Analytica Chimica Acta 804 (2013) 66–69

Contents lists available at ScienceDirect

Analytica Chimica Acta journal homepage: www.elsevier.com/locate/aca

Determination of thiomersal by flow injection coupled with microwave-assisted photochemical online oxidative decomposition of organic mercury and cold vapor atomic fluorescence spectroscopy Beatrice Campanella a , Massimo Onor a , Marco Carlo Mascherpa a , Alessandro D’Ulivo a , Carlo Ferrari b , Emilia Bramanti a,∗ a National Research Council of Italy, C.N.R., Istituto di Chimica dei Composti Organo Metallici-ICCOM-UOS Pisa, Area di Ricerca, Via G. Moruzzi 1, 56124 Pisa, Italy b National Research Council of Italy, C.N.R., Istituto Nazionale di Ottica, INO–UOS Pisa, Area di Ricerca, Via G. Moruzzi 1, 56124 Pisa, Italy

h i g h l i g h t s

g r a p h i c a l

a b s t r a c t

• Thiomersal was determined on line using FI-MW/UV-CVGAFS.

• MW/UV allows a “green” on line oxidation of organic mercury to HgII .

• Each measure requires less than 5 min with a LOD of 3 ng mL−1 (as mercury). • Hg concentration in commercial ophthalmic solutions ranges between 7.5 and 59.0 ␮g mL−1 .

a r t i c l e

i n f o

Article history: Received 5 August 2013 Received in revised form 7 October 2013 Accepted 10 October 2013 Available online 18 October 2013 Keywords: Thiomersal Atomic fluorescence spectrometry Flow injection analysis Organic mercury-compounds Mercury oxidation

a b s t r a c t We developed a flow injection (FI) method for the determination of thiomersal (sodium ethylmercurithiosalicylate, C9 H9 HgNaO2 S) based on the UV/microwave (MW) photochemical, online oxidation of organic mercury, followed by cold vapor generation atomic fluorescence spectrometry (CVG-AFS) detection. Thiomersal was quantitatively converted in the MW/UV process to Hg(II), with a yield of 97 ± 3%. This reaction was followed by the reduction of Hg(II) to Hg(0) performed in a knotted reaction coil with NaBH4 solution, and AFS detection in an Ar/H2 miniaturized flame. The method was linear in the 0.01–2 ␮g mL−1 range, with a LOD of 0.003 ␮g mL−1 . This method has been applied to the determination of thiomersal in ophthalmic solutions, with recoveries ranging between 97% and 101%. We found a mercury concentration in commercial ophthalmic solutions ranging between 7.5 and 59.0 ␮g mL−1 . © 2013 Elsevier B.V. All rights reserved.

1. Introduction Thiomersal (sodium ethylmercury thiosalicylate, C9 H9 HgNaO2 S, Fig. 1A [1]), also known as merthiolate or thimerosal is an organomercury compound widely used as antimicrobial agent and preservative in a variety of products including topical antiseptic solutions, cosmetics, cleaning solutions for contact lenses,

∗ Corresponding author. Tel.: +39 50 3152293; fax: +39 50 315 2555. E-mail address: [email protected] (E. Bramanti). 0003-2670/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.aca.2013.10.018

vaccines and other injectable biological products to prevent bacteriological contamination for up to 60 years [2]. The declared, unknown concentration of thiomersal used as antimicrobial agent in several commercial products varies from 0.01% in IM–IV–SC injection and topic preparations to 0.001–0.15% in ophthalmic solutions, suspensions and preparations [1]. Thiomersal is very toxic by inhalation, ingestion, and contact with skin (EC hazard symbol T+), with a hazard for cumulative effects, and it is toxic for aquatic organisms (EC hazard symbol N). Thiomersal contains 49.6% (w/w) mercury. In the body it is metabolized and degraded to thiosalicylic acid (thiosalicylate at

B. Campanella et al. / Analytica Chimica Acta 804 (2013) 66–69

67

Fig. 1. Chemical structure of thiomersal (A) and its hydrolysis ((B) and (C)) [3].

physiological pH, Fig. 1B) and ethylmercury (C2 H5 Hg+ , Fig. 1C) [3,4]. The disposition patterns of mercury after thiomersal injection are similar to those found after exposure to equivalent doses of ethylmercury chloride: the central nervous system and the kidneys are targets, the lack of motor coordination being a common sign. Similar signs and symptoms have been observed in accidental human poisonings [1]. Few studies of the toxicity of thiomersal in humans have been performed [5–7]. The detection of thiomersal at trace level in pharmaceutical products is of crucial importance. In the past various analytical techniques including high-performance liquid chromatography (HPLC) with electrochemical detection, enzymatic amperometry, colorimetry, atomic absorption spectrometry (AAS) and atomic fluorescence spectrometry (AFS) have been developed to detect thiomersal [2,4]. Table 1 summarizes the methods employed for the analytical determination of thiomersal and their limit of detection and dynamic linear range [2,4,8–12]. In this work we propose a flow injection (FI) method for the determination of thiomersal, based on a fully integrated online UV/microwave (MW) photochemical reactor for the digestion of organic mercury followed by chemical vapor generation atomic fluorescence spectrometry (CVGAFS) detection (FI-UV/MWCVGAFS). Although this apparatus has been previously employed for the determination of methylmercury, ethylmercury, parahydroxymercurybenzoate (pHMB) and their complexes with thiols [13,14], thiomersal determination using this system has never been optimized before. 2. Materials and methods 2.1. Chemicals Stock solution of 1000 ± 5 ␮g mL−1 of inorganic HgII in the form of Hg(NO3 )2 was purchased from Merck Laboratory Supplies (Poole, Dorset, UK). Thiomersal (CAS no. 54-64-8, 2(C2 H5 HgS)C6 H4 CO2 Na) was purchased from Sigma (Sigma-Aldrich, Milan, Italy). Sodium chloride (CAS no. 7647-14-5, NaCl) was purchased from J. T. Baker. Water deionized with a Milli-Q system (Millipore, Bedford, MA, U.S.A.) was used throughout. A stock solution of thiomersal (50 ␮g mL−1 ) was prepared by dissolving the salt in water, stored at 4 ◦ C and diluted freshly just before use. Working solutions used for calibration were prepared by diluting the stock solution just before use in 0.15 M NaCl at

Fig. 2. Schematic diagram of the FIA-MW/UV-CVG-AFS system. MW/UV generator system: MS, 100 W, 2450 MHz, solid-state source; BC, bidirectional coupler; FP, forward power probe-head; RP, reflected power probe-head; CC, flexible coaxial cable. MW/UV quartz photoreactor: ML, bulb of the coaxial electrodeless UV lamp; T, quartz coaxial tubing; MA, microwave antenna; F, point of maximum MW power emission, or feed point; G/L separator, gas/liquid separator; PP, peristaltic pump.

0.01–5 ␮g mL−1 concentration levels. All working solutions were stored at 4 ◦ C and in the dark when not in use. Solutions of tetrahydroborate (THB, NaBH4 ) more concentrated than 0.27 M (1% m/v) were prepared by dissolving the solid reagent (Merck & Co., Inc., N.J. USA, reagent for AAS, minimum assay >96%) into 0.3% m/v NaOH solution. The solutions were microfiltered through a 0.45 ␮m membrane and stored in a refrigerator. Diluted solutions of NaBH4 (0.05 mol L−1 ) were prepared daily by appropriate dilution of the stock solutions, the total NaOH concentration being kept at 0.3% (m/v). HCl diluted solution was prepared from 37% (m/m) HCl (Carlo Erba, Rodano, Milan, Italy). Six eye drop formulations were analyzed for thiomersal determination: one was provided as thiomersal free and five were provided with unknown thiomersal content. None of the analytical techniques used in this work required sample pre-treatment. The total concentration of mercury in thiomersal stock solution and in all real samples (expressed as mercury content) was determined usign a Milestone DMA80 (FKV s.r.l., Milan). 2.2. Instrumentation An HPLC gradient pump (P4000, ThermoQuest) equipped with a Rheodyne 7125 injector (Rheodyne, Cotati, CA, U.S.A.) and a 50 ␮L injection loop was used. The pump flow rate was 1.0 mL min−1 . Samples were eluted with 0.15 M NaCl. The effluent was mixed in a T with 2 M HCl (4 mL min−1 ) and passed through a Teflon coil coiled around a quartz ML bulb (Fig. 2), immersed in a flowing water bath

Table 1 Summary of recent methods for the analytical determination of thiomersal. Refs.

Matrix

Sample pretreatment

Detection mode

LOD (␮g mL−1 )

Linearity

[3] [8]

Vaccines Standard solutions

1:10 dilution in 10% HCOOH No pretreatment

0.0006 0.0007

0.0005–0.01 ␮g mL−1 Linear up to 0.2 ␮g mL−1

[11] [9] [12] [2]

Spiked natural waters Vaccines Pharmaceutical samples Vaccines

No pretreatment Acidification with HNO3 No pretreatment No pretreatment

0.0003 0.025 0.02 0.0003

Linear up to 0.1 ␮g mL−1 50–50,000 ng mL−1 0.1–1 ␮g mL−1 0.5–200 ␮g L−1

[10]

Ophthalmic solutions

1:2 dilution with H2 O2

Photo vapor generation (PVG)—ICP OES Solution cathode glow discharge—ICP atomic emission spectrometry (AES) FI-PVG-AAS Electrolyte cathode glow discharge—AES Graphite furnace—AAS Dielectric barrier discharge-plasma induced vaporization—AFS UV/H2 O2 -sono induced-CVG-AAS

0.04

Not specified

68

B. Campanella et al. / Analytica Chimica Acta 804 (2013) 66–69

Table 2 Summary of MW conditions and CVGAFS method.

Table 4 Limits of repeatability obtained for thiomersal at 0.02–0.2–2 ␮g mL−1 for ˛ = 0.05 and t1−˛/2 ,∞ = 1.96.

MW/reactor Optimize power Volume of post column reaction coil Length of post column reaction coil i.d of post column reaction coil Temperature of post column reaction coil Reaction time

30 W 3 mL 3.82 mm 1 mm 17 ± 1 ◦ C 50 s

Reagents 2 mol L−1 4 mL min−1 0.005 mol L−1 2.8 mL min−1

HCl concentration HCl flow rate NaBH4 concentration NaBH4 flow rate NDAFS detector

170 mL min−1 150/90 mL min−1 500 Hz 1s 3 mm above the burner top

Ar (stripping) Ar/H2 (flame) Modulation frequency for EDL RC time constant Observation height

kept at 17 ± 1 ◦ C. Further details of the MW/UV combined reactor with CVG-AFS detection system used for all the measurements has been previously described in details [13–15]. The conditions of MW/UV system and CVGAFS detection method are summarized in Table 2. The output data from the lock-in amplifier were collected with a personal computer equipped with a data acquisition card (DAC, National Instruments, Austin, TX) and its acquisition software (LabVIEW version 6, National Instruments). Experimental results obtained by FIA-MW/UV-CVGAFS were compared with that obtained with the DMA80. The heating cycle (Table 3) is carried out in a constant oxygen flow (200 mL min−1 ). The absorbance measurements were carried out at 253.7 nm. The total processing time is approximately 5 min per sample. 3. Results 3.1. Yield of the oxidative decomposition of thiomersal The efficiency of the oxidative decomposition for thiomersal (1 ␮g mL−1 injected in 0.15 mol L−1 NaCl solution flowing at 1 mL min−1 , N = 3 injections) was calculated on the basis of the FI peak area with respect to the FI peak area of an equimolar concentration of Hg(II) analyzed in the same operating conditions and it was 97 ± 3%. This result is not so obvious because mercurial compounds may have different on line oxidation conditions. Methyland ethylmercury do not need on line MW/UV oxidation [13], while thiomersal, as well as pHMB [14] require MW/UV system for their detection and determination. 3.2. Calibration and figures of merit Calibration was evaluated by analyzing three replicates of thiomersal solutions in the 0.01–5 ␮g mL−1 range of Hg. The sixpoint calibration curve, determined by a least squares regression Table 3 Programmed heating cycle used in DMA experiments.

Thiomersal concentration (␮g mL−1 )

R

RSD (%)

0.02 0.2 2

0.003 0.004 0.08

5 0.9 2

algorithm, was linear over the range of 0.01–2 ␮g mL−1 . The fitting equation was y = 0.257(SD = 0.003)104 x + 0.33(SD = 0.02)102 with a mean correlation coefficient of 0.9998. The limit of detection (LOD) and the limit of quantitation (LOQ) were calculated as 3sb /slope and 10sb /slope, when sb is the standard deviation of 10 blank measurements, and are 0.003 and 0.009 ␮g mL−1 , respectively. In the same figure the calibration curve of Hg(II) was reported as comparison (N = 3 replicates in the 0.01–2 ␮g mL−1 Hg(II) concentration range). The six-point calibration curve, determined by a least squares regression algorithm, was linear over the range of 0.01–2 ␮g mL−1 . The fitting equation was y = 0.259(SD = 0.004)104 x + 0.36(SD = 0.06)102 with a mean correlation coefficient of 0.9992. The two slopes were compared with t-Test: at the confidence level chosen (˛ = 0.05, N = 3) the two slopes did not differ significantly. 3.2.1. Repeatability The limit of repeatability was determined by the equation 1: √ r = r = t1− ˛ , 2r 2

where t1−˛/2, is the t-Student (two tail) for 0.95 level of confidence,  = (N − 1) with N = degrees of freedom and  r the standard deviation of 10 measurements of thiomersal solutions at 0.02, 0.2 and 2 ␮g mL−1 concentration level in conditions of repeatability. Table 4 shows the limits of repeatability obtained. The repeatability limit r is the maximum value, predictable at a certain level of confidence, of the absolute difference between two results obtained under repeatability conditions. 3.2.2. Recovery Recovery tests were performed analyzing the only sample declared as thiomersal free before and after spike with known concentrations of thiomersal. Three replicates at low (0.05 ␮g mL−1 ) and high (1 ␮g mL−1 ) concentration levels were prepared. The recoveries at 0.05 and 1 ␮g mL−1 were 97.4 ± 0.2% and 98.6 ± 0.3%, respectively. 3.3. Application of the method to real samples The method was used to analyze six ophthalmic solutions in triplicates. The matrix effects were evaluated by comparing via statistical t-test at 95% of confidence the slope of the external calibration with the slope of the analyte standard addition curves for the five samples not thiomersal-free (Table 5). For all samples at the level of confidence chosen, the slopes of external and internal curves did not differ significantly. Thus, the matrices of samples analyzed did not affect the analytical results. Table 5 Fitting parameters of the internal calibration curves obtained in 5 ophthalmic real samples.

Time (s)

Temperature (◦ C)

Sample

Slope (SD) × 104

Intercept (SD) × 103

R2

0 10 70 160 340

100 200 200 650 650

Artificial tear n.2 Antihistamine eye drops n. 1 Antihistamine eye drops n.2 Antihistamine eye drops n.3 Veterinarian eye drops

0.25 ± 0.01 0.249 ± 0.003 0.250 ± 0.006 0.249 ± 0.002 0.244 ± 0.001

0.28 ± 0.05 0.35 ± 0.01 0.20 ± 0.02 0.24 ± 0.01 0.185 ± 0.005

0.9940 0.9996 0.9986 0.9998 0.9999

B. Campanella et al. / Analytica Chimica Acta 804 (2013) 66–69

69

Table 6 Analytical results for the determination of thiomersal in ophthalmic solutions by FI-MW/UV-CVGAFS and DMA methods. Commercial sample

Artificial tear no. 1 Artificial tear no. 2 Antihistamine eye drops no. 1 Antihistamine eye drops no. 2 Antihistamine eye drops no. 3 Veterinarian eye drops a

FI-MW/UV-CVGAFS method (␮g mL−1 ) (RSD %) External calibration

Standard additions