Alexsandro dos Santos E. da
Cruz
*a,
Marcos V.
Puydinger dos Santos
a,
Raul B.
Campanelli
a,
Pascoal G.
Pagliuso
a,
Jefferson
Bettini
b,
Kleber R.
Pirota
a and
Fanny
Béron
a
aInstitute of Physics Gleb Wataghin (IFGW), University of Campinas (UNICAMP), Campinas, 13083-859 São Paulo, Brazil. E-mail: asec@ifi.unicamp.br
bBrazilian Center for Research in Energy and Materials (CNPEM), Brazilian Nanotechnology National Laboratory (LNNano), Campinas, 13085-903 São Paulo, Brazil
First published on 19th April 2021
Recently, core–shell nanowires have been proposed as potential electrical connectors for nanoelectronics components. A promising candidate is Mn5Si3 nanowires encapsulated in an oxide shell, due to their low reactivity and large flexibility. In this work, we investigate the use of the one-step metallic flux nanonucleation method to easily grow manganese silicide single crystal oxide-protected nanowires by performing their structural and electrical characterization. We find that the fabrication method yields a room-temperature hexagonal crystalline structure with the c-axis along the nanowire. Moreover, the obtained nanowires are metallic at low temperature and low sensitive to a strong external magnetic field. Finally, we observe an unknown electron scattering mechanism for small diameters. In conclusion, the one-step metallic flux nanonucleation method yields intermetallic nanowires suitable for both integration in flexible nanoelectronics as well as low-dimensionality transport experiments.
Another promising nanowire composition is Mn5Si3, a silicon-based intermetallic compound that exhibits some ceramic properties such as chemical inertness favoring stability in harsh environments required for nanocables.14,15 As it is commonly observed for other silicide nanowires, Mn5Si3 is expected to be a metallic conductor, while integrable into silicon-based devices. Sun et al. have demonstrated that a SiO2 protective layer around long Mn5Si3 nanocables allows conduction under high temperatures, in acidic, or oxidizing environments and permits good system flexibility without creating defects that would degrade their electrical performance.9
In the bulk form, this alloy presents a complex magnetic structure exhibiting two transitions: a non-collinear antiferromagnetic phase transition (AF1) at T1 ≈ 65 K, followed by a collinear one (AF2) occurring at T2 ≈ 105 K.16–18 Both transitions arise from crystallographic phase transitions: expanded orthorhombic to regular (AF1), until hexagonal (AF2). Due to its itinerant electron character, the Mn5Si3 compound electronic behavior is expected to be strongly influenced by its magnetic order, which is anisotropic below T2. Indeed, Vinokurova et al. have found a strong anisotropic resistivity behavior of Mn5Si3 single crystals by applying a current along the c-axis and in the hexagonal basal plane, the latter exhibiting a stronger non-monotonic temperature evolution.19 Besides, reducing the system dimensions can interplay with the magnetic transitions. For example, Sürgers et al.20 have shown that the AF1 and AF2 transitions are less pronounced in a low dimensionality (2D) 160 nm-thick layer, regarding the bulk counterpart.
Nowadays, the common method to grow manganese silicide nanowires is metal–organic chemical vapor deposition (MOCVD).9,21,22 MOCVD favors a single crystalline structure, unlike for Mn5Si3 nanorods obtained via solid-state reaction route.23 However, to achieve a core–shell configuration with MOCVD, a two-steps fabrication route is required.9 Moreover, a further additional mechanical removal of the nanocables from the substrate is mandatory. Consequently, simplifying the growth of conductive nanocables with a protective layer while maintaining their large aspect ratio and single crystalline nature is necessary.
In this work, we propose a new fabrication route for high quality oxide-shell-protected Mn5Si3 nanowires, called metallic flux nanonucleation method (MFNN).24 This fabrication technique, which combines the metal flux method with nanoporous alumina templates, has already been successfully employed on the fabrication of several intermetallic nanowires.24–27 It is a versatile and user-friendly method that presents a possible high yield of around millions of nanowires per squared millimeters. Here, this one-step nanowire production technique allowed us to reproducibly fabricate high crystalline quality manganese silicide oxide-coated nanowires with length up to several tens of microns. MFNN directly yields a core–shell structure that is highly desirable for nanoelectronics applications and allows benefiting from the promising electrical, mechanical, and chemical characteristics of the material as nanocables in harsh environments. Furthermore, MFNN results in high quality single crystals. This point is critical for nanostructures exhibiting properties arising from the reduced dimensionality, compared to the bulk. In this work, we focused on the dimension effects introduced by the mesoscopic diameter (d = 180 up to 800 nm) on the nanowire electrical behavior around the previously discussed bulk AFM transitions (temperature T = 2–300 K).
After confirming their core–shell and homogeneous structure through energy dispersive X-ray spectroscopy (EDS), we carried out further structural characterizations (selected-area electron diffraction (SAED) and transmission Kikuchi diffraction (TKD)) to investigate the crystalline quality of the obtained nanowires. We subsequently probed single nanowire resistivity behavior varying both the temperature and external magnetic field. Results show that the MFNN fabrication technique consistently yields nanowires with the necessary structural and electrical characteristics to be used as nanocables. In particular, their electrical behavior exhibits no notable effect under the application of a magnetic field up to 7 T. Additionally, preliminary measurements suggest that the diameter reduction modifies their electrical properties. Indeed, the smallest diameter of 180 nm presents the lowest resistivity at 300 K (275 μΩ cm). Furthermore, our data indicate a change in the resistivity mechanism for temperature above the detected magnon–electron scattering regime. Whereas thicker nanowires switch to a phonon–electron scattering regime, the smallest nanowire exhibits as an unidentified regime. This transformation occurs around 45–55 K, at lower temperature for thinner nanowires. In conclusion, we show that the MFNN technique is a promising fabrication route for diverse intermetallic nanowires to be used for nanoelectronics as well as for the investigation of physical phenomena in reduced dimension structures.
Fig. 1a depicts a scanning electron microscopy (SEM) image of the alumina surface after the MFNN process. The straight and long synthesized nanowires with length reaching up to 70 μm are encountered freely lying on top of the alumina template. Since the template pores are solely mediating the nanowire growth, they remained clean and empty after removing the remaining liquid elements (Fig. 1b). The histograms of the template pore and the nanowire diameter distributions, built based on the SEM images, are presented in Fig. 1c and d, respectively. While the average pore diameter is 170 ± 10 nm, we observe that the nanowire diameter does not directly follow the pore size distribution. Instead, the nanowire diameter is found to be between 85 and 850 nm, yielding large length to diameter aspect ratios from 100 up to 740. Remarkably, even for a diameter well above the largest pore diameter, the produced nanowires exhibit a homogenous shape (see ESI, Suppl. 1†). This fabrication process repeatedly yielded similar results.
A typical overview TEM image of a nanowire transferred to a grid is seen in Fig. 1e, along with a higher magnification of the marked area (Fig. 1f). The images reveal the nanowire core–shell configuration through a contrast difference, which should be related to regions of different densities.
![]() | ||
Fig. 3 Mn–Si core microstructural SAED analysis (same nanowire than presented in Fig. 2). (a) and (b) Experimental and (c) and (d) simulated SAED patterns for Mn5Si3 crystalline structure. Angle of the nanowire longitudinal direction in the bright-field TEM image (e) and of the [0002] direction in the respective SAED pattern (f). |
The nanowire microstructure characterization is completed with transmission Kikuchi diffraction (TKD), a powerful technique sensitive to the grain boundary, size, and orientation, as well as the presence of strain, preferred crystallographic orientation (texture), or crystalline phase distribution.33–36 Remarkably, within the technique resolution, the results indicate none of these aforementioned features, demonstrating the high crystalline quality of the synthesized core–shell nanowires by MFNN (see ESI, Suppl. 4†). Therefore, this new fabrication method turns out to be a promising alternative for achieving oxide-protected nanocables of different compositions for future use in nanoelectronics.
d (nm) | ρ (μΩ cm) @ 2 K | RRR | T* (K) | ln(A) |
---|---|---|---|---|
180 | 149.5 ± 3.0 | 2.0 | ≈30 | −21.4 ± 0.2 |
450 | 109.0 ± 2.2 | 4.3 | ≈45 | −21.4 ± 0.1 |
800 | 85.6 ± 4.2 | 4.8 | ≈54 | −21.0 ± 0.2 |
Bulk19 | 50.0 | 4.0 | — | — |
Furthermore, since sensitivity to external perturbations could impair the nanowire use in nanoelectronic devices, we tested their resistivity robustness under a high magnetic field (H) up to 7 T applied either perpendicular or parallel to the crystallographic c-axis. For all three diameters, we do not detect any notable resistivity change (inside a noise level evaluated to 2%) in the whole temperature range investigated (2–300 K), as shown in Fig. 4c and d for d = 450 nm. This behavior strongly differs from what has been reported in bulk Mn5Si3 single crystals, both on the magnetoresistance and transition temperature aspects. With a current flowing along the c-axis, C. Sürgers et al. have observed a magnetoresistance of 13% at about 52 K, when applying a perpendicular (along the b-axis) magnetic field up to 5 T, together with a T1 decrease.38
Since even the larger diameter nanowires exhibit the same behavior, this discrepancy may not be related to dimensionality effects but rather arising from the different synthesis conditions between the systems. As previously mentioned, the supersaturation melt in which the nanowires are grown may yield an inclusion of Ga atoms in the Mn5Si3 core. On the opposite, no doping is expected in the crystals produced in ref. 38, through a combined Bridgman and flux growth method using Mn self-flux. One hypothesis is that these possible Ga atoms could disturb the crystalline structure, which could modify its magnetic properties. Since the magnetic order strongly depends on the Mn–Mn atomic interactions, it can be altered through external or chemical pressure. In bulk single crystals, L. Vinokurova et al. have reported a T1 shift towards lower temperatures under hydrostatical pressure, before being fully suppressed under pressures around 5.5 kbar.39 B. Gopalakrishnan et al. have observed similar results in Mn5Si3Cx thin films, which have switched from antiferromagnetic to ferromagnetic for x = 0.4.40 Such negative chemical pressure has also been reported in MnSi crystals grown in a Ga flux, induced by the atomic substitution of around 1 at% of Ga atoms at the Si site.41 Therefore, the nanowires synthesized through MFNN might be under chemical pressure due to a Ga doping, to the point in which it either destabilizes the antiferromagnetic coupling or suppresses the AF1 transition, yielding adequate electrical performance for low-temperature application in various environments.
Several resistivity sources are temperature-dependent, such as electron–phonon, electron–magnon, and electron–electron scatterings, while the residual resistivity, originating from crystalline defects, is constant in temperature. All these terms contribute to the resistivity, adding up according to Matthiessen's rule.43 Regarding phonon scattering, it produces a resistivity evolution Tn at low temperatures that can either be with n = 3 (magnetic compound) or with n = 5 (non-magnetic material), which both reduce to a linear term at high enough temperature.44–46 On the other hand, both antiferromagnetic magnon–electron (for an isotropic system) and electron–electron scatterings yield a resistivity term that is proportional to the squared temperature, T2.44,47,48
From Fig. 4a, we observe that the resistivity curves are qualitatively similar to those obtained from bulk single crystals measured along the c-axis.19 The respective contributions to the electrical resistivity were extracted through their linearization, which exhibits two regions of interest below 90 K (ESI, Supp. 6† and Fig. 5a and b). The results are summarized in Fig. 5c.
![]() | ||
Fig. 5 (a) and (b) Linearization process for the 180 nm nanowire. (c) Power laws extracted from the data linearization (error bar represented by the height). (d) Resistivity derivative. |
In a first step, we focus on the variation of the transition temperatures between these two resistivity regimes. While there is a net transition for d = 180 and 450 nm, the 800 nm noisier resistivity curve leaves a temperature gap between these regimes (Fig. 5c). Therefore, we opted to define the transition temperature using the resistivity derivative performed after an adjacent-averaging smoothing process (Fig. 5d). We observe that the derivative maximum occurs at a temperature T* either at the linearization transition or inside the temperature interval between the linearized sections. Interestingly, this temperature transition, which may be related to the AF1 bulk transition, increases with the nanowire diameter while remaining below bulk T1 (see Table 1). Several hypotheses may account for the observed trend. On one side, it could be a consequence of the increased proportion of the disordered spins at the surface when increasing the surface to volume ratio. As raised by Kamalakar and co-workers,49 such disordered spins may not support a spin-wave propagation, thus impairing it and favoring a resistivity regime transition at lower temperature. On the other side, remembering that the c-axis coincides with the nanowire one, this expansion may remain sustainable until lower temperature for lower lateral width (i.e. diameter). Probing the temperature evolution of the nanowire magnetization and crystalline structure would help to understand the mechanisms underlying the transition temperature relationship with their dimension.
In a second step, we observe the linearization parameters of the different nanowires (Fig. 5c). In the low-temperature range, all nanowires behave according to an n = 2 mechanism, which arises either from electron–magnon or electron–electron scattering. Similar behavior was reported by Gopalakrishnan et al.40 for a 100 nm-thick polycrystalline Mn5Si3 thin film below 30 K, with a proportionality constant A = 0.6 μΩ cm K−2. They attributed this contribution to magnon–electron scattering, arguing that the resistivity behaviors of the Mn5Si3 and isotropic antiferromagnetic α-Mn compounds are quite similar, the latter having a strong magnon contribution of about 0.1 μΩ cm K−2.44,50 All nanowire A values (≈0.1 μΩ cm K−2) are of the same order of magnitude as those reported in ref. 38 and 44 (see Table 1). Since no diameter dependence is observed, it suggests that a long-range magnetic order occurs in the low-temperature regime for nanowires in the size range investigated.
As the temperature increases, a second regime with a different power law is observed (Fig. 5c). For both larger diameters (d = 450 and 800 nm), the resistivity can be assumed to be linear in this region. This behavior is related to an electron–phonon scattering process, meaning that the lattice vibration contribution becomes preponderant over other resistivity sources. In the interval between T1 and T2, a resistivity increase was also observed for both bulk and thin films of Mn5Si3, but not confirmed as being linear.19,40 However, this trend differs when reducing the diameter until 180 nm. First, the temperature region ends at lower temperature than for larger diameters (52 K vs. 88–90 K, respectively). Moreover, the extracted temperature exponent (n = 1.3 ± 0.1) does not indicate a physical scattering process. Further specific measurements are required to conclude on the consequences of the core–shell nanowire dimensionality reduction in this temperature interval. For example, we already observed a transformation of the magnetic order, from AFM to ferrimagnetic or a mixture of ferromagnetic and AFM, in the intermediary temperature region of Fe3Ga4 nanowires of similar diameter.26
On the other hand, this resistivity contribution modification could explain the smallest nanowire unexpectedly exhibiting the lowest resistivity compared to larger nanowires (see Fig. 4a), especially since it presents the highest residual resistivity and a RRR value less than half than for larger nanowires. From an application point of view, this electrical performance, coupled with a resistivity remaining almost constant for temperatures down to 100 K, yields the d = 180 nm nanowire as the best choice for a nanocable in this temperature range.
By investigating the electron scattering mechanism, we observed that the nanowire diameter reduction to 180 nm modifies the resistivity scattering contributions. At lower temperatures, all investigated nanowires exhibit a magnon–electron contribution. For diameters above 450 nm, it transforms into a phonon scattering mechanism. However, it evolves in an n = 1.3 ± 0.1 behavior for the smallest diameter (180 nm), which does not directly relate to a physical process and may indicate a more complex magnetic order. Remarkably, it presents the lowest resistivity value from 300 K down to 100 K, despite a larger residual resistivity and smaller RRR values than for larger nanowires, as expected. These preliminary observations should be completed by more specific phonon measurements to understand the consequences arising from the diameter reduction. In conclusion, we believe that the metallic flux nanonucleation method is promising for easily producing high quality intermetallic nanowires, either for flexible nanoelectronics or low-dimensionality investigation of emerging physical phenomena.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/d0na00809e |
This journal is © The Royal Society of Chemistry 2021 |