Oxalamide Based Fe(II)-MOFs as Potential Electrode Modifiers for Glucose Detection

Abstract

In an attempt to expand the coordination chemistry of N,N′-bis(2,4-dicarboxyphenyl)-oxalamide (H₆L) ligand, we isolated and structurally characterized two new Fe(II) Metal-Organic Frameworks (MOFs), namely [Fe₂(H₂L)(H₂O)₅] (3D-Fe-MOF) and [Fe(H₄L)(H₂O)₂]∙2H₂O, (2D-Fe-MOF) by carefully adjusting the reaction conditions to achieve the optimal degree of deprotonation of the bridging ligand. Both MOFs were found stable in water, as evidenced by powder X-ray diffraction data and their ability to sorb glucose (GLU) from either an aqueous solution or artificial sweat was investigated only to show negligible sorption. A graphite paste sensor (GPE) using the 3D-Fe-MOF as a modifier was fabricated. The 3D-Fe-MOF modified GPE was assessed for non-enzymatic GLU detection in aqueous solution at pH 6 via differential pulse voltammetry and the preliminary results were discussed.

1. Introduction

Diabetes mellitus is a serious disease, resulting from the insufficiency of insulin in the body, which causes elevated blood-glucose (GLU) levels (hyperglycemia) or reduced GLU concentrations (hypoglycemia). The continuous monitoring of blood GLU levels in diabetic patients is an important matter, as shown by the development of electrochemical GLU biosensors throughout the last two decades [1,2]. The majority of the electrochemical biosensors, reported in the literature, focus on monitoring the blood GLU levels, which includes the painful and inconvenient process of blood samples for the patients [3]. Nevertheless, more recently, scientists have also turned their interest to sweat sampling and wearable sensors, not only for GLU but for other biomarkers too, thanks to the ease of sampling from the skin surface, as the GLU blood levels are associated with the sweat levels [4,5,6]. GLU sensors are divided into enzymatic and non-enzymatic ones, with the first ones being based on the presence of enzymes such as GLU oxidase, hexokinase, and GLU dehydrogenase [1,7]. On the other hand, non-enzymatic sensors are broadly adaptable, providing stability, simplicity, and reproducibility, while are not affected by temperature, pH, or humidity [8,9,10].

The extended research around electrochemical sensors has led to the exploitation of different kinds of electrode modifiers, such as metal oxides, metal compounds, and recently, metal-organic frameworks (MOFs), in either enzymatic or non-enzymatic sensors [2,11,12]. MOFs are crystalline, porous materials composed of metal ions or metal clusters and polytopic bridging organic linkers, demonstrating fascinating properties such as high surface area, thermal stability, tunable pore size [13], and magnetic properties [14]. They have potential applications in many fields including catalysis, gas storage, energy storage, environmental pollution, sensing platforms, catalysis, photocatalysis, oxidation, hydrogenation, and separation due to their diverse active sites and latterly as electrode modifiers for GLU sensing in blood serum or sweat samples [13,15,16,17,18,19,20,21]. In the case of enzymatic sensors, MOFs are usually used as an immobilization matrix for the appropriate enzyme, with the redox abilities of the electrode to be enhanced by the metal centers of the MOFs. Such an example is the case of the HP-MIL-88, a three-dimensional (3D) Fe-based MOF, which was functionalized with boronic acid, affording an efficient immobilization matrix for GLU oxidase [22]. The non-enzymatic sensors exploit MOFs’ electrocatalytic ability for the electrooxidation of the GLU by mimicking the enzyme functionality, leading to self-supported electrodes for GLU detection with low detections limits and high sensitivities [21,23,24,25,26]. The most reported MOF-modified electrodes concern MOFs that are based on Cu and Ni and lesser on Co and Fe. MOF-14, a 3D porous Cu-MOF, was utilized as a non-enzymatic sensor for the electrochemical determination of GLU, by Zhang and his research team, exhibiting significant stability in alkaline solution and high electrocatalytic activity towards GLU oxidation [23]. Xuan et al. developed a non-enzymatic sweat GLU sensor utilizing a Ni-based MOF, achieving a low detection limit in sweat samples [21]. Fe-based MOFs have also been reported as modifiers in non-enzymatic electrochemical GLU sensors, however, only in blood or serum samples. MIL-53(Fe), an iron-based rod-MOF, was found to enhance the chemiluminescence (CL) of luminol in the presence of H₂O₂ in an alkaline medium [19]. The combination with the GLU oxidase led to an economical, sensitive, and selective CL method for GLU sensing. Furthermore, Fe-MIL-88-NH₂, a MOF based on a trinuclear iron cluster, demonstrates peroxidase-like activity and exhibits high catalytic efficiency and stability as a colorimetric detector for GLU [26] in serum samples.

Our research group focuses on the synthesis of new MOFs based on oxalamide ligands and utilizing them for different purposes such as heavy metal ion sorption, magnetism, and more recently as modifiers in electrochemical sensors [14,27,28,29,30,31,32,33,34]. Herein we report the synthesis, crystal structures, and characterization of a two-dimensional (2D) and a 3D Fe(II)-MOFs which are based on a previously reported oxalamide ligand, H₆L [27]. The two Fe(II)-MOFs were studied towards GLU sorption in aqueous solutions, as well as in artificial sweat solutions. In addition, the 3D-Fe-MOF was also tested as a modifier in a graphite paste electrode (GPE) for the detection of GLU. Preliminary results from the electrochemical application of the modified GPE are discussed.

2. Materials and Methods

2.1. Synthesis and Characterization

All procedures were performed under aerobic conditions. Solvents and reagents were obtained from commercial sources and used as received. The ligand, N,N′-bis(2,4-dicarboxyphenyl)-oxalamide (H₆L), was synthesized following our previously reported method [27]. IR spectra (Figures S1 and S2) were recorded on a Shimadzu FT/IR IRAffinity-1 spectrometer with samples prepared as KBr pellets. Powder X-ray diffraction (pxrd) patterns (Figures S3–S7) were collected on freshly prepared samples on a Bruker D8 Advance X-ray diffractometer (CuKa radiation, λ = 1.5418 Å). TGA diagrams (Figures S8 and S9) were recorded on a Mettler-Toledo TGA/DSC1 instrument under a N₂ flow of 50 mL/min.

Single Crystal Synthesis of [Fe₂(H₂L)(H₂O)₅], (3D-Fe-MOF)

Here, 0.06 mmol (24 mg) of Fe(NO₃)₃∙9H₂O was dissolved in 9 mL H₂O. The yellow clear solution was added to an autoclave containing an N-methyl-2-pyrrolidone (NMP) solution (3 mL) of H₆L 0.06 mmol (25 mg) and 0.24 mmol of LiOH. The autoclave of the yellow blur solution was sealed and placed in an oven at 140 °C. After 2 days yellow needles were formed and collected. FT-IR (KBr disks): 3505 (s), 3419 (s), 3193 (m), 2962 (m), 1703 (m), 1686 (m), 1668 (m), 1653 (m), 1559 (s), 1508 (s), 1451 (s), 1412 (s), 1371 (s), 1301 (m), 1224 (m), 1177 (m), 1080 (w), 961 (w), 922 (w), 908 (w), 895 (w), 866 (w), 813 (m), 777 (m), 692 (w), 557 (m), 527 (m), 497 (m), 419 (m), 402 (w).

Scale-up Synthesis of the 3D-Fe-MOF

Then, 0.8 mmol (323 mg) of Fe(NO₃)₃∙9H₂O was dissolved in 24 mL H₂O. The yellow clear solution was added to an autoclave containing an NMP solution (8 mL) of H₆L 0.4 mmol (194 mg) and 1.6 mmol of LiOH. The autoclave of the yellow blur solution was sealed and placed in an oven at 140 °C for 3 days to afford yellow needles. The needles were collected by vacuum filtration, washed extensively with H₂O (5 × 50 mL), and dried in air. Yield was 82%.

Single Crystal Synthesis of [Fe(H₄L)(H₂O)₂]∙2H₂O, (2D-Fe-MOF)

In the next step, 0.06 mmol (24 mg) of Fe(NO₃)₃∙9H₂O was dissolved in 9 mL H₂O. The yellow clear solution was added to an autoclave containing an NMP solution (3 mL) of H₆L 0.06 mmol (25 mg) and 0.24 mmol of Et₃N. The autoclave with the yellow blur solution was sealed and placed in an oven at 140 °C. After 2 days yellow needles were formed and collected. FT-IR (KBr disks): 3459 (s), 3119 (m), 2930 (m), 2798 (m), 2637 (m), 2538 (m), 2377 (m), 1714 (s), 1677 (m), 1619 (m), 1579 (s), 1508 (s), 1458 (s), 1420 (s), 1384 (s), 1294 (s), 1264 (m), -, 1222 (m), 1172 (m), 1129 (m), 1080 (m), 947 (m), 913 (m), 868 (m), 841 (m), 802 (m), 765 (m), 698 (m), 565 (m), 533 (m), 500 (m), 421 (w), 405 (w), 399 (w).

Scale-up Synthesis of the 2D-Fe-MOF

Finally, 0.2 mmol (81 mg) of Fe(NO₃)₃∙9H₂O was dissolved in 27 mL H₂O. The yellow clear solution was added to an autoclave containing an NMP solution (8 mL) of H₆L 0.2 mmol (98 mg) and 0.4 mmol of LiOH. The autoclave of the yellow blur solution was sealed and placed in an oven at 120 °C for 3 days to afford yellow needles. The needles were collected by vacuum filtration, washed extensively with H₂O (5 × 50 mL), and dried in air. Yield was 65%.

2.2. Single Crystal X-ray Crystallography

The data for the single crystals of the two complexes were collected on a Bruker D8 VENTURE diffractometer (University of Crete), equipped with a PHOTION II CPAD detector, utilizing Cu-Kα (λ = 1.54178 Å) radiation from an IμS micro-focus source. Suitable crystals were mounted on a Hampton cryoloop with Paratone-N oil and transferred to a goniostat for data collection. Data integration and reduction were performed using SaintPlus 7.68 A implemented in APEX3 [35]. Absorption correction was performed by the multi-scan method implemented in SADABS [36]. Space groups were determined using XPREP implemented in APEX3 [31]. The structures were solved by direct methods using SHELXT and refined by full-matrix least-squares techniques on F2 with SHELXL contained in the OLEX program package [37,38]. The non-H atoms were treated anisotropically, whereas most H atoms, except some that were found from the Fourier map, were placed in calculated, ideal positions and refined as riding on their respective atoms. Selected crystal data and parameters for data collection are reported in Table S1.

2.3. Sorption Studies

Experimental details concerning the preparation of the standards and the instrumental analysis, including selected reaction monitoring (SRM) transitions of GLU and GLU-¹³C (Table S2) and representative chromatograms of the samples (Figures S10–S12) are given in the supplementary materials.

A GLU solution (6 mM) was prepared by dissolving 0.2706 g of GLU in 250 mL ultra-pure H₂O and its pH value was adjusted to 4.0 with the appropriate addition of hydrochloric acid 0.01 M. The GLU solution was left for 48 h at 4 °C in order to complete its isomerization process. The same procedure was repeated for the artificial sweat sample. The artificial sweat contained 22 mmol L⁻¹ urea, 5.5 mmol L⁻¹ lactic acid, 3 mmol L⁻¹ NH₄Cl, 0.4 mmol L⁻¹ CaCl₂, 50 μmol L⁻¹ MgCl₂, 8 mmol L⁻¹ KCl, and 80 mmol L⁻¹ NaCl and 25 μmol L⁻¹ uric acid (UA). In the above solution, we added 400 μmol L⁻¹ GLU. Then, 20 mg of each Fe-MOF absorbent was weighed, and the addition of GLU’s solution (10 mL) was followed. The containers were sealed, and the solutions were stirred for four hours. The supernatants were received and centrifuged for 10 min at 3000 rpm. The supernatant solutions were filtered using a cotton filter and the filtrates were transferred to glass containers where they were stored in the refrigerator (4 °C). The samples were filtered using regenerated cellulose (RC) syringe filters. Appropriate dilutions of the samples (GLU solutions and solutions of artificial sweat) were made with an ACN:H₂O, (80:20, v/v) solvent mixture (ACN = acetonitrile) so that their concentrations were within the linear range of the GLU standards used for their quantification. The internal standard (IS) was used to correct for possible instrumental instability.

Six working solutions with various concentrations ranging from 0.5–40 mg L⁻¹ were prepared via appropriate dilution of a GLU solution (100 mg L⁻¹) in ACN:H₂O, (80:20, v/v), while GLU-¹³C (IS) was added to each solution at the same concentration (10 mg L⁻¹). The ratio of the peak area of GLU to the peak area of the IS was used for the construction of the calibration curve. Additionally, the matrix effect has been investigated through the preparation of matrix-matched samples at the concentration of 10 mg L⁻¹.

The final concentrations of the GLU solutions were determined with UPLCMS/MS. Finally, the sorption capacity (S in mg_GLU/g_MOF) was determined by the following equation:

$$\mathrm{S}=\frac{\left({\mathrm{C}}_{\mathrm{i}}{ - \mathrm{C}}_{\mathrm{f}}\right)·\mathrm{V}}{{\mathrm{m}}_{\mathrm{MOF}}}$$

(1)

where V is the volume of GLU solution (0.010 L) where the material was dispersed and m_MOF is the weight of the material (0.020 g).

2.4. Electrochemical Application

A stock solution of 0.1 M GLU was prepared in water and stored at 4 °C until use. Phosphate buffer 0.10 M, pH 6.0 was prepared by dissolving the appropriate amounts of Na₂HPO₄∙2H₂O (Chem Lab) and NaH₂PO₄∙2H₂O (Chem Lab) in water, and the pH value was adjusted using 1 M HCl solution. For the preparation of all solutions, Milli-Q-grade water was used.

Fabrication of the modified graphite paste electrode

Modified graphite paste electrode (GPE) was fabricated by mixing 0.95 g graphite powder (grade #38, Fisher Scientific, Pittsburgh, PA, USA), 0.05 g of 3D-Fe-MOF and 0.4 g paraffin (Merck, Rahway, NJ, USA). The resulting paste was filled up in a plastic syringe and a copper wire passing through the paste was used to achieve electric contact. The surface of the modified GPE was flattened by pressing the edge of the syringe on tissue paper, while the electrode surface was renewed via a slight pressure on the syringe plunger and polishing it on filter paper for further use.

Voltammetric measurements

All electrochemical measurements were carried out with a portable EmStat3 potentiostat (Palm Sens, Houten, The Netherlands) combined with the PS Trace 4.2 software (Palm Sens). A three-electrodes system was used with the modified GPE acting as the working electrode, a saturated calomel electrode (SCE) as a reference electrode and a Pt wire as the counter electrode. Differential pulse (DP) voltammetric measurements were carried out under the following conditions: the potential range was from −1.40 to 1.9 V, the modulation amplitude was 0.05 V, the modulation time was 0.02 s, the step potential was 2 mV, and the scan rate was 20 mV/s. All measurements were performed in a voltammetric cell containing 25.0 mL of the working solution. The peak currents of voltammograms were used for the quantitative detection of GLU.

3. Results and Discussion

3.1. Synthesis

As mentioned above [14,27,28,29,30,31,32,33,34], we investigated oxalamide ligands for the synthesis of new MOFs. Given that N,N′-bis(2,4-dicarboxyphenyl)-oxalamide (H₆L) has already been reported to react with alkaline earths, lanthanides, and 3d transition metal, we extended our studies with iron salts from, which we obtained the two new iron-based MOFs after conducting a large amount of solvothermal experiments. The solvents that were tested were dimethylformamide (DMF), dimethylacetamide (DMA), N-methyl-2-pyrrolidone (NMP), water (H₂O), methanol (MeOH) and ethanol (EtOH) and mixtures of them. To this end, we found that both iron-MOFs were best prepared by a 3:1 mixture of H₂O and NMP in the presence of a base.

For the deprotonation of the ligands, we tested both inorganic and organic bases, such as LiOH, NaOH, KOH, Et₄NOH, and Et₃N. In the case of the 3D-Fe-MOF, four equivalents of LiOH led to the tetra-anion of the ligand (H₂L⁴⁻) while the ligand in the 2D-Fe-MOF is only twice deprotonated (H₄L²⁻). It is worth noting that four equivalents of the weak base Et₃N were not enough to deprotonate the oxalamide ligand four times. The degree of deprotonation of the ligand is crucial for the formation of the final product since the H₄L²⁻ led to the 2D-Fe-MOF, while the H₂L⁴⁻ to the 3D-Fe-MOF. Besides the kind and amount of base used, concentration, as well as the temperature, affect the final product, as discovered during the scale-up experiments. For the scale-up synthesis of the 3D-Fe-MOF, only the adjustment of the equivalents of iron nitrate to H₆L to 2:1 (and four equivalents of LiOH) was needed to succeed with an 82% yield. However, when we tried to scale up the synthesis of the 2D-Fe-MOF the increased concentration, the four equivalents of Et₃N, and the temperature of 140 °C led to the 3D-Fe-MOF instead. To this end, after several attempts, we managed to scale up the synthesis of the 2D-Fe-MOF with the use of two equivalents of LiOH at a temperature of 120°C which led to pure 2D-Fe-MOF, with a yield of 65%.

Although it has been stated that the influence of metal-ligand ratios on the structure of the resulting compounds is important [39], in this case, the degree of deprotonation of the ligand has a stronger effect than the metal-ligand ratio.

The reduction of the trivalent iron that was used in the first place is also noteworthy. Trivalent iron nitrate was selected for the first experiments because of its good solubility in water and its low price. However, in both products, Fe(III) was reduced to Fe(II) as confirmed by the bond-valence sum (BVS) calculations (Table S3). This kind of iron reduction has already been reported in the literature several times and it is possibly attributed to the usage of NMP at high temperatures [40,41,42,43]. We also tried to obtain the products by reacting H₆L with divalent iron salts such as FeCl₂∙4H₂O and Fe(ClO₄)₂∙6H₂O. Although both salts afforded the two MOFs, yields were considerably reduced (i.e., 5–10%) and the scale-up experiments showed colored impurities, especially, in the case of the FeCl₂∙4H₂O. Therefore, the best results were obtained by the use of the trivalent iron salt, Fe(NO₃)₃∙9H₂O, with the pxrd analysis (Figures S5 and S6) revealing high purity and water stability (the products were extensively washed with H₂O during filtration) for each product.

3.2. Description of Structures

3D-Fe-MOF crystallizes in the triclinic space group P-1. The asymmetric unit comprises two crystallographically independent Fe(II) ions, a tetra-anion of the organic ligand (H₂L⁴⁻), and 5 coordinated water molecules (Figure S13). Both metal centers adopt a slightly distorted octahedral geometry. The Fe1 is surrounded by four carboxylate O atoms and two O atoms from two terminal H₂O (Figure S14). The coordination sphere of Fe2 contains three carboxylate O atoms and three terminal H₂O (Figure S14). The H₂L⁴⁻ ligand adopts a trans conformation bridging seven different iron ions (Figure S15). There are three kinds of carboxylates. Two of the four carboxylates (C1, C13) of the ligand act as μ₂-η¹:η¹ bridges leading to the formation of two alternating Fe1–Fe2 rods down the crystallographic axis a. The third carboxylate C18 acts, also, as a μ₂-η¹:η¹ bridge between Fe1 and Fe2 belonging to two different rods leading to a double rod ladder. The final carboxylate C8 acts as a terminal monodentate O donor. In this arrangement, a 3D framework is formed (Figure 1). For the topological analysis of the framework, we utilized six different nodes. Three of the four carboxylate carbon atoms C1, C13, and C18 serve as 5-coordinated nodes while C8 serves as a 3-coordinated node. Finally, the two centroids of the benzene rings of the ligand serve as two 3-coordinated nodes with the ligand serving as two fused 3-coordinated nodes. This leads to a new 3,3,3,5,5,5-c network (Figure 2) with the following point symbol {3.4².5.6².7.8³}{3.4².5.6³.7².8}{3.4³.5².6².7.8}{4.6.8}{6.8²}₂.

2D-Fe-MOF crystallizes in the monoclinic space group P2₁n. The asymmetric unit contains half Fe(II) ion, half organic ligand (H₄L²⁻) two times deprotonated, and two water molecules, one coordinated and one in the lattice (Figure S16). The metal center adopts a slightly distorted octahedral geometry, where the equatorial oxygens are originated from four carboxylates from four different ligands and the two axial oxygens from the two water molecules (Figure S17). The H₄L²⁻ ligand bridges two different iron ions adopting trans conformation (Figure S17) leading to a two-dimensional network (Figure 3). The deprotonated carboxylates act as μ₂-η¹:η¹ bridges forming an iron rod down the crystallographic axis b. In addition, hydrogen bonds can be observed in the structure. There are four different types of hydrogen bonds with the water molecule in the lattice that bind the two-dimensional layers extending the network to the third dimension. From the topological point of view, each deprotonated carboxylate serves as a four-coordinated node forming an sql network with a point symbol (4⁴.6²) (Figure S18).

3.3. Sorption Study

Although neither of the two MOFs has an open (porous) framework, we tried to study both materials for GLU sorption, given their water stability. To the best of our knowledge, GLU sorption by MOFs has never been reported before even when electrochemical research is included. Herein, we investigated the sorption ability of both Fe-MOFs of GLU’s solution at pH 4.0 as well as the sorption ability of artificial sweat at the same pH value.

The capacity of 3D-Fe-MOF sorption for the GLU solution and artificial sweat was 15 mg/g and 1.0 mg/g, respectively, whereas for 2D-Fe-MOF were 3.4 mg/g and 2.5 mg/g.

The quantitative results were derived from the standard calibration curve corrected with IS. The correlation coefficient of the standards calibration curve corrected with the IS being 0.9998, whereas the intercept did not differ significantly from zero, as presented in Equation (2). The instrumental Limit of Detection was 0.15 mg/L, which was calculated as the concentration at which the signal-to-noise ratio was above 3.3.

$$y=\left(0.0489\pm 0.0035\right)x+\left(-0.017\pm 0.086\right)$$

(2)

Additionally, neither suppression nor enhancement of the analytical signal was observed based on the obtained recoveries. As expected, the observed GLU sorption of both MOFs is negligible and whatever sorption is observed can only be described as superficial.

3.4. Preliminary Electrochemical Evaluation of the GPE Modified with 3D-Fe-MOF

The modified graphite paste electrode was tested for the enzyme-free detection of GLU. Figure 4 presents the DP voltammograms of the modified GPEs loaded with 1%, 2%, or 5% (w/w) 3D-Fe-MOF for the detection of 0.6 mM GLU in phosphate buffer 0.10 M, pH 6.0. Among the three synthesized GPEs, the 5% (w/w) exhibited a higher voltammetric response for GLU.

Figure 5 presents the background-subtracted DP voltammograms at the modified GPE with the 3D-Fe-MOF for the detection of three different concentrations of GLU between 100–300 μM in phosphate buffer 0.10 M (pH 6.0). The limit of detection (LOD) was conditionally calculated as 22 μM but work is in progress to further decrease this value. The performance of the GPE modified with the 3D-Fe-MOF was tested in artificial sweat (pH 4.0) containing 500 μM GLU and a well-defined voltammetric peak was obtained (Figure S19) indicating the potential scope of this sensor for GLU detection in sweat.

A possible mechanism for GLU detection is [44,45]:

2Fe(III) + Glucose → 2Fe(II) + Gluconolactone + H₂O

Derived from 3D-Fe-MOF, Fe(III) is in situ formed during the voltammetric scan initiated from −1.4 V.

4. Conclusions

Two new Fe(II) MOFs based on the oxalamide ligand N,N′-bis(2,4-dicarboxyphenyl)-oxalamide (H₆L) have been synthesized and structurally characterized by single crystal and powder X-ray diffraction methods. Depending on the reaction conditions and the degree of the deprotonation of the bridging ligand, a 2D and a 3D Fe(II)-MOF were isolated. Both MOFs were found to maintain their structure in an aqueous solution as evidenced by pxrd data. The investigation of the sorption ability of GLU from aqueous solutions by both Fe-MOFs indicated very little or negligible sorption. We fabricated a graphite paste electrode using the 3D-Fe-MOF as a modifier for the non-enzymatic GLU detection in an aqueous solution at pH 6. Preliminary results indicated that the 3D-Fe-MOF-modified GPE is capable of detecting GLU. Work is in progress for the optimization and development of a 3D-Fe-MOF-modified electrode for the quantitative determination of GLU in aqueous solutions.