel, manganese, strontium and lithium are selectively extracted into small (0-30 to 0.50 ml) volumes of isobutyl methyl ketone as their chelates or ion-association complexes, and are determined in the organic phases by atomic-absorption spectrophotometry.The enhance- ment effect of the organic solvent combined with the extraction and concen- tration of the metals results in average sensitivity increases of seven times that obtained by a direct determination on the aqueous solutions. The recovery of the metals added to blood is quantitative and, with two exceptions (lead and bismuth), a precision of better than 8 per cent. can be achieved at the 0.1 p.p.m. level. The results are given of an application of the method to a study of the problem of metal ingestion by children who have pica. TRACE-ELEMENT survey analysis of biological materials has shown that in addition to those metals known to be essential to man a further 20 metals are consistently present in human tissues.lS2 Many of these “non-essential” metals do not have any known or suspected bio- chemical function and are thought to be environmental contaminant^.^ The contamination of tissues with metals is a greater problem with children than with adults because normally developing children mouth and chew unfamiliar objects as a way of examining them.For some children the desire to chew such objects is uncontrollable, and these children are said to have pica. Such children have a higher incidence of lead poisoning,4 because of an increased oral ingestion of lead-containing materials, than children who do not have pica. In view of the wide range of metals used in the manufacture of materials, such as paints, plastics, rubbers and paper, likely to be chewed by children, it is possible that metals other than lead are excessively ingested by children.Multi-metal determination in blood samples taken from one group of children with pica and another group without pica would be of value in estab- lishing whether the excessive ingestion of metals, other than lead, is a problem in young children. The analytical methods that have been used successfully for multi-metal determination in biological materials are d.c. arc emission spectroscopy,l spark-source mass spectrometry5 and neutron-activation analysis.6 These techniques are, however, expen~ive,~ 9 6 difficult to apply to the routine analysis of a large number of ~arnples~.~ and require pre-concentration from large sample sizes1 Although simultaneous multi-element determination by atomic- absorption spectrophotometry is not yet feasible with commercially available instruments, the technique has good sensitivity for many elements and is easily adapted to routine analysis, and the basic equipment for single-element determinations is relatively cheap.This paper describes a method for the determination of eleven metals in a 1-ml sample solution of oxidised blood by sequential solvent extraction and atomic-absorption spectrophotometry. The metals determined were iron, copper, bismuth, zinc, cadmium, lead, cobalt, nickel, manganese, strontium and lithium. When investigating biochemical lesions in young children, it is wise to take the minimum amount of blood possible for the determinations required. Any method developed must therefore be capable of providing multi-metal determinations with as little as 1 to 2-ml samples for each of two duplicate determinations. 0 SAC and the authors.DELVES, SHEPHERD AND VINTER 261 With the exception of iron, zinc and copper, the normal physiological concentrations of the above metals are so low (0.005 to 0.05 p.p.m.) that a solvent-extraction and concentration stage is essential to obtain atomic-absorption signals that are accurately distinguishable from the background signals.A single extraction and 2-fold concentration from a 1-ml sample solution of oxidised blood would yield only 0.5 ml of solvent for the atomic-absorption deter- minations. Even by reducing the sample uptake rate of the nebuliser to one half of that recommended by the manufacturer (a Perkin-Elmer 303 instrument was used), this volume would be sufficient for only two determinations (Fig.1). The determination in duplicate of the above eleven metals by using a single “universal” extractant, if available, would therefore require more than 9ml of blood. However, a sequential separation of the metals would enable them to be determined by using 1 ml of sample solution and could therefore provide the basis of a method for multi-metal determination on the small samples of blood that can be taken from children. EXPERIMENTAL DEVELOPMENT OF A SEQUENTIAL EXTRACTION SCHEME- The solvents that can be used for atomic-absorption spectrophotometry with air-supported flames are limited to those which support the combustion processes of the flames. Fortunately, isobutyl methyl ketone, which has been shown by Allan’ to be an excellent solvent for flame atomic-absorption determinations, has also been proved useful for the extraction of metal chelates.8~~ This solvent was therefore used for each extraction stage in the separation scheme described. The extraction and concentration of metals from 1 ml of an aqueous sample solution into 0.5 ml of solvent yielded sufficient solution for one single atomic-absorption determination by using the sample uptake rate of the nebuliser recommended by the manufacturer, vDix., 5.6 ml minute-l of isobutyl methyl ketone. When this was reduced to 2.8 ml minute-l, the reduction in sensitivity was 14 per cent.relative, but determinations could be made with as little as 0.20ml of solvent. It was then possible to make determinations of two different elements with 0.5 ml of extract with little loss in sensitivity or precision (Fig.1). An extrac- tion scheme was therefore devised to separate the metals to be determined into groups of h 50- hl - 60- S 0 .- v) .- $, 70- 2 c-’ Q, m 80- Q) ? Q. A f l 9 0 1 #. B C D E F Fig. 1. Absorption signals from a continuous 20-s aspiration of a 0.2 pg ml-1 cadmium solution in isobutyl methyl ketone (A), and discrete aspira- tions (B to F) from 1.0 ml of the same solution until the entire sample was consumed; 1.0 ml of the solution was sufficient for four determinations (B to E) but not for the fifth deter- mination (F)262 DELVES, SHEPHERD AND VINTER : DETERMINATION OF [Analyst, Vol. 96 100 80 P E .w 2 60 al Ol (u 2 40 Y B 20 1 .o 2 .o 3 .O 0 Hydrochloric acid, N 0.2 0.4 0.6 0.8 1.0 DDDC in isobutyl methyl ketone, 1 per cent.w/v Hydrochloric acid, N PH PH :Fig. 2. The extraction of metals from aqueous solutions of oxidised blood samples: (a), cupferron; (b), diethylammonium diethyldithiocarbamate from 2 N hydrochloric acid ; (c), triisooctylamine ; (4, ammonium pyrrolidinedithiocarbamate ; (e) , 8-hydroxyquinoline ; and (f) , 2-thenoyl-3,3,3-tri- fluoroacetoneApril, 19711 ELEVEN METALS I N SMALL SAMPLES OF BLOOD Aqueous phase Organic phase Add 0.50 ml of 0.175 per cent. w/v DDDC in isobutyl methyl ketone and extract for 1 minute. Remove supernatant liquid 263 B Copper, loo% bismuth, loo% ; _3 (lead, 3y0) To 1.0 ml of sample solution (in 2.2 N HC1*) in a 7-ml polypropylene centrifuge tube, add 0.05 ml of freshly prepared 10 per cent. w/v aqueous cupferron solution.Mix and extract for 15 s with 0.50 ml of isobutyl methyl ketone. Remove and retain super- natant liquid, repeat twice with 0.5 ml of isobutyl methyl ketone for each extraction. Combine the three organic extracts I Extract with 0.50ml of 15 per cent. v/v T I 0 in isobutyl methyl ketone. Centrifuge and remove supernatant liquid I I A Iron, 100%; tin, 94y0;t --+ molybdenum, 94% ;t thallium, 100% ;t (copper. 2%) Add 0.3 ml of the 6 N ammonia solution - 5 per cent. w/v ammonium citrate reagent, 0.05 ml of bromophenol blue indicator solution and sufficient N ammonia solution to produce a blue colour without a red background (pH 2.8 to 3.2). Add 0.1 ml of 1 per cent. w/v of aqueous APDC solution, mix, add 0.50ml of isobutyl methyl ketone and extract for 1 minute.Remove supernatant liquid I D -+ + I I Add 0.1 ml of 6 N ammonia solution to bring pH to between 9.0 and 9.5. Extract for 1 minute with 0.5ml of 0.2 M HTTA in isobutyl methyl ketone. Centrifuge if necessary and remove super- natant liquid F __f C + Cadmium, 100%; zinc, 100%; lead, 84% Cobalt, 100% ; nickel, 100% Add 0.2 ml of 6 N ammonia solution to bring pH to between 8.5 and 9.0. Mix and extract for 1 minute with 0.30 ml of 0.1 M 8-hydroxyquinoline in isobutyl methyl ketone. Remove supernatant I liquid -1: Manganese, 100% Strontium, 100% ; lithium, 92% ; calcium, 100% ; t magnesium, 97 % t * The additions of aqueous cupferron solution reduce the concentration to 2 N hydrochloric acid.t Metals that are extracted but not determined in the final method. The precise adjustment of the pH of the aqueous phase for the APDC extraction (pH 2.8 to 3.2) enables the required pH values for the 8-hydroxyquinoline and HTTA extractions to be attained by adding the volumes of 6 N ammonia solution indicated. No further adjustment is necessary. HTTA undergoes hydrolysis into trifluoroacetate and acetylthiophene a t pH values greater than 8.0. The organic extract with this reagent must be removed as quickly as possible after extraction to avoid hydrolysis, which results in precipitation in both phases. This precipitate does not contain strontium or lithium but can cause trouble by blocking the nebuliser capillary if allowed to form or to remain in the organic phase.All the solutions of the organic reagents should be prepared immediately before use. Fig. 3. Procedure for the sequential separation of metals264 DELVES, SHEPHERD AND VINTER: DETERMINATION OF [Analyst, Vol. 96 two. The number of metals extracted in any one group could be increased to include iron or zinc, or both, as their high concentrations in blood (500 and 10 p.p.m., respectively) would enable determinations of them to be made with a dilution of a small, e.g., 50-p1, portion of the extract. Seventeen elements that are either essential or toxic to man, or are general environ- mental contaminants, were investigated and a scheme was developed for the quantitative extraction and separation of 16 of them. Five metals were not included in the final method: molybdenum, tin and thallium, because of their low sensitivity by atomic absorption; and calcium and magnesium, because they were not of interest in the present pica studies.The extraction efficiencies of these metals may be of interest to workers in the field of trace metals and biological systems and are shown in Figs. 2 and 3. The extraction systems investigated are shown in Fig. 2 (a to f). In each instance, the aqueous phase was prepared from a hydrochloric acid solution of the inorganic residues from a blood sample that had been wet oxidised with nitric, perchloric and sulphuric acids and evaporated to dryness. The volume of hydrochloric acid solution used to dissolve the inorganic residues was one-fifth of the original volume of blood. This five times concentrated solution was used to prepare the aqueous phases for the extraction studies, so that each aqueous phase contained all of the inorganic constituents of blood at their normal physio- logical concentrations.This took into account any effects that these constituents, e.g., phosphate ions, may have had on the extraction equilibria. Ammonium citrate solution (to give a final concentration of 1 per cent. w/v of ammonium citrate) was added to those aqueous phases the pH values of which were adjusted to greater than 1.0 for the extraction studies. This gave some buffering action and prevented the precipitation of metal hydroxides at high pH values. A 2: 1 ratio of the volumes of the aqueous to organic phases was used to ensure that a concentration stage was obtained.A single 60-s extraction was used for all systems except that with cupferron (Fig. 3 A) for which three successive 15-s extractions were necessary to remove completely the high concentration of iron present in blood. The metals to be investigated were added to the aqueous phases a t concentrations that were sufficiently high (2 to 10 pg ml-l) to permit their accurate determination, before and after extract ion, by at omic-absorpt ion spect ropho t ometry . The experimental conditions giving the optimum separation of the metals are indicated by the dotted lines intercepting the axis of abscissae in Fig. 2 (a tof). The procedure for the sequential separation of the metals prior to their determination is given in Fig. 3. Iron, which is present at high concentrations in blood (450 pg rnl-l), was separated in the first extraction stage (Fig.2 a) by extracting with cupferron and isobutyl methyl ketone- This extraction stage was used to remove iron from the aqueous solution of oxidised blood prior to investigating the other extraction systems (Fig. 2, b to f). This eliminated any interference from the high concentrations of iron in these solutions that could react with,. and thus remove, most of the added reagent or form chelates that are insoluble in isobutyl methyl ketone, such as the iron(II1) - ammonium pyrrolidinedithiocarbamate (APDC) chelate and iron(II1) 8-hydroxyquinolinate. Group A-Three successive extractions with isobutyl methyl ketone from an aqueous phase that was 2 N in hydrochloric acid and contained 0.05 per cent.w/v of cupferron gave a quantitative extraction of iron(II1) , molybdenum(VI), thallium(II1) and tin(I1) , with only 2 per cent. extraction of copper(I1). Group B-Quantitative extraction of copper and bismuth was easily achieved during preliminary tests with diethylammonium diethyldithiocarbamate (DDDC) in isobutyl methyl ketone, but it was not possible to separate bismuth from lead, or bismuth and lead from the other metals. This extraction system was therefore investigated by using a 6 x 4 factorial arrangement of six different reagent concentrations in isobutyl methyl ketone, and four different hydrochloric acid concentrations (0.5, 1.0, 2.0 and 4-0 N) in the aqueous phases. The extraction from 2 N hydrochloric acid solution into a 0.175 per cent.w/v solution of the reagent in isobutyl methyl ketone gave the best separation of copper(I1) and bismuth(III), both of which were completely extracted, from lead(II), 3 per cent. of which was extracted. Cadmium(I1) and all of the other metals studied were not extracted under these conditions. Group C-Triisooctylamine (TIO) fonns a quaternary salt in hydrochloric acid solutions, the chloride ion of which can undergo anion exchange with chloride ion-association complexesApril, 19711 ELEVEN METALS I N SMALL SAMPLES OF BLOOD 265 of metals to yield extractable species. Zinc(II), cadmium(I1) and lead(I1) were separated with this reagent from those metals which survived the first two extraction stages. Grou@ D-Ammonium pyrrolidinedithiocarbamate has poor selectivity but was used to separate quantitatively cobalt(I1) and nickel(I1) from manganese(I1) and the other metals remaining after the extractions for groups A, B and C.Groz@ E-Manganese( 11) was quantitatively extracted and separated from the alkali metals and the alkaline earths at pH 8-5 to 9.0 with 0.1 M 8-hydroxyquinoline in isobutyl methyl ketone. Grou@ F-Lithium(I), strontium(II), calcium(I1) and magnesium(I1) were all extracted with 0.2 M thenoyltrifluoroacetone (HTTA) in isobutyl methyl ketone (IBMK) at pH 9 to 9.5. The quantitative extraction of lithium( I) was unexpected but has been reported by HealylO Metal Fe c u Bi Cd Pb Zn co Ni Mn Sr Li 20 40 60 0 0 bservat ion heig ht/mm I I 3.0 4.0 CZH2 flow/l minute-' O 1 /PO 210 Fig. 4. Effect of (a) observation height and (b) acetylene flow-rate on flame absorbance for organic solutions of metals TABLE I EXPERIMENTAL CONDITIONS FOR ATOMIC ABSORPTION A/nm 302.0 324.7 223.1 228.8 283-3 213-9 240.7 232.0 280* 460.7 670.8 Slit Band Range width/mm pass/nm expansion 0.3 1.0 0.3 1.0 1.0 1.0 0.3 0.3 1.0 1.0 1.0 0.2 0.7 0.2 0.7 0.7 0.7 0.2 0.2 0.7 1.3 1.3 1 1 3 3 3 1 3 3 3 10 10 Observation height/mm 10.0 10.0 7-5 10.0 10.0 12.0 7.5 10.0 10.0 10.0 15.0 Acetylene flow/l minute-l 2-9 2.9 2.6 2-6 2-9 2.6 2.7 2.7 2-9 2.1 2.1 *Unresolved triplet 279.5, 279.8 and 280.1 nm.Sample uptake rate 2-8 ml minute-l of isobutyl methyl ketone. Air flow 22.8 1 minute-1, 30 p.s.i. ; acetylene pressure, 5 p.s.i. Noise suppression 2, i.e., 2-s time constant in amplifier output circuit.266 DELVES, SHEPHERD AND VINTER: DETERMINATION OF [Analyst, Vol. 96 to be the result of adduct formation, with isobutyl methyl ketone acting as a neutral donor ligand according to the reaction OPTIMUM FLAME CONDITIONS FOR THE ATOMIC-ABSORPTION DETERMINATION OF METALS IN The optimum observation height, i.e., the vertical distance between the optical axis of the monochromator and the top of the burner head, and the optimum fuel-to-oxidant flow ratio were determined for the eleven metals under investigation.An air - acetylene flame and a triple-slot (Boling) burner were used for the investigations. The acetylene and air flow meters of the Perkin-Elmer 303 instrument were calibrated in free litres per minute at 15 "C in accordance with the recommendations of Mansfield and Wine- fordnerll and Kirkbright and Sargent,12 who have criticised the publication of arbitrary and meaningless flow-rates of gases in atomic-absorption methods.The optimum flame conditions established from these observations and other instrumental data for the atomic-absorption determination of the metals are given in Table I. Strontium and lithium were the only metals that showed any marked dependence of free-atom concentration in the flame on observation height and acetylene-to-air ratio. Both metals form hydroxide and oxide species in air-supported flames and the reduction of these species depends on the atomic hydrogen concentration in the flame. The maximum absorption signals were obtained with observation heights just above the primary reaction zone, where the atomic hydrogen concentration is known to be greatest.l3 The response of the other metals (Fig.4, a and b) indicated varying degrees of oxidation in the higher regions of the flame and increased reduction to free metal atoms with fuel-rich flames. Li&) + HTTA,o,g) + BIBMK,, + LiTTA(IBMK),,,) + H&) ISOBUTYL METHYL KETONE SOLUTIONS- The results are given in Fig. 4 (a and b). METHOD APPARATUS- A Perkin-Elmer 303 atomic-absorption spectrophotometer fitted with a 10-cm long triple-slot air - acetylene burner head (Boling) was used. Hollow-cathode lamps were used as radiation sources for each of the metals determined. The absorption signals were recorded on a 10-mV chart recorder, Rikadenki B-24X. The instrumental settings for the atomic- absorption determinations are given in Table I.Silica conical $asks-25-ml capacity. Stoppered glass t.ubes-5-ml capacity, graduated in O-l-ml divisions. Glass tubes-10-ml capacity, 125 x 15 mm i.d. Polypropylene tubes-7-ml capacity, 100 x 100 mm id., and 3-ml capacity, 75 x 7 mm Pasteur pipettes. Glass syringes-2-ml capacity. Rotamixer-Made by Hook and Tucker. Sand-bath-Thermostatically controlled up to 300 "C. The concentrated acids and ammonia solution used were Aristar grade reagents. All Nitric acid, concentrated, sp.gr. 1.42. Perchloric acid, sp.gr. 1.54-This substance should be handled cautiously, and the wet Sulphuric acid, concentrated, sp.gr. 1-84. Dilute sulphuric acid, (1 + 4 v / v ) . Hydrochloric acid, concentrated, sp.gr. 1-18. Dilute hydrochloric acid solutions-Prepare 10 and 4.5 N hydrochloric acid solutions by diluting the concentrated acid with de-ionised water, and standardise by titration.Prepare a 1 + 1Ov/v dilute hydrochloric acid solution. Ammonia solution, concentrated, sp.gr. 0.88. Dilute ammonia solutions, 6 N and 1 N (approximately)-Prepare by diluting 42 and 7 ml, i.d. REAGENTS- the other reagents used were of analytical-reagent grade unless otherwise specified. ashing should be carried out in suitably designed fume cupboards. respectively, of the concentrated reagent solution to 100 ml with de-ionised water.April, 19711 ELEVEN METALS IN SMALL SAMPLES OF BLOOD 267 Phenol red indicator solution, 0.02 per cent. w/v. Bromophenol blue indicator solution, 0.04 per cent, w/v. Ammonia - ammonium citrate solution, 5 per cent.w/v ammonium citrate in 6 N ammonia solution-Dissolve 40 g of citric acid in about 200 ml of de-ionised water. Add 2 drops of phenol red indicator solution and sufficient concentrated ammonia solution (about 100 ml) to produce a red colour. Remove trace metals from this solution by extraction with a 0-1 per cent. w/v solution of dithizone in chloroform as previously described,14 add 5ml of con- centrated nitric acid solution and dilute to 500 ml with de-ionised water. Dilute 50 ml of this 10 per cent. w/v ammonium citrate solution to 100 ml with 42 ml of ammonia solution (sp.gr. 0.88) and de-ionised water. Isobutyl methyl ketone saturated with water. Cupferron, 10 per cent. w/v solution in de-ionised water-Dissolve 1.0 g of the reagent in 10 ml of water.Diethylammonium diethyldithiocarbamate, 0.176 per cent. w/v solution in is0 butyl methyl ketone-Dissolve 0-044 g of the reagent in 25 ml of isobutyl methyl ketone. Triisooctylamine, 15 per cent. v / v solution in isobutyl methyl ketofine-Dilute 4.5 ml of the reagent to 30 ml with isobutyl methyl ketone. Ammonium Pyrrolidinedithiocarbamate, 1.0 per cent. w/v aqueous solution-Dissolve 0.1 g of the reagent in 10 ml of de-ionised water. 8-Hydroxyquinoline, 0.1 M solution in isobutyl methyl ketone-Dissolve 0-36 g of the reagent in 25 ml of isobutyl methyl ketone. Standard solutions ofthe metals-Prepare from the pure metal, or a suitable salt, standard solutions of each metal, containing 5 mg ml-l for iron and 1 mg ml-l for the other metals, in 1 per cent.w/v hydrochloric acid. Mixed standard solution of metals-For each of the metals except iron, prepare dilute standard solutions containing 100 pg ml-l of metal in 1 per cent. v/v hydrochloric acid solution. Place the following volumes of the 100 pg ml-l standard solutions of the metals indicated into a 100-ml calibrated flask and dilute to volume with de-ionised water: lithium, 0.10 ml; cadmium, cobalt, nickel, manganese and strontium, 0.50 ml; lead and bismuth, 3.00 ml; copper, 5.00 ml; and zinc, 25-0 ml. Working standard solutions-To a series of six 10-ml calibrated flasks, add 0, 0.5, 1.0, 2.0, 3.0 and 4.0 ml of the mixed standard solution of metals. Add, in the same order, 0, 0.2, 0.4, 0.6, 0.8 and 1.0 ml of the 5 mg ml-l standard iron solution. Add 2.2 ml of 10 N hydro- chloric acid solution to each flask and dilute to volume with de-ionised water.These solutions are 2-2 N with respect to hydrochloric acid and contain: Solution I A \ Metal 1 2 3 4 5 6 0 0.5 1.0 2.0 3.0 4.0 Cd, Co, Ni, Mn and Sr 0 2.5 5.0 10.0 15.0 20.0 BiandPb pgper100ml 0 15 30 60 90 120 c u 0 25 50 100 150 200 Zn 0 125 250 500 750 1000 0 100 200 300 400 500 Fe, pg ml-l Li 1 PROCEDURE- With a pipette introduce 2.0 ml of haemolysed whole blood into a 25-ml silica conical flask. Rinse down the sides of the flask with water (about 2 ml) and add 5 ml of concen- trated nitric acid, 1.0 ml of perchloric acid (sp.gr. 1.54) and 0-1 ml of the dilute sulphuric acid (1 + 4) solution. Mix and allow to stand for 15 minutes. Place the flask on a sand-bath (or hot-plate) at a temperature of 150 "C.Watch care- fully for any signs of vigorous reaction and frothing and remove the flask if the frothing becomes excessive. When the initial reaction has subsided, leave the flask on the sand-bath until a clear yellow - brown solution is obtained. Increase the temperature to 200 "C and evaporate to fumes of perchloric acid, then increase the temperature to about 300 "C and evaporate to dryness. For convenience this can be done overnight. Remove the flask from the sand-bath. When cool add 0.5 ml of the 4-5 N hydrochloric acid solution and return it to the sand-bath at a temperature of 150 "C. Evaporate the solution just to dryness and remove the flask from the sand-bath. When cool, the residue may become yellow as a result of residual hydrochloric acid vapour in the flask, but no268 DELVES, SHEPHERD AND VINTER : DETERMINATION OF [Analyst, Vol.96 Fe, pg rnl-' 0 TOO 200 3@0 400 500 I I 1 I I i Cu, Zn, pg per 100 ml 100 - 80 - w 07 r .- 2 60- Y a a 40 - 20 - Mn, Cd, Sr, Li, Bi, Co, Pb, Ni, 1-19 per 100 ml Fig. 5. Calibration graphs (lithium concentration was one-fifth of that shown; zinc concentration was 5 times that shown; and lead and bismuth concentrations were 6 times those shown) liquid should remain. Add 1.00 ml of 4.5 N hydrochloric acid solution to the flask and place it on a sand-bath at a temperature of about 150 "C for about 2 minutes. Remove the flask from the sand-bath and allow it to cool. A clear yellow solution should be obtained. Transfer it to a 5-ml graduated tube, rinsing the flask three times with 0-2 to 0.3 ml of water, dilute to 2-00 ml with de-ionised water and mix.With a pipette place a 1.00-ml aliquot of the solution in a 7-ml polypropylene centrifuge test-tube for the sequential extraction and separation of the metals by the procedure given in Fig. 3. Mix the phases by holding the tubes against the rubber pad of a vibrator - mixer. When the phases have separated, transfer the upper organic phase into a 3-ml polypropylene tube with a syringe-controlled Pasteur pipette. After each extraction stage, determine the concentrations of the metals in the organic phases by using the experimental conditions given in Table I. For the determination of iron and zinc dilute 0.05ml of the appropriate organic phase with 2.00ml of isobutyl methyl ketone in a 7-ml polypropylene tube.Determine all other metals directly in the organic phases. Determine duplicate reagent blanks with each batch of samples by using the method as described, except that water is substituted for haemolysed whole blood. Establish calibration graphs by transferring with a pipette 2.00 ml of the mixed working standard solutions into 10-ml glass tubes and carry out the separation scheme as described but with twice the given volumes of the reagents. This provides enough solution for the atomic- absorption determinations before and during a run of samples. To obtain the best results with the method described it is necessary to exercise a little more care than would normally be required for atomic-absorption determinations with larger volumes of solution.The nebulisation of 0.2 ml from a 0.5-ml fraction is judged on a time basis (about 4 s) and some practice is needed to enable a new operator to become skilled at making (at least) four determinations with 1 ml of solvent. Losses of isobutyl methyl ketone by volatilisation from the extraction tubes are negligible during the time required for analysis (about 0.7 per cent. loss after 30 minutes).April, 19711 ELEVEN METALS IN SMALL SAMPLES OF BLOOD 269 RESULTS Calibration graphs and the results of recovery tests are given in Fig. 5 and Table 11. Recovery tests were carried out for both synthetic aqueous standards and for solutions of metals added to blood samples. With the former, the solution of the inorganic residues from the oxidation stage was divided into two portions.A l-ml portion was subjected to TABLE I1 RECOVERY OF METALS ADDED TO BLOOD ANALYSED BY THE DESCRIBED PROCEDURE Concentration Mean per cent. recovered of metal added, A --, Number r pg per 100 ml Bi Cd Co Ni Mn Sr 5 104 79 100 94 112 88 10 118 98 96 89 113 103 20 120 108 104 91 111 109 30 107 102 98 89 106 104 40 120 101 107 90 106 105 50 108 91 98 89 103 98 20 40 60 80 100 120 Mean per cent. recovered r Pb Cu Znt Fei 110 73 95 105 90 98 101 105 97 104 97 110 105 108 101 101 110 106 99 103 108 102 100 103 * Li added at one-fifth of concentration shown. t Zn added at 10 times concentration shown. 1 Fe added at 500 times concentration shown. Li* of tests 88 4 105 4 110 4 101 5 103 3 103 5 the separation scheme and the metals determined in the organic extracts (Table 111); the remaining 4-ml portion was analysed directly by atomic-absorption spectrophotometry for the same metals. The correlations between the concentrations found by both techniques are given in Table IV.TABLE I11 RECOVERY OF SYNTHETIC STANDARD SOLUTIONS OF METALS AFTER OXIDATION Results are expressed as regression equations of the form : concentration found = (concentration added) m + c. Ideally m should approach 1-000 and c should approach zero AND EXTRACTION Constants in regression equation (pg per 100 ml) Concentration 7 range studied, f A Metal* m G pg per 100 ml c u 0.9 19 - 2.2 10 to 400 Bi 1-075 - 4.399 2-5 to 100 Cd 1.009 + 0.806 2.5 to 100 Zn 1.001 -0.218 25 to 1000 Pb 0.945 + 2.086 2.5 to 100 c o 0.992 +0-115 2.5 to 100 Ni 0,889 +3*17 2.5 to 100 Sr 0.996 -0.138 2.5 to 100 Li 1.037 +0-515 0.5 to 20 * Only one result was obtained for manganese because of mechanical loss Iron was not of sample: added 100 p g per 100 ml; found 102 pg per 100 ml. included in these studies.Results for precision are given in Table V. The detection limits given in Table VI are twice the standard deviation of the results of replicate blank determinations. Iron and zinc are not included in this table because they are determined on dilutions of the organic extracts.270 DELVES, SHEPHERD AND VINTER: DETERMINATION OF [Amalyst, Vol. 96 TABLE IV CORRELATION COEFFICIENTS FOR CONCENTRATIONS FOUND AFTER EXTRACTION AND Metal . . . . . . Cu Bi Cd Zn Pb Co Ni Sr Li Correlation coefficient.. 0.998 * 0.998 0.992 * 0.994 0.990 0.994 0.993 BY DIRECT ANALYSIS OF AQUEOUS SOLUTIONS * No results for direct determinations because of insensitivity of atomic-absorption spectrophotometry for these metals in aqueous solutions. The average increase in sensitivity resulting from solvent extraction and a 2-fold con- centratlon was 6.7 times (range 5.9 t o 7.1) that obtained by direct analysis of the corre- sponding aqueous solution. The increased sensitivity factor was 9.1 for manganese, which resulted from a %fold concentration stage. APPL I c ATI ON- The method described was applied to the analysis of small samples of blood taken from two groups of children. One group had a history of pica whereas the other group did not. For brevity they are referred to as the “pica group” and the “control group,” respectively.The results are given in Tables VII and VIII. Metal Fe * c u Bi Cd Pb Zn co Ni Mn Sr Li Metal TABLE V PRECISION OF THE METHOD Mean concentration found, Standard Sample pg per 100 ml deviation Blood . . . . . . . . . . 343-3 16.49 Blood . . . . . . . . .. 81 2.73 Blood plus 100 pg of Cu per 100 ml 7-42 Blood . . 2.1 1.63 Blood plus 60’;g of Bi per i60 ml ’ * 2.63 Blood . . . . . . . . . . 2-4 0.22 Blood plus 10 pg of Cd per 100 ml. . 0.22 Blood . . .. . . . . .. 10.0 3.1 Blood plus 60 pg of Pb per 100 ml. . 1.8 Blood . . 382 27.4 Blood plus 50b’pg of‘ i n pe; i 0 0 ml’ 821 52.0 Blood . . .. . . . . . . 0-6 1 0.28 Blood plus 10 pg of Co per 100 ml 0-6 1 Blood . . . . . . 10.0 0.78 Blood plus 10 pg of Ni per ib0 ml’ ’ 1.84 Blood .. . . . . . . . . 1.37 0-17 Blood plus 10 pg of Mn per 100 ml 0.60 Blood . . . . . . . . .. 4.7 0.37 Blood plus 10 pg of Sr per 100 ml . . 0.36 Blood . . . . . . . . 0.45 0.09 Blood plus 2.0 pg of Li per 100 ml’ ’ 0-15 181 66.0 12.3 71.4 10.1 19.8 11.6 15.6 2.06 * Fe concentrations are in micrograms per millilitre. Coefficient of variation, per cent. 4.8 3-3 4.1 4.0 9.2 1-8 2.5 7.2 6.3 6.0 7.8 9.3 12.4 5.2 7.9 2.3 7.3 78 31 46 20 TABLE VI LIMITS OF DETECTION . Cu Bi Cd Pb Co Ni Mn Detection limit, p g per 100 ml . . 0.58 3.2 0.74 4.0* 0.44 4.2 0.2 Number of tests 19 10 9 10 9 10 10 10 10 10 10 10 9 10 9 10 10 10 9 6 10 Sr Li 0.56 0*1* * The blank signals were not distinguishable from the flame background signals. These values were twice the standard deviation of the flame background signals.April, 19711 ELEVEN METALS I N SMALL SAMPLES OF BLOOD TABLE VII CONCENTRATIONS OF METALS FOUND IN BLOOD SAMPLES FROM THE CONTROL GROUP COMPARED WITH OTHER REPORTED VALUES Found in this study From literature r A ’I Mean,* Mean, Standard Number Metal p g per 100 ml pg per 100 ml deviation of tests Fe 475 x 102 381 x lo2 47.2 88 c u 107 97 21-8 82 Bi 1.2 0.9 1.7 44 Zn 650 509 140 83 Ccl 0.7 0-5 0-5 88 Pb 20 11 6.5 37 co 0.03 0.4 0.6 65 Ni 4.6 2.2 2-2 76 Mn 2.6 1.2 0.9 90 Sr 0.95 2-9 2.5 75 Li 2.0 0.3 0.4 70 * Data from B ~ w e n .~ 27 1 TABLE VIII RAISED CONCENTRA4TIONS* OF METALS FOUND I N BLOOD SAMPLES Pb Zn Mn Fe Sr Cd Cu Bi Li Ni Co Upper concentration Number of children limit, pgper 100ml 36 930 3.8 240 x 102t 10.5 1.9 160 6.1 1-6 8.7 2-3 with raised concen- trations of metals in : Control group ., 0 1 1 0 2 0 0 1 3 1 1 Picagroup . . 60 22 19 10 9 7 1 3 3 0 0 * Greater than the mean plus three times the standard deviation of the mean of the control group. t Abnormal concentrations of iron in whole blood will be lowered, e.g., in anaemia. This lower concentration limit is the mean less three times the standard deviation of the mean of the control group. Samples listed are those below 240 p g ml-l of iron. There were 194 children in the “pica” group. DISCUSSION No investigations were made into the effects of the hollow-cathode lamp current or slit width on the intensity of the absorption signals. The primary aim in these analyses was to obtain both a sufficiently intense and stable emission of radiation from the hollow-cathode lamp, which would permit a range expansion of up to ten times and at the same time keep the photomultiplier gain to a minimum.This enabled a small time constant to be used in the smoothing circuit of the recorder amplifier, which allowed determinations to be made satisfactorily with 0.20 to 0.30 ml of the organic extract (Fig. 1). The lamp currents and slit widths used fulfilled these requirements, and were usually those recommended by the manufacturer. No study of potential interferences of sample concomitants on the determinations was made because the extraction equilibria were established in the presence of the inorganic constituents of blood $Zus large numbers of added metal ions at concentrations up to about 1000 times those normally present in blood.Further, the high specificity of atomic-absorption spectrophotometry would be complemented by the selective solvent extraction of the metals to be determined. A check for spectral interference between metals extracted together was negative, as expected. The number of operations in the extraction scheme was kept to a minimum, so that the method could easily be applied to large numbers of samples. For example, the hydro- chloric acid concentration of the aqueous solutions of oxidised blood was fixed at 2.2 N, and this enabled the first three extractions to be carried out without any direct adjustments of the acidity of the aqueous phases. Only one accurate adjustment of the pH was necessary. By adjusting the pH of the aqueous phase for the fourth extraction stage to 3.0 0.2 the pH values of the aqueous phases for the subsequent stages were obtained simply by adding the stated volumes of 6 N ammonia solution.272 DELVES, SHEPHERD AND VINTER: DETERMINATION OF [Analyst, Vol.96 For routine application to large numbers of samples the extractions were carried out batchwise over a period of 3 days. In this way two people could in 1 week analyse fifteen samples in duplicate for each of the eleven metals. This amounted to about 440 deter- minations including reagent blanks and standards. The recovery tests for synthetic aqueous standard solutions of metals were quantitative (Table 111) and good correlations were observed between the concentrations found by direct analysis of these solutions and of the extracted organic phases (Table IV).The latter results showed that the extraction procedure, which was established for solutions of oxidised blood samples, was also quantitative for synthetic aqueous standards. The recoveries of metals added to whole blood (Table V) were quantitative for all of the metals determined. The coefficients of variation were better than +S per cent. relative at a concentration of 10 pg per 100 ml for nine of the metals (Table V). With bismuth and lead, however, the precision was poor at this low concentration, which approached the detection limits for these metals. At higher concentrations of about 60 pg per 100 ml the precision was better, being 4.0 per cent. for bismuth and 2.5 per cent. for lead.The precision for lithium was good, being 7-3 per cent. even at the low concentration of 2.00 pg per 100 ml (Le., 0.02 p.p.m.). The extraction - concentration stages resulted in increased atomic-absorption sensitivities of about seven times that obtained by direct analysis of aqueous solutions, and eliminated interferences from any non-specific absorption signals from the constituents of oxidised blood solutions, and from the effect of phosphate ions on strontium. The increase in detection limits was of the same magnitude as the increased sensitivity because the variations in the flame background signals were about the same when aqueous or organic solvents were burned. It would have been possible to improve on the detection limits obtained (Table VI) by using increased range expansion and noise suppression.However, this would have required larger volumes of solution and hence larger sample sizes to avoid dilution. The method described recommends that 2 ml of blood be taken for analysis but only 1 ml is used for the extraction - atomic absorption determinations. In our study of children with pica the remaining 1 ml was used for the determination of chromium, which is not reported here. The sensitivity of the method could be improved by using the whole of the 2 ml of oxidised blood solution for the determinations. On the other hand, smaller volumes of blood, Le., 1 ml or less, could be used for the preparation of the oxidised blood solution, if only a small sample is available. The results for the “control group” (Table VII) agreed with published results for normal persons where these were available.It must be pointed out that the children from this group were those undergoing surgery and were assumed to be biochemically normal. A study of their clinical histories is necessary to confirm this. The abnormally high concentra- tions given in Table VIII showed that, in addition to lead, the metals zinc, manganese, stron- tium, cadmium and lithium could be excessively ingested by children with pica. A detailed survey of all of the results is being carried out, and the clinical histories of the children con- cerned are being evaluated by medically qualified colleagues. CONCLUSION The method described provides the means of trace-metal survey analysis of large numbers of blood samples at a capital cost far below that of other available techniques such as spark- source mass spectrometry, neutron-activation analysis, etc. The sequential separation of metals from a l-ml portion of oxidised blood solution, as described in this paper, enables eleven metals to be determined in 2 ml or less of blood, which is of value when the samples are taken from a child who is young or sick, or both. We thank Professor B. E. Clayton, consultant chemical pathologist at the Hospital for Sick Children for her interest in this work, and The Wellcome Trust for financial support. REFERENCES 1. 2. 3. Tipton, I. H., in Seven, M. J., and Johnson, L. A., Editors, “Metal Binding in Medicine,” J. B. Butt, E. M., Nusbaum, R. E., Gilmour, T. C., Didio, S. L., and Sister Mariano, Arch. Emir. Hlth, Bowen, H. J. M., “Trace Elements in Biochemistry,” Academic Press, London and New York, Lipincott Co., Philadelphia, Montreal, 1960, p. 27. 1964, 8, 60. p. 159.April, 19711 ELEVEN METALS IN SMALL SAMPLES OF BLOOD 273 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. Bicknell, J., Clayton, B. E., and Delves, H. T., J . Ment. Defic. Res., 1968, 12, 282. Evans, C. A., jun., and Morrison, G. H., Analyt. Chem., 1968, 40, 869. Samsahl, K., Brune, D., and Wester, P. O., Int. J . Appl. Radiat. Isotopes, 1965, 16, 273. Allan, J. E., Spectrochim. Acta, 1961, 17, 467. Starv, J., “The Solvent Extraction of Metal Chelates,” Pergamon Press, Oxford, 1964, p. 76. Sandell, E. B., “Colorimetric Determination of Traces of Metals,” Third Edition, Interscience Healy, T. V., J . Inorg. Nucl. Chem., 1968, 30, 1025. Mansfield, J. M., and Winefordner, J. D., Analytica Chim. Acta, 1968, 40, 357. Kirbbright, G. F., and Sargent, M., Analyst, 1968, 93, 552. Hermann, R., and Alkemade, C. Th. J., “Chemical Analysis by Flame Photometry,” Second Delves, H. T., and Vinter, P., J . Clin. Path., 1966, 19, 504. Received July 3rd, 1970 Accepted October 22nd, 1970 Publishers Inc., New York, 1959, p. 60. Edition, Interscience Publishers Inc., New York and London, 1963, p. 32.