% 30i^Bc 941 'S. i UNITED STATES ATOMIC ENERGY COMMISSION OAK RIDGE TENNESSEE DETERMINATION OF SMALL AMOUNTS OF BERYLLIUM BY FLUORESCENCE MEASUREMENT by A. B. Carlson W. F. Neuman A. L. Underwood I>ublished for use within the Atomic Energy Commission. Inquiries for additional copies and any questions regarding reproduction by recipients of this document may be referred to the Doomients Distribution Subsection, Publication Section, Technical Information Branch, Atomic Energy Commission, P. O. Box E, Oak Ridge, Tennessee. Inasmuch as a declassified docimient may differ materially from the original classified document by reason of deletions necessary to accomplish declassification, this copy does not constitute authority for declassification of classified copies of a similar document which may bear the same title and authors. Date of Manuscript: February 1947 Document Declassified: May 8, 1947 This document consists of 13 pages. / ! I MDDC 941 ABSTRACT The fluorescence of alkaline beryllium solutions with quinizarin has been studied in detail. pH, dye concentration and many common ions have been shown to influence the fluorescence of the complex. From these studies a procedure was developed for the de- termination of beryllium in amounts of 1-10 micrograms. There is good reason to believe that the method can be extended to determine even smaller amounts of beryllium. MDDC 941 DETERMINATION OF SMALL AMOUNTS OF BERYLLIUM BY FLUORESCENCE MEASUREMENT I. INTRODUCTION White and Lowe (1) described the fluorescence of alkaline beryllium solutions with 1- amino-4-hydroxjra.nthraquinone. Fairhall and his co-workers (2) found that this fluorescence was proportional to the beryllium concentration in the range of 0,05 to 10 micrograms. These investigators studied the fluorescence in ultra-violet light by visual comparison. It is further stated (2) that 1,4-dihydroxyanthraquinone (quinizarin) produced the fluorescence as well as the amino compound. Our attempts to reproduce the method of Fairhall (2) led to anomalous results; there- fore studies of the effects of pH, dye concentration, time of standing, and interfering ions were undertaken. Quinizarin was employed in these studies. A procedure based on fluores- cence measurement has been developed for determining beryllium in the range of 1 to 10 micrograms per 20 ml, with a standard error of about 10%. n. EXPERIMENTAL fostrument- Fluorescence intensities were measured by means of the fluorophotometer (Fig. 1) designed and built in this laboratory. This instrument employed two phototubes (and amplifiers), one (the control tube) receiving light directly from the ultra-violet lamp, the other being activated by fluorescence of the sample. A balanced circuit was thus ob- tained, although n\ill-point measurement with a slide- wire was abandoned in favor of direct galvanometer readings. The ultra-violet source was a General Electric Type H4 mercury discharge lamp. Heat resistant Corning ultra-violet filters (no. 5874) were employed to reduce visible light produced by the lamp; Corning filters (no. 3486) were interposed be- tween the sample chamber and the phototube to absorb ultra-violet light and to transmit the orange-red fluorescent light. The lamp was operated with a constant- voltage transformer; this, together with the balanced circuit mentioned above and shown in the diagram, rendered the instrument insensitive to changes in line voltage which are frequently encoimtered. The fluorophotometer circuit was operated with four commercial-type, lead storage batteries arranged in series. Cuvettes made of high- silica glass (Klett Mfg. Co.) were used to hold the solutions. The instrument was checked for stability and linearity of response by meas- uring the fluorescence of quinine sulfate solutions of various strengths (Fig. 2), Corning filters (no. 4303, 3385, and 3389) being used in place of no. 3496 to pass the bluish fluores- cence of quinine sulfate. The adsorption spectrum of quinizarin in alkaline solution was determined, using a Beckmann photoelectric spectrophotometer. From Fig. 3 it is seen that the solution does not strongly absorb light of the principle wave-length of the lamp output (3650 Angstroms). This is important, since penetration of the ultra-violet light is necessary to activate flores- cence throughout the solution when beryllium is present. It was found, however, that addi- tion of beryllium did not significantly alter the absorption spectrum of the solution. This may be disadvantageous, since absorption of the activating light is a prerequisite for fluores- cence. •3- MDDC 941 Pig. ; f L. uonoy HO TOM ere^ -4- MDDC 941 Fig. 2 riUORESCENCE OF Quinine Sulfate IT 20 i\i Quinine Sulf^tc SrecA Solv IN XD Ml ToTfIL VoLUMC TICN -5- MDDC 941 i»» 9» U 7* 5x. Fig. 3 Absorption Spectrum of quin/zarin in Alkaline Solution Jo* Jro MDDC 941 Reagents- 1 ,4-dihydroxyanthraquinone (qvdnizarin) , obtained as the technical grade re- agent from Eastman Kodak Co., was sublimed, and recrystallized from ethyl alcohol, yield- ing an orange-red powder having a melting point of 194-195^0. It is moderately soluble in alcohol, forming an orange solution, and is very soluble in alkalies, giving rise to a deep purple solution. In the studies reported below, a solution of 0.3 mg per ml 95% ethyl alcohol was used. Solution was effected by gentle warming over a steam-bath. Standard beryllium solutions were prepared by dissolving 1 g of the metal in dilute HCl, diluting to a liter, and using 1 ml of this solution to prepare a liter containing 1 microgram beryllium per ml. Diethylamine (Eastman Kodak Co.) was redistilled (B. P. ca. 55°C) and dissolved in water to give a IM solution. Effect of pH- A series of solutions containing 5 micrograms beryllium and 0.2 ml dye solution in a total volxune of 20 ml was prepared, using various amoimts of NaOH to vary the pH from 7.9 to 13.5. It is important to add the alkali before the addition of the dye; other- wise, a difficultly- soluble precipitate of dye forms. Blanks were prepared for each solu- tion. The pH of each solution was determined by means of a Beckmann pH meter. It was found that maximal fluorescence occurs at pH 11.3 - 11.4 (Table I, Fig. 4). Table I EFFECT OF pH ON FLUORESCENCE pH Galvanometer Reading (Sample - Blank) - 7.90 -11 9.60 -4 10.28 6 10.65 18 10.90 32 11.10 49 11.30 62 , 11.40 61 11.60 48 11.70 34 11.80 24 13.50 2 Because the pH was shown to be critical, the solutions were buffered in all further stud- ies. It was established by titration studies , that a combination of diethylamine and its hy- drochloride was a useful buffer at the pH 11.4. A solution of 2 ml of IM diethylamine plus 4 ml O.IN HCl diluted to 20 ml had a pH of about 11.4 which was unchanged by addition of small amounts of beryllium and dye. By comparison of solutions thus buffered with a solu- tion adjusted to the same pH with NaOH, it was established that the buffer does not inter- fere with the beryllium fluorescence. MDDC 941 ic So J» (9 (u ct: ft: i: o > ■J iOX3M0NV/V1V< X3 N MDDC 941 Effect of Dye Concentration- Solutions of constant pH and beryllivtm concentration, with varying amounts of dye, were prepared. Table U and Fig. 5 show that the dye concentration is critical. This was repeated for several beryllium levels, and it was found that for each beryllium concentration there was an optimal amoimt of dye which gave the maximal fluores- cence. Fig. 6 shows a plot of beryllium concentration (in micrograms per 20 ml) vs. the amoimt of dye in mg. giving maximal fluorescence). Table n FFECT OF DYE CONCENTRATION ON FLUOR Mg Dye Net Galv. (Sample-blank) 0.015 37 0.030 56 0.045 64 0.060 64 0.075 66 0.090 64 0.120 59 0.180 40 0.300 26 0.450 17 In preliminary test tube experiments, it was noted that ether extracted dye from the aqueous layer in alkaline solutions containing no beryllium, while if beryllium was present, dye remained in the aqueous layer. Therefore an attempt was made to obtain automatically the optimal dye concentration by adding excess dye to several solutions containing from 1 to 10 micrograms of beryllium and extracting the excess with ether, with the idea in mind that the beryllium might hold in the aqueous layer the appropriate amount of dye. This proved not to be the case. Although excess dye was extracted preferentially, as indicated by increased fluorescence of solutions low in beryllium, some essential dye was also re- moved, since solutions of greater beryllium concentration fluoresced less strongly after extraction. Effect of Time Standing - No measurable changes in fluorescence intensity occurred in solutions standing for as long as an hour-and-a-half. Interfering Substances - Fairhall (2) mentioned the interference of calcium and magnesium in determining beryllium in bone samples by the fluorescence method; these precipitated as phosphates from alkaline solution. Beryllium recoveries from such solutions were cor- rected by factors obtained with bone solutions having known beryllium content. In the present investigation, various common ions were added in ratios to beryllium of 10:1, 100:1, and 1000:1. Na+ , PO|, F , HCO3, CI7 NO3, and SO| were found not to inter- fere with the fluorescence in the above ratios. Ca++, Mg^"^ Fe**t Cu'^'^ and Mn*Mefinitely interfered in ratios of 10:1, fluorescence measurements being low by as much as 30-50%. -10- MDDC 941 O.09 0.*4 0.09 Optimal Dve in Mg -11- MDDC 941 t6 SO Fig. 7 Stamdarp Curve PC I,. < (- u J » J- £ Amount Bsnyt-i-i^M ii^_^ g -12- MDDC 941 In larger amounts, the hydroxides of these metals precipitated, making fluorescence meas- urements impossible. Procedure for Determining Beryllium in Pure Solution- The total volume selected as convenient for fluorimeter employed was 20 ml. Unknown saturations containing from 1 to 10 micrograms of beryllium were treated as follows: 2 ml IM diethylamine were added, followed by 0.3 ml dye solution and 4 ml O.IN HCl; the solutions were then diluted to 20 ml and the fluorescence measured. Beryllium content was read from a standard curve obtained by treating known beryllium solutions in the same manner. Larger amounts of beryllixim coiild be determined by dilutions to give samples within the concentration range required. A typical standard curve is shown in Figure 7. Results- Two groups of "unknowns", one with 24 samples, the other with 20 samples, were analyzed, using the procedure outlined above. Comparisons of the analytical recovery with the true values are shown in Table HI. The standard deviation (sigma) was computed according to the formula: a in % = ^(d ) where d is the percentage deviation for each sample and N is the total number of samples. TABLE ni Group 1 Analyst's Report Actual Be Content micrograms micrograms 3.2 3.4 3.5 4.0 7.9 6.9 4.3 5.1 4.3 4.6 5.3 6.0 9.0 9.0 10.0 7.0 1.9 2.0 8.6 8.0 3.6 4.0 5.2 5.4 4.8 5.5 1.3 1.0 3.4 2.5 4.3 4.5 2.9 3.0 6.8 8.5 3.5 3.5 5.5 6.0 1.5 1.5 5.3 5.6 4.0 5.0 8.2 10.0 ain9i :. = 17 Group 2 Analyst's Report Actual Be Content micrograms micrograms 5.6 3.4 10.0 1.0 8.3 2.4 4.5 3.2 7.8 6.0 4.3 10.0 3.7 10.0 5.4 7.4 10.0 5.7 4.0 6.2 5.0 3.0 10.0 1.0 6.2 2.0 4.1 3.0 7.0 5.0 4.4 8.0 3.5 9.0 5.0 6.0 7.0 5.5 4.0 6.0