Introduction

Arsenic exposure through drinking water sourced from groundwater, is a global public health problem that is particularly devastating in certain highly populated countries1,2. According to a 2000 to 2010 case study, 35 to 77 million people in areas of Bangladesh or India have been chronically exposed to arsenic in their drinking water in what described as the most significant mass poisoning in history3. Arsenic is a naturally occurring metalloid, also released to the environment via anthropogenic activities. Arsenic strongly binds to proteins, so traces of this element can cause severe health problems to all life forms4. The predominant forms of arsenic in the aquatic environment are AsIII (arsenite) and AsV (arsenate). AsIII is more hazardous than AsV, as it is more mobile/bioavailable, thus more toxic5. This toxicity is due to its dominant H3AsO3 form, i.e. the predominating species in a wide range of pH < 96, typically encountered in natural waters. So far, traditional ion-exchange materials and sorbents, e.g. zeolites, clays, layered double hydroxides, resins have been used as arsenic adsorbends, however, with limited efficiency vs. AsIII 7,8.

Given that AsV, which exists as anion at pH > 2, is easier to adsorb on cationic surfaces, to overcome the low efficiency of AsIII uptake, an extra step of oxidation of AsIII to AsV can be chosen before the application of various remediation technologies. Such oxidative pretreatment, however, suffers from the presence of multiple substances that interfere with AsIII oxidation9,10. Therefore, it will be desirable to develop sorbents that could directly capture AsIII in natural pH conditions, without the need for the oxidation to AsV. Several materials have been investigated for the direct removal of AsIII, including TiO2 nanoparticles11, iron-based nanoparticles, e.g. zero-valent Fe nanoparticles12,13,14, carbon nanotubes15. Such sorbents with high specific surface area and various functional groups seem promising for AsIII remediation16,17,18.

A further hurdle to overcome concerns cost-criteria, i.e. the required mass of an AsIII-uptaking material, should be considered on a cost-efficiency basis together with the ecological impact of such sorbents is natural or urban water bodies. So far, most of the available materials have sorption capacities of the range 60–150 mg g−1 19,20,21, while less than five materials achieving sorption capacities > 300 mg g−1 22,23. In one report in 201423, the highest –so far- AsIII uptake reported was 320 mg gr−1 by a hybrid consisting of Fe2O3 nanoparticles dispersed on a macroporous silica. Our systematic efforts during the last decade led to a series of nanomaterials with promising performances, e.g. a mesoporous spinel CoFe2O424 with an uptake of 252 mg AsIII g−1, magnetic carbon nanocages22 with a sorption capacity of 264 mg AsIII g−1 and MIL-100(Fe)16 showing uptake of 120 mg AsIII g−1.

Regarding the physicochemical AsIII-uptake mechanism, so far, in all well-understood cases, the underlying theoretical mechanism is that originally developed by Goldberg25,26 and Manning7,10, which entails that: [i] AsIII-uptake by solid materials is determined by surface complexation of the AsIII species. [ii] at pH range 5–8, i.e. typical for natural waters, the dominant species is the neutral form H3AsO327,28. Thus –so far – the strategy by all research groups, including us16,22,29, was to maximize the number and accessibility of surface sites based on diligent preparation protocols. In this way, it has been achieved a max AsIII-uptake capacity 320 mg gr−1 by γ-Fe2O3 nanoparticles encapsulated in a macroporous Silica23. Recent data show that certain carbon based materials i.e. graphene-based30 or more innovative graphydine31,32,33 have a promising potential for adsorbing heavy metals, metalloids and other pollutants from water. In addition, some of the metal-loaded materials can have enhanced catalytic fucntionalities34,35,36,37.

Within this frame of thinking, aiming at maximization of the surface sites, it becomes obvious that for any material, the theoretical upper limit would be determined by the site-density and the surface area:

$${N}_{max}(sites/gram)=SSA({m}^{2}/gr)\times Ns(sites/n{m}^{2})={10}^{18}\times SSA(n{m}^{2}/g{r}^{1})\times Ns(sites/n{m}^{2})$$
(1)

Using this expression, a material whose maximum- AsIII uptake capacity is determined by surface-adsorption has an upper theoretical limit, which under ideal conditions, would be determined by its specific surface area (SSA) and the number of surface sites (Ns). In real systems, this maximum-uptake would be further limited by the binding constant of AsIII-on the surface-sites. On the other hand, in natural systems, Soil Organic Matter is known to be able to sequester non-polar organics via a partitioning-sequestration mechanism38. This phenomenon is based on the fundamental concept of the partitioning of a non-polar organic between a polar and a non-polar solvent, e.g. for example, the partitioning of phenol on [octanol: water]39,40. Within a technological context, when partitioning is operating, a significant mass of sorbent can be transferred in the apolar matrix, thus resulting in cost-effective molecular-separation approaches41,42. So far, however, such a profitable concept has not been applied to attack problems such as AsIII-remediation. Herein we introduce a mind-changing approach by developing a Zirconium Metal-Organic Framework [[Zr6O4(OH)4(NH2BDC)6] (NH2-BDC2− = 2-amino-terephthalate) [herein codenamed ZrMOF] which is able to perform a partitioning-like AsIII-uptake thanks to the microenvironment of its pores. As we show, this approach allows an unprecedented AsIII-uptake efficiency of > 2000 mg AsIII per gram of ZrMOF.

Metal-organic frameworks (MOFs), which are crystalline porous materials are constituted by metal ions or metal clusters, and polytopic organic ligands, have emerged as a new class of sorbents with a promise for various remediation processes43,44,45. MOFs combine extremely high surface areas, well-defined pores, and a variety of functional groups. Furthermore, several MOFs show remarkable thermal (up to 400–500 °C) and chemical stability e.g. high resistance to acid or base46,47. Recently, several MOFs have been investigated as arsenic sorbents, mostly concerning investigations of forms of ionic-AsV sorption48,49,50. Only a few reports exist on the capture of AsIII by MOFs40,43,44,45. These sorbents, however, demonstrated only moderate sorption capacities (<150 mg AsIII g−1)51.

Apart from the challenge of AsIII-uptake efficiency, the efficient large-scale utilization of AsIII-uptaking materials52,53,54, requires their post-synthesis engineering to be usable for large-water body cleaning. The technology of grafting of the functional material on a macroscopic surface allows scale-up handling and usage. Herein we have used woven silk fibers [SFd] as a scaffold for grafting of the ZrMOF material. The so-obtained hybrid material ZrMOF@SFd retains the very high AsIII sorption capacity of ZrMOF, i.e. reaching 1800 mg AsIII per g−1 of material.

To understand the unprecedented high sorption efficiency of ZrMOF and ZrMOF@SFd, we have carried out a detailed study ofthe AsIII-sorption mechanism in conjunction with the dynamics of pore-filling and surface complexation. In a more general context, the present research exemplifies for the first time that a “partitioning-like” mechanism to be operating in for adsorption of metalloids, i.e., H3AsO3 by metal oxide materials, so far conceptualized only for synthetic polymers & natural organic matter (NOM)55 used to uptake apolar organics56,57.

Results and Discussion

Field Emission-Scanning electronic microscopy (FE-SEM) images showed that ZrMOF is composed of aggregated polyhedral-shape nanoparticles with size ~100–300 nm (Figure 1a1,a2). No obvious changes in shape and size of particles are observed for the material after the AsIII sorption (Fig. 1b1,b2). The SEM micrograph for the silk fiber (SFd) (Fig. 1c), shows the surface morphology of well-defined fibers of natural-silk. After covalent grafting of ZrMOF on the SFd fibers, we obtain well-dispersed ZrMOF particles on the silk-fibers (Fig. 1d). Adsorption of AsIII onto ZrMOF@SFd does not alter the particle morphology (Fig. 1e). Thermogravimetric (TGA) analysis, Fig. S1 in Supporting Information, shows that the ZrMOF@SFd hybrid contains 5.7% w:w of ZrMOF.

Figure 1
figure 1

FE-SEM images for ZrMOF (a1,a2), AsIII-loaded ZrMOF (b1,b2), natural-silk SFd (c),ZrMOF@SFd (d), AsIII-loaded ZrMOF@SFd (e).

Full size image

For the engineering of the ZrMOF grafting on the silk fiber, we have used natural silk tissue, which we have degummed according to established procedures (see Supporting Information, photos in Scheme S1)58. The degumming method of Gulrajani58, resulted in high-quality silk-fiber, as evidenced by the SEM micrography, see Fig. 1(c), as well as the XRD pattern, Fig. 2A, which reveals retention of the fibers’ order and physical integrity in the structure of the degummed silk (SFd). After grafting of the ZrMOF@SFd, the SEM data (Fig. 1d) shows a good dispersion of the ZrMOF particles on the silk fibers. XRD data for the ZrMOF@SFd hybrid, Fig. 2a(red) show the characteristic reflection at 7.3° and 8.5° originating from the ZrMOF particles grafted on the silk. Notice that, upon grafting, the crystallinity of the SFd is distorted, i.e. see the loss of the sharp peaks at 15°–17° in Fig. 2a. This result is due to molecular covalent grafting of the ZrMOF-silane on SFd (see also Scheme S2 in Supporting Information)

Figure 2
figure 2

(a) PXRD data for ZrMOF and ZrMOF @ SFd, (b) PXRD data for ZrMOF, and AsIII@ ZrMOF. (c)TGA (solid line) and first derivative (DTG) (dashed line) plots for ZrMOF,(d) FT-IR spectra for pristine (−), and AsIII-loaded ZrMOF material (), (e) XPS As3d analysis for ZrMOF after AsIII loading.

Full size image

TGA analysis, Fig. 2c, revealed a significant weight loss (~17.2%) in the temperature range 25 to 102 °C attributed to the release of solvent molecules, mostly MeOH. Then, there is a continuous weight loss (~7.9%) (with no discrete steps) till 302 °C, followed by an abrupt weight loss (38.8%) ending at 540 °C. Finally, there is a small weight loss (~2%) from 615 to 715 °C (Fig. 2c). The TGA residue is solid ZrO2, as confirmed by XRD data. For 100 g of ZrMOF 33.7 g of ZrO2 was obtained after calcination, which corresponds to ~24.9% Zr w:w. Based on this % Zr content found from TGA, the suggested formula for ZrMOF is [Zr6O4(OH)4(NH2-BDC)6]∙12MeOH ∙ 3H2O (calculated % Zr = 24.96).

The pristine silk fabric (SF) and degummed silk fibroin fibers (SFd) were characterized by FT-IR and thermogravimetric analysis (TG-DTA). The typical IR-peaks of SF are ≈1620–1700, 1511–1539, and 1226–1235 cm−1, characteristic for amide I (C=O stretching), amide II (N–H deformation, and C–N stretching) and amide III (C–N stretching and N–H deformation). FT-IR (cm−1, selected peaks) SF: 3533: ν(OH); 3072, 2980, 2936, 2880: ν(C-H); 1697: amide I (β-sheet); 1595: amide II (β-sheet); 1256: amide III (β-sheet), 1166: ν(C-ΟH). SFd 3477: ν(OH); 3075, 2980, 2936, 2880: ν(C-H); 1708, amide I (β-sheet); 1595: amide II (β-sheet); 1271: amide III (β-sheet), 1001: ν(C-ΟH) (see Fig. S2). Both TG-DTA curves (see Supporting Information Fig. S1) for the SF and the SFd show similar thermal-response behavior. At T < 110 °C the weight loss is attributed to the evaporation of water. The change from 170 °C to 275 °C can be assigned to the loss of other low-temperature volatile species, and the change from 275 °C to 400 °C is associated with the breakdown of side-chain groups of amino acid residues as well as the cleavage of peptide bonds of silk fiber, and at T > 400 °C it is attributed to fibroins’ degradation59. The TG% curves of SF and SFd exhibit a total weight loss of 98.4% and 98.9% respectively, in the range 20–700 °C with a broad exothermic peak at 400–600 °C assigned to the fibroins’ degradation. The DTA curve of SFd also shows an intense peak at 577 °C, originating from the amorphous sericin extraction59 and the degummed β-sheet fibroin degradation59.

The final ZrMOF@SFd hybrid was characterized by FT-IR and thermogravimetric analysis (TG-DTA) (see Fig. S1 in Supporting Information). The typical IR-peaks of SF are ≈1620, 1511, and 1226 cm−1, characteristic for amide I (C=O stretching), amide II (N–H deformation, and C–N stretching), and amide III (C–N stretching and N–H deformation). FT-IR (cm−1, selected peaks) ZrMOF@SFd: ν(OH); ν(C-H); ν(C=C); ν(C-OH); ν(C-Ο) (see Fig. S1 in Supporting Information). The TG-DTA curves for ZrMOF@SFd show a thermal-response profile similar to that of the SFd fibers. The TG% curve of ZrMOF@SFd exhibits a total weight loss of 97.3% in the range of 20–700 °C with a broad exothermic peak at 400–600 °C assigned to the fibroins’ degradation. The DTA curve of ZrMOF@SFd presents a shifting of the degradation temperature around T = 360 °C (compared to the SFd DTA curve at T = 320 °C), which is attributed to the combustion of the organic part of ZrMOF@SFd (estimated w:w ≈ 5.7%).

The FT-IR spectra for ZrMOF and AsIII-loaded ZrMOF (Fig. 2d) are very similar, indicating the retention of the structure of the ZrMOF after AsIII sorption. Noteworthy, in the IR spectrum of AsIII @ ZrMOF, there is a band around 740 cm−1 and 1040 cm−1, which is attributed to AsIII-O stretch from H3AsO325,48,60,61.

X-ray photoelectron (XPS) analysis was used to determine the AsIII-valence state and the eventual interaction between arsenic and the adsorbent. The high-resolution As3d XPS spectrum, shown in Fig. 2e, clearly indicates that AsIII is the only oxidation form adsorbed onto ZrMOF sorbent. The characteristic peak at 44.2 eV corresponds to AsIII in agreement with Sudhakar et al.62, while no peak corresponding to AsV is detected in As-loaded ZrMOF. This result shows that after adsorption of the AsIII on the ZrMOF, there is no oxidation event of AsIII,  thusall the bound As atoms on ZrMOF are in the AsIII  oxidation form. This information is in agreement with our FT-IR data, which detects the AsIII-O stretch, Fig. 2d, originating from H3AsO3. Also, the prevalence of the H3AsO3 form is corroborated hereafter by the adsorption-isotherms’ analysis, which shows that the adsorbed species is exclusively the neutral form of AsIII, i.e., H3AsIIIO3.

ZrMOF is a highly porous material with a 12-connected net based on [Zr6O4(OH)4] hexanuclear units interconnected via NH2-BDC2− ligands. We should note that pZrMOF is charged due to the protonation of amine groups (as the ZrMOF is prepared in acidified water). Prior to the AsIII sorption investigations, the ZrMOF was treated with MeOH/Et3N to deprotonate the  amine groups, thus resulting in a neutral framework.

A severe decrease of the Specific Surface Area of the ZrMOF is observed upon AsIII adsorption, see Fig. 3a,b. The non-linear [SSA vs. AsIII] trend in Fig. 3b, for ZrMOF can be analysed into two different domains: [i] at low AsIII-loading the SSA is decreased moderately, [ii] at high-AsIII loading there is an abrupt lowering of the SSA. This change in SSA is not due to alteration of the crystal structure of ZrMOF, as verified by PXRD, Fig. 2b. Thus, the severe decrease of SSA upon As-uptake provides important insight into the AsIII-uptake mechanism by ZrMOF as follows: the SSA of 610 m2 gr−1 for ZrMOF is equivalent to 6.1×1020 nm2 per gram of ZrMOF. The molecular volume of H3AsO3 in H2O has been estimated by Canaval et al.63 to be 75 ± 10Å3. Accordingly, each nm2 surface element of ZrMOF can accommodate not more than 2 H3AsO3 molecules, which gives an N-maximum of surface-adsorbed H3AsO3 molecules Nmax = 2 [H3AsO3 per nm2] × [6.1×1020 nm2 per gram of ZrMOF] ~ 1.5mmoles of H3AsO3 per gram of ZrMOF. When we compare this vs. the maximum AsIII-uptake capacity i.e. ~30 mmoles H3AsO3 per gram, we conclude that the experimental As-uptake is 20 times higher than the maximum AsIII-uptake capacity of 1.5 mmoles of H3AsO3 per gram, that would correspond to a mere surface coverage. Instead, the SSA drop vs. As-uptake data in Fig. 3b indicates a pore-filling mechanism, not a simple surface complexation. At the same time, the crystallinity of the ZrMOF material is retained after AsIII-uptake, see XRD in Fig. 2b. This result makes the ZrMOF behaving like an “AsIII-sponge” being capable of adsorbing unprecedented high-amounts, i.e., 2.2 grams of toxic AsIII per gram of ZrMOF.

Figure 3
figure 3

(a) Nitrogen sorption isotherms at 77 K for ZrMOF and AsIII@ZrMOF for different AsIII loadings (b) BET surface area vs. AsIII loading.

Full size image

AsIII-adsorption kinetics

Kinetic data of AsIII adsorption by ZrMOF, Fig. 4a, show fast kinetics with a non-linear time-profile. The kinetic data can be fitted by the Weber and Morris model, described by Eq. 264.

$$q(t)={K}_{in}{t}^{0.5}+C$$
(2)
Figure 4
figure 4

AsIII adsorption kinetics for (a) ZrMOF and (b) ZrMOF@SFd at pH 7. Symbols (, ) are experimental data. Lines are theoretical curves calculated using Eq. 2, with the parameters listed in Table S1. In (a) the added AsIII concentration was 50 mg L−1. In (b) the added AsIII concentration was 15 mg L−1.

Full size image

The Weber and Morris model64 is based on the key-assumption that diffusivity and mass- transfer phenomena are determining the adsorption process. In Eq. 2, the fittable parameters are the kinetic constant rate Kin (g g−1 h1/2), and C (g g−1) which is a constant depending on the type of the boundary layer64. Accordingly, the data in Fig. 4a can be fitted by considering two different sets of Kin and C, listed in Table S1 of the Supporting Information. At early adsorption times, (red circles in Fig. 4a), a fast kinetic constant Kin = 6 g g−1 h1/2 is obtained, with C = 0.11 g g−1 while at prolonged adsorption times, the kinetic constant is much lower Kin = 1.9 g g−1 h1/2, with C = 1.1 g g−1. This analysis reveals a two kinetic-phase phenomenon for AsIII uptake by ZrMOF. Taking into account the analysis of SSA data, we consider that the initial fast phase, corresponding to low-As uptake, is responsible for low decrease of SSA. At prolonged interaction times, where the adsorbed AsIII is high, a slower kinetic phase is operating, which corresponds to the sharp drop of SSA, i.e. the pores of ZrMOF are filled up with H3AsO3.

The same kinetic two-phase profile is observed for the ZrMOF@ SFd hybrid, Fig. 4b, indicating that the grafted ZrMOF particles operate similarly, i.e. surface adsorption of AsIII at low concentrations (red symbols in Fig. 4b) followed by pore filling at high AsIII-concentrations (green symbols in Fig. 4b). The silk fiber plays a minor role in As-uptake i.e. see adsorption isotherm in Fig. 5.

Figure 5
figure 5

AsIII adsorption isotherms for ZrMOF (, , •), ZrMOF@SFd (, , ■),pZrMOF(), and SFd (▲) at pH 7.

Full size image

AsIII-adsorption isotherms

A non-linear isotherm characterizes the uptake of AsIII by ZrMOF, see Fig. 5(, ), which can be analyzed in the two regions depending on the initial concentration of AsIII. [i] At low AsIII-concentrations (<25 mg AsIII Lt −1), the isotherm shows a Langmuir-like trend, see solid-red symbols in Fig. 5(). [ii] At increased initial AsIII concentrations, the isotherm data show a linear As-uptake isotherm, Fig. 5(). This trend continues up to 75–80 mg of added AsIII Lt −1. At even higher initial AsIII, the isotherm flattens, indicating a saturation of the AsIII-uptake by ZrMOF. Based on the data of Fig. 5, the maximum adsorbed AsIIIper gram of ZrMOF material corresponds to a maximum of 2200 mg AsIII gr−1 of ZrMOF. The ZrMOF@SFd material exhibited a similar two-isotherms profile, see Fig. 5(, ) When normalized [per mass of grafted ZrMOF], the AsIII -uptake data in Fig. 5(, ) show that the performance of the ZrMOF@SFd material is within ~10% comparable to ZrMOF. Thus, grafting of ZrMOF retains its AsIII -uptaking capacity. For reference, pZrMOF and the SFd alone Fig. 5(), (▲) show a very low AsIII -uptake, i.e. 0.260 g g−1, and 0.068 g g−1 respectevelly. The significant inhibitory effect of the cation sites in cationic pZrMOF, to AsIII uptake i.e. vs the neutral ZrMOF, reveal that the surficial NH2 sites play key role in the uptake mechanism. This is further analyzed in the following in the theoretical surface Complexation Modeling hereafter.

Theoretical modeling of the data in Fig. 5 can be done using two isotherm-adsorption models. [a] At low added-AsIII concentrations, a Langmuir equation (Eq. 3) describes adequately the process, see fit (red line in Fig. 5) where qm (mg g−1) is maximum AsIII adsorption, qLADS (mg g−1) is the surface concentration of adsorbed AsIII species in materials. Ce (mg L−1) is the initial As-concentration. KLangmuir is the Langmuir stability constant representing the strength of AsIII-binding of the ZrMOF surface65.

$${q}_{{\rm{ADS}}}^{L}=\frac{{{q}_{m}}^{\ast }{{K}_{{\rm{Langmuir}}}}^{\ast }{C}_{e}}{1+{{K}_{{\rm{Langmuir}}}}^{\ast }{C}_{e}}$$
(3)

[b] At increased added-AsIII concentrations, we consider a linear Freundlich-type isotherm (Eq. 4)

$${q}_{{\rm{Ads}}\,}^{{\rm{Part}}}={K}_{{\rm{part}}}{C}_{e}^{\frac{1}{n}}$$
(4)

where qPart gives the bound AsIII -moieties in the ZrMOF, in mg g−1. The index n reflects a constant related to the intensity of sorption or the degree of the dependence of sorption on concentration. The efficiency of uptake is classified according to the value of Kpart. The linear Freundlich-type adsorption isotherms can be used to describe pore filling/partitioning processes66 in hydrophobic/hydrophilic interfaces.

Using the two isotherms Eqs. 3 and 4, the data in Fig. 5 can be fitted (see solid lines in Fig. 5) with the parameters listed in Table 1. More specifically, at low AsIII-concentrations, the uptake capacity -due to surface complexion- can achieve a maximum of qm = 0.83 g g−1of AsIII at pH = 7. At high AsIII-concentration, where pore filling is operating, Fig. 5 (green line) a maximum As-uptake is attained of qe = 2.2 g g−1of AsIII at pH = 7.

Table 1 Parameters for Langmuir isotherms and Freundlich isotherms, used to fit the experimental data for AsIII binding onto pZrMOF, ZrMOF, SFd, and ZrMOF@ SFd at pH 7.
Full size table

Further analysis of the surface adsorption for the AsIII species can be done by modeling of the pH-dependent As-uptake on the ZrMOF. This analysis, detailed in our previous works16,24,67, is described in Supporting Information, Fig. S4. According to Fig. S4, the pH-dependent profile for low AsIII-concentrations shows that AsIII binds in its neutral form H3AsO3 at the neutral amino-sites of neutral ZrMOF. This result is in agreement with Georgiou et al.16, Gupta et al.27, Su and Puls28. The surface amines ≡NH2 act a specific binding sites for AsIII-uptake, see reaction (15), and (17) ≡NH2 + H3AsO3 ↔ ≡NH2-[H3AsO3] in Table S2. We underline that in the cationic pZrMOF, the protonated ≡NH3+ sites, reaction (12) in Table S2, do not favor adsorption of AsIII and this is the origin of the inferior performance of pZrMOF vs. ZrMOF. This is structurally described in Fig. S4d.

Since, in natural waters, several ions may coexist with arsenic, these can potentially compete with As-uptake68. In this context, the impact of competing ions such as PO43−, CO2−3, NO3−, SO2−4, Cl− and HCO3− on AsIII adsorption was studied, see Fig. S5 in the Supporting Information. The data in Fig. S5, indicate that ZrMOF and ZrMOF@ SFd effectively remove AsIII even in the presence of CO3, NO3−, SO42–, Cl− and HCO3−. The stronger inhibitory effect is exerted by PO43− ions which may inhibit AsIII adsorption by 40% and 60% for ZrMOF and ZrMOF@SFd, respectively. The results are well agreement with Sudhakar62 et al. and Jain and Loeppert69, which point out that the natural water ions do not affect the AsIII adsorption except for PO43 which destabilizes the MOF structure.

Finally, we have evaluated the possibility of reusing the ZrMOF and ZrMOF@SFd materials after regeneration. Thus, we have applied the regeneration protocol24,70, which involved high ionic-strength treatment. More particularly, the protocol involves incubation for 24 hours under stirring at a pure aqueous solution containing 1 M of KNO3. Our data show that the AsIII adsorbed on either ZrMOF or ZrMOF@SFd cannot be removed by this treatment, indicating the high stability of the bound AsIII, i.e. attributed to its irreversible penetration into the pores of the ZrMOF.

Comparison of AsIII-uptake with similar metal-organic framework materials

Figure 6 summarizes a comparison of AsIII sorption by the present materials vs. other pertinent MOF-based materials reported in the literature. According to Fig. 6, ZrMOF supersedes by far any known material.

Figure 6
figure 6

Maximum AsIII adsorption capacity (g g−1) at waters’ near-neutral pH of some adsorbents reported in the literature compared with the present materials: 1) ZIF-8(cubic)77, 2) ZIF-8(leaf)77, 3) ZIF-8 (dodecahedral)77, 4) Fe3O4@ZIF-878,5) HCl-UiO-66(SH)279, 6) CoFe2O4@MIL-100(Fe)48, 7) Fe3O4@MIL-10180, 8) MIL-100(Fe)16, 9) M60016, 10) M80016, 11) M90016, 12) ZrMOF(this work), 13) cationic pZrMOF (this work), 14) SFd (this work), 15)neutral ZrMOF@SFd (this work).

Full size image

This result is attributed to the fundamentally different mode of action of the neutral ZrMOF, i.e. the partitioning-like mechanism resulting in pore-filling allows exploitation of the full pore volume as a “sponge” for the uptake of the AsIII species form solution.

Conclusions

Using XPS, FTIR, BET-porosimetry data, with theoretical Surface-Complexation-Modeling (SCM), we report a two-step  phenomen non  which boosts high-AsΙΙΙ-uptake. First, at low AsIII-concentrations, surface-complexation of H3AsO3 results in AsIII-coated voids of neutral ZrMOF, and subsequently, at high AsIII-concentrations, the AsIII-coated voids of neutral ZrMOF are filled-up by H3AsO3 via a partitioning-like mechanism. Also, we present an innovative hybrid-material, ZrMOF@SFd operating like an “AsIII-sponge” with unprecedented efficiency of 1800 mg AsIII gr−1. ZrMOF@SFd consists of a Zirconium Metal-Organic Framework [ZrMOF] covalently grafted on SFd. ZrMOF itself exhibits AsIII adsorption of 2200 mg gr−1, which supersedes any -so far- known AsΙΙΙ-sorbent. The reference materials i.e. cationic-pZrMOF and SFd play secondary role in AsIII-adsorption with adsorption capacity 260 mg AsIII gr−1 and 68 mg AsIII gr−1 respectively. Finally, the present research exemplifies for the first time a novel concept of a “partitioning-like” mechanism, operating for adsorption of H3AsO3 , by neutral metal oxide materials. So far, such a mechanism has been conceptualized only for the uptake of non-polar organics by natural organic matter or synthetic polymers.

Methods

Materials

Sodium meta-arsenite NaAsO2 was obtained from Sigma-Aldrich, while HCl, NaOH, KNO3, and Cu (NO3) ∙ 3H2O obtained from Merck. 2-(N-Morpholino)ethanesulfonic acid hydrate, 4-Morpholineethanesulfonic acid (call MES hydrate) & 4-(2-Hydroxyethyl)piperazine-1-ethanesulfonic acid, N-(2-Hydroxyethyl)piperazine-N-(2-ethanesulfonic acid)(call HEPES), used for pH buffering, were obtained from Sigma-Aldrich. Milli-Q Academic system, Millipore produced ultrapure water.

The Silk Fabric (SF) provided by Tsiakiris Georgios Silk Company, Alexandroupoli, Greece. Sodium carbonate (Na2CO3) purchased from Riedel de Haën. The coupling agent 3-(chloropropyl)trimethoxysilane was provided by Fluka. Methanol and ethanol purchased from Merck and diethyl ether from Sigma Aldrich.

All reagents were of analytical reagent grade purity, and all solutions prepared using deionized water obtained with a Milli-Q system with a conductivity of 18.2 μS cm−1.

ZrMOF preparation

The protonated (cationic) [Zr6O4(OH)4(NH3+-BDC)6]Cl6 ∙ 35H2O (herein codenamed as pZrMOF) was synthesized as described previously71. To prepare the neutral material, pZrMOF (100 mg, 0.038 mmol) was treated with Et3N (72.6 mg, 0.7 mmol) in MeOH (4 mL) for 1 h. The resulting solid [Zr6O4(OH)4(NH2-BDC)6]∙xMeOH∙yH2O (herein codenamed as ZrMOF) was then isolated by filtration, washed with MeOH and dried in the air. Yield: 89%.

Degumming process of Silk Fabric (SFd)

The SF cut in pieces of 3.3 × 0.9 cm (≈15 mg), which were immersed into a 200 ml round-bottom flask and were degummed in a 0.05 wt.% Na2CO3/H2O solution at 90 °C for 30 min and then rinsed thoroughly with double distilled water to extract the sericin protein and other impurities. This process repeated three times to obtain pure silk fibroin fibers (SFd). The degummed silk fibroin fibers (SFd) dried at 40 °C under atmospheric pressure.

Covalent grafting of ZrMOF on SFd (ZrMOF@SFd)

A solution of ZrMOF (0.066 mmol) in 10 ml of methanol prepared for sonicating to achieve a good suspension in the dispersion media. After 0.022 mmol of 3-(chloropropyl) trimethoxysilane sonicated added in the solution, and finally, the reaction mixture refluxed at 60 °C for 48 h. The molar ratio of ZrMOF/silane was 3:1. To this, 30 mg of degummed SFd fibers and 5 ml of ethanol were added and refluxed at 60 °C for 24 h. The degummed SFd fibers (30 mg) immersed into 10 ml of ethanol for 24 h, before modification with ZrMOF/silane. After cooling at room temperature, the resulting material, ZrMOF@SFd was washed several times with methanol, ethanol, and diethyl ether and dried under vacuum at 40 °C for 24 h.

Physical characterization of materials

EDS analysis for ZrMOF showed no Cl confirming the complete deprotonation of ammonium groups. The powder X-ray diffraction (PXRD) measured at room temperature on an STOE-STADIMP powder diffractometer. PXRD equipped with an asymmetrically curved Germanium monochromator (CuKα1 radiation, λ = 1.54056 Å) and a one-dimensional silicon strip detector (MYTHEN2 1 K from DECTRIS). The line focused Cu X-ray tube operated at 40 kV and 40 mA. Powder of each sample was packed in a 1 mm diameter polyimide capillary (polymer substrate with neither Bragg reflections nor broad peaks above 10 °) and measured in Debye-Scherrer geometry on a spinning stage (~200 rpm). Intensity data from 3 to 125 degrees 2Θ collected for 17 h with a step of 0.005 degrees. The instrument calibrated against a NIST Silicon standard (640d) before the measurement.FT-IR spectra were recorded on KBr pellets in the 4000-400 cm−1 range using a Perkin-Elmer Spectrum GX spectrometer. Thermogravimetric analyses (TGA) were performed on a NETZSCH STA 449 C system. Sample analysis was conducted from 25 to 900 °C in an air atmosphere (50 mL min−1 flow rate) with a heating rate of 10 °C min−1. Scanning electron microscope (SEM) performed by a JEOL JSM-6390LV equipped with an Oxford INCA PentaFET-x3 energy-dispersive X-ray spectroscopy (EDS) detector. Data acquisition performed with an accelerating voltage of 20 kV and 120 s accumulation time. Then the images were taken with a field emission JEOL JSM 7000 F electron microscope operating at 15 kV accelerated voltage. The samples were sputter-coated with 5–10 nm also reduce charging by used Au film.

X-Ray Photoelectron Spectroscopy (XPS) measurements were using a SPECS GmbH. Instrument equipped with a monochromatic MgKα source (hν = 1253.6 eV) and a Phoibos-100 hemispherical analyzer. The spectra were recorded under ultra-high vacuum s with a base pressure of 2–5 × 10–10 mbar. Prior to measurement, the samples were placed on silicon substrates under high vacuum, before being placed in the main chamber for XPS measurement. The take-off angle was 45°. The recorded spectra were the average of three scans, with energy step 0.1–0.2 eV and a dwell time of 1 s. The As3d binding energy is calibrated based on the C1s core level at 284.5 eV. The spectral analysis included [i] a Shirley background subtraction, and [ii] peak deconvolution employing mixed Gaussian–Lorentzian functions in a least-squares curve-fitting program (WinSpec, Laboratoire Interdisciplinaire de Spectroscopie Electronique, University of Namur, Belgium)72,73.

N2 adsorption isotherms were measured at 77 K using a Quantachrome NOVAtouch LX2. Before analysis, all samples were degassed at 80 °C under vacuum (<10–5Torr) for 16 h. The specific surface areas were calculated by the Brumauer-Emmett-Teller (BET) method using the N2-adsorption data points, in the relative-pressure range P/Po of 0.05–0.35.

AsIII analytical determination

The concentration of AsIII in the aqueous solution determined by square wave Cathodic Stripping Voltammetry (SW-CSV) using a Trace Master5-MD150 polarograph by Radiometer Analytica. SW-CSV is well suited for the analytical determination of AsIII 22,74 with a low detection limit (0.5 μg L−1). The measuring borosilicate glass cells obtained from Radiometer Analytica. The working electrode was a hanging mercury drop electrode (HMDE) with drop diameter of 0.4 mm generated by a 70 μm capillary. An Ag/AgCl electrode with a double liquid junction used as the reference electrode with a Pt measuring electrode. Importantly, samples were not purged with N2 gas to avoid the loss of AsIII. During the stripping step, the solution stirred at 525 rpm. All measurements we performed using aliquots of 8.3 mL shifting at pH < 0.5 by 1.5 mL from 6.66 M of HCl and final 2 M concentration in the electrochemical cell, then 8 ppm of Cu2+ were added. In the following, AsIII was determined by SW-CSV with accumulation potential E = −400 mV and accumulation time in the 60 s. AsIII quantified by its signal at E1/2 = −670 mV16,19,22.

AsIII adsorption experiments

For the kinetic measurements, AsIII uptake from aqueous solutions studied in batch experiments. The kinetics of AsIII adsorption using ZrMOF studied as follows: 4 10–4 g L−1 of ZrMOF were dispersed in 50 mL buffered aqueous solution in polypropylene tubes at pH 7, in the presence of 50 mg L−1 AsIII. For samples, ZrMOF @ SFd 3.4 mg was dispersed in 25 mL buffered aqueous solution in polypropylene tubes at pH 7, in the presence of 15 mg L−1 AsIII. The time-evolution of AsIII concentration was monitored at contact times ranging between 0–240 min and 0–960 min, respectively. At the end of each contract period, all sample centrifugation and the supernatant solution analyzed for AsIII. To determine the adsorption rates of AsIII, the amount of AsIII adsorbed at time t, q (mg AsIII g−1), calculated from the mass-balance between the initial concentration and the concentration at time t onto the solid adsorbents16,29.

Adsorption isotherms for pZrMOF and ZrMOF were recorded at pH 7 in the presence of 0–100 mg L−1 NaAsO2 interacting with 0.1 g L−1 pZrMOF and 0 to 150 mg L−1 NaAsO2, interacting with 4 10–4 g L−1 of ZrMOF suspended in 50 mL buffer solution in polypropylene tubes. On the other hand, for SFd and ZrMOF @ SFd, 0 to 100 mg L−1 and 0 to 150 mg L−1 NaAsO2 and 0.2 g L−1,0.14 g L−1 respectively were suspended in 25 mL buffer solution in polypropylene tubes.

pH-dependent (pH-edge) experiments allow a detailed probing of the interfacial adsorption mechanisms16,29, while adsorption isotherms report the maximum uptake capacity. In this work, pH-edge adsorption experiments carried out for an initial concentration of 5, 5, and 15 mg L−1(NaAsO2) and also 0.1 g L−1,0.4 10–4 g L−1, 0.14 g L−1 of pZrMOF, ZrMOF and ZrMOF @ SFd respectively, suspended in 50 mL buffer solution whose pH adjusted in the range 4 to 8, in polypropylene tubes.

After metal addition, the suspension was allowed to equilibrate for 15 min ZrMOF, and pZrMOF at RT, while agitated using a magnetic stirrer. After completion of equilibration, the suspensions centrifuged at 6000 rpm for 10 min, and the supernatant solutions were analyzed for AsIII as described above. For SFd and ZrMOF @ SFd after metal addition, the suspension was allowed to equilibrate overnight at RT while using a magnetic stirrer. After completion of equilibration, the SFd or ZrMOF @ SFd suspension collects by metal tweezer.

Reuse experiments were also conducted for ZrMOF, which had adsorbed AsIII at pH 7. To reuse the samples, we had to desorb the adsorbed As. Thus, following the method used in ref. 30,75, after AsIII adsorption, the material was immersed in an aqueous solution of 1 M KNO3 for 24 h, and the supernatant was analyzed for AsIII release24,70,76. Similarly, the ZrMOF @ SFd once loaded with AsIII were washed at high ionic strength 1 M KNO3 and the supernatant was analyzed for AsIII release.

Control experiments (without ZrMOF, pZrMOF, SFd, and ZrMOF @ SFd) showed no loss of initial AsIII. The initial pH values of buffer solutions were adjusted to the requested using small volumes of 1 M HCl or 1 M NaOH. It should mention that HCl is inert towards AsIII in voltammetric measurements, the pH drift of each suspension, i.e. measured at the beginning and the end of incubation, was <0.2 pH units.