Magnetic properties of cubic FeCo nanoparticles with anisotropic long chain structure

Magnetic nanoparticles (NPs) have attracted researchers’ interests in the past decades due to their wide applications in hyperthermia,^1,2 drug delivery,^3,4 granular giant magnetoresistance,^5,6 catalysis recycling,^7,8 and so on. Magnetic NPs tend to agglomerate due to magnetostatic energy so the cluster or chain-like assemblies are likely to be formed. The magnetic properties could behave quite different for the cluster or chain-like assemblies,^9,10 where NP interactions could not be ignored. So further understanding on the NP assemblies with the presence of interactions is needed. Moreover, the NP assemblies and chain structure also have wide applications such as logic gates,¹¹ magnetic memory,¹² and shaping and diversifying components in magnetic devices.^13,14 So further understanding on the NPs assemblies with the presence of interactions is needed.

The interactions among magnetic NPs can have a significant impact on magnetic properties for both magnetically soft and hard NPs. O. Akdogan et al.¹⁵ reported the effect of exchange interaction on the coercivity of SmCo₅ NPs. Carbon layers were used to increase the distance of the SmCo₅ NPs in order to reduce the interactions among NPs. The coercivity of SmCo₅ NPs sample can increase from almost zero to more than 10 kOe at room temperature due to the decrease of interaction between of SmCo₅ NPs. For magnetic metals such as iron or iron cobalt alloy, they may show superparamagnetic behavior if the nanoparticle size is small enough. The critical size of Fe₆₅Co₃₅ to be superparamagnetic is around 17 nm when the anisotropy constant is K=2×10⁵ erg/cm³ (20 kJ/m³).¹⁶ However, when there are strong interactions among NPs such as occur at a high packing density, NP assemblies are no longer superparamagnetic and can show coercivity even though an individual NP from the assembly may be superparamagnetic.¹⁷ Mørup et al.¹⁸ first observed this phenomenon in 10∼100 nm goethite NPs. They invoked superferromagnetism (SFM) to explain their results, where strong exchange coupling among NPs (superspins) was expected.

Papers that reported on the nanoparticle SFM system were mainly focused on the high packing density case. One of the interesting structures of NP assemblies that have the SFM effect is a long chain structure, which shows unique magnetic properties. Veronica Salgueiriño-Maceira et al.¹⁹ reported the transition from SPM to ferromagnetism of cobalt NPs when they formed chain-like pearl necklaces. Core-shell nanocomposites could also have SFM. A. Zeleňáková et al.¹⁷ reported SFM Fe@SiO₂ nanoparticle assemblies with long chain structure. The size of Fe NPs was around 4 nm. The 4 nm Fe NPs were superparamagnetic at room temperature when they were well-dispersed. However, the assemblies showed ferromagnetic instead of superparamagnetic. The magnetic NP long chain structure needs a deeper understanding to broaden its applications. Long chain NP assemblies produced before were mainly prepared by chemical methods and the spherical shape of NPs was obtained with isotropic long chain structure. Few papers investigated the anisotropic long chain NP assemblies with cubic shape NPs, which have more contact surface and exchange coupling effect than that of NPs with sphere shape.

In this work, we will focus on the magnetic properties of anisotropic long chain structure of high magnetic moment FeCo alloy NP assemblies. High magnetic moment NPs can provide more magnetic signals which is very helpful for their applications.²⁰ Herein, sputter based gas-condensation method was used to prepare cubic shape FeCo NPs. Two magnets were loaded on a substrate holder to align NPs during their deposition process in order to get the anisotropic long chain structure (named S1). As an experiment control, well-dispersed FeCo NPs (named S2) were also prepared. The long chain FeCo NPs showed ferromagnetic behavior at a wide temperature range (10 K-400 K).

FeCo nanoparticles were prepared using dc sputter based gas-condensation method. Details of the system could be found in our group’s previous publication.²⁰ Fe₆₀Co₄₀ alloy target was used as the source material to make long chain and well-dispersed FeCo NPs. Base pressure of the system was around 10⁻⁷ Torr. Sputter current was fixed as 0.5 A with the sputter voltage around 250 V. Ultrapure Ar gas was used as sputter gas (pressure ∼350 mTorr) and carrier gas which helped carry NPs generated by dc sputter to the deposition chamber. Two magnets were fixed on a substrate holder to align NPs and obtain the anisotropic long chain NP assemblies during their soft landing process. The details of magnetic field assisted alignment could be found in the paper published previously by our group.²¹ Single crystal silicon wafer was used as the substrate to collect FeCo NPs. After NPs collection, Ti capping layer was deposited on the substrate by dc sputter to reduce further oxidation after taking them out of the system.

Morphology of the NPs (long chain and well-dispersed) and diffraction patterns were characterized by transmission electron microscope (TEM, FEI T12, 120 kV). X-ray diffraction (XRD, Bruker D8 Discover 2D, 45 kV, 40 mA) was used to measure the crystal structure. Physical Properties Measurement System (PPMS) integrated with the vibrating sample magnetometer (Quantum Design) was used to measure the magnetic properties. FeCo NPs were deposited on silicon wafer for XRD and PPMS measurement.

Figure 1 showed the TEM images and the selected area electron diffraction (SAED) patterns of FeCo NPs. Figure 1(a) was the zoom-out image of S1 with long chain structure. The chains were generally orientated due to magnetic field assisted alignment. The zoom-in image was shown in figure 1(b). FeCo NPs had cubic shape and the average size was 15 nm. Figure 1(d) and 1(e) were zoom-out and zoom-in images of well-dispersed sample S2. In sample S2, the size of NPs was similar as sample S1, which were well dispersed. The diffraction rings of SAED patterns for S1 (figure 1(c)) and S2 (figure 1(e)) indicated that S1 and S2 both had bcc phase. Figure 2 showed the size distributed of FeCo NPs. The mean size of the NPs was 15 nm. XRD pattern was also taken to characterize the phase information of the FeCo NPs. As shown in figure 3, FeCo NPs were bcc phase which was consistent with the SAED patterns. The background slope from 20 to 40 degree is due to silicon substrate. And the wide diffraction peak around 50 to 60 degree was background noise from silicon substrate but not FeCo samples. The composition of the NPs was around Fe₆₀Co₄₀ based our previous work.²² 

Hysteresis loops of these two samples were measured in-plane parallel to the substrate. The samples measured by PPMS had diamagnetic background due to silicon substrate. We measured the silicon substrate which showed diamagnetic. But when we used the measured slope as a reference before doing the subtraction, we found that the slope need to be modified a little bit. So we choose the slope to make sure the high field range (∼80 %) is almost flattened. Therefore, a line with positive slope was added to the measured M-H loops to compensate the diamagnetic background from silicon substrate. And then M-H loops shown in figure 4 were obtained. As shown in figure 4, shapes of two loops were quite different. S1 (open squares) showed ferromagnetic behavior with coercivity of 765 Oe at 300 K, while S2 (open circles) had much smaller coercivity ∼30 Oe. The FeCo NPs of S2 were mostly isolated and the small coercivity was from NPs with larger size due to their size distribution as shown in figure 2, and some short NP chains as shown in figure 1(e). The coercivity of the well-dispersed sample S2 is large compared with bulk FeCo alloy whose coercivity is around 2 Oe.²³ The coercivity of S1 was much larger than that of S2. Hysteresis loop of S1 at 10 K as shown in figure 4 was very similar to the one at 300 K. At 10 K M_r/M_s ratio was 0.56 and 0.53 at 300 K, where M_r is the remanent magnetization and M_s is the saturation magnetization, and coercivity increased a little bit from 765 Oe to 810 Oe. So for S1, thermal energy just played a minor role while exchange coupling interaction was the dominant factor.

Other researchers also reported long chain NP assemblies with ferromagnetic properties even though the chains were isotropic instead of anisotropic as we reported in this work. Zeleňáková et al.¹⁷ reported that the 4 nm Fe NP chain-like structure with SiO₂ shell showing hysteresis at room temperature though 4 nm Fe NP itself was superparamagnetic. The long chain Fe@SiO₂ sample had coercivity of 200 Oe at 300 K and the remanence ratio M_r/M_s was around 0.28, indicating that even the superparamagnetic Fe NPs could show ferromagnetic behavior due to strong exchange interactions between NPs in the long chain structure. The SFM was used to explain ferromagnetic behavior of superparamagnetic NP long chains, where every single NP worked as a superspin and the superspins were ferromagnetically exchange coupled together to form the SFM state which displayed ferromagnetism. The magnetic properties of long chain FeCo NP assemblies were similar to Fe@SiO₂ chain-like nanocompostie and their ferromagnetic performance (shown in figure 4) was also due to SFM effect between superspins’ exchange interaction. The FeCo NPs with anisotropic long chain structure in this work had higher remanence ratio (∼0.53) than that of the isotropic one (∼0.28),¹⁷ indicating the orientation and strong exchange coupling between NPs in anisotropic long chains. In figure 1, well crystallized FeCo alloy NPs show cubic shape which can have more contact area with other NPs compared to the spherical ones. Therefore higher exchange effect between NPs was expected, which was consistent with its higher coercivity and M_r/M_s ratio as shown in figure 4.

In order to further understand the magnetic properties of the anisotropic long chain FeCo NP assemblies, temperature dependent magnetization zero-field-cooled (ZFC) and field-cooled (FC) curves were measured at temperatures ranging from 10 K to 400 K. The ZFC/FC curves of sample S1 and S2 were measured as follows. Both samples were heated to 400 K with zero magnetic field. Samples were then cooled down to 10 K at zero field. Then temperature was increased to 400 K and meanwhile small magnetic field were applied (200 Oe for S1 and 10 Oe for S2). These two samples had much different coercivity, 765 Oe for S1 and 30 Oe for S2 at 300 K. So we choose 200 Oe for S1 and 10 Oe for S2. The reason why we do not choose 10 Oe for S1 is that the field is too small for S1 and the ZFC/FC curves are almost merged together. So the ZFC curves M vs. T were obtained as shown in figure 5 open squares. For the FC curves, small magnetic fields (200 Oe for S1 and 10 Oe for S2) were applied during the cooling process from 400 K to 10 K and then M vs. T curves were obtained when heating samples from 10 K to 400 K with the same applied magnetic field and then FC curves were collected as shown in figure 5 open circles. For sample S1, the shape of ZFC/FC curves were similar while FC curve displayed higher magnetic moment, indicating that S1 showed strong interactions between NPs in the temperature range from 10 K to 400 K and thermal energy is less than the interaction energy (exchange/dipole and Zeeman) which was consistent with the discussion on the hysteresis loops mentioned above. For the well-dispersed sample S2, the shape of ZFC/FC curves were different as shown in figure 5(b) and the two curves appear to merge together when the temperature is a greater than 400 K. For ZFC curve of S2, the magnetic moment was almost constant from 10 K to 150 K, where the magnetic moments were freezed in low temperature region. As the temperature increased further, the thermal energy increased and the magnetic moments got more freedom to rotate. So the magnetic moments tend to partially follow external field and the total magnetic moment of S2 increased. As the temperature kept increasing, the thermal energy played a dominant role and the magnetic moments became more disorder which made the total magnetic moment of S2 decrease. Therefore, well-dispersed sample S2 was more thermal sensitive than the long chain structure sample S1.

Time dependent remanent magnetic moment M_r curves were also measured to further investigate the magnetic properties of sample S1 and S2. Samples were firstly saturated by applying 10000 Oe field and then set the field to zero. The remanent magnetic moment was collected every 20 sec. In figure 6, M_r vs. time curves of the long chain sample S1 were shown at four temperature points: 300 K, 250 K, 200 K and 150 K. Figure 7 was the Mr vs. time curves for the well-dispersed sample S2. As shown in figure 6 and 7, sample S1 had longer magnetic relaxation time than that of sample S2 at all of the four temperature points recorded, indicating more time stability for the long chain sample. Meanwhile, sample S2 had relatively larger error bar than that of S1 (the large error bar might also be due to noise floor on the instrument), demonstrating that S2 was more thermal sensitive than S1, which was consistent with ZFC/FC curves discussed before. So the long chain assemblies had stronger exchange coupling and thermal stability.

Well crystalized cubic shape FeCo NPs were successfully synthesized using sputter based gas-condensation method. Two types of FeCo NPs samples, anisotropic long chain assemblies and well-dispersed NPs, were prepared. Strong exchange coupling interaction between NPs in the long chain sample made its magnetic performance quite different from that of the well-dispersed one. The long chain assemblies showed 765 Oe coercivity at 300 K which was much higher than that of the well-dispersed sample (∼30 Oe). The long chain sample had M_r/M_s ratio up to 0.53 due to the aligned anisotropic chains. From ZFC/FC curves and time dependent remanent magnetic moment curves, long chain sample had higher thermal stability due to SFM exchange coupling between NPs.

Parts of this work were carried out using the Characterization Facility, which receives partial support from NSF through the NSF Minnesota MRSEC program under Award Number DMR-0819885. Authors thank the partial support from the Institute for Rock Magnetism, Department of Earth Science, University of Minnesota, Twin Cities, for the use of instruments. The authors thank Prof. Bin Ma and Dr. Junyang Chen for enlightening discussions.