Magnetic Behaviour of Iron Oxide/Dextran Nanoparticles in a Keratin Matrix

Featured Application

Development of a method to characterize size distributions of magnetic nanoparticles and their magnetic behavior in polymeric matrices.

Abstract

Magnetic nanoparticles (MNPs) are interesting for their potential employment in biomedical and environmental technologies. Although they have been characterized by many techniques, there are some issues that need to be solved. For instance, it is not yet possible to finely characterize their size distribution or to detect their local magnetic properties. In this work, commercial MNPs were employed, which were made of iron oxide cores with a mean diameter of 8 nm embedded in a matrix of dextran to form skeins with a mean diameter of 20 nm. These MNPs have been dispersed in keratin, a natural protein extracted from wool. Thin films have been realized by spin coating water solutions with various MNP concentrations. Analysis was conducted using a set of techniques, namely Atomic Force Microscopy, Environmental Scanning Electron Microscopy, a Physical Property Measurement System–Vibrating Sample Magnetometer, and the spatially resolved Magneto-Optic Kerr Effect (NanoMOKE). These MNPs show superparamagnetic properties, although a wide distribution of blocking temperature values indicates that the cores are not isolated and interact with others. NanoMOKE not only allows us to map the magnetic behavior of MNP clusters, but also to detect the presence of isolated MNPs dispersed in the keratin matrix.

1. Introduction

Nowadays, magnetic nanoparticles (MNPs) are becoming day by day ever more considered for their employment in a variety of technological applications. These fields span from biomedical to ‘green’ areas [1,2,3,4]. In recent years, there has been intense activity from researchers regarding topics ranging from fabrication to structural characterization [5,6,7]. Therefore, MNPs have been investigated with many different instruments to determine their chemical and physical properties [8]. However, there are some issues that still need to be clearly solved. For instance, there is not a definitive method allowing one to finely determine their size distribution. This lack of knowledge at the moment may not seem as crucial as the phenomena studied are statistically observed on a macroscopic scale that considers other phenomena at lower scales as minor effects. However, the microscopic and nanoscopic regimes are likely to contain a wealth of information that could be essential for future finer developments of these applications [9].

Up to now, the size distribution of an ensemble of MNPs can be measured with different methods depending on their characteristics [10]. If they are soluble in a liquid, Dynamic Light Scattering (DLS) can be fruitfully employed [11,12]. DLS provides a reliable result, especially if the particles have a hard structure, i.e., they do not incorporate solvent molecules and do not swell. However, if the distribution is not monomodal, Atomic Force Microscopy (AFM) [13] can be more effective than DLS [14]. Using AFM, the size distribution can, however, be affected by the deposition method and the analysis may only represent a small portion of the whole ensemble.

If the MNPs are not soluble and can only be handled in powder, they can be deposited on a conductive surface for analysis with Scanning Electron Microscopy (SEM) [15] or on a grid to be subjected to Transmission Electron Microscopy (TEM) [16]. In this case, the same problems described above for AFM can affect distribution analysis [17]. Finally, a thermomagnetic method has been proposed that is based on studying the dependence of the magnetic properties on the temperature (T) [18]. This procedure can provide reliable information if the MNPs present superparamagnetic behavior [19].

Whatever method one employs, it is not simple to determine the size distribution precisely and uniquely. In addition, it is also necessary to characterize their magnetic properties when dispersed in a polymer matrix. To tackle these issues, the authors recently performed an investigation of a system made of keratin mixed with nanometric hydrotalcites (HTlc) containing Mg and Fe atoms [20]. The data gathered have proven that a combination of techniques allows one to characterize the magnetic properties of HTlc from the powder down to small portions of a thin film. This approach opens new perspectives in the characterization of composite materials. However, those HTlc do not show superparamagnetic behavior as expected considering the small dimensions. In order to assess the capabilities of the procedure proposed, a more defined system is thus needed.

To further test our approach, some other commercial MNPs were chosen for these experiments. According to the manufacturer, they are made of small iron oxide cores with a magnetite structure. These cores are decorated with dextran strains, which can be polymerized in a second instance to embed the magnetic parts in a web of dextran within a nanometric skein. In the end, iron oxide/dextran MNPs can thus be produced. The iron oxide cores have a nominal size distribution as determined by Transmission Electron Microscopy (TEM), with a mean size of 8 nm. On the other hand, the MNPs present a size distribution determined by DLS of a mean size of 20 nm and a polydispersity index < 0.25. In Figure 1a, a sketch of the material purchased is presented; the number of cores is not necessarily two, depending on the actual size of the MNPs and other factors. Each one of these skeins can contain a variable number of magnetic cores, whereas the mean distance between adjacent cores may also vary. To our knowledge, there is no technique to observe the inner structure of these MNPs. They are suitable for assessing the validity of our approach to characterize the magnetic properties of bulk materials and small fractions of them dispersed in a polymer matrix.

In the present work, the purchased MNPs were dispersed in a matrix of keratin, a natural protein extracted from wool [21]. Thin films, with a thickness of around 200 nm, were deposited on square cut pieces of silicon wafers via spin coating a water solution of keratin and MNPs (Figure 1b). Keratin was first dispersed in water with a 10% weight concentration to form the basic solution. Then, the MNPs were added to it with a concentration ranging from 2.5 to 20% in order to deposit films with a variable density of MNPs.

The analysis of the samples was conducted using a set of techniques that are suitable to achieving our aim [20], namely Atomic Force Microscopy, Environmental Scanning Electron Microscopy, a Physical Property Measurement System–Vibrating Sample Magnetometer, and the spatially resolved Magneto-Optic Kerr Effect (NanoMOKE). The data collected are herein presented and discussed. In particular, the MNPs do not present a defined value of blocking temperature but a wide distribution of values. This is an indication that the magnetic cores may interact with each other, forming clusters of various dimensions. Similarly, in the films, clusters of various dimensions could be identified with all the techniques employed. However, single MNPs or small clusters can only be detected with NanoMOKE. This may be attributed to the fact that their numbers are below the detection limit of the other techniques.

2. Materials and Methods

The commercial magnetic nanoparticles (MNPs) employed had a diameter of 20 nm (#WHM-G053) with a polydispersity index <0.25 [22]. They contained iron oxide cores with a mean diameter of 8 nm, with the producer reporting that they have a magnetite structure (Fe₃O₄). No further details on the size distribution are provided by the manufacturer. These MNPs were designed with an unmodified dextran surface, thus making them soluble in water. According to the manufacturer, MNP density in the purchased solution is equal to 4.3 × 10¹⁵ particles/mL. One can calculate that there are 4300 MNPs per 1 μm³ of pristine solution.

High molecular weight keratin powder (≈50 kDa) was extracted from raw wool and kindly provided by Kerline Srl (Bologna, Italy) for these experiments. The MNPs/keratin blend films were prepared by spin coating a water solution in air on Si/SiO₂ substrates that were plasma-treated for 3 min. The basic solution contained water and 10% w/v of keratin. The content of MNPs was varied, with 2.5, 5, 10, or 20% w/w vs. keratin of the MNP solution added to the pristine solution. For a concentration > 20%, the solution becomes unstable and precipitation occurs. The spinning conditions chosen for the deposition were the following: final speed 4000 rpm, acceleration 4000 rpm/s, time duration of 60 s. Using a profilometer, the thickness of the film was found to be around 200 nm. All the samples were analyzed as prepared with no further treatment [23].

The MNP powder was characterized using a Physical Property Measurement System (PPMS, by Quantum Design, Tokyo, Japan) equipped with a Vibrating Sample Magnetometer (VSM) insert. The magnetization per unit mass (M, emu/g) has been measured as a function of the temperature M (T) and the magnetic field M (H). To perform the M (T) measurement, the sample was cooled down to 5 K in the absence of a magnetic field. After this, a magnetic field of 1000 Oe was applied, and the sample was warmed up to 300 K to obtain the Zero Field Cooling curve (ZFC). Subsequently, it was cooled down to 5 K to obtain the Field Cooling curve (FC). For the M (H) measurements, the target temperature was fixed and the magnetic field was swept from 0 to 90 kOe, from 90 to −90 kOe, and again from −90 to 90 kOe.

NanoMOKE^® is an instrument based on the Magneto-Optic Kerr Effect (MOKE). It combines the capabilities of the Durham Magneto-Optics NanoMOKE3 with the flexibility of a closed-cycle optical cryostation. This system allows for characterization of the magnetic properties of the sample surface from room temperature down to about 8 K within a few micrometers of lateral resolution. The maximum H value that can be applied is 4000 Oe, and the frequency range can span from 0.1 to 20 Hz depending on the maximum H value. The typical values employed are 2000 Oe and 2.1 Hz. The experimental setup is based on a scanning laser microscope operating at λ = 660 nm whose position is managed by means of a pair of galvanometric mirrors. The measurements were performed in a polar configuration with the external magnetic field applied perpendicular to the sample surface (out-of-plane magnetization) [20].

Atomic Force Microscopy (AFM) was performed using a hybrid system that was assembled using a commercial head (SMENA, NT-MDT), home-built electronics, home-developed software, and a commercial digital lock-in amplifier (Zurich HF2LI). The setup was operated in intermittent contact mode (ICM). The cantilevers used are available from MikroMasch, the HQ:NSC35 model, with a nominal force constant from 5.4 to 16 N/m and a resonance frequency from 130 to 300 kHz.

The Environmental Scanning Electron Microscopy (ESEM) data of the MNPs/keratin films were collected with a Zeiss EVO LS 10 LaB6 (Carl Zeiss, Milano, Italy).

3. Results

To start, an analysis of the iron oxide/dextran MNPs alone was performed. When purchased, they come in a small container dispersed in a water solution. It is therefore necessary to extract a portion of them in order to carry out an investigation with a PPMS-VSM. The specimen can be obtained through the complete evaporation of water from a few drops of the pristine keratin solution placed on a piece of silicon wafer at a concentration of 20%. The result is a brown flake that can be detached, reduced into powder, and then placed in the PPMS-VSM sample holder.

Figure 2 presents a graph showing the magnetization (M) versus the temperature (T). This type of measurement is used to perform a preliminary characterization of the magnetic properties of a sample of MNPs. In particular, the temperature associated with the maximum value (T_MAX) of the ZFC branch is closely correlated to the so-called blocking temperature (T_B) [8,24,25]. Usually, if the size distribution is narrow, T_MAX represents a good approximation of the smean T_B; otherwise, it may indicate the value above which the majority of MNPs activate [24,25]. The graph reported in Figure 2 does not show a maximum for the ZFC branch. The curve monotonically increases from 5 up to 250 K and then reaches a plateau. On the contrary, the FC branch increases slightly from 300 to 150 K, and remains constant down to 5 K.

In Figure 3, two graphs are shown of M versus the magnetic field (H), recorded at 300 and 5 K. These graphs can provide further information on the magnetic behavior of the MNPs. It is possible to notice that the branches, recorded for increasing and decreasing H, are fully superposed at room T with a coercive field (H_C) that is almost null. This represents a clear indication that the sample behaves as a superparamagnet. However, the branches do not superpose at 5 K with H at around 200 Oe, indicating that the magnetic momenta associated with the MNPs are completely blocked (see inset).

The films spin-coated at various MNP concentrations on silicon can be characterized by techniques such as Atomic Force Microscopy (AFM), Environmental Scanning Electron Microscopy (ESEM), and the spatially resolved Magneto-Optic Kerr Effect (NanoMOKE). One can thus determine and correlate the morphology, the chemical composition, and the local magnetic behavior of all samples as a function of MNP density.

The AFM images of all samples are presented in Figure 4. From them, one can derive the 3D morphology, i.e., the height values as a function of the lateral coordinates. The surface is flat for the concentration of 2.5%, with some fractal features due to keratin crystallization that are a few nm thick [20]. At a concentration of 5%, some isolated clusters start to appear. The presence of clusters further increases at concentrations of 10 and 20%. This trend in morphology may indicate that the MNPs are separated, mainly scattered, and isolated at concentrations below 5%. On the other hand, above 5%, they tend to organize in clusters of various dimensions. This does not exclude the presence of single nanoparticles deeply embedded in the flat regions.

In Figure 5, some ESEM data obtained on a sample with a 20% concentration are reported. The morphology is very similar to the one observed with AFM: flat regions alternate with some clusters and fractal features. Whereas AFM provides the vertical dimension, ESEM allows one to probe the chemical composition correlated with morphology. The energy dispersive X-ray (EDX) analysis shows that the clusters contain both iron and oxygen. On the contrary, the flat areas present no evidence of the presence of iron or oxygen. The same results have been obtained for a sample with a concentration of 10%, whereas below 5% it was not possible to detect neither iron nor oxygen. Combined with the AFM data, this might indicate that the MNPs are present in clusters of various dimensions. In addition, they are either completely missing in the flat regions or they are scattered and isolated with a density below the detection limit of EDX.

Figure 6 presents some NanoMOKE maps obtained for the same sample as in Figure 5 and recorded at 300 and 15 K. In (a) and (c), the reflectivity maps are reported, i.e., the local reflection of the laser beam. Keratin is largely transparent, and the laser can therefore penetrate the whole film and access the MNPs dispersed throughout the whole thickness of the film [20]. In (b) and (d), the maximum Kerr rotation values at 2000 Oe are plotted using a color scale. It is possible to notice that the more reflective areas do not show any noticeable magnetic contrast. However, in these areas, the average Kerr signal increases at 15 K. At 300 K, the dark regions that can be identified with clusters present some magnetic contrast with stronger Kerr rotation values that also increase at 15 K. The spatial distribution of the Kerr signal provides a complex picture, i.e., it appears that not all MNPs in a cluster align perpendicular to the surface as the setup works in a polar configuration. NanoMOKE can thus provide a fine map of the local magnetization.

Figure 7 presents some NanoMOKE maps obtained for a pure keratin sample: the maximum Kerr rotation values at 2000 Oe are plotted using a color scale. These maps are homogeneous, both in reflectivity and Kerr rotation values. The average Kerr signal is slightly negative at 300 K, though the difference in relation to 15 K is very small. More interestingly, it is always lower than the average Kerr signal obtained from the bright regions of the 20% sample at both room and low T. This observation may indicate that the bright regions, made prevalently of keratin, may have the presence of some isolated MNPs. These MNPs must be dispersed as single ones or form small clusters, all of them well diluted and in depth, as neither AFM nor ESEM can detect their presence.

In Figure 8, some NanoMOKE data are presented for the same 20% sample. Specifically, some graphs are shown of Kerr rotation versus H obtained at 300 and 15 K. The Kerr rotation value is proportional to the local magnetization. The curves were recorded at three spots in Figure 6: one within a cluster and two within the bright areas. Pure keratin data are also reported as a reference. As expected, the data of pure keratin do not depend on T. At 300 K, the curves recorded in the cluster have a stronger signal compared to the curves recorded in the bright region. One spot of the bright regions is similar to pure keratin, whereas the other one shows signals slightly stronger than those obtained on pure keratin. At 15 K, the signal increases in the cluster region and also in the bright regions, though to a lesser extent. This is an indication of the presence of MNPs not only in the clusters, as observed with AFM and ESEM, but also in the bright areas. Additionally, the local density is not homogeneous.

It must be underlined that the Kerr signal is noisier at 15 K due to the vibration of the cooling system. Although the signal can be averaged over several seconds (typically 1 min), the vibration cannot be completely cancelled. This is the reason why the curves are less smooth compared to room T. Finally, comparing Figure 3 and Figure 8, one can notice that the hysteresis at 300 K is accentuated for NanoMOKE, whereas for the PPMS-VSM graphs it is null. One needs to keep in mind that the measurements have not been carried out in equivalent ways. In the case of NanoMOKE, the field varies with a rate of 16 kOe/sec, whereas in the case of the PPMS-VSM the rate is equal to 1 Oe/sec. The variation in rate is more than four orders of magnitude. This large difference is likely to be the cause of the discrepancies observed since, for the PPMS-VSM measurements, the MNPs can be considered to be in a quasi-static condition, whereas for the NanoMOKE measurements they cannot.

Another important aspect may be represented by the difference in the number of MNPs probed with the two systems. In particular, the PPMS-VSM can measure the whole ensemble (i.e., it probes clusters with all the different dimensions), while NanoMOKE can only probe a tiny portion of them, namely those present in the film area that the laser is shining on.

4. Discussion

According to the ESEM and AFM data, the MNPs can be individuated only within the clusters. Single MNPs, if present, are not detectable in the regions with high reflectivity where keratin is predominant if not exclusively present. Starting from the manufacturer datasheet, an estimation of the number of MNPs dispersed in the films can be attempted. The density of the purchased solution is equal to 4 × 10¹² MNPs/μL. The number of MNPs in the pristine solution is further reduced when added to the water solution with keratin, depending on the percentage chosen. The percentage in the final solution was selected to be in the range between 2.5 and 20%, the maximum limit before precipitation occurs.

As previously described, the films were deposited by spin coating on square-cut silicon wafer pieces with an area of 2 × 2 cm². The quantity of solution placed on them was about 100 μL. For a sample with a 10% concentration, the volume of MNPs in the solution is roughly 10 μL, and the number of MNPs is thus about 4 × 10¹³. If all MNPs are deposited, one can calculate the number of MNPs per unit area, i.e., 10⁵ MNPs/μm². This is a relatively large number and, if one evaporates a drop of the final solution, the color of the residual is brown indeed. However, the quantity of solution that remains on the substrate dramatically reduces when spinning starts and a 200 nm film of keratin plus MNPs forms. The color of the thin films is still relatively transparent. Therefore, MNP density cannot be precisely calculated, though it can be reasonably estimated to within a few hundred per μm². From the AFM analysis, the number and size of clusters is in reasonable agreement with the estimation made for the 10% sample. The detection of a low density of single MNPs always presents a sensitivity problem, even for Magnetic Force Microscopy (MFM) or Electrostatic AFM (EFM). It is also difficult to select a microscopy technique sensitive to subsurface features (like Ultrasonic Force Microscopy (UFM) or Peak Force AFM) that is capable of clearly detecting MNPs, especially if they are placed at different depths. The weak signal obtained with NanoMOKE is therefore indicative of the presence of MNPs, although a quantitative analysis cannot be conducted yet.

The PPMS-VSM data suggest that the MNPs, designed to be used in a liquid environment [22], are not isolated and interact with each other when they are mixed with keratin in a solid sample. In fact, the ideal magnetic behavior of a sample of non-interacting MNPs would show a monotonic increase in FC magnetization with decreasing T [18]. As observed in Figure 2, the magnetization of the FC curve first increases at high T values and then approaches a constant value at lower T values. As reported in Figure 9b, the T_B distribution can be extracted from the first derivative of the difference between the ZFC and FC curves, as described by Micha et al. [18,25]. From this graph, a broad peak at 118 K can be noticed as well as a smaller one at 201 K. A possible explanation of the presence of two peaks in the T_B distribution is that the MNPs have a bimodal size distribution [26]. According to the information provided by the producer, the MNPs used have a monomodal size distribution peaked at 8 nm. Thus, the behavior shown in Figure 9b needs a different interpretation. From the literature, one can advance further hypotheses such as the presence of clusters mixed to non-interacting nanoparticles [27,28,29,30] or an anisotropic effect appearing at low temperature [31], such as spin-freezing [32].

Considering the geometry of MNPs, schematically drawn in Figure 1, the magnetic cores can agglomerate in different ways. First, they can be close enough within single MNPs (Figure 1a). During the process of dextran polymerization, they can be trapped and located at different distances and in various numbers. Second, the MNPs can be stuck next to each other, and cores from adjacent MNPs can thus interact as they assemble in clusters (Figure 1b). This agglomeration can already occur in the purchased solution, an effect possibly enhanced by ageing. The magnetic cores of the different MNPs can be located on the perimeter and thus close to external cores. The whole situation can thus give rise to a wide range of core-to-core distances.

In Figure 10, some other NanoMOKE maps are finally reported, obtained for the same 20% sample. These maps were recorded upon zooming in to two clusters that are visible on the top right of the map shown in Figure 6. This map is much more detailed as the field view is restricted to a lateral dimension of 100 × 100 μm² and it provides more insights into the local magnetization of the clusters. It can be further noticed that each cluster does not present a homogeneous magnetization. Specifically, one can notice that the Kerr rotation varies from high to low and even exhibits slightly negative values within a single cluster. This may depend on the local superposition of the magnetic fields from different MNPs or from different clusters. The Kerr rotation also depends on T, i.e., it increases with lowering T. This is an expected result, as one can deduct from the graphs reported in Figure 3 for the powder.

5. Conclusions

In this work, some experiments were devised to assess the capabilities of an approach developed to characterize the magnetic properties of nanoparticles from powder to films where they are dispersed in a polymeric matrix. The analysis was conducted using Atomic Force Microscopy, Environmental Scanning Electron Microscopy, a Physical Property Measurement System–Vibrating Sample Magnetometer, and the spatially resolved Magneto-Optic Kerr Effect. For this purpose, commercial magnetic nanoparticles were employed, realized from small iron oxide cores with a mean diameter of 8 nm embedded in a web of dextran to create a skein with a mean diameter of 20 nm. These nanoparticles were dispersed in a water solution with keratin, and thin films of around 200 nm were realized via spin coating solutions with various concentrations. The data collected allow us to state that the nanoparticles have superparamagnetic behavior. However, they show a wide distribution of blocking temperature values, indicating that the magnetic cores are not isolated but rather interact with a variable number of other cores. NanoMOKE allows us to detect local magnetization due to clusterization and additionally detects the presence of single nanoparticles or small clusters dispersed in the keratin matrix at concentrations not detectable by other techniques, such as Atomic Force and Environmental Scanning Electron Microscopy. This approach can be extended to other MNPs dispersed in different polymeric matrices, provided that the polymers are transparent. In the end, it can be stated that the characterization of magnetic nanoparticles with regards to their size distribution and magnetic properties in a specific context requires the utilization of different techniques, with the more techniques used the better. The selection of these techniques depends also on the final application the magnetic nanoparticles are devised and fabricated for, i.e., dispersed in a liquid or a solid matrix.