Irregularly Shaped Iron Nitride Nanoparticles as a Potential Candidate for Biomedical Applications: From Synthesis to Characterization.

Magnetic nanoparticles (MNPs) have been extensively used in drug/gene delivery, hyperthermia therapy, magnetic particle imaging (MPI), magnetic resonance imaging (MRI), magnetic bioassays, and so forth. With proper surface chemical modifications, physicochemically stable and nontoxic MNPs are emerging contrast agents and tracers for in vivo MRI and MPI applications. Herein, we report the high magnetic moment, irregularly shaped γ'-Fe4N nanoparticles for enhanced hyperthermia therapy and T2 contrast agent for MRI application. The static and dynamic magnetic properties of γ'-Fe4N nanoparticles are characterized by a vibrating sample magnetometer (VSM) and a magnetic particle spectroscopy (MPS) system, respectively. Compared to the γ-Fe2O3 nanoparticles, γ'-Fe4N nanoparticles show at least three times higher saturation magnetization, which, as a result, gives rise to the stronger dynamic magnetic responses as proved in the MPS measurement results. In addition, γ'-Fe4N nanoparticles are functionalized with an oleic acid layer by a wet mechanical milling process. The morphologies of as-milled nanoparticles are characterized by transmission electron microscopy (TEM), dynamic light scattering (DLS), and nanoparticle tracking analyzer (NTA). We report that with proper surface chemical modification and tuning on morphologies, γ'-Fe4N nanoparticles could be used as tiny heating sources for hyperthermia and contrast agents for MRI applications with minimum dose.

Introduction

Nowadays, magnetic nanoparticles (MNPs), with proper surface chemical modifications, are emerging nanomaterials that have been exploited in the areas of magnetic resonance imaging (MRI) and magnetic particle imaging (MPI),[1−10] drug/gene delivery,[11−16] hyperthermia,[17−25] bioassays,[10,26,27,27−31] cell sorting and separation,[32−36] and so forth. For different applications, high magnetic moment MNPs are demanded for large magnetic torques in drug/gene delivery and cell sorting and separation applications, for high sensitivity magnetic bioassays, as well as for efficient and minimum dose usage in MRI, MPI, and hyperthermia applications. In view of this demand, γ′-Fe4N nanoparticles are magnetically soft,[37] chemically stable,[38] cheap, and possess high saturation magnetization (182 emu/g).[38,39] Since year 2000, many groups have reported the facile synthesis of high purity γ′-Fe4N nanoparticles.[38−40] Based on the Fe–N phase diagram, the γ′-Fe4N phase forms at the temperature range from 200 to 680 °C.[41] As proposed by the Lehrer diagram, the most stable iron nitride phase could be modified by tuning the nitriding potential that is controlled by the partial pressure of hydrogen and ammonia gas and the nitridation temperature.[42] Thus, different iron nitride phases are obtained using the gas nitridation process, such as α″-Fe16N2, γ′-Fe4N, γ-FeN, ε-Fe-3N, and so forth.[43−47] In this paper, we report the gas nitridation method to synthesize γ′-Fe4N nanoparticles. During the nitridation process, ammonia gas provides nitrogen atoms and hydrogen gas is applied to tune the nitriding potential to obtain γ′-Fe4N nanoparticles. Figure shows the experimental setup for preparing γ′-Fe4N nanoparticles. Hydrogen gas is used for reducing the starting materials, γ′-Fe2O3 nanoparticles, in a tube furnace. Then, the nitridation is proceeded in the same furnace under a mixture of ammonia and hydrogen gas to obtain γ′-Fe4N nanoparticles, as shown in Figure a. The reduction and nitridation reactions are illustrated in Figure b.

Figure 1

γ′-Fe4N nanoparticles prepared by a gas nitridation approach. (a) Schematic drawing of the gas nitridation set up. The starting material γ-Fe2O3 nanoparticles are placed in a tube furnace. Hydrogen and ammonia gas cylinders provide high purity gas for the reduction and nitridation processes. (b) Summary on the working principle of gas nitridation. The γ-Fe2O3 nanoparticles, the starting materials, are reduced by hydrogen gas in a tube furnace. Then, a mixture of hydrogen and ammonia is applied to synthesize γ′-Fe4N nanoparticles. These synthesized γ′-Fe4N nanoparticles are transferred into a glove box to avoid oxidation. (i) Photograph of γ-Fe2O3 powder, (ii) photograph of γ′-Fe4N powder, (iii) crystal structure of γ-Fe2O3, (iv) crystal structure of α-Fe, and (v) crystal structure of γ′-Fe4N.

After the synthesis of γ′-Fe4N nanoparticles from γ-Fe2O3 nanoparticles, the crystalline structures of both nanoparticles are characterized by X-ray diffraction (XRD), and the high purity γ′-Fe4N phase is confirmed from our synthesized γ′-Fe4N nanoparticles. The γ-Fe2O3 and synthesized γ′-Fe4N powders are wet mechanically milled with oleic acid (OA) to functionalize the nanoparticle surface with OA chemical groups and to effectively separate nanoparticles. The static and dynamic magnetic responses of γ-Fe2O3 and γ′-Fe4N nanoparticles in OA solution are measured and compared using the vibrating sample magnetometer (VSM) and magnetic particle spectroscopy (MPS) systems, respectively. In addition, the morphologies of γ-Fe2O3 and γ′-Fe4N nanoparticles are characterized by transmission electron microscopy (TEM). It is confirmed that γ-Fe2O3 nanoparticles, with an average magnetic core size of 20 nm, are sintered into larger, irregularly shaped γ′-Fe4N nanoparticles of around 100 nm. The dynamic light scattering (DLS) and a nanoparticle tracking analyzer (NTA) are used to measure the hydrodynamic size of both nanoparticles. The irregularly shaped γ′-Fe4N nanoparticles, with high saturation magnetizations, provide a new way for enhanced T2 relaxivity in MRI and efficient hyperthermia treatment with minimum dose requirements.

Results and Discussion

XRD of γ-Fe2O3 and Synthesized γ′-Fe4N MNPs

The structure of γ-Fe2O3 and synthesized γ′-Fe4N are investigated by XRD. As shown in Figure a, the XRD pattern of the starting material matches the γ-Fe2O3 phase. After the hydrogen reduction and gas nitridation, nanoparticles show that γ′-Fe4N is the main phase. The γ′-Fe4N nanoparticles are successfully synthesized by the gas nitridation method. There is also a diffraction peak at around 2θ 36° that is from iron oxide, which might be due to the oxidation when transferring the powder sample from the tube furnace to a glove box. The crystal structures of γ′-Fe4N and γ-Fe2O3 are also plotted in Figure b,c, respectively.

Figure 2

XRD patterns and crystal structures of γ-Fe2O3 and synthesized γ′-Fe4N nanoparticles. (a) XRD patterns of γ-Fe2O3 nanoparticles (red solid line) and nitride nanoparticles (black solid line). The powder diffraction files (PDFs) are also plotted, as shown in the panels (i) γ′-Fe4N and (ii) γ-Fe2O3. The XRD pattern of the starting material, iron oxide nanoparticles, matches well with the γ-Fe2O3 (PDF card no. 00-004-0755). Based on the PDF of γ′-Fe4N (PDF card no. 00-006-0627), the main phase of the synthesized nitride nanoparticles is γ′-Fe4N. A tiny iron oxide peak around 36° is observed from the synthesized nitride nanoparticle sample. The oxidation might happen during the sample transfer process such as from the tube to the glove box. (b,c) Crystal structures of γ′-Fe4N and γ-Fe2O3 are revealed.

Morphology Characterization on γ-Fe2O3 and γ′-Fe4N Nanoparticles

The morphologies of γ-Fe2O3 and γ′-Fe4N nanoparticles are obtained using TEM. From each nanoparticle sample, three different samples are prepared for the TEM characterization: the wet mechanical milled γ-Fe2O3 and γ′-Fe4N nanoparticles in OA, named γ-Fe2O3@BM and γ′-Fe4N@BM, respectively; the supernatant of these nanoparticle suspensions after a ultra-centrifugation step (10,000 rpm for 20 min), named γ-Fe2O3@Ultra and γ′-Fe4N@Ultra; the supernatant from these nanoparticle suspension after keeping at room temperature for 24 h, named as γ-Fe2O3@Sup and γ′-Fe4N@Sup. The TEM images, illustration of each TEM sample preparation process, and schematic drawings of different shape nanoparticles are summarized in Figure a,b. Both γ-Fe2O3@Ultra and γ′-Fe4N@Ultra show well dispersed nanoparticles with diameters below 20 nm. Most of these nanoparticles have irregular shapes. TEM images of the as-milled nanoparticles, γ-Fe2O3@BM and γ′-Fe4N@BM, show that nanoparticles tend to aggregate together to minimize their surface energy. Around 60% of the nanoparticles from the γ′-Fe4N@BM sample are sintered bodies and aggregate together, which might happen during the reduction and nitridation which are handled at a relatively high temperature (400 °C). These sintered bodies are also observed from the γ′-Fe4N@Sup sample. The γ-Fe2O3@Sup sample does not have any sintered bodies, but aggregations are still observed. Various shapes of nanoparticles are highlighted in the TEM images shown in Figure by dashed lines. The corresponding schematic shapes of single nanoparticles and sintered bodies are drawn in Figure (i)–(viii). Micromagnetic simulations on the static magnetic responses of these γ-Fe2O3 and γ′-Fe4N nanoparticles (sintered bodies) are given in Figures and 8.

Figure 3

TEM sample preparation and bright field images of γ-Fe2O3 and γ′-Fe4N nanoparticles. Three different samples are prepared for the TEM measurements: original nanoparticle suspensions of as-milled nanoparticles in OA are named BM; original nanoparticle suspensions ultra-centrifuged at 10,000 rpm for 20 min are named Ultra; original nanoparticle suspensions suspended for 24 h are named Sup. A drop of each suspension is obtained from supernatant and dropped onto TEM grid (copper meshes with amorphous carbon coating). The TEM grids are air-dried at room temperature, which will be used for TEM measurements. (a) TEM characterization on γ-Fe2O3 nanoparticle samples. Three different samples are characterized: γ-Fe2O3@Ultra, γ-Fe2O3@BM, and γ-Fe2O3@Sup. The corresponding TEM images are shown on the right panel. Most of the nanoparticles show irregular shapes with some spherical nanoparticles, as shown in (i–iv). Less aggregations of nanoparticles are observed from the supernatant of the sample γ-Fe2O3@Ultra. (b) TEM characterization on γ′-Fe4N nanoparticle samples. Three different samples are characterized: γ′-Fe4N@Ultra, γ′-Fe4N@BM, and γ′-Fe4N@Sup. Well-dispersed nanoparticles are observed from the γ′-Fe4N@Ultra sample. For the samples γ′-Fe4N@BM and γ′-Fe4N@Sup, most of the nanoparticles show irregular shapes and the size of the nanoparticles are larger than that of the starting material, γ-Fe2O3 nanoparticles, as shown in (v–viii). More aggregations are observed which is due to the sintering of nanoparticles during the reduction and nitridation processes as well as the non-superparamagnetic properties of these sintered bodies.

Figure 7

Mumuax3 simulation models. (a) Spherical γ-Fe2O3 nanoparticle with a diameter of 15 nm. (b) Spherical γ-Fe2O3 nanoparticle with a diameter of 25 nm. (c) Cubic γ-Fe2O3 nanoparticle with a side length of 15 nm. (d) Ellipsoid γ-Fe2O3 nanoparticle with a long axis of 30 nm and a short axis of 10 nm. (e) Sintered body, spherical γ′-Fe4N nanoparticle with a diameter of 100 nm. (f) Sintered body, ellipsoid γ′-Fe4N nanoparticle with a long axis of 200 nm and a short axis of 50 nm. (g) γ′-Fe4N nanoparticle cluster, an array of 5 × 5 spherical γ′-Fe4N nanoparticles (diameter of 100 nm).

Figure 8

Evolution of magnetizations in different γ-Fe2O3 and γ′-Fe4N nanoparticles under different dc magnetic fields. Figure a–g corresponds to the Mumax3 simulation models from Figure a–g (a) spherical γ-Fe2O3 nanoparticles with a diameter of 15 nm; (b) spherical γ-Fe2O3 nanoparticles with a diameter of 25 nm; (c) cubic γ-Fe2O3 nanoparticles with a side length of 15 nm; (d) ellipsoid γ-Fe2O3 nanoparticles with a long axis of 30 nm and a short axis of 10 nm; (e) the sintered body, spherical γ′-Fe4N nanoparticle with a diameter of 100 nm; (f) sintered body, ellipsoid γ′-Fe4N nanoparticles with a long axis of 200 nm and a short axis of 50 nm; (g) a 5 × 5 array of 100 nm spherical γ′-Fe4N nanoparticles clustered together. Blue and red colors represent +z and −z components of magnetization, respectively. The arrows represent the x–y components of magnetization.

Hydrodynamic Size of γ-Fe2O3 and γ′-Fe4N Nanoparticles

Figure a shows the hydrodynamic size distribution of the γ′-Fe2O3@BM sample measured by the DLS, where the size distribution peaks around 25 nm, and this too correlates with the TEM image Figure (i) added in the subset. In comparison to γ′-Fe4N@BM, the γ′-Fe2O3@BM sample shows far less aggregations, and this suggests that they have formed well-dispersed nanoparticle sample solution in Isopar G fluid. Figure b shows the hydrodynamic size distribution of γ′-Fe4N@BM sample where the peak is around 100 nm, agreeing with the TEM image Figure (ii) added in the subset. The DLS result of the γ′-Fe4N@BM show another peak at around 1000 nm, this finding suggests that the γ′-Fe4N MNPs are aggregated into clusters. A photograph of γ′-Fe4N@BM sample is given in the glass bottle, where the uppermost part is the supernatant, while the sediments are nanoparticle clusters. Another interesting point to be noted is that the hydrodynamic size distribution in the DLS results appear slightly larger than the observed TEM images. The reason behind this being that the γ-Fe2O3@BM and γ′-Fe4N@BM nanoparticles are formed as a result of wet ball milling in OA. Thus, these nanoparticles are coated with a thin layer of OA (around 2 nm thick), and this makes the hydrodynamic size distribution from DLS appears to be slightly larger than that observed from the TEM images. The schematic drawing of OA coated MNPs is given in Figure c.

Figure 4

Hydrodynamic size distribution of (a) γ-Fe2O3@BM and (b) γ′-Fe4N@BM samples measured by DLS. γ-Fe2O3@BM shows a peak of 25 nm which corroborates with the TEM image in the subset. In addition, it also shows very small peaks at >100 nm suggesting a very small amount of clustering of the nanoparticles in the sample. γ′-Fe4N@BM peaks around 100 nm implying significant sintering of the nanoparticles, and the peaks near 1000 nm signifies the clustering of these sintered bodies. (c) A schematic drawing of OA surfactant conjugation on the nanoparticles. The thickness of the OA layer is around 2 nm, which increased the hydrodynamic size of nanoparticles by 4 nm.

Hydrodynamic Size and Concentration of the γ′-Fe4N@BM Nanoparticles in Isopar G Fluid

The DLS results give a slight discrepancy with that of the TEM images, although we justified the discrepancy by two explanations in the last section: first, as the samples were ball milled, they added an extra 2 nm OA coating around the nanoparticles and made the hydrodynamic size distribution slightly larger than observed from the TEM images; second, the hydrodynamic peaks around 100 and 1000 nm from the γ′-Fe4N@BM sample helped us infer that the particles might have sintered and aggregated. Thus, in order to further characterize the size of the synthesized γ′-Fe4N nanoparticles, another competent optical characterization method, the NTA (details regarding the techniques and models have been mentioned in the Materials and Methods section), is used. An added advantage of characterizing the nanoparticles by the NTA over the DLS is that they also give an information about the concentration of the particles from the solution. Figure shows that the γ′-Fe4N@BM sample has a concentration of the order of 107 particles/mL. Five independent NTA measurements, each having a time span of 60 s, are carried out and labeled as curves I–V in Figure . The nanoparticle concentration is averaged over five measurements and represented by the shadowed curve in Figure . For each measurement, a 1 min video is recorded using the camera of the NTA (videos are provided in the Videos S1–S5). Snapshots at the 15th, 30th, 45th, and 60th seconds are summarized on the right panel of Figure . The NTA results also confirm that the synthesized γ′-Fe4N nanoparticles show both sintering (peaks at 46, 136 and 187 nm) and clustering (peaks at 609 nm and over 1000 nm) as was concluded from the DLS results. It is to be noted here that, the characterization of γ′-Fe4N@BM samples by NTA have been reported and the same by γ-Fe2O3@BM have not been made. This argument has been addressed in the Materials and Methods Section of this paper.

Figure 5

Hydrodynamic size distribution of γ′-Fe4N@BM nanoparticles measured by the NTA. Five independent runs are carried out on the sample. The snapshots of the γ′-Fe4N@BM sample collected under the 403 nm wavelength light projected under the microscope for the NTA. The background color-codes of the snapshots corresponds to the color codes of the hydrodynamic size distribution curves from the 5 consecutive runs, each of 60 s at the 15th, 30th, 45th and 60th second. Bright spots from the snapshots are nanoparticles with larger spots representing clustered nanoparticles. The NTA size distribution corroborates well with DLS results of the γ′-Fe4N@BM, as discussed in Figure b.

Static Magnetization Curves of γ-Fe2O3 and γ′-Fe4N Nanoparticles Measured using a Vibrating Sample Magnetometer

The static (dc) hysteresis loops of γ-Fe2O3@BM and γ′-Fe4N@BM samples are measured at room temperature using a VSM system, as plotted in Figure . The magnetic field is swept from −5000 to +5000 Oe with a step width of 5 Oe (or −2000 to +2000 Oe with a step width of 2 Oe), and the averaging time for each data point is 100 ms so that the magnetizations of nanoparticles are able to relax and align to the field direction. The γ-Fe2O3@BM nanoparticles show a saturation magnetization Ms of 51 emu/g and superparamagnetic property with negligible coercivity of 22 Oe. On the other hand, the γ′-Fe4N@BM nanoparticles show a hysteresis loop with a coercivity of 166 Oe and a saturation magnetization Ms of 164 emu/g. This large coercivity explains the sintered bodies (sizes around 100 nm) from the TEM, DLS, and NTA results in Figures b, 4b, and 5, respectively. At zero field, the remanent magnetization is around 26% of Ms, which explains the severe aggregations from the γ′-Fe4N@BM sample in Figures b and 5. The larger coercivity and higher saturation magnetization Ms of γ′-Fe4N over γ-Fe2O3 suggest that γ′-Fe4N could be a potential candidate for magnetic hyperthermia. Because of this non-superparamagnetic property and sintered bodies of synthesized γ′-Fe4N nanoparticles, the stability of γ′-Fe4N nanoparticles in OA is examined, and given in Supporting Information S5, where the γ′-Fe4N@BM, γ-Fe2O3@BM, γ′-Fe4N@Ultra, and γ-Fe2O3@Ultra samples are placed at room temperature for 7 days of continuous observations.

Figure 6

dc (static) magnetization curves of dried γ-Fe2O3 and γ′-Fe4N nanoparticles measured by VSM under external field ranging from (a) −5000 to 5000 and (b) −2000 to 2000 Oe. The γ-Fe2O3 nanoparticles saturate at 2000 Oe field with a saturation magnetization of 51 emu/g and coercivity of 22 Oe. The γ′-Fe4N nanoparticles saturate at 4000 Oe field with a saturation magnetization of 164 emu/g and coercivity of 166 Oe.

Mumax3 Simulation and Magnetic Properties Analysis

Micromagnetic simulations conforming the different nanoparticle shapes observed in Figure (i)–(viii) are carried out on Mumax3. The simulation models shown in Figure a–d are the evenly dispersed γ-Fe2O3 nanoparticles of different shapes and sizes. Figure e,f shows the γ′-Fe4N sintered body models. Figure g is the γ′-Fe4N nanoparticle cluster model.

The simulation results show that all the γ-Fe2O3 nanoparticles modeled in this work are superparamagnetic and their magnetic moments align to the external dc field as macro-spins (Figure a–d). On the other hand, the 100 nm γ′-Fe4N nanoparticles show domain walls, and their remanent magnetization is non-negligible (see the magnetization in Figure e at 0 Oe). In addition, the nanoparticle clusters are also simulated in this work (γ′-Fe4N nanoparticles with average size of above 500 nm from Figures b and 5) as shown in Figure g. It was found that the remanent magnetization is very significant, which contributes to the hysteresis loop in the VSM result. At zero external field (0 Oe), the 5 × 5 spherical γ′-Fe4N nanoparticle clusters show very strange domain patterns because of the following: (1) each γ′-Fe4N nanoparticle shows domain walls similar to the situation in Figure e; (2) the interparticle interactions are negligible because of the clustering. The collective magnetic responses of γ′-Fe4N nanoparticle clusters result in non-zero remanent magnetization observed in Figure .

Dynamic Magnetic Responses of γ-Fe2O3 and γ′-Fe4N Nanoparticles in Aqueous Solutions

The dynamic magnetic responses are characterized by a homebuilt MPS system.[10,26,48−50,50−54] Where the γ-Fe2O3 and γ′-Fe4N nanoparticles suspended in OA solution (overall volume of 200 μL, concentration of 67 mg/mL) are characterized and recorded by this MPS system. This MPS system generates an alternating current (ac) magnetic field to magnetize MNPs, and as a result, the magnetic moments of nanoparticles relax to align with the external driving field through the Néel- or Brownian-relaxation dominated process or through the joint Néel–Brownian relaxation process (Supporting Information S1).[26,48,55,56] In this work, the magnetic moments of both γ-Fe2O3 and γ′-Fe4N nanoparticles in OA relax along the ac magnetic field through a Néel relaxation-dominated process. Theoretical analysis can be found from Supporting Information S2. The ac magnetic driving field can be tuned with varying frequencies f from 50 to 2850 Hz, and the amplitude of the driving field is set at 170 Oe. The relaxation of magnetic moments of MNPs subjected to ac field is dynamic magnetic responses. This time-varying magnetic moment causes electromotive force (EMF) in a pair of differently wound pick-up coils (Faraday’s law of induction). As a result, the dynamic magnetic responses of γ-Fe2O3 and γ′-Fe4N nanoparticles are recorded (Supporting Information S1) as real-time voltage. Because of the nonlinear magnetic responses of nanoparticles under driving field f, higher odd harmonics at 3f, 5f, 7f, and so forth are found from the frequency domain of the collected voltage signal.[10,49,51,57]Figure a–c summarizes the amplitudes of the 3rd, the 5th, and the 7th harmonics recorded at 3f, 5f, and 7f, respectively, as we vary the driving field frequency from 50 to 2850 Hz. Under all driving field frequencies, the γ′-Fe4N@BM sample shows stronger magnetic responses (higher harmonic amplitudes) over γ-Fe2O3, which indicates that γ′-Fe4N nanoparticles show higher magnetic moment per particle compared to γ-Fe2O3 (discussed in Supporting Information S3). The amplitudes of all the odd harmonics collected from both γ-Fe2O3@BM and γ′-Fe4N@BM samples show a similar trend as we vary the driving field frequency: the harmonic amplitude increases as the driving field frequency f increases, it reaches to a plateau at a critical frequency fcrit (marked by stars in Figures and 12), and then it slowly decays as we further increase the driving field frequency. As the driving field frequency f increases from 50 to 2850 Hz, the γ-Fe2O3 and γ′-Fe4N nanoparticles go through three different regions (labeled as I, II, and III in Figure ): in region I (f-dominant region), the dynamic magnetic responses are dependent on the driving field frequency f, and the magnetic responses increase as f increases; in region III (ϕ-dominant region), the dynamic magnetic responses are dependent on the phase lag ϕ between the magnetic moments of nanoparticles and the fast-changing ac field; in region II (f–ϕ co-led), the transitional stage between regions I and III, both f and ϕ impact the dynamic magnetic responses and the harmonic amplitude curves reach to their maxima.

Figure 9

Recorded (a) 3rd, (b) 5th, and (c) 7th harmonic amplitudes as the driving field frequency varies from 50 to 2850 Hz. The red and black stars mark the critical frequencies fcrit, where γ-Fe2O3 and γ′-Fe4N nanoparticles show highest dynamic magnetic responses. (d) In f-dominant region, the magnetic moments of nanoparticles are almost synchronized with the ac magnetic field, and the detected harmonic amplitude increases as the driving field frequency f increases. (e) In f–ϕ co-led region, the dynamic magnetic responses reach to maxima, where the enhancement effect of f and the attenuation effect of ϕ are equally important. (f) In ϕ-dominant region, the magnetic moments of nanoparticles cannot synchronize with the fast-changing ac magnetic field; thus, a phase lag ϕ between the magnetic moments and field causes attenuated harmonic amplitudes detected by the pick-up coils (discussed in Supporting Information S3).

Figure 12

(a–c) Normalized 3rd, 5th, and 7th harmonics from γ′-Fe4N and γ-Fe2O3 nanoparticles at varying driving field frequencies, respectively. The critical frequencies fcrit at which the harmonics reach to maxima are labeled by stars. (d–k) Harmonic amplitude ratios calculated from γ′-Fe4N and γ-Fe2O3 nanoparticles at varying driving field frequencies.

Real-Time Dynamic Magnetic Responses Recorded by MPS

The relaxation of magnetic moments of MNPs subjected to ac driving field causes detectable EMF in the pick-up coils, and this EMF is a time-varying voltage signal. This analog voltage signal collected from pick-up coils is sampled at a sampling rate of 500 kHz, and the higher odd harmonics (due to the nonlinear dynamic magnetic responses of γ-Fe2O3 and γ′-Fe4N nanoparticles) are extracted. The discrete-time total voltage signal, the 3rd, the 5th, and the 7th harmonics are replotted in Figure a–l. Figure a–f and g–l corresponds to the dynamic magnetic responses of γ′-Fe4N and γ-Fe2O3 nanoparticles, respectively. Each time window records the voltage signals within one period of ac driving field (i.e., 1/f second). The first to the sixth rows correspond to the scenarios, where driving field frequency f = 350, 650, 950, 1250, 1850, and 2450 Hz, respectively. Distortions in the voltage signal are observed from both γ-Fe2O3@BM and γ′-Fe4N@BM samples (highlighted in grey in Figure ), where the voltage signal from γ′-Fe4N@BM sample shows severer distortions over the γ-Fe2O3@BM sample under the same driving field condition. These distortions are caused by the periodically synchronized higher odd harmonics (the 3f, 5f, and 7f harmonic voltage signals denoted in red, green, and blue solid curves). Whenever the crests and troughs of higher odd harmonics are synchronized, the cumulative effect causes small convex and concave in the total signal curve, respectively, as highlighted in Figure a–l.

Figure 10

(a–l) Real-time magnetic responses recorded by MPS system. The total voltage signal (black solid lines) received a pair of pick-up coils, the 3rd (red solid lines), the 5th (green solid lines), and the 7th (blue solid lines) harmonics are plotted in a time window of one period of ac driving field (1/f second). (a–f) and (g–l) Are the summarized dynamic magnetic responses from 200 μL, 67 mg/mL γ′-Fe4N@BM and γ-Fe2O3@BM samples, respectively. The distortions in total voltage signal curves (highlighted in grey) are caused by the periodically synchronized higher odd harmonics (3f, 5f, and 7f). (m–r) Total voltage signal from γ-Fe2O3@BM (red solid lines) and γ′-Fe4N@BM (black solid lines) samples plotted along with the ac driving field (orange dotted lines) in time domain. The phase differences between the voltage and ac field under different driving field frequencies are marked by the blue arrows.

Phase Lags between Dynamic Magnetic Responses and ac Magnetic Fields

According to Faraday’s law of induction, the time-varying magnetic flux induces EMF in the pick-up coils. During one relaxation process of magnetic moments of nanoparticles to align with the direction of external ac field, the stray field causes the EMF in the coils, and there is a 90° phase shift between magnetic moment and detected voltage from pick-up coils. As shown in Figure m–r, the real-time voltages collected from γ-Fe2O3@BM and γ′-Fe4N@BM samples along with the ac fields are plotted during a time window of one period of ac driving field. On top of the 90° phase shift due to the law of induction, we observed phase differences of 51.9, 31.8, 19.5, 13.5, 8.6, and −4.4° between the voltage and field under the driving field frequencies of 350, 650, 950, 1250, 1850, and 2450 Hz, respectively. The detected voltages from γ-Fe2O3@BM and γ′-Fe4N@BM samples are quite synchronous, indicating the identical phase lags of both types of nanoparticles to the ac driving fields. The calculated phase lags of magnetic moments of MNPs to different driving field frequencies are summarized in Figure e.

Figure 11

Measured field-dependent dynamic (ac) magnetization curves of γ-Fe2O3 and γ′-Fe4N nanoparticles in OA subjected to (a) 350 and (b) 2450 Hz driving fields. (c,d) Transformations of dynamic magnetization curves as the driving field frequency increases from 350 to 2450 Hz, for γ′-Fe4N@BM and γ-Fe2O3@BM samples, respectively, are shown. (e) Calculated phase lags of both MNPs to different driving field frequencies.

Dynamic Magnetization Curves of γ-Fe2O3 and γ′-Fe4N Nanoparticles

The dynamic magnetization responses of γ-Fe2O3@BM and γ′-Fe4N@BM samples are calculated using the real-time voltage signals from Figure m–r. Figure a,b shows the normalized magnetization curves of γ-Fe2O3@BM and γ′-Fe4N@BM samples subjected to driving field frequencies of 350 and 2450 Hz, respectively. At both low and high driving field frequencies, the γ′-Fe4N nanoparticles show higher dynamic magnetic responses than γ-Fe2O3. In addition, as we gradually increase the frequency of driving field, the dynamic hysteresis loops transform from long ellipses to flat ovals for both samples, as shown in Figure c,d. This is due to the fact that, as the ac field sweeps faster, both types of nanoparticles are unable to synchronize with the fast-changing ac fields; thus, a larger phase lag of magnetic moment to external driving field is induced.

For magnetic hyperthermia treatments, when MNPs are subjected to the ac field, the area of their magnetic hysteresis loop, A, corresponds to the dissipated energy.[58−60] The power generated by these MNPs or specific absorption rate (SAR), is evaluated by the equation, SAR = A·f. Because the maximum SAR achievable is directly proportional to the saturation magnetization of MNPs, γ′-Fe4N nanoparticles reported in this work can enhance the SAR and meanwhile minimize the dose. As shown in Figure a,b, the dynamic magnetization curves of γ′-Fe4N@BM and γ-Fe2O3@BM are compared under different driving field frequencies, f. The magnetizations are normalized to the magnetizations of γ′-Fe4N@BM. Under both driving field conditions, γ′-Fe4N@BM shows a larger hysteresis loop area A over γ-Fe2O3@BM, indicating that γ′-Fe4N@BM could be potentially applied as high-performance heating sources in hyperthermia treatment with minimum dose.

Normalized Harmonics and Harmonic Ratios

Because the harmonic amplitudes resulted from the dynamic magnetic responses of nanoparticles are dependent on the quantity of nanoparticles from the testing sample and the pick-up coil design (winding number, width and diameter), as discussed in Supporting Information S3.[10,49] Thus, the normalized magnetic responses and the harmonic ratios are used as nanoparticle quantity-independent metrics for characterizing the dynamic magnetic properties of nanoparticles.[26,51]Figure a–c shows the normalized 3rd, 5th, and 7th harmonics from γ′-Fe4N@BM and γ-Fe2O3@BM samples under varying driving field frequencies, corresponding to the recorded harmonic amplitudes in Figure a–c. The normalized harmonic curve of γ-Fe2O3@BM shows a sharper peak compared to γ′-Fe4N@BM. The harmonic ratios are summarized in Figure d–k under driving field frequencies of 150, 350, 650, 850, 1250, 1650, 2250, and 2650 Hz. The harmonics of the γ′-Fe4N@BM sample decays at a slower rate as the harmonic number increases (black lines in Figure d–k).

Stability and Biocompatibility

The stability of OA-coated γ-Fe2O3 and γ′-Fe4N nanoparticles are investigated in Supporting Information S5. With smaller sizes and less aggregations, γ-Fe2O3 nanoparticles show better stability. In this work, OA is used as surfactant on γ′-Fe4N nanoparticles, these lipophilic MNPs show very good dissolvability in polar liquids such as oil. In addition, it is reported that OA can form a dense protective monolayer that binds firmly to the MNP surfaces with enhanced colloidal stability.[61−63] However, for biomedical applications, the lipophilic substances (i.e., OA-coated MNPs) are not good candidates, and thus, the practical use of these MNPs are limited. In the future, we can further functionalize OA-coated MNPs with trialkoxysilanes. Thus, the functionalized nanoparticles can be dispersed in various aqueous solutions such as human serum and plasma.[64] Another option is to improve the biocompatibility of γ′-Fe4N nanoparticles such as the synthesize of core@shell nanoparticles where the high Ms γ′-Fe4N material could be the core and biocompatible materials such as Au, Ag, and SiO2, and so forth serve as the shells.[65−68] Thus, the core@shell nanoparticles would have both high magnetic moments and biocompatibility. In addition, the biocompatibility and colloidal stability of γ′-Fe4N nanoparticles can be further enhanced by conjugating chemical compounds such as chitosan, polyethylene glycol, amino acids, citric acid, and so forth, so that the water solubility of MNPs can increase significantly.[69−72]

Conclusions

In this paper, we reported high magnetic moment, irregularly shaped γ′-Fe4N nanoparticles with OA surfactant. The crystalline structure, static (dc) and dynamic (ac) magnetic properties are characterized by VSM and MPS systems and compared to the γ-Fe2O3 nanoparticles. Our γ′-Fe4N nanoparticles show superior magnetic properties with more than 3 times higher saturation magnetization when compared to γ-Fe2O3, which makes them promising candidates for applications that require high magnetic moment per particle such as in vitro magnetic biosensing, magnetic separation, drug delivery, and so forth.[65] The larger coercivity and higher saturation magnetization Ms of γ′-Fe4N compared to γ-Fe2O3 suggest that γ′-Fe4N could be a potential candidate for magnetic hyperthermia. In addition, because of the sintering in the nitradation process, the synthesized γ′-Fe4N nanoparticles, with irregular shapes, hold great promise for enhancing T2 relaxivity as contrast agents in MRI applications. Nanoparticles with a larger hydrodynamic size distribution generate greater magnetic field gradient, leading to a higher order of proton dephasing.[73−76] The γ′-Fe4N nanoparticles synthesized and characterized in this paper are irregularly shaped and have formed sintered bodies. This will cause them to yield both magnetic fields coupling induced inhomogeneous magnetic field distribution and artificially enhanced magnetic field inhomogeneity. This inhomogeneity is extremely advantageous for MRI applications as they pave the way for varied relaxation rates (R1 and R2) and hence varied relaxation times (R1 = 1/T1 and R2 = 1/T2) over a specific area of tissue to be imaged. This phenomenon enhances the contrast efficiency between adjacent tissues for MRI applications.

Materials and Methods

Chemicals

γ-Fe2O3 powder (purity 99.5%, size 20 nm) is purchased from MTI Corporation, Richmond, CA. OA is purchased from Fisher Scientific, Hampton, NH. Isopar G fluid is purchased from ExxoMobil, Irving, TX.

Wet Mechanical Milling

Four hundred milligram of γ-Fe2O3 and γ′-Fe4N powders are prepared for wet mechanical milling process, respectively. In the case of wet milling, 400 mg powder is dispersed in 6 mL of OA (CH3(CH2)7CH=CH(CH2)7COOH) under an argon atmosphere in a glove box. The milling conditions are 14 mm ball diameter and a vial rotation of 8000 rpm for 8 h.

XRD Characterization

Samples for the XRD measurement are prepared in the glove box. Certain amounts of γ-Fe2O3 and γ′-Fe4N nanoparticles are put on a piece of glass and sealed by epoxy to avoid oxidation when the samples are taken out of the glove box for the XRD measurement. XRD pattern are measured using a Bruker D8 discover 2D diffractometer (40 kV and 35 mA). A cobalt radiation source (wavelength ≈ 1.79 Å) is used to get better signal. The XRD patterns are converted to copper radiation for a convenient comparison.

Sample Preparation and TEM Characterization

The as-milled γ-Fe2O3 and γ′-Fe4N nanoparticles in OA (labeled as γ-Fe2O3@BM and γ′-Fe4N@BM in this paper, with a concentration of 67 mg/mL) are ultra-sonicated for 1 h to make the nanoparticles evenly dispersed. The γ-Fe2O3@BM and γ′-Fe4N@BM suspensions are diluted by 100 times in OA, and then, 10 μL of each suspension (concentration of 0.67 mg/mL) is dropped onto TEM grids. The γ-Fe2O3@BM and γ′-Fe4N@BM in OA are ultra-centrifuged at 10,000 rpm, 9,300 g for 20 min, and then, 10 μL supernatant (labeled as γ-Fe2O3@Ultra and γ′-Fe4N@Ultra in this paper) is drew from each sample and dropped onto TEM grids. The γ-Fe2O3@BM and γ′-Fe4N@BM in OA are sealed and placed at room temperature for 24 h; the larger MNP clusters precipitate to the bottom of suspension, and 10 μL supernatant (labeled as γ-Fe2O3@Sup and γ′-Fe4N@Sup in this paper) is drew from each sample and dropped onto TEM grids. All the six TEM grids are air-dried before the TEM characterizations. A FEI Tecnai T12 TEM (T12, 120 kV) is used to characterize the samples.

DLS and NTA Characterization

Wet mechanically milled γ-Fe2O3 (transparent, refractive index, r.i = 2.91) and γ′-Fe4N (absorbing, irregularly shaped) (50 μL) in OA are diluted 8 times in a synthetic isoparaffinic fluid, Isopar G fluid (r.i = 1.49). The hydrodynamic size distribution of the both γ-Fe2O3 and γ′-Fe4N nanoparticles are characterized using the DLS particle analyzer (Model name: Microtrac NanoFlex).

To establish a stronger corroboration with that of the TEM images and DLS results, the samples of the same order of dilution were characterized in NTA (model: Nanosight LM-10). We would like to clarify here that the hydrodynamic size distribution obtained from DLS for both γ-Fe2O3 and γ′-Fe4N are in correlation with the TEM images. However, only γ′-Fe4N could be characterized by NTA. Although the reason behind why γ-Fe2O3 could not be characterized by NTA is not clear, our estimation is that as NTA uses a wavelength of 403 nm, γ-Fe2O3 having an extremely high transparent refractive index (r.i. = 2.91) in Isorpar G fluid media of much lower refractive index must cause total internal reflection for most of the light. Nevertheless, NTA gives us the concentration of the γ′-Fe4N nanoparticles which in turn specifies the γ-Fe2O3 concentration as both the nanoparticle samples for hydrodynamic size characterization were prepared in a similar manner. The sample volume prepared for both DLS and the NTA particle analyzer is 1.5 mL.

VSM Characterization

γ-Fe2O3 and γ′-Fe4N nanoparticles (25 μL) in OA suspensions are dropped on a filter paper and air-dried and then fit into a 5 mm diameter 12 mm long gelatin capsule (gelcap). During the VSM measurement, the gelcap is inserted into a sample tube and affixed to the sample-rod. The magnetic field is swept from −5000 to +5000 Oe with a step width of 5 Oe (or −2000 to +2000 Oe with a step width of 2 Oe), and the averaging time for each measurement is 100 ms.

Mumax3 Simulation

The TEM images and DLS and NTA characterizations give us a legitimate knowledge on the idea about the shapes and sizes of γ-Fe2O3 and synthesized γ′-Fe4N nanoparticles. We simulate the shapes of the nanoparticles using micromagnetic framework, Mumax3, and observe the magnetization distribution within the γ-Fe2O3 and γ′-Fe4N nanoparticles.[77] The magnetic properties of the nanoparticles were obtained from our experimental and previously reported results listed in Tables and 2. A total of seven different nanoparticle shapes are modeled and plotted in Figure . Four different shapes of γ-Fe2O3 nanoparticles are modeled: spherical with a diameter of 15 nm (Figures a and 8a), spherical with a diameter of 25 nm (Figures b and 8b), cubic with a side length of 15 nm (Figures c and 8c), and ellipsoid with a long axis of 30 nm and a short axis of 10 nm (Figures d and 8d). Three different shapes of γ′-Fe4N nanoparticles are modeled: sintered body, spherical with a diameter of 100 nm (Figures e and 8e), ellipsoid with a long axis of 200 nm and a short axis of 50 nm (Figures f and 8f), a 5 × 5 array of 100 nm spherical γ′-Fe4N nanoparticles clustered together (Figures g and 8g). The uniaxial anisotropy γ-Fe2O3 nanoparticles with easy axis align along [1 1 1] and cubic anisotropy γ′-Fe4N nanoparticles with easy axis align along [1 0 0] are assumed, and external magnetic field is applied along [1 1 0] direction (see Figure ). The mathematical models for uniaxial and cubic anisotropy energy distributions are given in Supporting Information S4.

Table 1

Micromagnetic Simulation Parameters for γ-Fe2O3 MNPs

parameters	description	values
MNP dimension	spherical diameter	(a) 15 nm
		(b) 25 nm
	cubic dimensions	(c) 15 nm × 15 nm × 15 nm
	ellipsoid dimensions	(d) 30 nm × 10 nm × 10 nm
cell size	length × width × thickness	1 nm × 1 nm × 1 nm
A	Gilbert damping factor[78]	0.2
A	exchange constant[79]	10–11 J/m
Ms	saturation magnetization	280 kA/m
Ku	uniaxial anisotropy[80]	4.6 kJ/m3

Table 2

Micromagnetic Simulation Parameters for γ′-Fe4N MNPs

parameters	description	values
MNP dimension	sintered body, spherical diameter	(e) 100 nm
	ellipsoid dimensions	(f) 200 × 50 nm × 50 nm
	nanoparticle cluster composed of 25 spherical MNPs	(g) 100 nm, each arranged in a 5 by 5 square
cell size	length × width × thickness	2 nm × 2 nm × 2 nm
a	Gilbert damping factor[81]	1
A	exchange constant[81]	15 × 10–12 J/m
Ms	saturation magnetization[81]	1430 × kA/m
Kc1	cubic asnisotropy[81]	3 × 104 J/m3

MPS Measurement

200 μL γ-Fe2O3 and γ′-Fe4N nanoparticles in OA suspensions (concentration of 67 mg/mL) are sealed in a plastic vial for MPS measurements. The ac magnetic field frequency f is varied from 50 to 2850 Hz, with amplitude set at 170 Oe. For each run, the vial containing nanoparticles is inserted into the pick-up coils (see Supporting Information S1) and real time voltage signal is collected for 10 s. The analogue voltage signal is sampled at a sampling rate of 500 kHz.