Determination of chiral pharmaceuticals and illicit drugs in wastewater and sludge using microwave assisted extraction, solid-phase extraction and chiral liquid chromatography coupled with tandem mass spectrometry

This is the first method for enantiomeric profiling of chiral drugs (cPACs) in solid & liquid matrices. Analysis of both liquid & solid matrix is critical to do mass balance of cPACs in wastewater. Chiral PACs are often non-racemic in wastewater matrices. Enantiomeric composition of PACs differs in liquid and solid wastewater matrices. Not analysing the solid fraction of wastewater may lead to over-estimation of cPAC’s removal rates. A R T I C L E I N F O

Determination of chiral pharmaceuticals and illicit drugs in wastewater and sludge using microwave assisted extraction, solid-phase extraction and chiral liquid chromatography coupled with tandem mass spectrometry Sian E. Evans, Paul Davies, Anneke Lubben, Barbara Kasprzyk-Hordern * University of Bath, Department of Chemistry, Faculty of Science, Bath BA2 7AY, UK

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
This is the first method for enantiomeric profiling of chiral drugs (cPACs) in solid & liquid matrices. Analysis of both liquid & solid matrix is critical to do mass balance of cPACs in wastewater. Chiral PACs are often non-racemic in wastewater matrices. Enantiomeric composition of PACs differs in liquid and solid wastewater matrices. Not analysing the solid fraction of wastewater may lead to over-estimation of cPAC's removal rates.

Introduction
Developments in pharmacology and an ageing population have led to a steady increase in the prescription of pharmaceuticals worldwide. Once metabolised and excreted pharmacologically active compounds (PACs) and their metabolites enter the sewage system and if not removed or degraded, they are then pumped into effluent receiving water bodies and/or are incinerated or spread on agricultural land if adsorbed on to solids.
Pharmaceuticals and illicit drugs are routinely identified in wastewater. The data produced is often used to inform risk assessments [1,2] and develop and test prediction models [3]. However, these concentrate on pharmacologically active compounds (PACs) present in the aqueous matrix and they do not consider the fraction of PACs adsorbed on to the sediments and suspended solids which may still be bioavailable to organisms and/ or may desorb in the future, possibly due to a change in matrix or environmental conditions. Methods have been developed recently in order to tackle this knowledge gap and to verify concentration levels of PACs in sediments [4][5][6][7][8], sludge [7], colloids [9], settable particulate matter [10] and suspended matter [6,11] as well as other solid matrices such as soil [12]. Corresponding models have also emerged to take into account this additional data [13].
Many PACs, prescribed, over the counter and illicit, are chiral. Enantiomers often have widely disparate pharmaco-dynamics and -kinetics in humans [14]. This may result in the excreted parent drug and/or metabolites having a different enantiomeric composition to the ingested drugs [14]. The fate of chiral PACs (cPACs) within the environment can be affected by microorganisms and this process has been shown to be, at least in part, stereoselective. The enantiomeric composition of cPACs has also been demonstrated to be significant in the toxicity of several environmentally relevant species. S(+)-fluoxetine and S(À)-atenolol were found to significantly inhibit the growth of Tetrahymena thermophila, a freshwater protozoan, more than R (À)-fluoxetine and R(+)-atenolol. Whilst R(+)-atenolol increased mortality rates in Pseudokirchneriella subcapitata, an algae, significantly more than S(À)-atenolol [16]. Stanley et al. [17,18] have also demonstrated enantiomer selective toxicity in fluoxetine and propranolol towards Pimephales promelas and fathead minnow, with S(+)-fluoxetine and S(À)-propranolol being the more toxic enantiomers [17,18]. All three of these studies also exposed Daphnia magna to the enantiomers of atenolol, fluoxetine and propranolol but found no enantioselective toxicity towards this species.
Consequently, in more recent years, attention has been placed on the verification of enantiomeric composition of cPACs within wastewater and the effluent receiving waters [19][20][21][22][23][24][25][26][27][28][29][30][31][32][33][34]. However, these studies all concern themselves with the liquid fraction of the environment. Only one study has extracted and quantified antifungals from wastewater solid material [35]. This gap in the methodology may result in an under estimation of the effects of cPACs within an aquatic environment, if they adsorb to solid material. It may also be the case that bacterial populations associated with the solid material, and the process under which it is digested at wastewater treatment plants (WWTPs), may alter the enantiomeric fraction compared to both what it was on solid material in the influent as well as what is currently being recorded in the liquid fraction. Therefore, it is of the greatest importance to develop analytical capability to undertake stereoselective analysis of chiral compounds in solid matrices.
This study aims to develop a novel comprehensive analytical method for multi-residue identification and quantification of 23 cPACs (including several beta-blockers, antidepressants and amphetamines, Table S1) at enantiomeric level in both the liquid and solid fractions of wastewater. To the authors knowledge, such a method has never been previously reported. It is anticipated that this technique could then be validated and adapted for other solid matrices such as biological matrices and soils where these PACs may also be found due to, for example, the disposal of digested sludge on agricultural land. The targeted analytes have been identified based on high prescription/consumption rates within the UK, widespread environmental occurrence and biological effects, their chirality and the ability to resolve enantiomers via chiral HPLC.

Sample collection and preparation
Digested cake and grab wastewater samples were collected in a single visit to a UK wastewater treatment plant, whose characteristics are shown in Table S2. They were transported back to the laboratory in cool boxes packed with ice blocks and frozen immediately upon arrival. All influent, effluent and digested sludge samples were collected simultaneously and therefore should not be interpreted as representing the same body of water. The cake was frozen, then dry frozen (Scanvac dryer, Scientific Laboratory Supplies, UK) and stored at À20 C. Wastewater was also stored at À20 C.

Microwave assisted extraction (MAE) for solid matrix
Method development for MAE firstly identified conditions which had been reported as successful within the literature for related compounds, methods and matrices. Mass of sample, solvent volume and type, temperature and exposure time as well as pH were identified as crucial parameters. Table 1 summarises the iterations trialled.
The following conditions resulted in the most effective MAE recovery: 20 mL methanol:water (1:1), heated to 120 C for 30 min. Both 1 g and 3 g of freeze-dried sample were used for method validation purposes. were filtered and then spiked with SS/IS (1 mg L À1 ) before being passed through Oasis HLB cartridges, with reversed phase sorbent, conditioned with 2 mL methanol, followed by 2 mL deionised water and eluted in 4 mL methanol.
2.2.2.2. SPE for solid matrix. Method development for the solid matrix SPE included trialling 3 types of SPE adsorbents (Table 2). Initially Oasis HLB 60 mg cartridges were used in accordance with the liquid matrix methodology described above, with the sample eluent diluted to no more than 10% methanol.
MCX cartridges were also trialled using a method previously successful in non-chiral analysis of pharmaceuticals on suspended particulate matter [11], these are packed with a mixed-mode polymeric sorbent for extracting basic compounds with cationexchange groups. In brief, cartridges were primed with 2% formic acid in water, and after extraction of cPACs from wastewater they were washed with 2% formic acid in water and then again with 0.6% formic acid in water prior to elution with 7% ammonium hydroxide in methanol.
Following this, Oasis MAX cartridges were trialled, using the same methodology as the HLB cartridges described above. These are packed with a mixed-mode polymeric sorbent for extracting acidic compounds with anion-exchange groups.
In all cases, the sample (wastewater or MAE extract) was deposited on the SPE cartridges at a rate of <6 mL min À1 , eluents were evaporated to dryness with a TurboVap evaporator (Caliper, UK, 40 C, N 2 , <5 psi) and reconstituted in 0.5 mL of mobile phase.
The CBH column is relatively restrictive regarding mobile phase composition e.g. no more than 20% organic modifier is recommended and pHs between 3 and 7 are allowed [36]. However, it shows the best chiral recognition towards amphetamine-related compounds. A successful method for the resolution of amphetamines has already been developed by Kasprzyk-Hordern and Baker [20] and it was utilised in this work without any further development.
The best enantiomeric separation and sensitivity were observed in the following conditions: methanol, 4 mM ammonium acetate and 0.005% formic acid.
All analytes were identified and quantified using a Xevo TQD Triple Quadrupole Mass Spectrometer (Waters, UK), equipped with an electrospray ionisation source in positive ion mode. Nitrogen was used as the nebulising gas, supplied by a high purity nitrogen generator (Peak Scientific, UK). Argon (99.998%) was the collision gas supplied by a BOC cylinder. MassLynx 4.1 (Waters, UK) was used to control the Waters ACQUITY system and the Xevo TQD. Data processing was carried out using the TargetLynx software (Waters, UK). A valley drop method was used to integrate partially resolved peaks. The sensitivity was monitored whilst adjusting the nebulising gas flow rate to 100, 250 and 500 L h À1 as well as the source temperature between 90 and 250 C to achieve the greatest sensitivity with the chosen mobile phase, each injection was 20 mL. Optimised MS parameters were as follows: the capillary voltage set at 3.49 kV, source temperature at 150 C, desolvation gas flow at 300 L h À1 . Nitrogen was used as nebulising and desolvation gas, while argon was used as a collision gas.

Quantification and confirmation
Each compound was quantified in multiple reaction monitoring (MRM mode), using the protonated molecular ion as the precursor ion ( Table 3). The most abundant product ion was used for quantification (in most cases), whilst confirmation was carried out using the lesser (second) abundant product ion. In addition the ratio of quantifier to confirmatory ion was used according to limits set by EC guidelines [37] i.e. ratios !50% AE 20%, >20-50% AE 25%, >10-20% AE 30% and 10% AE 50%. Deuterated surrogate/internal standards (SS/IS) were used to compensate for ion suppression/enhancement, loss during intraday precision and accuracy was determined at 0.5 mg L À1 , 5 mg L À1 , 50 mg L À1 and 250 mg L À1 , by repeated injection within 24 h and across three days, respectively. Acceptable tolerance limits were placed at the EC guideline levels [37] i.e. for concentrations 10-100 mg L À1 the RSD 20%, for standards >100 mg L À1 RSD 15%, and accuracy should not exceed À20% or +10% of the standard concentration. Instrumental detection limits (IDL S/N ) and instrumental quantification limits (IQL S/N ) were calculated using the signal to noise approach i.e. IDL was the lowest concentration at which the quantification ion S/N ! 3 assuming a linear response, IQL was the lowest concentration at which the quantification ion S/N ! 10 and the confirmatory ion S/N ! 3 assuming a linear response.
Resolution was calculated using Eq. (1),R s ! 1 was deemed adequate for quantification. Enantiomeric fraction (EF) was calculated using Eq. (2), EF accuracy was considered acceptable if AE0.05 of the true value. Retention times were recorded and relative retention times were also calculated (see Eq. (3)).
where R s is the resolution of two enantiomers, t r2 is the retention time of the second eluting enantiomer, t r1 is the retention time of the first eluting enantiomer, w1 is the base width of the first eluting enantiomer, w2 is the base width of the second eluting enantiomer.
where EF is the enantiomeric fraction. When calculating relative enantiomeric fractions, E1 is the concentration of (+)-enantiomer or first eluted enantiomer, E2 is the concentration of (À)-enantiomer or second eluting enantiomer. Peak areas were also used instead of concentrations to calculate absolute enantiomer fractions (not adjusted with surrogate/standard).
where t r rel is the relative retention time, t r is the retention time recorded for the analyte, IS t r is the retention time recorded for the internal standard.

Validation of analytical method -MAE-SPE-LC-MS/MS method
The method was validated at 5 ng g À1 , 50 ng g À1 and 100 ng g À1 of target analytes (each enantiomer) for digested sludge, and 5 ng L À1 , 50 ng L À1 and 250 ng L À1 in wastewater. At each concentration triplicate samples were either spiked before MAE, before SPE or after SPE to assess the impact of these steps on analyte recovery. This was carried out on 1 g and 3 g of matrix or 50 mL of wastewater. Sludge and wastewater samples were also run with only deuterated compounds to assess concentrations of target analytes in studied matrices.
In addition, the relative ion intensities were assessed according to the EC Directive guidelines [37] i.e. for relative intensities !50% AE 20%, >20% to 50% AE 25%, >10% to 20% AE 30%, 10% AE 50%. Resolution, enantiomeric fractions and relative retention times were calculated, using Eqs. (1)-(3), for three whole drug concentrations (10 ng L À1 , 100 ng L À1 , 500 ng L À1 in the case of wastewater or 10 ng g À1 , 100 ng g À1 or 200 ng g À1 for sludge) spiked into wastewater influent, effluent and digested sludge. Method detection limits (MDLs) and method quantification limits (MQLs) were determined using Eqs. (4)- (7). The IDL S/N , IQL S/N were used to generate MDL calc and MQL calc .   (4) where S.IDL is the IDL S/N multiplied by the solvent volume used, in this case 30 mL, divided by the mass of matrix used, in this case 1 or 3 g. Av.Rec is the average absolute recovery for that compound. The concentration factor is the ratio between the initial volume (30 mL) and the final volume in mobile phase, in this case 0.5 (mL), therefore CF = 60. The discarded fraction is to compensate for taking a known volume post centrifugation; in this case 50% was discarded so the concentration factor is multiplied by 2.
This equation is set out the same as Eq. (4), however the S.IQL is the IQL S/N multiplied by the solvent volume used, in this case 30 mL, divided by the mass of matrix used, in this case 1 or 3 g.
where Av. Rec is the average absolute recovery recorded and CF is the concentration factor between the sample taken and the final volume in mobile phase, in this case 50-0.5 mL so CF = 100.
This is set out the same as Eq. (6). Analyte recovery was calculated using Eq. (8) by comparing the initial spike concentration with the measured concentration after extraction and analysis. Known concentrations of racemic standards were applied to the liquid matrices before SPE and also after SPE, whereas solid matrix samples were spiked before MAE and after MAE (before SPE) to identify the analyte recovery associated with each step of the process.
where PA SPE is the peak area of an analyte from the sample spiked before SPE, PA b is the peak area of the (unspiked) blank sample, PA c is the peak area from the quality control sample (at the same concentration of that which PA SPE was spiked at). Alternatively, concentrations were also used, instead of peak areas, to calculate relative recoveries.

Analysis of wastewater samples
Grab samples were collected from the influent, effluent and digested sludge at the WWTP. The liquid grab samples were collected in a clean container and transferred to plastic bottles for transportation, on ice, and storage, at À20 C. Digested sludge was collected in a plastic sampling bag, sealed and transported and stored under the same conditions as the liquid matrices. All the samples were collected at the same time and should not be interpreted as the same body of waste. The validated analytical procedures described in Section 3 and shown in Fig. 1, were applied to samples prepared in triplicate, which were then injected and analysed in triplicate.

Results and discussion
3.1. Sample preparation 3.1.1. Microwave assisted extraction for solid matrix MAE was demonstrated to successfully extract the chosen analytes from complex matrices such as digested sludge. In brief, increased temperature did appear to improve recovery with complex matrices from an average recovery across all the compounds analysed of 22.5% at 90 C for 30 min to 37.4% recovery at 120 C. Although interestingly initial trials using silica as a matrix found lower temperatures performed better, 120 C for 30 min resulted in 50.6% recovery with 90 C resulting in 102%.
With both matrices 150 C resulted in lower recoveries for some compounds, with an average recovery of only 51% in silica, suggesting degradation was occurring at this temperature. For complex matrices 120 C was the most successful temperature. At 120 C, as well as lower temperatures, 30 min extraction time had improved recovery as compared to 15 min, 37% and 29%, respectively.
The mass of the sample did not appear to have a great effect on the recovery. In the case of analysis of samples where lower concentrations are expected, 3 g of matrix would be suggested to improve the MDL; however, 3 g of matrix is more difficult to process. After centrifugation a semi-solid hydrophobic layer was formed between the solid matrix and the solvent, whilst the dry matrix also absorbed the solvent; a larger mass of sample increased these challenges making it difficult to decant a known volume for SPE. Therefore, 1 g was preferred for this method. Compounds with lower log P values tended to be best extracted with 1:3 methanol:water mixture, amphetamine based compounds average recovery was 101% at 1:3 methanol:water and 91% at a ratio of 1:1; however, this had a significant effect on the recovery of more hydrophobic compounds, with recoveries of the anti-depressants falling from 80% to 34%. Therefore, a 1:1 mixture was settled upon as this had limited impact on the recovery of the more hydrophilic compounds. In general it was found the less solvent used the better the recovery, however, the difference was minimal (1 g with 10 mL had an average recovery of 30% and with 30 mL 27%) and less solvent made it considerably harder to cleanly remove the known volume for SPE without disturbing the matrix and/or removing the hydrophobic layer on top of it. It was decided that 30 mL of solvent was the lowest volume which could be used consistently with both 3 g and 1 g of matrix.
The final method is as follows (Fig. 1): replicates of digested sludge (1 g each), after being spiked with SS/IS (at concentration 1 mg g À1 ), were placed into the MAE tubes with 15 mL of methanol and 15 mL deionised water and microwaved (CEM 1400 microwave, CEM, UK) between 0 W and 1200 W to allow for a controlled ramp to 120 C, this was held for 30 min before allowing to cool to room temperature. The samples were transferred to a centrifuge, the MAE tubes were rinsed out with 10 mL of deionised water, which was added to the centrifuged containers. The samples were then centrifuged (VWR, Radnor, USA) for 15 min at 4800 rpm. A fixed volume (20 mL) of solvent was decanted and diluted with 65 mL of deionised water ready for SPE.
If analysis of individual chiral drugs was to be carried out, solvents, exposure time etc. could be tailored more to specific analytes, however, a compromise was made here to ensure extraction of the widest range of analytes possible.

Solid phase extraction
3.1.2.1. SPE for MAE extracts. The chromatography was successful when Oasis HLB cartridges were used after MAE (for solid matrix), i.e. resolution of enantiomers was sufficient for quantification (see Fig. S1a), however, the backpressure recorded in the column increased with each subsequent injection, (see Fig. S2). This was not observed in the case of liquid matrices where extraction with HLB was carried out. This back pressure increase suggested matrix clean-up by SPE using Oasis HLB cartridges had not been sufficient for MAE extracts and interfering compounds were retained on the chiral LC column causing the increased backpressure.
MCX sorbents were also found to be successful in removing the interfering compounds (from MAE extracts) which had been retained in the column, and therefore no increase in backpressure was observed, however, cPACs have to be eluted from MCX cartridges in methanol modified with ammonium hydroxide. Despite evaporation and re-elution in mobile phase the high pH resulted in very poor resolution for all the compounds as well as loss of chiral recognition (see Fig. S1b). This has been demonstrated not to be an issue for C 18 stationary phase and non-chiral reversed phase chromatography [11], however, chiral columns are particularly sensitive to even small pH changes and the presence of charged molecules in the mobile phase impacts on both retention time and resolution of enantiomers.
MAX cartridges proved to be the most effective when coupled with chiral LC-MS. Despite low recoveries, good chromatographic separation of enantiomers was achieved and no build-up of back pressure was observed (see Fig. S1c).
The optimised method for the extraction of analytes from MAE extracts is therefore as follows (Fig. 1): 20 mL of MAE centrifuged supernatant, diluted with 65 mL of deionised water, was passed through Oasis MAX cartridge pre-conditioned with 4 mL of methanol followed by 4 mL of deionised water and eluted with 4 mL methanol.

SPE for liquid matrix.
Oasis HLB cartridges provided successful extraction of analytes from wastewater. The method involved filtering 50 mL wastewater samples (GF/F, Whatman, UK) then spiking with SS/IS (1 mg L À1 ). These were then passed through Oasis HLB cartridges, conditioned with 2 mL methanol, followed by 2 mL deionised water and eluted in 4 mL methanol (Fig. 1). However, it has to be remembered that this method, due to the use of GF/F filters, allows for the measurement of the analytes present in the liquid phase of wastewater and not any compounds adsorbed onto suspended particulate material (SPM). Previous analysis of SPM suggests that the amphetamine based compounds, venlafaxine and tramadol, are not significantly bound to SPM ( 3.4% of the total concentration was found on SPM), however, fluoxetine was largely found on the SPM (59.1% of the total concentration) so further analysis of SPM is required to determine an accurate whole matrix concentration [11].

Chiral LC-MS/MS with CBH column
The method for separation of enantiomers on the CBH column was used as reported elsewhere (Kasprzyk-Hordern and Baker [21]). Separation was undertaken using isocratic conditions, with a mobile phase of 90% H 2 O, 10% 2-propanol and 1 mM ammonium acetate at a flow rate of 0.075 mL min À1 . The column was maintained at 25 C and the injection volume was 20 mL (Fig. 1).

Chiral LC-MS/MS with Chirobiotic V column
Increased water content in the mobile phase increased elution of the compounds, however, also decreased the signal to noise ratio and the enantiomeric resolution in most compounds, with the notable exception of norfluoxetine. Increased ammonium acetate concentration increased the retention times of most compounds and with up to 10 mM the resolution was also improved, however, no compounds which were previously below the quantification threshold (resolution ! 1) improved enough to become quantifiable. Retention times and resolution were best at 0.001% formic acid, but again none of the compounds whose resolution was too poor to quantify became quantifiable between 0.001 and 0.005%. There was no trend related to signal to noise for ammonium acetate and formic acid at these concentrations. Therefore, the final mobile phase composition chosen for the Chirobiotic V column 6 S.E. Evans et al. / Analytica Chimica Acta xxx (2015) xxx-xxx   Note: (1) performance parameters (linearity retention time, resolution, enantiomeric fractions and ion ratios) were calculated for average values for the calibration protocol as a whole, above the ID. Whole calibration is 63 injections in total (2) accuracy and precision were calculated for two concentrations 50 and 250 mg L À1 ; elution orders shown have either been demonstrated by comparing single enantiomer t r or assumed to be the same as in other published work, using a comparable mobile phase and same column type. These are: alprenolol [38], citalopram, desmethylcitalopram [39], venlafaxine, desmethylvenlafaxine [40], metoprolol [41], mirtazapine [42], salbutamol [43] and terbutaline [44]. Linearity range for individual compounds: IDL-250 mg L À1 . Note: all performance parameters (retention time, resolution, accuracy, precision, ion ratios and SPE recoveries) were calculated for two concentrations 50 and 250 ng L À1 which were prepared in triplicate, n = 6.  Note: all performance parameters (retention time, resolution, accuracy, precision, ion ratios and SPE recoveries) were calculated for two concentrations 50 and 250 ng L À1 which were prepared in triplicate, n = 6.
In general the method development results highlight how longer retention times can produce better resolution, however, these two factors must be balanced in order to maintain a practical analytical technique, the current method for the target analytes is 80 min long. Although many iterations of additives and mobile phases were trialled, and trends were noted, none of the compounds previously unsuccessfully resolved with methanol, 0.005% formic acid and 4 mM ammonium acetate could be brought to above the relevant quantification thresholds and the method therefore remained unchanged.
The optimised chromatographic conditions used for separation of enantiomers in Chirobiotic V utilised a mobile phase composed of methanol, 0.005% formic acid and 4 mM ammonium acetate. Separations were undertaken under isocratic conditions and a flow rate of 0.1 mL min À1 . The column temperature was 25 C and the injection volume was 20 mL (Fig. 1).

Validation of instrumental parameters
Results of the calibration curves are detailed in Table 4. Good linearity of response (R 2 ! 0.997) within the studied concentration range was achieved for most compounds with the exception of terbutaline and norfluoxetine. These analytes consistently fell outside of tolerance limits set and will therefore not be considered further (results are included in the relevant table for illustrative purposes only).
Resolution (R s ! 1.00) was also achieved for all the enantiomers except MDEA (R s = 0.59-0.65), norephedrine (0.81-0.94) and norfluoxetine (0.83-0.87). For a few compounds, including methamphetamine, desmethylvenlafaxine, fluoxetine and citalopram, higher concentrations, resulting in larger peak areas, resulted in reduced resolution, however, this effect is considered to be negligible at environmentally relevant concentrations. Absolute EF values were usually within the range of 0.45-0.55 with the exception of amphetamine, ephedrine, norfluoxetine and tramadol, however, all of these were brought to within this range when average relative EF values were calculated, with the exception of tramadol. However, several compounds' relative enantiomeric fractions were not between 0.45 and 0.55, although their absolute enantiomeric fractions were. These include salbutamol, mirtazapine and desmethylvenlafaxine. Accuracy was good throughout and both inter-and intra-day precision was generally <10%. The ion ratio standard deviations were consistently well within tolerance limits set for all compounds. IDLs were usually between 0.01 and 1.83 mg L À1 and IQLs between 0.03 and 1.16 mg L À1 .

Validation of analytical methods
Almost all of the compounds were successfully calibrated, passing all of the criteria set out in Section 2.5, only exceptions were MDEA, terbutaline and norfluoxetine which were therefore not carried forward to validation. These three compounds all suffer from poor chromatographic resolution. Terbutaline co-eluted with another similarly structured compound resulting in multiple, overlapping peaks which impacted identification and quantification. MDEA and norfluoxetine did not achieve adequate resolution. During method development it was known that norfluoxetine requires 10% water in the mobile phase to achieve good enantiomeric resolution, however, this had significant impacts on the resolution and sensitivity of many other compounds, therefore, this compound could not be analysed alongside the other compounds resolved by the Chirobiotic V.  Tables 5 and 6, respectively. In brief, the retention time for all the compounds from the liquid matrices is not impacted by more than 4 min compared to sample diluent, except in the cases of citalopram, desmethylcitalpram and MDA in influent and effluent as well as S(+)-MDMA and S(+)-amphetamine in influent only. Changes in retention times were compensated for by using internal standards (see relative retention times in Tables 5 and 6). Enantiomeric resolution was maintained above 1 for all compounds at the three studied concentrations in wastewater influent and effluent except alprenolol (R s = 0.9), MDMA (R s = 0.9), tramadol (R s = 0.8) and methamphetamine (R s = 0.7). In order to undertake quantification of enantiomers, their resolution has to be !1. Therefore, compounds having R s < 1 were treated on a semiquantitative basis.
Several absolute EFs deviated from 0.5, however, their relative EF values were close to 0.5, with the exception of citalopram (EF = 0.8), tramadol (0.6), desmethylvenlafaxine (0.4) in influent, as well as citalopram (0.7) and pseudoephedrine (0.4) in effluent. In addition, although their absolute EF values were close to 0.5, desmethylvenlafaxine (0.4) and fluoxetine (0.6) EF values deviated from racemic in influent and ephedrine (0.6) and tramadol (0.4) in effluent. No deuterated analogues were available for these compounds and this result suggests that although there is no stereoselective degradation of the target analytes stereoselective matrix effects do appear to have occurred in the internal standard and therefore impacted the relative enantiomeric fraction. Interday and intraday precision for both matrices was high, ranging from 1.5% (MDMA) to 74.6% (atenolol), although most were less than 10%. Relative ion intensities were consistently within tolerance levels for both wastewater influent and effluent. Absolute SPE recovery in influent ranged from 30.4% (salbutamol) to 292.9% (desmethylvenlafaxine). Relative SPE recoveries in influent were all adequate between 56.6% (pseudoephedrine) and 138.8% (tramadol), although there were three exceptions: alprenolol (159.3 and 198.9% for each enantiomer), atenolol (21.4 and 19.5% for each enantiomer) and salbutamol (46.9 and 51.4% for each enantiomer). Effluent absolute SPE recoveries ranged between 25.9% (atenolol) and 252.4% (alprenolol). Several of the absolute recoveries were greater than 150%, although the corresponding relative recoveries were much lower suggesting significant ion enhancement took place. This was not the case in the influent which highlights the importance of validating all types of matrices to be analysed as large discrepancies can occur between comparable matrices and relatively cleaner matrices should not be assumed to pose fewer challenges for analysis. Relative SPE recoveries were generally acceptable, ranging between 56.4% (pseudoephedrine) and 142.0% (tramadol), the only exceptions to this range was atenolol (55.1% and 54.8 for each enantiomer), alprenolol (187.3% and 170.3% for each enantiomer) and salbutamol (46.9 and 64.2 for each enantiomer). There are some cases of absolute recoveries being quite different between enantiomers. This is thought to be a result of temporal separation during the chromatography followed by ion suppression and/or enhancement of one or both of the enantiomers.
Ion ratios for all the compounds in the liquid matrices are within the thresholds suggesting no co-eluting compounds are compromising the analysis. MDLs in wastewater influent were between 0.05 ng L À1 and 28.74 ng L À1 , MQLs ranged from 0.03 ng L À1 to 95.81 ng L À1 . MDLs in wastewater effluent were between 0.01 ng L À1 and 32.73 ng L À1 , MQLs ranged from 0.07 ng L À1 to 109.08 ng L À1 . The majority of MDLs and MQLs were below 1 ng L À1 , isolated cases, notably atenolol, citalopram and salbutamol, were significantly high though. The MQLs in the liquid matrices are all suitable for wastewater analysis, illustrating that despite a complex matrix this sample preparation and analytical method are suitable for wastewater concentrations of this magnitude.
The validation of the analytical method illustrates that the sample preparation methods chosen are appropriate.  Table 7. Ephedrine, pseudoephedrine and norephedrine all suffered from several issues including poor accuracy and precision and they are not considered validated for this method; the results are included for illustrative purposes only, in addition salbutamol's recovery was insufficient to be quantifiable. In brief, the retention times are impacted by this matrix and/or the method more than for the liquid matrices, however, due to the use of surrogate standards, changes in retention times are compensated for. Five compounds' EF values deviated from racemic in solid matrices including citalopram and desmethylcitalopram (0.6), mirtazapine (0.3), fluoxetine (0.2), tramadol (0.6) and venlafaxine (0.4). Intraday precision ranged between 4.3 and 33.4%, however, was generally below 15%. Inter-day precision ranged between 5.1 and 45.9%, although was generally less than 30%. Relative ion intensities were consistently within tolerance levels. Absolute recoveries are generally, relatively low, often between 10% and 50%, however, most relative recoveries were between 65% and 140%, exceptions however include atenolol (25.8 and 18.6% for each enantiomer), desmethylcitalopram (55.6 and 26.2% for each enenatiomer), fluoxetine (9.0 and 47.9% for each enantiomer), mirtazapine (243.3 and 172.9% for each enantiomer), and tramadol (209.3 and 323.7% for each enantiomer). Compounds in standard solutions were also passed through the cartridges without matrix (Table S3) which confirmed that the MAX cartridges are poor at retaining and/or eluting these compounds.
The sample preparation was suitable to remove co-eluting compounds whilst maintaining the resolution at a variety of concentrations. The cartridges, as discussed above, do not retain and/or elute the basic compounds efficiently, resulting in poor absolute recoveries. However, despite this, the majority are well compensated for by their internal standards, highlighting the importance of these internal standards. Compounds which do not meet the standards set for relative recoveries and/or relative enantiomeric fractions did not have direct deuterated analogues as surrogate standards, and therefore these discrepancies may be addressed with the introduction of new surrogate standards (if available), being direct deuterated or C13-analogues of target analytes. MDLs ranged between 0.080 and 7.12 ng g À1 . The MQLs, ranging from 0.080 ng g À1 (S(+)-venlafaxine) to 25.2 ng g À1 (R(+)-atenolol), appear to be appropriate for the concentrations found in the sludge samples suggesting the sample size chosen is adequate.

Wastewater analysis
The methods described above were applied to influent and effluent wastewater and digested sludge, and represent the first example of analysis of a suite of compounds from all three matrices in the wastewater treatment system at the enantiomeric level. Mass chromatograms of chiral drugs present in wastewater influent and analysed with both CBH and Vancomycin V columns are shown in Figs. 2 and 3 as examples. Concentrations in the influent ranged between 1.7 ng L À1 (metoprolol) and 1320.7 ng L À1 (tramadol), however, effluent levels appeared to be lower for all compounds apart from MDA, MDMA and salbutamol ( Table 8). The samples were all taken simultaneously, so it is impossible to comment on removal rates as they do not represent the same body of wastewater as it flows through the WWTP. However, the results do indicate that even closely related compounds do not appear to have similar degradation pathways. For example, atenolol and sotalol appear at relatively high concentrations in the influent and effluent, with little having been adsorbed to the digested sludge, whereas a relatively high concentration of propranolol has adsorbed to the digested sludge.
Enantiomeric fractions of ten compounds in the influent significantly varied from the EF recorded in the validation, suggesting human metabolism and biological action within the sewerage system had altered the enantiomeric fraction from the either racemic or single enantiomer ingested drug. In the effluent, after the biological action of activated sludge, EFs of 8 compounds were significantly different from those measured in a racemic mixture. Nine compounds: amphetamine, desmethylcitalopram, ephedrine, MDA, MDMA, methamphetamine, mirtazapine, norephedrine and pseudoephedrine had significantly different EF values in the influent and effluent, which indicates stereoselective processes occurring during activated sludge treatment.
The digested sludge, which had undergone activated sludge action and anaerobic digestions as well, contained all the validated compounds at quantifiable concentrations, except ephedrine, pseudoephedrine and salbutamol; concentrations ranged from 0.4 ng g À1 (metoprolol) to 275.2 ng g À1 (citalopram) even though many are polar. This analysis of digested sludge does highlight the potential for an overestimation in the success rate of activated sludge removing pharmaceuticals, such as in the cases of citalopram and fluoxetine where low concentrations are found in the effluent, however, relatively high concentrations are in the digested sludge. For the first time the examination of the enantiomeric fraction of these compounds in the solid matrices indicates that EFs for four compounds (alprenolol, MDA, MDMA, norephedrine and tramadol) significantly deviated from the racemic EF. In addition 9 compounds analysed had significantly different EFs in the effluent and the digested sludge demonstrating that biological processes which occur in the sludge, either during activated sludge treatment or during anaerobic digestion, result in a different EF when compared to the one in the water fraction. This indicates that either differing microorganism populations or the change in conditions instigates a different metabolic pathway which results in a different EF. It is worth noting that different microbial communities are present during aerobic activated sludge treatment and anaerobic digestion and therefore different stereoselective metabolic pathways were expected in this work. This is particularly stark in comparing effluent and digested sludge enantiomeric fractions of MDA (0.5 and 0.3, respectively), MDMA (0.9 and 0.4, respectively), alprenolol (0.5 and 0.7) and mirtazapine (0.2 and 0.5, respectively). Further work is currently being undertaken to understand stereoselective degradation of chiral PACs during wastewater treatment.

Conclusions
This work has demonstrated for the first time a method which can identify and quantify a range of chiral pharmaceuticals and illicit drugs at enantiomeric level from wastewater influent, effluent and digested sludge. These methods were evaluated through a series of parameters recommended by the EC Directive [37] and in addition the resolution and enantiomeric fraction, both absolute and relative, were assessed. To the authors' knowledge this is also one of the most rigorous validation tests of chiral wastewater analysis published to date. A combination of excellent instrument sensitivity, concentration steps within the sample preparation and good recoveries have resulted in low quantification levels within the anticipated environmental range.
Based on the results above, the following compounds are considered to have been successfully calibrated and validated for all three matrices: amphetamine, methamphetamine, MDMA, MDA, venlafaxine, desmethylvenlafaxine, citalopram, metoprolol, propranolol and sotalol. Mirtazapine, salbutamol, fluoxetine, desmethylcitalopram, atenolol, ephedrine and pseudoephedrine passed all validation criteria in liquid matrices, although in solid matrices they did not pass all the criteria and should be considered on a semi-quantitative basis only. In addition, alprenolol did not pass all the criteria in the liquid matrices and should be considered semi-quantitative in this matrix, although fully quantitative in the solid matrix.
The results indicate that chiral PACs are present in all wastewater matrices and that the EF values are often not racemic, suggesting current risk assessment procedures assessing whole drugs for toxicity in environmental matrices may not be appropriate, and that biological wastewater treatments are capable of changing the enantiomeric fraction.