20 Analyst, January, 1967, Vol. 92, Pe. 20-26 The Spectrofluorimetric Determination of Magnesium with ~N’-Bis-salicylidene-2,3-diaminobenzofuran* BY R. M. DAGNALL, K. SMITH AND T. S. WEST (Depavtment of Chewzistvy, Imperial College, London, S . W.7) A~Ar’-Bis-salicylidene-2,3-diaminobenzofuran, SABF, has been synthe- sised and is proposed as a stable and inexpensive spectrofluorimetric reagent for determining magnesium in the range 0.1 to 6 p g (2 x to 1 x 10-1 p.p.m.). The method is both rapid and sensitive, and has a detection limit of about p.p.m. in 50 per cent. aqueous methanol a t an apparent pH of 10.5. The orange fluorescence (545 mp) of the 1 : 1 complex is excited a t 475 mp and can be measured within 15 to 45 minutes of mixing the solutions. Few other ions yield a fluorescence with the reagent and those that do, or that interfere in other ways, may be tolerated in 100-fold molar proportions by simple addition of masking agents.The fluorescent magnesium - SABF complex can be extracted quantitatively into isobutyl methyl ketone without interference from 2000-fold amounts of calcium, thus providing an extremely useful method for determining magnesium in technical materials, etc. To demonstrate this, nine municipal water samples and fourteen blood plasma samples have been analysed successfully for magnesium by the proposed method in a 50 per cent. aqueous methanol medium. THERE are few reagents currently available for the spectrofluorimetric determination of submicrogram amounts of magnesium in the presence of calcium.Lumomagneson IREA (2-hydroxy-3-sulpho-5-chloro-benzeneazo barbituric acid) 1 and 2,2’-dihydroxyazobenzene2 give fluorescence reactions with magnesium, but both reagents form coloured calcium complexes which give rise to inner-filter quenching when calcium is present in greater than a 10-fold molar excess over magnesium. The magnesium complexes of these two reagents can only be extracted into organic solvents with difficulty. The most sensitive fluorimetric magnesium reagent proposed to date is NN‘-bis-salicylidene eth~lenediamine~ which will detect about However, the use of this reagent is restricted to anhydrous dimethylformamide solutions. This presents severe difficulties to its practical application and, in addition, calcium forms a coloured, non-fluorescent complex.Other reagents that have been proposed for the spectrofluorimetric determination of magnesium include 8-hydroxyquinoline,4 methyl salicylate5 and calcein.6 Although these reagents are sensitive, they also give a strong fluorescence reaction with calcium and the other alkaline earths. NN’-Bis-salicylidene-2,3-diaminobenzofuran, SABF, (I) was first prepared in this labora- tory7 and contains the same reactive groupings as NN’-bis-salicylidene ethylenediamine.8 p.p.m. of magnesium. (1) HY5 / Investigations showed that SABF gave a fluorescence reaction with magnesium in aqueous methanol solutions, and that the fluorescence intensity was only slightly influenced by the degree of reagent hydrolysis. This latter factor is of prime importance with this type of * Paper first presented by R.Smith at the meeting of the Society at Bristol on May 4th, 1966.DAGNALL, SMITH AND WEST 21 Schiff-base reagent, because hydrolysis occurs easily and the reaction products are fluorescent, thus giving high blanks. The detection limit for magnesium with this reagent p.p.m.) is at least as good as that obtained with lumomagneson,l and probably better than that obtained with 2,2'-dihydroxyazobenzene, although Diehl, Olsen, Spielholtz and Jensen2 have not quoted sensitivity figures for the latter. Similar excesses of calcium can be tolerated with both of these reagents and SABF. In contrast, however, SABF is easily prepared and purified. Further, traces of magnesium can be extracted quantitatively from aqueous solution into isobutyl methyl ketone.This extraction procedure is unique in that only a single extraction is necessary, c f. dihydroxyazobenzene, and that a non-fluorescent calcium complex is extracted, cf, 8-hydroxyquinoline. The extraction of the coloured calcium - SABF complex can be suppressed by the addition of strontium - EDTA. APPARATUS- A Farrand Optical Co. spectrofluorimeter (Catalogue No. 104244)-This was fitted with a 150-watt d.c. xenon arc lamp and R.C.A. IP28 photomultiplier. The excitation mono- chromator was used with 20-mp or 10-mp half band-width slits and the emission monochro- mator was used with 10-mp half band-width slits. Fused quartz cells (10 x 20 x 30 mm) were used throughout. The spectral response curves for the lamp and photomultiplier have been given el~ewhere.~ A Unicam SP9OOA atomic-absorption jame-emission spectrop~otometer-This was used with an air - acetylene flame at 2852 A.Calibrated jasks-These were coated periodically with silicone "Repelcote," a 2 per cent. solution of dimethyldichlorosilane in carbon tetrachloride (Hopkin and Williams Ltd.), to minimise adsorption of ions on the glass. REAGENTS- NN'-Bis-saZicylidene-2,3-diaminobenzofu~an, SABF-De-aerate a mixture of 50 ml of salicylaldehyde and 80ml of ethanol by passing nitrogen through it for 20 minutes, and then add 15 ml of concentrated ammonia solution. A white or pale lemon precipitate is formed immediately. Place the mixture in an ice-bath and, while stirring, add to it 6 g of potassium cyanide dissolved in 9 ml of water. During the course of the reaction the pre- cipitate changes from lemon to orange - yellow, and after about two hours sets to a solid mass.Break up the mass and add 50ml of ethanol. After a further 90 minutes add 15ml of concentrated ammonia solution and continue stirring for a further 30 minutes. Filter the yellow or orange product and wash it well with 50 per cent. v/v water - methanol solution to remove the cyanide, followed by methanol, until the filtrate is clear yellow. Remove the residue from the filter, triturate with about 300ml of methanol and filter. Dry in a vacuum over calcium chloride. [Yield 2.95 g (90 per cent. based on potassium cyanide). M.p. 181" to 183" C.] Although the product obtained above is satisfactory for analytical purposes, the com- pound can be recrystallised from benzene - light petroleum or dioxan - water.[M.p. 182" to 183" C.] METHOD Microanalysis results obtained for C,,HI,N20, (molecular weight = 356) Found, per cent. . . . . . . C 74.0 H 4.4 N 7-8 Calculatcd, per cent. . . .. . . C 74.2 H 4-5 N 7.8 Molecular weight 356 (mass spectrum). Dissolve 0.2 g of the reagent in 1 litre of dioxan (general-purpose reagent). This solution is stable for several months. +H 10.5 bufer-Add 60 ml of diethylamine (general-purpose reagent) to 500 ml of 50 per cent. v/v methanol - water solution and adjust the apparent pH to 10.5 & 0.1 with about 25 ml of concentrated hydrochloric acid. Standard magnesium solution-Dry analytical-reagent grade magnesium oxide in an oven at 120" C for 3 hours. Dissolve 1.663 g in the minimum amount of 2 N hydrochloric acid and dilute the solution to 1 litre with de-ionised water.This gives a solution containing Prepare by dilution a stock solution containing 100 p.p.m. of magnesium from which working solutions of 1 p.p.m. or 0.1 p.p.m. can be obtained. Solutions containing 1 p.p.m. or less of magnesium should not be stored for longer than two weeks because of losses caused by adsorption on the walls of the vessel. p.p.m. of magnesium.22 DAGNALL, SMITH AND WEST : SPECTROFLUORIMETRIC [Analyst, Vol. 92 Methanol, analytical-reagent grade. Isobutyl methyl ketone, general-purpose reagent grade. Pyridine, general-purpose reagent grade. Diethylamine, general-purpose reagent grade. Distilled water from an all-glass distillation apparatus was de-ionised by passing it through a 15 x 5-em column of Zeo-Karb 225 (Permutit Co.Ltd., London) cation-exchange resin in the hydrogen ion form. PROCEDURE- Development of JEuorescence ;% 50 per cent. methanol-To each 50-ml calibrated flask add 25ml of methanol, 1 ml of pH 10.5 buffer and a solution containing between 0.1 and 6 pg of magnesium in about 15 ml. Add masking agents where necessary (see Interferences) at this point and allow the solutions to stand for 5 minutes. Add 5ml of 0.02 per cent. SABF reagent in dioxan and dilute to the mark with de-ionised water. Measure the fluorescence intensity at 545mp, with an excitation wavelength of 475mp, between 15 and 45 minutes after the addition of reagent. Analysis of blood plasma samples-The procedure above was followed by using 50 pl of untreated blood plasma and standard calibration solutions containing 0, 1 and 2 pg of magnesium.No masking agent was used, and it was not found necessary to de-proteinate the samples. Analysis of water samples-The above procedure was followed by using 0.5, 0.25 or 0.05 ml of water and calibration solutions containing 0, 2, 4 and 6 pg of magnesium. Use 2 ml of strontium - EDTA masking agent, made by adding 50 ml of 10-1 M EDTA solution to 100 ml of 10-1 M strontium bromide, to overcome interference arising from the calcium. Extraction procedure-To about 25 ml of aqueous solution containing 0.1 pg to 6 pg of magnesium in a 250-ml separating funnel, add masking agent as required (see Interferences), followed by 2 ml of pyridine and 1 ml of diethylamine. Add 25 ml of isobutyl methyl ketone and 2 ml of 0.1 per cent.SABF reagent in dioxan. Shake the mixture thoroughly for 1 minute, separate and discard the aqueous layer. Dilute the organic layer to 50 ml in a graduated flask and measure the fluorescence intensity at 555 mp, with an excitation wavelength of 525 mp. RESULTS AND DISCUSSION SPECTRAL CHARACTERISTICS- All Schiff bases hydrolyse to some extent in aqueous solution and the effects of hydrolysis have been carefully evaluated for SABF. The uncorrected fluorescence excitation and emission spectra for the complex and blank are shown in Fig. 1. Correction curves for spectral variations Fig. 1. Fluorescence emission and excitation spectra for: A, 1 ml of M magnesium solution in the presence of 5 ml of 0-02 per cent. reagent; B, reagent blank measured in 50 per cent.aqueous methanol a t an apparent pH of 10.5January, 19671 DETERMINATION OF MAGNESIUM 23 in lamp emission and photomultiplier response have been given el~ewhere.~ The fluorescence of this blank was measured in the presence of EDTA and is, therefore, caused entirely by the hydrolysis products rather than by the presence of foreign cations. The emission maximum of the blank fortuitously coincides with the excitation maximum of the magnesium complex and, as a result, the blue fluorescence of the hydrolysis products does not interfere with the measurement of the orange fluorescence of the magnesium complex. The corrected fluorescence excitation spectrum for the magnesium - SABF complex is compared with the absorption spectrum for the complex in Fig.2. The fluorescence excitation maximum for the isobutyl methyl ketone extract is at 525 mp. This peak is not the same as that used for excitation in 50 per cent. aqueous methanol where it occurs at 485 mp (Fig. 1). The fluorescence emission maximum is shifted from 545 mp in 50 per cent. aqueous methanol to 555 mp in the ketone extract. These may be genuine solvent effects or may be the result of inner-filter quenching caused by the excess of reagent. 1 I 1 I I I 300 400 500 Wavelength mp - Be - I I I 1 I Apparent pH 9.0 9.5 10.0 10.5 11.0 I 5 Fig. 2. Comparison of A, the fluorescence Fig. 3. Variation of fluorescence intensity excitation spectrum (after correction for spectral with apparent pH in a 50 per cent. aqueous variations in xenon arc lamp output) with B, the methanol medium for: A, magnesium complex; absorption spectrum of the complex by using 1 ml of a M magnesium solution and 5 ml of a 0.02 per cent.reagent solution B, reagent blank INFLUENCE OF PH- The fluorescence intensity was measured over a range of apparent pH values for 50 per cent. aqueous methanol. The pH was adjusted by the addition of diethylamine and hydro- chloric acid (Fig. 3). The corresponding blanks were obtained by adding one drop of 10-1 M EDTA solution to each of the solutions, after the fluorescence intensity of the magnesium complex had been measured. Maximum fluorescence intensity was obtained at pH 10.5 0-2. This pH value can be adequately maintained with the buffer described and the fluorescence intensity is independent of the volume of buffer used.REAGENT CONCENTRATION- Maximum fluorescence intensity was obtained by using 4 to 10 ml of a 0.02 per cent. solution of the reagent in dioxan. The recommended volume of reagent, 5 ml, corresponds to a concentration of about 5.6 x M in the final solution. This represents a 10-fold molar excess over the highest point on the calibration curve, i.e., 6 pg of magnesium in 50 ml of solution. There is a tendency for precipitation t o occur when larger excesses of reagent are used or if the pH falls below about pH 10. EFFECT OF METHANOL- Methanol has been used with other reagents1y2 to improve their sensitivity towards magnesium. These reagents were found to give results that were dependent on the methanol concentration.With SABF, however, there is no difference in fluorescence intensity for24 DAGNALL, SMITH AND WEST: SPECTROFLUORIMETRIC [APzabysi!, VOl. 92 methanol concentrations between 80 and 40 per cent. Below 40 per cent. of methanol, precipi- tation of the reagent occurs. The use of methanol concentrations higher than .50 per cent. reduces hydrolysis of the excess of reagent. DEVELOPMENT TIME- The fluorescence intensity increases during the first 10 minutes after addition of the reagent. The heat generated on mixing methanol and water is sufficient to raise the tempera- ture of the solution to above 30" C. As the solution cools, the fluorescence intensity increases as a result of the negative temperature coefficient which is common to most fluorescence systems.A temperature effect also applies to the hydrolysis of the excess of reagent. In this instance the speed of hydrolysis increases with increase of temperature. As the reagent becomes hydrolysed, its inner-filter effect is reduced and the fluorescence intensity rises. If the solutions are allowed to reach room temperature before the addition of reagent, the rate of hydrolysis of excess of reagent is slow. As a result of these two opposing temperature effects, no meaningful temperature coefficient could be evaluated for the system. If no cooling time is allowed, the fluorescence intensity, relative to a quinine sulphate standard, is constant for between 15 and 45 minutes after addition of the reagent. With the extraction procedure, on the other hand, the fluorescence intensity may be measured immediately after dilution and the fluorescence is stable for at least 2 hours.INTERFERENCES- The ions that react with SABF to form complexes have been summarised in a previous communication from this laboratory.' At pH 10.5 in 50 per cent. aqueous methanol, only magnesium, zinc and aluminium form fluorescent complexes with SABF. Many metals of ionic radius of less than 1 react to give coloured complexes which give rise to inner-filter action. The theoretical background to this has been discussed already' and, therefore, is not repeated here. The effects of 100-fold molar excesses of foreign ions on the fluorescence intensity pro- duced by 5 ml of M magnesium solution (about 1.2 pg) were examined. The limiting error was taken empirically as +5 per cent.of a magnesium standard containing no foreign ion. The following ions caused no interference in the 50 per cent. aqueous methanol procedure: As(III), As(V), Ba, Be, B0,3-, Cr(VI), Ga, Ge(IV), Hg(II), K, Mo(VI), Na, Sr, Ti(IV), Tl(1) and V(V). In the absence of masking agents, 100-fold amounts of Ca, Cd, Co(II), Cr(III), Cu(II), Fe(II), Fe(III), Mn(II), Ni, Pd(I1) and Zn interfered by formation of coloured, non-fluorescent complexes, which gave rise to low results. Aluminium formed a fluorescent complex at this pH; the complex was only weakly fluorescent, but gave rise to high results. The even more weakly fluorescent zinc complex gave rise to low results because the large excess of zinc over reagent prevents the formation of the magnesium complex.The following ions interfered by precipitation as chlorides or hydroxides: Ag, Hg(I), Pb, Sn(II), Sn(IV), Th and Zr. In addition, Bi, In, La, Sb(III), Se(IV), Te(1V) and Y gave results that were low by about 10 per cent., although they did not show any visible signs of precipitation or colour formation. The effects of various masking agents on selected interferences were, therefore, examined. Calcium could be tolerated in a 10-fold molar excess without the use of masking agents, and the interference arising from larger amounts of calcium was completely suppressed by a mixture of 10-1 M strontium and EDTA solutions in the ratio 2 to 1. Two hundred-fold molar excesses of calcium could be tolerated when 2 ml of the above mixture was used as a masking agent.The interference of 100-fold excesses of lanthanum and yttrium was also prevented in this way. Some metals that form very stable EDTA complexes in slightly acid solution, e.g., aluminium, transition metals, etc., were not masked with strontium - EDTA because of the high pH of the determination. The interferences arising from Ag, Cd, Co(II), Cu(II), Ni and Pd (11) were completely suppressed by the addition of 1 ml of 3 per cent. potassium cyanide, but Fe(II), Fe(II1) and Zn were not masked in this way. Marginal interferences were examined in more detail.January, 19671 DETERMINATION OF MAGNESIUM 25 The addition of 5ml of a 10-1 M sodium fluoride solution completely suppressed the interference of 100-fold excess of selenium(1V) or tellurium(1V). However, if this masking agent is used in the presence of large amounts of calcium (about 100 pg) co-precipitation of calcium fluoride and magnesium fluoride occurs and low readings are obtained.The interference arising from 100-fold excesses of iron(III), aluminium or indium could be reduced by about 50 per cent. by using 0.5 ml of triethanolamine as a masking agent. This reagent may, however, be useful for complete suppression of the effects of lesser amounts of these metal ions. The effects of other masking agents were considered. One hundred-fold molar excesses of oxalate, citrate, tartrate, nitrilotriacetic acid, tram-l,2-diaminocyclohexane-NNN’N’- tetra-acetic acid and EDTA reduced the fluorescence intensity of the magnesium complex to below the blank value. A 100-fold molar excess of orthophosphate gave a reduction in signal of about 40 per cent. No interference, however, was noted from acetate, bromide, chloride, cyanide, fluoride, iodide, nitrate, perchlorate, sulphate or sulphide when present in 100-fold molar excess over magnesium.Similar concentrations of oxidising agents such as chromate or vanadate caused no interference. Of these materials interfering in the 50 per cent. aqueous methanol method, no.inter- ference was noted in the extraction procedure from aluminium, mercury(I), manganese(II), lead and antimony(II1) in 100-fold molar excess over magnesium. All the other interfering metal ions resulted in low recoveries. Various masking agents, e.g., thiourea, tiron, sodium sulphide, o-phenanthroline, sodium fluoride and potassium cyanide, were used in an attempt to improve the selectivity of the extraction. Cyanide prevented the interference of copper only, and the other masking agents proved ineffective.However, with strontium - EDTA solution, calcium could be tolerated in 2000-fold molar excess over magnesium. This is a considerable improvement on the 50 per cent. aqueous methanol procedure. Large amounts of the above interfering ions could probably be quantitatively removed by one or two pre-extractions from aqueous solution with either cupferron or sodium diet hyldit hiocarbamate.1° Manganese could not be prevented from interfering. STRUCTURE OF THE COMPLEX- A spectrofluorimetric study of complex formation with solution techniques7 indicated the formation of a 1 :1 magnesium - SABF complex. Further examination of the solid complex by infrared spectroscopy and therm~gravimetry~ indicated the structure as [Mg(H,O) ,(SABF)].This accounts for the slight solubility of the complex in pyridine, diethylamine, dimethyl- sulphoxide and dioxan, and its insolubility in chloroform, carbon tetrachloride and aromatic or aliphatic hydrocarbon solvents. Presumably the process of dissolution in any of these solvents corresponds to replacement of the water molecules by polar solvent molecules. No precipitation occurs when saturated solutions of the complex in polar solvents [Mg(solvent) ,(SABF)] are added to light petroleum, despite the fact that the solid complex [Mg(H,O) ,(SABF)] is completely insoluble in this solvent. The structure proposed also accounts for the necessity of the presence of pyridine in the extraction procedure.APPLICATIONS TO PRACTICAL SAMPLES- Water samfiZes-Nine water samples have been analysed for magnesium by the procedure described. The results (Table I) have been compared with those obtained by flame photo- metry on undiluted samples. The approximate calcium concentration of the samples was determined by EDTA titration with Acid Alizarin black SN as indicator. Plasma sam@es-Fourteen plasma samples were analysed for magnesium by using 50 p1 of sample and without subjecting this to any pre-treatment. The results are summarised in Table I1 and some flame-photometric results are also included. The flame-photometric method, however, requires 1 ml of sample and because of this, analysis of all the samples by this method was not possible.It is probable that the fluorimetric procedure could be adapted to the direct analysis of other biological materials. The extraction procedure cannot be used for this determination because of protein precipitation during extraction.26 DAGNALL, SMITH AND WEST TABLE I DETERMINATION OF MAGNESIUM IN BLOOD PLASMA Sample l% Per ml t-G Per ml Found, Flame photometry, 1 2 3 4 5 G 7 8 9 10 11 12 13 14 Origin Gloucester . . .. PontvDool . . . . Witney . . .. 28 28 26 28 30 28 23 30 23 21 22 22 27 26 27 - 26 26 27 - 25 23 25 - 21 26 24 - 29 26 27 29 20 - - 24 26 - 14 15 - 19 20 - 25 26 - 27 27 - - - - - - - TABLE I1 ANALYSIS OF WATER SAMPLES Magnesium, Magnesium, Calcium, photometry metry titration I% Per ml Pg Per ml CLg_per ml Flame Fluori- bDTA . . .. 5.8 . . .. 14.2 .. .. 6.3 5.8 14.6 5-8 170 115 36 Marl6;ook (treated well-water) . . 1.8 1.5 32 Clacton . . .. . . .. 46 46 210 London . . .. .. . . 6.8 6.9 260 Bagshot . . .. .. . . 7.5 7.7 135 Marlow . . .. .. . . 2.0 2-3 110 Exetcr . . .. .. .. 6.0 6.0 64 We thank Dr. I. Macintyre of the Postgraduate Medical School, Hammersmith, for supply- ing blood plasma samples, and to the Science Research Council for the award of a Research Studentship to one of us (R.S.). Thanks are also given to I.C.I. (Mond Division) for the loan of the Farrand spectrofluorimeter used in this work. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. REFERENCES Serebryakova, C. V., Lukin, A. M., and Bozhevol’nov, E. A., U.S.S.R. Patent 129,273, June 15th, Diehl, H., Olsen, R., Spielholtz, G. I., and Jensen, R., Analyt. Chem., 1963, 35, 1144. White, C. E., and Cuttitta, F., Ibid., 1959, 31, 2083. Watanabe, S., Frantz, W., and Trottier, D., Analyt. Biochern., 1963, 5, 345. Aurivillius, B., and Stenson, P., Ark. Kenzi, 1965, 23, 551; Chern. Abstr., 1965, 63, 1680f. Wallach, D. F. H., and Steck, T. L., Analyt. Chern., 1963, 35, 1035. Dagnall, R. M., Smith, R., and West, T. S., J . Chem. Soc., 1966, in the press. Baldwin, J. E., Blythin, D. J., Cooper, R., and Smith, R., J . Chem. Soc., submitted for publication. Dagnall, R. M., Smith, R., and West, T. S., Talanta, 1966, 13, 609. Stary, J., “The Solvent Extraction of Metal Chelates,” Pergamon Press, Oxford, 1964, p. 158. 1960; Chem. Abstr., 1961, 55P, 2365a. Received July 18th, 1966