Stable aqueous dispersions of bare and double layer functionalized superparamagnetic iron oxide nanoparticles for biomedical applications

All analytical reagents required for the synthesis were used without further purification. For the synthesis of magnetic nanofluids containing BIONs, as well as nanoparticles coated with organic molecules (OA, poly(ethylene glycol) monooleate average Mn ~ 460 – PEGMO₄₆₀, poly(ethylene glycol) monooleate average Mn ~ 860 – PEGMO₈₆₀) we used: iron (III) chloride hexahydrate (FeCl₃·6H₂O) (≥ 98% – Sigma-Aldrich, Germany), iron (II) sulfate heptahydrate (FeSO₄·7H₂O) (≥ 99% – Carl Roth Gmbh + Co. KG, Germany), solution of ammonium hydroxide (NH₄OH) (≥ 25% – Carl Roth Gmbh + Co. KG – Germany), OA (technical grade, 90% Sigma-Aldrich – Germany), PEGMO₄₆₀ (Sigma-Aldrich, Germany), PEGMO₈₆₀ (Sigma-Aldrich – Germany). The distilled water was used during the whole process of synthesis.

Water-based magnetic nanofluids containing bare IONPs as well as OA-coated IONPs, and subsequently modified by PEGMO (OA-SPIONs, OA-PEGMO-SPION), were obtained by modifying a standard chemical coprecipitation procedure, and by performing the controlled coprecipitation in a reactor under a low-vacuum environment (1 kPa). The vacuum environment was used to prevent the undesired oxidation of divalent iron ions (Fe²⁺).

Two independent syntheses were performed in the present study: magnetic nanofluids containing bare IONPs at close to physiological pH were obtained in Synthesis 1, while in Synthesis 2, IONPs were coated with OA and modified by OA–PEGMO₄₆₀ and OA–PEGMO₈₆₀. Both syntheses are described in detail below.

An effective indicator of the stability of a colloidal dispersion is Zeta-potential, the parameter associated with the surface charge of particles suspended in a dispersion medium [6, 16]. The experimental data for ELS and DLS study are collected in Table 1, while Figure 1 shows the distribution of the size and Zeta-potential of the synthesized samples.

Fig. 1

As can be seen from Figure 1, the ζ potential of the magnetic nanofluid containing BIONs at pH = 7.0 is positive and its average value is 21 mV, although most scientific literature indicates that the zero-point charge (ZPC) for magnetite (Fe₃O₄) in the water dispersion medium is in the region of 6.5, and the isoelectric point (IEP) is observed at pH ≈ 6–6.8. Therefore, magnetite particles in an aqueous dispersion medium are losing stability and flocculated near this point due to the low density of the surface charge [17,18,19].

Thus, with an increase in pH above the IEP, the potential of bare magnetite in an aqueous dispersed medium turns negative due to deprotonation, and vice versa, at a low pH, the total charge is positive [6, 19].

However, in our case, bare magnetite-containing magnetic fluid near pH 7.0 does not have a low zeta potential and, accordingly, does not aggregate even for several months, which indicates its electrostatic stabilization. We believe that this is due to the following reason: as is known, magnetite nanoparticles exhibit amphoteric surface activity and are capable of forming charges on their surface by protonation of Fe − OH reaction centers (Fe−OH+H+ Fe−OH2+) \left( {{\rm{Fe}} - {\rm{OH}} + {{\rm{H}}^ + }\;{\rm{Fe}} - {\rm{OH}}_2^ + } \right) and deprotonation (Fe − OH Fe − O⁻ + H⁺). These reactions on the surface can be interpreted as the specific adsorption of H⁺ and OH⁻ ions on the interface of water – solid IONP (see Figure 2) [20].

Fig. 2

Accordingly, depending on the pH of aqueous solution, magnetite has a positive or negative charged surface. In addition to pH value, the ionic strength of the dispersion medium (the presence of multivalent cations in the dispersion medium) also affects the ζ potential [17].

In our opinion, such a high value of the zeta potential (22 mV) in the pH ≈ 7 region is associated with the adsorption of divalent/trivalent iron ions (possibly divalent ions are turned to trivalent ions) remaining in the aqueous solution after the synthesis of nanoparticles. This results from the stoichiometry of the reaction mixture, since the ratio Fe⁺³/Fe⁺² = 1.84. The surface-bonded ions increase the ionic strength and the magnitude of the zeta potential.

A similar mechanism was suggested by Sun et al. [21]. They studied natural and synthetic magnetite of different compositions in aqueous suspension, and found that in the presence of Fe⁺² or Fe⁺³ cations, specific adsorption on the surface of magnetite nanoparticles, even in very small quantities, strongly affects the magnitude of the zeta potential.

Thus, a positive zeta potential with a magnitude of 22 mV in the region of pH = (6.5–7.0) leads to the sedimentation stability of nanofluid containing bare magnetite nanoparticles and indicates effective electrostatic stabilization, which also affected the hydrodynamic size of magnetite nanoparticles in a dispersion medium.

As for the samples OA-SPIONs, PEGMO₄₆₀-OA-SPIONs, and PEGMO₈₆₀-OA-SPIONs, their ζ-potential is −24.2, −23.3, and −34.5 mV, respectively (Figure 1). As described above in Synthesis 2, a magnetic nanofluid containing IONPs stabilized with OA was first prepared and then modified with PEG monooleate (PEGMO₄₆₀and PEGMO₈₆₀, respectively). Indeed, it is known that OA has a polar head of the carboxyl group and a non-polar hydrocarbon tail (Figure 3). It is also known that carboxylate anions coordinate on the surface of magnetite, preferably around iron atoms, with both carboxylate oxygen atoms (COO). Thus, the polar head is attached to the magnetite surface, and the nonpolar tail extends into dispersion medium; therefore, such single-layer coated magnetite nanoparticles are hydrophobic and well dispersible in organic solvents [22, 23]. In order for the OA coated magnetite nanoparticles to become hydrophilic, they must form a second shell. Such a shell is formed from free OA molecules (present in a disperse medium) by van der Waals forces. They coordinate with the hydrophobic tails around the first oleate layer so that the carboxyl head remains in the water phase, forming a hydrophilic bi-layer around the nanoparticles (Figure 3). Thus, the coating of nanoparticles with OA molecules should lead to negative zeta potentials [24].

Fig. 3

Apparently, in our case, the samples were completely coated with OA double shells. Thus, during the process of modifying magnetic nanofluid with PEGMO molecules, oleic molecules were replaced by PEG monooleate molecules in the outer layer, thereby completely encapsulating the nuclei of the magnetite NP with the formation of a two-layer hydrophilic shell, which leads to an electrostatic and steric stabilization (see Figure 3). This mechanism has also been confirmed by IR spectroscopy, as further explained below. The high zeta potential of these samples provides high colloidal stability of the colloidal system. The hydrodynamic dimensions of the samples are 101.5 nm and 108.9 nm, respectively.

XRD patterns of the samples are shown in Figure 4. All diffraction patterns show the same characteristic reflections of the cubic crystal structure of the spinel: (220), (311), (400), (422), (511), (440), (533), (642), (731) (JCPDS No. 019-0629), at near to the following angles 2Θ (Cu Kα): 30, 35, 43, 53, 57, 63, 74, 87, 90, respectively.

Fig. 4

It is known that magnetite and maghemite have the inverse spinel structure, and the difference between their permanent lattices is small. The lattice constant of magnetite, depending on the stoichiometry during the synthesis, as well as on the temperature treatment, varies between 8.348 and 8.396, while for maghemite it is 8.33–8.336 Å [21]. The lattice parameters of the synthesized samples correspond to the magnetite parameters (Fe₃O₄) (Table 2). The absence of other interference maxima indicates that the sample does not contain crystalline impurities. In addition, the dark black color and strong magnetic properties of dried powders of magnetic fluids are a hallmark of magnetite, in comparison with other iron oxides.

When the crystallite size of the sample does not exceed 150 nm, the width of the peak of XRD of the powders is inversely proportional to the size of the crystals. In this case, it is possible to calculate the effective crystallite sizes, which are close to the size of the nanoparticles. We calculated the effective dimensions of the crystalline grains using the Scherrer equation: D=Kλβcosθ D = {{K\lambda } \over {\beta \cos \theta }} where D – average crystallite size; K – a dimensionless shape factor (Scherrer's constant), taken as 0.95 for spherical shape; λ – the X-ray wavelength (λ (Fe Kα) = 1.937283 Å; θ – diffraction angle (Bragg angle); and β – the line broadening at half the maximum intensity (FWHM) (in radians, and in 2θ units – Δ (2θ)). The calculated crystallite sizes are shown in Table 2.

The average size distribution of the particle agglomerates in the synthesized samples obtained by SAXS measurements is shown below. Data processing and modeling from the SAXS curve were performed as in our previous work, in which we prepared magnetite nanoparticles coated with folic acid using pulsed electrohydraulic discharges, in an automated chemical reactor [25].

It should be noted that the radiation scattering methods, in contrast to transmission electron microscopy, provide information on the size distribution of nanoparticles throughout the whole sample, and not for several hundred nanoparticles typically observed in TEM images. Thus, the results obtained by this method are integral characteristics and apply to the entire sample.

In order to get information about scattering of the particles only, extraction from scatter data was performed using the program Primus 3.3 [26]. Analysis of SAXS data such as radius of gyration (Rg) was carried out via Guinier fit; calculation of inverse Fourier transformations (IFTs) via GNOM was performed by a GUI based, open-source Python program BioXTAS RAW 2.0.3 [27]. Ab-initio shape determination by simulated annealing using a bead model was performed using DAMMIN [28]. These software packages have been developed by Svergun and co-workers at the European Molecular Biology Laboratory (EMBL) Group [26, 28].

The size distributions were calculated by the Monte Carlo regression package in McSAS software [29].

Figure 5 shows the scattering data in Guinier representation (ln I against q²) for evaluating radius of gyration, Rg (Figure 6). Deviation from the linear behavior can indicate aggregation or a large polydispersity of the scattering particles.

Fig. 5

Fig. 6

Model parameter distributions in absolute volume fraction from the scattering data with reasonable uncertainty estimates were performed with software McSAS (Figure 7). The following parameter values were used: lower and upper q cut-off: 0.05–3.9 nm⁻¹, respectively; the number of repetitions: 10; model: sphere; and distribution of sphere radii: between 1 nm and 60 nm. The red dashed line indicates the minimum level required for each bin to contribute in a measurable amount to the scattering pattern (i.e., the sensitivity limit).

Fig. 7

The obtained distributions are nearly similar for all the samples, with some variations in the radius range of 1–20 nm, and most of the particles have radii 6–10 nm, which is in good agreement with the single-particle or crystallite size obtained from the XRD data. The average size of particles and agglomerates, obtained by SAXS measurement, is shown in Table 3.

Using the output data of GNOM, we performed an ab-initio shape determination by simulated annealing using a single-phase dummy atom model (DAMMIN). The program restores ab-initio low-resolution shape of randomly oriented particles in solution (Figure 8). The reconstructed particle shapes resemble aggregates of nearly spherical particles.

Fig. 8

Figure 9 shows the magnetization curves of the synthesized samples BIONs, OA-SPIONs, PEGMO₄₆₀-OA-SPIONs, and PEGMO₈₆₀-OA-SPIONs. As can be seen from the graph, no hysteresis loop was detected on the magnetization curves, which indicates that the ferromagnetic nanofluid contains single domain nanoparticles that exhibit superparamagnetism at room temperature.

Fig. 9

In Figure 9, it is seen that for all samples at small values of the external magnetic field H_a, the magnetization increases rapidly with increasing magnetic field, reaching saturation at H_a ≈ 0.5 T. The shape of the M(H) curves follow the Langevin magnetization behavior of superparamagnetic particles, as commonly observed for magnetite nanoparticles suspended in aqueous [30] and in oily [31] media.

The saturation magnetization of the samples is given in Table 4. As expected, the BIONs sample has the highest value (≈ 64 emu/g), while the samples modified with OA and PEG monooleate have lower saturation magnetization values.

It should be noted that the saturation magnetization value for the BIONs is less than the magnetization characteristic for bulk magnetite phase (M_s = 88 emu/g) [32]. The decrease in the saturation magnetization of the BIONs compared to the bulk phase can be explained by the surface effect, the so-called core-shell model of the magnetite nanoparticles. Indeed, a decrease in the size of nanoparticles sharply increases the number of atoms on the surface. The surface atoms have lower coordination numbers than intra-volume atoms, which leads to a violation of the lattice symmetry on the surface of nanoparticles, as well as to a violation of exchange bonds between surface spins, which in turn leads to spin disorder and, subsequently, to a general decrease in magnetism [33].

It is known that the effect of organic residues on the surface of nanoparticles contribute to the behavior of surface spins. This process can be interpreted as the effect of surface pinning. In fact, the field created by the ligand environment can cause the very strong pinning of magnetic moments on metallic ions in the surface layer, thereby affecting the magnetic properties [34].

Thus, compared to the BION sample, a decrease in the saturation magnetization of the samples modified with OA and PEG monooleate indicates that the magnetic nanoparticles are coated with organic molecules.

In the FTIR spectrum of BIONs, a characteristic peak of magnetite is observed in the range 550–630 cm⁻¹; in our case, a peak at 580 cm⁻¹ indicates the presence of magnetite with a face-centered, reverse spinel structure, which was also confirmed by X-ray analysis. In this range, the absorption band is due to vibrations of the Fe – O bonds in the tetrahedral and octahedral positions. The vibrational bands at 3,431 cm⁻¹ and 1,630 cm⁻¹, corresponding to stretching (transverse) vibrations and O–H bending vibrations, are due to OH groups (coordinated on the surface of magnetite nanoparticles) and water molecules on unsaturated surface iron atoms [35].

The FTIR spectra of powder samples OA-SPIONs, PEGMO₄₆₀–OA–SPIONs, PEGMO₈₆₀–OA–SPIONs, and pure OA, PEGMO₄₆₀, PEGMO₈₆₀ in liquid form, are also presented in Figure 10.

Fig. 10

In the FTIR spectrum in the range of 3,500–2,800 cm⁻¹, for all three powder samples, a wide absorption band was observed due to stretching vibrations of O–H bonds both in the carboxyl group (in the OA molecule) and in the hydroxyl group (coordinated around magnetite nanoparticles). In addition, the main peaks characteristic of pure OA was observed in the spectrum of the OA-SPIONs sample (Figure 10). In particular, the sharp peak (with low intensity) at 3,004 cm⁻¹, on which the band of stretching vibrations of O–H bonds is superimposed, is due to stretching vibrations of the C–H bond in the vinyl fragment [36], and peaks at 2,924 cm⁻¹, 2,855 cm⁻¹, and 1,459 cm⁻¹ are caused by stretching (both symmetric and asymmetric) and deformation vibrations of (– CH₂₋) bonds in the hydrophobic molecular chain of OA, respectively.

The average peak in intensity at 1,540 cm⁻¹ is due to the symmetric stretching vibration of the carbon group [–COO–]. The sharp intense peak in the spectrum of pure OA at 1,707 cm⁻¹ is associated with the C=O stretching vibration of the carbonyl group of OA molecules. This peak for the OA–SPIONs sample is shifted toward high wavenumbers (high energies) at 1,747 cm⁻¹, which clearly indicates that free OA molecules in the double layer around magnetite nanoparticles are connected by physical (hydrophobic) bonds with oleate molecules (primary layer) chemisorbed on the surface of IONPs (Figure 10).

This opinion is supported by the fact that, in contrast to the FTIR spectrum of pure OA, other peaks are also shifted in the spectrum of the OA-SPIONs sample; in particular, the absorption band at 1,289 cm⁻¹, which corresponds to stretching vibrations bond C–O, is shifted to the position of 1,240 cm⁻¹, and the peak at 935 cm⁻¹ caused by the angular deformation of the bond (–CH₂₋) is shifted by 894 cm⁻¹, respectively [37].

The above discussion of the existence of free OA molecules on a double layer around magnetite nanoparticles is confirmed by the fact that on the FTIR spectra of samples PEGMO₄₆₀–OA–SPIONs and PEGMO₈₆₀–OA–SPIONs, characteristic absorption bands of physically bounded OA molecules completely disappeared or significantly decreased.

In addition, there is a shift or the appearance of new bands. In particular, in the FTIR spectra of samples PEGMO₄₆₀–OA–SPIONs and PEGMO₈₆₀–OA–SPIONs, broad absorption bands appeared with an extremum at 3,140–3,141 cm⁻¹. This is most likely due to the superposition of the stretching vibrations of the OH group of pure magnetite and PEG monooleate molecules in this spectral band (characteristic bands of pure PEGMO_460/480 at 3,447/3,474 cm⁻¹, respectively). The absorption bands at 2,924 cm⁻¹ and 2,855 cm⁻¹ are shifted to 2,922 cm⁻¹ and 2,852 cm⁻¹, respectively.

The displacement of the peaks is likely due to the physical interaction of PEG-monooleate with oleate molecules in the bilayer around magnetite nanoparticles.

In addition, a strong absorption band (at 1,117/1,099 cm⁻¹) and a characteristic absorption band at 1,735/1,734 cm⁻¹ in the spectrum of pure PEGMO_460/860 (Figure 3), caused by the stretching vibration of the PEG ether group and by C = O stretching vibration [13], respectively, with small shifts at 1,109/1,108 cm⁻¹ and 1,739/1,736 cm⁻¹, appear in both FTIR spectra of PEGMO-OA-SPION samples, respectively.

It should also be noted that when OA is chemisorbed on the surface of magnetite nanoparticles in the form of a carboxylate, COO- gives absorption band of C = O vibration at 1,635 cm⁻¹ [38], which is observed for both samples. Thus, in the IR spectra of samples PEGMO₄₆₀ – OA-SPION and PEGMO₈₆₀ – OA-SPION, the slight shift of peaks of OA molecules and new peaks at 3,140/3,141, 1,636 cm⁻¹ indicate chemisorbed oleate molecules on the surface of Fe₃O₄ nanoparticles.