Determination of photoirradiated tetracyclines in water by high-performance liquid chromatography with chemiluminescence detection based reaction of rhodamine B with cerium (IV)

Journal of Chromatography A, 1167 (2007) 85–94

Determination of photoirradiated tetracyclines in water by high-performance liquid chromatography with chemiluminescence detection based reaction of rhodamine B with cerium (IV) R. Santiago Valverde a , I. S´anchez P´erez a , F. Franceschelli b , M. Mart´ınez Galera a , M.D. Gil Garc´ıa a,∗ a

Department of Analytical Chemistry, Faculty of Experimental Sciences, University of Almeria, 04061 Almeria, Spain b Instituto di Scienze Chimiche, Universit` a di Bologna, Via San Donato 15, I-40127 Bologna, Italy Received 5 June 2007; received in revised form 13 August 2007; accepted 15 August 2007 Available online 21 August 2007

Abstract A simple, selective and sensitive method has been developed for the simultaneous determination of tetracycline, oxytetracycline, chlorotetracycline, demeclocycline, doxycycline and meclocycline based on reversed-phase high-performance liquid chromatography with chemiluminescence detection. The procedure was based on the chemiluminescent enhancement by photoirradiated tetracyclines of the cerium (IV)–rhodamine B system in sulphuric acid medium. The six tetracyclines were separated on an Aquasil-C18 column with a gradient elution using a mixture of acetonitrile and 0.1 mol L−1 phosphate buffer as mobile phase, photoderivatized using a photoreactor consisting of a tube reactor coil of PFA and a 8W Xenon lamp. Under the optimized conditions, the method was validated with respect to linearity, precision, limits of detection and quantification and accuracy. The relative standard deviation (RSD) on intra-day precision was below 10% and detection limits ranged between 0.12 and 0.34 ␮g L−1 . The proposed method has been successfully applied to the determination of tetracyclines in surface water samples. A possible mechanism of the chemiluminescence in the system is discussed. © 2007 Elsevier B.V. All rights reserved. Keywords: Chemiluminescence detection; Photochemical derivatization; HPLC; Cerium (IV); Rhodamine B; Tetracyclines; Water

1. Introduction In recent years, concern about the presence of pharmaceutical compounds in the environment, their possible adverse effects on humans and ecological systems, and new strains of resistant bacteria, has considerably increased. After they have been administered, a considerable amount of some of these pharmaceuticals is excreted unmetabolized and can remain in the environment [1]. Tetracyclines (TCs) constitute a group of antibiotics, which are naturally obtained by fermentation with some fungi or by semi-synthetic processes, characterized by a broad spectrum of activity against pathogenic microorganisms, acting by inhibiting the formation of proteins within bacteria. These therapeutic compounds are used to control human and animal bacterial infec-

∗

Corresponding author. E-mail address: [email protected] (M.D. Gil Garc´ıa).

0021-9673/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.chroma.2007.08.043

tions and have also found applications in preserving harvested fruit and vegetables, exterminating insect pest and supplementing animal feed [2], in such way that they are frequently used in an indiscriminate manner and ends up contaminating the environment. Possible pathways that allow antibiotics to reach surface waters have been described by Hirch et al. [3]. According to these authors, after intake, antibiotics are subject to metabolic reactions, but a significant amount of the original substances leave the organism unmetabolized via urine or faeces and would therefore enter raw sewage or manure. In addition, the excreted metabolites can even be transformed back to the original active drugs. The polar antibiotics may not be eliminated effectively in sewage treatment plants, as a large part of elimination is achieved by absorption on activated sludge which is partly mediated through hydrophobic interactions and, hence, one can expect to find antibiotic substances in surface waters. Also, an effective pathway into surface water is via runoff, derived from the dispersion of manure on fields as fertilizer. Thus, residues of TCs

86

R. Santiago Valverde et al. / J. Chromatogr. A 1167 (2007) 85–94

have been found in concentrations ranging from 0.11 ␮g L−1 in surface water in the United States [4] to 4.2 ␮g L−1 in surface waters in Germany [5]. Their widespread use and potential adverse effects in the environment have increased interest in their determination. TCs are determined by a number of spectrophotometric [6,7] and fluorimetric [8–11] methods. In addition, several chemiluminescent (CL) methods [12–17] have been exploited for their determination, as well as capillary electrophoretics [18,19] and mass spectrometry methods [1,20,21]. Chemiluninescence is becoming an attractive technique to be used as detection in LC as its good selectivity, high sensitivity, and simple and inexpensive instrumentation, as well as the rapid time of analysis, render it a highly promising tool in the analysis of organic compounds [22]. Several alternative methods have emerged offering the ability to generate light through simple or more complex chemical reactions. Simple methods include direct light emission after oxidation of the target analytes with a suitable oxidant. The most popular agents, are permanganate, periodate, hydrogen peroxide, hypoclorite anions and tetravalent cerium (IV), which preferably oxidize conjugate double bonds [23]. The Ce(IV)-involved CL reactions with sensitizers have been studied and used for the detection of a number of compounds [24–26]. For example, the CL reaction between Ce(IV) and carbaryl in sulphuric acid medium sensitized by rhodamine 6G has been reported for the determination of carbaryl [27]. Rhodamine compounds, as a series of xanthane dyes, have been applied in spectrophometry and fluorescence, and have also been used as CL reagents [28]. Rhodamine B (RhB) was often thought to be a sensitizer for the CL system, but Ma et al. [28] showed that the oxidation of RhB could produce CL and investigated the behaviour of this CL system in an acidic medium. In this way, RhB was used as a CL reagent in the determination of flavonoids [29], l-ascorbic acid [30] and TCs [31] by flow-injection CL detection method. On the other hand, it has been established that the irradiation of photoreactive analytes leads to the formation of species that can be detected providing very sensitive procedures [32], but no works dealing with photodegradation and CL detection have been published in determining antibiotics. This process, has however been widely used for the determination of pesticides [33–35]. For analytical purposes, photochemical derivatization is extremely useful because of its selectivity and sensitivity and many of these reactions have been adapted as post-column detection systems in HPLC [36–42]. The main advantages of the post-column derivatization are that the analytes are separated in their original form, without the need for a complete derivatization reaction (assuming reproducibility) and the reaction products need no stability for a long period of time [43]. The goal of our study was to check the usefulness of the CL reaction of Ce(IV)–RhB system for the detection of TCs antibiotics. We found that after irradiation with UV light the TCs produced a great enhancement on the CL emission from the RhB oxidation by Ce(IV) in sulphuric medium. This enhancement in the CL emission is proportional to the concentration of the selected compounds, which can be determined by measur-

ing the increase in the CL intensity. Based on these findings, a new HPLC-CL method has been developed for the sensitive determination of TCs, which has been satisfactorily applied in surface water samples. 2. Experimental 2.1. Chemical and solvents Analytical standards (pestanal quality) of tetracycline hydrochloride (TTC, 97.3%), oxytetracycline hydrochloride (OTC, 96.2%), chlorotetracycline hydrochloride (CTC, 85.7%) and demeclocycline hydrochloride hemydrate (DMC, 92.8%) were obtained from Riedel-de Ha¨en (Germany); doxycycline hyclate (DC, 98%) was obtained from Fluka (Switzerland) and meclocycline sulfosalicylate salt (MCC, 98%) was obtained from Sigma–Aldrich (Germany). Fig. 1 shows the structural formulas of the six tetracyclines. Acetonitrile (ACN) and methanol (MeOH) of HPLC grade were obtained from J.T. Baker (Holland). Ortho phosphoric acid (H3 PO4 , 85%), sulphuric acid (H2 SO4 , 96%), hydrochloric acid (HCl, 38%) and ethylenediaminetraacetic acid disodium salt2hydrate (Na2 -EDTA) were obtained from Panreac (Spain) and potassium dihydrogen phosphate (KH2 PO4 ) and rhodamine B N,N,N N -tetraacetylrhodamine chlorohydrate (RhB) for analysis were obtained from Merck (Germany). Cerium (IV) sulphate tetrahydrate (Ce(SO4 )2 ·4H2 O) was supplied by Riedel-de Ha¨en (Germany). Ultra pure water, obtained from a Milli-Q water purification system from Millipore (Bedford, MA, USA), was used. Mobile phases were filtered through a 0.45 ␮m cellulose acetate (water) or polytetrafluoroethylene (PTFE) (organic solvents) and degassed with helium prior to and during use. The solid phase pre-concentration (SPE) of water samples was carried out using Oasis HLB (hydrophilic lipophilic bal-

Fig. 1. Structural formulas of the six tetracycline antibiotics studied.

R. Santiago Valverde et al. / J. Chromatogr. A 1167 (2007) 85–94

ance) 6 cm3 cartridges containing 0.5 g of stationary phase from Waters (Milford, MA, USA). The Ce(IV) sulphate solution 0.1 mmol L−1 , prepared in H2 SO4 0.25 mol L−1 aqueous solution, RhB solution 10 mg L−1 were filtered through a Millipore membrane of cellulose acetate (0.45 ␮m particle size) before pumping it into the chromatographic system.

87

digital venturis FP 575 pentium personal computer using a Millenium 32 (Chromatography Manager, Waters, Milford, MA, USA) software was used for acquisition and treatment of data. A vacuum system B¨uchi Vac V-500 and a vacuum controller V-800 (Switzerland) connected to an extraction manifold Waters (Miniford, USA) was used as pre-concentration water system. 2.3. Preparation of standards and spiked samples

2.2. Instrumentation The HPLC-CL system consisted of a Waters (Milford, MA, USA) HPLC equipment, composed of a Model 600E multisolvent delivery system and a Rheodyne 7725i manual injector valve with a 200 ␮L sample loop. The photochemical step was carried out on a photochemical reactor (Supelco, USA) fitted with a knitted open tube reactor coil (4 m × 1.6 mm O.D. and 0.8 mm I.D.) of perfluoroalkoxy (PFA) and a 8W Xenon lamp. CL detection was conducted on a CL detector from Jasco CL-2027 (Tokyo, Japan), which incorporated a modification consisting of placing the mixing chamber as near as possible to the detection cell, being 2.5 cm the distance from the confluence point to the inlet of the flow cell detection. The CL detector was equipped with a uniquely designed coiled PTFE tube flow cell (67 cm × 1.6 mm O.D. and 0.5 mm I.D.), placed immediately in front of a highly sensitive photomultiplier (PMT) which maximizes the CL signal. The HPLC-CL detection system was as depicted in Fig. 2. The CL detector was connected to the HPLC equipment through an interface (Waters busSAT/IN Module). The reagent solution containing Ce(SO4 )2 /H2 SO4 was propelled by a Gilson Minipuls 3 peristaltic pump and RhB solution was pumped with a system Water Model 510. The RhB solution was firstly mixed with the column effluent and Ce(IV) solution was after mixed with the resulting effluent inside the box containing the reaction cell and the CL detector. HPLC separations were performed with an Aquasil C18 150 mm × 4.6 mm (5 ␮m particle size) column from Thermo Electron Corporation (Waltham, MA, USA). A

Fig. 2. Schematic diagram of the HPLC-CL system used in the determination of TCs. P, HPLC pump; I, injector; PHR-UV, photochemical UV reactor; CL, chemiluminescence detector; PC, personal computer; MP, mixing point device; PMT, photomultiplier; W, waste.

Individual analytical standard solutions of TCs (400 mg L−1 ) were prepared by exactly weighing and dissolving the corresponding compounds in ACN. Furthermore, the standard solutions were protected against light and stored at 4 ◦ C in a refrigerator. In these conditions, they were stable for at least 3 months. Working standard solutions of the analytes were prepared daily in H3 PO4 /KH2 PO4 buffer solution at pH 2.6 with 20% of MeOH (v/v) and were filtered through Millipore membrane PTFE filters (0.45 ␮m particle size), before injection into the chromatographic system. 500 mL of aqueous samples were spiked with the appropriate volumes of standards and prepared for extraction by adding 2 mL of 5% Na2 EDTA and adjusting the pH to 2.8 with HCl immediately prior to extraction. 2.4. Procedure 2.4.1. Procedure to pre-concentrate tetracyclines from surface water The Oasis HLB cartridges were pre-conditioned with 5 mL of MeOH and 5 mL of Milli-Q water and then 500 mL of water sample, treated as described above, were passed through them at a flow rate of 8 mL min−1 . The analytes retained on the SPE cartridge were eluted with 5 mL of MeOH, and the extracts were concentrated to dryness under a stream of N2 , redissolved in 1 mL H3 PO4 /KH2 PO4 (pH 2.6):ACN 80:20 (v/v) and then filtered through a 0.45 ␮m PTFE filter before injection in the chromatographic system. This PTFE filter was cleaned with 1 mL H3 PO4 /KH2 PO4 (pH 2.6):ACN 80:20 (v/v) before its use to avoid interferences. 2.4.2. HPLC procedure Samples were chromatographed by a binary (A:B, v/v) mobile phase with a gradient elution for 40 min at a flow rate of 1 mL min−1 . Solvent A was 0.1 mol L−1 KH2 PO4 and H3 PO4 to adjust the pH at 2.6 and solvent B was ACN. The gradient program was as follows: initially A:B (90:10, v/v), then 2 min linear gradient to A:B (85:15, v/v), followed by 6 min linear gradient to A:B (75:25, v/v) and 20 min isocratic with A:B (75:25, v/v); then an additional period of 4 min linear gradient to the initial conditions and finally 12 min in the initial conditions was sufficient time before subsequent analysis runs. 200 ␮L of A:B (80:20, v/v) sample solutions were analyzed under the above conditions. After separation of TCs into the analytical column and photochemical derivatization with a Xenon lamp, the resulting photoproducts were mixed with RhB (10 mg L−1 ) solution

88

R. Santiago Valverde et al. / J. Chromatogr. A 1167 (2007) 85–94

pumped at a flow rate of 0.5 mL min−1 . Before CL detection, the eluent was mixed with Ce(IV) solution (0.1 mmol L−1 ) in H2 SO4 0.25 mol L−1 pumped at a flow rate of 3 mL min−1 inside the box containing the reaction cell and the CL detector. 3. Results and discussion TCs have been determined by a large number of CL methods, most of them employing a luminometer or a flow-injection CL system, whereas only a method has been proposed using HPLC-CL [12]. The lack of HPLC-CL methods may be due to the fact that TCs interacts with the silanol groups and trace metals present in silica packing materials, being unavoidable the use of acid medium for separation; thus, CL detection for these compounds is restricted to reactions developed in acid medium. Taking into account this limitation, experiments were carried out with TCs in a batch system, using a luminometer and microplates. The application of strong oxidizing agents such as KMnO4 /H+ , Ce(IV)/H+ , did not show a positive reaction and only the Ce(IV)–RhB/H+ system proposed by Xiong et al. [31] showed a weak CL. In this way, we attempted to take advantage of the possibility of changing their molecular structure by irradiation with UV light, and then, to try the CL reaction. As a result, the photoirradiated TCs generated an appreciable increase of CL signal for the Ce(IV)–RhB/H2 SO4 system. Two parameters affect the yield in photochemical reactions, namely UV irradiation time and the solvent used. In our study, batch experiments showed that the weak CL signal obtained without irradiation was greatly increased by irradiation up to 15 min with a Xenon lamp. This behaviour occurred in binary aqueous mixtures of MeOH and ACN, but for the purpose of our study, ACN was selected because of its higher elutropic power for separation. Besides, the CL intensity increased when the percentage of ACN decreased in both photoderivatization step and CL reaction media. 3.1. Kinetic curve Previous studies showed that the last added reactive must be Ce(IV) solution and the rate of the CL reaction was fast (from the reagent mixing to the peak maximum in the kinetic curve, only 500 ms elapsed and it took about 5 s for the signal reach near to the baseline). Initially, the commercial CL detector was equipped with an external mixing chamber placed upstream, just outside the dark box containing the detection cell, in such a way that light intensity was emitted in this mixer rather than in the cell. Therefore, a modification, which consists of placing the mixing chamber as near as possible to the cell, inside the dark box, was carried out (Fig. 2). 3.2. Effect of chemiluminescent reagents To obtain the maximal CL intensity in the determination of TCs by HPLC-CL, the effects of RhB, Ce(IV) and sulphuric acid concentrations were investigated (Fig. 3). The concentra-

tion of RhB versus signal intensity was studied at concentrations ranging from 1 to 50 mg L−1 at a flow rate of 0.6 mL min−1 , the flow rate and concentration of Ce(IV) reagent being kept at 2 mL min−1 and 0.1 mmol L−1 Ce(IV) in 0.5 mol L−1 H2 SO4 . Fig. 3a shows that CL intensities increased with the concentration of RhB up to 10 mg L−1 , and then CL intensity slightly decreased. Thus, 10 mg L−1 was selected as the optimum concentration of RhB in order to carry out the remaining experiments. The effects of Ce(IV) concentration on the CL intensities of TCs (Fig. 3b) was investigated in the range of 0.01 to 0.5 mmol L−1 , in 0.5 mol L−1 H2 SO4 , at a flow rate of 2 mL min−1 and remaining constant the flow and concentration of RhB solution (0.6 mL min−1 of 10 mg L−1 RhB). The CL intensity increased with the Ce(IV) concentration through the range checked. However, 0.1 mmol L−1 was chosen as a compromise because higher concentrations could cause problems due to its precipitation in the tubes of flow devices. This optimal value was used in all remaining experiments. Finally, the concentration of sulphuric acid versus CL intensity was studied from 0.1 to 0.75 mol L−1 at concentration values previously optimized for RhB and Ce(IV) solutions (Fig. 3c). CL intensities increased with the sulphuric acid concentration, up to 0.25 mol L−1 , whereas higher concentrations keep the CL signal constant. The selected concentration of sulphuric acid to be used in further experiments was 0.25 mol L−1 . 3.3. Effect of the flow rate of the chemiluminescent reagents The flow rates of reagents solutions are very important to the CL reaction and should be regulated; thus when too slow or too high flow rates are used, CL may be not emitted in the flow cell and hence, the emitter might not be detected. The effect of flow rates of Ce(SO4 )2 /H2 SO4 on the CL intensities of TCs was studied over the range 1–3 mL min−1 , while the RhB flow rate was maintained at 0.6 mL min−1 , as shown in Fig. 4a. The results obtained showed that the CL intensities increased when the flow rates increased not reaching a defined maximum value. The selected flow rate for the Ce(IV) solution was 3 mL min−1 , because higher flow rates caused too much pressure in the flow tubes and connections, as well as an excessive consumption of reagents. The effect of flow rates of RhB was studied from 0.2 to 1 mL min−1 with the Ce(SO4 )2 /H2 SO4 flow rate kept at 3 mL min−1 . As can be seen in Fig. 4b CL intensities slightly increased up to 0.5 mL min−1 , and after CL intensity decreased slightly. Therefore, 0.3 mL min−1 was selected as the most optimal flow. 3.4. Extraction As antibiotics appear at low concentrations in the environment, pre-concentration steps are needed. In this sense, in order to determine TCs at the concentration levels expected in surface water samples, a SPE procedure using Oasis HLB cartridges was performed. Oasis HLB have been widely used in literature to pre-concentrate TCs [1,18] and have shown better results

R. Santiago Valverde et al. / J. Chromatogr. A 1167 (2007) 85–94

89

Fig. 3. Effect of the concentration of reagents on the CL intensity of a standard solution containing 2 ␮g mL−1 of: (䊉) OTC; () TTC; () DMC; () DC; () CTC and () MCC. (a) Effect of the RhB concentration; (b) effect of Ce(IV) concentration and (c) effect of H2 SO4 concentration.

than C18 or STRATA-X [44], which is a modified styrenedivinylbenzene polymer. TCs have been shown to chelate metals, requiring the addition of EDTA to chelate metal ions to improve their recovery [45]. Thus, adding 2 mL of 5% Na2 -EDTA was considered suitable, as described elsewhere [46]. The pH of the sample was adjusted to 2.8 (HCl) to increase the retention of the TCs on the SPE cartridges, as recommended in the literature [1]. On the other hand, because TCs are not stable in solutions at pH < 2.0 [47], extraction using SPE cartridges was performed with sample pH adjusted immediately prior to extraction. 3.5. Optimization of HPLC system The mobile phase of HPLC must not only be suitable for the separation of the TCs, but also compatible with the CL reaction. Several mobile phases have been described for the separation of

TCs in HPLC. Most methods used an isocratic mobile phase, typically consisting of ACN combined with phosphoric acid [12], phosphate/phosphoric buffer [48], oxalic acid [49], citric acid [50] and acetic acid [6], although MeOH is used sometimes in conjunction with ACN [51,52]. With the aim of checking the effect of different acidic media on the yield of the CL reaction, we carried out some experiments with the six TCs in a batch system, using a luminometer and microplates, with different CL reaction mediums. The results obtained showed that CL intensity was inhibited by oxalic acid, citric acid and acetic acid whereas ACN-phosphoric acid and ACN-phosphate/phosphoric buffer, considered mobile phases suitable for separation of TCs, were compatible with the Ce(IV)–RhB system in an acidic medium. In addition, the CL reaction showed a higher intensity when carried out in ACN-phosphate/phosphoric. Therefore, the effect of its concentration on the CL intensity was studied in

90

R. Santiago Valverde et al. / J. Chromatogr. A 1167 (2007) 85–94

Fig. 4. Effect of the flow rate of reagents on the CL intensity of a standard containing 2 ␮g mL−1 of: (䊉) OTC; () TTC; () DMC; () DC; () CTC and () MCC. (a) Effect of flow rates of Ce(IV)/H2 SO4 solution and (b) effect of flow rates of RhB solution.

the HPLC system from 0.01 to 0.1 mol L−1 , using an Aquasil C18 column. When the concentration of the phosphate buffer was 0.1 mol L−1 , the separation was good and CL intensity was maximum. TCs are typically separated using C18 analytical columns. Therefore, HPLC separation was also tried with a Gemini C18 150 mm × 4.6 mm (5 ␮m particle size) column from Phenomenex (USA). This column provided a higher degree of retention (larger time of analysis), whereas the Aquasil column allowed satisfactory separation and lower time of analysis. One parameter affecting the yield in photochemical reactions is the UV irradiation time. This value is dependent on the flow rate and the tube reactor coil (tube length, material, O.D. and I.D.) The flow rate was conditioned by the separation between analytes. Different tube reactor coils were assayed: a tube reactor coil of PTFE (0.3 mm I.D. 0.6 mm O.D.) of 10 m and a tube reactor coil of PFA (0.8 mm I.D. 1.6 mm O.D.) with different dimensions (4, 8 and 10 m). Better results were found with PFA (0.8 O.D. and 1.6 I.D.) and 4 m of length. Fig. 5 shows a chromatogram of the six TCs in the optimized conditions. In all cases, the peak area was used as analytical signal for quantification because of its higher repeatability.

The linear range was established for each TC, the lower limit being the LOQ calculated according the latest criterion and the upper limit, the concentration for which the signal deviates from linearity by 3–5% [55]. Calibration curves were obtained with eight standards covering the whole linear range and each point in triplicate. They showed a good linear relationship (R2 > 0.992) between 0.4 and 15.0 ␮g mL−1 for the OTC, TTC, DMC, DC and between 0.6 and 15 ␮g mL−1 for CTC and MCC. The intra-day precision was tested with 10 repeated injections of different standard solutions containing the analytes at two concentration levels (0.8 and 2.0 ␮g mL−1 ) for each TC, the RSDs being lower than 8.6% in all cases (Table 1). Fig. 6 shows the HPLC-CL chromatograms of a surface water blank extract (a river water sample previously analyzed not containing the target analytes) and of the same surface water blank extract spiked with the six TCs at concentration levels corresponding to the LOQs. Peaks of TCs were well resolved and

3.6. Validation The analytical figures of merit, obtained under the optimum conditions described above, are summarized in Table 1. Limits of detection (LODs) and limits of quantification (LOQs) were calculated statistically [53] as 3.78 and 10 times, respectively, the standard deviation of the signals corresponding to 10 blank solutions divided by the slope of the calibration curve. The LOQs were also calculated, according to the EURACHEM Guidance [54], as the lowest concentration of the analyte for which the relative standard deviation (RSD) of the signal is equal to a fixed percentage (10% in our case).

Fig. 5. HPLC-CL chromatogram of a standard solution containing 2 ␮g mL−1 of: (1) OTC, (2) TTC, (3) DMC, (4) DC, (5) CTC and (6) MCC.

R. Santiago Valverde et al. / J. Chromatogr. A 1167 (2007) 85–94

91

Table 1 Analytical figures of merit obtained using solvent-based standards Compound

OTC TTC DMC DC CTC MCC a b c

Linear range (␮g mL−1 )

0.4–15.0 0.4–15.0 0.4–15.0 0.4–15.0 0.5–15.0 0.5–15.0

R2

0.994 0.997 0.994 0.989 0.995 0.998

LODa (␮g mL−1 )

0.05 0.08 0.04 0.08 0.10 0.10

LOQa (␮g mL−1 )

0.14 0.21 0.16 0.23 0.31 0.29

LOQb (␮g mL−1 )

0.4 0.4 0.4 0.4 0.5 0.5

Repeatability RSD (%)c 0.8 ␮g mL−1

2.0 ␮g mL−1

6.8 7.0 8.6 7.9 5.2 6.9

3.6 6.3 5.9 4.7 2.1 7.6

IUPAC criterion. EURACHEM criterion (RSD 10%). n = 10.

showed no interferences with the real water matrix. However, when solvent-based and matrix-matched calibration graphs were compared, a decrease effect on the analytical signal, which was due to the matrix, was observed for all compounds in the real water sample (Fig. 7). With the aim of checking the presence of matrix effect, the slopes of calibration graphs obtained in both ways were compared for each TC and matrix by means of a ttest [56]. The results showed significant differences between the slopes of the two calibration curves for all TCs in this matrix. Thus, analytical figures of merit were calculated using standards prepared by spiking blank extracts of real surface water. Method detection limits (MDLs) were calculated according to the procedure proposed by de U.S. EPA [57]. The MDL is defined by this organism as “the minimum concentration of a substance that can be measured and reported with 99% confidence that the analyte concentration is greater than zero and it is determined from analysis of a sample in a given matrix containing the analyte. It is essential that all sample processing steps of the analytical method be included in the determination of the method detection limit”. In this way, the MDL takes into account,

Fig. 6. HPLC-CL chromatograms of: (- - - -) a surface water blank extract and (—) the same water sample extract spiked at LOQs concentration levels: (1) OTC, (2) TTC, (3) DMC, (4) DC, (5) CTC and (6) MCC.

not only the matrix effect, but also the variability introduced by all the sample processing steps. This procedure recommends the analysis of a minimum of seven samples containing low enough analyte concentrations (between one and five times the expected MDL when analyzing spiked blank samples and less than 10 times the expected MDL when analyzing samples containing the analyte). Each replicate must be processed through the entire analytical method. An initial estimate of the MDL is then calculated by multiplying the standard deviation of the results by the appropriate t statistic. MDL = t(n−1,α=0.001) × S where n is the number of replicate analyses, S is the standard deviation of the replicate analyses, and t is the student’s t value for n − 1 degrees of freedom at 99% confidence level. The method points out that the variance of the analytical method may change with concentration. If this happens, the estimated MDL will also vary depending on the concentration of the measured replicated samples. The U.S. EPA suggests checking this by analyzing seven replicate aliquots at a slightly different concentration (but still lower than the recommended limit) in order to very the reasonableness of the estimate of the MDL. If

Fig. 7. Calibration curves of DMC in pure solvent (—) and water matrix (- - - -).

92

R. Santiago Valverde et al. / J. Chromatogr. A 1167 (2007) 85–94

Table 2 Calculated EPA method detection limit (MDL) Compound (␮g mL−1 )

Concentration Number of analysis Degrees of freedom (ν) Standard deviation (S) t(n − 1, α = 0.01) SA2 /SB2 Fcrit(nA −1,nB −1,α=0.01) Spooled t(nA −nB ,α=0.01) MDL (␮g mL−1 )

OTC

TTC

DMC

DC

CTC

MCC

0.8 7 6 9173.7 3.143 2.48 3.79 22844.2 4.318 0.06

0.8 7 6 10128.0 3.143 2.43 3.79 25230.0 4.318 0.10

0.8 7 6 4411.8 3.143 2.46 3.79 11863.2 4.318 0.17

0.8 7 6 4351.9 3.143 2.56 3.79 11742.6 4.318 0.16

1.2 7 6 4778.1 3.143 2.13 3.79 12701.7 4.318 0.10

1.2 7 6 5950.0 3.143 2.48 3.79 16013.9 4.318 0.17

the difference between the variance obtained in both cases (SA and SB ) is not statistically significant (based on the F statistic of their ratio), these two variances are pooled to obtain a single estimated S2 as follows: 2 Spooled

Table 4 Recovery percentages and RSD (%) for the determination of TCs in real water spiked at two concentration levels, using calibration graphs built with matrix based standards for quantification Compound

(nA − 1)SA2 + (nB − 1)SB2 = nA + nB − 2

where nA and nB are the number of samples analyzed in each set The MDL is then calculated using the pooled standard deviation as MDL = t(12−1,α=0.001) × Spooled

OTC TTC DMC DC CTC MCC a

Table 2 summarizes the MDL calculations for the six TCs in real surface water samples, as well as the values obtained for this parameter. The MDL values range between 0.06 and 0.17 ␮g mL−1 , being slightly higher than the LODs obtained in solvent. Table 3 contains the LOQs calculated according to the EURACHEM Guidance [54], with a RSD percentage fixed at 10%, along with the intra-day precision, established at 0.8 and 2.0 ␮g mL−1 , both parameters being obtained by using blank surface water extracts to prepare the standards. The RSDs values obtained for repeatability were lower than 10% for all TCs at both concentration levels. 3.7. Recovery studies In order to establish the accuracy and precision of the total method, six replicates of real water samples were spiked at

LOQ (mg kg−1 )

2 × LOQ (␮g kg−1 )

Mean recovery (%)a

RSD (%)a

Mean recovery (%)a

RSD (%)a

98.6 90.3 102.5 89.6 86.2 98.7

7.4 8.5 6.5 8.9 7.4 5.7

108.8 104.3 99.4 87.0 92.8 102.8

4.4 5.7 4.8 6.3 7.0 4.6

n = 6.

two concentration levels of each tetracycline (corresponding to the LOQs in blank water and two times LOQs), extracted and analyzed by using the described method. The mean recovery percentages and the RSDs (%) of the six replicate samples are shown in Table 4. Recoveries for target TCs were between 86.2 and 108.8%, which can be considered satisfactory. 3.8. CL mechanism The reaction of Ce(IV) with RhB in an acidic medium yields significant CL. The CL behaviour of RhB was studied by Ma et al. [28] and a possible mechanism for this CL reaction, based on UV–vis, fluorescence and IR spectra was also proposed by these authors. In addition, this reaction was used by the same authors to determine l-ascorbic acid [30]. The same changes

Table 3 Analytical figures of merit obtained using matrix-matched standards Compound

OTC TTC DMC DC CTC MCC a b c

Linear range (␮g mL−1 )

0.4–15.0 0.4–15.0 0.4–15.0 0.4–15.0 0.6–15.0 0.6–15.0

EPA method detection limit. EURACHEM criterion (RSD 10%). n = 10.

R2

0.995 0.987 0.995 0.989 0.992 0.986

MDLa (␮g mL−1 )

0.06 0.10 0.17 0.16 0.10 0.17

LOQb (␮g mL−1 )

0.4 0.4 0.4 0.4 0.6 0.6

Repeatability RSD (%)c 0.8 ␮g mL−1

2.0 ␮g mL−1

5.8 4.6 7.8 8.7 8.6 6.5

4.2 5.1 6.6 5.2 1.9 2.4

R. Santiago Valverde et al. / J. Chromatogr. A 1167 (2007) 85–94

reported for UV–vis and fluorescence spectra were observed by us in the CL reaction involving TCs. The maximum absorption of RhB at 577 nm (phosphate medium) and the violet colour of RhB faded away simultaneously when a Ce(IV) solution was added. Simultaneously, maximum absorption was observed at 372 nm for the Ce(IV)–RhB mixture, which was higher when the TCs were present at the CL medium. This demonstrates that irradiated TCs act as sensitizers. Finally, the explanation of why TCs act as sensitizers after UV irradiation is that radicals TC• are yielded after irradiation, which takes part in the CL reaction, in the same way as proposed by Ma et al. for the determination of l-ascorbic acid [30]. Based on the above considerations, the possible mechanism of the CL reaction could be explained as follows: Ce(IV)/H2 SO4 • ∗ −→ [ RhB− ]OX hυ •

RhB

→ [• RHB− ]OX + hυ

TC−→TC

[• RhB− ]OX + TC•

Ce(IV)/H2 SO4 •

−→

∗

[ RhB− ]OX + TC

4. Conclusions In this paper, a new HPLC-CL detection method has been developed for the determination of six TCs based on the enhancement of Ce(IV)–RhB CL in sulphuric acid medium, previous post-column photolysis. The method allows the simultaneous and sensitive detection of oxytetracline, tetracycline, demeclocycline, doxycycline, chlortetracycline and meclocycline in surface waters. The proposed method is simple, rapid, fairly sensitive, selective and sufficiently accurate and precise. In addition, it requires no sophisticated instruments. Moreover, the CL reaction is compatible with the mobile phase of HPLC. The tandem photodegradation-CL detection offers an interesting strategy for improving sensitivity and permits to increase the number of compounds to be determined by direct CL (compounds without or with very weak CL behaviour). References [1] S. Revert´e, F. Borrull, E. Pocurull, R.M. Marc´e, J. Chromatogr. A 1010 (2003) 225. [2] M. Hern´andez, F. Borrull, M. Callul, Trends Anal. Chem. 22 (2003) 416. [3] R. Hirch, T. Ternes, K. Haberer, K.-L. Kratz, Sci. Total Environ. 225 (1999) 109. [4] D.W. Kolpin, E.T. Furlong, M.T. Meyer, E.M. Thurman, S.D. Zaugg, L.B. Barber, H.T. Buxton, Environ. Sci. Technol. 36 (2002) 1202. [5] A. Mulroy, Water Environ. Technol. 13 (2001) 32. [6] N. Furusawa, Talanta 59 (2003) 155. [7] P.A. Blackwell, H.C. Holten L¨utzhøft, H.P. Ma, B. Halling-Sørensen, A.B.A. Boxall, P. Kay, J. Chromatogr. A 1045 (2004) 111. [8] M.C. Mahedero, D. Bohoyo, F. Salinas, T. Ardila, D. Airado, B. Rold´an, J. Pharm. Biomed. Anal. 37 (2005) 1101. [9] B. Ferraz Spisso, A.L. de Oliveira e Jesus, M.A. Gonc¸alves de Ara´ujo J´unior, M. Alves Monteiro, Anal. Chim. Acta 581 (2007) 108. [10] C.Z. Huang, Y. Liu, Y.F. Li, J. Pharm. Biomed. Anal. 34 (2004) 103. [11] A. Pena, A. Carmona, A. Barbosa, C. Lino, I. Silveira, B. Castillo, Pharm. Biomed. Anal. 18 (1998) 839.

93

[12] G.H. Wan, H. Cui, H.S. Zheng, J. Zhou, L.J. Liu, X.F. Yu, J. Chromatogr. B 824 (2005) 57. [13] S. Marczynki, Biopolymers 57 (2000) 365. [14] Y. Huang, Z. Zhang, J. Pharm. Biomed. Anal. 35 (2004) 1293. [15] Ch. Lau, J. Lu, M. Kai, Anal. Chim. Acta 503 (2004) 235. [16] A. Townshend, W. Ruengsitagoon, C. Thongpoon, S. Liawruangrath, Anal. Chim. Acta 541 (2005) 103. [17] S. Han, E. Liu, H. Li, Luminescence 21 (2006) 106. [18] L. Nozal, L. Arce, B.M. Simonet, A. R´ıos, M. Valc´arcel, Anal. Chim. Acta 517 (2004) 89. [19] J. Tjø´ rnelund, S.H. Hansen, J. Chromatogr. A 779 (1997) 235. [20] S. Yang, J. Cha, K. Carlson, Rapid Commun. Mass Spectrom. 18 (2004) 2131. [21] M. Cherlet, M. Schelkens, S. Croubels, P. De Backer, Anal. Chim. Acta 492 (2003) 199. [22] P. Fletcher, K.N. Andrew, A.C. Calokerinos, S. Forbes, P.J. Worsfold, Luminescence 16 (2001) 1. [23] G.Z. Tsogas, D.L. Giokas, A.G. Vlessidis, N.P. Evmiridis, J. Chromatogr. A 1107 (2006) 208. [24] Q. Zhang, M. Lian, L. Liu, H. Cui, Anal. Chim. Acta 537 (2005) 31. [25] Q. Zhang, H. Cui, J. Sep. Sci. 28 (2005) 1171. [26] L.F. Capitan-Vallvey, M.C.V. Miron, R.A. Acosta, Talanta 51 (2000) 1155. [27] J.A. Murillo-Pulgar´ın, A. Ala˜no´ n-Molina, P. Fern´andez-L´opez, Talanta 68 (2006) 586. [28] Y. Ma, X. Jin, M. Zhou, Z. Zhang, X. Teng, H. Chen, Anal. Chim. Acta 489 (2003) 173. [29] H. Yao, X. Yang, H. Li, Anal. Lett. 39 (2006) 2007. [30] Y. Ma, M. Zhou, X. Jin, B. Zhang, H. Chen, N. Guo, Anal. Chim. Acta 464 (2002) 289. [31] Y. Xiong, H. Zhou, Z. Zhang, D. He, C. He, Analyst 131 (2006) 829. [32] C. Gonz´alez-Barreiro, M. Lores, M.C. Casais, R. Cela, J. Chromatogr. A 993 (2003) 29. [33] I. Sahuquillo Ricart, G.M. Ant´on-Fos, M.J. Duart, J.V. Garc´ıa Mateo, L. Lahuerta Zamora, J. Mart´ınez Calatayud, Talanta 72 (2007) 378. [34] M.D. Gil Garc´ıa, M. Mart´ınez Galera, R. Santiago Valverde, Anal. Bioanal. Chem. 387 (2007) 1973. [35] M. Mart´ınez Galera, M.D. Gil Garc´ıa, R. Santiago Valverde, J. Chromatogr. A 1113 (2006) 191. [36] B.M. Patel, H.A. Moye, R. Weinberger, Talanta 38 (1991) 913. [37] I.S. Lurie, D.A. Cooper, I.S. Krull, J. Chromatogr. 629 (1993) 143. [38] K.M.S. Sundaram, J. Curry, J. Chromatogr. A 672 (1994) 117. [39] H. Liu, Ch.T. Duda, T. Huang, W.O. Aruda, P.T. Kissinger, J. Chromatogr. A 818 (1998) 69. [40] M. Mart´ınez Galera, T. L´opez L´opez, M.D. Gil Garc´ıa, J.L. Mart´ınez Vidal, P. Parrilla V´azquez, J. Chomatogr. A 918 (2001) 79. [41] T. L´opez L´opez, M.D. Gil Garc´ıa, J.L. Mart´ınez Vidal, M. Mart´ınez Galera, Anal. Chim. Acta 447 (2001) 101. [42] A. Mu˜noz de la Pe˜na, M.C. Mahedero, A. Bautista S´anchez, J. Chromatogr. A 950 (2002) 287. [43] T. P´erez-Ruiz, C. Mart´ınez-Lozano, V. Tom´as, J. Mart´ın, J. Chromatogr. A 1026 (2004) 57. [44] C.R. Anderson, H.S. Rupp, W.-H. Wu, J. Chromatogr. A 1075 (2005) 23. [45] R. Hirsch, T. Ternes, K. Haberer, K.L. Kratz, Sci. Total Environ. 225 (1999) 109. [46] S.M. Croubels, K.E.I. Vanoosthuyze, C.H. Van Peteghem, J. Chromatogr. B 690 (1997) 173. [47] S. Yang, K. Carlson, Water Res. 38 (2004) 3155. [48] F. Zhao, X. Zhang, Y. Gan, J. Chromatogr. A 1055 (2004) 109. [49] P. Vi˜nas, N. Balsalobre, C. L´opez-Erroz, M. Hern´andez C´ordoba, J. Chromatogr. A 1022 (2004) 125. [50] J. Zhu, D.D. Snow, D.A. Cassada, S.J. Monson, R.F. Spalding, J. Chromatogr. A 928 (2001) 177. [51] C.G. Smyrniotakis, H.A. Archontaki, J. Pharm. Biomed. Anal. 43 (2007) 506. [52] A.L. Cinquina, F. Longo, G. Anastasi, L. Giannetti, R. Cozzani, J. Chromatogr. A 987 (2003) 227. [53] L.A. Currie, Anal. Chim. Acta 391 (1999) 105.

94

R. Santiago Valverde et al. / J. Chromatogr. A 1167 (2007) 85–94

[54] EURACHEM Guidance Document No. 1 WELAC Guidance Document No. WGD 2 “Accreditation for chemical laboratories: guidance on the interpretation of the EN 45000 series of standard and ISO/IEC guide 25”, Available from the EURACHEM Secretariat, P.O. Box 46, Teddington, Middlesex TW11 ONH, UK, 1993. [55] D.L. Massart, B.G.M. Vandeginste, S.N. Deming, Y. Michotte, L. Kaufman, Chemometric: A Textbook, Part 2, Elsevier, Amsterdam, 1988, p. 110.

[56] D.L. Massart, B.G.M. Vandeginste, L.M.C. Buydens, S. De Jong, P.J. Lewi, J. Smeyers-Verbeke, Handbook of Chemometrics and Qualimetrics, Part A, Elsevier, Amsterdam, 1997, p. 190, 209, 436. [57] Definition and Procedure for the Determination of the Method Detection Limit, Appendix B to Part 136, Title 40, Revision 1.11, Statistics for Environmental Engineers, Paul Mac Berthovex, Linfield C. Brown, Lewis Pubishers, USA, 1994.