Frustules or the rigid amorphous silica cell wall of unicellular, photosynthetic microalgae with unique porous architecture has been used to synthesize a composite by immobilizing zirconium and iron oxides on its surface and in the pores. This was effective for removal of Se from water, which is an emerging contaminant that is a micronutrient at low concentrations but toxic at high concentrations. The adsorption isotherms followed both Langmuir and Freundlich models, and the composite was regenerable. The Langmuir maximum adsorption capacity for Se(IV) was 227 mg/g, which is among the highest ever reported. The research findings highlight the synthesis of bimetallic composite as well as the potential of diatoms as hosts for nanomaterials for use in water treatment.

1. Introduction

Selenium is an important micronutrient for animals and humans but is toxic in excess [1]. Higher Se concentration can lower reproduction rates and increase birth defects [2, 3]. In water, selenium exists predominantly in inorganic forms selenite () where Se is present as the Se4+ and selenate () where selenium is present as the Se6+ [4]. The toxicity of selenium depends on its oxidation state and Se(IV) is considerably more toxic than Se(VI) [5]. Drinking water is a primary source for selenium exposure and the US Environmental Protection Agency has set the maximum contaminant level in drinking water to be 0.05 mg/L. Sources of Se include agricultural and mine drainage, residues from fossil fuel thermoelectric power plants, oil refineries, and metal ores [6].

A variety of treatment technologies have been explored for the remediation of both selenium oxoanions in water. These include bacterial reduction, membrane filtration, chemical reduction, reverse osmosis, and solar ponds [79]. These technologies have their limitations and alternative treatment techniques are being explored. Adsorption by metal oxides of iron and aluminum has shown promise in selenium removal [10]. Other materials like Mg/Fe hydrotalcite type compounds, hematite, magnetite, iron-coated GAC, and magnetic Fe/Mn oxide nanomaterials also have high affinity for selenium [1115]. Adsorbents such as sulphuric acid-treated peanut shell, hydrocalumite, ettringite, AlPO4, biopolymeric materials, aluminum-based water treatment chemicals, hardened cement paste, cement minerals, aluminum oxides, iron oxyhydroxides, iron-coated sand, and zero valent iron [1623] have also been tested for Se removal.

The search for more efficient water treatment media has led to the development of nanostructured adsorbents where metals, metal oxides, and ion-exchange medium are immobilized on supports that promote nanostructuring [24]. Along with the active sorbent(s), the support plays an important role. Properties like thermal, mechanical, and chemical aid in structuring of the active ingredient [24]. Consequently, composite or hybrid structures are promising materials for water treatment. The widely used host materials for nanocomposite include carbonaceous materials like granular activated carbon [25], silica [2629], cellulose [30, 31], chitosan [32, 33], sand [34, 35], and polymers [3638]. Some examples include the immobilization of metal oxides on carbon nanotubes for lead, arsenic, and fluoride removal [3941], activated carbon immobilized on carbon nanotubes for chromium removal [25], and carbohydrate and iron oxide on multiwalled carbon nanotubes for zinc removal [42].

Diatoms are unicellular, photosynthetic microalgae that are widely distributed in fresh and seawater. Diatoms can self-replicate and can further be engineered to provide cost-effective and programmable industrialized system [43]. There are over eleven thousand known species whose size ranges from 2 µm to 2 mm and they also have diverse morphology [44]. Naturally occurring diatom frustules are a source of nanomaterials. Frustules, or the rigid amorphous silica cell wall of the diatoms, have unique porous architecture [45, 46] and high surface area [47]. Recently, they have found a variety of applications including water filtration membrane [48], gas sensor [49], electroluminescent display device [50], lithium battery electrode [50], dye sensitized solar cells [50], biochip [51], and drug delivery [52]. It is possible that the diatoms can serve as hosts for immobilizing active sorbent particles on their surface. The surface of the diatom can have different functionalities such as –COOH, –NH2, –OH, and –SiOH, and different compounds can be immobilized on their surface by interactions with these functional groups [53, 54]. They have been successfully used as templates for the synthesis of advanced nanostructured biohybrids. Diatom can be used in water filtration and purification due to different interesting properties likes filtration of microorganisms, homogenous permeability, and fixed pore size, USEPA approved [43]. Not much research is done on exploring diatom for water purification applications.

The objective of this research is to develop nanosorbents by immobilizing metal oxides on diatom surface and in pores of diatom frustules. It is well known that mixed metal oxides usually exhibit better sorption abilities than individual oxides in terms of higher capacities, pH tolerance, and faster kinetics [55, 56]. Hydrous zirconium oxide is known to retain various oxo metal anions, especially those that can form weak conjugate acids [57]. It is also known to be chemically stable and nontoxic and does not dissolve in water at a wide range of pH [58]. At the same time, the iron oxides/hydroxides have been gaining popularity in water treatment and are known to adsorb Se [56]. Therefore, the specific objective is to develop a bimetallic oxide composite by immobilizing iron and zirconium oxides on diatom frustules for effective removal of Se from water.

2. Materials and Methods

2.1. Synthesis of Bimetallic Diatom Composite

Diatom Phaeodactylum tricornutum was cultured and maintained in artificial sea water Aquil [59] using a diurnal chamber with 12-hour day/night cycles at 19°C ± 1°C. 6 L of diatom culture was flocculated with 6 g of ZrOCl2·8H2O and 60 g of FeCL36H2O purchased from Sigma Aldrich at pH 9. One molar NaOH was used to adjust the pH to 9.

After flocculation, the culture was conditioned overnight with shaking (150 rpm, 2880 VWR orbital shaker). The resulting bimetallic diatom mixture was separated by gravitational settling and membrane filtration (<5 psi, 0.2 µm PTFE filter) and was washed with 500 mL Milli-Q water. The resulting slurry was transferred to 50 mL centrifuge tube and thermally treated at 70°C in an oven for 6 h. Then, it was treated with 10 mL of concentrated H2SO4 and heated for 2 hrs at 200°C, vacuum-filtered using 0.2 µm filter, washed with Milli-Q water to neutral pH, and then dried at 200°C in a vacuum oven. The bimetallic oxide-diatom composite (BMDC) was characterized using transmission electron (HRTEM), scanning electron microscope equipped with energy dispersive X-ray spectrometer (SEM, LEO 1530 VP), thermogravimetric analyzer (TGA using Pyris 1 from PerkinElmer Inc.), BET surface area analyzer (Quantachrome Autosorb-1), and FTIR (IR Affinity-1, Shimadzu).

2.2. Adsorption Studies

The kinetics of adsorption were performed as follows. 50 millilitres of 5 mg/L of Se(IV) and (VI) solutions was contacted with 0.010 g and 0.025 g of adsorbent in polycarbonate bottles and the samples were collected at 5, 15, and 30 minutes followed by 1, 3, 6, and 24 hours. 5 mL aliquot was withdrawn at different time intervals and filtered using 0.2 µm membrane filter and residual Se in the media was quantified using Agilent 7500 ICP-MS. All standards were prepared from multielement solution 2A, 10 mg/L (Spex Certiprep), with addition of internal standard mix (Li6, Ge, Y, In, Tb, and Bi). Multielement instrument calibration standard, 1 and 20 mg/L (Spex Certiprep), was used for the verification of calibration. The adsorption capacity and adsorption isotherms were obtained by varying the mass of adsorbent from 0.002 to 0.015 g (Se IV) and 0.002 to 0.050 g Se(VI) with 50 mL of 5 mg/L Se containing water at pH 6. The samples were collected for analysis after 0, 6, and 24 hours of adsorption.

Adsorption was studied at pH values 2, 4, 6, and 8 using 50 millilitres of 1 mg/L Se(IV) solution that was contacted with 0.010 g of adsorbent for 24 h. Desorption of selenium from the spent bimetallic diatom composite was carried out using 1 M NaCl and 0.1 M NaOH solution. After selenium adsorption, the composite was washed with 1 M NaCl, followed by distilled water and 0.1 M NaOH.

3. Results and Discussion

3.1. Characterization of Bimetallic Oxide-Diatom Composite

Diatom Phaeodactylum tricornutum was grown till the late exponential phase [60]. After adding ZrOCl28H2O, FeCl38H2O, and NaOH to the diatom, Zr and Fe oxides were precipitated on the diatom surface and chemically attached with diatom functional groups such as –COOH, –NH2, –OH, and –SiOH. This was enhanced by heat treatment at 70°C for 6 h. Furthermore, H2SO4 treatment decomposed all the organic matter leaving algal core with Zr and Fe oxides on its surface. Mesoporous zirconia can be obtained by contacting amorphous hydrous zirconia obtained by alkaline hydrolysis with sulphuric acid followed by calcinations [61].

The presence of Zr and Fe particles on diatom surface was studied using SEM. Figure 1(a) shows SEM of the original diatom, Figure 1(b) shows bimetallic composite, and Figures 1(c) and 1(d) show mapping of Zr and Fe from BMDC. Similarly, Figures 1(e) and 1(f) show TEM image of the diatom and BMDC. SEM and TEM images showed that the original diatom was reduced to porous nanobiosilica. EDX analysis using SEM confirmed the presence of Zr (89.87%) and Fe (8.75%) in BMDC.

The specific surface area of bimetallic diatom composite based on BET measurements was found to be 128 m2/g and pore diameter was 9.39 Å. The FTIR (Figure 2) was used to confirm the functional groups. The observed peak at 374–400 cm−1 was attributed to Zr-O vibration, which confirmed the ZrO2 structure [62]. The observed peak at 551 cm−1 and around 700 cm−1 was attributed to Fe [63]. Both spectra showed peaks at 1108 and between 3300 cm−1 and 3500 cm−1. These were due to stretching of siloxane (Si–O–Si) and free silanol group (Si–OH) [64].

TGA was used to test the thermal stability of the BMDC (Figure 3). The weight loss below 120°C was attributed to the removal of physisorbed water while the loss between 120 and 300°C was due to chemisorbed water [65]. In the 300–400°C range, the weight loss was from the oxidation and decomposition of mercaptopropyl or aminopropyl groups on diatom surface. The broad exothermic weight loss in the range of 400–600°C was due to the decomposition of strongly tethered organosilanes and dehydration of silanol groups, and the weight reduction between 400 and 800°C was due to the dehydroxylation of the silica surface [66, 67]. TGA curve indicated that bimetallic diatom composite exhibited good thermal stability. It is worth noting that the observed weight loss (25%) below 300°C was mainly ascribed to the evaporation of water. Only 15% of the mass remained as residue beyond 600°C.

3.2. Selenium Removal Using BMDC

It was observed that no selenium was adsorbed on the pure diatom, but the BMDC was effective in removing selenium from water. Selenite and selenate uptake by BMDC was studied as a function of time and is presented in Figure 4. Selenium sorption increased as a function of contact time. 0.010 g of bimetallic composite showed Se(IV) uptake of 50 and 95% at 15 mins and 3 hrs, respectively, whereas 0.025 g of bimetallic composite showed removal of only 40% Se(VI) after 24 hrs of contact. This shows that bimetallic diatom composite shows better Se(IV) removal efficiency compared with Se(VI).

The amount of selenium adsorbed was calculated using the following equation:where and (mg/L) are the liquid-phase concentrations of selenium at initial and equilibrium concentrations, respectively, is the volume of solution in (L), and is the mass of the adsorbent used (g). Adsorbent dosage is an important parameter because this determines the capacity of an adsorbent. It was observed that, at equilibrium, the percentage removal increased with the increase in the adsorbent concentration while the adsorption capacity decreased. Table 1 presents at different adsorbent dosage. It was seen that the initial adsorbent dose affects the adsorption capacity; dropped from 56.24 mg/g for 0.001 g of adsorbent to 6.45 mg/g for 0.015 g for Se(IV) and 10.6 mg/g for 0.002 g of adsorbent to 0.89 mg/g for 0.05 g for Se(VI).

The kinetics of selenium uptake were studied using Lagergren [68] as well as Ho and McKay kinetic models [69]. The former models the rate of adsorption of pollutants based on pseudo-first equation to describe the kinetics of liquid-solid phase adsorption [70].The rate constant was calculated from a linear plot of log verses , where and are the sorption capacity (mg/g) of the adsorbent at equilibrium and at time (h−1), and is the rate constant of pseudo-first order. The second-order equation from Ho and McKay is based on the assumption that adsorption may be of second order and the rate limiting step may be from chemical adsorption involving exchange of valence electrons [71]The rate constant was calculated from a linear plot of versus . Here is the pseudo-second-order sorption rate constant (g/h/mg) and is time (h−1). The applicability of first- and second-order models was tested for adsorption of selenium on BMDC. Table 2 presents the kinetic data. The best fit was selected based on the linear regression coefficient . The models were fitted with the experimental data. Ho and McKay’s second-order equation was found to be a better fit as compared with the first-order equation. The adsorption rates obtained from first- and second-order kinetic models are given in Table 2. In the first-order model, a larger adsorption rate constant usually represents quicker adsorption whereas in the second-order model, a lower value of represents faster adsorption. This is due to the availability of large number of binding sites.

The effect of pH was studied at 2, 4, 6, 8, and 10 (Figure 5). The variation in pH in the range of 2 to 6 did not show much effect on percent adsorption. With increase in pH from 6 to 10, the adsorption capacity decreased. At low pH conditions, the H3O+ ion concentration was optimum to make the surface of the BMDC positively charged and hence accessible for selenium ions. With an increase in pH, especially in alkaline medium, the selenium uptake was reduced due to the competition with OH-, which also had high affinity for the zirconium and iron ion.

Langmuir [72] and Freundlich [73] isotherms provided an insight into the surface coverage via physisorption and/or chemisorption. Langmuir isotherm best describes the chemisorption process. The adsorption involves the attachment of monolayer of molecules on the surface. The linear form of Langmuir adsorption isotherm, which involves a plot of versus , is represented asHere, (mg/g) is the maximum sorption capacity of the sorbent, (mg/L) is the equilibrium selenium ion concentration, and Langmuir constant (L/mg) is indirectly related to the enthalpy of adsorption. The essential features of Langmuir adsorption parameters can be used to predict the affinity between the sorbent and the sorbate using the dimensionless separation factor which is expressed as follows:where is Langmuir constant and is initial concentration. indicates the nature of adsorption ( between 0 and 1 is considered favourable, is unfavourable, and is reversible) [74, 75].

The correlation coefficient for the Langmuir model was 0.956, which demonstrated a high degree of correlation. The Langmuir parameters are presented in Table 3. The maximum adsorption capacity for Se(IV) on the composite was 277 mg/g and Se(VI) was 0.48 mg/g at pH 6. The Langmuir constant , the ratio of adsorption rate constant to desorption rate constant, is an indication of affinity of the sorbent material towards selenium. Since was less than 1, the adsorption was considered favourable. Hence Langmuir model effectively explained the selenium uptake by the bimetallic diatom composite.

Freundlich adsorption was also tested. In its linearized form, the Freundlich isotherm involves a plot of log qe and log Ce: The values for log and were obtained as the intercept and slope, respectively (Table 3). A measure of adsorption capacity and adsorption intensity was provided by the Freundlich constants (mg g−1) and , respectively. Here, was an indicator of the degree of nonlinearity between water concentration and sorption ( denotes linear adsorption, a chemisorption, and physisorption [76]). The bond energy increases proportionally with surface density for and vice versa for . Values of being 0.47 and 0.45 implied that the adsorption was a chemical process and that the bond energy increased with surface coverage [73]. The correlation coefficient, , was 0.95.

Based on the second-order kinetics, Langmuir model, and Freundlich model, the adsorption appeared to be via a chemical process. Adsorption of Se can occur due to both outer and inner sphere complexation on metal oxides [77]. Selenate is known to be more strongly adsorbed than selenite at a wide range of pH. As per the recent review, mixed metal oxide, double layered materials, and adsorbents based on natural materials have shown good sorption capacities and relative fast kinetics. It is worth mentioning that the sorption capacity from our study is comparable to those reported before. For example, of 2.38 mg/g was reported for selenate removal using Fe3O4 [78], 26.3 mg/g using FeOOH [77, 79], 29.0 mg/g using Fe-Mn hydrous oxide [24], and 74.9 mg/g Se(IV) using green alga Cladophora hutchinsia [80]. Recent study on selenium removal using MgO sheets showed adsorption capacity of Se(IV) 103.52 mg/g and Se(VI) 10.58 mg/g [22]. Se(IV) Langmuir adsorption capacity of [email protected] is 25.0 mg/g, MNP core 15.3 mg/g, magnetic nanoparticle-graphene oxide 23.8 mg/g [81], Al(III)/SiO2 20.4 mg/g [82], a-Fe2O3 17.9 mg/g [83], Fe/Si coprecipitates 17.4 mg/g [20], MIO/MWCNTs 13.1 mg/g [21], CuFe2O4 14.1 mg/g [19], and iron manganese oxide 6.57 mg/g [15].

Desorption of Se(IV) and regeneration of BMDC were studied by monitoring the effluent washes from a Se(IV) sorbed BMDC. 0.050 g of adsorbent was contacted with 50 mL of 2 mg/L of Se(IV) solution. After 24 hours of exposure, the solution was vacuum-filtered using 0.2 um filter and washed with 1 M NaCl followed by distilled water and 0.1 M NaOH solutions. Brine wash resulted only in small amounts of Se(IV) desorption indicating that only a small fraction of ions are held by ion-exchange/electrostatic forces. A nearly complete removal of adsorbed Se(IV) was achieved by alkali wash which overcame all interactions of the sorbent with Se(IV) through a dominating competition for surface active sites. In short, Se(IV) could be regenerated relatively easily by treating with an alkaline solution (e.g., 0.1 M NaOH). Sodium hydroxide quantitatively desorbed all bound Se(IV) without damaging the BMDC which could be reused after washing it with 0.1 M H2SO4 acid followed by distilled water till neutral pH. Two successive regenerations showed more than 90% removal efficiency.

4. Conclusion

Diatom offers unique architecture with excellent mechanical strength. This paper highlights the potential of diatom as the host for immobilizing nanomaterials to form composite. A bimetallic composite was successfully synthesized for removal of Se from water. This was achieved by precipitating zirconium and iron oxides on the diatom and then oxidizing the organic mass. Maximum Se(IV) adsorption capacity of 277 mg/g adsorbent was calculated using the Langmuir adsorption isotherm. A variation in pH between 2 and 6 did not alter the adsorption capacity, but a significant reduction in capacity was seen at pH 8. The BMDC had a higher adsorption capacity for Se(IV) when compared with Se(VI).


Any opinions, findings, and conclusions or recommendations expressed in this paper are those of the authors and do not necessarily reflect the views of the NJWRRI.

Conflicts of Interest

The authors declare that they have no conflicts of interest.


The authors would like to thank New Jersey Water Resources Research Institute of Technology (NJWRRI) for partial funding under Grant no. 2016NJ383B.