Anal. Chem. 2003, 75, 922-926
Liquid Chromatography with On-Line Electrochemical Derivatization and Fluorescence Detection for the Determination of Phenols Jo 1 rg Meyer,† Andre´ Liesener, Sebastian Go 1 tz, Heiko Hayen, and Uwe Karst*
Department of Chemical Analysis and MESA+ Research Institute, University of Twente, P.O. Box 217, 7500 AE Enschede, The Netherlands
A new methodological approach for the determination of monosubstituted phenols is described. After liquid chromatographic separation of the analytes, an on-line electrochemical derivatization is carried out and the reaction products are detected fluorometrically. Phenols are oxidized in the electrochemical cell to form fluorescent dimers and higher oligomers, which were identified by on-line electrochemistry/mass spectrometry. Major advantages of the proposed method include enhanced selectivity and sensitivity. Without prior enrichment of the analytes, limits of detection down to 2 × 10-9 M (20 fmol) may be reached for selected phenols, e.g., for 4-octylphenol, 4-ethylphenol, and 4-(1-indanyl)phenol. Only readily available instrumentation is required for these measurements. Phenolic compounds are important analytes in environmental and food chemistry. Several publications deal with the determination of phenols in highly diverse matrixes including water,1-8 biological fluids,9-11 beverages,12-14 wood extracts,15 and aviation jet fuel.16 Most authors rely on high-performance liquid chromatography (HPLC) with UV/visible, fluorescence, and electrochemical detection or combinations thereof.1,3,4,9-11,14-16 More * Corresponding author. E-mail:
[email protected]. † Current address: Philips Research Laboratories, Weisshausstrasse 2, 52066 Aachen, Germany. (1) Engelsma, M.; Kok, W. Th.; Smit, H. C. J. Chromatogr. 1990, 506, 201210. (2) Chao, Y.-C.; Whang, C.-W. J. Chromatogr., A 1994, 663, 229-237. (3) Lamprecht, G.; Huber, J. F. K. J. Chromatogr., A 1994, 667, 47-57. (4) Galceran, M. T.; Ja´uregui, O. Anal. Chim. Acta 1995, 304, 75-84. (5) Whang, J.; Chatrathi, M. P.; Tian, B. Anal. Chim. Acta 2000, 416, 9-14. (6) Tsai, C.-Y.; Her, G.-R. J. Chromatogr., A 1996, 743, 315-321. (7) Rodrı´guez, I.; Llompart, M. P.; Cela, R. J. Chromatogr., A 2000, 885, 291304. (8) Wissiack, R.; Rosenberg, E.; Grasserbauer, M. J. Chromatogr., A 2000, 896, 159-170. (9) Lores, M.; Garcı´a, C. M.; Cela, R. J. Chromatogr., A 1994, 683, 31-44. (10) Vin ˜as, P.; Lo´pez-Erroz, C.; Marı´n-Herna´ndes, J. J.; Herna´ndes-Co´rdoba, M. J. Chromatogr., A 2000, 871, 85-93. (11) Montanari, L.; Perretti, G.; Natella, F.; Guidi, A.; Fantozzi, P. Lebensm.-Wiss. Technol. 1999, 32, 535-539. (12) Aly, F. A.; Al-Tamimi, S. A.; Alwarthan, A. A. J. AOAC Int. 2000, 83, 12991305. (13) Mardones, C.; Rı´os, A.; Valca´rel, M. Electrophoresis 1999, 20, 2922-2929. (14) Jen, J.-F.; Tsai, M.-Y. J. Chromatogr., B 1994, 658, 87-92. (15) Baran, H.; Schwedt, G. Z. Lebensm. Unters. Forsch. 1993, 196, 370-374. (16) Bernabei, M.; Bocchinfuso, G.; Carrozzo, P.; De Angelis, C. J. Chromatogr., A 2000, 871, 235-241.
922 Analytical Chemistry, Vol. 75, No. 4, February 15, 2003
recently, high-performance liquid chromatography with mass spectrometric detection (HPLC/MS) has also been used for phenol determination.8 Capillary electrophoresis (CE) with fluorescence,2 amperometric,5 UV/visible,13 or MS6 detection has been described frequently for this purpose as well. The known detection techniques suffer from severe limitations. UV/visible detection even when a diode array detector (DAD) and multiwavelength detection is used can provide neither the selectivity nor the sensitivity required for direct real sample measurement; thus, cumbersome sample preparation has to be carried out.10,11,13,15 Direct fluorescence detection (FL) is restricted to those phenols that show native fluorescence and the detection limits are dependent on their unfavorable fluorescence properties.1,10,14,15 Electrochemical detection (ED) is highly sensitive but is adversely influenced by oxidizable matrix components, which are oxidized at the relatively high detection potentials. Additionally, when LC separation is desired, gradient elution is not compatible with ED, reducing the separation capacity and lengthening analysis times in LC.1,4,5,16 LC/MS and CE/MS, although highest selectivity is obtained, still require sample preparation to enrich the analytes and instrumentation is comparably expensive.6,8 Some authors use dedicated detection schemes tailored for phenolic compounds or special subgroups thereof, including the following examples: CE with laser-induced indirect fluorometry is very sensitive but lacks the selectivity of direct fluorometry.2 HPLC with postcolumn photochemical derivatization can be applied to the sensitive and selective determination of phenolic aldehydes only.9 Another method uses HPLC with on-line reaction detection, where phenols are oxidized by cerium(IV) and the increase of cerium(III) fluorescence is detected successively. This technique allows for ultratrace analysis of this group of analytes, but any compound oxidized by cerium(IV) will also be detected.3 The coupling of phenols in the presence of peroxidase enzymes and hydrogen peroxide has been used in Analytical Chemistry since the late 1960s.17 The reaction products are claimed to be dimers of the respective phenols.17-19 These biphenols exhibit favorable fluorescence properties when compared to the monomers. This reaction has been used recently for the determination (17) Guilbault, G. G.; Brignac, P. J.; Juneau, M. Anal. Chem. 1968, 40, 12561263. (18) Casella, L.; Poli, S.; Selvaggini, C.; Beringhelli, T.; Marchesini, A. Biochemistry 1994, 33, 6377-6386. (19) Eickhoff, H.; Jung, G.; Rieker, A. Tetrahedron 2001, 57, 353-364. 10.1021/ac0204808 CCC: $25.00
© 2003 American Chemical Society Published on Web 01/15/2003
Table 1. Methanol/Water Binary Gradient Profile Used for the Separation of Phenols (Flow Rate 0.7 mL/min) time (min) MeOH content (%, v/v)
Figure 1. Scheme of the instrumental setup.
of paracetamol by HPLC with postcolumn derivatization.20 Based on this method, we developed a new technique for the determination of phenols: The phenols are separated by HPLC, postcolumn oxidation is carried out electrochemically, not enzymatically by peroxidase/H2O2, and the reaction products are detected fluorometrically. The substitution of the peroxidase reaction by an electrochemical oxidation has the following advantages: The complexicity of the required instrumentation is reduced, no unstable chemicals (enzymes, hydrogen peroxide) have to be added, and the selectivity can be expanded to a greater variety of phenols by using an electrode instead of an enzyme for the oxidation. To our knowledge, the only related technique reported so far is dedicated to the determination of nitro-substituted polynuclear aromatic hydrocarbons: The nitro group of the analytes is reduced in the analytical cell of a coulometric detector and the formed highly fluorescent amines are detected fluorometrically.21 EXPERIMENTAL SECTION Apparatus. Figure 1 depicts the chromatographic system with electrochemical postcolumn derivatization. Gradient elution was performed on a Shimadzu (Duisburg, Germany) HPLC system consisting of two LC-10AS pumps, GT-154 degasser unit, SIL-10A autosampler, SPD-M10Avp diode array detector, RF-10AXL fluorescence detector, and CBM-10A controller unit with class LC-10 software version 1.6. The equipment used for postcolumn electrochemical derivatization was obtained from ESA Inc. (Chelmsford, MA). It comprised a GuardStat potentiostat and a model 5021 conditioning cell. The working electrode material was glassy carbon with a Pd counter and a Pd/H2 reference electrode. The working potential applied throughout this work was 0.9 V versus Pd/H2. For protection of the working electrode, a PEEK in-line filter (ESA Inc.) was mounted between column and electrode. Rebuffering of the eluent was carried out via a low-void-volume mixing tee (Upchurch Oak Harbor, WA). Alkaline buffer (NH3/ NH4Cl, pH 9.5, 0.39 M) was supplied by a Knauer (Berlin, Germany) 64 HPLC pump. The mixing tee was situated between the electrochemical cell and the fluorescence detector. For HPLC/MS measurements, the following equipment from Shimadzu was used: SCL-10Avp controller unit, DGU-14A degas(20) Meyer, J.; Karst, U. Chromatographia 2001, 54, 163-167. (21) Murayama, M.; Dasgupta, P. K. Anal. Chem. 1996, 68, 1226-1232.
0 10
0.5 10
4 30
11 30
16 80
17 90
23 90
27 10
28 stop
ser, two LC-10ADvp pumps, SIL-10A autosampler, SPD10AV UV/ visible detector, LCMS QP8000 single quadrupole mass spectrometer with electrospray ionization probe, and Class 8000 software version 1.20. Postcolumn electrochemical derivatization was carried out as described above for the fluorometric measurements. HPLC Conditions. The mobile phase consisted of a binary gradient of the following solutions: The first (A) was a formate buffer containing 250 mg of ammonium formate 97% (Aldrich, Steinheim, Germany) and 0.6 mL of formic acid p.a. (Fluka, Buchs, Switzerland) in 1 L of deionized water with 0.01% v/v trifluoroacetic acid (TFA) 99+%, spectrophotometric grade (Aldrich). The second (B) was methanol Chromanorm HPLC grade (Merck, Darmstadt, Germany) also containing 0.01% v/v TFA. Separation was performed on a 150 mm × 4.6 mm i.d., 5-µm particle size, 100-Å pore size Discovery C18 column (Supelco, Deisenhofen, Germany) equipped with a 20 mm × 4.6 mm guard column of the same material. The profile of the gradient is shown in Table 1. Injection volume was 10 µL. For flow injection analysis/mass spectrometry (FIA-MS) measurements, an eluent consisting of 30% v/v A and 70% v/v B was delivered at a flow rate of 0.3 mL min-1. MS Conditions. For ESI-MS, the curved desolvation line (CDL, a stainless steel capillary serving both for evaporation of remaining excess solvent and as a vacuum restrictor, respectively) temperature was set to be 210 °C and the CDL voltage was -5 V. Deflector voltages were -50 V, probe voltage was -5 kV, and the detector voltage was 1.7 kV. The flow rate of the nebulizer gas was 4 L min-1. Reagents. DL-Octopamine and DL-p-hydroxyphenyllactic acid were from Sigma (Deisenhofen, Germany), 4-hydroxyphenylacetic acid and 4-hydroxyphenylpropionic acid were from Fluka, 4-nnonylphenol was from Lancaster (Mu¨hlheim am Main, Germany), and all other phenols were from Aldrich. Phenols were purchased in the highest purity available. 4-Hydroxyphenylacetic acid and 4-hydroxyphenylpropionic acid were recrystallized from water twice. RESULTS AND DISCUSSION The coupling of phenol radicals is well known to analytical17 and synthetic19 chemists. The nature of the reaction products depends on the oxidizing agent used. For analytical purposes, the fluorescence properties of the products and the redox potential of the respective phenols are of particular interest. When biphenols or higher oligomers are formed, the fluorescence properties change dramatically compared with the monomeric phenols. Thus, by oxidation of phenols under suitable conditions (e.g., reagent, potential, etc.), highly fluorescent compounds are formed. These may readily be detected by fluorometry. The major approach of this work is to oxidize phenols electrochemically after liquid chromatographic separation and to detect the oxidation products fluorometrically. The resulting Analytical Chemistry, Vol. 75, No. 4, February 15, 2003
923
technique should be more selective than detection by native fluorescence (if present at all), coulometry, or UV/visible spectroscopy and more sensitive than native fluorescence or UV/ visible detection. Different groups of phenols were selected as model compounds for this work: The first group was known fluorogenic peroxidase substrates and related compounds: 4-hydroxyphenylacetic acid (pHPA), 4-hydroxyphenylpropionic acid (pHPPA), 4-hydroxyphenyllactic acid (pHPLA), 3-(4-hydroxyphenyl)-1-propanol (pHPPol), and 4-hydroxybenzyl alcohol (pHBol). The second group comprised para-substituted alkylphenols: 4-ethylphenol (pEP), 4-octylphenol (pOP), and 4-nonylphenol (pNP). The third group consisted of alkylphenols showing a different substitution pattern or more sterically demanding structures: o-cresol (oMP) and 4-(1indanyl)phenol (pIP). The last group was made up of two biologically relevant amines with phenolic groups as substituents: octopamine (Oct) and tyramine (Tyr). Finally, phenol (P) itself was also part of the investigations. This selection should by no means be understood as a complete list of the substances amenable to this detection scheme. We attempted to investigate a range of substances as broad as possible and to cover some of the species of environmental, biological, or food chemical relevance. Using the gradient provided in Table 1, baseline separation of all listed phenols could be carried out within 28 min. The concentration of the organic phase was not raised above 90% to ensure sufficient concentration of the base electrolyte. The detection scheme had to be optimized in several steps. Optimization was carried out with four selected phenols: Oct, pHPLA, pHPA, and pHPPA. This choice was made for the following considerations: First, the fluorescence wavelengths for the enzymatic oxidation products of pHPA and pHPPA were known; second, the high polarity of the substances ensured low retention times, and thus, relatively fast separation was possible. All parameters were varied independently and considered to be optimal when the peak height reached the highest values for a majority or all of the substances. The potential of the coulometric cell was varied in the range between 500 and 1800 mV by 100 mV. An electrode potential of 900 mV versus Pd/H2 resulted in the largest value for the peak height of all peaks. For most substances, the fluorescence signal is strongly dependent on the pH value. In this context, it was recognized that while the buffer used for separation was acidic (pH 20-fold for some compounds) only by rebuffering of the eluent. As the concentrations were similar for all substances (∼10-5 M), it is obvious that the overall fluorescence yield, which results from the combined conversion rate of fluorophor formation, the molar absorptivity, and the quantum yield, is dependent on the structure of the phenol. This value is high especially for the alkylphenols with bulky substituents. The comparison between UV/visible and fluorescence detection clearly demonstrates the sensitivity of the new technique, as is depicted in Figure 3. Elucidation of the nature of the oxidation products was attempted by two different approaches. The first one was to record the UV/visible spectra of the reaction products on-line by use of the DAD. For this purpose, the detector was connected down-
(22) Meyer, J.; Karst, U. Analyst 2001, 126, 175-178.
(23) Meyer, J., unpublished results.
924 Analytical Chemistry, Vol. 75, No. 4, February 15, 2003
Figure 2. Chromatograms for the separation of 11 phenols with different PCD conditions. Peaks: 1, Oct; 2, pHBol; 3, pHPA; 4, P; 5, pHPPol; 6, pHPPA; 7, oMP; 8, pEP; 9, pIP; 10, pOP; 11, pNP. Peaks marked with an asterisk belong to unidentified substances.
Table 2. Calibration Data for the Examined Phenols substance
limit of detection (M)/(fmol)
limit of quantification (M)/(fmol)
linear range (M)
regression coefficient
octopamine tyramine 4-hydroxybenzyl alcohol 4-hydroxyphenyllactic acid 3-(4-hydroxyphenyl)-1-propanol 4-hydroxyphenylpropionic acid 4-hydroxyphenylacetic acid o-cresol 4-ethylphenol 4-nonylphenol 4-octylphenol 4-(1-indanyl)phenol
6.6 × 10-8/660 5.4 × 10-7/5400 4.2 × 10-7/4200 9 × 10-6/90000 4.1 × 10-7/4100 2.4 × 10-7/2400 2.7 × 10-7/2700 1.5 × 10-7/1500 1.1 × 10-9/11 1.1 × 10-8/110 2.4 × 10-9/24 1.6 × 10-9/16
2.2 × 10-7/2200 1.8 × 10-6/18000 1.4 × 10-6/14000 3 × 10-5/300000 1.4 × 10-6/14000 7.9 × 10-7/7900 9 × 10-7/9000 4.9 × 10-7/4900 5.3 × 10-9/53 3.5 × 10-8/350 8.1 × 10-9/81 5.4 × 10-9/81
1.1 × 10-7-1.1 × 10-5 3.4 × 10-7-1.1 × 10-5 4.2 × 10-7-1.4 × 10-5 9 × 10-6-1.6 × 10-5 4.1 × 10-7-1.2 × 10-5 1.3 × 10-6-1.3 × 10-5 3.5 × 10-7-1.2 × 10-5 1.6 × 10-7-1.6 × 10-5 5.3 × 10-9-2 × 10-5 2.6 × 10-8-2.6 × 10-5 8.1 × 10-9-1.3 × 10-5 5.4 × 10-9-1.1 × 10-5
0.997 0.999 0.997 0.997 0.992 0.998 0.998 0.997 0.999 0.999 0.993 0.998
Figure 4. ESI-MS spectrum of the oxidation products of pEP.
Figure 3. Separation of the phenols listed in the caption for Figure 2 with UV/visible and PCD/fluorescence detection.
stream of the electrochemical cell. The UV/visible spectra of the phenols change considerably upon oxidation: Absorption maximums were shifted from ∼280 nm to ∼400 nm. This clearly shows that a major transformation has taken place, which leads to an enlarged chromophoric system. The second approach to obtain structural information consisted of LC/MS or FIA-MS measurements, respectively. The ionization of the phenols in the separation buffer turned out to be low in yield. Therefore, it was decided to use the analytes in comparably high concentration (∼1 mM) and to perform FIA-MS measurements of each compound separately. The reason for this was that the LC conditions of the fluorescence detection should be altered only as little as possible in order to elucidate the nature of the oxidation products under exactly the same conditions. One consequence of this was that a large number of parameters (type of buffer, buffer concentration, pH, cell potential, etc.) were not varied during the MS measurements. Due
to the low ionization yields of the phenols, MS spectra could not be obtained for all phenols. Figure 4 shows the MS spectrum of the oxidation products of pEP. Peaks were observed at m/z ) 241, 361, and 481. The mass of pEP is 122 g/mol. This leads to the following conclusion: At the electrode, phenol radicals are formed under the loss of a hydrogen atom. The recombination of two of these radicals leads to a dimer with the mass 242 g/mol. This may couple with another radical, which leads to a trimer of 362 g/mol. A third coupling leads to a tetramer of 482 g/mol. The ESI-MS signals can be traced back to the [M - H]- pseudomolecular ions of the dimer, trimer, and tetramer of pEP. The exact position of the coupling cannot be determined by MS: Both C-C and C-O coupling have been reported in the literature18,19 for coupling reactions of the phenols. We propose that at least some of the product should be formed through C-C coupling, as the fluorescence properties indicate the presence of a bi- (tri- or tetra-) phenolic compound.22 Under the given conditions, the presence of dimers could also be observed for pIP and pHBol. When the electrode potential was switched off, the oligomers could no longer be detected in all cases. Calibration has been carried out for Oct, Tyr, pHPLA, pHPA, pHPPA, pEP, pOP, pNP, pHBol, pIP, pHPPol, and oMP. The results are strongly dependent on the nature of the analytes: For several phenols, the limit of detection is very low, reaching into the 10-8 M (Oct, pNP) and even the 10-9 M (pEP, pIP, pOP) concentration range. Only for pHPLA, the limit of detection is unsatisfactory with concentrations in the upper micromolar range. The calibration data for the phenols (limit of detection, limit of Analytical Chemistry, Vol. 75, No. 4, February 15, 2003
925
quantification, linear range, and regression coefficient for the plot of signal vs concentration) are summarized in Table 2. After ∼200 injections of a mixture of 10-4 M standards of various phenols, a decrease of the fluorescence signal was observed. This can be traced back to the formation of a layer of oxidation products on the electrode surface, thus resulting in reduced effectiveness of the oxidation. Simple rinsing of the electrode with dimethyl sulfoxide failed to restore the initial effectiveness, but rinsing with toluene brought the electrode back into operational status. Cleaning with half-concentrated nitric acid cleans the electrode surface efficiently, but this aggressive method is recommended by the manufacturer of the electrochemical cell only as the last choice. However, within the relevant submicromolar concentration range of the phenols, several hundred chromatographic runs can be performed before a deactivation of the electrode surface is observed. Under appropriate precautions, the method can therefore be considered to be comparably robust. Selectivity is very high due to the combination of HPLC separation, electrochemical conversion, and fluorescence detection. In the case of complex matrixes, it may be helpful to run samples once with and once without the electrochemical cell to detect sub-
926
Analytical Chemistry, Vol. 75, No. 4, February 15, 2003
stances that exhibit native fluorescence at the wavelengths under discussion and therefore to further increase selectivity. In summary, an entirely new and very powerful approach for the determination of phenols has been developed. On-line oxidation of the analytes leads to the formation of dimers, trimers, and tetramers as was elucidated by ESI-MS. The oxidation products are detected fluorometrically. The overall detection scheme is highly selective and sensitive. The range of phenols that is accessible by this detection scheme comprises compounds with several different substituents in para position as well as o-cresol and unsubstituted phenol. Therefore, it seems likely that an even wider range of phenols can be detected by this technique. Other phenols shall be examined for detectability soon. ACKNOWLEDGMENT Financial support by the Deutsche Forschungsgemeinschaft (DFG, Bonn, Germany) is gratefully acknowledged. Received for review July 23, 2002. Accepted November 18, 2002. AC0204808
