Science of the Total Environment 473–474 (2014) 207–214

Contents lists available at ScienceDirect

Science of the Total Environment

j ourna l homepage: www.e lsev ie r .com/ locate /sc i totenv

γ-Al2O3-based nanocomposite adsorbents for arsenic(V) removal:Assessing performance, toxicity and particle leakage

LindaÖnnby a,⁎, Christian Svensson b, LubindaMbundi c,d, Rosa Busquets c,e, AndrewCundy c, Harald Kirsebom a

a Department of Biotechnology, Lund University, P.O. Box 124, SE-221 00 Lund, Swedenb Ergonomics and Aerosol Technology, Lund University, Sölvegatan 26, SE-221 00 Lund, Swedenc Faculty of Science and Engineering, University of Brighton, Brighton BN2 4GJ, UKd Blond McIndoe Research Foundation, Queen Victoria Hospital NHS Trust East Grinstead, West Sussex RH19 3DZ, UKe Kingston University, Faculty of Science, Engineering and Computing, School of Pharmacy and Chemistry, Penrhyn Road, Kingston Upon Thames, Surrey, KT1 2EE, 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

• Adsorption performance and environ-mental risks with nanomaterial wereinvestigated.

• Nanocomposites embeddedwithγ-Al2O3

nanoparticles show no particle leakage.• High stability in flow-through experi-ments was confirmed from examinedmaterials.

• Monomer leakage was linked to reduc-tion in viability on epithelial cells.

⁎ Corresponding author.E-mail address: Linda.Onnby@biotek.lu.se (L. Önnby).

0048-9697/$ – see front matter © 2013 Elsevier B.V. All rihttp://dx.doi.org/10.1016/j.scitotenv.2013.12.020

a b s t r a c t

a r t i c l e i n f o

Article history:Received 5 September 2013Received in revised form 22 November 2013Accepted 5 December 2013Available online 24 December 2013

Keywords:CompositesAdsorbentsMacroporous polymersNanoparticlesAluminium oxideRisk assessment

The generation and development of effective adsorptionmaterials for arsenic removal are urgently neededdue toacute arsenic contamination of water sources in many regions around the world. In the search for these new ad-sorbents, the application of nanomaterials or nanocomposites, and especially the use of nanoparticles (NPs), hasproven increasingly attractive.While the adsorptive performance of a range of nanocomposite andnanomaterial-based systems has been extensively reviewed in previously-published literature, the stability of these systems interms of NP release, i.e. the ability of the nanomaterial or nanocomposite to retain incorporated NPs, is less wellunderstood. Here we examine the performance of nanocomposites comprised of aluminium oxide nanoparticles(AluNPs) incorporated in macroporous polyacrylamide-based cryogels (n-Alu-cryo, where n indicates the per-centage of AluNPs in the polymer material (n = 0–6%, w/v)) for As(V) adsorption, and evaluate AluNP leakagebefore and after the use of thesematerials. A range of techniques is utilised and assessed (SEM, TEM,massweightchange, PIXE and in vitro toxicity studies). The 4-Alu-cryo nanocomposite was shown to be optimal forminimising AluNP losses while maximising As(V) removal. From the same nanocomposite we were furtherable to show that NP losses were not detectable at the AluNP concentrations used in the study. Toxicity tests re-vealed that no cytotoxic effects could be observed. The cryogel-AluNPs composites were not only effective inAs(V) removal but also in immobilising the AluNPs.More challenging flow-through conditions for the evaluationofNP leakage could be included as a next step in a continued study assessing particle loss and subsequent toxicity.

© 2013 Elsevier B.V. All rights reserved.

ghts reserved.

1. Introduction

Contamination of water sources by arsenic is a global environmentalproblem. While particularly severe in Bangladesh, arsenic contamina-tion is also amajor issue inmanyother regions: e.g. SouthAsia, southern

208 L. Önnby et al. / Science of the Total Environment 473–474 (2014) 207–214

parts of Europe, the USA, South America and China (Chowdhury et al.,2000; Nickson et al., 1998; Nordstrom, 2002; Winkel et al., 2008). Dueto different groundwater conditions (Hug et al., 2008), and indeed vary-ing levels of economic development andwater pre-treatment, there arecurrently no universal treatment solutions for As-contaminated water.In order to provide regional and local solutions, a range of technologieshave to be developed and implemented. Arsenic remediation is chal-lenging; arsenic can for instance be present in different oxidation states,show high temporal and spatial variability in concentration, and co-exist with other (competing) ions depending on the pollution sourceas well as the sedimentary and geochemical characteristics of thegroundwater source area (Chowdhury et al., 2000; Hug et al., 2008;Nickson et al., 1998; Smith et al., 2000). While dissolved concentrationsof arsenic can vary significantly, the guide line value concentration forarsenic in drinking water is set to 10 μg/l (WHO, 2001). From a treat-ment point of view, considering the environmentally relevant concen-trations for arsenic in groundwater, arsenic is most optimally removedusing adsorption processes (Mohan and Pittman, 2007; Nordstrom,2002). A wide range of adsorbent-based solutions for As removal havebeen proposed (see review in Mohan and Pittman, 2007), including re-cently a range of nanomaterial-based sorbents (Ali, 2012; Khin et al.,2012). These nanomaterials, incorporating different metal nanoparti-cles (NPs, e.g. Al, Fe, Mn, Zr, and Ti oxides and oxyhydroxides) are inter-esting since they offer a much greater surface area per unit volume incomparison to the corresponding micro-sized metal (and consequentlymay show enhanced reactivity and adsorption capacity), and theadsorption kinetics are often improved for particles of smaller size(Khin et al., 2012). There remain, however, a number of concerns overthe toxicity of metal (and other) NPs, particularly in a “free” or mobileform in solution and/or suspension.

Evidently, NPs of different elements (or compounds of one or moreelements) may give different toxicity responses. In addition, the directtoxicity from the NP itself must be distinguished from any toxic solublecontaminant derived from theNP, such as ions or surfactants (Ali, 2012;Klaine et al., 2008). There have been a number of discussions regardingsolubility aspects of NPmaterials involved in toxicity studies, especiallyfor aluminium ions (Al3+), which show phytotoxicity and are releasedand dissolved at higher pH-values (Murashov, 2006; Yang and Watts,2005). Comparing toxicity for different bulk concentrations of NPs canhowever be difficult. Using concentration as a metric is not straightfor-ward as surface area, particle charge, size and coating are also importantparameters (Arvidsson et al., 2011; Klaine et al., 2008; Teeguarden et al.,2007). As an example, the available surface area of NPs is related tothe surrounding conditions, such as pH and ionic strength. More gener-ally, both the environmental toxicity and fate of NPs remain poorlyconstrained, even for more toxic (and more intensely studied) NPtypes such as AgNPs. Recent observations indicate that AgNPs showlittle discharge into surface waters as a result of the urban wastewater cycle (Kaegi et al., 2013). In contrast, in other studies syntheticnanoparticles have been observed to readily distribute through environ-mental systems, e.g. after an exposure period of 12 days in an aquaticmesocosm, gold nano-rods could be found in biofilm, the water column,clams and other biota (Ferry et al., 2009).

Polymers and othermaterials can be used as a support for adsorbentparticles or NPs, i.e. forming nanocomposites (Gupta et al., 2012; Onnbyet al., 2012; Sankar et al., 2013; Savina et al., 2011; Zhu et al., 2012)which have been proposed as water clean-up devices. These canharness the high reactivity of NPswhile preventing their environmentalrelease, and also allow application of these adsorbent devices in avariety of end-use configurations (e.g. as beads, column filters, and sur-face coatings). Besides applying nanocomposites for adsorption of toxicions, nanocomposites have also been studied for disinfection purposesor used as membranes in water treatment. In the case of the latter,NPs have been shown to result in less bio-fouling and improved perme-ate quality (Kim and Van der Bruggen, 2010; Loo et al., 2013). For mostresearch dedicated to nanocomposites, however, little attention is paid

to possible environmental consequences arising from their application,e.g. if NPs are lost to, or dissolved in, the treated water. Despite theprototype status of many nanocomposite devices, such environmentalrisk assessments are a key to characterize any potential risk from thedevices prior to up-scaling and practical application. Here we examineadsorbent nanocomposite materials based on highly porous polymers.These polymers, so-called cryogels, are produced at temperaturesbelow the freezing point of water allowing them to possess a beneficialthree-dimensional interconnected continuous pore system of micro-metre size (Lozinsky, 2002). Ideally, the use of nanocompositematerialsshould reduce the release and potential toxic effects of NPs while theadsorption of the target contaminant(s) by the nanocompositeshould be comparable (or higher) than that obtained from the freeNPs. Herein, we test the As(V) removal performance of a nanocom-posite of AluNPs incorporated in a polyacrylamide (pAAm) cryogelmatrix, referred to subsequently as Alu-cryo. In addition, we investi-gate the possible loss of incorporated NPs from the Alu-cryo with re-spect to robustness of the material following application as a waterfilter, and examine the release of AluNPs to filtered water by meansof particle induced X-ray emission (PIXE) mass analysis. The potentialtoxicity of the filtrate using human epithelial cells in cell culture isalso evaluated.

2. Material and methods

2.1. Chemicals

Acrylamide (AAm, N99%), D3-acrylamide (98%), N′,N′,-methylene-bisacrylamide (MBAAm), N,N,N′,N′-tetramethylethylenediamine(TEMED), ammoniumpersulphate (APS) and aluminiumoxide nanopar-ticles (AluNPs, nanopowder, γ-Al2O3, b50 nm, N40 m2/g, 4 g/cm3) werepurchased from Sigma Aldrich. Arsenic standard stock solution, As(V),and palladium nitrate were from analytical standards (Germany).Ultrapure water (18.2 MΩ cm at 25 °C) was used throughout thestudy (Millipore system, France) except for the cytotoxicity studieswhere water from an ELGA Purelab flex purification system (Veolia,UK) was used. The ultrapure water is referred to subsequently asMilli-Q water.

The cell media used in the cytotoxicity studies, was a DMEM/Ham'sF12 medium supplemented with 10% foetal bovine serum, 2.5 mML-glutamine, and 100 IU/ml penicillin–streptomycin; (PAA, UK).

2.2. Preparation of Alu-cryo nanocomposite

Nanocomposites were prepared following (Onnby et al., 2012) butwith slight modifications, hence we prepared a pAAm cryogel withoutaddition of allylamine, as the stability was improved with the pAAmcryogel (Onnby et al., 2012). In detail, different amounts of AluNPswere added to a 6% monomer solution with 1:7 molar ratio ofMBAAm:AAm, resulting in a nanocomposite with 1–6% (w/v) particleloading. The monomer and NP suspension were activated with TEMEDand further initiated with APS, both added as 1 wt.% of the total mono-mer concentration respectively. Prior to the initiation of the reaction, themonomer solution was degassed with nitrogen while cooling on ice. Al-iquots of monomer-nanoparticle suspension (0.5 ml) were thereafteradded to glass tubes (inner diameter = 10 mm, length 30 mm) andallowed to freeze in an ethanol bath for 1 h at −12 °C. Samples werethereafter kept overnight in a dry freezer (−12 °C).Wehereby obtained0.5 ml monoliths, which we have chosen to denote 1-, 2-, 3-, 4-, 5- and6-Alu-cryo respectively (depending on the corresponding percentage ofNP loading). The small size of the monoliths was chosen to facilitatelaboratory handling and minimise volumes of As-spiked water used inthe subsequent tests. Themost optimally loaded adsorbentwas thereaf-ter screened for potential leakage of NPs as well as toxicity. Before anysamples were applied for adsorption studies, all composites were thor-oughly washed with 20 ml Milli-Q water (40 bed volumes) in order to

209L. Önnby et al. / Science of the Total Environment 473–474 (2014) 207–214

remove excess and/or unreacted monomers and NPs, following the re-sults obtained in the preparation of pAAm cryogels earlier in our group(Plieva et al., 2004).

2.3. Characterization

The swelling of the polymer was determined using the followingequation:

Swellinggg

� �¼ mwet−mdry

mdryð1Þ

where m is the mass of the composite in the wet as well as in the drystate.

To verify particle incorporation, the preparedmaterials (compositesand the pAAm cryogel) were studied using transmission electronmicroscopy, TEM, (Tecnai™ G2 Spirit BioTWIN (120 kv) from FEI Com-pany (USA)). For TEM, fixing of the sampleswas carried out over 24 h ina 0.1 M sodium cacodylate buffer (Na(CH3)2 AsO2·3H2O) containing2.5% (v/v) glutaraldehyde and 2% (v/v) paraformaldehyde. Sampleswere dehydrated with absolute ethanol (100%) in a series of differentconcentrations starting with the lower (i.e. from 50%, 70%, 96% to100%) and finally in acetone. All samples were thereafter infiltratedusing an EPON resin (EPON™ Resin 862, Momentive, USA). Infiltrationis done to ensure that the dehydrating solvent is replaced. Small thindiscs (approximately 50 nm)were finally cut from the samplemonolithand stained with lead citrate and uranyl acetate. A 50% ethanol suspen-sion of AluNPs (20 mg/l) was also included in the same sample seriesand used for particle distribution studies.

2.4. Adsorption and leakage study

All adsorption experiments were conducted in sealed plastic tubescontaining 10 ml of 5 mg/l As(V) solution for 2 h (25 °C) on a rockingtable at neutral pH (7 ± 0.5), a pH-value realistic for drinking water.Prepared nanocomposites (1-, 2-, 3-, 4-, 5- and 6-Alu-cryo) were alltested for As(V) adsorption where 4-Alu-cryo is equal to an AluNPload of 2 g/l (for 1, 2, 3, 5 and 6-Alu-cryo, the load is 0.5, 1, 1.5, 2.5and 3 g/l respectively). Loss of AluNPs in the production phase was in-vestigated on freshly thawed Alu-cryo samples (1–6% AluNPs added)by washing them individually with 20 mlMilli-Q water which was fur-ther collected and filtered through a Nuclepore™ filter (Whatman®,Track-Etched Membranes, polycarbonate, pore size 0.05 μm). Bothinitial and final weights were determined on dried (at 110 °C for 2 h)filter papers and the final wash was squeezed out of the Alu-cryo sam-ples, to remove water held within the porous network. The 4-Alu-cryocomposite was thereafter evaluated for NP-leakage. Here, a Milli-Qwater volume of 0, 500, 1000, 2000 and 4000 ml was allowed to passthrough a 0.5 ml monolith of 4-Alu-cryo at a flow rate of 0.2 ml/min(giving water: nanocomposite ratios of 0, 1000, 2000, 4000, and8000). Both initial and final weights were documented for the treated4-Alu-cryo. Adsorption tests were also performed after the leakagestudy. The concentration of As(V) was determined using graphite AAS(AnalytikJena 700, Germany). Arsenic concentrations presented are av-erage values of triplicate measurements.

2.5. Characterisation of AluNPs

Transmission electron microscopy images of AluNPs in suspensionsand the 4-Alu-cryo composite were studied regarding their size distri-bution and shape using the open software ImageJ and the function his-togram from this programme. The mean particle size (as length in nm)was also determined for each specific condition, i.e. in suspension orembedded in pAAm cryogels.

2.6. Particle induced X-ray emission studies

PIXE is an analogue to energy dispersive X-ray analysis (EDX), but iscapable of multi-elemental analysis at the ppm level. The methoddetects characteristic x-rays produced by high energy protons as theyinteract with the sample (Johansson et al., 1970). With PIXE analysis,detection of AluNPs is possible from its particulate form, unlike anICP-based technique which requires that the element be ionized priorto analysis, often achieved using a strong acid. This is difficult forAluNPs, as these are relatively resistant to acid attack and PIXE washence chosen as an analysis method.

Water samples were collected from the leakage study (ratios 2000,4000 and 8000). The water samples were thereafter filtered throughNuclepore™ filters with a pore size of 50 nm. In addition to the leakagestudies a filter sample was collected from a known concentration of30 mg/l AluNP dispersed in Milli-Q water. A blank was also collected(control), all samples were collected in duplicate. All filters wereanalysed using PIXE at Lund Nuclear Microprobe Laboratory, LundUniversity.

2.7. In vitro cytotoxicity studies

The toxicity of water filtered through the composites was tested onheterogeneous human epithelial colorectal adenocarcinoma cells(Caco-2) from ATCC, USA. Briefly, Milli-Q water was filtered throughthe most optimal cryogel, the hydrated 4-Alu-cryo nanocomposites(10 mm diameter × 14 mm length) at a flow rate of 0.4 ml/minand 10 ml fractions of the filtrate were collected at different intervals(0–10, 10–20, 20–30, 30–40, 40–50, 490–500, 690–700, 840–850and 990–1000 ml) in non-sterile conditions. The nanocompositesused tofilter thewater for toxicity studieswere never pre-loadedwith ar-senic. The filtrates and controls (non-filtered water) were lyophilised,UV-sterilised for 1 h and reconstituted in 10 ml sterile culture mediumdescribed in Section 2.1.

The effect of thewaterfiltratewas assessed on the viability of Caco-2cells. Cells were first cultured in filtrate-free medium to 85% confluence(5 days), with media replacement every 3 days. The cells were trans-ferred into 96 cell well plates at a seeding density of 2000 cells/well infiltrate-treated culture medium (0.2 ml) and incubated (37 °C, 5% CO2,95% humidity) for 24 h.

The culture medium was then collected, centrifuged to remove de-bris and stored for cytotoxicity analysis. Cells that remained in cultureplates were given fresh (untreated) culture medium immediately. Theamount of live cells, as a measure of cell viability, was then assessedby studying the ability of themitochondrial enzymes to reduce tetrazo-lium dye (3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium) inner salt, (MTS) to soluble formazan(purple colour) in culture medium using the CellTiter 96® AQueousNon-Radioactive Cell Proliferation Assay following kit instructions forthe MTS assay (Promega, 2001). The amount of formazan produced inculture was determined spectrophotometrically at an absorbancewavelength of 490 nm. The control cultures for the MTS assay weretreated with culture medium without filtered water.

The number of dead cells, as a measure of cytotoxicity, was thenassessed by determining the amount of intracellular lactic dehydroge-nase (LDH) released into the culture medium collected prior to theMTS assay. The amount of LDH release in culture is directly proportionalto the number of dead cells and this was performed using CytoTox 96Non-Radioactive Cytotoxicity Assay Kit following themanufacturer's in-structions (Promega, 2001). Briefly, the culture medium was collected,centrifuged to remove debris and the amount of LDH in the supernatantdetermined spectrophotometrically at an absorbance wavelength of490 nm. The positive control cultures for the LDH assay had all thecells deliberately lysed to release total LDH and the negative controlcultures were given culture medium not treated with water filtrate(Promega, 2012).

210 L. Önnby et al. / Science of the Total Environment 473–474 (2014) 207–214

2.8. Analysis of acrylamide in the filtrate water

AAm was determined in fractions of water filtered through thesynthetized cryogels by liquid chromatography–mass spectrometry(LC–MS) working in full scan mode (scan range 35–300 m/z). Isotopicdilution was used as quantification strategy; d3-atazine (25 μl) inmethanol, at a concentration of 8 mg/l, was spiked to aliquots (1 ml)of water filtrate fractions at concentration level of 200 μg/l. The systemused for the analysis was an HPLC Agilent 1200 (Agilent technologiesInc., USA), coupled to an ion trap model HCT Esquire (Bruker Daltonics,UK), which had an off-axis electrospray ionisation source. The optimalconditions of the electrospray were nebuliser gas (N2) 55 psi, dry gas(N2) 12 l/min and 360 °C as dry temperature. The voltages applied forthe optimal transmission of ions in the working scan range werecapillary−3516 V; end plate offset−500 V; skimmer 31.7 V; capillaryexit 62.3 V; octapole 1 DC 15.60 V; octapole 2 DC 0.65 V; octapoleRF 62.3 V; lens 1, −1.7 V; lens 2, −41 V trap drive −41.0 V. Themaximum accumulation time was 50 ms and the chromatographicseparation was carried out with an Ascentis Express® C18 column(2.1 × 100 mm) with porous shell particles of 2.7 μm from Supelco(Sigma-Aldrich, UK). The separation was carried out under isocraticconditions of 5% methanol in water at 0.1 ml/min and 40 °C. The injec-tion volume was 5 μl, which led to a detection limit (signal-to-noiseratio of 3) of AAm in filtrate water of 2 μg AAm/l. When higher sensitiv-ity was needed, a sample injection volume of 75 μl made it possible todetect AAm in water at 0.1 μg/l with a peak still separated from thedead volume.

3. Results and discussion

For the composite adsorbent preparation, a range of different AluNPloadingswas screened in order tofind the optimal AluNP loadingwithinthe polymer (cryogel) support as well as a stable adsorption perfor-mance for As(V). Results indicate that higher particle incorporationresulted in higher dry weights. For the highest mass load of AluNPs,the polymer weight is more than five times higher than for the 1-Alu-cryo (see Table 1). The swelling degree decreased by more than 50%from an initial 21.9 g/g in the non-loaded pAAm cryogel to 9.7 g/g forthe highest AluNPs loading, which is reasonable considering that withmore AluNPs the polymer incorporates a metal oxide mass whichis not able to adsorb water and hence affects swelling negatively(i.e. swelling decreases). For the calculation of swelling degree, thedry weight, which is the denominator in the formula, increases,explainingwhy the swelling degree decreases. Similar results regardingreduction in swelling withmore particle incorporation in cryogels havebeen reported previously (Koç et al., 2011).

The polymer support without particles has a porosity close to 90% asmost of thematerial is comprised of pores (Ozmen andOkay, 2008). TheNPs are physically embedded in the thin polymer “walls” within the

Table 1Characteristics of the polymer support (pAAm cryogel) and n-Alu-cryogels with differentAluNP loads (n = 1–6%, w/v).

Material AluNPs% (w/v)

Loss of AluNPs (mg) afterthawing, (relative loss %)b

Massincorporated(mg)c

Swelling(g/g)

pAAm cryogel 0 na.a 0 21.9 ± 1.21-Alu-cryo 1 0.20 ± 0.17 (4.0 ± 3.5) 4.4 ± 0.2 19.1 ± 0.62-Alu-cryo 2 0.27 ± 0.29 (2.7 ± 2.9) 6.9 ± 0.6 16.6 ± 0.13-Alu-cryo 3 0.65 ± 0.49 (4.3 ± 3.3) 11.7 ± 0.1 13.7 ± 0.54-Alu-cryo 4 0.40 ± 0.07 (1.8 ± 0.3) 16.7 ± 0.2 12.8 ± 0.05-Alu-cryo 5 0.50 ± 0.10 (2.0 ± 0.4) 21.7 ± 0.3 9.6 ± 1.16-Alu-cryo 6 0.87 ± 0.38 (3.5 ± 1.5) 23.6 ± 0.8 9.7 ± 0.2

a na., not applicable.b Incorporation loss is expressed as loss in mass vs initial mass added of AluNPs.c It is assumed that the weight of the polymer is the same in all composites as the aver-

age for the pAAm cryogel.

composite and at least part of the active NP surface is covered by thepolymer. Adsorption will therefore be driven by direct surface contactand also short diffusion pathways through the polymer (i.e. where theparticle surface is covered by the polymer film).With more mass incor-porated, however, the swelling degree is decreased and with the NPsembedded in the walls, it is reasonable that a limit for mass incorpora-tion is obtained at 6-Alu-cryo (Table 1). We have chosen the optimalnanocomposite for further study based on AluNP load (mass input)versus pollutant removal, i.e. where high adsorption of As(V) is gainedwith minimal incorporation of AluNPs. Physically embedding particlesin cryogels may generate practical limitations in terms of particle load-ing and adsorption kinetics and capacity, and different types of incorpo-ration such as generating surface coatings of NPs could potentiallyimprove NP reactivity and thereby improve adsorption capacity, asshown for chitosan spheres coated with zero valent iron NPs for As(III)adsorption (Gupta et al., 2012). This may however increase the risk ofNP wash-out during operation.

For the preparation of the nanocomposites, a highermass incorpora-tion can result in higher probability of collapse or leakage of particlesimmediately after the production, i.e. during thawing. The mass loss ofAluNPs (in mg) indicates that the loss increased with a higher portionof incorporated AluNPs; for 6-Alu-cryo, 0.87 mg is lost whereas only0.2 mg is lost for the 1-Alu-cryo (Table 1). If the incorporation loss(relative loss) is taken into account, the most optimal loading isobtained somewhere between 4-Alu-cryo and 5-Alu-cryo (Table 1).

The adsorption of As(V) indicated that the 4-Alu-cryowas optimal interms of added mass of AluNP versus As(V) removal (Fig. 1). Overall,higher As(V) removal is obtained with higher loads of AluNPs up to4% loading, with a slight drop observed at 5% and greater loadings.This slight reduction in As(V) removal is likely due to the tendency ofNPs to cluster and aggregate (due to gravitational effects during cryogelfreezing) at the base of the manufactured composite, which is visuallyapparent in the production phase and occurs particularly in the Alu-cryowith higher NP loadings (especially the 5- and 6Alu-cryo). In a pre-vious study, we examined As(V) removal using “free”Alu-NPs aswell asAluNPs incorporated into cryogels (Onnby et al., 2012) and, for the pre-vailing conditions, there are no significant differences in As(V) removalbetween the 4-Alu-cryo and “free” AluNPs added in the same relativeproportion (2 g/l, Fig. 1). This indicates that the embedding of the parti-cles in the cryogel did not negatively affect the contact between the sur-face of the NPs in the nanocomposite and the target ions in the solution.It should be noted that the removal is expressed as a percentage andtherefore does not take into consideration that the amount of AluNPdiffers between the six different Alu-cryo samples. The highest uptakeof As(V) expressed per mass unit of AluNP incorporated is obtained by1-Alu-cryo for this initial concentration. It should be stressed that this

Fig. 1. Removal of As(V) compared between free AluNPs in suspension, the pAAm cryogel(control) and for each respective Alu-cryo nanocomposite (n-Alu-cryo) where n denotesweight percentage of AluNPs added in the cryogel, (n = 1–6).

211L. Önnby et al. / Science of the Total Environment 473–474 (2014) 207–214

removal capacity is the maximum obtained at this specific concentra-tion. Not surprisingly, for a lower amount of NPs (as in the case of 1-,2- and 3-Alu-cryo respectively), the removed As per mass of NP ishigher. The slight difference obtained is due to the loss of NPs in theproduction step (see Table 1) since only ca 85% of the initial AluNPsstay in the polymer support after production. The loss in capacitybetween embedded NPs in a cryogel and the same “free”NPs in suspen-sion were also studied by (Savina et al., 2011). In that study, it was seenthat the adsorption capacity for incorporated α-Fe2O3 and Fe3O4 NPs ina cryogel support dropped to approximately one third of the corre-sponding capacity of the free particles when tested for As(III) removal.For filtration applications, packed beds with nanoparticles have narrowchannels that hinder the circulation of water and eventually becomeblocked. In contrast, the use of cryogels, with a macroporous matrixand with incorporated NPs, facilitates the flow-through of water. In-deed, for water treatment applications, e.g. use in membranes, cryogelscome with benefits such as extremely low back pressures (Elcik et al.,2013). Due to the accessibility of the NPs in the matrix, pollutants canbe removed very effectively.

To investigate the distribution of the AluNPs in the pAAm cryogel,samples were studied and compared with the non Al-loaded pAAmcryogel via TEM, as shown in Fig. 2. The polymer wall of the pAAmcryogel without particles is shown in Fig. 2a. The physical incorporationof AluNPs can be confirmed by comparing the unmodified pAAmcryogelwith 4-Alu-cryo (Fig. 2a and b). The AluNPs appear as agglomer-ates (Fig. 2b), which are widely distributed across the polymer withvaried density and coverage. At higher magnification, the individual ag-glomerates show crystalline AluNPs, appearing as black needle-likeareas (Fig. 2c). With these small particles, however, we were able to

Fig. 2. TEM image of the polymer support (a), the needle-shaped AluNPs studied from suspensi(c) are shown in (d).

obtain a wide distribution of the NPs over the polymer surface, whichwill be beneficial for adsorption performance. This effect is also seenfor catalytic applications, where it was shown that an improved disper-sion of Pd/Fe NPs, obtained via composite supports of cellulose, couldyield more than 15 times faster destruction of chlorinated solventscompared to non-supported NPs, which was partly explained by an im-proved suspension dispersion (He et al., 2010).

Based on supplier data, the AluNP particle size is estimated to beN50 nm, but as the size of NPs is dependent on their surrounding condi-tions, we estimated the particle distribution for AluNPs in suspensionand for AluNPs in 4-Alu-cryo from TEM data (ref. Section 2.5), asshown in Fig. 2d. The mean particle size obtained was larger for AluNPsin suspension (162.4 ± 53 nm) as compared to AluNPs embedded inthe pAAm cryogels (141.8 ± 48 nm). As expected, under both condi-tions the AluNPs have formed larger aggregates, as the particle sizehas increased almost three-fold compared to the supplier data. Theparticle size distribution is greater for the embedded AluNPs than forthe particles in suspension, indicating that AluNPs do, to some extent,de-aggregate during the embedding process. Considering the strategyused for loading particles to the cryogel support, aggregate formationis likely to take place. This could be avoided by a more sophisticatedapproach, e.g. by surface coating or a stabilizing agent in the AluNP +pAAm suspension prior to freezing and polymerization of the material.More even and fine dispersions of particles could possibly improve theavailable surface area and enhance the adsorption performance. Evenlydistributed NPs of TiO2 attached to the surface of a glass support haveshown enhanced activity in photocatalytic reactions (Yu et al., 2002).

Due to concerns over potential environmental effects from particleleakage, the possible loss of NPs from the Alu-cryo composite was

on (b) and from the 4-Alu-cryo composite (c). Particle distributions obtained from (b) and

Table 2Loss of weight from the 4-Alu-cryo composite for different water: adsorbent ratiosexpressed in mass (mg) as well as in percentage loss (%).

Water:nanocompositeratio (v:v)

Weight loss(mg)

Loss(%)

0 0 0.7 ± 0.91000 0.70 ± 0.29 1.1 ± 0.72000 0.30 ± 0.14 0.5 ± 0.34000 0.30 ± 0.14 0.5 ± 0.38000 0.60 ± 0.57 1.4 ± 0.1

212 L. Önnby et al. / Science of the Total Environment 473–474 (2014) 207–214

studied by evaluating the effect caused by passing Milli-Q water throughthe composite, mimicking its operation as a water filter. Differentvolumes of water were washed through the composite, representingwater to nanocomposite volume ratio between 0 and 8000. If the mea-sured As(V) adsorption capacity was changed after water flow-through,it was assumed to be due to particle loss. We could not detect any signif-icant losses of material from the flow-through study based on residualadsorption (Fig. 3). In more practical terms, if we assume:

(a) that themaximumadsorption capacity of theAluNP-cryogel com-posites follows our previous results, i.e. adsorption capacity =20.3 mg As(V)/g AluNP (Onnby et al., 2012), and

(b) that the adsorption capacity is independent of initial concentra-tion, then for groundwater containing e.g. 22 μg/l of As, whichlies within reasonable As-concentrations (Nordstrom, 2002),treated with 1 g of AluNPs supported in 25 ml pAAm-cryogels(4% AluNPs (w/v)), with a permissible As limit of 10 μg/l, the re-sults indicate that 200 l of water could theoretically be treatedwithout any particles leaching (i.e. where the water volume tonanocomposite ratio is 8000). Even after 200 l the adsorbent isonly loaded to approximately 12% of the maximum adsorptioncapacity, as the 25 ml composite of 4-Alu-cryo can treat a watervolume about 1700 l under these conditions, which is equivalentto awater volume for a family of five for 170 days (averagewaterconsumption assumed to be 10 l/day).

In addition to studying the retained capacity for adsorption, the lossof NPs was also investigated via weight loss of the 4-Alu-cryo materialafter the flow-through experiment had been conducted. For all exam-ined ratios, no evidence was found for any significant loss of NPs usingthemeasured composite weights (see Table 2).With the standard devi-ation taken into account, no significant changes in weight were seencompared to the control. These data correspond to the data from thecapacity measurements (Fig. 3), which suggest no, or at worst onlyslight, losses of NPs.

Possible AluNP loss from the cryogels during the filtration experi-ments was also examined using PIXE multi elemental analysis. Filtersamples were collected for water:adsorbent ratios of 0, 2000, 4000,and 8000, and from a known AluNP suspension of 30 mg/l. Whilethere is a statistically significant difference (one-sided T-test, n = 1,p b 0.05) in Al detected between (a) the known (30 mg/l) sample and(b) the filters from the 4-Alu-cryo composite, no statistically significantdifference was found between the PIXE responses for the differentwater:adsorbent ratios examined (Fig. 4). These results thereforeshow that (a) PIXE is a viable tool for the detection of AluNPs from thefilteredwatermass, due to the increased response of the filtered samplewith known (i.e. 30 mg/l) AluNP concentration, and (b) there is no sig-nificant increase in elemental Al in any of the filtrates from the AluNPcryogels compared to the ratio 0 (which represents conditions on

Fig. 3. The capacity of As(V) removal (mg As/g AluNP) following repeated wash-throughwith Milli-Q water, for water:nanocomposite volume ratios (v:v) between 0 and 8000.

commencement of the filtration study), i.e. there was no detectable re-lease of AluNP from the 4-Alu-cryogel for any of the studied volume ra-tios (water: nanocomposite).

Toxicity studies on water filtered through the composites were car-ried out using epithelial cells because this is the first cell type thatwouldcome into contactwith particles released from adsorbents used inwaterpurification (i.e. in the case of NP leaching and subsequent ingestion).The MTS assay showed a significant reduction in the viability of thecells cultured in the medium treated with the first 10 ml filtrate com-pared to the control sample (p b 0.05) (Fig. 5a). However, all othertreatment groups (i.e. volume fractions) are not significantly differentfrom each other or from the control (p N 0.05). Although the LDHassay result shows a 50% toxicity level as compared with the positivecontrol (Fig. 5b), all the fraction treatment groupswere not significantlydifferent from the negative control (cell in filtrate-free cultures). TheLDH assay indicates leakage of cytosolic material from the cells whilecell viability is a more definite measure. However, if the viability testis too short, cells that are inhibited can be judged as being dead whenthey are simply inhibited and have the potential to recover. A drop inviability was observed but at the same time no increase in cell death.The reduction in cell metabolism was only observed after exposure tothe first (and to a lesser degree to the second) 10 ml fraction and isfollowed by a continued increase in viability.

The potential presence of unreacted acrylamide in eluates fromthe optimal cryogel 4-Alu-cryo was examined by LC–MS using d3-acrylamide as an internal standard. Acrylamide concentrations weredetermined not only in the eluted fractions for which cytotoxicity wasassayed (Fig. 5b) but also in fractions from the initial 20 ml of Milli-Qwater used to wash the freshly prepared Alu-cryo nanocomposites(see Section 2.2). Concentrations of acrylamide gradually decreasedduring initial washing of the cryogel (Table 3, prewash volume data),but they remained above detection limits (0.1 μg/l) throughout thepre-washing phase as well as for the following 20 ml of filtrate, the lat-ter corresponding to thefirst 2 volume fractions seen in Fig. 5 (0–10 and10–20 ml) for which toxicity was observed. Hence, the presence of ac-rylamide in thewater assayedwith the cells could at least partly explainthe reduction of viability observed in the MTS assay for the first twovolume fractions (see Fig. 5b). The EU 98/83 Drinking Water Directiveregulates the level of acrylamide (which is a probable carcinogenaccording to the International Agency for Research on Cancer (IARC,1994)), to amaximum level of 0.1 μg/l. The toxicity of AAm is well stud-ied and findings show for instance that \SH groups and \NH2 groupson proteins and nucleic acids are modified and changed after AAm ex-posure, as evidenced from both in vitro and in vivo studies (Friedman,1967, 2003). In addition, cell changes have earlier been confirmed andlinked to alterations in glutathione content (Odland et al., 1994), whichcould be one reason for the toxicity induction obtained in this study.Therefore, in this regard, washing the cryogels with 80 bed volumesprior to use is recommended.

Materials containing nanoparticles need to be carefully assessedespecially if their use could cause direct exposure and contact betweenNPs and the environment. However, there is no official guideline (todate) on testing the toxicity of such products. We chose to investigateany possible leakage of AluNPs from our Alu-cryo nanocomposite filtersfrom different perspectives: by assessing the nanocomposite itself, after

Fig. 4. The relative amount of elemental Al released (A.U.) during flow-through studies onthe cryogels at different water to nanocomposite volume ratios (v:v). The known amountof AluNPs was 30 mg/l in suspension.

Table 3Quantification of AAm in water filtered with 4-Alu-cryo by LC–MS. Thedetermination has been carried out in three independent cryogels(n = 3). The filtration flow rate was 0.4 ml/min.

Water filtrate AAm in the filtrate (μg/l)

Prewash volume (ml)0–5 205 ± 69.96–10 31 ± 15.511–15 14 ± 3.816–20 13 ± 2.0

Filtrate fraction (ml)0–5 16 ± 1.46–10 10 ± 2.911–20 6 ± 2.4N20 bLODa

a LOD = 0.1 μg/l.

213L. Önnby et al. / Science of the Total Environment 473–474 (2014) 207–214

production and after use, as well as the water which has flowedthrough the nanocomposite. A range of methods and techniqueswas used (e.g. weight loss, SEM, TEM, PIXE, in vitro toxicity studies)in order to thoroughly investigate the nanocomposite. Given theconcentrations of AluNPs incorporated in our materials, we havebeen able to show, for the first time, that no AluNP losses could bedetected from these AluNP composites using these techniques. Theresults from the toxicity study indicate a change between an inhibitionphase (in the first volume fractions) and a recovered phase (for the lat-ter volume fractions). The toxicity observed in the first 40 bed volumes(which was obtained after a previous wash with 40 bed volumes) canbe explained to some extent by the levels of unreacted monomerAAm being eluted. Based on this, future studies could examine replace-ment of the current pAAm cryogel with amore environmentally friend-ly polymer, provided that its stability against NP leakage is retained.

Fig. 5. (a)MTS viability assay and (b) LDH cytotoxicity assay obtained for different volumefractions. Negative control in (a) and in (b) is the assay tested in the absence of testmaterial, whereas the positive control in (b) defines dead cells. Presented data ismean ± standard deviation, n = 6.

4. Conclusion

From a range of different AluNP loads, we demonstrate that 4-Alu-cryo (i.e. 4% NP loading) is the optimal NP load for AluNPs embeddedinto a pAAm cryogel. Here, little loss of NPs during production, and neg-ligible (undetectable) loss during application, is obtained combinedwith confirmed As(V) removal. With higher AluNP incorporation wealso show that the relative loss of AluNPs increases and the swelling de-gree decreases. Additionally, we do not find any indications of AluNPleakage, either from determining the weight loss after production orfrom examining filter papers used during the thawing of the materials.More sophisticated analysis using PIXE mass analysis confirmed thesefindings. A loss of NPs could be one parameter increasing or inducing tox-icity. The high amount of water filtered through the nanocompositeshowedno evidence for NP presence andno evident cell death in exposedCaco-2 cells was observed, but cell viability was slightly affected sinceAAm traces were found in the first 40 bed volumes of the composite elu-ates. To further assess the stability of thesematerials, particularly in termsof NP release during practical operation, their performance at higher flowspeeds and composite volumes, under more aggressive pH conditions,andwith higher ionic strengthwaters, should be evaluated as a next step.

Conflict of interest

The manuscript is an original work of all included authors and themanuscript has not been submitted to any other journal prior to thissubmission. There is no conflict of interest including financial, personalor other relationships with other people or organizations for any of theco-authors.

Acknowledgements

The EC Marie-Curie Industry-Academia Partnerships and PathwaysProgram (grant PIAP-GA-2008-230676 CARBOSORB) is acknowledgedfor financial support.

R. Busquets acknowledges her IEF Marie Curie fellowship (Polarclean project, n274985) from the FP7 People program. Department ofNuclear Physics at Lund University is acknowledged for assistancewith PIXE-analysis. The authors also acknowledge three anonymousreviewers for their useful and thorough comments which improvedthe overall quality of the manuscript.

References

Ali I. New generation adsorbents for water treatment. Chem Rev 2012;112:5073–91.Arvidsson R, Molander S, Sandén BA, Hassellöv M. Challenges in exposure modeling of

nanoparticles in aquatic environments. Hum Ecol Risk Assess Int J 2011;17:245–62.Chowdhury UK, Biswas BK, Chowdhury TR, Samanta G, Mandal BK, Basu GC, et al.

Groundwater arsenic contamination in Bangladesh and West Bengal, India. EnvironHealth Perspect 2000;108:393–7.

214 L. Önnby et al. / Science of the Total Environment 473–474 (2014) 207–214

Elcik H, Cakmakci M, Sahinkaya E, Ozkaya B. Arsenic removal from drinking water usinglow pressure membranes. Ind Eng Chem Res 2013;52:9958–64.

Ferry JL, Craig P, Hexel C, Sisco P, Frey R, Pennington PL, et al. Transfer of gold nanoparti-cles from the water column to the estuarine food web. Nat Nanotechnol 2009;4:441–4.

Friedman M. Solvent effects in reactions of amino groups in amino acids peptides andproteins and proteins with alpha beta-unsaturated compounds. J Am Chem Soc1967;89:4709–13.

Friedman M. Chemistry, biochemistry, and safety of acrylamide. A review. J Agric FoodChem 2003;51:4504–26.

Gupta A, Yunus M, Sankararamakrishnan N. Zerovalent iron encapsulated chitosan nano-spheres—a novel adsorbent for the removal of total inorganic Arsenic from aqueoussystems. Chemosphere 2012;86:150–5.

He F, Zhao D, Paul C. Field assessment of carboxymethyl cellulose stabilized iron nanopar-ticles for in situ destruction of chlorinated solvents in source zones. Water Res2010;44:2360–70.

Hug SJ, Leupin OX, BergM. Bangladesh and Vietnam: different groundwater compositionsrequire different approaches to arsenic mitigation. Environ Sci Technol 2008;42:6318–23.

IARC. IARC monographs on the evaluation of carcinogenic risks to humans. Some tradi-tional herbal medicines, some mycotoxins, naphthalene and styrene, 60. Lyon,France: WHO; 1994389–433.

Johansson TB, Akselsson R, Johansson SAE. X-ray analysis: elemental trace analysis at the10–12 g level. Nucl Inst Methods 1970;84:141–3.

Kaegi R, Voegelin A, Ort C, Sinnet B, ThalmannB, Krismer J, et al. Fate and transformation ofsilver nanoparticles in urban wastewater systems. Water Res 2013;47(12):3866–77.

Khin MM, Nair AS, Babu VJ, Murugan R, Ramakrishna S. A review on nanomaterials forenvironmental remediation. Energy Environ Sci 2012;5:8075–109.

Kim J, Van der Bruggen B. The use of nanoparticles in polymeric and ceramic membranestructures: review of manufacturing procedures and performance improvement forwater treatment. Environ Pollut 2010;158:2335–49.

Klaine SJ, Alvarez PJJ, Batley GE, Fernandes TF, Handy RD, Lyon DY, et al. Nanomaterials inthe environment: behavior, fate, bioavailability, and effects. Environ Toxicol Chem2008;27:1825–51.

Koç İ, Baydemir G, Bayram E, Yavuz H, Denizli A. Selective removal of 17β-estradiolwith molecularly imprinted particle-embedded cryogel systems. J Hazard Mater2011;192:1819–26.

Loo S-L, Fane AG, Lim T-T, Krantz WB, Liang Y-N, Liu X, et al. Superabsorbent cryogelsdecorated with silver nanoparticles as a novel water technology for point-of-use dis-infection. Environ Sci Technol 2013;47:9363–71.

Lozinsky V. Cryogels on the basis of natural and synthetic polymers: preparation, proper-ties and application. Russ Chem Rev 2002;71:489–511.

Mohan D, Pittman JCU. Arsenic removal from water/wastewater using adsorbents—acritical review. J Hazard Mater 2007;142:1–53.

Murashov V. Comments on “Particle surface characteristics may play an important role inphytotoxicity of alumina nanoparticles” by Yang, L., Watts, D.J., Toxicology Letters,2005, 158, 122–132. Toxicol Lett 2006;164:185–7.

Nickson R, McArthur J, Burgess W, Ahmed KM, Ravenscroft P, Rahmann M. Arsenic poi-soning of Bangladesh groundwater. Nature 1998;395:338.

Nordstrom DK. Worldwide occurrences of arsenic in ground water. Science 2002;296:2143–5.

Odland L, Romert L, Clemedson C,Walum E. Glutathione content, glutathione transferase-activity and lipid-peroxidation in acrylamide-treated neuroblastoma N1E-115 cells.Toxicol in Vitro 1994;8:263–7.

Onnby L, Pakade V, Mattiasson B, Kirsebom H. Polymer composite adsorbents using par-ticles of molecularly imprinted polymers or aluminium oxide nanoparticles for treat-ment of arsenic contaminated waters. Water Res 2012;46:4111–20.

OzmenMM,OkayO. Formation ofmacroporous poly(acrylamide) hydrogels inDMSO/watermixture: transition fromcryogelation tophase separation copolymerization. React FunctPolym 2008;68:1467–75.

Plieva FM, Andersson J, Galaev IY, Mattiasson B. Characterization of polyacrylamide basedmonolithic columns. J Sep Sci 2004;27:828–36.

Promega. CellTiter 96® AQueous non-radioactive cell proliferation assay. In: Promega, ed-itor. Technical Bulletin No. 169. Madison, USA: Promega Cooperation; 2001. p. 10.

Promega. CytoTox 96® non-radioactive cytotoxicity assay. Instructions for use of productG1780. Madison, USA: Promega; 201219.

Sankar MU, Aigal S, Maliyekkal SM, Chaudhary A, Anshup, Kumar AA, et al. Biopolymer-reinforced synthetic granular nanocomposites for affordable point-of-use water puri-fication. Proc Natl Acad Sci U S A 2013;110:8459–64.

Savina IN, English CJ, Whitby RLD, Zheng Y, Leistner A, Mikhalovsky SV, et al. High effi-ciency removal of dissolved As(III) using iron nanoparticle-embedded macroporouspolymer composites. J Hazard Mater 2011;192:1002–8.

Smith AH, Lingas EO, Rahman M. Contamination of drinking-water by arsenic inBangladesh: a public health emergency. BullWorld Health Organ 2000;78:1093–103.

Teeguarden JG, Hinderliter PM,OrrG, Thrall BD, Pounds JG. Particokinetics in vitro: dosimetryconsiderations for in vitro nanoparticle toxicity assessments. Toxicol Sci 2007;95:300–12.

WHO. WHO environmental health criteria 224. Arsenic and arsenic compounds. Geneva,Switzerland: World Health Organization; 2001.

Winkel L, Berg M, Amini M, Hug SJ, Annette Johnson C. Predicting groundwater arseniccontamination in Southeast Asia from surface parameters. Nat Geosci 2008;1:536–42.

Yang L, Watts DJ. Particle surface characteristics may play an important role in phytotox-icity of alumina nanoparticles. Toxicol Lett 2005;158:122–32.

Yu JC, Yu J, Zhao J. Enhanced photocatalytic activity of mesoporous and ordinary TiO2 thinfilms by sulfuric acid treatment. Appl Catal Environ 2002;36:31–43.

Zhu JH, Wei SY, Gu HB, Rapole SB, Wang Q, Luo ZP, et al. One-pot synthesis of magneticgraphene nanocomposites decorated with core@double-shell nanoparticles for fastchromium removal. Environ Sci Technol 2012;46:977–85.