Determination of Acidic Pharmaceuticals in Municipal Wastewater by Using Solid-Phase Extraction Followed by Gas Chromatography-Mass Spectrometry * * 1

Pharmaceuticals and their transformation products have been more frequently detected in aquatic environments over recent years. The presence of pharmaceuticals in environment systems for the rst time was observed in 1970s in United States, when clo bric acid, salicylic acid and ca eine were detected in sewage e uent and in ground water (clo bric acid) [9]. Yet, it was only in the second half of the 1990s that scienti c interest has grown signi cantly as a consequence of huge developments in analytical techniques. Moreover, the presence of pharmaceuticals in the environment has raised awareness of the risk that pharmaceuticals may pose on aquatic organisms due to their biological activity and possible long-term exposure [23]. Drugs, after oral administration are excreted from the body either as metabolites or as the parent compounds and through the domestic sewage system they reach municipal wastewater treatment plant. Recent studies showed, that conventional wastewater treatment procedures consisting of physical, chemical and biological processes are insu cient to eliminate these micro contaminants completely, thus e uents have been recognized as a major source of pharmaceuticals entering the aquatic environment [3, 10, 11, 26]. Residues of pharmaceuticals are being detected mainly in Western Europe and USA water environment at the level of ppt to ppb in sewage e uents [25, 26], in surface water [2, 30] and even in drinking water sources [1, 8]. So far, there has been just few publications on their occurrence in Polish water bodies. In 2005, the presence of 23 pharmaceuticals from di erent therapeutic classes in raw wastewater from Zabrze up to 27 g/L was revealed [7]. In the same year, scientists from Gdansk University of Technology determined the presence of non-steroidal anti-in ammatory drugs (NSAIDs) in surface water up to ~0.5 g/L [5]. Kasprzyk-Hordern et al. [13, 14] published results of determination of numerous


Introduction
Pharmaceuticals and their transformation products have been more frequently detected in aquatic environments over recent years.The presence of pharmaceuticals in environment systems for the rst time was observed in 1970s in United States, when clo bric acid, salicylic acid and ca eine were detected in sewage e uent and in ground water (clo bric acid) [9].Yet, it was only in the second half of the 1990s that scienti c interest has grown signi cantly as a consequence of huge developments in analytical techniques.Moreover, the presence of pharmaceuticals in the environment has raised awareness of the risk that pharmaceuticals may pose on aquatic organisms due to their biological activity and possible long-term exposure [23].
Drugs, after oral administration are excreted from the body either as metabolites or as the parent compounds and through the domestic sewage system they reach municipal wastewater treatment plant.Recent studies showed, that conventional wastewater treatment procedures consisting of physical, chemical and biological processes are insu cient to eliminate these micro contaminants completely, thus e uents have been recognized as a major source of pharmaceuticals entering the aquatic environment [3,10,11,26].Residues of pharmaceuticals are being detected mainly in Western Europe and USA water environment at the level of ppt to ppb in sewage e uents [25,26], in surface water [2,30] and even in drinking water sources [1,8].So far, there has been just few publications on their occurrence in Polish water bodies.In 2005, the presence of 23 pharmaceuticals from di erent therapeutic classes in raw wastewater from Zabrze up to 27 g/L was revealed [7].In the same year, scientists from Gdansk University of Technology determined the presence of non-steroidal anti-in ammatory drugs (NSAIDs) in surface water up to ~0.5 g/L [5].Kasprzyk-Hordern et al. [13,14] published results of determination of numerous group of pharmaceuticals in the Warta river, in uent and e uent from wastewater treatment plants.The most studied pharmaceuticals were found at a concentration of up to 2.1 g/L, 10 g/L and 2.8 g/L, respectively.
In general, residues of pharmaceuticals are continuously delivered to environment in trace levels but the problem may grow due to the increasing consumption and world production of pharmaceuticals.Global pharmaceutical market was estimated to reach $ 808 billion in 2009 with 5-7% growth rate in the last years.According to the latest reports (PMR Publications) Polish pharmaceutical market is the sixth one in the Europe with the sales of nearly PLN 30 billion in 2010 [17].Around 34% of pharmacy sales are generated by OTC (over the counter) products whereas 26% of OTC market comes from analgesic drugs [16].Because of general accessibility and overusing, OTC drugs are most often detected in aquatic systems in arelatively high concentration level [6].
So far, there is no systematic monitoring of pharmaceutical residues in waters and no standardized analytical methods available.For these reasons the aim of this work was to present sensitive analytical method for detection of selected nonsteroidal anti-in ammatory drugs in wastewater by solid phase extraction followed by gas chromatography mass-spectrometry.

Materials and Chemicals
In this study ve selected pharmaceuticals or chemicals associated with pharmaceuticals were selected for analysis.The selection included several nonsteroidal anti-in ammatory drugs -naproxen, ketoprofen and diclofenac.Bisphenol A, a chemical used to manufacture polycarbonate plastics, suspected to be also an endocrine disruptor with potential to mimic the role of the natural hormones, was also included.Moreover, triclosan, an antibacterial and antifungal agent, which with chlorine can form carcinogenic chloroform, was also taken into consideration.The structures of the pharmaceuticals considered in this study are shown in gure 1.
Standards were of the following origins: naproxen and ketoprofen from Sigma-Aldrich (St. Louis, MO, USA), diclofenac sodium and bisphenol A from Dr. Ehrenstorfer (Augsburg, Germany) and triclosan from Merck (Darmstadt, Germany).1-hydroxypiren used as internal standard was obtained from Sigma-Aldrich (St. Louis, MO, USA).Stock solutions of all compounds were prepared in ethyl acetate.Silylating agent, N-Methyl-N-(trimethylsilyl)tri uoroacetamide (MSTFA) was from Sigma-Aldrich (St. Louis, MO, USA).
All solvents were of analytical grade of high residue purity.Methanol, n-hexane and hydrochloric acid were purchased from POCH (Gliwice, Poland).Ethyl acetate was purchased from Merck (Darmstadt, Germany).Ultrapure water was obtained from an HLP5 system (Hydrolab, Poland).Glass-ber lter papers GF-4 and GF-1 (Ø = 45 mm) used in samples pre-treatment were purchased from Macherey-Nagel (Düren, Germany).The cartridges used for solid phase extraction were Oasis HLB 60mg from Waters (Ireland).

Sample Collection and Pre-Preparation
Wastewater samples were collected in January 2011 from Plaszow Wastewater Treatment Plant in Krakow.This is a conventional sewage plant consisting of pre--treatment followed by a secondary activated sludge treatment.The design capacity of the plant is 656 thousand m 3 per day but it threats approximately 165 thousand m 3 of urban wastewater per day, what is over 70% of the total volume of the city's wastewater.E uents are discharged to the Drwina river and then to the Vistula river.
Both, in uent and e uent samples were taken early in the morning and collected in 5L plastic containers.Samples were stored at 4°C until further processing.Next, the samples were ltered with glass ber lters and extraction was performed during the day of the sampling.

Solid-Phase Extraction
Because of low concentration of pharmaceuticals in environmental water samples pre-concentration steps are necessary before analysis.
Pre-concentration and clean-up steps were accomplished by solid-phase extraction using Oasis HLB extraction cartridges.The SPE procedure was based mostly on our own previous investigations and optimization experience reported in the literature [19,21].Oasis HLB is a universal polymeric sorbent for acidic, neutral and basic compounds.In order to enhance recoveries, analytes (all weak acids) were conducted to non-ionic form through adjusting pH of wastewater samples to 2.0 with 2M hydrochloric acid.Extraction was carried out on a SPE-12G (JT Baker) vacuum manifold system.
Before sample loading, SPE cartridges were conditioned with 3 mL of ethyl acetate, 3 mL methanol and 3 mL deionized water at adjusted pH 2.0.250 mL of in uent, and 500 mL of e uent were then transferred to the SPE cartridges at ow rate up to 10 mL/min.After the extraction, cartridges were washed with 3 mL of methanol:water (5:95, v/v) and 1 mL of n-hexane and then dried for 15 minutes under a vacuum.The elution was performed with 3 mL of ethyl acetate at ow rate of about 1 mL/min.The extracts were evaporated to dryness with a Six Port Mini-Vap Evaporator from Supelco (Bellefonte, PA, USA) at 45°C in a water bath under a gentle ow of argon.Finally, dry residues were dissolved in 0.1 mL of 5 g/mL of 1-hydroxypyrene in ethyl acetate.1-hydroxypyrene was used to compensate possible loss of the sample during injection.
For recovery studies, ultrapure water was spiked with each of target compounds before the SPE process and before the derivatization step at a concentration of 0.5 g/L and 1.0 g/L.

Derivatization for GC/MS Detection
To increase the volatility and stability silylation of analytes with the N-methy-N-(trimethylsilyl) tri uoroacetoamide (MSTFA) was performed.Silylation is the prevalent type of derivatization used in GC/MS analysis of hydroxyl compounds.Therefore, 50 L of concentrated wastewater was mixed with 15 L of MSTFA [24].Samples were heated at 65° for 35 minutes using digital dry bath thermo-block Labnet (NJ, USA) and subsequently analyzed in a GC/MS system.

Calibration Method
Standard addition calibration was used for quanti cation due to existing matrix e ects.1-hydroxypyrene was used as an internal standard to control sample volume during manual injection.
Repeatability of parallel injection expressed as a variation coe cient (n = 3) ranged from 0.2% for naproxen to 9% for diclofenac, which seems to be satisfactory for wastewater matrix.Limits of detection (LOD) and limits of quanti cation (LOQ) for the presented method were determined experimentally, both in in uent and efuent samples, according to standard procedures, by measuring the signal to noise ratio (S/N).LOD was estimated as three times noise value measured around the analyte retention time (±0.5 min) and LOQ was estimated as three times of the LOD value.For calibration, ve standard mixtures containing internal standard and test compounds in concentration within 0.5-15.0g/mL were analyzed.The expected model for the data is linear regression y = bx + a, where a is the theoretical y-intercept and b is the theoretical slope (x -analyte concentration in the sample, g/mL, y -peak area ratio: analyte to internal standard).Before accepting linear model, the ing of a linear regression was considered using R (software for statistical computing and graphics), by comparing linear (y ~ x) and quadratic model (y ~ x + I(x 2 )) -Mandel test [4].The null hypothesis in this case is H 0 : I(x 2 ) = 0, that quadratic model does not t.Additional criteria is AIC (Akaike's Information Criterion) -a measure of the relative goodness of t of a statistical model (AIC = 2k -2ln(L), where k is the number of parameters in the statistical model and L is the value of the likelihood for the estimated model).The model to choose is the one for which AIC is least (Tab.1).
After choosing a linear model as a be er option for the evaluated data, linear regression analysis (least squares method) was held to predict the unknown concentrations of tested compounds in wastewater samples (Tab.1).In order to ensure the accuracy of estimates, the signi cance of model regression was checked by the Student's t-test [15].Hypothesis tests were conducted to show statistically signi cant di erence between slope b and zero, and no signi cant di erence between theoretical y-intercept a and experimental |a 0 | (y value at x = 0).In the end, the strong linear relationships among variables x and y was statistically con rmed.

GC/MS Analysis
The chromatographic separation and analysis were performed with the HP 5890 series II gas chromatograph equipped with a split/splitless injector, HP -5 MS capillary column (30 m × 0.25 mm i.d., 0.25 m lm thickness, 5% phenyl 95% dimethylpolysiloxane stationary phase) and HP 5971 quadrupole mass selective detector (Agilent Technologies, USA).
The injection port temperature was set to 250°C and transfer line temperature was 260°C.The GC oven temperature started at 70°C and was held at 70°C for 2 Then, the temperature was increased to 280°C at a heating rate 15°C/min and held at 280°C for 7 min (23 min.total run time).The samples in 1 L aliquots were manually injected in splitless mode and then carried to the column by helium gas in constant ow rate of 1 mL/min.The electron impact ionization potential was 70 eV.Identi cation and quanti cation of analytes were performed in selected ion monitoring (SIM) mode on the basis of time retention and characteristic fragment ions (m/z, Tab. 2) of target compounds.

Results
A statistical analysis con rmed linearity of data in whole tested concentration range giving R 2 values from 0.9913 to 0.9993.For diclofenac linearity is limited to the range of 0.5-5.0g/mL with R 2 0.9766 for e uent and 0.9925 for in uent.For each pharmaceutical compound, the statistically signi cance of slope b con rmed the obvious relationship between the concentration and chromatographic signal, both for in uent and e uent.Theoretical intercept a in every case is statistically insigni cant meaning that our quantitative method works properly.
Results of the determination of pharmaceuticals in wastewater samples with validation parameters are shown in table 3. Analyte recoveries are considered to be satisfactory, as they ranged from 61% for triclosan to 78% for diclofenac.In treated wastewater LOD were varied from 0.002 g/L to 0.016 g/L whereas in raw wastewater LOD were between 0.006 g/L to 0.121 g/L.
The data show that all of the tested pharmaceutical substances enter the wastewater treatment plant at trace concentrations ranging from 0.166 g/L to 4.0 g/L.After all the stages of treatment, reduction in the concentrations of naproxen, ketoprofen and bisphenol A appear to be signi cant whereas diclofenac was detected in e uent in a similar amount as in uent.The results are comparable to those of previous studies conducted at A. Mickiewicz University in Pozna [13].Similarly, naproxen and ketoprofen were determined in raw wastewater at the highest concentration level (7-10 g/L), but for these compounds conventional treatment processes were observed to be e cient (signi cantly lower concentration in e uent comparing to in uent), while diclofenac was found to be the most resistant towards the treatment (concentration in in uent is comparable to e uent).Low susceptibility of diclofenac to the removal mechanisms was also revealed by other authors [12,20,25].Anti-bacterial triclosan was detected both in in uent and e uent samples at concentration 0.166 g/L and 0.093 g/L, respectively and these results are comparable to those reported by Weigel and Thomas [27,29].According to the literature data, conventional treatment plants remove pharmaceuticals compounds mainly during biological treatment with activated sludge [18,22,28].Here, biodegradation or sorption to sludge is the probable mechanism of elimination [31].Removal of acidic compounds is achieved mostly by the biodegradation process.More lipophilic substances (triclosan) may be additionally eliminated with fat separation, during primary treatment and adsorbed onto solid surfaces (sludge) [31].Low removal rates of diclofenac or even higher contents in e uent might be a ributed to the hydrolysis of diclofenac glucuronides under treatment procedures and/or its desorption from particles [20].
The performance of the wastewater treatment plant towards the tested pharmaceuticals is not fully representative, because wastewater samples were collected only at one time (early morning).Nevertheless, the results allow us to estimate approximate the pharmaceutical contaminants load in wastewater generated by an urban community and an average amount of pharmaceutical discharge daily into surface water.It should be noted that the main aim of this work was to present an e ective and sensitive method enabling us to determine pharmaceutical trace compounds in a complex matrix.Obviously, the preliminary results should be supported with extended studies.

Conclusion
The paper presents a versatile method useful for the determination of selected emerging contaminants in aquatic environment.Such a tool is highly required since the presence of residual pharmaceutics in the environment is becoming more and more an important ecological issue.A version of the analytical method described in this study, based on SPE and GC/MS is suitable for the analysis of acidic pharmaceuticals residues in wastewaters samples.An analysis of trace compounds in wastewater is di cult due to complexity of the matrix and high organic content.A method has been successfully applied to the analysis of several popular pharmaceutical compounds in wastewater samples even at the low g/L level and provided satisfactory recoveries and precision.

FF
both in uent and e uent -H 0 cannot be rejected meaning that the linear model is suitable.AIC values con rm that the linear model ts be er than the quadratic model both in uent and e uent -the di erence between b and 0 R > r cr -both in uent and e uent -there is statistically signi cant dependence between variables x and y t < t cr -both in uent and e uent -there is no statistically signi cant di erence between a and |a 0 |

Table 1 .
Standard addition calibration curves of tested compounds with statistical analysis (equation format: y = bx + a, n = 6 -number of calibration points, f -degrees of freedom, -signi cance level, F -Snedecor's test value, t -Student's test value, AIC lin -measure of relative goodness of t of linear model, AIC quad -measure of relative goodness of t of quadratic model) cr F F -both in uent and e uent -H 0 cannot be rejected meaning that the linear model is suitable.AIC values con rm that the linear model ts be er than the quadratic model cr F F -both in uent and e uent -H 0 cannot be rejected meaning that the linear model is suitable.AIC values con rm that the linear model ts be er than the quadratic model cr -both in uent and e uent -there is statistically signi cant dependence between variables x and yTable 1 cont.cr F F -both in uent and e uent H 0 -cannot be rejected meaning that the linear model is suitable.AIC values con rm the linear model ts be er than the quadratic model cr F F -both in uent and e uent H 0 -cannot be rejected meaning that the linear model is suitable.AIC values con rm the linear model ts be er than the quadratic model Calibration curves Calibration curves Table 1 cont.cr -both in uent and e uent -there is statistically signi cant dependence between variables x and y Table 1 cont.

Table 2 .
Retention time and m/z ratio of characteristic ions of target compounds

Table 3 .
Concentration of investigated pharmaceuticals in wastewater * ± value -standard error of predicted x-value (converted to g/L) for y = 0 in the regression.