Drug abuse has been well chronicled for decades, but until recently it has not played a big role in environmental controls. In the last few years, drugs of abuse, as well as some common legal drugs such as caffeine and nicotine have been increasingly identified in water testing. It has reached the point where these drugs have become a new class of environmental contaminants identified in the aquatic environment, and have thus received an increased interest from water specialists.

The presence of these compounds, either unaltered or as their main human metabolites, in wastewater has been reported in many countries. Moreover, it has been proven that they are often only partially removed by wastewater treatment plants (WWTPs) that use conventional treatments. This incomplete elimination leads to the release of these compounds into surface receiving waters. Since these waters can be used for drinking water production it is important that drugs are eliminated through drinking water treatments.

Table 1: Properties of the studied compounds

The use of reverse osmosis (RO) has proved to be effective at removing emerging contaminants such as pharmaceuticals by greater than 95 percent in wastewater. Researchers at Hokkaido University in Sapporo, Japan, explored the capability of a polyamide membrane (XLE) and a cellulose acetate membrane (SC-3100) to eliminate pharmaceutical neutral compounds. Rejection by XLE membrane was largely related to molecular weight, whereas SC-3100 membrane rejection was highly dependant on polarity.

The aim of this work was to examine the ability of different types of RO membranes to eliminate drugs of abuse under real operational conditions. Experiments were performed in a small RO pilot plant, placed in a WWTP in Northeast Spain, which was treating urban wastewater. The secondary effluent after biological treatment was used as an RO feed. Feed and permeate samples were studied in order to obtain rejection values. Three types of RO membranes in parallel have been tested for caffeine, nicotine, its metabolite cotinine, codeine, its metabolite norcodeine and amphetamine-type compounds. The rejection of the target compounds present in the feed has also been studied seasonally.

Table 2: Details of the used membrane elements

Studied compounds

Compounds included in this study, their metabolites and some of their chemical characteristics can be found in Table 1. Except caffeine, all of the studied compounds were listed as commonly abused drugs by the U.S. National Institute on Drug Abuse (NIDA) that can have potential health consequences. Although caffeine might be categorized as a stimulant drug, it was included in the study as it is often considered an indicator of pollution of aquatic environments with wastewater effluent. AMP, METH, MDA, MDMA (ecstasy) and MDEA are stimulant drugs, abused due to their ability to cause euphoria. AMP and METH are also approved for restricted medical use. Nicotine is found in common tobacco, cigar and smoking cessation assistance products. In humans, nicotine readily metabolizes to cotinine. Codeine is an opiate, commonly used narcotic pain medicine to treat mild to moderate pain. Following ingestion, codeine metabolizes to norcodeine.

Figure 1: Above is a flow diagram of the pilot plant.

Membranes

Details of the membranes are presented in Table 2. Three different polyamide membranes were used in this study. The LE membrane was a low-energy membrane widely used for industrial and municipal applications that operate at low pressure. The BW30 membrane was an industry-standard, high-rejection and high-productivity brackish water membrane. XFR was a Dow Chemical membrane with advanced fouling resistance, targeted to wastewater applications. The elements were 2.5 inches in diameter and 14 inches in length.

Rejection was specified in following conditions: 2,000 ppm NaCl, 25°C, 5-percent recovery and pressures of 10.3 bar (LE) and 15.5 bar (BW30 and XFR). Two sets of new membranes were used, one set for summer sampling and a second set for the winter sampling.

Figure 2: RO inlet concentrations of norcodeine and codeine.

Reverse osmosis pilot plant and experimental conditions

The pilot plant used in this experiment was located in a Spanish wastewater treatment plant treating municipal wastewater (16,500 m³/d) originating from the near by villages. The existing treatment scheme consists of a primary treatment, a secondary biological treatment and a tertiary treatment including chlorination, coagulation/flocculation, lamellar clarification and sand filtration. In this experiment the feed water to RO was taken after secondary treatment. The process is shown schematically in Figure 1.

In order to assess seasonal variations in the presence and rejection of the studied compounds, three sampling campaigns were done; June, August and March. Details of the feed water and experimental conditions at the sampling time can be found in Table 3. One-liter samples were taken in amber-colored glass bottles from the outlet of the secondary treatment (RO pilot plant feed) and three permeates.

Table 3: Feed water composition and experimental conditions during sampling

Standards, reagents and methodology

Standard solutions (1 mg/mL in methanol) of AMP, METH, MDA, MDMA, MDEA, caffeine, codeine, norcodeine, nicotine and cotinine were used. Standard solutions of each compound and deuterated analogues, AMP-d8, METH-d9, MDA-d5, MDMA-d5, MDEA-d5, 13C3-caffeine, codeine-d6, nicotine-d4 and cotinine-d3 were purchased from Cerilliant of Austin, Texas.

The target compounds were analyzed following the method developed by researchers at Spain's University of Santiago de Compostela, based on solid-phase extraction ultra-performance liquid chromatography-tandem mass spectrometry (SPE-UPLC/MS/MS). Briefly, 100 mL of water spiked with deuterated standards were enriched by SPE on Oasis-HLB cartridges from Waters Corp., Milford, Mass. Cartridges were washed with a 5-percent methanol aqueous solution, dried with nitrogen and eluted using methanol.

Compounds were analyzed by UPLC-MS/MS. An Acquity BEH C18 column (100 mm × 2.1 mm, 1.7 µm) and solvent A: acetonitrile with 0.1 percent formic acid and solvent B: 30 mM formic acid/ammonium formate (pH 3.5), were used for optimum separation of target compounds. The UPLC was coupled to a Quattro Micro triple quadruple mass spectrometer operating in positive electrospray ionization mode. Acquisition was performed in selected reaction monitoring (SRM) mode. The precursor ion of each compound was the protonated molecular ion. Two transitions per compound were employed as required by the European Union. The quantification and confirmation transitions have been published elsewhere.

Figure 3: RO inlet concentrations of Nicotine and Cotinine.

Presence of studied compounds in the WWTP effluent by season

Ten different compounds (see Table 1) were studied; the observed concentrations in the treated wastewater during three campaigns are presented by compound group in Figures 2-5. Except for AMP, all studied compounds were detected in the secondary treated wastewater, indicating that the conventional biological process alone was not sufficient to remove them from waste streams. Most of the studied compounds showed a clear increase during the summer season (June and August). This can be related to the fact that the region is a very popular tourist area and the population (and thus the wastewater load) significantly increases during the holiday period.

Codeine was found in high concentrations in every campaign. Its metabolite, norcodeine was also found to be present, but around 20 to 33 times lower concentrations compared to codeine. Very high concentrations, up to four times (for codeine) and six times (norcodeine) higher were observed during the summer season. These concentrations are significantly higher than previously reported in studies.

Figure 4: RO inlet concentrations of MDA, MDMA, MDEA, AMP, and METH

Concentrations of nicotine and its metabolite cotinine are shown in Figure 3. In all four campaigns, cotinine was found at higher concentrations than nicotine. This can be related to the half-life of both compounds. For nicotine in humans, it is only around two hours, where for cotinine the half-life is around 20 hours, and it is typically detected for several days in blood and urine. For this reason, cotinine, rather than nicotine, is used as indicator to assess exposure to tobacco. As can be seen in Figure 3, the presence of both compounds in the treated wastewater is fairly stable throughout the campaigns with an important increase of cotinine in August. The found concentrations were within the limits reported by Boleda et al, from similar WWTPs.

The concentrations of amphetamine-type compounds are given in Figure 4. Amphetamine was not found in the tested feed samples and methamphetamine was found to be present only at very low concentrations, between 0.6 and 1 ng/L. The highest concentration was found during the summer season. MDA and MDMA were found to be present in the outlet of the WWTP throughout the year. Similarly to METH, both illicit drugs showed an increase in concentration during the tourist season. MDEA was also found in all tested samples, but at lower concentrations.

Caffeine was found in high concentrations, between 446-4,106 ng/L in every campaign. These results are in line with Buerge et al., reporting concentrations between 30-9,500 ng/L found in the outlet of Swiss WWTPs while the removal efficiency of the WWTPs was already very good, between 81 and 99.9 percent. Similarly, as with other tested compounds, higher concentrations were observed during the summer time, especially in the August sample, when the WWTP outlet concentration was around nine times higher than in the winter sample.

Figure 5: RO inlet concentrations of caffeine.

Elimination percentage

Concentrations of studied compounds in the RO permeate samples during three sampling campaigns are presented in Table 4. Elimination percentage, or rejection, of every compound was calculated as presented below:

Rejection =
Where:	Cp = permeate concentration
	Cf = feed concentration

Results are summarized by membrane type, LE, BW30 and XFR in Table 5 and Figures 6-8, respectively. The rejection of different compound groups are discussed separately below, but some general trends observed for all the compounds are included here:

• In some cases the concentrations of the studied compounds in the permeate samples were below the quantification limits (Table 4). In these cases, the rejection was calculated by using permeate concentration, which is half of the limit of quantification (LOQ). These values are presented in italics in Table 5.
• In general the obtained rejection values are high, indicating that reverse osmosis is a suitable technology to remove drug contaminants from waste streams.
• No major differences between the three studied membranes were observed, which is important since low energy membranes can efficiently be used for these compounds.

The rejection by season and membrane for nicotine, cotinine and caffeine is presented in Figure 6. For cotinine, a very high rejection, between 92 and 97 percent was obtained in all the campaigns and membranes. In contrast, nicotine rejections were lower, at between 28 and 75 percent. The lowest rejection was obtained during the winter sampling. During the first two campaigns (June and August), the rejection was significantly better, between 59 and 75 percent. A significant rejection difference between the three membranes was observed in the winter sample, as the obtained rejection for BW30 was around 15 percent higher than that for the LE and XFR membranes. This greater performance of the BW30 membrane was not observed with any other tested compound and thus it could be considered as an excluder.

Table 4: Observed concentrations (ng/L) in the RO feed and permeate of LE, BW30 and XFR membranes

Caffeine was eliminated by RO very well, between 96 and 99 percent during the summer campaigns, but significantly lower rejections, 55 to 70 percent, were obtained during the winter sampling. As seen with nicotine, the winter sample showed variation between the tested membranes, but the results were reversed. The rejection of the BW30 membrane was around 20-percent lower than obtained with LE or XFR membranes. This correlation was not observed with any other tested compound, and the relationship was not consistent throughout the sampling, thus it can also be considered to be an excluder. Previously published literature suggests rejection of around 90 percent for caffeine with a BW30 membrane, which is in line with the values obtained during summer sampling.

The reason rejection was higher in summer is not clear. Based on the standard RO solute transport model, one would expect the relationship to be reversed. The change of membranes could be related to the phenomenon, but then one would expect to see a similar behavior with all tested compounds and, other than these two compounds, no reduced rejection during winter sampling was observed.

Table 5: Elimination percentage for three permeates tested in three campaigns.

Generally, codeine and norcodiene were not detected in the studied permeate samples, even when they were present at high concentration in the RO inlet water. Thus, the elimination percentage by RO was high, in the range of 99 percent for codeine and 90 to 98 percent for norcodeine with all tested membranes.

The elimination percentages of MDA, MDMA, MDEA and METH are given in Figure 7. MDMA was rejected at the range of 93 to 98 percent, and no correlation between the observed rejection and the membrane type or season could be observed. In contrast, MDA and MDEA showed seasonal variations in rejection. For MDA, the rejection in the June campaign was around 48 to 50 percent, whereas rejection in the August campaign was slightly higher, around 63 to 65 percent. The winter campaign was conclusive compared to the earlier campaigns, as the rejection varied from 46 to 76 percent.

The obtained results for MDEA rejection were questionable, as the obtained rejection values were generally very low and varied – from 9 to 73 percent and 11 to 22 percent in the June and August campaigns, respectively, and 51 percent in the March campaign. In general, the observed concentrations were very close to the LOQ, which might affect the precision of the results.

Methamphetamine was present in the feed water at low concentrations and could not be detected in of the tested permeate samples. The calculated elimination percentages were in the range of 61 to 91 percent for methamphetamine. PE

Figure 6: Rejection of nicotine, cotinine and caffeine with RO membranes.

Acknowledgement

Figure 7: Rejection of MDA, MDMA, MDEA and METH with RO membranes.

This study has been carried out within the framework of the Sostaqua project, led by Agbar, aiming on developing technologies towards the sustainability of urban water cycle. Authors would like to acknowledge the support and funding by the Spanish Government through CDTI organism (Centro para el Desarrollo Tecnológico Industrial) under the Ingenio 2010 program. Authors also would like to acknowledge the staff at Sorea (member of Agbar group) WWTP for their contribution on the practical execution of the study as well as ACA (Agencia Catalana de l'Aigua) for authorizing the pilot plant installation at their premises.

References for this study are available upon written request.