PEG size effect and its interaction with Fe3O4 nanoparticles synthesized by solvothermal method: morphology and effect of pH on the stability

The synthesis and characterization of Fe3O4 magnetic nanoparticles (MNPs) obtained by the solvothermal method in ethyleneglycol with the addition of polyethyleneglycol (PEG) with molar mass of 4000, 8000 and 20000 g mol−1 are described, aimed at evaluating its effect on the size, morphology and stability of the nanoparticle. The syntheses were carried out by solubilizing the precursors at 85 and 140 °C, providing smaller nanoparticles as well as smaller crystallites at higher temperatures, while the effect of PEG was less evident. Measurements of nanoparticle surface areas synthesized with PEG 4000 and 20000 g mol−1 at 140 °C provided values of 76 and 14 m2 g−1, respectively, indicating that PEG 4000 surrounds the crystallites, while PEG 20000 preferably surrounds the whole MNP. As a consequence, MNP with very dissimilar porosities were obtained. Electron energy loss spectroscopy (EELS) indicated that MNP synthesized with PEG 20000 possesses higher electronic density than those obtained with PEG 4000, in agreement with the surface area results. Infrared spectroscopy and thermogravimetric analysis demonstrated the presence of PEG in the particles, whose amount increased as the particle size decreased. Dynamic Light Scattering (DLS) measurements showed that MNP hydrodynamic radius increases with the PEG size and stability in solution increases from pH 5.0 to 9.0 for smaller NP, while polymer presents slight effect on stability for the larger particles. The results obtained in this work show that properties of MNP can be tuned by the dissolution temperature of the chemical precursors and the PEG molar mass, changing their porosity and stability in solution, that are important variables in processes of adsorption, drug delivery and sensor developing.


Introduction
Since the discovery of the magnetic properties and biocompatibility of magnetic nanoparticles (MNPs), those constituted of iron oxides have shown significant uses in biomedical applications [1,2], mainly in processes of magnetic resonance imaging [3,4], drug delivery [5] and cancer treatment by hyperthermia [6,7]. In addition, the magnetic properties of these nanoparticles have facilitated processes of separation, allowing the detection of metal ions [8][9][10] and removal of pollutants [11][12][13] in different media.
MNPs of iron oxide can be obtained by different methods, such as co-precipitation [14], thermal decomposition [15], microemulsion [16], sonochemical [17], solvothermal [18], among others. The preparation method is fundamental to determine the characteristics of MNPs, such as morphology and size distribution. Coprecipitation is the most used method to prepare iron oxide MNPs, due to its simplicity and low cost. This method produces crystallites with average size around 10 nm, and, when synthesized with the addition of polyphosphate [19] and mannitol [20], respectively, they present improvement in colloidal stability, resistance to the presence of proteins and, even so, strong responsiveness to the magnetic field. Other surfactants as oleic acid, sodium citrate and Triton X-100 [21] also have been used. Besides bringing great advantages concerning the dispersion of MNPs [22], oleic acid and trisodium citrate can also allow better control of hyperthermia temperature treatment [23]. The MNPs with dimensions around 10-20 nm present better performance as they exhibit superparamagnetic behavior, which allow them to rapidly respond to an external magnetic field, with remanence and coercitivity of low magnitudes [24]. The MNPs with dimensions>20 nm can present an enhancement in the magnetic saturation; however, the transition from superparamagnetic to ferromagnetic behaviors will be also induced [25].
Crystallites of iron oxide with dimensions <20 nm can form clusters, resulting in secondary structures, which present magnetic saturation values much more intense, because each crystallite is able to maintain its superparamagnetic behavior [26]. Due to this characteristic, clusters of crystallites with different dimensions have been explored for several applications, mainly for the development of SERS substrates [27], processes of catalysis [28] and bioseparation [26]. These secondary structures, constituted by several crystallites, can be achieved by the solvothermal method, which has the advantage of providing monodisperse and hydrophilic nanoparticles with a degree of homogeneity higher than those provided by other methods [25,26]. Dong et al [29] synthesized Fe 3 O 4 MNPs with high porosity, high surface area and strong magnetic response. The nanoparticles were synthesized from FeCl 3 ·6H 2 O by the solvothermal method, using ammonium acetate and sodium citrate in ethyleneglycol. Ammonium acetate acted as a porogenic agent, source of alkali and reducing reagent, in conjunction with ethyleneglycol, while sodium citrate acted as MNP stabilizer and modifier [29]. The effect of sodium citrate was investigated, producing smaller crystallites and MNPs with higher surface area when higher concentration was employed. The amount of ammonium acetate also affected the surface area of the nanoparticle, which increased as the concentration of the reagent increased, reaching a maximum, and then decreasing with further increasing of reagent concentration. This result is supposedly related to the pH of the medium, which increased according to ammonium acetate amount, therefore increasing the reaction rate, producing denser clusters. The nature of electrostatic stabilizer also affected MNP and crystallite sizes. Kim et al [30] verified the effect of sodium citrate, poly-ethyleneglycol (PEG 8000) and didodecyldimethylammonium bromide (DMAB), respectively, on the solvothermal synthesis of Fe 3 O 4 MNPs in ethyleneglycol, using sodium acetate as stabilizer to avoid agglomeration of primary particles. Nanoparticles with sizes ranging from 174 to 336 nm and constituted of crystallites ranging from 11.6 to 21.3 nm were obtained, including those from the synthesis without the addition of stabilizer. Interestingly, the use of sodium citrate provided the largest particles with smallest crystallites, while DMAB produced the smallest particle size and narrowest size distribution. In addition, nanoparticles presented good dispersibility in water. The addition of 4-poly(vinylpyridine) in the synthesis of Fe 3 O 4 in sodium acetate/ethyleneglycol solution produced mesoporous nanoparticles, with size of 240 nm and crystallites of 35.5 nm, able to remove Congo red from water, as described by Zeng et al [31]. According to the authors, 4-poly(vinylpyridine) can interact with the crystalline nuclei and reduce their growth during the hydrothermal reaction. In addition, this polymer can modify the surface of the crystalline nucleus, favoring the assembling of porous regular structures, with a uniform size distribution.
The stability of the magnetic nanoparticles dispersed in water is of great concern because agglomeration impairs their application in many areas, such as drug delivery, extraction and sensors. Agglomeration can occur due to the strong dipole-dipole interactions among the particles and to their high surface energies [32]. Therefore, surface modification of the nanoparticle is necessary in order to overcome this drawback, providing stabilization of MNP, which are mainly based on electrostatic (van der Waals and double layer), steric (polymeric, surfactant) and electrosteric forces [32]. Polymers, such as polyvinylalcohol (PVA), polyacrylic acid, polyethyleneimine (PEI), dextran, polyethyleneglycol (PEG) and surfactants, such as cetyltrimethylammonium bromide (CTAB), are usual modifiers to improve the stability of MNP. PEG, being biocompatible and hydrophilic, is often used as modifier/stabilizer of MNPs [32]. Despite its favorable characteristics, the modification of MNP with PEG is not straightforward, being usually performed after the synthesis of the nanoparticles and usually accomplished either by incorporating an appropriate functional group in the MNP or in the polymer [32]. The use of 3-aminopropyltrimethoxysilane (APTES) [33] has been described to modify the nanoparticles before PEGylation, while trimethoxysilanePEG [34] and dicarboxyl-terminated [35] have been used for modification of MNP. In spite of the strategies above mentioned, surface modification has been also made with as-prepared MNP and regular PEG [36].
The solvothermal method for preparation of MNP from iron(III) chloride frequently employs ethyleneglycol as a solvent, which also acts as reducing agent, and a precipitating agent, such as sodium acetate. Stabilizers are also added to the reaction medium to avoid particles agglomeration, as mentioned above. Polyethyleneglycol is used in the synthesis of iron oxide MNP as a stabilizer and polymers with different molar masses such as 8000 [30], 4000 [37] and 2000 g mol −1 [38] have been used. However, to the best of our knowledge, there is no systematic study on the effect of PEG size on MNPs morphology and its stability in different pH values. Therefore, this work encompasses the synthesis of iron oxide MNP using ethyleneglycol as a solvent, in the presence of PEG with molar masses of 4000, 8000 and 20000 g mol −1 , evaluating their effect on the size, morphology and stability of the particles. In addition, syntheses were performed in two different temperatures (85 and 140°C), as this variable also affects the size of the nanoparticles [37].

Synthesis of magnetic nanoparticles (MNPs)
A volume of 40 ml of ethyleneglycol was heated at a 140°C in a 100-ml beaker. Afterwards, 8 mmol of FeCl 3 was added and the mixture was kept under stirring for 15 min after its temperature reaching again the temperature of 140°C, thus making the salt soluble. Then, 1.0 g of PEG was added and, after stirring for 5 min, 26.5 mmol of NaAc·3H 2 O were added, keeping the stirring for more 30 min. Finally, the solution was transferred to a PTFEcoated stainless steel autoclave (80 ml), which was placed in an oven at 200°C for 8 h. After this period, the produced MNPs were separated from the solution with a magneto, washed with deionized water followed of ethanol and then dried in an oven at 60°C for 8 h. This procedure was carried out using PEG 4000, PEG 8000 and PEG 20000, providing MNPs named 4K140, 8K140 and 20K140, respectively. Similar syntheses were performed at solubilization temperature of 85°C, producing MNPs named 4K85, 8K85 and 20K85, respectively.

Transmission Electron Microscopy (TEM)
Morphology, and particle and crystallite sizes were evaluated with a Carl Zeiss microscope, model Libra 120 equipped with an omega 'Ω' filter in-column, operating at 80 or 120 kV, depending on the analysis conditions for obtaining optimized images. The samples were suspended in water, dripped on a copper grid (Ultrathin carbon film on lacey carbon support grid, 400 mesh, copper) and dried at room temperature. Electron Energy Loss Spectroscopy analysis were recorded for all spectra using fixed experimental conditions such as i) objective aperture of 90 μm resulting in a collection semi-angle around 15 mrad, ii) fixed spectrometer entrance aperture preserving same sample area size and iii) same dispersion magnification on spectrometer and on the digital recording system. Evaluation of a relative thickness (t/λ Ine ) was done using Log-Ratio method [40]. Olympus Cantega G2 (2048×2048 pixels) camera and iTEM software were used to register the images and spectra.

Analysis of the surface area and determination of pore size and volume
The nitrogen adsorption/desorption measurements were performed in a surface area and pore size analyzer ASAP 2020 Micromeritics. Samples were heated at 120°C for 12 h before analysis. From these results, it was possible to determine the surface area using the BET method (Brunauer-Emmett-Teller). The pore diameter and volume were calculated using the Barrett-Joyner-Halenda (BJH) method.

Fourier transform infrared (FT-IR)
Measurements in the transmission mode were performed in KBr pellets at ambient temperature with an Agilent Cary 630 spectrometer. Spectra were acquired in the region from 400 to 4000 cm −1 as an average of 64 runs, with resolution of 4 cm −1 .

Thermogravimetric analysis (TGA)
A mass of 10-20 mg of sample was placed in a platinum crucible and analysis was performed with a TGA 2050 TA Instruments, at a heating rate of 10°C min −1 , from 25 to 800°C, in flowing argon atmosphere (100 ml min −1 ).

Dynamic light scattering (DLS)
The particle sizes were measured with a Malvern equipment, model ZetasizerNano Zs-Zen3600. Samples (2.5 mg) were transferred to a 100-ml volumetric flask filled with deionized water and dispersed using at ultrasound bath for 25 min An aliquot of 2.5 ml of suspension was mixed with 2.5 ml of deionized water, acetate buffer (pH 5.0), phosphate buffer (pH 7.0) or carbonate buffer (pH 9.0) and 1.0 ml of this mixture was transferred to the measuring cell. The suspension was homogenized and measurements were performed during 1200 s, in intervals of 300 s.

Results and discussion
Iron oxide nanoparticles can be obtained from several methods of synthesis. The solvothermal method is a versatile chemical route to prepare hydrophilic Fe 3 O 4 MNPs, with a relatively narrow size distribution [25]. In addition, the temperature employed to solubilize the reagents allows obtaining MNPs with different sizes [37]. The MNPs obtained in the present study have similar size range of those described in a previous work, in which no significant variation in the magnetization values was observed [37]. The MNPs, whose precursors were solubilized at 85 and 140°C and synthesized with PEG 4000, 8000 and 20000, were characterized by x ray diffraction and the effect of these variables on the particle size as well as on the crystallite size was evaluated. Figure 1 shows the x ray diffractograms of the MNPs, with diffraction lines indexed to the structure of the inverted spinel type according to magnetite catalog card JCPDS 19-629, suggesting that MNPs has a facecentered cubic structure.
In addition, it can be noted that increasing the dissolution temperature occurs a decrease in the relative intensities of the peaks. In order to verify this effect, the crystallite sizes were calculated by the Scherrer equation    Table 1 lists the values obtained in this calculation. Electron diffraction and dark field images corroborate the analysis of the variation on grain size (see Supporting Information (available online at stacks.iop.org/NANOX/2/020022/mmedia), Figures S1 and S2 and table S1). Further electron diffraction confirms the high crystallinity degree of the crystallite obtained by solvothermal synthesis [41]. Therefore, the decrease of x ray peak intensity is associated to the decrease of the crystallite size. No significant effect was observed on the crystallite size when PEGs with different molar masses were used.
The reduction of the crystallite size as the dissolution temperature of precursors is decreased is associated to nucleation and growing steps. Higher temperatures increase the nucleation rate, favoring the formation of smaller crystallites (and also smaller NPs), while lower temperatures favor the growing step, producing higher crystallites, as described by Carvalho et al [37].
Observing the TEM images in figure 3, it can be noted the formation of nanostructured magnetic spheres produced by the solvothermal method, composed by several crystallites which aggregated to form larger secondary structures, therefore more stable than primary crystallites. In addition, size distribution is narrow for MNPs obtained from precursor dissolution performed at 140°C and those obtained from dissolution at 85°C seems to present a bimodal distribution. The formation of secondary structures from primary crystallites is also affected by the dissolution temperature and PEG plays an important role on the MNP porosity. PEG is supposed to surround the crystallites and its molar mass, and consequently its chain size, plays a fundamental role on the particle morphology and size distribution. For dissolution at 140°C, the increase of PEG chain increases the particle average size and narrows its distribution (figures 3(a)-(c)), indicating the synergistic effect between PEG chain size and higher temperatures of precursors dissolution. Fine tuning of these parameters can allow size and distribution controls for several biomedical and sensing applications. Measurements of BET surface areas of 4K140 and 20K140 MNPs provided values of 76 and 14 m 2 g −1 , respectively, indicating that PEG 4K surrounds the crystallites, while PEG 20K preferably surrounds the whole MNP, conducting to MNP with very dissimilar porosities. This assumption is also supported by the results obtained by BJH pore volume and pore diameter measurements, whose values are (0.22 cm 3 g −1 and 115 Å) and (0.12 cm 3 g −1 and 647 Å) for 4K140 and 20K140 MNPs, respectively. This result is important for the development of MNPs for removal of heavy metals, which requires porous particles in order to carry considerable amounts of these ions from aqueous systems [42,43].  The effect of the PEG polymeric chain size on the electronic density of the NPs was also investigated by EELS, a characterization technique that does not require a specific crystalline structure, being able of analyzing materials with thicknesses from 15 to 500 nm [40]. Therefore, the effect of PEG on the morphology was investigated only for the MNPs smaller than 500 nm, that is, the 140°C MNPs, which have diameter around 160 nm. A set of five similar size MNPs prepared with each type of PEG was chosen for this study. Table 2 shows the relative thickness (t/λ Ine ) values obtained for each MNP, while figure 4 indicates a direct relationship between PEG molar mass and this parameter.
Since the size parameter is not varying, the decrease in (t/λ Ine ) as a function of the PEG chain reflects an increase in the mean free path as the PEG molar mass increases. It means that electronic density is proportional to the PEG molar mass, therefore indicating that denser MNPs are obtained with the use of PEG with higher molar mass, corroborating the results obtained by surface area measurements. Considering the experimental conditions, any difference in energy loss spectra and relative thickness parameter can be related to the electronic density (see Supporting Information). Thus, the more intense interaction, the higher is the electronic density in the sample, which means higher amount of iron in the particle and shorter the effective mean free path.  The MNPs were also analyzed by FT-IR spectroscopy, which provided information about the chemical groups on the particle surface, as shown in figure 5. It can be observed absorption bands around 3435 cm −1 and 1629 cm −1 , which are attributed to stretching [44] and bending [45] vibrations of the -O-H groups presented on the MNP surface, respectively. Low intensity absorption bands for MNPs obtained by using dissolution temperature of 140°C are observed around 2924 cm −1 and 2854 cm −1 , which can be attributed to asymmetric and symmetric stretching vibrations of -C-H bonds [44,46]. The absorption band centered at 1051 cm −1 can be attributed to stretching vibration of -C-C bond, while bands around 1420 cm −1 and 1090 cm −1 indicate stretching vibrations of -CO. These bands are more evident for the MNPs obtained when 140°C was used for precursor dissolution, signalizing the PEG:Fe 3 O 4 mass ratio is higher in these particles. Regarding the nanoparticles obtained from 85°C, either these bands are not observed or they are weaker, probably because the larger size of the NPs that decreases the PEG:Fe 3 O 4 mass ratio and causes a significant absorption of the radiation in the 4000-2000 cm −1 region. The absorption bands in the 700-400 cm −1 region appear in all spectra and can be ascribed to the Fe-O bond vibrations in the tetrahedral and octahedral sites [44,[47][48][49].
The PEG:Fe 3 O 4 mass ratio was evaluated by thermogravimetric analysis, in inert atmosphere and temperature ranging from 25 to 800°C. Figure 6 shows the thermogravimetric curves of nanoparticles synthesized in the presence of PEG 4000. The 140°C MNPs present a higher mass loss than 85°C MNPs, which are associated to the amount of polymer [50] and to the process of nanoparticle formation during the synthesis. The smaller nucleus formed at 140°C can interact more effectively with PEG, therefore carrying a higher amount of polymer, results that corroborates those obtained by FT-IR spectroscopy. In addition, 140°C MNPs have higher surface area, which can increase the PEG:Fe 3 O 4 mass ratio for these samples. It can be noted that nanoparticles adsorb water efficiently via hydrogen bonds and decomposition of the polymer starts at a lower The results obtained by FT-IR and TG confirm the presence of -OH groups of water molecules physiosorbed on the MNP, which can provide hydrophilic character to them. In order to assess this stability, DLS measurements were performed in deionized water and buffer solutions at pH 5.0, 7.0 and 9.0. The suspensions of MNPs were prepared at the concentration of 25 mg L −1 , with the aid of ultrasound bath, in order to maintain  similar interactions between the particles, as they tend to aggregate and form clusters of different sizes [52]. Figure 7(a) shows the smaller size of the MNPs obtained with precursors dissolved at 140°C compared to those obtained at 85°C, while the hydrodynamic radius increases as a function of the molar mass of PEG, corroborating the result obtained by electronic density measurements that indicates that PEG 20K surrounds the MNP, while PEG 4K remains among the crystallites. In addition, a lower polydispersity is obtained when higher dissolution temperature is used, although it is also evident that increasing the PEG chain the polydispersity index increases, as shown in figure 7(b). The term 'polydispersity' (PDI or 'dispersity' as recommended by IUPAC) is used to describe the non-uniformity degree of the particle size distribution. Also known as heterogeneity index, PDI is a number calculated from a two-parameter fit to the correlation data (the cumulants analysis). This index is dimensionless and values smaller than 0.05 are mainly seen with highly monodisperse standards. PDI values higher than 0.7 indicate the sample has a very broad particle size distribution and is probably not suitable to be analyzed by the DLS technique. Different size distribution algorithms work with data that fit between these two extreme values of PDI (i.e., 0.05-0.7) [53]. Therefore, according to these values, it can be considered that 4K140 MNPs have a monodisperse distribution in suspension, while other MNPs present a population of low-to-medium polydispersity.
The stability of the MNPs in buffer solutions was evaluated by measuring their size increase as a function of time, demonstrating the aggregation rate. Buffer solution was added to MNP aqueous suspensions at the beginning of the measurements, therefore producing a change in pH as well as in the ionic strength of the medium. Figure 8 summarizes the results obtained, which also shows normalized values obtained by taking as reference the particle sizes at the beginning of the experiments (t=0). Figure 8(a) shows that 85°C NPs decrease in size at short times, presenting almost a random behavior as the measurement progresses. The size decrease can be explained by the effect of ionic strength that avoids the agglomeration of particles, whose size trend is summarized in figure 8(c). However, a more plausible reason concerns the sedimentation of the particles as a function of time due to their sizes, while the smaller ones remain in suspension. On the other hand, 140°C MNPs undergo a significant effect of the solution pH, as shown in figures 8(b) and (d), in which particles increase their size almost three times at pH 5.0, due to agglomeration. Considering these results, it is clear that size of MNPs plays an important role on the stability, while the PEG chain size does not seem to affect it significantly in buffered solutions, as described for MNPs modified with PEG after their synthesis [54]. Figure 8(d) indicates that sizes of the 140°C MNPs present virtually a linear behavior as a function of time, independently of the solution pH. This curve pattern is associated to the size of the MNPs as well as the surface characteristics, such as the presence of -OH groups from water molecules, that is practically absent on the surface of 85°C NPs, as demonstrated by FT-IR. Thus, the aggregation rate for 140°C NPs can be calculated as a function of time, as shown in figure 9. It can be noted that aggregation rate decreases as the solution pH increases, as in acidic pH MNPs are more susceptible to aggregate due to the neutralization of the negative charge on their surfaces. In addition, increasing the size of PEG chain, the stability of the MNPs are usually higher due to the steric hindrance, which keeps the particles apart from each other. It seems from figure 9 that steric hindrance plays an important role against aggregation for PEG 8K and 20K, as solution pH has little effect on the NPs stability above the value of 7.0. At pH 9.0, however, the 4K140 MNP, having a small amount of PEG on the surface, is highly charged, presenting the lowest aggregation rate. The polydispersity index of the MNPs obtained with precursors dissolution at 140°C also present a small increase as a function of time, while those obtained from 85°C vary significantly as shown in figure S5 (Supplementary Information). Therefore, 20K140 MNPs, followed of 8K140, seem to be useful for bioanalytical, clinic and medical applications, where the pH of fluids are between 7.0 and 8.0, as well as for the development of sensors for metal ions, that usually work in the 7 to 9 pH range.

Conclusion
Iron oxide NPs were synthesized by the solvothermal method in ethyleneglycol, by using PEG with molar masses of 4000, 8000 and 20000 g mol −1 and solubilizing the chemical precursors at 85 and 140°C. Results obtained by TEM and BET showed that dissolution temperature of reagents affected the particle sizes and their distribution as well as the formation of respective aggregates. EELS showed that electronic density of particles synthesized by solubilizing the precursors at 140°C increased as the PEG molar mass also increased, indicating PEG with high molar mass cover the NPs while it stands in between the crystallites when its molar mass is low. Infrared spectroscopy and thermogravimetric analysis demonstrated the presence of PEG in the particles, whose amount increased as the particle size decreased. DLS measurements showed that MNP hydrodynamic radius increase with the PEG size and stability in solution increases from pH 5.0 to 9.0 for smaller particles, while it presents slight effect on stability for larger ones. The results obtained in this work show MNPs morphological and dispersity properties can be tuned by the temperature of dissolution of the chemical precursors and the PEG molar mass, allowing to vary their porosity and stability in suspension. These features enable the use of the synthesized MNPs for applications where controlled sorption processes are important variables, such as drug delivery, magnetic separation and sensor developing.