Anthropogenic organic micropollutants (OMP) occur along the whole urban water cycle including drinking water. Various OMP can be efficiently adsorbed onto activated carbon. In the present study a commercial monolithic adsorber (MA), originally developed for gas treatment, was examined for the removal of OMP from drinking water. As a promising advantage, the adsorber can be introduced into existing pipes without causing substantial pressure losses. The MA was first characterized with scanning electron microscopy and energy dispersive X-ray spectroscopy. Weight loss during incineration at 550 °C indicated an activated carbon content of around 25%. Adsorption isotherms were recorded with milled material of the MA to estimate the capacity of the embedded adsorbent. Long-term flow-through experiments with two different flow rates were conducted to estimate the in-line removal efficiency. At low flow rates removals of 60% benzotriazole, 40% carbamazepine and 30% diclofenac were observed initially followed by a linear increase of effluent concentrations. Calculated loadings after 60 days of operation were e.g. 30 μg benzotriazole per g adsorber or 110 μg benzotriazole per g activated carbon as constituent of the adsorber. For specific applications, MA might be useful for the in-line elimination of OMP within drinking water distribution systems.
The widespread occurrence of organic micropollutants (OMP) such as pharmaceuticals and contrast agents in wastewater treatment plant effluents, surface waters (Ternes 1998; Reemtsma et al. 2006; Kasprzyk-Hordern et al. 2008; Loos et al. 2009; Loos et al. 2013) and in ground water (Reddersen et al. 2002; Loos et al. 2010; Wiese et al. 2011) has been confirmed by manifold investigations. Therefore the occurrence of OMP in drinking water cannot be excluded in partially closed urban water cycles. Granular activated carbon is successfully applied in fixed bed adsorption filters for the adsorptive elimination of OMP in waterworks (Moore et al. 2001; Yu et al. 2008; Scheurer et al. 2010). Recent investigations revealed that powdered activated carbon can also be applied in fixed beds (Ruhl et al. 2014a, 2014b; Altmann et al. 2015). However, another type of activated carbon fixed bed adsorber referred to as a monolithic adsorber (MA) or activated carbon monolith or activated carbon honeycomb has not been tested for the removal of widespread OMP from drinking water yet, to our knowledge.
Different types of MA have been developed for the adsorption of gas phase contaminants such as volatile organic compounds (Yates et al. 2002; Moreno-Castilla & Perez-Cadenas 2010). One type of MA is produced by activating a monolith consisting of phenolic resin (Gadkaree & Jaroniec 2000) while another type of MA is obtained by coating ceramic monoliths with polymers and subsequent activation (Vergunst et al. 1998). Liu et al. (2006) developed a method to extrude an MA on the basis of milled coal. Powdered activated carbon in combination with a mineral binder such as silicate clay can also be extruded to MA (Yates et al. 2002).
The different types of MA have been tested for the adsorptive removal of different contaminants, mainly from the gas phase. Agueda et al. (2011) studied the adsorption of volatile organic compounds onto MA with different degrees of activation. Through a high degree of activation an increase in the average pore size and a wider pore size distribution could be attained, which led to an increase in the adsorption capacity for dichloromethane. A mesoporous carbon monolith has also been used as a catalyst support and for the treatment of effluent gas purification Yates et al. (2000).
An advantage of MA is the convenient handling, as they are quick to replace in cases of saturation. Due to their open channel structure lower pressure drops occur at high flow rates in comparison to granular materials (Crittenden et al. 2005). However, Yates et al. (2000) reported that the dynamic adsorption capacity is limited by the width of the open channels and that a decrease of 20% of the channel width led to a fourfold amount of adsorbed toluene while an increase of the wall thickness did not lead to a significant increase. Gatica et al. (2010) examined the influence of the activation process on the adsorption capacity of coal-based MA for the removal of methylene blue and p-nitrophenol from liquid streams. The pore structures of the adsorber and thus the removal efficiencies of certain pollutants differed depending on the activation process (Gatica et al. 2013). Cheah et al. (2013) investigated acid modified carbon coated monoliths for methyl orange adsorption.
The main objectives of this work were: (1) to characterize the monolithic activated carbon adsorber, (2) to determine the adsorption capacity of milled MA material in batch tests in order to estimate the capacity loss of the activated carbon due to embedding, and (3) to evaluate how efficiently OMP can be removed from aqueous solutions in continuous in-line applications.
A set of MA (Aktiwa-w 100, Hydrotec, Germany) was received in cylindrical shape with a diameter of 8 mm and 60 mm length. According to the supplier the MA consisted of powdered activated carbon made of coconut shell and a binder. The outer surface area of the MA was calculated to be ca. 60 cm2 by considering the number of channels (ca. 24), each with an approximate width of 1.0 mm and the length of the MA. The adsorber volume (corresponding to the bed volume of a filter) was 3.0 cm3 and the total weight was 2.43 g. The weight loss after drying at 105 °C for 24 h revealed an initial water content of 6%. An additional weight loss of 28% was determined after combustion at 550 °C for 6 h. The residue, consisting of the inorganic binder material, was still in the shape of the MA.
A test solution was prepared by adding benzotriazole (BTA), carbamazepine (CBZ) and diclofenac (DCF) as representative indicator OMP (Jekel et al. 2015) to tap water. Stock solutions of each compound with concentrations of 20 mg/L were added to Berlin tap water (ca. 4.5 mg/L dissolved organic carbon, DOC) to achieve target concentrations of 1 μg/L. All compounds were supplied by Sigma-Aldrich.
The MA was milled for batch experiments in order to estimate equilibrium loadings of the activated carbon and to compare it with loadings obtained on the embedded activated carbon. A stock suspension of 1 g/L pulverized MA was prepared. Defined volumes were then pipetted to volumes of 100 mL of the test solution to achieve doses of 1, 2, 4 and 8 mg/L of the pulverized MA. The flasks were then shaken for 72 h to obtain equilibrium.
The MA was placed in a customized holder with minimal space between the outer adsorber wall and the inner holder wall. The adsorber was fed in up-flow with contaminated tap water from a 500 L tank using a peristaltic pump. The effluent was collected and weighed to adjust flow rates of 100 or 500 mL per hour, corresponding with ca. 1.8 or 0.4 minutes residence times. The tests were operated over 60 and 26 days, respectively. Samples for chemical analyses were collected at the effluent and references were taken from the tank.
The surface morphology of the MA was investigated by scanning electron microscopy (SEM) with cold field emission cathode and detectors for secondary electrons (SE) and back-scattered electrons (BSE). The SEM was equipped with energy dispersive X-ray spectroscopy (EDX) that was used to determine the chemical composition of the MA. Prior to analyses, the monolith was fractured to analyze the inner composition of the adsorber material.
A dual beam spectrophotometer (type Smartline UV Detector 200, Knauer, Germany) was used to measure the UV absorption at a wavelength of 254 nm (UVA254). OMP were quantified by high-performance liquid chromatography with tandem mass spectrometry (HPLC-MS/MS) with a method described elsewhere in detail (Zietzschmann et al. 2014a).
RESULTS AND DISCUSSION
Characterization of the monolith
The UVA254 as bulk parameter was measured to estimate the breakthrough of competing bulk organic constituents and as a surrogate parameter that has been reported to correlate with OMP removals (Zietzschmann et al. 2014a; Anumol et al. 2015). At the beginning the UVA254 was reduced to around 60% but quickly increased to almost the influent UVA254. The early breakthrough of UVA254 indicates that the adsorption capacity for bulk organic constituents is largely exhausted after a throughput of 100 L or 30,000 bed volumes, respectively. However, as long as 5–10% UVA254 abatement was observed, OMP removals were still significant. A much earlier breakthrough of UVA254 compared to OMP was also reported for granulated activated carbon (Corwin & Summers 2012; Zietzschmann et al. 2014b; Frank et al. 2015). The target OMP show a greater affinity towards the activated carbon surface compared to the bulk water constituents that cause the UVA254 (otherwise adsorption onto activated carbon would not be an efficient method for OMP removal). Due to the low concentration of 1 μg/L compared to a DOC concentration of 4.5 mg/L in the Berlin tap water, the UVA254 caused by the OMP does not significantly contribute to the overall UVA254. However, UV light absorption of OMP at higher concentrations (above 1 mg/L) has been reported to enable targeted activated carbon testing (Ruhl et al. 2014a, 2014b).
Comparison between batch and continuous experiments
The OMP load on the pulverized MA with its liberated activated carbon is several times greater than on the MA after the long-term experiment. A BTA solid phase concentration of almost 600 μg/g was obtained in the batch experiment with a dose of 1 mg/L pulverized MA while the solid phase concentration in the long-term experiment was below 30 μg/g. The difference can be attributed to the higher specific surface area of the pulverized MA and the better contact between the adsorber and the OMP. A large fraction of the activated carbon in the MA is probably shielded by the binding material and therefore not available for OMP adsorption. CBZ loadings were lower but the difference by a factor of 20 was almost identical. DCF revealed an unexpected behavior as loadings were low for doses of 1 and 2 mg/L pulverized MA. Competition for adsorption sites might have caused the comparably low DCF removal or solid phase concentration, respectively. The embedding of activated carbon in a binder material therefore leads to a loss of adsorption capacity.
Analysis of a MA by electron microscopy and X-ray spectroscopy revealed the presence of activated carbon particles surrounded by a mineral binder mainly consisting of silicon. Batch tests with milled MA material (and liberated activated carbon) achieved BTA loadings of 0.2 to 0.6 mg/g for doses of 4 to 1 mg/L. The two long-term tests were carried out with different flow rates and thus retention times. As expected, a longer retention time within the adsorber led to a greater OMP removal. A residence time of only 0.4 min was too short to obtain great OMP removals. The minimum contact time (depending on the flow rate within the drinking water pipe) can be easily increased by additional MA. Although lower OMP loadings were obtained with the embedded activated carbon in the MA compared to the pulverized MA, notable removals were obtained for more than 10,000 bed volumes. As the MA can be easily integrated into existing pipes and replaced after breakthrough, MA might be an interesting option for e.g. short-term application in the case of emergency. The pressure loss is expected to be negligible due to the geometry shown in the SEM images and according to Crittenden et al. (2005) but needs some further investigation with larger MA.
The investigations within the TestTools project (contract 02WRS1368A) were supported by the Federal Ministry of Education and Research (BMBF) as part of the funding measure RiSKWa. We thank F. Roeske for providing MA and U. Gernert (ZELMI, Technische Universität Berlin) for SEM and EDX analyses.