The effect of organic coatings in the magnetization of CoFe₂O₄ nanoparticles

Magnetic-oxide nanoparticles (NPs), especially magnetite, are of central importance to a large array of developing bio-medical applications including hyperthermic cancer treatment,¹ cancer detection,² in vivo imaging,³ and directed drug delivery.^4,5 The ability to control the magnetic signal strength and magnetic morphology of the NP tags, however, is of critical importance. It is generally accepted that oleic acid (OA) coating is protective for the intrinsic magnetization of ferrite nanoparticles.^6–12 However, it remains unclear why the magnetization of these NPs decreases so dramatically in the absence of a protective surfactant. For example, does a self-terminating magnetically dead layer form only at the NP surface or does the absence of a protective shell affect the core magnetization? Is the magnetic decrease related to oxidation or from the ability of the NPs to retain their ferrimagnetic alignment?¹³ Are other coating types with biologically desirable properties, such as the polysaccharide fucan (FC) (sulfated polysaccharides extracted from seaweeds), as magnetically protective as oleic acid and IGEPAL^® CO-520?¹⁴ 

Comparing the effect of coating on NP magnetization across different NP systems is fraught with challenges. Subtle differences in NP size, shape, material purity, surface roughness, and crystallinity are all dependent on the synthesis specifics—sol–gel methods,¹⁵ citrate precursor techniques,¹⁶ electrochemical synthesis,¹⁷ combustion methods,¹⁸ solid state reaction,¹⁹ mechanical alloying,²⁰ and chemical precipitation,^21–23 which can, in turn, greatly affect the magnetization. By examining a consistently synthesized system of ∼20 nm diameter CoFe₂O₄ NPs based on an improved co-precipitation technique²⁴ and coated with oleic acid (OA), the polysaccharide fucan (FC), or left uncoated (bare), our aim is to address these questions.¹⁰ By including spatially sensitive neutron scattering alongside conventional magnetometry, this work is sensitive to differences in magnetization between interior and surface regions. Additionally, we ascertain the intrinsic NP magnetization under a range of applied magnetic fields, including remanence, which mimics the actual condition under which these NPs may be utilized in real-world devices.

The CoFe₂O₄ nanoparticles were synthesized by the co-precipitation of Co²⁺ and Fe³⁺ or Fe²⁺ with NaOH.^24,25 In brief, an aqueous solution was prepared with 50 ml of 1M FeCl₃·6H₂O (Alfa Aesar, 97%) and 50 ml of 0.5M CoCl₂·6H₂O (Alfa Aesar, 97%) were used as received, without further purification. The mixture was then heated at 100 ± 3 °C for 1 h and homogenized under vigorous stirring (∼116 Hz). 0.5 ml of oleic acid (C₁₇H₃₃COOH, Alfa Aesar, practical 90%–95%) was used as the surfactant and, along with sodium hydroxide, was dropped over 30 min until obtaining a final pH of 13. The liquid precipitate was then brought to a reaction temperature of 80 °C and stirred for 1 h. The product was allowed to cool to room temperature. The precipitate was then washed twice with distilled water and then with ethanol, to remove the excess surfactant from the solution, and then dried at 100 °C for 24 h to obtain the bare NPs. The sample used for the OA coating was thoroughly washed with distilled water until it reached a neutral pH, and then, the material was dried at 60 °C overnight. For obtaining the fucan coated sample, the resulting nanoparticles coated with OA were added to a 2% by volume aqueous solution of the fucan polysaccharide¹⁴ prepared in distilled water and again homogenized with vigorous stirring (∼116 Hz) while being heated to 60 °C and then dried overnight at 60 °C. The final product obtained from this process for the bare, OA coated, and FC coated NPs appeared dense and black in color and presented magnetic characteristics.

Magnetic hysteresis measurements [Fig. 1(a)] reveal that oleic acid coated CoFe₂O₄ NPs have a magnetic moment of ∼60 A m²/kg, significantly greater than the fucan coated and bare CoFe₂O₄ NPs corroborating the previously observed increase in NP magnetization due to organic coatings.⁷ At room temperature (where all further analyses were performed), the NPs are not fully magnetically saturated by 70 000 Oe (7 T), but they are reasonably close to the full magnetic moment by 1.5 T where our “high field” neutron scattering studies were conducted. The magnetization values recorded at 1.5 T and room temperature are listed in Table I for ease of comparison with Small Angle Neutron Scattering (SANS) measurements.

X-ray powder diffraction (XRD) data were collected using a commercially available x-ray diffractometer with Bragg–Brentano geometry and with a Cu-Kα source (λ = 1.5418 Å) at 40 kV and 40 mA and angular variation range from 20° to 70° in steps of 0.05°. XRD data show the characteristic cubic spinel structure, in accordance with standards JCPDS International Centre for Diffraction Data file No. 22–1086 [Fig. 1(b)]. The broadening of peaks can be attributed to the small particle size of the prepared sample. The Scherrer broadening for the x-ray peak average is given by the following equation:

where β is the peak full-width half-maximum in radians (adjusted for the value of the instrument broadening contribution), K is the Scherrer constant (constant of proportionality), λ is the x-ray wavelength, θ is the scattering angle, and L is the crystallite grain size; results in a calculated peak average grain size (using K = 0.94) of 17.37 (±0.12) for the bare CoFe₂O₄, 17.46 (±0.12) for the CoFe₂O₄ + OA, and 17.60 ± 0.10 for the CoFe₂O₄ + FC NPs (Table I).

Scanning electron microscopy (SEM) is sensitive to the total NP diameter including surfactant, while transmission electron microscopy (TEM) measurements show a contrast between the NP core as well as evidence of the coating layer. Both SEM and TEM measurements were done at the National Institute of Standards and Technology’s (NIST) Center for Nanoscale Science and Technology (CNST). TEM images were taken on a FEI Titan 80–300 analytical TEM operated at an accelerating voltage of 300 kV and with a spatial resolution of 0.20 nm at 300 kV in bright-field mode using a Gatan Orius digital camera. Samples were prepared for TEM imaging onto carbon-coated copper TEM grids. Different coatings of the CoFe₂O₄ nanoparticles show evidence of changes in the particle size as observed through SEM. SEM images revealed a broad morphology and particle size around 20 nm in diameter for the bare CoFe₂O₄ [Fig. 2(a)]. For the cobalt ferrite nanoparticles coated with fucan and oleic acid, the TEM results show particle diameter sizes of 32 and 36 nm, respectively; moreover, it was observed that the particles are more dispersed and uniform in size [Fig. 2(b)]. Transmission electron microscopy (TEM) measurements corroborate the systematic increase in particle size with different coatings as can be seen in Table I.

Small Angle Neutron Scattering (SANS) is an ideal tool for measuring the ensemble-averaged structural and magnetic morphology within samples.^26–28 For our samples, SANS was performed at the NIST Center for Neutron Research at beam line NG7 SANS using neutron with a wavelength of 5.5 Å for the case of polarized data for detector distances of 3 and 14 m, while the unpolarized neutron data were collected at 6 and 8 Å at detector distances of 4 and 15.31 m, respectively. In each case (polarized or unpolarized), the configurations were joined to form one continuous I vs Q plot. SANS probes both structural and magnetic morphology of nanostructures as well as average NP-to-NP spacing as shown in Fig. 3. Here, the intensity, I, is shown as a function of momentum transfer, q, which is roughly 2π/d, where d is the distance between repeating features. The structural scattering (N²) dominates, as expected for CoFe₂O₄,¹² which is evidenced by the fact that the scattering profile remains nearly indistinguishable as a function of applied field, H, from 0.005 to 1.5 T (not shown).

The structural scattering analysis was performed in two ways. First, the inflection points were determined from the I vs Q plot to approximate NP-to-NP distance as 2π/Q (as shown in the supplementary material and recorded in Table I). The NP-to-NP spacing measured in SANS is slightly larger than the measured single NP diameter determined by SEM and TEM (Fig. 2), as expected since it represents the average whole particle sizing plus gap between adjacent particles; nevertheless, the trends between NP coatings are consistent (Table I). Second, an addition of models was employed: Guinier Porod (to extract radius of gyration, Rg) + correlation length (to capture the fact that the nanoparticles are clustered). Note that the structural SANS is a complex combination of surface roughness (≈Q⁻⁴), Porod scattering,^29,30 NP surface coatings, spherical CoFe₂O₄ NPs, and any inter-particle periodic correlations. The Rg values of the bare, FC, and OA samples were determined to be 5.8, 6.4, and 12.2 nm, respectively. Since the average radius of structurally uniform, spherical particles is R = Rg × √(5/3), we can infer from the bare NPs an average CoFe₂O₄ radius of 15.0 nm, which is slightly less than, but approximates the 17.5 nm diameter determined from x-ray Scherrer broadening (Table I).

Spin analysis of the scattering neutrons was additionally applied in order to measure the magnetic-only scattering of the sample.^27,31,32 Typically, the SANS fits would be performed with a nuclear scattering length density of 6.07 × 10⁻⁶ (Å⁻²) and the magnetic scattering length density of up to 1.42 × 10⁻⁶ (Å⁻²) for CoFe₂O₄. However, since the sample packing volumes of our powdered samples were unknown, we were instead able to extract the relative magnetic scattering length densities for moments aligned with the applied field at 1.5 T and for moments at 0.0005 T where the net moment direction for any given nanoparticle is nearly random with respect to the applied magnetic field. Specifically, for the cases of CoFe₂O₄ + OA and CoFe₂O₄ + FC, the magnetic scattering length density ratios at 0.005 T (normalized by the magnetic scattering length density at 1.5 T) were determined to be 0.56 and 0.87, respectively, from the SANS fits. When neutrons encounter an external field (even a very small one), half the neutrons align with the field (↑) and the other half anti-parallel to the field (↓). By using a polarizing supermirror (i.e., a FeSi multilayer diffraction grating) and a pi spin-flipper, ↑ or ↓ neutrons are selected prior to interaction with the sample (a polarized neutron beam) [Fig. 4(a)]. For a structurally isotropic sample, the scattering from moments aligned with the applied field parallel to the x-direction (M_X||H) with the neutron beam directed along Z can be measured using the following equation:³³ 

Here, I is the intensity of the ↑, ↓ neutrons measured from a ±15° sector cut along the θ direction specified in the subscripts, where θ is the scattering angle measured from the positive X axis (||H) within the X–Y detector plane, as shown in Fig. 4(a).

The results are shown in Fig. 4(b). The low-q turnover for the samples with the weakest signal is a known artifact that can result from the approximation that the ↑ + ↓ intensity of N² + M² ∼ N² since M ≪ N¹² and does not affect the analysis of the results. Spherical model fits³⁴ of the magnetic core to the data with the FWHM polydispersity in the core diameter held constant at 50% (see Discussion) are shown in Fig. 4(b), with the resulting magnetic diameters shown in Table I.

Obtaining the scattering from the component of moments aligned ⊥ H (i.e., M oriented along Y and Z) requires an additional ³He neutron spin analyzer placed after interaction with the sample.³⁵ Assuming M_Y⊥H = M_Z⊥H = M_⊥H, such full spin analysis yields³³ 

Here, the double arrows refer to spin analysis before and after sample scattering, respectively, and the intensity is measured from a circular average (all summed over θ). The results of this technique are plotted in Fig. 4(c). The OA coated NPs show almost the same $M⊥→H$ and $M∥→H$ diameters, while $M⊥→H$ is just slightly reduced from $M∥→H$ for the FC coated NP sample, where we continue to use a model polydispersity of 50%. The $M⊥→H$ signal from the bare sample is not shown since it was too small to fit. Figure 4(d) additionally demonstrates that this calculation is sensitive to an average magnetic radius with a sensitivity of ±0.4 nm.

X-ray diffraction indicates that the crystalline size of the CoFe₂O₄ cores, for any coating or lack thereof, is about 17.5 ± 0.1 nm in diameter. This is in reasonable agreement with SEM and TEM, both indicating a particle size of about 20 nm. It is also in reasonable agreement with unpolarized SANS, which places the bare CoFe₂O₄ nanoparticle diameter at 15.0 nm from Rg fitting (the least realizable technique of these given the complication of the structural SANS model). However, the magnetic SANS modeling is far more precise than the structural SANS modeling given that the latter is only sensitive to magnetic scattering arising from the CoFe₂O₄ cores and can thus be fit with a simple spherical model. Furthermore, measurement of $M∥→H$ for the OA NPs yields a magnetic core of 13.79 ± 0.01 nm with a polydispersity of 50%. Thus, the largest NPs would, according to this fit, be 1.5 × 13.79 nm² = 20.7 nm, which is in good agreement with the TEM and SEM results. The fact that x-ray diffraction measures an average crystallite size of about 17.5 nm (based on finite-size peak broadening) is not contradictory given that the NPs are polydisperse and may not be single crystal throughout their CoFe₂O₄ volumes.

The similarity between $M⊥→H$ at 0.005 T and $M∥→H$ at 1.5 T diameters for OA (13.79 vs 13.86 nm) and for FC (11.99 vs 11.90 nm) (Table I) suggests that the magnetic NP volume is intrinsic to the NP itself, rather than being dependent upon having surface spins aligned by the presence of an applied magnetic field. In the case of the bare NPs, it is unclear whether the lack of a distinctive $M⊥→H$ at 0.005 T scattering is due to internal magnetic disordering or if the signal is simply too small to observe compared to the flat, incoherent background.

The remaining question is whether the superconducting quantum interference device (SQUID)-measured reduction in magnetization per mass with a change in NP coating (OA > FC > bare) is attributable to a simple reduction in the active magnetic area (typically associated with surface-limited disordering) or it is largely attributable to an average reduction in the magnetization per volume throughout the NP. Relative to the OA NP M || H radii of 13.79 ± 0.01 nm measured at 1.5 T, the FC and bare NPs have slightly smaller magnetic diameters of 11.99 ± 0.03 and 11.90 ± 0.16 nm, respectively (Fig. 4 and Table I). However, the reduction in the average measured magnetic volume of $11.99313.793=0.65$ and $11.90313.863=0.63$⁠, respectively, cannot account for the total magnetization reduction measured by SQUID at 1.5 T of 0.49 (FC) and 0.32 (bare), listed in Table II. It is worthwhile to note that the difference in overall magnetization between coatings is probably even more significant than that obtained from magnetic hysteresis alone [Fig. 1(a)] since the extra volume occupied by the coatings has not been accounted for. Moreover, SANS magnetic modeling could be performed with radial polydispersity anywhere in the range between 20% and 50%, where we have chosen the latter for demonstration because it represents the smallest decrease in the magnetization density required to account for the magnetization change measured by SQUID magnetometry. For example, magnetic modeling with 50% polydispersity yields average magnetic radii at 1.5 T of 15.4 ± 0.1 nm (OA), 14.5 ± 0.3 nm (FC), and 13.2 ± 0.5 nm (bare), or relative magnetic volumes of 1.0 (OA), 0.83 (FC), and 0.63 (bare). Since the SQUID magnetometry magnetization drops faster than the magnetically active area per nanoparticle with any polydispersity within the plausible range 20%–50%, we conclude that the average magnetization density must decrease as the OA is replaced by FC and even more so when the nanoparticles are uncoated (bare). While there may be some reduction in the magnetically active area between OA coating and the others, it is this further reduction in the average magnetization density that indicates the coating protects against more than a surface-limited, 1–2 nm at most, magnetic disordering.

In summary, using the neutron scattering technique, we have demonstrated that a change in coating type can be used to tune the intrinsic magnetization of ∼20 nm CoFe₂O₄ nanoparticles, while still nearly preserving the spatial extent of their magnetically active regions. Thus, the coating-dictated reduction in NP magnetization is not simply the result of forming a magnetically dead or oxidized layer on the CoFe₂O₄ core. This means the magnetization of the NPs could be an intrinsic property that does not depend on an external applied magnetic field in order to align the magnetic spins. Neutron scattering results show that the ability to tune NP magnetization, independent of the applied magnetic field and without having to alter the NP dimensions (either physical or magnetic aspects), as a function of coating type, should be explored more closely as compared to magnetization data and could be highly relevant in the optimization of many bio-medical applications.

The data that further substantiate the findings of this study are available in its supplementary material.

This work was supported by Fundação de Amparo a Ciência e Tecnologia do Estado de Pernambuco FACEPE under Funding No. BFP-0070-2.08/13.

The authors would like to thank Shannon Watson, Wangchun Chen, Jeff Krzywon, Paul Butler, and Cedric Gagnon for their invaluable assistance during SANS measurements.

The identification of any commercial product or trade name does not imply endorsement or recommendation by the National Institute of Standards and Technology.

The authors have no conflicts to disclose.

The data that support the findings of this study are available from the corresponding author upon reasonable request.