Occurrence and fate of pharmaceutical and personal care products in a sewage treatment works

Abstract

The occurrence and fate of eight pharmaceutical and personal care products (PPCPs) during sewage treatment has been studied in a pilot-scale treatment plant, comprising a primary settler (2.85 m³), an aeration tank (1.845 m³) and a secondary clarifier (0.5 m³), placed on site at a wastewater treatment works in the north west of the UK. It was fed both with raw sewage and the return liquor produced after sludge centrifugation, thus representing the most common configuration for a municipal sewage treatment plant based on the activated sludge process. Samples were taken at six different locations, including the return liquor stream, and analysed for musk fragrances and pharmaceutically active compounds belonging to various therapeutic groups such as anti-inflammatory drugs, tranquillisers and antiepileptics. Mass balances were conducted for those PPCPs that were quantifiable. The fate of the PPCPs was found to differ according to their physical-chemical characteristics. Anti-inflammatories underwent a degradation process and were almost completely removed from sewage during the biological treatment step. Musk fragrances were only partially removed, through adsorption onto the primary suspended solids and the biomass in the aerobic process, due to their strong lipophilic characteristics. The results of this study provide increasing evidence that the partial removal of these substances through the sewage treatment process contribute to the environmental occurrence of PPCPs. Consequently, existing STPs should be upgraded in order to attenuate the release of these substances into the aquatic environment.

---

Environmental impact

Sewage treatment plant (STP) effluents have been identified as the main pathway of pharmaceutical and personal care products (PPCPs) present in sewage into the aquatic environment but reported data tends to be contradictory. This study contributes to extend the understanding of the processes involved in PPCP removal along the different treatment units using a fully instrumented pilot plant unit, which was placed at the premises of a STP and simulated similar conditions to those typically established in full-scale STPs. The mass flow of five PPCPs was examined and different fates were observed depending on their physical-chemical properties. With the data obtained, the unwanted release of the anti-inflammatory diclofenac or the cosmetic ingredient galaxolide in the full-scale plant was estimated at rates of 1.5 kg per day, confirming that existing STPs should be upgraded in order to attenuate the release of these substances into the environment.

1. Introduction

Over the last decade, the occurrence of trace amounts of pharmaceutical products and other chemical ingredients from cosmetics in lakes, rivers and even tap water has become of increasing concern.^1–3 These chemicals are often referred to as pharmaceutical and personal care products (PPCPs) and are used in large quantities globally. Only in the last 10 years have analytical methods become sufficiently sensitive to detect and quantify PPCPs at the low concentrations at which they arise (ppb or ppt level). Since then, a few chronic ecotoxicological effects on organisms have been reported. A well known example is the dramatic decrease of vulture species populations in India, caused by traces of the anti-inflammatory drug diclofenac which is present on carrion.⁴ This has led to the supposition that similar effects might be occurring in surface waters, where aquatic organisms are continuously exposed to complex mixtures of micropollutants and their metabolites. However, there remains a paucity of available information regarding the potential impacts of PPCPs on the aquatic environment at environmentally relevant concentrations. Consequently, the general knowledge about PPCPs fate has gradually improved. It is now recognised that these compounds are released into the environment through different pathways. Most relevant are excretions via urine or faeces into the sewage system of unmetabolized fractions of drugs. The fate of these chemicals at sewage treatment plants (STPs) depends on their individual physical-chemical properties and biodegradability.⁵ Presently, recalcitrant PPCPs are released into surface water and are not significantly removed by a classical STP.²

Most of the research on this field has been carried out during normal operation of full-scale STPs considering only the liquid phase of both raw influent and final effluent to estimate overall removal rates without consideration of the influence of the different removal mechanisms. A few studies considered different sampling points along the studied STP, involving intensive sampling campaigns to generate mass balances,⁶ but the size of these locations and the daily variations of incoming crude sewage flow makes the generation of consistent results challenging. Pilot or lab scale studies in better controlled conditions, working with synthetic sewage where micropollutants are spiked, have also been performed,⁷ improving the general knowledge about biodegradability of these substances by estimating their biodegradation constants (k_biol).

Whilst providing more precise data, lab-based studies are not necessarily representative of processes operated at larger scale. Moreover, the use of synthetic media in some cases may not be representative of the behaviour of these substances in sewage treatment since actual full scale works would also be expected to receive both PPCP parent compounds and their metabolites and conjugates, the fate of which may differ from the parent compounds. Additionally, many studies have been carried out using significantly different values of operational parameters such as hydraulic retention time (HRT) and sludge retention time (SRT) which are known to influence the removal capacity of the treatment processes, making it difficult to draw general conclusions when comparing different research works. Therefore, many of the gathered results are subject to high uncertainty, as can be easily observed by comparing removal efficiencies from different studies. For example, ibuprofen (IBP) and naproxen (NPX) are considered polar substances that easily undergo biological transformation, but reported removal rates ranges between 60–90% and 40–90%, respectively.^3,6,8–11 The available data for diclofenac (DCF) is even more dispersed and contradictory since reported removal rates range from 0% up to 75%.^3,9–12 In the case of musk fragrances, they are substances characterized by a high lipophilicity which enables their removal from the liquid phase following a sorption mechanism onto either suspended solids or biological sludge. However, further degradation or volatilisation might be achieved due to their retention inside the aeration tank but there is not a general consensus about this possibility and again, reported removal rates are affected by a high variability (39–90% for galaxolide (HHCB) and 53–96% for tonalide (AHTN).^13–15 It is thus important to confirm previous research and to extend understanding of the processes involved in PPCP removal along the different treatment steps in STPs. Consequently, this work was carried out using similar conditions and operational parameters to those typically set on conventional sewage works, considering every stream entering or leaving the different units involved in the sewage treatment process. Simultaneous treatment of the return liquor produced after sludge centrifugation, also known as sludge reject water, is of particular interest since increasingly stricter environmental legislation requires that many existing STPs improve their final effluent quality by incorporating technologies able to cope with the simultaneous elimination of organic matter and nutrients, mainly nitrogen and phosphorus. These streams are characterized by high ammonia content and therefore a low COD/N ratio and its treatment might mean a potential improvement for STPs since reject waters from sludge digestion might contain around 10–30% of the nitrogen load entering the treatment plant. Examples of typical sludge reject water composition can be found at Ghyoot et al.¹⁶ and Wett et al.¹⁷ However, the composition of such streams in terms of micropollutants and their hypothetical influence in the final effluent quality is usually missed. This work aims to provide more extensive knowledge on the occurrence and fate of PPCPs in sewage treatment processes under strictly controlled conditions by means of a fully instrumented pilot plant operated at the premises of a full-scale STP. The pilot plant treated simultaneously raw sewage and a stream of the liquor produced after primary and excess humus sludge treatment, which was recycled into the aeration tank. This stream was considered a relevant sampling point due to its possibly high PPCPs content.

2. Experimental

2.1. Pilot-scale activated sludge plant

A diagram of the pilot plant used in this work is shown in Fig. 1. The activated sludge tank received both settled sewage and the return liquor which were brought in to the plant weekly from the on-site works centrifuge. The composition of the sludge before centrifugation was 59% of primary sludge, 23.4% of secondary sludge and 17% of sludge from the intermediate settlement tanks. To assist centrifugation, liquid polymer (Allied Colloids) was added to the sludge.

Fig. 1 Flow sheet of the pilot plant and considered sampling points.

In order to operate the biological unit with a HRT of 6 h, the flow of crude sewage into the primary clarifier was maintained at around 300 L h⁻¹, return liquor stream was fed at 9 L h⁻¹ and return activated sludge (RAS) rate was set at 1.

Throughout the sampling campaign, samples for PPCPs analysis were collected twice a day on each one of the six sampling locations, during two alternate days. Sampling points were the crude sewage, settled sewage, mixed liquor suspended solids (MLSS) supernatant, RAS, final effluent and return liquor. Routine physical-chemical analysis was carried out on a daily basis to assess the performance of the pilot plant with respect to aerobic carbonaceous removal and nitrification. These analyses included total and soluble biochemical oxygen demand (BODt and BODs), total and soluble chemical oxygen demand (CODt and CODs), total suspended solid (TSS) and ammonia. On-site test facilities were available to carry out operational testing such as conductivity, pH and dissolved oxygen for plant monitoring purposes.

2.2. Analytical methods

TSS content was determined according to standard methods.¹⁸ Total and soluble COD and ammonia (NH₄⁺) were determined using a Spectroquant Cell Test and measured on a Nova 60 model spectrophotometer (Merck, West Drayton, UK). Total and soluble BOD were determined according to standard methods.¹⁹ Conductivity and pH were measured using a Jenway 3540 pH & Conductivity Meter (Jenway, Dunmow, UK) according to standard methods.²⁰ Substances considered in this work were galaxolide (HHCB), tonalide (AHTN), celestolide (ADBI), ibuprofen (IBP), naproxen (NPX), diclofenac (DCF), carbamazepine (CBZ) and diazepam (DZP). Sample treatment for PPCP analysis consisted of a pre-filtration step through glass-fibre filters (APFC04700 or AP4004705, Millipore) immediately after sample collection, followed by filtration through nitrate cellulose membrane filters and addition of sodium diazide (Sigma-Aldrich, UK) as biocide, in order to avoid further biological degradation. PPCP content was determined after solid-phase extraction (SPE) of 100 mL pre-treated samples for sewage (crude or settled) and return liquor, or 250 mL samples for MLSS or RAS supernatant and final effluent samples, using 60 mg OASIS HLB cartridges (Waters, Milford, MA, USA). Cartridges were eluted with 3 mL of ethyl acetate. SPE extract was divided in two fractions for the direct determination of the soluble content of carbamazepine, diazepam and fragrances. The second fraction was used for the determination of anti-inflammatory drugs following silylation. GC-MS (Varian Saturn 2100 T) was used to determine the concentration of the investigated compounds in the SPE extract. Every sample was analyzed by duplicate in the GC-MS and the results were averaged. More detailed information about the analysis of the soluble content of anti-inflammatory compounds, CBZ, DZP and musk fragrances can be found at Rodríguez et al.²¹

2.3. Mass balance calculations

Prior to determining mass balances, PPCP concentrations sorbed onto sludge were estimated using only solid–water distribution coefficients from the literature (K_d in L kg⁻¹). This parameter, defined as the ratio between the concentration in the solid and liquid phases at equilibrium conditions, can reasonably predict PPCP sorption in STP processes.²² Due to the high variability of reported distribution coefficients and considering that these parameters might be matrix dependent, a selection criteria was followed: chosen K_d values were always experimentally determined, avoiding the selection of coefficients estimated with theoretical calculations. Table 1 shows the parameters that were chosen, for primary or secondary sludge and returning liquor, and other physical-chemical properties.

Table 1 PPCPs detected and physical-chemical properties.⁵ (s: solubility in water in mg L⁻¹; H: Henry's law constant); K_d values from: Ternes et al.²² and Urase and Kikuta.²³K_biol values (L g⁻¹ (TSS) d⁻¹) from Joss et al.⁷

PPCP	CAS	pKa	s	H	K biol	log Kd
						Primary	Secondary	Return Liquor
Ibuprofen (IBP)	15687-21-1	4.9–5.2	21	6.1 × 10^−6	9–35	<1.3	0.9	1.1
Naproxen (NPX)	22204-53-1	4.2	16	1.4 × 10^−8	0.4–1.9	1.1	1.1	1.1
Diclofenac (DCF)	15307-86-5	4.1–4.2	2.4	1.9 × 10^−10	<0.1	2.7	1.2	1.95
Galaxolide (HHCB)	1222-05-5	—	1.8	5.4 × 10^−3	<0.03	3.7	3.3	3.5
Tonalide (AHTN)	1506-02-1	—	1.2	5.1 × 10^−3	<0.02	3.7	3.4	3.55

---

In carrying out mass balances, some assumptions had to be made:

• Since no data were found in the literature, PPCP concentrations sorbed onto return liquor were estimated using a mean K_d value calculated from the ones reported for both primary and secondary sludge.

• The primary K_d value for NPX was assumed to be the same as that for secondary.

• Since the IBP, DCF and AHTN concentrations measured in certain samples were below the detection or quantification limit, mass balances which depended on these samples were calculated using these limit values.

Total mass fluxes for PPCPs load on each stream (Fig. 2) were calculated according to the next expression:

m = Q_in(S + X) (Eq 1)

Where m is the mass flux of PPCP (μg (PPCP) d⁻¹) entering or leaving a specific unit of the STP, Q_in is the sum of incoming flows (L d⁻¹), S is the PPCP concentration in the liquid phase (μg (PPCP) L⁻¹) and X (μg (PPCP) L⁻¹) is the amount of PPCP estimated to be sorbed onto the sludge phase.

3. Results and discussion

3.1. Conventional parameters

The pilot plant was operated in similar conditions to correctly operated full-scale STPs. Table 2 shows the values for conventional parameters measured in different sampling points and general sludge quality parameters during the sampling week.

Table 2 Characterization of conventional parameters in the sampling points

Stream	BODt		BODs		CODt		CODs		NH^4+		TSS
	Mean	SD	Mean	SD	Mean	SD	Mean	SD	Mean	SD	Mean	SD
Settled Sewage	111.5	34.8	55.8	13.9	299.2	44.1	132.8	24.7	24.0	3.7	83.4	27.8
Final Effluent	10.5	12.9	2.7	0.9	56.0	46.7	25.9	4.9	0.4	0.0	32.0	29.4
RAS	—	—	—	—	—	—	—	—	—	—	9818.0	1643.1
Aeration Tank	1922.2	683.3	47.4	21.2	6499.0	780.5	59.5	35.1	—	—	4870.0	567.1

---

	T/°C	pH	DO	ORP/mV	SVI
Aeration Tank	16.60	7.20	6.00	66.40	101.30
Secondary Settler	16.60	7.10	2.20	16.40	—

---

8 weeks after seeding with sludge from the on-site activated sludge bioreactor, stable MLSS concentration and acclimatisation conditions (>96% nitrification) were achieved and the sampling campaign was carried out. Dissolved oxygen content was always kept high enough (∼6 mg L⁻¹) to guarantee the development of a stable population of heterotrophic and nitrifying bacteria. Temperature and pH were not controlled, but their values (16 °C and 7, respectively) were representative of the normal situation in full-scale STPs. The sludge volumetric index (SVI) was in the common range (50–150 mL g⁻¹) for these systems. Therefore, the concentration of solids in the final effluent was low (<76 mg L⁻¹), which helped to maintain a SRT above 150 days (optimum for the development of slowly growing bacteria such as nitrifiers). Carbon removal and nitrification occurred efficiently since COD (total and soluble) and ammonia overall removals were 80 and 98% respectively.

3.2. Occurrence of selected PPCPs in the pilot plant

Table 3 displays average concentrations measured for PPCPs detected during the two sampling days in the different locations, together with the detection/quantification limits and the removal rates from the liquid phase.

Table 3 Concentrations of PPCPs (μg L⁻¹) detected along the different units of the pilot plant, removal rates and standard deviations. (LOD: detection limit; LOQ: quantification limit; n: number of samples; n.a.: not available)

Sampling Location	IBP	n	NPX	n	DCF	n	HHCB	n	AHTN	n
Crude Sewage	7.5	3	3	3	<0.1	3	1.59	3	0.7	2
SD	0.67		0.42		n.a.		0.34		0.2
Primary Effluent	7.5	8	3	8	<0.1	8	1.54	8	0.7	4
SD	1.4		0.5		n.a.		0.36		0.2
Return Liquor	4.6	4	1.7	4	<0.1	4	0.69	4	<0.023	4
SD	0.9		0.3		n.a.		0.29		n.a.
MLSS Supernatant	<0.08	3	0.2	1	1.2	3	1.06	3	0.37	3
SD	n.a.		n.a.		0.309		0.08		0.03
RAS	<0.08	3	0.2	1	1.2	3	0.96	3	0.36	3
SD	n.a.		n.a.		0.43		0.05		0.01
Final Effluent	0.2	2	0.2	3	1.1	4	1.07	4	0.37	4
SD	0.08		0.01		0.33		0.06		0.01
Removal Rate (%)	98		93		—		33		48
LOD	0.03		0.03		0.1		0.02		0.02
LOQ	0.08		0.08		0.3		0.07		0.07

---

The musk fragrance ADBI was not found at any sample and, on the contrary, HHCB and AHTN were found at substantial levels (2.0 and 0.9 μg L⁻¹ respectively). These two fragrances comprise about 95% of the EU market and 90% of the USA market for all polycyclic musks.²⁴ The ratio between the detected levels of HHCB and AHTN (2–3) is slightly lower compared with many previous works. Reiner et al.²⁵ compared concentrations detected in two different STPs for both musk fragrances and found influent concentrations of HHCB 4.5 to 6 times higher than AHTN. However, those results and the ones presented on this work are indicative of the greater production and use of HHCB compared with AHTN.

Similar to ADBI, the tranquilliser DZP was not found at any sample. This substance is not normally detected in STPs, and very few authors managed to detect concentrations even in the low ng L⁻¹ range.²⁶ CBZ was found in some locations during the first day of sampling, but the levels detected were always below the quantification limit of the analytical method (1.4 μg L⁻¹), which is particularly high for this substance. Considering that typically reported concentrations for CBZ are similar or below this value,²⁷ this result might be expected. However, it is interesting to mention that the detected levels were always higher in the final effluent, MLSS and RAS compared with the crude and settled sewage streams, leading to apparent negative removal rates. This behaviour through the sewage treatment process has been reported for CBZ and other substances such as antibiotics or β-blockers.²⁸ Pharmaceuticals enter STPs as either the original compound or as one of its metabolites, for example, glucoronide conjugates which remains undetected in the wastewater matrix. If these products are transformed during the treatment process to liberate the original compound, calculated removal rates might be underestimated. Considering that CBZ is considered a persistent substance,²⁷ in case its conjugates are transformed back to the parent compound during the biological treatment step, it would not undergo a further degradation process leading to higher outlet concentrations.

IBP was detected at the highest levels in sewage samples compared to the rest of the targeted pharmaceuticals, which is consistent with consumption rates reported for many EU countries.²⁹ On the other hand, DCF concentrations were below the detection limit in the sewage (crude and settled) and return liquor samples, whereas it was correctly quantified in the MLSS, RAS and final effluent samples. Consequently, increased effluent concentrations were found leading to estimated negative removal efficiencies. This trend has already been observed by other authors. Lishman et al.²⁸ pointed out that some deconjugation of acidic drugs within the collection and treatment system of STPs might be expected and also mentioned that, during the analytical process, extraction recoveries were lower for raw influent because the elution of co-extractives from the SPE cartridges could have caused the derivatization efficiency to drop. Moreover, Reddersen et al.³⁰ reported analytical issues for DCF measures based on acidic (pH = 2) solid-phase extraction of matrix-prone samples such as STP influents. As a consequence, it was not feasible to calculate accurately its removal efficiency and to estimate mass balances. In this work, DCF was found to have the second highest mean concentration in the final effluent samples. Ashton et al.¹ investigated the occurrence of several pharmaceuticals, including IBP and DCF, in several STP effluents and surface waters from the UK. The range of DCF concentrations detected in the final effluent of the pilot plant (0.8–1.4 μg L⁻¹) is consistent with this previous research in STPs from the UK. On the other hand, IBP concentrations, which ranged from 0.1–0.3 μg L⁻¹, were significantly lower than the ones reported by Ashton et al.¹ Comparing with treated wastewaters of different countries such as France, Greece, Italy, Sweden or Canada,^31,32 strong variations are observed between median and maximum IBP concentrations among the different countries. Whereas prescription rates and usage profiles may differ strongly from country to country, different factors (operational parameters, technology of the studied STPs, analytical methodologies and sampling protocols) are indeed decisive in the data variability, confirming the necessity of carrying out works in this field with better controlled conditions and following similar operational criteria.

Regarding the removal efficiencies from the liquid phase, IBP (98%) and NPX (93%) were almost completely eliminated. In the case of DCF, typically reported concentrations in raw sewage are usually in the range of 0.5–2 μg L⁻¹.^33,34 Comparing these reported levels with the ones measured in the effluent, it might be considered that the removal of DCF was low or negligible. However, the lack of DCF concentration data in the sewage stream makes it difficult to draw definite conclusions regarding the fate of this substance. Musk fragrance removal from the liquid phase was from moderate to intermediate (30–50%). These substances have high solid–water distribution coefficients (Table 1) and consequently, they tend to be attached onto the particulate phase, representing a good example of the importance of considering both liquid and solid phase in order to determine overall removal efficiencies, as will be discussed in the following section.

3.3. Mass balances

The calculation of mass balances in every unit permits the identification of removal mechanisms involved for each PPCP along sewage treatment and to estimate overall removal efficiencies for each compound. Main removal mechanisms are biodegradation, sorption and volatilization.⁵ The latter is influenced by the Henry's coefficient, only relevant for ADBI, which was not detected. In the remaining compounds, volatilization only accounts for less than 2% of removal.⁷ Therefore, volatilization will not be considered in this work. Fig. 2 shows calculated mass balances considering both liquid and solid phase, which were calculated with the average values from the two sampling days (Table 2), and Fig. 3 shows the overall removal efficiencies.

Fig. 2 Mean mass balances of PPCPs calculated along the different units of the studied STP.

Fig. 3 Global removal efficiencies calculated for PPCPs.

Crude sewage represents the load of PPCPs in the crude sewage stream (Fig. 2). Primary effluent load is calculated based on the settled sewage stream. RAS, settled sewage and return liquor are incorporated into the calculation of the activated sludge tank influent stream (biological in), MLSS supernatant is the only stream considered for the activated sludge tank effluent (biological out), which match up with the incoming load into the secondary settler (not shown). The load following secondary clarification (settler out) is calculated from the final effluent and RAS samples and the final effluent stream was calculated considering only the values from the final effluent samples.

Ibuprofen incoming load (54.6 mg d⁻¹) was the highest of all PPCPs and its elimination took place mainly along biological treatment (98%), confirming the biological degradation as its main removal mechanism. No differences were observed when comparing its removal rate from the liquid phase with the overall removal calculated after incorporating solid-phase data (Fig. 3), which permits the confirmation that IBP is a polar substance with no tendency to be sorbed onto solids. A similar behaviour was observed for NPX in terms of overall removal and sorption behaviour. Despite an incoming load of 21.9 mg d⁻¹, half the amount of IBP, NPX removal rate was slightly lower (93%). According to Joss et al.⁷ the biological degradation constant of this pharmaceutical is moderate (1–1.9 L g⁻¹_ss d⁻¹), one order of magnitude below the IBP constant (21–35 L g⁻¹_ss d⁻¹). Therefore, longer HRTs or higher MLSS concentrations are necessary for achieving significant NPX removal. In this work, the established HRT of 6 h might be low, but the MLSS concentration (∼5 g L⁻¹) was high enough to remove this substance in a significant rate.

Diclofenac mass flow load in the final effluent stream was considerably higher (8.4 mg d⁻¹) in comparison with IBP and NPX. Considering previously mentioned occurrence data, it can be assumed that its incoming load was lower compared to those of IBP and NPX and therefore its removal might be estimated as low or negligible. This finding is also confirmed by its low degradation constant and distribution coefficient (Table 1). Considering that MLSS and SRT were high enough, the only possibility to enhance the DCF removal along biological treatment might be to establish considerably longer HRTs, which indeed might affect the overall output of the STP, mainly in economic terms.

For both polycyclic musk fragrances, a certain degree of removal was achieved after primary settling. This behaviour might be due to their strong lipophillic character and indicates that removal following a sorption mechanism occurs along every stream which contains suspended solids. A marked increase in HHCB and AHTN incoming loads into the aeration tank was observed because of the influence of the return liquor and particularly the RAS stream. As a difference with the pharmaceuticals, a significant reduction of the fragrances load was observed comparing the biological out and final effluent streams, due to the solids separation after secondary settling. After this final step, the total removal rates achieved in the pilot plant were 68% for HHCB and 75% for AHTN (Fig. 3), significantly higher compared with the liquid phase data. A certain degree of biodegradation might also be achieved in despite of the low K_biol values reported for musk fragrances (Table 1), as a consequence of longer retention times inside the reactor due to their association with solids. This supposition is based considering works which detected HHCB-lactone, product of HHCB oxidation, in treated effluents.²⁵ Nevertheless, the calculated mass balances of this work indicate that the key removal mechanism for fragrances is sorption onto the particulate phase, primarily during the biological treatment step. Despite the degree of removal achieved, the loads in the final effluent for HHCB and AHTN were 8.3 and 2.9 mg d⁻¹ respectively, which are considerably higher than the ones measured for IBP and NPX (1.2 and 1.6 mg d⁻¹, respectively) and similar to DCF load, in the case of HHCB. Considering that the average flow rate of the full-scale STP where this research was conducted is 54 000 m³ d⁻¹, the estimated release of HHCB or DCF in the discharged liquid stream would be of 1.5 kg per day. In order to draw additional conclusions about the strategies that should be followed in order to attenuate the release of PPCPs into the aquatic environment, a direct comparison among removal rates reported in this work and previous research carried out by Carballa et al.⁶ in a STP located in north west Spain can be done, since most of the studied substances and analytical methodologies were similar. Interestingly, there are significant differences: IBP and NPX eliminations from the liquid phase were somewhat lower and fragrances removal rates were particularly high compared with our work. The elimination of HHCB and AHTN along pre-treatment and primary treatment steps accounted for half of their overall removal along the STP whereas in the present work most of the removal took place mainly along secondary treatment. As a feasible explanation, the full-scale STP incorporated an additional pre-treatment step (based on screenings and grit-fat removal) and more efficient primary settlers which were enabled to work with longer HRTs along primary treatment. On the other hand, the removal rates of IBP and NPX were considerably higher in our work (65 and 50% versus 93 and 99%, respectively). In this case, the main difference between both biological treatments is based on the SRT, considered an influencing parameter in terms of PPCP removal.¹² In the full-scale STP, no ammonia removal was achieved since a low SRTs of 1–3 days was established, whereas the correct design and operation of the studied pilot plant enabled to work with extended SRTs of 150 days. This operational strategy permitted to enhance the overall treatment quality, improving significantly the removal rates of IBP and NPX. The results clearly show the necessity to enhance the sewage treatment quality in order to attenuate the release of micropollutants into the aquatic environment.

3.4. Return liquor treatment

Table 4 shows conventional parameters analysis performed during the sampling week on the return liquor which were directly fed into the aeration tank. Comparing with data reported for reject water from sludge digestion,^16,17 nitrogen concentrations were lower but, on the contrary, COD (total and soluble) and TSS values were significantly higher.

Table 4 Conventional analysis of the return liquor in mg L⁻¹

Day	TSS	BODt	BODs	CODt	CODs	Ammonia
Monday	1110	1540	777	2940	1680	89.8
Tuesday	840	1880	789	3000	1560	88.2
Wednesday	780	1520	458	2950	1540	84.8
Thursday	790	1250	582	2810	1800	92
Friday	1060	1020	581	3010	1820	86.8
Mean	916	1442	637.4	2942	1680	88.32
SD	156.94	325	142.2	79.81	130.38	2.76

---

With respect to the treated settled sewage stream, considerably higher values were detected for the measured parameters in the return liquor. Despite this, the overall treatment capacity of the pilot plant was always excellent in terms of COD (total and soluble) and ammonia removal with no apparent impact on its normal operation, which confirms the benefits of treating this kind of stream in the conventional biological treatment. Regarding the studied PPCPs, their concentrations in the return liquor were roughly half that measured in the crude sewage stream (Table 3) with the exception of the musk fragrance AHTN, which had a concentration that was below the detection limit. Therefore, the detected levels confirm that a biological treatment of the return liquor is beneficial also in terms of PPCP removal and its influence on the overall treatment can be considered negligible, considering that the flow rate of this stream was significantly lower compared with the main stream of settled sewage coming into the aeration tank.

4. Conclusions

The pilot plant described in this work, based on the activated sludge system, has proven to be very effective for a combined treatment of both urban sewage and return liquor from sludge centrifugation. Moreover, COD (total and soluble) removal was always high, a nitrification rate up to 96% was easily achieved and no adverse effects were observed after treating the return liquor stream. Eight different PPCPs were analyzed in sewage samples and along the inflow/outflow of the different units of the pilot plant. Only AHTN and DZP remained below the detection limit, and CBZ was found in a few discrete samples. After estimating amounts of PPCPs sorbed onto solids, complete mass balances were calculated. The analysis of PPCP behaviour along the different units helped to ascertain the two main removal mechanisms involved. Anti-inflammatory drugs were mainly removed inside the activated sludge tank, most probably by biological degradation, whereas musk fragrance removal occurred by sorption onto solids and arose in the primary or secondary sludge. PPCPs levels in the return liquor from sludge centrifugation were approximately half that of the crude sewage, which confirms the benefits of treating such streams since no apparent influence on the overall removal of conventional parameters or PPCPs was observed.

In general, results observed in this work corroborate some of those previously reported after intensive sampling in full-scale treatment plants.^6,35 Almost complete removal rates were observed for IBP, NPX, and intermediate in the case of HHCB and AHTN. DCF concentration in the final effluent was the highest of this work (up to 1.1 μg L⁻¹), which might be indicative of a low or negligible removal potential. However, this substance could not be detected in sewage samples and therefore, it was not possible to confirm this trend with the available data. The calculated mass balances for the outflow load permitted estimates of release of PPCPs in the range 0.3–1.5 kg per day, depending on the substance considered. This estimation confirms that the development of enhancement strategies in existing plants should be a priority since it might help to attenuate the release of micropollutants in the water cycle, though new treatment and post-treatment technologies such as membrane bioreactors or ozonation are likely to continue to be explored.

Acknowledgements

The work described was supported by Spanish Ministry of Science projects FARMEDAR (CTM2004-04475), MICROFARM 2007–2010 (CTQ2007-66265) and NOVEDAR_Consolider (CSD2007-00055). Special thanks to Peter Hillis from United Utilities for his support.

6. References

1. D. Ashton, M. Hilton and K. V. Thomas, Sci. Total Environ., 2004, 333, 167–184.
2. T. Heberer, Toxicol. Lett., 2002, 131(1–2), 5–17.
3. T. A. Ternes, Water Res., 1998, 32(11), 3245–3260.
4. J. L. Oaks, M. Gilbert, M. Z. Virani, R. T. Watson, C. U. Meteyer, B. A. Rideout, H. L. Shivaprasad, S. Ahmed, M. J. Iqbal, C. M. Arshad, S. Mahmood, A. Ali and A. A. Khan, Nature, 2004, 427(6975), 630–633.
5. S. Suarez, M. Carballa, F. Omil and J. M. Lema, Rev. Environ. Sci. Bio/Technol., 2008, 7, 125–138.
6. M. Carballa, F. Omil, J. M. Lema, M. Llompart, C. Garcia-Jares, I. Rodriguez, M. Gomez and T. A. Ternes, Water Res., 2004, 38(12), 2918–2926.
7. A. Joss, S. Zabczynski, A. Göbel, B. Hoffmann, D. Loffler, C. S. McArdell, T. A. Ternes, A. Thomsen and H. Siegrist, Water Res., 2006, 40(8), 1686–1696.
8. N. Nakada, T. Tanishima, H. Shinohara, K. Kiri and H. Takada, Water Res., 2006, 40(17), 3297–3303.
9. M. J. Gomez, M. J. Martinez Bueno, S. Lacorte, A. R. Fernandez-Alba and A. Agüera, Chemosphere, 2007, 66(6), 993–1002.
10. M. Stumpf, T. A. Ternes, R. D. Wilken, S. V. Rodrigues and W. Baumann, Sci. Total Environ., 1999, 225(1–2), 135–141.
11. C. Zwiener, T. Glauner and F. H. Frimmel, J. High Resolut. Chromatogr., 2000, 23(7–8), 474–478.
12. M. Clara, B. Strenn, O. Gans, E. Martinez, N. Kreuzinger and H. Kroiss, Water Res., 2005, 39(19), 4797–4807.
13. R. Kanda, P. Griffin, H. A. James and J. Fothergill, J. Environ. Monit., 2003, 5(5), 823–30.
14. K. Bester, Chemosphere, 2004, 57(8), 863–870.
15. T. Kupper, C. Plagellat, R. C. Braendli, L. F. de Alencastro, D. Grandjean and J. Tarradellas, Water Res., 2006, 40(14), 2603–2612.
16. W. Ghyoot, S. Vandaele and W. Verstraete, Water Res., 1999, 33, 23–32.
17. B. Wett, R. Rostek, W. Rauch and K. Ingerle, Water Sci. Technol., 1998, 37(12), 165–172.
18. Methods for the Examination of Waters and Associated Materials (105, HMSO, 1980).
19. Methods for the Examination of Waters and Associated Materials (130, HMSO, 1988).
20. Methods for the Examination of Waters and Associated Materials (14, HMSO, 1978).
21. I. Rodríguez, J. B. Quintana, J. Carpinteiro, A. M. Carro, R. A. Lorenzo and R. Cela, J. Chromatogr., A, 2003, 985, 265–274.
22. T. A. Ternes, N. Herrmann, M. Bonerz, T. Knacker, H. Siegrist and A. Joss, Water Res., 2004, 38, 4075–4084.
23. T. Urase and T. Kikuta, Water Res., 2005, 39, 1289–1300.
24. Human and environmental risk assessment (HERA) on ingredients of household cleaning products, polycyclic musks AHTN (CAS 1506-02-1) and HHCB (CAS 1222-05-05), Environmental Section, version 2.0., Belgium, 2004.
25. J. L. Reiner, J. D. Berset and K. Kannan, Arch. Environ. Contam. Toxicol., 2007, 52, 451–457.
26. S. Castiglioni, R. Bagnati, R. Fanelli, F. Pomati, D. Calamari and E. Zuccato, Environ. Sci. Technol., 2006, 40(1), 357–363.
27. M. Clara, B. Strenn and N. Kreuzinger, Water Res., 2004, 38(4), 947–954.
28. L. Lishman, S. A. Smyth, K. Sarafin, S. Kleywegt, J. Toito, T. Peart, B. Lee, M. Servos, M. Beland and P. Seto, Sci. Total Environ., 2006, 367, 544–558.
29. M. Carballa, F. Omil and J. M. Lema, Chemosphere, 2008, 72, 1118–1123.
30. K. Reddersen and T. Heberer, J. Chromatogr., A, 2003, 1011, 221–226.
31. R. Andreozzi, M. Raffaele and P. Nicklas, Chemosphere, 2003, 50(10), 1319–1330.
32. C. D. Metcalfe, X. S. Miao, W. Hua, R. Letcher and M. Servos in Pharmaceuticals in the Canadian environment. Sources, fate, effects and risks, ed. K. Kümmerer, Springer, Berlín, Heidelberg, New York, 2nd edn, 2004, ch. 6, pp. 67–90.
33. A. C. Alder, A. Bruchet, M. Carballa, M. Clara, A. Joss, D. Löffler, C. S. McArdell, K. Miksch, F. Omil, T. Tuhkanen and T. A. Ternes in Human Pharmaceuticals, Hormones and Fragrances: The challenge of micropollutants in urban water management, ed. T. A. Ternes and A. Joss, IWA Publishing, London, 1st edn, 2006, ch. 2, pp. 15–54.
34. R. Rosal, A. Rodríguez, J. A. Perdigón-Meló, A. Petre, E. García-Calvo, M. J. Gómez, A. Agüera and A. R. Fernández-Alba, Water Res., 2010, 44, 578–588.
35. E. Artola-Garicano, I. Borkent, J. L. M. Hermens and W. H. J. Vaes, Environ. Sci. Technol., 2003, 37, 3111–3116.

---