Graphene membranes as novel preconcentration platforms for chromium speciation by total reflection X-ray fluorescence

Abstract

Fabrication of unmodified graphene membranes for their application as selective sorptive platforms of hexavalent chromium [Cr(VI)] is described for the first time. Multilayer graphene membranes are synthesized by drop-casting of graphene oxide (GO) onto a glass substrate followed by mild thermal reduction. As-prepared membranes are formatted to fit the measurement area of total reflection X-ray fluorescence (TXRF). Structural and morphological characterization by atomic force microscopy (AFM) and X-ray photoelectron spectroscopy (XPS) shows that graphene membranes are 122 nm height and contain non-reduced functional groups resulting in lattice defects. Adsorption isotherm models and characterization by time-of-flight secondary ion mass spectrometry (TOF-SIMS) indicate that adsorption sites on graphene membranes are uniformly distributed and bind Cr(VI) as a monolayer, both by electrostatic interaction and chemisorption. Graphene membranes display high flexibility and become conical-shaped when immersed into stirred liquid samples. When combining graphene membrane preconcentration and TXRF, a detection limit of 0.08 μg L⁻¹ Cr(VI) is obtained. Repeatability expressed as relative standard deviation is 3% (N = 5). Two certified reference materials, i.e. CASS-4 seawater and NWTM-27.2 lake water, are used for testing accuracy. The proposed method is simple, solvent-free and sensitive, being suitable for Cr speciation in water including high salinity samples.

---

1. Introduction

Graphene is a planar sheet of sp²-bonded carbon atoms packed in a honeycomb crystal lattice, which has attracted much attention in the field of material science due to its electrical, mechanical, thermal and structural properties.^1–3 The most used method for the production of graphene in large-scale is the Hummers' method based on the oxidation of graphite.⁴ Graphene oxide (GO) nanosheets can be easily prepared following exfoliation of graphite oxide. GO nanosheets can be reduced (chemical or thermally) in order to obtain graphene films.^5,6

In the last few years, GO and graphene have been reported as new sorptive materials for extraction and preconcentration.^7,8 In contrast to other carbon-based nanomaterials, e.g. carbon nanotubes (CNTs), GO and graphene display sorptive properties in both surfaces of the planar sheet resulting in larger specific surface area and higher adsorption kinetics. Besides, unlike CNTs, chemical synthesis of GO and graphene from graphite does not require metal catalysts, which prevents sorbents from metal impurification. Both GO and graphene provide plenty of sites for functional modification.⁹ For sorption of trace metal ions, GO is usually preferable due to the abundance of functional groups that can interact with analytes by a variety of mechanisms (electrostatic interactions, cation–π interactions, etc.).¹⁰ GO can be used for in-column extraction, e.g. SiO₂@GO composite can be loaded into a glass column or coated hollow fibers.^11,12 However, low flow-rate is required in order to avoid overpressure problems and/or losses of the solid phase. Thus, most reported applications use dispersive micro-solid phase extraction (D-μ-SPE) involving GO nanosheets as nanocomposites (e.g., Fe₃O₄@SiO₂@GO and Fe₃O₄@GO),^13,14 which are further modified with specific functional groups (e.g. 3-mercaptopropyltrimethoxysilane-modified GO),¹⁵ or unmodified.¹⁶ After extraction, GO can be separated by filtration, aggregation by adding NaCl or by applying an external magnetic field. Graphene nanosheets have also been reported as a new solid phase for D-μ-SPE of trace metals. Given the extremely hydrophobic properties of graphene, complexation of metal ions with suitable chelating agents, e.g. ammonium pyrrolidinedithiocarbamate (APDC), is required for efficient sorption.^17,18 It should be noted that before measurement, GO and graphene nanosheets have to be isolated from the sample solution. Losses during separation (filtration, centrifugation or decantation) can occur due to the small size of nanosheets.

New formats of carbon nanomaterials used as preconcentration platforms, such as easy-to-handle graphene membranes, would avoid centrifugation or filtration steps thus simplifying sorption procedures. In addition, graphene membranes should be well suited for direct measurement of metal species by total reflection X-ray fluorescence (TXRF) without the need for an elution step.

Trace metal speciation in environmental and biological samples is required since metal bioavailability and toxicity depends on the different chemical forms. Cr occurs primarily in two oxidation states, namely, trivalent chromium [Cr(III)] and hexavalent chromium [Cr(VI)]. Cr(III) is considered to be essential for carbohydrate and lipid metabolism, its toxicity being very low.¹⁹ On the contrary, Cr(VI) is highly toxic for humans and it is considered to be a strong carcinogen.²⁰ Cr speciation can be performed by a variety of techniques. High performance liquid chromatography (HPLC) coupled to atomic spectrometry detection provides comprehensive information on the presence of Cr species in a variety of matrices.²¹ For speciation of Cr(III) and Cr(VI), derivatization with ethylenediaminetetraacetic acid followed by anion-exchange chromatography is commonly performed.²² Apart from hyphenated techniques, the development of non-chromatographic approaches for chemical speciation is attractive since they can provide simpler and less expensive ways to perform Cr speciation. Among the different strategies, solid phase extraction and microextraction are the most used techniques.²³ Selective extraction can be achieved by ion-exchange resins,²⁴ molecularly imprinted polymers (MIPs),^25,26 or by complexing chromium species with specific organic groups before extraction, e.g., complexation with APDC before extraction onto CNTs,²⁷ or complexation with ethylenediaminetetraacetic acid (EDTA) before extraction onto silver nanoparticles.²⁸

In this work, a novel method for the preparation of multilayer graphene membranes by drop-casting of GO onto a glass substrate followed by mild thermal reduction is described. As-prepared graphene membranes are applied as preconcentration platforms for the detection of Cr(VI) and total Cr at trace level in waters by TXRF. Morphology and structure of the novel sorptive platforms are fully characterized and sorption mechanisms for Cr(VI) outlined.

2. Experimental section

2.1. Apparatus

TXRF measurements are carried out by means of a TXRF spectrometer model S2 Picofox™ (Bruker AXS Microanalysis GmbH, Berlin, Germany). The spectrometer is equipped with a molybdenum X-ray tube operating at 50 kV, 1000 μA (air-cooled, max. power 50 W), a Si-drift detector (area 10 mm², resolution < 160 eV (Mn K_alpha)) and a multilayer Ni/C monochromator (17.5 eV, beam rectangular shape about 8 × 0.1 mm²). TXRF spectra are collected within a counting time of 500 s.

A PowerPac™ basic power supply (programmable 10–300 V, 4–400 mA) equipped with a Sub-Cell® GT (Bio-Rad, California, USA) was used for gel electrophoresis. Graphene oxide aliquots (5 μL) are loaded in the agarose gel 0.8% (m/v) buffered at pH 1.5. Electric field with a voltage programmed at 90 V was applied for 30 min.

An Agimatic-N magnetic stirrer with heating (550 W) equipped with an electronic heater energy control from 0–100% (JP Selecta, Barcelona, Spain) was used for membrane preparation. Heater energy is programmed to 15% (T ≈ 100 °C) for thermal reduction of GO deposited onto glass substrates.

A JP Selecta model Ultrasons-1L ultrasonic bath (Barcelona, Spain) with a maximum output power of 50 W and an operating frequency of 20 kHz was employed for ultrasound-assisted exfoliation of graphite oxide.

A field emission transmission electron microscope (FETEM) model JEOL JEM-1010 (JEOL, Seoul, Korea) operating at an acceleration voltage of 100 kV was used for obtaining TEM images. A 10 μL aliquot of freshly prepared GO colloid was deposited onto a copper mesh cover (with a film of carbon/formvar) and left to dryness in a laminar flow hood.

A high resolution stylus profilometer (vertical resolution of 1 Å) model Dektak XT (Bruker GmbH, Berlin, Germany) and an AFM instrument model Multimode 8 (Veeco, Digital Instruments, Santa Barbara, CA) equipped with Nanoscope V controller were used for obtaining information about the membrane surface morphology, roughness and thickness. For XPS measurements, a Thermo Scientific K-alpha ESCA instrument (Thermo Scientific, UK) equipped with and aluminum K-alpha 1,2 monochromatized radiation X-ray source (1486.6 eV) was used. For XPS, AFM and profilometry measurement, samples were deposited onto pre-cleaned silicon wafers (orientation 〈111〉, type P, roughness 2 nm) and left to dryness in a laminar flow hood. Samples for XPS characterization were stored under vacuum to avoid contamination by atmospheric oxygen.

A time-of-flight secondary ion mass spectrometry (TOF-SIMS) instrument model TOF.SIMS⁴ (IonTof, Münster, Germany) was used for obtaining information about chromium distribution onto the graphene membrane. Bismuth liquid metal ion gum mode (BiHCBU) at 25 kev pulsed beam was used.

2.2. Reagents and chemicals

All chemicals were of analytical reagent grade. High-purity deionised water obtained from an Ultra Clear™ TWF EDI UV TM water system (Siemens, Barsbüttel, Germany) was used throughout. A 1000 mg L⁻¹ Ga stock standard solution (Fluka, Buchs, Switzerland) was used to prepare the internal standard solution (1 mg L⁻¹ Ga in 1.5% v/v HNO₃) for TXRF measurements.

Graphite powder (grain size < 75 μm) 99% (Scharlau, Barcelona, Spain), KNO₃ 99% (Probus, Badalona, Spain), KMnO₄ 99.5% (Analyticals Carlo Erba, Milano, Italy), H₂O₂ 30% w/w (Merck, Darmstadt, Germany) and H₂SO₄ 95% w/v (Prolabo, Fontenay-sous-Bois, France) were used for synthesis of GO.

HNO₃ 65% w/v (Prolabo, Fontenay-sous-Bois, France) was used to adjust the pH of the extraction medium to 1.0. Stock standard solutions of 1000 mg L⁻¹ Cr(III) and Cr(VI)were prepared by dissolving 6.5 mg of Cr(NO₃)₃·9H₂O 99% (Aldrich, Steinheim, Germany) and 0.14 g of K₂Cr₂O₇ 99.5% (Prolabo, Fontenay-sous-Bois, France), respectively, in 50 mL of ultrapure water. Diluted working standard solutions were prepared fresh daily from the stock solution.

The following reagents were employed for studying interferent effects: MgCl₂·6H₂O 98% (Prolabo, Fontenay-sous-Bois, France), Na₂SO₄ (Probus, Badalona, Spain), KH₂PO₄ 99.5% (Panreac, Barcelona, Spain), Na₂CO₃ anhydrous 99.5% (Panreac, Barcelona, Spain), KNO₃ 99% (Probus, Badalona, Spain) and humic acid (Fluka, Buchs, Switzerland). Ammonium pyrrolidinedithiocarbamate (APDC) 98% (Fluka, Buchs, Switzerland) was used for sorption experiments of the Cr(VI)-PDC chelate onto graphene membranes.

Agarose standard low EEO (electroendosmosis) (Ecogen, Barcelona, Spain) was used for gel electrophoresis experiments.

Certified reference materials CASS-4 coastal Atlantic surface seawater (National Research Council of Canada) and NWTM-27.2 fortified lake water (National Water Research Institute of Canada) were used for testing accuracy of the method.

2.3. Graphene membrane synthesis

First, graphite oxide was synthesized by Hummers' method with few modifications.⁴ Briefly, graphite powder (1 g), KNO₃ (0.55 g), concentrated H₂SO₄ (25 mL) and KMnO₄ (3 g) were mixed and stirred for 30 min at 0 °C. As the chemical oxidation proceeded, there was a change in color from black to brownish grey. Then, ultrapure water (75 mL) was added and stirring was maintained for 30 min at ambient temperature. Finally, H₂O₂ 3% (v/v) was dropped into the reaction medium to reduce the residual permanganate. Upon addition of H₂O₂, the reaction medium color turned bright yellow. Synthesized graphite oxide was separated by centrifugation. The supernatant was discarded and the yellowish-brown solid (graphite oxide) was washed with ultrapure water. Then, GO nanosheets were obtained following ultrasound-assisted exfoliation (1 h) of graphite oxide. After that, the solid was separated by centrifugation (7800 rpm, 1 h) and the supernatant (GO, light yellow) was collected for membrane preparation.

For membrane preparation, thermal reduction was followed. In order to obtain 10 mm diameter membranes, glass substrates (Ø 10 mm) are used as support for preparation. First, glass substrates were cleaned with propanol and acetone under ultrasonic irradiation and dried. After that, they were heated at 95 °C in a hot plate. Drop-casting method was used to prepare membranes. For this, three aliquots of 100 μL GO were added onto the pre-heated glass substrates, water was evaporated and color of the GO film on the glass substrate changed from yellow to black, thus indicating the formation of a multilayer graphene membrane. Since graphene has hydrophobic behavior in contrast to glass substrates, membranes can be easily separated by rinsing with ultrapure water.

2.4. Procedure for Cr(VI) sorption onto graphene membranes and TXRF analysis

For sorption of Cr(VI), a graphene membrane is suspended in an extraction vessel containing 50 mL of sample/standard solution with a composition of 1.5% (v/v) HNO₃ (pH < 1). Then, magnetic stirring (900 rpm) is applied so that the graphene membrane remains in the centre of the formed vortex. Due to its high flexibility, the membrane becomes conical-shaped in the vortex occurring when immersed into the stirred sample solution. Extraction equilibrium is reached within 20 min. After that, the stirring is stopped and the graphene membrane returns to its initial flat shape. Then the membrane is washed with ultrapure water and deposited onto a quartz substrate. A 10 μL aliquot of internal standard solution is added on the surface of the membrane for TXRF measurements. Membranes are left to dry in a laminar flow hood. Finally, the quartz substrate with the graphene membrane is assembled onto the sample holder of the TXRF instrument prior to measurement.

For total Cr detection, samples are treated with 2.5% m/v potassium peroxydisulphate at 80 °C for 30 min in order to ensure the oxidation of Cr(III) to Cr(VI) before microextraction. The Cr(III) content can be calculated as the difference between the total Cr and Cr(VI) contents.

3. Results and discussion

3.1. Optimization of graphene membrane fabrication

Mild thermal reduction provides a facile, low-cost and greener route compared to chemical reduction for obtaining graphene membranes. Besides, topological defects, i.e. non-reduced functional groups such as hydroxyl, epoxy and carboxyl, are present in the graphene membrane, thus allowing the interaction with ionic species in the extraction medium. Main parameters for membrane preparation are the mass of precursor (GO) and temperature for thermal reduction.

Graphene oxide mass. The amount of functional groups as well as the thickness of the graphene membrane directly depends on the total mass of GO used as precursor. In order to estimate the concentration of synthesized GO, an aliquot of dispersed GO is dried in an oven and then the solid is weighed. A concentration of 1.5 mg mL⁻¹ of GO is obtained.

Taking into account this concentration, different mass of GO are tried by varying the volume of the aliquots during drop-casting operation. Three aliquots in the range of 20–200 μL are deposited onto the glass substrates for thermal reduction. As can be seen in Fig. 1a, the analytical signal for Cr(VI) increases on increasing GO mass. This behavior can be ascribed to the increased amount of functional groups that can interact with the analyte. However, it should be noted that the background signal in the TXRF spectra increases with increasing GO mass (Fig. 1a). Larger GO masses give rise to thicker graphene membranes, which results in higher background. For TXRF measurements, the limit of detection (LOD) directly depends on the background signal: the lower spectral background, the better LOD. A mass of 450 μg of GO (3 aliquots of 100 μL) was selected for the preparation of graphene membranes as a compromise between high analytical signal and low background.

Fig. 1 Optimization of experimental parameters for graphene membrane preparation. (a) GO mass (b) Thermal reduction temperature.

Thermal reduction temperature and reduction time. Generally, the reduction of GO to graphene is carried out by chemical methods using different reducing agents.²⁹ However, the main disadvantages are the use of hazardous reagents and time consuming procedures. As a greener alternative, thermal treatment can be applied. Conventional approaches for thermal reduction of GO to obtain pristine graphene require the rapid heating of GO up to temperatures around 1000 °C under inert atmosphere.^30,31 In contrast to earlier approaches, we perform a facile mild thermal reduction at ‘low-temperature’ under air atmosphere. Results are shown in Fig. 1b. For temperatures below 70 °C no changes are observed. Color of dried GO does not change to black, hence indicating that reduction of GO to graphene is not achieved. For temperatures in the range of 70–100 °C, GO color changes to black and highly flexible and hydrophobic membranes are obtained. For temperatures above 110 °C, aggregation is observed in the glass substrate and the resulting membranes cannot be handled because they break easily.

Following thermal treatment, oxygen-containing functional groups, e.g. hydroxyl, carboxy and epoxy, tend to be reduced. However, incomplete reduction to pristine graphene may occur. The obtained graphene membrane may contain lattice defects originated from the formation of carbon oxides.³² The presence of non-reduced functional groups in the structure promotes the formation of holes along the layers of the membrane. Besides, oxygen functional groups attached to graphene keep the distance between layers³³ and also can increase the membrane permeability for analyte retention. During the sorption process, Cr(VI) can pass through the graphene membrane and interact with non-oxidized groups. At pH of extraction (≈0.9), Cr(VI) exists as dichromate anion (Cr₂O₇²⁻) and oxygen-containing functional groups in the membrane can have positive charge. Therefore, Cr(VI) may be retained by electrostatic interaction.

In addition, the reduction time may also influence the flexibility of graphene membranes. In this way, reduction times for thermal treatment at 70 °C and 100 °C are tried. Flexible graphene membranes that can be easily handled are obtained after 15 min of thermal treatment. On the other hand, when glass substrates are heated to 100 °C, shorter times are necessary (Fig. S1†). After drying aliquots of GO, color changes from yellow to black in 2 min, but membranes cannot be easily separated from the glass substrate. Membranes can break resulting in impaired precision for Cr(VI) extraction. For reduction times longer than 4 min, highly flexible graphene membranes are obtained, which can be readily separated from glass substrates and handled for sorption and measurement steps. Then, thermal reduction at 100 °C for 4 min is fixed for preparation of graphene membranes.

3.2. Characterization of graphene membranes

Different studies have been performed for the characterization of graphene membranes (see ESI†). Gel electrophoresis studies provide information about the charge of the functional groups present in the graphene membranes under sorption conditions (Fig. S2†). It can be assumed that functional groups are positively charged.

The synthesized GO was characterized by TEM. As can be see in Fig. S3,† GO is well-exfoliated giving rise to irregularly-shaped nanosheets (60 nm in length on average).

Besides, AFM studies provide information about morphology, thickness and roughness (Fig. S4†). Graphene membranes synthesized under optimal conditions are 122 nm height with a surface nanoroughness of 22.14 nm. Furthermore, XPS studies provide information about chemical composition of graphene membranes. Results show that complete reduction to pristine graphene is not achieved, so residual non-reduced functional groups (e.g. hydroxyl, carboxy and epoxy) are present in the as-prepared graphene membranes (Fig. S5†). Finally, TOF-SIMS studies provide information about the distribution of Cr(VI) onto the membrane. Results show that Cr(VI) is well-distributed onto the graphene membrane (Fig. S6†).

3.3. Selectivity studies

A preliminary screening for the application of unmodified graphene membranes as novel preconcentration platforms for 19 different metal species is performed. Sorption of different metal ions at a concentration of 20 μg L⁻¹ shows that the as-prepared graphene membranes are highly selective toward Cr(VI) (Fig. 2). The analytical signal for Cr(VI) is around 2700 times higher than that of Cr(III). It is also 200 times higher compared to that of other metal ions at high oxidation state, i.e. As(V), Sb(V) and Se(VI), and 500 times higher than that of Mn(VII). Besides, compared to divalent metal ions, e.g. Co(II), Ni(II), Zn(II), Hg(II), and Pb(II), the analytical signal for Cr(VI) is around 2000 times higher. Hence, unmodified graphene membranes are suitable for selective sorption of Cr(VI) from water in the presence of high concentration of other metal ions.

Fig. 2 Analytical response for different metal ions following sorption onto graphene membranes. Sorption conditions: pH, 0.9; sorption time, 10 min; stirring rate, 900 rpm; metal ion concentration, 20 μg L⁻¹.

Moreover, the sorption of Cr(VI) in the presence of the aforementioned metal ions has also been attempted using APDC at pH 1. APDC–metal complexes are formed with all metal ions, which can interact with the graphene membrane by π–π interactions. This system displayed less selectivity as compared to the use of unmodified graphene membranes at acidic pH for Cr(VI) microextraction. Thus, direct interaction of the membrane graphene (i.e., without addition of chelating agents) with Cr(VI) is selected for further experiments.

3.4. Optimization of Cr(VI) sorption onto graphene membranes

An optimization of parameters affecting the sorption process of Cr(VI) is required in order to reach optimal sensitivity and LOD for analytical applications. Effect of pH, sample volume, stirring rate and sorption time was examined.

Effect of pH. The sorption of Cr(VI) depends on the pH of the extraction medium. Functional groups acting as binding sites can be positively charged, negatively charged or in neutral form, depending on the pH. Since Cr(VI) in solution exists as an anion, functional groups need to be positively charged for efficient interaction to occur. Given the pK_a of carboxyl (4.2–6, depending on its position) and hydroxyl (9) groups in GO, strong acidic media must be used to facilitate the protonation of oxygen-containing groups.³⁴ In this way, different pH values are tried for sorption of Cr(VI) (Fig. 3a). Best results are obtained for pH ≤ 1. For pH > 1, most acidic functional groups are uncharged, and consequently, they do not interact with Cr(VI), resulting in a decreased analytical signal. For pH ≥ 6, carboxyl groups are deprotonated (pH > pK_a) and hydroxyl groups are as neutral species (pH ≈ pK_a) hence reducing the ability of interaction with Cr(VI). Under these conditions, poor Cr(VI) adsorption is obtained.

Fig. 3 Optimization of experimental parameters for sorption of Cr(VI). (a) Effect of pH in the sample (sorption time: 10 min). (b) Effect of HNO₃ concentration ( – HNO₃ concentration; – pH of the extraction medium) (sample volume: 25 mL; stirring rate: 500 rpm; sorption time: 10 min). (c) Effect of sample volume () (pH: 0.9; [HNO₃]: 1 mol L⁻¹; stirring rate: 500 rpm; sorption time: 10 min) and stirring rate () (pH: 0.9; [HNO₃]: 1 mol L⁻¹; sample volume: 50 mL; sorption time: 10 min) (d) Effect of sorption time (pH: 0.9; [HNO₃]: 1 mol L⁻¹; sample volume: 50 mL; stirring rate: 900 rpm). The Cr(VI) concentration was 20 μg L⁻¹ in all experiments.

Different HNO₃ concentrations, i.e. 0.02, 0.1, 0.4, 1, 2.5 and 4.5 mol L⁻¹, were tried for pH adjustment (Fig. 3b). For concentrations above 1 mol L⁻¹ HNO₃, the analytical signal reaches a plateau.

Sample volume and stirring rate. In order to evaluate the effect of sample volume, different volumes from 5 to 100 mL are tried. As can be seen in Fig. 3c, the analytical signal rises as the sample volume increases. For a sample volume higher than 50 mL, this increase is negligible. Taking into account that longer extraction times are needed in order to achieve equilibrium conditions as the sample volume increases, a 50 mL sample volume is selected.

In addition, stirring rate of sample solution is a relevant parameter in order to allow fast mass transfer of the target analyte to the graphene membrane. The time required to reach equilibrium conditions can be reduced by using agitation, so the more effective the stirring is, the shorter the sorption times is required. For a fixed microextraction time (10 min), different stirring rates in the range of 0–900 rpm were tested (Fig. 3c). For stirring rates higher than 900 rpm with a 50 mL sample volume, it is difficult to keep the graphene membrane in the center of vortex. The graphene membrane can move to the bottom of the extraction vessel and hit the magnetic stirrer resulting in breakage.

Sorption time. The sorption process can be extended to equilibrium to reach the highest sensitivity. In order to evaluate this parameter, sorption times from 30 s to 60 min were tried (Fig. 3d). As can be seen, there is a rapid increase in the amount of extracted analyte immediately after placing the graphene membrane in the extraction vessel. For a sorption time longer than 20 min, the analytical signal levels off, hence indicating that equilibrium conditions are achieved. Graphene membranes are thinner than commercial solid-phase microextraction fibers, and therefore, the equilibrium time is shortened. Besides, graphene provides a high area-to-volume ratio resulting in enhanced extraction efficiency, and hence, the sorption time can be reduced as compared to conventional adsorbents.

In order to obtain good sensitivity and repeatability, a sorption time of 20 min is used for further experiments. Besides, it should be noted that if higher sample throughput is required, shorter sorption times can be applied by using pre-equilibrium conditions, but in this case, a precise control of sorption time is required to ensure good repeatability.

3.5. Adsorption isotherms for Cr(VI)

The adsorption of Cr(VI) onto the graphene membranes is evaluated using linear Langmuir (eqn (1)) and Freundlich (eqn (2)) isotherm models:³⁵where q_e is the amount of analyte adsorbed onto the graphene membrane, q_max is the maximum amount of analyte adsorbed per unit weight of graphene membrane (mg g⁻¹), K_L is the Langmuir isotherm constant related to the free energy adsorption (L mg⁻¹), C_e is the concentration of the analyte in the sample at equilibrium (mg L⁻¹), K_F is Freundlich constant related to the adsorption capacity (mg¹⁻ⁿ Lⁿ g⁻¹) and 1/n is constant related to adsorption intensity. To obtain isotherm parameters, experimental data for adsorption equilibrium at different concentrations of Cr(VI) are fitted using the equations mentioned above.

Isotherms parameters indicate that the correlation coefficient for the adjustment of the experimental data are R = 0.9926 and R = 0.8706, for Langmuir and Freundlich model, respectively. Results clearly indicate that experimental data fit better using the Langmuir model. We can conclude that the adsorption sites on graphene are assumed to be uniformly distributed and bind Cr(VI) as a monolayer, in which positively charged oxygen-containing groups can interact with Cr(VI) as dichromate anion. Besides, localized adsorption is thus seen to be very plausible for chemisorption. Due to the strong oxidizing power of Cr(VI) in acid medium, hydroxyl groups present in the graphene membrane can be oxidized to the ketone form, with the subsequent reduction of Cr(VI) to Cr(III).^36,37

Therefore, we can assume that Cr(VI) is retained both by electrostatic interaction with positively charged functional groups and by chemisorption upon reaction with hydroxyl groups.

In addition, the slope of linear adjustment to Langmuir isotherm (1/q_max) provides information about the maximum adsorption capacity of the graphene membrane (q_max), obtaining a result of 22 μg Cr(VI) per mg of graphene membrane. The adsorption capacity for Cr(VI) can vary in a wide range depending on the sorbent.^38–40 Unmodified CNTs (used as dispersions for batch extraction or packed into a column) provide an adsorption capacity in the range 1–11 μg Cr(VI) per mg of CNTs.^27,41–43 Besides, non-nanostructured membranes, e.g. modified silica membrane, shows an adsorption capacity around 3 μg Cr(VI) per mg of adsorbent.⁴⁴ Moreover, using CNTs large extraction times are usually needed to reach equilibrium (several hours), and also poor selectivity is achieved. Using graphene membranes, highly selective extraction of Cr(VI) can be achieved with high adsorption capacity and short extraction time. Moreover, operations for isolating the sorbent by centrifugation or filtration are avoided.

3.6. Study of potential interferences

Waters may contain organic matter and salts, which could influence the sorption of Cr(VI). The effect of potential interferents is shown in Table 1. An interference effect is considered to be significant if a variation beyond ±10% in the measurement is observed. Experiments at different interferent concentration levels were carried out using: NaCl (range: 100–30 000 mg L⁻¹); MgCl₂, KH₂PO₄, Na₂SO₄ and KNO₃ (range: 10–200 mg L⁻¹); Na₂CO₃ (range 50–500 mg L⁻¹). No interferent effect was observed in the ranges studied. It is remarkable the absence of matrix effects due to high levels of NaCl, thus showing that our approach is well suited for the detection of Cr(VI) in seawater. Typical contents of humic acid in waters could vary from 1 to 20 mg L⁻¹. The effect of natural organic matter was studied using humic acid as model compound in the range of 1–100 mg L⁻¹. No interference was observed in the studied range. The presence of the other metal species, e.g. Fe(III), Cu(II), Zn(II), Ni(II), Co(II), As(III), Se(IV), Hg(II), etc., do not cause any interference effect up to at least 2000 μg L⁻¹. Results show that the proposed method can be successfully applied to the detection of Cr(VI) in different kinds of waters without interference effects.

Table 1 Effect of foreign substances on the sorption of Cr(VI) using graphene membranes

Interferent	Studied level (mg L^−1)	Cr(VI) recoveries (%)	Effect (%)
NaCl	100	99	+1.1
	1000	100	+2
	20 000	98	+4.8
	30 000	101	+7.5
MgCl2	10	101	+0.6
	100	99	−0.4
	200	100	−0.9
KH2PO4	10	100	+5.6
	100	101	+7.1
	200	99	+9.5
NaSO4	10	100	+0.5
	100	99	+1.5
	200	99	+3.5
KNO3	10	95	+0.6
	100	100	−0.9
	200	100	−2.3
Na2CO3	50	100	+0.6
	100	97	+0.8
	500	101	+1.3
Humic acid	1	99	+2.3
	50	101	+4
	100	98	+5

---

3.7. Analytical characteristics and analysis of certified reference materials

The LOD, based on the statistical inspection of the fluorescence peak area and the subjacent spectral background, was 0.08 μg L⁻¹ of Cr(VI). The calibration curve for Cr(VI) was linear at least up to 5000 μg L⁻¹ Cr(VI). The repeatability of the analytical response expressed as the relative standard deviation was 3% (N = 5), whereas the reproducibility estimated as between-day precision was 5.5% (N = 3).

In addition, the accuracy of the proposed method for detection of Cr(VI) was evaluated by analyzing two certified reference materials (CRMs), i.e. CASS-4 seawater and NWTM-27.2 lake water (Table 2). Prior to preconcentration, samples are treated with 2.5% m/v potassium peroxydisulphate at 80 °C for 30 min in order to ensure the oxidation of Cr(III) to Cr(VI). The found content for NWTM-27.2 is in good agreement with the certified value for Cr, a recovery of 101% being achieved. The condition t_exp < t_crit (t_crit = 4.303 for N − 1 = 2 freedom degrees) was fulfilled for the studied certified material and consequently, non-significant differences (p = 0.05) occurred between the certified and the found concentration. For the certified material CASS-4, the Cr content was below the limit of quantification (LOQ). In addition, CRMs were spiked with Cr(VI) and Cr(III) at different levels, recoveries for Cr(VI) and total Cr being in the range of 99–102% and 99–101%, respectively.

Table 2 Analytical results for analysis of certified reference materials and recovery studies^a

CRM^b	Spiked content (μg L^−1)		Found content (μg L^−1) ± standard deviation (N = 3)		Recovery (%)
	Cr(VI)	Cr(III)	Cr(VI)	Total Cr	Cr(VI)	Total Cr
a [Total Cr] = [Cr(VI)] + [Cr(III)].b Certified values for total Cr content: NWTM-27.2 (1.9 μg L^−1); CASS-4 (0.144 μg L^−1).c The limit of quantification (LOQ) was 0.25 μg L^−1 of Cr(VI).
NWTM-27.2	—	—	—	1.9 ± 0.1	—	101
	10	10	9.9 ± 0.1	19.8 ± 0.5	99.4	99.3
	50	50	50.7 ± 0.4	101 ± 0.7	99.5	101
CASS-4	—	—	—	<LOQ^c	—	—
	10	10	10.1 ± 0.2	19.9 ± 0.2	100	99.5
	50	50	50.6 ± 0.4	101 ± 1	102	101

---

4. Conclusions

A novel method for Cr speciation based on sorption of Cr(VI) onto multilayer graphene membranes prepared by the drop-casting method has been developed. In contrast to earlier analytical approaches using carbon nanomaterials for trace metal sorption, no filtration, centrifugation or elution are required prior to analysis. Unlike GO or graphene nanosheets, our approach does not need the use of surfactants for efficient dispersion in the extraction medium. Characterization by AFM and XPS shows that graphene membranes display a multilayer organization, which contains non-reduced functional groups resulting in lattice defects. Cr(VI) sorption onto graphene membranes does not require prior complexation, being retained both by electrostatic interaction with positively charged functional groups and by chemisorption due to the strong oxidizing power of Cr(VI) in acidic solutions. Besides, graphene membranes show excellent flexibility and softness resulting in smooth, thin and easy-to-handle sorptive platforms for TXRF analysis. A LOD of 0.08 μg L⁻¹ Cr(VI) is obtained. This LOD is far below the maximum content level established by EPA (100 μg L⁻¹ as total chromium). The use of transportable TXRF instrumentation can open the door to field measurements, thus avoiding sampling and preservation operations needed when using centralized analytical techniques.

Acknowledgements

Financial support from Spanish Ministry of Economy and Competitiveness (Project CTQ2012-32788) and the European Commission (FEDER) is gratefully acknowledged. The Spanish Ministry of Education, Culture and Sport is acknowledged for financial support through a FPU predoctoral grant to V. Romero. C. Serra (CACTI, University of Vigo) is thanked for her assistance with AFM, XPS and TOF-SIMS measurements.

References

1. A. A. Balandin, S. Ghosh, W. Bao, I. Calizo, D. Teweldebrhan, F. Miao and C. N. Lau, Nano Lett., 2008, 8, 902–907.
2. G. M. Rutter, J. N. Crain, N. P. Guisinger, T. Li, P. N. First and J. A. Stroscio, Science, 2007, 317, 219–222.
3. K. S. Novoselov, A. K. Geim, S. V. Morozov, D. Jiang, M. I. Katsnelson, I. V. Grigorieva, S. V. Dubonos and A. A. Firsov, Nature, 2005, 438, 197–200.
4. W. S. Hummers and R. E. Offeman, J. Am. Chem. Soc., 1958, 80, 1339.
5. S. Stankovich, D. a. Dikin, R. D. Piner, K. a. Kohlhaas, A. Kleinhammes, Y. Jia, Y. Wu, S. T. Nguyen and R. S. Ruoff, Carbon, 2007, 45, 1558–1565.
6. X. Wang, L. Zhi and K. Müllen, Nano Lett., 2008, 8, 323–327.
7. R. Sitko, B. Zawisza and E. Malicka, TrAC, Trends Anal. Chem., 2013, 51, 33–43.
8. Q. Liu, J. Shi and G. Jiang, TrAC, Trends Anal. Chem., 2012, 37, 1–11.
9. T. Kuila, S. Bose, A. K. Mishra, P. Khanra, N. H. Kim and J. H. Lee, Prog. Mater. Sci., 2012, 57, 1061–1105.
10. S. C. Smith and D. F. Rodrigues, Carbon, 2015, 91, 122–143.
11. S. Su, B. Chen, M. He and B. Hu, Talanta, 2014, 123, 1–9.
12. M. R. Pourjavid, M. Arabieh, S. R. Yousefi, M. R. Jamali, M. Rezaee, M. H. Hosseini and A. A. Sehat, Mater. Sci. Eng., C, 2015, 47, 114–122.
13. S. Su, B. Chen, M. He, B. Hu and Z. Xiao, Talanta, 2014, 119, 458–466.
14. J. Sun, Q. Liang, Q. Han, X. Zhang and M. Ding, Talanta, 2015, 132, 557–563.
15. R. Sitko, P. Janik, B. Zawisza, E. Talik, E. Margui and I. Queralt, Anal. Chem., 2015, 87, 3535–3542.
16. D. Deng, X. Jiang, L. Yang, X. Hou and C. Zheng, Anal. Chem., 2014, 86, 758–765.
17. K. Kocot and R. Sitko, Spectrochim. Acta, Part B, 2014, 94–95, 7–13.
18. K. Kocot, R. Leardi, B. Walczak and R. Sitko, Talanta, 2015, 134, 360–365.
19. B. J. Clodfelder, J. Emamaullee, D. D. D. Hepburn, N. E. Chakov, H. S. Nettles and J. B. Vincent, J. Biol. Inorg. Chem., 2001, 6, 608–617.
20. A. Zhitkovich, Chem. Res. Toxicol., 2011, 24, 1617–1629.
21. J. Ščančar and R. Milačič, J. Anal. At. Spectrom., 2014, 29, 427.
22. K. Pyrzynska, TrAC, Trends Anal. Chem., 2012, 32, 100–112.
23. D. Das, U. Gupta and A. K. Das, TrAC, Trends Anal. Chem., 2012, 38, 163–171.
24. S. A. Steiner, M. D. Porter and J. S. Fritz, J. Chromatogr. A, 2006, 1118, 62–67.
25. B. Leśniewska, B. Godlewska-Zyłkiewicz and A. Z. Wilczewska, Microchem. J., 2012, 105, 88–93.
26. Z. Ren, D. Kong, K. Wang and W. Zhang, J. Mater. Chem. A, 2014, 2, 17952–17961.
27. M. Tuzen and M. Soylak, J. Hazard. Mater., 2007, 147, 219–225.
28. I. López-García, Y. Vicente-Martínez and M. Hernández-Córdoba, Talanta, 2015, 132, 23–28.
29. C. K. Chua and M. Pumera, Chem. Soc. Rev., 2014, 43, 291–312.
30. W. Chen and L. Yan, Nanoscale, 2010, 2, 559–563.
31. W. Chen, L. Yan and P. R. Bangal, Carbon, 2010, 48, 1146–1152.
32. B. Dai, L. Fu, L. Liao, N. Liu, K. Yan, Y. Chen and Z. Liu, Nano Res., 2011, 4, 434–439.
33. R. R. Nair, H. A. Wu, P. N. Jayaram, I. V. Grigorieva and A. K. Geim, Science, 2012, 335, 442–444.
34. B. Konkena and S. Vasudevan, J. Phys. Chem. Lett., 2012, 3, 867–872.
35. V. Jegatheesan, J. Virkutyte, L. Shu, J. Allen, Y. Wang, E. Searston, Z. P. Xu, J. Naylor, S. Pinchon, C. Teil, D. Navaratna and H. K. Shon, in Wastewater Treatment, Advanced Processes and Technologies, ed. D. G. Rao, R. Senthilkumar, J. A. Byrne and S. Feroz, CRC Press, Boca Raton, 2013, pp. 275–318.
36. F. H. Westheimer, Chem. Rev., 1949, 45, 419–451.
37. V. G. Chandrasekaran, in Comprehensive Organic Synthesis, ed. P. Knochel and G. A. Molander, Elsevier, Oxford, 2014, pp. 277–294.
38. J. Jachuła and Z. Hubicki, Appl. Water Sci., 2013, 3, 653–664.
39. M. Owlad, M. K. Aroua, W. A. W. Daud and S. Baroutian, Water, Air, Soil Pollut., 2009, 200, 59–77.
40. M. N. Sahmoune, K. Lauhab and A. Boukhiar, Environ. Prog. Sustainable Energy, 2011, 30, 284–293.
41. C. Jung, J. Heo, J. Han, N. Her, S. J. Lee, J. Oh, J. Ryu and Y. Yoon, Sep. Purif. Technol., 2013, 106, 63–71.
42. Ş. S. Bayazit and Ö. Kerkez, Chem. Eng. Res. Des., 2014, 2, 2725–2733.
43. M. H. Dehghani, M. M. Taher, A. K. Bajpai, B. Heibati, I. Tyagi, M. Asif, S. Agarwal and V. K. Gupta, Chem. Eng. J., 2015, 279, 344–352.
44. Y. Bao, X. Yan, W. Du, X. Xie, Z. Pan, J. Zhou and L. Li, Chem. Eng. J., 2015, 281, 460–467.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra23116g

---