Dependence of phase configurations, microstructures and magnetic properties of iron-nickel (Fe-Ni) alloy nanoribbons on deoxidization temperature in hydrogen

Iron-nickel (Fe-Ni) alloy nanoribbons were reported for the first time by deoxidizing NiFe2O4 nanoribbons, which were synthesized through a handy route of electrospinning followed by air-annealing at 450 °C, in hydrogen (H2) at different temperatures. It was demonstrated that the phase configurations, microstructures and magnetic properties of the as-deoxidized samples closely depended upon the deoxidization temperature. The spinel NiFe2O4 ferrite of the precursor nanoribbons were firstly deoxidized into the body-centered cubic (bcc) Fe-Ni alloy and then transformed into the face-centered cubic (fcc) Fe-Ni alloy of the deoxidized samples with the temperature increasing. When the deoxidization temperature was in the range of 300 ~ 500 °C, although each sample possessed its respective morphology feature, all of them completely reserved the ribbon-like structures. When it was further increased to 600 °C, the nanoribbons were evolved completely into the fcc Fe-Ni alloy nanochains. Additionally, all samples exhibited typical ferromagnetism. The saturation magnetization (Ms) firstly increased, then decreased, and finally increased with increasing the deoxidization temperature, while the coercivity (Hc) decreased monotonously firstly and then basically stayed unchanged. The largest Ms (~145.7 emu·g−1) and the moderate Hc (~132 Oe) were obtained for the Fe-Ni alloy nanoribbons with a mixed configuration of bcc and fcc phases.

at% and it reached about 879.5 K at 75 at% Ni 12 . Furthermore, it is noteworthy that the preparation condition and topographic structure (shape and size) are also significant to the magnetic nanomaterials. By controlling the oxidation kinetics, in 2013, Ucar et al. successfully tuned the T c of the ball milled γ -Fe 70 Ni 30 and γ -Fe 72 Ni 28 nanoparticles (NPs), and severally calculated their refrigeration capacities to be about 470 and 250 J·kg at 5 T 6 . Moghimi et al. revealed in 2014 that the maximum magnetization of the Fe 63 Ni 37 concave cubic (CC) NPs (85 emu·g −1 ) at 0.5 T was higher than that of the cuboctahedron (COh) NPs (40 emu·g −1 ) while the H c for the CC NPs (100 Oe) was smaller than that for the COh NPs (188 Oe) although both of them had same composition 13 . The discrepancy was ascribed to different shapes and phase (fcc, bcc) combinations.
One-dimensional (1D) Fe-Ni alloys (nanowires, nanochains and nanofibers) recently are of particularly scientific and technical interest amongst the numerous magnetic nanostructures. This is surely attributed to their unique natures of effectively conquering the abnormally agglomeration between NPs and exhibiting many useful magnetic properties. For instance, in 2005, Liu et al. found that the H c of the electrodeposited Fe 1-x Ni x nanowires assisted by AAO templates didn't monotonically decrease as Ni content increased at 300 K 14 . In 2012, Jia et al. 15 and Lu et al. 16 respectively prepared FeNi 3 nanochains through a template-free microwave-hydrothermal method and a CTAB mediated aqueous solution method to obtain the enhanced H c and M s , respectively compared with that of bulk FeNi 3 and monodispersed FeNi 3 nanospheres. Lu also revealed that the imaginary permeability of the FeNi 3 nanochains showed a broad resonance peak in the frequency range of 150 MHz to 4 GHz 16 . Thus it is believed that preparing a greater variety of 1D Fe-Ni alloy nanostructures with more satisfactory magnetic properties should be significantly meaningful.
To prepare 1D Fe-Ni alloy nanostructures, electrospinning technique is intensively recommended due to its simplicity, efficiency and versatility [17][18][19] . Earlier reports have certified that nanofibers 20 , nanotubes 21 and nanoribbons 22 could be directly obtained via the electrospinning route combined with a series subsequent treatments, and that all of these 1D nanostructures were tens of micrometers in length and their diameters or width can be flexibly tuned from nanometers to micrometers. To date, however, only Lee et al. prepared the Fe-Ni nanofibers using electrospinning route followed by air-calcination and H 2 -deoxidization, and studied the electromagnetic wave absorption property in the GHz frequency range 23 . Unlike nanofibers and nanotubes possessing round cross sections, on the other hand, novel nanoribbons are slightly different due to their rectangular cross sections. Nanoribbons could also be regarded as a development by cutting a finite-width slice from the two-dimensional nanosheets. In recently works, we have demonstrated that the electrospun SrFe 12 O 19 and CoFe 2 O 4 nanoribbons both exhibited deeply impressed magnetic properties 22,24 . To the best of our knowledge, there is still no public report on Fe-Ni alloy nanoribbons. Furthermore, the worth emphasizing is that this work is an indispensable part in research system of iron-based nanoribbons in the range of ferrite to iron-based alloys. Therefore our main objective in this study is to synthesize Fe-Ni alloy nanoribbons through the electrospinning route followed by air-annealing and H 2 -deoxidizing successively, and to systematically explore the effect of deoxidization temperature on their morphologies, phase configurations and room temperature magnetic properties.

Experimental Section
Synthesis of samples. Chemical agents of polyvinylpyrrolidone (PVP), nickel nitrate hexahydrate (Ni(NO 3 ) 2 ·6 H 2 O), iron nitrate nonahydrate (Fe(NO 3 ) 3 ·9 H 2 O) with analytical-grade were used without any purification. The Fe-Ni alloy nanoribbons were synthesized through an electrospinning route followed by air-annealing firstly and then H 2 -deoxidation successively. The detailed procedure is described as below. Firstly, 0.132 g Ni(NO 3 ) 2 ·6 H 2 O, 0.368 g Fe(NO 3 ) 3 ·9 H 2 O and 0.4 g PVP were sufficiently dissolved in a mixed solution of 1.5 g deionized water and 2.4 g C 2 H 5 OH under vigorous stirring for overnight to form a steady and viscous spinning solution. Secondly, the obtained solution was placed in a glass syringe equipped with a stainless steel needle (0.4 mm inner diameter). Subsequently, the electrospinning was carried out at a DC high-voltage of 15 kV, a spacing of 20 cm between the needle tip and metal collector, and a flow rate of 0.3 mL·h −1 . Several hours later, the PVP/Fe(NO 3 ) 3 /Ni(NO 3 ) 2 precursor nanoribbons were deposited on the collector with a form of an uniform membrane. Thirdly, the as-spun precursor membrane was aged initially at 60 °C in a drying box for a few hours followed by annealing at 450 °C for 2 h in air with a heating rate of 1 °C·min −1 . Then the as-annealed products S0 was divided into four groups, which were further deoxidized at 300, 400, 500 and 600 °C in H 2 atmosphere with a constant flow for 1 h respectively. Finally, the corresponding Fe-Ni alloy nanoribbons S1-S4 were obtained.
Characterization of samples. The morphologies and microstructures of all samples were observed by field emission scanning electron microscopy (FESEM, Hitachi S4800, Japan) and transmission electron microscopy (TEM, FEI Tecnai TM G 2 F30, USA) equipped with an energy dispersive X-ray spectroscopy (EDX, Oxford Instrument, UK). The phase components, elements and corresponding chemical states were detected by powder X-ray diffraction on a X-ray diffractometer (XRD, Philips X'Pert Pro MPD, the Netherlands) with Cu-Kα irradiation (λ = 1.54056 Å), high-resolution transmission electron microscopy (HRTEM), EDX and X-ray photoelectron spectroscopy (XPS, Kratos AXIS Ultra DLD , 600 W, UK) with an monochrome Al-Kα probe beam. All the recorded XPS spectra were calibrated using C 1 s peak with the binding energy of 284.8 eV and were analyzed through Gaussian-fitting. A vibrating sample magnetometer (VSM, Lakeshore 7403, USA) was used to investigate the room temperature magnetic properties of the samples.

Results and Discussion
Phase configurations and microstructures of samples. Figure 1 shows the XRD patterns of the H 2 -deoxidized samples S1-S4 as well as the air-annealed sample S0. For S0, the sample is certified as a single phase material. All the observed diffraction peaks marked by the black diamonds can be unambiguously indexed to the standard spinel NiFe 2  That is, NiFe 2 O 4 in S0 is partially deoxidized to fcc and bcc Fe-Ni alloys in S1. Then sample S1 is a complex of the cubic NiFe 2 O 4 , fcc Fe-Ni and bcc Fe-Ni. However, the peak intensities of fcc Fe-Ni is weaker than that of bcc Fe-Ni. When S0 is H 2 -deoxidized at 400 °C, the peaks of NiFe 2 O 4 phase disappear completely in sample S2, only the peaks of fcc and bcc Fe-Ni phases are observed and the peak intensities of fcc Fe-Ni are strengthened obviously. Thus sample S2 is a pure alloy just with a mixed phase of fcc and bcc Fe-Ni. When the deoxidization temperature is further risen to 500 °C, the peaks of fcc Fe-Ni strengthen further but that of bcc Fe-Ni disappear. Therefore, the as-deoxidized sample S3 is a pure fcc Fe-Ni alloy. Seeing from the phase transformation from S0 to S3, it can be concluded that the cubic NiFe 2 O 4 is deoxidized to bcc Fe-Ni at first and then transformed into fcc Fe-Ni with the deoxidization temperature increasing. For S4 resulted from deoxidizing S0 at 600 °C, its phase is fcc Fe-Ni undoubtedly. Using Debye-Scherrer equation 25 , the average grain sizes (D) of fcc Fe-Ni alloy in S1, S2, S3 and S4 are estimated to be about 17.2 ± 0.30, 29.8 ± 0.32, 34.1 ± 0.27 and 39.9 ± 0.35 nm respectively. It indicates that the as-deoxidized fcc Fe-Ni alloy grains gradually grow with the increased deoxidization temperature.
The morphology features of all samples S0-S4 with the defined phase configurations are examined by FESEM firstly. As presented in Fig. 2(a-c), a large number of continuously ribbon-like structures with random direction are found for sample S0 (NiFe 2 O 4 ). The ribbon-width is ranged from 269 nm to 512 nm and the average width is about 348.4 nm. Each of nanoribbon is quite uniform in width along the length direction and is constructed by small nanoparticles (NPs). When the NiFe 2 O 4 nanoribbons are deoxidized at 300 to 600 °C, some changes are also occurred on the microstructures beyond on the phase configurations of the resultant products, which are shown in Fig. 2(d-o). Firstly, in Fig. 2(d-f), when the deoxidization temperature is 300 °C, the obtained composite nanoribbons S1 of NiFe 2 O 4 + fcc Fe-Ni + bcc Fe-Ni well keep the continuous structures of the NiFe 2 O 4 nanoribbons. Secondly, when the temperature is risen up to 400 and 500 °C, the NPs contained within the composite nanoribbons S2 of fcc Fe-Ni + bcc Fe-Ni ( Fig. 2(g-i)) and pure nanoribbons S3 of fcc Fe-Ni ( Fig. 2(j-l)) increase sharply and fuse together. And the nanoribbons S3 become uneven and bent. Compared to S0, however, the average ribbon-widths of S1, S2 and S3 don't change much and they are about 347.9, 345.1 and 345.4 nm respectively. Thirdly, when the deoxidization temperature is further increased to 600 °C, basically no nanoribbons except for NP-chain (or nanochains) are found in the final fcc Fe-Ni sample (S4) (Fig. 2(m-o)). Due to the excessive growth and fusion together of their NPs, the nanochains S4 are constructed by the NP-clusters as one-by-one and the average diameter is about 241 nm. Moreover, unlike the nanoribbons of S0-S3, the nanochains of S4 connect with each other by lots of junctions, which were marked by the red circles. Hence the whole sample S4 looks more like a three-dimensional network of the nanochains.
TEM analysis was carried out to further explore the microstructures of S0-S4. Firstly, Fig. 3(a,e,i,m,q) describe the typical high-magnification TEM images of the random individuals of all samples. These closer TEM examinations powerfully certify the ribbon-like structures of S0-S3 and the chain-like structure of S4. Owing to the continued particle growth during the second annealing in H 2 , NPs of the deoxidized S1-S4 are bigger than that of their precursor nanoribbons S0. For S1-S4, the particle size increase gradually with the deoxidization temperature increasing. Secondly, Fig. 3(b-d,f-h,j-l,n-p,r-t) respectively display the corresponding HRTEM images. In Fig. 3(b-d), the lattice fringes scanned from the selected NPs of S0 are measured to be about 0.209, 0.249 and 0.250 nm, which are indexed well to (400), (311) and (311) planes of NiFe 2 O 4 (JCPDS card no. 54-0964). In Fig. 3(f-h)  respectively. Thus it proves that the chemical phase of S0 nanoribbons is pure NiFe 2 O 4 and the S1 nanoribbons are assembled by the fcc Fe-Ni, bcc Fe-Ni and spinel NiFe 2 O 4 NPs. In other words, it further demonstrates that NiFe 2 O 4 of S0 transforms to the complex of fcc Fe-Ni alloys, bcc Fe-Ni alloys and spinel NiFe 2 O 4 ferrite after annealing in H 2 at 300 °C for S1. Similarly, from Fig. 3(j-l,n-p,r-t), it also notifies that the nanoribbons S2 are constructed by the fcc and bcc Fe-Ni NPs, the nanoribbons S3 and nanochains S4 are both purely constructed by the fcc Fe-Ni NPs, respectively.
To identify the chemical elements and corresponding surface chemical states of S0-S4, the TEM-EDX and XPS spectra are presented. With the exception of the extra signals of C and Cu elements descended from the carbon-coated copper grid clamping the samples for TEM measurements, as shown in Fig. 4a, the peaks of Fe, Ni and O elements are detected together in the spectra of all samples S0-S4. Of which, the atomic ratio of Fe: Ni (38.6: 19.1) of S0 is basically equivalent to the stoichiometric ratio (n(Fe): n(Ni) = 2: 1) of NiFe 2 O 4 . Although S1-S4 are respectively the complex of NiFe 2 O 4 + fcc Fe-Ni + bcc Fe-Ni (S1), the complex of fcc Fe-Ni + bcc Fe-Ni (S2) and the pure fcc Fe-Ni (S3 and S4), their total atomic ratios of Fe: Ni (30.9: 15.6, 50.9: 25.3, 41.5: 20.3 and 49.6: 24.7) are still very close to 2: 1. Differently, however, the O contents are found to have a decrease in S2-S4 compared to S0 and S1. For S0 and S1, O elements are mainly belonged to the NiFe 2 O 4 , but the small amounts of O elements detected in S2-S4 are probably ascribed to the self-oxidized surfaces. This assumption will be carefully discussed as below. According to the obtained atomic ratios of Fe: Ni, additionally, the S2-S4 can be roughly expressed as Fe 67 Ni 33 . Fig. 4(b,d,f,h,j) and (c,e,g,i,k) reveal the recorded high-resolution XPS spectra of Ni 2p 3/2 and Fe 2p 3/2 , respectively. All the binding energy scales are calibrated using the C 1 s peak with a binding energy of 284.8 eV. For precursor nanoribbons S0, the only peak located at about 854.50 eV in its Ni 2p 3/2 spectrum (Fig. 4(b)) is assigned to the Ni 2+ of O-Ni bonds 26,27 . The peak observed in Fe 2p 3/2 spectrum (Fig. 4(c)) can be fitted into two peaks located at about 711.81 and 709.91 eV, which are respectively assigned to Fe 3+ in tetrahedral-site and Fe 3+ in octahedral-site of the inverse spinel NiFe 2 O 4 28,29 . For reduced nanoribbons S1, besides the peak of Ni 2+ with a higher binding energy presented in its Ni 2p 3/2 spectrum (Fig. 4(d)), another peak with a lower binding energy at about 852.11 eV is also observed and assigned to the metallic Ni 26,27,30 . By fitting, similarly, its Fe 2p 3/2 spectrum (Fig. 4(e)) contains four peaks, which can be fallen into two categories. The first and second peaks located at higher binding energies are surely assigned to Fe 3+ of the NiFe 2 O 4 , but the third and fourth peaks located at lower binding energies of about 707.24 and 706.03 eV are both assigned to metallic Fe 29,31,32 . Based on the experiment process and above analysis, the metallic Ni and Fe should be derived from the partial deoxidized NiFe 2 O 4 and they are further compounded to fcc and bcc Fe-Ni alloy NPs. For S2-S4 (Fe-Ni alloys), the peaks of Ni 2+ and metallic Ni are also found in their Ni 2p 3/2 spectra (Fig. 4(f,h,j)) as well as the peaks of Fe 3+ and metallic Fe found in Fe 2p 3/2 spectra (Fig. 4(g,i,k)). However, the peak signals of Ni 2+ in S2-S4 are much weaker than that of metallic Ni while the peaks of Fe 3+ are much stronger than that of metallic Fe. It means that the reduced Fe element is oxidized more easily than Ni element in air. Then Ni element in surfaces of S2-S4 mainly exists in metallic state but Fe element exists in oxide state. Because no any obvious Ni-or Fe-oxides are observed from the XRD patterns and HRTEM images of S2-S4, it suggests that the detected signals of Ni 2+ and Fe 3+ should be ascribed to the slightly oxidization shells on their surfaces.  35,36 . In our case, thus, the experimental value is very close to the estimated bulk value. The above investigations of XRD and XPS have indicated that a thin oxide shells present at the surface, in which the magnetic spins are disordered and pinned 37,38 . This could account for the slight difference of M s between our result and the calculated bulk value. On the other hand, except for S1, the H c values of S2-S4 (pure Fe 67 Ni 33 ) are all lower than 150 Oe. Based on the above results, hence, the Fe-Ni alloy nanoribbon is a considerable soft magnetic nanomaterial. Additionally, the most significant feature of the Fe-Ni alloy nanoribbons is their unique geometry of the ribbon-like structure self-assembled by NPs. Compared with the zero-dimensional alloy materials, Fe-Ni alloy nanoribbons have an amazing advantage is that they can be used in the miniaturizing and high density electronic components such as antennas and stealth devices. Therefore, the high-frequency properties can probably be a major future research content and direction. To further understand the effect of deoxidization temperature on the magnetic properties of S1-S4, the M s and H c values as functions of the deoxidization temperature are shown in Fig. 5(b). It is found that M s of these Fe-Ni alloy samples firstly increases, then decreases, and finally increases with the deoxidization temperature increasing, but the H c monotonously decreases firstly and then basically stays unchanged. To understand this, the plausible contributing reasons are clarified. For M s , firstly, because the deoxidization temperature of 300 °C in H 2 for S1 is too low to completely deoxidize NiFe 2 O 4 of S0 into Fe-Ni alloys but that of 400 °C for S2 is big enough to deoxidize NiFe 2 O 4 into Fe-Ni alloys completely, a certain amount of residual NiFe 2 O 4 in S1 certainly pulls down the total M s . Besides, the grain size of S1 (D 1 ≈ 17.2 nm) is much smaller than that of S2 (D 2 ≈ 29.8 nm). Hence even if S2 has a mix combination of fcc and bcc Fe-Ni phases, its M s is still increased considerably compared to that of S1. Secondly, when the deoxidization temperature is risen up to 500 °C for S3, bcc Fe-Ni phase is transformed into fcc Fe-Ni phase and then the obtained S3 is pure fcc Fe-Ni with an increased grain size (   works have pointed out that the M s of bcc phase is universally larger than that of fcc phase in Invar Fe-Ni alloys (30 ~ 38 at% Ni) 13,39,40 , and that if the Fe-Ni alloy is a combination of bcc and fcc phases, its M s linearly decreases with the fcc phase increasing 13 . Then S3 has a lower M s than that of S2. Thirdly, when the temperature is sequentially increased to 600 °C, the composition of obtained S4 is still fcc Fe-Ni. Due to the same composition of Fe 67 Ni 33 45 . Using the suggested method, similarly, we also respectively estimate the A, K 1 and L ex of Fe 67 Ni 33 alloys to be about 6.68 × 10 -12 J·m −1 , 3.18 × 10 4 J·m −3 and 14.5 nm. In our case, hence, all of the average grain sizes of Fe 67 Ni 33 NPs of S2-S4 are larger than the L ex . Furthermore, the inset in Fig. 5(b) shows that the variation of H c with the grain size (D) from S1 to S3 except for S4 generally conforms to the H c ∝ 1/D situation of the RAM theory. Therefore, the decrease of H c from S1 to S2 to S3 is caused by the increase of grain size. Although all deoxidized samples are self-assembled by the Fe-Ni NPs, thirdly, the nanochains of S4 are still slightly different from the nanoribbons of S2-S3. On one hand, the nanochains are connected by the big clusters of Fe-Ni NPs. Within these clusters range, Fe-Ni NPs contact with each other tightly. That way, the exchange interactions between Fe-Ni NPs are increased at a certain extent. On the other hand, most of the nanochains of S4 connect with each other through the junctions, being marked by the red circles in Fig. 2m, but the nanoribbons of S2 and S4 are more isolated individuals. So there are still increased exchange interactions between the nanochains in S4 compared to S2 and S3. Hence the above two aspects of increased exchange interactions in S4 effectively compensate the decrease of H c caused by the increase of grain size, and then the net H c dose not decrease. Consequently, it is concluded that the deoxidization temperature has a great influence on the microstructures of Fe-Ni samples and then further affects the magnetic properties.

Conclusions
Briefly, Fe-Ni alloy nanoribbons were successfully synthesized through an electrospinning route followed by air-annealing and H 2 -deoxidizing orderly. The results revealed that the deoxidization temperature had a significant influence on the phase components, morphologies and magnetic properties of the final samples. On the one hand, the air-annealed cubic NiFe 2 O 4 ferrite could be firstly deoxidized into bcc Fe-Ni alloy and then transformed into fcc Fe-Ni alloy in H 2 with the deoxidization temperature increasing. When the temperature is in the range of 300~500 °C, all the obtained samples possessed the ribbon-like structures perfectly. While when the temperature was further risen to 600 °C, the Fe-Ni alloys nanoribbons were evolved into the Fe-Ni alloy nanochains. On the other hand, the Fe-Ni alloys nanoribbons and nanochains both performed excellent ferromagnetism. Their M s first increased, then decreased, and finally increased with the deoxidization temperature increasing, but their H c monotonously decreased firstly and then basically stayed unchanged. Last but not least is that these results can provide a meaningful reference to related fundamental research and practical application in future.