Anal. Chem. 2003, 75, 413-419
Selective Determination of Beryllium(II) Ion at Picomole per Decimeter Cubed Levels by Kinetic Differentiation Mode Reversed-Phase High-Performance Liquid Chromatography with Fluorometric Detection Using 2-(2′-Hydroxyphenyl)-10-hydroxybenzo[h]quinoline as Precolumn Chelating Reagent Hiroaki Matsumiya* and Hitoshi Hoshino*
Department of Applied Chemistry, Graduate School of Engineering, Tohoku University, Aoba, Aramaki, Aoba-ku, Sendai 980-8579, Japan
A highly sensitive and selective method for the determination of the Be(II) ion has been developed by the use of reversed-phase high-performance liquid chromatography (HPLC) with fluorometric detection using 2-(2′-hydroxyphenyl)-10-hydroxybenzo[h]quinoline (HPHBQ) as a precolumn (off-line) chelating reagent. The reagent HPHBQ has been designed to form the kinetically inert Be chelate compatible with high fluorescence yield, which is appropriate to the HPLC-fluorometric detection system. The Be-HPHBQ chelate is efficiently separated on a LiChrospher 100 RP-18(e) column with a methanol (58.3 wt %)-water eluent containing 20 mmol kg-1 of tartaric acid and is fluorometrically detected at 520 nm with the excitation at 420 nm. Under the conditions used, the concentration range of 20-8000 pmol dm-3 of Be(II) ion can be determined without interferences from 10 µmol dm-3 each of common metal ions, typically Al(III), Cu(II), Fe(III), and Zn(II), and still more coexistence of Ca(II) and Mg(II) ions at 0.50 mmol dm-3 and 5.0 mmol dm-3, respectively, is tolerated. The detection limit (3σ baseline fluctuation) is 4.3 pmol dm-3 (39 fg cm-3). The extraordinarily high sensitivity with toughness toward the matrix influence was demonstrated with the successful application to environmental Be analyses, such as determination of Be in rainwater and tap water. Nowadays, the high-performance liquid chromatography (HPLC) of metal chelate compounds has been one of the most powerful methodologies for trace-metal analysis.1 Since we reported the first application of ion-pair HPLC to the separation of some anionic chelate compounds,2 we have been engaged in systematic re* Correspondence may be addressed to either author. Phone: +81-(0)22217-7221. Fax: +81-(0)22-217-7223. E-mail (H.M.):
[email protected]. tohoku.ac.jp. E-mail (H.H.):
[email protected]. (1) Wang, P.; Lee, H. K. J. Chromatogr., A 1997, 789, 437-451. (2) Hoshino, H.; Yotsuyanagi, T.; Aomura, K. Bunseki Kagaku 1978, 27, 315346. 10.1021/ac0260847 CCC: $25.00 Published on Web 01/04/2003
© 2003 American Chemical Society
search on ultratrace determinations of metal ions using a precolumn (off-line) chelation technique for the reversed-phase (RP) HPLC system with a photometric or fluorometric detection.3-8 Compared with the on- and postcolumn chelation systems in which a reagent stream is used for chelation, there are two attractive features in the precolumn chelation system in which no labeling reagent is added in the mobile phase. The actual analytical sensitivity can be significantly enhanced with a detector commonly available on the market. Such a detector equipped with a flow-cell unit inherently gives a stable baseline signal with a very low short-term noise level (e.g., (2 × 10-5 absorbance unit). The absence of labeling reagent in the eluent stream allows one to pick up only the chelate signals without any background caused by the reagent, provided that the chelates are completely separated from the excess reagent added at the off-line chelation stage. Therefore, measurement of the chelate peaks with high signal-to-noise (S/N) ratio is readily possible. The second feature is that a unique detection selectivity to certain metal ions can be attained. Contrary to ordinary organic compounds possessing stable covalent bonds, the coordination bonds between a metal ion and ligating atoms in some metal chelates cannot afford to resist the overwhelming force to dissociation, which essentially arises from a very steep decrease in reagent concentration near the chelate bands caused by the RP-HPLC separation. Consequently, only kinetically stable (inert) chelates surviving during elution can reach a detector cell, whereas kinetically unstable (labile) ones are compelled to be dissociated in a column. That (3) Hoshino, H.; Saitoh, Y.; Nakano, K.; Takahashi, T.; Yotsuyanagi, T. Bull. Chem. Soc. Jpn. 2001, 74, 1279-1284 and references therein. (4) Iki, N.; Hoshino, H.; Yotsuyanagi, T. Mikrochim. Acta 1994, 113, 137152. (5) Kaneko, E.; Hoshino, H.; Yotsuyanagi, T.; Gunji, N.; Sato, M.; Kikuta, T.; Yuasa, M. Anal. Chem. 1991, 63, 2219-2222. (6) Sato, M.; Yoshimura, H.; Shimmura. T.; Obi, H.; Hatakeyama, S.; Kaneko, E.; Hoshino, H.; Yotsuyanagi, T. J. Chromatogr., A 1997, 789, 361-367. (7) Iki, N.; Irie, K.; Hoshino, H.; Yotsuyanagi, T. Environ. Sci. Technol. 1997, 31, 12-16. (8) Hoshino, H.; Nomura, T.; Nakano, K.; Yostuyanagi, T. Anal. Chem. 1996, 68, 1960-1965.
Analytical Chemistry, Vol. 75, No. 3, February 1, 2003 413
Scheme 1. Synthetic Pathway to 2-(2′-Hydroxyphenyl)-10-hydroxybenzo[h]quinolinea
a
Reagents: i, Me2SO4, KOH, 1,4-dioxane; ii, n-BuLi, n-hexane-THF; iii, n-hexane-THF; iv, HBr, AcOH.
is, the HPLC column works not only as a conventional mutual separation device but also as a powerful kinetic discriminator to detect selectively the kinetically inert chelates. Thereby, we have proposed that this HPLC system should be termed kinetic differentiation mode HPLC (KD-HPLC).4 Here, it should be emphasized that in the KD-HPLC system, the complete elimination of all the undesirable signals (e.g., the reagent blank and interfering chelate peaks) from the peak position of the target metal chelates is readily accessible using the synergic interactions of four origins of selectivity, that is, the precolumn chelation, chromatographic separation, dissociation kinetics, and spectral selectivity. Hence, KD-HPLC often provides great sensitivity with compatibility to matrix toughness, and its practical applications have been extended to a wide field, for example, clinical analysis of Al5,6 and municipal air monitoring of Ni and V.7 In this work, the specific determination of picomolar levels of the beryllium(II) ion has been successfully achieved with a coupled scheme of KD-HPLC and fluorometric detection using 2-(2′-hydroxyphenyl)-10-hydroxybenzo[h]quinoline (HPHBQ, H2L, see Scheme 1) as a precolumn chelating reagent. The reagent HPHBQ has been designed to satisfy the requirements for KDHPLC-fluorometry scheme, that is, to form a kinetically inert chelate that is compatible with high fluorescence yield, on the basis of our recent resultful findings: (1) The azo dye, 1-(2,4dihydroxy-1-phenylazo)-8-hydroxy-3,6-naphthalenedisulfonate (Hresorcinol), forms a chelate with Be(II) ion that is so inert as to be employed for KD-HPLC-photometry, resulting in giving a detection limit of 0.8 nM (hereafter, 1 M ≡ 1 mol dm-3).8 The inertness of the Be-H-resorcinol chelate seems to arise from its chelating environment to form two aromatic six-membered chelate rings bearing a dinegative charge, which is well-suited to very small metal ions such as Be(II), (2) The Be chelate of 10hydroxybenzo[h]quinoline (HBQ) fluoresces so strongly as to allow nanomolar-level detection, even by conventional solution fluorometry (one-batch method), owing to its large π-electron system.9 The utilization of such great fluorescence of the BeHBQ chelate in a KD-HPLC-fluorometric detection system was attempted for accessing to the pM levels of Be in solution; however, the lability of the Be-HBQ chelate unfortunately hindered this approach. Thus, to solve the problem associated with this kinetic instability, a HBQ derivative having an O,N,O (9) Matsumiya, H.; Hoshino, H.; Yotsuyanagi, T. Analyst 2001, 126, 20822086.
414
Analytical Chemistry, Vol. 75, No. 3, February 1, 2003
donor set to provide a chelating environment similar to that of H-resorcinol, that is, HPHBQ, was prepared by the introduction of a phenol unit into the 2-position of HBQ (see Scheme 1). This ligand is reasonably expected to give the highly fluorescent Be chelate owing to its expanded π-electron system. The KD-HPLC-fluorometric detection system using HPHBQ is proposed in view of the application to environmental Be monitoring, which has grown into an urgent demand not only for the surveillance of environmental pollution but also for the investigation of its influence on ecological systems because of its high toxicity, especially its carcinogenic nature.10 In recent years, a variety of methods have been employed in ultratrace determination of Be: for example, spectrophotometry,11-13 fluorometry,14,15 graphite furnace (GF) atomic absorption spectrometry (AAS),16-21 inductively coupled plasma (ICP) atomic emission spectrometry (AES),22-24 ICP mass spectrometry (MS),25-28 and gas chromatography (GC)-electron capture detection (ECD).29 However, most of the above methods do not seem to satisfy requirements for (10) Skilleter, D. N. Chem. Br. 1990, 26-30. (11) Nukatsuka, I.; Ohba, T.; Ishida, H.; Satoh, H.; Ohzeki, K.; Ishida, R. Analyst 1992, 117, 1513-1517. (12) Valencia, M. C.; Boudra, S.; Bosque-Sendra, J. M. Anal. Chim. Acta 1996, 327, 73-82. (13) Amin, A. S. Anal. Chim. Acta 2001, 437, 265-272. (14) Pal, B. K.; Baksi, K. Mikrochim. Acta 1992, 108, 275-283. (15) Capita´n, F.; Manzano, E.; Navalo´n, A.; Vilchez, J. L.; Capita´n-Vallvey, L. F. Talanta 1992, 39, 21-27. (16) Okutani, T.; Tsuruta, Y.; Sakuragawa, A. Anal. Chem. 1993, 65, 12731276. (17) Nukatsuka, I.; Sakai, K.; Kudo, R.; Ohzeki, K. Analyst 1995, 120, 28192822. (18) Shimizu, T.; Ohya, K.; Kawaguchi, H.; Shijo, Y. Bull. Chem. Soc. Jpn. 1999, 72, 249-252. (19) Peng, H. W.; Kuo, M. S. Anal. Sci. 2000, 16, 157-161. (20) Burguera, J. L.; Burguera, M.; Rondo´n, C.; Carrero, P.; Brunetto, M. R.; de Pen ˜a, Y. P. Talanta 2000, 52, 27-37. (21) Goto, T.; Tange, H.; Yasuda, M. Bunseki Kagaku 2001, 50, 193-196. (22) Tao, S.; Okamoto, Y.; Kumanura, T. Anal. Chim. Acta 1995, 309, 379385. (23) Wu, Y.; Hu, B.; Peng, T.; Jiang, Z. Anal. Chim. Acta 2001, 439, 153-158. (24) Okamoto, Y.; Nakata, F.; Obata, Y.; Takahashi, T.; Fujiwara, T.; Yamamoto, M. J. Anal. At. Spectrom. 2002, 17, 277-279. (25) Suzuki, C.; Yoshinaga, J.; Morita, M. Anal. Sci. 1991 (Suppl. 7), 997-1000. (26) Becotte-Haigh, P.; Tyson, J. F.; Denoyer, E. J. Anal. At. Spectrom. 1998, 13, 1327-1331. (27) Feldmann, I.; Jakubowski, N.; Thomas, C.; Stuewer, D. Fresenius’ J. Anal. Chem. 1999, 365, 422-428. (28) Frengstad, B.; Skrede, A. K. M.; Banks, D.; Krog, J. R.; Siewers, U. Sci. Total Environ. 2000, 246, 21-40. (29) Measures, C. I.; Edmond, J. M. Anal. Chem. 1986, 58, 2065-2069.
environmental Be monitoring in terms of sensitivity, matrix toughness, facility, and cost-effectiveness; preliminary separation and enrichment stages are most often necessary to detect picogram-per-cubic-centimeter levels or less of Be in the presence of complicated matrixes. Additionally, only a few laboratories have access to some powerful, but expensive and sophisticated instrumentations, such as ICPMS. The KD-HPLC method proposed here enables one to determine Be with the extremely low detection limit (S/N ) 3) of 4.3 pM (39 fg cm-3), even in the presence of complicated matrixes, hence, being an alternative promising approach for environmental Be monitoring. EXPERIMENTAL SECTION Apparatus. The fluorescence spectra were recorded on an AMINCO-Bowman model FA-257 luminescence spectrometer (SLM-Aminco, New York) equipped with a 150-W xenon lamp and a 1-cm quartz cell. All fluorescence data in this work, except for quantum yield estimation, were uncorrected for the light source and the photomultiplier response, being represented in the arbitrary units for the instrument employed. The corrected fluorescence data for quantum yield estimation were measured with a model F-4500 spectrofluorometer (Hitach, Tokyo, Japan). The pH’s of the solutions were measured with a model M-13 pH meter from Horiba Corp. (Kyoto, Japan). The HPLC setup consisted of a model LC-9A pump unit, a model RF-10AXL fluorescence detector equipped with a 150-W xenon lamp and a 12-mm3 flow cell of 1-cm light-path length from Shimadzu Corp. (Kyoto, Japan), and a Reodyne model 7125 sample injection valve with a 100-mm3 sample loop. The analytical column used was a LiChrospher 100 RP-18(e) (4.6 mm bore, 125 mm in length, packed with 5-µm particles of octadecylsilanized silica) from CicaMerck Japan (Tokyo, Japan). Reagents and Solutions. The standard solution of Be(II) ion (10.0 mM) was prepared from anhydrous beryllium chloride (>99% from Mitsuwa Pure Chemical Co., Tokyo, Japan) in 10 mM hydrochloric acid solution. The stock solutions of other metal ions (10.0 mM), except for V(V) and Mo(VI), that were used for the interference studies were prepared from the chlorides or the nitrates. The stock solutions of V(V) and Mo(VI) were prepared from ammonium metavanadate and ammonium molybdate, respectively. The synthetic standard solution for air dust was prepared to give the composition and concentrations shown in Table 1. This element composition of the standard solution is similar to that commonly found in air-dust digestion solution.30 The chelating reagent, HPHBQ, was synthesized as described later. The stock solution of HPHBQ (0.50 mM) was prepared by dissolving the solid in an aqueous solution containing 0.10 g cm-3 of a nonionic surfactant, polyoxyethylene 4-nonylphenoxy ether with 20 oxyethylene units (PONPE-20), to facilitate dissolution of HPHBQ and its metal chelates. The pH buffer solution used for precolumn chelation was tris(hydroxymethyl)aminomethane (Tris)hydrochloric acid (pH 8.0; 1.0 M). All other reagents and solvents used were of guaranteed reagent grade. Doubly distilled water was used throughout this study. Preparation of 2-(2′-Hydroxyphenyl)-10-hydroxybenzo[h]quinoline (HPHBQ). HPHBQ was synthesized in a manner (30) Schlieckmann, F.; Umland, F. Fresenius’ Z. Anal. Chem. 1984, 318, 495497.
Table 1. Composition and Concentrations of the Synthetic Standard Solution Containing Main Metals Found in Air Dusta metal ion K(I) Ca(II) Al(III) Fe(III) Na(I) Mg(II) Zn(II) Pb(II) Mn(II) Cu(II) Ba(II) V(V) Ni(II) Cd(II) Co(II) Se(IV) Be(II) a
concn; µmol dm-3
concn; µg dm-3
5000 1500 1300 700 400 330 80 30 20 10 10 3.0 1.4 0.70 0.40 0.30 0.0250
200000 60000 35000 39000 9200 8000 5200 6200 110 640 1400 150 82 79 24 24 0.225
Acidified with HCl (0.01 mol dm-3).
similar to that for 2-(2′-hydroxyphenyl)-8-hydroxyquinoline31 (see Scheme 1). The starting material, 10-hydroxybenzo[h]quinoline (HBQ), was purchased from Tokyo Kasei Co. Ltd. (Tokyo, Japan) and used without further purification. 10-Methoxybenzo[h]quinoline (MBQ). HBQ (9.76 g, 50.0 mmol) was added to a suspension of KOH (3.37 g, 60.0 mmol) in 1,4dioxane (40 cm3). After the powder of KOH and HBQ were dissolved, a solution of dimethyl sulfate (6.31 g, 50.0 mmol) in 1,4-dioxane (25 cm3) was added portionwise with stirring at room temperature. After the orange reaction mixture was stirred for an additional 10 h, the solid residue was removed by filtration. Subsequently, the solution was poured over ∼1000 cm3 of water and then made basic by adding Na2CO3 to give a crude product as a yellow precipitate (8.96 g). The crude product, MBQ, was collected by filtration and then washed well with an ethanol-water mixture (1/1 v/v). Finally, recrystallization from benzene by gradually adding n-hexane gave a pure sample of MBQ as a yellow powder (5.84 g; yield, 55.8% based on HBQ). 2-(2′-Hydroxyphenyl)-10-methoxybenzo[h]quinoline (HPMBQ). A solution of n-butyllithium (∼0.16 mol) in anhydrous n-hexane (100 cm3) was added to a solution of 2-bromophenol (10.7 g, 61.8 mmol) in anhydrous tetrahydrofuran (100 cm3) portionwise with stirring under a nitrogen atmosphere at room temperature. After the reaction mixture was stirred for an additional 50 min to form 2-lithium-lithium phenoxide, a solution of MBQ (7.85 g, 37.5 mmol) in tetrahydrofuran (75 cm3) was added dropwise with stirring over a period of 2 h. After 3 h, the orange reaction mixture was poured over 400 g of crushed ice and neutralized by adding 6 M HCl solution, and then the organic solvents, n-hexane and tetrahydrofuran, were removed by evaporation to give a crude product as a dark-red solid (11.1 g). The crude product, HPMBQ, was collected by filtration and then washed well with a hot ethanol-water mixture (1/1 v/v). Finally, recrystallization from benzene gave a pure sample of HPMBQ as an orange powder (4.86 g; yield 43.0% based on MBQ). (31) Corsini, A.; Cassidy, R. M. Talanta 1974, 21, 273-278.
Analytical Chemistry, Vol. 75, No. 3, February 1, 2003
415
2-(2′-Hydroxyphenyl)-10-hydroxybenzo[h]quinoline (HPHBQ). HPMBQ (2.00 g, 6.64 mmol) was dissolved in a mixture of 30% HBr (90 cm3) and acetic acid (30 cm3). After refluxing for 3 days, the yellow reaction mixture was poured into 250 cm3 of water and neutralized by adding Na2CO3 to give a crude product as a yellow precipitate (1.96 g). The crude product, HPHBQ, was collected by filtration and then washed well with an ethanol-water mixture (1/1 v/v). Finally, recrystallization from benzene by gradually adding n-hexane gave a pure sample of HPHBQ as a yellow powder (1.31 g; yield 68.7% based on HPMBQ). Recommended Procedure for the Determination of Beryllium. To a slightly acidic aqueous solution containing Be(II) ion in a 25-cm3 volumetric flask was added 0.250 cm3 of disodium tartrate solution (0.5 M), 2.50 cm3 of HPHBQ solution (0.50 mM), and 0.250 cm3 of the Tris-HCl pH buffer solution (pH 8.0; 1.0 M). The mixture was diluted with doubly distilled water to the volume, transferred into a polypropylene-capped tube, and then heated for 5 min in a boiling water bath. After cooling to room temperature, an aliquot of the resulting solution was injected into the HPLC system through a 100-mm3 sample loop. As the mobile phase, a methanol (58.3 wt %)-water mixture containing 20 mmol kg-1 of tartaric acid was used at a flow rate of 0.7 cm3 min-1. The apparent pH of the mobile phase was 3.1. The emission at 520 nm from the eluate was monitored using the appropriate sensitivity settings with excitation at 420 nm. Application to Environmental Waters. The sampling was made at the Graduate School of Engineering, Tohoku University (Aobayama campus), in Sendai City on January 13, 2000. The water samples, rainwater and tap water, were collected in polyethylene vessels that had been carefully cleaned with HNO3. Immediately after sampling, 100 cm3 of the sample was boiled for 15 min with an addition of 1 cm3 of 16 mol dm-3 HNO3. After cooling to room temperature, the solution was filtered through a G4 glass filter (5-10-µm pore size). The filtrate was made up to 100 cm3 and stored in a precleaned polyethylene bottle. An aliquot (10.0 or 20.0 cm3) of this solution was subjected to the precolumn chelation protocol. In the recovery tests, the standard solution of Be(II) was added at this stage. The strong acidity of the sample solutions was neutralized with 3 M NH3 solution before the precolumn chelation. Caution. Be and the compounds are very harmful. Handling of these compounds should be done in a ventilating hood. In addition, the waste solutions of the salt and the chelate are to be stored in large polyethylene or Teflon bottles. At this stage, physiological effects of HPHBQ are not known; hence, careful handling of this compound is strongly recommended. RESULTS AND DISCUSSION Spectra and Precolumn Complexation Reaction. Unless otherwise noted, the spectral and complexation studies were performed with PONPE-20 micellar solutions in conformity with the precolumn chelation procedure. The excitation and emission spectra of the Be-HPHBQ chelate are shown in Figure 1a in comparison with those of Be-HBQ chelate. The emission maximum of the Be-HPHBQ chelate is 496 nm, with the excitation maximum at 420 nm. It should be noted that the fluorescence intensity of the HPHBQ chelate (spectrum 1) is ∼3 times greater than that of the parent HBQ chelate (spectrum 2). To evaluate the fluorescence properties of 416 Analytical Chemistry, Vol. 75, No. 3, February 1, 2003
Figure 1. Fluorescence spectra of (a) the Be chelates of HPHBQ and HBQ and (b) HPHBQ alone in aqueous solution containing 10 mg cm-3 of PONPE-20. (a) Chelate spectra: (1) [Be-HPHBQ] ) 10 µmol dm-3, pH 8.0, λex ) 420 nm, λem ) 496 nm; (2) [Be-HBQ] ) 10 µmol dm-3, pH 12.0, λex ) 388 nm, λem ) 494 nm. (b) HPHBQ spectra at 10 µmol dm-3: (1) pH 1.0, λex ) 420 nm, λem ) 625 nm; (2) pH 7.0, λex ) 380 nm, λem ) 625 nm; (3) pH 13.0, λex ) 397 nm, λem ) 590 nm. Table 2. Photophysical Properties of Be Chelates of HBQ and HPHBQ
Be-HBQ chelatea Be-HPHBQ chelatea quinineb
pH
λex nm
/103 dm3 mol-1 cm-1
Φf
12.0 8.0 -
388 420 365
5.15 15.3 3.55
0.460 0.483 0.546c
a Measured in 10 mg cm-3 PONPE-20 solution. b Measured in 0.5 mol dm-3 H2SO4 solution. c Data from ref 32.
the Be chelates, the fluorescence quantum yields (Φf) of the Be chelates were estimated by a conventional comparison method using quinine sulfate in 0.5 M H2SO4 solution as a reference, assuming a yield of 0.546 with excitation at 365 nm.32 As shown in Table 2, the estimated Φf values of the Be chelates are so high as to be >0.4 and are very close each other. On the other hand, the molar absorptivity () of the Be-HPHBQ chelate is ∼3 times greater than that of the Be-HBQ chelate. Judging from these photophysical properties of the Be chelates, the increase in the value of the chelate, which probably arises from the expansion of the π-electron system by the introduction of a phenol unit into the 2-position of HBQ, seems to be responsible for this enhancement of the fluorescence intensity. The excitation and emission spectra of HPHBQ at various pH values are shown in Figure 1b. In contrast with the Be chelate, (32) Demas, J. N.; Crosby, G. A. J. Phys. Chem. 1971, 75, 991-1024.
Figure 2. Effects of pH on the fluorescence intensity of the Be(II)-HPHBQ system in aqueous solution containing 10 mg cm-3 of PONPE-20. CBe ) 10.0 µmol dm-3, CHPHBQ ) 10 µmol dm-3, λex ) 420 nm, λem ) 496 nm.
the reagent HPHBQ has a much poorer fluorescence yield. The spectra of HPHBQ vary with the solution pH because of the acid dissociation reactions. The equilibrium constants, determined by the conventional spectrophotometric titration method as Ka1 ) ([H+][H2L])/[H3L+], Ka2 ) ([H+][HL-])/[H2L], and Ka3 ) ([H+][L2-])/[HL-], are 10-3.39, 10-10.20 and