638 Analyst, October, 1968, Vol. 93, $p. 638-642 The BY Fluorescence Quantum Efficiencies of Some Analytically Useful Chelate Complexes R. M. DAGNALL, S. J. PRATT, R. SMITH* AND T. S. WEST (Chemistry Department, Imperial College, London, S. W.7) A computer method is described for calculating the quantum efficiencies of fluorescent compounds and chelate complexes in both aqueous and organic solvents. The quantum efficiency values can be incorporated into a sensi- tivity factor, S, which may be used to evaluate and assess analytically useful fluorescent complexes. LIMITS of detection in spectrofluorimetric analysis depend to a large extent on the techniques and apparatus used and, therefore, a true measure of the sensitivity of a fluorimetric reagent cannot be obtained merely by comparison of limits of detection derived from diverse sources.An instance of this is the determination of the alkaline earth metals, calcium and magnesium. Here, the common occurrence of these elements frequently affects the magnitude and the reproducibility of the blank and, consequently, the limits of detection quoted by various authors ar reflections of the extent to which the solvents have been purified. Thus, a reliable g de to the sensitivity of an organic, fluorimetric reagent can only be obtained from physical measurements of fluorescence quantum efficiency, extinction coefficient and half-band width of the fluorescence-emission spectrum. The total fluorescence-emission intensity of a species present in a dilute solution is given by- .. * . - * (1) F = 2.303 K#I,E.c.d .. .. where K is a constant depending on instrumental factors, 4 is the quantum efficiency of fluorescence, E is the molar extinction coefficient of the fluorescent species at the wavelength of excitation (which need not correspond to a maximum value in the excitation spectrum), c is the molar concentration of the species, d is the excitation path length and I . is the incident radiation intensity in units of quanta. Only the values of # and E are affected by the choice of spectrofluorimetric reagent. However, it is normal analytical practice to observe only a narrow band width of the total fluorescence-emission spectrum, hence the shape of the spectrum may be allowed for by introducing H , the half-band width of the emission spectrum. The sensitivity factor, S , for the metallofluorescent reagent is then given by- .... .. .. . . (2). S =- 4 H Parker and Reesl have suggested the use of such a sensitivity index, for comparison of fluorescence, but have used values of H in wavenumber units of cm-l and E in units based on mass (pg per ml). As most commercial spectrofluorimeters are now calibrated in wave- length units of nm, the authors recommend the use of H in units of nm. Also, the use of the molar extinction coefficient E would appear to have more significance than a mass-based E . If sufficient instrumental parameters are available, the absolute fluorescence signal in microamps can be calculated.2 Further, if a noise factor can be calculated or assumed, the limit of detection of the compound can be obtained.2 For both of these calculations a knowledge of the S factor is necessary.* Present address : Department of Chemistry, University of Florida, Gainesville 32601, U.S.A. 8 SAC and the authors.DAGNALL, PRATT, SMITH AND WEST 639 REAGENTS- Quinine suZ;bhate standards-These were prepared from quinine (British Drug Houses Ltd.), which was recrystallised from aqueous ethanol and dried to constant weight over phosphorus pentoxide before being dissolved in 0.1 M sulphuric acid to give a 1000 p.p.m. solution. More dilute solutions were prepared by dilution with 0.1 M sulphuric acid. Aromatic hydrocarbons-These were obtained from a collection of purified specimens held by the Organic Chemistry Department, Imperial College. Dyestufls artd indicators-B.D.H.Indicator grade were used without further purification. Azomethine reagents-These reagents were prepared in this laboratory for a previous Calcein-As supplied by G. F. Smith Chemical Co., Columbus, Ohio. All solvents used were of analytical-reagent grade purity; water was distilled from an s t ~ d y . ~ all-glass apparatus. APPARATUS- The photomultiplier and exit monochromator were calibrated for spectral response by using a 1-kW tungsten lamp, run under specified conditions such that the filament operated at a colour temperature of 2854" K. Measurements were carried out with a Farrand spectrofluorimeter. EXPERIMENTAL Determinations of quantum efficiency were carried out by using the relative method described elsewhere in detai1.l~~ Quinine sulphate was used as the standard (4 = O G ) , and the correction procedure used was the same as outlined by the above authors.The photomultiplier (RCA 1P 28) and exit monochromator were calibrated for their relative quantum spectral response by using a standard tungsten lamp (380 to 650 nm). Fluorescent screens of rhodamine B,5 rhodamine B with a~riflavine,~ and fluorescein,l were used to calibrate the xenon arc lamp of the apparatus (250 to 400 nm) ; the xenon lamp was then used to calibrate the detector system over the lower wavelength region. A method similar to that previously described by Parker6 was used. A calibrated xenon lamp is not necessary for the evaluation of the results; it is only used here as a secondary standard for calibration of the lower wavelength range of the photomultiplier and emission monochromator. Fluorescence-emission spectra were obtained for quinine sulphate and unknown solutions by using the same wavelength of excitation.This eliminates the need for a calibrated excita- tion system and depends on the constancy of the quantum yield of the standard or unknown for excitation wavelengths lower than that of the long wavelength absorbance maximum. Absorbances were measured by placing the photomultiplier of the spectrofluorimeter so that the transmitted incident radiation was viewed directly. In this way, errors that might have arisen from differences in the spectral band widths of the spectrofluorimeter and absorption spectrophotometer were eliminated. Absorbances were measured on concentrated solutions (about 10-5 M), and the results used to calculate the absorbances of diluted solutions, assuming Beer's law was obeyed.Ten fluorescence spectra were obtained for each complex, covering a concentration range of two orders of magnitude. Where absorbances exceed 0.02, the filter effect on the excitation light was corrected for, by using a factor, f, which was empirically evaluated from standards. In instances when there is an overlap of excitation and emission spectra ( k , a filter effect on the emission, as well as excitation radiation), a plot of - versus concentration would have given a shallow curve instead of a straight line parallel to the concentration axis. However, we did not observe any self-absorption of this nature with any of our dilute solutions. The values of log f were proportional to concentration, as would be expected from Beer's law.At the concentrations used, self-quenching by collisional de-activation can be assumed to be negligible5 because of the large intermolecular distances involved. Oxygen quenching was considerable with some of the organic compounds listed in Table I, although none of the compounds in Table 11, or quinine sulphate, exhibited this to any appreciable extent. Oxygen quenching was eliminated by bubbling nitrogen through the solutions for 10 to 15 minutes before measurement. The fluorescence of all of the solutions was measured at 20" C. 1 4640 DAGNALL et al. : FLUORESCENCE QUANTUM EFFICIENCIES [Analyst, Vol. 93 The complexes examined had metal-to-ligand ratios of 1 : 1 or 2 : 1. Because no com- plexes of the ratio 1 : 2 were formed, a large excess of metal ion could be used to ensure com- plete reaction of the chelating reagents (about lo2 to 103-fold molar excesses of metal were used).The pH of aqueous alkaline solutions was adjusted to the recommended value with diethylamine. With non-aqueous solutions a few drops of diethylamine were added. CALCULATION- The equation used to derive the result from the experimental values is given below1s4- F.(€.C.d),C#q .. * * (3) .. .. .. = Fq.(E.c.d) where the subscripts, q, refer to the quinine standard, all other symbols having been defined earlier. F is the total fluorescence emission and is equal to the area under the corrected emission spectrum. Fluorescence-emission spectra were digitised at equally spaced frequency intervals, and then corrected for spectral sensitivity variations in the detection system by using the cor- rection factors described earlier.This is normally a tedious and time-consuming process, but was facilitated by the use of a computer. Input to the computer consisted of one card for each frequency. On the card were punched the frequency, the correction factor and up to ten intensity data points for that frequency, taken from different fluorescence spectra. Thirty- one cards constituted a set and the programme was designed to correct the data, integrate the corrected spectrum by using Simpson's rule, normalise the corrected and uncorrected data and finally print out the correct and uncorrected spectra, together with the result of the integra- tion.Calculations were performed on the Imperial College I.B.M. 7090 computer, with a FORTRAN IV programme.' Drushel, Sommers and Cox4 have described the use of a com- puter programme for the same purpose, although their programme differed considerably from ours. The results for F and F, obtained from the computer were substituted into equation 3 to give 4. Previous authors* have noted that comparisons of fluorescence between substances in different solvents must allow for effects arising from the different refractive indices of the media used. These corrections are only of second order importance and differ according to the optical system used. The main correction is to account for the difference in solid angle viewed by the exit monochromator; for our apparatus this was obtained by multiplying the quantum efficiency of fluorescence by the where n is the refractive index of the solvent containing the unknown and n, is the refractive index of 0.1 M sulphuric acid.RESULTS AND DISCUSSION To test the experimental technique and calculation, the results for the quantum effi- ciencies of some standard compounds are given in Table I. These results are for single deter- minations only. Several values obtained by both relative and absolute methods are quoted in the literature, and our results appear to be in good agreement. Values for rhodamine B given in the literature range from about 0.6 to nearly unity, and are probably a result of the different ionisation modes of this dyestuff. Reproducible values would only be expected when buffered, aqueous solvents are used.In addition, the dyestuffs rhodamine B and eosin are difficult to obtain in a pure form. Acriflavine, although a mixture of methylated amino- acridines, appears to give reproducible values of 4, whenever these have been quoted in the literature. In Tables I and 11, values are those which have been corrected for refractive index of the solvent. The results of quantum efficiencies of fluorescence for the magnesium - azomethine complexes are given in Table 11. Standard deviations have been calculated after neglecting the highest and lowest values of 4 within each set of ten. We have also corroborated earlier results3 obtained in this laboratory by finding that the corresponding calcium complexes had 4 values that were immeasurably small. The free reagents were also non-fluorescent, although they gave fluorescent hydrolysis products in aqueous solution.It is probable that the most important factor in determining the quantum efficiencies of these complexes is the energy of the n - n* transition, relative to the v - n* transition. If n - n* transitions due to the azo- methine nitrogen atoms occur at lower energies than the n - n* transitions, a non-fluorescent or weakly fluorescent species will result as a consequence of increased inter-system crossingOctober, 19681 OF SOME ANALYTICALLY USEFUL CHELATE COMPLEXES 641 to the triplet state.s On complexing the reagent with a non-transition metal, the T- n* transition is generally unchanged relative to the 7~ - T* transition of the fully ionised reagent. However, the n - 7 ~ * transition is always shifted to higher energies and, if shifted sufficiently, a fluorescent species may result.This effect has even been observed on formation of hydrogen bonds and has been used to explain solvent-dependent fluorescence.l* Because of the low extinction coefficients of n - 7c* transitions ( E = 10 to loo), they can be experimentally observed only in certain instances. Other authors have used this mechanism to account for the fluorescence of the complexes of 8-hydroxyquinolinell and also for the fluorescence of this reagent in strongly acidic solutions.12 In view of this, it is unlikely that the reagents are non-fluorescent at room temperature solely because of internal conversion de-activation processes caused by free rotation of the molecule; this would not account for the fluorescence of the chemically similar calcium complexes.One more useful observation can be made from the results for the magnesium complexes listed in Table 11. For several years the reagent bis-salicylidene ethylenediamine has been regarded as the most sensitive reagent for magnesium, no doubt because of the careful control of acidity and solvent purity.13 Table I1 shows that other reagents may well approach, or even exceed, the limit of determination quoted by White and Cuttitta.l3 In addition, these reagents may have different properties with regard to selectivity, which may render them even more suitable for a particular application. TABLE I QUANTUM EFFICIENCIES OF SOME STANDARD ORGANIC COMPOUNDS Compound Acriflavine Anthracene Coronene Eosin Fluorescein Rhodamine B Rubrene Solvent Water Ethanol Concentration, 4.0 0.5 tG Per ml Chloroform 10.0 Sodium hydroxide, 0-lw 2.0 Sodium carbonate, 0.1111 1.0 Sodium hydrogen carbonate, Ethanol 1.0 Benzene 10.0 0*1M 4 0.57 0.28 0.26 0.30 0.12 0.82 0.81 0.66 1.02 dcorr.0.57 0.30 0.28 0-33 0-12 0-82 0-81 0.70 1-17 Reference 0~56~4, 0*5416 0 ~ 2 7 ~ , 0-3014 0.31s 0*7717, 0 ~ 7 9 ~ ’ 0.1217, 0.1518 0.85l, 0-8018 0*69l, 0.764 1.0217 TABLE I1 QUANTUM EFFICIENCIES OF FLUORESCENCE OF SOME COMPLEXES COMMONLY USED DETERMINATION OF THE ALKALINE EARTH METALS Complex Mg - SABF Mg - SOPD Mg - SED Mg - Cal. Ca - Cal. Sr - Cal. Ba - Cal. d 0.2 1 0.16 0.39 0.79 1.01 0.74 0.81 dcorr. 0-23 0.18 0.45 0-79 1.01 0.74 0.81 6 6 per x cent.8-7 4-1 18 11 23 5.9 55 7-0 80 8.0 63 8.5 77 9.0 Eh x 108 23.0 50-0 13.5 63-7 52.2 50.3 52.2 A, nm 492 410 360 485 485 485 485 H , nm 47.2 70.1 85.6 54.8 43.5 52-6 54.4 I N THE S 112 86 114 92 130 71 78 SABF is NN’-bis-salicylidene-2,3-diaminobenzofuran in dimeth ylformamide. SOPD is NN’-bis-salicylidene-o-phenylenediamine in dimethylformamide. SED is NN’-bis-salicylidene ethylenediamine in dimethylformamide. Cal. is calcein in water. y5 = quantum efficiency. dc,-,rr. = quantum efficiency corrected for refractive index of solvent. 6 = standard deviation. E = molar extinction coefficient of fluorescent species a t wavelength of excitation. A = absorption maximum of complex (not wavelength of excitation a t which 4 is measured). H = half-band width of fluorescence-emission spectrum.s = sensitivity factor (equal tog). more difficult to explain than with the azomethine reagents. The generation of fluorescence in calcein on complexing with alkaline earth metals is This is mainly because of the642 DAGNALL, PRATT, SMITH AND WEST large number of ionisation modes of this reagent in the ground-state; each of these ground- state ionisation modes will have corresponding excited-state ionisation modes occurring at different pH values. Calcein is essentially a fluorescein molecule, substituted in the 2- and 4-positions with a carboxymethylaminomethyl group, and it is worth noting that the fluor- escence efficiencies of the calcein complexes are of the same order as that of the fluorescein molecules. The sensitivity values quoted for the calcein complexes suggest that calcein would be the most sensitive fluorimetric reagent for determination of the alkaline earth metals.However, as we have already pointed out, the higher degrees of selectivity obtainable with other reagents may well prove to be the more important factors in analysis. Since submitting this paper we have noted a similar comparisonlg of common reagents for aluminium, in which a sensitivity factor defined by S = e+ was used. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. REFERENCES Parker, C. A., and Rees, W. T., Analyst, 1960, 85, 587. St. John, P. A., McCarthy, W. J., and Winefordner, J. D., Analyt. Chem., 1966, 38, 1828. Dagnall, R. M., Smith, R., and West, T. S., J. Chem. Soc., ( A ) , 1966, 1595. Drushel, H. V., Sommers, A. L., and Cox, R. C . , Analyt. Chem., 1963, 35, 2166. Melhuish, W. H., J. Phys. Chem., 1961, 65, 229. Parker, C. A., Analyt. Chem., 1962,34, 502. Smith, R., Ph.D. Thesis, 1967, University of London. Gilmore, E. H., Gibson, G. E., and McClure, D. S., J. Chem. Phys., 1955,23, 399. Kasha, M., Discuss. Faraday Soc., 1950, 9, 14. van Duuren, B. L., Chem. Rev., 1963, 63, 325. Popovych, O., and Rogers, L. B., Spectrochim. Acta, 1959, 15, 584. Ballard, R. E., and Edwards, J. W., J . Chem. SOC., 1964, 4868. White, C. E., and Cuttitta, F., Analyt. Chem., 1959, 31, 2083. Hercules, D. M., and Frankel, H., Science, N . Y . , 1960, 131, 1611. Weber, G., and Teale, F. W., Trans. Faraday Soc., 1957, 53, 646. Bowen, E. J., Ibid., 1954, 50, 97. Forster, L. S., and Livingston, R., J . Chem. Phys.. 1952,20, 1315. Vavilov, S. J., Physik, 1924,22, 266. Babko, A. K., Volkova, A. I., and Get’man, T. E., Zh. Analit. Khim., 1967, 22, 1004; J. Analyt. Received Afiril 30th, 1968 Chem. USSR, 1967.22, 842.